feat(auth): implement credential management modals and hooks

- Added  component for handling delete confirmation and edit modals.
- Introduced  and  for displaying credentials.
- Created custom hooks  and  to manage credential actions and modal states.
- Refactored  component to utilize new hooks and components for improved readability and maintainability.

.claude/settings.json

@@ -5,5 +5,18 @@
     "typescript-lsp@claude-plugins-official": true,
     "pyright-lsp@claude-plugins-official": true,
     "ralph-loop@claude-plugins-official": true
+  },
+  "hooks": {
+    "PreToolUse": [
+      {
+        "matcher": "Bash",
+        "hooks": [
+          {
+            "type": "command",
+            "command": "npx -y block-no-verify@1.1.1"
+          }
+        ]
+      }
+    ]
   }
 }

.github/workflows/api-tests.yml

@@ -39,12 +39,6 @@ jobs:
       - name: Install dependencies
         run: uv sync --project api --dev
 
-      - name: Run pyrefly check
-        run: |
-          cd api
-          uv add --dev pyrefly
-          uv run pyrefly check || true
-
       - name: Run dify config tests
         run: uv run --project api dev/pytest/pytest_config_tests.py
 

.github/workflows/deploy-hitl.yml

@@ -0,0 +1,29 @@
+name: Deploy HITL
+
+on:
+  workflow_run:
+    workflows: ["Build and Push API & Web"]
+    branches:
+      - "feat/hitl-frontend"
+      - "feat/hitl-backend"
+    types:
+      - completed
+
+jobs:
+  deploy:
+    runs-on: ubuntu-latest
+    if: |
+      github.event.workflow_run.conclusion == 'success' &&
+      (
+        github.event.workflow_run.head_branch == 'feat/hitl-frontend' ||
+        github.event.workflow_run.head_branch == 'feat/hitl-backend'
+      )
+    steps:
+      - name: Deploy to server
+        uses: appleboy/ssh-action@v0.1.8
+        with:
+          host: ${{ secrets.HITL_SSH_HOST }}
+          username: ${{ secrets.SSH_USER }}
+          key: ${{ secrets.SSH_PRIVATE_KEY }}
+          script: |
+            ${{ vars.SSH_SCRIPT || secrets.SSH_SCRIPT }}

.nvmrc

@@ -1 +0,0 @@
-22.11.0

api/.env.example

@@ -589,6 +589,7 @@ ENABLE_CLEAN_UNUSED_DATASETS_TASK=false
 ENABLE_CREATE_TIDB_SERVERLESS_TASK=false
 ENABLE_UPDATE_TIDB_SERVERLESS_STATUS_TASK=false
 ENABLE_CLEAN_MESSAGES=false
+ENABLE_WORKFLOW_RUN_CLEANUP_TASK=false
 ENABLE_MAIL_CLEAN_DOCUMENT_NOTIFY_TASK=false
 ENABLE_DATASETS_QUEUE_MONITOR=false
 ENABLE_CHECK_UPGRADABLE_PLUGIN_TASK=true

api/commands.py

@@ -1,4 +1,5 @@
 import base64
+import datetime
 import json
 import logging
 import secrets
@@ -34,7 +35,7 @@ from libs.rsa import generate_key_pair
 from models import Tenant
 from models.dataset import Dataset, DatasetCollectionBinding, DatasetMetadata, DatasetMetadataBinding, DocumentSegment
 from models.dataset import Document as DatasetDocument
-from models.model import Account, App, AppAnnotationSetting, AppMode, Conversation, MessageAnnotation, UploadFile
+from models.model import App, AppAnnotationSetting, AppMode, Conversation, MessageAnnotation, UploadFile
 from models.oauth import DatasourceOauthParamConfig, DatasourceProvider
 from models.provider import Provider, ProviderModel
 from models.provider_ids import DatasourceProviderID, ToolProviderID
@@ -45,6 +46,7 @@ from services.clear_free_plan_tenant_expired_logs import ClearFreePlanTenantExpi
 from services.plugin.data_migration import PluginDataMigration
 from services.plugin.plugin_migration import PluginMigration
 from services.plugin.plugin_service import PluginService
+from services.retention.workflow_run.clear_free_plan_expired_workflow_run_logs import WorkflowRunCleanup
 from tasks.remove_app_and_related_data_task import delete_draft_variables_batch
 
 logger = logging.getLogger(__name__)
@@ -62,8 +64,10 @@ def reset_password(email, new_password, password_confirm):
     if str(new_password).strip() != str(password_confirm).strip():
         click.echo(click.style("Passwords do not match.", fg="red"))
         return
+    normalized_email = email.strip().lower()
+
     with sessionmaker(db.engine, expire_on_commit=False).begin() as session:
-        account = session.query(Account).where(Account.email == email).one_or_none()
+        account = AccountService.get_account_by_email_with_case_fallback(email.strip(), session=session)
 
         if not account:
             click.echo(click.style(f"Account not found for email: {email}", fg="red"))
@@ -84,7 +88,7 @@ def reset_password(email, new_password, password_confirm):
         base64_password_hashed = base64.b64encode(password_hashed).decode()
         account.password = base64_password_hashed
         account.password_salt = base64_salt
-        AccountService.reset_login_error_rate_limit(email)
+        AccountService.reset_login_error_rate_limit(normalized_email)
         click.echo(click.style("Password reset successfully.", fg="green"))
 
 
@@ -100,20 +104,22 @@ def reset_email(email, new_email, email_confirm):
     if str(new_email).strip() != str(email_confirm).strip():
         click.echo(click.style("New emails do not match.", fg="red"))
         return
+    normalized_new_email = new_email.strip().lower()
+
     with sessionmaker(db.engine, expire_on_commit=False).begin() as session:
-        account = session.query(Account).where(Account.email == email).one_or_none()
+        account = AccountService.get_account_by_email_with_case_fallback(email.strip(), session=session)
 
         if not account:
             click.echo(click.style(f"Account not found for email: {email}", fg="red"))
             return
 
         try:
-            email_validate(new_email)
+            email_validate(normalized_new_email)
         except:
             click.echo(click.style(f"Invalid email: {new_email}", fg="red"))
             return
 
-        account.email = new_email
+        account.email = normalized_new_email
         click.echo(click.style("Email updated successfully.", fg="green"))
 
 
@@ -658,7 +664,7 @@ def create_tenant(email: str, language: str | None = None, name: str | None = No
         return
 
     # Create account
-    email = email.strip()
+    email = email.strip().lower()
 
     if "@" not in email:
         click.echo(click.style("Invalid email address.", fg="red"))
@@ -852,6 +858,61 @@ def clear_free_plan_tenant_expired_logs(days: int, batch: int, tenant_ids: list[
     click.echo(click.style("Clear free plan tenant expired logs completed.", fg="green"))
 
 
+@click.command("clean-workflow-runs", help="Clean expired workflow runs and related data for free tenants.")
+@click.option("--days", default=30, show_default=True, help="Delete workflow runs created before N days ago.")
+@click.option("--batch-size", default=200, show_default=True, help="Batch size for selecting workflow runs.")
+@click.option(
+    "--start-from",
+    type=click.DateTime(formats=["%Y-%m-%d", "%Y-%m-%dT%H:%M:%S"]),
+    default=None,
+    help="Optional lower bound (inclusive) for created_at; must be paired with --end-before.",
+)
+@click.option(
+    "--end-before",
+    type=click.DateTime(formats=["%Y-%m-%d", "%Y-%m-%dT%H:%M:%S"]),
+    default=None,
+    help="Optional upper bound (exclusive) for created_at; must be paired with --start-from.",
+)
+@click.option(
+    "--dry-run",
+    is_flag=True,
+    help="Preview cleanup results without deleting any workflow run data.",
+)
+def clean_workflow_runs(
+    days: int,
+    batch_size: int,
+    start_from: datetime.datetime | None,
+    end_before: datetime.datetime | None,
+    dry_run: bool,
+):
+    """
+    Clean workflow runs and related workflow data for free tenants.
+    """
+    if (start_from is None) ^ (end_before is None):
+        raise click.UsageError("--start-from and --end-before must be provided together.")
+
+    start_time = datetime.datetime.now(datetime.UTC)
+    click.echo(click.style(f"Starting workflow run cleanup at {start_time.isoformat()}.", fg="white"))
+
+    WorkflowRunCleanup(
+        days=days,
+        batch_size=batch_size,
+        start_from=start_from,
+        end_before=end_before,
+        dry_run=dry_run,
+    ).run()
+
+    end_time = datetime.datetime.now(datetime.UTC)
+    elapsed = end_time - start_time
+    click.echo(
+        click.style(
+            f"Workflow run cleanup completed. start={start_time.isoformat()} "
+            f"end={end_time.isoformat()} duration={elapsed}",
+            fg="green",
+        )
+    )
+
+
 @click.option("-f", "--force", is_flag=True, help="Skip user confirmation and force the command to execute.")
 @click.command("clear-orphaned-file-records", help="Clear orphaned file records.")
 def clear_orphaned_file_records(force: bool):

api/configs/feature/__init__.py

@@ -1101,6 +1101,10 @@ class CeleryScheduleTasksConfig(BaseSettings):
         description="Enable clean messages task",
         default=False,
     )
+    ENABLE_WORKFLOW_RUN_CLEANUP_TASK: bool = Field(
+        description="Enable scheduled workflow run cleanup task",
+        default=False,
+    )
     ENABLE_MAIL_CLEAN_DOCUMENT_NOTIFY_TASK: bool = Field(
         description="Enable mail clean document notify task",
         default=False,

api/configs/middleware/storage/volcengine_tos_storage_config.py

@@ -4,7 +4,7 @@ from pydantic_settings import BaseSettings
 
 class VolcengineTOSStorageConfig(BaseSettings):
     """
-    Configuration settings for Volcengine Tinder Object Storage (TOS)
+    Configuration settings for Volcengine Torch Object Storage (TOS)
     """
 
     VOLCENGINE_TOS_BUCKET_NAME: str | None = Field(

api/controllers/console/auth/activate.py

@@ -63,10 +63,9 @@ class ActivateCheckApi(Resource):
         args = ActivateCheckQuery.model_validate(request.args.to_dict(flat=True))  # type: ignore
 
         workspaceId = args.workspace_id
-        reg_email = args.email
         token = args.token
 
-        invitation = RegisterService.get_invitation_if_token_valid(workspaceId, reg_email, token)
+        invitation = RegisterService.get_invitation_with_case_fallback(workspaceId, args.email, token)
         if invitation:
             data = invitation.get("data", {})
             tenant = invitation.get("tenant", None)
@@ -100,11 +99,12 @@ class ActivateApi(Resource):
     def post(self):
         args = ActivatePayload.model_validate(console_ns.payload)
 
-        invitation = RegisterService.get_invitation_if_token_valid(args.workspace_id, args.email, args.token)
+        normalized_request_email = args.email.lower() if args.email else None
+        invitation = RegisterService.get_invitation_with_case_fallback(args.workspace_id, args.email, args.token)
         if invitation is None:
             raise AlreadyActivateError()
 
-        RegisterService.revoke_token(args.workspace_id, args.email, args.token)
+        RegisterService.revoke_token(args.workspace_id, normalized_request_email, args.token)
 
         account = invitation["account"]
         account.name = args.name

api/controllers/console/auth/email_register.py

@@ -1,7 +1,6 @@
 from flask import request
 from flask_restx import Resource
 from pydantic import BaseModel, Field, field_validator
-from sqlalchemy import select
 from sqlalchemy.orm import Session
 
 from configs import dify_config
@@ -62,6 +61,7 @@ class EmailRegisterSendEmailApi(Resource):
     @email_register_enabled
     def post(self):
         args = EmailRegisterSendPayload.model_validate(console_ns.payload)
+        normalized_email = args.email.lower()
 
         ip_address = extract_remote_ip(request)
         if AccountService.is_email_send_ip_limit(ip_address):
@@ -70,13 +70,12 @@ class EmailRegisterSendEmailApi(Resource):
         if args.language in languages:
             language = args.language
 
-        if dify_config.BILLING_ENABLED and BillingService.is_email_in_freeze(args.email):
+        if dify_config.BILLING_ENABLED and BillingService.is_email_in_freeze(normalized_email):
             raise AccountInFreezeError()
 
         with Session(db.engine) as session:
-            account = session.execute(select(Account).filter_by(email=args.email)).scalar_one_or_none()
-        token = None
-        token = AccountService.send_email_register_email(email=args.email, account=account, language=language)
+            account = AccountService.get_account_by_email_with_case_fallback(args.email, session=session)
+        token = AccountService.send_email_register_email(email=normalized_email, account=account, language=language)
         return {"result": "success", "data": token}
 
 
@@ -88,9 +87,9 @@ class EmailRegisterCheckApi(Resource):
     def post(self):
         args = EmailRegisterValidityPayload.model_validate(console_ns.payload)
 
-        user_email = args.email
+        user_email = args.email.lower()
 
-        is_email_register_error_rate_limit = AccountService.is_email_register_error_rate_limit(args.email)
+        is_email_register_error_rate_limit = AccountService.is_email_register_error_rate_limit(user_email)
         if is_email_register_error_rate_limit:
             raise EmailRegisterLimitError()
 
@@ -98,11 +97,14 @@ class EmailRegisterCheckApi(Resource):
         if token_data is None:
             raise InvalidTokenError()
 
-        if user_email != token_data.get("email"):
+        token_email = token_data.get("email")
+        normalized_token_email = token_email.lower() if isinstance(token_email, str) else token_email
+
+        if user_email != normalized_token_email:
             raise InvalidEmailError()
 
         if args.code != token_data.get("code"):
-            AccountService.add_email_register_error_rate_limit(args.email)
+            AccountService.add_email_register_error_rate_limit(user_email)
             raise EmailCodeError()
 
         # Verified, revoke the first token
@@ -113,8 +115,8 @@ class EmailRegisterCheckApi(Resource):
             user_email, code=args.code, additional_data={"phase": "register"}
         )
 
-        AccountService.reset_email_register_error_rate_limit(args.email)
-        return {"is_valid": True, "email": token_data.get("email"), "token": new_token}
+        AccountService.reset_email_register_error_rate_limit(user_email)
+        return {"is_valid": True, "email": normalized_token_email, "token": new_token}
 
 
 @console_ns.route("/email-register")
@@ -141,22 +143,23 @@ class EmailRegisterResetApi(Resource):
         AccountService.revoke_email_register_token(args.token)
 
         email = register_data.get("email", "")
+        normalized_email = email.lower()
 
         with Session(db.engine) as session:
-            account = session.execute(select(Account).filter_by(email=email)).scalar_one_or_none()
+            account = AccountService.get_account_by_email_with_case_fallback(email, session=session)
 
             if account:
                 raise EmailAlreadyInUseError()
             else:
-                account = self._create_new_account(email, args.password_confirm)
+                account = self._create_new_account(normalized_email, args.password_confirm)
                 if not account:
                     raise AccountNotFoundError()
                 token_pair = AccountService.login(account=account, ip_address=extract_remote_ip(request))
-                AccountService.reset_login_error_rate_limit(email)
+                AccountService.reset_login_error_rate_limit(normalized_email)
 
         return {"result": "success", "data": token_pair.model_dump()}
 
-    def _create_new_account(self, email, password) -> Account | None:
+    def _create_new_account(self, email: str, password: str) -> Account | None:
         # Create new account if allowed
         account = None
         try:

api/controllers/console/auth/forgot_password.py

@@ -4,7 +4,6 @@ import secrets
 from flask import request
 from flask_restx import Resource, fields
 from pydantic import BaseModel, Field, field_validator
-from sqlalchemy import select
 from sqlalchemy.orm import Session
 
 from controllers.console import console_ns
@@ -21,7 +20,6 @@ from events.tenant_event import tenant_was_created
 from extensions.ext_database import db
 from libs.helper import EmailStr, extract_remote_ip
 from libs.password import hash_password, valid_password
-from models import Account
 from services.account_service import AccountService, TenantService
 from services.feature_service import FeatureService
 
@@ -76,6 +74,7 @@ class ForgotPasswordSendEmailApi(Resource):
     @email_password_login_enabled
     def post(self):
         args = ForgotPasswordSendPayload.model_validate(console_ns.payload)
+        normalized_email = args.email.lower()
 
         ip_address = extract_remote_ip(request)
         if AccountService.is_email_send_ip_limit(ip_address):
@@ -87,11 +86,11 @@ class ForgotPasswordSendEmailApi(Resource):
             language = "en-US"
 
         with Session(db.engine) as session:
-            account = session.execute(select(Account).filter_by(email=args.email)).scalar_one_or_none()
+            account = AccountService.get_account_by_email_with_case_fallback(args.email, session=session)
 
         token = AccountService.send_reset_password_email(
             account=account,
-            email=args.email,
+            email=normalized_email,
             language=language,
             is_allow_register=FeatureService.get_system_features().is_allow_register,
         )
@@ -122,9 +121,9 @@ class ForgotPasswordCheckApi(Resource):
     def post(self):
         args = ForgotPasswordCheckPayload.model_validate(console_ns.payload)
 
-        user_email = args.email
+        user_email = args.email.lower()
 
-        is_forgot_password_error_rate_limit = AccountService.is_forgot_password_error_rate_limit(args.email)
+        is_forgot_password_error_rate_limit = AccountService.is_forgot_password_error_rate_limit(user_email)
         if is_forgot_password_error_rate_limit:
             raise EmailPasswordResetLimitError()
 
@@ -132,11 +131,16 @@ class ForgotPasswordCheckApi(Resource):
         if token_data is None:
             raise InvalidTokenError()
 
-        if user_email != token_data.get("email"):
+        token_email = token_data.get("email")
+        if not isinstance(token_email, str):
+            raise InvalidEmailError()
+        normalized_token_email = token_email.lower()
+
+        if user_email != normalized_token_email:
             raise InvalidEmailError()
 
         if args.code != token_data.get("code"):
-            AccountService.add_forgot_password_error_rate_limit(args.email)
+            AccountService.add_forgot_password_error_rate_limit(user_email)
             raise EmailCodeError()
 
         # Verified, revoke the first token
@@ -144,11 +148,11 @@ class ForgotPasswordCheckApi(Resource):
 
         # Refresh token data by generating a new token
         _, new_token = AccountService.generate_reset_password_token(
-            user_email, code=args.code, additional_data={"phase": "reset"}
+            token_email, code=args.code, additional_data={"phase": "reset"}
         )
 
-        AccountService.reset_forgot_password_error_rate_limit(args.email)
-        return {"is_valid": True, "email": token_data.get("email"), "token": new_token}
+        AccountService.reset_forgot_password_error_rate_limit(user_email)
+        return {"is_valid": True, "email": normalized_token_email, "token": new_token}
 
 
 @console_ns.route("/forgot-password/resets")
@@ -187,9 +191,8 @@ class ForgotPasswordResetApi(Resource):
         password_hashed = hash_password(args.new_password, salt)
 
         email = reset_data.get("email", "")
-
         with Session(db.engine) as session:
-            account = session.execute(select(Account).filter_by(email=email)).scalar_one_or_none()
+            account = AccountService.get_account_by_email_with_case_fallback(email, session=session)
 
             if account:
                 self._update_existing_account(account, password_hashed, salt, session)

api/controllers/console/auth/login.py

@@ -90,32 +90,38 @@ class LoginApi(Resource):
     def post(self):
         """Authenticate user and login."""
         args = LoginPayload.model_validate(console_ns.payload)
+        request_email = args.email
+        normalized_email = request_email.lower()
 
-        if dify_config.BILLING_ENABLED and BillingService.is_email_in_freeze(args.email):
+        if dify_config.BILLING_ENABLED and BillingService.is_email_in_freeze(normalized_email):
             raise AccountInFreezeError()
 
-        is_login_error_rate_limit = AccountService.is_login_error_rate_limit(args.email)
+        is_login_error_rate_limit = AccountService.is_login_error_rate_limit(normalized_email)
         if is_login_error_rate_limit:
             raise EmailPasswordLoginLimitError()
 
+        invite_token = args.invite_token
         invitation_data: dict[str, Any] | None = None
-        if args.invite_token:
-            invitation_data = RegisterService.get_invitation_if_token_valid(None, args.email, args.invite_token)
+        if invite_token:
+            invitation_data = RegisterService.get_invitation_with_case_fallback(None, request_email, invite_token)
+            if invitation_data is None:
+                invite_token = None
 
         try:
             if invitation_data:
                 data = invitation_data.get("data", {})
                 invitee_email = data.get("email") if data else None
-                if invitee_email != args.email:
+                invitee_email_normalized = invitee_email.lower() if isinstance(invitee_email, str) else invitee_email
+                if invitee_email_normalized != normalized_email:
                     raise InvalidEmailError()
-                account = AccountService.authenticate(args.email, args.password, args.invite_token)
-            else:
-                account = AccountService.authenticate(args.email, args.password)
+            account = _authenticate_account_with_case_fallback(
+                request_email, normalized_email, args.password, invite_token
+            )
         except services.errors.account.AccountLoginError:
             raise AccountBannedError()
-        except services.errors.account.AccountPasswordError:
-            AccountService.add_login_error_rate_limit(args.email)
-            raise AuthenticationFailedError()
+        except services.errors.account.AccountPasswordError as exc:
+            AccountService.add_login_error_rate_limit(normalized_email)
+            raise AuthenticationFailedError() from exc
         # SELF_HOSTED only have one workspace
         tenants = TenantService.get_join_tenants(account)
         if len(tenants) == 0:
@@ -130,7 +136,7 @@ class LoginApi(Resource):
                 }
 
         token_pair = AccountService.login(account=account, ip_address=extract_remote_ip(request))
-        AccountService.reset_login_error_rate_limit(args.email)
+        AccountService.reset_login_error_rate_limit(normalized_email)
 
         # Create response with cookies instead of returning tokens in body
         response = make_response({"result": "success"})
@@ -170,18 +176,19 @@ class ResetPasswordSendEmailApi(Resource):
     @console_ns.expect(console_ns.models[EmailPayload.__name__])
     def post(self):
         args = EmailPayload.model_validate(console_ns.payload)
+        normalized_email = args.email.lower()
 
         if args.language is not None and args.language == "zh-Hans":
             language = "zh-Hans"
         else:
             language = "en-US"
         try:
-            account = AccountService.get_user_through_email(args.email)
+            account = _get_account_with_case_fallback(args.email)
         except AccountRegisterError:
             raise AccountInFreezeError()
 
         token = AccountService.send_reset_password_email(
-            email=args.email,
+            email=normalized_email,
             account=account,
             language=language,
             is_allow_register=FeatureService.get_system_features().is_allow_register,
@@ -196,6 +203,7 @@ class EmailCodeLoginSendEmailApi(Resource):
     @console_ns.expect(console_ns.models[EmailPayload.__name__])
     def post(self):
         args = EmailPayload.model_validate(console_ns.payload)
+        normalized_email = args.email.lower()
 
         ip_address = extract_remote_ip(request)
         if AccountService.is_email_send_ip_limit(ip_address):
@@ -206,13 +214,13 @@ class EmailCodeLoginSendEmailApi(Resource):
         else:
             language = "en-US"
         try:
-            account = AccountService.get_user_through_email(args.email)
+            account = _get_account_with_case_fallback(args.email)
         except AccountRegisterError:
             raise AccountInFreezeError()
 
         if account is None:
             if FeatureService.get_system_features().is_allow_register:
-                token = AccountService.send_email_code_login_email(email=args.email, language=language)
+                token = AccountService.send_email_code_login_email(email=normalized_email, language=language)
             else:
                 raise AccountNotFound()
         else:
@@ -229,14 +237,17 @@ class EmailCodeLoginApi(Resource):
     def post(self):
         args = EmailCodeLoginPayload.model_validate(console_ns.payload)
 
-        user_email = args.email
+        original_email = args.email
+        user_email = original_email.lower()
         language = args.language
 
         token_data = AccountService.get_email_code_login_data(args.token)
         if token_data is None:
             raise InvalidTokenError()
 
-        if token_data["email"] != args.email:
+        token_email = token_data.get("email")
+        normalized_token_email = token_email.lower() if isinstance(token_email, str) else token_email
+        if normalized_token_email != user_email:
             raise InvalidEmailError()
 
         if token_data["code"] != args.code:
@@ -244,7 +255,7 @@ class EmailCodeLoginApi(Resource):
 
         AccountService.revoke_email_code_login_token(args.token)
         try:
-            account = AccountService.get_user_through_email(user_email)
+            account = _get_account_with_case_fallback(original_email)
         except AccountRegisterError:
             raise AccountInFreezeError()
         if account:
@@ -275,7 +286,7 @@ class EmailCodeLoginApi(Resource):
             except WorkspacesLimitExceededError:
                 raise WorkspacesLimitExceeded()
         token_pair = AccountService.login(account, ip_address=extract_remote_ip(request))
-        AccountService.reset_login_error_rate_limit(args.email)
+        AccountService.reset_login_error_rate_limit(user_email)
 
         # Create response with cookies instead of returning tokens in body
         response = make_response({"result": "success"})
@@ -309,3 +320,22 @@ class RefreshTokenApi(Resource):
             return response
         except Exception as e:
             return {"result": "fail", "message": str(e)}, 401
+
+
+def _get_account_with_case_fallback(email: str):
+    account = AccountService.get_user_through_email(email)
+    if account or email == email.lower():
+        return account
+
+    return AccountService.get_user_through_email(email.lower())
+
+
+def _authenticate_account_with_case_fallback(
+    original_email: str, normalized_email: str, password: str, invite_token: str | None
+):
+    try:
+        return AccountService.authenticate(original_email, password, invite_token)
+    except services.errors.account.AccountPasswordError:
+        if original_email == normalized_email:
+            raise
+        return AccountService.authenticate(normalized_email, password, invite_token)

api/controllers/console/auth/oauth.py

@@ -3,7 +3,6 @@ import logging
 import httpx
 from flask import current_app, redirect, request
 from flask_restx import Resource
-from sqlalchemy import select
 from sqlalchemy.orm import Session
 from werkzeug.exceptions import Unauthorized
 
@@ -118,7 +117,10 @@ class OAuthCallback(Resource):
             invitation = RegisterService.get_invitation_by_token(token=invite_token)
             if invitation:
                 invitation_email = invitation.get("email", None)
-                if invitation_email != user_info.email:
+                invitation_email_normalized = (
+                    invitation_email.lower() if isinstance(invitation_email, str) else invitation_email
+                )
+                if invitation_email_normalized != user_info.email.lower():
                     return redirect(f"{dify_config.CONSOLE_WEB_URL}/signin?message=Invalid invitation token.")
 
             return redirect(f"{dify_config.CONSOLE_WEB_URL}/signin/invite-settings?invite_token={invite_token}")
@@ -175,7 +177,7 @@ def _get_account_by_openid_or_email(provider: str, user_info: OAuthUserInfo) ->
 
     if not account:
         with Session(db.engine) as session:
-            account = session.execute(select(Account).filter_by(email=user_info.email)).scalar_one_or_none()
+            account = AccountService.get_account_by_email_with_case_fallback(user_info.email, session=session)
 
     return account
 
@@ -197,9 +199,10 @@ def _generate_account(provider: str, user_info: OAuthUserInfo) -> tuple[Account,
                 tenant_was_created.send(new_tenant)
 
     if not account:
+        normalized_email = user_info.email.lower()
         oauth_new_user = True
         if not FeatureService.get_system_features().is_allow_register:
-            if dify_config.BILLING_ENABLED and BillingService.is_email_in_freeze(user_info.email):
+            if dify_config.BILLING_ENABLED and BillingService.is_email_in_freeze(normalized_email):
                 raise AccountRegisterError(
                     description=(
                         "This email account has been deleted within the past "
@@ -210,7 +213,11 @@ def _generate_account(provider: str, user_info: OAuthUserInfo) -> tuple[Account,
                 raise AccountRegisterError(description=("Invalid email or password"))
         account_name = user_info.name or "Dify"
         account = RegisterService.register(
-            email=user_info.email, name=account_name, password=None, open_id=user_info.id, provider=provider
+            email=normalized_email,
+            name=account_name,
+            password=None,
+            open_id=user_info.id,
+            provider=provider,
         )
 
         # Set interface language

api/controllers/console/datasets/datasets_document.py

@@ -7,7 +7,7 @@ from typing import Literal, cast
 import sqlalchemy as sa
 from flask import request
 from flask_restx import Resource, fields, marshal, marshal_with
-from pydantic import BaseModel
+from pydantic import BaseModel, Field
 from sqlalchemy import asc, desc, select
 from werkzeug.exceptions import Forbidden, NotFound
 
@@ -104,6 +104,15 @@ class DocumentRenamePayload(BaseModel):
     name: str
 
 
+class DocumentDatasetListParam(BaseModel):
+    page: int = Field(1, title="Page", description="Page number.")
+    limit: int = Field(20, title="Limit", description="Page size.")
+    search: str | None = Field(None, alias="keyword", title="Search", description="Search keyword.")
+    sort_by: str = Field("-created_at", alias="sort", title="SortBy", description="Sort by field.")
+    status: str | None = Field(None, title="Status", description="Document status.")
+    fetch_val: str = Field("false", alias="fetch")
+
+
 register_schema_models(
     console_ns,
     KnowledgeConfig,
@@ -225,14 +234,16 @@ class DatasetDocumentListApi(Resource):
     def get(self, dataset_id):
         current_user, current_tenant_id = current_account_with_tenant()
         dataset_id = str(dataset_id)
-        page = request.args.get("page", default=1, type=int)
-        limit = request.args.get("limit", default=20, type=int)
-        search = request.args.get("keyword", default=None, type=str)
-        sort = request.args.get("sort", default="-created_at", type=str)
-        status = request.args.get("status", default=None, type=str)
+        raw_args = request.args.to_dict()
+        param = DocumentDatasetListParam.model_validate(raw_args)
+        page = param.page
+        limit = param.limit
+        search = param.search
+        sort = param.sort_by
+        status = param.status
         # "yes", "true", "t", "y", "1" convert to True, while others convert to False.
         try:
-            fetch_val = request.args.get("fetch", default="false")
+            fetch_val = param.fetch_val
             if isinstance(fetch_val, bool):
                 fetch = fetch_val
             else:

api/controllers/console/setup.py

@@ -84,10 +84,11 @@ class SetupApi(Resource):
             raise NotInitValidateError()
 
         args = SetupRequestPayload.model_validate(console_ns.payload)
+        normalized_email = args.email.lower()
 
         # setup
         RegisterService.setup(
-            email=args.email,
+            email=normalized_email,
             name=args.name,
             password=args.password,
             ip_address=extract_remote_ip(request),

api/controllers/console/workspace/account.py

@@ -41,7 +41,7 @@ from fields.member_fields import account_fields
 from libs.datetime_utils import naive_utc_now
 from libs.helper import EmailStr, TimestampField, extract_remote_ip, timezone
 from libs.login import current_account_with_tenant, login_required
-from models import Account, AccountIntegrate, InvitationCode
+from models import AccountIntegrate, InvitationCode
 from services.account_service import AccountService
 from services.billing_service import BillingService
 from services.errors.account import CurrentPasswordIncorrectError as ServiceCurrentPasswordIncorrectError
@@ -536,7 +536,8 @@ class ChangeEmailSendEmailApi(Resource):
         else:
             language = "en-US"
         account = None
-        user_email = args.email
+        user_email = None
+        email_for_sending = args.email.lower()
         if args.phase is not None and args.phase == "new_email":
             if args.token is None:
                 raise InvalidTokenError()
@@ -546,16 +547,24 @@ class ChangeEmailSendEmailApi(Resource):
                 raise InvalidTokenError()
             user_email = reset_data.get("email", "")
 
-            if user_email != current_user.email:
+            if user_email.lower() != current_user.email.lower():
                 raise InvalidEmailError()
+
+            user_email = current_user.email
         else:
             with Session(db.engine) as session:
-                account = session.execute(select(Account).filter_by(email=args.email)).scalar_one_or_none()
+                account = AccountService.get_account_by_email_with_case_fallback(args.email, session=session)
             if account is None:
                 raise AccountNotFound()
+            email_for_sending = account.email
+            user_email = account.email
 
         token = AccountService.send_change_email_email(
-            account=account, email=args.email, old_email=user_email, language=language, phase=args.phase
+            account=account,
+            email=email_for_sending,
+            old_email=user_email,
+            language=language,
+            phase=args.phase,
         )
         return {"result": "success", "data": token}
 
@@ -571,9 +580,9 @@ class ChangeEmailCheckApi(Resource):
         payload = console_ns.payload or {}
         args = ChangeEmailValidityPayload.model_validate(payload)
 
-        user_email = args.email
+        user_email = args.email.lower()
 
-        is_change_email_error_rate_limit = AccountService.is_change_email_error_rate_limit(args.email)
+        is_change_email_error_rate_limit = AccountService.is_change_email_error_rate_limit(user_email)
         if is_change_email_error_rate_limit:
             raise EmailChangeLimitError()
 
@@ -581,11 +590,13 @@ class ChangeEmailCheckApi(Resource):
         if token_data is None:
             raise InvalidTokenError()
 
-        if user_email != token_data.get("email"):
+        token_email = token_data.get("email")
+        normalized_token_email = token_email.lower() if isinstance(token_email, str) else token_email
+        if user_email != normalized_token_email:
             raise InvalidEmailError()
 
         if args.code != token_data.get("code"):
-            AccountService.add_change_email_error_rate_limit(args.email)
+            AccountService.add_change_email_error_rate_limit(user_email)
             raise EmailCodeError()
 
         # Verified, revoke the first token
@@ -596,8 +607,8 @@ class ChangeEmailCheckApi(Resource):
             user_email, code=args.code, old_email=token_data.get("old_email"), additional_data={}
         )
 
-        AccountService.reset_change_email_error_rate_limit(args.email)
-        return {"is_valid": True, "email": token_data.get("email"), "token": new_token}
+        AccountService.reset_change_email_error_rate_limit(user_email)
+        return {"is_valid": True, "email": normalized_token_email, "token": new_token}
 
 
 @console_ns.route("/account/change-email/reset")
@@ -611,11 +622,12 @@ class ChangeEmailResetApi(Resource):
     def post(self):
         payload = console_ns.payload or {}
         args = ChangeEmailResetPayload.model_validate(payload)
+        normalized_new_email = args.new_email.lower()
 
-        if AccountService.is_account_in_freeze(args.new_email):
+        if AccountService.is_account_in_freeze(normalized_new_email):
             raise AccountInFreezeError()
 
-        if not AccountService.check_email_unique(args.new_email):
+        if not AccountService.check_email_unique(normalized_new_email):
             raise EmailAlreadyInUseError()
 
         reset_data = AccountService.get_change_email_data(args.token)
@@ -626,13 +638,13 @@ class ChangeEmailResetApi(Resource):
 
         old_email = reset_data.get("old_email", "")
         current_user, _ = current_account_with_tenant()
-        if current_user.email != old_email:
+        if current_user.email.lower() != old_email.lower():
             raise AccountNotFound()
 
-        updated_account = AccountService.update_account_email(current_user, email=args.new_email)
+        updated_account = AccountService.update_account_email(current_user, email=normalized_new_email)
 
         AccountService.send_change_email_completed_notify_email(
-            email=args.new_email,
+            email=normalized_new_email,
         )
 
         return updated_account
@@ -645,8 +657,9 @@ class CheckEmailUnique(Resource):
     def post(self):
         payload = console_ns.payload or {}
         args = CheckEmailUniquePayload.model_validate(payload)
-        if AccountService.is_account_in_freeze(args.email):
+        normalized_email = args.email.lower()
+        if AccountService.is_account_in_freeze(normalized_email):
             raise AccountInFreezeError()
-        if not AccountService.check_email_unique(args.email):
+        if not AccountService.check_email_unique(normalized_email):
             raise EmailAlreadyInUseError()
         return {"result": "success"}

api/controllers/console/workspace/members.py

@@ -116,26 +116,31 @@ class MemberInviteEmailApi(Resource):
             raise WorkspaceMembersLimitExceeded()
 
         for invitee_email in invitee_emails:
+            normalized_invitee_email = invitee_email.lower()
             try:
                 if not inviter.current_tenant:
                     raise ValueError("No current tenant")
                 token = RegisterService.invite_new_member(
-                    inviter.current_tenant, invitee_email, interface_language, role=invitee_role, inviter=inviter
+                    tenant=inviter.current_tenant,
+                    email=invitee_email,
+                    language=interface_language,
+                    role=invitee_role,
+                    inviter=inviter,
                 )
-                encoded_invitee_email = parse.quote(invitee_email)
+                encoded_invitee_email = parse.quote(normalized_invitee_email)
                 invitation_results.append(
                     {
                         "status": "success",
-                        "email": invitee_email,
+                        "email": normalized_invitee_email,
                         "url": f"{console_web_url}/activate?email={encoded_invitee_email}&token={token}",
                     }
                 )
             except AccountAlreadyInTenantError:
                 invitation_results.append(
-                    {"status": "success", "email": invitee_email, "url": f"{console_web_url}/signin"}
+                    {"status": "success", "email": normalized_invitee_email, "url": f"{console_web_url}/signin"}
                 )
             except Exception as e:
-                invitation_results.append({"status": "failed", "email": invitee_email, "message": str(e)})
+                invitation_results.append({"status": "failed", "email": normalized_invitee_email, "message": str(e)})
 
         return {
             "result": "success",

api/controllers/web/forgot_password.py

@@ -4,7 +4,6 @@ import secrets
 from flask import request
 from flask_restx import Resource
 from pydantic import BaseModel, Field, field_validator
-from sqlalchemy import select
 from sqlalchemy.orm import Session
 
 from controllers.common.schema import register_schema_models
@@ -22,7 +21,7 @@ from controllers.web import web_ns
 from extensions.ext_database import db
 from libs.helper import EmailStr, extract_remote_ip
 from libs.password import hash_password, valid_password
-from models import Account
+from models.account import Account
 from services.account_service import AccountService
 
 
@@ -70,6 +69,9 @@ class ForgotPasswordSendEmailApi(Resource):
     def post(self):
         payload = ForgotPasswordSendPayload.model_validate(web_ns.payload or {})
 
+        request_email = payload.email
+        normalized_email = request_email.lower()
+
         ip_address = extract_remote_ip(request)
         if AccountService.is_email_send_ip_limit(ip_address):
             raise EmailSendIpLimitError()
@@ -80,12 +82,12 @@ class ForgotPasswordSendEmailApi(Resource):
             language = "en-US"
 
         with Session(db.engine) as session:
-            account = session.execute(select(Account).filter_by(email=payload.email)).scalar_one_or_none()
+            account = AccountService.get_account_by_email_with_case_fallback(request_email, session=session)
         token = None
         if account is None:
             raise AuthenticationFailedError()
         else:
-            token = AccountService.send_reset_password_email(account=account, email=payload.email, language=language)
+            token = AccountService.send_reset_password_email(account=account, email=normalized_email, language=language)
 
         return {"result": "success", "data": token}
 
@@ -104,9 +106,9 @@ class ForgotPasswordCheckApi(Resource):
     def post(self):
         payload = ForgotPasswordCheckPayload.model_validate(web_ns.payload or {})
 
-        user_email = payload.email
+        user_email = payload.email.lower()
 
-        is_forgot_password_error_rate_limit = AccountService.is_forgot_password_error_rate_limit(payload.email)
+        is_forgot_password_error_rate_limit = AccountService.is_forgot_password_error_rate_limit(user_email)
         if is_forgot_password_error_rate_limit:
             raise EmailPasswordResetLimitError()
 
@@ -114,11 +116,16 @@ class ForgotPasswordCheckApi(Resource):
         if token_data is None:
             raise InvalidTokenError()
 
-        if user_email != token_data.get("email"):
+        token_email = token_data.get("email")
+        if not isinstance(token_email, str):
+            raise InvalidEmailError()
+        normalized_token_email = token_email.lower()
+
+        if user_email != normalized_token_email:
             raise InvalidEmailError()
 
         if payload.code != token_data.get("code"):
-            AccountService.add_forgot_password_error_rate_limit(payload.email)
+            AccountService.add_forgot_password_error_rate_limit(user_email)
             raise EmailCodeError()
 
         # Verified, revoke the first token
@@ -126,11 +133,11 @@ class ForgotPasswordCheckApi(Resource):
 
         # Refresh token data by generating a new token
         _, new_token = AccountService.generate_reset_password_token(
-            user_email, code=payload.code, additional_data={"phase": "reset"}
+            token_email, code=payload.code, additional_data={"phase": "reset"}
         )
 
-        AccountService.reset_forgot_password_error_rate_limit(payload.email)
-        return {"is_valid": True, "email": token_data.get("email"), "token": new_token}
+        AccountService.reset_forgot_password_error_rate_limit(user_email)
+        return {"is_valid": True, "email": normalized_token_email, "token": new_token}
 
 
 @web_ns.route("/forgot-password/resets")
@@ -174,7 +181,7 @@ class ForgotPasswordResetApi(Resource):
         email = reset_data.get("email", "")
 
         with Session(db.engine) as session:
-            account = session.execute(select(Account).filter_by(email=email)).scalar_one_or_none()
+            account = AccountService.get_account_by_email_with_case_fallback(email, session=session)
 
             if account:
                 self._update_existing_account(account, password_hashed, salt, session)

api/controllers/web/login.py

@@ -197,25 +197,29 @@ class EmailCodeLoginApi(Resource):
         )
         args = parser.parse_args()
 
-        user_email = args["email"]
+        user_email = args["email"].lower()
 
         token_data = WebAppAuthService.get_email_code_login_data(args["token"])
         if token_data is None:
             raise InvalidTokenError()
 
-        if token_data["email"] != args["email"]:
+        token_email = token_data.get("email")
+        if not isinstance(token_email, str):
+            raise InvalidEmailError()
+        normalized_token_email = token_email.lower()
+        if normalized_token_email != user_email:
             raise InvalidEmailError()
 
         if token_data["code"] != args["code"]:
             raise EmailCodeError()
 
         WebAppAuthService.revoke_email_code_login_token(args["token"])
-        account = WebAppAuthService.get_user_through_email(user_email)
+        account = WebAppAuthService.get_user_through_email(token_email)
         if not account:
             raise AuthenticationFailedError()
 
         token = WebAppAuthService.login(account=account)
-        AccountService.reset_login_error_rate_limit(args["email"])
+        AccountService.reset_login_error_rate_limit(user_email)
         response = make_response({"result": "success", "data": {"access_token": token}})
         # set_access_token_to_cookie(request, response, token, samesite="None", httponly=False)
         return response

api/core/agent/base_agent_runner.py

@@ -1,6 +1,7 @@
 import json
 import logging
 import uuid
+from decimal import Decimal
 from typing import Union, cast
 
 from sqlalchemy import select
@@ -41,6 +42,7 @@ from core.tools.tool_manager import ToolManager
 from core.tools.utils.dataset_retriever_tool import DatasetRetrieverTool
 from extensions.ext_database import db
 from factories import file_factory
+from models.enums import CreatorUserRole
 from models.model import Conversation, Message, MessageAgentThought, MessageFile
 
 logger = logging.getLogger(__name__)
@@ -289,6 +291,7 @@ class BaseAgentRunner(AppRunner):
         thought = MessageAgentThought(
             message_id=message_id,
             message_chain_id=None,
+            tool_process_data=None,
             thought="",
             tool=tool_name,
             tool_labels_str="{}",
@@ -296,20 +299,20 @@ class BaseAgentRunner(AppRunner):
             tool_input=tool_input,
             message=message,
             message_token=0,
-            message_unit_price=0,
-            message_price_unit=0,
+            message_unit_price=Decimal(0),
+            message_price_unit=Decimal("0.001"),
             message_files=json.dumps(messages_ids) if messages_ids else "",
             answer="",
             observation="",
             answer_token=0,
-            answer_unit_price=0,
-            answer_price_unit=0,
+            answer_unit_price=Decimal(0),
+            answer_price_unit=Decimal("0.001"),
             tokens=0,
-            total_price=0,
+            total_price=Decimal(0),
             position=self.agent_thought_count + 1,
             currency="USD",
             latency=0,
-            created_by_role="account",
+            created_by_role=CreatorUserRole.ACCOUNT,
             created_by=self.user_id,
         )
 
@@ -342,7 +345,8 @@ class BaseAgentRunner(AppRunner):
             raise ValueError("agent thought not found")
 
         if thought:
-            agent_thought.thought += thought
+            existing_thought = agent_thought.thought or ""
+            agent_thought.thought = f"{existing_thought}{thought}"
 
         if tool_name:
             agent_thought.tool = tool_name
@@ -440,21 +444,30 @@ class BaseAgentRunner(AppRunner):
             agent_thoughts: list[MessageAgentThought] = message.agent_thoughts
             if agent_thoughts:
                 for agent_thought in agent_thoughts:
-                    tools = agent_thought.tool
-                    if tools:
-                        tools = tools.split(";")
+                    tool_names_raw = agent_thought.tool
+                    if tool_names_raw:
+                        tool_names = tool_names_raw.split(";")
                         tool_calls: list[AssistantPromptMessage.ToolCall] = []
                         tool_call_response: list[ToolPromptMessage] = []
-                        try:
-                            tool_inputs = json.loads(agent_thought.tool_input)
-                        except Exception:
-                            tool_inputs = {tool: {} for tool in tools}
-                        try:
-                            tool_responses = json.loads(agent_thought.observation)
-                        except Exception:
-                            tool_responses = dict.fromkeys(tools, agent_thought.observation)
+                        tool_input_payload = agent_thought.tool_input
+                        if tool_input_payload:
+                            try:
+                                tool_inputs = json.loads(tool_input_payload)
+                            except Exception:
+                                tool_inputs = {tool: {} for tool in tool_names}
+                        else:
+                            tool_inputs = {tool: {} for tool in tool_names}
 
-                        for tool in tools:
+                        observation_payload = agent_thought.observation
+                        if observation_payload:
+                            try:
+                                tool_responses = json.loads(observation_payload)
+                            except Exception:
+                                tool_responses = dict.fromkeys(tool_names, observation_payload)
+                        else:
+                            tool_responses = dict.fromkeys(tool_names, observation_payload)
+
+                        for tool in tool_names:
                             # generate a uuid for tool call
                             tool_call_id = str(uuid.uuid4())
                             tool_calls.append(
@@ -484,7 +497,7 @@ class BaseAgentRunner(AppRunner):
                                 *tool_call_response,
                             ]
                         )
-                    if not tools:
+                    if not tool_names_raw:
                         result.append(AssistantPromptMessage(content=agent_thought.thought))
             else:
                 if message.answer:

api/core/agent/fc_agent_runner.py

@@ -188,7 +188,7 @@ class FunctionCallAgentRunner(BaseAgentRunner):
                     ),
                 )
 
-            assistant_message = AssistantPromptMessage(content="", tool_calls=[])
+            assistant_message = AssistantPromptMessage(content=response, tool_calls=[])
             if tool_calls:
                 assistant_message.tool_calls = [
                     AssistantPromptMessage.ToolCall(
@@ -200,8 +200,6 @@ class FunctionCallAgentRunner(BaseAgentRunner):
                     )
                     for tool_call in tool_calls
                 ]
-            else:
-                assistant_message.content = response
 
             self._current_thoughts.append(assistant_message)
 

api/core/app/app_config/entities.py

@@ -1,4 +1,3 @@
-import json
 from collections.abc import Sequence
 from enum import StrEnum, auto
 from typing import Any, Literal
@@ -121,7 +120,7 @@ class VariableEntity(BaseModel):
     allowed_file_types: Sequence[FileType] | None = Field(default_factory=list)
     allowed_file_extensions: Sequence[str] | None = Field(default_factory=list)
     allowed_file_upload_methods: Sequence[FileTransferMethod] | None = Field(default_factory=list)
-    json_schema: str | None = Field(default=None)
+    json_schema: dict | None = Field(default=None)
 
     @field_validator("description", mode="before")
     @classmethod
@@ -135,17 +134,11 @@ class VariableEntity(BaseModel):
 
     @field_validator("json_schema")
     @classmethod
-    def validate_json_schema(cls, schema: str | None) -> str | None:
+    def validate_json_schema(cls, schema: dict | None) -> dict | None:
         if schema is None:
             return None
-
         try:
-            json_schema = json.loads(schema)
-        except json.JSONDecodeError:
-            raise ValueError(f"invalid json_schema value {schema}")
-
-        try:
-            Draft7Validator.check_schema(json_schema)
+            Draft7Validator.check_schema(schema)
         except SchemaError as e:
             raise ValueError(f"Invalid JSON schema: {e.message}")
         return schema

api/core/app/apps/advanced_chat/app_config_manager.py

@@ -26,7 +26,6 @@ class AdvancedChatAppConfigManager(BaseAppConfigManager):
     @classmethod
     def get_app_config(cls, app_model: App, workflow: Workflow) -> AdvancedChatAppConfig:
         features_dict = workflow.features_dict
-
         app_mode = AppMode.value_of(app_model.mode)
         app_config = AdvancedChatAppConfig(
             tenant_id=app_model.tenant_id,

api/core/app/apps/advanced_chat/app_runner.py

@@ -24,7 +24,7 @@ from core.app.layers.conversation_variable_persist_layer import ConversationVari
 from core.db.session_factory import session_factory
 from core.moderation.base import ModerationError
 from core.moderation.input_moderation import InputModeration
-from core.variables.variables import VariableUnion
+from core.variables.variables import Variable
 from core.workflow.enums import WorkflowType
 from core.workflow.graph_engine.command_channels.redis_channel import RedisChannel
 from core.workflow.graph_engine.layers.base import GraphEngineLayer
@@ -149,8 +149,8 @@ class AdvancedChatAppRunner(WorkflowBasedAppRunner):
                 system_variables=system_inputs,
                 user_inputs=inputs,
                 environment_variables=self._workflow.environment_variables,
-                # Based on the definition of `VariableUnion`,
-                # `list[Variable]` can be safely used as `list[VariableUnion]` since they are compatible.
+                # Based on the definition of `Variable`,
+                # `VariableBase` instances can be safely used as `Variable` since they are compatible.
                 conversation_variables=conversation_variables,
             )
 
@@ -318,7 +318,7 @@ class AdvancedChatAppRunner(WorkflowBasedAppRunner):
             trace_manager=app_generate_entity.trace_manager,
         )
 
-    def _initialize_conversation_variables(self) -> list[VariableUnion]:
+    def _initialize_conversation_variables(self) -> list[Variable]:
         """
         Initialize conversation variables for the current conversation.
 
@@ -343,7 +343,7 @@ class AdvancedChatAppRunner(WorkflowBasedAppRunner):
             conversation_variables = [var.to_variable() for var in existing_variables]
 
             session.commit()
-            return cast(list[VariableUnion], conversation_variables)
+            return cast(list[Variable], conversation_variables)
 
     def _load_existing_conversation_variables(self, session: Session) -> list[ConversationVariable]:
         """

api/core/app/apps/advanced_chat/generate_task_pipeline.py

@@ -358,25 +358,6 @@ class AdvancedChatAppGenerateTaskPipeline(GraphRuntimeStateSupport):
         if node_finish_resp:
             yield node_finish_resp
 
-        # For ANSWER nodes, check if we need to send a message_replace event
-        # Only send if the final output differs from the accumulated task_state.answer
-        # This happens when variables were updated by variable_assigner during workflow execution
-        if event.node_type == NodeType.ANSWER and event.outputs:
-            final_answer = event.outputs.get("answer")
-            if final_answer is not None and final_answer != self._task_state.answer:
-                logger.info(
-                    "ANSWER node final output '%s' differs from accumulated answer '%s', sending message_replace event",
-                    final_answer,
-                    self._task_state.answer,
-                )
-                # Update the task state answer
-                self._task_state.answer = str(final_answer)
-                # Send message_replace event to update the UI
-                yield self._message_cycle_manager.message_replace_to_stream_response(
-                    answer=str(final_answer),
-                    reason="variable_update",
-                )
-
     def _handle_node_failed_events(
         self,
         event: Union[QueueNodeFailedEvent, QueueNodeExceptionEvent],

api/core/app/apps/base_app_generator.py

@@ -1,4 +1,3 @@
-import json
 from collections.abc import Generator, Mapping, Sequence
 from typing import TYPE_CHECKING, Any, Union, final
 
@@ -76,12 +75,24 @@ class BaseAppGenerator:
         user_inputs = {**user_inputs, **files_inputs, **file_list_inputs}
 
         # Check if all files are converted to File
-        if any(filter(lambda v: isinstance(v, dict), user_inputs.values())):
-            raise ValueError("Invalid input type")
-        if any(
-            filter(lambda v: isinstance(v, dict), filter(lambda item: isinstance(item, list), user_inputs.values()))
-        ):
-            raise ValueError("Invalid input type")
+        invalid_dict_keys = [
+            k
+            for k, v in user_inputs.items()
+            if isinstance(v, dict)
+            and entity_dictionary[k].type not in {VariableEntityType.FILE, VariableEntityType.JSON_OBJECT}
+        ]
+        if invalid_dict_keys:
+            raise ValueError(f"Invalid input type for {invalid_dict_keys}")
+
+        invalid_list_dict_keys = [
+            k
+            for k, v in user_inputs.items()
+            if isinstance(v, list)
+            and any(isinstance(item, dict) for item in v)
+            and entity_dictionary[k].type != VariableEntityType.FILE_LIST
+        ]
+        if invalid_list_dict_keys:
+            raise ValueError(f"Invalid input type for {invalid_list_dict_keys}")
 
         return user_inputs
 
@@ -178,12 +189,8 @@ class BaseAppGenerator:
                     elif value == 0:
                         value = False
             case VariableEntityType.JSON_OBJECT:
-                if not isinstance(value, str):
-                    raise ValueError(f"{variable_entity.variable} in input form must be a string")
-                try:
-                    json.loads(value)
-                except json.JSONDecodeError:
-                    raise ValueError(f"{variable_entity.variable} in input form must be a valid JSON object")
+                if value and not isinstance(value, dict):
+                    raise ValueError(f"{variable_entity.variable} in input form must be a dict")
             case _:
                 raise AssertionError("this statement should be unreachable.")
 

api/core/app/layers/conversation_variable_persist_layer.py

@@ -1,6 +1,6 @@
 import logging
 
-from core.variables import Variable
+from core.variables import VariableBase
 from core.workflow.constants import CONVERSATION_VARIABLE_NODE_ID
 from core.workflow.conversation_variable_updater import ConversationVariableUpdater
 from core.workflow.enums import NodeType
@@ -44,7 +44,7 @@ class ConversationVariablePersistenceLayer(GraphEngineLayer):
             if selector[0] != CONVERSATION_VARIABLE_NODE_ID:
                 continue
             variable = self.graph_runtime_state.variable_pool.get(selector)
-            if not isinstance(variable, Variable):
+            if not isinstance(variable, VariableBase):
                 logger.warning(
                     "Conversation variable not found in variable pool. selector=%s",
                     selector,

api/core/helper/ssrf_proxy.py

@@ -33,6 +33,10 @@ class MaxRetriesExceededError(ValueError):
     pass
 
 
+request_error = httpx.RequestError
+max_retries_exceeded_error = MaxRetriesExceededError
+
+
 def _create_proxy_mounts() -> dict[str, httpx.HTTPTransport]:
     return {
         "http://": httpx.HTTPTransport(

api/core/model_runtime/entities/message_entities.py

@@ -251,10 +251,7 @@ class AssistantPromptMessage(PromptMessage):
 
         :return: True if prompt message is empty, False otherwise
         """
-        if not super().is_empty() and not self.tool_calls:
-            return False
-
-        return True
+        return super().is_empty() and not self.tool_calls
 
 
 class SystemPromptMessage(PromptMessage):

api/core/ops/aliyun_trace/aliyun_trace.py

@@ -1,6 +1,7 @@
 import logging
 from collections.abc import Sequence
 
+from opentelemetry.trace import SpanKind
 from sqlalchemy.orm import sessionmaker
 
 from core.ops.aliyun_trace.data_exporter.traceclient import (
@@ -54,7 +55,7 @@ from core.ops.entities.trace_entity import (
     ToolTraceInfo,
     WorkflowTraceInfo,
 )
-from core.repositories import SQLAlchemyWorkflowNodeExecutionRepository
+from core.repositories import DifyCoreRepositoryFactory
 from core.workflow.entities import WorkflowNodeExecution
 from core.workflow.enums import NodeType, WorkflowNodeExecutionMetadataKey
 from extensions.ext_database import db
@@ -151,6 +152,7 @@ class AliyunDataTrace(BaseTraceInstance):
             ),
             status=status,
             links=trace_metadata.links,
+            span_kind=SpanKind.SERVER,
         )
         self.trace_client.add_span(message_span)
 
@@ -273,7 +275,7 @@ class AliyunDataTrace(BaseTraceInstance):
         service_account = self.get_service_account_with_tenant(app_id)
 
         session_factory = sessionmaker(bind=db.engine)
-        workflow_node_execution_repository = SQLAlchemyWorkflowNodeExecutionRepository(
+        workflow_node_execution_repository = DifyCoreRepositoryFactory.create_workflow_node_execution_repository(
             session_factory=session_factory,
             user=service_account,
             app_id=app_id,
@@ -456,6 +458,7 @@ class AliyunDataTrace(BaseTraceInstance):
                 ),
                 status=status,
                 links=trace_metadata.links,
+                span_kind=SpanKind.SERVER,
             )
             self.trace_client.add_span(message_span)
 
@@ -475,6 +478,7 @@ class AliyunDataTrace(BaseTraceInstance):
             ),
             status=status,
             links=trace_metadata.links,
+            span_kind=SpanKind.SERVER if message_span_id is None else SpanKind.INTERNAL,
         )
         self.trace_client.add_span(workflow_span)
 

api/core/ops/aliyun_trace/data_exporter/traceclient.py

@@ -166,7 +166,7 @@ class SpanBuilder:
             attributes=span_data.attributes,
             events=span_data.events,
             links=span_data.links,
-            kind=trace_api.SpanKind.INTERNAL,
+            kind=span_data.span_kind,
             status=span_data.status,
             start_time=span_data.start_time,
             end_time=span_data.end_time,

api/core/ops/aliyun_trace/entities/aliyun_trace_entity.py

@@ -4,7 +4,7 @@ from typing import Any
 
 from opentelemetry import trace as trace_api
 from opentelemetry.sdk.trace import Event
-from opentelemetry.trace import Status, StatusCode
+from opentelemetry.trace import SpanKind, Status, StatusCode
 from pydantic import BaseModel, Field
 
 
@@ -34,3 +34,4 @@ class SpanData(BaseModel):
     status: Status = Field(default=Status(StatusCode.UNSET), description="The status of the span.")
     start_time: int | None = Field(..., description="The start time of the span in nanoseconds.")
     end_time: int | None = Field(..., description="The end time of the span in nanoseconds.")
+    span_kind: SpanKind = Field(default=SpanKind.INTERNAL, description="The OpenTelemetry SpanKind for this span.")

api/core/tools/workflow_as_tool/tool.py

@@ -7,8 +7,8 @@ from typing import Any, cast
 
 from flask import has_request_context
 from sqlalchemy import select
-from sqlalchemy.orm import Session
 
+from core.db.session_factory import session_factory
 from core.file import FILE_MODEL_IDENTITY, File, FileTransferMethod
 from core.model_runtime.entities.llm_entities import LLMUsage, LLMUsageMetadata
 from core.tools.__base.tool import Tool
@@ -20,7 +20,6 @@ from core.tools.entities.tool_entities import (
     ToolProviderType,
 )
 from core.tools.errors import ToolInvokeError
-from extensions.ext_database import db
 from factories.file_factory import build_from_mapping
 from libs.login import current_user
 from models import Account, Tenant
@@ -230,30 +229,32 @@ class WorkflowTool(Tool):
         """
         Resolve user from database (worker/Celery context).
         """
+        with session_factory.create_session() as session:
+            tenant_stmt = select(Tenant).where(Tenant.id == self.runtime.tenant_id)
+            tenant = session.scalar(tenant_stmt)
+            if not tenant:
+                return None
+
+            user_stmt = select(Account).where(Account.id == user_id)
+            user = session.scalar(user_stmt)
+            if user:
+                user.current_tenant = tenant
+                session.expunge(user)
+                return user
+
+            end_user_stmt = select(EndUser).where(EndUser.id == user_id, EndUser.tenant_id == tenant.id)
+            end_user = session.scalar(end_user_stmt)
+            if end_user:
+                session.expunge(end_user)
+                return end_user
 
-        tenant_stmt = select(Tenant).where(Tenant.id == self.runtime.tenant_id)
-        tenant = db.session.scalar(tenant_stmt)
-        if not tenant:
             return None
 
-        user_stmt = select(Account).where(Account.id == user_id)
-        user = db.session.scalar(user_stmt)
-        if user:
-            user.current_tenant = tenant
-            return user
-
-        end_user_stmt = select(EndUser).where(EndUser.id == user_id, EndUser.tenant_id == tenant.id)
-        end_user = db.session.scalar(end_user_stmt)
-        if end_user:
-            return end_user
-
-        return None
-
     def _get_workflow(self, app_id: str, version: str) -> Workflow:
         """
         get the workflow by app id and version
         """
-        with Session(db.engine, expire_on_commit=False) as session, session.begin():
+        with session_factory.create_session() as session, session.begin():
             if not version:
                 stmt = (
                     select(Workflow)
@@ -265,22 +266,24 @@ class WorkflowTool(Tool):
                 stmt = select(Workflow).where(Workflow.app_id == app_id, Workflow.version == version)
                 workflow = session.scalar(stmt)
 
-        if not workflow:
-            raise ValueError("workflow not found or not published")
+            if not workflow:
+                raise ValueError("workflow not found or not published")
 
-        return workflow
+            session.expunge(workflow)
+            return workflow
 
     def _get_app(self, app_id: str) -> App:
         """
         get the app by app id
         """
         stmt = select(App).where(App.id == app_id)
-        with Session(db.engine, expire_on_commit=False) as session, session.begin():
+        with session_factory.create_session() as session, session.begin():
             app = session.scalar(stmt)
-        if not app:
-            raise ValueError("app not found")
+            if not app:
+                raise ValueError("app not found")
 
-        return app
+            session.expunge(app)
+            return app
 
     def _transform_args(self, tool_parameters: dict) -> tuple[dict, list[dict]]:
         """

api/core/variables/__init__.py

@@ -30,6 +30,7 @@ from .variables import (
     SecretVariable,
     StringVariable,
     Variable,
+    VariableBase,
 )
 
 __all__ = [
@@ -62,4 +63,5 @@ __all__ = [
     "StringSegment",
     "StringVariable",
     "Variable",
+    "VariableBase",
 ]

api/core/variables/segments.py

@@ -232,7 +232,7 @@ def get_segment_discriminator(v: Any) -> SegmentType | None:
 # - All variants in `SegmentUnion` must inherit from the `Segment` class.
 # - The union must include all non-abstract subclasses of `Segment`, except:
 #   - `SegmentGroup`, which is not added to the variable pool.
-#   - `Variable` and its subclasses, which are handled by `VariableUnion`.
+#   - `VariableBase` and its subclasses, which are handled by `Variable`.
 SegmentUnion: TypeAlias = Annotated[
     (
         Annotated[NoneSegment, Tag(SegmentType.NONE)]

api/core/variables/variables.py

@@ -27,7 +27,7 @@ from .segments import (
 from .types import SegmentType
 
 
-class Variable(Segment):
+class VariableBase(Segment):
     """
     A variable is a segment that has a name.
 
@@ -45,23 +45,23 @@ class Variable(Segment):
     selector: Sequence[str] = Field(default_factory=list)
 
 
-class StringVariable(StringSegment, Variable):
+class StringVariable(StringSegment, VariableBase):
     pass
 
 
-class FloatVariable(FloatSegment, Variable):
+class FloatVariable(FloatSegment, VariableBase):
     pass
 
 
-class IntegerVariable(IntegerSegment, Variable):
+class IntegerVariable(IntegerSegment, VariableBase):
     pass
 
 
-class ObjectVariable(ObjectSegment, Variable):
+class ObjectVariable(ObjectSegment, VariableBase):
     pass
 
 
-class ArrayVariable(ArraySegment, Variable):
+class ArrayVariable(ArraySegment, VariableBase):
     pass
 
 
@@ -89,16 +89,16 @@ class SecretVariable(StringVariable):
         return encrypter.obfuscated_token(self.value)
 
 
-class NoneVariable(NoneSegment, Variable):
+class NoneVariable(NoneSegment, VariableBase):
     value_type: SegmentType = SegmentType.NONE
     value: None = None
 
 
-class FileVariable(FileSegment, Variable):
+class FileVariable(FileSegment, VariableBase):
     pass
 
 
-class BooleanVariable(BooleanSegment, Variable):
+class BooleanVariable(BooleanSegment, VariableBase):
     pass
 
 
@@ -139,13 +139,13 @@ class RAGPipelineVariableInput(BaseModel):
     value: Any
 
 
-# The `VariableUnion`` type is used to enable serialization and deserialization with Pydantic.
-# Use `Variable` for type hinting when serialization is not required.
+# The `Variable` type is used to enable serialization and deserialization with Pydantic.
+# Use `VariableBase` for type hinting when serialization is not required.
 #
 # Note:
-# - All variants in `VariableUnion` must inherit from the `Variable` class.
-# - The union must include all non-abstract subclasses of `Segment`, except:
-VariableUnion: TypeAlias = Annotated[
+# - All variants in `Variable` must inherit from the `VariableBase` class.
+# - The union must include all non-abstract subclasses of `VariableBase`.
+Variable: TypeAlias = Annotated[
     (
         Annotated[NoneVariable, Tag(SegmentType.NONE)]
         | Annotated[StringVariable, Tag(SegmentType.STRING)]

api/core/workflow/conversation_variable_updater.py

@@ -1,7 +1,7 @@
 import abc
 from typing import Protocol
 
-from core.variables import Variable
+from core.variables import VariableBase
 
 
 class ConversationVariableUpdater(Protocol):
@@ -20,12 +20,12 @@ class ConversationVariableUpdater(Protocol):
     """
 
     @abc.abstractmethod
-    def update(self, conversation_id: str, variable: "Variable"):
+    def update(self, conversation_id: str, variable: "VariableBase"):
         """
         Updates the value of the specified conversation variable in the underlying storage.
 
         :param conversation_id: The ID of the conversation to update. Typically references `ConversationVariable.id`.
-        :param variable: The `Variable` instance containing the updated value.
+        :param variable: The `VariableBase` instance containing the updated value.
         """
         pass
 

api/core/workflow/enums.py

@@ -211,6 +211,10 @@ class WorkflowExecutionStatus(StrEnum):
     def is_ended(self) -> bool:
         return self in _END_STATE
 
+    @classmethod
+    def ended_values(cls) -> list[str]:
+        return [status.value for status in _END_STATE]
+
 
 _END_STATE = frozenset(
     [

api/core/workflow/graph_engine/entities/commands.py

@@ -11,7 +11,7 @@ from typing import Any
 
 from pydantic import BaseModel, Field
 
-from core.variables.variables import VariableUnion
+from core.variables.variables import Variable
 
 
 class CommandType(StrEnum):
@@ -46,7 +46,7 @@ class PauseCommand(GraphEngineCommand):
 class VariableUpdate(BaseModel):
     """Represents a single variable update instruction."""
 
-    value: VariableUnion = Field(description="New variable value")
+    value: Variable = Field(description="New variable value")
 
 
 class UpdateVariablesCommand(GraphEngineCommand):

api/core/workflow/nodes/http_request/executor.py

@@ -17,6 +17,7 @@ from core.helper import ssrf_proxy
 from core.variables.segments import ArrayFileSegment, FileSegment
 from core.workflow.runtime import VariablePool
 
+from ..protocols import FileManagerProtocol, HttpClientProtocol
 from .entities import (
     HttpRequestNodeAuthorization,
     HttpRequestNodeData,
@@ -78,6 +79,8 @@ class Executor:
         timeout: HttpRequestNodeTimeout,
         variable_pool: VariablePool,
         max_retries: int = dify_config.SSRF_DEFAULT_MAX_RETRIES,
+        http_client: HttpClientProtocol = ssrf_proxy,
+        file_manager: FileManagerProtocol = file_manager,
     ):
         # If authorization API key is present, convert the API key using the variable pool
         if node_data.authorization.type == "api-key":
@@ -104,6 +107,8 @@ class Executor:
         self.data = None
         self.json = None
         self.max_retries = max_retries
+        self._http_client = http_client
+        self._file_manager = file_manager
 
         # init template
         self.variable_pool = variable_pool
@@ -200,7 +205,7 @@ class Executor:
                     if file_variable is None:
                         raise FileFetchError(f"cannot fetch file with selector {file_selector}")
                     file = file_variable.value
-                    self.content = file_manager.download(file)
+                    self.content = self._file_manager.download(file)
                 case "x-www-form-urlencoded":
                     form_data = {
                         self.variable_pool.convert_template(item.key).text: self.variable_pool.convert_template(
@@ -239,7 +244,7 @@ class Executor:
                             ):
                                 file_tuple = (
                                     file.filename,
-                                    file_manager.download(file),
+                                    self._file_manager.download(file),
                                     file.mime_type or "application/octet-stream",
                                 )
                                 if key not in files:
@@ -332,19 +337,18 @@ class Executor:
         do http request depending on api bundle
         """
         _METHOD_MAP = {
-            "get": ssrf_proxy.get,
-            "head": ssrf_proxy.head,
-            "post": ssrf_proxy.post,
-            "put": ssrf_proxy.put,
-            "delete": ssrf_proxy.delete,
-            "patch": ssrf_proxy.patch,
+            "get": self._http_client.get,
+            "head": self._http_client.head,
+            "post": self._http_client.post,
+            "put": self._http_client.put,
+            "delete": self._http_client.delete,
+            "patch": self._http_client.patch,
         }
         method_lc = self.method.lower()
         if method_lc not in _METHOD_MAP:
             raise InvalidHttpMethodError(f"Invalid http method {self.method}")
 
         request_args = {
-            "url": self.url,
             "data": self.data,
             "files": self.files,
             "json": self.json,
@@ -357,8 +361,12 @@ class Executor:
         }
         # request_args = {k: v for k, v in request_args.items() if v is not None}
         try:
-            response: httpx.Response = _METHOD_MAP[method_lc](**request_args, max_retries=self.max_retries)
-        except (ssrf_proxy.MaxRetriesExceededError, httpx.RequestError) as e:
+            response: httpx.Response = _METHOD_MAP[method_lc](
+                url=self.url,
+                **request_args,
+                max_retries=self.max_retries,
+            )
+        except (self._http_client.max_retries_exceeded_error, self._http_client.request_error) as e:
             raise HttpRequestNodeError(str(e)) from e
         # FIXME: fix type ignore, this maybe httpx type issue
         return response

api/core/workflow/nodes/http_request/node.py

@@ -1,10 +1,11 @@
 import logging
 import mimetypes
-from collections.abc import Mapping, Sequence
-from typing import Any
+from collections.abc import Callable, Mapping, Sequence
+from typing import TYPE_CHECKING, Any
 
 from configs import dify_config
-from core.file import File, FileTransferMethod
+from core.file import File, FileTransferMethod, file_manager
+from core.helper import ssrf_proxy
 from core.tools.tool_file_manager import ToolFileManager
 from core.variables.segments import ArrayFileSegment
 from core.workflow.enums import NodeType, WorkflowNodeExecutionStatus
@@ -13,6 +14,7 @@ from core.workflow.nodes.base import variable_template_parser
 from core.workflow.nodes.base.entities import VariableSelector
 from core.workflow.nodes.base.node import Node
 from core.workflow.nodes.http_request.executor import Executor
+from core.workflow.nodes.protocols import FileManagerProtocol, HttpClientProtocol
 from factories import file_factory
 
 from .entities import (
@@ -30,10 +32,35 @@ HTTP_REQUEST_DEFAULT_TIMEOUT = HttpRequestNodeTimeout(
 
 logger = logging.getLogger(__name__)
 
+if TYPE_CHECKING:
+    from core.workflow.entities import GraphInitParams
+    from core.workflow.runtime import GraphRuntimeState
+
 
 class HttpRequestNode(Node[HttpRequestNodeData]):
     node_type = NodeType.HTTP_REQUEST
 
+    def __init__(
+        self,
+        id: str,
+        config: Mapping[str, Any],
+        graph_init_params: "GraphInitParams",
+        graph_runtime_state: "GraphRuntimeState",
+        *,
+        http_client: HttpClientProtocol = ssrf_proxy,
+        tool_file_manager_factory: Callable[[], ToolFileManager] = ToolFileManager,
+        file_manager: FileManagerProtocol = file_manager,
+    ) -> None:
+        super().__init__(
+            id=id,
+            config=config,
+            graph_init_params=graph_init_params,
+            graph_runtime_state=graph_runtime_state,
+        )
+        self._http_client = http_client
+        self._tool_file_manager_factory = tool_file_manager_factory
+        self._file_manager = file_manager
+
     @classmethod
     def get_default_config(cls, filters: Mapping[str, object] | None = None) -> Mapping[str, object]:
         return {
@@ -71,6 +98,8 @@ class HttpRequestNode(Node[HttpRequestNodeData]):
                 timeout=self._get_request_timeout(self.node_data),
                 variable_pool=self.graph_runtime_state.variable_pool,
                 max_retries=0,
+                http_client=self._http_client,
+                file_manager=self._file_manager,
             )
             process_data["request"] = http_executor.to_log()
 
@@ -199,7 +228,7 @@ class HttpRequestNode(Node[HttpRequestNodeData]):
         mime_type = (
             content_disposition_type or content_type or mimetypes.guess_type(filename)[0] or "application/octet-stream"
         )
-        tool_file_manager = ToolFileManager()
+        tool_file_manager = self._tool_file_manager_factory()
 
         tool_file = tool_file_manager.create_file_by_raw(
             user_id=self.user_id,

api/core/workflow/nodes/iteration/iteration_node.py

@@ -11,7 +11,7 @@ from typing_extensions import TypeIs
 from core.model_runtime.entities.llm_entities import LLMUsage
 from core.variables import IntegerVariable, NoneSegment
 from core.variables.segments import ArrayAnySegment, ArraySegment
-from core.variables.variables import VariableUnion
+from core.variables.variables import Variable
 from core.workflow.constants import CONVERSATION_VARIABLE_NODE_ID
 from core.workflow.enums import (
     NodeExecutionType,
@@ -240,7 +240,7 @@ class IterationNode(LLMUsageTrackingMixin, Node[IterationNodeData]):
                         datetime,
                         list[GraphNodeEventBase],
                         object | None,
-                        dict[str, VariableUnion],
+                        dict[str, Variable],
                         LLMUsage,
                     ]
                 ],
@@ -308,7 +308,7 @@ class IterationNode(LLMUsageTrackingMixin, Node[IterationNodeData]):
         item: object,
         flask_app: Flask,
         context_vars: contextvars.Context,
-    ) -> tuple[datetime, list[GraphNodeEventBase], object | None, dict[str, VariableUnion], LLMUsage]:
+    ) -> tuple[datetime, list[GraphNodeEventBase], object | None, dict[str, Variable], LLMUsage]:
         """Execute a single iteration in parallel mode and return results."""
         with preserve_flask_contexts(flask_app=flask_app, context_vars=context_vars):
             iter_start_at = datetime.now(UTC).replace(tzinfo=None)
@@ -515,11 +515,11 @@ class IterationNode(LLMUsageTrackingMixin, Node[IterationNodeData]):
 
         return variable_mapping
 
-    def _extract_conversation_variable_snapshot(self, *, variable_pool: VariablePool) -> dict[str, VariableUnion]:
+    def _extract_conversation_variable_snapshot(self, *, variable_pool: VariablePool) -> dict[str, Variable]:
         conversation_variables = variable_pool.variable_dictionary.get(CONVERSATION_VARIABLE_NODE_ID, {})
         return {name: variable.model_copy(deep=True) for name, variable in conversation_variables.items()}
 
-    def _sync_conversation_variables_from_snapshot(self, snapshot: dict[str, VariableUnion]) -> None:
+    def _sync_conversation_variables_from_snapshot(self, snapshot: dict[str, Variable]) -> None:
         parent_pool = self.graph_runtime_state.variable_pool
         parent_conversations = parent_pool.variable_dictionary.get(CONVERSATION_VARIABLE_NODE_ID, {})
 

api/core/workflow/nodes/node_factory.py

@@ -1,16 +1,21 @@
-from collections.abc import Sequence
+from collections.abc import Callable, Sequence
 from typing import TYPE_CHECKING, final
 
 from typing_extensions import override
 
 from configs import dify_config
+from core.file import file_manager
+from core.helper import ssrf_proxy
 from core.helper.code_executor.code_executor import CodeExecutor
 from core.helper.code_executor.code_node_provider import CodeNodeProvider
+from core.tools.tool_file_manager import ToolFileManager
 from core.workflow.enums import NodeType
 from core.workflow.graph import NodeFactory
 from core.workflow.nodes.base.node import Node
 from core.workflow.nodes.code.code_node import CodeNode
 from core.workflow.nodes.code.limits import CodeNodeLimits
+from core.workflow.nodes.http_request.node import HttpRequestNode
+from core.workflow.nodes.protocols import FileManagerProtocol, HttpClientProtocol
 from core.workflow.nodes.template_transform.template_renderer import (
     CodeExecutorJinja2TemplateRenderer,
     Jinja2TemplateRenderer,
@@ -43,6 +48,9 @@ class DifyNodeFactory(NodeFactory):
         code_providers: Sequence[type[CodeNodeProvider]] | None = None,
         code_limits: CodeNodeLimits | None = None,
         template_renderer: Jinja2TemplateRenderer | None = None,
+        http_request_http_client: HttpClientProtocol = ssrf_proxy,
+        http_request_tool_file_manager_factory: Callable[[], ToolFileManager] = ToolFileManager,
+        http_request_file_manager: FileManagerProtocol = file_manager,
     ) -> None:
         self.graph_init_params = graph_init_params
         self.graph_runtime_state = graph_runtime_state
@@ -61,6 +69,9 @@ class DifyNodeFactory(NodeFactory):
             max_object_array_length=dify_config.CODE_MAX_OBJECT_ARRAY_LENGTH,
         )
         self._template_renderer = template_renderer or CodeExecutorJinja2TemplateRenderer()
+        self._http_request_http_client = http_request_http_client
+        self._http_request_tool_file_manager_factory = http_request_tool_file_manager_factory
+        self._http_request_file_manager = http_request_file_manager
 
     @override
     def create_node(self, node_config: dict[str, object]) -> Node:
@@ -113,6 +124,7 @@ class DifyNodeFactory(NodeFactory):
                 code_providers=self._code_providers,
                 code_limits=self._code_limits,
             )
+
         if node_type == NodeType.TEMPLATE_TRANSFORM:
             return TemplateTransformNode(
                 id=node_id,
@@ -122,6 +134,17 @@ class DifyNodeFactory(NodeFactory):
                 template_renderer=self._template_renderer,
             )
 
+        if node_type == NodeType.HTTP_REQUEST:
+            return HttpRequestNode(
+                id=node_id,
+                config=node_config,
+                graph_init_params=self.graph_init_params,
+                graph_runtime_state=self.graph_runtime_state,
+                http_client=self._http_request_http_client,
+                tool_file_manager_factory=self._http_request_tool_file_manager_factory,
+                file_manager=self._http_request_file_manager,
+            )
+
         return node_class(
             id=node_id,
             config=node_config,

api/core/workflow/nodes/protocols.py

@@ -0,0 +1,29 @@
+from typing import Protocol
+
+import httpx
+
+from core.file import File
+
+
+class HttpClientProtocol(Protocol):
+    @property
+    def max_retries_exceeded_error(self) -> type[Exception]: ...
+
+    @property
+    def request_error(self) -> type[Exception]: ...
+
+    def get(self, url: str, max_retries: int = ..., **kwargs: object) -> httpx.Response: ...
+
+    def head(self, url: str, max_retries: int = ..., **kwargs: object) -> httpx.Response: ...
+
+    def post(self, url: str, max_retries: int = ..., **kwargs: object) -> httpx.Response: ...
+
+    def put(self, url: str, max_retries: int = ..., **kwargs: object) -> httpx.Response: ...
+
+    def delete(self, url: str, max_retries: int = ..., **kwargs: object) -> httpx.Response: ...
+
+    def patch(self, url: str, max_retries: int = ..., **kwargs: object) -> httpx.Response: ...
+
+
+class FileManagerProtocol(Protocol):
+    def download(self, f: File, /) -> bytes: ...

api/core/workflow/nodes/start/start_node.py

@@ -1,4 +1,3 @@
-import json
 from typing import Any
 
 from jsonschema import Draft7Validator, ValidationError
@@ -43,25 +42,22 @@ class StartNode(Node[StartNodeData]):
             if value is None and variable.required:
                 raise ValueError(f"{key} is required in input form")
 
+            # If no value provided, skip further processing for this key
+            if not value:
+                continue
+
+            if not isinstance(value, dict):
+                raise ValueError(f"JSON object for '{key}' must be an object")
+
+            # Overwrite with normalized dict to ensure downstream consistency
+            node_inputs[key] = value
+
+            # If schema exists, then validate against it
             schema = variable.json_schema
             if not schema:
                 continue
 
-            if not value:
-                continue
-
             try:
-                json_schema = json.loads(schema)
-            except json.JSONDecodeError as e:
-                raise ValueError(f"{schema} must be a valid JSON object")
-
-            try:
-                json_value = json.loads(value)
-            except json.JSONDecodeError as e:
-                raise ValueError(f"{value} must be a valid JSON object")
-
-            try:
-                Draft7Validator(json_schema).validate(json_value)
+                Draft7Validator(schema).validate(value)
             except ValidationError as e:
                 raise ValueError(f"JSON object for '{key}' does not match schema: {e.message}")
-            node_inputs[key] = json_value

api/core/workflow/nodes/variable_assigner/v1/node.py

@@ -1,7 +1,7 @@
 from collections.abc import Mapping, Sequence
 from typing import TYPE_CHECKING, Any
 
-from core.variables import SegmentType, Variable
+from core.variables import SegmentType, VariableBase
 from core.workflow.constants import CONVERSATION_VARIABLE_NODE_ID
 from core.workflow.entities import GraphInitParams
 from core.workflow.enums import NodeType, WorkflowNodeExecutionStatus
@@ -33,6 +33,15 @@ class VariableAssignerNode(Node[VariableAssignerData]):
             graph_runtime_state=graph_runtime_state,
         )
 
+    def blocks_variable_output(self, variable_selectors: set[tuple[str, ...]]) -> bool:
+        """
+        Check if this Variable Assigner node blocks the output of specific variables.
+
+        Returns True if this node updates any of the requested conversation variables.
+        """
+        assigned_selector = tuple(self.node_data.assigned_variable_selector)
+        return assigned_selector in variable_selectors
+
     @classmethod
     def version(cls) -> str:
         return "1"
@@ -64,7 +73,7 @@ class VariableAssignerNode(Node[VariableAssignerData]):
         assigned_variable_selector = self.node_data.assigned_variable_selector
         # Should be String, Number, Object, ArrayString, ArrayNumber, ArrayObject
         original_variable = self.graph_runtime_state.variable_pool.get(assigned_variable_selector)
-        if not isinstance(original_variable, Variable):
+        if not isinstance(original_variable, VariableBase):
             raise VariableOperatorNodeError("assigned variable not found")
 
         match self.node_data.write_mode:

api/core/workflow/nodes/variable_assigner/v2/node.py

@@ -2,7 +2,7 @@ import json
 from collections.abc import Mapping, MutableMapping, Sequence
 from typing import TYPE_CHECKING, Any
 
-from core.variables import SegmentType, Variable
+from core.variables import SegmentType, VariableBase
 from core.variables.consts import SELECTORS_LENGTH
 from core.workflow.constants import CONVERSATION_VARIABLE_NODE_ID
 from core.workflow.enums import NodeType, WorkflowNodeExecutionStatus
@@ -118,7 +118,7 @@ class VariableAssignerNode(Node[VariableAssignerNodeData]):
                 # ==================== Validation Part
 
                 # Check if variable exists
-                if not isinstance(variable, Variable):
+                if not isinstance(variable, VariableBase):
                     raise VariableNotFoundError(variable_selector=item.variable_selector)
 
                 # Check if operation is supported
@@ -192,7 +192,7 @@ class VariableAssignerNode(Node[VariableAssignerNodeData]):
 
         for selector in updated_variable_selectors:
             variable = self.graph_runtime_state.variable_pool.get(selector)
-            if not isinstance(variable, Variable):
+            if not isinstance(variable, VariableBase):
                 raise VariableNotFoundError(variable_selector=selector)
             process_data[variable.name] = variable.value
 
@@ -213,7 +213,7 @@ class VariableAssignerNode(Node[VariableAssignerNodeData]):
     def _handle_item(
         self,
         *,
-        variable: Variable,
+        variable: VariableBase,
         operation: Operation,
         value: Any,
     ):

api/core/workflow/runtime/variable_pool.py

@@ -9,10 +9,10 @@ from typing import Annotated, Any, Union, cast
 from pydantic import BaseModel, Field
 
 from core.file import File, FileAttribute, file_manager
-from core.variables import Segment, SegmentGroup, Variable
+from core.variables import Segment, SegmentGroup, VariableBase
 from core.variables.consts import SELECTORS_LENGTH
 from core.variables.segments import FileSegment, ObjectSegment
-from core.variables.variables import RAGPipelineVariableInput, VariableUnion
+from core.variables.variables import RAGPipelineVariableInput, Variable
 from core.workflow.constants import (
     CONVERSATION_VARIABLE_NODE_ID,
     ENVIRONMENT_VARIABLE_NODE_ID,
@@ -32,7 +32,7 @@ class VariablePool(BaseModel):
     # The first element of the selector is the node id, it's the first-level key in the dictionary.
     # Other elements of the selector are the keys in the second-level dictionary. To get the key, we hash the
     # elements of the selector except the first one.
-    variable_dictionary: defaultdict[str, Annotated[dict[str, VariableUnion], Field(default_factory=dict)]] = Field(
+    variable_dictionary: defaultdict[str, Annotated[dict[str, Variable], Field(default_factory=dict)]] = Field(
         description="Variables mapping",
         default=defaultdict(dict),
     )
@@ -46,13 +46,13 @@ class VariablePool(BaseModel):
         description="System variables",
         default_factory=SystemVariable.empty,
     )
-    environment_variables: Sequence[VariableUnion] = Field(
+    environment_variables: Sequence[Variable] = Field(
         description="Environment variables.",
-        default_factory=list[VariableUnion],
+        default_factory=list[Variable],
     )
-    conversation_variables: Sequence[VariableUnion] = Field(
+    conversation_variables: Sequence[Variable] = Field(
         description="Conversation variables.",
-        default_factory=list[VariableUnion],
+        default_factory=list[Variable],
     )
     rag_pipeline_variables: list[RAGPipelineVariableInput] = Field(
         description="RAG pipeline variables.",
@@ -105,7 +105,7 @@ class VariablePool(BaseModel):
                 f"got {len(selector)} elements"
             )
 
-        if isinstance(value, Variable):
+        if isinstance(value, VariableBase):
             variable = value
         elif isinstance(value, Segment):
             variable = variable_factory.segment_to_variable(segment=value, selector=selector)
@@ -114,9 +114,9 @@ class VariablePool(BaseModel):
             variable = variable_factory.segment_to_variable(segment=segment, selector=selector)
 
         node_id, name = self._selector_to_keys(selector)
-        # Based on the definition of `VariableUnion`,
-        # `list[Variable]` can be safely used as `list[VariableUnion]` since they are compatible.
-        self.variable_dictionary[node_id][name] = cast(VariableUnion, variable)
+        # Based on the definition of `Variable`,
+        # `VariableBase` instances can be safely used as `Variable` since they are compatible.
+        self.variable_dictionary[node_id][name] = cast(Variable, variable)
 
     @classmethod
     def _selector_to_keys(cls, selector: Sequence[str]) -> tuple[str, str]:

api/core/workflow/variable_loader.py

@@ -2,7 +2,7 @@ import abc
 from collections.abc import Mapping, Sequence
 from typing import Any, Protocol
 
-from core.variables import Variable
+from core.variables import VariableBase
 from core.variables.consts import SELECTORS_LENGTH
 from core.workflow.runtime import VariablePool
 
@@ -26,7 +26,7 @@ class VariableLoader(Protocol):
     """
 
     @abc.abstractmethod
-    def load_variables(self, selectors: list[list[str]]) -> list[Variable]:
+    def load_variables(self, selectors: list[list[str]]) -> list[VariableBase]:
         """Load variables based on the provided selectors. If the selectors are empty,
         this method should return an empty list.
 
@@ -36,7 +36,7 @@ class VariableLoader(Protocol):
         :param: selectors: a list of string list, each inner list should have at least two elements:
             - the first element is the node ID,
             - the second element is the variable name.
-        :return: a list of Variable objects that match the provided selectors.
+        :return: a list of VariableBase objects that match the provided selectors.
         """
         pass
 
@@ -46,7 +46,7 @@ class _DummyVariableLoader(VariableLoader):
     Serves as a placeholder when no variable loading is needed.
     """
 
-    def load_variables(self, selectors: list[list[str]]) -> list[Variable]:
+    def load_variables(self, selectors: list[list[str]]) -> list[VariableBase]:
         return []
 
 

api/core/workflow/workflow_entry.py

@@ -19,6 +19,7 @@ from core.workflow.graph_engine.protocols.command_channel import CommandChannel
 from core.workflow.graph_events import GraphEngineEvent, GraphNodeEventBase, GraphRunFailedEvent
 from core.workflow.nodes import NodeType
 from core.workflow.nodes.base.node import Node
+from core.workflow.nodes.node_factory import DifyNodeFactory
 from core.workflow.nodes.node_mapping import NODE_TYPE_CLASSES_MAPPING
 from core.workflow.runtime import GraphRuntimeState, VariablePool
 from core.workflow.system_variable import SystemVariable
@@ -136,13 +137,11 @@ class WorkflowEntry:
         :param user_inputs: user inputs
         :return:
         """
-        node_config = workflow.get_node_config_by_id(node_id)
+        node_config = dict(workflow.get_node_config_by_id(node_id))
         node_config_data = node_config.get("data", {})
 
-        # Get node class
+        # Get node type
         node_type = NodeType(node_config_data.get("type"))
-        node_version = node_config_data.get("version", "1")
-        node_cls = NODE_TYPE_CLASSES_MAPPING[node_type][node_version]
 
         # init graph init params and runtime state
         graph_init_params = GraphInitParams(
@@ -158,12 +157,12 @@ class WorkflowEntry:
         graph_runtime_state = GraphRuntimeState(variable_pool=variable_pool, start_at=time.perf_counter())
 
         # init workflow run state
-        node = node_cls(
-            id=str(uuid.uuid4()),
-            config=node_config,
+        node_factory = DifyNodeFactory(
             graph_init_params=graph_init_params,
             graph_runtime_state=graph_runtime_state,
         )
+        node = node_factory.create_node(node_config)
+        node_cls = type(node)
 
         try:
             # variable selector to variable mapping

api/extensions/ext_celery.py

@@ -163,6 +163,13 @@ def init_app(app: DifyApp) -> Celery:
             "task": "schedule.clean_workflow_runlogs_precise.clean_workflow_runlogs_precise",
             "schedule": crontab(minute="0", hour="2"),
         }
+    if dify_config.ENABLE_WORKFLOW_RUN_CLEANUP_TASK:
+        # for saas only
+        imports.append("schedule.clean_workflow_runs_task")
+        beat_schedule["clean_workflow_runs_task"] = {
+            "task": "schedule.clean_workflow_runs_task.clean_workflow_runs_task",
+            "schedule": crontab(minute="0", hour="0"),
+        }
     if dify_config.ENABLE_WORKFLOW_SCHEDULE_POLLER_TASK:
         imports.append("schedule.workflow_schedule_task")
         beat_schedule["workflow_schedule_task"] = {

api/extensions/ext_commands.py

@@ -4,6 +4,7 @@ from dify_app import DifyApp
 def init_app(app: DifyApp):
     from commands import (
         add_qdrant_index,
+        clean_workflow_runs,
         cleanup_orphaned_draft_variables,
         clear_free_plan_tenant_expired_logs,
         clear_orphaned_file_records,
@@ -56,6 +57,7 @@ def init_app(app: DifyApp):
         setup_datasource_oauth_client,
         transform_datasource_credentials,
         install_rag_pipeline_plugins,
+        clean_workflow_runs,
     ]
     for cmd in cmds_to_register:
         app.cli.add_command(cmd)

api/extensions/ext_logstore.py

@@ -10,6 +10,7 @@ import os
 
 from dotenv import load_dotenv
 
+from configs import dify_config
 from dify_app import DifyApp
 
 logger = logging.getLogger(__name__)
@@ -19,12 +20,17 @@ def is_enabled() -> bool:
     """
     Check if logstore extension is enabled.
 
+    Logstore is considered enabled when:
+    1. All required Aliyun SLS environment variables are set
+    2. At least one repository configuration points to a logstore implementation
+
     Returns:
-        True if all required Aliyun SLS environment variables are set, False otherwise
+        True if logstore should be initialized, False otherwise
     """
     # Load environment variables from .env file
     load_dotenv()
 
+    # Check if Aliyun SLS connection parameters are configured
     required_vars = [
         "ALIYUN_SLS_ACCESS_KEY_ID",
         "ALIYUN_SLS_ACCESS_KEY_SECRET",
@@ -33,24 +39,32 @@ def is_enabled() -> bool:
         "ALIYUN_SLS_PROJECT_NAME",
     ]
 
-    all_set = all(os.environ.get(var) for var in required_vars)
+    sls_vars_set = all(os.environ.get(var) for var in required_vars)
 
-    if not all_set:
-        logger.info("Logstore extension disabled: required Aliyun SLS environment variables not set")
+    if not sls_vars_set:
+        return False
 
-    return all_set
+    # Check if any repository configuration points to logstore implementation
+    repository_configs = [
+        dify_config.CORE_WORKFLOW_EXECUTION_REPOSITORY,
+        dify_config.CORE_WORKFLOW_NODE_EXECUTION_REPOSITORY,
+        dify_config.API_WORKFLOW_NODE_EXECUTION_REPOSITORY,
+        dify_config.API_WORKFLOW_RUN_REPOSITORY,
+    ]
+
+    uses_logstore = any("logstore" in config.lower() for config in repository_configs)
+
+    if not uses_logstore:
+        return False
+
+    logger.info("Logstore extension enabled: SLS variables set and repository configured to use logstore")
+    return True
 
 
 def init_app(app: DifyApp):
     """
     Initialize logstore on application startup.
-
-    This function:
-    1. Creates Aliyun SLS project if it doesn't exist
-    2. Creates logstores (workflow_execution, workflow_node_execution) if they don't exist
-    3. Creates indexes with field configurations based on PostgreSQL table structures
-
-    This operation is idempotent and only executes once during application startup.
+    If initialization fails, the application continues running without logstore features.
 
     Args:
         app: The Dify application instance
@@ -58,17 +72,23 @@ def init_app(app: DifyApp):
     try:
         from extensions.logstore.aliyun_logstore import AliyunLogStore
 
-        logger.info("Initializing logstore...")
+        logger.info("Initializing Aliyun SLS Logstore...")
 
-        # Create logstore client and initialize project/logstores/indexes
+        # Create logstore client and initialize resources
         logstore_client = AliyunLogStore()
         logstore_client.init_project_logstore()
 
-        # Attach to app for potential later use
         app.extensions["logstore"] = logstore_client
 
         logger.info("Logstore initialized successfully")
+
     except Exception:
-        logger.exception("Failed to initialize logstore")
-        # Don't raise - allow application to continue even if logstore init fails
-        # This ensures that the application can still run if logstore is misconfigured
+        logger.exception(
+            "Logstore initialization failed. Configuration: endpoint=%s, region=%s, project=%s, timeout=%ss. "
+            "Application will continue but logstore features will NOT work.",
+            os.environ.get("ALIYUN_SLS_ENDPOINT"),
+            os.environ.get("ALIYUN_SLS_REGION"),
+            os.environ.get("ALIYUN_SLS_PROJECT_NAME"),
+            os.environ.get("ALIYUN_SLS_CHECK_CONNECTIVITY_TIMEOUT", "30"),
+        )
+        # Don't raise - allow application to continue even if logstore setup fails

api/extensions/logstore/aliyun_logstore.py

@@ -2,6 +2,7 @@ from __future__ import annotations
 
 import logging
 import os
+import socket
 import threading
 import time
 from collections.abc import Sequence
@@ -179,9 +180,18 @@ class AliyunLogStore:
         self.region: str = os.environ.get("ALIYUN_SLS_REGION", "")
         self.project_name: str = os.environ.get("ALIYUN_SLS_PROJECT_NAME", "")
         self.logstore_ttl: int = int(os.environ.get("ALIYUN_SLS_LOGSTORE_TTL", 365))
-        self.log_enabled: bool = os.environ.get("SQLALCHEMY_ECHO", "false").lower() == "true"
+        self.log_enabled: bool = (
+            os.environ.get("SQLALCHEMY_ECHO", "false").lower() == "true"
+            or os.environ.get("LOGSTORE_SQL_ECHO", "false").lower() == "true"
+        )
         self.pg_mode_enabled: bool = os.environ.get("LOGSTORE_PG_MODE_ENABLED", "true").lower() == "true"
 
+        # Get timeout configuration
+        check_timeout = int(os.environ.get("ALIYUN_SLS_CHECK_CONNECTIVITY_TIMEOUT", 30))
+
+        # Pre-check endpoint connectivity to prevent indefinite hangs
+        self._check_endpoint_connectivity(self.endpoint, check_timeout)
+
         # Initialize SDK client
         self.client = LogClient(
             self.endpoint, self.access_key_id, self.access_key_secret, auth_version=AUTH_VERSION_4, region=self.region
@@ -199,6 +209,49 @@ class AliyunLogStore:
 
         self.__class__._initialized = True
 
+    @staticmethod
+    def _check_endpoint_connectivity(endpoint: str, timeout: int) -> None:
+        """
+        Check if the SLS endpoint is reachable before creating LogClient.
+        Prevents indefinite hangs when the endpoint is unreachable.
+
+        Args:
+            endpoint: SLS endpoint URL
+            timeout: Connection timeout in seconds
+
+        Raises:
+            ConnectionError: If endpoint is not reachable
+        """
+        # Parse endpoint URL to extract hostname and port
+        from urllib.parse import urlparse
+
+        parsed_url = urlparse(endpoint if "://" in endpoint else f"http://{endpoint}")
+        hostname = parsed_url.hostname
+        port = parsed_url.port or (443 if parsed_url.scheme == "https" else 80)
+
+        if not hostname:
+            raise ConnectionError(f"Invalid endpoint URL: {endpoint}")
+
+        sock = None
+        try:
+            # Create socket and set timeout
+            sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
+            sock.settimeout(timeout)
+            sock.connect((hostname, port))
+        except Exception as e:
+            # Catch all exceptions and provide clear error message
+            error_type = type(e).__name__
+            raise ConnectionError(
+                f"Cannot connect to {hostname}:{port} (timeout={timeout}s): [{error_type}] {e}"
+            ) from e
+        finally:
+            # Ensure socket is properly closed
+            if sock:
+                try:
+                    sock.close()
+                except Exception:  # noqa: S110
+                    pass  # Ignore errors during cleanup
+
     @property
     def supports_pg_protocol(self) -> bool:
         """Check if PG protocol is supported and enabled."""
@@ -220,19 +273,16 @@ class AliyunLogStore:
         try:
             self._use_pg_protocol = self._pg_client.init_connection()
             if self._use_pg_protocol:
-                logger.info("Successfully connected to project %s using PG protocol", self.project_name)
+                logger.info("Using PG protocol for project %s", self.project_name)
                 # Check if scan_index is enabled for all logstores
                 self._check_and_disable_pg_if_scan_index_disabled()
                 return True
             else:
-                logger.info("PG connection failed for project %s. Will use SDK mode.", self.project_name)
+                logger.info("Using SDK mode for project %s", self.project_name)
                 return False
         except Exception as e:
-            logger.warning(
-                "Failed to establish PG connection for project %s: %s. Will use SDK mode.",
-                self.project_name,
-                str(e),
-            )
+            logger.info("Using SDK mode for project %s", self.project_name)
+            logger.debug("PG connection details: %s", str(e))
             self._use_pg_protocol = False
             return False
 
@@ -246,10 +296,6 @@ class AliyunLogStore:
         if self._use_pg_protocol:
             return
 
-        logger.info(
-            "Attempting delayed PG connection for newly created project %s ...",
-            self.project_name,
-        )
         self._attempt_pg_connection_init()
         self.__class__._pg_connection_timer = None
 
@@ -284,11 +330,7 @@ class AliyunLogStore:
         if project_is_new:
             # For newly created projects, schedule delayed PG connection
             self._use_pg_protocol = False
-            logger.info(
-                "Project %s is newly created. Will use SDK mode and schedule PG connection attempt in %d seconds.",
-                self.project_name,
-                self.__class__._pg_connection_delay,
-            )
+            logger.info("Using SDK mode for project %s (newly created)", self.project_name)
             if self.__class__._pg_connection_timer is not None:
                 self.__class__._pg_connection_timer.cancel()
             self.__class__._pg_connection_timer = threading.Timer(
@@ -299,7 +341,6 @@ class AliyunLogStore:
             self.__class__._pg_connection_timer.start()
         else:
             # For existing projects, attempt PG connection immediately
-            logger.info("Project %s already exists. Attempting PG connection...", self.project_name)
             self._attempt_pg_connection_init()
 
     def _check_and_disable_pg_if_scan_index_disabled(self) -> None:
@@ -318,9 +359,9 @@ class AliyunLogStore:
             existing_config = self.get_existing_index_config(logstore_name)
             if existing_config and not existing_config.scan_index:
                 logger.info(
-                    "Logstore %s has scan_index=false, USE SDK mode for read/write operations. "
-                    "PG protocol requires scan_index to be enabled.",
+                    "Logstore %s requires scan_index enabled, using SDK mode for project %s",
                     logstore_name,
+                    self.project_name,
                 )
                 self._use_pg_protocol = False
                 # Close PG connection if it was initialized
@@ -748,7 +789,6 @@ class AliyunLogStore:
             reverse=reverse,
         )
 
-        # Log query info if SQLALCHEMY_ECHO is enabled
         if self.log_enabled:
             logger.info(
                 "[LogStore] GET_LOGS | logstore=%s | project=%s | query=%s | "
@@ -770,7 +810,6 @@ class AliyunLogStore:
             for log in logs:
                 result.append(log.get_contents())
 
-            # Log result count if SQLALCHEMY_ECHO is enabled
             if self.log_enabled:
                 logger.info(
                     "[LogStore] GET_LOGS RESULT | logstore=%s | returned_count=%d",
@@ -845,7 +884,6 @@ class AliyunLogStore:
                 query=full_query,
             )
 
-            # Log query info if SQLALCHEMY_ECHO is enabled
             if self.log_enabled:
                 logger.info(
                     "[LogStore-SDK] EXECUTE_SQL | logstore=%s | project=%s | from_time=%d | to_time=%d | full_query=%s",
@@ -853,8 +891,7 @@ class AliyunLogStore:
                     self.project_name,
                     from_time,
                     to_time,
-                    query,
-                    sql,
+                    full_query,
                 )
 
             try:
@@ -865,7 +902,6 @@ class AliyunLogStore:
                 for log in logs:
                     result.append(log.get_contents())
 
-                # Log result count if SQLALCHEMY_ECHO is enabled
                 if self.log_enabled:
                     logger.info(
                         "[LogStore-SDK] EXECUTE_SQL RESULT | logstore=%s | returned_count=%d",

api/extensions/logstore/aliyun_logstore_pg.py

@@ -7,8 +7,7 @@ from contextlib import contextmanager
 from typing import Any
 
 import psycopg2
-import psycopg2.pool
-from psycopg2 import InterfaceError, OperationalError
+from sqlalchemy import create_engine
 
 from configs import dify_config
 
@@ -16,11 +15,7 @@ logger = logging.getLogger(__name__)
 
 
 class AliyunLogStorePG:
-    """
-    PostgreSQL protocol support for Aliyun SLS LogStore.
-
-    Handles PG connection pooling and operations for regions that support PG protocol.
-    """
+    """PostgreSQL protocol support for Aliyun SLS LogStore using SQLAlchemy connection pool."""
 
     def __init__(self, access_key_id: str, access_key_secret: str, endpoint: str, project_name: str):
         """
@@ -36,24 +31,11 @@ class AliyunLogStorePG:
         self._access_key_secret = access_key_secret
         self._endpoint = endpoint
         self.project_name = project_name
-        self._pg_pool: psycopg2.pool.SimpleConnectionPool | None = None
+        self._engine: Any = None  # SQLAlchemy Engine
         self._use_pg_protocol = False
 
     def _check_port_connectivity(self, host: str, port: int, timeout: float = 2.0) -> bool:
-        """
-        Check if a TCP port is reachable using socket connection.
-
-        This provides a fast check before attempting full database connection,
-        preventing long waits when connecting to unsupported regions.
-
-        Args:
-            host: Hostname or IP address
-            port: Port number
-            timeout: Connection timeout in seconds (default: 2.0)
-
-        Returns:
-            True if port is reachable, False otherwise
-        """
+        """Fast TCP port check to avoid long waits on unsupported regions."""
         try:
             sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
             sock.settimeout(timeout)
@@ -65,166 +47,101 @@ class AliyunLogStorePG:
             return False
 
     def init_connection(self) -> bool:
-        """
-        Initialize PostgreSQL connection pool for SLS PG protocol support.
-
-        Attempts to connect to SLS using PostgreSQL protocol. If successful, sets
-        _use_pg_protocol to True and creates a connection pool. If connection fails
-        (region doesn't support PG protocol or other errors), returns False.
-
-        Returns:
-            True if PG protocol is supported and initialized, False otherwise
-        """
+        """Initialize SQLAlchemy connection pool with pool_recycle and TCP keepalive support."""
         try:
-            # Extract hostname from endpoint (remove protocol if present)
             pg_host = self._endpoint.replace("http://", "").replace("https://", "")
 
-            # Get pool configuration
-            pg_max_connections = int(os.environ.get("ALIYUN_SLS_PG_MAX_CONNECTIONS", 10))
+            # Pool configuration
+            pool_size = int(os.environ.get("ALIYUN_SLS_PG_POOL_SIZE", 5))
+            max_overflow = int(os.environ.get("ALIYUN_SLS_PG_MAX_OVERFLOW", 5))
+            pool_recycle = int(os.environ.get("ALIYUN_SLS_PG_POOL_RECYCLE", 3600))
+            pool_pre_ping = os.environ.get("ALIYUN_SLS_PG_POOL_PRE_PING", "false").lower() == "true"
 
-            logger.debug(
-                "Check PG protocol connection to SLS: host=%s, project=%s",
-                pg_host,
-                self.project_name,
-            )
+            logger.debug("Check PG protocol connection to SLS: host=%s, project=%s", pg_host, self.project_name)
 
-            # Fast port connectivity check before attempting full connection
-            # This prevents long waits when connecting to unsupported regions
+            # Fast port check to avoid long waits
             if not self._check_port_connectivity(pg_host, 5432, timeout=1.0):
-                logger.info(
-                    "USE SDK mode for read/write operations, host=%s",
-                    pg_host,
-                )
+                logger.debug("Using SDK mode for host=%s", pg_host)
                 return False
 
-            # Create connection pool
-            self._pg_pool = psycopg2.pool.SimpleConnectionPool(
-                minconn=1,
-                maxconn=pg_max_connections,
-                host=pg_host,
-                port=5432,
-                database=self.project_name,
-                user=self._access_key_id,
-                password=self._access_key_secret,
-                sslmode="require",
-                connect_timeout=5,
-                application_name=f"Dify-{dify_config.project.version}",
+            # Build connection URL
+            from urllib.parse import quote_plus
+
+            username = quote_plus(self._access_key_id)
+            password = quote_plus(self._access_key_secret)
+            database_url = (
+                f"postgresql+psycopg2://{username}:{password}@{pg_host}:5432/{self.project_name}?sslmode=require"
             )
 
-            # Note: Skip test query because SLS PG protocol only supports SELECT/INSERT on actual tables
-            # Connection pool creation success already indicates connectivity
+            # Create SQLAlchemy engine with connection pool
+            self._engine = create_engine(
+                database_url,
+                pool_size=pool_size,
+                max_overflow=max_overflow,
+                pool_recycle=pool_recycle,
+                pool_pre_ping=pool_pre_ping,
+                pool_timeout=30,
+                connect_args={
+                    "connect_timeout": 5,
+                    "application_name": f"Dify-{dify_config.project.version}-fixautocommit",
+                    "keepalives": 1,
+                    "keepalives_idle": 60,
+                    "keepalives_interval": 10,
+                    "keepalives_count": 5,
+                },
+            )
 
             self._use_pg_protocol = True
             logger.info(
-                "PG protocol initialized successfully for SLS project=%s. Will use PG for read/write operations.",
+                "PG protocol initialized for SLS project=%s (pool_size=%d, pool_recycle=%ds)",
                 self.project_name,
+                pool_size,
+                pool_recycle,
             )
             return True
 
         except Exception as e:
-            # PG connection failed - fallback to SDK mode
             self._use_pg_protocol = False
-            if self._pg_pool:
+            if self._engine:
                 try:
-                    self._pg_pool.closeall()
+                    self._engine.dispose()
                 except Exception:
-                    logger.debug("Failed to close PG connection pool during cleanup, ignoring")
-            self._pg_pool = None
+                    logger.debug("Failed to dispose engine during cleanup, ignoring")
+            self._engine = None
 
-            logger.info(
-                "PG protocol connection failed (region may not support PG protocol): %s. "
-                "Falling back to SDK mode for read/write operations.",
-                str(e),
-            )
-            return False
-
-    def _is_connection_valid(self, conn: Any) -> bool:
-        """
-        Check if a connection is still valid.
-
-        Args:
-            conn: psycopg2 connection object
-
-        Returns:
-            True if connection is valid, False otherwise
-        """
-        try:
-            # Check if connection is closed
-            if conn.closed:
-                return False
-
-            # Quick ping test - execute a lightweight query
-            # For SLS PG protocol, we can't use SELECT 1 without FROM,
-            # so we just check the connection status
-            with conn.cursor() as cursor:
-                cursor.execute("SELECT 1")
-                cursor.fetchone()
-            return True
-        except Exception:
+            logger.debug("Using SDK mode for region: %s", str(e))
             return False
 
     @contextmanager
     def _get_connection(self):
-        """
-        Context manager to get a PostgreSQL connection from the pool.
+        """Get connection from SQLAlchemy pool. Pool handles recycle, invalidation, and keepalive automatically."""
+        if not self._engine:
+            raise RuntimeError("SQLAlchemy engine is not initialized")
 
-        Automatically validates and refreshes stale connections.
-
-        Note: Aliyun SLS PG protocol does not support transactions, so we always
-        use autocommit mode.
-
-        Yields:
-            psycopg2 connection object
-
-        Raises:
-            RuntimeError: If PG pool is not initialized
-        """
-        if not self._pg_pool:
-            raise RuntimeError("PG connection pool is not initialized")
-
-        conn = self._pg_pool.getconn()
+        connection = self._engine.raw_connection()
         try:
-            # Validate connection and get a fresh one if needed
-            if not self._is_connection_valid(conn):
-                logger.debug("Connection is stale, marking as bad and getting a new one")
-                # Mark connection as bad and get a new one
-                self._pg_pool.putconn(conn, close=True)
-                conn = self._pg_pool.getconn()
-
-            # Aliyun SLS PG protocol does not support transactions, always use autocommit
-            conn.autocommit = True
-            yield conn
+            connection.autocommit = True  # SLS PG protocol does not support transactions
+            yield connection
+        except Exception:
+            raise
         finally:
-            # Return connection to pool (or close if it's bad)
-            if self._is_connection_valid(conn):
-                self._pg_pool.putconn(conn)
-            else:
-                self._pg_pool.putconn(conn, close=True)
+            connection.close()
 
     def close(self) -> None:
-        """Close the PostgreSQL connection pool."""
-        if self._pg_pool:
+        """Dispose SQLAlchemy engine and close all connections."""
+        if self._engine:
             try:
-                self._pg_pool.closeall()
-                logger.info("PG connection pool closed")
+                self._engine.dispose()
+                logger.info("SQLAlchemy engine disposed")
             except Exception:
-                logger.exception("Failed to close PG connection pool")
+                logger.exception("Failed to dispose engine")
 
     def _is_retriable_error(self, error: Exception) -> bool:
-        """
-        Check if an error is retriable (connection-related issues).
-
-        Args:
-            error: Exception to check
-
-        Returns:
-            True if the error is retriable, False otherwise
-        """
-        # Retry on connection-related errors
-        if isinstance(error, (OperationalError, InterfaceError)):
+        """Check if error is retriable (connection-related issues)."""
+        # Check for psycopg2 connection errors directly
+        if isinstance(error, (psycopg2.OperationalError, psycopg2.InterfaceError)):
             return True
 
-        # Check error message for specific connection issues
         error_msg = str(error).lower()
         retriable_patterns = [
             "connection",
@@ -234,34 +151,18 @@ class AliyunLogStorePG:
             "reset by peer",
             "no route to host",
             "network",
+            "operational error",
+            "interface error",
         ]
         return any(pattern in error_msg for pattern in retriable_patterns)
 
     def put_log(self, logstore: str, contents: Sequence[tuple[str, str]], log_enabled: bool = False) -> None:
-        """
-        Write log to SLS using PostgreSQL protocol with automatic retry.
-
-        Note: SLS PG protocol only supports INSERT (not UPDATE). This uses append-only
-        writes with log_version field for versioning, same as SDK implementation.
-
-        Args:
-            logstore: Name of the logstore table
-            contents: List of (field_name, value) tuples
-            log_enabled: Whether to enable logging
-
-        Raises:
-            psycopg2.Error: If database operation fails after all retries
-        """
+        """Write log to SLS using INSERT with automatic retry (3 attempts with exponential backoff)."""
         if not contents:
             return
 
-        # Extract field names and values from contents
         fields = [field_name for field_name, _ in contents]
         values = [value for _, value in contents]
-
-        # Build INSERT statement with literal values
-        # Note: Aliyun SLS PG protocol doesn't support parameterized queries,
-        # so we need to use mogrify to safely create literal values
         field_list = ", ".join([f'"{field}"' for field in fields])
 
         if log_enabled:
@@ -272,67 +173,40 @@ class AliyunLogStorePG:
                 len(contents),
             )
 
-        # Retry configuration
         max_retries = 3
-        retry_delay = 0.1  # Start with 100ms
+        retry_delay = 0.1
 
         for attempt in range(max_retries):
             try:
                 with self._get_connection() as conn:
                     with conn.cursor() as cursor:
-                        # Use mogrify to safely convert values to SQL literals
                         placeholders = ", ".join(["%s"] * len(fields))
                         values_literal = cursor.mogrify(f"({placeholders})", values).decode("utf-8")
                         insert_sql = f'INSERT INTO "{logstore}" ({field_list}) VALUES {values_literal}'
                         cursor.execute(insert_sql)
-                # Success - exit retry loop
                 return
 
             except psycopg2.Error as e:
-                # Check if error is retriable
                 if not self._is_retriable_error(e):
-                    # Not a retriable error (e.g., data validation error), fail immediately
-                    logger.exception(
-                        "Failed to put logs to logstore %s via PG protocol (non-retriable error)",
-                        logstore,
-                    )
+                    logger.exception("Failed to put logs to logstore %s (non-retriable error)", logstore)
                     raise
 
-                # Retriable error - log and retry if we have attempts left
                 if attempt < max_retries - 1:
                     logger.warning(
-                        "Failed to put logs to logstore %s via PG protocol (attempt %d/%d): %s. Retrying...",
+                        "Failed to put logs to logstore %s (attempt %d/%d): %s. Retrying...",
                         logstore,
                         attempt + 1,
                         max_retries,
                         str(e),
                     )
                     time.sleep(retry_delay)
-                    retry_delay *= 2  # Exponential backoff
+                    retry_delay *= 2
                 else:
-                    # Last attempt failed
-                    logger.exception(
-                        "Failed to put logs to logstore %s via PG protocol after %d attempts",
-                        logstore,
-                        max_retries,
-                    )
+                    logger.exception("Failed to put logs to logstore %s after %d attempts", logstore, max_retries)
                     raise
 
     def execute_sql(self, sql: str, logstore: str, log_enabled: bool = False) -> list[dict[str, Any]]:
-        """
-        Execute SQL query using PostgreSQL protocol with automatic retry.
-
-        Args:
-            sql: SQL query string
-            logstore: Name of the logstore (for logging purposes)
-            log_enabled: Whether to enable logging
-
-        Returns:
-            List of result rows as dictionaries
-
-        Raises:
-            psycopg2.Error: If database operation fails after all retries
-        """
+        """Execute SQL query with automatic retry (3 attempts with exponential backoff)."""
         if log_enabled:
             logger.info(
                 "[LogStore-PG] EXECUTE_SQL | logstore=%s | project=%s | sql=%s",
@@ -341,20 +215,16 @@ class AliyunLogStorePG:
                 sql,
             )
 
-        # Retry configuration
         max_retries = 3
-        retry_delay = 0.1  # Start with 100ms
+        retry_delay = 0.1
 
         for attempt in range(max_retries):
             try:
                 with self._get_connection() as conn:
                     with conn.cursor() as cursor:
                         cursor.execute(sql)
-
-                        # Get column names from cursor description
                         columns = [desc[0] for desc in cursor.description]
 
-                        # Fetch all results and convert to list of dicts
                         result = []
                         for row in cursor.fetchall():
                             row_dict = {}
@@ -372,36 +242,31 @@ class AliyunLogStorePG:
                         return result
 
             except psycopg2.Error as e:
-                # Check if error is retriable
                 if not self._is_retriable_error(e):
-                    # Not a retriable error (e.g., SQL syntax error), fail immediately
                     logger.exception(
-                        "Failed to execute SQL query on logstore %s via PG protocol (non-retriable error): sql=%s",
+                        "Failed to execute SQL on logstore %s (non-retriable error): sql=%s",
                         logstore,
                         sql,
                     )
                     raise
 
-                # Retriable error - log and retry if we have attempts left
                 if attempt < max_retries - 1:
                     logger.warning(
-                        "Failed to execute SQL query on logstore %s via PG protocol (attempt %d/%d): %s. Retrying...",
+                        "Failed to execute SQL on logstore %s (attempt %d/%d): %s. Retrying...",
                         logstore,
                         attempt + 1,
                         max_retries,
                         str(e),
                     )
                     time.sleep(retry_delay)
-                    retry_delay *= 2  # Exponential backoff
+                    retry_delay *= 2
                 else:
-                    # Last attempt failed
                     logger.exception(
-                        "Failed to execute SQL query on logstore %s via PG protocol after %d attempts: sql=%s",
+                        "Failed to execute SQL on logstore %s after %d attempts: sql=%s",
                         logstore,
                         max_retries,
                         sql,
                     )
                     raise
 
-        # This line should never be reached due to raise above, but makes type checker happy
         return []

api/extensions/logstore/repositories/__init__.py

@@ -0,0 +1,29 @@
+"""
+LogStore repository utilities.
+"""
+
+from typing import Any
+
+
+def safe_float(value: Any, default: float = 0.0) -> float:
+    """
+    Safely convert a value to float, handling 'null' strings and None.
+    """
+    if value is None or value in {"null", ""}:
+        return default
+    try:
+        return float(value)
+    except (ValueError, TypeError):
+        return default
+
+
+def safe_int(value: Any, default: int = 0) -> int:
+    """
+    Safely convert a value to int, handling 'null' strings and None.
+    """
+    if value is None or value in {"null", ""}:
+        return default
+    try:
+        return int(float(value))
+    except (ValueError, TypeError):
+        return default

api/extensions/logstore/repositories/logstore_api_workflow_node_execution_repository.py

@@ -14,6 +14,8 @@ from typing import Any
 from sqlalchemy.orm import sessionmaker
 
 from extensions.logstore.aliyun_logstore import AliyunLogStore
+from extensions.logstore.repositories import safe_float, safe_int
+from extensions.logstore.sql_escape import escape_identifier, escape_logstore_query_value
 from models.workflow import WorkflowNodeExecutionModel
 from repositories.api_workflow_node_execution_repository import DifyAPIWorkflowNodeExecutionRepository
 
@@ -52,9 +54,8 @@ def _dict_to_workflow_node_execution_model(data: dict[str, Any]) -> WorkflowNode
     model.created_by_role = data.get("created_by_role") or ""
     model.created_by = data.get("created_by") or ""
 
-    # Numeric fields with defaults
-    model.index = int(data.get("index", 0))
-    model.elapsed_time = float(data.get("elapsed_time", 0))
+    model.index = safe_int(data.get("index", 0))
+    model.elapsed_time = safe_float(data.get("elapsed_time", 0))
 
     # Optional fields
     model.workflow_run_id = data.get("workflow_run_id")
@@ -130,6 +131,12 @@ class LogstoreAPIWorkflowNodeExecutionRepository(DifyAPIWorkflowNodeExecutionRep
             node_id,
         )
         try:
+            # Escape parameters to prevent SQL injection
+            escaped_tenant_id = escape_identifier(tenant_id)
+            escaped_app_id = escape_identifier(app_id)
+            escaped_workflow_id = escape_identifier(workflow_id)
+            escaped_node_id = escape_identifier(node_id)
+
             # Check if PG protocol is supported
             if self.logstore_client.supports_pg_protocol:
                 # Use PG protocol with SQL query (get latest version of each record)
@@ -138,10 +145,10 @@ class LogstoreAPIWorkflowNodeExecutionRepository(DifyAPIWorkflowNodeExecutionRep
                         SELECT *, 
                             ROW_NUMBER() OVER (PARTITION BY id ORDER BY log_version DESC) as rn
                         FROM "{AliyunLogStore.workflow_node_execution_logstore}"
-                        WHERE tenant_id = '{tenant_id}' 
-                          AND app_id = '{app_id}' 
-                          AND workflow_id = '{workflow_id}' 
-                          AND node_id = '{node_id}'
+                        WHERE tenant_id = '{escaped_tenant_id}' 
+                          AND app_id = '{escaped_app_id}' 
+                          AND workflow_id = '{escaped_workflow_id}' 
+                          AND node_id = '{escaped_node_id}'
                           AND __time__ > 0
                     ) AS subquery WHERE rn = 1
                     LIMIT 100
@@ -153,7 +160,8 @@ class LogstoreAPIWorkflowNodeExecutionRepository(DifyAPIWorkflowNodeExecutionRep
             else:
                 # Use SDK with LogStore query syntax
                 query = (
-                    f"tenant_id: {tenant_id} and app_id: {app_id} and workflow_id: {workflow_id} and node_id: {node_id}"
+                    f"tenant_id: {escaped_tenant_id} and app_id: {escaped_app_id} "
+                    f"and workflow_id: {escaped_workflow_id} and node_id: {escaped_node_id}"
                 )
                 from_time = 0
                 to_time = int(time.time())  # now
@@ -227,6 +235,11 @@ class LogstoreAPIWorkflowNodeExecutionRepository(DifyAPIWorkflowNodeExecutionRep
             workflow_run_id,
         )
         try:
+            # Escape parameters to prevent SQL injection
+            escaped_tenant_id = escape_identifier(tenant_id)
+            escaped_app_id = escape_identifier(app_id)
+            escaped_workflow_run_id = escape_identifier(workflow_run_id)
+
             # Check if PG protocol is supported
             if self.logstore_client.supports_pg_protocol:
                 # Use PG protocol with SQL query (get latest version of each record)
@@ -235,9 +248,9 @@ class LogstoreAPIWorkflowNodeExecutionRepository(DifyAPIWorkflowNodeExecutionRep
                         SELECT *, 
                             ROW_NUMBER() OVER (PARTITION BY id ORDER BY log_version DESC) as rn
                         FROM "{AliyunLogStore.workflow_node_execution_logstore}"
-                        WHERE tenant_id = '{tenant_id}' 
-                          AND app_id = '{app_id}' 
-                          AND workflow_run_id = '{workflow_run_id}'
+                        WHERE tenant_id = '{escaped_tenant_id}' 
+                          AND app_id = '{escaped_app_id}' 
+                          AND workflow_run_id = '{escaped_workflow_run_id}'
                           AND __time__ > 0
                     ) AS subquery WHERE rn = 1
                     LIMIT 1000
@@ -248,7 +261,10 @@ class LogstoreAPIWorkflowNodeExecutionRepository(DifyAPIWorkflowNodeExecutionRep
                 )
             else:
                 # Use SDK with LogStore query syntax
-                query = f"tenant_id: {tenant_id} and app_id: {app_id} and workflow_run_id: {workflow_run_id}"
+                query = (
+                    f"tenant_id: {escaped_tenant_id} and app_id: {escaped_app_id} "
+                    f"and workflow_run_id: {escaped_workflow_run_id}"
+                )
                 from_time = 0
                 to_time = int(time.time())  # now
 
@@ -313,16 +329,24 @@ class LogstoreAPIWorkflowNodeExecutionRepository(DifyAPIWorkflowNodeExecutionRep
         """
         logger.debug("get_execution_by_id: execution_id=%s, tenant_id=%s", execution_id, tenant_id)
         try:
+            # Escape parameters to prevent SQL injection
+            escaped_execution_id = escape_identifier(execution_id)
+
             # Check if PG protocol is supported
             if self.logstore_client.supports_pg_protocol:
                 # Use PG protocol with SQL query (get latest version of record)
-                tenant_filter = f"AND tenant_id = '{tenant_id}'" if tenant_id else ""
+                if tenant_id:
+                    escaped_tenant_id = escape_identifier(tenant_id)
+                    tenant_filter = f"AND tenant_id = '{escaped_tenant_id}'"
+                else:
+                    tenant_filter = ""
+
                 sql_query = f"""
                     SELECT * FROM (
                         SELECT *, 
                             ROW_NUMBER() OVER (PARTITION BY id ORDER BY log_version DESC) as rn
                         FROM "{AliyunLogStore.workflow_node_execution_logstore}"
-                        WHERE id = '{execution_id}' {tenant_filter} AND __time__ > 0
+                        WHERE id = '{escaped_execution_id}' {tenant_filter} AND __time__ > 0
                     ) AS subquery WHERE rn = 1
                     LIMIT 1
                 """
@@ -332,10 +356,14 @@ class LogstoreAPIWorkflowNodeExecutionRepository(DifyAPIWorkflowNodeExecutionRep
                 )
             else:
                 # Use SDK with LogStore query syntax
+                # Note: Values must be quoted in LogStore query syntax to prevent injection
                 if tenant_id:
-                    query = f"id: {execution_id} and tenant_id: {tenant_id}"
+                    query = (
+                        f"id:{escape_logstore_query_value(execution_id)} "
+                        f"and tenant_id:{escape_logstore_query_value(tenant_id)}"
+                    )
                 else:
-                    query = f"id: {execution_id}"
+                    query = f"id:{escape_logstore_query_value(execution_id)}"
 
                 from_time = 0
                 to_time = int(time.time())  # now

api/extensions/logstore/repositories/logstore_api_workflow_run_repository.py

@@ -10,6 +10,7 @@ Key Features:
 - Optimized deduplication using finished_at IS NOT NULL filter
 - Window functions only when necessary (running status queries)
 - Multi-tenant data isolation and security
+- SQL injection prevention via parameter escaping
 """
 
 import logging
@@ -22,6 +23,8 @@ from typing import Any, cast
 from sqlalchemy.orm import sessionmaker
 
 from extensions.logstore.aliyun_logstore import AliyunLogStore
+from extensions.logstore.repositories import safe_float, safe_int
+from extensions.logstore.sql_escape import escape_identifier, escape_logstore_query_value, escape_sql_string
 from libs.infinite_scroll_pagination import InfiniteScrollPagination
 from models.enums import WorkflowRunTriggeredFrom
 from models.workflow import WorkflowRun
@@ -63,10 +66,9 @@ def _dict_to_workflow_run(data: dict[str, Any]) -> WorkflowRun:
     model.created_by_role = data.get("created_by_role") or ""
     model.created_by = data.get("created_by") or ""
 
-    # Numeric fields with defaults
-    model.total_tokens = int(data.get("total_tokens", 0))
-    model.total_steps = int(data.get("total_steps", 0))
-    model.exceptions_count = int(data.get("exceptions_count", 0))
+    model.total_tokens = safe_int(data.get("total_tokens", 0))
+    model.total_steps = safe_int(data.get("total_steps", 0))
+    model.exceptions_count = safe_int(data.get("exceptions_count", 0))
 
     # Optional fields
     model.graph = data.get("graph")
@@ -101,7 +103,8 @@ def _dict_to_workflow_run(data: dict[str, Any]) -> WorkflowRun:
     if model.finished_at and model.created_at:
         model.elapsed_time = (model.finished_at - model.created_at).total_seconds()
     else:
-        model.elapsed_time = float(data.get("elapsed_time", 0))
+        # Use safe conversion to handle 'null' strings and None values
+        model.elapsed_time = safe_float(data.get("elapsed_time", 0))
 
     return model
 
@@ -165,16 +168,26 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
             status,
         )
         # Convert triggered_from to list if needed
-        if isinstance(triggered_from, WorkflowRunTriggeredFrom):
+        if isinstance(triggered_from, (WorkflowRunTriggeredFrom, str)):
             triggered_from_list = [triggered_from]
         else:
             triggered_from_list = list(triggered_from)
 
-        # Build triggered_from filter
-        triggered_from_filter = " OR ".join([f"triggered_from='{tf.value}'" for tf in triggered_from_list])
+        # Escape parameters to prevent SQL injection
+        escaped_tenant_id = escape_identifier(tenant_id)
+        escaped_app_id = escape_identifier(app_id)
 
-        # Build status filter
-        status_filter = f"AND status='{status}'" if status else ""
+        # Build triggered_from filter with escaped values
+        # Support both enum and string values for triggered_from
+        triggered_from_filter = " OR ".join(
+            [
+                f"triggered_from='{escape_sql_string(tf.value if isinstance(tf, WorkflowRunTriggeredFrom) else tf)}'"
+                for tf in triggered_from_list
+            ]
+        )
+
+        # Build status filter with escaped value
+        status_filter = f"AND status='{escape_sql_string(status)}'" if status else ""
 
         # Build last_id filter for pagination
         # Note: This is simplified. In production, you'd need to track created_at from last record
@@ -188,8 +201,8 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
             SELECT * FROM (
                 SELECT *, ROW_NUMBER() OVER (PARTITION BY id ORDER BY log_version DESC) AS rn
                 FROM {AliyunLogStore.workflow_execution_logstore}
-                WHERE tenant_id='{tenant_id}'
-                  AND app_id='{app_id}'
+                WHERE tenant_id='{escaped_tenant_id}'
+                  AND app_id='{escaped_app_id}'
                   AND ({triggered_from_filter})
                   {status_filter}
                   {last_id_filter}
@@ -232,6 +245,11 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
         logger.debug("get_workflow_run_by_id: tenant_id=%s, app_id=%s, run_id=%s", tenant_id, app_id, run_id)
 
         try:
+            # Escape parameters to prevent SQL injection
+            escaped_run_id = escape_identifier(run_id)
+            escaped_tenant_id = escape_identifier(tenant_id)
+            escaped_app_id = escape_identifier(app_id)
+
             # Check if PG protocol is supported
             if self.logstore_client.supports_pg_protocol:
                 # Use PG protocol with SQL query (get latest version of record)
@@ -240,7 +258,10 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
                         SELECT *, 
                             ROW_NUMBER() OVER (PARTITION BY id ORDER BY log_version DESC) as rn
                         FROM "{AliyunLogStore.workflow_execution_logstore}"
-                        WHERE id = '{run_id}' AND tenant_id = '{tenant_id}' AND app_id = '{app_id}' AND __time__ > 0
+                        WHERE id = '{escaped_run_id}' 
+                          AND tenant_id = '{escaped_tenant_id}' 
+                          AND app_id = '{escaped_app_id}' 
+                          AND __time__ > 0
                     ) AS subquery WHERE rn = 1
                     LIMIT 100
                 """
@@ -250,7 +271,12 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
                 )
             else:
                 # Use SDK with LogStore query syntax
-                query = f"id: {run_id} and tenant_id: {tenant_id} and app_id: {app_id}"
+                # Note: Values must be quoted in LogStore query syntax to prevent injection
+                query = (
+                    f"id:{escape_logstore_query_value(run_id)} "
+                    f"and tenant_id:{escape_logstore_query_value(tenant_id)} "
+                    f"and app_id:{escape_logstore_query_value(app_id)}"
+                )
                 from_time = 0
                 to_time = int(time.time())  # now
 
@@ -323,6 +349,9 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
         logger.debug("get_workflow_run_by_id_without_tenant: run_id=%s", run_id)
 
         try:
+            # Escape parameter to prevent SQL injection
+            escaped_run_id = escape_identifier(run_id)
+
             # Check if PG protocol is supported
             if self.logstore_client.supports_pg_protocol:
                 # Use PG protocol with SQL query (get latest version of record)
@@ -331,7 +360,7 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
                         SELECT *, 
                             ROW_NUMBER() OVER (PARTITION BY id ORDER BY log_version DESC) as rn
                         FROM "{AliyunLogStore.workflow_execution_logstore}"
-                        WHERE id = '{run_id}' AND __time__ > 0
+                        WHERE id = '{escaped_run_id}' AND __time__ > 0
                     ) AS subquery WHERE rn = 1
                     LIMIT 100
                 """
@@ -341,7 +370,8 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
                 )
             else:
                 # Use SDK with LogStore query syntax
-                query = f"id: {run_id}"
+                # Note: Values must be quoted in LogStore query syntax
+                query = f"id:{escape_logstore_query_value(run_id)}"
                 from_time = 0
                 to_time = int(time.time())  # now
 
@@ -410,6 +440,11 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
             triggered_from,
             status,
         )
+        # Escape parameters to prevent SQL injection
+        escaped_tenant_id = escape_identifier(tenant_id)
+        escaped_app_id = escape_identifier(app_id)
+        escaped_triggered_from = escape_sql_string(triggered_from)
+
         # Build time range filter
         time_filter = ""
         if time_range:
@@ -418,6 +453,8 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
 
         # If status is provided, simple count
         if status:
+            escaped_status = escape_sql_string(status)
+
             if status == "running":
                 # Running status requires window function
                 sql = f"""
@@ -425,9 +462,9 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
                     FROM (
                         SELECT *, ROW_NUMBER() OVER (PARTITION BY id ORDER BY log_version DESC) AS rn
                         FROM {AliyunLogStore.workflow_execution_logstore}
-                        WHERE tenant_id='{tenant_id}'
-                          AND app_id='{app_id}'
-                          AND triggered_from='{triggered_from}'
+                        WHERE tenant_id='{escaped_tenant_id}'
+                          AND app_id='{escaped_app_id}'
+                          AND triggered_from='{escaped_triggered_from}'
                           AND status='running'
                           {time_filter}
                     ) t
@@ -438,10 +475,10 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
                 sql = f"""
                     SELECT COUNT(DISTINCT id) as count
                     FROM {AliyunLogStore.workflow_execution_logstore}
-                    WHERE tenant_id='{tenant_id}'
-                      AND app_id='{app_id}'
-                      AND triggered_from='{triggered_from}'
-                      AND status='{status}'
+                    WHERE tenant_id='{escaped_tenant_id}'
+                      AND app_id='{escaped_app_id}'
+                      AND triggered_from='{escaped_triggered_from}'
+                      AND status='{escaped_status}'
                       AND finished_at IS NOT NULL
                       {time_filter}
                 """
@@ -467,13 +504,14 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
         # No status filter - get counts grouped by status
         # Use optimized query for finished runs, separate query for running
         try:
+            # Escape parameters (already escaped above, reuse variables)
             # Count finished runs grouped by status
             finished_sql = f"""
                 SELECT status, COUNT(DISTINCT id) as count
                 FROM {AliyunLogStore.workflow_execution_logstore}
-                WHERE tenant_id='{tenant_id}'
-                  AND app_id='{app_id}'
-                  AND triggered_from='{triggered_from}'
+                WHERE tenant_id='{escaped_tenant_id}'
+                  AND app_id='{escaped_app_id}'
+                  AND triggered_from='{escaped_triggered_from}'
                   AND finished_at IS NOT NULL
                   {time_filter}
                 GROUP BY status
@@ -485,9 +523,9 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
                 FROM (
                     SELECT *, ROW_NUMBER() OVER (PARTITION BY id ORDER BY log_version DESC) AS rn
                     FROM {AliyunLogStore.workflow_execution_logstore}
-                    WHERE tenant_id='{tenant_id}'
-                      AND app_id='{app_id}'
-                      AND triggered_from='{triggered_from}'
+                    WHERE tenant_id='{escaped_tenant_id}'
+                      AND app_id='{escaped_app_id}'
+                      AND triggered_from='{escaped_triggered_from}'
                       AND status='running'
                       {time_filter}
                 ) t
@@ -546,7 +584,13 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
         logger.debug(
             "get_daily_runs_statistics: tenant_id=%s, app_id=%s, triggered_from=%s", tenant_id, app_id, triggered_from
         )
-        # Build time range filter
+
+        # Escape parameters to prevent SQL injection
+        escaped_tenant_id = escape_identifier(tenant_id)
+        escaped_app_id = escape_identifier(app_id)
+        escaped_triggered_from = escape_sql_string(triggered_from)
+
+        # Build time range filter (datetime.isoformat() is safe)
         time_filter = ""
         if start_date:
             time_filter += f" AND __time__ >= to_unixtime(from_iso8601_timestamp('{start_date.isoformat()}'))"
@@ -557,9 +601,9 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
         sql = f"""
             SELECT DATE(from_unixtime(__time__)) as date, COUNT(DISTINCT id) as runs
             FROM {AliyunLogStore.workflow_execution_logstore}
-            WHERE tenant_id='{tenant_id}'
-              AND app_id='{app_id}'
-              AND triggered_from='{triggered_from}'
+            WHERE tenant_id='{escaped_tenant_id}'
+              AND app_id='{escaped_app_id}'
+              AND triggered_from='{escaped_triggered_from}'
               AND finished_at IS NOT NULL
               {time_filter}
             GROUP BY date
@@ -601,7 +645,13 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
             app_id,
             triggered_from,
         )
-        # Build time range filter
+
+        # Escape parameters to prevent SQL injection
+        escaped_tenant_id = escape_identifier(tenant_id)
+        escaped_app_id = escape_identifier(app_id)
+        escaped_triggered_from = escape_sql_string(triggered_from)
+
+        # Build time range filter (datetime.isoformat() is safe)
         time_filter = ""
         if start_date:
             time_filter += f" AND __time__ >= to_unixtime(from_iso8601_timestamp('{start_date.isoformat()}'))"
@@ -611,9 +661,9 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
         sql = f"""
             SELECT DATE(from_unixtime(__time__)) as date, COUNT(DISTINCT created_by) as terminal_count
             FROM {AliyunLogStore.workflow_execution_logstore}
-            WHERE tenant_id='{tenant_id}'
-              AND app_id='{app_id}'
-              AND triggered_from='{triggered_from}'
+            WHERE tenant_id='{escaped_tenant_id}'
+              AND app_id='{escaped_app_id}'
+              AND triggered_from='{escaped_triggered_from}'
               AND finished_at IS NOT NULL
               {time_filter}
             GROUP BY date
@@ -655,7 +705,13 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
             app_id,
             triggered_from,
         )
-        # Build time range filter
+
+        # Escape parameters to prevent SQL injection
+        escaped_tenant_id = escape_identifier(tenant_id)
+        escaped_app_id = escape_identifier(app_id)
+        escaped_triggered_from = escape_sql_string(triggered_from)
+
+        # Build time range filter (datetime.isoformat() is safe)
         time_filter = ""
         if start_date:
             time_filter += f" AND __time__ >= to_unixtime(from_iso8601_timestamp('{start_date.isoformat()}'))"
@@ -665,9 +721,9 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
         sql = f"""
             SELECT DATE(from_unixtime(__time__)) as date, SUM(total_tokens) as token_count
             FROM {AliyunLogStore.workflow_execution_logstore}
-            WHERE tenant_id='{tenant_id}'
-              AND app_id='{app_id}'
-              AND triggered_from='{triggered_from}'
+            WHERE tenant_id='{escaped_tenant_id}'
+              AND app_id='{escaped_app_id}'
+              AND triggered_from='{escaped_triggered_from}'
               AND finished_at IS NOT NULL
               {time_filter}
             GROUP BY date
@@ -709,7 +765,13 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
             app_id,
             triggered_from,
         )
-        # Build time range filter
+
+        # Escape parameters to prevent SQL injection
+        escaped_tenant_id = escape_identifier(tenant_id)
+        escaped_app_id = escape_identifier(app_id)
+        escaped_triggered_from = escape_sql_string(triggered_from)
+
+        # Build time range filter (datetime.isoformat() is safe)
         time_filter = ""
         if start_date:
             time_filter += f" AND __time__ >= to_unixtime(from_iso8601_timestamp('{start_date.isoformat()}'))"
@@ -726,9 +788,9 @@ class LogstoreAPIWorkflowRunRepository(APIWorkflowRunRepository):
                     created_by,
                     COUNT(DISTINCT id) AS interactions
                 FROM {AliyunLogStore.workflow_execution_logstore}
-                WHERE tenant_id='{tenant_id}'
-                  AND app_id='{app_id}'
-                  AND triggered_from='{triggered_from}'
+                WHERE tenant_id='{escaped_tenant_id}'
+                  AND app_id='{escaped_app_id}'
+                  AND triggered_from='{escaped_triggered_from}'
                   AND finished_at IS NOT NULL
                   {time_filter}
                 GROUP BY date, created_by

api/extensions/logstore/repositories/logstore_workflow_execution_repository.py

@@ -10,6 +10,7 @@ from sqlalchemy.orm import sessionmaker
 from core.repositories.sqlalchemy_workflow_execution_repository import SQLAlchemyWorkflowExecutionRepository
 from core.workflow.entities import WorkflowExecution
 from core.workflow.repositories.workflow_execution_repository import WorkflowExecutionRepository
+from core.workflow.workflow_type_encoder import WorkflowRuntimeTypeConverter
 from extensions.logstore.aliyun_logstore import AliyunLogStore
 from libs.helper import extract_tenant_id
 from models import (
@@ -22,18 +23,6 @@ from models.enums import WorkflowRunTriggeredFrom
 logger = logging.getLogger(__name__)
 
 
-def to_serializable(obj):
-    """
-    Convert non-JSON-serializable objects into JSON-compatible formats.
-
-    - Uses `to_dict()` if it's a callable method.
-    - Falls back to string representation.
-    """
-    if hasattr(obj, "to_dict") and callable(obj.to_dict):
-        return obj.to_dict()
-    return str(obj)
-
-
 class LogstoreWorkflowExecutionRepository(WorkflowExecutionRepository):
     def __init__(
         self,
@@ -79,7 +68,7 @@ class LogstoreWorkflowExecutionRepository(WorkflowExecutionRepository):
 
         # Control flag for dual-write (write to both LogStore and SQL database)
         # Set to True to enable dual-write for safe migration, False to use LogStore only
-        self._enable_dual_write = os.environ.get("LOGSTORE_DUAL_WRITE_ENABLED", "true").lower() == "true"
+        self._enable_dual_write = os.environ.get("LOGSTORE_DUAL_WRITE_ENABLED", "false").lower() == "true"
 
         # Control flag for whether to write the `graph` field to LogStore.
         # If LOGSTORE_ENABLE_PUT_GRAPH_FIELD is "true", write the full `graph` field;
@@ -113,6 +102,9 @@ class LogstoreWorkflowExecutionRepository(WorkflowExecutionRepository):
         # Generate log_version as nanosecond timestamp for record versioning
         log_version = str(time.time_ns())
 
+        # Use WorkflowRuntimeTypeConverter to handle complex types (Segment, File, etc.)
+        json_converter = WorkflowRuntimeTypeConverter()
+
         logstore_model = [
             ("id", domain_model.id_),
             ("log_version", log_version),  # Add log_version field for append-only writes
@@ -127,19 +119,19 @@ class LogstoreWorkflowExecutionRepository(WorkflowExecutionRepository):
             ("version", domain_model.workflow_version),
             (
                 "graph",
-                json.dumps(domain_model.graph, ensure_ascii=False, default=to_serializable)
+                json.dumps(json_converter.to_json_encodable(domain_model.graph), ensure_ascii=False)
                 if domain_model.graph and self._enable_put_graph_field
                 else "{}",
             ),
             (
                 "inputs",
-                json.dumps(domain_model.inputs, ensure_ascii=False, default=to_serializable)
+                json.dumps(json_converter.to_json_encodable(domain_model.inputs), ensure_ascii=False)
                 if domain_model.inputs
                 else "{}",
             ),
             (
                 "outputs",
-                json.dumps(domain_model.outputs, ensure_ascii=False, default=to_serializable)
+                json.dumps(json_converter.to_json_encodable(domain_model.outputs), ensure_ascii=False)
                 if domain_model.outputs
                 else "{}",
             ),

api/extensions/logstore/repositories/logstore_workflow_node_execution_repository.py

@@ -24,6 +24,8 @@ from core.workflow.enums import NodeType
 from core.workflow.repositories.workflow_node_execution_repository import OrderConfig, WorkflowNodeExecutionRepository
 from core.workflow.workflow_type_encoder import WorkflowRuntimeTypeConverter
 from extensions.logstore.aliyun_logstore import AliyunLogStore
+from extensions.logstore.repositories import safe_float, safe_int
+from extensions.logstore.sql_escape import escape_identifier
 from libs.helper import extract_tenant_id
 from models import (
     Account,
@@ -73,7 +75,7 @@ def _dict_to_workflow_node_execution(data: dict[str, Any]) -> WorkflowNodeExecut
         node_execution_id=data.get("node_execution_id"),
         workflow_id=data.get("workflow_id", ""),
         workflow_execution_id=data.get("workflow_run_id"),
-        index=int(data.get("index", 0)),
+        index=safe_int(data.get("index", 0)),
         predecessor_node_id=data.get("predecessor_node_id"),
         node_id=data.get("node_id", ""),
         node_type=NodeType(data.get("node_type", "start")),
@@ -83,7 +85,7 @@ def _dict_to_workflow_node_execution(data: dict[str, Any]) -> WorkflowNodeExecut
         outputs=outputs,
         status=status,
         error=data.get("error"),
-        elapsed_time=float(data.get("elapsed_time", 0.0)),
+        elapsed_time=safe_float(data.get("elapsed_time", 0.0)),
         metadata=domain_metadata,
         created_at=created_at,
         finished_at=finished_at,
@@ -147,7 +149,7 @@ class LogstoreWorkflowNodeExecutionRepository(WorkflowNodeExecutionRepository):
 
         # Control flag for dual-write (write to both LogStore and SQL database)
         # Set to True to enable dual-write for safe migration, False to use LogStore only
-        self._enable_dual_write = os.environ.get("LOGSTORE_DUAL_WRITE_ENABLED", "true").lower() == "true"
+        self._enable_dual_write = os.environ.get("LOGSTORE_DUAL_WRITE_ENABLED", "false").lower() == "true"
 
     def _to_logstore_model(self, domain_model: WorkflowNodeExecution) -> Sequence[tuple[str, str]]:
         logger.debug(
@@ -274,16 +276,34 @@ class LogstoreWorkflowNodeExecutionRepository(WorkflowNodeExecutionRepository):
         Save or update the inputs, process_data, or outputs associated with a specific
         node_execution record.
 
-        For LogStore implementation, this is similar to save() since we always write
-        complete records. We append a new record with updated data fields.
+        For LogStore implementation, this is a no-op for the LogStore write because save()
+        already writes all fields including inputs, process_data, and outputs. The caller
+        typically calls save() first to persist status/metadata, then calls save_execution_data()
+        to persist data fields. Since LogStore writes complete records atomically, we don't
+        need a separate write here to avoid duplicate records.
+
+        However, if dual-write is enabled, we still need to call the SQL repository's
+        save_execution_data() method to properly update the SQL database.
 
         Args:
             execution: The NodeExecution instance with data to save
         """
-        logger.debug("save_execution_data: id=%s, node_execution_id=%s", execution.id, execution.node_execution_id)
-        # In LogStore, we simply write a new complete record with the data
-        # The log_version timestamp will ensure this is treated as the latest version
-        self.save(execution)
+        logger.debug(
+            "save_execution_data: no-op for LogStore (data already saved by save()): id=%s, node_execution_id=%s",
+            execution.id,
+            execution.node_execution_id,
+        )
+        # No-op for LogStore: save() already writes all fields including inputs, process_data, and outputs
+        # Calling save() again would create a duplicate record in the append-only LogStore
+
+        # Dual-write to SQL database if enabled (for safe migration)
+        if self._enable_dual_write:
+            try:
+                self.sql_repository.save_execution_data(execution)
+                logger.debug("Dual-write: saved node execution data to SQL database: id=%s", execution.id)
+            except Exception:
+                logger.exception("Failed to dual-write node execution data to SQL database: id=%s", execution.id)
+                # Don't raise - LogStore write succeeded, SQL is just a backup
 
     def get_by_workflow_run(
         self,
@@ -292,8 +312,8 @@ class LogstoreWorkflowNodeExecutionRepository(WorkflowNodeExecutionRepository):
     ) -> Sequence[WorkflowNodeExecution]:
         """
         Retrieve all NodeExecution instances for a specific workflow run.
-        Uses LogStore SQL query with finished_at IS NOT NULL filter for deduplication.
-        This ensures we only get the final version of each node execution.
+        Uses LogStore SQL query with window function to get the latest version of each node execution.
+        This ensures we only get the most recent version of each node execution record.
         Args:
             workflow_run_id: The workflow run ID
             order_config: Optional configuration for ordering results
@@ -304,16 +324,19 @@ class LogstoreWorkflowNodeExecutionRepository(WorkflowNodeExecutionRepository):
             A list of NodeExecution instances
 
         Note:
-            This method filters by finished_at IS NOT NULL to avoid duplicates from
-            version updates. For complete history including intermediate states,
-            a different query strategy would be needed.
+            This method uses ROW_NUMBER() window function partitioned by node_execution_id
+            to get the latest version (highest log_version) of each node execution.
         """
         logger.debug("get_by_workflow_run: workflow_run_id=%s, order_config=%s", workflow_run_id, order_config)
-        # Build SQL query with deduplication using finished_at IS NOT NULL
-        # This optimization avoids window functions for common case where we only
-        # want the final state of each node execution
+        # Build SQL query with deduplication using window function
+        # ROW_NUMBER() OVER (PARTITION BY node_execution_id ORDER BY log_version DESC)
+        # ensures we get the latest version of each node execution
 
-        # Build ORDER BY clause
+        # Escape parameters to prevent SQL injection
+        escaped_workflow_run_id = escape_identifier(workflow_run_id)
+        escaped_tenant_id = escape_identifier(self._tenant_id)
+
+        # Build ORDER BY clause for outer query
         order_clause = ""
         if order_config and order_config.order_by:
             order_fields = []
@@ -327,16 +350,23 @@ class LogstoreWorkflowNodeExecutionRepository(WorkflowNodeExecutionRepository):
             if order_fields:
                 order_clause = "ORDER BY " + ", ".join(order_fields)
 
-        sql = f"""
-            SELECT *
-            FROM {AliyunLogStore.workflow_node_execution_logstore}
-            WHERE workflow_run_id='{workflow_run_id}'
-              AND tenant_id='{self._tenant_id}'
-              AND finished_at IS NOT NULL
-        """
-
+        # Build app_id filter for subquery
+        app_id_filter = ""
         if self._app_id:
-            sql += f" AND app_id='{self._app_id}'"
+            escaped_app_id = escape_identifier(self._app_id)
+            app_id_filter = f" AND app_id='{escaped_app_id}'"
+
+        # Use window function to get latest version of each node execution
+        sql = f"""
+            SELECT * FROM (
+                SELECT *, ROW_NUMBER() OVER (PARTITION BY node_execution_id ORDER BY log_version DESC) AS rn
+                FROM {AliyunLogStore.workflow_node_execution_logstore}
+                WHERE workflow_run_id='{escaped_workflow_run_id}'
+                  AND tenant_id='{escaped_tenant_id}'
+                  {app_id_filter}
+            ) t
+            WHERE rn = 1
+        """
 
         if order_clause:
             sql += f" {order_clause}"

api/extensions/logstore/sql_escape.py

@@ -0,0 +1,134 @@
+"""
+SQL Escape Utility for LogStore Queries
+
+This module provides escaping utilities to prevent injection attacks in LogStore queries.
+
+LogStore supports two query modes:
+1. PG Protocol Mode: Uses SQL syntax with single quotes for strings
+2. SDK Mode: Uses LogStore query syntax (key: value) with double quotes
+
+Key Security Concerns:
+- Prevent tenant A from accessing tenant B's data via injection
+- SLS queries are read-only, so we focus on data access control
+- Different escaping strategies for SQL vs LogStore query syntax
+"""
+
+
+def escape_sql_string(value: str) -> str:
+    """
+    Escape a string value for safe use in SQL queries.
+
+    This function escapes single quotes by doubling them, which is the standard
+    SQL escaping method. This prevents SQL injection by ensuring that user input
+    cannot break out of string literals.
+
+    Args:
+        value: The string value to escape
+
+    Returns:
+        Escaped string safe for use in SQL queries
+
+    Examples:
+        >>> escape_sql_string("normal_value")
+        "normal_value"
+        >>> escape_sql_string("value' OR '1'='1")
+        "value'' OR ''1''=''1"
+        >>> escape_sql_string("tenant's_id")
+        "tenant''s_id"
+
+    Security:
+        - Prevents breaking out of string literals
+        - Stops injection attacks like: ' OR '1'='1
+        - Protects against cross-tenant data access
+    """
+    if not value:
+        return value
+
+    # Escape single quotes by doubling them (standard SQL escaping)
+    # This prevents breaking out of string literals in SQL queries
+    return value.replace("'", "''")
+
+
+def escape_identifier(value: str) -> str:
+    """
+    Escape an identifier (tenant_id, app_id, run_id, etc.) for safe SQL use.
+
+    This function is for PG protocol mode (SQL syntax).
+    For SDK mode, use escape_logstore_query_value() instead.
+
+    Args:
+        value: The identifier value to escape
+
+    Returns:
+        Escaped identifier safe for use in SQL queries
+
+    Examples:
+        >>> escape_identifier("550e8400-e29b-41d4-a716-446655440000")
+        "550e8400-e29b-41d4-a716-446655440000"
+        >>> escape_identifier("tenant_id' OR '1'='1")
+        "tenant_id'' OR ''1''=''1"
+
+    Security:
+        - Prevents SQL injection via identifiers
+        - Stops cross-tenant access attempts
+        - Works for UUIDs, alphanumeric IDs, and similar identifiers
+    """
+    # For identifiers, use the same escaping as strings
+    # This is simple and effective for preventing injection
+    return escape_sql_string(value)
+
+
+def escape_logstore_query_value(value: str) -> str:
+    """
+    Escape value for LogStore query syntax (SDK mode).
+
+    LogStore query syntax rules:
+    1. Keywords (and/or/not) are case-insensitive
+    2. Single quotes are ordinary characters (no special meaning)
+    3. Double quotes wrap values: key:"value"
+    4. Backslash is the escape character:
+       - \" for double quote inside value
+       - \\ for backslash itself
+    5. Parentheses can change query structure
+
+    To prevent injection:
+    - Wrap value in double quotes to treat special chars as literals
+    - Escape backslashes and double quotes using backslash
+
+    Args:
+        value: The value to escape for LogStore query syntax
+
+    Returns:
+        Quoted and escaped value safe for LogStore query syntax (includes the quotes)
+
+    Examples:
+        >>> escape_logstore_query_value("normal_value")
+        '"normal_value"'
+        >>> escape_logstore_query_value("value or field:evil")
+        '"value or field:evil"'  # 'or' and ':' are now literals
+        >>> escape_logstore_query_value('value"test')
+        '"value\\"test"'  # Internal double quote escaped
+        >>> escape_logstore_query_value('value\\test')
+        '"value\\\\test"'  # Backslash escaped
+
+    Security:
+        - Prevents injection via and/or/not keywords
+        - Prevents injection via colons (:)
+        - Prevents injection via parentheses
+        - Protects against cross-tenant data access
+
+    Note:
+        Escape order is critical: backslash first, then double quotes.
+        Otherwise, we'd double-escape the escape character itself.
+    """
+    if not value:
+        return '""'
+
+    # IMPORTANT: Escape backslashes FIRST, then double quotes
+    # This prevents double-escaping (e.g., " -> \" -> \\" incorrectly)
+    escaped = value.replace("\\", "\\\\")  # \ -> \\
+    escaped = escaped.replace('"', '\\"')  # " -> \"
+
+    # Wrap in double quotes to treat as literal string
+    # This prevents and/or/not/:/() from being interpreted as operators
+    return f'"{escaped}"'

api/factories/variable_factory.py

@@ -38,7 +38,7 @@ from core.variables.variables import (
     ObjectVariable,
     SecretVariable,
     StringVariable,
-    Variable,
+    VariableBase,
 )
 from core.workflow.constants import (
     CONVERSATION_VARIABLE_NODE_ID,
@@ -72,25 +72,25 @@ SEGMENT_TO_VARIABLE_MAP = {
 }
 
 
-def build_conversation_variable_from_mapping(mapping: Mapping[str, Any], /) -> Variable:
+def build_conversation_variable_from_mapping(mapping: Mapping[str, Any], /) -> VariableBase:
     if not mapping.get("name"):
         raise VariableError("missing name")
     return _build_variable_from_mapping(mapping=mapping, selector=[CONVERSATION_VARIABLE_NODE_ID, mapping["name"]])
 
 
-def build_environment_variable_from_mapping(mapping: Mapping[str, Any], /) -> Variable:
+def build_environment_variable_from_mapping(mapping: Mapping[str, Any], /) -> VariableBase:
     if not mapping.get("name"):
         raise VariableError("missing name")
     return _build_variable_from_mapping(mapping=mapping, selector=[ENVIRONMENT_VARIABLE_NODE_ID, mapping["name"]])
 
 
-def build_pipeline_variable_from_mapping(mapping: Mapping[str, Any], /) -> Variable:
+def build_pipeline_variable_from_mapping(mapping: Mapping[str, Any], /) -> VariableBase:
     if not mapping.get("variable"):
         raise VariableError("missing variable")
     return mapping["variable"]
 
 
-def _build_variable_from_mapping(*, mapping: Mapping[str, Any], selector: Sequence[str]) -> Variable:
+def _build_variable_from_mapping(*, mapping: Mapping[str, Any], selector: Sequence[str]) -> VariableBase:
     """
     This factory function is used to create the environment variable or the conversation variable,
     not support the File type.
@@ -100,7 +100,7 @@ def _build_variable_from_mapping(*, mapping: Mapping[str, Any], selector: Sequen
     if (value := mapping.get("value")) is None:
         raise VariableError("missing value")
 
-    result: Variable
+    result: VariableBase
     match value_type:
         case SegmentType.STRING:
             result = StringVariable.model_validate(mapping)
@@ -134,7 +134,7 @@ def _build_variable_from_mapping(*, mapping: Mapping[str, Any], selector: Sequen
         raise VariableError(f"variable size {result.size} exceeds limit {dify_config.MAX_VARIABLE_SIZE}")
     if not result.selector:
         result = result.model_copy(update={"selector": selector})
-    return cast(Variable, result)
+    return cast(VariableBase, result)
 
 
 def build_segment(value: Any, /) -> Segment:
@@ -285,8 +285,8 @@ def segment_to_variable(
     id: str | None = None,
     name: str | None = None,
     description: str = "",
-) -> Variable:
-    if isinstance(segment, Variable):
+) -> VariableBase:
+    if isinstance(segment, VariableBase):
         return segment
     name = name or selector[-1]
     id = id or str(uuid4())
@@ -297,7 +297,7 @@ def segment_to_variable(
 
     variable_class = SEGMENT_TO_VARIABLE_MAP[segment_type]
     return cast(
-        Variable,
+        VariableBase,
         variable_class(
             id=id,
             name=name,

api/fields/workflow_fields.py

@@ -1,7 +1,7 @@
 from flask_restx import fields
 
 from core.helper import encrypter
-from core.variables import SecretVariable, SegmentType, Variable
+from core.variables import SecretVariable, SegmentType, VariableBase
 from fields.member_fields import simple_account_fields
 from libs.helper import TimestampField
 
@@ -21,7 +21,7 @@ class EnvironmentVariableField(fields.Raw):
                 "value_type": value.value_type.value,
                 "description": value.description,
             }
-        if isinstance(value, Variable):
+        if isinstance(value, VariableBase):
             return {
                 "id": value.id,
                 "name": value.name,

api/migrations/versions/2026_01_09_1630-905527cc8fd3_add_workflow_run_created_at_id_idx.py

@@ -0,0 +1,30 @@
+"""add workflow_run_created_at_id_idx
+
+Revision ID: 905527cc8fd3
+Revises: 7df29de0f6be
+Create Date: 2025-01-09 16:30:02.462084
+
+"""
+from alembic import op
+import models as models
+
+# revision identifiers, used by Alembic.
+revision = '905527cc8fd3'
+down_revision = '7df29de0f6be'
+branch_labels = None
+depends_on = None
+
+
+def upgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    with op.batch_alter_table('workflow_runs', schema=None) as batch_op:
+        batch_op.create_index('workflow_run_created_at_id_idx', ['created_at', 'id'], unique=False)
+
+    # ### end Alembic commands ###
+
+
+def downgrade():
+    # ### commands auto generated by Alembic - please adjust! ###
+    with op.batch_alter_table('workflow_runs', schema=None) as batch_op:
+        batch_op.drop_index('workflow_run_created_at_id_idx')
+    # ### end Alembic commands ###

api/models/model.py

@@ -1843,7 +1843,7 @@ class MessageChain(TypeBase):
     )
 
 
-class MessageAgentThought(Base):
+class MessageAgentThought(TypeBase):
     __tablename__ = "message_agent_thoughts"
     __table_args__ = (
         sa.PrimaryKeyConstraint("id", name="message_agent_thought_pkey"),
@@ -1851,34 +1851,42 @@ class MessageAgentThought(Base):
         sa.Index("message_agent_thought_message_chain_id_idx", "message_chain_id"),
     )
 
-    id = mapped_column(StringUUID, default=lambda: str(uuid4()))
-    message_id = mapped_column(StringUUID, nullable=False)
-    message_chain_id = mapped_column(StringUUID, nullable=True)
+    id: Mapped[str] = mapped_column(
+        StringUUID, insert_default=lambda: str(uuid4()), default_factory=lambda: str(uuid4()), init=False
+    )
+    message_id: Mapped[str] = mapped_column(StringUUID, nullable=False)
     position: Mapped[int] = mapped_column(sa.Integer, nullable=False)
-    thought = mapped_column(LongText, nullable=True)
-    tool = mapped_column(LongText, nullable=True)
-    tool_labels_str = mapped_column(LongText, nullable=False, default=sa.text("'{}'"))
-    tool_meta_str = mapped_column(LongText, nullable=False, default=sa.text("'{}'"))
-    tool_input = mapped_column(LongText, nullable=True)
-    observation = mapped_column(LongText, nullable=True)
+    created_by_role: Mapped[str] = mapped_column(String(255), nullable=False)
+    created_by: Mapped[str] = mapped_column(StringUUID, nullable=False)
+    message_chain_id: Mapped[str | None] = mapped_column(StringUUID, nullable=True, default=None)
+    thought: Mapped[str | None] = mapped_column(LongText, nullable=True, default=None)
+    tool: Mapped[str | None] = mapped_column(LongText, nullable=True, default=None)
+    tool_labels_str: Mapped[str] = mapped_column(LongText, nullable=False, default=sa.text("'{}'"))
+    tool_meta_str: Mapped[str] = mapped_column(LongText, nullable=False, default=sa.text("'{}'"))
+    tool_input: Mapped[str | None] = mapped_column(LongText, nullable=True, default=None)
+    observation: Mapped[str | None] = mapped_column(LongText, nullable=True, default=None)
     # plugin_id = mapped_column(StringUUID, nullable=True)  ## for future design
-    tool_process_data = mapped_column(LongText, nullable=True)
-    message = mapped_column(LongText, nullable=True)
-    message_token: Mapped[int | None] = mapped_column(sa.Integer, nullable=True)
-    message_unit_price = mapped_column(sa.Numeric, nullable=True)
-    message_price_unit = mapped_column(sa.Numeric(10, 7), nullable=False, server_default=sa.text("0.001"))
-    message_files = mapped_column(LongText, nullable=True)
-    answer = mapped_column(LongText, nullable=True)
-    answer_token: Mapped[int | None] = mapped_column(sa.Integer, nullable=True)
-    answer_unit_price = mapped_column(sa.Numeric, nullable=True)
-    answer_price_unit = mapped_column(sa.Numeric(10, 7), nullable=False, server_default=sa.text("0.001"))
-    tokens: Mapped[int | None] = mapped_column(sa.Integer, nullable=True)
-    total_price = mapped_column(sa.Numeric, nullable=True)
-    currency = mapped_column(String(255), nullable=True)
-    latency: Mapped[float | None] = mapped_column(sa.Float, nullable=True)
-    created_by_role = mapped_column(String(255), nullable=False)
-    created_by = mapped_column(StringUUID, nullable=False)
-    created_at = mapped_column(sa.DateTime, nullable=False, server_default=sa.func.current_timestamp())
+    tool_process_data: Mapped[str | None] = mapped_column(LongText, nullable=True, default=None)
+    message: Mapped[str | None] = mapped_column(LongText, nullable=True, default=None)
+    message_token: Mapped[int | None] = mapped_column(sa.Integer, nullable=True, default=None)
+    message_unit_price: Mapped[Decimal | None] = mapped_column(sa.Numeric, nullable=True, default=None)
+    message_price_unit: Mapped[Decimal] = mapped_column(
+        sa.Numeric(10, 7), nullable=False, default=Decimal("0.001"), server_default=sa.text("0.001")
+    )
+    message_files: Mapped[str | None] = mapped_column(LongText, nullable=True, default=None)
+    answer: Mapped[str | None] = mapped_column(LongText, nullable=True, default=None)
+    answer_token: Mapped[int | None] = mapped_column(sa.Integer, nullable=True, default=None)
+    answer_unit_price: Mapped[Decimal | None] = mapped_column(sa.Numeric, nullable=True, default=None)
+    answer_price_unit: Mapped[Decimal] = mapped_column(
+        sa.Numeric(10, 7), nullable=False, default=Decimal("0.001"), server_default=sa.text("0.001")
+    )
+    tokens: Mapped[int | None] = mapped_column(sa.Integer, nullable=True, default=None)
+    total_price: Mapped[Decimal | None] = mapped_column(sa.Numeric, nullable=True, default=None)
+    currency: Mapped[str | None] = mapped_column(String(255), nullable=True, default=None)
+    latency: Mapped[float | None] = mapped_column(sa.Float, nullable=True, default=None)
+    created_at: Mapped[datetime] = mapped_column(
+        sa.DateTime, nullable=False, init=False, server_default=sa.func.current_timestamp()
+    )
 
     @property
     def files(self) -> list[Any]:

api/models/workflow.py

@@ -1,11 +1,9 @@
-from __future__ import annotations
-
 import json
 import logging
 from collections.abc import Generator, Mapping, Sequence
 from datetime import datetime
 from enum import StrEnum
-from typing import TYPE_CHECKING, Any, Union, cast
+from typing import TYPE_CHECKING, Any, Optional, Union, cast
 from uuid import uuid4
 
 import sqlalchemy as sa
@@ -46,7 +44,7 @@ if TYPE_CHECKING:
 
 from constants import DEFAULT_FILE_NUMBER_LIMITS, HIDDEN_VALUE
 from core.helper import encrypter
-from core.variables import SecretVariable, Segment, SegmentType, Variable
+from core.variables import SecretVariable, Segment, SegmentType, VariableBase
 from factories import variable_factory
 from libs import helper
 
@@ -69,7 +67,7 @@ class WorkflowType(StrEnum):
     RAG_PIPELINE = "rag-pipeline"
 
     @classmethod
-    def value_of(cls, value: str) -> WorkflowType:
+    def value_of(cls, value: str) -> "WorkflowType":
         """
         Get value of given mode.
 
@@ -82,7 +80,7 @@ class WorkflowType(StrEnum):
         raise ValueError(f"invalid workflow type value {value}")
 
     @classmethod
-    def from_app_mode(cls, app_mode: Union[str, AppMode]) -> WorkflowType:
+    def from_app_mode(cls, app_mode: Union[str, "AppMode"]) -> "WorkflowType":
         """
         Get workflow type from app mode.
 
@@ -178,12 +176,12 @@ class Workflow(Base):  # bug
         graph: str,
         features: str,
         created_by: str,
-        environment_variables: Sequence[Variable],
-        conversation_variables: Sequence[Variable],
+        environment_variables: Sequence[VariableBase],
+        conversation_variables: Sequence[VariableBase],
         rag_pipeline_variables: list[dict],
         marked_name: str = "",
         marked_comment: str = "",
-    ) -> Workflow:
+    ) -> "Workflow":
         workflow = Workflow()
         workflow.id = str(uuid4())
         workflow.tenant_id = tenant_id
@@ -447,7 +445,7 @@ class Workflow(Base):  # bug
 
         # decrypt secret variables value
         def decrypt_func(
-            var: Variable,
+            var: VariableBase,
         ) -> StringVariable | IntegerVariable | FloatVariable | SecretVariable:
             if isinstance(var, SecretVariable):
                 return var.model_copy(update={"value": encrypter.decrypt_token(tenant_id=tenant_id, token=var.value)})
@@ -463,7 +461,7 @@ class Workflow(Base):  # bug
         return decrypted_results
 
     @environment_variables.setter
-    def environment_variables(self, value: Sequence[Variable]):
+    def environment_variables(self, value: Sequence[VariableBase]):
         if not value:
             self._environment_variables = "{}"
             return
@@ -487,7 +485,7 @@ class Workflow(Base):  # bug
                 value[i] = origin_variables_dictionary[variable.id].model_copy(update={"name": variable.name})
 
         # encrypt secret variables value
-        def encrypt_func(var: Variable) -> Variable:
+        def encrypt_func(var: VariableBase) -> VariableBase:
             if isinstance(var, SecretVariable):
                 return var.model_copy(update={"value": encrypter.encrypt_token(tenant_id=tenant_id, token=var.value)})
             else:
@@ -517,7 +515,7 @@ class Workflow(Base):  # bug
         return result
 
     @property
-    def conversation_variables(self) -> Sequence[Variable]:
+    def conversation_variables(self) -> Sequence[VariableBase]:
         # TODO: find some way to init `self._conversation_variables` when instance created.
         if self._conversation_variables is None:
             self._conversation_variables = "{}"
@@ -527,7 +525,7 @@ class Workflow(Base):  # bug
         return results
 
     @conversation_variables.setter
-    def conversation_variables(self, value: Sequence[Variable]):
+    def conversation_variables(self, value: Sequence[VariableBase]):
         self._conversation_variables = json.dumps(
             {var.name: var.model_dump() for var in value},
             ensure_ascii=False,
@@ -597,6 +595,7 @@ class WorkflowRun(Base):
     __table_args__ = (
         sa.PrimaryKeyConstraint("id", name="workflow_run_pkey"),
         sa.Index("workflow_run_triggerd_from_idx", "tenant_id", "app_id", "triggered_from"),
+        sa.Index("workflow_run_created_at_id_idx", "created_at", "id"),
     )
 
     id: Mapped[str] = mapped_column(StringUUID, default=lambda: str(uuid4()))
@@ -621,7 +620,7 @@ class WorkflowRun(Base):
     finished_at: Mapped[datetime | None] = mapped_column(DateTime)
     exceptions_count: Mapped[int] = mapped_column(sa.Integer, server_default=sa.text("0"), nullable=True)
 
-    pause: Mapped[WorkflowPause | None] = orm.relationship(
+    pause: Mapped[Optional["WorkflowPause"]] = orm.relationship(
         "WorkflowPause",
         primaryjoin="WorkflowRun.id == foreign(WorkflowPause.workflow_run_id)",
         uselist=False,
@@ -691,7 +690,7 @@ class WorkflowRun(Base):
         }
 
     @classmethod
-    def from_dict(cls, data: dict[str, Any]) -> WorkflowRun:
+    def from_dict(cls, data: dict[str, Any]) -> "WorkflowRun":
         return cls(
             id=data.get("id"),
             tenant_id=data.get("tenant_id"),
@@ -843,7 +842,7 @@ class WorkflowNodeExecutionModel(Base):  # This model is expected to have `offlo
     created_by: Mapped[str] = mapped_column(StringUUID)
     finished_at: Mapped[datetime | None] = mapped_column(DateTime)
 
-    offload_data: Mapped[list[WorkflowNodeExecutionOffload]] = orm.relationship(
+    offload_data: Mapped[list["WorkflowNodeExecutionOffload"]] = orm.relationship(
         "WorkflowNodeExecutionOffload",
         primaryjoin="WorkflowNodeExecutionModel.id == foreign(WorkflowNodeExecutionOffload.node_execution_id)",
         uselist=True,
@@ -853,13 +852,13 @@ class WorkflowNodeExecutionModel(Base):  # This model is expected to have `offlo
 
     @staticmethod
     def preload_offload_data(
-        query: Select[tuple[WorkflowNodeExecutionModel]] | orm.Query[WorkflowNodeExecutionModel],
+        query: Select[tuple["WorkflowNodeExecutionModel"]] | orm.Query["WorkflowNodeExecutionModel"],
     ):
         return query.options(orm.selectinload(WorkflowNodeExecutionModel.offload_data))
 
     @staticmethod
     def preload_offload_data_and_files(
-        query: Select[tuple[WorkflowNodeExecutionModel]] | orm.Query[WorkflowNodeExecutionModel],
+        query: Select[tuple["WorkflowNodeExecutionModel"]] | orm.Query["WorkflowNodeExecutionModel"],
     ):
         return query.options(
             orm.selectinload(WorkflowNodeExecutionModel.offload_data).options(
@@ -934,7 +933,7 @@ class WorkflowNodeExecutionModel(Base):  # This model is expected to have `offlo
                     )
         return extras
 
-    def _get_offload_by_type(self, type_: ExecutionOffLoadType) -> WorkflowNodeExecutionOffload | None:
+    def _get_offload_by_type(self, type_: ExecutionOffLoadType) -> Optional["WorkflowNodeExecutionOffload"]:
         return next(iter([i for i in self.offload_data if i.type_ == type_]), None)
 
     @property
@@ -1048,7 +1047,7 @@ class WorkflowNodeExecutionOffload(Base):
         back_populates="offload_data",
     )
 
-    file: Mapped[UploadFile | None] = orm.relationship(
+    file: Mapped[Optional["UploadFile"]] = orm.relationship(
         foreign_keys=[file_id],
         lazy="raise",
         uselist=False,
@@ -1066,7 +1065,7 @@ class WorkflowAppLogCreatedFrom(StrEnum):
     INSTALLED_APP = "installed-app"
 
     @classmethod
-    def value_of(cls, value: str) -> WorkflowAppLogCreatedFrom:
+    def value_of(cls, value: str) -> "WorkflowAppLogCreatedFrom":
         """
         Get value of given mode.
 
@@ -1183,7 +1182,7 @@ class ConversationVariable(TypeBase):
     )
 
     @classmethod
-    def from_variable(cls, *, app_id: str, conversation_id: str, variable: Variable) -> ConversationVariable:
+    def from_variable(cls, *, app_id: str, conversation_id: str, variable: VariableBase) -> "ConversationVariable":
         obj = cls(
             id=variable.id,
             app_id=app_id,
@@ -1192,7 +1191,7 @@ class ConversationVariable(TypeBase):
         )
         return obj
 
-    def to_variable(self) -> Variable:
+    def to_variable(self) -> VariableBase:
         mapping = json.loads(self.data)
         return variable_factory.build_conversation_variable_from_mapping(mapping)
 
@@ -1336,7 +1335,7 @@ class WorkflowDraftVariable(Base):
     )
 
     # Relationship to WorkflowDraftVariableFile
-    variable_file: Mapped[WorkflowDraftVariableFile | None] = orm.relationship(
+    variable_file: Mapped[Optional["WorkflowDraftVariableFile"]] = orm.relationship(
         foreign_keys=[file_id],
         lazy="raise",
         uselist=False,
@@ -1506,7 +1505,7 @@ class WorkflowDraftVariable(Base):
         node_execution_id: str | None,
         description: str = "",
         file_id: str | None = None,
-    ) -> WorkflowDraftVariable:
+    ) -> "WorkflowDraftVariable":
         variable = WorkflowDraftVariable()
         variable.id = str(uuid4())
         variable.created_at = naive_utc_now()
@@ -1529,7 +1528,7 @@ class WorkflowDraftVariable(Base):
         name: str,
         value: Segment,
         description: str = "",
-    ) -> WorkflowDraftVariable:
+    ) -> "WorkflowDraftVariable":
         variable = cls._new(
             app_id=app_id,
             node_id=CONVERSATION_VARIABLE_NODE_ID,
@@ -1550,7 +1549,7 @@ class WorkflowDraftVariable(Base):
         value: Segment,
         node_execution_id: str,
         editable: bool = False,
-    ) -> WorkflowDraftVariable:
+    ) -> "WorkflowDraftVariable":
         variable = cls._new(
             app_id=app_id,
             node_id=SYSTEM_VARIABLE_NODE_ID,
@@ -1573,7 +1572,7 @@ class WorkflowDraftVariable(Base):
         visible: bool = True,
         editable: bool = True,
         file_id: str | None = None,
-    ) -> WorkflowDraftVariable:
+    ) -> "WorkflowDraftVariable":
         variable = cls._new(
             app_id=app_id,
             node_id=node_id,
@@ -1669,7 +1668,7 @@ class WorkflowDraftVariableFile(Base):
     )
 
     # Relationship to UploadFile
-    upload_file: Mapped[UploadFile] = orm.relationship(
+    upload_file: Mapped["UploadFile"] = orm.relationship(
         foreign_keys=[upload_file_id],
         lazy="raise",
         uselist=False,
@@ -1736,7 +1735,7 @@ class WorkflowPause(DefaultFieldsMixin, Base):
     state_object_key: Mapped[str] = mapped_column(String(length=255), nullable=False)
 
     # Relationship to WorkflowRun
-    workflow_run: Mapped[WorkflowRun] = orm.relationship(
+    workflow_run: Mapped["WorkflowRun"] = orm.relationship(
         foreign_keys=[workflow_run_id],
         # require explicit preloading.
         lazy="raise",
@@ -1792,7 +1791,7 @@ class WorkflowPauseReason(DefaultFieldsMixin, Base):
     )
 
     @classmethod
-    def from_entity(cls, pause_reason: PauseReason) -> WorkflowPauseReason:
+    def from_entity(cls, pause_reason: PauseReason) -> "WorkflowPauseReason":
         if isinstance(pause_reason, HumanInputRequired):
             return cls(
                 type_=PauseReasonType.HUMAN_INPUT_REQUIRED, form_id=pause_reason.form_id, node_id=pause_reason.node_id

api/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "dify-api"
-version = "1.11.2"
+version = "1.11.3"
 requires-python = ">=3.11,<3.13"
 
 dependencies = [
@@ -189,7 +189,7 @@ storage = [
     "opendal~=0.46.0",
     "oss2==2.18.5",
     "supabase~=2.18.1",
-    "tos~=2.7.1",
+    "tos~=2.9.0",
 ]
 
 ############################################################

api/repositories/api_workflow_run_repository.py

@@ -34,11 +34,14 @@ Example:
     ```
 """
 
-from collections.abc import Sequence
+from collections.abc import Callable, Sequence
 from datetime import datetime
 from typing import Protocol
 
+from sqlalchemy.orm import Session
+
 from core.workflow.entities.pause_reason import PauseReason
+from core.workflow.enums import WorkflowType
 from core.workflow.repositories.workflow_execution_repository import WorkflowExecutionRepository
 from libs.infinite_scroll_pagination import InfiniteScrollPagination
 from models.enums import WorkflowRunTriggeredFrom
@@ -253,6 +256,44 @@ class APIWorkflowRunRepository(WorkflowExecutionRepository, Protocol):
         """
         ...
 
+    def get_runs_batch_by_time_range(
+        self,
+        start_from: datetime | None,
+        end_before: datetime,
+        last_seen: tuple[datetime, str] | None,
+        batch_size: int,
+        run_types: Sequence[WorkflowType] | None = None,
+        tenant_ids: Sequence[str] | None = None,
+    ) -> Sequence[WorkflowRun]:
+        """
+        Fetch ended workflow runs in a time window for archival and clean batching.
+        """
+        ...
+
+    def delete_runs_with_related(
+        self,
+        runs: Sequence[WorkflowRun],
+        delete_node_executions: Callable[[Session, Sequence[WorkflowRun]], tuple[int, int]] | None = None,
+        delete_trigger_logs: Callable[[Session, Sequence[str]], int] | None = None,
+    ) -> dict[str, int]:
+        """
+        Delete workflow runs and their related records (node executions, offloads, app logs,
+        trigger logs, pauses, pause reasons).
+        """
+        ...
+
+    def count_runs_with_related(
+        self,
+        runs: Sequence[WorkflowRun],
+        count_node_executions: Callable[[Session, Sequence[WorkflowRun]], tuple[int, int]] | None = None,
+        count_trigger_logs: Callable[[Session, Sequence[str]], int] | None = None,
+    ) -> dict[str, int]:
+        """
+        Count workflow runs and their related records (node executions, offloads, app logs,
+        trigger logs, pauses, pause reasons) without deleting data.
+        """
+        ...
+
     def create_workflow_pause(
         self,
         workflow_run_id: str,

api/repositories/sqlalchemy_api_workflow_node_execution_repository.py

@@ -7,13 +7,18 @@ using SQLAlchemy 2.0 style queries for WorkflowNodeExecutionModel operations.
 
 from collections.abc import Sequence
 from datetime import datetime
-from typing import cast
+from typing import TypedDict, cast
 
-from sqlalchemy import asc, delete, desc, select
+from sqlalchemy import asc, delete, desc, func, select, tuple_
 from sqlalchemy.engine import CursorResult
 from sqlalchemy.orm import Session, sessionmaker
 
-from models.workflow import WorkflowNodeExecutionModel
+from models.enums import WorkflowRunTriggeredFrom
+from models.workflow import (
+    WorkflowNodeExecutionModel,
+    WorkflowNodeExecutionOffload,
+    WorkflowNodeExecutionTriggeredFrom,
+)
 from repositories.api_workflow_node_execution_repository import DifyAPIWorkflowNodeExecutionRepository
 
 
@@ -44,6 +49,26 @@ class DifyAPISQLAlchemyWorkflowNodeExecutionRepository(DifyAPIWorkflowNodeExecut
         """
         self._session_maker = session_maker
 
+    @staticmethod
+    def _map_run_triggered_from_to_node_triggered_from(triggered_from: str) -> str:
+        """
+        Map workflow run triggered_from values to workflow node execution triggered_from values.
+        """
+        if triggered_from in {
+            WorkflowRunTriggeredFrom.APP_RUN.value,
+            WorkflowRunTriggeredFrom.DEBUGGING.value,
+            WorkflowRunTriggeredFrom.SCHEDULE.value,
+            WorkflowRunTriggeredFrom.PLUGIN.value,
+            WorkflowRunTriggeredFrom.WEBHOOK.value,
+        }:
+            return WorkflowNodeExecutionTriggeredFrom.WORKFLOW_RUN.value
+        if triggered_from in {
+            WorkflowRunTriggeredFrom.RAG_PIPELINE_RUN.value,
+            WorkflowRunTriggeredFrom.RAG_PIPELINE_DEBUGGING.value,
+        }:
+            return WorkflowNodeExecutionTriggeredFrom.RAG_PIPELINE_RUN.value
+        return ""
+
     def get_node_last_execution(
         self,
         tenant_id: str,
@@ -290,3 +315,119 @@ class DifyAPISQLAlchemyWorkflowNodeExecutionRepository(DifyAPIWorkflowNodeExecut
             result = cast(CursorResult, session.execute(stmt))
             session.commit()
             return result.rowcount
+
+    class RunContext(TypedDict):
+        run_id: str
+        tenant_id: str
+        app_id: str
+        workflow_id: str
+        triggered_from: str
+
+    @staticmethod
+    def delete_by_runs(session: Session, runs: Sequence[RunContext]) -> tuple[int, int]:
+        """
+        Delete node executions (and offloads) for the given workflow runs using indexed columns.
+
+        Uses the composite index on (tenant_id, app_id, workflow_id, triggered_from, workflow_run_id)
+        by filtering on those columns with tuple IN.
+        """
+        if not runs:
+            return 0, 0
+
+        tuple_values = [
+            (
+                run["tenant_id"],
+                run["app_id"],
+                run["workflow_id"],
+                DifyAPISQLAlchemyWorkflowNodeExecutionRepository._map_run_triggered_from_to_node_triggered_from(
+                    run["triggered_from"]
+                ),
+                run["run_id"],
+            )
+            for run in runs
+        ]
+
+        node_execution_ids = session.scalars(
+            select(WorkflowNodeExecutionModel.id).where(
+                tuple_(
+                    WorkflowNodeExecutionModel.tenant_id,
+                    WorkflowNodeExecutionModel.app_id,
+                    WorkflowNodeExecutionModel.workflow_id,
+                    WorkflowNodeExecutionModel.triggered_from,
+                    WorkflowNodeExecutionModel.workflow_run_id,
+                ).in_(tuple_values)
+            )
+        ).all()
+
+        if not node_execution_ids:
+            return 0, 0
+
+        offloads_deleted = (
+            cast(
+                CursorResult,
+                session.execute(
+                    delete(WorkflowNodeExecutionOffload).where(
+                        WorkflowNodeExecutionOffload.node_execution_id.in_(node_execution_ids)
+                    )
+                ),
+            ).rowcount
+            or 0
+        )
+
+        node_executions_deleted = (
+            cast(
+                CursorResult,
+                session.execute(
+                    delete(WorkflowNodeExecutionModel).where(WorkflowNodeExecutionModel.id.in_(node_execution_ids))
+                ),
+            ).rowcount
+            or 0
+        )
+
+        return node_executions_deleted, offloads_deleted
+
+    @staticmethod
+    def count_by_runs(session: Session, runs: Sequence[RunContext]) -> tuple[int, int]:
+        """
+        Count node executions (and offloads) for the given workflow runs using indexed columns.
+        """
+        if not runs:
+            return 0, 0
+
+        tuple_values = [
+            (
+                run["tenant_id"],
+                run["app_id"],
+                run["workflow_id"],
+                DifyAPISQLAlchemyWorkflowNodeExecutionRepository._map_run_triggered_from_to_node_triggered_from(
+                    run["triggered_from"]
+                ),
+                run["run_id"],
+            )
+            for run in runs
+        ]
+        tuple_filter = tuple_(
+            WorkflowNodeExecutionModel.tenant_id,
+            WorkflowNodeExecutionModel.app_id,
+            WorkflowNodeExecutionModel.workflow_id,
+            WorkflowNodeExecutionModel.triggered_from,
+            WorkflowNodeExecutionModel.workflow_run_id,
+        ).in_(tuple_values)
+
+        node_executions_count = (
+            session.scalar(select(func.count()).select_from(WorkflowNodeExecutionModel).where(tuple_filter)) or 0
+        )
+        offloads_count = (
+            session.scalar(
+                select(func.count())
+                .select_from(WorkflowNodeExecutionOffload)
+                .join(
+                    WorkflowNodeExecutionModel,
+                    WorkflowNodeExecutionOffload.node_execution_id == WorkflowNodeExecutionModel.id,
+                )
+                .where(tuple_filter)
+            )
+            or 0
+        )
+
+        return int(node_executions_count), int(offloads_count)

api/repositories/sqlalchemy_api_workflow_run_repository.py

@@ -21,7 +21,7 @@ Implementation Notes:
 
 import logging
 import uuid
-from collections.abc import Sequence
+from collections.abc import Callable, Sequence
 from datetime import datetime
 from decimal import Decimal
 from typing import Any, cast
@@ -32,7 +32,7 @@ from sqlalchemy.engine import CursorResult
 from sqlalchemy.orm import Session, selectinload, sessionmaker
 
 from core.workflow.entities.pause_reason import HumanInputRequired, PauseReason, SchedulingPause
-from core.workflow.enums import WorkflowExecutionStatus
+from core.workflow.enums import WorkflowExecutionStatus, WorkflowType
 from extensions.ext_storage import storage
 from libs.datetime_utils import naive_utc_now
 from libs.helper import convert_datetime_to_date
@@ -40,8 +40,14 @@ from libs.infinite_scroll_pagination import InfiniteScrollPagination
 from libs.time_parser import get_time_threshold
 from libs.uuid_utils import uuidv7
 from models.enums import WorkflowRunTriggeredFrom
-from models.workflow import WorkflowPause as WorkflowPauseModel
-from models.workflow import WorkflowPauseReason, WorkflowRun
+from models.workflow import (
+    WorkflowAppLog,
+    WorkflowPauseReason,
+    WorkflowRun,
+)
+from models.workflow import (
+    WorkflowPause as WorkflowPauseModel,
+)
 from repositories.api_workflow_run_repository import APIWorkflowRunRepository
 from repositories.entities.workflow_pause import WorkflowPauseEntity
 from repositories.types import (
@@ -314,6 +320,171 @@ class DifyAPISQLAlchemyWorkflowRunRepository(APIWorkflowRunRepository):
         logger.info("Total deleted %s workflow runs for app %s", total_deleted, app_id)
         return total_deleted
 
+    def get_runs_batch_by_time_range(
+        self,
+        start_from: datetime | None,
+        end_before: datetime,
+        last_seen: tuple[datetime, str] | None,
+        batch_size: int,
+        run_types: Sequence[WorkflowType] | None = None,
+        tenant_ids: Sequence[str] | None = None,
+    ) -> Sequence[WorkflowRun]:
+        """
+        Fetch ended workflow runs in a time window for archival and clean batching.
+
+        Query scope:
+        - created_at in [start_from, end_before)
+        - type in run_types (when provided)
+        - status is an ended state
+        - optional tenant_id filter and cursor (last_seen) for pagination
+        """
+        with self._session_maker() as session:
+            stmt = (
+                select(WorkflowRun)
+                .where(
+                    WorkflowRun.created_at < end_before,
+                    WorkflowRun.status.in_(WorkflowExecutionStatus.ended_values()),
+                )
+                .order_by(WorkflowRun.created_at.asc(), WorkflowRun.id.asc())
+                .limit(batch_size)
+            )
+            if run_types is not None:
+                if not run_types:
+                    return []
+                stmt = stmt.where(WorkflowRun.type.in_(run_types))
+
+            if start_from:
+                stmt = stmt.where(WorkflowRun.created_at >= start_from)
+
+            if tenant_ids:
+                stmt = stmt.where(WorkflowRun.tenant_id.in_(tenant_ids))
+
+            if last_seen:
+                stmt = stmt.where(
+                    or_(
+                        WorkflowRun.created_at > last_seen[0],
+                        and_(WorkflowRun.created_at == last_seen[0], WorkflowRun.id > last_seen[1]),
+                    )
+                )
+
+            return session.scalars(stmt).all()
+
+    def delete_runs_with_related(
+        self,
+        runs: Sequence[WorkflowRun],
+        delete_node_executions: Callable[[Session, Sequence[WorkflowRun]], tuple[int, int]] | None = None,
+        delete_trigger_logs: Callable[[Session, Sequence[str]], int] | None = None,
+    ) -> dict[str, int]:
+        if not runs:
+            return {
+                "runs": 0,
+                "node_executions": 0,
+                "offloads": 0,
+                "app_logs": 0,
+                "trigger_logs": 0,
+                "pauses": 0,
+                "pause_reasons": 0,
+            }
+
+        with self._session_maker() as session:
+            run_ids = [run.id for run in runs]
+            if delete_node_executions:
+                node_executions_deleted, offloads_deleted = delete_node_executions(session, runs)
+            else:
+                node_executions_deleted, offloads_deleted = 0, 0
+
+            app_logs_result = session.execute(delete(WorkflowAppLog).where(WorkflowAppLog.workflow_run_id.in_(run_ids)))
+            app_logs_deleted = cast(CursorResult, app_logs_result).rowcount or 0
+
+            pause_ids = session.scalars(
+                select(WorkflowPauseModel.id).where(WorkflowPauseModel.workflow_run_id.in_(run_ids))
+            ).all()
+            pause_reasons_deleted = 0
+            pauses_deleted = 0
+
+            if pause_ids:
+                pause_reasons_result = session.execute(
+                    delete(WorkflowPauseReason).where(WorkflowPauseReason.pause_id.in_(pause_ids))
+                )
+                pause_reasons_deleted = cast(CursorResult, pause_reasons_result).rowcount or 0
+                pauses_result = session.execute(delete(WorkflowPauseModel).where(WorkflowPauseModel.id.in_(pause_ids)))
+                pauses_deleted = cast(CursorResult, pauses_result).rowcount or 0
+
+            trigger_logs_deleted = delete_trigger_logs(session, run_ids) if delete_trigger_logs else 0
+
+            runs_result = session.execute(delete(WorkflowRun).where(WorkflowRun.id.in_(run_ids)))
+            runs_deleted = cast(CursorResult, runs_result).rowcount or 0
+
+            session.commit()
+
+            return {
+                "runs": runs_deleted,
+                "node_executions": node_executions_deleted,
+                "offloads": offloads_deleted,
+                "app_logs": app_logs_deleted,
+                "trigger_logs": trigger_logs_deleted,
+                "pauses": pauses_deleted,
+                "pause_reasons": pause_reasons_deleted,
+            }
+
+    def count_runs_with_related(
+        self,
+        runs: Sequence[WorkflowRun],
+        count_node_executions: Callable[[Session, Sequence[WorkflowRun]], tuple[int, int]] | None = None,
+        count_trigger_logs: Callable[[Session, Sequence[str]], int] | None = None,
+    ) -> dict[str, int]:
+        if not runs:
+            return {
+                "runs": 0,
+                "node_executions": 0,
+                "offloads": 0,
+                "app_logs": 0,
+                "trigger_logs": 0,
+                "pauses": 0,
+                "pause_reasons": 0,
+            }
+
+        with self._session_maker() as session:
+            run_ids = [run.id for run in runs]
+            if count_node_executions:
+                node_executions_count, offloads_count = count_node_executions(session, runs)
+            else:
+                node_executions_count, offloads_count = 0, 0
+
+            app_logs_count = (
+                session.scalar(
+                    select(func.count()).select_from(WorkflowAppLog).where(WorkflowAppLog.workflow_run_id.in_(run_ids))
+                )
+                or 0
+            )
+
+            pause_ids = session.scalars(
+                select(WorkflowPauseModel.id).where(WorkflowPauseModel.workflow_run_id.in_(run_ids))
+            ).all()
+            pauses_count = len(pause_ids)
+            pause_reasons_count = 0
+            if pause_ids:
+                pause_reasons_count = (
+                    session.scalar(
+                        select(func.count())
+                        .select_from(WorkflowPauseReason)
+                        .where(WorkflowPauseReason.pause_id.in_(pause_ids))
+                    )
+                    or 0
+                )
+
+            trigger_logs_count = count_trigger_logs(session, run_ids) if count_trigger_logs else 0
+
+            return {
+                "runs": len(runs),
+                "node_executions": node_executions_count,
+                "offloads": offloads_count,
+                "app_logs": int(app_logs_count),
+                "trigger_logs": trigger_logs_count,
+                "pauses": pauses_count,
+                "pause_reasons": int(pause_reasons_count),
+            }
+
     def create_workflow_pause(
         self,
         workflow_run_id: str,

api/repositories/sqlalchemy_workflow_trigger_log_repository.py

@@ -4,8 +4,10 @@ SQLAlchemy implementation of WorkflowTriggerLogRepository.
 
 from collections.abc import Sequence
 from datetime import UTC, datetime, timedelta
+from typing import cast
 
-from sqlalchemy import and_, select
+from sqlalchemy import and_, delete, func, select
+from sqlalchemy.engine import CursorResult
 from sqlalchemy.orm import Session
 
 from models.enums import WorkflowTriggerStatus
@@ -84,3 +86,37 @@ class SQLAlchemyWorkflowTriggerLogRepository(WorkflowTriggerLogRepository):
         )
 
         return list(self.session.scalars(query).all())
+
+    def delete_by_run_ids(self, run_ids: Sequence[str]) -> int:
+        """
+        Delete trigger logs associated with the given workflow run ids.
+
+        Args:
+            run_ids: Collection of workflow run identifiers.
+
+        Returns:
+            Number of rows deleted.
+        """
+        if not run_ids:
+            return 0
+
+        result = self.session.execute(delete(WorkflowTriggerLog).where(WorkflowTriggerLog.workflow_run_id.in_(run_ids)))
+        return cast(CursorResult, result).rowcount or 0
+
+    def count_by_run_ids(self, run_ids: Sequence[str]) -> int:
+        """
+        Count trigger logs associated with the given workflow run ids.
+
+        Args:
+            run_ids: Collection of workflow run identifiers.
+
+        Returns:
+            Number of rows matched.
+        """
+        if not run_ids:
+            return 0
+
+        count = self.session.scalar(
+            select(func.count()).select_from(WorkflowTriggerLog).where(WorkflowTriggerLog.workflow_run_id.in_(run_ids))
+        )
+        return int(count or 0)

api/repositories/workflow_trigger_log_repository.py

@@ -109,3 +109,15 @@ class WorkflowTriggerLogRepository(Protocol):
             A sequence of recent WorkflowTriggerLog instances
         """
         ...
+
+    def delete_by_run_ids(self, run_ids: Sequence[str]) -> int:
+        """
+        Delete trigger logs for workflow run IDs.
+
+        Args:
+            run_ids: Workflow run IDs to delete
+
+        Returns:
+            Number of rows deleted
+        """
+        ...

api/schedule/clean_workflow_runs_task.py

@@ -0,0 +1,43 @@
+from datetime import UTC, datetime
+
+import click
+
+import app
+from configs import dify_config
+from services.retention.workflow_run.clear_free_plan_expired_workflow_run_logs import WorkflowRunCleanup
+
+
+@app.celery.task(queue="retention")
+def clean_workflow_runs_task() -> None:
+    """
+    Scheduled cleanup for workflow runs and related records (sandbox tenants only).
+    """
+    click.echo(
+        click.style(
+            (
+                "Scheduled workflow run cleanup starting: "
+                f"cutoff={dify_config.SANDBOX_EXPIRED_RECORDS_RETENTION_DAYS} days, "
+                f"batch={dify_config.SANDBOX_EXPIRED_RECORDS_CLEAN_BATCH_SIZE}"
+            ),
+            fg="green",
+        )
+    )
+
+    start_time = datetime.now(UTC)
+
+    WorkflowRunCleanup(
+        days=dify_config.SANDBOX_EXPIRED_RECORDS_RETENTION_DAYS,
+        batch_size=dify_config.SANDBOX_EXPIRED_RECORDS_CLEAN_BATCH_SIZE,
+        start_from=None,
+        end_before=None,
+    ).run()
+
+    end_time = datetime.now(UTC)
+    elapsed = end_time - start_time
+    click.echo(
+        click.style(
+            f"Scheduled workflow run cleanup finished. start={start_time.isoformat()} "
+            f"end={end_time.isoformat()} duration={elapsed}",
+            fg="green",
+        )
+    )

api/services/account_service.py

@@ -8,7 +8,7 @@ from hashlib import sha256
 from typing import Any, cast
 
 from pydantic import BaseModel
-from sqlalchemy import func
+from sqlalchemy import func, select
 from sqlalchemy.orm import Session
 from werkzeug.exceptions import Unauthorized
 
@@ -748,6 +748,21 @@ class AccountService:
         cls.email_code_login_rate_limiter.increment_rate_limit(email)
         return token
 
+    @staticmethod
+    def get_account_by_email_with_case_fallback(email: str, session: Session | None = None) -> Account | None:
+        """
+        Retrieve an account by email and fall back to the lowercase email if the original lookup fails.
+
+        This keeps backward compatibility for older records that stored uppercase emails while the
+        rest of the system gradually normalizes new inputs.
+        """
+        query_session = session or db.session
+        account = query_session.execute(select(Account).filter_by(email=email)).scalar_one_or_none()
+        if account or email == email.lower():
+            return account
+
+        return query_session.execute(select(Account).filter_by(email=email.lower())).scalar_one_or_none()
+
     @classmethod
     def get_email_code_login_data(cls, token: str) -> dict[str, Any] | None:
         return TokenManager.get_token_data(token, "email_code_login")
@@ -1363,16 +1378,22 @@ class RegisterService:
         if not inviter:
             raise ValueError("Inviter is required")
 
+        normalized_email = email.lower()
+
         """Invite new member"""
         with Session(db.engine) as session:
-            account = session.query(Account).filter_by(email=email).first()
+            account = AccountService.get_account_by_email_with_case_fallback(email, session=session)
 
         if not account:
             TenantService.check_member_permission(tenant, inviter, None, "add")
-            name = email.split("@")[0]
+            name = normalized_email.split("@")[0]
 
             account = cls.register(
-                email=email, name=name, language=language, status=AccountStatus.PENDING, is_setup=True
+                email=normalized_email,
+                name=name,
+                language=language,
+                status=AccountStatus.PENDING,
+                is_setup=True,
             )
             # Create new tenant member for invited tenant
             TenantService.create_tenant_member(tenant, account, role)
@@ -1394,7 +1415,7 @@ class RegisterService:
         # send email
         send_invite_member_mail_task.delay(
             language=language,
-            to=email,
+            to=account.email,
             token=token,
             inviter_name=inviter.name if inviter else "Dify",
             workspace_name=tenant.name,
@@ -1493,6 +1514,16 @@ class RegisterService:
             invitation: dict = json.loads(data)
             return invitation
 
+    @classmethod
+    def get_invitation_with_case_fallback(
+        cls, workspace_id: str | None, email: str | None, token: str
+    ) -> dict[str, Any] | None:
+        invitation = cls.get_invitation_if_token_valid(workspace_id, email, token)
+        if invitation or not email or email == email.lower():
+            return invitation
+        normalized_email = email.lower()
+        return cls.get_invitation_if_token_valid(workspace_id, normalized_email, token)
+
 
 def _generate_refresh_token(length: int = 64):
     token = secrets.token_hex(length)

api/services/conversation_variable_updater.py

@@ -1,7 +1,7 @@
 from sqlalchemy import select
 from sqlalchemy.orm import Session, sessionmaker
 
-from core.variables.variables import Variable
+from core.variables.variables import VariableBase
 from models import ConversationVariable
 
 
@@ -13,7 +13,7 @@ class ConversationVariableUpdater:
     def __init__(self, session_maker: sessionmaker[Session]) -> None:
         self._session_maker: sessionmaker[Session] = session_maker
 
-    def update(self, conversation_id: str, variable: Variable) -> None:
+    def update(self, conversation_id: str, variable: VariableBase) -> None:
         stmt = select(ConversationVariable).where(
             ConversationVariable.id == variable.id, ConversationVariable.conversation_id == conversation_id
         )

api/services/enterprise/base.py

@@ -1,9 +1,14 @@
+import logging
 import os
 from collections.abc import Mapping
 from typing import Any
 
 import httpx
 
+from core.helper.trace_id_helper import generate_traceparent_header
+
+logger = logging.getLogger(__name__)
+
 
 class BaseRequest:
     proxies: Mapping[str, str] | None = {
@@ -38,6 +43,15 @@ class BaseRequest:
         headers = {"Content-Type": "application/json", cls.secret_key_header: cls.secret_key}
         url = f"{cls.base_url}{endpoint}"
         mounts = cls._build_mounts()
+
+        try:
+            # ensure traceparent even when OTEL is disabled
+            traceparent = generate_traceparent_header()
+            if traceparent:
+                headers["traceparent"] = traceparent
+        except Exception:
+            logger.debug("Failed to generate traceparent header", exc_info=True)
+
         with httpx.Client(mounts=mounts) as client:
             response = client.request(method, url, json=json, params=params, headers=headers)
         return response.json()

api/services/plugin/plugin_service.py

@@ -3,6 +3,7 @@ from collections.abc import Mapping, Sequence
 from mimetypes import guess_type
 
 from pydantic import BaseModel
+from sqlalchemy import select
 from yarl import URL
 
 from configs import dify_config
@@ -25,7 +26,9 @@ from core.plugin.entities.plugin_daemon import (
 from core.plugin.impl.asset import PluginAssetManager
 from core.plugin.impl.debugging import PluginDebuggingClient
 from core.plugin.impl.plugin import PluginInstaller
+from extensions.ext_database import db
 from extensions.ext_redis import redis_client
+from models.provider import ProviderCredential
 from models.provider_ids import GenericProviderID
 from services.errors.plugin import PluginInstallationForbiddenError
 from services.feature_service import FeatureService, PluginInstallationScope
@@ -506,6 +509,33 @@ class PluginService:
     @staticmethod
     def uninstall(tenant_id: str, plugin_installation_id: str) -> bool:
         manager = PluginInstaller()
+
+        # Get plugin info before uninstalling to delete associated credentials
+        try:
+            plugins = manager.list_plugins(tenant_id)
+            plugin = next((p for p in plugins if p.installation_id == plugin_installation_id), None)
+
+            if plugin:
+                plugin_id = plugin.plugin_id
+                logger.info("Deleting credentials for plugin: %s", plugin_id)
+
+                # Delete provider credentials that match this plugin
+                credentials = db.session.scalars(
+                    select(ProviderCredential).where(
+                        ProviderCredential.tenant_id == tenant_id,
+                        ProviderCredential.provider_name.like(f"{plugin_id}/%"),
+                    )
+                ).all()
+
+                for cred in credentials:
+                    db.session.delete(cred)
+
+                db.session.commit()
+                logger.info("Deleted %d credentials for plugin: %s", len(credentials), plugin_id)
+        except Exception as e:
+            logger.warning("Failed to delete credentials: %s", e)
+            # Continue with uninstall even if credential deletion fails
+
         return manager.uninstall(tenant_id, plugin_installation_id)
 
     @staticmethod

api/services/rag_pipeline/rag_pipeline.py

@@ -36,7 +36,7 @@ from core.rag.entities.event import (
 )
 from core.repositories.factory import DifyCoreRepositoryFactory
 from core.repositories.sqlalchemy_workflow_node_execution_repository import SQLAlchemyWorkflowNodeExecutionRepository
-from core.variables.variables import Variable
+from core.variables.variables import VariableBase
 from core.workflow.entities.workflow_node_execution import (
     WorkflowNodeExecution,
     WorkflowNodeExecutionStatus,
@@ -270,8 +270,8 @@ class RagPipelineService:
         graph: dict,
         unique_hash: str | None,
         account: Account,
-        environment_variables: Sequence[Variable],
-        conversation_variables: Sequence[Variable],
+        environment_variables: Sequence[VariableBase],
+        conversation_variables: Sequence[VariableBase],
         rag_pipeline_variables: list,
     ) -> Workflow:
         """

api/services/retention/workflow_run/clear_free_plan_expired_workflow_run_logs.py

@@ -0,0 +1,301 @@
+import datetime
+import logging
+from collections.abc import Iterable, Sequence
+
+import click
+from sqlalchemy.orm import Session, sessionmaker
+
+from configs import dify_config
+from enums.cloud_plan import CloudPlan
+from extensions.ext_database import db
+from models.workflow import WorkflowRun
+from repositories.api_workflow_run_repository import APIWorkflowRunRepository
+from repositories.sqlalchemy_api_workflow_node_execution_repository import (
+    DifyAPISQLAlchemyWorkflowNodeExecutionRepository,
+)
+from repositories.sqlalchemy_workflow_trigger_log_repository import SQLAlchemyWorkflowTriggerLogRepository
+from services.billing_service import BillingService, SubscriptionPlan
+
+logger = logging.getLogger(__name__)
+
+
+class WorkflowRunCleanup:
+    def __init__(
+        self,
+        days: int,
+        batch_size: int,
+        start_from: datetime.datetime | None = None,
+        end_before: datetime.datetime | None = None,
+        workflow_run_repo: APIWorkflowRunRepository | None = None,
+        dry_run: bool = False,
+    ):
+        if (start_from is None) ^ (end_before is None):
+            raise ValueError("start_from and end_before must be both set or both omitted.")
+
+        computed_cutoff = datetime.datetime.now() - datetime.timedelta(days=days)
+        self.window_start = start_from
+        self.window_end = end_before or computed_cutoff
+
+        if self.window_start and self.window_end <= self.window_start:
+            raise ValueError("end_before must be greater than start_from.")
+
+        if batch_size <= 0:
+            raise ValueError("batch_size must be greater than 0.")
+
+        self.batch_size = batch_size
+        self._cleanup_whitelist: set[str] | None = None
+        self.dry_run = dry_run
+        self.free_plan_grace_period_days = dify_config.SANDBOX_EXPIRED_RECORDS_CLEAN_GRACEFUL_PERIOD
+        self.workflow_run_repo: APIWorkflowRunRepository
+        if workflow_run_repo:
+            self.workflow_run_repo = workflow_run_repo
+        else:
+            # Lazy import to avoid circular dependencies during module import
+            from repositories.factory import DifyAPIRepositoryFactory
+
+            session_maker = sessionmaker(bind=db.engine, expire_on_commit=False)
+            self.workflow_run_repo = DifyAPIRepositoryFactory.create_api_workflow_run_repository(session_maker)
+
+    def run(self) -> None:
+        click.echo(
+            click.style(
+                f"{'Inspecting' if self.dry_run else 'Cleaning'} workflow runs "
+                f"{'between ' + self.window_start.isoformat() + ' and ' if self.window_start else 'before '}"
+                f"{self.window_end.isoformat()} (batch={self.batch_size})",
+                fg="white",
+            )
+        )
+        if self.dry_run:
+            click.echo(click.style("Dry run mode enabled. No data will be deleted.", fg="yellow"))
+
+        total_runs_deleted = 0
+        total_runs_targeted = 0
+        related_totals = self._empty_related_counts() if self.dry_run else None
+        batch_index = 0
+        last_seen: tuple[datetime.datetime, str] | None = None
+
+        while True:
+            run_rows = self.workflow_run_repo.get_runs_batch_by_time_range(
+                start_from=self.window_start,
+                end_before=self.window_end,
+                last_seen=last_seen,
+                batch_size=self.batch_size,
+            )
+            if not run_rows:
+                break
+
+            batch_index += 1
+            last_seen = (run_rows[-1].created_at, run_rows[-1].id)
+            tenant_ids = {row.tenant_id for row in run_rows}
+            free_tenants = self._filter_free_tenants(tenant_ids)
+            free_runs = [row for row in run_rows if row.tenant_id in free_tenants]
+            paid_or_skipped = len(run_rows) - len(free_runs)
+
+            if not free_runs:
+                click.echo(
+                    click.style(
+                        f"[batch #{batch_index}] skipped (no sandbox runs in batch, {paid_or_skipped} paid/unknown)",
+                        fg="yellow",
+                    )
+                )
+                continue
+
+            total_runs_targeted += len(free_runs)
+
+            if self.dry_run:
+                batch_counts = self.workflow_run_repo.count_runs_with_related(
+                    free_runs,
+                    count_node_executions=self._count_node_executions,
+                    count_trigger_logs=self._count_trigger_logs,
+                )
+                if related_totals is not None:
+                    for key in related_totals:
+                        related_totals[key] += batch_counts.get(key, 0)
+                sample_ids = ", ".join(run.id for run in free_runs[:5])
+                click.echo(
+                    click.style(
+                        f"[batch #{batch_index}] would delete {len(free_runs)} runs "
+                        f"(sample ids: {sample_ids}) and skip {paid_or_skipped} paid/unknown",
+                        fg="yellow",
+                    )
+                )
+                continue
+
+            try:
+                counts = self.workflow_run_repo.delete_runs_with_related(
+                    free_runs,
+                    delete_node_executions=self._delete_node_executions,
+                    delete_trigger_logs=self._delete_trigger_logs,
+                )
+            except Exception:
+                logger.exception("Failed to delete workflow runs batch ending at %s", last_seen[0])
+                raise
+
+            total_runs_deleted += counts["runs"]
+            click.echo(
+                click.style(
+                    f"[batch #{batch_index}] deleted runs: {counts['runs']} "
+                    f"(nodes {counts['node_executions']}, offloads {counts['offloads']}, "
+                    f"app_logs {counts['app_logs']}, trigger_logs {counts['trigger_logs']}, "
+                    f"pauses {counts['pauses']}, pause_reasons {counts['pause_reasons']}); "
+                    f"skipped {paid_or_skipped} paid/unknown",
+                    fg="green",
+                )
+            )
+
+        if self.dry_run:
+            if self.window_start:
+                summary_message = (
+                    f"Dry run complete. Would delete {total_runs_targeted} workflow runs "
+                    f"between {self.window_start.isoformat()} and {self.window_end.isoformat()}"
+                )
+            else:
+                summary_message = (
+                    f"Dry run complete. Would delete {total_runs_targeted} workflow runs "
+                    f"before {self.window_end.isoformat()}"
+                )
+            if related_totals is not None:
+                summary_message = f"{summary_message}; related records: {self._format_related_counts(related_totals)}"
+            summary_color = "yellow"
+        else:
+            if self.window_start:
+                summary_message = (
+                    f"Cleanup complete. Deleted {total_runs_deleted} workflow runs "
+                    f"between {self.window_start.isoformat()} and {self.window_end.isoformat()}"
+                )
+            else:
+                summary_message = (
+                    f"Cleanup complete. Deleted {total_runs_deleted} workflow runs before {self.window_end.isoformat()}"
+                )
+            summary_color = "white"
+
+        click.echo(click.style(summary_message, fg=summary_color))
+
+    def _filter_free_tenants(self, tenant_ids: Iterable[str]) -> set[str]:
+        tenant_id_list = list(tenant_ids)
+
+        if not dify_config.BILLING_ENABLED:
+            return set(tenant_id_list)
+
+        if not tenant_id_list:
+            return set()
+
+        cleanup_whitelist = self._get_cleanup_whitelist()
+
+        try:
+            bulk_info = BillingService.get_plan_bulk_with_cache(tenant_id_list)
+        except Exception:
+            bulk_info = {}
+            logger.exception("Failed to fetch billing plans in bulk for tenants: %s", tenant_id_list)
+
+        eligible_free_tenants: set[str] = set()
+        for tenant_id in tenant_id_list:
+            if tenant_id in cleanup_whitelist:
+                continue
+
+            info = bulk_info.get(tenant_id)
+            if info is None:
+                logger.warning("Missing billing info for tenant %s in bulk resp; treating as non-free", tenant_id)
+                continue
+
+            if info.get("plan") != CloudPlan.SANDBOX:
+                continue
+
+            if self._is_within_grace_period(tenant_id, info):
+                continue
+
+            eligible_free_tenants.add(tenant_id)
+
+        return eligible_free_tenants
+
+    def _expiration_datetime(self, tenant_id: str, expiration_value: int) -> datetime.datetime | None:
+        if expiration_value < 0:
+            return None
+
+        try:
+            return datetime.datetime.fromtimestamp(expiration_value, datetime.UTC)
+        except (OverflowError, OSError, ValueError):
+            logger.exception("Failed to parse expiration timestamp for tenant %s", tenant_id)
+            return None
+
+    def _is_within_grace_period(self, tenant_id: str, info: SubscriptionPlan) -> bool:
+        if self.free_plan_grace_period_days <= 0:
+            return False
+
+        expiration_value = info.get("expiration_date", -1)
+        expiration_at = self._expiration_datetime(tenant_id, expiration_value)
+        if expiration_at is None:
+            return False
+
+        grace_deadline = expiration_at + datetime.timedelta(days=self.free_plan_grace_period_days)
+        return datetime.datetime.now(datetime.UTC) < grace_deadline
+
+    def _get_cleanup_whitelist(self) -> set[str]:
+        if self._cleanup_whitelist is not None:
+            return self._cleanup_whitelist
+
+        if not dify_config.BILLING_ENABLED:
+            self._cleanup_whitelist = set()
+            return self._cleanup_whitelist
+
+        try:
+            whitelist_ids = BillingService.get_expired_subscription_cleanup_whitelist()
+        except Exception:
+            logger.exception("Failed to fetch cleanup whitelist from billing service")
+            whitelist_ids = []
+
+        self._cleanup_whitelist = set(whitelist_ids)
+        return self._cleanup_whitelist
+
+    def _delete_trigger_logs(self, session: Session, run_ids: Sequence[str]) -> int:
+        trigger_repo = SQLAlchemyWorkflowTriggerLogRepository(session)
+        return trigger_repo.delete_by_run_ids(run_ids)
+
+    def _count_trigger_logs(self, session: Session, run_ids: Sequence[str]) -> int:
+        trigger_repo = SQLAlchemyWorkflowTriggerLogRepository(session)
+        return trigger_repo.count_by_run_ids(run_ids)
+
+    @staticmethod
+    def _build_run_contexts(
+        runs: Sequence[WorkflowRun],
+    ) -> list[DifyAPISQLAlchemyWorkflowNodeExecutionRepository.RunContext]:
+        return [
+            {
+                "run_id": run.id,
+                "tenant_id": run.tenant_id,
+                "app_id": run.app_id,
+                "workflow_id": run.workflow_id,
+                "triggered_from": run.triggered_from,
+            }
+            for run in runs
+        ]
+
+    @staticmethod
+    def _empty_related_counts() -> dict[str, int]:
+        return {
+            "node_executions": 0,
+            "offloads": 0,
+            "app_logs": 0,
+            "trigger_logs": 0,
+            "pauses": 0,
+            "pause_reasons": 0,
+        }
+
+    @staticmethod
+    def _format_related_counts(counts: dict[str, int]) -> str:
+        return (
+            f"node_executions {counts['node_executions']}, "
+            f"offloads {counts['offloads']}, "
+            f"app_logs {counts['app_logs']}, "
+            f"trigger_logs {counts['trigger_logs']}, "
+            f"pauses {counts['pauses']}, "
+            f"pause_reasons {counts['pause_reasons']}"
+        )
+
+    def _count_node_executions(self, session: Session, runs: Sequence[WorkflowRun]) -> tuple[int, int]:
+        run_contexts = self._build_run_contexts(runs)
+        return DifyAPISQLAlchemyWorkflowNodeExecutionRepository.count_by_runs(session, run_contexts)
+
+    def _delete_node_executions(self, session: Session, runs: Sequence[WorkflowRun]) -> tuple[int, int]:
+        run_contexts = self._build_run_contexts(runs)
+        return DifyAPISQLAlchemyWorkflowNodeExecutionRepository.delete_by_runs(session, run_contexts)

api/services/webapp_auth_service.py

@@ -12,6 +12,7 @@ from libs.passport import PassportService
 from libs.password import compare_password
 from models import Account, AccountStatus
 from models.model import App, EndUser, Site
+from services.account_service import AccountService
 from services.app_service import AppService
 from services.enterprise.enterprise_service import EnterpriseService
 from services.errors.account import AccountLoginError, AccountNotFoundError, AccountPasswordError
@@ -32,7 +33,7 @@ class WebAppAuthService:
     @staticmethod
     def authenticate(email: str, password: str) -> Account:
         """authenticate account with email and password"""
-        account = db.session.query(Account).filter_by(email=email).first()
+        account = AccountService.get_account_by_email_with_case_fallback(email)
         if not account:
             raise AccountNotFoundError()
 
@@ -52,7 +53,7 @@ class WebAppAuthService:
 
     @classmethod
     def get_user_through_email(cls, email: str):
-        account = db.session.query(Account).where(Account.email == email).first()
+        account = AccountService.get_account_by_email_with_case_fallback(email)
         if not account:
             return None
 

api/services/workflow_draft_variable_service.py

@@ -15,7 +15,7 @@ from sqlalchemy.sql.expression import and_, or_
 from configs import dify_config
 from core.app.entities.app_invoke_entities import InvokeFrom
 from core.file.models import File
-from core.variables import Segment, StringSegment, Variable
+from core.variables import Segment, StringSegment, VariableBase
 from core.variables.consts import SELECTORS_LENGTH
 from core.variables.segments import (
     ArrayFileSegment,
@@ -77,14 +77,14 @@ class DraftVarLoader(VariableLoader):
     # Application ID for which variables are being loaded.
     _app_id: str
     _tenant_id: str
-    _fallback_variables: Sequence[Variable]
+    _fallback_variables: Sequence[VariableBase]
 
     def __init__(
         self,
         engine: Engine,
         app_id: str,
         tenant_id: str,
-        fallback_variables: Sequence[Variable] | None = None,
+        fallback_variables: Sequence[VariableBase] | None = None,
     ):
         self._engine = engine
         self._app_id = app_id
@@ -94,12 +94,12 @@ class DraftVarLoader(VariableLoader):
     def _selector_to_tuple(self, selector: Sequence[str]) -> tuple[str, str]:
         return (selector[0], selector[1])
 
-    def load_variables(self, selectors: list[list[str]]) -> list[Variable]:
+    def load_variables(self, selectors: list[list[str]]) -> list[VariableBase]:
         if not selectors:
             return []
 
-        # Map each selector (as a tuple via `_selector_to_tuple`) to its corresponding Variable instance.
-        variable_by_selector: dict[tuple[str, str], Variable] = {}
+        # Map each selector (as a tuple via `_selector_to_tuple`) to its corresponding variable instance.
+        variable_by_selector: dict[tuple[str, str], VariableBase] = {}
 
         with Session(bind=self._engine, expire_on_commit=False) as session:
             srv = WorkflowDraftVariableService(session)
@@ -145,7 +145,7 @@ class DraftVarLoader(VariableLoader):
 
         return list(variable_by_selector.values())
 
-    def _load_offloaded_variable(self, draft_var: WorkflowDraftVariable) -> tuple[tuple[str, str], Variable]:
+    def _load_offloaded_variable(self, draft_var: WorkflowDraftVariable) -> tuple[tuple[str, str], VariableBase]:
         # This logic is closely tied to `WorkflowDraftVaribleService._try_offload_large_variable`
         # and must remain synchronized with it.
         # Ideally, these should be co-located for better maintainability.

api/services/workflow_service.py

@@ -13,8 +13,8 @@ from core.app.apps.advanced_chat.app_config_manager import AdvancedChatAppConfig
 from core.app.apps.workflow.app_config_manager import WorkflowAppConfigManager
 from core.file import File
 from core.repositories import DifyCoreRepositoryFactory
-from core.variables import Variable
-from core.variables.variables import VariableUnion
+from core.variables import VariableBase
+from core.variables.variables import Variable
 from core.workflow.entities import WorkflowNodeExecution
 from core.workflow.enums import ErrorStrategy, WorkflowNodeExecutionMetadataKey, WorkflowNodeExecutionStatus
 from core.workflow.errors import WorkflowNodeRunFailedError
@@ -198,8 +198,8 @@ class WorkflowService:
         features: dict,
         unique_hash: str | None,
         account: Account,
-        environment_variables: Sequence[Variable],
-        conversation_variables: Sequence[Variable],
+        environment_variables: Sequence[VariableBase],
+        conversation_variables: Sequence[VariableBase],
     ) -> Workflow:
         """
         Sync draft workflow
@@ -1044,7 +1044,7 @@ def _setup_variable_pool(
     workflow: Workflow,
     node_type: NodeType,
     conversation_id: str,
-    conversation_variables: list[Variable],
+    conversation_variables: list[VariableBase],
 ):
     # Only inject system variables for START node type.
     if node_type == NodeType.START or node_type.is_trigger_node:
@@ -1070,9 +1070,9 @@ def _setup_variable_pool(
         system_variables=system_variable,
         user_inputs=user_inputs,
         environment_variables=workflow.environment_variables,
-        # Based on the definition of `VariableUnion`,
-        # `list[Variable]` can be safely used as `list[VariableUnion]` since they are compatible.
-        conversation_variables=cast(list[VariableUnion], conversation_variables),  #
+        # Based on the definition of `Variable`,
+        # `VariableBase` instances can be safely used as `Variable` since they are compatible.
+        conversation_variables=cast(list[Variable], conversation_variables),  #
     )
 
     return variable_pool

api/tests/fixtures/workflow/test_streaming_conversation_variables_v1_overwrite.yml

@@ -0,0 +1,158 @@
+app:
+  description: Validate v1 Variable Assigner blocks streaming until conversation variable is updated.
+  icon: 🤖
+  icon_background: '#FFEAD5'
+  mode: advanced-chat
+  name: test_streaming_conversation_variables_v1_overwrite
+  use_icon_as_answer_icon: false
+dependencies: []
+kind: app
+version: 0.5.0
+workflow:
+  conversation_variables:
+  - description: ''
+    id: 6ddf2d7f-3d1b-4bb0-9a5e-9b0c87c7b5e6
+    name: conv_var
+    selector:
+    - conversation
+    - conv_var
+    value: default
+    value_type: string
+  environment_variables: []
+  features:
+    file_upload:
+      allowed_file_extensions:
+      - .JPG
+      - .JPEG
+      - .PNG
+      - .GIF
+      - .WEBP
+      - .SVG
+      allowed_file_types:
+      - image
+      allowed_file_upload_methods:
+      - local_file
+      - remote_url
+      enabled: false
+      fileUploadConfig:
+        audio_file_size_limit: 50
+        batch_count_limit: 5
+        file_size_limit: 15
+        image_file_size_limit: 10
+        video_file_size_limit: 100
+        workflow_file_upload_limit: 10
+      image:
+        enabled: false
+        number_limits: 3
+        transfer_methods:
+        - local_file
+        - remote_url
+      number_limits: 3
+    opening_statement: ''
+    retriever_resource:
+      enabled: true
+    sensitive_word_avoidance:
+      enabled: false
+    speech_to_text:
+      enabled: false
+    suggested_questions: []
+    suggested_questions_after_answer:
+      enabled: false
+    text_to_speech:
+      enabled: false
+      language: ''
+      voice: ''
+  graph:
+    edges:
+    - data:
+        isInIteration: false
+        isInLoop: false
+        sourceType: start
+        targetType: assigner
+      id: start-source-assigner-target
+      source: start
+      sourceHandle: source
+      target: assigner
+      targetHandle: target
+      type: custom
+      zIndex: 0
+    - data:
+        isInLoop: false
+        sourceType: assigner
+        targetType: answer
+      id: assigner-source-answer-target
+      source: assigner
+      sourceHandle: source
+      target: answer
+      targetHandle: target
+      type: custom
+      zIndex: 0
+    nodes:
+    - data:
+        desc: ''
+        selected: false
+        title: Start
+        type: start
+        variables: []
+      height: 54
+      id: start
+      position:
+        x: 30
+        y: 253
+      positionAbsolute:
+        x: 30
+        y: 253
+      selected: false
+      sourcePosition: right
+      targetPosition: left
+      type: custom
+      width: 244
+    - data:
+        answer: 'Current Value Of `conv_var` is:{{#conversation.conv_var#}}'
+        desc: ''
+        selected: false
+        title: Answer
+        type: answer
+        variables: []
+      height: 106
+      id: answer
+      position:
+        x: 638
+        y: 253
+      positionAbsolute:
+        x: 638
+        y: 253
+      selected: true
+      sourcePosition: right
+      targetPosition: left
+      type: custom
+      width: 244
+    - data:
+        assigned_variable_selector:
+        - conversation
+        - conv_var
+        desc: ''
+        input_variable_selector:
+        - sys
+        - query
+        selected: false
+        title: Variable Assigner
+        type: assigner
+        write_mode: over-write
+      height: 84
+      id: assigner
+      position:
+        x: 334
+        y: 253
+      positionAbsolute:
+        x: 334
+        y: 253
+      selected: false
+      sourcePosition: right
+      targetPosition: left
+      type: custom
+      width: 244
+    viewport:
+      x: 0
+      y: 0
+      zoom: 0.7

api/tests/test_containers_integration_tests/services/test_agent_service.py

@@ -230,7 +230,6 @@ class TestAgentService:
 
         # Create first agent thought
         thought1 = MessageAgentThought(
-            id=fake.uuid4(),
             message_id=message.id,
             position=1,
             thought="I need to analyze the user's request",
@@ -257,7 +256,6 @@ class TestAgentService:
 
         # Create second agent thought
         thought2 = MessageAgentThought(
-            id=fake.uuid4(),
             message_id=message.id,
             position=2,
             thought="Based on the analysis, I can provide a response",
@@ -545,7 +543,6 @@ class TestAgentService:
 
         # Create agent thought with tool error
         thought_with_error = MessageAgentThought(
-            id=fake.uuid4(),
             message_id=message.id,
             position=1,
             thought="I need to analyze the user's request",
@@ -759,7 +756,6 @@ class TestAgentService:
 
         # Create agent thought with multiple tools
         complex_thought = MessageAgentThought(
-            id=fake.uuid4(),
             message_id=message.id,
             position=1,
             thought="I need to use multiple tools to complete this task",
@@ -877,7 +873,6 @@ class TestAgentService:
 
         # Create agent thought with files
         thought_with_files = MessageAgentThought(
-            id=fake.uuid4(),
             message_id=message.id,
             position=1,
             thought="I need to process some files",
@@ -957,7 +952,6 @@ class TestAgentService:
 
         # Create agent thought with empty tool data
         empty_thought = MessageAgentThought(
-            id=fake.uuid4(),
             message_id=message.id,
             position=1,
             thought="I need to analyze the user's request",
@@ -999,7 +993,6 @@ class TestAgentService:
 
         # Create agent thought with malformed JSON
         malformed_thought = MessageAgentThought(
-            id=fake.uuid4(),
             message_id=message.id,
             position=1,
             thought="I need to analyze the user's request",

api/tests/unit_tests/controllers/console/auth/test_account_activation.py

@@ -40,7 +40,7 @@ class TestActivateCheckApi:
             "tenant": tenant,
         }
 
-    @patch("controllers.console.auth.activate.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.activate.RegisterService.get_invitation_with_case_fallback")
     def test_check_valid_invitation_token(self, mock_get_invitation, app, mock_invitation):
         """
         Test checking valid invitation token.
@@ -66,7 +66,7 @@ class TestActivateCheckApi:
         assert response["data"]["workspace_id"] == "workspace-123"
         assert response["data"]["email"] == "invitee@example.com"
 
-    @patch("controllers.console.auth.activate.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.activate.RegisterService.get_invitation_with_case_fallback")
     def test_check_invalid_invitation_token(self, mock_get_invitation, app):
         """
         Test checking invalid invitation token.
@@ -88,7 +88,7 @@ class TestActivateCheckApi:
         # Assert
         assert response["is_valid"] is False
 
-    @patch("controllers.console.auth.activate.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.activate.RegisterService.get_invitation_with_case_fallback")
     def test_check_token_without_workspace_id(self, mock_get_invitation, app, mock_invitation):
         """
         Test checking token without workspace ID.
@@ -109,7 +109,7 @@ class TestActivateCheckApi:
         assert response["is_valid"] is True
         mock_get_invitation.assert_called_once_with(None, "invitee@example.com", "valid_token")
 
-    @patch("controllers.console.auth.activate.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.activate.RegisterService.get_invitation_with_case_fallback")
     def test_check_token_without_email(self, mock_get_invitation, app, mock_invitation):
         """
         Test checking token without email parameter.
@@ -130,6 +130,20 @@ class TestActivateCheckApi:
         assert response["is_valid"] is True
         mock_get_invitation.assert_called_once_with("workspace-123", None, "valid_token")
 
+    @patch("controllers.console.auth.activate.RegisterService.get_invitation_with_case_fallback")
+    def test_check_token_normalizes_email_to_lowercase(self, mock_get_invitation, app, mock_invitation):
+        """Ensure token validation uses lowercase emails."""
+        mock_get_invitation.return_value = mock_invitation
+
+        with app.test_request_context(
+            "/activate/check?workspace_id=workspace-123&email=Invitee@Example.com&token=valid_token"
+        ):
+            api = ActivateCheckApi()
+            response = api.get()
+
+        assert response["is_valid"] is True
+        mock_get_invitation.assert_called_once_with("workspace-123", "Invitee@Example.com", "valid_token")
+
 
 class TestActivateApi:
     """Test cases for account activation endpoint."""
@@ -212,7 +226,7 @@ class TestActivateApi:
         mock_revoke_token.assert_called_once_with("workspace-123", "invitee@example.com", "valid_token")
         mock_db.session.commit.assert_called_once()
 
-    @patch("controllers.console.auth.activate.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.activate.RegisterService.get_invitation_with_case_fallback")
     def test_activation_with_invalid_token(self, mock_get_invitation, app):
         """
         Test account activation with invalid token.
@@ -241,7 +255,7 @@ class TestActivateApi:
             with pytest.raises(AlreadyActivateError):
                 api.post()
 
-    @patch("controllers.console.auth.activate.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.activate.RegisterService.get_invitation_with_case_fallback")
     @patch("controllers.console.auth.activate.RegisterService.revoke_token")
     @patch("controllers.console.auth.activate.db")
     def test_activation_sets_interface_theme(
@@ -290,7 +304,7 @@ class TestActivateApi:
             ("es-ES", "Europe/Madrid"),
         ],
     )
-    @patch("controllers.console.auth.activate.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.activate.RegisterService.get_invitation_with_case_fallback")
     @patch("controllers.console.auth.activate.RegisterService.revoke_token")
     @patch("controllers.console.auth.activate.db")
     def test_activation_with_different_locales(
@@ -336,7 +350,7 @@ class TestActivateApi:
         assert mock_account.interface_language == language
         assert mock_account.timezone == timezone
 
-    @patch("controllers.console.auth.activate.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.activate.RegisterService.get_invitation_with_case_fallback")
     @patch("controllers.console.auth.activate.RegisterService.revoke_token")
     @patch("controllers.console.auth.activate.db")
     def test_activation_returns_success_response(
@@ -376,7 +390,7 @@ class TestActivateApi:
         # Assert
         assert response == {"result": "success"}
 
-    @patch("controllers.console.auth.activate.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.activate.RegisterService.get_invitation_with_case_fallback")
     @patch("controllers.console.auth.activate.RegisterService.revoke_token")
     @patch("controllers.console.auth.activate.db")
     def test_activation_without_workspace_id(
@@ -415,3 +429,37 @@ class TestActivateApi:
         # Assert
         assert response["result"] == "success"
         mock_revoke_token.assert_called_once_with(None, "invitee@example.com", "valid_token")
+
+    @patch("controllers.console.auth.activate.RegisterService.get_invitation_with_case_fallback")
+    @patch("controllers.console.auth.activate.RegisterService.revoke_token")
+    @patch("controllers.console.auth.activate.db")
+    def test_activation_normalizes_email_before_lookup(
+        self,
+        mock_db,
+        mock_revoke_token,
+        mock_get_invitation,
+        app,
+        mock_invitation,
+        mock_account,
+    ):
+        """Ensure uppercase emails are normalized before lookup and revocation."""
+        mock_get_invitation.return_value = mock_invitation
+
+        with app.test_request_context(
+            "/activate",
+            method="POST",
+            json={
+                "workspace_id": "workspace-123",
+                "email": "Invitee@Example.com",
+                "token": "valid_token",
+                "name": "John Doe",
+                "interface_language": "en-US",
+                "timezone": "UTC",
+            },
+        ):
+            api = ActivateApi()
+            response = api.post()
+
+        assert response["result"] == "success"
+        mock_get_invitation.assert_called_once_with("workspace-123", "Invitee@Example.com", "valid_token")
+        mock_revoke_token.assert_called_once_with("workspace-123", "invitee@example.com", "valid_token")

api/tests/unit_tests/controllers/console/auth/test_authentication_security.py

@@ -34,7 +34,7 @@ class TestAuthenticationSecurity:
     @patch("controllers.console.auth.login.AccountService.authenticate")
     @patch("controllers.console.auth.login.AccountService.add_login_error_rate_limit")
     @patch("controllers.console.auth.login.dify_config.BILLING_ENABLED", False)
-    @patch("controllers.console.auth.login.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.login.RegisterService.get_invitation_with_case_fallback")
     def test_login_invalid_email_with_registration_allowed(
         self, mock_get_invitation, mock_add_rate_limit, mock_authenticate, mock_is_rate_limit, mock_features, mock_db
     ):
@@ -67,7 +67,7 @@ class TestAuthenticationSecurity:
     @patch("controllers.console.auth.login.AccountService.authenticate")
     @patch("controllers.console.auth.login.AccountService.add_login_error_rate_limit")
     @patch("controllers.console.auth.login.dify_config.BILLING_ENABLED", False)
-    @patch("controllers.console.auth.login.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.login.RegisterService.get_invitation_with_case_fallback")
     def test_login_wrong_password_returns_error(
         self, mock_get_invitation, mock_add_rate_limit, mock_authenticate, mock_is_rate_limit, mock_db
     ):
@@ -100,7 +100,7 @@ class TestAuthenticationSecurity:
     @patch("controllers.console.auth.login.AccountService.authenticate")
     @patch("controllers.console.auth.login.AccountService.add_login_error_rate_limit")
     @patch("controllers.console.auth.login.dify_config.BILLING_ENABLED", False)
-    @patch("controllers.console.auth.login.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.login.RegisterService.get_invitation_with_case_fallback")
     def test_login_invalid_email_with_registration_disabled(
         self, mock_get_invitation, mock_add_rate_limit, mock_authenticate, mock_is_rate_limit, mock_features, mock_db
     ):

api/tests/unit_tests/controllers/console/auth/test_email_register.py

@@ -0,0 +1,177 @@
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+import pytest
+from flask import Flask
+
+from controllers.console.auth.email_register import (
+    EmailRegisterCheckApi,
+    EmailRegisterResetApi,
+    EmailRegisterSendEmailApi,
+)
+from services.account_service import AccountService
+
+
+@pytest.fixture
+def app():
+    flask_app = Flask(__name__)
+    flask_app.config["TESTING"] = True
+    return flask_app
+
+
+class TestEmailRegisterSendEmailApi:
+    @patch("controllers.console.auth.email_register.Session")
+    @patch("controllers.console.auth.email_register.AccountService.get_account_by_email_with_case_fallback")
+    @patch("controllers.console.auth.email_register.AccountService.send_email_register_email")
+    @patch("controllers.console.auth.email_register.BillingService.is_email_in_freeze")
+    @patch("controllers.console.auth.email_register.AccountService.is_email_send_ip_limit", return_value=False)
+    @patch("controllers.console.auth.email_register.extract_remote_ip", return_value="127.0.0.1")
+    def test_send_email_normalizes_and_falls_back(
+        self,
+        mock_extract_ip,
+        mock_is_email_send_ip_limit,
+        mock_is_freeze,
+        mock_send_mail,
+        mock_get_account,
+        mock_session_cls,
+        app,
+    ):
+        mock_send_mail.return_value = "token-123"
+        mock_is_freeze.return_value = False
+        mock_account = MagicMock()
+
+        mock_session = MagicMock()
+        mock_session_cls.return_value.__enter__.return_value = mock_session
+        mock_get_account.return_value = mock_account
+
+        feature_flags = SimpleNamespace(enable_email_password_login=True, is_allow_register=True)
+        with (
+            patch("controllers.console.auth.email_register.db", SimpleNamespace(engine="engine")),
+            patch("controllers.console.auth.email_register.dify_config", SimpleNamespace(BILLING_ENABLED=True)),
+            patch("controllers.console.wraps.dify_config", SimpleNamespace(EDITION="CLOUD")),
+            patch("controllers.console.wraps.FeatureService.get_system_features", return_value=feature_flags),
+        ):
+            with app.test_request_context(
+                "/email-register/send-email",
+                method="POST",
+                json={"email": "Invitee@Example.com", "language": "en-US"},
+            ):
+                response = EmailRegisterSendEmailApi().post()
+
+        assert response == {"result": "success", "data": "token-123"}
+        mock_is_freeze.assert_called_once_with("invitee@example.com")
+        mock_send_mail.assert_called_once_with(email="invitee@example.com", account=mock_account, language="en-US")
+        mock_get_account.assert_called_once_with("Invitee@Example.com", session=mock_session)
+        mock_extract_ip.assert_called_once()
+        mock_is_email_send_ip_limit.assert_called_once_with("127.0.0.1")
+
+
+class TestEmailRegisterCheckApi:
+    @patch("controllers.console.auth.email_register.AccountService.reset_email_register_error_rate_limit")
+    @patch("controllers.console.auth.email_register.AccountService.generate_email_register_token")
+    @patch("controllers.console.auth.email_register.AccountService.revoke_email_register_token")
+    @patch("controllers.console.auth.email_register.AccountService.add_email_register_error_rate_limit")
+    @patch("controllers.console.auth.email_register.AccountService.get_email_register_data")
+    @patch("controllers.console.auth.email_register.AccountService.is_email_register_error_rate_limit")
+    def test_validity_normalizes_email_before_checks(
+        self,
+        mock_rate_limit_check,
+        mock_get_data,
+        mock_add_rate,
+        mock_revoke,
+        mock_generate_token,
+        mock_reset_rate,
+        app,
+    ):
+        mock_rate_limit_check.return_value = False
+        mock_get_data.return_value = {"email": "User@Example.com", "code": "4321"}
+        mock_generate_token.return_value = (None, "new-token")
+
+        feature_flags = SimpleNamespace(enable_email_password_login=True, is_allow_register=True)
+        with (
+            patch("controllers.console.auth.email_register.db", SimpleNamespace(engine="engine")),
+            patch("controllers.console.wraps.dify_config", SimpleNamespace(EDITION="CLOUD")),
+            patch("controllers.console.wraps.FeatureService.get_system_features", return_value=feature_flags),
+        ):
+            with app.test_request_context(
+                "/email-register/validity",
+                method="POST",
+                json={"email": "User@Example.com", "code": "4321", "token": "token-123"},
+            ):
+                response = EmailRegisterCheckApi().post()
+
+        assert response == {"is_valid": True, "email": "user@example.com", "token": "new-token"}
+        mock_rate_limit_check.assert_called_once_with("user@example.com")
+        mock_generate_token.assert_called_once_with(
+            "user@example.com", code="4321", additional_data={"phase": "register"}
+        )
+        mock_reset_rate.assert_called_once_with("user@example.com")
+        mock_add_rate.assert_not_called()
+        mock_revoke.assert_called_once_with("token-123")
+
+
+class TestEmailRegisterResetApi:
+    @patch("controllers.console.auth.email_register.AccountService.reset_login_error_rate_limit")
+    @patch("controllers.console.auth.email_register.AccountService.login")
+    @patch("controllers.console.auth.email_register.EmailRegisterResetApi._create_new_account")
+    @patch("controllers.console.auth.email_register.Session")
+    @patch("controllers.console.auth.email_register.AccountService.get_account_by_email_with_case_fallback")
+    @patch("controllers.console.auth.email_register.AccountService.revoke_email_register_token")
+    @patch("controllers.console.auth.email_register.AccountService.get_email_register_data")
+    @patch("controllers.console.auth.email_register.extract_remote_ip", return_value="127.0.0.1")
+    def test_reset_creates_account_with_normalized_email(
+        self,
+        mock_extract_ip,
+        mock_get_data,
+        mock_revoke_token,
+        mock_get_account,
+        mock_session_cls,
+        mock_create_account,
+        mock_login,
+        mock_reset_login_rate,
+        app,
+    ):
+        mock_get_data.return_value = {"phase": "register", "email": "Invitee@Example.com"}
+        mock_create_account.return_value = MagicMock()
+        token_pair = MagicMock()
+        token_pair.model_dump.return_value = {"access_token": "a", "refresh_token": "r"}
+        mock_login.return_value = token_pair
+
+        mock_session = MagicMock()
+        mock_session_cls.return_value.__enter__.return_value = mock_session
+        mock_get_account.return_value = None
+
+        feature_flags = SimpleNamespace(enable_email_password_login=True, is_allow_register=True)
+        with (
+            patch("controllers.console.auth.email_register.db", SimpleNamespace(engine="engine")),
+            patch("controllers.console.wraps.dify_config", SimpleNamespace(EDITION="CLOUD")),
+            patch("controllers.console.wraps.FeatureService.get_system_features", return_value=feature_flags),
+        ):
+            with app.test_request_context(
+                "/email-register",
+                method="POST",
+                json={"token": "token-123", "new_password": "ValidPass123!", "password_confirm": "ValidPass123!"},
+            ):
+                response = EmailRegisterResetApi().post()
+
+        assert response == {"result": "success", "data": {"access_token": "a", "refresh_token": "r"}}
+        mock_create_account.assert_called_once_with("invitee@example.com", "ValidPass123!")
+        mock_reset_login_rate.assert_called_once_with("invitee@example.com")
+        mock_revoke_token.assert_called_once_with("token-123")
+        mock_extract_ip.assert_called_once()
+        mock_get_account.assert_called_once_with("Invitee@Example.com", session=mock_session)
+
+
+def test_get_account_by_email_with_case_fallback_uses_lowercase_lookup():
+    mock_session = MagicMock()
+    first_query = MagicMock()
+    first_query.scalar_one_or_none.return_value = None
+    expected_account = MagicMock()
+    second_query = MagicMock()
+    second_query.scalar_one_or_none.return_value = expected_account
+    mock_session.execute.side_effect = [first_query, second_query]
+
+    account = AccountService.get_account_by_email_with_case_fallback("Case@Test.com", session=mock_session)
+
+    assert account is expected_account
+    assert mock_session.execute.call_count == 2

api/tests/unit_tests/controllers/console/auth/test_forgot_password.py

@@ -0,0 +1,176 @@
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+import pytest
+from flask import Flask
+
+from controllers.console.auth.forgot_password import (
+    ForgotPasswordCheckApi,
+    ForgotPasswordResetApi,
+    ForgotPasswordSendEmailApi,
+)
+from services.account_service import AccountService
+
+
+@pytest.fixture
+def app():
+    flask_app = Flask(__name__)
+    flask_app.config["TESTING"] = True
+    return flask_app
+
+
+class TestForgotPasswordSendEmailApi:
+    @patch("controllers.console.auth.forgot_password.Session")
+    @patch("controllers.console.auth.forgot_password.AccountService.get_account_by_email_with_case_fallback")
+    @patch("controllers.console.auth.forgot_password.AccountService.send_reset_password_email")
+    @patch("controllers.console.auth.forgot_password.AccountService.is_email_send_ip_limit", return_value=False)
+    @patch("controllers.console.auth.forgot_password.extract_remote_ip", return_value="127.0.0.1")
+    def test_send_normalizes_email(
+        self,
+        mock_extract_ip,
+        mock_is_ip_limit,
+        mock_send_email,
+        mock_get_account,
+        mock_session_cls,
+        app,
+    ):
+        mock_account = MagicMock()
+        mock_get_account.return_value = mock_account
+        mock_send_email.return_value = "token-123"
+        mock_session = MagicMock()
+        mock_session_cls.return_value.__enter__.return_value = mock_session
+
+        wraps_features = SimpleNamespace(enable_email_password_login=True, is_allow_register=True)
+        controller_features = SimpleNamespace(is_allow_register=True)
+        with (
+            patch("controllers.console.auth.forgot_password.db", SimpleNamespace(engine="engine")),
+            patch(
+                "controllers.console.auth.forgot_password.FeatureService.get_system_features",
+                return_value=controller_features,
+            ),
+            patch("controllers.console.wraps.dify_config", SimpleNamespace(EDITION="CLOUD")),
+            patch("controllers.console.wraps.FeatureService.get_system_features", return_value=wraps_features),
+        ):
+            with app.test_request_context(
+                "/forgot-password",
+                method="POST",
+                json={"email": "User@Example.com", "language": "zh-Hans"},
+            ):
+                response = ForgotPasswordSendEmailApi().post()
+
+        assert response == {"result": "success", "data": "token-123"}
+        mock_get_account.assert_called_once_with("User@Example.com", session=mock_session)
+        mock_send_email.assert_called_once_with(
+            account=mock_account,
+            email="user@example.com",
+            language="zh-Hans",
+            is_allow_register=True,
+        )
+        mock_is_ip_limit.assert_called_once_with("127.0.0.1")
+        mock_extract_ip.assert_called_once()
+
+
+class TestForgotPasswordCheckApi:
+    @patch("controllers.console.auth.forgot_password.AccountService.reset_forgot_password_error_rate_limit")
+    @patch("controllers.console.auth.forgot_password.AccountService.generate_reset_password_token")
+    @patch("controllers.console.auth.forgot_password.AccountService.revoke_reset_password_token")
+    @patch("controllers.console.auth.forgot_password.AccountService.add_forgot_password_error_rate_limit")
+    @patch("controllers.console.auth.forgot_password.AccountService.get_reset_password_data")
+    @patch("controllers.console.auth.forgot_password.AccountService.is_forgot_password_error_rate_limit")
+    def test_check_normalizes_email(
+        self,
+        mock_rate_limit_check,
+        mock_get_data,
+        mock_add_rate,
+        mock_revoke_token,
+        mock_generate_token,
+        mock_reset_rate,
+        app,
+    ):
+        mock_rate_limit_check.return_value = False
+        mock_get_data.return_value = {"email": "Admin@Example.com", "code": "4321"}
+        mock_generate_token.return_value = (None, "new-token")
+
+        wraps_features = SimpleNamespace(enable_email_password_login=True)
+        with (
+            patch("controllers.console.wraps.dify_config", SimpleNamespace(EDITION="CLOUD")),
+            patch("controllers.console.wraps.FeatureService.get_system_features", return_value=wraps_features),
+        ):
+            with app.test_request_context(
+                "/forgot-password/validity",
+                method="POST",
+                json={"email": "ADMIN@Example.com", "code": "4321", "token": "token-123"},
+            ):
+                response = ForgotPasswordCheckApi().post()
+
+        assert response == {"is_valid": True, "email": "admin@example.com", "token": "new-token"}
+        mock_rate_limit_check.assert_called_once_with("admin@example.com")
+        mock_generate_token.assert_called_once_with(
+            "Admin@Example.com",
+            code="4321",
+            additional_data={"phase": "reset"},
+        )
+        mock_reset_rate.assert_called_once_with("admin@example.com")
+        mock_add_rate.assert_not_called()
+        mock_revoke_token.assert_called_once_with("token-123")
+
+
+class TestForgotPasswordResetApi:
+    @patch("controllers.console.auth.forgot_password.ForgotPasswordResetApi._update_existing_account")
+    @patch("controllers.console.auth.forgot_password.Session")
+    @patch("controllers.console.auth.forgot_password.AccountService.get_account_by_email_with_case_fallback")
+    @patch("controllers.console.auth.forgot_password.AccountService.revoke_reset_password_token")
+    @patch("controllers.console.auth.forgot_password.AccountService.get_reset_password_data")
+    def test_reset_fetches_account_with_original_email(
+        self,
+        mock_get_reset_data,
+        mock_revoke_token,
+        mock_get_account,
+        mock_session_cls,
+        mock_update_account,
+        app,
+    ):
+        mock_get_reset_data.return_value = {"phase": "reset", "email": "User@Example.com"}
+        mock_account = MagicMock()
+        mock_get_account.return_value = mock_account
+
+        mock_session = MagicMock()
+        mock_session_cls.return_value.__enter__.return_value = mock_session
+
+        wraps_features = SimpleNamespace(enable_email_password_login=True)
+        with (
+            patch("controllers.console.auth.forgot_password.db", SimpleNamespace(engine="engine")),
+            patch("controllers.console.wraps.dify_config", SimpleNamespace(EDITION="CLOUD")),
+            patch("controllers.console.wraps.FeatureService.get_system_features", return_value=wraps_features),
+        ):
+            with app.test_request_context(
+                "/forgot-password/resets",
+                method="POST",
+                json={
+                    "token": "token-123",
+                    "new_password": "ValidPass123!",
+                    "password_confirm": "ValidPass123!",
+                },
+            ):
+                response = ForgotPasswordResetApi().post()
+
+        assert response == {"result": "success"}
+        mock_get_reset_data.assert_called_once_with("token-123")
+        mock_revoke_token.assert_called_once_with("token-123")
+        mock_get_account.assert_called_once_with("User@Example.com", session=mock_session)
+        mock_update_account.assert_called_once()
+
+
+def test_get_account_by_email_with_case_fallback_uses_lowercase_lookup():
+    mock_session = MagicMock()
+    first_query = MagicMock()
+    first_query.scalar_one_or_none.return_value = None
+    expected_account = MagicMock()
+    second_query = MagicMock()
+    second_query.scalar_one_or_none.return_value = expected_account
+    mock_session.execute.side_effect = [first_query, second_query]
+
+    account = AccountService.get_account_by_email_with_case_fallback("Mixed@Test.com", session=mock_session)
+
+    assert account is expected_account
+    assert mock_session.execute.call_count == 2

api/tests/unit_tests/controllers/console/auth/test_login_logout.py

@@ -76,7 +76,7 @@ class TestLoginApi:
     @patch("controllers.console.wraps.db")
     @patch("controllers.console.auth.login.dify_config.BILLING_ENABLED", False)
     @patch("controllers.console.auth.login.AccountService.is_login_error_rate_limit")
-    @patch("controllers.console.auth.login.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.login.RegisterService.get_invitation_with_case_fallback")
     @patch("controllers.console.auth.login.AccountService.authenticate")
     @patch("controllers.console.auth.login.TenantService.get_join_tenants")
     @patch("controllers.console.auth.login.AccountService.login")
@@ -120,7 +120,7 @@ class TestLoginApi:
             response = login_api.post()
 
         # Assert
-        mock_authenticate.assert_called_once_with("test@example.com", "ValidPass123!")
+        mock_authenticate.assert_called_once_with("test@example.com", "ValidPass123!", None)
         mock_login.assert_called_once()
         mock_reset_rate_limit.assert_called_once_with("test@example.com")
         assert response.json["result"] == "success"
@@ -128,7 +128,7 @@ class TestLoginApi:
     @patch("controllers.console.wraps.db")
     @patch("controllers.console.auth.login.dify_config.BILLING_ENABLED", False)
     @patch("controllers.console.auth.login.AccountService.is_login_error_rate_limit")
-    @patch("controllers.console.auth.login.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.login.RegisterService.get_invitation_with_case_fallback")
     @patch("controllers.console.auth.login.AccountService.authenticate")
     @patch("controllers.console.auth.login.TenantService.get_join_tenants")
     @patch("controllers.console.auth.login.AccountService.login")
@@ -182,7 +182,7 @@ class TestLoginApi:
     @patch("controllers.console.wraps.db")
     @patch("controllers.console.auth.login.dify_config.BILLING_ENABLED", False)
     @patch("controllers.console.auth.login.AccountService.is_login_error_rate_limit")
-    @patch("controllers.console.auth.login.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.login.RegisterService.get_invitation_with_case_fallback")
     def test_login_fails_when_rate_limited(self, mock_get_invitation, mock_is_rate_limit, mock_db, app):
         """
         Test login rejection when rate limit is exceeded.
@@ -230,7 +230,7 @@ class TestLoginApi:
     @patch("controllers.console.wraps.db")
     @patch("controllers.console.auth.login.dify_config.BILLING_ENABLED", False)
     @patch("controllers.console.auth.login.AccountService.is_login_error_rate_limit")
-    @patch("controllers.console.auth.login.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.login.RegisterService.get_invitation_with_case_fallback")
     @patch("controllers.console.auth.login.AccountService.authenticate")
     @patch("controllers.console.auth.login.AccountService.add_login_error_rate_limit")
     def test_login_fails_with_invalid_credentials(
@@ -269,7 +269,7 @@ class TestLoginApi:
     @patch("controllers.console.wraps.db")
     @patch("controllers.console.auth.login.dify_config.BILLING_ENABLED", False)
     @patch("controllers.console.auth.login.AccountService.is_login_error_rate_limit")
-    @patch("controllers.console.auth.login.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.login.RegisterService.get_invitation_with_case_fallback")
     @patch("controllers.console.auth.login.AccountService.authenticate")
     def test_login_fails_for_banned_account(
         self, mock_authenticate, mock_get_invitation, mock_is_rate_limit, mock_db, app
@@ -298,7 +298,7 @@ class TestLoginApi:
     @patch("controllers.console.wraps.db")
     @patch("controllers.console.auth.login.dify_config.BILLING_ENABLED", False)
     @patch("controllers.console.auth.login.AccountService.is_login_error_rate_limit")
-    @patch("controllers.console.auth.login.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.login.RegisterService.get_invitation_with_case_fallback")
     @patch("controllers.console.auth.login.AccountService.authenticate")
     @patch("controllers.console.auth.login.TenantService.get_join_tenants")
     @patch("controllers.console.auth.login.FeatureService.get_system_features")
@@ -343,7 +343,7 @@ class TestLoginApi:
     @patch("controllers.console.wraps.db")
     @patch("controllers.console.auth.login.dify_config.BILLING_ENABLED", False)
     @patch("controllers.console.auth.login.AccountService.is_login_error_rate_limit")
-    @patch("controllers.console.auth.login.RegisterService.get_invitation_if_token_valid")
+    @patch("controllers.console.auth.login.RegisterService.get_invitation_with_case_fallback")
     def test_login_invitation_email_mismatch(self, mock_get_invitation, mock_is_rate_limit, mock_db, app):
         """
         Test login failure when invitation email doesn't match login email.
@@ -371,6 +371,52 @@ class TestLoginApi:
             with pytest.raises(InvalidEmailError):
                 login_api.post()
 
+    @patch("controllers.console.wraps.db")
+    @patch("controllers.console.auth.login.dify_config.BILLING_ENABLED", False)
+    @patch("controllers.console.auth.login.AccountService.is_login_error_rate_limit")
+    @patch("controllers.console.auth.login.RegisterService.get_invitation_with_case_fallback")
+    @patch("controllers.console.auth.login.AccountService.authenticate")
+    @patch("controllers.console.auth.login.AccountService.add_login_error_rate_limit")
+    @patch("controllers.console.auth.login.TenantService.get_join_tenants")
+    @patch("controllers.console.auth.login.AccountService.login")
+    @patch("controllers.console.auth.login.AccountService.reset_login_error_rate_limit")
+    def test_login_retries_with_lowercase_email(
+        self,
+        mock_reset_rate_limit,
+        mock_login_service,
+        mock_get_tenants,
+        mock_add_rate_limit,
+        mock_authenticate,
+        mock_get_invitation,
+        mock_is_rate_limit,
+        mock_db,
+        app,
+        mock_account,
+        mock_token_pair,
+    ):
+        """Test that login retries with lowercase email when uppercase lookup fails."""
+        mock_db.session.query.return_value.first.return_value = MagicMock()
+        mock_is_rate_limit.return_value = False
+        mock_get_invitation.return_value = None
+        mock_authenticate.side_effect = [AccountPasswordError("Invalid"), mock_account]
+        mock_get_tenants.return_value = [MagicMock()]
+        mock_login_service.return_value = mock_token_pair
+
+        with app.test_request_context(
+            "/login",
+            method="POST",
+            json={"email": "Upper@Example.com", "password": encode_password("ValidPass123!")},
+        ):
+            response = LoginApi().post()
+
+        assert response.json["result"] == "success"
+        assert mock_authenticate.call_args_list == [
+            (("Upper@Example.com", "ValidPass123!", None), {}),
+            (("upper@example.com", "ValidPass123!", None), {}),
+        ]
+        mock_add_rate_limit.assert_not_called()
+        mock_reset_rate_limit.assert_called_once_with("upper@example.com")
+
 
 class TestLogoutApi:
     """Test cases for the LogoutApi endpoint."""

api/tests/unit_tests/controllers/console/auth/test_oauth.py

@@ -12,6 +12,7 @@ from controllers.console.auth.oauth import (
 )
 from libs.oauth import OAuthUserInfo
 from models.account import AccountStatus
+from services.account_service import AccountService
 from services.errors.account import AccountRegisterError
 
 
@@ -215,6 +216,34 @@ class TestOAuthCallback:
         assert status_code == 400
         assert response["error"] == expected_error
 
+    @patch("controllers.console.auth.oauth.dify_config")
+    @patch("controllers.console.auth.oauth.get_oauth_providers")
+    @patch("controllers.console.auth.oauth.RegisterService")
+    @patch("controllers.console.auth.oauth.redirect")
+    def test_invitation_comparison_is_case_insensitive(
+        self,
+        mock_redirect,
+        mock_register_service,
+        mock_get_providers,
+        mock_config,
+        resource,
+        app,
+        oauth_setup,
+    ):
+        mock_config.CONSOLE_WEB_URL = "http://localhost:3000"
+        oauth_setup["provider"].get_user_info.return_value = OAuthUserInfo(
+            id="123", name="Test User", email="User@Example.com"
+        )
+        mock_get_providers.return_value = {"github": oauth_setup["provider"]}
+        mock_register_service.is_valid_invite_token.return_value = True
+        mock_register_service.get_invitation_by_token.return_value = {"email": "user@example.com"}
+
+        with app.test_request_context("/auth/oauth/github/callback?code=test_code&state=invite123"):
+            resource.get("github")
+
+        mock_register_service.get_invitation_by_token.assert_called_once_with(token="invite123")
+        mock_redirect.assert_called_once_with("http://localhost:3000/signin/invite-settings?invite_token=invite123")
+
     @pytest.mark.parametrize(
         ("account_status", "expected_redirect"),
         [
@@ -395,12 +424,12 @@ class TestAccountGeneration:
         account.name = "Test User"
         return account
 
-    @patch("controllers.console.auth.oauth.db")
-    @patch("controllers.console.auth.oauth.Account")
+    @patch("controllers.console.auth.oauth.AccountService.get_account_by_email_with_case_fallback")
     @patch("controllers.console.auth.oauth.Session")
-    @patch("controllers.console.auth.oauth.select")
+    @patch("controllers.console.auth.oauth.Account")
+    @patch("controllers.console.auth.oauth.db")
     def test_should_get_account_by_openid_or_email(
-        self, mock_select, mock_session, mock_account_model, mock_db, user_info, mock_account
+        self, mock_db, mock_account_model, mock_session, mock_get_account, user_info, mock_account
     ):
         # Mock db.engine for Session creation
         mock_db.engine = MagicMock()
@@ -410,15 +439,31 @@ class TestAccountGeneration:
         result = _get_account_by_openid_or_email("github", user_info)
         assert result == mock_account
         mock_account_model.get_by_openid.assert_called_once_with("github", "123")
+        mock_get_account.assert_not_called()
 
-        # Test fallback to email
+        # Test fallback to email lookup
         mock_account_model.get_by_openid.return_value = None
         mock_session_instance = MagicMock()
-        mock_session_instance.execute.return_value.scalar_one_or_none.return_value = mock_account
         mock_session.return_value.__enter__.return_value = mock_session_instance
+        mock_get_account.return_value = mock_account
 
         result = _get_account_by_openid_or_email("github", user_info)
         assert result == mock_account
+        mock_get_account.assert_called_once_with(user_info.email, session=mock_session_instance)
+
+    def test_get_account_by_email_with_case_fallback_uses_lowercase_lookup(self):
+        mock_session = MagicMock()
+        first_result = MagicMock()
+        first_result.scalar_one_or_none.return_value = None
+        expected_account = MagicMock()
+        second_result = MagicMock()
+        second_result.scalar_one_or_none.return_value = expected_account
+        mock_session.execute.side_effect = [first_result, second_result]
+
+        result = AccountService.get_account_by_email_with_case_fallback("Case@Test.com", session=mock_session)
+
+        assert result == expected_account
+        assert mock_session.execute.call_count == 2
 
     @pytest.mark.parametrize(
         ("allow_register", "existing_account", "should_create"),
@@ -466,6 +511,35 @@ class TestAccountGeneration:
                     mock_register_service.register.assert_called_once_with(
                         email="test@example.com", name="Test User", password=None, open_id="123", provider="github"
                     )
+                else:
+                    mock_register_service.register.assert_not_called()
+
+    @patch("controllers.console.auth.oauth._get_account_by_openid_or_email", return_value=None)
+    @patch("controllers.console.auth.oauth.FeatureService")
+    @patch("controllers.console.auth.oauth.RegisterService")
+    @patch("controllers.console.auth.oauth.AccountService")
+    @patch("controllers.console.auth.oauth.TenantService")
+    @patch("controllers.console.auth.oauth.db")
+    def test_should_register_with_lowercase_email(
+        self,
+        mock_db,
+        mock_tenant_service,
+        mock_account_service,
+        mock_register_service,
+        mock_feature_service,
+        mock_get_account,
+        app,
+    ):
+        user_info = OAuthUserInfo(id="123", name="Test User", email="Upper@Example.com")
+        mock_feature_service.get_system_features.return_value.is_allow_register = True
+        mock_register_service.register.return_value = MagicMock()
+
+        with app.test_request_context(headers={"Accept-Language": "en-US"}):
+            _generate_account("github", user_info)
+
+        mock_register_service.register.assert_called_once_with(
+            email="upper@example.com", name="Test User", password=None, open_id="123", provider="github"
+        )
 
     @patch("controllers.console.auth.oauth._get_account_by_openid_or_email")
     @patch("controllers.console.auth.oauth.TenantService")

api/tests/unit_tests/controllers/console/auth/test_password_reset.py

@@ -28,6 +28,22 @@ from controllers.console.auth.forgot_password import (
 from controllers.console.error import AccountNotFound, EmailSendIpLimitError
 
 
+@pytest.fixture(autouse=True)
+def _mock_forgot_password_session():
+    with patch("controllers.console.auth.forgot_password.Session") as mock_session_cls:
+        mock_session = MagicMock()
+        mock_session_cls.return_value.__enter__.return_value = mock_session
+        mock_session_cls.return_value.__exit__.return_value = None
+        yield mock_session
+
+
+@pytest.fixture(autouse=True)
+def _mock_forgot_password_db():
+    with patch("controllers.console.auth.forgot_password.db") as mock_db:
+        mock_db.engine = MagicMock()
+        yield mock_db
+
+
 class TestForgotPasswordSendEmailApi:
     """Test cases for sending password reset emails."""
 
@@ -47,20 +63,16 @@ class TestForgotPasswordSendEmailApi:
         return account
 
     @patch("controllers.console.wraps.db")
-    @patch("controllers.console.auth.forgot_password.db")
     @patch("controllers.console.auth.forgot_password.AccountService.is_email_send_ip_limit")
-    @patch("controllers.console.auth.forgot_password.Session")
-    @patch("controllers.console.auth.forgot_password.select")
+    @patch("controllers.console.auth.forgot_password.AccountService.get_account_by_email_with_case_fallback")
     @patch("controllers.console.auth.forgot_password.AccountService.send_reset_password_email")
     @patch("controllers.console.auth.forgot_password.FeatureService.get_system_features")
     def test_send_reset_email_success(
         self,
         mock_get_features,
         mock_send_email,
-        mock_select,
-        mock_session,
+        mock_get_account,
         mock_is_ip_limit,
-        mock_forgot_db,
         mock_wraps_db,
         app,
         mock_account,
@@ -75,11 +87,8 @@ class TestForgotPasswordSendEmailApi:
         """
         # Arrange
         mock_wraps_db.session.query.return_value.first.return_value = MagicMock()
-        mock_forgot_db.engine = MagicMock()
         mock_is_ip_limit.return_value = False
-        mock_session_instance = MagicMock()
-        mock_session_instance.execute.return_value.scalar_one_or_none.return_value = mock_account
-        mock_session.return_value.__enter__.return_value = mock_session_instance
+        mock_get_account.return_value = mock_account
         mock_send_email.return_value = "reset_token_123"
         mock_get_features.return_value.is_allow_register = True
 
@@ -125,20 +134,16 @@ class TestForgotPasswordSendEmailApi:
         ],
     )
     @patch("controllers.console.wraps.db")
-    @patch("controllers.console.auth.forgot_password.db")
     @patch("controllers.console.auth.forgot_password.AccountService.is_email_send_ip_limit")
-    @patch("controllers.console.auth.forgot_password.Session")
-    @patch("controllers.console.auth.forgot_password.select")
+    @patch("controllers.console.auth.forgot_password.AccountService.get_account_by_email_with_case_fallback")
     @patch("controllers.console.auth.forgot_password.AccountService.send_reset_password_email")
     @patch("controllers.console.auth.forgot_password.FeatureService.get_system_features")
     def test_send_reset_email_language_handling(
         self,
         mock_get_features,
         mock_send_email,
-        mock_select,
-        mock_session,
+        mock_get_account,
         mock_is_ip_limit,
-        mock_forgot_db,
         mock_wraps_db,
         app,
         mock_account,
@@ -154,11 +159,8 @@ class TestForgotPasswordSendEmailApi:
         """
         # Arrange
         mock_wraps_db.session.query.return_value.first.return_value = MagicMock()
-        mock_forgot_db.engine = MagicMock()
         mock_is_ip_limit.return_value = False
-        mock_session_instance = MagicMock()
-        mock_session_instance.execute.return_value.scalar_one_or_none.return_value = mock_account
-        mock_session.return_value.__enter__.return_value = mock_session_instance
+        mock_get_account.return_value = mock_account
         mock_send_email.return_value = "token"
         mock_get_features.return_value.is_allow_register = True
 
@@ -229,8 +231,46 @@ class TestForgotPasswordCheckApi:
         assert response["email"] == "test@example.com"
         assert response["token"] == "new_token"
         mock_revoke_token.assert_called_once_with("old_token")
+        mock_generate_token.assert_called_once_with(
+            "test@example.com", code="123456", additional_data={"phase": "reset"}
+        )
         mock_reset_rate_limit.assert_called_once_with("test@example.com")
 
+    @patch("controllers.console.wraps.db")
+    @patch("controllers.console.auth.forgot_password.AccountService.is_forgot_password_error_rate_limit")
+    @patch("controllers.console.auth.forgot_password.AccountService.get_reset_password_data")
+    @patch("controllers.console.auth.forgot_password.AccountService.revoke_reset_password_token")
+    @patch("controllers.console.auth.forgot_password.AccountService.generate_reset_password_token")
+    @patch("controllers.console.auth.forgot_password.AccountService.reset_forgot_password_error_rate_limit")
+    def test_verify_code_preserves_token_email_case(
+        self,
+        mock_reset_rate_limit,
+        mock_generate_token,
+        mock_revoke_token,
+        mock_get_data,
+        mock_is_rate_limit,
+        mock_db,
+        app,
+    ):
+        mock_db.session.query.return_value.first.return_value = MagicMock()
+        mock_is_rate_limit.return_value = False
+        mock_get_data.return_value = {"email": "User@Example.com", "code": "999888"}
+        mock_generate_token.return_value = (None, "fresh-token")
+
+        with app.test_request_context(
+            "/forgot-password/validity",
+            method="POST",
+            json={"email": "user@example.com", "code": "999888", "token": "upper_token"},
+        ):
+            response = ForgotPasswordCheckApi().post()
+
+        assert response == {"is_valid": True, "email": "user@example.com", "token": "fresh-token"}
+        mock_generate_token.assert_called_once_with(
+            "User@Example.com", code="999888", additional_data={"phase": "reset"}
+        )
+        mock_revoke_token.assert_called_once_with("upper_token")
+        mock_reset_rate_limit.assert_called_once_with("user@example.com")
+
     @patch("controllers.console.wraps.db")
     @patch("controllers.console.auth.forgot_password.AccountService.is_forgot_password_error_rate_limit")
     def test_verify_code_rate_limited(self, mock_is_rate_limit, mock_db, app):
@@ -355,20 +395,16 @@ class TestForgotPasswordResetApi:
         return account
 
     @patch("controllers.console.wraps.db")
-    @patch("controllers.console.auth.forgot_password.db")
     @patch("controllers.console.auth.forgot_password.AccountService.get_reset_password_data")
     @patch("controllers.console.auth.forgot_password.AccountService.revoke_reset_password_token")
-    @patch("controllers.console.auth.forgot_password.Session")
-    @patch("controllers.console.auth.forgot_password.select")
+    @patch("controllers.console.auth.forgot_password.AccountService.get_account_by_email_with_case_fallback")
     @patch("controllers.console.auth.forgot_password.TenantService.get_join_tenants")
     def test_reset_password_success(
         self,
         mock_get_tenants,
-        mock_select,
-        mock_session,
+        mock_get_account,
         mock_revoke_token,
         mock_get_data,
-        mock_forgot_db,
         mock_wraps_db,
         app,
         mock_account,
@@ -383,11 +419,8 @@ class TestForgotPasswordResetApi:
         """
         # Arrange
         mock_wraps_db.session.query.return_value.first.return_value = MagicMock()
-        mock_forgot_db.engine = MagicMock()
         mock_get_data.return_value = {"email": "test@example.com", "phase": "reset"}
-        mock_session_instance = MagicMock()
-        mock_session_instance.execute.return_value.scalar_one_or_none.return_value = mock_account
-        mock_session.return_value.__enter__.return_value = mock_session_instance
+        mock_get_account.return_value = mock_account
         mock_get_tenants.return_value = [MagicMock()]
 
         # Act
@@ -475,13 +508,11 @@ class TestForgotPasswordResetApi:
                 api.post()
 
     @patch("controllers.console.wraps.db")
-    @patch("controllers.console.auth.forgot_password.db")
     @patch("controllers.console.auth.forgot_password.AccountService.get_reset_password_data")
     @patch("controllers.console.auth.forgot_password.AccountService.revoke_reset_password_token")
-    @patch("controllers.console.auth.forgot_password.Session")
-    @patch("controllers.console.auth.forgot_password.select")
+    @patch("controllers.console.auth.forgot_password.AccountService.get_account_by_email_with_case_fallback")
     def test_reset_password_account_not_found(
-        self, mock_select, mock_session, mock_revoke_token, mock_get_data, mock_forgot_db, mock_wraps_db, app
+        self, mock_get_account, mock_revoke_token, mock_get_data, mock_wraps_db, app
     ):
         """
         Test password reset for non-existent account.
@@ -491,11 +522,8 @@ class TestForgotPasswordResetApi:
         """
         # Arrange
         mock_wraps_db.session.query.return_value.first.return_value = MagicMock()
-        mock_forgot_db.engine = MagicMock()
         mock_get_data.return_value = {"email": "nonexistent@example.com", "phase": "reset"}
-        mock_session_instance = MagicMock()
-        mock_session_instance.execute.return_value.scalar_one_or_none.return_value = None
-        mock_session.return_value.__enter__.return_value = mock_session_instance
+        mock_get_account.return_value = None
 
         # Act & Assert
         with app.test_request_context(

api/tests/unit_tests/controllers/console/test_setup.py

@@ -0,0 +1,39 @@
+from types import SimpleNamespace
+from unittest.mock import patch
+
+from controllers.console.setup import SetupApi
+
+
+class TestSetupApi:
+    def test_post_lowercases_email_before_register(self):
+        """Ensure setup registration normalizes email casing."""
+        payload = {
+            "email": "Admin@Example.com",
+            "name": "Admin User",
+            "password": "ValidPass123!",
+            "language": "en-US",
+        }
+        setup_api = SetupApi(api=None)
+
+        mock_console_ns = SimpleNamespace(payload=payload)
+
+        with (
+            patch("controllers.console.setup.console_ns", mock_console_ns),
+            patch("controllers.console.setup.get_setup_status", return_value=False),
+            patch("controllers.console.setup.TenantService.get_tenant_count", return_value=0),
+            patch("controllers.console.setup.get_init_validate_status", return_value=True),
+            patch("controllers.console.setup.extract_remote_ip", return_value="127.0.0.1"),
+            patch("controllers.console.setup.request", object()),
+            patch("controllers.console.setup.RegisterService.setup") as mock_register,
+        ):
+            response, status = setup_api.post()
+
+        assert response == {"result": "success"}
+        assert status == 201
+        mock_register.assert_called_once_with(
+            email="admin@example.com",
+            name=payload["name"],
+            password=payload["password"],
+            ip_address="127.0.0.1",
+            language=payload["language"],
+        )

api/tests/unit_tests/controllers/console/test_workspace_account.py

@@ -0,0 +1,247 @@
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+import pytest
+from flask import Flask, g
+
+from controllers.console.workspace.account import (
+    AccountDeleteUpdateFeedbackApi,
+    ChangeEmailCheckApi,
+    ChangeEmailResetApi,
+    ChangeEmailSendEmailApi,
+    CheckEmailUnique,
+)
+from models import Account
+from services.account_service import AccountService
+
+
+@pytest.fixture
+def app():
+    app = Flask(__name__)
+    app.config["TESTING"] = True
+    app.config["RESTX_MASK_HEADER"] = "X-Fields"
+    app.login_manager = SimpleNamespace(_load_user=lambda: None)
+    return app
+
+
+def _mock_wraps_db(mock_db):
+    mock_db.session.query.return_value.first.return_value = MagicMock()
+
+
+def _build_account(email: str, account_id: str = "acc", tenant: object | None = None) -> Account:
+    tenant_obj = tenant if tenant is not None else SimpleNamespace(id="tenant-id")
+    account = Account(name=account_id, email=email)
+    account.email = email
+    account.id = account_id
+    account.status = "active"
+    account._current_tenant = tenant_obj
+    return account
+
+
+def _set_logged_in_user(account: Account):
+    g._login_user = account
+    g._current_tenant = account.current_tenant
+
+
+class TestChangeEmailSend:
+    @patch("controllers.console.wraps.db")
+    @patch("controllers.console.workspace.account.current_account_with_tenant")
+    @patch("controllers.console.workspace.account.AccountService.get_change_email_data")
+    @patch("controllers.console.workspace.account.AccountService.send_change_email_email")
+    @patch("controllers.console.workspace.account.AccountService.is_email_send_ip_limit", return_value=False)
+    @patch("controllers.console.workspace.account.extract_remote_ip", return_value="127.0.0.1")
+    @patch("libs.login.check_csrf_token", return_value=None)
+    @patch("controllers.console.wraps.FeatureService.get_system_features")
+    def test_should_normalize_new_email_phase(
+        self,
+        mock_features,
+        mock_csrf,
+        mock_extract_ip,
+        mock_is_ip_limit,
+        mock_send_email,
+        mock_get_change_data,
+        mock_current_account,
+        mock_db,
+        app,
+    ):
+        _mock_wraps_db(mock_db)
+        mock_features.return_value = SimpleNamespace(enable_change_email=True)
+        mock_account = _build_account("current@example.com", "acc1")
+        mock_current_account.return_value = (mock_account, None)
+        mock_get_change_data.return_value = {"email": "current@example.com"}
+        mock_send_email.return_value = "token-abc"
+
+        with app.test_request_context(
+            "/account/change-email",
+            method="POST",
+            json={"email": "New@Example.com", "language": "en-US", "phase": "new_email", "token": "token-123"},
+        ):
+            _set_logged_in_user(_build_account("tester@example.com", "tester"))
+            response = ChangeEmailSendEmailApi().post()
+
+        assert response == {"result": "success", "data": "token-abc"}
+        mock_send_email.assert_called_once_with(
+            account=None,
+            email="new@example.com",
+            old_email="current@example.com",
+            language="en-US",
+            phase="new_email",
+        )
+        mock_extract_ip.assert_called_once()
+        mock_is_ip_limit.assert_called_once_with("127.0.0.1")
+        mock_csrf.assert_called_once()
+
+
+class TestChangeEmailValidity:
+    @patch("controllers.console.wraps.db")
+    @patch("controllers.console.workspace.account.current_account_with_tenant")
+    @patch("controllers.console.workspace.account.AccountService.reset_change_email_error_rate_limit")
+    @patch("controllers.console.workspace.account.AccountService.generate_change_email_token")
+    @patch("controllers.console.workspace.account.AccountService.revoke_change_email_token")
+    @patch("controllers.console.workspace.account.AccountService.add_change_email_error_rate_limit")
+    @patch("controllers.console.workspace.account.AccountService.get_change_email_data")
+    @patch("controllers.console.workspace.account.AccountService.is_change_email_error_rate_limit")
+    @patch("libs.login.check_csrf_token", return_value=None)
+    @patch("controllers.console.wraps.FeatureService.get_system_features")
+    def test_should_validate_with_normalized_email(
+        self,
+        mock_features,
+        mock_csrf,
+        mock_is_rate_limit,
+        mock_get_data,
+        mock_add_rate,
+        mock_revoke_token,
+        mock_generate_token,
+        mock_reset_rate,
+        mock_current_account,
+        mock_db,
+        app,
+    ):
+        _mock_wraps_db(mock_db)
+        mock_features.return_value = SimpleNamespace(enable_change_email=True)
+        mock_account = _build_account("user@example.com", "acc2")
+        mock_current_account.return_value = (mock_account, None)
+        mock_is_rate_limit.return_value = False
+        mock_get_data.return_value = {"email": "user@example.com", "code": "1234", "old_email": "old@example.com"}
+        mock_generate_token.return_value = (None, "new-token")
+
+        with app.test_request_context(
+            "/account/change-email/validity",
+            method="POST",
+            json={"email": "User@Example.com", "code": "1234", "token": "token-123"},
+        ):
+            _set_logged_in_user(_build_account("tester@example.com", "tester"))
+            response = ChangeEmailCheckApi().post()
+
+        assert response == {"is_valid": True, "email": "user@example.com", "token": "new-token"}
+        mock_is_rate_limit.assert_called_once_with("user@example.com")
+        mock_add_rate.assert_not_called()
+        mock_revoke_token.assert_called_once_with("token-123")
+        mock_generate_token.assert_called_once_with(
+            "user@example.com", code="1234", old_email="old@example.com", additional_data={}
+        )
+        mock_reset_rate.assert_called_once_with("user@example.com")
+        mock_csrf.assert_called_once()
+
+
+class TestChangeEmailReset:
+    @patch("controllers.console.wraps.db")
+    @patch("controllers.console.workspace.account.current_account_with_tenant")
+    @patch("controllers.console.workspace.account.AccountService.send_change_email_completed_notify_email")
+    @patch("controllers.console.workspace.account.AccountService.update_account_email")
+    @patch("controllers.console.workspace.account.AccountService.revoke_change_email_token")
+    @patch("controllers.console.workspace.account.AccountService.get_change_email_data")
+    @patch("controllers.console.workspace.account.AccountService.check_email_unique")
+    @patch("controllers.console.workspace.account.AccountService.is_account_in_freeze")
+    @patch("libs.login.check_csrf_token", return_value=None)
+    @patch("controllers.console.wraps.FeatureService.get_system_features")
+    def test_should_normalize_new_email_before_update(
+        self,
+        mock_features,
+        mock_csrf,
+        mock_is_freeze,
+        mock_check_unique,
+        mock_get_data,
+        mock_revoke_token,
+        mock_update_account,
+        mock_send_notify,
+        mock_current_account,
+        mock_db,
+        app,
+    ):
+        _mock_wraps_db(mock_db)
+        mock_features.return_value = SimpleNamespace(enable_change_email=True)
+        current_user = _build_account("old@example.com", "acc3")
+        mock_current_account.return_value = (current_user, None)
+        mock_is_freeze.return_value = False
+        mock_check_unique.return_value = True
+        mock_get_data.return_value = {"old_email": "OLD@example.com"}
+        mock_account_after_update = _build_account("new@example.com", "acc3-updated")
+        mock_update_account.return_value = mock_account_after_update
+
+        with app.test_request_context(
+            "/account/change-email/reset",
+            method="POST",
+            json={"new_email": "New@Example.com", "token": "token-123"},
+        ):
+            _set_logged_in_user(_build_account("tester@example.com", "tester"))
+            ChangeEmailResetApi().post()
+
+            mock_is_freeze.assert_called_once_with("new@example.com")
+            mock_check_unique.assert_called_once_with("new@example.com")
+            mock_revoke_token.assert_called_once_with("token-123")
+            mock_update_account.assert_called_once_with(current_user, email="new@example.com")
+            mock_send_notify.assert_called_once_with(email="new@example.com")
+            mock_csrf.assert_called_once()
+
+
+class TestAccountDeletionFeedback:
+    @patch("controllers.console.wraps.db")
+    @patch("controllers.console.workspace.account.BillingService.update_account_deletion_feedback")
+    def test_should_normalize_feedback_email(self, mock_update, mock_db, app):
+        _mock_wraps_db(mock_db)
+        with app.test_request_context(
+            "/account/delete/feedback",
+            method="POST",
+            json={"email": "User@Example.com", "feedback": "test"},
+        ):
+            response = AccountDeleteUpdateFeedbackApi().post()
+
+        assert response == {"result": "success"}
+        mock_update.assert_called_once_with("User@Example.com", "test")
+
+
+class TestCheckEmailUnique:
+    @patch("controllers.console.wraps.db")
+    @patch("controllers.console.workspace.account.AccountService.check_email_unique")
+    @patch("controllers.console.workspace.account.AccountService.is_account_in_freeze")
+    def test_should_normalize_email(self, mock_is_freeze, mock_check_unique, mock_db, app):
+        _mock_wraps_db(mock_db)
+        mock_is_freeze.return_value = False
+        mock_check_unique.return_value = True
+
+        with app.test_request_context(
+            "/account/change-email/check-email-unique",
+            method="POST",
+            json={"email": "Case@Test.com"},
+        ):
+            response = CheckEmailUnique().post()
+
+        assert response == {"result": "success"}
+        mock_is_freeze.assert_called_once_with("case@test.com")
+        mock_check_unique.assert_called_once_with("case@test.com")
+
+
+def test_get_account_by_email_with_case_fallback_uses_lowercase_lookup():
+    session = MagicMock()
+    first = MagicMock()
+    first.scalar_one_or_none.return_value = None
+    second = MagicMock()
+    expected_account = MagicMock()
+    second.scalar_one_or_none.return_value = expected_account
+    session.execute.side_effect = [first, second]
+
+    result = AccountService.get_account_by_email_with_case_fallback("Mixed@Test.com", session=session)
+
+    assert result is expected_account
+    assert session.execute.call_count == 2

api/tests/unit_tests/controllers/console/test_workspace_members.py

@@ -0,0 +1,82 @@
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+import pytest
+from flask import Flask, g
+
+from controllers.console.workspace.members import MemberInviteEmailApi
+from models.account import Account, TenantAccountRole
+
+
+@pytest.fixture
+def app():
+    flask_app = Flask(__name__)
+    flask_app.config["TESTING"] = True
+    flask_app.login_manager = SimpleNamespace(_load_user=lambda: None)
+    return flask_app
+
+
+def _mock_wraps_db(mock_db):
+    mock_db.session.query.return_value.first.return_value = MagicMock()
+
+
+def _build_feature_flags():
+    placeholder_quota = SimpleNamespace(limit=0, size=0)
+    workspace_members = SimpleNamespace(is_available=lambda count: True)
+    return SimpleNamespace(
+        billing=SimpleNamespace(enabled=False),
+        workspace_members=workspace_members,
+        members=placeholder_quota,
+        apps=placeholder_quota,
+        vector_space=placeholder_quota,
+        documents_upload_quota=placeholder_quota,
+        annotation_quota_limit=placeholder_quota,
+    )
+
+
+class TestMemberInviteEmailApi:
+    @patch("controllers.console.workspace.members.FeatureService.get_features")
+    @patch("controllers.console.workspace.members.RegisterService.invite_new_member")
+    @patch("controllers.console.workspace.members.current_account_with_tenant")
+    @patch("controllers.console.wraps.db")
+    @patch("libs.login.check_csrf_token", return_value=None)
+    def test_invite_normalizes_emails(
+        self,
+        mock_csrf,
+        mock_db,
+        mock_current_account,
+        mock_invite_member,
+        mock_get_features,
+        app,
+    ):
+        _mock_wraps_db(mock_db)
+        mock_get_features.return_value = _build_feature_flags()
+        mock_invite_member.return_value = "token-abc"
+
+        tenant = SimpleNamespace(id="tenant-1", name="Test Tenant")
+        inviter = SimpleNamespace(email="Owner@Example.com", current_tenant=tenant, status="active")
+        mock_current_account.return_value = (inviter, tenant.id)
+
+        with patch("controllers.console.workspace.members.dify_config.CONSOLE_WEB_URL", "https://console.example.com"):
+            with app.test_request_context(
+                "/workspaces/current/members/invite-email",
+                method="POST",
+                json={"emails": ["User@Example.com"], "role": TenantAccountRole.EDITOR.value, "language": "en-US"},
+            ):
+                account = Account(name="tester", email="tester@example.com")
+                account._current_tenant = tenant
+                g._login_user = account
+                g._current_tenant = tenant
+                response, status_code = MemberInviteEmailApi().post()
+
+        assert status_code == 201
+        assert response["invitation_results"][0]["email"] == "user@example.com"
+
+        assert mock_invite_member.call_count == 1
+        call_args = mock_invite_member.call_args
+        assert call_args.kwargs["tenant"] == tenant
+        assert call_args.kwargs["email"] == "User@Example.com"
+        assert call_args.kwargs["language"] == "en-US"
+        assert call_args.kwargs["role"] == TenantAccountRole.EDITOR
+        assert call_args.kwargs["inviter"] == inviter
+        mock_csrf.assert_called_once()

api/tests/unit_tests/controllers/web/test_forgot_password.py

@@ -1,195 +0,0 @@
-"""Unit tests for controllers.web.forgot_password endpoints."""
-
-from __future__ import annotations
-
-import base64
-import builtins
-from types import SimpleNamespace
-from unittest.mock import MagicMock, patch
-
-import pytest
-from flask import Flask
-from flask.views import MethodView
-
-# Ensure flask_restx.api finds MethodView during import.
-if not hasattr(builtins, "MethodView"):
-    builtins.MethodView = MethodView  # type: ignore[attr-defined]
-
-
-def _load_controller_module():
-    """Import controllers.web.forgot_password using a stub package."""
-
-    import importlib
-    import importlib.util
-    import sys
-    from types import ModuleType
-
-    parent_module_name = "controllers.web"
-    module_name = f"{parent_module_name}.forgot_password"
-
-    if parent_module_name not in sys.modules:
-        from flask_restx import Namespace
-
-        stub = ModuleType(parent_module_name)
-        stub.__file__ = "controllers/web/__init__.py"
-        stub.__path__ = ["controllers/web"]
-        stub.__package__ = "controllers"
-        stub.__spec__ = importlib.util.spec_from_loader(parent_module_name, loader=None, is_package=True)
-        stub.web_ns = Namespace("web", description="Web API", path="/")
-        sys.modules[parent_module_name] = stub
-
-    return importlib.import_module(module_name)
-
-
-forgot_password_module = _load_controller_module()
-ForgotPasswordCheckApi = forgot_password_module.ForgotPasswordCheckApi
-ForgotPasswordResetApi = forgot_password_module.ForgotPasswordResetApi
-ForgotPasswordSendEmailApi = forgot_password_module.ForgotPasswordSendEmailApi
-
-
-@pytest.fixture
-def app() -> Flask:
-    """Configure a minimal Flask app for request contexts."""
-
-    app = Flask(__name__)
-    app.config["TESTING"] = True
-    return app
-
-
-@pytest.fixture(autouse=True)
-def _enable_web_endpoint_guards():
-    """Stub enterprise and feature toggles used by route decorators."""
-
-    features = SimpleNamespace(enable_email_password_login=True)
-    with (
-        patch("controllers.console.wraps.dify_config.ENTERPRISE_ENABLED", True),
-        patch("controllers.console.wraps.dify_config.EDITION", "CLOUD"),
-        patch("controllers.console.wraps.FeatureService.get_system_features", return_value=features),
-    ):
-        yield
-
-
-@pytest.fixture(autouse=True)
-def _mock_controller_db():
-    """Replace controller-level db reference with a simple stub."""
-
-    fake_db = SimpleNamespace(engine=MagicMock(name="engine"))
-    fake_wraps_db = SimpleNamespace(
-        session=MagicMock(query=MagicMock(return_value=MagicMock(first=MagicMock(return_value=True))))
-    )
-    with (
-        patch("controllers.web.forgot_password.db", fake_db),
-        patch("controllers.console.wraps.db", fake_wraps_db),
-    ):
-        yield fake_db
-
-
-@patch("controllers.web.forgot_password.AccountService.send_reset_password_email", return_value="reset-token")
-@patch("controllers.web.forgot_password.Session")
-@patch("controllers.web.forgot_password.AccountService.is_email_send_ip_limit", return_value=False)
-@patch("controllers.web.forgot_password.extract_remote_ip", return_value="203.0.113.10")
-def test_send_reset_email_success(
-    mock_extract_ip: MagicMock,
-    mock_is_ip_limit: MagicMock,
-    mock_session: MagicMock,
-    mock_send_email: MagicMock,
-    app: Flask,
-):
-    """POST /forgot-password returns token when email exists and limits allow."""
-
-    mock_account = MagicMock()
-    session_ctx = MagicMock()
-    mock_session.return_value.__enter__.return_value = session_ctx
-    session_ctx.execute.return_value.scalar_one_or_none.return_value = mock_account
-
-    with app.test_request_context(
-        "/forgot-password",
-        method="POST",
-        json={"email": "user@example.com"},
-    ):
-        response = ForgotPasswordSendEmailApi().post()
-
-    assert response == {"result": "success", "data": "reset-token"}
-    mock_extract_ip.assert_called_once()
-    mock_is_ip_limit.assert_called_once_with("203.0.113.10")
-    mock_send_email.assert_called_once_with(account=mock_account, email="user@example.com", language="en-US")
-
-
-@patch("controllers.web.forgot_password.AccountService.reset_forgot_password_error_rate_limit")
-@patch("controllers.web.forgot_password.AccountService.generate_reset_password_token", return_value=({}, "new-token"))
-@patch("controllers.web.forgot_password.AccountService.revoke_reset_password_token")
-@patch("controllers.web.forgot_password.AccountService.get_reset_password_data")
-@patch("controllers.web.forgot_password.AccountService.is_forgot_password_error_rate_limit", return_value=False)
-def test_check_token_success(
-    mock_is_rate_limited: MagicMock,
-    mock_get_data: MagicMock,
-    mock_revoke: MagicMock,
-    mock_generate: MagicMock,
-    mock_reset_limit: MagicMock,
-    app: Flask,
-):
-    """POST /forgot-password/validity validates the code and refreshes token."""
-
-    mock_get_data.return_value = {"email": "user@example.com", "code": "123456"}
-
-    with app.test_request_context(
-        "/forgot-password/validity",
-        method="POST",
-        json={"email": "user@example.com", "code": "123456", "token": "old-token"},
-    ):
-        response = ForgotPasswordCheckApi().post()
-
-    assert response == {"is_valid": True, "email": "user@example.com", "token": "new-token"}
-    mock_is_rate_limited.assert_called_once_with("user@example.com")
-    mock_get_data.assert_called_once_with("old-token")
-    mock_revoke.assert_called_once_with("old-token")
-    mock_generate.assert_called_once_with(
-        "user@example.com",
-        code="123456",
-        additional_data={"phase": "reset"},
-    )
-    mock_reset_limit.assert_called_once_with("user@example.com")
-
-
-@patch("controllers.web.forgot_password.hash_password", return_value=b"hashed-value")
-@patch("controllers.web.forgot_password.secrets.token_bytes", return_value=b"0123456789abcdef")
-@patch("controllers.web.forgot_password.Session")
-@patch("controllers.web.forgot_password.AccountService.revoke_reset_password_token")
-@patch("controllers.web.forgot_password.AccountService.get_reset_password_data")
-def test_reset_password_success(
-    mock_get_data: MagicMock,
-    mock_revoke_token: MagicMock,
-    mock_session: MagicMock,
-    mock_token_bytes: MagicMock,
-    mock_hash_password: MagicMock,
-    app: Flask,
-):
-    """POST /forgot-password/resets updates the stored password when token is valid."""
-
-    mock_get_data.return_value = {"email": "user@example.com", "phase": "reset"}
-    account = MagicMock()
-    session_ctx = MagicMock()
-    mock_session.return_value.__enter__.return_value = session_ctx
-    session_ctx.execute.return_value.scalar_one_or_none.return_value = account
-
-    with app.test_request_context(
-        "/forgot-password/resets",
-        method="POST",
-        json={
-            "token": "reset-token",
-            "new_password": "StrongPass123!",
-            "password_confirm": "StrongPass123!",
-        },
-    ):
-        response = ForgotPasswordResetApi().post()
-
-    assert response == {"result": "success"}
-    mock_get_data.assert_called_once_with("reset-token")
-    mock_revoke_token.assert_called_once_with("reset-token")
-    mock_token_bytes.assert_called_once_with(16)
-    mock_hash_password.assert_called_once_with("StrongPass123!", b"0123456789abcdef")
-    expected_password = base64.b64encode(b"hashed-value").decode()
-    assert account.password == expected_password
-    expected_salt = base64.b64encode(b"0123456789abcdef").decode()
-    assert account.password_salt == expected_salt
-    session_ctx.commit.assert_called_once()

api/tests/unit_tests/controllers/web/test_web_forgot_password.py

@@ -0,0 +1,226 @@
+import base64
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+import pytest
+from flask import Flask
+
+from controllers.web.forgot_password import (
+    ForgotPasswordCheckApi,
+    ForgotPasswordResetApi,
+    ForgotPasswordSendEmailApi,
+)
+
+
+@pytest.fixture
+def app():
+    flask_app = Flask(__name__)
+    flask_app.config["TESTING"] = True
+    return flask_app
+
+
+@pytest.fixture(autouse=True)
+def _patch_wraps():
+    wraps_features = SimpleNamespace(enable_email_password_login=True)
+    dify_settings = SimpleNamespace(ENTERPRISE_ENABLED=True, EDITION="CLOUD")
+    with (
+        patch("controllers.console.wraps.db") as mock_db,
+        patch("controllers.console.wraps.dify_config", dify_settings),
+        patch("controllers.console.wraps.FeatureService.get_system_features", return_value=wraps_features),
+    ):
+        mock_db.session.query.return_value.first.return_value = MagicMock()
+        yield
+
+
+class TestForgotPasswordSendEmailApi:
+    @patch("controllers.web.forgot_password.AccountService.send_reset_password_email")
+    @patch("controllers.web.forgot_password.AccountService.get_account_by_email_with_case_fallback")
+    @patch("controllers.web.forgot_password.AccountService.is_email_send_ip_limit", return_value=False)
+    @patch("controllers.web.forgot_password.extract_remote_ip", return_value="127.0.0.1")
+    @patch("controllers.web.forgot_password.Session")
+    def test_should_normalize_email_before_sending(
+        self,
+        mock_session_cls,
+        mock_extract_ip,
+        mock_rate_limit,
+        mock_get_account,
+        mock_send_mail,
+        app,
+    ):
+        mock_account = MagicMock()
+        mock_get_account.return_value = mock_account
+        mock_send_mail.return_value = "token-123"
+        mock_session = MagicMock()
+        mock_session_cls.return_value.__enter__.return_value = mock_session
+
+        with patch("controllers.web.forgot_password.db", SimpleNamespace(engine="engine")):
+            with app.test_request_context(
+                "/web/forgot-password",
+                method="POST",
+                json={"email": "User@Example.com", "language": "zh-Hans"},
+            ):
+                response = ForgotPasswordSendEmailApi().post()
+
+        assert response == {"result": "success", "data": "token-123"}
+        mock_get_account.assert_called_once_with("User@Example.com", session=mock_session)
+        mock_send_mail.assert_called_once_with(account=mock_account, email="user@example.com", language="zh-Hans")
+        mock_extract_ip.assert_called_once()
+        mock_rate_limit.assert_called_once_with("127.0.0.1")
+
+
+class TestForgotPasswordCheckApi:
+    @patch("controllers.web.forgot_password.AccountService.reset_forgot_password_error_rate_limit")
+    @patch("controllers.web.forgot_password.AccountService.generate_reset_password_token")
+    @patch("controllers.web.forgot_password.AccountService.revoke_reset_password_token")
+    @patch("controllers.web.forgot_password.AccountService.add_forgot_password_error_rate_limit")
+    @patch("controllers.web.forgot_password.AccountService.get_reset_password_data")
+    @patch("controllers.web.forgot_password.AccountService.is_forgot_password_error_rate_limit")
+    def test_should_normalize_email_for_validity_checks(
+        self,
+        mock_is_rate_limit,
+        mock_get_data,
+        mock_add_rate,
+        mock_revoke_token,
+        mock_generate_token,
+        mock_reset_rate,
+        app,
+    ):
+        mock_is_rate_limit.return_value = False
+        mock_get_data.return_value = {"email": "User@Example.com", "code": "1234"}
+        mock_generate_token.return_value = (None, "new-token")
+
+        with app.test_request_context(
+            "/web/forgot-password/validity",
+            method="POST",
+            json={"email": "User@Example.com", "code": "1234", "token": "token-123"},
+        ):
+            response = ForgotPasswordCheckApi().post()
+
+        assert response == {"is_valid": True, "email": "user@example.com", "token": "new-token"}
+        mock_is_rate_limit.assert_called_once_with("user@example.com")
+        mock_add_rate.assert_not_called()
+        mock_revoke_token.assert_called_once_with("token-123")
+        mock_generate_token.assert_called_once_with(
+            "User@Example.com",
+            code="1234",
+            additional_data={"phase": "reset"},
+        )
+        mock_reset_rate.assert_called_once_with("user@example.com")
+
+    @patch("controllers.web.forgot_password.AccountService.reset_forgot_password_error_rate_limit")
+    @patch("controllers.web.forgot_password.AccountService.generate_reset_password_token")
+    @patch("controllers.web.forgot_password.AccountService.revoke_reset_password_token")
+    @patch("controllers.web.forgot_password.AccountService.get_reset_password_data")
+    @patch("controllers.web.forgot_password.AccountService.is_forgot_password_error_rate_limit")
+    def test_should_preserve_token_email_case(
+        self,
+        mock_is_rate_limit,
+        mock_get_data,
+        mock_revoke_token,
+        mock_generate_token,
+        mock_reset_rate,
+        app,
+    ):
+        mock_is_rate_limit.return_value = False
+        mock_get_data.return_value = {"email": "MixedCase@Example.com", "code": "5678"}
+        mock_generate_token.return_value = (None, "fresh-token")
+
+        with app.test_request_context(
+            "/web/forgot-password/validity",
+            method="POST",
+            json={"email": "mixedcase@example.com", "code": "5678", "token": "token-upper"},
+        ):
+            response = ForgotPasswordCheckApi().post()
+
+        assert response == {"is_valid": True, "email": "mixedcase@example.com", "token": "fresh-token"}
+        mock_generate_token.assert_called_once_with(
+            "MixedCase@Example.com",
+            code="5678",
+            additional_data={"phase": "reset"},
+        )
+        mock_revoke_token.assert_called_once_with("token-upper")
+        mock_reset_rate.assert_called_once_with("mixedcase@example.com")
+
+
+class TestForgotPasswordResetApi:
+    @patch("controllers.web.forgot_password.ForgotPasswordResetApi._update_existing_account")
+    @patch("controllers.web.forgot_password.AccountService.get_account_by_email_with_case_fallback")
+    @patch("controllers.web.forgot_password.Session")
+    @patch("controllers.web.forgot_password.AccountService.revoke_reset_password_token")
+    @patch("controllers.web.forgot_password.AccountService.get_reset_password_data")
+    def test_should_fetch_account_with_fallback(
+        self,
+        mock_get_reset_data,
+        mock_revoke_token,
+        mock_session_cls,
+        mock_get_account,
+        mock_update_account,
+        app,
+    ):
+        mock_get_reset_data.return_value = {"phase": "reset", "email": "User@Example.com", "code": "1234"}
+        mock_account = MagicMock()
+        mock_get_account.return_value = mock_account
+        mock_session = MagicMock()
+        mock_session_cls.return_value.__enter__.return_value = mock_session
+
+        with patch("controllers.web.forgot_password.db", SimpleNamespace(engine="engine")):
+            with app.test_request_context(
+                "/web/forgot-password/resets",
+                method="POST",
+                json={
+                    "token": "token-123",
+                    "new_password": "ValidPass123!",
+                    "password_confirm": "ValidPass123!",
+                },
+            ):
+                response = ForgotPasswordResetApi().post()
+
+        assert response == {"result": "success"}
+        mock_get_account.assert_called_once_with("User@Example.com", session=mock_session)
+        mock_update_account.assert_called_once()
+        mock_revoke_token.assert_called_once_with("token-123")
+
+    @patch("controllers.web.forgot_password.hash_password", return_value=b"hashed-value")
+    @patch("controllers.web.forgot_password.secrets.token_bytes", return_value=b"0123456789abcdef")
+    @patch("controllers.web.forgot_password.Session")
+    @patch("controllers.web.forgot_password.AccountService.revoke_reset_password_token")
+    @patch("controllers.web.forgot_password.AccountService.get_reset_password_data")
+    @patch("controllers.web.forgot_password.AccountService.get_account_by_email_with_case_fallback")
+    def test_should_update_password_and_commit(
+        self,
+        mock_get_account,
+        mock_get_reset_data,
+        mock_revoke_token,
+        mock_session_cls,
+        mock_token_bytes,
+        mock_hash_password,
+        app,
+    ):
+        mock_get_reset_data.return_value = {"phase": "reset", "email": "user@example.com"}
+        account = MagicMock()
+        mock_get_account.return_value = account
+        mock_session = MagicMock()
+        mock_session_cls.return_value.__enter__.return_value = mock_session
+
+        with patch("controllers.web.forgot_password.db", SimpleNamespace(engine="engine")):
+            with app.test_request_context(
+                "/web/forgot-password/resets",
+                method="POST",
+                json={
+                    "token": "reset-token",
+                    "new_password": "StrongPass123!",
+                    "password_confirm": "StrongPass123!",
+                },
+            ):
+                response = ForgotPasswordResetApi().post()
+
+        assert response == {"result": "success"}
+        mock_get_reset_data.assert_called_once_with("reset-token")
+        mock_revoke_token.assert_called_once_with("reset-token")
+        mock_token_bytes.assert_called_once_with(16)
+        mock_hash_password.assert_called_once_with("StrongPass123!", b"0123456789abcdef")
+        expected_password = base64.b64encode(b"hashed-value").decode()
+        assert account.password == expected_password
+        expected_salt = base64.b64encode(b"0123456789abcdef").decode()
+        assert account.password_salt == expected_salt
+        mock_session.commit.assert_called_once()

api/tests/unit_tests/controllers/web/test_web_login.py

@@ -0,0 +1,91 @@
+import base64
+from types import SimpleNamespace
+from unittest.mock import MagicMock, patch
+
+import pytest
+from flask import Flask
+
+from controllers.web.login import EmailCodeLoginApi, EmailCodeLoginSendEmailApi
+
+
+def encode_code(code: str) -> str:
+    return base64.b64encode(code.encode("utf-8")).decode()
+
+
+@pytest.fixture
+def app():
+    flask_app = Flask(__name__)
+    flask_app.config["TESTING"] = True
+    return flask_app
+
+
+@pytest.fixture(autouse=True)
+def _patch_wraps():
+    wraps_features = SimpleNamespace(enable_email_password_login=True)
+    console_dify = SimpleNamespace(ENTERPRISE_ENABLED=True, EDITION="CLOUD")
+    web_dify = SimpleNamespace(ENTERPRISE_ENABLED=True)
+    with (
+        patch("controllers.console.wraps.db") as mock_db,
+        patch("controllers.console.wraps.dify_config", console_dify),
+        patch("controllers.console.wraps.FeatureService.get_system_features", return_value=wraps_features),
+        patch("controllers.web.login.dify_config", web_dify),
+    ):
+        mock_db.session.query.return_value.first.return_value = MagicMock()
+        yield
+
+
+class TestEmailCodeLoginSendEmailApi:
+    @patch("controllers.web.login.WebAppAuthService.send_email_code_login_email")
+    @patch("controllers.web.login.WebAppAuthService.get_user_through_email")
+    def test_should_fetch_account_with_original_email(
+        self,
+        mock_get_user,
+        mock_send_email,
+        app,
+    ):
+        mock_account = MagicMock()
+        mock_get_user.return_value = mock_account
+        mock_send_email.return_value = "token-123"
+
+        with app.test_request_context(
+            "/web/email-code-login",
+            method="POST",
+            json={"email": "User@Example.com", "language": "en-US"},
+        ):
+            response = EmailCodeLoginSendEmailApi().post()
+
+        assert response == {"result": "success", "data": "token-123"}
+        mock_get_user.assert_called_once_with("User@Example.com")
+        mock_send_email.assert_called_once_with(account=mock_account, language="en-US")
+
+
+class TestEmailCodeLoginApi:
+    @patch("controllers.web.login.AccountService.reset_login_error_rate_limit")
+    @patch("controllers.web.login.WebAppAuthService.login", return_value="new-access-token")
+    @patch("controllers.web.login.WebAppAuthService.get_user_through_email")
+    @patch("controllers.web.login.WebAppAuthService.revoke_email_code_login_token")
+    @patch("controllers.web.login.WebAppAuthService.get_email_code_login_data")
+    def test_should_normalize_email_before_validating(
+        self,
+        mock_get_token_data,
+        mock_revoke_token,
+        mock_get_user,
+        mock_login,
+        mock_reset_login_rate,
+        app,
+    ):
+        mock_get_token_data.return_value = {"email": "User@Example.com", "code": "123456"}
+        mock_get_user.return_value = MagicMock()
+
+        with app.test_request_context(
+            "/web/email-code-login/validity",
+            method="POST",
+            json={"email": "User@Example.com", "code": encode_code("123456"), "token": "token-123"},
+        ):
+            response = EmailCodeLoginApi().post()
+
+        assert response.get_json() == {"result": "success", "data": {"access_token": "new-access-token"}}
+        mock_get_user.assert_called_once_with("User@Example.com")
+        mock_revoke_token.assert_called_once_with("token-123")
+        mock_login.assert_called_once()
+        mock_reset_login_rate.assert_called_once_with("user@example.com")

api/tests/unit_tests/core/app/apps/advanced_chat/test_generate_task_pipeline_answer_node.py

@@ -1,390 +0,0 @@
-"""
-Tests for AdvancedChatAppGenerateTaskPipeline._handle_node_succeeded_event method,
-specifically testing the ANSWER node message_replace logic.
-"""
-
-from datetime import datetime
-from types import SimpleNamespace
-from unittest.mock import MagicMock, Mock, patch
-
-import pytest
-
-from core.app.entities.app_invoke_entities import AdvancedChatAppGenerateEntity
-from core.app.entities.queue_entities import QueueNodeSucceededEvent
-from core.workflow.enums import NodeType
-from models import EndUser
-from models.model import AppMode
-
-
-class TestAnswerNodeMessageReplace:
-    """Test cases for ANSWER node message_replace event logic."""
-
-    @pytest.fixture
-    def mock_application_generate_entity(self):
-        """Create a mock application generate entity."""
-        entity = Mock(spec=AdvancedChatAppGenerateEntity)
-        entity.task_id = "test-task-id"
-        entity.app_id = "test-app-id"
-        entity.workflow_run_id = "test-workflow-run-id"
-        # minimal app_config used by pipeline internals
-        entity.app_config = SimpleNamespace(
-            tenant_id="test-tenant-id",
-            app_id="test-app-id",
-            app_mode=AppMode.ADVANCED_CHAT,
-            app_model_config_dict={},
-            additional_features=None,
-            sensitive_word_avoidance=None,
-        )
-        entity.query = "test query"
-        entity.files = []
-        entity.extras = {}
-        entity.trace_manager = None
-        entity.inputs = {}
-        entity.invoke_from = "debugger"
-        return entity
-
-    @pytest.fixture
-    def mock_workflow(self):
-        """Create a mock workflow."""
-        workflow = Mock()
-        workflow.id = "test-workflow-id"
-        workflow.features_dict = {}
-        return workflow
-
-    @pytest.fixture
-    def mock_queue_manager(self):
-        """Create a mock queue manager."""
-        manager = Mock()
-        manager.listen.return_value = []
-        manager.graph_runtime_state = None
-        return manager
-
-    @pytest.fixture
-    def mock_conversation(self):
-        """Create a mock conversation."""
-        conversation = Mock()
-        conversation.id = "test-conversation-id"
-        conversation.mode = "advanced_chat"
-        return conversation
-
-    @pytest.fixture
-    def mock_message(self):
-        """Create a mock message."""
-        message = Mock()
-        message.id = "test-message-id"
-        message.query = "test query"
-        message.created_at = Mock()
-        message.created_at.timestamp.return_value = 1234567890
-        return message
-
-    @pytest.fixture
-    def mock_user(self):
-        """Create a mock end user."""
-        user = MagicMock(spec=EndUser)
-        user.id = "test-user-id"
-        user.session_id = "test-session-id"
-        return user
-
-    @pytest.fixture
-    def mock_draft_var_saver_factory(self):
-        """Create a mock draft variable saver factory."""
-        return Mock()
-
-    @pytest.fixture
-    def pipeline(
-        self,
-        mock_application_generate_entity,
-        mock_workflow,
-        mock_queue_manager,
-        mock_conversation,
-        mock_message,
-        mock_user,
-        mock_draft_var_saver_factory,
-    ):
-        """Create an AdvancedChatAppGenerateTaskPipeline instance with mocked dependencies."""
-        from core.app.apps.advanced_chat.generate_task_pipeline import AdvancedChatAppGenerateTaskPipeline
-
-        with patch("core.app.apps.advanced_chat.generate_task_pipeline.db"):
-            pipeline = AdvancedChatAppGenerateTaskPipeline(
-                application_generate_entity=mock_application_generate_entity,
-                workflow=mock_workflow,
-                queue_manager=mock_queue_manager,
-                conversation=mock_conversation,
-                message=mock_message,
-                user=mock_user,
-                stream=True,
-                dialogue_count=1,
-                draft_var_saver_factory=mock_draft_var_saver_factory,
-            )
-            # Initialize workflow run id to avoid validation errors
-            pipeline._workflow_run_id = "test-workflow-run-id"
-            # Mock the message cycle manager methods we need to track
-            pipeline._message_cycle_manager.message_replace_to_stream_response = Mock()
-            return pipeline
-
-    def test_answer_node_with_different_output_sends_message_replace(self, pipeline, mock_application_generate_entity):
-        """
-        Test that when an ANSWER node's final output differs from accumulated answer,
-        a message_replace event is sent.
-        """
-        # Arrange: Set initial accumulated answer
-        pipeline._task_state.answer = "initial answer"
-
-        # Create ANSWER node succeeded event with different final output
-        event = QueueNodeSucceededEvent(
-            node_execution_id="test-node-execution-id",
-            node_id="test-answer-node",
-            node_type=NodeType.ANSWER,
-            start_at=datetime.now(),
-            outputs={"answer": "updated final answer"},
-        )
-
-        # Mock the workflow response converter to avoid extra processing
-        pipeline._workflow_response_converter.workflow_node_finish_to_stream_response = Mock(return_value=None)
-        pipeline._save_output_for_event = Mock()
-
-        # Act
-        responses = list(pipeline._handle_node_succeeded_event(event))
-
-        # Assert
-        assert pipeline._task_state.answer == "updated final answer"
-        # Verify message_replace was called
-        pipeline._message_cycle_manager.message_replace_to_stream_response.assert_called_once_with(
-            answer="updated final answer", reason="variable_update"
-        )
-
-    def test_answer_node_with_same_output_does_not_send_message_replace(self, pipeline):
-        """
-        Test that when an ANSWER node's final output is the same as accumulated answer,
-        no message_replace event is sent.
-        """
-        # Arrange: Set initial accumulated answer
-        pipeline._task_state.answer = "same answer"
-
-        # Create ANSWER node succeeded event with same output
-        event = QueueNodeSucceededEvent(
-            node_execution_id="test-node-execution-id",
-            node_id="test-answer-node",
-            node_type=NodeType.ANSWER,
-            start_at=datetime.now(),
-            outputs={"answer": "same answer"},
-        )
-
-        # Mock the workflow response converter
-        pipeline._workflow_response_converter.workflow_node_finish_to_stream_response = Mock(return_value=None)
-        pipeline._save_output_for_event = Mock()
-
-        # Act
-        list(pipeline._handle_node_succeeded_event(event))
-
-        # Assert: answer should remain unchanged
-        assert pipeline._task_state.answer == "same answer"
-        # Verify message_replace was NOT called
-        pipeline._message_cycle_manager.message_replace_to_stream_response.assert_not_called()
-
-    def test_answer_node_with_none_output_does_not_send_message_replace(self, pipeline):
-        """
-        Test that when an ANSWER node's output is None or missing 'answer' key,
-        no message_replace event is sent.
-        """
-        # Arrange: Set initial accumulated answer
-        pipeline._task_state.answer = "existing answer"
-
-        # Create ANSWER node succeeded event with None output
-        event = QueueNodeSucceededEvent(
-            node_execution_id="test-node-execution-id",
-            node_id="test-answer-node",
-            node_type=NodeType.ANSWER,
-            start_at=datetime.now(),
-            outputs={"answer": None},
-        )
-
-        # Mock the workflow response converter
-        pipeline._workflow_response_converter.workflow_node_finish_to_stream_response = Mock(return_value=None)
-        pipeline._save_output_for_event = Mock()
-
-        # Act
-        list(pipeline._handle_node_succeeded_event(event))
-
-        # Assert: answer should remain unchanged
-        assert pipeline._task_state.answer == "existing answer"
-        # Verify message_replace was NOT called
-        pipeline._message_cycle_manager.message_replace_to_stream_response.assert_not_called()
-
-    def test_answer_node_with_empty_outputs_does_not_send_message_replace(self, pipeline):
-        """
-        Test that when an ANSWER node has empty outputs dict,
-        no message_replace event is sent.
-        """
-        # Arrange: Set initial accumulated answer
-        pipeline._task_state.answer = "existing answer"
-
-        # Create ANSWER node succeeded event with empty outputs
-        event = QueueNodeSucceededEvent(
-            node_execution_id="test-node-execution-id",
-            node_id="test-answer-node",
-            node_type=NodeType.ANSWER,
-            start_at=datetime.now(),
-            outputs={},
-        )
-
-        # Mock the workflow response converter
-        pipeline._workflow_response_converter.workflow_node_finish_to_stream_response = Mock(return_value=None)
-        pipeline._save_output_for_event = Mock()
-
-        # Act
-        list(pipeline._handle_node_succeeded_event(event))
-
-        # Assert: answer should remain unchanged
-        assert pipeline._task_state.answer == "existing answer"
-        # Verify message_replace was NOT called
-        pipeline._message_cycle_manager.message_replace_to_stream_response.assert_not_called()
-
-    def test_answer_node_with_no_answer_key_in_outputs(self, pipeline):
-        """
-        Test that when an ANSWER node's outputs don't contain 'answer' key,
-        no message_replace event is sent.
-        """
-        # Arrange: Set initial accumulated answer
-        pipeline._task_state.answer = "existing answer"
-
-        # Create ANSWER node succeeded event without 'answer' key in outputs
-        event = QueueNodeSucceededEvent(
-            node_execution_id="test-node-execution-id",
-            node_id="test-answer-node",
-            node_type=NodeType.ANSWER,
-            start_at=datetime.now(),
-            outputs={"other_key": "some value"},
-        )
-
-        # Mock the workflow response converter
-        pipeline._workflow_response_converter.workflow_node_finish_to_stream_response = Mock(return_value=None)
-        pipeline._save_output_for_event = Mock()
-
-        # Act
-        list(pipeline._handle_node_succeeded_event(event))
-
-        # Assert: answer should remain unchanged
-        assert pipeline._task_state.answer == "existing answer"
-        # Verify message_replace was NOT called
-        pipeline._message_cycle_manager.message_replace_to_stream_response.assert_not_called()
-
-    def test_non_answer_node_does_not_send_message_replace(self, pipeline):
-        """
-        Test that non-ANSWER nodes (e.g., LLM, END) don't trigger message_replace events.
-        """
-        # Arrange: Set initial accumulated answer
-        pipeline._task_state.answer = "existing answer"
-
-        # Test with LLM node
-        llm_event = QueueNodeSucceededEvent(
-            node_execution_id="test-llm-execution-id",
-            node_id="test-llm-node",
-            node_type=NodeType.LLM,
-            start_at=datetime.now(),
-            outputs={"answer": "different answer"},
-        )
-
-        # Mock the workflow response converter
-        pipeline._workflow_response_converter.workflow_node_finish_to_stream_response = Mock(return_value=None)
-        pipeline._save_output_for_event = Mock()
-
-        # Act
-        list(pipeline._handle_node_succeeded_event(llm_event))
-
-        # Assert: answer should remain unchanged
-        assert pipeline._task_state.answer == "existing answer"
-        # Verify message_replace was NOT called
-        pipeline._message_cycle_manager.message_replace_to_stream_response.assert_not_called()
-
-    def test_end_node_does_not_send_message_replace(self, pipeline):
-        """
-        Test that END nodes don't trigger message_replace events even with 'answer' output.
-        """
-        # Arrange: Set initial accumulated answer
-        pipeline._task_state.answer = "existing answer"
-
-        # Create END node succeeded event with answer output
-        event = QueueNodeSucceededEvent(
-            node_execution_id="test-end-execution-id",
-            node_id="test-end-node",
-            node_type=NodeType.END,
-            start_at=datetime.now(),
-            outputs={"answer": "different answer"},
-        )
-
-        # Mock the workflow response converter
-        pipeline._workflow_response_converter.workflow_node_finish_to_stream_response = Mock(return_value=None)
-        pipeline._save_output_for_event = Mock()
-
-        # Act
-        list(pipeline._handle_node_succeeded_event(event))
-
-        # Assert: answer should remain unchanged
-        assert pipeline._task_state.answer == "existing answer"
-        # Verify message_replace was NOT called
-        pipeline._message_cycle_manager.message_replace_to_stream_response.assert_not_called()
-
-    def test_answer_node_with_numeric_output_converts_to_string(self, pipeline):
-        """
-        Test that when an ANSWER node's final output is numeric,
-        it gets converted to string properly.
-        """
-        # Arrange: Set initial accumulated answer
-        pipeline._task_state.answer = "text answer"
-
-        # Create ANSWER node succeeded event with numeric output
-        event = QueueNodeSucceededEvent(
-            node_execution_id="test-node-execution-id",
-            node_id="test-answer-node",
-            node_type=NodeType.ANSWER,
-            start_at=datetime.now(),
-            outputs={"answer": 12345},
-        )
-
-        # Mock the workflow response converter
-        pipeline._workflow_response_converter.workflow_node_finish_to_stream_response = Mock(return_value=None)
-        pipeline._save_output_for_event = Mock()
-
-        # Act
-        list(pipeline._handle_node_succeeded_event(event))
-
-        # Assert: answer should be converted to string
-        assert pipeline._task_state.answer == "12345"
-        # Verify message_replace was called with string
-        pipeline._message_cycle_manager.message_replace_to_stream_response.assert_called_once_with(
-            answer="12345", reason="variable_update"
-        )
-
-    def test_answer_node_files_are_recorded(self, pipeline):
-        """
-        Test that ANSWER nodes properly record files from outputs.
-        """
-        # Arrange
-        pipeline._task_state.answer = "existing answer"
-
-        # Create ANSWER node succeeded event with files
-        event = QueueNodeSucceededEvent(
-            node_execution_id="test-node-execution-id",
-            node_id="test-answer-node",
-            node_type=NodeType.ANSWER,
-            start_at=datetime.now(),
-            outputs={
-                "answer": "same answer",
-                "files": [
-                    {"type": "image", "transfer_method": "remote_url", "remote_url": "http://example.com/img.png"}
-                ],
-            },
-        )
-
-        # Mock the workflow response converter
-        pipeline._workflow_response_converter.fetch_files_from_node_outputs = Mock(return_value=event.outputs["files"])
-        pipeline._workflow_response_converter.workflow_node_finish_to_stream_response = Mock(return_value=None)
-        pipeline._save_output_for_event = Mock()
-
-        # Act
-        list(pipeline._handle_node_succeeded_event(event))
-
-        # Assert: files should be recorded
-        assert len(pipeline._recorded_files) == 1
-        assert pipeline._recorded_files[0] == event.outputs["files"][0]