version to 0.5.11 (#3038)

Co-authored-by: crazywoola <427733928@qq.com>

CONTRIBUTING.md

@@ -155,4 +155,4 @@ And that's it! Once your PR is merged, you will be featured as a contributor in
 
 ## Getting Help
 
-If you ever get stuck or got a burning question while contributing, simply shoot your queries our way via the related GitHub issue, or hop onto our [Discord](https://discord.gg/AhzKf7dNgk) for a quick chat. 
+If you ever get stuck or got a burning question while contributing, simply shoot your queries our way via the related GitHub issue, or hop onto our [Discord](https://discord.gg/8Tpq4AcN9c) for a quick chat. 

CONTRIBUTING_CN.md

@@ -152,4 +152,4 @@ Dify的后端使用Python编写，使用[Flask](https://flask.palletsprojects.co
 
 ## 获取帮助
 
-如果你在贡献过程中遇到困难或者有任何问题，可以通过相关的 GitHub 问题提出你的疑问，或者加入我们的 [Discord](https://discord.gg/AhzKf7dNgk) 进行快速交流。
+如果你在贡献过程中遇到困难或者有任何问题，可以通过相关的 GitHub 问题提出你的疑问，或者加入我们的 [Discord](https://discord.gg/8Tpq4AcN9c) 进行快速交流。

README.md

@@ -100,10 +100,12 @@ docker compose up -d
 
 After running, you can access the Dify dashboard in your browser at [http://localhost/install](http://localhost/install) and start the initialization installation process.
 
-### Helm Chart
+#### Deploy with Helm Chart
 
-Big thanks to @BorisPolonsky for providing us with a [Helm Chart](https://helm.sh/) version, which allows Dify to be deployed on Kubernetes.
-You can go to https://github.com/BorisPolonsky/dify-helm for deployment information.
+[Helm Chart](https://helm.sh/) version, which allows Dify to be deployed on Kubernetes.
+
+- [Helm Chart by @LeoQuote](https://github.com/douban/charts/tree/master/charts/dify)
+- [Helm Chart by @BorisPolonsky](https://github.com/BorisPolonsky/dify-helm)
 
 ### Configuration
 
@@ -120,6 +122,7 @@ For those who'd like to contribute code, see our [Contribution Guide](https://gi
 
 At the same time, please consider supporting Dify by sharing it on social media and at events and conferences.
 
+
 ### Contributors
 
 <a href="https://github.com/langgenius/dify/graphs/contributors">
@@ -128,7 +131,7 @@ At the same time, please consider supporting Dify by sharing it on social media
 
 ### Translations
 
-We are looking for contributors to help with translating Dify to languages other than Mandarin or English. If you are interested in helping, please see the [i18n README](https://github.com/langgenius/dify/blob/main/web/i18n/README_EN.md) for more information, and leave us a comment in the `global-users` channel of our [Discord Community Server](https://discord.gg/AhzKf7dNgk).
+We are looking for contributors to help with translating Dify to languages other than Mandarin or English. If you are interested in helping, please see the [i18n README](https://github.com/langgenius/dify/blob/main/web/i18n/README.md) for more information, and leave us a comment in the `global-users` channel of our [Discord Community Server](https://discord.gg/8Tpq4AcN9c).
 
 ## Community & Support
 

README_CN.md

@@ -94,10 +94,12 @@ docker compose up -d
 
 运行后，可以在浏览器上访问 [http://localhost/install](http://localhost/install) 进入 Dify 控制台并开始初始化安装操作。
 
-### Helm Chart
+#### 使用 Helm Chart 部署
 
-非常感谢 @BorisPolonsky 为我们提供了一个 [Helm Chart](https://helm.sh/) 版本，可以在 Kubernetes 上部署 Dify。
-您可以前往 https://github.com/BorisPolonsky/dify-helm 来获取部署信息。
+使用 [Helm Chart](https://helm.sh/) 版本，可以在 Kubernetes 上部署 Dify。
+
+- [Helm Chart by @LeoQuote](https://github.com/douban/charts/tree/master/charts/dify)
+- [Helm Chart by @BorisPolonsky](https://github.com/BorisPolonsky/dify-helm)
 
 ### 配置
 

api/.env.example

@@ -39,7 +39,7 @@ DB_DATABASE=dify
 
 # Storage configuration
 # use for store upload files, private keys...
-# storage type: local, s3
+# storage type: local, s3, azure-blob
 STORAGE_TYPE=local
 STORAGE_LOCAL_PATH=storage
 S3_ENDPOINT=https://your-bucket-name.storage.s3.clooudflare.com
@@ -47,6 +47,11 @@ S3_BUCKET_NAME=your-bucket-name
 S3_ACCESS_KEY=your-access-key
 S3_SECRET_KEY=your-secret-key
 S3_REGION=your-region
+# Azure Blob Storage configuration
+AZURE_BLOB_ACCOUNT_NAME=your-account-name
+AZURE_BLOB_ACCOUNT_KEY=your-account-key
+AZURE_BLOB_CONTAINER_NAME=yout-container-name
+AZURE_BLOB_ACCOUNT_URL=https://<your_account_name>.blob.core.windows.net
 
 # CORS configuration
 WEB_API_CORS_ALLOW_ORIGINS=http://127.0.0.1:3000,*

api/commands.py

@@ -109,19 +109,20 @@ def reset_encrypt_key_pair():
         click.echo(click.style('Sorry, only support SELF_HOSTED mode.', fg='red'))
         return
 
-    tenant = db.session.query(Tenant).first()
-    if not tenant:
-        click.echo(click.style('Sorry, no workspace found. Please enter /install to initialize.', fg='red'))
-        return
+    tenants = db.session.query(Tenant).all()
+    for tenant in tenants:
+        if not tenant:
+            click.echo(click.style('Sorry, no workspace found. Please enter /install to initialize.', fg='red'))
+            return
 
-    tenant.encrypt_public_key = generate_key_pair(tenant.id)
+        tenant.encrypt_public_key = generate_key_pair(tenant.id)
 
-    db.session.query(Provider).filter(Provider.provider_type == 'custom').delete()
-    db.session.query(ProviderModel).delete()
-    db.session.commit()
+        db.session.query(Provider).filter(Provider.provider_type == 'custom', Provider.tenant_id == tenant.id).delete()
+        db.session.query(ProviderModel).filter(ProviderModel.tenant_id == tenant.id).delete()
+        db.session.commit()
 
-    click.echo(click.style('Congratulations! '
-                           'the asymmetric key pair of workspace {} has been reset.'.format(tenant.id), fg='green'))
+        click.echo(click.style('Congratulations! '
+                               'the asymmetric key pair of workspace {} has been reset.'.format(tenant.id), fg='green'))
 
 
 @click.command('vdb-migrate', help='migrate vector db.')
@@ -254,7 +255,7 @@ def migrate_knowledge_vector_database():
         for dataset in datasets:
             total_count = total_count + 1
             click.echo(f'Processing the {total_count} dataset {dataset.id}. '
-                       + f'{create_count} created, ${skipped_count} skipped.')
+                       + f'{create_count} created, {skipped_count} skipped.')
             try:
                 click.echo('Create dataset vdb index: {}'.format(dataset.id))
                 if dataset.index_struct_dict:

api/config.py

@@ -22,6 +22,7 @@ DEFAULTS = {
     'SERVICE_API_URL': 'https://api.dify.ai',
     'APP_WEB_URL': 'https://udify.app',
     'FILES_URL': '',
+    'S3_ADDRESS_STYLE': 'auto',
     'STORAGE_TYPE': 'local',
     'STORAGE_LOCAL_PATH': 'storage',
     'CHECK_UPDATE_URL': 'https://updates.dify.ai',
@@ -59,7 +60,8 @@ DEFAULTS = {
     'CAN_REPLACE_LOGO': 'False',
     'ETL_TYPE': 'dify',
     'KEYWORD_STORE': 'jieba',
-    'BATCH_UPLOAD_LIMIT': 20
+    'BATCH_UPLOAD_LIMIT': 20,
+    'TOOL_ICON_CACHE_MAX_AGE': 3600,
 }
 
 
@@ -90,7 +92,7 @@ class Config:
         # ------------------------
         # General Configurations.
         # ------------------------
-        self.CURRENT_VERSION = "0.5.10"
+        self.CURRENT_VERSION = "0.5.11"
         self.COMMIT_SHA = get_env('COMMIT_SHA')
         self.EDITION = "SELF_HOSTED"
         self.DEPLOY_ENV = get_env('DEPLOY_ENV')
@@ -180,6 +182,11 @@ class Config:
         self.S3_ACCESS_KEY = get_env('S3_ACCESS_KEY')
         self.S3_SECRET_KEY = get_env('S3_SECRET_KEY')
         self.S3_REGION = get_env('S3_REGION')
+        self.S3_ADDRESS_STYLE = get_env('S3_ADDRESS_STYLE')
+        self.AZURE_BLOB_ACCOUNT_NAME = get_env('AZURE_BLOB_ACCOUNT_NAME')
+        self.AZURE_BLOB_ACCOUNT_KEY = get_env('AZURE_BLOB_ACCOUNT_KEY')
+        self.AZURE_BLOB_CONTAINER_NAME = get_env('AZURE_BLOB_CONTAINER_NAME')
+        self.AZURE_BLOB_ACCOUNT_URL = get_env('AZURE_BLOB_ACCOUNT_URL')
 
         # ------------------------
         # Vector Store Configurations.
@@ -294,6 +301,7 @@ class Config:
         self.BATCH_UPLOAD_LIMIT = get_env('BATCH_UPLOAD_LIMIT')
 
         self.API_COMPRESSION_ENABLED = get_bool_env('API_COMPRESSION_ENABLED')
+        self.TOOL_ICON_CACHE_MAX_AGE = get_env('TOOL_ICON_CACHE_MAX_AGE')
 
 
 class CloudEditionConfig(Config):

api/controllers/console/workspace/tool_providers.py

@@ -1,6 +1,6 @@
 import io
 
-from flask import send_file
+from flask import current_app, send_file
 from flask_login import current_user
 from flask_restful import Resource, reqparse
 from werkzeug.exceptions import Forbidden
@@ -80,7 +80,8 @@ class ToolBuiltinProviderIconApi(Resource):
     @setup_required
     def get(self, provider):
         icon_bytes, minetype = ToolManageService.get_builtin_tool_provider_icon(provider)
-        return send_file(io.BytesIO(icon_bytes), mimetype=minetype)
+        icon_cache_max_age = int(current_app.config.get('TOOL_ICON_CACHE_MAX_AGE'))
+        return send_file(io.BytesIO(icon_bytes), mimetype=minetype, max_age=icon_cache_max_age)
 
 class ToolModelProviderIconApi(Resource):
     @setup_required

api/controllers/service_api/dataset/segment.py

@@ -197,11 +197,11 @@ class DatasetSegmentApi(DatasetApiResource):
 
         # validate args
         parser = reqparse.RequestParser()
-        parser.add_argument('segments', type=dict, required=False, nullable=True, location='json')
+        parser.add_argument('segment', type=dict, required=False, nullable=True, location='json')
         args = parser.parse_args()
 
-        SegmentService.segment_create_args_validate(args, document)
-        segment = SegmentService.update_segment(args, segment, document, dataset)
+        SegmentService.segment_create_args_validate(args['segment'], document)
+        segment = SegmentService.update_segment(args['segment'], segment, document, dataset)
         return {
             'data': marshal(segment, segment_fields),
             'doc_form': document.doc_form

api/core/extension/extensible.py

@@ -1,5 +1,5 @@
 import enum
-import importlib.util
+import importlib
 import json
 import logging
 import os

api/core/features/assistant_base_runner.py

@@ -145,7 +145,7 @@ class BaseAssistantApplicationRunner(AppRunner):
                 result += f"result link: {response.message}. please tell user to check it."
             elif response.type == ToolInvokeMessage.MessageType.IMAGE_LINK or \
                  response.type == ToolInvokeMessage.MessageType.IMAGE:
-                result += "image has been created and sent to user already, you should tell user to check it now."
+                result += "image has been created and sent to user already, you do not need to create it, just tell the user to check it now."
             else:
                 result += f"tool response: {response.message}."
 
@@ -566,7 +566,11 @@ class BaseAssistantApplicationRunner(AppRunner):
                         tools = tools.split(';')
                         tool_calls: list[AssistantPromptMessage.ToolCall] = []
                         tool_call_response: list[ToolPromptMessage] = []
-                        tool_inputs = json.loads(agent_thought.tool_input)
+                        try:
+                            tool_inputs = json.loads(agent_thought.tool_input)
+                        except Exception as e:
+                            logging.warning("tool execution error: {}, tool_input: {}.".format(str(e), agent_thought.tool_input))
+                            tool_inputs = { agent_thought.tool: agent_thought.tool_input }
                         for tool in tools:
                             # generate a uuid for tool call
                             tool_call_id = str(uuid.uuid4())
@@ -599,4 +603,4 @@ class BaseAssistantApplicationRunner(AppRunner):
 
         db.session.close()
 
-        return result
+        return result

api/core/features/assistant_cot_runner.py

@@ -182,7 +182,7 @@ class AssistantCotApplicationRunner(BaseAssistantApplicationRunner):
                             delta=LLMResultChunkDelta(
                                 index=0,
                                 message=AssistantPromptMessage(
-                                    content=json.dumps(chunk)
+                                    content=json.dumps(chunk, ensure_ascii=False) # if ensure_ascii=True, the text in webui maybe garbled text
                                 ),
                                 usage=None
                             )
@@ -667,4 +667,4 @@ class AssistantCotApplicationRunner(BaseAssistantApplicationRunner):
         try:
             return json.dumps(tools, ensure_ascii=False)
         except json.JSONDecodeError:
-            return json.dumps(tools)
+            return json.dumps(tools)

api/core/indexing_runner.py

@@ -1,3 +1,4 @@
+import concurrent.futures
 import datetime
 import json
 import logging
@@ -650,17 +651,44 @@ class IndexingRunner:
         # chunk nodes by chunk size
         indexing_start_at = time.perf_counter()
         tokens = 0
-        chunk_size = 100
+        chunk_size = 10
 
         embedding_model_type_instance = None
         if embedding_model_instance:
             embedding_model_type_instance = embedding_model_instance.model_type_instance
             embedding_model_type_instance = cast(TextEmbeddingModel, embedding_model_type_instance)
+        with concurrent.futures.ThreadPoolExecutor(max_workers=10) as executor:
+            futures = []
+            for i in range(0, len(documents), chunk_size):
+                chunk_documents = documents[i:i + chunk_size]
+                futures.append(executor.submit(self._process_chunk, current_app._get_current_object(), index_processor,
+                                               chunk_documents, dataset,
+                                               dataset_document, embedding_model_instance,
+                                               embedding_model_type_instance))
 
-        for i in range(0, len(documents), chunk_size):
+            for future in futures:
+                tokens += future.result()
+
+        indexing_end_at = time.perf_counter()
+
+        # update document status to completed
+        self._update_document_index_status(
+            document_id=dataset_document.id,
+            after_indexing_status="completed",
+            extra_update_params={
+                DatasetDocument.tokens: tokens,
+                DatasetDocument.completed_at: datetime.datetime.utcnow(),
+                DatasetDocument.indexing_latency: indexing_end_at - indexing_start_at,
+            }
+        )
+
+    def _process_chunk(self, flask_app, index_processor, chunk_documents, dataset, dataset_document,
+                       embedding_model_instance, embedding_model_type_instance):
+        with flask_app.app_context():
             # check document is paused
             self._check_document_paused_status(dataset_document.id)
-            chunk_documents = documents[i:i + chunk_size]
+
+            tokens = 0
             if dataset.indexing_technique == 'high_quality' or embedding_model_type_instance:
                 tokens += sum(
                     embedding_model_type_instance.get_num_tokens(
@@ -670,9 +698,9 @@ class IndexingRunner:
                     )
                     for document in chunk_documents
                 )
+
             # load index
             index_processor.load(dataset, chunk_documents)
-            db.session.add(dataset)
 
             document_ids = [document.metadata['doc_id'] for document in chunk_documents]
             db.session.query(DocumentSegment).filter(
@@ -687,18 +715,7 @@ class IndexingRunner:
 
             db.session.commit()
 
-        indexing_end_at = time.perf_counter()
-
-        # update document status to completed
-        self._update_document_index_status(
-            document_id=dataset_document.id,
-            after_indexing_status="completed",
-            extra_update_params={
-                DatasetDocument.tokens: tokens,
-                DatasetDocument.completed_at: datetime.datetime.utcnow(),
-                DatasetDocument.indexing_latency: indexing_end_at - indexing_start_at,
-            }
-        )
+            return tokens
 
     def _check_document_paused_status(self, document_id: str):
         indexing_cache_key = 'document_{}_is_paused'.format(document_id)

api/core/model_runtime/docs/en_US/schema.md

@@ -147,7 +147,7 @@
 
 - `input` (float) Input price, i.e., Prompt price
 - `output` (float) Output price, i.e., returned content price
-- `unit` (float) Pricing unit, e.g., per 100K price is `0.000001`
+- `unit` (float) Pricing unit, e.g., if the price is meausred in 1M tokens, the corresponding token amount for the unit price is `0.000001`.
 - `currency` (string) Currency unit
 
 ### ProviderCredentialSchema

api/core/model_runtime/docs/zh_Hans/schema.md

@@ -149,7 +149,7 @@
 
 - `input` (float) 输入单价，即 Prompt 单价
 - `output` (float) 输出单价，即返回内容单价
-- `unit` (float) 价格单位，如：每 100K 的单价为 `0.000001`
+- `unit` (float) 价格单位，如以 1M tokens 计价，则单价对应的单位 token 数为 `0.000001`
 - `currency` (string) 货币单位
 
 ### ProviderCredentialSchema

api/core/model_runtime/entities/defaults.py

@@ -73,8 +73,8 @@ PARAMETER_RULE_TEMPLATE: dict[DefaultParameterName, dict] = {
         },
         'type': 'int',
         'help': {
-            'en_US': 'The maximum number of tokens to generate. Requests can use up to 2048 tokens shared between prompt and completion.',
-            'zh_Hans': '要生成的标记的最大数量。请求可以使用最多2048个标记，这些标记在提示和完成之间共享。',
+            'en_US': 'Specifies the upper limit on the length of generated results. If the generated results are truncated, you can increase this parameter.',
+            'zh_Hans': '指定生成结果长度的上限。如果生成结果截断，可以调大该参数。',
         },
         'required': False,
         'default': 64,

api/core/model_runtime/model_providers/__base/model_provider.py

@@ -1,4 +1,3 @@
-import importlib
 import os
 from abc import ABC, abstractmethod
 
@@ -7,6 +6,7 @@ import yaml
 from core.model_runtime.entities.model_entities import AIModelEntity, ModelType
 from core.model_runtime.entities.provider_entities import ProviderEntity
 from core.model_runtime.model_providers.__base.ai_model import AIModel
+from core.utils.module_import_helper import get_subclasses_from_module, import_module_from_source
 
 
 class ModelProvider(ABC):
@@ -104,17 +104,10 @@ class ModelProvider(ABC):
 
         # Dynamic loading {model_type_name}.py file and find the subclass of AIModel
         parent_module = '.'.join(self.__class__.__module__.split('.')[:-1])
-        spec = importlib.util.spec_from_file_location(f"{parent_module}.{model_type_name}.{model_type_name}", model_type_py_path)
-        mod = importlib.util.module_from_spec(spec)
-        spec.loader.exec_module(mod)
-
-        model_class = None
-        for name, obj in vars(mod).items():
-            if (isinstance(obj, type) and issubclass(obj, AIModel) and not obj.__abstractmethods__
-                    and obj != AIModel and obj.__module__ == mod.__name__):
-                model_class = obj
-                break
-
+        mod = import_module_from_source(
+            f'{parent_module}.{model_type_name}.{model_type_name}', model_type_py_path)
+        model_class = next(filter(lambda x: x.__module__ == mod.__name__ and not x.__abstractmethods__,
+                                  get_subclasses_from_module(mod, AIModel)), None)
         if not model_class:
             raise Exception(f'Missing AIModel Class for model type {model_type} in {model_type_py_path}')
 

api/core/model_runtime/model_providers/_position.yaml

@@ -11,6 +11,8 @@
 - groq
 - replicate
 - huggingface_hub
+- xinference
+- triton_inference_server
 - zhipuai
 - baichuan
 - spark
@@ -20,7 +22,6 @@
 - moonshot
 - jina
 - chatglm
-- xinference
 - yi
 - openllm
 - localai

api/core/model_runtime/model_providers/bedrock/bedrock.py

@@ -17,9 +17,11 @@ class BedrockProvider(ModelProvider):
         """
         try:
             model_instance = self.get_model_instance(ModelType.LLM)
-            bedrock_validate_model_name = credentials.get('model_for_validation', 'amazon.titan-text-lite-v1')
+
+            # Use `amazon.titan-text-lite-v1` model by default for validating credentials
+            model_for_validation = credentials.get('model_for_validation', 'amazon.titan-text-lite-v1')
             model_instance.validate_credentials(
-                model=bedrock_validate_model_name,
+                model=model_for_validation,
                 credentials=credentials
             )
         except CredentialsValidateFailedError as ex:

api/core/model_runtime/model_providers/bedrock/bedrock.yaml

@@ -74,7 +74,7 @@ provider_credential_schema:
       label:
         en_US: Available Model Name
         zh_Hans: 可用模型名称
-      type: text-input
+      type: secret-input
       placeholder:
         en_US: A model you have access to (e.g. amazon.titan-text-lite-v1) for validation.
         zh_Hans: 为了进行验证，请输入一个您可用的模型名称 (例如：amazon.titan-text-lite-v1)

api/core/model_runtime/model_providers/bedrock/llm/anthropic.claude-instant-v1.yaml

@@ -1,33 +1,50 @@
 model: anthropic.claude-instant-v1
 label:
-  en_US: Claude Instant V1
+  en_US: Claude Instant 1
 model_type: llm
 model_properties:
   mode: chat
   context_size: 100000
 parameter_rules:
-  - name: temperature
-    use_template: temperature
-  - name: topP
-    use_template: top_p
-  - name: topK
-    label:
-      zh_Hans: 取样数量
-      en_US: Top K
-    type: int
-    help:
-      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
-      en_US: Only sample from the top K options for each subsequent token.
-    required: false
-    default: 250
-    min: 0
-    max: 500
-  - name: max_tokens_to_sample
+  - name: max_tokens
     use_template: max_tokens
     required: true
+    type: int
     default: 4096
     min: 1
     max: 4096
+    help:
+      zh_Hans: 停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
+      en_US: The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
+  - name: temperature
+    use_template: temperature
+    required: false
+    type: float
+    default: 1
+    min: 0.0
+    max: 1.0
+    help:
+      zh_Hans: 生成内容的随机性。
+      en_US: The amount of randomness injected into the response.
+  - name: top_p
+    required: false
+    type: float
+    default: 0.999
+    min: 0.000
+    max: 1.000
+    help:
+      zh_Hans: 在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
+      en_US: In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
+  - name: top_k
+    required: false
+    type: int
+    default: 0
+    min: 0
+    # tip docs from aws has error, max value is 500
+    max: 500
+    help:
+      zh_Hans: 对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
+      en_US: Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
 pricing:
   input: '0.0008'
   output: '0.0024'

api/core/model_runtime/model_providers/bedrock/llm/anthropic.claude-v1.yaml

@@ -1,33 +1,50 @@
 model: anthropic.claude-v1
 label:
-  en_US: Claude V1
+  en_US: Claude 1
 model_type: llm
 model_properties:
   mode: chat
   context_size: 100000
 parameter_rules:
-  - name: temperature
-    use_template: temperature
-  - name: top_p
-    use_template: top_p
-  - name: top_k
-    label:
-      zh_Hans: 取样数量
-      en_US: Top K
-    type: int
-    help:
-      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
-      en_US: Only sample from the top K options for each subsequent token.
-    required: false
-    default: 250
-    min: 0
-    max: 500
-  - name: max_tokens_to_sample
+  - name: max_tokens
     use_template: max_tokens
     required: true
+    type: int
     default: 4096
     min: 1
     max: 4096
+    help:
+      zh_Hans: 停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
+      en_US: The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
+  - name: temperature
+    use_template: temperature
+    required: false
+    type: float
+    default: 1
+    min: 0.0
+    max: 1.0
+    help:
+      zh_Hans: 生成内容的随机性。
+      en_US: The amount of randomness injected into the response.
+  - name: top_p
+    required: false
+    type: float
+    default: 0.999
+    min: 0.000
+    max: 1.000
+    help:
+      zh_Hans: 在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
+      en_US: In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
+  - name: top_k
+    required: false
+    type: int
+    default: 0
+    min: 0
+    # tip docs from aws has error, max value is 500
+    max: 500
+    help:
+      zh_Hans: 对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
+      en_US: Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
 pricing:
   input: '0.008'
   output: '0.024'

api/core/model_runtime/model_providers/bedrock/llm/anthropic.claude-v2.1.yaml

@@ -1,33 +1,50 @@
 model: anthropic.claude-v2:1
 label:
-  en_US: Claude V2.1
+  en_US: Claude 2.1
 model_type: llm
 model_properties:
   mode: chat
   context_size: 200000
 parameter_rules:
-  - name: temperature
-    use_template: temperature
-  - name: top_p
-    use_template: top_p
-  - name: top_k
-    label:
-      zh_Hans: 取样数量
-      en_US: Top K
-    type: int
-    help:
-      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
-      en_US: Only sample from the top K options for each subsequent token.
-    required: false
-    default: 250
-    min: 0
-    max: 500
-  - name: max_tokens_to_sample
+  - name: max_tokens
     use_template: max_tokens
     required: true
+    type: int
     default: 4096
     min: 1
     max: 4096
+    help:
+      zh_Hans: 停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
+      en_US: The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
+  - name: temperature
+    use_template: temperature
+    required: false
+    type: float
+    default: 1
+    min: 0.0
+    max: 1.0
+    help:
+      zh_Hans: 生成内容的随机性。
+      en_US: The amount of randomness injected into the response.
+  - name: top_p
+    required: false
+    type: float
+    default: 0.999
+    min: 0.000
+    max: 1.000
+    help:
+      zh_Hans: 在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
+      en_US: In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
+  - name: top_k
+    required: false
+    type: int
+    default: 0
+    min: 0
+    # tip docs from aws has error, max value is 500
+    max: 500
+    help:
+      zh_Hans: 对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
+      en_US: Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
 pricing:
   input: '0.008'
   output: '0.024'

api/core/model_runtime/model_providers/bedrock/llm/anthropic.claude-v2.yaml

@@ -1,33 +1,50 @@
 model: anthropic.claude-v2
 label:
-  en_US: Claude V2
+  en_US: Claude 2
 model_type: llm
 model_properties:
   mode: chat
   context_size: 100000
 parameter_rules:
-  - name: temperature
-    use_template: temperature
-  - name: top_p
-    use_template: top_p
-  - name: top_k
-    label:
-      zh_Hans: 取样数量
-      en_US: Top K
-    type: int
-    help:
-      zh_Hans: 仅从每个后续标记的前 K 个选项中采样。
-      en_US: Only sample from the top K options for each subsequent token.
-    required: false
-    default: 250
-    min: 0
-    max: 500
-  - name: max_tokens_to_sample
+  - name: max_tokens
     use_template: max_tokens
     required: true
+    type: int
     default: 4096
     min: 1
     max: 4096
+    help:
+      zh_Hans: 停止前生成的最大令牌数。请注意，Anthropic Claude 模型可能会在达到 max_tokens 的值之前停止生成令牌。不同的 Anthropic Claude 模型对此参数具有不同的最大值。
+      en_US: The maximum number of tokens to generate before stopping. Note that Anthropic Claude models might stop generating tokens before reaching the value of max_tokens. Different Anthropic Claude models have different maximum values for this parameter.
+  - name: temperature
+    use_template: temperature
+    required: false
+    type: float
+    default: 1
+    min: 0.0
+    max: 1.0
+    help:
+      zh_Hans: 生成内容的随机性。
+      en_US: The amount of randomness injected into the response.
+  - name: top_p
+    required: false
+    type: float
+    default: 0.999
+    min: 0.000
+    max: 1.000
+    help:
+      zh_Hans: 在核采样中，Anthropic Claude 按概率递减顺序计算每个后续标记的所有选项的累积分布，并在达到 top_p 指定的特定概率时将其切断。您应该更改温度或top_p，但不能同时更改两者。
+      en_US: In nucleus sampling, Anthropic Claude computes the cumulative distribution over all the options for each subsequent token in decreasing probability order and cuts it off once it reaches a particular probability specified by top_p. You should alter either temperature or top_p, but not both.
+  - name: top_k
+    required: false
+    type: int
+    default: 0
+    min: 0
+    # tip docs from aws has error, max value is 500
+    max: 500
+    help:
+      zh_Hans: 对于每个后续标记，仅从前 K 个选项中进行采样。使用 top_k 删除长尾低概率响应。
+      en_US: Only sample from the top K options for each subsequent token. Use top_k to remove long tail low probability responses.
 pricing:
   input: '0.008'
   output: '0.024'

api/core/model_runtime/model_providers/bedrock/llm/llm.py

@@ -72,16 +72,16 @@ class BedrockLargeLanguageModel(LargeLanguageModel):
         :return: full response or stream response chunk generator result
         """
 
-        # invoke claude 3 models via anthropic official SDK
-        if "anthropic.claude-3" in model:
-            return self._invoke_claude3(model, credentials, prompt_messages, model_parameters, stop, stream, user)
-        # invoke model
+        # invoke anthropic models via anthropic official SDK
+        if "anthropic" in model:
+            return self._generate_anthropic(model, credentials, prompt_messages, model_parameters, stop, stream, user)
+        # invoke other models via boto3 client
         return self._generate(model, credentials, prompt_messages, model_parameters, stop, stream, user)
 
-    def _invoke_claude3(self, model: str, credentials: dict, prompt_messages: list[PromptMessage], model_parameters: dict,
+    def _generate_anthropic(self, model: str, credentials: dict, prompt_messages: list[PromptMessage], model_parameters: dict,
                 stop: Optional[list[str]] = None, stream: bool = True, user: Optional[str] = None) -> Union[LLMResult, Generator]:
         """
-        Invoke Claude3 large language model
+        Invoke Anthropic large language model
 
         :param model: model name
         :param credentials: model credentials
@@ -114,7 +114,7 @@ class BedrockLargeLanguageModel(LargeLanguageModel):
             # ref: https://github.com/anthropics/anthropic-sdk-python/blob/e84645b07ca5267066700a104b4d8d6a8da1383d/src/anthropic/resources/messages.py#L465
             # extra_model_kwargs['metadata'] = message_create_params.Metadata(user_id=user)
 
-        system, prompt_message_dicts = self._convert_claude3_prompt_messages(prompt_messages)
+        system, prompt_message_dicts = self._convert_claude_prompt_messages(prompt_messages)
 
         if system:
             extra_model_kwargs['system'] = system
@@ -128,11 +128,11 @@ class BedrockLargeLanguageModel(LargeLanguageModel):
         )
 
         if stream:
-            return self._handle_claude3_stream_response(model, credentials, response, prompt_messages)
+            return self._handle_claude_stream_response(model, credentials, response, prompt_messages)
 
-        return self._handle_claude3_response(model, credentials, response, prompt_messages)
+        return self._handle_claude_response(model, credentials, response, prompt_messages)
 
-    def _handle_claude3_response(self, model: str, credentials: dict, response: Message,
+    def _handle_claude_response(self, model: str, credentials: dict, response: Message,
                                 prompt_messages: list[PromptMessage]) -> LLMResult:
         """
         Handle llm chat response
@@ -172,7 +172,7 @@ class BedrockLargeLanguageModel(LargeLanguageModel):
 
         return response
 
-    def _handle_claude3_stream_response(self, model: str, credentials: dict, response: Stream[MessageStreamEvent],
+    def _handle_claude_stream_response(self, model: str, credentials: dict, response: Stream[MessageStreamEvent],
                                         prompt_messages: list[PromptMessage], ) -> Generator:
         """
         Handle llm chat stream response
@@ -231,7 +231,7 @@ class BedrockLargeLanguageModel(LargeLanguageModel):
         except Exception as ex:
             raise InvokeError(str(ex))
 
-    def _calc_claude3_response_usage(self, model: str, credentials: dict, prompt_tokens: int, completion_tokens: int) -> LLMUsage:
+    def _calc_claude_response_usage(self, model: str, credentials: dict, prompt_tokens: int, completion_tokens: int) -> LLMUsage:
         """
         Calculate response usage
 
@@ -275,7 +275,7 @@ class BedrockLargeLanguageModel(LargeLanguageModel):
 
         return usage
 
-    def _convert_claude3_prompt_messages(self, prompt_messages: list[PromptMessage]) -> tuple[str, list[dict]]:
+    def _convert_claude_prompt_messages(self, prompt_messages: list[PromptMessage]) -> tuple[str, list[dict]]:
         """
         Convert prompt messages to dict list and system
         """
@@ -295,11 +295,11 @@ class BedrockLargeLanguageModel(LargeLanguageModel):
         prompt_message_dicts = []
         for message in prompt_messages:
             if not isinstance(message, SystemPromptMessage):
-                prompt_message_dicts.append(self._convert_claude3_prompt_message_to_dict(message))
+                prompt_message_dicts.append(self._convert_claude_prompt_message_to_dict(message))
 
         return system, prompt_message_dicts
 
-    def _convert_claude3_prompt_message_to_dict(self, message: PromptMessage) -> dict:
+    def _convert_claude_prompt_message_to_dict(self, message: PromptMessage) -> dict:
         """
         Convert PromptMessage to dict
         """
@@ -405,7 +405,7 @@ class BedrockLargeLanguageModel(LargeLanguageModel):
 
         if "anthropic.claude-3" in model:
             try:
-                self._invoke_claude3(model=model,
+                self._invoke_claude(model=model,
                                         credentials=credentials,
                                         prompt_messages=[{"role": "user", "content": "ping"}],
                                         model_parameters={},

api/core/model_runtime/model_providers/jina/rerank/jina-colbert-v1-en.yaml

@@ -0,0 +1,4 @@
+model: jina-colbert-v1-en
+model_type: rerank
+model_properties:
+  context_size: 8192

api/core/model_runtime/model_providers/model_provider_factory.py

@@ -1,4 +1,3 @@
-import importlib
 import logging
 import os
 from typing import Optional
@@ -10,6 +9,7 @@ from core.model_runtime.entities.provider_entities import ProviderConfig, Provid
 from core.model_runtime.model_providers.__base.model_provider import ModelProvider
 from core.model_runtime.schema_validators.model_credential_schema_validator import ModelCredentialSchemaValidator
 from core.model_runtime.schema_validators.provider_credential_schema_validator import ProviderCredentialSchemaValidator
+from core.utils.module_import_helper import load_single_subclass_from_source
 from core.utils.position_helper import get_position_map, sort_to_dict_by_position_map
 
 logger = logging.getLogger(__name__)
@@ -229,15 +229,10 @@ class ModelProviderFactory:
 
             # Dynamic loading {model_provider_name}.py file and find the subclass of ModelProvider
             py_path = os.path.join(model_provider_dir_path, model_provider_name + '.py')
-            spec = importlib.util.spec_from_file_location(f'core.model_runtime.model_providers.{model_provider_name}.{model_provider_name}', py_path)
-            mod = importlib.util.module_from_spec(spec)
-            spec.loader.exec_module(mod)
-
-            model_provider_class = None
-            for name, obj in vars(mod).items():
-                if isinstance(obj, type) and issubclass(obj, ModelProvider) and obj != ModelProvider:
-                    model_provider_class = obj
-                    break
+            model_provider_class = load_single_subclass_from_source(
+                module_name=f'core.model_runtime.model_providers.{model_provider_name}.{model_provider_name}',
+                script_path=py_path,
+                parent_type=ModelProvider)
 
             if not model_provider_class:
                 logger.warning(f"Missing Model Provider Class that extends ModelProvider in {py_path}, Skip.")

api/core/model_runtime/model_providers/spark/llm/_client.py

@@ -148,7 +148,8 @@ class SparkLLMClient:
         data = {
             "header": {
                 "app_id": self.app_id,
-                "uid": user_id
+                # resolve this error message => $.header.uid' length must be less or equal than 32
+                "uid": user_id[:32] if user_id else None
             },
             "parameter": {
                 "chat": {

api/core/model_runtime/model_providers/triton_inference_server/_assets/icon_s_en.svg

@@ -0,0 +1,3 @@
+<svg width="567" height="376" viewBox="0 0 567 376" fill="none" xmlns="http://www.w3.org/2000/svg">
+<path d="M58.0366 161.868C58.0366 161.868 109.261 86.2912 211.538 78.4724V51.053C98.2528 60.1511 0.152344 156.098 0.152344 156.098C0.152344 156.098 55.7148 316.717 211.538 331.426V302.282C97.1876 287.896 58.0366 161.868 58.0366 161.868ZM211.538 244.32V271.013C125.114 255.603 101.125 165.768 101.125 165.768C101.125 165.768 142.621 119.799 211.538 112.345V141.633C211.486 141.633 211.449 141.617 211.406 141.617C175.235 137.276 146.978 171.067 146.978 171.067C146.978 171.067 162.816 227.949 211.538 244.32ZM211.538 0.47998V51.053C214.864 50.7981 218.189 50.5818 221.533 50.468C350.326 46.1273 434.243 156.098 434.243 156.098C434.243 156.098 337.861 273.296 237.448 273.296C228.245 273.296 219.63 272.443 211.538 271.009V302.282C218.695 303.201 225.903 303.667 233.119 303.675C326.56 303.675 394.134 255.954 459.566 199.474C470.415 208.162 514.828 229.299 523.958 238.55C461.745 290.639 316.752 332.626 234.551 332.626C226.627 332.626 219.018 332.148 211.538 331.426V375.369H566.701V0.47998H211.538ZM211.538 112.345V78.4724C214.829 78.2425 218.146 78.0672 221.533 77.9602C314.148 75.0512 374.909 157.548 374.909 157.548C374.909 157.548 309.281 248.693 238.914 248.693C228.787 248.693 219.707 247.065 211.536 244.318V141.631C247.591 145.987 254.848 161.914 276.524 198.049L324.737 157.398C324.737 157.398 289.544 111.243 230.219 111.243C223.768 111.241 217.597 111.696 211.538 112.345Z" fill="#77B900"/>
+</svg>

api/core/model_runtime/model_providers/triton_inference_server/llm/llm.py

@@ -0,0 +1,267 @@
+from collections.abc import Generator
+
+from httpx import Response, post
+from yarl import URL
+
+from core.model_runtime.entities.common_entities import I18nObject
+from core.model_runtime.entities.llm_entities import LLMMode, LLMResult, LLMResultChunk, LLMResultChunkDelta, LLMUsage
+from core.model_runtime.entities.message_entities import (
+    AssistantPromptMessage,
+    PromptMessage,
+    PromptMessageTool,
+    SystemPromptMessage,
+    UserPromptMessage,
+)
+from core.model_runtime.entities.model_entities import (
+    AIModelEntity,
+    FetchFrom,
+    ModelPropertyKey,
+    ModelType,
+    ParameterRule,
+    ParameterType,
+)
+from core.model_runtime.errors.invoke import (
+    InvokeAuthorizationError,
+    InvokeBadRequestError,
+    InvokeConnectionError,
+    InvokeError,
+    InvokeRateLimitError,
+    InvokeServerUnavailableError,
+)
+from core.model_runtime.errors.validate import CredentialsValidateFailedError
+from core.model_runtime.model_providers.__base.large_language_model import LargeLanguageModel
+
+
+class TritonInferenceAILargeLanguageModel(LargeLanguageModel):
+    def _invoke(self, model: str, credentials: dict, prompt_messages: list[PromptMessage], 
+                model_parameters: dict, tools: list[PromptMessageTool] | None = None, 
+                stop: list[str] | None = None, stream: bool = True, user: str | None = None) \
+        -> LLMResult | Generator:
+        """
+            invoke LLM
+
+            see `core.model_runtime.model_providers.__base.large_language_model.LargeLanguageModel._invoke`
+        """
+        return self._generate(
+            model=model, credentials=credentials, prompt_messages=prompt_messages, model_parameters=model_parameters,
+            tools=tools, stop=stop, stream=stream, user=user,
+        )
+
+    def validate_credentials(self, model: str, credentials: dict) -> None:
+        """
+            validate credentials
+        """
+        if 'server_url' not in credentials:
+            raise CredentialsValidateFailedError('server_url is required in credentials')
+        
+        try:
+            self._invoke(model=model, credentials=credentials, prompt_messages=[
+                UserPromptMessage(content='ping')
+            ], model_parameters={}, stream=False)
+        except InvokeError as ex:
+            raise CredentialsValidateFailedError(f'An error occurred during connection: {str(ex)}')
+
+    def get_num_tokens(self, model: str, credentials: dict, prompt_messages: list[PromptMessage],
+                       tools: list[PromptMessageTool] | None = None) -> int:
+        """
+            get number of tokens
+
+            cause TritonInference LLM is a customized model, we could net detect which tokenizer to use
+            so we just take the GPT2 tokenizer as default
+        """
+        return self._get_num_tokens_by_gpt2(self._convert_prompt_message_to_text(prompt_messages))
+    
+    def _convert_prompt_message_to_text(self, message: list[PromptMessage]) -> str:
+        """
+            convert prompt message to text
+        """
+        text = ''
+        for item in message:
+            if isinstance(item, UserPromptMessage):
+                text += f'User: {item.content}'
+            elif isinstance(item, SystemPromptMessage):
+                text += f'System: {item.content}'
+            elif isinstance(item, AssistantPromptMessage):
+                text += f'Assistant: {item.content}'
+            else:
+                raise NotImplementedError(f'PromptMessage type {type(item)} is not supported')
+        return text
+
+    def get_customizable_model_schema(self, model: str, credentials: dict) -> AIModelEntity | None:
+        """
+            used to define customizable model schema
+        """
+        rules = [
+            ParameterRule(
+                name='temperature',
+                type=ParameterType.FLOAT,
+                use_template='temperature',
+                label=I18nObject(
+                    zh_Hans='温度',
+                    en_US='Temperature'
+                ),
+            ),
+            ParameterRule(
+                name='top_p',
+                type=ParameterType.FLOAT,
+                use_template='top_p',
+                label=I18nObject(
+                    zh_Hans='Top P',
+                    en_US='Top P'
+                )
+            ),
+            ParameterRule(
+                name='max_tokens',
+                type=ParameterType.INT,
+                use_template='max_tokens',
+                min=1,
+                max=int(credentials.get('context_length', 2048)),
+                default=min(512, int(credentials.get('context_length', 2048))),
+                label=I18nObject(
+                    zh_Hans='最大生成长度',
+                    en_US='Max Tokens'
+                )
+            )
+        ]
+
+        completion_type = None
+
+        if 'completion_type' in credentials:
+            if credentials['completion_type'] == 'chat':
+                completion_type = LLMMode.CHAT.value
+            elif credentials['completion_type'] == 'completion':
+                completion_type = LLMMode.COMPLETION.value
+            else:
+                raise ValueError(f'completion_type {credentials["completion_type"]} is not supported')
+        
+        entity = AIModelEntity(
+            model=model,
+            label=I18nObject(
+                en_US=model
+            ),
+            parameter_rules=rules,
+            fetch_from=FetchFrom.CUSTOMIZABLE_MODEL,
+            model_type=ModelType.LLM,
+            model_properties={
+                ModelPropertyKey.MODE: completion_type,
+                ModelPropertyKey.CONTEXT_SIZE: int(credentials.get('context_length', 2048)),
+            },
+        )
+
+        return entity
+    
+    def _generate(self, model: str, credentials: dict, prompt_messages: list[PromptMessage], 
+                 model_parameters: dict,
+                 tools: list[PromptMessageTool] | None = None, 
+                 stop: list[str] | None = None, stream: bool = True, user: str | None = None) \
+            -> LLMResult | Generator:
+        """
+            generate text from LLM
+        """
+        if 'server_url' not in credentials:
+            raise CredentialsValidateFailedError('server_url is required in credentials')
+        
+        if 'stream' in credentials and not bool(credentials['stream']) and stream:
+            raise ValueError(f'stream is not supported by model {model}')
+
+        try:
+            parameters = {}
+            if 'temperature' in model_parameters:
+                parameters['temperature'] = model_parameters['temperature']
+            if 'top_p' in model_parameters:
+                parameters['top_p'] = model_parameters['top_p']
+            if 'top_k' in model_parameters:
+                parameters['top_k'] = model_parameters['top_k']
+            if 'presence_penalty' in model_parameters:
+                parameters['presence_penalty'] = model_parameters['presence_penalty']
+            if 'frequency_penalty' in model_parameters:
+                parameters['frequency_penalty'] = model_parameters['frequency_penalty']
+
+            response = post(str(URL(credentials['server_url']) / 'v2' / 'models' / model / 'generate'), json={
+                'text_input': self._convert_prompt_message_to_text(prompt_messages),
+                'max_tokens': model_parameters.get('max_tokens', 512),
+                'parameters': {
+                    'stream': False,
+                    **parameters
+                },
+            }, timeout=(10, 120))
+            response.raise_for_status()
+            if response.status_code != 200:
+                raise InvokeBadRequestError(f'Invoke failed with status code {response.status_code}, {response.text}')
+            
+            if stream:
+                return self._handle_chat_stream_response(model=model, credentials=credentials, prompt_messages=prompt_messages,
+                                                        tools=tools, resp=response)
+            return self._handle_chat_generate_response(model=model, credentials=credentials, prompt_messages=prompt_messages,
+                                                        tools=tools, resp=response)
+        except Exception as ex:
+            raise InvokeConnectionError(f'An error occurred during connection: {str(ex)}')
+        
+    def _handle_chat_generate_response(self, model: str, credentials: dict, prompt_messages: list[PromptMessage],
+                                        tools: list[PromptMessageTool],
+                                        resp: Response) -> LLMResult:
+        """
+            handle normal chat generate response
+        """
+        text = resp.json()['text_output']
+
+        usage = LLMUsage.empty_usage()
+        usage.prompt_tokens = self.get_num_tokens(model, credentials, prompt_messages)
+        usage.completion_tokens = self._get_num_tokens_by_gpt2(text)
+
+        return LLMResult(
+            model=model,
+            prompt_messages=prompt_messages,
+            message=AssistantPromptMessage(
+                content=text
+            ),
+            usage=usage
+        )
+
+    def _handle_chat_stream_response(self, model: str, credentials: dict, prompt_messages: list[PromptMessage],
+                                        tools: list[PromptMessageTool],
+                                        resp: Response) -> Generator:
+        """
+            handle normal chat generate response
+        """
+        text = resp.json()['text_output']
+
+        usage = LLMUsage.empty_usage()
+        usage.prompt_tokens = self.get_num_tokens(model, credentials, prompt_messages)
+        usage.completion_tokens = self._get_num_tokens_by_gpt2(text)
+
+        yield LLMResultChunk(
+            model=model,
+            prompt_messages=prompt_messages,
+            delta=LLMResultChunkDelta(
+                index=0,
+                message=AssistantPromptMessage(
+                    content=text
+                ),
+                usage=usage
+            )
+        )
+
+    @property
+    def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:
+        """
+        Map model invoke error to unified error
+        The key is the error type thrown to the caller
+        The value is the error type thrown by the model,
+        which needs to be converted into a unified error type for the caller.
+
+        :return: Invoke error mapping
+        """
+        return {
+            InvokeConnectionError: [
+            ],
+            InvokeServerUnavailableError: [
+            ],
+            InvokeRateLimitError: [
+            ],
+            InvokeAuthorizationError: [
+            ],
+            InvokeBadRequestError: [
+                ValueError
+            ]
+        }

api/core/model_runtime/model_providers/triton_inference_server/triton_inference_server.py

@@ -0,0 +1,9 @@
+import logging
+
+from core.model_runtime.model_providers.__base.model_provider import ModelProvider
+
+logger = logging.getLogger(__name__)
+
+class XinferenceAIProvider(ModelProvider):
+    def validate_provider_credentials(self, credentials: dict) -> None:
+        pass

api/core/model_runtime/model_providers/triton_inference_server/triton_inference_server.yaml

@@ -0,0 +1,84 @@
+provider: triton_inference_server
+label:
+  en_US: Triton Inference Server
+icon_small:
+  en_US: icon_s_en.svg
+icon_large:
+  en_US: icon_l_en.png
+background: "#EFFDFD"
+help:
+  title:
+    en_US: How to deploy Triton Inference Server
+    zh_Hans: 如何部署 Triton Inference Server
+  url:
+    en_US: https://github.com/triton-inference-server/server
+supported_model_types:
+  - llm
+configurate_methods:
+  - customizable-model
+model_credential_schema:
+  model:
+    label:
+      en_US: Model Name
+      zh_Hans: 模型名称
+    placeholder:
+      en_US: Enter your model name
+      zh_Hans: 输入模型名称
+  credential_form_schemas:
+    - variable: server_url
+      label:
+        zh_Hans: 服务器URL
+        en_US: Server url
+      type: secret-input
+      required: true
+      placeholder:
+        zh_Hans: 在此输入 Triton Inference Server 的服务器地址，如 http://192.168.1.100:8000
+        en_US: Enter the url of your Triton Inference Server, e.g. http://192.168.1.100:8000
+    - variable: context_size
+      label:
+        zh_Hans: 上下文大小
+        en_US: Context size
+      type: text-input
+      required: true
+      placeholder:
+        zh_Hans: 在此输入您的上下文大小
+        en_US: Enter the context size
+      default: 2048
+    - variable: completion_type
+      label:
+        zh_Hans: 补全类型
+        en_US: Model type
+      type: select
+      required: true
+      default: chat
+      placeholder:
+        zh_Hans: 在此输入您的补全类型
+        en_US: Enter the completion type
+      options:
+        - label:
+            zh_Hans: 补全模型
+            en_US: Completion model
+          value: completion
+        - label:
+            zh_Hans: 对话模型
+            en_US: Chat model
+          value: chat
+    - variable: stream
+      label:
+        zh_Hans: 流式输出
+        en_US: Stream output
+      type: select
+      required: true
+      default: true
+      placeholder:
+        zh_Hans: 是否支持流式输出
+        en_US: Whether to support stream output
+      options:
+        - label:
+            zh_Hans: 是
+            en_US: Yes
+          value: true
+        - label:
+            zh_Hans: 否
+            en_US: No
+          value: false

api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-4k-0205.yaml

@@ -0,0 +1,37 @@
+model: ernie-3.5-8k
+label:
+  en_US: Ernie-3.5-8K
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 4096
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    min: 0.1
+    max: 1.0
+    default: 0.8
+  - name: top_p
+    use_template: top_p
+  - name: max_tokens
+    use_template: max_tokens
+    default: 1024
+    min: 2
+    max: 2048
+  - name: presence_penalty
+    use_template: presence_penalty
+  - name: frequency_penalty
+    use_template: frequency_penalty
+  - name: response_format
+    use_template: response_format
+  - name: disable_search
+    label:
+      zh_Hans: 禁用搜索
+      en_US: Disable Search
+    type: boolean
+    help:
+      zh_Hans: 禁用模型自行进行外部搜索。
+      en_US: Disable the model to perform external search.
+    required: false

api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-8k-0205.yaml

@@ -0,0 +1,37 @@
+model: ernie-3.5-8k-0205
+label:
+  en_US: Ernie-3.5-8K-0205
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    min: 0.1
+    max: 1.0
+    default: 0.8
+  - name: top_p
+    use_template: top_p
+  - name: max_tokens
+    use_template: max_tokens
+    default: 1024
+    min: 2
+    max: 2048
+  - name: presence_penalty
+    use_template: presence_penalty
+  - name: frequency_penalty
+    use_template: frequency_penalty
+  - name: response_format
+    use_template: response_format
+  - name: disable_search
+    label:
+      zh_Hans: 禁用搜索
+      en_US: Disable Search
+    type: boolean
+    help:
+      zh_Hans: 禁用模型自行进行外部搜索。
+      en_US: Disable the model to perform external search.
+    required: false

api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-8k-1222.yaml

@@ -0,0 +1,37 @@
+model: ernie-3.5-8k-1222
+label:
+  en_US: Ernie-3.5-8K-1222
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    min: 0.1
+    max: 1.0
+    default: 0.8
+  - name: top_p
+    use_template: top_p
+  - name: max_tokens
+    use_template: max_tokens
+    default: 1024
+    min: 2
+    max: 2048
+  - name: presence_penalty
+    use_template: presence_penalty
+  - name: frequency_penalty
+    use_template: frequency_penalty
+  - name: response_format
+    use_template: response_format
+  - name: disable_search
+    label:
+      zh_Hans: 禁用搜索
+      en_US: Disable Search
+    type: boolean
+    help:
+      zh_Hans: 禁用模型自行进行外部搜索。
+      en_US: Disable the model to perform external search.
+    required: false

api/core/model_runtime/model_providers/wenxin/llm/ernie-3.5-8k.yaml

@@ -0,0 +1,37 @@
+model: ernie-3.5-8k
+label:
+  en_US: Ernie-3.5-8K
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    min: 0.1
+    max: 1.0
+    default: 0.8
+  - name: top_p
+    use_template: top_p
+  - name: max_tokens
+    use_template: max_tokens
+    default: 1024
+    min: 2
+    max: 2048
+  - name: presence_penalty
+    use_template: presence_penalty
+  - name: frequency_penalty
+    use_template: frequency_penalty
+  - name: response_format
+    use_template: response_format
+  - name: disable_search
+    label:
+      zh_Hans: 禁用搜索
+      en_US: Disable Search
+    type: boolean
+    help:
+      zh_Hans: 禁用模型自行进行外部搜索。
+      en_US: Disable the model to perform external search.
+    required: false

api/core/model_runtime/model_providers/wenxin/llm/ernie-4.0-8k.yaml

@@ -0,0 +1,37 @@
+model: ernie-4.0-8k
+label:
+  en_US: Ernie-4.0-8K
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    min: 0.1
+    max: 1.0
+    default: 0.8
+  - name: top_p
+    use_template: top_p
+  - name: max_tokens
+    use_template: max_tokens
+    default: 1024
+    min: 2
+    max: 2048
+  - name: presence_penalty
+    use_template: presence_penalty
+  - name: frequency_penalty
+    use_template: frequency_penalty
+  - name: response_format
+    use_template: response_format
+  - name: disable_search
+    label:
+      zh_Hans: 禁用搜索
+      en_US: Disable Search
+    type: boolean
+    help:
+      zh_Hans: 禁用模型自行进行外部搜索。
+      en_US: Disable the model to perform external search.
+    required: false

api/core/model_runtime/model_providers/wenxin/llm/ernie-bot-4.yaml

@@ -36,3 +36,4 @@ parameter_rules:
       zh_Hans: 禁用模型自行进行外部搜索。
       en_US: Disable the model to perform external search.
     required: false
+deprecated: true

api/core/model_runtime/model_providers/wenxin/llm/ernie-bot-8k.yaml

@@ -36,3 +36,4 @@ parameter_rules:
       zh_Hans: 禁用模型自行进行外部搜索。
       en_US: Disable the model to perform external search.
     required: false
+deprecated: true

api/core/model_runtime/model_providers/wenxin/llm/ernie-bot-turbo.yaml

@@ -27,3 +27,4 @@ parameter_rules:
     use_template: frequency_penalty
   - name: response_format
     use_template: response_format
+deprecated: true

api/core/model_runtime/model_providers/wenxin/llm/ernie-bot.yaml

@@ -36,3 +36,4 @@ parameter_rules:
     required: false
   - name: response_format
     use_template: response_format
+deprecated: true

api/core/model_runtime/model_providers/wenxin/llm/ernie-lite-8k-0308.yaml

@@ -0,0 +1,30 @@
+model: ernie-lite-8k-0308
+label:
+  en_US: ERNIE-Lite-8K-0308
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    min: 0.1
+    max: 1.0
+    default: 0.95
+  - name: top_p
+    use_template: top_p
+    min: 0
+    max: 1.0
+    default: 0.7
+  - name: max_tokens
+    use_template: max_tokens
+    default: 1024
+    min: 2
+    max: 1024
+  - name: presence_penalty
+    use_template: presence_penalty
+    default: 1.0
+    min: 1.0
+    max: 2.0

api/core/model_runtime/model_providers/wenxin/llm/ernie-lite-8k-0922.yaml

@@ -0,0 +1,30 @@
+model: ernie-lite-8k-0922
+label:
+  en_US: ERNIE-Lite-8K-0922
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    min: 0.1
+    max: 1.0
+    default: 0.95
+  - name: top_p
+    use_template: top_p
+    min: 0
+    max: 1.0
+    default: 0.7
+  - name: max_tokens
+    use_template: max_tokens
+    default: 1024
+    min: 2
+    max: 1024
+  - name: presence_penalty
+    use_template: presence_penalty
+    default: 1.0
+    min: 1.0
+    max: 2.0

api/core/model_runtime/model_providers/wenxin/llm/ernie-speed-128k.yaml

@@ -0,0 +1,30 @@
+model: ernie-speed-128k
+label:
+  en_US: ERNIE-Speed-128K
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 128000
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    min: 0.1
+    max: 1.0
+    default: 0.95
+  - name: top_p
+    use_template: top_p
+    min: 0
+    max: 1.0
+    default: 0.7
+  - name: max_tokens
+    use_template: max_tokens
+    default: 1024
+    min: 2
+    max: 1024
+  - name: presence_penalty
+    use_template: presence_penalty
+    default: 1.0
+    min: 1.0
+    max: 2.0

api/core/model_runtime/model_providers/wenxin/llm/ernie-speed-8k.yaml

@@ -0,0 +1,30 @@
+model: ernie-speed-8k
+label:
+  en_US: ERNIE-Speed-8K
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    min: 0.1
+    max: 1.0
+    default: 0.95
+  - name: top_p
+    use_template: top_p
+    min: 0
+    max: 1.0
+    default: 0.7
+  - name: max_tokens
+    use_template: max_tokens
+    default: 1024
+    min: 2
+    max: 1024
+  - name: presence_penalty
+    use_template: presence_penalty
+    default: 1.0
+    min: 1.0
+    max: 2.0

api/core/model_runtime/model_providers/wenxin/llm/ernie-speed-appbuilder.yaml

@@ -0,0 +1,25 @@
+model: ernie-speed-appbuilder
+label:
+  en_US: ERNIE-Speed-AppBuilder
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    min: 0.1
+    max: 1.0
+    default: 0.95
+  - name: top_p
+    use_template: top_p
+    min: 0
+    max: 1.0
+    default: 0.7
+  - name: presence_penalty
+    use_template: presence_penalty
+    default: 1.0
+    min: 1.0
+    max: 2.0

api/core/model_runtime/model_providers/wenxin/llm/ernie_bot.py

@@ -121,15 +121,29 @@ class ErnieMessage:
 
 class ErnieBotModel:
     api_bases = {
-        'ernie-bot': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions',
+        'ernie-bot': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie-3.5-4k-0205',
         'ernie-bot-4': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions_pro',
-        'ernie-bot-8k': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie_bot_8k',
+        'ernie-bot-8k': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions',
         'ernie-bot-turbo': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/eb-instant',
+        'ernie-3.5-8k': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions',
+        'ernie-3.5-8k-0205': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie-3.5-8k-0205',
+        'ernie-3.5-8k-1222': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie-3.5-8k-1222',
+        'ernie-3.5-4k-0205': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie-3.5-4k-0205',
+        'ernie-4.0-8k': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/completions_pro',
+        'ernie-speed-8k': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie_speed',
+        'ernie-speed-128k': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie-speed-128k',
+        'ernie-speed-appbuilder': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ai_apaas',
+        'ernie-lite-8k-0922': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/eb-instant',
+        'ernie-lite-8k-0308': 'https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop/chat/ernie-lite-8k',
     }
 
     function_calling_supports = [
         'ernie-bot',
         'ernie-bot-8k',
+        'ernie-3.5-8k',
+        'ernie-3.5-8k-0205',
+        'ernie-3.5-8k-1222',
+        'ernie-3.5-4k-0205'
     ]
 
     api_key: str = ''
@@ -285,6 +299,12 @@ class ErnieBotModel:
             **parameters
         }
 
+        if 'max_tokens' in parameters and type(parameters['max_tokens']) == int:
+            body['max_output_tokens'] = parameters['max_tokens']
+
+        if 'presence_penalty' in parameters and type(parameters['presence_penalty']) == float:
+            body['penalty_score'] = parameters['presence_penalty']
+
         if system_message:
             body['system'] = system_message
 

api/core/model_runtime/model_providers/yi/llm/llm.py

@@ -1,30 +1,119 @@
 from collections.abc import Generator
 from typing import Optional, Union
+from urllib.parse import urlparse
+
+import tiktoken
 
 from core.model_runtime.entities.llm_entities import LLMResult
 from core.model_runtime.entities.message_entities import (
     PromptMessage,
     PromptMessageTool,
+    SystemPromptMessage,
 )
-from core.model_runtime.model_providers.openai_api_compatible.llm.llm import OAIAPICompatLargeLanguageModel
+from core.model_runtime.model_providers.openai.llm.llm import OpenAILargeLanguageModel
 
 
-class YiLargeLanguageModel(OAIAPICompatLargeLanguageModel):
+class YiLargeLanguageModel(OpenAILargeLanguageModel):
     def _invoke(self, model: str, credentials: dict,
                 prompt_messages: list[PromptMessage], model_parameters: dict,
                 tools: Optional[list[PromptMessageTool]] = None, stop: Optional[list[str]] = None,
                 stream: bool = True, user: Optional[str] = None) \
             -> Union[LLMResult, Generator]:
         self._add_custom_parameters(credentials)
+
+        # yi-vl-plus not support system prompt yet.
+        if model == "yi-vl-plus":
+            prompt_message_except_system: list[PromptMessage] = []
+            for message in prompt_messages:
+                if not isinstance(message, SystemPromptMessage):
+                    prompt_message_except_system.append(message)
+            return super()._invoke(model, credentials, prompt_message_except_system, model_parameters, tools, stop, stream)
+
         return super()._invoke(model, credentials, prompt_messages, model_parameters, tools, stop, stream)
 
     def validate_credentials(self, model: str, credentials: dict) -> None:
         self._add_custom_parameters(credentials)
         super().validate_credentials(model, credentials)
 
+    # refactored from openai model runtime, use cl100k_base for calculate token number
+    def _num_tokens_from_string(self, model: str, text: str,
+                                tools: Optional[list[PromptMessageTool]] = None) -> int:
+        """
+        Calculate num tokens for text completion model with tiktoken package.
+
+        :param model: model name
+        :param text: prompt text
+        :param tools: tools for tool calling
+        :return: number of tokens
+        """
+        encoding = tiktoken.get_encoding("cl100k_base")
+        num_tokens = len(encoding.encode(text))
+
+        if tools:
+            num_tokens += self._num_tokens_for_tools(encoding, tools)
+
+        return num_tokens
+
+    # refactored from openai model runtime, use cl100k_base for calculate token number
+    def _num_tokens_from_messages(self, model: str, messages: list[PromptMessage],
+                                  tools: Optional[list[PromptMessageTool]] = None) -> int:
+        """Calculate num tokens for gpt-3.5-turbo and gpt-4 with tiktoken package.
+
+        Official documentation: https://github.com/openai/openai-cookbook/blob/
+        main/examples/How_to_format_inputs_to_ChatGPT_models.ipynb"""
+        encoding = tiktoken.get_encoding("cl100k_base")
+        tokens_per_message = 3
+        tokens_per_name = 1
+
+        num_tokens = 0
+        messages_dict = [self._convert_prompt_message_to_dict(m) for m in messages]
+        for message in messages_dict:
+            num_tokens += tokens_per_message
+            for key, value in message.items():
+                # Cast str(value) in case the message value is not a string
+                # This occurs with function messages
+                # TODO: The current token calculation method for the image type is not implemented,
+                #  which need to download the image and then get the resolution for calculation,
+                #  and will increase the request delay
+                if isinstance(value, list):
+                    text = ''
+                    for item in value:
+                        if isinstance(item, dict) and item['type'] == 'text':
+                            text += item['text']
+
+                    value = text
+
+                if key == "tool_calls":
+                    for tool_call in value:
+                        for t_key, t_value in tool_call.items():
+                            num_tokens += len(encoding.encode(t_key))
+                            if t_key == "function":
+                                for f_key, f_value in t_value.items():
+                                    num_tokens += len(encoding.encode(f_key))
+                                    num_tokens += len(encoding.encode(f_value))
+                            else:
+                                num_tokens += len(encoding.encode(t_key))
+                                num_tokens += len(encoding.encode(t_value))
+                else:
+                    num_tokens += len(encoding.encode(str(value)))
+
+                if key == "name":
+                    num_tokens += tokens_per_name
+
+        # every reply is primed with <im_start>assistant
+        num_tokens += 3
+
+        if tools:
+            num_tokens += self._num_tokens_for_tools(encoding, tools)
+
+        return num_tokens
+
     @staticmethod
     def _add_custom_parameters(credentials: dict) -> None:
         credentials['mode'] = 'chat'
-
+        credentials['openai_api_key']=credentials['api_key']
         if 'endpoint_url' not in credentials or credentials['endpoint_url'] == "":
-            credentials['endpoint_url'] = 'https://api.lingyiwanwu.com/v1'
+            credentials['openai_api_base']='https://api.lingyiwanwu.com'
+        else:
+            parsed_url = urlparse(credentials['endpoint_url'])
+            credentials['openai_api_base']=f"{parsed_url.scheme}://{parsed_url.netloc}"

api/core/model_runtime/model_providers/yi/llm/yi-34b-chat-0205.yaml

@@ -37,7 +37,7 @@ parameter_rules:
       zh_Hans: 控制生成结果的随机性。数值越小，随机性越弱；数值越大，随机性越强。一般而言，top_p 和 temperature 两个参数选择一个进行调整即可。
       en_US: Control the randomness of generated results. The smaller the value, the weaker the randomness; the larger the value, the stronger the randomness. Generally speaking, you can adjust one of the two parameters top_p and temperature.
 pricing:
-  input: '0.0025'
-  output: '0.0025'
-  unit: '0.00001'
+  input: '2.5'
+  output: '2.5'
+  unit: '0.000001'
   currency: RMB

api/core/model_runtime/model_providers/yi/llm/yi-34b-chat-200k.yaml

@@ -37,7 +37,7 @@ parameter_rules:
       zh_Hans: 控制生成结果的随机性。数值越小，随机性越弱；数值越大，随机性越强。一般而言，top_p 和 temperature 两个参数选择一个进行调整即可。
       en_US: Control the randomness of generated results. The smaller the value, the weaker the randomness; the larger the value, the stronger the randomness. Generally speaking, you can adjust one of the two parameters top_p and temperature.
 pricing:
-  input: '0.012'
-  output: '0.012'
-  unit: '0.00001'
+  input: '12'
+  output: '12'
+  unit: '0.000001'
   currency: RMB

api/core/model_runtime/model_providers/yi/llm/yi-vl-plus.yaml

@@ -37,7 +37,7 @@ parameter_rules:
       zh_Hans: 控制生成结果的随机性。数值越小，随机性越弱；数值越大，随机性越强。一般而言，top_p 和 temperature 两个参数选择一个进行调整即可。
       en_US: Control the randomness of generated results. The smaller the value, the weaker the randomness; the larger the value, the stronger the randomness. Generally speaking, you can adjust one of the two parameters top_p and temperature.
 pricing:
-  input: '0.01'
-  output: '0.03'
-  unit: '0.001'
-  currency: USD
+  input: '6'
+  output: '6'
+  unit: '0.000001'
+  currency: RMB

api/core/model_runtime/model_providers/zhipuai/llm/llm.py

@@ -5,6 +5,7 @@ from core.model_runtime.entities.llm_entities import LLMResult, LLMResultChunk,
 from core.model_runtime.entities.message_entities import (
     AssistantPromptMessage,
     PromptMessage,
+    PromptMessageContent,
     PromptMessageContentType,
     PromptMessageRole,
     PromptMessageTool,
@@ -31,6 +32,7 @@ And you should always end the block with a "```" to indicate the end of the JSON
 
 ```JSON"""
 
+
 class ZhipuAILargeLanguageModel(_CommonZhipuaiAI, LargeLanguageModel):
 
     def _invoke(self, model: str, credentials: dict,
@@ -159,7 +161,7 @@ class ZhipuAILargeLanguageModel(_CommonZhipuaiAI, LargeLanguageModel):
 
         if len(prompt_messages) == 0:
             raise ValueError('At least one message is required')
-        
+
         if prompt_messages[0].role == PromptMessageRole.SYSTEM:
             if not prompt_messages[0].content:
                 prompt_messages = prompt_messages[1:]
@@ -185,7 +187,7 @@ class ZhipuAILargeLanguageModel(_CommonZhipuaiAI, LargeLanguageModel):
                     continue
 
                 if new_prompt_messages and new_prompt_messages[-1].role == PromptMessageRole.USER and \
-                    copy_prompt_message.role == PromptMessageRole.USER:
+                        copy_prompt_message.role == PromptMessageRole.USER:
                     new_prompt_messages[-1].content += "\n\n" + copy_prompt_message.content
                 else:
                     if copy_prompt_message.role == PromptMessageRole.USER:
@@ -205,31 +207,7 @@ class ZhipuAILargeLanguageModel(_CommonZhipuaiAI, LargeLanguageModel):
                     new_prompt_messages.append(copy_prompt_message)
 
         if model == 'glm-4v':
-            params = {
-                'model': model,
-                'messages': [{
-                    'role': prompt_message.role.value,
-                    'content': 
-                        [
-                            {
-                                'type': 'text',
-                                'text': prompt_message.content
-                            }
-                        ] if isinstance(prompt_message.content, str) else 
-                        [
-                            {
-                                'type': 'image',
-                                'image_url': {
-                                    'url': content.data
-                                }
-                            } if content.type == PromptMessageContentType.IMAGE else {
-                                'type': 'text',
-                                'text': content.data
-                            } for content in prompt_message.content
-                        ],
-                } for prompt_message in new_prompt_messages],
-                **model_parameters
-            }
+            params = self._construct_glm_4v_parameter(model, new_prompt_messages, model_parameters)
         else:
             params = {
                 'model': model,
@@ -277,8 +255,8 @@ class ZhipuAILargeLanguageModel(_CommonZhipuaiAI, LargeLanguageModel):
                 for prompt_message in new_prompt_messages:
                     # merge system message to user message
                     if prompt_message.role == PromptMessageRole.SYSTEM or \
-                        prompt_message.role == PromptMessageRole.TOOL or \
-                        prompt_message.role == PromptMessageRole.USER:
+                            prompt_message.role == PromptMessageRole.TOOL or \
+                            prompt_message.role == PromptMessageRole.USER:
                         if len(params['messages']) > 0 and params['messages'][-1]['role'] == 'user':
                             params['messages'][-1]['content'] += "\n\n" + prompt_message.content
                         else:
@@ -306,8 +284,44 @@ class ZhipuAILargeLanguageModel(_CommonZhipuaiAI, LargeLanguageModel):
 
         response = client.chat.completions.create(**params, **extra_model_kwargs)
         return self._handle_generate_response(model, credentials_kwargs, tools, response, prompt_messages)
-        
-    def _handle_generate_response(self, model: str, 
+
+    def _construct_glm_4v_parameter(self, model: str, prompt_messages: list[PromptMessage],
+                                    model_parameters: dict):
+        messages = [
+            {
+                'role': message.role.value,
+                'content': self._construct_glm_4v_messages(message.content)
+            }
+            for message in prompt_messages
+        ]
+
+        params = {
+            'model': model,
+            'messages': messages,
+            **model_parameters
+        }
+
+        return params
+
+    def _construct_glm_4v_messages(self, prompt_message: Union[str | list[PromptMessageContent]]) -> list[dict]:
+        if isinstance(prompt_message, str):
+            return [{'type': 'text', 'text': prompt_message}]
+
+        return [
+            {'type': 'image_url', 'image_url': {'url': self._remove_image_header(item.data)}}
+            if item.type == PromptMessageContentType.IMAGE else
+            {'type': 'text', 'text': item.data}
+
+            for item in prompt_message
+        ]
+
+    def _remove_image_header(self, image: str) -> str:
+        if image.startswith('data:image'):
+            return image.split(',')[1]
+
+        return image
+
+    def _handle_generate_response(self, model: str,
                                   credentials: dict,
                                   tools: Optional[list[PromptMessageTool]],
                                   response: Completion,
@@ -338,7 +352,7 @@ class ZhipuAILargeLanguageModel(_CommonZhipuaiAI, LargeLanguageModel):
                         )
 
             text += choice.message.content or ''
-          
+
         prompt_usage = response.usage.prompt_tokens
         completion_usage = response.usage.completion_tokens
 
@@ -358,7 +372,7 @@ class ZhipuAILargeLanguageModel(_CommonZhipuaiAI, LargeLanguageModel):
 
         return result
 
-    def _handle_generate_stream_response(self, model: str, 
+    def _handle_generate_stream_response(self, model: str,
                                          credentials: dict,
                                          tools: Optional[list[PromptMessageTool]],
                                          responses: Generator[ChatCompletionChunk, None, None],
@@ -380,7 +394,7 @@ class ZhipuAILargeLanguageModel(_CommonZhipuaiAI, LargeLanguageModel):
 
             if delta.finish_reason is None and (delta.delta.content is None or delta.delta.content == ''):
                 continue
-            
+
             assistant_tool_calls: list[AssistantPromptMessage.ToolCall] = []
             for tool_call in delta.delta.tool_calls or []:
                 if tool_call.type == 'function':
@@ -454,8 +468,8 @@ class ZhipuAILargeLanguageModel(_CommonZhipuaiAI, LargeLanguageModel):
 
         return message_text
 
-
-    def _convert_messages_to_prompt(self, messages: list[PromptMessage], tools: Optional[list[PromptMessageTool]] = None) -> str:
+    def _convert_messages_to_prompt(self, messages: list[PromptMessage],
+                                    tools: Optional[list[PromptMessageTool]] = None) -> str:
         """
         :param messages: List of PromptMessage to combine.
         :return: Combined string with necessary human_prompt and ai_prompt tags.
@@ -473,4 +487,4 @@ class ZhipuAILargeLanguageModel(_CommonZhipuaiAI, LargeLanguageModel):
                 text += f"\n{tool.json()}"
 
         # trim off the trailing ' ' that might come from the "Assistant: "
-        return text.rstrip()
+        return text.rstrip()

api/core/rag/datasource/vdb/milvus/milvus_vector.py

@@ -144,6 +144,16 @@ class MilvusVector(BaseVector):
             utility.drop_collection(self._collection_name, None, using=alias)
 
     def text_exists(self, id: str) -> bool:
+        alias = uuid4().hex
+        if self._client_config.secure:
+            uri = "https://" + str(self._client_config.host) + ":" + str(self._client_config.port)
+        else:
+            uri = "http://" + str(self._client_config.host) + ":" + str(self._client_config.port)
+        connections.connect(alias=alias, uri=uri, user=self._client_config.user, password=self._client_config.password)
+
+        from pymilvus import utility
+        if not utility.has_collection(self._collection_name, using=alias):
+            return False
 
         result = self._client.query(collection_name=self._collection_name,
                                     filter=f'metadata["doc_id"] == "{id}"',

api/core/rag/datasource/vdb/qdrant/qdrant_vector.py

@@ -275,6 +275,13 @@ class QdrantVector(BaseVector):
             )
 
     def text_exists(self, id: str) -> bool:
+        all_collection_name = []
+        collections_response = self._client.get_collections()
+        collection_list = collections_response.collections
+        for collection in collection_list:
+            all_collection_name.append(collection.name)
+        if self._collection_name not in all_collection_name:
+            return False
         response = self._client.retrieve(
             collection_name=self._collection_name,
             ids=[id]

api/core/rag/datasource/vdb/vector_factory.py

@@ -66,7 +66,7 @@ class Vector:
                     raise ValueError('Dataset Collection Bindings is not exist!')
             else:
                 if self._dataset.index_struct_dict:
-                    class_prefix: str = self.dataset.index_struct_dict['vector_store']['class_prefix']
+                    class_prefix: str = self._dataset.index_struct_dict['vector_store']['class_prefix']
                     collection_name = class_prefix
                 else:
                     dataset_id = self._dataset.id
@@ -128,8 +128,8 @@ class Vector:
         if kwargs.get('duplicate_check', False):
             documents = self._filter_duplicate_texts(documents)
         embeddings = self._embeddings.embed_documents([document.page_content for document in documents])
-        self._vector_processor.add_texts(
-            documents=documents,
+        self._vector_processor.create(
+            texts=documents,
             embeddings=embeddings,
             **kwargs
         )

api/core/rag/datasource/vdb/weaviate/weaviate_vector.py

@@ -134,6 +134,11 @@ class WeaviateVector(BaseVector):
 
     def text_exists(self, id: str) -> bool:
         collection_name = self._collection_name
+        schema = self._default_schema(self._collection_name)
+
+        # check whether the index already exists
+        if not self._client.schema.contains(schema):
+            return False
         result = self._client.query.get(collection_name).with_additional(["id"]).with_where({
             "path": ["doc_id"],
             "operator": "Equal",

api/core/rag/extractor/unstructured/unstructured_doc_extractor.py

@@ -53,7 +53,7 @@ class UnstructuredWordExtractor(BaseExtractor):
             elements = partition_docx(filename=self._file_path)
 
         from unstructured.chunking.title import chunk_by_title
-        chunks = chunk_by_title(elements, max_characters=2000, combine_text_under_n_chars=0)
+        chunks = chunk_by_title(elements, max_characters=2000, combine_text_under_n_chars=2000)
         documents = []
         for chunk in chunks:
             text = chunk.text.strip()

api/core/rag/extractor/unstructured/unstructured_eml_extractor.py

@@ -43,7 +43,7 @@ class UnstructuredEmailExtractor(BaseExtractor):
             pass
 
         from unstructured.chunking.title import chunk_by_title
-        chunks = chunk_by_title(elements, max_characters=2000, combine_text_under_n_chars=0)
+        chunks = chunk_by_title(elements, max_characters=2000, combine_text_under_n_chars=2000)
         documents = []
         for chunk in chunks:
             text = chunk.text.strip()

api/core/rag/extractor/unstructured/unstructured_markdown_extractor.py

@@ -38,7 +38,7 @@ class UnstructuredMarkdownExtractor(BaseExtractor):
 
         elements = partition_md(filename=self._file_path, api_url=self._api_url)
         from unstructured.chunking.title import chunk_by_title
-        chunks = chunk_by_title(elements, max_characters=2000, combine_text_under_n_chars=0)
+        chunks = chunk_by_title(elements, max_characters=2000, combine_text_under_n_chars=2000)
         documents = []
         for chunk in chunks:
             text = chunk.text.strip()

api/core/rag/extractor/unstructured/unstructured_msg_extractor.py

@@ -28,7 +28,7 @@ class UnstructuredMsgExtractor(BaseExtractor):
 
         elements = partition_msg(filename=self._file_path, api_url=self._api_url)
         from unstructured.chunking.title import chunk_by_title
-        chunks = chunk_by_title(elements, max_characters=2000, combine_text_under_n_chars=0)
+        chunks = chunk_by_title(elements, max_characters=2000, combine_text_under_n_chars=2000)
         documents = []
         for chunk in chunks:
             text = chunk.text.strip()

api/core/rag/extractor/unstructured/unstructured_text_extractor.py

@@ -28,7 +28,7 @@ class UnstructuredTextExtractor(BaseExtractor):
 
         elements = partition_text(filename=self._file_path, api_url=self._api_url)
         from unstructured.chunking.title import chunk_by_title
-        chunks = chunk_by_title(elements, max_characters=2000, combine_text_under_n_chars=0)
+        chunks = chunk_by_title(elements, max_characters=2000, combine_text_under_n_chars=2000)
         documents = []
         for chunk in chunks:
             text = chunk.text.strip()

api/core/rag/extractor/unstructured/unstructured_xml_extractor.py

@@ -28,7 +28,7 @@ class UnstructuredXmlExtractor(BaseExtractor):
 
         elements = partition_xml(filename=self._file_path, xml_keep_tags=True, api_url=self._api_url)
         from unstructured.chunking.title import chunk_by_title
-        chunks = chunk_by_title(elements, max_characters=2000, combine_text_under_n_chars=0)
+        chunks = chunk_by_title(elements, max_characters=2000, combine_text_under_n_chars=2000)
         documents = []
         for chunk in chunks:
             text = chunk.text.strip()

api/core/tools/provider/builtin/devdocs/_assets/icon.svg

@@ -0,0 +1,4 @@
+<?xml version="1.0" encoding="utf-8"?>
+<svg width="800px" height="800px" viewBox="0 0 24 24" fill="none" xmlns="http://www.w3.org/2000/svg">
+<path fill-rule="evenodd" clip-rule="evenodd" d="M15.6111 1.5837C17.2678 1.34703 18.75 2.63255 18.75 4.30606V5.68256C19.9395 6.31131 20.75 7.56102 20.75 9.00004V19C20.75 21.0711 19.0711 22.75 17 22.75H7C4.92893 22.75 3.25 21.0711 3.25 19V5.00004C3.25 4.99074 3.25017 4.98148 3.2505 4.97227C3.25017 4.95788 3.25 4.94344 3.25 4.92897C3.25 4.02272 3.91638 3.25437 4.81353 3.12621L15.6111 1.5837ZM4.75 6.75004V19C4.75 20.2427 5.75736 21.25 7 21.25H17C18.2426 21.25 19.25 20.2427 19.25 19V9.00004C19.25 7.7574 18.2426 6.75004 17 6.75004H4.75ZM5.07107 5.25004H17.25V4.30606C17.25 3.54537 16.5763 2.96104 15.8232 3.06862L5.02566 4.61113C4.86749 4.63373 4.75 4.76919 4.75 4.92897C4.75 5.10629 4.89375 5.25004 5.07107 5.25004ZM7.25 12C7.25 11.5858 7.58579 11.25 8 11.25H16C16.4142 11.25 16.75 11.5858 16.75 12C16.75 12.4143 16.4142 12.75 16 12.75H8C7.58579 12.75 7.25 12.4143 7.25 12ZM7.25 15.5C7.25 15.0858 7.58579 14.75 8 14.75H13.5C13.9142 14.75 14.25 15.0858 14.25 15.5C14.25 15.9143 13.9142 16.25 13.5 16.25H8C7.58579 16.25 7.25 15.9143 7.25 15.5Z" fill="#1C274D"/>
+</svg>

api/core/tools/provider/builtin/devdocs/devdocs.py

@@ -0,0 +1,21 @@
+from core.tools.errors import ToolProviderCredentialValidationError
+from core.tools.provider.builtin.devdocs.tools.searchDevDocs import SearchDevDocsTool
+from core.tools.provider.builtin_tool_provider import BuiltinToolProviderController
+
+
+class DevDocsProvider(BuiltinToolProviderController):
+    def _validate_credentials(self, credentials: dict) -> None:
+        try:
+            SearchDevDocsTool().fork_tool_runtime(
+                meta={
+                    "credentials": credentials,
+                }
+            ).invoke(
+                user_id='',
+                tool_parameters={
+                    "doc": "python~3.12",
+                    "topic": "library/code",
+                },
+            )
+        except Exception as e:
+            raise ToolProviderCredentialValidationError(str(e))

api/core/tools/provider/builtin/devdocs/devdocs.yaml

@@ -0,0 +1,10 @@
+identity:
+  author: Richards Tu
+  name: devdocs
+  label:
+    en_US: DevDocs
+    zh_Hans: DevDocs
+  description:
+    en_US: Get official developer documentations on DevDocs.
+    zh_Hans: 从DevDocs获取官方开发者文档。
+  icon: icon.svg

api/core/tools/provider/builtin/devdocs/tools/searchDevDocs.py

@@ -0,0 +1,42 @@
+from typing import Any, Union
+
+import requests
+from pydantic import BaseModel, Field
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+
+
+class SearchDevDocsInput(BaseModel):
+    doc: str = Field(..., description="The name of the documentation.")
+    topic: str = Field(..., description="The path of the section/topic.")
+
+
+class SearchDevDocsTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> Union[ToolInvokeMessage, list[ToolInvokeMessage]]:
+        """
+        Invokes the DevDocs search tool with the given user ID and tool parameters.
+
+        Args:
+            user_id (str): The ID of the user invoking the tool.
+            tool_parameters (dict[str, Any]): The parameters for the tool, including 'doc' and 'topic'.
+
+        Returns:
+            ToolInvokeMessage | list[ToolInvokeMessage]: The result of the tool invocation, which can be a single message or a list of messages.
+        """
+        doc = tool_parameters.get('doc', '')
+        topic = tool_parameters.get('topic', '')
+
+        if not doc:
+            return self.create_text_message('Please provide the documentation name.')
+        if not topic:
+            return self.create_text_message('Please provide the topic path.')
+
+        url = f"https://documents.devdocs.io/{doc}/{topic}.html"
+        response = requests.get(url)
+
+        if response.status_code == 200:
+            content = response.text
+            return self.create_text_message(self.summary(user_id=user_id, content=content))
+        else:
+            return self.create_text_message(f"Failed to retrieve the documentation. Status code: {response.status_code}")

api/core/tools/provider/builtin/devdocs/tools/searchDevDocs.yaml

@@ -0,0 +1,34 @@
+identity:
+  name: searchDevDocs
+  author: Richards Tu
+  label:
+    en_US: Search Developer Docs
+    zh_Hans: 搜索开发者文档
+description:
+  human:
+    en_US: A tools for searching for a specific topic and path in DevDocs based on the provided documentation name and topic. Don't for get to add some shots in the system prompt; for example, the documentation name should be like \"vuex~4\", \"css\", or \"python~3.12\", while the topic should be like \"guide/actions\" for Vuex 4, \"display-box\" for CSS, or \"library/code\" for Python 3.12.
+    zh_Hans: 一个用于根据提供的文档名称和主题，在DevDocs中搜索特定主题和路径的工具。不要忘记在系统提示词中添加一些示例；例如，文档名称应该是\"vuex~4\"、\"css\"或\"python~3.12\"，而主题应该是\"guide/actions\"用于Vuex 4，\"display-box\"用于CSS，或\"library/code\"用于Python 3.12。
+  llm: A tools for searching for specific developer documentation in DevDocs based on the provided documentation name and topic.
+parameters:
+  - name: doc
+    type: string
+    required: true
+    label:
+      en_US: Documentation name
+      zh_Hans: 文档名称
+    human_description:
+      en_US: The name of the documentation.
+      zh_Hans: 文档名称。
+    llm_description: The name of the documentation, such as \"vuex~4\", \"css\", or \"python~3.12\". The exact value should be identified by the user.
+    form: llm
+  - name: topic
+    type: string
+    required: true
+    label:
+      en_US: Topic name
+      zh_Hans: 主题名称
+    human_description:
+      en_US: The path of the section/topic.
+      zh_Hans: 文档主题的路径。
+    llm_description: The path of the section/topic, such as \"guide/actions\" for Vuex 4, \"display-box\" for CSS, or \"library/code\" for Python 3.12.
+    form: llm

api/core/tools/provider/builtin/stackexchange/_assets/icon.svg

@@ -0,0 +1 @@
+<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 120 120"><style>.st0{fill:#376db6}.st1{fill:#4ca2da}.st2{fill:#91d8f4}.st3{fill:#1e5397}</style><path class="st0" d="M22.4 57.5h74.8v15.4H22.4z"/><path class="st1" d="M22.4 37.6h74.8V53H22.4z"/><path class="st2" d="M85.5 17H34.4c-6.6 0-12 5.5-12 12.3v4h74.8v-4C97.2 22.5 92 17 85.5 17z"/><path class="st3" d="M22.4 77.3v4c0 6.8 5.4 12.3 12 12.3h32v16.3l15.8-16.3h3.5c6.6 0 12-5.5 12-12.3v-4H22.4z"/></svg>

api/core/tools/provider/builtin/stackexchange/stackexchange.py

@@ -0,0 +1,25 @@
+from core.tools.errors import ToolProviderCredentialValidationError
+from core.tools.provider.builtin.stackexchange.tools.searchStackExQuestions import SearchStackExQuestionsTool
+from core.tools.provider.builtin_tool_provider import BuiltinToolProviderController
+
+
+class StackExchangeProvider(BuiltinToolProviderController):
+    def _validate_credentials(self, credentials: dict) -> None:
+        try:
+            SearchStackExQuestionsTool().fork_tool_runtime(
+                meta={
+                    "credentials": credentials,
+                }
+            ).invoke(
+                user_id='',
+                tool_parameters={
+                    "intitle": "Test",
+                    "sort": "relevance",  
+                    "order": "desc",
+                    "site": "stackoverflow",
+                    "accepted": True,
+                    "pagesize": 1
+                },
+            )
+        except Exception as e:
+            raise ToolProviderCredentialValidationError(str(e))

api/core/tools/provider/builtin/stackexchange/stackexchange.yaml

@@ -0,0 +1,10 @@
+identity:
+  author: Richards Tu
+  name: stackexchange
+  label:
+    en_US: Stack Exchange
+    zh_Hans: Stack Exchange
+  description:
+    en_US: Access questions and answers from the Stack Exchange and its sub-sites.
+    zh_Hans: 从Stack Exchange和其子论坛获取问题和答案。
+  icon: icon.svg

api/core/tools/provider/builtin/stackexchange/tools/fetchAnsByStackExQuesID.py

@@ -0,0 +1,37 @@
+from typing import Any, Union
+
+import requests
+from pydantic import BaseModel, Field
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+
+
+class FetchAnsByStackExQuesIDInput(BaseModel):
+    id: int = Field(..., description="The question ID")
+    site: str = Field(..., description="The Stack Exchange site")
+    order: str = Field(..., description="asc or desc")
+    sort: str = Field(..., description="activity, votes, creation")
+    pagesize: int = Field(..., description="Number of answers per page")
+    page: int = Field(..., description="Page number")
+
+
+class FetchAnsByStackExQuesIDTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> Union[ToolInvokeMessage, list[ToolInvokeMessage]]:
+        input = FetchAnsByStackExQuesIDInput(**tool_parameters)
+
+        params = {
+            "site": input.site,
+            "filter": "!nNPvSNdWme",
+            "order": input.order,
+            "sort": input.sort,
+            "pagesize": input.pagesize,
+            "page": input.page
+        }
+
+        response = requests.get(f"https://api.stackexchange.com/2.3/questions/{input.id}/answers", params=params)
+
+        if response.status_code == 200:
+            return self.create_text_message(self.summary(user_id=user_id, content=response.text))
+        else:
+            return self.create_text_message(f"API request failed with status code {response.status_code}")

api/core/tools/provider/builtin/stackexchange/tools/fetchAnsByStackExQuesID.yaml

@@ -0,0 +1,189 @@
+identity:
+  name: fetchAnsByStackExQuesID
+  author: Richards Tu
+  label:
+    en_US: Fetch Stack Exchange Answers
+    zh_Hans: 获取 Stack Exchange 答案
+description:
+  human:
+    en_US: A tool for retrieving answers for a specific Stack Exchange question ID. Specify the question ID, Stack Exchange site, sorting order, number of results per page, and page number. Must be used with the searchStackExQuesID tool.
+    zh_Hans: 用于检索特定Stack Exchange问题ID的答案的工具。指定问题ID、Stack Exchange站点、排序顺序、每页结果数和页码。 必须与searchStackExQuesID工具一起使用。
+  llm: A tool for retrieving answers for a specific Stack Exchange question ID based on the provided parameters.
+parameters:
+  - name: id
+    type: string
+    required: true
+    label:
+      en_US: Question ID
+      zh_Hans: 问题ID
+    human_description:
+      en_US: The ID of the Stack Exchange question to fetch answers for.
+      zh_Hans: 要获取答案的Stack Exchange问题的ID。
+    llm_description: The ID of the Stack Exchange question.
+    form: llm
+  - name: site
+    type: string
+    required: true
+    label:
+      en_US: Stack Exchange site
+      zh_Hans: Stack Exchange站点
+    human_description:
+      en_US: The Stack Exchange site the question is from, e.g. stackoverflow, unix, etc.
+      zh_Hans: 问题所在的Stack Exchange站点，例如stackoverflow、unix等。
+    llm_description: The Stack Exchange site identifier.
+    options:
+      - value: stackoverflow
+        label:
+          en_US: stackoverflow
+      - value: serverfault
+        label:
+          en_US: serverfault
+      - value: superuser
+        label:
+          en_US: superuser
+      - value: askubuntu
+        label:
+          en_US: askubuntu
+      - value: unix
+        label:
+          en_US: unix
+      - value: cs
+        label:
+          en_US: cs
+      - value: softwareengineering
+        label:
+          en_US: softwareengineering
+      - value: codegolf
+        label:
+          en_US: codegolf
+      - value: codereview
+        label:
+          en_US: codereview
+      - value: cstheory
+        label:
+          en_US: cstheory
+      - value: security
+        label:
+          en_US: security
+      - value: cryptography
+        label:
+          en_US: cryptography
+      - value: reverseengineering
+        label:
+          en_US: reverseengineering
+      - value: datascience
+        label:
+          en_US: datascience
+      - value: devops
+        label:
+          en_US: devops
+      - value: ux
+        label:
+          en_US: ux
+      - value: dba
+        label:
+          en_US: dba
+      - value: gis
+        label:
+          en_US: gis
+      - value: webmasters
+        label:
+          en_US: webmasters
+      - value: arduino
+        label:
+          en_US: arduino
+      - value: raspberrypi
+        label:
+          en_US: raspberrypi
+      - value: networkengineering
+        label:
+          en_US: networkengineering
+      - value: iot
+        label:
+          en_US: iot
+      - value: tor
+        label:
+          en_US: tor
+      - value: sqa
+        label:
+          en_US: sqa
+      - value: mathoverflow
+        label:
+          en_US: mathoverflow
+      - value: math
+        label:
+          en_US: math
+      - value: mathematica
+        label:
+          en_US: mathematica
+      - value: dsp
+        label:
+          en_US: dsp
+      - value: gamedev
+        label:
+          en_US: gamedev
+      - value: robotics
+        label:
+          en_US: robotics
+      - value: genai
+        label:
+          en_US: genai
+      - value: computergraphics
+        label:
+          en_US: computergraphics
+    form: form
+  - name: filter
+    type: string
+    required: true
+    label:
+      en_US: Filter
+      zh_Hans: 过滤器
+    human_description:
+      en_US: This is required in order to actually get the body of the answer.
+      zh_Hans: 为了实际获取答案的正文，这是必需的。
+    llm_description: Required in order to actually get the body of the answer. Must be \"!nNPvSNdWme\".
+    form: llm
+  - name: order
+    type: string
+    required: true
+    label:
+      en_US: Sort direction
+      zh_Hans: 排序方向
+    human_description:
+      en_US: The direction to sort the answers - ascending or descending.
+      zh_Hans: 答案的排序方向 - 升序或降序。
+    llm_description: asc for ascending, desc for descending.
+    form: llm
+  - name: sort
+    type: string
+    required: true
+    label:
+      en_US: Sort order
+      zh_Hans: 排序
+    human_description:
+      en_US: The sort order for the answers - activity, votes, or creation date.
+      zh_Hans: 答案的排序顺序 - 活动、投票或创建日期。
+    llm_description: activity, votes, or creation.
+    form: llm
+  - name: pagesize
+    type: number
+    required: true
+    label:
+      en_US: Results per page
+      zh_Hans: 每页结果数
+    human_description:
+      en_US: The number of answers to return per page.
+      zh_Hans: 每页返回的答案数。
+    llm_description: The number of answers per page.
+    form: llm
+  - name: page
+    type: number
+    required: true
+    label:
+      en_US: Page number
+      zh_Hans: 页码
+    human_description:
+      en_US: The page number of answers to retrieve.
+      zh_Hans: 要检索的答案的页码。
+    llm_description: The page number to retrieve.
+    form: llm

api/core/tools/provider/builtin/stackexchange/tools/searchStackExQuestions.py

@@ -0,0 +1,43 @@
+from typing import Any, Union
+
+import requests
+from pydantic import BaseModel, Field
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+
+
+class SearchStackExQuestionsInput(BaseModel):
+    intitle: str = Field(..., description="The search query.")
+    sort: str = Field(..., description="The sort order - relevance, activity, votes, creation.")  
+    order: str = Field(..., description="asc or desc")
+    site: str = Field(..., description="The Stack Exchange site.")
+    tagged: str = Field(None, description="Semicolon-separated tags to include.")
+    nottagged: str = Field(None, description="Semicolon-separated tags to exclude.")
+    accepted: bool = Field(..., description="true for only accepted answers, false otherwise") 
+    pagesize: int = Field(..., description="Number of results per page")
+
+
+class SearchStackExQuestionsTool(BuiltinTool):
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> Union[ToolInvokeMessage, list[ToolInvokeMessage]]:
+        input = SearchStackExQuestionsInput(**tool_parameters)
+
+        params = {
+            "intitle": input.intitle,
+            "sort": input.sort,
+            "order": input.order,  
+            "site": input.site,
+            "accepted": input.accepted,
+            "pagesize": input.pagesize
+        }
+        if input.tagged:
+            params["tagged"] = input.tagged
+        if input.nottagged:
+            params["nottagged"] = input.nottagged
+
+        response = requests.get("https://api.stackexchange.com/2.3/search", params=params)
+
+        if response.status_code == 200:
+            return self.create_text_message(self.summary(user_id=user_id, content=response.text))
+        else:
+            return self.create_text_message(f"API request failed with status code {response.status_code}")

api/core/tools/provider/builtin/stackexchange/tools/searchStackExQuestions.yaml

@@ -0,0 +1,200 @@
+identity:
+  name: searchStackExQuestions
+  author: Richards Tu
+  label:
+    en_US: Search Stack Exchange Questions
+    zh_Hans: 搜索Stack Exchange问题
+description:
+  human:
+    en_US: A tool for searching questions on a Stack Exchange site. Specify the search query, sorting order, tags to include or exclude, whether to search only for questions with accepted answers, the Stack Exchange site, and number of results per page.
+    zh_Hans: 在Stack Exchange站点上搜索问题的工具。指定搜索查询、排序顺序、要包含或排除的标签、是否仅搜索有已接受答案的问题、Stack Exchange站点以及每页结果数。
+  llm: A tool for searching questions on a Stack Exchange site based on the provided parameters.
+parameters:
+  - name: intitle
+    type: string
+    required: true
+    label:
+      en_US: Search query
+      zh_Hans: 搜索查询
+    human_description:
+      en_US: The search query to use for finding questions.
+      zh_Hans: 用于查找问题的搜索查询。
+    llm_description: The search query to use.
+    form: llm
+  - name: sort
+    type: string
+    required: true
+    label:
+      en_US: Sort order
+      zh_Hans: 排序
+    human_description:
+      en_US: The sort order for the search results - relevance, activity, votes, or creation date.
+      zh_Hans: 搜索结果的排序顺序 - 相关性、活动、投票或创建日期。
+    llm_description: The sort order - relevance, activity, votes, or creation.
+    form: llm
+  - name: order
+    type: string
+    required: true
+    label:
+      en_US: Sort direction
+      zh_Hans: 排序方向
+    human_description:
+      en_US: The direction to sort - ascending or descending.
+      zh_Hans: 排序方向 - 升序或降序。
+    llm_description: asc for ascending, desc for descending.
+    form: llm
+  - name: site
+    type: string
+    required: true
+    label:
+      en_US: Stack Exchange site
+      zh_Hans: Stack Exchange 站点
+    human_description:
+      en_US: The Stack Exchange site to search, e.g. stackoverflow, unix, etc.
+      zh_Hans: 要搜索的Stack Exchange站点，例如stackoverflow、unix等。
+    llm_description: The Stack Exchange site identifier.
+    options:
+      - value: stackoverflow
+        label:
+          en_US: stackoverflow
+      - value: serverfault
+        label:
+          en_US: serverfault
+      - value: superuser
+        label:
+          en_US: superuser
+      - value: askubuntu
+        label:
+          en_US: askubuntu
+      - value: unix
+        label:
+          en_US: unix
+      - value: cs
+        label:
+          en_US: cs
+      - value: softwareengineering
+        label:
+          en_US: softwareengineering
+      - value: codegolf
+        label:
+          en_US: codegolf
+      - value: codereview
+        label:
+          en_US: codereview
+      - value: cstheory
+        label:
+          en_US: cstheory
+      - value: security
+        label:
+          en_US: security
+      - value: cryptography
+        label:
+          en_US: cryptography
+      - value: reverseengineering
+        label:
+          en_US: reverseengineering
+      - value: datascience
+        label:
+          en_US: datascience
+      - value: devops
+        label:
+          en_US: devops
+      - value: ux
+        label:
+          en_US: ux
+      - value: dba
+        label:
+          en_US: dba
+      - value: gis
+        label:
+          en_US: gis
+      - value: webmasters
+        label:
+          en_US: webmasters
+      - value: arduino
+        label:
+          en_US: arduino
+      - value: raspberrypi
+        label:
+          en_US: raspberrypi
+      - value: networkengineering
+        label:
+          en_US: networkengineering
+      - value: iot
+        label:
+          en_US: iot
+      - value: tor
+        label:
+          en_US: tor
+      - value: sqa
+        label:
+          en_US: sqa
+      - value: mathoverflow
+        label:
+          en_US: mathoverflow
+      - value: math
+        label:
+          en_US: math
+      - value: mathematica
+        label:
+          en_US: mathematica
+      - value: dsp
+        label:
+          en_US: dsp
+      - value: gamedev
+        label:
+          en_US: gamedev
+      - value: robotics
+        label:
+          en_US: robotics
+      - value: genai
+        label:
+          en_US: genai
+      - value: computergraphics
+        label:
+          en_US: computergraphics
+    form: form
+  - name: tagged
+    type: string
+    required: false
+    label:
+      en_US: Include tags
+      zh_Hans: 包含标签
+    human_description:
+      en_US: A semicolon-separated list of tags that questions must have.
+      zh_Hans: 问题必须具有的标签的分号分隔列表。
+    llm_description: Semicolon-separated tags to include. Leave blank if not needed.
+    form: llm
+  - name: nottagged
+    type: string
+    required: false
+    label:
+      en_US: Exclude tags
+      zh_Hans: 排除标签
+    human_description:
+      en_US: A semicolon-separated list of tags to exclude from the search.
+      zh_Hans: 从搜索中排除的标签的分号分隔列表。
+    llm_description: Semicolon-separated tags to exclude. Leave blank if not needed.
+    form: llm
+  - name: accepted
+    type: boolean
+    required: true
+    label:
+      en_US: Has accepted answer
+      zh_Hans: 有已接受的答案
+    human_description:
+      en_US: Whether to limit to only questions that have an accepted answer.
+      zh_Hans: 是否限制为只有已接受答案的问题。
+    llm_description: true to limit to only questions with accepted answers, false otherwise.
+    form: llm
+  - name: pagesize
+    type: number
+    required: true
+    label:
+      en_US: Results per page
+      zh_Hans: 每页结果数
+    human_description:
+      en_US: The number of results to return per page.
+      zh_Hans: 每页返回的结果数。
+    llm_description: The number of results per page.
+    form: llm

api/core/tools/provider/builtin_tool_provider.py

@@ -1,4 +1,3 @@
-import importlib
 from abc import abstractmethod
 from os import listdir, path
 from typing import Any
@@ -16,6 +15,7 @@ from core.tools.errors import (
 from core.tools.provider.tool_provider import ToolProviderController
 from core.tools.tool.builtin_tool import BuiltinTool
 from core.tools.tool.tool import Tool
+from core.utils.module_import_helper import load_single_subclass_from_source
 
 
 class BuiltinToolProviderController(ToolProviderController):
@@ -63,16 +63,11 @@ class BuiltinToolProviderController(ToolProviderController):
                 tool_name = tool_file.split(".")[0]
                 tool = load(f.read(), FullLoader)
                 # get tool class, import the module
-                py_path = path.join(path.dirname(path.realpath(__file__)), 'builtin', provider, 'tools', f'{tool_name}.py')
-                spec = importlib.util.spec_from_file_location(f'core.tools.provider.builtin.{provider}.tools.{tool_name}', py_path)
-                mod = importlib.util.module_from_spec(spec)
-                spec.loader.exec_module(mod)
-
-                # get all the classes in the module
-                classes = [x for _, x in vars(mod).items() 
-                    if isinstance(x, type) and x not in [BuiltinTool, Tool] and issubclass(x, BuiltinTool)
-                ]
-                assistant_tool_class = classes[0]
+                assistant_tool_class = load_single_subclass_from_source(
+                    module_name=f'core.tools.provider.builtin.{provider}.tools.{tool_name}',
+                    script_path=path.join(path.dirname(path.realpath(__file__)),
+                                           'builtin', provider, 'tools', f'{tool_name}.py'),
+                    parent_type=BuiltinTool)
                 tools.append(assistant_tool_class(**tool))
 
         self.tools = tools

api/core/tools/tool/api_tool.py

@@ -148,7 +148,7 @@ class ApiTool(Tool):
                 value = ''
                 if parameter['name'] in parameters:
                     value = parameters[parameter['name']]
-                elif parameter['required']:
+                elif parameter.get('required', False):
                     raise ToolParameterValidationError(f"Missing required parameter {parameter['name']}")
                 else:
                     value = (parameter.get('schema', {}) or {}).get('default', '')
@@ -158,7 +158,7 @@ class ApiTool(Tool):
                 value = ''
                 if parameter['name'] in parameters:
                     value = parameters[parameter['name']]
-                elif parameter['required']:
+                elif parameter.get('required', False):
                     raise ToolParameterValidationError(f"Missing required parameter {parameter['name']}")
                 else:
                     value = (parameter.get('schema', {}) or {}).get('default', '')
@@ -168,7 +168,7 @@ class ApiTool(Tool):
                 value = ''
                 if parameter['name'] in parameters:
                     value = parameters[parameter['name']]
-                elif parameter['required']:
+                elif parameter.get('required', False):
                     raise ToolParameterValidationError(f"Missing required parameter {parameter['name']}")
                 else:
                     value = (parameter.get('schema', {}) or {}).get('default', '')

api/core/tools/tool/tool.py

@@ -232,7 +232,7 @@ class Tool(BaseModel, ABC):
                 result += f"result link: {response.message}. please tell user to check it."
             elif response.type == ToolInvokeMessage.MessageType.IMAGE_LINK or \
                  response.type == ToolInvokeMessage.MessageType.IMAGE:
-                result += "image has been created and sent to user already, you should tell user to check it now."
+                result += "image has been created and sent to user already, you do not need to create it, just tell the user to check it now."
             elif response.type == ToolInvokeMessage.MessageType.BLOB:
                 if len(response.message) > 114:
                     result += str(response.message[:114]) + '...'

api/core/tools/tool_manager.py

@@ -1,4 +1,3 @@
-import importlib
 import json
 import logging
 import mimetypes
@@ -34,6 +33,7 @@ from core.tools.utils.configuration import (
     ToolParameterConfigurationManager,
 )
 from core.tools.utils.encoder import serialize_base_model_dict
+from core.utils.module_import_helper import load_single_subclass_from_source
 from extensions.ext_database import db
 from models.tools import ApiToolProvider, BuiltinToolProvider
 
@@ -72,21 +72,11 @@ class ToolManager:
 
         if provider_entity is None:
             # fetch the provider from .provider.builtin
-            py_path = path.join(path.dirname(path.realpath(__file__)), 'builtin', provider, f'{provider}.py')
-            spec = importlib.util.spec_from_file_location(f'core.tools.provider.builtin.{provider}.{provider}', py_path)
-            mod = importlib.util.module_from_spec(spec)
-            spec.loader.exec_module(mod)
-
-            # get all the classes in the module
-            classes = [ x for _, x in vars(mod).items() 
-                       if isinstance(x, type) and x != ToolProviderController and issubclass(x, ToolProviderController)
-            ]
-            if len(classes) == 0:
-                raise ToolProviderNotFoundError(f'provider {provider} not found')
-            if len(classes) > 1:
-                raise ToolProviderNotFoundError(f'multiple providers found for {provider}')
-            
-            provider_entity = classes[0]()
+            provider_class = load_single_subclass_from_source(
+                module_name=f'core.tools.provider.builtin.{provider}.{provider}',
+                script_path=path.join(path.dirname(path.realpath(__file__)), 'builtin', provider, f'{provider}.py'),
+                parent_type=ToolProviderController)
+            provider_entity = provider_class()
 
         return provider_entity.invoke(tool_id, tool_name, tool_parameters, credentials, prompt_messages)
     
@@ -330,23 +320,12 @@ class ToolManager:
                 if provider.startswith('__'):
                     continue
 
-                py_path = path.join(path.dirname(path.realpath(__file__)), 'provider', 'builtin', provider, f'{provider}.py')
-                spec = importlib.util.spec_from_file_location(f'core.tools.provider.builtin.{provider}.{provider}', py_path)
-                mod = importlib.util.module_from_spec(spec)
-                spec.loader.exec_module(mod)
-
-                # load all classes
-                classes = [
-                    obj for name, obj in vars(mod).items() 
-                        if isinstance(obj, type) and obj != BuiltinToolProviderController and issubclass(obj, BuiltinToolProviderController)
-                ]
-                if len(classes) == 0:
-                    raise ToolProviderNotFoundError(f'provider {provider} not found')
-                if len(classes) > 1:
-                    raise ToolProviderNotFoundError(f'multiple providers found for {provider}')
-                
                 # init provider
-                provider_class = classes[0]
+                provider_class = load_single_subclass_from_source(
+                    module_name=f'core.tools.provider.builtin.{provider}.{provider}',
+                    script_path=path.join(path.dirname(path.realpath(__file__)),
+                                           'provider', 'builtin', provider, f'{provider}.py'),
+                    parent_type=BuiltinToolProviderController)
                 builtin_providers.append(provider_class())
 
         # cache the builtin providers

api/core/utils/module_import_helper.py

@@ -0,0 +1,62 @@
+import importlib.util
+import logging
+import sys
+from types import ModuleType
+from typing import AnyStr
+
+
+def import_module_from_source(
+        module_name: str,
+        py_file_path: AnyStr,
+        use_lazy_loader: bool = False
+) -> ModuleType:
+    """
+    Importing a module from the source file directly
+    """
+    try:
+        existed_spec = importlib.util.find_spec(module_name)
+        if existed_spec:
+            spec = existed_spec
+        else:
+            # Refer to: https://docs.python.org/3/library/importlib.html#importing-a-source-file-directly
+            spec = importlib.util.spec_from_file_location(module_name, py_file_path)
+            if use_lazy_loader:
+                # Refer to: https://docs.python.org/3/library/importlib.html#implementing-lazy-imports
+                spec.loader = importlib.util.LazyLoader(spec.loader)
+        module = importlib.util.module_from_spec(spec)
+        if not existed_spec:
+            sys.modules[module_name] = module
+        spec.loader.exec_module(module)
+        return module
+    except Exception as e:
+        logging.exception(f'Failed to load module {module_name} from {py_file_path}: {str(e)}')
+        raise e
+
+
+def get_subclasses_from_module(mod: ModuleType, parent_type: type) -> list[type]:
+    """
+    Get all the subclasses of the parent type from the module
+    """
+    classes = [x for _, x in vars(mod).items()
+               if isinstance(x, type) and x != parent_type and issubclass(x, parent_type)]
+    return classes
+
+
+def load_single_subclass_from_source(
+        module_name: str,
+        script_path: AnyStr,
+        parent_type: type,
+        use_lazy_loader: bool = False,
+) -> type:
+    """
+    Load a single subclass from the source
+    """
+    module = import_module_from_source(module_name, script_path, use_lazy_loader)
+    subclasses = get_subclasses_from_module(module, parent_type)
+    match len(subclasses):
+        case 1:
+            return subclasses[0]
+        case 0:
+            raise Exception(f'Missing subclass of {parent_type.__name__} in {script_path}')
+        case _:
+            raise Exception(f'Multiple subclasses of {parent_type.__name__} in {script_path}')

api/extensions/ext_compress.py

@@ -5,6 +5,12 @@ def init_app(app: Flask):
     if app.config.get('API_COMPRESSION_ENABLED', False):
         from flask_compress import Compress
 
+        app.config['COMPRESS_MIMETYPES'] = [
+            'application/json',
+            'image/svg+xml',
+            'text/html',
+        ]
+
         compress = Compress()
         compress.init_app(app)
 

api/extensions/ext_storage.py

@@ -2,9 +2,12 @@ import os
 import shutil
 from collections.abc import Generator
 from contextlib import closing
+from datetime import datetime, timedelta
 from typing import Union
 
 import boto3
+from azure.storage.blob import AccountSasPermissions, BlobServiceClient, ResourceTypes, generate_account_sas
+from botocore.client import Config
 from botocore.exceptions import ClientError
 from flask import Flask
 
@@ -25,8 +28,21 @@ class Storage:
                 aws_secret_access_key=app.config.get('S3_SECRET_KEY'),
                 aws_access_key_id=app.config.get('S3_ACCESS_KEY'),
                 endpoint_url=app.config.get('S3_ENDPOINT'),
-                region_name=app.config.get('S3_REGION')
+                region_name=app.config.get('S3_REGION'),
+                config=Config(s3={'addressing_style': app.config.get('S3_ADDRESS_STYLE')})
             )
+        elif self.storage_type == 'azure-blob':
+            self.bucket_name = app.config.get('AZURE_BLOB_CONTAINER_NAME')
+            sas_token = generate_account_sas(
+                account_name=app.config.get('AZURE_BLOB_ACCOUNT_NAME'),
+                account_key=app.config.get('AZURE_BLOB_ACCOUNT_KEY'),
+                resource_types=ResourceTypes(service=True, container=True, object=True),
+                permission=AccountSasPermissions(read=True, write=True, delete=True, list=True, add=True, create=True),
+                expiry=datetime.utcnow() + timedelta(hours=1)
+            )
+            self.client = BlobServiceClient(account_url=app.config.get('AZURE_BLOB_ACCOUNT_URL'),
+                                            credential=sas_token)
+
         else:
             self.folder = app.config.get('STORAGE_LOCAL_PATH')
             if not os.path.isabs(self.folder):
@@ -35,6 +51,9 @@ class Storage:
     def save(self, filename, data):
         if self.storage_type == 's3':
             self.client.put_object(Bucket=self.bucket_name, Key=filename, Body=data)
+        elif self.storage_type == 'azure-blob':
+            blob_container = self.client.get_container_client(container=self.bucket_name)
+            blob_container.upload_blob(filename, data)
         else:
             if not self.folder or self.folder.endswith('/'):
                 filename = self.folder + filename
@@ -63,6 +82,10 @@ class Storage:
                     raise FileNotFoundError("File not found")
                 else:
                     raise
+        elif self.storage_type == 'azure-blob':
+            blob = self.client.get_container_client(container=self.bucket_name)
+            blob = blob.get_blob_client(blob=filename)
+            data = blob.download_blob().readall()
         else:
             if not self.folder or self.folder.endswith('/'):
                 filename = self.folder + filename
@@ -90,6 +113,11 @@ class Storage:
                         raise FileNotFoundError("File not found")
                     else:
                         raise
+            elif self.storage_type == 'azure-blob':
+                blob = self.client.get_blob_client(container=self.bucket_name, blob=filename)
+                with closing(blob.download_blob()) as blob_stream:
+                    while chunk := blob_stream.readall(4096):
+                        yield chunk
             else:
                 if not self.folder or self.folder.endswith('/'):
                     filename = self.folder + filename
@@ -109,6 +137,11 @@ class Storage:
         if self.storage_type == 's3':
             with closing(self.client) as client:
                 client.download_file(self.bucket_name, filename, target_filepath)
+        elif self.storage_type == 'azure-blob':
+            blob = self.client.get_blob_client(container=self.bucket_name, blob=filename)
+            with open(target_filepath, "wb") as my_blob:
+                blob_data = blob.download_blob()
+                blob_data.readinto(my_blob)
         else:
             if not self.folder or self.folder.endswith('/'):
                 filename = self.folder + filename
@@ -128,6 +161,9 @@ class Storage:
                     return True
                 except:
                     return False
+        elif self.storage_type == 'azure-blob':
+            blob = self.client.get_blob_client(container=self.bucket_name, blob=filename)
+            return blob.exists()
         else:
             if not self.folder or self.folder.endswith('/'):
                 filename = self.folder + filename

api/libs/login.py

@@ -53,7 +53,7 @@ def login_required(func):
     def decorated_view(*args, **kwargs):
         auth_header = request.headers.get('Authorization')
         admin_api_key_enable = os.getenv('ADMIN_API_KEY_ENABLE', default='False')
-        if admin_api_key_enable:
+        if admin_api_key_enable.lower() == 'true':
             if auth_header:
                 if ' ' not in auth_header:
                     raise Unauthorized('Invalid Authorization header format. Expected \'Bearer <api-key>\' format.')

api/libs/smtp.py

@@ -16,7 +16,7 @@ class SMTPClient:
         smtp = smtplib.SMTP(self.server, self.port)
         if self._use_tls:
             smtp.starttls()
-        if (self.username):
+        if self.username and self.password:
             smtp.login(self.username, self.password)
         msg = MIMEMultipart()
         msg['Subject'] = mail['subject']

api/requirements.txt

@@ -28,8 +28,8 @@ scikit-learn==1.2.2
 sentry-sdk[flask]~=1.39.2
 sympy==1.12
 jieba==0.42.1
-celery==5.2.7
-redis~=4.5.4
+celery~=5.3.6
+redis[hiredis]~=5.0.3
 openpyxl==3.1.2
 chardet~=5.1.0
 python-docx~=1.1.0
@@ -67,8 +67,10 @@ yfinance~=0.2.35
 pydub~=0.25.1
 gmpy2~=2.1.5
 numexpr~=2.9.0
-duckduckgo-search==4.4.3
+duckduckgo-search==5.1.0
 arxiv==2.1.0
 yarl~=1.9.4
 twilio==9.0.0
 qrcode~=7.4.2
+azure-storage-blob==12.9.0
+azure-identity==1.15.0

api/services/account_service.py

@@ -435,11 +435,13 @@ class RegisterService:
 
             if open_id is not None or provider is not None:
                 AccountService.link_account_integrate(provider, open_id, account)
+            if current_app.config['EDITION'] != 'SELF_HOSTED':
+                tenant = TenantService.create_tenant(f"{account.name}'s Workspace")
 
-            tenant = TenantService.create_tenant(f"{account.name}'s Workspace")
+                TenantService.create_tenant_member(tenant, account, role='owner')
+                account.current_tenant = tenant
 
-            TenantService.create_tenant_member(tenant, account, role='owner')
-            account.current_tenant = tenant
+                tenant_was_created.send(tenant)
 
             db.session.commit()
         except Exception as e:
@@ -447,8 +449,6 @@ class RegisterService:
             logging.error(f'Register failed: {e}')
             raise AccountRegisterError(f'Registration failed: {e}') from e
 
-        tenant_was_created.send(tenant)
-
         return account
 
     @classmethod
@@ -461,7 +461,6 @@ class RegisterService:
             name = email.split('@')[0]
 
             account = cls.register(email=email, name=name, language=language, status=AccountStatus.PENDING)
-
             # Create new tenant member for invited tenant
             TenantService.create_tenant_member(tenant, account, role)
             TenantService.switch_tenant(account, tenant.id)

api/tasks/annotation/disable_annotation_reply_task.py

@@ -9,7 +9,7 @@ from core.rag.datasource.vdb.vector_factory import Vector
 from extensions.ext_database import db
 from extensions.ext_redis import redis_client
 from models.dataset import Dataset
-from models.model import App, AppAnnotationSetting
+from models.model import App, AppAnnotationSetting, MessageAnnotation
 
 
 @shared_task(queue='dataset')
@@ -25,7 +25,7 @@ def disable_annotation_reply_task(job_id: str, app_id: str, tenant_id: str):
         App.tenant_id == tenant_id,
         App.status == 'normal'
     ).first()
-
+    annotations_count = db.session.query(MessageAnnotation).filter(MessageAnnotation.app_id == app_id).count()
     if not app:
         raise NotFound("App not found")
 
@@ -49,8 +49,9 @@ def disable_annotation_reply_task(job_id: str, app_id: str, tenant_id: str):
         )
 
         try:
-            vector = Vector(dataset, attributes=['doc_id', 'annotation_id', 'app_id'])
-            vector.delete_by_metadata_field('app_id', app_id)
+            if annotations_count > 0:
+                vector = Vector(dataset, attributes=['doc_id', 'annotation_id', 'app_id'])
+                vector.delete_by_metadata_field('app_id', app_id)
         except Exception:
             logging.exception("Delete annotation index failed when annotation deleted.")
         redis_client.setex(disable_app_annotation_job_key, 600, 'completed')

api/tests/integration_tests/model_runtime/wenxin/test_llm.py

@@ -154,7 +154,7 @@ def test_invoke_stream_model():
     model = ErnieBotLargeLanguageModel()
 
     response = model.invoke(
-        model='ernie-bot',
+        model='ernie-3.5-8k',
         credentials={
             'api_key': os.environ.get('WENXIN_API_KEY'),
             'secret_key': os.environ.get('WENXIN_SECRET_KEY')

api/tests/integration_tests/utils/child_class.py

@@ -0,0 +1,7 @@
+from tests.integration_tests.utils.parent_class import ParentClass
+
+
+class ChildClass(ParentClass):
+    def __init__(self, name: str):
+        super().__init__(name)
+        self.name = name

api/tests/integration_tests/utils/lazy_load_class.py

@@ -0,0 +1,7 @@
+from tests.integration_tests.utils.parent_class import ParentClass
+
+
+class LazyLoadChildClass(ParentClass):
+    def __init__(self, name: str):
+        super().__init__(name)
+        self.name = name

api/tests/integration_tests/utils/parent_class.py

@@ -0,0 +1,6 @@
+class ParentClass:
+    def __init__(self, name):
+        self.name = name
+
+    def get_name(self):
+        return self.name

api/tests/integration_tests/utils/test_module_import_helper.py

@@ -0,0 +1,32 @@
+import os
+
+from core.utils.module_import_helper import load_single_subclass_from_source, import_module_from_source
+from tests.integration_tests.utils.parent_class import ParentClass
+
+
+def test_loading_subclass_from_source():
+    current_path = os.getcwd()
+    module = load_single_subclass_from_source(
+        module_name='ChildClass',
+        script_path=os.path.join(current_path, 'child_class.py'),
+        parent_type=ParentClass)
+    assert module and module.__name__ == 'ChildClass'
+
+
+def test_load_import_module_from_source():
+    current_path = os.getcwd()
+    module = import_module_from_source(
+        module_name='ChildClass',
+        py_file_path=os.path.join(current_path, 'child_class.py'))
+    assert module and module.__name__ == 'ChildClass'
+
+
+def test_lazy_loading_subclass_from_source():
+    current_path = os.getcwd()
+    clz = load_single_subclass_from_source(
+        module_name='LazyLoadChildClass',
+        script_path=os.path.join(current_path, 'lazy_load_class.py'),
+        parent_type=ParentClass,
+        use_lazy_loader=True)
+    instance = clz('dify')
+    assert instance.get_name() == 'dify'

docker/docker-compose.middleware.yaml

@@ -1,4 +1,4 @@
-version: '3.1'
+version: '3'
 services:
   # The postgres database.
   db:

docker/docker-compose.yaml

@@ -1,8 +1,8 @@
-version: '3.1'
+version: '3'
 services:
   # API service
   api:
-    image: langgenius/dify-api:0.5.10
+    image: langgenius/dify-api:0.5.11
     restart: always
     environment:
       # Startup mode, 'api' starts the API server.
@@ -70,7 +70,7 @@ services:
       # If you want to enable cross-origin support,
       # you must use the HTTPS protocol and set the configuration to `SameSite=None, Secure=true, HttpOnly=true`.
       #
-      # The type of storage to use for storing user files. Supported values are `local` and `s3`, Default: `local`
+      # The type of storage to use for storing user files. Supported values are `local` and `s3` and `azure-blob`, Default: `local`
       STORAGE_TYPE: local
       # The path to the local storage directory, the directory relative the root path of API service codes or absolute path. Default: `storage` or `/home/john/storage`.
       # only available when STORAGE_TYPE is `local`.
@@ -81,6 +81,11 @@ services:
       S3_ACCESS_KEY: 'ak-difyai'
       S3_SECRET_KEY: 'sk-difyai'
       S3_REGION: 'us-east-1'
+      # The Azure Blob storage configurations, only available when STORAGE_TYPE is `azure-blob`.
+      AZURE_BLOB_ACCOUNT_NAME: 'difyai'
+      AZURE_BLOB_ACCOUNT_KEY: 'difyai'
+      AZURE_BLOB_CONTAINER_NAME: 'difyai-container'
+      AZURE_BLOB_ACCOUNT_URL: 'https://<your_account_name>.blob.core.windows.net'
       # The type of vector store to use. Supported values are `weaviate`, `qdrant`, `milvus`.
       VECTOR_STORE: weaviate
       # The Weaviate endpoint URL. Only available when VECTOR_STORE is `weaviate`.
@@ -135,7 +140,7 @@ services:
   # worker service
   # The Celery worker for processing the queue.
   worker:
-    image: langgenius/dify-api:0.5.10
+    image: langgenius/dify-api:0.5.11
     restart: always
     environment:
       # Startup mode, 'worker' starts the Celery worker for processing the queue.
@@ -164,9 +169,20 @@ services:
       REDIS_USE_SSL: 'false'
       # The configurations of celery broker.
       CELERY_BROKER_URL: redis://:difyai123456@redis:6379/1
-      # The type of storage to use for storing user files. Supported values are `local` and `s3`, Default: `local`
+      # The type of storage to use for storing user files. Supported values are `local` and `s3` and `azure-blob`, Default: `local`
       STORAGE_TYPE: local
       STORAGE_LOCAL_PATH: storage
+      # The S3 storage configurations, only available when STORAGE_TYPE is `s3`.
+      S3_ENDPOINT: 'https://xxx.r2.cloudflarestorage.com'
+      S3_BUCKET_NAME: 'difyai'
+      S3_ACCESS_KEY: 'ak-difyai'
+      S3_SECRET_KEY: 'sk-difyai'
+      S3_REGION: 'us-east-1'
+      # The Azure Blob storage configurations, only available when STORAGE_TYPE is `azure-blob`.
+      AZURE_BLOB_ACCOUNT_NAME: 'difyai'
+      AZURE_BLOB_ACCOUNT_KEY: 'difyai'
+      AZURE_BLOB_CONTAINER_NAME: 'difyai-container'
+      AZURE_BLOB_ACCOUNT_URL: 'https://<your_account_name>.blob.core.windows.net'
       # The type of vector store to use. Supported values are `weaviate`, `qdrant`, `milvus`.
       VECTOR_STORE: weaviate
       # The Weaviate endpoint URL. Only available when VECTOR_STORE is `weaviate`.
@@ -206,7 +222,7 @@ services:
 
   # Frontend web application.
   web:
-    image: langgenius/dify-web:0.5.10
+    image: langgenius/dify-web:0.5.11
     restart: always
     environment:
       EDITION: SELF_HOSTED
@@ -310,8 +326,10 @@ services:
       - ./nginx/nginx.conf:/etc/nginx/nginx.conf
       - ./nginx/proxy.conf:/etc/nginx/proxy.conf
       - ./nginx/conf.d:/etc/nginx/conf.d
+      #- ./nginx/ssl:/etc/ssl
     depends_on:
       - api
       - web
     ports:
       - "80:80"
+      #- "443:443"

docker/nginx/conf.d/default.conf

@@ -26,4 +26,13 @@ server {
       proxy_pass http://web:3000;
       include proxy.conf;
     }
-}
+
+    # If you want to support HTTPS, please uncomment the code snippet below
+    #listen 443 ssl;
+    #ssl_certificate ./../ssl/your_cert_file.cer;
+    #ssl_certificate_key ./../ssl/your_cert_key.key;
+    #ssl_protocols TLSv1.1 TLSv1.2 TLSv1.3;
+    #ssl_prefer_server_ciphers on;
+    #ssl_session_cache shared:SSL:10m;
+    #ssl_session_timeout 10m;
+}

docker/nginx/ssl/.gitkeep

@@ -0,0 +1 @@
+

web/app/(commonLayout)/apps/AppCard.tsx

@@ -167,7 +167,7 @@ const AppCard = ({ app, onRefresh }: AppCardProps) => {
                 style.actionIconWrapper,
               )
             }
-            className={'!w-[128px] h-fit !z-20'}
+            className={'!w-[128px] h-fit !z-0'}
             manualClose
           />}
         </div>

web/app/(commonLayout)/datasets/DatasetCard.tsx

@@ -46,7 +46,7 @@ const DatasetCard = ({
 
   return (
     <>
-      <Link href={`/datasets/${dataset.id}/documents`} className={cn(style.listItem)}>
+      <Link href={`/datasets/${dataset.id}/documents`} className={cn(style.listItem)} data-disable-nprogress={true}>
         <div className={style.listItemTitle}>
           <AppIcon size='small' className={cn(!dataset.embedding_available && style.unavailable)} />
           <div className={cn(style.listItemHeading, !dataset.embedding_available && style.unavailable)}>