version to 0.6.3 (#3519)

Co-authored-by: miendinh <miendinh@users.noreply.github.com>

.devcontainer/Dockerfile

@@ -1,8 +1,5 @@
 FROM mcr.microsoft.com/devcontainers/python:3.10
 
-COPY . .
-
-
 # [Optional] Uncomment this section to install additional OS packages.
 # RUN apt-get update && export DEBIAN_FRONTEND=noninteractive \
 #     && apt-get -y install --no-install-recommends <your-package-list-here>

.github/workflows/api-tests.yml

@@ -32,15 +32,22 @@ jobs:
       - name: Checkout code
         uses: actions/checkout@v4
 
+      - name: Install APT packages
+        uses: awalsh128/cache-apt-pkgs-action@v1
+        with:
+          packages: ffmpeg
+
       - name: Set up Python
         uses: actions/setup-python@v5
         with:
           python-version: '3.10'
           cache: 'pip'
-          cache-dependency-path: ./api/requirements.txt
+          cache-dependency-path: |
+            ./api/requirements.txt
+            ./api/requirements-dev.txt
 
       - name: Install dependencies
-        run: pip install -r ./api/requirements.txt
+        run: pip install -r ./api/requirements.txt -r ./api/requirements-dev.txt
 
       - name: Run ModelRuntime
         run: pytest api/tests/integration_tests/model_runtime/anthropic api/tests/integration_tests/model_runtime/azure_openai api/tests/integration_tests/model_runtime/openai api/tests/integration_tests/model_runtime/chatglm api/tests/integration_tests/model_runtime/google api/tests/integration_tests/model_runtime/xinference api/tests/integration_tests/model_runtime/huggingface_hub/test_llm.py

README.md

@@ -4,7 +4,7 @@
   <a href="https://cloud.dify.ai">Dify Cloud</a> ·
   <a href="https://docs.dify.ai/getting-started/install-self-hosted">Self-hosting</a> ·
   <a href="https://docs.dify.ai">Documentation</a> ·
-  <a href="https://cal.com/guchenhe/30min">Commercial inquiry</a>
+  <a href="https://cal.com/guchenhe/60-min-meeting">Enterprise inquiry</a>
 </p>
 
 <p align="center">
@@ -56,7 +56,7 @@ Dify is an open-source LLM app development platform. Its intuitive interface com
 **2. Comprehensive model support**: 
   Seamless integration with hundreds of proprietary / open-source LLMs from dozens of inference providers and self-hosted solutions, covering GPT, Mistral, Llama2, and any OpenAI API-compatible models. A full list of supported model providers can be found [here](https://docs.dify.ai/getting-started/readme/model-providers).
 
-![providers-v3](https://github.com/langgenius/dify/assets/13230914/55fab860-d818-4c95-95a2-7ac39f6aea83)
+![providers-v5](https://github.com/langgenius/dify/assets/13230914/5a17bdbe-097a-4100-8363-40255b70f6e3)
 
 
 **3. Prompt IDE**: 
@@ -75,7 +75,7 @@ Dify is an open-source LLM app development platform. Its intuitive interface com
   All of Dify's offerings come with corresponding APIs, so you could effortlessly integrate Dify into your own business logic.
 
 
-## Feature Comparison
+## Feature comparison
 <table style="width: 100%;">
   <tr>
     <th align="center">Feature</th>
@@ -151,7 +151,7 @@ We host a [Dify Cloud](https://dify.ai) service for anyone to try with zero setu
 Quickly get Dify running in your environment with this [starter guide](#quick-start).
 Use our [documentation](https://docs.dify.ai) for further references and more in-depth instructions.
 
-- **Dify for Enterprise / Organizations</br>**
+- **Dify for enterprise / organizations</br>**
 We provide additional enterprise-centric features. [Schedule a meeting with us](https://cal.com/guchenhe/30min) or [send us an email](mailto:business@dify.ai?subject=[GitHub]Business%20License%20Inquiry) to discuss enterprise needs. </br>
   > For startups and small businesses using AWS, check out [Dify Premium on AWS Marketplace](https://aws.amazon.com/marketplace/pp/prodview-t22mebxzwjhu6) and deploy it to your own AWS VPC with one-click. It's an affordable AMI offering with the option to create apps with custom logo and branding.
 
@@ -164,7 +164,7 @@ Star Dify on GitHub and be instantly notified of new releases.
 
 
 
-## Quick Start
+## Quick start
 > Before installing Dify, make sure your machine meets the following minimum system requirements:
 > 
 >- CPU >= 2 Core
@@ -207,7 +207,7 @@ At the same time, please consider supporting Dify by sharing it on social media
   <img src="https://contrib.rocks/image?repo=langgenius/dify" />
 </a>
 
-## Community & Contact
+## Community & contact
 
 * [Github Discussion](https://github.com/langgenius/dify/discussions). Best for: sharing feedback and asking questions.
 * [GitHub Issues](https://github.com/langgenius/dify/issues). Best for: bugs you encounter using Dify.AI, and feature proposals. See our [Contribution Guide](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md).
@@ -232,12 +232,12 @@ Or, schedule a meeting directly with a team member:
   </tr>
 </table>
 
-## Star History
+## Star history
 
 [![Star History Chart](https://api.star-history.com/svg?repos=langgenius/dify&type=Date)](https://star-history.com/#langgenius/dify&Date)
 
 
-## Security Disclosure
+## Security disclosure
 
 To protect your privacy, please avoid posting security issues on GitHub. Instead, send your questions to security@dify.ai and we will provide you with a more detailed answer.
 

README_CN.md

@@ -1,82 +1,167 @@
-[![](./images/describe.png)](https://dify.ai)
-<p align="center">
-  <a href="./README.md">English</a> |
-  <a href="./README_CN.md">简体中文</a> |
-  <a href="./README_JA.md">日本語</a> |
-  <a href="./README_ES.md">Español</a> |
-  <a href="./README_KL.md">Klingon</a> |
-  <a href="./README_FR.md">Français</a>
-</p>
+![cover-v5-optimized](https://github.com/langgenius/dify/assets/13230914/f9e19af5-61ba-4119-b926-d10c4c06ebab)
+
+<div align="center">
+  <a href="https://cloud.dify.ai">Dify 云服务</a> ·
+  <a href="https://docs.dify.ai/getting-started/install-self-hosted">自托管</a> ·
+  <a href="https://docs.dify.ai">文档</a> ·
+  <a href="https://cal.com/guchenhe/dify-demo">预约演示</a>
+</div>
 
 <p align="center">
     <a href="https://dify.ai" target="_blank">
-        <img alt="Static Badge" src="https://img.shields.io/badge/AI-Dify?logo=AI&logoColor=%20%23f5f5f5&label=Dify&labelColor=%20%23155EEF&color=%23EAECF0"></a>
+        <img alt="Static Badge" src="https://img.shields.io/badge/Product-F04438"></a>
+    <a href="https://dify.ai/pricing" target="_blank">
+        <img alt="Static Badge" src="https://img.shields.io/badge/free-pricing?logo=free&color=%20%23155EEF&label=pricing&labelColor=%20%23528bff"></a>
     <a href="https://discord.gg/FngNHpbcY7" target="_blank">
-        <img src="https://img.shields.io/discord/1082486657678311454?logo=discord"
+        <img src="https://img.shields.io/discord/1082486657678311454?logo=discord&labelColor=%20%235462eb&logoColor=%20%23f5f5f5&color=%20%235462eb"
             alt="chat on Discord"></a>
     <a href="https://twitter.com/intent/follow?screen_name=dify_ai" target="_blank">
-        <img src="https://img.shields.io/twitter/follow/dify_ai?style=social&logo=X"
+        <img src="https://img.shields.io/twitter/follow/dify_ai?logo=X&color=%20%23f5f5f5"
             alt="follow on Twitter"></a>
     <a href="https://hub.docker.com/u/langgenius" target="_blank">
-        <img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/langgenius/dify-web"></a>
+        <img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/langgenius/dify-web?labelColor=%20%23FDB062&color=%20%23f79009"></a>
+    <a href="https://github.com/langgenius/dify/graphs/commit-activity" target="_blank">
+        <img alt="Commits last month" src="https://img.shields.io/github/commit-activity/m/langgenius/dify?labelColor=%20%2332b583&color=%20%2312b76a"></a>
+    <a href="https://github.com/langgenius/dify/" target="_blank">
+        <img alt="Issues closed" src="https://img.shields.io/github/issues-search?query=repo%3Alanggenius%2Fdify%20is%3Aclosed&label=issues%20closed&labelColor=%20%237d89b0&color=%20%235d6b98"></a>
+    <a href="https://github.com/langgenius/dify/discussions/" target="_blank">
+        <img alt="Discussion posts" src="https://img.shields.io/github/discussions/langgenius/dify?labelColor=%20%239b8afb&color=%20%237a5af8"></a>
 </p>
 
-<p align="center">
-<a href="https://trendshift.io/repositories/2152" target="_blank"><img src="https://trendshift.io/api/badge/repositories/2152" alt="langgenius%2Fdify | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
-</p>
-
-<p align="center">
-   <a href="https://mp.weixin.qq.com/s/TnyfIuH-tPi9o1KNjwVArw" target="_blank">
-   Dify 发布 AI Agent 能力：基于不同的大型语言模型构建 GPTs 和 Assistants
-  </a>
-</p>
-
-Dify 是一个 LLM 应用开发平台，已经有超过 10 万个应用基于 Dify.AI 构建。它融合了 Backend as Service 和 LLMOps 的理念，涵盖了构建生成式 AI 原生应用所需的核心技术栈，包括一个内置 RAG 引擎。使用 Dify，你可以基于任何模型自部署类似 Assistants API 和 GPTs 的能力。
-
-![](./images/demo.png)
-
-## 使用云端服务
-
-使用 [Dify.AI Cloud](https://dify.ai) 提供开源版本的所有功能,并包含 200 次 GPT 试用额度。
-
-## 为什么选择 Dify
-
-Dify 具有模型中立性，相较 LangChain 等硬编码开发库 Dify 是一个完整的、工程化的技术栈，而相较于 OpenAI 的 Assistants API 你可以完全将服务部署在本地。
-
-| 功能 | Dify.AI | Assistants API | LangChain |
-| --- | --- | --- | --- |
-| 编程方式 | 面向 API | 面向 API | 面向 Python 代码 |
-| 生态策略 | 开源 | 封闭且商用 | 开源 |
-| RAG 引擎 | 支持 | 支持 | 不支持 |
-| Prompt IDE | 包含 | 包含 | 没有 |
-| 支持的 LLMs | 丰富 | 仅 GPT | 丰富 |
-| 本地部署 | 支持 | 不支持 | 不适用 |
+<div align="center">
+  <a href="./README.md"><img alt="上个月的提交次数" src="https://img.shields.io/badge/英文-d9d9d9"></a>
+  <a href="./README_CN.md"><img alt="上个月的提交次数" src="https://img.shields.io/badge/简体中文-d9d9d9"></a>
+  <a href="./README_JA.md"><img alt="上个月的提交次数" src="https://img.shields.io/badge/日本語-d9d9d9"></a>
+  <a href="./README_ES.md"><img alt="上个月的提交次数" src="https://img.shields.io/badge/西班牙语-d9d9d9"></a>
+  <a href="./README_KL.md"><img alt="上个月的提交次数" src="https://img.shields.io/badge/法语-d9d9d9"></a>
+  <a href="./README_FR.md"><img alt="上个月的提交次数" src="https://img.shields.io/badge/克林贡语-d9d9d9"></a>
+</div>
 
 
-## 特点
+#
 
-![](./images/models.png)
+<div align="center">
+  <a href="https://trendshift.io/repositories/2152" target="_blank"><img src="https://trendshift.io/api/badge/repositories/2152" alt="langgenius%2Fdify | 趋势转变" style="width: 250px; height: 55px;" width="250" height="55"/></a>
+</div>
 
-**1. LLM支持**：与 OpenAI 的 GPT 系列模型集成,或者与开源的 Llama2 系列模型集成。事实上，Dify支持主流的商业模型和开源模型(本地部署或基于 MaaS)。
+Dify 是一个开源的LLM应用开发平台。其直观的界面结合了AI工作流程、RAG管道、代理功能、模型管理、可观察性功能等，让您可以快速从原型到生产。以下是其核心功能列表：
+</br> </br>
 
-**2. Prompt IDE**：和团队一起在 Dify 协作，通过可视化的 Prompt 和应用编排工具开发 AI 应用。 支持无缝切换多种大型语言模型。
+**1. 工作流**: 
+  在视觉画布上构建和测试功能强大的AI工作流程，利用以下所有功能以及更多功能。
 
-**3. RAG引擎**：包括各种基于全文索引或向量数据库嵌入的 RAG 能力，允许直接上传 PDF、TXT 等各种文本格式。
 
-**4. AI Agent**：基于 Function Calling 和 ReAct 的 Agent 推理框架，允许用户自定义工具，所见即所得。Dify 提供了十多种内置工具调用能力，如谷歌搜索、DELL·E、Stable Diffusion、WolframAlpha 等。
+  https://github.com/langgenius/dify/assets/13230914/356df23e-1604-483d-80a6-9517ece318aa
 
-**5. 持续运营**：监控和分析应用日志和性能，使用生产数据持续改进 Prompt、数据集或模型。
 
-## 在开始之前
 
-**关注我们，您将立即收到 GitHub 上所有新发布版本的通知！**
+**2. 全面的模型支持**: 
+  与数百种专有/开源LLMs以及数十种推理提供商和自托管解决方案无缝集成，涵盖GPT、Mistral、Llama2以及任何与OpenAI API兼容的模型。完整的支持模型提供商列表可在[此处](https://docs.dify.ai/getting-started/readme/model-providers)找到。
 
-![star-us](https://github.com/langgenius/dify/assets/100913391/95f37259-7370-4456-a9f0-0bc01ef8642f)
+![providers-v5](https://github.com/langgenius/dify/assets/13230914/5a17bdbe-097a-4100-8363-40255b70f6e3)
 
-- [网站](https://dify.ai)
-- [文档](https://docs.dify.ai)
-- [部署文档](https://docs.dify.ai/getting-started/install-self-hosted)
-- [常见问题](https://docs.dify.ai/getting-started/faq)
+
+**3. Prompt IDE**: 
+  用于制作提示、比较模型性能以及向基于聊天的应用程序添加其他功能（如文本转语音）的直观界面。
+
+**4. RAG Pipeline**: 
+  广泛的RAG功能，涵盖从文档摄入到检索的所有内容，支持从PDF、PPT和其他常见文档格式中提取文本的开箱即用的支持。
+
+**5. Agent 智能体**: 
+  您可以基于LLM函数调用或ReAct定义代理，并为代理添加预构建或自定义工具。Dify为AI代理提供了50多种内置工具，如谷歌搜索、DELL·E、稳定扩散和WolframAlpha等。
+
+**6. LLMOps**: 
+  随时间监视和分析应用程序日志和性能。您可以根据生产数据和注释持续改进提示、数据集和模型。
+
+**7. 后端即服务**: 
+  所有Dify的功能都带有相应的API，因此您可以轻松地将Dify集成到自己的业务逻辑中。
+
+
+## 功能比较
+<table style="width: 100%;">
+  <tr>
+    <th align="center">功能</th>
+    <th align="center">Dify.AI</th>
+    <th align="center">LangChain</th>
+    <th align="center">Flowise</th>
+    <th align="center">OpenAI助理API</th>
+  </tr>
+  <tr>
+    <td align="center">编程方法</td>
+    <td align="center">API + 应用程序导向</td>
+    <td align="center">Python代码</td>
+    <td align="center">应用程序导向</td>
+    <td align="center">API导向</td>
+  </tr>
+  <tr>
+    <td align="center">支持的LLMs</td>
+    <td align="center">丰富多样</td>
+    <td align="center">丰富多样</td>
+    <td align="center">丰富多样</td>
+    <td align="center">仅限OpenAI</td>
+  </tr>
+  <tr>
+    <td align="center">RAG引擎</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+  </tr>
+  <tr>
+    <td align="center">代理</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+  </tr>
+  <tr>
+    <td align="center">工作流程</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+  </tr>
+  <tr>
+    <td align="center">可观察性</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+    <td align="center">❌</td>
+  </tr>
+  <tr>
+    <td align="center">企业功能（SSO/访问控制）</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+    <td align="center">❌</td>
+    <td align="center">❌</td>
+  </tr>
+  <tr>
+    <td align="center">本地部署</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+  </tr>
+</table>
+
+## 使用 Dify
+
+- **云 </br>**
+我们提供[ Dify 云服务](https://dify.ai)，任何人都可以零设置尝试。它提供了自部署版本的所有功能，并在沙盒计划中包含 200 次免费的 GPT-4 调用。
+
+- **自托管 Dify 社区版</br>**
+使用这个[入门指南](#quick-start)快速在您的环境中运行 Dify。
+使用我们的[文档](https://docs.dify.ai)进行进一步的参考和更深入的说明。
+
+- **面向企业/组织的 Dify</br>**
+我们提供额外的面向企业的功能。[与我们安排会议](https://cal.com/guchenhe/30min)或[给我们发送电子邮件](mailto:business@dify.ai?subject=[GitHub]Business%20License%20Inquiry)讨论企业需求。 </br>
+  > 对于使用 AWS 的初创公司和中小型企业，请查看 [AWS Marketplace 上的 Dify 高级版](https://aws.amazon.com/marketplace/pp/prodview-t22mebxzwjhu6)，并使用一键部署到您自己的 AWS VPC。它是一个价格实惠的 AMI 产品，提供了使用自定义徽标和品牌创建应用程序的选项。
+
+## 保持领先
+
+在 GitHub 上给 Dify Star，并立即收到新版本的通知。
+
+![star-us](https://github.com/langgenius/dify/assets/13230914/b823edc1-6388-4e25-ad45-2f6b187adbb4)
 
 ## 安装社区版
 
@@ -114,6 +199,19 @@ docker compose up -d
 [![Star History Chart](https://api.star-history.com/svg?repos=langgenius/dify&type=Date)](https://star-history.com/#langgenius/dify&Date)
 
 
+## Contributing
+
+对于那些想要贡献代码的人，请参阅我们的[贡献指南](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md)。
+同时，请考虑通过社交媒体、活动和会议来支持Dify的分享。
+
+> 我们正在寻找贡献者来帮助将Dify翻译成除了中文和英文之外的其他语言。如果您有兴趣帮助，请参阅我们的[i18n README](https://github.com/langgenius/dify/blob/main/web/i18n/README.md)获取更多信息，并在我们的[Discord社区服务器](https://discord.gg/8Tpq4AcN9c)的`global-users`频道中留言。
+
+**Contributors**
+
+<a href="https://github.com/langgenius/dify/graphs/contributors">
+  <img src="https://contrib.rocks/image?repo=langgenius/dify" />
+</a>
+
 ## 社区与支持
 
 我们欢迎您为 Dify 做出贡献，以帮助改善 Dify。包括：提交代码、问题、新想法，或分享您基于 Dify 创建的有趣且有用的 AI 应用程序。同时，我们也欢迎您在不同的活动、会议和社交媒体上分享 Dify。

README_ES.md

@@ -1,119 +1,245 @@
-[![](./images/describe.png)](https://dify.ai)
+![cover-v5-optimized](https://github.com/langgenius/dify/assets/13230914/f9e19af5-61ba-4119-b926-d10c4c06ebab)
+
 <p align="center">
-  <a href="./README.md">English</a> |
-  <a href="./README_CN.md">简体中文</a> |
-  <a href="./README_JA.md">日本語</a> |
-  <a href="./README_ES.md">Español</a> |
-  <a href="./README_KL.md">Klingon</a> |
-  <a href="./README_FR.md">Français</a>
+  <a href="https://cloud.dify.ai">Dify Cloud</a> ·
+  <a href="https://docs.dify.ai/getting-started/install-self-hosted">Auto-alojamiento</a> ·
+  <a href="https://docs.dify.ai">Documentación</a> ·
+  <a href="https://cal.com/guchenhe/dify-demo">Programar demostración</a>
 </p>
 
 <p align="center">
     <a href="https://dify.ai" target="_blank">
-        <img alt="Static Badge" src="https://img.shields.io/badge/AI-Dify?logo=AI&logoColor=%20%23f5f5f5&label=Dify&labelColor=%20%23155EEF&color=%23EAECF0"></a>
+        <img alt="Insignia Estática" src="https://img.shields.io/badge/Producto-F04438"></a>
+    <a href="https://dify.ai/pricing" target="_blank">
+        <img alt="Insignia Estática" src="https://img.shields.io/badge/gratis-precios?logo=gratis&color=%20%23155EEF&label=precios&labelColor=%20%23528bff"></a>
     <a href="https://discord.gg/FngNHpbcY7" target="_blank">
-        <img src="https://img.shields.io/discord/1082486657678311454?logo=discord"
-            alt="chat on Discord"></a>
+        <img src="https://img.shields.io/discord/1082486657678311454?logo=discord&labelColor=%20%235462eb&logoColor=%20%23f5f5f5&color=%20%235462eb"
+            alt="chat en Discord"></a>
     <a href="https://twitter.com/intent/follow?screen_name=dify_ai" target="_blank">
-        <img src="https://img.shields.io/twitter/follow/dify_ai?style=social&logo=X"
-            alt="follow on Twitter"></a>
+        <img src="https://img.shields.io/twitter/follow/dify_ai?logo=X&color=%20%23f5f5f5"
+            alt="seguir en Twitter"></a>
     <a href="https://hub.docker.com/u/langgenius" target="_blank">
-        <img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/langgenius/dify-web"></a>
+        <img alt="Descargas de Docker" src="https://img.shields.io/docker/pulls/langgenius/dify-web?labelColor=%20%23FDB062&color=%20%23f79009"></a>
+    <a href="https://github.com/langgenius/dify/graphs/commit-activity" target="_blank">
+        <img alt="Actividad de Commits el último mes" src="https://img.shields.io/github/commit-activity/m/langgenius/dify?labelColor=%20%2332b583&color=%20%2312b76a"></a>
+    <a href="https://github.com/langgenius/dify/" target="_blank">
+        <img alt="Issues cerrados" src="https://img.shields.io/github/issues-search?query=repo%3Alanggenius%2Fdify%20is%3Aclosed&label=issues%20cerrados&labelColor=%20%237d89b0&color=%20%235d6b98"></a>
+    <a href="https://github.com/langgenius/dify/discussions/" target="_blank">
+        <img alt="Publicaciones de discusión" src="https://img.shields.io/github/discussions/langgenius/dify?labelColor=%20%239b8afb&color=%20%237a5af8"></a>
 </p>
 
 <p align="center">
-   <a href="https://dify.ai/blog/dify-ai-unveils-ai-agent-creating-gpts-and-assistants-with-various-llms" target="_blank">
-   Dify.AI Unveils AI Agent: Creating GPTs and Assistants with Various LLMs
-  </a>
+  <a href="./README.md"><img alt="Actividad de Commits el último mes" src="https://img.shields.io/badge/Inglés-d9d9d9"></a>
+  <a href="./README_CN.md"><img alt="Actividad de Commits el último mes" src="https://img.shields.io/badge/简体中文-d9d9d9"></a>
+  <a href="./README_JA.md"><img alt="Actividad de Commits el último mes" src="https://img.shields.io/badge/日本語-d9d9d9"></a>
+  <a href="./README_ES.md"><img alt="Actividad de Commits el último mes" src="https://img.shields.io/badge/Español-d9d9d9"></a>
+  <a href="./README_KL.md"><img alt="Actividad de Commits el último mes" src="https://img.shields.io/badge/Français-d9d9d9"></a>
+  <a href="./README_FR.md"><img alt="Actividad de Commits el último mes" src="https://img.shields.io/badge/Klingon-d9d9d9"></a>
 </p>
 
-**Dify** es una plataforma de desarrollo de aplicaciones para modelos de lenguaje de gran tamaño (LLM) que ya ha visto la creación de más de **100,000** aplicaciones basadas en Dify.AI. Integra los conceptos de Backend como Servicio y LLMOps, cubriendo el conjunto de tecnologías esenciales requerido para construir aplicaciones nativas de inteligencia artificial generativa, incluyendo un motor RAG incorporado. Con Dify, **puedes auto-desplegar capacidades similares a las de Assistants API y GPTs basadas en cualquier LLM.**
+#
 
-![](./images/demo.png)
+<p align="center">
+  <a href="https://trendshift.io/repositories/2152" target="_blank"><img src="https://trendshift.io/api/badge/repositories/2152" alt="langgenius%2Fdify | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
+</p>
+Dify es una plataforma de desarrollo de aplicaciones de LLM de código abierto. Su interfaz intuitiva combina flujo de trabajo de IA, pipeline RAG, capacidades de agente, gestión de modelos, características de observabilidad y más, lo que le permite pasar rápidamente de un prototipo a producción. Aquí hay una lista de las características principales:
+</br> </br>
 
-## Utilizar Servicios en la Nube
+**1. Flujo de trabajo**: 
+  Construye y prueba potentes flujos de trabajo de IA en un lienzo visual, aprovechando todas las siguientes características y más.
 
-Usar [Dify.AI Cloud](https://dify.ai) proporciona todas las capacidades de la versión de código abierto, e incluye un complemento de 200 créditos de prueba para GPT.
 
-## Por qué Dify
+  https://github.com/langgenius/dify/assets/13230914/356df23e-1604-483d-80a6-9517ece318aa
 
-Dify se caracteriza por su neutralidad de modelo y es un conjunto tecnológico completo e ingenierizado, en comparación con las bibliotecas de desarrollo codificadas como LangChain. A diferencia de la API de Assistants de OpenAI, Dify permite el despliegue local completo de los servicios.
 
-| Característica | Dify.AI | API de Assistants | LangChain |
-|----------------|---------|------------------|-----------|
-| **Enfoque de Programación** | Orientado a API | Orientado a API | Orientado a Código en Python |
-| **Estrategia del Ecosistema** | Código Abierto | Cerrado y Comercial | Código Abierto |
-| **Motor RAG** | Soportado | Soportado | No Soportado |
-| **IDE de Prompts** | Incluido | Incluido | Ninguno |
-| **LLMs Soportados** | Gran Variedad | Solo GPT | Gran Variedad |
-| **Despliegue Local** | Soportado | No Soportado | No Aplicable |
 
-## Características
+**2. Soporte de modelos completo**: 
+  Integración perfecta con cientos de LLMs propietarios / de código abierto de docenas de proveedores de inferencia y soluciones auto-alojadas, que cubren GPT, Mistral, Llama2 y cualquier modelo compatible con la API de OpenAI. Se puede encontrar una lista completa de proveedores de modelos admitidos [aquí](https://docs.dify.ai/getting-started/readme/model-providers).
 
-![](./images/models.png)
+![proveedores-v5](https://github.com/langgenius/dify/assets/13230914/5a17bdbe-097a-4100-8363-40255b70f6e3)
 
-**1. Soporte LLM**: Integración con la familia de modelos GPT de OpenAI, o los modelos de la familia Llama2 de código abierto. De hecho, Dify soporta modelos comerciales convencionales y modelos de código abierto (desplegados localmente o basados en MaaS).
 
-**2. IDE de Prompts**: Orquestación visual de aplicaciones y servicios basados en LLMs con tu equipo.
+**3. IDE de prompt**: 
+  Interfaz intuitiva para crear prompts, comparar el rendimiento del modelo y agregar características adicionales como texto a voz a una aplicación basada en chat. 
 
-**3. Motor RAG**: Incluye varias capacidades RAG basadas en indexación de texto completo o incrustaciones de base de datos vectoriales, permitiendo la carga directa de PDFs, TXTs y otros formatos de texto.
+**4. Pipeline RAG**: 
+  Amplias capacidades de RAG que cubren todo, desde la ingestión de documentos hasta la recuperación, con soporte listo para usar para la extracción de texto de PDF, PPT y otros formatos de documento comunes.
 
-**4. Agente de IA**: Basado en la llamada de funciones y ReAct, el marco de inferencia del Agente permite a los usuarios personalizar las herramientas, lo que ves es lo que obtienes. Dify proporciona más de una docena de capacidades de llamada de herramientas incorporadas, como Búsqueda de Google, DELL·E, Difusión Estable, WolframAlpha, etc.
+**5. Capacidades de agente**: 
+  Puedes definir agent
 
-**5. Operaciones Continuas**: Monitorear y analizar registros de aplicaciones y rendimiento, mejorando continuamente Prompts, conjuntos de datos o modelos usando datos de producción.
+es basados en LLM Function Calling o ReAct, y agregar herramientas preconstruidas o personalizadas para el agente. Dify proporciona más de 50 herramientas integradas para agentes de IA, como Búsqueda de Google, DELL·E, Difusión Estable y WolframAlpha.
 
-## Antes de Empezar
+**6. LLMOps**: 
+  Supervisa y analiza registros de aplicaciones y rendimiento a lo largo del tiempo. Podrías mejorar continuamente prompts, conjuntos de datos y modelos basados en datos de producción y anotaciones.
 
-**¡Danos una estrella, y recibirás notificaciones instantáneas de todos los nuevos lanzamientos en GitHub!**
+**7. Backend como servicio**: 
+  Todas las ofertas de Dify vienen con APIs correspondientes, por lo que podrías integrar Dify sin esfuerzo en tu propia lógica empresarial.
 
-![star-us](https://github.com/langgenius/dify/assets/100913391/95f37259-7370-4456-a9f0-0bc01ef8642f)
 
-- [Sitio web](https://dify.ai)
-- [Documentación](https://docs.dify.ai)
-- [Documentación de Implementación](https://docs.dify.ai/getting-started/install-self-hosted)
-- [Preguntas Frecuentes](https://docs.dify.ai/getting-started/faq)
+## Comparación de características
+<table style="width: 100%;">
+  <tr>
+    <th align="center">Característica</th>
+    <th align="center">Dify.AI</th>
+    <th align="center">LangChain</th>
+    <th align="center">Flowise</th>
+    <th align="center">API de Asistentes de OpenAI</th>
+  </tr>
+  <tr>
+    <td align="center">Enfoque de programación</td>
+    <td align="center">API + orientado a la aplicación</td>
+    <td align="center">Código Python</td>
+    <td align="center">Orientado a la aplicación</td>
+    <td align="center">Orientado a la API</td>
+  </tr>
+  <tr>
+    <td align="center">LLMs admitidos</td>
+    <td align="center">Gran variedad</td>
+    <td align="center">Gran variedad</td>
+    <td align="center">Gran variedad</td>
+    <td align="center">Solo OpenAI</td>
+  </tr>
+  <tr>
+    <td align="center">Motor RAG</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+  </tr>
+  <tr>
+    <td align="center">Agente</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+  </tr>
+  <tr>
+    <td align="center">Flujo de trabajo</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+  </tr>
+  <tr>
+    <td align="center">Observabilidad</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+    <td align="center">❌</td>
+  </tr>
+  <tr>
+    <td align="center">Característica empresarial (SSO/Control de acceso)</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+    <td align="center">❌</td>
+    <td align="center">❌</td>
+  </tr>
+  <tr>
+    <td align="center">Implementación local</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+  </tr>
+</table>
 
-## Instalar la Edición Comunitaria
+## Usando Dify
 
-### Requisitos del Sistema
+- **Nube </br>**
+Hospedamos un servicio [Dify Cloud](https://dify.ai) para que cualquiera lo pruebe sin configuración. Proporciona todas las capacidades de la versión autoimplementada e incluye 200 llamadas gratuitas a GPT-4 en el plan sandbox.
 
-Antes de instalar Dify, asegúrate de que tu máquina cumpla con los siguientes requisitos mínimos del sistema:
+- **Auto-alojamiento de Dify Community Edition</br>**
+Pon rápidamente Dify en funcionamiento en tu entorno con esta [guía de inicio rápido](#quick-start).
+Usa nuestra [documentación](https://docs.dify.ai) para más referencias e instrucciones más detalladas.
 
-- CPU >= 2 núcleos
-- RAM >= 4GB
+- **Dify para Empresas / Organizaciones</br>**
+Proporcionamos características adicionales centradas en la empresa. [Programa una reunión con nosotros](https://cal.com/guchenhe/30min) o [envíanos un correo electrónico](mailto:business@dify.ai?subject=[GitHub]Business%20License%20Inquiry) para discutir las necesidades empresariales. </br>
+  > Para startups y pequeñas empresas que utilizan AWS, echa un vistazo a [Dify Premium en AWS Marketplace](https://aws.amazon.com/marketplace/pp/prodview-t22mebxzwjhu6) e impleméntalo en tu propio VPC de AWS con un clic. Es una AMI asequible que ofrece la opción de crear aplicaciones con logotipo y marca personalizados.
 
-### Inicio Rápido
 
-La forma más sencilla de iniciar el servidor de Dify es ejecutar nuestro archivo [docker-compose.yml](docker/docker-compose.yaml). Antes de ejecutar el comando de instalación, asegúrate de que [Docker](https://docs.docker.com/get-docker/) y [Docker Compose](https://docs.docker.com/compose/install/) estén instalados en tu máquina:
+## Manteniéndote al tanto
+
+Dale estrella a Dify en GitHub y serás notificado instantáneamente de las nuevas versiones.
+
+![danos estrella](https://github.com/langgenius/dify/assets/13230914/b823edc1-6388-4e25-ad45-2f6b187adbb4)
+
+
+
+## Inicio Rápido
+> Antes de instalar Dify, asegúrate de que tu máquina cumpla con los siguientes requisitos mínimos del sistema:
+> 
+>- CPU >= 2 núcleos
+>- RAM >= 4GB
+
+</br>
+
+La forma más fácil de iniciar el servidor de Dify es ejecutar nuestro archivo [docker-compose.yml](docker/docker-compose.yaml). Antes de ejecutar el comando de instalación, asegúrate de que [Docker](https://docs.docker.com/get-docker/) y [Docker Compose](https://docs.docker.com/compose/install/) estén instalados en tu máquina:
 
 ```bash
 cd docker
 docker compose up -d
 ```
 
-Después de ejecutarlo, puedes acceder al panel de control de Dify en tu navegador en [http://localhost/install](http://localhost/install) y comenzar el proceso de instalación de inicialización.
+Después de ejecutarlo, puedes acceder al panel de control de Dify en tu navegador en [http://localhost/install](http://localhost/install) y comenzar el proceso de inicialización.
 
-### Gráfico Helm
+> Si deseas contribuir a Dify o realizar desarrollo adicional, consulta nuestra [guía para implementar desde el código fuente](https://docs.dify.ai/getting-started/install-self-hosted/local-source-code)
 
-Un gran agradecimiento a @BorisPolonsky por proporcionarnos una versión del [Gráfico Helm](https://helm.sh/), que permite implementar Dify en Kubernetes. Puedes visitar https://github.com/BorisPolonsky/dify-helm para obtener información sobre la implementación.
+## Próximos pasos
 
-### Configuración
+Si necesitas personalizar la configuración, consulta los comentarios en nuestro archivo [docker-compose.yml](docker/docker-compose.yaml) y configura manualmente la configuración del entorno
 
-Si necesitas personalizar la configuración, consulta los comentarios en nuestro archivo [docker-compose.yml](docker/docker-compose.yaml) y configura manualmente la configuración del entorno. Después de realizar los cambios, ejecuta nuevamente `docker-compose up -d`. Puedes ver la lista completa de variables de entorno en nuestra [documentación](https://docs.dify.ai/getting-started/install-self-hosted/environments).
+. Después de realizar los cambios, ejecuta `docker-compose up -d` nuevamente. Puedes ver la lista completa de variables de entorno [aquí](https://docs.dify.ai/getting-started/install-self-hosted/environments).
+
+Si deseas configurar una instalación altamente disponible, hay [Gráficos Helm](https://helm.sh/) contribuidos por la comunidad que permiten implementar Dify en Kubernetes.
+
+- [Gráfico Helm por @LeoQuote](https://github.com/douban/charts/tree/master/charts/dify)
+- [Gráfico Helm por @BorisPolonsky](https://github.com/BorisPolonsky/dify-helm)
+
+
+## Contribuir
+
+Para aquellos que deseen contribuir con código, consulten nuestra [Guía de contribución](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md). 
+Al mismo tiempo, considera apoyar a Dify compartiéndolo en redes sociales y en eventos y conferencias.
+
+
+> Estamos buscando colaboradores para ayudar con la traducción de Dify a idiomas que no sean el mandarín o el inglés. Si estás interesado en ayudar, consulta el [README de i18n](https://github.com/langgenius/dify/blob/main/web/i18n/README.md) para obtener más información y déjanos un comentario en el canal `global-users` de nuestro [Servidor de Comunidad en Discord](https://discord.gg/8Tpq4AcN9c).
+
+**Contribuidores**
+
+<a href="https://github.com/langgenius/dify/graphs/contributors">
+  <img src="https://contrib.rocks/image?repo=langgenius/dify" />
+</a>
+
+## Comunidad y Contacto
+
+* [Discusión en GitHub](https://github.com/langgenius/dify/discussions). Lo mejor para: compartir comentarios y hacer preguntas.
+* [Reporte de problemas en GitHub](https://github.com/langgenius/dify/issues). Lo mejor para: errores que encuentres usando Dify.AI y propuestas de características. Consulta nuestra [Guía de contribución](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md).
+* [Correo electrónico](mailto:support@dify.ai?subject=[GitHub]Questions%20About%20Dify). Lo mejor para: preguntas que tengas sobre el uso de Dify.AI.
+* [Discord](https://discord.gg/FngNHpbcY7). Lo mejor para: compartir tus aplicaciones y pasar el rato con la comunidad.
+* [Twitter](https://twitter.com/dify_ai). Lo mejor para: compartir tus aplicaciones y pasar el rato con la comunidad.
+
+O, programa una reunión directamente con un miembro del equipo:
+
+<table>
+  <tr>
+    <th>Punto de Contacto</th>
+    <th>Propósito</th>
+  </tr>
+  <tr>
+    <td><a href='https://cal.com/guchenhe/15min' target='_blank'><img class="schedule-button" src='https://github.com/langgenius/dify/assets/13230914/9ebcd111-1205-4d71-83d5-948d70b809f5' alt='Git-Hub-README-Button-3x' style="width: 180px; height: auto; object-fit: contain;"/></a></td>
+    <td>Consultas comerciales y retroalimentación del producto</td>
+  </tr>
+  <tr>
+    <td><a href='https://cal.com/pinkbanana' target='_blank'><img class="schedule-button" src='https://github.com/langgenius/dify/assets/13230914/d1edd00a-d7e4-4513-be6c-e57038e143fd' alt='Git-Hub-README-Button-2x' style="width: 180px; height: auto; object-fit: contain;"/></a></td>
+    <td>Contribuciones, problemas y solicitudes de características</td>
+  </tr>
+</table>
 
 ## Historial de Estrellas
 
 [![Gráfico de Historial de Estrellas](https://api.star-history.com/svg?repos=langgenius/dify&type=Date)](https://star-history.com/#langgenius/dify&Date)
 
-## Comunidad y Soporte
-
-Te damos la bienvenida a contribuir a Dify para ayudar a hacer que Dify sea mejor de diversas maneras, enviando código, informando problemas, proponiendo nuevas ideas o compartiendo las aplicaciones de inteligencia artificial interesantes y útiles que hayas creado basadas en Dify. Al mismo tiempo, también te invitamos a compartir Dify en diferentes eventos, conferencias y redes sociales.
-
-- [Problemas en GitHub](https://github.com/langgenius/dify/issues). Lo mejor para: errores y problemas que encuentres al usar Dify.AI, consulta la [Guía de Contribución](CONTRIBUTING.md).
-- [Soporte por Correo Electrónico](mailto:hello@dify.ai?subject=[GitHub]Preguntas%20sobre%20Dify). Lo mejor para: preguntas que tengas sobre el uso de Dify.AI.
-- [Discord](https://discord.gg/FngNHpbcY7). Lo mejor para: compartir tus aplicaciones y socializar con la comunidad.
-- [Twitter](https://twitter.com/dify_ai). Lo mejor para: compartir tus aplicaciones y socializar con la comunidad.
-- [Licencia Comercial](mailto:business@dify.ai?subject=[GitHub]Consulta%20de%20Licencia%20Comercial). Lo mejor para: consultas comerciales sobre la licencia de Dify.AI para uso comercial.
 
 ## Divulgación de Seguridad
 
@@ -121,4 +247,4 @@ Para proteger tu privacidad, evita publicar problemas de seguridad en GitHub. En
 
 ## Licencia
 
-Este repositorio está disponible bajo la [Licencia de Código Abierto Dify](LICENSE), que es esencialmente Apache 2.0 con algunas restricciones adicionales.
+Este repositorio está disponible bajo la [Licencia de Código Abierto de Dify](LICENSE), que es esencialmente Apache 2.0 con algunas restricciones adicionales.

README_FR.md

@@ -1,127 +1,250 @@
-[![](./images/describe.png)](https://dify.ai)
+![cover-v5-optimized](https://github.com/langgenius/dify/assets/13230914/f9e19af5-61ba-4119-b926-d10c4c06ebab)
+
 <p align="center">
-  <a href="./README.md">English</a> |
-  <a href="./README_CN.md">简体中文</a> |
-  <a href="./README_JA.md">日本語</a> |
-  <a href="./README_ES.md">Español</a> |
-  <a href="./README_KL.md">Klingon</a> |
-  <a href="./README_FR.md">Français</a>
+  <a href="https://cloud.dify.ai">Dify Cloud</a> ·
+  <a href="https://docs.dify.ai/getting-started/install-self-hosted">Auto-hébergement</a> ·
+  <a href="https://docs.dify.ai">Documentation</a> ·
+  <a href="https://cal.com/guchenhe/dify-demo">Planifier une démo</a>
 </p>
 
 <p align="center">
     <a href="https://dify.ai" target="_blank">
-        <img alt="Static Badge" src="https://img.shields.io/badge/AI-Dify?logo=AI&logoColor=%20%23f5f5f5&label=Dify&labelColor=%20%23155EEF&color=%23EAECF0"></a>
+        <img alt="Badge statique" src="https://img.shields.io/badge/Produit-F04438"></a>
+    <a href="https://dify.ai/pricing" target="_blank">
+        <img alt="Badge statique" src="https://img.shields.io/badge/gratuit-Tarification?logo=free&color=%20%23155EEF&label=pricing&labelColor=%20%23528bff"></a>
     <a href="https://discord.gg/FngNHpbcY7" target="_blank">
-        <img src="https://img.shields.io/discord/1082486657678311454?logo=discord"
-            alt="chat on Discord"></a>
+        <img src="https://img.shields.io/discord/1082486657678311454?logo=discord&labelColor=%20%235462eb&logoColor=%20%23f5f5f5&color=%20%235462eb"
+            alt="chat sur Discord"></a>
     <a href="https://twitter.com/intent/follow?screen_name=dify_ai" target="_blank">
-        <img src="https://img.shields.io/twitter/follow/dify_ai?style=social&logo=X"
-            alt="follow on Twitter"></a>
+        <img src="https://img.shields.io/twitter/follow/dify_ai?logo=X&color=%20%23f5f5f5"
+            alt="suivre sur Twitter"></a>
     <a href="https://hub.docker.com/u/langgenius" target="_blank">
-        <img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/langgenius/dify-web"></a>
+        <img alt="Tirages Docker" src="https://img.shields.io/docker/pulls/langgenius/dify-web?labelColor=%20%23FDB062&color=%20%23f79009"></a>
+    <a href="https://github.com/langgenius/dify/graphs/commit-activity" target="_blank">
+        <img alt="Commits le mois dernier" src="https://img.shields.io/github/commit-activity/m/langgenius/dify?labelColor=%20%2332b583&color=%20%2312b76a"></a>
+    <a href="https://github.com/langgenius/dify/" target="_blank">
+        <img alt="Problèmes fermés" src="https://img.shields.io/github/issues-search?query=repo%3Alanggenius%2Fdify%20is%3Aclosed&label=issues%20closed&labelColor=%20%237d89b0&color=%20%235d6b98"></a>
+    <a href="https://github.com/langgenius/dify/discussions/" target="_blank">
+        <img alt="Messages de discussion" src="https://img.shields.io/github/discussions/langgenius/dify?labelColor=%20%239b8afb&color=%20%237a5af8"></a>
 </p>
 
 <p align="center">
-   <a href="https://dify.ai/blog/dify-ai-unveils-ai-agent-creating-gpts-and-assistants-with-various-llms" target="_blank">
-   Dify.AI Unveils AI Agent: Creating GPTs and Assistants with Various LLMs
-  </a>
+  <a href="./README.md"><img alt="Commits le mois dernier" src="https://img.shields.io/badge/Anglais-d9d9d9"></a>
+  <a href="./README_CN.md"><img alt="Commits le mois dernier" src="https://img.shields.io/badge/简体中文-d9d9d9"></a>
+  <a href="./README_JA.md"><img alt="Commits le mois dernier" src="https://img.shields.io/badge/日本語-d9d9d9"></a>
+  <a href="./README_ES.md"><img alt="Commits le mois dernier" src="https://img.shields.io/badge/Español-d9d9d9"></a>
+  <a href="./README_KL.md"><img alt="Commits le mois dernier" src="https://img.shields.io/badge/Français-d9d9d9"></a>
+  <a href="./README_FR.md"><img alt="Commits le mois dernier" src="https://img.shields.io/badge/Klingon-d9d9d9"></a>
 </p>
 
+#
 
-**Dify** est une plateforme de développement d'applications LLM qui a déjà vu plus de **100,000** applications construites sur Dify.AI. Elle intègre les concepts de Backend as a Service et LLMOps, couvrant la pile technologique de base requise pour construire des applications natives d'IA générative, y compris un moteur RAG intégré. Avec Dify, **vous pouvez auto-déployer des capacités similaires aux API Assistants et GPT basées sur n'importe quels LLM.**
+<p align="center">
+  <a href="https://trendshift.io/repositories/2152" target="_blank"><img src="https://trendshift.io/api/badge/repositories/2152" alt="langgenius%2Fdify | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
+</p>
+Dify est une plateforme de développement d'applications LLM open source. Son interface intuitive combine un flux de travail d'IA, un pipeline RAG, des capacités d'agent, une gestion de modèles, des fonctionnalités d'observabilité, et plus encore, vous permettant de passer rapidement du prototype à la production. Voici une liste des fonctionnalités principales:
+</br> </br>
 
-![](./images/demo.png)
-
-## Utiliser les services cloud  
-
-L'utilisation de [Dify.AI Cloud](https://dify.ai) fournit toutes les capacités de la version open source, et comprend un essai gratuit de 200 crédits GPT.
-
-## Pourquoi Dify
-
-Dify présente une neutralité de modèle et est une pile technologique complète et conçue par rapport à des bibliothèques de développement codées en dur comme LangChain. Contrairement à l'API Assistants d'OpenAI, Dify permet un déploiement local complet des services.
-
-| Fonctionnalité | Dify.AI | API Assistants | LangChain |
-|---------------|----------|-----------------|------------|
-| **Approche de programmation** | Orientée API | Orientée API | Orientée code Python |  
-| **Stratégie écosystème** | Open source | Fermé et commercial | Open source |
-| **Moteur RAG** | Pris en charge | Pris en charge | Non pris en charge |
-| **IDE d'invite** | Inclus | Inclus | Aucun |
-| **LLM pris en charge** | Grande variété | Seulement GPT | Grande variété |
-| **Déploiement local** | Pris en charge | Non pris en charge | Non applicable |
-
- ## Fonctionnalités  
-
-![](./images/models.png)
-
-**1\. Support LLM**: Intégration avec la famille de modèles GPT d'OpenAI, ou les modèles de la famille open source Llama2. En fait, Dify prend en charge les modèles commerciaux grand public et les modèles open source (déployés localement ou basés sur MaaS).  
-
-**2\. IDE d'invite**: Orchestration visuelle d'applications et de services basés sur LLMs avec votre équipe.  
-
-**3\. Moteur RAG**: Comprend diverses capacités RAG basées sur l'indexation de texte intégral ou les embeddings de base de données vectorielles, permettant le chargement direct de PDF, TXT et autres formats de texte.
-
-**4\. AI Agent**: Basé sur l'appel de fonction et ReAct, le framework d'inférence de l'Agent permet aux utilisateurs de personnaliser les outils, ce que vous voyez est ce que vous obtenez. Dify propose plus d'une douzaine de capacités d'appel d'outils intégrées, telles que la recherche Google, DELL·E, Diffusion Stable, WolframAlpha, etc.
-
-**5\. Opérations continues**: Surveillez et analysez les journaux et les performances des applications, améliorez en continu les invites, les datasets ou les modèles à l'aide de données de production.  
-
-## Avant de commencer
-
-**Étoilez-nous, et vous recevrez des notifications instantanées pour toutes les nouvelles sorties sur GitHub !**
-![star-us](https://github.com/langgenius/dify/assets/100913391/95f37259-7370-4456-a9f0-0bc01ef8642f)
-
-- [Site web](https://dify.ai)  
-- [Documentation](https://docs.dify.ai)  
-- [Documentation de déploiement](https://docs.dify.ai/getting-started/install-self-hosted)   
-- [FAQ](https://docs.dify.ai/getting-started/faq)  
+**1. Flux de travail**: 
+  Construisez et testez des flux de travail d'IA puissants sur un canevas visuel, en utilisant toutes les fonctionnalités suivantes et plus encore.
 
 
-## Installer la version Communauté  
+  https://github.com/langgenius/dify/assets/13230914/356df23e-1604-483d-80a6-9517ece318aa
 
-### Configuration système  
 
-Avant d'installer Dify, assurez-vous que votre machine répond aux exigences minimales suivantes:  
 
-- CPU >= 2 cœurs
-- RAM >= 4 Go 
+**2. Prise en charge complète des modèles**: 
+  Intégration transparente avec des centaines de LLM propriétaires / open source provenant de dizaines de fournisseurs d'inférence et de solutions auto-hébergées, couvrant GPT, Mistral, Llama2, et tous les modèles compatibles avec l'API OpenAI. Une liste complète des fournisseurs de modèles pris en charge se trouve [ici](https://docs.dify.ai/getting-started/readme/model-providers).
 
-### Démarrage rapide 
+![providers-v5](https://github.com/langgenius/dify/assets/13230914/5a17bdbe-097a-4100-8363-40255b70f6e3)
 
-La façon la plus simple de démarrer le serveur Dify est d'exécuter notre fichier [docker-compose.yml](docker/docker-compose.yaml). Avant d'exécuter la commande d'installation, assurez-vous que [Docker](https://docs.docker.com/get-docker/) et [Docker Compose](https://docs.docker.com/compose/install/) sont installés sur votre machine:  
+
+**3. IDE de prompt**: 
+  Interface intuitive pour créer des prompts, comparer les performances des modèles et ajouter des fonctionnalités supplémentaires telles que la synthèse vocale à une application basée sur des chats. 
+
+**4. Pipeline RAG**: 
+  Des capacités RAG étendues qui couvrent tout, de l'ingestion de documents à la récupération, avec un support prêt à l'emploi pour l'extraction de texte à partir de PDF, PPT et autres formats de document courants.
+
+**5. Capac
+
+ités d'agent**: 
+  Vous pouvez définir des agents basés sur l'appel de fonction LLM ou ReAct, et ajouter des outils pré-construits ou personnalisés pour l'agent. Dify fournit plus de 50 outils intégrés pour les agents d'IA, tels que la recherche Google, DELL·E, Stable Diffusion et WolframAlpha.
+
+**6. LLMOps**: 
+  Surveillez et analysez les journaux d'application et les performances au fil du temps. Vous pouvez continuellement améliorer les prompts, les ensembles de données et les modèles en fonction des données de production et des annotations.
+
+**7. Backend-as-a-Service**: 
+  Toutes les offres de Dify sont accompagnées d'API correspondantes, vous permettant d'intégrer facilement Dify dans votre propre logique métier.
+
+
+## Comparaison des fonctionnalités
+<table style="width: 100%;">
+  <tr>
+    <th align="center">Fonctionnalité</th>
+    <th align="center">Dify.AI</th>
+    <th align="center">LangChain</th>
+    <th align="center">Flowise</th>
+    <th align="center">OpenAI Assistants API</th>
+  </tr>
+  <tr>
+    <td align="center">Approche de programmation</td>
+    <td align="center">API + Application</td>
+    <td align="center">Code Python</td>
+    <td align="center">Application</td>
+    <td align="center">API</td>
+  </tr>
+  <tr>
+    <td align="center">LLMs pris en charge</td>
+    <td align="center">Grande variété</td>
+    <td align="center">Grande variété</td>
+    <td align="center">Grande variété</td>
+    <td align="center">Uniquement OpenAI</td>
+  </tr>
+  <tr>
+    <td align="center">Moteur RAG</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+  </tr>
+  <tr>
+    <td align="center">Agent</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+  </tr>
+  <tr>
+    <td align="center">Flux de travail</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+  </tr>
+  <tr>
+    <td align="center">Observabilité</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+    <td align="center">❌</td>
+  </tr>
+  <tr>
+    <td align="center">Fonctionnalité d'entreprise (SSO/Contrôle d'accès)</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+    <td align="center">❌</td>
+    <td align="center">❌</td>
+  </tr>
+  <tr>
+    <td align="center">Déploiement local</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+  </tr>
+</table>
+
+## Utiliser Dify
+
+- **Cloud </br>**
+Nous hébergeons un service [Dify Cloud](https://dify.ai) pour que tout le monde puisse l'essayer sans aucune configuration. Il fournit toutes les capacités de la version auto-hébergée et comprend 200 appels GPT-4 gratuits dans le plan bac à sable.
+
+- **Auto-hébergement Dify Community Edition</br>**
+Lancez rapidement Dify dans votre environnement avec ce [guide de démarrage](#quick-start).
+Utilisez notre [documentation](https://docs.dify.ai) pour plus de références et des instructions plus détaillées.
+
+- **Dify pour les entreprises / organisations</br>**
+Nous proposons des fonctionnalités supplémentaires adaptées aux entreprises. [Planifiez une réunion avec nous](https://cal.com/guchenhe/30min) ou [envoyez-nous un e-mail](mailto:business@dify.ai?subject=[GitHub]Business%20License%20Inquiry) pour discuter des besoins de l'entreprise. </br>
+  > Pour les startups et les petites entreprises utilisant AWS, consultez [Dify Premium sur AWS Marketplace](https://aws.amazon.com/marketplace/pp/prodview-t22mebxzwjhu6) et déployez-le dans votre propre VPC AWS en un clic. C'est une offre AMI abordable avec la possibilité de créer des applications avec un logo et une marque personnalisés.
+
+
+## Rester en avance
+
+Mettez une étoile à Dify sur GitHub et soyez instantanément informé des nouvelles versions.
+
+![star-us](https://github.com/langgenius/dify/assets/13230914/b823edc1-6388-4e25-ad45-2f6b187adbb4)
+
+
+
+## Démarrage rapide
+> Avant d'installer Dify, assurez-vous que votre machine répond aux exigences système minimales suivantes:
+> 
+>- CPU >= 2 cœurs
+>- RAM >= 4 Go
+
+</br>
+
+La manière la plus simple de démarrer le serveur Dify est d'exécuter notre fichier [docker-compose.yml](docker/docker-compose.yaml). Avant d'exécuter la commande d'installation, assurez-vous que [Docker](https://docs.docker.com/get-docker/) et [Docker Compose](https://docs.docker.com/compose/install/) sont installés sur votre machine:
 
 ```bash
 cd docker
 docker compose up -d
 ```
 
-Après l'exécution, vous pouvez accéder au tableau de bord Dify dans votre navigateur à l'adresse [http://localhost/install](http://localhost/install) et démarrer le processus d'installation initiale.  
+Après l'exécution, vous pouvez accéder au tableau de bord Dify dans votre navigateur à [http://localhost/install](http://localhost/install) et commencer le processus d'initialisation.
 
-### Chart Helm 
+> Si vous souhaitez contribuer à Dify ou effectuer un développement supplémentaire, consultez notre [guide de déploiement à partir du code source](https://docs.dify.ai/getting-started/install-self-hosted/local-source-code)
 
-Un grand merci à @BorisPolonsky pour nous avoir fourni une version [Helm Chart](https://helm.sh/) qui permet le déploiement de Dify sur Kubernetes.
-Vous pouvez accéder à https://github.com/BorisPolonsky/dify-helm pour des informations de déploiement.  
+## Prochaines étapes
 
-### Configuration  
+Si vous devez personnaliser la configuration, veuillez
 
-Si vous avez besoin de personnaliser la configuration, veuillez vous référer aux commentaires de notre fichier [docker-compose.yml](docker/docker-compose.yaml) et définir manuellement la configuration de l'environnement. Après avoir apporté les modifications, veuillez exécuter à nouveau `docker-compose up -d`. Vous trouverez la liste complète des variables d'environnement dans notre [documentation](https://docs.dify.ai/getting-started/install-self-hosted/environments).  
+ vous référer aux commentaires dans notre fichier [docker-compose.yml](docker/docker-compose.yaml) et définir manuellement la configuration de l'environnement. Après avoir apporté les modifications, veuillez exécuter à nouveau `docker-compose up -d`. Vous pouvez voir la liste complète des variables d'environnement [ici](https://docs.dify.ai/getting-started/install-self-hosted/environments).
 
-## Historique d'étoiles  
+Si vous souhaitez configurer une installation hautement disponible, il existe des [Helm Charts](https://helm.sh/) contribués par la communauté qui permettent de déployer Dify sur Kubernetes.
 
-[![Diagramme de l'historique des étoiles](https://api.star-history.com/svg?repos=langgenius/dify&type=Date)](https://star-history.com/#langgenius/dify&Date)  
+- [Helm Chart par @LeoQuote](https://github.com/douban/charts/tree/master/charts/dify)
+- [Helm Chart par @BorisPolonsky](https://github.com/BorisPolonsky/dify-helm)
 
 
-## Communauté & Support  
+## Contribuer
 
-Nous vous invitons à contribuer à Dify pour aider à améliorer Dify de diverses manières, en soumettant du code, des problèmes, de nouvelles idées ou en partageant les applications d'IA intéressantes et utiles que vous avez créées sur la base de Dify. En même temps, nous vous invitons également à partager Dify lors de différents événements, conférences et réseaux sociaux.  
+Pour ceux qui souhaitent contribuer du code, consultez notre [Guide de contribution](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md). 
+Dans le même temps, veuillez envisager de soutenir Dify en le partageant sur les réseaux sociaux et lors d'événements et de conférences.
 
-- [Problèmes GitHub](https://github.com/langgenius/dify/issues). Idéal pour : les bogues et les erreurs que vous rencontrez en utilisant Dify.AI, voir le [Guide de contribution](CONTRIBUTING.md).  
-- [Support par courriel](mailto:hello@dify.ai?subject=[GitHub]Questions%20About%20Dify). Idéal pour : les questions que vous avez au sujet de l'utilisation de Dify.AI.   
-- [Discord](https://discord.gg/FngNHpbcY7). Idéal pour : partager vos applications et discuter avec la communauté.   
-- [Twitter](https://twitter.com/dify_ai). Idéal pour : partager vos applications et discuter avec la communauté.
-- [Licence commerciale](mailto:business@dify.ai?subject=[GitHub]Business%20License%20Inquiry). Idéal pour : les demandes commerciales de licence de Dify.AI pour un usage commercial.  
 
-## Divulgation de la sécurité  
+> Nous recherchons des contributeurs pour aider à traduire Dify dans des langues autres que le mandarin ou l'anglais. Si vous êtes intéressé à aider, veuillez consulter le [README i18n](https://github.com/langgenius/dify/blob/main/web/i18n/README.md) pour plus d'informations, et laissez-nous un commentaire dans le canal `global-users` de notre [Serveur communautaire Discord](https://discord.gg/8Tpq4AcN9c).
 
-Pour protéger votre vie privée, veuillez éviter de publier des problèmes de sécurité sur GitHub. Envoyez plutôt vos questions à security@dify.ai et nous vous fournirons une réponse plus détaillée.  
+**Contributeurs**
 
-## Licence  
+<a href="https://github.com/langgenius/dify/graphs/contributors">
+  <img src="https://contrib.rocks/image?repo=langgenius/dify" />
+</a>
 
-Ce référentiel est disponible sous la [Licence open source Dify](LICENSE), qui est essentiellement Apache 2.0 avec quelques restrictions supplémentaires.
+## Communauté & Contact
+
+* [Discussion GitHub](https://github.com/langgenius/dify/discussions). Meilleur pour: partager des commentaires et poser des questions.
+* [Problèmes GitHub](https://github.com/langgenius/dify/issues). Meilleur pour: les bogues que vous rencontrez en utilisant Dify.AI et les propositions de fonctionnalités. Consultez notre [Guide de contribution](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md).
+* [E-mail](mailto:support@dify.ai?subject=[GitHub]Questions%20About%20Dify). Meilleur pour: les questions que vous avez sur l'utilisation de Dify.AI.
+* [Discord](https://discord.gg/FngNHpbcY7). Meilleur pour: partager vos applications et passer du temps avec la communauté.
+* [Twitter](https://twitter.com/dify_ai). Meilleur pour: partager vos applications et passer du temps avec la communauté.
+
+Ou, planifiez directement une réunion avec un membre de l'équipe:
+
+<table>
+  <tr>
+    <th>Point de contact</th>
+    <th>Objectif</th>
+  </tr>
+  <tr>
+    <td><a href='https://cal.com/guchenhe/15min' target='_blank'><img class="schedule-button" src='https://github.com/langgenius/dify/assets/13230914/9ebcd111-1205-4d71-83d5-948d70b809f5' alt='Git-Hub-README-Button-3x' style="width: 180px; height: auto; object-fit: contain;"/></a></td>
+    <td>Demandes commerciales & retours produit</td>
+  </tr>
+  <tr>
+    <td><a href='https://cal.com/pinkbanana' target='_blank'><img class="schedule-button" src='https://github.com/langgenius/dify/assets/13230914/d1edd00a-d7e4-4513-be6c-e57038e143fd' alt='Git-Hub-README-Button-2x' style="width: 180px; height: auto; object-fit: contain;"/></a></td>
+    <td>Contributions, problèmes & demandes de fonctionnalités</td>
+  </tr>
+</table>
+
+## Historique des étoiles
+
+[![Graphique de l'historique des étoiles](https://api.star-history.com/svg?repos=langgenius/dify&type=Date)](https://star-history.com/#langgenius/dify&Date)
+
+
+## Divulgation de sécurité
+
+Pour protéger votre vie privée, veuillez éviter de publier des problèmes de sécurité sur GitHub. Au lieu de cela, envoyez vos questions à security@dify.ai et nous vous fournirons une réponse plus détaillée.
+
+## Licence
+
+Ce référentiel est disponible sous la [Licence open source Dify](LICENSE), qui est essentiellement l'Apache 2.0 avec quelques restrictions supplémentaires.

README_JA.md

@@ -1,131 +1,249 @@
-[![](./images/describe.png)](https://dify.ai)
+![cover-v5-optimized](https://github.com/langgenius/dify/assets/13230914/f9e19af5-61ba-4119-b926-d10c4c06ebab)
+
 <p align="center">
-  <a href="./README.md">English</a> |
-  <a href="./README_CN.md">简体中文</a> |
-  <a href="./README_JA.md">日本語</a> |
-  <a href="./README_ES.md">Español</a> |
-  <a href="./README_KL.md">Klingon</a> |
-  <a href="./README_FR.md">Français</a>
+  <a href="https://cloud.dify.ai">Dify Cloud</a> ·
+  <a href="https://docs.dify.ai/getting-started/install-self-hosted">自己ホスティング</a> ·
+  <a href="https://docs.dify.ai">ドキュメント</a> ·
+  <a href="https://cal.com/guchenhe/dify-demo">デモのスケジュール</a>
 </p>
 
 <p align="center">
     <a href="https://dify.ai" target="_blank">
-        <img alt="Static Badge" src="https://img.shields.io/badge/AI-Dify?logo=AI&logoColor=%20%23f5f5f5&label=Dify&labelColor=%20%23155EEF&color=%23EAECF0"></a>
+        <img alt="Static Badge" src="https://img.shields.io/badge/Product-F04438"></a>
+    <a href="https://dify.ai/pricing" target="_blank">
+        <img alt="Static Badge" src="https://img.shields.io/badge/free-pricing?logo=free&color=%20%23155EEF&label=pricing&labelColor=%20%23528bff"></a>
     <a href="https://discord.gg/FngNHpbcY7" target="_blank">
-        <img src="https://img.shields.io/discord/1082486657678311454?logo=discord"
-            alt="chat on Discord"></a>
+        <img src="https://img.shields.io/discord/1082486657678311454?logo=discord&labelColor=%20%235462eb&logoColor=%20%23f5f5f5&color=%20%235462eb"
+            alt="Discordでチャット"></a>
     <a href="https://twitter.com/intent/follow?screen_name=dify_ai" target="_blank">
-        <img src="https://img.shields.io/twitter/follow/dify_ai?style=social&logo=X"
-            alt="follow on Twitter"></a>
+        <img src="https://img.shields.io/twitter/follow/dify_ai?logo=X&color=%20%23f5f5f5"
+            alt="Twitterでフォロー"></a>
     <a href="https://hub.docker.com/u/langgenius" target="_blank">
-        <img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/langgenius/dify-web"></a>
+        <img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/langgenius/dify-web?labelColor=%20%23FDB062&color=%20%23f79009"></a>
+    <a href="https://github.com/langgenius/dify/graphs/commit-activity" target="_blank">
+        <img alt="先月のコミット" src="https://img.shields.io/github/commit-activity/m/langgenius/dify?labelColor=%20%2332b583&color=%20%2312b76a"></a>
+    <a href="https://github.com/langgenius/dify/" target="_blank">
+        <img alt="クローズされた問題" src="https://img.shields.io/github/issues-search?query=repo%3Alanggenius%2Fdify%20is%3Aclosed&label=issues%20closed&labelColor=%20%237d89b0&color=%20%235d6b98"></a>
+    <a href="https://github.com/langgenius/dify/discussions/" target="_blank">
+        <img alt="ディスカッション投稿" src="https://img.shields.io/github/discussions/langgenius/dify?labelColor=%20%239b8afb&color=%20%237a5af8"></a>
 </p>
 
 <p align="center">
-   <a href="https://dify.ai/blog/dify-ai-unveils-ai-agent-creating-gpts-and-assistants-with-various-llms" target="_blank">
-   Dify.AI Unveils AI Agent: Creating GPTs and Assistants with Various LLMs
-  </a>
+  <a href="./README.md"><img alt="先月のコミット" src="https://img.shields.io/badge/English-d9d9d9"></a>
+  <a href="./README_CN.md"><img alt="先月のコミット" src="https://img.shields.io/badge/简体中文-d9d9d9"></a>
+  <a href="./README_JA.md"><img alt="先月のコミット" src="https://img.shields.io/badge/日本語-d9d9d9"></a>
+  <a href="./README_ES.md"><img alt="先月のコミット" src="https://img.shields.io/badge/Español-d9d9d9"></a>
+  <a href="./README_KL.md"><img alt="先月のコミット" src="https://img.shields.io/badge/Français-d9d9d9"></a>
+  <a href="./README_FR.md"><img alt="先月のコミット" src="https://img.shields.io/badge/Klingon-d9d9d9"></a>
 </p>
 
+#
 
-"Difyは、既にDify.AI上で10万以上のアプリケーションが構築されているLLMアプリケーション開発プラットフォームです。バックエンド・アズ・ア・サービスとLLMOpsの概念を統合し、組み込みのRAGエンジンを含む、生成AIネイティブアプリケーションを構築するためのコアテックスタックをカバーしています。Difyを使用すると、どのLLMに基づいても、Assistants APIやGPTのような機能を自己デプロイすることができます。"
+<p align="center">
+  <a href="https://trendshift.io/repositories/2152" target="_blank"><img src="https://trendshift.io/api/badge/repositories/2152" alt="langgenius%2Fdify | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
+</p>
 
-Please note that translating complex technical terms can sometimes result in slight variations in meaning due to differences in language nuances.
+DifyはオープンソースのLLMアプリケーション開発プラットフォームです。直感的なインターフェースには、AIワークフロー、RAGパイプライン、エージェント機能、モデル管理、観測機能などが組み合わさっており、プロトタイプから本番までの移行を迅速に行うことができます。以下は、主要機能のリストです：
+</br> </br>
 
-![](./images/demo.png)
-
-## クラウドサービスの利用
-
-[Dify.AI Cloud](https://dify.ai) を使用すると、オープンソース版の全機能を利用でき、さらに200GPTのトライアルクレジットが無料で提供されます。
-
-## Difyの利点
-
-Difyはモデルニュートラルであり、LangChainのようなハードコードされた開発ライブラリと比較して、完全にエンジニアリングされた技術スタックを特徴としています。OpenAIのAssistants APIとは異なり、Difyではサービスの完全なローカルデプロイメントが可能です。
-
-| 機能 | Dify.AI | Assistants API | LangChain |
-|---------|---------|----------------|-----------|
-| **プログラミングアプローチ** | API指向 | API指向 | Pythonコード指向 |
-| **エコシステム戦略** | オープンソース | 閉鎖的かつ商業的 | オープンソース |
-| **RAGエンジン** | サポート済み | サポート済み | 非サポート |
-| **プロンプトIDE** | 含まれる | 含まれる | なし |
-| **サポートされるLLMs** | 豊富な種類 | GPTのみ | 豊富な種類 |
-| **ローカルデプロイメント** | サポート済み | 非サポート | 該当なし |
-
- ## 機能
-
-![](./images/models.png)
-
-**1\. LLMサポート**: OpenAIのGPTファミリーモデルやLlama2ファミリーのオープンソースモデルとの統合。 実際、Difyは主要な商用モデルとオープンソースモデル(ローカルでデプロイまたはMaaSベース)をサポートしています。
-
-**2\. プロンプトIDE**: チームとのLLMベースのアプリケーションとサービスの視覚的なオーケストレーション。
-
-**3\. RAGエンジン**: フルテキストインデックスまたはベクトルデータベース埋め込みに基づくさまざまなRAG機能を含み、PDF、TXT、その他のテキストフォーマットの直接アップロードを可能にします。
-
-**4. AIエージェント**: 関数呼び出しとReActに基づくAgent推論フレームワークにより、ユーザーはツールをカスタマイズすることができます。Difyは、Google検索、DELL·E、Stable Diffusion、WolframAlphaなど、十数種類の組み込みツール呼び出し機能を提供しています。
-
-**5\. 継続的運用**: アプリケーションログとパフォーマンスを監視および分析し、運用データを使用してプロンプト、データセット、またはモデルを継続的に改善します。
-
-## 開始する前に
-
-**私たちをスターして、GitHub上でのすべての新しいリリースに対する即時通知を受け取ります！**
-
-![私たちをスターして](https://github.com/langgenius/dify/assets/100913391/95f37259-7370-4456-a9f0-0bc01ef8642f)
-
-- [Website](https://dify.ai)
-- [Docs](https://docs.dify.ai)
-- [Deployment Docs](https://docs.dify.ai/getting-started/install-self-hosted)
-- [FAQ](https://docs.dify.ai/getting-started/faq) 
+**1. ワークフロー**: 
+  ビジュアルキャンバス上で強力なAIワークフローを構築してテストし、以下の機能を活用してプロトタイプを超えることができます。
 
 
-## コミュニティエディションのインストール
+  https://github.com/langgenius/dify/assets/13230914/356df23e-1604-483d-80a6-9517ece318aa
 
-### システム要件
 
-Difyをインストールする前に、以下の最低限のシステム要件を満たしていることを確認してください：
 
-- CPU >= 2コア
-- RAM >= 4GB
+**2. 網羅的なモデルサポート**: 
+  数百のプロプライエタリ/オープンソースのLLMと、数十の推論プロバイダーおよびセルフホスティングソリューションとのシームレスな統合を提供します。GPT、Mistral、Llama2、およびOpenAI API互換のモデルをカバーします。サポートされているモデルプロバイダーの完全なリストは[こちら](https://docs
 
-### クイックスタート
+.dify.ai/getting-started/readme/model-providers)をご覧ください。
 
-Difyサーバーを始める最も簡単な方法は、[docker-compose.yml](docker/docker-compose.yaml) ファイルを実行することです。インストールコマンドを実行する前に、マシンに [Docker](https://docs.docker.com/get-docker/) と [Docker Compose](https://docs.docker.com/compose/install/) がインストールされていることを確認してください：
+![providers-v5](https://github.com/langgenius/dify/assets/13230914/5a17bdbe-097a-4100-8363-40255b70f6e3)
+
+
+**3. プロンプトIDE**: 
+  チャットベースのアプリにテキスト読み上げなどの追加機能を追加するプロンプトを作成し、モデルのパフォーマンスを比較する直感的なインターフェース。
+
+**4. RAGパイプライン**: 
+  文書の取り込みから取得までをカバーする幅広いRAG機能で、PDF、PPTなどの一般的なドキュメント形式からのテキスト抽出に対するアウトオブボックスのサポートを提供します。
+
+**5. エージェント機能**: 
+  LLM関数呼び出しまたはReActに基づいてエージェントを定義し、エージェント向けの事前構築済みまたはカスタムのツールを追加できます。Difyには、Google検索、DELL·E、Stable Diffusion、WolframAlphaなどのAIエージェント用の50以上の組み込みツールが用意されています。
+
+**6. LLMOps**: 
+  アプリケーションログとパフォーマンスを時間の経過とともにモニタリングおよび分析します。本番データと注釈に基づいて、プロンプト、データセット、およびモデルを継続的に改善できます。
+
+**7. Backend-as-a-Service**: 
+  Difyのすべての提供には、それに対応するAPIが付属しており、独自のビジネスロジックにDifyをシームレスに統合できます。
+
+
+## 機能比較
+<table style="width: 100%;">
+  <tr>
+    <th align="center">機能</th>
+    <th align="center">Dify.AI</th>
+    <th align="center">LangChain</th>
+    <th align="center">Flowise</th>
+    <th align="center">OpenAI Assistants API</th>
+  </tr>
+  <tr>
+    <td align="center">プログラミングアプローチ</td>
+    <td align="center">API + アプリ指向</td>
+    <td align="center">Pythonコード</td>
+    <td align="center">アプリ指向</td>
+    <td align="center">API指向</td>
+  </tr>
+  <tr>
+    <td align="center">サポートされているLLM</td>
+    <td align="center">豊富なバリエーション</td>
+    <td align="center">豊富なバリエーション</td>
+    <td align="center">豊富なバリエーション</td>
+    <td align="center">OpenAIのみ</td>
+  </tr>
+  <tr>
+    <td align="center">RAGエンジン</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+  </tr>
+  <tr>
+    <td align="center">エージェント</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+  </tr>
+  <tr>
+    <td align="center">ワークフロー</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+  </tr>
+  <tr>
+    <td align="center">観測性</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+    <td align="center">❌</td>
+  </tr>
+  <tr>
+    <td align="center">エンタープライズ機能（SSO/アクセス制御）</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+    <td align="center">❌</td>
+    <td align="center">❌</td>
+  </tr>
+  <tr>
+    <td align="center">ローカル展開</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+  </tr>
+</table>
+
+## Difyの使用方法
+
+- **クラウド </br>**
+[こちら](https://dify.ai)のDify Cloudサービスを利用して、セットアップが不要で誰でも試すことができます。サンドボックスプランでは、200回の無料のGPT-4呼び出しが含まれています。
+
+- **Dify Community Editionのセルフホスティング</br>**
+この[スターターガイド](#quick-start)を使用して、環境でDifyをすばやく実行できます。
+さらなる参照や詳細な手順については、[ドキュメント](https://docs.dify.ai)をご覧ください。
+
+- **エンタープライズ/組織向けのDify</br>**
+追加のエンタープライズ向け機能を提供しています。[こちらからミーティ
+
+ングを予約](https://cal.com/guchenhe/30min)したり、[メールを送信](mailto:business@dify.ai?subject=[GitHub]Business%20License%20Inquiry)してエンタープライズのニーズについて相談してください。 </br>
+  > AWSを使用しているスタートアップや中小企業の場合は、[AWS Marketplace](https://aws.amazon.com/marketplace/pp/prodview-t22mebxzwjhu6)のDify Premiumをチェックして、ワンクリックで独自のAWS VPCにデプロイできます。カスタムロゴとブランディングでアプリを作成するオプションを備えた手頃な価格のAMIオファリングです。
+
+
+## 先を見る
+
+GitHubでDifyにスターを付け、新しいリリースをすぐに通知されます。
+
+![star-us](https://github.com/langgenius/dify/assets/13230914/b823edc1-6388-4e25-ad45-2f6b187adbb4)
+
+
+
+## クイックスタート
+> Difyをインストールする前に、マシンが以下の最小システム要件を満たしていることを確認してください：
+> 
+>- CPU >= 2コア
+>- RAM >= 4GB
+
+</br>
+
+Difyサーバーを起動する最も簡単な方法は、当社の[docker-compose.yml](docker/docker-compose.yaml)ファイルを実行することです。インストールコマンドを実行する前に、マシンに[Docker](https://docs.docker.com/get-docker/)と[Docker Compose](https://docs.docker.com/compose/install/)がインストールされていることを確認してください。
 
 ```bash
 cd docker
 docker compose up -d
 ```
 
-実行後、ブラウザで [http://localhost/install](http://localhost/install) にアクセスし、初期化インストールプロセスを開始できます。
+実行後、ブラウザで[http://localhost/install](http://localhost/install)にアクセスし、初期化プロセスを開始できます。
 
-### Helm Chart
+> Difyに貢献したり、追加の開発を行う場合は、[ソースコードからのデプロイガイド](https://docs.dify.ai/getting-started/install-self-hosted/local-source-code)を参照してください。
 
-@BorisPolonskyによる[Helm Chart](https://helm.sh/) バージョンを提供してくれて、大変感謝しています。これにより、DifyはKubernetes上にデプロイすることができます。
-デプロイ情報については、https://github.com/BorisPolonsky/dify-helm をご覧ください。
+## 次のステップ
 
-### 設定
+環境設定をカスタマイズする場合は、[docker-compose.yml](docker/docker-compose.yaml)ファイル内のコメントを参照して、環境設定を手動で設定してください。変更を加えた後は、再び `docker-compose up -d` を実行してください。環境変数の完全なリストは[こちら](https://docs.dify.ai/getting-started/install-self-hosted/environments)をご覧ください。
 
-設定をカスタマイズする必要がある場合は、[docker-compose.yml](docker/docker-compose.yaml) ファイルのコメントを参照し、環境設定を手動で行ってください。変更を行った後は、もう一度 `docker-compose up -d` を実行してください。環境変数の完全なリストは、[ドキュメント](https://docs.dify.ai/getting-started/install-self-hosted/environments)で確認できます。
+高可用性のセットアップを構成する場合は、コミュニティによって提供されている[Helm Charts](https://helm.sh/)があり、これによりKubernetes上にDifyを展開できます。
+
+- [Helm Chart by @LeoQuote](https://github.com/douban/charts/tree/master/charts/dify)
+- [Helm Chart by @BorisPolonsky](https://github.com/BorisPolonsky/dify-helm)
 
 
-## スターヒストリー
+## 貢献
 
-[![Star History Chart](https://api.star-history.com/svg?repos=langgenius/dify&type=Date)](https://star-history.com/#langgenius/dify&Date)
+コードに貢献したい方は、[Contribution Guide](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md)を参照してください。
+同時に、DifyをSNSやイベント、カンファレンスで共有してサポートしていただけると幸いです。
 
-## コミュニティとサポート
 
-Difyに貢献していただき、コードの提出、問題の報告、新しいアイデアの提供、またはDifyを基に作成した興味深く有用なAIアプリケーションの共有により、Difyをより良いものにするお手伝いを歓迎します。同時に、さまざまなイベント、会議、ソーシャルメディアでDifyを共有することも歓迎します。
+> Difyを英語または中国語以外の言語に翻訳してくれる貢献者を募集しています。興味がある場合は、詳細については[i18n README](https://github.com/langgenius/dify/blob/main/web/i18n/README.md)を参照してください。また、[Discordコミュニティサーバー](https://discord.gg/8Tpq4AcN9c)の`global-users`チャンネルにコメントを残してください。
 
-- [Github Discussion](https://github.com/langgenius/dify/discussions). 👉：アプリを共有し、コミュニティとコミュニケーション。
-- [GitHub Issues](https://github.com/langgenius/dify/issues)。最適な使用法：Dify.AIの使用中に遭遇するバグやエラー、[貢献ガイド](CONTRIBUTING.md)を参照。
-- [Email サポート](mailto:hello@dify.ai?subject=[GitHub]Questions%20About%20Dify)。最適な使用法：Dify.AIの使用に関する質問。
-- [Discord](https://discord.gg/FngNHpbcY7)。最適な使用法：アプリケーションの共有とコミュニティとの交流。
-- [Twitter](https://twitter.com/dify_ai)。最適な使用法：アプリケーションの共有とコミュニティとの交流。
-- [ビジネスライセンス](mailto:business@dify.ai?subject=[GitHub]Business%20License%20Inquiry)。最適な使用法：Dify.AIを商業利用するためのビジネス関連の問い合わせ。
+**貢献者**
 
-## セキュリティ
+<a href="https://github.com/langgenius/dify/graphs/contributors">
+  <img src="https://contrib.rocks/image?repo=langgenius/dify" />
+</a>
+
+## コミュニティ & お問い合わせ
+
+* [Github Discussion](https://github.com/langgenius/dify/discussions). 主に: フィードバックの共有や質問。
+* [GitHub Issues](https://github.com/langgenius/dify/issues). 主に: Dify.AIの使用中に遭遇したバグや機能提案。
+* [Email](mailto:support@dify.ai?subject=[GitHub]Questions%20About%20Dify). 主に: Dify.AIの使用に関する質問。
+* [Discord](https://discord.gg/FngNHpbcY7). 主に: アプリケーションの共有やコミュニティとの交流。
+* [Twitter](https://twitter.com/dify_ai). 主に: アプリケーションの共有やコミュニティとの交流。
+
+または、直接チームメンバーとミーティングをスケジュールします：
+
+<table>
+  <tr>
+    <th>連絡先</th>
+    <th>目的</th>
+  </tr>
+  <tr>
+    <td><a href='https://cal.com
+
+/guchenhe/30min'>ミーティング</a></td>
+    <td>無料の30分間のミーティングをスケジュールしてください。</td>
+  </tr>
+  <tr>
+    <td><a href='mailto:support@dify.ai?subject=[GitHub]Technical%20Support'>技術サポート</a></td>
+    <td>技術的な問題やサポートに関する質問</td>
+  </tr>
+  <tr>
+    <td><a href='mailto:business@dify.ai?subject=[GitHub]Business%20License%20Inquiry'>営業担当</a></td>
+    <td>法人ライセンスに関するお問い合わせ</td>
+  </tr>
+</table>
 
-プライバシー保護のため、GitHub へのセキュリティ問題の投稿は避けてください。代わりに、あなたの質問を security@dify.ai に送ってください。より詳細な回答を提供します。
 
 ## ライセンス
 
- このリポジトリは、基本的にApache 2.0にいくつかの追加制限を加えた[Difyオープンソースライセンス](LICENSE)の下で利用できます。
+プロジェクトはMITライセンスの下で利用可能です。[LICENSE](LICENSE)をご参照ください。

README_KL.md

@@ -1,119 +1,250 @@
-[![](./images/describe.png)](https://dify.ai)
+![cover-v5-optimized](https://github.com/langgenius/dify/assets/13230914/f9e19af5-61ba-4119-b926-d10c4c06ebab)
+
 <p align="center">
-  <a href="./README.md">English</a> |
-  <a href="./README_CN.md">简体中文</a> |
-  <a href="./README_JA.md">日本語</a> |
-  <a href="./README_ES.md">Español</a> |
-  <a href="./README_KL.md">Klingon</a> |
-  <a href="./README_FR.md">Français</a>
+  <a href="https://cloud.dify.ai">Dify Cloud</a> ·
+  <a href="https://docs.dify.ai/getting-started/install-self-hosted">Self-hosting</a> ·
+  <a href="https://docs.dify.ai">Documentation</a> ·
+  <a href="https://cal.com/guchenhe/dify-demo">Schedule demo</a>
 </p>
 
 <p align="center">
     <a href="https://dify.ai" target="_blank">
-        <img alt="Static Badge" src="https://img.shields.io/badge/AI-Dify?logo=AI&logoColor=%20%23f5f5f5&label=Dify&labelColor=%20%23155EEF&color=%23EAECF0"></a>
+        <img alt="Static Badge" src="https://img.shields.io/badge/Product-F04438"></a>
+    <a href="https://dify.ai/pricing" target="_blank">
+        <img alt="Static Badge" src="https://img.shields.io/badge/free-pricing?logo=free&color=%20%23155EEF&label=pricing&labelColor=%20%23528bff"></a>
     <a href="https://discord.gg/FngNHpbcY7" target="_blank">
-        <img src="https://img.shields.io/discord/1082486657678311454?logo=discord"
+        <img src="https://img.shields.io/discord/1082486657678311454?logo=discord&labelColor=%20%235462eb&logoColor=%20%23f5f5f5&color=%20%235462eb"
             alt="chat on Discord"></a>
     <a href="https://twitter.com/intent/follow?screen_name=dify_ai" target="_blank">
-        <img src="https://img.shields.io/twitter/follow/dify_ai?style=social&logo=X"
+        <img src="https://img.shields.io/twitter/follow/dify_ai?logo=X&color=%20%23f5f5f5"
             alt="follow on Twitter"></a>
     <a href="https://hub.docker.com/u/langgenius" target="_blank">
-        <img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/langgenius/dify-web"></a>
+        <img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/langgenius/dify-web?labelColor=%20%23FDB062&color=%20%23f79009"></a>
+    <a href="https://github.com/langgenius/dify/graphs/commit-activity" target="_blank">
+        <img alt="Commits last month" src="https://img.shields.io/github/commit-activity/m/langgenius/dify?labelColor=%20%2332b583&color=%20%2312b76a"></a>
+    <a href="https://github.com/langgenius/dify/" target="_blank">
+        <img alt="Issues closed" src="https://img.shields.io/github/issues-search?query=repo%3Alanggenius%2Fdify%20is%3Aclosed&label=issues%20closed&labelColor=%20%237d89b0&color=%20%235d6b98"></a>
+    <a href="https://github.com/langgenius/dify/discussions/" target="_blank">
+        <img alt="Discussion posts" src="https://img.shields.io/github/discussions/langgenius/dify?labelColor=%20%239b8afb&color=%20%237a5af8"></a>
 </p>
 
-**Dify** Hoch LLM qorwI' pIqoDvam pagh laHta' je **100,000** pIqoDvamvam Dify.AI De'wI'. Dify leghpu' Backend chu' a Service teH LLMOps vItlhutlh, generative AI-native pIqoD teq wa'vam, vIyoD Built-in RAG engine. Dify, **'ej chenmoHmoH Hoch 'oHna' Assistant API 'ej GPTmey HoStaHbogh LLMmey.**
+<p align="center">
+  <a href="./README.md"><img alt="Commits last month" src="https://img.shields.io/badge/English-d9d9d9"></a>
+  <a href="./README_CN.md"><img alt="Commits last month" src="https://img.shields.io/badge/简体中文-d9d9d9"></a>
+  <a href="./README_JA.md"><img alt="Commits last month" src="https://img.shields.io/badge/日本語-d9d9d9"></a>
+  <a href="./README_ES.md"><img alt="Commits last month" src="https://img.shields.io/badge/Español-d9d9d9"></a>
+  <a href="./README_KL.md"><img alt="Commits last month" src="https://img.shields.io/badge/Français-d9d9d9"></a>
+  <a href="./README_FR.md"><img alt="Commits last month" src="https://img.shields.io/badge/Klingon-d9d9d9"></a>
+</p>
 
-![](./images/demo.png)
+#
 
-## ngIl QaQ
+<p align="center">
+  <a href="https://trendshift.io/repositories/2152" target="_blank"><img src="https://trendshift.io/api/badge/repositories/2152" alt="langgenius%2Fdify | Trendshift" style="width: 250px; height: 55px;" width="250" height="55"/></a>
+</p>
+Dify is an open-source LLM app development platform. Its intuitive interface combines AI workflow, RAG pipeline, agent capabilities, model management, observability features and more, letting you quickly go from prototype to production. Here's a list of the core features:
+</br> </br>
 
-[Dify.AI ngIl](https://dify.ai) pIm neHlaH 'ej ghaH. cha'logh wa' DIvI' 200 GPT trial credits.
+**1. Workflow**: 
+  Build and test powerful AI workflows on a visual canvas, leveraging all the following features and beyond.
 
-## Dify WovmoH
 
-Dify Daq rIn neutrality 'ej Hoch, LangChain tInHar HubwI'. maH Daqbe'law' Qawqar, OpenAI's Assistant API Daq local neH deployment.
+  https://github.com/langgenius/dify/assets/13230914/356df23e-1604-483d-80a6-9517ece318aa
 
-| Qo'logh | Dify.AI | Assistants API | LangChain |
-|---------|---------|----------------|-----------|
-| **qet QaS** | API-oriented | API-oriented | Python Code-oriented |
-| **Ecosystem Strategy** | Open Source | Closed and Commercial | Open Source |
-| **RAG Engine** | Ha'qu' | Ha'qu' | ghoS Ha'qu' |
-| **Prompt IDE** | jaH Include | jaH Include | qeylIS qaq |
-| **qet LLMmey** | bo'Degh Hoch | GPTmey tIn | bo'Degh Hoch |
-| **local deployment** | Ha'qu' | tInHa'qu' | tInHa'qu' ghogh |
 
-## ruch
 
-![](./images/models.png)
+**2. Comprehensive model support**: 
+  Seamless integration with hundreds of proprietary / open-source LLMs from dozens of inference providers and self-hosted solutions, covering GPT, Mistral, Llama2, and any OpenAI API-compatible models. A full list of supported model providers can be found [here](https://docs.dify.ai/getting-started/readme/model-providers).
 
-**1. LLM tIq**: OpenAI's GPT Hur nISmoHvam neH vIngeH, wa' Llama2 Hur nISmoHvam. Heghlu'lu'pu' Dify mIw 'oH choH qay'be'.Daq commercial Hurmey 'ej Open Source Hurmey (maqtaHvIS pagh locally neH neH deployment HoSvam).
+![providers-v5](https://github.com/langgenius/dify/assets/13230914/5a17bdbe-097a-4100-8363-40255b70f6e3)
 
-**2. Prompt IDE**: cha'logh wa' LLMmey Hoch janlu'pu' 'ej lughpu' choH qay'be'.
 
-**3. RAG Engine**: RAG vaD tIqpu' lo'taH indexing qor neH vector database wa' embeddings wIj, PDFs, TXTs, 'ej ghojmoHmoH HIq qorlIj je upload.
+**3. Prompt IDE**: 
+  Intuitive interface for crafting prompts, comparing model performance, and adding additional features such as text-to-speech to a chat-based app. 
 
-**4. AI Agent**: Function Calling 'ej ReAct Daq Hurmey, Agent inference framework Hoch users customize tools, vaj 'oH QaQ. Dify Hoch loS ghaH 'ej wa'vatlh built-in tool calling capabilities, Google Search, DELL·E, Stable Diffusion, WolframAlpha, 'ej.
+**4. RAG Pipeline**: 
+  Extensive RAG capabilities that cover everything from document ingestion to retrieval, with out-of-box support for text extraction from PDFs, PPTs, and other common document formats.
 
-**5. QaS muDHa'wI': cha'logh wa' pIq mI' logs 'ej quv yIn, vItlhutlh tIq 'e'wIj lo'taHmoHmoH Prompts, vItlhutlh, Hurmey ghaH production data jatlh.
+**5. Agent capabilities**: 
+  You can define agents based on LLM Function Calling or ReAct, and add pre-built or custom tools for the agent. Dify provides 50+ built-in tools for AI agents, such as Google Search, DELL·E, Stable Diffusion and WolframAlpha.
 
-## Do'wI' qabmey lo'taH
+**6. LLMOps**: 
+  Monitor and analyze application logs and performance over time. You could continuously improve prompts, datasets, and models based on production data and annotations.
 
-**maHvaD jatlhchugh, GitHub Daq Hoch chu' ghompu'vam tIqel yInob!**
+**7. Backend-as-a-Service**: 
+  All of Dify's offerings come with corresponding APIs, so you could effortlessly integrate Dify into your own business logic.
 
-![star-us](https://github.com/langgenius/dify/assets/100913391/95f37259-7370-4456-a9f0-0bc01ef8642f)
 
-- [Website](https://dify.ai)
-- [Docs](https://docs.dify.ai)
-- [lo'taHmoH Docs](https://docs.dify.ai/getting-started/install-self-hosted)
-- [FAQ](https://docs.dify.ai/getting-started/faq) 
+## Feature Comparison
+<table style="width: 100%;">
+  <tr
 
-## Community Edition tu' yo'
+>
+    <th align="center">Feature</th>
+    <th align="center">Dify.AI</th>
+    <th align="center">LangChain</th>
+    <th align="center">Flowise</th>
+    <th align="center">OpenAI Assistants API</th>
+  </tr>
+  <tr>
+    <td align="center">Programming Approach</td>
+    <td align="center">API + App-oriented</td>
+    <td align="center">Python Code</td>
+    <td align="center">App-oriented</td>
+    <td align="center">API-oriented</td>
+  </tr>
+  <tr>
+    <td align="center">Supported LLMs</td>
+    <td align="center">Rich Variety</td>
+    <td align="center">Rich Variety</td>
+    <td align="center">Rich Variety</td>
+    <td align="center">OpenAI-only</td>
+  </tr>
+  <tr>
+    <td align="center">RAG Engine</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+  </tr>
+  <tr>
+    <td align="center">Agent</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+  </tr>
+  <tr>
+    <td align="center">Workflow</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+  </tr>
+  <tr>
+    <td align="center">Observability</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+    <td align="center">❌</td>
+  </tr>
+  <tr>
+    <td align="center">Enterprise Feature (SSO/Access control)</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+    <td align="center">❌</td>
+    <td align="center">❌</td>
+  </tr>
+  <tr>
+    <td align="center">Local Deployment</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">✅</td>
+    <td align="center">❌</td>
+  </tr>
+</table>
 
-### System Qab
+## Using Dify
 
-Dify yo' yo' qaqmeH SuS chenmoH 'oH qech!
+- **Cloud </br>**
+We host a [Dify Cloud](https://dify.ai) service for anyone to try with zero setup. It provides all the capabilities of the self-deployed version, and includes 200 free GPT-4 calls in the sandbox plan.
 
-- CPU >= 2 Cores
-- RAM >= 4GB
+- **Self-hosting Dify Community Edition</br>**
+Quickly get Dify running in your environment with this [starter guide](#quick-start).
+Use our [documentation](https://docs.dify.ai) for further references and more in-depth instructions.
 
-### Quick Start
+- **Dify for Enterprise / Organizations</br>**
+We provide additional enterprise-centric features. [Schedule a meeting with us](https://cal.com/guchenhe/30min) or [send us an email](mailto:business@dify.ai?subject=[GitHub]Business%20License%20Inquiry) to discuss enterprise needs. </br>
+  > For startups and small businesses using AWS, check out [Dify Premium on AWS Marketplace](https://aws.amazon.com/marketplace/pp/prodview-t22mebxzwjhu6) and deploy it to your own AWS VPC with one-click. It's an affordable AMI offering with the option to create apps with custom logo and branding.
 
-Dify server luHoHtaHlu' vIngeH lo'laHbe'chugh vIyoD [docker-compose.yml](docker/docker-compose.yaml) QorwI'ghach. toH yItlhutlh chenmoH luH!chugh 'ay' vaj vIneHmeH, 'ej [Docker](https://docs.docker.com/get-docker/) 'ej [Docker Compose](https://docs.docker.com/compose/install/) vaj 'oH 'e' vIneHmeH:
+
+## Staying ahead
+
+Star Dify on GitHub and be instantly notified of new releases.
+
+![star-us](https://github.com/langgenius/dify/assets/13230914/b823edc1-6388-4e25-ad45-2f6b187adbb4)
+
+
+
+## Quick Start
+> Before installing Dify, make sure your machine meets the following minimum system requirements:
+> 
+>- CPU >= 2 Core
+>- RAM >= 4GB
+
+</br>
+
+The easiest way to start the Dify server is to run our [docker-compose.yml](docker/docker-compose.yaml) file. Before running the installation command, make sure that [Docker](https://docs.docker.com/get-docker/) and [Docker Compose](https://docs.docker.com/compose/install/) are installed on your machine:
 
 ```bash
 cd docker
 docker compose up -d
 ```
 
-luHoHtaHmeH HoHtaHvIS, Dify dashboard vIneHmeH vIngeH lI'wI' [http://localhost/install](http://localhost/install) 'ej 'oH initialization 'e' vIneHmeH.
+After running, you can access the Dify dashboard in your browser at [http://localhost/install](http://localhost/install) and start the initialization process.
 
-### Helm Chart
+> If you'd like to contribute to Dify or do additional development, refer to our [guide to deploying from source code](https://docs.dify.ai/getting-started/install-self-hosted/local-source-code)
 
-@BorisPolonsky Dify wIq tIq ['ay'var (Helm Chart)](https://helm.sh/) version Hur yIn chu' Dify luHoHchu'. Heghlu'lu' vIneHmeH [https://github.com/BorisPolonsky/dify-helm](https://github.com/BorisPolonsky/dify-helm) 'ej vaj QaS deployment information.
+## Next steps
 
-### veS config
+If you need to customize the configuration, please refer to the comments in our [docker-compose.yml](docker/docker-compose.yaml) file and manually set the environment configuration. After making the changes, please run `docker-compose up -d` again. You can see the full list of environment variables [here](https://docs.dify.ai/getting-started/install-self-hosted/environments).
 
-chenmoHDI' config lo'taH ghaH, vItlhutlh HIq wIgharghbe'lu'pu'. toH lo'taHvIS pagh vay' vIneHmeH, 'ej `docker-compose up -d` wa'DIch. tIqmoHmeH list full wa' lo'taHvo'lu'pu' ghaH [docs](https://docs.dify.ai/getting-started/install-self-hosted/environments).
+If you'd like to configure a highly-available setup, there are community-contributed [Helm Charts](https://helm.sh/) which allow Dify to be deployed on Kubernetes.
 
-## tIng qem
+- [Helm Chart by @LeoQuote](https://github.com/douban/charts/tree/master/charts/dify)
+- [Helm Chart by @BorisPolonsky](https://github.com/BorisPolonsky/dify-helm)
 
-[![tIng qem Hur Chart](https://api.star-history.com/svg?repos=langgenius/dify&type=Date)](https://star-history.com/#langgenius/dify&Date)
 
-## choHmoH 'ej vItlhutlh
+## Contributing
 
-Dify choHmoH je mIw Dify puqloD, Dify ghaHta'bogh vItlhutlh, HurDI' code, ghItlh, ghItlh qo'lu'pu'pu' qej. tIqmeH, Hurmey je, Dify Hur tIqDI' woDDaj, DuD QangmeH 'ej HInobDaq vItlhutlh HImej Dify'e'.
+For those who'd like to contribute code, see our [Contribution Guide](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md). 
+At the same time, please consider supporting Dify by sharing it on social media and at events and conferences.
 
-- [GitHub vItlhutlh](https://github.com/langgenius/dify/issues). Hurmey: bugs 'ej errors Dify.AI tIqmeH. yImej [Contribution Guide](CONTRIBUTING.md).
-- [Email QaH](mailto:hello@dify.ai?subject=[GitHub]Questions%20About%20Dify). Hurmey: questions vItlhutlh Dify.AI chaw'.
-- [Discord](https://discord.gg/FngNHpbcY7). Hurmey: jIpuv 'ej jImej mIw Dify vItlhutlh.
-- [Twitter](https://twitter.com/dify_ai). Hurmey: jIpuv 'ej jImej mIw Dify vItlhutlh.
-- [Business License](mailto:business@dify.ai?subject=[GitHub]Business%20License%20Inquiry). Hurmey: qurgh vItlhutlh Hurmey Dify.AI tIqbe'law'.
 
-## bIQDaqmey bom
+> 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).
 
-taghlI' vIngeH'a'? pong security 'oH posting GitHub. yItlhutlh, toH security@dify.ai 'ej vIngeH'a'.
+**Contributors**
+
+<a href="https://github.com/langgenius/dify/graphs/contributors">
+  <img src="https://contrib.rocks/image?repo=langgenius/dify" />
+</a>
+
+## Community & Contact
+
+* [Github Discussion](https://github.com/langgenius/dify/discussions
+
+). Best for: sharing feedback and asking questions.
+* [GitHub Issues](https://github.com/langgenius/dify/issues). Best for: bugs you encounter using Dify.AI, and feature proposals. See our [Contribution Guide](https://github.com/langgenius/dify/blob/main/CONTRIBUTING.md).
+* [Email](mailto:support@dify.ai?subject=[GitHub]Questions%20About%20Dify). Best for: questions you have about using Dify.AI.
+* [Discord](https://discord.gg/FngNHpbcY7). Best for: sharing your applications and hanging out with the community.
+* [Twitter](https://twitter.com/dify_ai). Best for: sharing your applications and hanging out with the community.
+
+Or, schedule a meeting directly with a team member:
+
+<table>
+  <tr>
+    <th>Point of Contact</th>
+    <th>Purpose</th>
+  </tr>
+  <tr>
+    <td><a href='https://cal.com/guchenhe/15min' target='_blank'><img class="schedule-button" src='https://github.com/langgenius/dify/assets/13230914/9ebcd111-1205-4d71-83d5-948d70b809f5' alt='Git-Hub-README-Button-3x' style="width: 180px; height: auto; object-fit: contain;"/></a></td>
+    <td>Business enquiries & product feedback</td>
+  </tr>
+  <tr>
+    <td><a href='https://cal.com/pinkbanana' target='_blank'><img class="schedule-button" src='https://github.com/langgenius/dify/assets/13230914/d1edd00a-d7e4-4513-be6c-e57038e143fd' alt='Git-Hub-README-Button-2x' style="width: 180px; height: auto; object-fit: contain;"/></a></td>
+    <td>Contributions, issues & feature requests</td>
+  </tr>
+</table>
+
+## Star History
+
+[![Star History Chart](https://api.star-history.com/svg?repos=langgenius/dify&type=Date)](https://star-history.com/#langgenius/dify&Date)
+
+
+## Security Disclosure
+
+To protect your privacy, please avoid posting security issues on GitHub. Instead, send your questions to security@dify.ai and we will provide you with a more detailed answer.
 
 ## License
 
-ghItlh puqloD chenmoH [Dify vItlhutlh Hur](LICENSE), ghaH nIvbogh Apache 2.0.
-
+This repository is available under the [Dify Open Source License](LICENSE), which is essentially Apache 2.0 with a few additional restrictions.

api/.env.example

@@ -57,7 +57,7 @@ AZURE_BLOB_ACCOUNT_URL=https://<your_account_name>.blob.core.windows.net
 WEB_API_CORS_ALLOW_ORIGINS=http://127.0.0.1:3000,*
 CONSOLE_CORS_ALLOW_ORIGINS=http://127.0.0.1:3000,*
 
-# Vector database configuration, support: weaviate, qdrant, milvus
+# Vector database configuration, support: weaviate, qdrant, milvus, relyt
 VECTOR_STORE=weaviate
 
 # Weaviate configuration
@@ -78,6 +78,13 @@ MILVUS_USER=root
 MILVUS_PASSWORD=Milvus
 MILVUS_SECURE=false
 
+# Relyt configuration
+RELYT_HOST=127.0.0.1
+RELYT_PORT=5432
+RELYT_USER=postgres
+RELYT_PASSWORD=postgres
+RELYT_DATABASE=postgres
+
 # Upload configuration
 UPLOAD_FILE_SIZE_LIMIT=15
 UPLOAD_FILE_BATCH_LIMIT=5
@@ -149,3 +156,7 @@ TEMPLATE_TRANSFORM_MAX_LENGTH=80000
 CODE_MAX_STRING_ARRAY_LENGTH=30
 CODE_MAX_OBJECT_ARRAY_LENGTH=30
 CODE_MAX_NUMBER_ARRAY_LENGTH=1000
+
+# API Tool configuration
+API_TOOL_DEFAULT_CONNECT_TIMEOUT=10
+API_TOOL_DEFAULT_READ_TIMEOUT=60

api/app.py

@@ -1,16 +1,13 @@
 import os
 
-from werkzeug.exceptions import Unauthorized
-
 if not os.environ.get("DEBUG") or os.environ.get("DEBUG").lower() != 'true':
     from gevent import monkey
+
     monkey.patch_all()
     # if os.environ.get("VECTOR_STORE") == 'milvus':
     import grpc.experimental.gevent
-    grpc.experimental.gevent.init_gevent()
 
-    import langchain
-    langchain.verbose = True
+    grpc.experimental.gevent.init_gevent()
 
 import json
 import logging
@@ -21,6 +18,7 @@ import warnings
 from flask import Flask, Response, request
 from flask_cors import CORS
 
+from werkzeug.exceptions import Unauthorized
 from commands import register_commands
 from config import CloudEditionConfig, Config
 from extensions import (
@@ -44,6 +42,7 @@ from services.account_service import AccountService
 # DO NOT REMOVE BELOW
 from events import event_handlers
 from models import account, dataset, model, source, task, tool, tools, web
+
 # DO NOT REMOVE ABOVE
 
 
@@ -51,7 +50,7 @@ warnings.simplefilter("ignore", ResourceWarning)
 
 # fix windows platform
 if os.name == "nt":
-    os.system('tzutil /s "UTC"')    
+    os.system('tzutil /s "UTC"')
 else:
     os.environ['TZ'] = 'UTC'
     time.tzset()
@@ -60,6 +59,7 @@ else:
 class DifyApp(Flask):
     pass
 
+
 # -------------
 # Configuration
 # -------------
@@ -67,6 +67,7 @@ class DifyApp(Flask):
 
 config_type = os.getenv('EDITION', default='SELF_HOSTED')  # ce edition first
 
+
 # ----------------------------
 # Application Factory Function
 # ----------------------------
@@ -192,7 +193,6 @@ def register_blueprints(app):
 app = create_app()
 celery = app.extensions["celery"]
 
-
 if app.config['TESTING']:
     print("App is running in TESTING mode")
 

api/commands.py

@@ -297,6 +297,14 @@ def migrate_knowledge_vector_database():
                         "vector_store": {"class_prefix": collection_name}
                     }
                     dataset.index_struct = json.dumps(index_struct_dict)
+                elif vector_type == "relyt":
+                    dataset_id = dataset.id
+                    collection_name = Dataset.gen_collection_name_by_id(dataset_id)
+                    index_struct_dict = {
+                        "type": 'relyt',
+                        "vector_store": {"class_prefix": collection_name}
+                    }
+                    dataset.index_struct = json.dumps(index_struct_dict)
                 else:
                     raise ValueError(f"Vector store {config.get('VECTOR_STORE')} is not supported.")
 

api/config.py

@@ -64,8 +64,8 @@ DEFAULTS = {
     'ETL_TYPE': 'dify',
     'KEYWORD_STORE': 'jieba',
     'BATCH_UPLOAD_LIMIT': 20,
-    'CODE_EXECUTION_ENDPOINT': '',
-    'CODE_EXECUTION_API_KEY': '',
+    'CODE_EXECUTION_ENDPOINT': 'http://sandbox:8194',
+    'CODE_EXECUTION_API_KEY': 'dify-sandbox',
     'TOOL_ICON_CACHE_MAX_AGE': 3600,
     'MILVUS_DATABASE': 'default',
     'KEYWORD_DATA_SOURCE_TYPE': 'database',
@@ -99,7 +99,7 @@ class Config:
         # ------------------------
         # General Configurations.
         # ------------------------
-        self.CURRENT_VERSION = "0.6.2"
+        self.CURRENT_VERSION = "0.6.3"
         self.COMMIT_SHA = get_env('COMMIT_SHA')
         self.EDITION = "SELF_HOSTED"
         self.DEPLOY_ENV = get_env('DEPLOY_ENV')
@@ -198,7 +198,7 @@ class Config:
 
         # ------------------------
         # Vector Store Configurations.
-        # Currently, only support: qdrant, milvus, zilliz, weaviate
+        # Currently, only support: qdrant, milvus, zilliz, weaviate, relyt
         # ------------------------
         self.VECTOR_STORE = get_env('VECTOR_STORE')
         self.KEYWORD_STORE = get_env('KEYWORD_STORE')
@@ -221,6 +221,13 @@ class Config:
         self.WEAVIATE_GRPC_ENABLED = get_bool_env('WEAVIATE_GRPC_ENABLED')
         self.WEAVIATE_BATCH_SIZE = int(get_env('WEAVIATE_BATCH_SIZE'))
 
+        # relyt settings
+        self.RELYT_HOST = get_env('RELYT_HOST')
+        self.RELYT_PORT = get_env('RELYT_PORT')
+        self.RELYT_USER = get_env('RELYT_USER')
+        self.RELYT_PASSWORD = get_env('RELYT_PASSWORD')
+        self.RELYT_DATABASE = get_env('RELYT_DATABASE')
+
         # ------------------------
         # Mail Configurations.
         # ------------------------

api/controllers/console/app/conversation.py

@@ -1,4 +1,4 @@
-from datetime import datetime
+from datetime import datetime, timezone
 
 import pytz
 from flask_login import current_user
@@ -262,7 +262,7 @@ def _get_conversation(app_model, conversation_id):
         raise NotFound("Conversation Not Exists.")
 
     if not conversation.read_at:
-        conversation.read_at = datetime.utcnow()
+        conversation.read_at = datetime.now(timezone.utc).replace(tzinfo=None)
         conversation.read_account_id = current_user.id
         db.session.commit()
 

api/controllers/console/auth/activate.py

@@ -1,6 +1,6 @@
 import base64
+import datetime
 import secrets
-from datetime import datetime
 
 from flask_restful import Resource, reqparse
 
@@ -66,7 +66,7 @@ class ActivateApi(Resource):
         account.timezone = args['timezone']
         account.interface_theme = 'light'
         account.status = AccountStatus.ACTIVE.value
-        account.initialized_at = datetime.utcnow()
+        account.initialized_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
         db.session.commit()
 
         return {'result': 'success'}

api/controllers/console/auth/oauth.py

@@ -1,5 +1,5 @@
 import logging
-from datetime import datetime
+from datetime import datetime, timezone
 from typing import Optional
 
 import requests
@@ -73,7 +73,7 @@ class OAuthCallback(Resource):
 
         if account.status == AccountStatus.PENDING.value:
             account.status = AccountStatus.ACTIVE.value
-            account.initialized_at = datetime.utcnow()
+            account.initialized_at = datetime.now(timezone.utc).replace(tzinfo=None)
             db.session.commit()
 
         TenantService.create_owner_tenant_if_not_exist(account)

api/controllers/console/datasets/data_source.py

@@ -80,7 +80,7 @@ class DataSourceApi(Resource):
         if action == 'enable':
             if data_source_binding.disabled:
                 data_source_binding.disabled = False
-                data_source_binding.updated_at = datetime.datetime.utcnow()
+                data_source_binding.updated_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
                 db.session.add(data_source_binding)
                 db.session.commit()
             else:
@@ -89,7 +89,7 @@ class DataSourceApi(Resource):
         if action == 'disable':
             if not data_source_binding.disabled:
                 data_source_binding.disabled = True
-                data_source_binding.updated_at = datetime.datetime.utcnow()
+                data_source_binding.updated_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
                 db.session.add(data_source_binding)
                 db.session.commit()
             else:

api/controllers/console/datasets/datasets_document.py

@@ -1,4 +1,4 @@
-from datetime import datetime
+from datetime import datetime, timezone
 
 from flask import request
 from flask_login import current_user
@@ -637,7 +637,7 @@ class DocumentProcessingApi(DocumentResource):
                 raise InvalidActionError('Document not in indexing state.')
 
             document.paused_by = current_user.id
-            document.paused_at = datetime.utcnow()
+            document.paused_at = datetime.now(timezone.utc).replace(tzinfo=None)
             document.is_paused = True
             db.session.commit()
 
@@ -717,7 +717,7 @@ class DocumentMetadataApi(DocumentResource):
                     document.doc_metadata[key] = value
 
         document.doc_type = doc_type
-        document.updated_at = datetime.utcnow()
+        document.updated_at = datetime.now(timezone.utc).replace(tzinfo=None)
         db.session.commit()
 
         return {'result': 'success', 'message': 'Document metadata updated.'}, 200
@@ -755,7 +755,7 @@ class DocumentStatusApi(DocumentResource):
             document.enabled = True
             document.disabled_at = None
             document.disabled_by = None
-            document.updated_at = datetime.utcnow()
+            document.updated_at = datetime.now(timezone.utc).replace(tzinfo=None)
             db.session.commit()
 
             # Set cache to prevent indexing the same document multiple times
@@ -772,9 +772,9 @@ class DocumentStatusApi(DocumentResource):
                 raise InvalidActionError('Document already disabled.')
 
             document.enabled = False
-            document.disabled_at = datetime.utcnow()
+            document.disabled_at = datetime.now(timezone.utc).replace(tzinfo=None)
             document.disabled_by = current_user.id
-            document.updated_at = datetime.utcnow()
+            document.updated_at = datetime.now(timezone.utc).replace(tzinfo=None)
             db.session.commit()
 
             # Set cache to prevent indexing the same document multiple times
@@ -789,9 +789,9 @@ class DocumentStatusApi(DocumentResource):
                 raise InvalidActionError('Document already archived.')
 
             document.archived = True
-            document.archived_at = datetime.utcnow()
+            document.archived_at = datetime.now(timezone.utc).replace(tzinfo=None)
             document.archived_by = current_user.id
-            document.updated_at = datetime.utcnow()
+            document.updated_at = datetime.now(timezone.utc).replace(tzinfo=None)
             db.session.commit()
 
             if document.enabled:
@@ -808,7 +808,7 @@ class DocumentStatusApi(DocumentResource):
             document.archived = False
             document.archived_at = None
             document.archived_by = None
-            document.updated_at = datetime.utcnow()
+            document.updated_at = datetime.now(timezone.utc).replace(tzinfo=None)
             db.session.commit()
 
             # Set cache to prevent indexing the same document multiple times

api/controllers/console/datasets/datasets_segments.py

@@ -1,5 +1,5 @@
 import uuid
-from datetime import datetime
+from datetime import datetime, timezone
 
 import pandas as pd
 from flask import request
@@ -192,7 +192,7 @@ class DatasetDocumentSegmentApi(Resource):
                 raise InvalidActionError("Segment is already disabled.")
 
             segment.enabled = False
-            segment.disabled_at = datetime.utcnow()
+            segment.disabled_at = datetime.now(timezone.utc).replace(tzinfo=None)
             segment.disabled_by = current_user.id
             db.session.commit()
 

api/controllers/console/explore/completion.py

@@ -1,5 +1,5 @@
 import logging
-from datetime import datetime
+from datetime import datetime, timezone
 
 from flask_login import current_user
 from flask_restful import reqparse
@@ -47,7 +47,7 @@ class CompletionApi(InstalledAppResource):
         streaming = args['response_mode'] == 'streaming'
         args['auto_generate_name'] = False
 
-        installed_app.last_used_at = datetime.utcnow()
+        installed_app.last_used_at = datetime.now(timezone.utc).replace(tzinfo=None)
         db.session.commit()
 
         try:
@@ -110,7 +110,7 @@ class ChatApi(InstalledAppResource):
 
         args['auto_generate_name'] = False
 
-        installed_app.last_used_at = datetime.utcnow()
+        installed_app.last_used_at = datetime.now(timezone.utc).replace(tzinfo=None)
         db.session.commit()
 
         try:

api/controllers/console/explore/installed_app.py

@@ -1,4 +1,4 @@
-from datetime import datetime
+from datetime import datetime, timezone
 
 from flask_login import current_user
 from flask_restful import Resource, inputs, marshal_with, reqparse
@@ -81,7 +81,7 @@ class InstalledAppsListApi(Resource):
                 tenant_id=current_tenant_id,
                 app_owner_tenant_id=app.tenant_id,
                 is_pinned=False,
-                last_used_at=datetime.utcnow()
+                last_used_at=datetime.now(timezone.utc).replace(tzinfo=None)
             )
             db.session.add(new_installed_app)
             db.session.commit()

api/controllers/console/workspace/account.py

@@ -1,4 +1,4 @@
-from datetime import datetime
+import datetime
 
 import pytz
 from flask import current_app, request
@@ -59,7 +59,7 @@ class AccountInitApi(Resource):
                 raise InvalidInvitationCodeError()
 
             invitation_code.status = 'used'
-            invitation_code.used_at = datetime.utcnow()
+            invitation_code.used_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
             invitation_code.used_by_tenant_id = account.current_tenant_id
             invitation_code.used_by_account_id = account.id
 
@@ -67,7 +67,7 @@ class AccountInitApi(Resource):
         account.timezone = args['timezone']
         account.interface_theme = 'light'
         account.status = 'active'
-        account.initialized_at = datetime.utcnow()
+        account.initialized_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
         db.session.commit()
 
         return {'result': 'success'}

api/controllers/service_api/wraps.py

@@ -1,5 +1,5 @@
 from collections.abc import Callable
-from datetime import datetime
+from datetime import datetime, timezone
 from enum import Enum
 from functools import wraps
 from typing import Optional
@@ -183,7 +183,7 @@ def validate_and_get_api_token(scope=None):
     if not api_token:
         raise Unauthorized("Access token is invalid")
 
-    api_token.last_used_at = datetime.utcnow()
+    api_token.last_used_at = datetime.now(timezone.utc).replace(tzinfo=None)
     db.session.commit()
 
     return api_token

api/core/agent/base_agent_runner.py

@@ -1,7 +1,7 @@
 import json
 import logging
 import uuid
-from datetime import datetime
+from datetime import datetime, timezone
 from typing import Optional, Union, cast
 
 from core.agent.entities import AgentEntity, AgentToolEntity
@@ -440,7 +440,7 @@ class BaseAgentRunner(AppRunner):
             ToolConversationVariables.conversation_id == self.message.conversation_id,
         ).first()
 
-        db_variables.updated_at = datetime.utcnow()
+        db_variables.updated_at = datetime.now(timezone.utc).replace(tzinfo=None)
         db_variables.variables_str = json.dumps(jsonable_encoder(tool_variables.pool))
         db.session.commit()
         db.session.close()

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

@@ -1,4 +1,5 @@
 import logging
+import os
 import threading
 import uuid
 from collections.abc import Generator
@@ -189,6 +190,8 @@ class AdvancedChatAppGenerator(MessageBasedAppGenerator):
                 logger.exception("Validation Error when generating")
                 queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
             except (ValueError, InvokeError) as e:
+                if os.environ.get("DEBUG") and os.environ.get("DEBUG").lower() == 'true':
+                    logger.exception("Error when generating")
                 queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
             except Exception as e:
                 logger.exception("Unknown Error when generating")

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

@@ -98,6 +98,7 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
         )
 
         self._stream_generate_routes = self._get_stream_generate_routes()
+        self._conversation_name_generate_thread = None
 
     def process(self) -> Union[ChatbotAppBlockingResponse, Generator[ChatbotAppStreamResponse, None, None]]:
         """
@@ -108,6 +109,12 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
         db.session.refresh(self._user)
         db.session.close()
 
+        # start generate conversation name thread
+        self._conversation_name_generate_thread = self._generate_conversation_name(
+            self._conversation,
+            self._application_generate_entity.query
+        )
+
         generator = self._process_stream_response()
         if self._stream:
             return self._to_stream_response(generator)
@@ -278,6 +285,9 @@ class AdvancedChatAppGenerateTaskPipeline(BasedGenerateTaskPipeline, WorkflowCyc
             else:
                 continue
 
+        if self._conversation_name_generate_thread:
+            self._conversation_name_generate_thread.join()
+
     def _save_message(self) -> None:
         """
         Save message.

api/core/app/apps/agent_chat/app_generator.py

@@ -1,4 +1,5 @@
 import logging
+import os
 import threading
 import uuid
 from collections.abc import Generator
@@ -198,6 +199,8 @@ class AgentChatAppGenerator(MessageBasedAppGenerator):
                 logger.exception("Validation Error when generating")
                 queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
             except (ValueError, InvokeError) as e:
+                if os.environ.get("DEBUG") and os.environ.get("DEBUG").lower() == 'true':
+                    logger.exception("Error when generating")
                 queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
             except Exception as e:
                 logger.exception("Unknown Error when generating")

api/core/app/apps/chat/app_generator.py

@@ -1,4 +1,5 @@
 import logging
+import os
 import threading
 import uuid
 from collections.abc import Generator
@@ -195,6 +196,8 @@ class ChatAppGenerator(MessageBasedAppGenerator):
                 logger.exception("Validation Error when generating")
                 queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
             except (ValueError, InvokeError) as e:
+                if os.environ.get("DEBUG") and os.environ.get("DEBUG").lower() == 'true':
+                    logger.exception("Error when generating")
                 queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
             except Exception as e:
                 logger.exception("Unknown Error when generating")

api/core/app/apps/completion/app_generator.py

@@ -1,4 +1,5 @@
 import logging
+import os
 import threading
 import uuid
 from collections.abc import Generator
@@ -184,6 +185,8 @@ class CompletionAppGenerator(MessageBasedAppGenerator):
                 logger.exception("Validation Error when generating")
                 queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
             except (ValueError, InvokeError) as e:
+                if os.environ.get("DEBUG") and os.environ.get("DEBUG").lower() == 'true':
+                    logger.exception("Error when generating")
                 queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
             except Exception as e:
                 logger.exception("Unknown Error when generating")

api/core/app/apps/workflow/app_generator.py

@@ -1,4 +1,5 @@
 import logging
+import os
 import threading
 import uuid
 from collections.abc import Generator
@@ -137,6 +138,8 @@ class WorkflowAppGenerator(BaseAppGenerator):
                 logger.exception("Validation Error when generating")
                 queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
             except (ValueError, InvokeError) as e:
+                if os.environ.get("DEBUG") and os.environ.get("DEBUG").lower() == 'true':
+                    logger.exception("Error when generating")
                 queue_manager.publish_error(e, PublishFrom.APPLICATION_MANAGER)
             except Exception as e:
                 logger.exception("Unknown Error when generating")

api/core/app/task_pipeline/easy_ui_based_generate_task_pipeline.py

@@ -97,6 +97,8 @@ class EasyUIBasedGenerateTaskPipeline(BasedGenerateTaskPipeline, MessageCycleMan
             )
         )
 
+        self._conversation_name_generate_thread = None
+
     def process(self) -> Union[
         ChatbotAppBlockingResponse,
         CompletionAppBlockingResponse,
@@ -110,6 +112,13 @@ class EasyUIBasedGenerateTaskPipeline(BasedGenerateTaskPipeline, MessageCycleMan
         db.session.refresh(self._message)
         db.session.close()
 
+        if self._application_generate_entity.app_config.app_mode != AppMode.COMPLETION:
+            # start generate conversation name thread
+            self._conversation_name_generate_thread = self._generate_conversation_name(
+                self._conversation,
+                self._application_generate_entity.query
+            )
+
         generator = self._process_stream_response()
         if self._stream:
             return self._to_stream_response(generator)
@@ -256,6 +265,9 @@ class EasyUIBasedGenerateTaskPipeline(BasedGenerateTaskPipeline, MessageCycleMan
             else:
                 continue
 
+        if self._conversation_name_generate_thread:
+            self._conversation_name_generate_thread.join()
+
     def _save_message(self) -> None:
         """
         Save message.

api/core/app/task_pipeline/message_cycle_manage.py

@@ -1,5 +1,8 @@
+from threading import Thread
 from typing import Optional, Union
 
+from flask import Flask, current_app
+
 from core.app.entities.app_invoke_entities import (
     AdvancedChatAppGenerateEntity,
     AgentChatAppGenerateEntity,
@@ -19,9 +22,10 @@ from core.app.entities.task_entities import (
     MessageReplaceStreamResponse,
     MessageStreamResponse,
 )
+from core.llm_generator.llm_generator import LLMGenerator
 from core.tools.tool_file_manager import ToolFileManager
 from extensions.ext_database import db
-from models.model import MessageAnnotation, MessageFile
+from models.model import AppMode, Conversation, MessageAnnotation, MessageFile
 from services.annotation_service import AppAnnotationService
 
 
@@ -34,6 +38,59 @@ class MessageCycleManage:
     ]
     _task_state: Union[EasyUITaskState, AdvancedChatTaskState]
 
+    def _generate_conversation_name(self, conversation: Conversation, query: str) -> Optional[Thread]:
+        """
+        Generate conversation name.
+        :param conversation: conversation
+        :param query: query
+        :return: thread
+        """
+        is_first_message = self._application_generate_entity.conversation_id is None
+        extras = self._application_generate_entity.extras
+        auto_generate_conversation_name = extras.get('auto_generate_conversation_name', True)
+
+        if auto_generate_conversation_name and is_first_message:
+            # start generate thread
+            thread = Thread(target=self._generate_conversation_name_worker, kwargs={
+                'flask_app': current_app._get_current_object(),
+                'conversation_id': conversation.id,
+                'query': query
+            })
+
+            thread.start()
+
+            return thread
+
+        return None
+
+    def _generate_conversation_name_worker(self,
+                                           flask_app: Flask,
+                                           conversation_id: str,
+                                           query: str):
+        with flask_app.app_context():
+            # get conversation and message
+            conversation = (
+                db.session.query(Conversation)
+                .filter(Conversation.id == conversation_id)
+                .first()
+            )
+
+            if conversation.mode != AppMode.COMPLETION.value:
+                app_model = conversation.app
+                if not app_model:
+                    return
+
+                # generate conversation name
+                try:
+                    name = LLMGenerator.generate_conversation_name(app_model.tenant_id, query)
+                    conversation.name = name
+                except:
+                    pass
+
+                db.session.merge(conversation)
+                db.session.commit()
+                db.session.close()
+
     def _handle_annotation_reply(self, event: QueueAnnotationReplyEvent) -> Optional[MessageAnnotation]:
         """
         Handle annotation reply.

api/core/app/task_pipeline/workflow_cycle_manage.py

@@ -1,6 +1,6 @@
 import json
 import time
-from datetime import datetime
+from datetime import datetime, timezone
 from typing import Any, Optional, Union, cast
 
 from core.app.entities.app_invoke_entities import AdvancedChatAppGenerateEntity, InvokeFrom, WorkflowAppGenerateEntity
@@ -120,7 +120,7 @@ class WorkflowCycleManage:
         workflow_run.elapsed_time = time.perf_counter() - start_at
         workflow_run.total_tokens = total_tokens
         workflow_run.total_steps = total_steps
-        workflow_run.finished_at = datetime.utcnow()
+        workflow_run.finished_at = datetime.now(timezone.utc).replace(tzinfo=None)
 
         db.session.commit()
         db.session.refresh(workflow_run)
@@ -149,7 +149,7 @@ class WorkflowCycleManage:
         workflow_run.elapsed_time = time.perf_counter() - start_at
         workflow_run.total_tokens = total_tokens
         workflow_run.total_steps = total_steps
-        workflow_run.finished_at = datetime.utcnow()
+        workflow_run.finished_at = datetime.now(timezone.utc).replace(tzinfo=None)
 
         db.session.commit()
         db.session.refresh(workflow_run)
@@ -223,7 +223,7 @@ class WorkflowCycleManage:
         workflow_node_execution.outputs = json.dumps(outputs) if outputs else None
         workflow_node_execution.execution_metadata = json.dumps(jsonable_encoder(execution_metadata)) \
             if execution_metadata else None
-        workflow_node_execution.finished_at = datetime.utcnow()
+        workflow_node_execution.finished_at = datetime.now(timezone.utc).replace(tzinfo=None)
 
         db.session.commit()
         db.session.refresh(workflow_node_execution)
@@ -251,7 +251,7 @@ class WorkflowCycleManage:
         workflow_node_execution.status = WorkflowNodeExecutionStatus.FAILED.value
         workflow_node_execution.error = error
         workflow_node_execution.elapsed_time = time.perf_counter() - start_at
-        workflow_node_execution.finished_at = datetime.utcnow()
+        workflow_node_execution.finished_at = datetime.now(timezone.utc).replace(tzinfo=None)
         workflow_node_execution.inputs = json.dumps(inputs) if inputs else None
         workflow_node_execution.process_data = json.dumps(process_data) if process_data else None
         workflow_node_execution.outputs = json.dumps(outputs) if outputs else None

api/core/entities/message_entities.py

@@ -1,19 +1,8 @@
 import enum
-from typing import Any, cast
+from typing import Any
 
-from langchain.schema import AIMessage, BaseMessage, FunctionMessage, HumanMessage, SystemMessage
 from pydantic import BaseModel
 
-from core.model_runtime.entities.message_entities import (
-    AssistantPromptMessage,
-    ImagePromptMessageContent,
-    PromptMessage,
-    SystemPromptMessage,
-    TextPromptMessageContent,
-    ToolPromptMessage,
-    UserPromptMessage,
-)
-
 
 class PromptMessageFileType(enum.Enum):
     IMAGE = 'image'
@@ -38,98 +27,3 @@ class ImagePromptMessageFile(PromptMessageFile):
 
     type: PromptMessageFileType = PromptMessageFileType.IMAGE
     detail: DETAIL = DETAIL.LOW
-
-
-class LCHumanMessageWithFiles(HumanMessage):
-    # content: Union[str, list[Union[str, Dict]]]
-    content: str
-    files: list[PromptMessageFile]
-
-
-def lc_messages_to_prompt_messages(messages: list[BaseMessage]) -> list[PromptMessage]:
-    prompt_messages = []
-    for message in messages:
-        if isinstance(message, HumanMessage):
-            if isinstance(message, LCHumanMessageWithFiles):
-                file_prompt_message_contents = []
-                for file in message.files:
-                    if file.type == PromptMessageFileType.IMAGE:
-                        file = cast(ImagePromptMessageFile, file)
-                        file_prompt_message_contents.append(ImagePromptMessageContent(
-                            data=file.data,
-                            detail=ImagePromptMessageContent.DETAIL.HIGH
-                            if file.detail.value == "high" else ImagePromptMessageContent.DETAIL.LOW
-                        ))
-
-                prompt_message_contents = [TextPromptMessageContent(data=message.content)]
-                prompt_message_contents.extend(file_prompt_message_contents)
-
-                prompt_messages.append(UserPromptMessage(content=prompt_message_contents))
-            else:
-                prompt_messages.append(UserPromptMessage(content=message.content))
-        elif isinstance(message, AIMessage):
-            message_kwargs = {
-                'content': message.content
-            }
-
-            if 'function_call' in message.additional_kwargs:
-                message_kwargs['tool_calls'] = [
-                    AssistantPromptMessage.ToolCall(
-                        id=message.additional_kwargs['function_call']['id'],
-                        type='function',
-                        function=AssistantPromptMessage.ToolCall.ToolCallFunction(
-                            name=message.additional_kwargs['function_call']['name'],
-                            arguments=message.additional_kwargs['function_call']['arguments']
-                        )
-                    )
-                ]
-
-            prompt_messages.append(AssistantPromptMessage(**message_kwargs))
-        elif isinstance(message, SystemMessage):
-            prompt_messages.append(SystemPromptMessage(content=message.content))
-        elif isinstance(message, FunctionMessage):
-            prompt_messages.append(ToolPromptMessage(content=message.content, tool_call_id=message.name))
-
-    return prompt_messages
-
-
-def prompt_messages_to_lc_messages(prompt_messages: list[PromptMessage]) -> list[BaseMessage]:
-    messages = []
-    for prompt_message in prompt_messages:
-        if isinstance(prompt_message, UserPromptMessage):
-            if isinstance(prompt_message.content, str):
-                messages.append(HumanMessage(content=prompt_message.content))
-            else:
-                message_contents = []
-                for content in prompt_message.content:
-                    if isinstance(content, TextPromptMessageContent):
-                        message_contents.append(content.data)
-                    elif isinstance(content, ImagePromptMessageContent):
-                        message_contents.append({
-                            'type': 'image',
-                            'data': content.data,
-                            'detail': content.detail.value
-                        })
-
-                messages.append(HumanMessage(content=message_contents))
-        elif isinstance(prompt_message, AssistantPromptMessage):
-            message_kwargs = {
-                'content': prompt_message.content
-            }
-
-            if prompt_message.tool_calls:
-                message_kwargs['additional_kwargs'] = {
-                    'function_call': {
-                        'id': prompt_message.tool_calls[0].id,
-                        'name': prompt_message.tool_calls[0].function.name,
-                        'arguments': prompt_message.tool_calls[0].function.arguments
-                    }
-                }
-
-            messages.append(AIMessage(**message_kwargs))
-        elif isinstance(prompt_message, SystemPromptMessage):
-            messages.append(SystemMessage(content=prompt_message.content))
-        elif isinstance(prompt_message, ToolPromptMessage):
-            messages.append(FunctionMessage(name=prompt_message.tool_call_id, content=prompt_message.content))
-
-    return messages

api/core/entities/provider_configuration.py

@@ -203,7 +203,7 @@ class ProviderConfiguration(BaseModel):
         if provider_record:
             provider_record.encrypted_config = json.dumps(credentials)
             provider_record.is_valid = True
-            provider_record.updated_at = datetime.datetime.utcnow()
+            provider_record.updated_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
             db.session.commit()
         else:
             provider_record = Provider(
@@ -351,7 +351,7 @@ class ProviderConfiguration(BaseModel):
         if provider_model_record:
             provider_model_record.encrypted_config = json.dumps(credentials)
             provider_model_record.is_valid = True
-            provider_model_record.updated_at = datetime.datetime.utcnow()
+            provider_model_record.updated_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
             db.session.commit()
         else:
             provider_model_record = ProviderModel(

api/core/helper/code_executor/code_executor.py

@@ -1,17 +1,17 @@
-from os import environ
 from typing import Literal, Optional
 
 from httpx import post
 from pydantic import BaseModel
 from yarl import URL
 
+from config import get_env
 from core.helper.code_executor.javascript_transformer import NodeJsTemplateTransformer
 from core.helper.code_executor.jina2_transformer import Jinja2TemplateTransformer
 from core.helper.code_executor.python_transformer import PythonTemplateTransformer
 
 # Code Executor
-CODE_EXECUTION_ENDPOINT = environ.get('CODE_EXECUTION_ENDPOINT', '')
-CODE_EXECUTION_API_KEY = environ.get('CODE_EXECUTION_API_KEY', '')
+CODE_EXECUTION_ENDPOINT = get_env('CODE_EXECUTION_ENDPOINT')
+CODE_EXECUTION_API_KEY = get_env('CODE_EXECUTION_API_KEY')
 
 CODE_EXECUTION_TIMEOUT= (10, 60)
 
@@ -27,6 +27,7 @@ class CodeExecutionResponse(BaseModel):
     message: str
     data: Data
 
+
 class CodeExecutor:
     @classmethod
     def execute_code(cls, language: Literal['python3', 'javascript', 'jinja2'], code: str, inputs: dict) -> dict:

api/core/indexing_runner.py

@@ -81,7 +81,7 @@ class IndexingRunner:
             except ProviderTokenNotInitError as e:
                 dataset_document.indexing_status = 'error'
                 dataset_document.error = str(e.description)
-                dataset_document.stopped_at = datetime.datetime.utcnow()
+                dataset_document.stopped_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
                 db.session.commit()
             except ObjectDeletedError:
                 logging.warning('Document deleted, document id: {}'.format(dataset_document.id))
@@ -89,7 +89,7 @@ class IndexingRunner:
                 logging.exception("consume document failed")
                 dataset_document.indexing_status = 'error'
                 dataset_document.error = str(e)
-                dataset_document.stopped_at = datetime.datetime.utcnow()
+                dataset_document.stopped_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
                 db.session.commit()
 
     def run_in_splitting_status(self, dataset_document: DatasetDocument):
@@ -140,13 +140,13 @@ class IndexingRunner:
         except ProviderTokenNotInitError as e:
             dataset_document.indexing_status = 'error'
             dataset_document.error = str(e.description)
-            dataset_document.stopped_at = datetime.datetime.utcnow()
+            dataset_document.stopped_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
             db.session.commit()
         except Exception as e:
             logging.exception("consume document failed")
             dataset_document.indexing_status = 'error'
             dataset_document.error = str(e)
-            dataset_document.stopped_at = datetime.datetime.utcnow()
+            dataset_document.stopped_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
             db.session.commit()
 
     def run_in_indexing_status(self, dataset_document: DatasetDocument):
@@ -202,13 +202,13 @@ class IndexingRunner:
         except ProviderTokenNotInitError as e:
             dataset_document.indexing_status = 'error'
             dataset_document.error = str(e.description)
-            dataset_document.stopped_at = datetime.datetime.utcnow()
+            dataset_document.stopped_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
             db.session.commit()
         except Exception as e:
             logging.exception("consume document failed")
             dataset_document.indexing_status = 'error'
             dataset_document.error = str(e)
-            dataset_document.stopped_at = datetime.datetime.utcnow()
+            dataset_document.stopped_at = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
             db.session.commit()
 
     def indexing_estimate(self, tenant_id: str, extract_settings: list[ExtractSetting], tmp_processing_rule: dict,
@@ -382,7 +382,7 @@ class IndexingRunner:
             after_indexing_status="splitting",
             extra_update_params={
                 DatasetDocument.word_count: sum([len(text_doc.page_content) for text_doc in text_docs]),
-                DatasetDocument.parsing_completed_at: datetime.datetime.utcnow()
+                DatasetDocument.parsing_completed_at: datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
             }
         )
 
@@ -467,7 +467,7 @@ class IndexingRunner:
         doc_store.add_documents(documents)
 
         # update document status to indexing
-        cur_time = datetime.datetime.utcnow()
+        cur_time = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
         self._update_document_index_status(
             document_id=dataset_document.id,
             after_indexing_status="indexing",
@@ -482,7 +482,7 @@ class IndexingRunner:
             dataset_document_id=dataset_document.id,
             update_params={
                 DocumentSegment.status: "indexing",
-                DocumentSegment.indexing_at: datetime.datetime.utcnow()
+                DocumentSegment.indexing_at: datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
             }
         )
 
@@ -685,7 +685,7 @@ class IndexingRunner:
             after_indexing_status="completed",
             extra_update_params={
                 DatasetDocument.tokens: tokens,
-                DatasetDocument.completed_at: datetime.datetime.utcnow(),
+                DatasetDocument.completed_at: datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None),
                 DatasetDocument.indexing_latency: indexing_end_at - indexing_start_at,
             }
         )
@@ -706,7 +706,7 @@ class IndexingRunner:
                 ).update({
                     DocumentSegment.status: "completed",
                     DocumentSegment.enabled: True,
-                    DocumentSegment.completed_at: datetime.datetime.utcnow()
+                    DocumentSegment.completed_at: datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
                 })
 
                 db.session.commit()
@@ -739,7 +739,7 @@ class IndexingRunner:
             ).update({
                 DocumentSegment.status: "completed",
                 DocumentSegment.enabled: True,
-                DocumentSegment.completed_at: datetime.datetime.utcnow()
+                DocumentSegment.completed_at: datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
             })
 
             db.session.commit()
@@ -838,7 +838,7 @@ class IndexingRunner:
         doc_store.add_documents(documents)
 
         # update document status to indexing
-        cur_time = datetime.datetime.utcnow()
+        cur_time = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
         self._update_document_index_status(
             document_id=dataset_document.id,
             after_indexing_status="indexing",
@@ -853,7 +853,7 @@ class IndexingRunner:
             dataset_document_id=dataset_document.id,
             update_params={
                 DocumentSegment.status: "indexing",
-                DocumentSegment.indexing_at: datetime.datetime.utcnow()
+                DocumentSegment.indexing_at: datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None)
             }
         )
         pass

api/core/llm_generator/llm_generator.py

@@ -1,8 +1,7 @@
 import json
 import logging
 
-from langchain.schema import OutputParserException
-
+from core.llm_generator.output_parser.errors import OutputParserException
 from core.llm_generator.output_parser.rule_config_generator import RuleConfigGeneratorOutputParser
 from core.llm_generator.output_parser.suggested_questions_after_answer import SuggestedQuestionsAfterAnswerOutputParser
 from core.llm_generator.prompts import CONVERSATION_TITLE_PROMPT, GENERATOR_QA_PROMPT

api/core/llm_generator/output_parser/errors.py

@@ -0,0 +1,2 @@
+class OutputParserException(Exception):
+    pass

api/core/llm_generator/output_parser/rule_config_generator.py

@@ -1,12 +1,11 @@
 from typing import Any
 
-from langchain.schema import BaseOutputParser, OutputParserException
-
+from core.llm_generator.output_parser.errors import OutputParserException
 from core.llm_generator.prompts import RULE_CONFIG_GENERATE_TEMPLATE
 from libs.json_in_md_parser import parse_and_check_json_markdown
 
 
-class RuleConfigGeneratorOutputParser(BaseOutputParser):
+class RuleConfigGeneratorOutputParser:
 
     def get_format_instructions(self) -> str:
         return RULE_CONFIG_GENERATE_TEMPLATE

api/core/llm_generator/output_parser/suggested_questions_after_answer.py

@@ -2,12 +2,10 @@ import json
 import re
 from typing import Any
 
-from langchain.schema import BaseOutputParser
-
 from core.llm_generator.prompts import SUGGESTED_QUESTIONS_AFTER_ANSWER_INSTRUCTION_PROMPT
 
 
-class SuggestedQuestionsAfterAnswerOutputParser(BaseOutputParser):
+class SuggestedQuestionsAfterAnswerOutputParser:
 
     def get_format_instructions(self) -> str:
         return SUGGESTED_QUESTIONS_AFTER_ANSWER_INSTRUCTION_PROMPT

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

@@ -343,8 +343,12 @@ class AzureOpenAILargeLanguageModel(_CommonAzureOpenAI, LargeLanguageModel):
 
             delta = chunk.choices[0]
 
-            if delta.finish_reason is None and (delta.delta.content is None or delta.delta.content == '') and \
-                delta.delta.function_call is None:
+            # Handling exceptions when content filters' streaming mode is set to asynchronous modified filter
+            if delta.delta is None or (
+                delta.finish_reason is None
+                and (delta.delta.content is None or delta.delta.content == '')
+                and delta.delta.function_call is None
+            ):
                 continue
             
             # assistant_message_tool_calls = delta.delta.tool_calls

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

@@ -15,6 +15,7 @@ help:
     en_US: https://console.aws.amazon.com/
 supported_model_types:
   - llm
+  - text-embedding
 configurate_methods:
   - predefined-model
 provider_credential_schema:

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

@@ -0,0 +1,3 @@
+- amazon.titan-embed-text-v1
+- cohere.embed-english-v3
+- cohere.embed-multilingual-v3

api/core/model_runtime/model_providers/bedrock/text_embedding/amazon.titan-embed-text-v1.yaml

@@ -0,0 +1,8 @@
+model: amazon.titan-embed-text-v1
+model_type: text-embedding
+model_properties:
+  context_size: 8192
+pricing:
+  input: '0.0001'
+  unit: '0.001'
+  currency: USD

api/core/model_runtime/model_providers/bedrock/text_embedding/cohere.embed-english-v3.yaml

@@ -0,0 +1,8 @@
+model: cohere.embed-english-v3
+model_type: text-embedding
+model_properties:
+  context_size: 512
+pricing:
+  input: '0.1'
+  unit: '0.000001'
+  currency: USD

api/core/model_runtime/model_providers/bedrock/text_embedding/cohere.embed-multilingual-v3.yaml

@@ -0,0 +1,8 @@
+model: cohere.embed-multilingual-v3
+model_type: text-embedding
+model_properties:
+  context_size: 512
+pricing:
+  input: '0.1'
+  unit: '0.000001'
+  currency: USD

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

@@ -0,0 +1,234 @@
+import json
+import logging
+import time
+from typing import Optional
+
+import boto3
+from botocore.config import Config
+from botocore.exceptions import (
+    ClientError,
+    EndpointConnectionError,
+    NoRegionError,
+    ServiceNotInRegionError,
+    UnknownServiceError,
+)
+
+from core.model_runtime.entities.model_entities import PriceType
+from core.model_runtime.entities.text_embedding_entities import EmbeddingUsage, TextEmbeddingResult
+from core.model_runtime.errors.invoke import (
+    InvokeAuthorizationError,
+    InvokeBadRequestError,
+    InvokeConnectionError,
+    InvokeError,
+    InvokeRateLimitError,
+    InvokeServerUnavailableError,
+)
+from core.model_runtime.model_providers.__base.text_embedding_model import TextEmbeddingModel
+
+logger = logging.getLogger(__name__)
+
+class BedrockTextEmbeddingModel(TextEmbeddingModel):
+
+
+    def _invoke(self, model: str, credentials: dict,
+                texts: list[str], user: Optional[str] = None) \
+            -> TextEmbeddingResult:
+        """
+        Invoke text embedding model
+
+        :param model: model name
+        :param credentials: model credentials
+        :param texts: texts to embed
+        :param user: unique user id
+        :return: embeddings result
+        """
+        client_config = Config(
+            region_name=credentials["aws_region"]
+        )
+
+        bedrock_runtime = boto3.client(
+            service_name='bedrock-runtime',
+            config=client_config,
+            aws_access_key_id=credentials["aws_access_key_id"],
+            aws_secret_access_key=credentials["aws_secret_access_key"]
+        )
+
+        embeddings = []
+        token_usage = 0
+        
+        model_prefix = model.split('.')[0]
+         
+        if model_prefix == "amazon" :
+           for text in texts:
+              body = {
+                 "inputText": text,
+              }
+              response_body = self._invoke_bedrock_embedding(model, bedrock_runtime, body)
+              embeddings.extend([response_body.get('embedding')])
+              token_usage += response_body.get('inputTextTokenCount')
+           logger.warning(f'Total Tokens: {token_usage}')
+           result = TextEmbeddingResult(
+                model=model,
+                embeddings=embeddings,
+                usage=self._calc_response_usage(
+                    model=model,
+                    credentials=credentials,
+                    tokens=token_usage
+                )
+           )
+           return result
+           
+        if model_prefix == "cohere" :
+           input_type = 'search_document' if len(texts) > 1 else 'search_query'
+           for text in texts:
+              body = {
+                 "texts": [text],
+                 "input_type": input_type,
+              }
+              response_body = self._invoke_bedrock_embedding(model, bedrock_runtime, body)
+              embeddings.extend(response_body.get('embeddings'))
+              token_usage += len(text)
+           result = TextEmbeddingResult(
+                model=model,
+                embeddings=embeddings,
+                usage=self._calc_response_usage(
+                    model=model,
+                    credentials=credentials,
+                    tokens=token_usage
+                )
+           )
+           return result
+        
+        #others
+        raise ValueError(f"Got unknown model prefix {model_prefix} when handling block response")
+
+
+    def get_num_tokens(self, model: str, credentials: dict, texts: list[str]) -> int:
+        """
+        Get number of tokens for given prompt messages
+
+        :param model: model name
+        :param credentials: model credentials
+        :param texts: texts to embed
+        :return:
+        """
+        num_tokens = 0
+        for text in texts:
+            num_tokens += self._get_num_tokens_by_gpt2(text)
+        return num_tokens
+
+    def validate_credentials(self, model: str, credentials: dict) -> None:
+        """
+        Validate model credentials
+
+        :param model: model name
+        :param credentials: model credentials
+        :return:
+        """
+    
+    @property
+    def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:
+        """
+        Map model invoke error to unified error
+        The key is the ermd = genai.GenerativeModel(model)ror type thrown to the caller
+        The value is the md = genai.GenerativeModel(model)error type thrown by the model,
+        which needs to be converted into a unified error type for the caller.
+
+        :return: Invoke emd = genai.GenerativeModel(model)rror mapping
+        """
+        return {
+            InvokeConnectionError: [],
+            InvokeServerUnavailableError: [],
+            InvokeRateLimitError: [],
+            InvokeAuthorizationError: [],
+            InvokeBadRequestError: []
+        }
+    
+    def _create_payload(self, model_prefix: str, texts: list[str], model_parameters: dict, stop: Optional[list[str]] = None, stream: bool = True):
+        """
+        Create payload for bedrock api call depending on model provider
+        """
+        payload = dict()
+
+        if model_prefix == "amazon":
+            payload['inputText'] = texts
+
+    
+    def _calc_response_usage(self, model: str, credentials: dict, tokens: int) -> EmbeddingUsage:
+        """
+        Calculate response usage
+
+        :param model: model name
+        :param credentials: model credentials
+        :param tokens: input tokens
+        :return: usage
+        """
+        # get input price info
+        input_price_info = self.get_price(
+            model=model,
+            credentials=credentials,
+            price_type=PriceType.INPUT,
+            tokens=tokens
+        )
+
+        # transform usage
+        usage = EmbeddingUsage(
+            tokens=tokens,
+            total_tokens=tokens,
+            unit_price=input_price_info.unit_price,
+            price_unit=input_price_info.unit,
+            total_price=input_price_info.total_amount,
+            currency=input_price_info.currency,
+            latency=time.perf_counter() - self.started_at
+        )
+
+        return usage
+    
+    def _map_client_to_invoke_error(self, error_code: str, error_msg: str) -> type[InvokeError]:
+        """
+        Map client error to invoke error
+
+        :param error_code: error code
+        :param error_msg: error message
+        :return: invoke error
+        """
+
+        if error_code == "AccessDeniedException":
+            return InvokeAuthorizationError(error_msg)
+        elif error_code in ["ResourceNotFoundException", "ValidationException"]:
+            return InvokeBadRequestError(error_msg)
+        elif error_code in ["ThrottlingException", "ServiceQuotaExceededException"]:
+            return InvokeRateLimitError(error_msg)
+        elif error_code in ["ModelTimeoutException", "ModelErrorException", "InternalServerException", "ModelNotReadyException"]:
+            return InvokeServerUnavailableError(error_msg)
+        elif error_code == "ModelStreamErrorException":
+            return InvokeConnectionError(error_msg)
+
+        return InvokeError(error_msg)
+    
+
+    def _invoke_bedrock_embedding(self, model: str, bedrock_runtime, body: dict, ):
+        accept = 'application/json' 
+        content_type = 'application/json'
+        try:
+            response = bedrock_runtime.invoke_model(
+                body=json.dumps(body), 
+                modelId=model, 
+                accept=accept, 
+                contentType=content_type
+            )
+            response_body = json.loads(response.get('body').read().decode('utf-8'))
+            return response_body
+        except ClientError as ex:
+            error_code = ex.response['Error']['Code']
+            full_error_msg = f"{error_code}: {ex.response['Error']['Message']}"
+            raise self._map_client_to_invoke_error(error_code, full_error_msg)
+        
+        except (EndpointConnectionError, NoRegionError, ServiceNotInRegionError) as ex:
+            raise InvokeConnectionError(str(ex))
+
+        except UnknownServiceError as ex:
+            raise InvokeServerUnavailableError(str(ex))
+
+        except Exception as ex:
+            raise InvokeError(str(ex))

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

@@ -602,7 +602,7 @@ class CohereLargeLanguageModel(LargeLanguageModel):
             parameter_definitions = {}
             for p_key, p_val in properties.items():
                 required = False
-                if property in required_properties:
+                if p_key in required_properties:
                     required = True
 
                 desc = p_val['description']

api/core/model_runtime/model_providers/cohere/rerank/_position.yaml

@@ -0,0 +1,4 @@
+- rerank-english-v2.0
+- rerank-english-v3.0
+- rerank-multilingual-v2.0
+- rerank-multilingual-v3.0

api/core/model_runtime/model_providers/cohere/rerank/rerank-english-v3.0.yaml

@@ -0,0 +1,4 @@
+model: rerank-english-v3.0
+model_type: rerank
+model_properties:
+  context_size: 5120

api/core/model_runtime/model_providers/cohere/rerank/rerank-multilingual-v3.0.yaml

@@ -0,0 +1,4 @@
+model: rerank-multilingual-v3.0
+model_type: rerank
+model_properties:
+  context_size: 5120

api/core/model_runtime/model_providers/google/llm/gemini-1.5-pro-latest.yaml

@@ -5,6 +5,8 @@ model_type: llm
 features:
   - agent-thought
   - vision
+  - tool-call
+  - stream-tool-call
 model_properties:
   mode: chat
   context_size: 1048576

api/core/model_runtime/model_providers/google/llm/gemini-pro.yaml

@@ -4,6 +4,8 @@ label:
 model_type: llm
 features:
   - agent-thought
+  - tool-call
+  - stream-tool-call
 model_properties:
   mode: chat
   context_size: 30720

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

@@ -1,7 +1,9 @@
+import json
 import logging
 from collections.abc import Generator
 from typing import Optional, Union
 
+import google.ai.generativelanguage as glm
 import google.api_core.exceptions as exceptions
 import google.generativeai as genai
 import google.generativeai.client as client
@@ -13,9 +15,9 @@ from core.model_runtime.entities.message_entities import (
     AssistantPromptMessage,
     PromptMessage,
     PromptMessageContentType,
-    PromptMessageRole,
     PromptMessageTool,
     SystemPromptMessage,
+    ToolPromptMessage,
     UserPromptMessage,
 )
 from core.model_runtime.errors.invoke import (
@@ -62,7 +64,7 @@ class GoogleLargeLanguageModel(LargeLanguageModel):
         :return: full response or stream response chunk generator result
         """
         # invoke model
-        return self._generate(model, credentials, prompt_messages, model_parameters, stop, stream, user)
+        return self._generate(model, credentials, prompt_messages, model_parameters, tools, stop, stream, user)
     
     def get_num_tokens(self, model: str, credentials: dict, prompt_messages: list[PromptMessage],
                        tools: Optional[list[PromptMessageTool]] = None) -> int:
@@ -94,6 +96,32 @@ class GoogleLargeLanguageModel(LargeLanguageModel):
         )
 
         return text.rstrip()
+    
+    def _convert_tools_to_glm_tool(self, tools: list[PromptMessageTool]) -> glm.Tool:
+        """
+        Convert tool messages to glm tools
+
+        :param tools: tool messages
+        :return: glm tools
+        """
+        return glm.Tool(
+            function_declarations=[
+                glm.FunctionDeclaration(
+                    name=tool.name,
+                    parameters=glm.Schema(
+                        type=glm.Type.OBJECT,
+                        properties={
+                            key: {
+                                'type_': value.get('type', 'string').upper(),
+                                'description': value.get('description', ''),
+                                'enum': value.get('enum', [])
+                            } for key, value in tool.parameters.get('properties', {}).items()
+                        },
+                        required=tool.parameters.get('required', [])
+                    ),
+                ) for tool in tools
+            ]
+        )
 
     def validate_credentials(self, model: str, credentials: dict) -> None:
         """
@@ -105,7 +133,7 @@ class GoogleLargeLanguageModel(LargeLanguageModel):
         """
         
         try:
-            ping_message = PromptMessage(content="ping", role="system")
+            ping_message = SystemPromptMessage(content="ping")
             self._generate(model, credentials, [ping_message], {"max_tokens_to_sample": 5})
             
         except Exception as ex:
@@ -114,8 +142,9 @@ class GoogleLargeLanguageModel(LargeLanguageModel):
 
     def _generate(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]:
+                  tools: Optional[list[PromptMessageTool]] = None, stop: Optional[list[str]] = None, 
+                  stream: bool = True, user: Optional[str] = None
+        ) -> Union[LLMResult, Generator]:
         """
         Invoke large language model
 
@@ -153,7 +182,6 @@ class GoogleLargeLanguageModel(LargeLanguageModel):
                 else:
                     history.append(content)
 
-
         # Create a new ClientManager with tenant's API key
         new_client_manager = client._ClientManager()
         new_client_manager.configure(api_key=credentials["google_api_key"])
@@ -167,14 +195,15 @@ class GoogleLargeLanguageModel(LargeLanguageModel):
             HarmCategory.HARM_CATEGORY_SEXUALLY_EXPLICIT: HarmBlockThreshold.BLOCK_NONE,
             HarmCategory.HARM_CATEGORY_DANGEROUS_CONTENT: HarmBlockThreshold.BLOCK_NONE,
         }
-        
+
         response = google_model.generate_content(
             contents=history,
             generation_config=genai.types.GenerationConfig(
                 **config_kwargs
             ),
             stream=stream,
-            safety_settings=safety_settings
+            safety_settings=safety_settings,
+            tools=self._convert_tools_to_glm_tool(tools) if tools else None,
         )
 
         if stream:
@@ -228,43 +257,61 @@ class GoogleLargeLanguageModel(LargeLanguageModel):
         """
         index = -1
         for chunk in response:
-            content = chunk.text
-            index += 1
-           
-            assistant_prompt_message = AssistantPromptMessage(
-                content=content if content else '',
-            )
-  
-            if not response._done:
-                
-                # transform assistant message to prompt message
-                yield LLMResultChunk(
-                    model=model,
-                    prompt_messages=prompt_messages,
-                    delta=LLMResultChunkDelta(
-                        index=index,
-                        message=assistant_prompt_message
-                    )
+            for part in chunk.parts:
+                assistant_prompt_message = AssistantPromptMessage(
+                    content=''
                 )
-            else:
-                
-                # calculate num tokens
-                prompt_tokens = self.get_num_tokens(model, credentials, prompt_messages)
-                completion_tokens = self.get_num_tokens(model, credentials, [assistant_prompt_message])
 
-                # transform usage
-                usage = self._calc_response_usage(model, credentials, prompt_tokens, completion_tokens)
-                
-                yield LLMResultChunk(
-                    model=model,
-                    prompt_messages=prompt_messages,
-                    delta=LLMResultChunkDelta(
-                        index=index,
-                        message=assistant_prompt_message,
-                        finish_reason=chunk.candidates[0].finish_reason,
-                        usage=usage
+                if part.text:
+                    assistant_prompt_message.content += part.text
+
+                if part.function_call:
+                    assistant_prompt_message.tool_calls = [
+                        AssistantPromptMessage.ToolCall(
+                            id=part.function_call.name,
+                            type='function',
+                            function=AssistantPromptMessage.ToolCall.ToolCallFunction(
+                                name=part.function_call.name,
+                                arguments=json.dumps({
+                                    key: value 
+                                    for key, value in part.function_call.args.items()
+                                })
+                            )
+                        )
+                    ]
+
+                index += 1
+    
+                if not response._done:
+                    
+                    # transform assistant message to prompt message
+                    yield LLMResultChunk(
+                        model=model,
+                        prompt_messages=prompt_messages,
+                        delta=LLMResultChunkDelta(
+                            index=index,
+                            message=assistant_prompt_message
+                        )
+                    )
+                else:
+                    
+                    # calculate num tokens
+                    prompt_tokens = self.get_num_tokens(model, credentials, prompt_messages)
+                    completion_tokens = self.get_num_tokens(model, credentials, [assistant_prompt_message])
+
+                    # transform usage
+                    usage = self._calc_response_usage(model, credentials, prompt_tokens, completion_tokens)
+                    
+                    yield LLMResultChunk(
+                        model=model,
+                        prompt_messages=prompt_messages,
+                        delta=LLMResultChunkDelta(
+                            index=index,
+                            message=assistant_prompt_message,
+                            finish_reason=chunk.candidates[0].finish_reason,
+                            usage=usage
+                        )
                     )
-                )
 
     def _convert_one_message_to_text(self, message: PromptMessage) -> str:
         """
@@ -288,6 +335,8 @@ class GoogleLargeLanguageModel(LargeLanguageModel):
             message_text = f"{ai_prompt} {content}"
         elif isinstance(message, SystemPromptMessage):
             message_text = f"{human_prompt} {content}"
+        elif isinstance(message, ToolPromptMessage):
+            message_text = f"{human_prompt} {content}"
         else:
             raise ValueError(f"Got unknown type {message}")
 
@@ -300,26 +349,53 @@ class GoogleLargeLanguageModel(LargeLanguageModel):
         :param message: one PromptMessage
         :return: glm Content representation of message
         """
-
-        parts = []
-        if (isinstance(message.content, str)):
-            parts.append(to_part(message.content))
+        if isinstance(message, UserPromptMessage):
+            glm_content = {
+                "role": "user",
+                "parts": []
+            }
+            if (isinstance(message.content, str)):
+                glm_content['parts'].append(to_part(message.content))
+            else:
+                for c in message.content:
+                    if c.type == PromptMessageContentType.TEXT:
+                        glm_content['parts'].append(to_part(c.data))
+                    else:
+                        metadata, data = c.data.split(',', 1)
+                        mime_type = metadata.split(';', 1)[0].split(':')[1]
+                        blob = {"inline_data":{"mime_type":mime_type,"data":data}}
+                        glm_content['parts'].append(blob)
+            return glm_content
+        elif isinstance(message, AssistantPromptMessage):
+            glm_content = {
+                "role": "model",
+                "parts": []
+            }
+            if message.content:
+                glm_content['parts'].append(to_part(message.content))
+            if message.tool_calls:
+                glm_content["parts"].append(to_part(glm.FunctionCall(
+                    name=message.tool_calls[0].function.name,
+                    args=json.loads(message.tool_calls[0].function.arguments),
+                )))
+            return glm_content
+        elif isinstance(message, SystemPromptMessage):
+            return {
+                "role": "user",
+                "parts": [to_part(message.content)]
+            }
+        elif isinstance(message, ToolPromptMessage):
+            return {
+                "role": "function",
+                "parts": [glm.Part(function_response=glm.FunctionResponse(
+                    name=message.name,
+                    response={
+                        "response": message.content
+                    }
+                ))]
+            }
         else:
-            for c in message.content:
-                if c.type == PromptMessageContentType.TEXT:
-                    parts.append(to_part(c.data))
-                else:
-                    metadata, data = c.data.split(',', 1)
-                    mime_type = metadata.split(';', 1)[0].split(':')[1]
-                    blob = {"inline_data":{"mime_type":mime_type,"data":data}}
-                    parts.append(blob)
-
-        glm_content = {
-            "role": "user" if message.role in (PromptMessageRole.USER, PromptMessageRole.SYSTEM) else "model",
-            "parts": parts
-        }
-        
-        return glm_content
+            raise ValueError(f"Got unknown type {message}")
     
     @property
     def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:

api/core/model_runtime/model_providers/nvidia/llm/_position.yaml

@@ -1,4 +1,5 @@
 - google/gemma-7b
+- google/codegemma-7b
 - meta/llama2-70b
 - mistralai/mixtral-8x7b-instruct-v0.1
 - fuyu-8b

api/core/model_runtime/model_providers/nvidia/llm/codegemma-7b.yaml

@@ -0,0 +1,30 @@
+model: google/codegemma-7b
+label:
+  zh_Hans: google/codegemma-7b
+  en_US: google/codegemma-7b
+model_type: llm
+features:
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+  - name: top_p
+    use_template: top_p
+  - name: max_tokens
+    use_template: max_tokens
+    default: 1024
+    min: 1
+    max: 1024
+  - name: frequency_penalty
+    use_template: frequency_penalty
+    min: -2
+    max: 2
+    default: 0
+  - name: presence_penalty
+    use_template: presence_penalty
+    min: -2
+    max: 2
+    default: 0

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

@@ -24,6 +24,7 @@ class NVIDIALargeLanguageModel(OAIAPICompatLargeLanguageModel):
         'fuyu-8b': 'vlm/adept/fuyu-8b',
         'mistralai/mixtral-8x7b-instruct-v0.1': '',
         'google/gemma-7b': '',
+        'google/codegemma-7b': '',
         'meta/llama2-70b': ''
     }
 

api/core/model_runtime/model_providers/nvidia/nvidia.yaml

@@ -1,6 +1,6 @@
 provider: nvidia
 label:
-  en_US: NVIDIA
+  en_US: API Catalog
 icon_small:
   en_US: icon_s_en.svg
 icon_large:

api/core/model_runtime/model_providers/nvidia/rerank/rerank-qa-mistral-4b.yaml

@@ -1,4 +1,4 @@
 model: nv-rerank-qa-mistral-4b:1
 model_type: rerank
 model_properties:
-  context_size: 8192
+  context_size: 512

api/core/model_runtime/model_providers/ollama/ollama.yaml

@@ -90,9 +90,9 @@ model_credential_schema:
       options:
         - value: 'true'
           label:
-            en_US: Yes
+            en_US: 'Yes'
             zh_Hans: 是
         - value: 'false'
           label:
-            en_US: No
+            en_US: 'No'
             zh_Hans: 否

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

@@ -150,6 +150,11 @@ class OAIAPICompatLargeLanguageModel(_CommonOAI_API_Compat, LargeLanguageModel):
             except json.JSONDecodeError as e:
                 raise CredentialsValidateFailedError('Credentials validation failed: JSON decode error')
 
+            if (completion_type is LLMMode.CHAT and json_result['object'] == ''):
+                json_result['object'] = 'chat.completion'
+            elif (completion_type is LLMMode.COMPLETION and json_result['object'] == ''):
+                json_result['object'] = 'text_completion'
+                
             if (completion_type is LLMMode.CHAT
                     and ('object' not in json_result or json_result['object'] != 'chat.completion')):
                 raise CredentialsValidateFailedError(
@@ -170,13 +175,14 @@ class OAIAPICompatLargeLanguageModel(_CommonOAI_API_Compat, LargeLanguageModel):
         features = []
 
         function_calling_type = credentials.get('function_calling_type', 'no_call')
-        if function_calling_type == 'function_call':
+        if function_calling_type in ['function_call']:
             features.append(ModelFeature.TOOL_CALL)
-        endpoint_url = credentials["endpoint_url"]
-        # if not endpoint_url.endswith('/'):
-        #     endpoint_url += '/'
-        # if 'https://api.openai.com/v1/' == endpoint_url:
-        #     features.append(ModelFeature.STREAM_TOOL_CALL)
+        elif function_calling_type in ['tool_call']:
+            features.append(ModelFeature.MULTI_TOOL_CALL)
+
+        stream_function_calling = credentials.get('stream_function_calling', 'supported')
+        if stream_function_calling == 'supported':
+            features.append(ModelFeature.STREAM_TOOL_CALL)
 
         vision_support = credentials.get('vision_support', 'not_support')
         if vision_support == 'support':
@@ -386,29 +392,37 @@ class OAIAPICompatLargeLanguageModel(_CommonOAI_API_Compat, LargeLanguageModel):
 
         def increase_tool_call(new_tool_calls: list[AssistantPromptMessage.ToolCall]):
             def get_tool_call(tool_call_id: str):
-                tool_call = next(
-                    (tool_call for tool_call in tools_calls if tool_call.id == tool_call_id), None
-                )
+                if not tool_call_id:
+                    return tools_calls[-1]
+
+                tool_call = next((tool_call for tool_call in tools_calls if tool_call.id == tool_call_id), None)
                 if tool_call is None:
                     tool_call = AssistantPromptMessage.ToolCall(
-                        id='', 
-                        type='function', 
+                        id=tool_call_id,
+                        type="function",
                         function=AssistantPromptMessage.ToolCall.ToolCallFunction(
-                            name='',
-                            arguments=''
+                            name="",
+                            arguments=""
                         )
                     )
                     tools_calls.append(tool_call)
+
                 return tool_call
 
             for new_tool_call in new_tool_calls:
                 # get tool call
-                tool_call = get_tool_call(new_tool_call.id)
+                tool_call = get_tool_call(new_tool_call.function.name)
                 # update tool call
-                tool_call.id = new_tool_call.id
-                tool_call.type = new_tool_call.type
-                tool_call.function.name = new_tool_call.function.name
-                tool_call.function.arguments += new_tool_call.function.arguments
+                if new_tool_call.id:
+                    tool_call.id = new_tool_call.id
+                if new_tool_call.type:
+                    tool_call.type = new_tool_call.type
+                if new_tool_call.function.name:
+                    tool_call.function.name = new_tool_call.function.name
+                if new_tool_call.function.arguments:
+                    tool_call.function.arguments += new_tool_call.function.arguments
+
+        finish_reason = 'Unknown'
 
         for chunk in response.iter_lines(decode_unicode=True, delimiter=delimiter):
             if chunk:
@@ -438,7 +452,17 @@ class OAIAPICompatLargeLanguageModel(_CommonOAI_API_Compat, LargeLanguageModel):
                     delta = choice['delta']
                     delta_content = delta.get('content')
 
-                    assistant_message_tool_calls = delta.get('tool_calls', None)
+                    assistant_message_tool_calls = None
+
+                    if 'tool_calls' in delta and credentials.get('function_calling_type', 'no_call') == 'tool_call':
+                        assistant_message_tool_calls = delta.get('tool_calls', None)
+                    elif 'function_call' in delta and credentials.get('function_calling_type', 'no_call') == 'function_call':
+                        assistant_message_tool_calls = [{
+                            'id': 'tool_call_id',
+                            'type': 'function',
+                            'function': delta.get('function_call', {})
+                        }]
+
                     # assistant_message_function_call = delta.delta.function_call
 
                     # extract tool calls from response
@@ -449,15 +473,13 @@ class OAIAPICompatLargeLanguageModel(_CommonOAI_API_Compat, LargeLanguageModel):
                     if delta_content is None or delta_content == '':
                         continue
 
-                    # function_call = self._extract_response_function_call(assistant_message_function_call)
-                    # tool_calls = [function_call] if function_call else []
-
                     # transform assistant message to prompt message
                     assistant_prompt_message = AssistantPromptMessage(
                         content=delta_content,
-                        tool_calls=tool_calls if assistant_message_tool_calls else []
                     )
 
+                    # reset tool calls
+                    tool_calls = []
                     full_assistant_content += delta_content
                 elif 'text' in choice:
                     choice_text = choice.get('text', '')
@@ -470,37 +492,36 @@ class OAIAPICompatLargeLanguageModel(_CommonOAI_API_Compat, LargeLanguageModel):
                 else:
                     continue
 
-                # check payload indicator for completion
-                if finish_reason is not None:
-                    yield LLMResultChunk(
-                        model=model,
-                        prompt_messages=prompt_messages,
-                        delta=LLMResultChunkDelta(
-                            index=chunk_index,
-                            message=AssistantPromptMessage(
-                                tool_calls=tools_calls,
-                            ),
-                            finish_reason=finish_reason
-                        )
-                    )
-
-                    yield create_final_llm_result_chunk(
+                yield LLMResultChunk(
+                    model=model,
+                    prompt_messages=prompt_messages,
+                    delta=LLMResultChunkDelta(
                         index=chunk_index,
                         message=assistant_prompt_message,
-                        finish_reason=finish_reason
-                    )
-                else:
-                    yield LLMResultChunk(
-                        model=model,
-                        prompt_messages=prompt_messages,
-                        delta=LLMResultChunkDelta(
-                            index=chunk_index,
-                            message=assistant_prompt_message,
-                        )
                     )
+                )
 
             chunk_index += 1
 
+        if tools_calls:
+            yield LLMResultChunk(
+                model=model,
+                prompt_messages=prompt_messages,
+                delta=LLMResultChunkDelta(
+                    index=chunk_index,
+                    message=AssistantPromptMessage(
+                        tool_calls=tools_calls,
+                        content=""
+                    ),
+                )
+            )
+
+        yield create_final_llm_result_chunk(
+            index=chunk_index,
+            message=AssistantPromptMessage(content=""),
+            finish_reason=finish_reason
+        )
+
     def _handle_generate_response(self, model: str, credentials: dict, response: requests.Response,
                                   prompt_messages: list[PromptMessage]) -> LLMResult:
 
@@ -757,13 +778,13 @@ class OAIAPICompatLargeLanguageModel(_CommonOAI_API_Compat, LargeLanguageModel):
         if response_tool_calls:
             for response_tool_call in response_tool_calls:
                 function = AssistantPromptMessage.ToolCall.ToolCallFunction(
-                    name=response_tool_call["function"]["name"],
-                    arguments=response_tool_call["function"]["arguments"]
+                    name=response_tool_call.get("function", {}).get("name", ""),
+                    arguments=response_tool_call.get("function", {}).get("arguments", "")
                 )
 
                 tool_call = AssistantPromptMessage.ToolCall(
-                    id=response_tool_call["id"],
-                    type=response_tool_call["type"],
+                    id=response_tool_call.get("id", ""),
+                    type=response_tool_call.get("type", ""),
                     function=function
                 )
                 tool_calls.append(tool_call)
@@ -781,14 +802,14 @@ class OAIAPICompatLargeLanguageModel(_CommonOAI_API_Compat, LargeLanguageModel):
         tool_call = None
         if response_function_call:
             function = AssistantPromptMessage.ToolCall.ToolCallFunction(
-                name=response_function_call['name'],
-                arguments=response_function_call['arguments']
+                name=response_function_call.get('name', ''),
+                arguments=response_function_call.get('arguments', '')
             )
 
             tool_call = AssistantPromptMessage.ToolCall(
-                id=response_function_call['name'],
+                id=response_function_call.get('id', ''),
                 type="function",
                 function=function
             )
 
-        return tool_call
+        return tool_call

api/core/model_runtime/model_providers/openai_api_compatible/openai_api_compatible.yaml

@@ -86,14 +86,32 @@ model_credential_schema:
       default: no_call
       options:
         - value: function_call
+          label:
+            en_US: Function Call
+            zh_Hans: Function Call
+        - value: tool_call
+          label:
+            en_US: Tool Call
+            zh_Hans: Tool Call
+        - value: no_call
+          label:
+            en_US: Not Support
+            zh_Hans: 不支持
+    - variable: stream_function_calling
+      show_on:
+        - variable: __model_type
+          value: llm
+      label:
+        en_US: Stream function calling
+      type: select
+      required: false
+      default: not_supported
+      options:
+        - value: supported
           label:
             en_US: Support
             zh_Hans: 支持
-#        - value: tool_call
-#          label:
-#            en_US: Tool Call
-#            zh_Hans: Tool Call
-        - value: no_call
+        - value: not_supported
           label:
             en_US: Not Support
             zh_Hans: 不支持

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

@@ -1,82 +0,0 @@
-from typing import Any, Optional
-
-from langchain.callbacks.manager import CallbackManagerForLLMRun
-from langchain.llms import Tongyi
-from langchain.llms.tongyi import generate_with_retry, stream_generate_with_retry
-from langchain.schema import Generation, LLMResult
-
-
-class EnhanceTongyi(Tongyi):
-    @property
-    def _default_params(self) -> dict[str, Any]:
-        """Get the default parameters for calling OpenAI API."""
-        normal_params = {
-            "top_p": self.top_p,
-            "api_key": self.dashscope_api_key
-        }
-
-        return {**normal_params, **self.model_kwargs}
-
-    def _generate(
-        self,
-        prompts: list[str],
-        stop: Optional[list[str]] = None,
-        run_manager: Optional[CallbackManagerForLLMRun] = None,
-        **kwargs: Any,
-    ) -> LLMResult:
-        generations = []
-        params: dict[str, Any] = {
-            **{"model": self.model_name},
-            **self._default_params,
-            **kwargs,
-        }
-        if self.streaming:
-            if len(prompts) > 1:
-                raise ValueError("Cannot stream results with multiple prompts.")
-            params["stream"] = True
-            text = ''
-            for stream_resp in stream_generate_with_retry(
-                self, prompt=prompts[0], **params
-            ):
-                if not generations:
-                    current_text = stream_resp["output"]["text"]
-                else:
-                    current_text = stream_resp["output"]["text"][len(text):]
-
-                text = stream_resp["output"]["text"]
-
-                generations.append(
-                    [
-                        Generation(
-                            text=current_text,
-                            generation_info=dict(
-                                finish_reason=stream_resp["output"]["finish_reason"],
-                            ),
-                        )
-                    ]
-                )
-
-                if run_manager:
-                    run_manager.on_llm_new_token(
-                        current_text,
-                        verbose=self.verbose,
-                        logprobs=None,
-                    )
-        else:
-            for prompt in prompts:
-                completion = generate_with_retry(
-                    self,
-                    prompt=prompt,
-                    **params,
-                )
-                generations.append(
-                    [
-                        Generation(
-                            text=completion["output"]["text"],
-                            generation_info=dict(
-                                finish_reason=completion["output"]["finish_reason"],
-                            ),
-                        )
-                    ]
-                )
-        return LLMResult(generations=generations)

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

@@ -1,8 +1,13 @@
+import base64
+import os
+import tempfile
+import uuid
 from collections.abc import Generator
-from typing import Optional, Union
+from http import HTTPStatus
+from typing import Optional, Union, cast
 
-from dashscope import get_tokenizer
-from dashscope.api_entities.dashscope_response import DashScopeAPIResponse
+from dashscope import Generation, MultiModalConversation, get_tokenizer
+from dashscope.api_entities.dashscope_response import GenerationResponse
 from dashscope.common.error import (
     AuthenticationError,
     InvalidParameter,
@@ -11,17 +16,21 @@ from dashscope.common.error import (
     UnsupportedHTTPMethod,
     UnsupportedModel,
 )
-from langchain.llms.tongyi import generate_with_retry, stream_generate_with_retry
 
 from core.model_runtime.callbacks.base_callback import Callback
 from core.model_runtime.entities.llm_entities import LLMMode, LLMResult, LLMResultChunk, LLMResultChunkDelta
 from core.model_runtime.entities.message_entities import (
     AssistantPromptMessage,
+    ImagePromptMessageContent,
     PromptMessage,
+    PromptMessageContentType,
     PromptMessageTool,
     SystemPromptMessage,
+    TextPromptMessageContent,
+    ToolPromptMessage,
     UserPromptMessage,
 )
+from core.model_runtime.entities.model_entities import ModelFeature
 from core.model_runtime.errors.invoke import (
     InvokeAuthorizationError,
     InvokeBadRequestError,
@@ -33,10 +42,9 @@ from core.model_runtime.errors.invoke import (
 from core.model_runtime.errors.validate import CredentialsValidateFailedError
 from core.model_runtime.model_providers.__base.large_language_model import LargeLanguageModel
 
-from ._client import EnhanceTongyi
-
 
 class TongyiLargeLanguageModel(LargeLanguageModel):
+    tokenizers = {}
 
     def _invoke(self, model: str, credentials: dict,
                 prompt_messages: list[PromptMessage], model_parameters: dict,
@@ -57,13 +65,13 @@ class TongyiLargeLanguageModel(LargeLanguageModel):
         :return: full response or stream response chunk generator result
         """
         # invoke model
-        return self._generate(model, credentials, prompt_messages, model_parameters, stop, stream, user)
-    
-    def _code_block_mode_wrapper(self, model: str, credentials: dict, 
-                                 prompt_messages: list[PromptMessage], model_parameters: dict, 
-                                 tools: list[PromptMessageTool] | None = None, stop: list[str] | None = None, 
+        return self._generate(model, credentials, prompt_messages, model_parameters, tools, stop, stream, user)
+
+    def _code_block_mode_wrapper(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, callbacks: list[Callback] = None) \
-                            -> LLMResult | Generator:
+            -> LLMResult | Generator:
         """
         Wrapper for code block mode
         """
@@ -88,7 +96,7 @@ if you are not sure about the structure.
                 stream=stream,
                 user=user
             )
-        
+
         model_parameters.pop("response_format")
         stop = stop or []
         stop.extend(["\n```", "```\n"])
@@ -99,13 +107,13 @@ if you are not sure about the structure.
             # override the system message
             prompt_messages[0] = SystemPromptMessage(
                 content=block_prompts
-                    .replace("{{instructions}}", prompt_messages[0].content)
+                .replace("{{instructions}}", prompt_messages[0].content)
             )
         else:
             # insert the system message
             prompt_messages.insert(0, SystemPromptMessage(
                 content=block_prompts
-                    .replace("{{instructions}}", f"Please output a valid {code_block} object.")
+                .replace("{{instructions}}", f"Please output a valid {code_block} object.")
             ))
 
         mode = self.get_model_mode(model, credentials)
@@ -138,7 +146,7 @@ if you are not sure about the structure.
                 prompt_messages=prompt_messages,
                 input_generator=response
             )
-        
+
         return response
 
     def get_num_tokens(self, model: str, credentials: dict, prompt_messages: list[PromptMessage],
@@ -152,7 +160,14 @@ if you are not sure about the structure.
         :param tools: tools for tool calling
         :return:
         """
-        tokenizer = get_tokenizer(model)
+        if model in ['qwen-turbo-chat', 'qwen-plus-chat']:
+            model = model.replace('-chat', '')
+
+        if model in self.tokenizers:
+            tokenizer = self.tokenizers[model]
+        else:
+            tokenizer = get_tokenizer(model)
+            self.tokenizers[model] = tokenizer
 
         # convert string to token ids
         tokens = tokenizer.encode(self._convert_messages_to_prompt(prompt_messages))
@@ -184,6 +199,7 @@ if you are not sure about the structure.
 
     def _generate(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]:
         """
@@ -192,24 +208,27 @@ if you are not sure about the structure.
         :param model: model name
         :param credentials: credentials
         :param prompt_messages: prompt messages
+        :param tools: tools for tool calling
         :param model_parameters: model parameters
         :param stop: stop words
         :param stream: is stream response
         :param user: unique user id
         :return: full response or stream response chunk generator result
         """
-        extra_model_kwargs = {}
-        if stop:
-            extra_model_kwargs['stop'] = stop
-
         # transform credentials to kwargs for model instance
         credentials_kwargs = self._to_credential_kwargs(credentials)
 
-        client = EnhanceTongyi(
-            model_name=model,
-            streaming=stream,
-            dashscope_api_key=credentials_kwargs['api_key'],
-        )
+        mode = self.get_model_mode(model, credentials)
+
+        if model in ['qwen-turbo-chat', 'qwen-plus-chat']:
+            model = model.replace('-chat', '')
+
+        extra_model_kwargs = {}
+        if tools:
+            extra_model_kwargs['tools'] = self._convert_tools(tools)
+
+        if stop:
+            extra_model_kwargs['stop'] = stop
 
         params = {
             'model': model,
@@ -218,30 +237,27 @@ if you are not sure about the structure.
             **extra_model_kwargs,
         }
 
-        mode = self.get_model_mode(model, credentials)
+        model_schema = self.get_model_schema(model, credentials)
+        if ModelFeature.VISION in (model_schema.features or []):
+            params['messages'] = self._convert_prompt_messages_to_tongyi_messages(prompt_messages, rich_content=True)
 
-        if mode == LLMMode.CHAT:
-            params['messages'] = self._convert_prompt_messages_to_tongyi_messages(prompt_messages)
+            response = MultiModalConversation.call(**params, stream=stream)
         else:
-            params['prompt'] = self._convert_messages_to_prompt(prompt_messages)
+            if mode == LLMMode.CHAT:
+                params['messages'] = self._convert_prompt_messages_to_tongyi_messages(prompt_messages)
+            else:
+                params['prompt'] = prompt_messages[0].content.rstrip()
+
+            response = Generation.call(**params,
+                                       result_format='message',
+                                       stream=stream)
 
         if stream:
-            responses = stream_generate_with_retry(
-                client, 
-                stream=True,
-                incremental_output=True,
-                **params
-            )
+            return self._handle_generate_stream_response(model, credentials, response, prompt_messages)
 
-            return self._handle_generate_stream_response(model, credentials, responses, prompt_messages)
-
-        response = generate_with_retry(
-            client,
-            **params,
-        )
         return self._handle_generate_response(model, credentials, response, prompt_messages)
-        
-    def _handle_generate_response(self, model: str, credentials: dict, response: DashScopeAPIResponse,
+
+    def _handle_generate_response(self, model: str, credentials: dict, response: GenerationResponse,
                                   prompt_messages: list[PromptMessage]) -> LLMResult:
         """
         Handle llm response
@@ -254,7 +270,7 @@ if you are not sure about the structure.
         """
         # transform assistant message to prompt message
         assistant_prompt_message = AssistantPromptMessage(
-            content=response.output.text
+            content=response.output.choices[0].message.content,
         )
 
         # transform usage
@@ -270,32 +286,65 @@ if you are not sure about the structure.
 
         return result
 
-    def _handle_generate_stream_response(self, model: str, credentials: dict, responses: Generator,
+    def _handle_generate_stream_response(self, model: str, credentials: dict,
+                                         responses: Generator[GenerationResponse, None, None],
                                          prompt_messages: list[PromptMessage]) -> Generator:
         """
         Handle llm stream response
 
         :param model: model name
         :param credentials: credentials
-        :param response: response
+        :param responses: response
         :param prompt_messages: prompt messages
         :return: llm response chunk generator result
         """
+        full_text = ''
+        tool_calls = []
         for index, response in enumerate(responses):
-            resp_finish_reason = response.output.finish_reason
-            resp_content = response.output.text
-            usage = response.usage
+            if response.status_code != 200 and response.status_code != HTTPStatus.OK:
+                raise ServiceUnavailableError(
+                    f"Failed to invoke model {model}, status code: {response.status_code}, "
+                    f"message: {response.message}"
+                )
 
-            if resp_finish_reason is None and (resp_content is None or resp_content == ''):
-                continue
+            resp_finish_reason = response.output.choices[0].finish_reason
 
-            # transform assistant message to prompt message
-            assistant_prompt_message = AssistantPromptMessage(
-                content=resp_content if resp_content else '',
-            )
+            if resp_finish_reason is not None and resp_finish_reason != 'null':
+                resp_content = response.output.choices[0].message.content
+
+                assistant_prompt_message = AssistantPromptMessage(
+                    content='',
+                )
+
+                if 'tool_calls' in response.output.choices[0].message:
+                    tool_calls = response.output.choices[0].message['tool_calls']
+                elif resp_content:
+                    # special for qwen-vl
+                    if isinstance(resp_content, list):
+                        resp_content = resp_content[0]['text']
+
+                    # transform assistant message to prompt message
+                    assistant_prompt_message.content = resp_content.replace(full_text, '', 1)
+
+                    full_text = resp_content
+
+                if tool_calls:
+                    message_tool_calls = []
+                    for tool_call_obj in tool_calls:
+                        message_tool_call = AssistantPromptMessage.ToolCall(
+                            id=tool_call_obj['function']['name'],
+                            type='function',
+                            function=AssistantPromptMessage.ToolCall.ToolCallFunction(
+                                name=tool_call_obj['function']['name'],
+                                arguments=tool_call_obj['function']['arguments']
+                            )
+                        )
+                        message_tool_calls.append(message_tool_call)
+
+                    assistant_prompt_message.tool_calls = message_tool_calls
 
-            if resp_finish_reason is not None:
                 # transform usage
+                usage = response.usage
                 usage = self._calc_response_usage(model, credentials, usage.input_tokens, usage.output_tokens)
 
                 yield LLMResultChunk(
@@ -309,6 +358,23 @@ if you are not sure about the structure.
                     )
                 )
             else:
+                resp_content = response.output.choices[0].message.content
+                if not resp_content:
+                    if 'tool_calls' in response.output.choices[0].message:
+                        tool_calls = response.output.choices[0].message['tool_calls']
+                    continue
+
+                # special for qwen-vl
+                if isinstance(resp_content, list):
+                    resp_content = resp_content[0]['text']
+
+                # transform assistant message to prompt message
+                assistant_prompt_message = AssistantPromptMessage(
+                    content=resp_content.replace(full_text, '', 1),
+                )
+
+                full_text = resp_content
+
                 yield LLMResultChunk(
                     model=model,
                     prompt_messages=prompt_messages,
@@ -343,11 +409,20 @@ if you are not sure about the structure.
         content = message.content
 
         if isinstance(message, UserPromptMessage):
-            message_text = f"{human_prompt} {content}"
+            if isinstance(content, str):
+                message_text = f"{human_prompt} {content}"
+            else:
+                message_text = ""
+                for sub_message in content:
+                    if sub_message.type == PromptMessageContentType.TEXT:
+                        message_text = f"{human_prompt} {sub_message.data}"
+                        break
         elif isinstance(message, AssistantPromptMessage):
             message_text = f"{ai_prompt} {content}"
         elif isinstance(message, SystemPromptMessage):
             message_text = content
+        elif isinstance(message, ToolPromptMessage):
+            message_text = content
         else:
             raise ValueError(f"Got unknown type {message}")
 
@@ -370,7 +445,8 @@ if you are not sure about the structure.
         # trim off the trailing ' ' that might come from the "Assistant: "
         return text.rstrip()
 
-    def _convert_prompt_messages_to_tongyi_messages(self, prompt_messages: list[PromptMessage]) -> list[dict]:
+    def _convert_prompt_messages_to_tongyi_messages(self, prompt_messages: list[PromptMessage],
+                                                    rich_content: bool = False) -> list[dict]:
         """
         Convert prompt messages to tongyi messages
 
@@ -382,23 +458,118 @@ if you are not sure about the structure.
             if isinstance(prompt_message, SystemPromptMessage):
                 tongyi_messages.append({
                     'role': 'system',
-                    'content': prompt_message.content,
+                    'content': prompt_message.content if not rich_content else [{"text": prompt_message.content}],
                 })
             elif isinstance(prompt_message, UserPromptMessage):
-                tongyi_messages.append({
-                    'role': 'user',
-                    'content': prompt_message.content,
-                })
+                if isinstance(prompt_message.content, str):
+                    tongyi_messages.append({
+                        'role': 'user',
+                        'content': prompt_message.content if not rich_content else [{"text": prompt_message.content}],
+                    })
+                else:
+                    sub_messages = []
+                    for message_content in prompt_message.content:
+                        if message_content.type == PromptMessageContentType.TEXT:
+                            message_content = cast(TextPromptMessageContent, message_content)
+                            sub_message_dict = {
+                                "text": message_content.data
+                            }
+                            sub_messages.append(sub_message_dict)
+                        elif message_content.type == PromptMessageContentType.IMAGE:
+                            message_content = cast(ImagePromptMessageContent, message_content)
+
+                            image_url = message_content.data
+                            if message_content.data.startswith("data:"):
+                                # convert image base64 data to file in /tmp
+                                image_url = self._save_base64_image_to_file(message_content.data)
+
+                            sub_message_dict = {
+                                "image": image_url
+                            }
+                            sub_messages.append(sub_message_dict)
+
+                    # resort sub_messages to ensure text is always at last
+                    sub_messages = sorted(sub_messages, key=lambda x: 'text' in x)
+
+                    tongyi_messages.append({
+                        'role': 'user',
+                        'content': sub_messages
+                    })
             elif isinstance(prompt_message, AssistantPromptMessage):
+                content = prompt_message.content
+                if not content:
+                    content = ' '
                 tongyi_messages.append({
                     'role': 'assistant',
-                    'content': prompt_message.content,
+                    'content': content if not rich_content else [{"text": content}],
+                })
+            elif isinstance(prompt_message, ToolPromptMessage):
+                tongyi_messages.append({
+                    "role": "tool",
+                    "content": prompt_message.content,
+                    "name": prompt_message.tool_call_id
                 })
             else:
                 raise ValueError(f"Got unknown type {prompt_message}")
 
         return tongyi_messages
 
+    def _save_base64_image_to_file(self, base64_image: str) -> str:
+        """
+        Save base64 image to file
+        'data:{upload_file.mime_type};base64,{encoded_string}'
+
+        :param base64_image: base64 image data
+        :return: image file path
+        """
+        # get mime type and encoded string
+        mime_type, encoded_string = base64_image.split(',')[0].split(';')[0].split(':')[1], base64_image.split(',')[1]
+
+        # save image to file
+        temp_dir = tempfile.gettempdir()
+
+        file_path = os.path.join(temp_dir, f"{uuid.uuid4()}.{mime_type.split('/')[1]}")
+
+        with open(file_path, "wb") as image_file:
+            image_file.write(base64.b64decode(encoded_string))
+
+        return f"file://{file_path}"
+
+    def _convert_tools(self, tools: list[PromptMessageTool]) -> list[dict]:
+        """
+        Convert tools
+        """
+        tool_definitions = []
+        for tool in tools:
+            properties = tool.parameters['properties']
+            required_properties = tool.parameters['required']
+
+            properties_definitions = {}
+            for p_key, p_val in properties.items():
+                desc = p_val['description']
+                if 'enum' in p_val:
+                    desc += (f"; Only accepts one of the following predefined options: "
+                             f"[{', '.join(p_val['enum'])}]")
+
+                properties_definitions[p_key] = {
+                    'description': desc,
+                    'type': p_val['type'],
+                }
+
+            tool_definition = {
+                "type": "function",
+                "function": {
+                    "name": tool.name,
+                    "description": tool.description,
+                    "parameters": properties_definitions,
+                    "required": required_properties
+                }
+            }
+
+            tool_definitions.append(tool_definition)
+
+        return tool_definitions
+
     @property
     def _invoke_error_mapping(self) -> dict[type[InvokeError], list[type[Exception]]]:
         """

api/core/model_runtime/model_providers/tongyi/llm/qwen-max-0403.yaml

@@ -0,0 +1,81 @@
+model: qwen-max-0403
+label:
+  en_US: qwen-max-0403
+model_type: llm
+features:
+  - multi-tool-call
+  - agent-thought
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.3
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 2000
+    min: 1
+    max: 2000
+    help:
+      zh_Hans: 用于指定模型在生成内容时token的最大数量，它定义了生成的上限，但不保证每次都会生成到这个数量。
+      en_US: It is used to specify the maximum number of tokens when the model generates content. It defines the upper limit of generation, but does not guarantee that this number will be generated every time.
+  - name: top_p
+    use_template: top_p
+    type: float
+    default: 0.8
+    min: 0.1
+    max: 0.9
+    help:
+      zh_Hans: 生成过程中核采样方法概率阈值，例如，取值为0.8时，仅保留概率加起来大于等于0.8的最可能token的最小集合作为候选集。取值范围为（0,1.0)，取值越大，生成的随机性越高；取值越低，生成的确定性越高。
+      en_US: The probability threshold of the kernel sampling method during the generation process. For example, when the value is 0.8, only the smallest set of the most likely tokens with a sum of probabilities greater than or equal to 0.8 is retained as the candidate set. The value range is (0,1.0). The larger the value, the higher the randomness generated; the lower the value, the higher the certainty generated.
+  - name: top_k
+    type: int
+    min: 0
+    max: 99
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    help:
+      zh_Hans: 生成时，采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。
+      en_US: The size of the sample candidate set when generated. For example, when the value is 50, only the 50 highest-scoring tokens in a single generation form a randomly sampled candidate set. The larger the value, the higher the randomness generated; the smaller the value, the higher the certainty generated.
+  - name: seed
+    required: false
+    type: int
+    default: 1234
+    label:
+      zh_Hans: 随机种子
+      en_US: Random seed
+    help:
+      zh_Hans: 生成时使用的随机数种子，用户控制模型生成内容的随机性。支持无符号64位整数，默认值为 1234。在使用seed时，模型将尽可能生成相同或相似的结果，但目前不保证每次生成的结果完全相同。
+      en_US: The random number seed used when generating, the user controls the randomness of the content generated by the model. Supports unsigned 64-bit integers, default value is 1234. When using seed, the model will try its best to generate the same or similar results, but there is currently no guarantee that the results will be exactly the same every time.
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+  - name: enable_search
+    type: boolean
+    default: false
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.12'
+  output: '0.12'
+  unit: '0.001'
+  currency: RMB

api/core/model_runtime/model_providers/tongyi/llm/qwen-max-1201.yaml

@@ -2,6 +2,10 @@ model: qwen-max-1201
 label:
   en_US: qwen-max-1201
 model_type: llm
+features:
+  - multi-tool-call
+  - agent-thought
+  - stream-tool-call
 model_properties:
   mode: chat
   context_size: 8192
@@ -9,7 +13,7 @@ parameter_rules:
   - name: temperature
     use_template: temperature
     type: float
-    default: 0.85
+    default: 0.3
     min: 0.0
     max: 2.0
     help:

api/core/model_runtime/model_providers/tongyi/llm/qwen-max-longcontext.yaml

@@ -2,6 +2,10 @@ model: qwen-max-longcontext
 label:
   en_US: qwen-max-longcontext
 model_type: llm
+features:
+  - multi-tool-call
+  - agent-thought
+  - stream-tool-call
 model_properties:
   mode: chat
   context_size: 32768
@@ -9,7 +13,7 @@ parameter_rules:
   - name: temperature
     use_template: temperature
     type: float
-    default: 0.85
+    default: 0.3
     min: 0.0
     max: 2.0
     help:

api/core/model_runtime/model_providers/tongyi/llm/qwen-max.yaml

@@ -2,6 +2,10 @@ model: qwen-max
 label:
   en_US: qwen-max
 model_type: llm
+features:
+  - multi-tool-call
+  - agent-thought
+  - stream-tool-call
 model_properties:
   mode: chat
   context_size: 8192
@@ -9,7 +13,7 @@ parameter_rules:
   - name: temperature
     use_template: temperature
     type: float
-    default: 0.85
+    default: 0.3
     min: 0.0
     max: 2.0
     help:

api/core/model_runtime/model_providers/tongyi/llm/qwen-plus-chat.yaml

@@ -0,0 +1,81 @@
+model: qwen-plus-chat
+label:
+  en_US: qwen-plus-chat
+model_type: llm
+features:
+  - multi-tool-call
+  - agent-thought
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 32768
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.3
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 1500
+    min: 1
+    max: 1500
+    help:
+      zh_Hans: 用于指定模型在生成内容时token的最大数量，它定义了生成的上限，但不保证每次都会生成到这个数量。
+      en_US: It is used to specify the maximum number of tokens when the model generates content. It defines the upper limit of generation, but does not guarantee that this number will be generated every time.
+  - name: top_p
+    use_template: top_p
+    type: float
+    default: 0.8
+    min: 0.1
+    max: 0.9
+    help:
+      zh_Hans: 生成过程中核采样方法概率阈值，例如，取值为0.8时，仅保留概率加起来大于等于0.8的最可能token的最小集合作为候选集。取值范围为（0,1.0)，取值越大，生成的随机性越高；取值越低，生成的确定性越高。
+      en_US: The probability threshold of the kernel sampling method during the generation process. For example, when the value is 0.8, only the smallest set of the most likely tokens with a sum of probabilities greater than or equal to 0.8 is retained as the candidate set. The value range is (0,1.0). The larger the value, the higher the randomness generated; the lower the value, the higher the certainty generated.
+  - name: top_k
+    type: int
+    min: 0
+    max: 99
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    help:
+      zh_Hans: 生成时，采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。
+      en_US: The size of the sample candidate set when generated. For example, when the value is 50, only the 50 highest-scoring tokens in a single generation form a randomly sampled candidate set. The larger the value, the higher the randomness generated; the smaller the value, the higher the certainty generated.
+  - name: seed
+    required: false
+    type: int
+    default: 1234
+    label:
+      zh_Hans: 随机种子
+      en_US: Random seed
+    help:
+      zh_Hans: 生成时使用的随机数种子，用户控制模型生成内容的随机性。支持无符号64位整数，默认值为 1234。在使用seed时，模型将尽可能生成相同或相似的结果，但目前不保证每次生成的结果完全相同。
+      en_US: The random number seed used when generating, the user controls the randomness of the content generated by the model. Supports unsigned 64-bit integers, default value is 1234. When using seed, the model will try its best to generate the same or similar results, but there is currently no guarantee that the results will be exactly the same every time.
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+  - name: enable_search
+    type: boolean
+    default: false
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.02'
+  output: '0.02'
+  unit: '0.001'
+  currency: RMB

api/core/model_runtime/model_providers/tongyi/llm/qwen-plus.yaml

@@ -2,6 +2,8 @@ model: qwen-plus
 label:
   en_US: qwen-plus
 model_type: llm
+features:
+  - agent-thought
 model_properties:
   mode: completion
   context_size: 32768
@@ -9,7 +11,7 @@ parameter_rules:
   - name: temperature
     use_template: temperature
     type: float
-    default: 0.85
+    default: 0.3
     min: 0.0
     max: 2.0
     help:

api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo-chat.yaml

@@ -0,0 +1,81 @@
+model: qwen-turbo-chat
+label:
+  en_US: qwen-turbo-chat
+model_type: llm
+features:
+  - multi-tool-call
+  - agent-thought
+  - stream-tool-call
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: temperature
+    use_template: temperature
+    type: float
+    default: 0.3
+    min: 0.0
+    max: 2.0
+    help:
+      zh_Hans: 用于控制随机性和多样性的程度。具体来说，temperature值控制了生成文本时对每个候选词的概率分布进行平滑的程度。较高的temperature值会降低概率分布的峰值，使得更多的低概率词被选择，生成结果更加多样化；而较低的temperature值则会增强概率分布的峰值，使得高概率词更容易被选择，生成结果更加确定。
+      en_US: Used to control the degree of randomness and diversity. Specifically, the temperature value controls the degree to which the probability distribution of each candidate word is smoothed when generating text. A higher temperature value will reduce the peak value of the probability distribution, allowing more low-probability words to be selected, and the generated results will be more diverse; while a lower temperature value will enhance the peak value of the probability distribution, making it easier for high-probability words to be selected. , the generated results are more certain.
+  - name: max_tokens
+    use_template: max_tokens
+    type: int
+    default: 1500
+    min: 1
+    max: 1500
+    help:
+      zh_Hans: 用于指定模型在生成内容时token的最大数量，它定义了生成的上限，但不保证每次都会生成到这个数量。
+      en_US: It is used to specify the maximum number of tokens when the model generates content. It defines the upper limit of generation, but does not guarantee that this number will be generated every time.
+  - name: top_p
+    use_template: top_p
+    type: float
+    default: 0.8
+    min: 0.1
+    max: 0.9
+    help:
+      zh_Hans: 生成过程中核采样方法概率阈值，例如，取值为0.8时，仅保留概率加起来大于等于0.8的最可能token的最小集合作为候选集。取值范围为（0,1.0)，取值越大，生成的随机性越高；取值越低，生成的确定性越高。
+      en_US: The probability threshold of the kernel sampling method during the generation process. For example, when the value is 0.8, only the smallest set of the most likely tokens with a sum of probabilities greater than or equal to 0.8 is retained as the candidate set. The value range is (0,1.0). The larger the value, the higher the randomness generated; the lower the value, the higher the certainty generated.
+  - name: top_k
+    type: int
+    min: 0
+    max: 99
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    help:
+      zh_Hans: 生成时，采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。
+      en_US: The size of the sample candidate set when generated. For example, when the value is 50, only the 50 highest-scoring tokens in a single generation form a randomly sampled candidate set. The larger the value, the higher the randomness generated; the smaller the value, the higher the certainty generated.
+  - name: seed
+    required: false
+    type: int
+    default: 1234
+    label:
+      zh_Hans: 随机种子
+      en_US: Random seed
+    help:
+      zh_Hans: 生成时使用的随机数种子，用户控制模型生成内容的随机性。支持无符号64位整数，默认值为 1234。在使用seed时，模型将尽可能生成相同或相似的结果，但目前不保证每次生成的结果完全相同。
+      en_US: The random number seed used when generating, the user controls the randomness of the content generated by the model. Supports unsigned 64-bit integers, default value is 1234. When using seed, the model will try its best to generate the same or similar results, but there is currently no guarantee that the results will be exactly the same every time.
+  - name: repetition_penalty
+    required: false
+    type: float
+    default: 1.1
+    label:
+      en_US: Repetition penalty
+    help:
+      zh_Hans: 用于控制模型生成时的重复度。提高repetition_penalty时可以降低模型生成的重复度。1.0表示不做惩罚。
+      en_US: Used to control the repeatability when generating models. Increasing repetition_penalty can reduce the duplication of model generation. 1.0 means no punishment.
+  - name: enable_search
+    type: boolean
+    default: false
+    help:
+      zh_Hans: 模型内置了互联网搜索服务，该参数控制模型在生成文本时是否参考使用互联网搜索结果。启用互联网搜索，模型会将搜索结果作为文本生成过程中的参考信息，但模型会基于其内部逻辑“自行判断”是否使用互联网搜索结果。
+      en_US: The model has a built-in Internet search service. This parameter controls whether the model refers to Internet search results when generating text. When Internet search is enabled, the model will use the search results as reference information in the text generation process, but the model will "judge" whether to use Internet search results based on its internal logic.
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.008'
+  output: '0.008'
+  unit: '0.001'
+  currency: RMB

api/core/model_runtime/model_providers/tongyi/llm/qwen-turbo.yaml

@@ -2,6 +2,8 @@ model: qwen-turbo
 label:
   en_US: qwen-turbo
 model_type: llm
+features:
+  - agent-thought
 model_properties:
   mode: completion
   context_size: 8192
@@ -9,7 +11,7 @@ parameter_rules:
   - name: temperature
     use_template: temperature
     type: float
-    default: 0.85
+    default: 0.3
     min: 0.0
     max: 2.0
     help:

api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-max.yaml

@@ -0,0 +1,47 @@
+model: qwen-vl-max
+label:
+  en_US: qwen-vl-max
+model_type: llm
+features:
+  - vision
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 8192
+parameter_rules:
+  - name: top_p
+    use_template: top_p
+    type: float
+    default: 0.8
+    min: 0.1
+    max: 0.9
+    help:
+      zh_Hans: 生成过程中核采样方法概率阈值，例如，取值为0.8时，仅保留概率加起来大于等于0.8的最可能token的最小集合作为候选集。取值范围为（0,1.0)，取值越大，生成的随机性越高；取值越低，生成的确定性越高。
+      en_US: The probability threshold of the kernel sampling method during the generation process. For example, when the value is 0.8, only the smallest set of the most likely tokens with a sum of probabilities greater than or equal to 0.8 is retained as the candidate set. The value range is (0,1.0). The larger the value, the higher the randomness generated; the lower the value, the higher the certainty generated.
+  - name: top_k
+    type: int
+    min: 0
+    max: 99
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    help:
+      zh_Hans: 生成时，采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。
+      en_US: The size of the sample candidate set when generated. For example, when the value is 50, only the 50 highest-scoring tokens in a single generation form a randomly sampled candidate set. The larger the value, the higher the randomness generated; the smaller the value, the higher the certainty generated.
+  - name: seed
+    required: false
+    type: int
+    default: 1234
+    label:
+      zh_Hans: 随机种子
+      en_US: Random seed
+    help:
+      zh_Hans: 生成时使用的随机数种子，用户控制模型生成内容的随机性。支持无符号64位整数，默认值为 1234。在使用seed时，模型将尽可能生成相同或相似的结果，但目前不保证每次生成的结果完全相同。
+      en_US: The random number seed used when generating, the user controls the randomness of the content generated by the model. Supports unsigned 64-bit integers, default value is 1234. When using seed, the model will try its best to generate the same or similar results, but there is currently no guarantee that the results will be exactly the same every time.
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.02'
+  output: '0.02'
+  unit: '0.001'
+  currency: RMB

api/core/model_runtime/model_providers/tongyi/llm/qwen-vl-plus.yaml

@@ -0,0 +1,47 @@
+model: qwen-vl-plus
+label:
+  en_US: qwen-vl-plus
+model_type: llm
+features:
+  - vision
+  - agent-thought
+model_properties:
+  mode: chat
+  context_size: 32768
+parameter_rules:
+  - name: top_p
+    use_template: top_p
+    type: float
+    default: 0.8
+    min: 0.1
+    max: 0.9
+    help:
+      zh_Hans: 生成过程中核采样方法概率阈值，例如，取值为0.8时，仅保留概率加起来大于等于0.8的最可能token的最小集合作为候选集。取值范围为（0,1.0)，取值越大，生成的随机性越高；取值越低，生成的确定性越高。
+      en_US: The probability threshold of the kernel sampling method during the generation process. For example, when the value is 0.8, only the smallest set of the most likely tokens with a sum of probabilities greater than or equal to 0.8 is retained as the candidate set. The value range is (0,1.0). The larger the value, the higher the randomness generated; the lower the value, the higher the certainty generated.
+  - name: top_k
+    type: int
+    min: 0
+    max: 99
+    label:
+      zh_Hans: 取样数量
+      en_US: Top k
+    help:
+      zh_Hans: 生成时，采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。
+      en_US: The size of the sample candidate set when generated. For example, when the value is 50, only the 50 highest-scoring tokens in a single generation form a randomly sampled candidate set. The larger the value, the higher the randomness generated; the smaller the value, the higher the certainty generated.
+  - name: seed
+    required: false
+    type: int
+    default: 1234
+    label:
+      zh_Hans: 随机种子
+      en_US: Random seed
+    help:
+      zh_Hans: 生成时使用的随机数种子，用户控制模型生成内容的随机性。支持无符号64位整数，默认值为 1234。在使用seed时，模型将尽可能生成相同或相似的结果，但目前不保证每次生成的结果完全相同。
+      en_US: The random number seed used when generating, the user controls the randomness of the content generated by the model. Supports unsigned 64-bit integers, default value is 1234. When using seed, the model will try its best to generate the same or similar results, but there is currently no guarantee that the results will be exactly the same every time.
+  - name: response_format
+    use_template: response_format
+pricing:
+  input: '0.008'
+  output: '0.008'
+  unit: '0.001'
+  currency: RMB

api/core/model_runtime/model_providers/tongyi/text_embedding/text_embedding.py

@@ -37,8 +37,11 @@ class TongyiTextEmbeddingModel(_CommonTongyi, TextEmbeddingModel):
         :return: embeddings result
         """
         credentials_kwargs = self._to_credential_kwargs(credentials)
-        dashscope.api_key = credentials_kwargs["dashscope_api_key"]
-        embeddings, embedding_used_tokens = self.embed_documents(model, texts)
+        embeddings, embedding_used_tokens = self.embed_documents(
+            credentials_kwargs=credentials_kwargs,
+            model=model,
+            texts=texts
+        )
 
         return TextEmbeddingResult(
             embeddings=embeddings,
@@ -74,17 +77,19 @@ class TongyiTextEmbeddingModel(_CommonTongyi, TextEmbeddingModel):
         try:
             # transform credentials to kwargs for model instance
             credentials_kwargs = self._to_credential_kwargs(credentials)
-            dashscope.api_key = credentials_kwargs["dashscope_api_key"]
+
             # call embedding model
-            self.embed_documents(model=model, texts=["ping"])
+            self.embed_documents(credentials_kwargs=credentials_kwargs, model=model, texts=["ping"])
         except Exception as ex:
             raise CredentialsValidateFailedError(str(ex))
 
     @staticmethod
-    def embed_documents(model: str, texts: list[str]) -> tuple[list[list[float]], int]:
+    def embed_documents(credentials_kwargs: dict, model: str, texts: list[str]) -> tuple[list[list[float]], int]:
         """Call out to Tongyi's embedding endpoint.
 
         Args:
+            credentials_kwargs: The credentials to use for the call.
+            model: The model to use for embedding.
             texts: The list of texts to embed.
 
         Returns:
@@ -93,7 +98,12 @@ class TongyiTextEmbeddingModel(_CommonTongyi, TextEmbeddingModel):
         embeddings = []
         embedding_used_tokens = 0
         for text in texts:
-            response = dashscope.TextEmbedding.call(model=model, input=text, text_type="document")
+            response = dashscope.TextEmbedding.call(
+                api_key=credentials_kwargs["dashscope_api_key"],
+                model=model,
+                input=text,
+                text_type="document"
+            )
             data = response.output["embeddings"][0]
             embeddings.append(data["embedding"])
             embedding_used_tokens += response.usage["total_tokens"]

api/core/model_runtime/model_providers/tongyi/tts/tts.py

@@ -118,7 +118,6 @@ class TongyiText2SpeechModel(_CommonTongyi, TTSModel):
         :param content_text: text content to be translated
         :return: text translated to audio file
         """
-        dashscope.api_key = credentials.get('dashscope_api_key')
         word_limit = self._get_model_word_limit(model, credentials)
         audio_type = self._get_model_audio_type(model, credentials)
         tts_file_id = self._get_file_name(content_text)
@@ -127,6 +126,7 @@ class TongyiText2SpeechModel(_CommonTongyi, TTSModel):
             sentences = list(self._split_text_into_sentences(text=content_text, limit=word_limit))
             for sentence in sentences:
                 response = dashscope.audio.tts.SpeechSynthesizer.call(model=voice, sample_rate=48000,
+                                                                      api_key=credentials.get('dashscope_api_key'),
                                                                       text=sentence.strip(),
                                                                       format=audio_type, word_timestamp_enabled=True,
                                                                       phoneme_timestamp_enabled=True)
@@ -146,8 +146,8 @@ class TongyiText2SpeechModel(_CommonTongyi, TTSModel):
         :param audio_type: audio file type
         :return: text translated to audio file
         """
-        dashscope.api_key = credentials.get('dashscope_api_key')
         response = dashscope.audio.tts.SpeechSynthesizer.call(model=voice, sample_rate=48000,
+                                                              api_key=credentials.get('dashscope_api_key'),
                                                               text=sentence.strip(),
                                                               format=audio_type)
         if isinstance(response.get_audio_data(), bytes):

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

@@ -43,7 +43,7 @@ model_credential_schema:
       placeholder:
         zh_Hans: 在此输入您的上下文大小
         en_US: Enter the context size
-      default: 2048
+      default: '2048'
     - variable: completion_type
       label:
         zh_Hans: 补全类型
@@ -69,16 +69,16 @@ model_credential_schema:
         en_US: Stream output
       type: select
       required: true
-      default: 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
+            en_US: 'Yes'
+          value: 'true'
         - label:
             zh_Hans: 否
-            en_US: No
-          value: false
+            en_US: 'No'
+          value: 'false'

api/core/rag/datasource/vdb/relyt/relyt_vector.py

@@ -0,0 +1,169 @@
+import logging
+from typing import Any
+
+from pgvecto_rs.sdk import PGVectoRs, Record
+from pydantic import BaseModel, root_validator
+from sqlalchemy import text as sql_text
+from sqlalchemy.orm import Session
+
+from core.rag.datasource.vdb.vector_base import BaseVector
+from core.rag.models.document import Document
+from extensions.ext_redis import redis_client
+
+logger = logging.getLogger(__name__)
+
+class RelytConfig(BaseModel):
+    host: str
+    port: int
+    user: str
+    password: str
+    database: str
+
+    @root_validator()
+    def validate_config(cls, values: dict) -> dict:
+        if not values['host']:
+            raise ValueError("config RELYT_HOST is required")
+        if not values['port']:
+            raise ValueError("config RELYT_PORT is required")
+        if not values['user']:
+            raise ValueError("config RELYT_USER is required")
+        if not values['password']:
+            raise ValueError("config RELYT_PASSWORD is required")
+        if not values['database']:
+            raise ValueError("config RELYT_DATABASE is required")
+        return values
+
+
+class RelytVector(BaseVector):
+
+    def __init__(self, collection_name: str, config: RelytConfig, dim: int):
+        super().__init__(collection_name)
+        self._client_config = config
+        self._url = f"postgresql+psycopg2://{config.user}:{config.password}@{config.host}:{config.port}/{config.database}"
+        self._client = PGVectoRs(
+            db_url=self._url,
+            collection_name=self._collection_name,
+            dimension=dim
+        )
+        self._fields = []
+
+    def get_type(self) -> str:
+        return 'relyt'
+
+    def create(self, texts: list[Document], embeddings: list[list[float]], **kwargs):
+        index_params = {}
+        metadatas = [d.metadata for d in texts]
+        self.create_collection(len(embeddings[0]))
+        self.add_texts(texts, embeddings)
+
+    def create_collection(self, dimension: int):
+        lock_name = 'vector_indexing_lock_{}'.format(self._collection_name)
+        with redis_client.lock(lock_name, timeout=20):
+            collection_exist_cache_key = 'vector_indexing_{}'.format(self._collection_name)
+            if redis_client.get(collection_exist_cache_key):
+                return
+            index_name = f"{self._collection_name}_embedding_index"
+            with Session(self._client._engine) as session:
+                drop_statement = sql_text(f"DROP TABLE IF EXISTS collection_{self._collection_name}")
+                session.execute(drop_statement)
+                create_statement = sql_text(f"""
+                    CREATE TABLE IF NOT EXISTS collection_{self._collection_name} (
+                        id UUID PRIMARY KEY,
+                        text TEXT NOT NULL,
+                        meta JSONB NOT NULL,
+                        embedding vector({dimension}) NOT NULL
+                    ) using heap; 
+                """)
+                session.execute(create_statement)
+                index_statement = sql_text(f"""
+                        CREATE INDEX {index_name}
+                        ON collection_{self._collection_name} USING vectors(embedding vector_l2_ops)
+                        WITH (options = $$
+                                optimizing.optimizing_threads = 30
+                                segment.max_growing_segment_size = 2000
+                                segment.max_sealed_segment_size = 30000000
+                                [indexing.hnsw]
+                                m=30
+                                ef_construction=500
+                                $$);
+                    """)
+                session.execute(index_statement)
+                session.commit()
+            redis_client.set(collection_exist_cache_key, 1, ex=3600)
+
+    def add_texts(self, documents: list[Document], embeddings: list[list[float]], **kwargs):
+        records = [Record.from_text(d.page_content, e, d.metadata) for d, e in zip(documents, embeddings)]
+        pks = [str(r.id) for r in records]
+        self._client.insert(records)
+        return pks
+
+    def delete_by_document_id(self, document_id: str):
+        ids = self.get_ids_by_metadata_field('document_id', document_id)
+        if ids:
+            self._client.delete_by_ids(ids)
+
+    def get_ids_by_metadata_field(self, key: str, value: str):
+        result = None
+        with Session(self._client._engine) as session:
+            select_statement = sql_text(
+                f"SELECT id FROM collection_{self._collection_name} WHERE meta->>'{key}' = '{value}'; "
+            )
+            result = session.execute(select_statement).fetchall()
+        if result:
+            return [item[0] for item in result]
+        else:
+            return None
+
+    def delete_by_metadata_field(self, key: str, value: str):
+
+        ids = self.get_ids_by_metadata_field(key, value)
+        if ids:
+            self._client.delete_by_ids(ids)
+
+    def delete_by_ids(self, doc_ids: list[str]) -> None:
+        with Session(self._client._engine) as session:
+            select_statement = sql_text(
+                f"SELECT id FROM collection_{self._collection_name} WHERE meta->>'doc_id' in ('{doc_ids}'); "
+            )
+            result = session.execute(select_statement).fetchall()
+        if result:
+            ids = [item[0] for item in result]
+            self._client.delete_by_ids(ids)
+
+    def delete(self) -> None:
+        with Session(self._client._engine) as session:
+            session.execute(sql_text(f"DROP TABLE IF EXISTS collection_{self._collection_name}"))
+            session.commit()
+
+    def text_exists(self, id: str) -> bool:
+        with Session(self._client._engine) as session:
+            select_statement = sql_text(
+                f"SELECT id FROM collection_{self._collection_name} WHERE meta->>'doc_id' = '{id}' limit 1; "
+            )
+            result = session.execute(select_statement).fetchall()
+        return len(result) > 0
+
+    def search_by_vector(self, query_vector: list[float], **kwargs: Any) -> list[Document]:
+        from pgvecto_rs.sdk import filters
+        filter_condition = filters.meta_contains(kwargs.get('filter'))
+        results = self._client.search(
+            top_k=int(kwargs.get('top_k')),
+            embedding=query_vector,
+            filter=filter_condition
+        )
+
+        # Organize results.
+        docs = []
+        for record, dis in results:
+            metadata = record.meta
+            metadata['score'] = dis
+            score_threshold = kwargs.get('score_threshold') if kwargs.get('score_threshold') else 0.0
+            if dis > score_threshold:
+                doc = Document(page_content=record.text,
+                               metadata=metadata)
+                docs.append(doc)
+        return docs
+
+    def search_by_full_text(self, query: str, **kwargs: Any) -> list[Document]:
+        # milvus/zilliz/relyt doesn't support bm25 search
+        return []

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

@@ -113,6 +113,31 @@ class Vector:
                     database=config.get('MILVUS_DATABASE'),
                 )
             )
+        elif vector_type == "relyt":
+            from core.rag.datasource.vdb.relyt.relyt_vector import RelytConfig, RelytVector
+            if self._dataset.index_struct_dict:
+                class_prefix: str = self._dataset.index_struct_dict['vector_store']['class_prefix']
+                collection_name = class_prefix
+            else:
+                dataset_id = self._dataset.id
+                collection_name = Dataset.gen_collection_name_by_id(dataset_id)
+                index_struct_dict = {
+                    "type": 'relyt',
+                    "vector_store": {"class_prefix": collection_name}
+                }
+                self._dataset.index_struct = json.dumps(index_struct_dict)
+            dim = len(self._embeddings.embed_query("hello relyt"))
+            return RelytVector(
+                collection_name=collection_name,
+                config=RelytConfig(
+                    host=config.get('RELYT_HOST'),
+                    port=config.get('RELYT_PORT'),
+                    user=config.get('RELYT_USER'),
+                    password=config.get('RELYT_PASSWORD'),
+                    database=config.get('RELYT_DATABASE'),
+                ),
+                dim=dim
+            )
         else:
             raise ValueError(f"Vector store {config.get('VECTOR_STORE')} is not supported.")
 

api/core/rag/extractor/blod/blod.py

@@ -159,7 +159,7 @@ class BlobLoader(ABC):
     def yield_blobs(
         self,
     ) -> Iterable[Blob]:
-        """A lazy loader for raw data represented by LangChain's Blob object.
+        """A lazy loader for raw data represented by Blob object.
 
         Returns:
             A generator over blobs

api/core/rag/extractor/excel_extractor.py

@@ -2,6 +2,7 @@
 from typing import Optional
 
 import pandas as pd
+import xlrd
 
 from core.rag.extractor.extractor_base import BaseExtractor
 from core.rag.models.document import Document
@@ -27,10 +28,37 @@ class ExcelExtractor(BaseExtractor):
         self._autodetect_encoding = autodetect_encoding
 
     def extract(self) -> list[Document]:
+        """ parse excel file"""
+        if self._file_path.endswith('.xls'):
+            return self._extract4xls()
+        elif self._file_path.endswith('.xlsx'):
+            return self._extract4xlsx()
+
+    def _extract4xls(self) -> list[Document]:
+        wb = xlrd.open_workbook(filename=self._file_path)
+        documents = []
+        # loop over all sheets
+        for sheet in wb.sheets():
+            for row_index, row in enumerate(sheet.get_rows(), start=1):
+                row_header = None
+                if self.is_blank_row(row):
+                    continue
+                if row_header is None:
+                    row_header = row
+                    continue
+                item_arr = []
+                for index, cell in enumerate(row):
+                    txt_value = str(cell.value)
+                    item_arr.append(f'{row_header[index].value}:{txt_value}')
+                item_str = "\n".join(item_arr)
+                document = Document(page_content=item_str, metadata={'source': self._file_path})
+                documents.append(document)
+        return documents
+
+    def _extract4xlsx(self) -> list[Document]:
         """Load from file path using Pandas."""
         data = []
-
-        # 使用 Pandas 读取 Excel 文件的每个工作表
+        # Read each worksheet of an Excel file using Pandas
         xls = pd.ExcelFile(self._file_path)
         for sheet_name in xls.sheet_names:
             df = pd.read_excel(xls, sheet_name=sheet_name)
@@ -43,5 +71,18 @@ class ExcelExtractor(BaseExtractor):
                 item = ';'.join(f'{k}:{v}' for k, v in row.items() if pd.notna(v))
                 document = Document(page_content=item, metadata={'source': self._file_path})
                 data.append(document)
-
         return data
+
+    @staticmethod
+    def is_blank_row(row):
+        """
+
+        Determine whether the specified line is a blank line.
+        :param row: row object。
+        :return: Returns True if the row is blank, False otherwise.
+        """
+        # Iterates through the cells and returns False if a non-empty cell is found
+        for cell in row:
+            if cell.value is not None and cell.value != '':
+                return False
+        return True

api/core/rag/extractor/extract_processor.py

@@ -16,6 +16,7 @@ from core.rag.extractor.pdf_extractor import PdfExtractor
 from core.rag.extractor.text_extractor import TextExtractor
 from core.rag.extractor.unstructured.unstructured_doc_extractor import UnstructuredWordExtractor
 from core.rag.extractor.unstructured.unstructured_eml_extractor import UnstructuredEmailExtractor
+from core.rag.extractor.unstructured.unstructured_epub_extractor import UnstructuredEpubExtractor
 from core.rag.extractor.unstructured.unstructured_markdown_extractor import UnstructuredMarkdownExtractor
 from core.rag.extractor.unstructured.unstructured_msg_extractor import UnstructuredMsgExtractor
 from core.rag.extractor.unstructured.unstructured_ppt_extractor import UnstructuredPPTExtractor
@@ -83,7 +84,7 @@ class ExtractProcessor:
                 etl_type = current_app.config['ETL_TYPE']
                 unstructured_api_url = current_app.config['UNSTRUCTURED_API_URL']
                 if etl_type == 'Unstructured':
-                    if file_extension == '.xlsx':
+                    if file_extension == '.xlsx' or file_extension == '.xls':
                         extractor = ExcelExtractor(file_path)
                     elif file_extension == '.pdf':
                         extractor = PdfExtractor(file_path)
@@ -106,12 +107,14 @@ class ExtractProcessor:
                         extractor = UnstructuredPPTXExtractor(file_path, unstructured_api_url)
                     elif file_extension == '.xml':
                         extractor = UnstructuredXmlExtractor(file_path, unstructured_api_url)
+                    elif file_extension == 'epub':
+                        extractor = UnstructuredEpubExtractor(file_path, unstructured_api_url)
                     else:
                         # txt
                         extractor = UnstructuredTextExtractor(file_path, unstructured_api_url) if is_automatic \
                             else TextExtractor(file_path, autodetect_encoding=True)
                 else:
-                    if file_extension == '.xlsx':
+                    if file_extension == '.xlsx' or file_extension == '.xls':
                         extractor = ExcelExtractor(file_path)
                     elif file_extension == '.pdf':
                         extractor = PdfExtractor(file_path)
@@ -123,6 +126,8 @@ class ExtractProcessor:
                         extractor = WordExtractor(file_path)
                     elif file_extension == '.csv':
                         extractor = CSVExtractor(file_path, autodetect_encoding=True)
+                    elif file_extension == 'epub':
+                        extractor = UnstructuredEpubExtractor(file_path)
                     else:
                         # txt
                         extractor = TextExtractor(file_path, autodetect_encoding=True)

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

@@ -0,0 +1,37 @@
+import logging
+
+from core.rag.extractor.extractor_base import BaseExtractor
+from core.rag.models.document import Document
+
+logger = logging.getLogger(__name__)
+
+
+class UnstructuredEpubExtractor(BaseExtractor):
+    """Load epub files.
+
+
+    Args:
+        file_path: Path to the file to load.
+    """
+
+    def __init__(
+        self,
+        file_path: str,
+        api_url: str = None,
+    ):
+        """Initialize with file path."""
+        self._file_path = file_path
+        self._api_url = api_url
+
+    def extract(self) -> list[Document]:
+        from unstructured.partition.epub import partition_epub
+
+        elements = partition_epub(filename=self._file_path, xml_keep_tags=True)
+        from unstructured.chunking.title import chunk_by_title
+        chunks = chunk_by_title(elements, max_characters=2000, combine_text_under_n_chars=2000)
+        documents = []
+        for chunk in chunks:
+            text = chunk.text.strip()
+            documents.append(Document(page_content=text))
+
+        return documents

api/core/rag/retrieval/dataset_retrieval.py

@@ -2,7 +2,6 @@ import threading
 from typing import Optional, cast
 
 from flask import Flask, current_app
-from langchain.tools import BaseTool
 
 from core.app.app_config.entities import DatasetEntity, DatasetRetrieveConfigEntity
 from core.app.entities.app_invoke_entities import InvokeFrom, ModelConfigWithCredentialsEntity
@@ -19,6 +18,7 @@ from core.rag.retrieval.router.multi_dataset_function_call_router import Functio
 from core.rag.retrieval.router.multi_dataset_react_route import ReactMultiDatasetRouter
 from core.rerank.rerank import RerankRunner
 from core.tools.tool.dataset_retriever.dataset_multi_retriever_tool import DatasetMultiRetrieverTool
+from core.tools.tool.dataset_retriever.dataset_retriever_base_tool import DatasetRetrieverBaseTool
 from core.tools.tool.dataset_retriever.dataset_retriever_tool import DatasetRetrieverTool
 from extensions.ext_database import db
 from models.dataset import Dataset, DatasetQuery, DocumentSegment
@@ -383,7 +383,7 @@ class DatasetRetrieval:
                                   return_resource: bool,
                                   invoke_from: InvokeFrom,
                                   hit_callback: DatasetIndexToolCallbackHandler) \
-            -> Optional[list[BaseTool]]:
+            -> Optional[list[DatasetRetrieverBaseTool]]:
         """
         A dataset tool is a tool that can be used to retrieve information from a dataset
         :param tenant_id: tenant id

api/core/rag/retrieval/output_parser/react_output.py

@@ -0,0 +1,25 @@
+from __future__ import annotations
+
+from dataclasses import dataclass
+from typing import NamedTuple, Union
+
+
+@dataclass
+class ReactAction:
+    """A full description of an action for an ReactAction to execute."""
+
+    tool: str
+    """The name of the Tool to execute."""
+    tool_input: Union[str, dict]
+    """The input to pass in to the Tool."""
+    log: str
+    """Additional information to log about the action."""
+
+
+class ReactFinish(NamedTuple):
+    """The final return value of an ReactFinish."""
+
+    return_values: dict
+    """Dictionary of return values."""
+    log: str
+    """Additional information to log about the return value"""

api/core/rag/retrieval/output_parser/structured_chat.py

@@ -2,28 +2,24 @@ import json
 import re
 from typing import Union
 
-from langchain.agents.structured_chat.output_parser import StructuredChatOutputParser as LCStructuredChatOutputParser
-from langchain.agents.structured_chat.output_parser import logger
-from langchain.schema import AgentAction, AgentFinish, OutputParserException
+from core.rag.retrieval.output_parser.react_output import ReactAction, ReactFinish
 
 
-class StructuredChatOutputParser(LCStructuredChatOutputParser):
-    def parse(self, text: str) -> Union[AgentAction, AgentFinish]:
+class StructuredChatOutputParser:
+    def parse(self, text: str) -> Union[ReactAction, ReactFinish]:
         try:
             action_match = re.search(r"```(\w*)\n?({.*?)```", text, re.DOTALL)
             if action_match is not None:
                 response = json.loads(action_match.group(2).strip(), strict=False)
                 if isinstance(response, list):
-                    # gpt turbo frequently ignores the directive to emit a single action
-                    logger.warning("Got multiple action responses: %s", response)
                     response = response[0]
                 if response["action"] == "Final Answer":
-                    return AgentFinish({"output": response["action_input"]}, text)
+                    return ReactFinish({"output": response["action_input"]}, text)
                 else:
-                    return AgentAction(
+                    return ReactAction(
                         response["action"], response.get("action_input", {}), text
                     )
             else:
-                return AgentFinish({"output": text}, text)
+                return ReactFinish({"output": text}, text)
         except Exception as e:
-            raise OutputParserException(f"Could not parse LLM output: {text}")
+            raise ValueError(f"Could not parse LLM output: {text}")

api/core/rag/retrieval/router/multi_dataset_react_route.py

@@ -1,20 +1,21 @@
 from collections.abc import Generator, Sequence
-from typing import Optional, Union
-
-from langchain import PromptTemplate
-from langchain.agents.structured_chat.base import HUMAN_MESSAGE_TEMPLATE
-from langchain.agents.structured_chat.prompt import PREFIX, SUFFIX
-from langchain.schema import AgentAction
+from typing import Union
 
 from core.app.entities.app_invoke_entities import ModelConfigWithCredentialsEntity
 from core.model_manager import ModelInstance
 from core.model_runtime.entities.llm_entities import LLMUsage
 from core.model_runtime.entities.message_entities import PromptMessage, PromptMessageRole, PromptMessageTool
 from core.prompt.advanced_prompt_transform import AdvancedPromptTransform
-from core.prompt.entities.advanced_prompt_entities import ChatModelMessage
+from core.prompt.entities.advanced_prompt_entities import ChatModelMessage, CompletionModelPromptTemplate
+from core.rag.retrieval.output_parser.react_output import ReactAction
 from core.rag.retrieval.output_parser.structured_chat import StructuredChatOutputParser
 from core.workflow.nodes.llm.llm_node import LLMNode
 
+PREFIX = """Respond to the human as helpfully and accurately as possible. You have access to the following tools:"""
+
+SUFFIX = """Begin! Reminder to ALWAYS respond with a valid json blob of a single action. Use tools if necessary. Respond directly if appropriate. Format is Action:```$JSON_BLOB```then Observation:.
+Thought:"""
+
 FORMAT_INSTRUCTIONS = """Use a json blob to specify a tool by providing an action key (tool name) and an action_input key (tool input).
 The nouns in the format of "Thought", "Action", "Action Input", "Final Answer" must be expressed in English.
 Valid "action" values: "Final Answer" or {tool_names}
@@ -86,7 +87,6 @@ class ReactMultiDatasetRouter:
             tenant_id: str,
             prefix: str = PREFIX,
             suffix: str = SUFFIX,
-            human_message_template: str = HUMAN_MESSAGE_TEMPLATE,
             format_instructions: str = FORMAT_INSTRUCTIONS,
     ) -> Union[str, None]:
         if model_config.mode == "chat":
@@ -95,7 +95,6 @@ class ReactMultiDatasetRouter:
                 tools=tools,
                 prefix=prefix,
                 suffix=suffix,
-                human_message_template=human_message_template,
                 format_instructions=format_instructions,
             )
         else:
@@ -103,7 +102,6 @@ class ReactMultiDatasetRouter:
                 tools=tools,
                 prefix=prefix,
                 format_instructions=format_instructions,
-                input_variables=None
             )
         stop = ['Observation:']
         # handle invoke result
@@ -127,9 +125,9 @@ class ReactMultiDatasetRouter:
             tenant_id=tenant_id
         )
         output_parser = StructuredChatOutputParser()
-        agent_decision = output_parser.parse(result_text)
-        if isinstance(agent_decision, AgentAction):
-            return agent_decision.tool
+        react_decision = output_parser.parse(result_text)
+        if isinstance(react_decision, ReactAction):
+            return react_decision.tool
         return None
 
     def _invoke_llm(self, completion_param: dict,
@@ -139,7 +137,6 @@ class ReactMultiDatasetRouter:
                     ) -> tuple[str, LLMUsage]:
         """
             Invoke large language model
-            :param node_data: node data
             :param model_instance: model instance
             :param prompt_messages: prompt messages
             :param stop: stop
@@ -197,7 +194,6 @@ class ReactMultiDatasetRouter:
             tools: Sequence[PromptMessageTool],
             prefix: str = PREFIX,
             suffix: str = SUFFIX,
-            human_message_template: str = HUMAN_MESSAGE_TEMPLATE,
             format_instructions: str = FORMAT_INSTRUCTIONS,
     ) -> list[ChatModelMessage]:
         tool_strings = []
@@ -227,16 +223,13 @@ class ReactMultiDatasetRouter:
             tools: Sequence[PromptMessageTool],
             prefix: str = PREFIX,
             format_instructions: str = FORMAT_INSTRUCTIONS,
-            input_variables: Optional[list[str]] = None,
-    ) -> PromptTemplate:
+    ) -> CompletionModelPromptTemplate:
         """Create prompt in the style of the zero shot agent.
 
         Args:
             tools: List of tools the agent will have access to, used to format the
                 prompt.
             prefix: String to put before the list of tools.
-            input_variables: List of input variables the final prompt will expect.
-
         Returns:
             A PromptTemplate with the template assembled from the pieces here.
         """
@@ -249,6 +242,4 @@ Thought: {agent_scratchpad}
         tool_names = ", ".join([tool.name for tool in tools])
         format_instructions = format_instructions.format(tool_names=tool_names)
         template = "\n\n".join([prefix, tool_strings, format_instructions, suffix])
-        if input_variables is None:
-            input_variables = ["input", "agent_scratchpad"]
-        return PromptTemplate(template=template, input_variables=input_variables)
+        return CompletionModelPromptTemplate(text=template)

api/core/tools/provider/_position.yaml

@@ -1,6 +1,7 @@
 - google
 - bing
 - duckduckgo
+- searxng
 - dalle
 - azuredalle
 - wikipedia
@@ -12,6 +13,7 @@
 - pubmed
 - stablediffusion
 - webscraper
+- jina
 - model.zhipuai
 - aippt
 - youtube

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

@@ -0,0 +1,4 @@
+<svg width="24" height="24" viewBox="0 0 24 24" fill="none" xmlns="http://www.w3.org/2000/svg">
+<path d="M6.56053 21.4486C9.07925 21.4486 11.1211 19.4068 11.1211 16.8882C11.1211 14.3696 9.07925 12.3279 6.56053 12.3279C4.04182 12.3279 2 14.3696 2 16.8882C2 19.4068 4.04182 21.4486 6.56053 21.4486Z" fill="#EB6161"/>
+<path d="M22.0002 3.59467L21.9406 12.3279C21.9406 17.3055 17.9464 21.3591 12.9685 21.4485L12.8789 12.3577L12.8791 3.62447C12.8791 3.02835 13.356 2.55145 13.9522 2.55145H20.9271C21.5233 2.55145 22.0002 2.99854 22.0002 3.59467Z" fill="#009191"/>
+</svg>

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

@@ -0,0 +1,12 @@
+from typing import Any
+
+from core.tools.errors import ToolProviderCredentialValidationError
+from core.tools.provider.builtin_tool_provider import BuiltinToolProviderController
+
+
+class GoogleProvider(BuiltinToolProviderController):
+    def _validate_credentials(self, credentials: dict[str, Any]) -> None:
+        try:
+            pass
+        except Exception as e:
+            raise ToolProviderCredentialValidationError(str(e))

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

@@ -0,0 +1,13 @@
+identity:
+  author: Dify
+  name: jina
+  label:
+    en_US: JinaReader
+    zh_Hans: JinaReader
+    pt_BR: JinaReader
+  description:
+    en_US: Convert any URL to an LLM-friendly input. Experience improved output for your agent and RAG systems at no cost.
+    zh_Hans: 将任何 URL 转换为 LLM 友好的输入。无需付费即可体验为您的 Agent 和 RAG 系统提供的改进输出。
+    pt_BR: Converta qualquer URL em uma entrada amigável ao LLM. Experimente uma saída aprimorada para seus sistemas de agente e RAG sem custo.
+  icon: icon.svg
+credentials_for_provider:

api/core/tools/provider/builtin/jina/tools/jina_reader.py

@@ -0,0 +1,35 @@
+from typing import Any, Union
+
+from yarl import URL
+
+from core.helper import ssrf_proxy
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+
+
+class JinaReaderTool(BuiltinTool):
+    _jina_reader_endpoint = 'https://r.jina.ai/'
+
+    def _invoke(self, 
+                user_id: str,
+               tool_parameters: dict[str, Any], 
+        ) -> Union[ToolInvokeMessage, list[ToolInvokeMessage]]:
+        """
+            invoke tools
+        """
+        url = tool_parameters['url']
+
+        headers = {
+            'Accept': 'text/event-stream'
+        }
+
+        response = ssrf_proxy.get(
+            str(URL(self._jina_reader_endpoint + url)), 
+            headers=headers,
+            timeout=(10, 60)
+        )
+
+        if tool_parameters.get('summary', False):
+            return self.create_text_message(self.summary(user_id, response.text))
+        
+        return self.create_text_message(response.text)

api/core/tools/provider/builtin/jina/tools/jina_reader.yaml

@@ -0,0 +1,41 @@
+identity:
+  name: jina_reader
+  author: Dify
+  label:
+    en_US: JinaReader
+    zh_Hans: JinaReader
+    pt_BR: JinaReader
+description:
+  human:
+    en_US: Convert any URL to an LLM-friendly input. Experience improved output for your agent and RAG systems at no cost.
+    zh_Hans: 将任何 URL 转换为 LLM 友好的输入。无需付费即可体验为您的 Agent 和 RAG 系统提供的改进输出。
+    pt_BR: Converta qualquer URL em uma entrada amigável ao LLM. Experimente uma saída aprimorada para seus sistemas de agente e RAG sem custo.
+  llm: A tool for scraping webpages. Input should be a URL.
+parameters:
+  - name: url
+    type: string
+    required: true
+    label:
+      en_US: URL
+      zh_Hans: 网页链接
+      pt_BR: URL
+    human_description:
+      en_US: used for linking to webpages
+      zh_Hans: 用于链接到网页
+      pt_BR: used for linking to webpages
+    llm_description: url for scraping
+    form: llm
+  - name: summary
+    type: boolean
+    required: false
+    default: false
+    label:
+      en_US: Enable summary
+      zh_Hans: 是否启用摘要
+      pt_BR: Habilitar resumo
+    human_description:
+      en_US: Enable summary for the output
+      zh_Hans: 为输出启用摘要
+      pt_BR: Habilitar resumo para a saída
+    llm_description: enable summary
+    form: form

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

@@ -0,0 +1,56 @@
+<?xml version="1.0" encoding="UTF-8" standalone="no"?>
+<svg
+   xmlns:dc="http://purl.org/dc/elements/1.1/"
+   xmlns:cc="http://creativecommons.org/ns#"
+   xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
+   xmlns:svg="http://www.w3.org/2000/svg"
+   xmlns="http://www.w3.org/2000/svg"
+   id="svg8"
+   version="1.1"
+   viewBox="0 0 92 92"
+   height="92mm"
+   width="92mm">
+  <defs
+     id="defs2" />
+  <metadata
+     id="metadata5">
+    <rdf:RDF>
+      <cc:Work
+         rdf:about="">
+        <dc:format>image/svg+xml</dc:format>
+        <dc:type
+           rdf:resource="http://purl.org/dc/dcmitype/StillImage" />
+        <dc:title></dc:title>
+      </cc:Work>
+    </rdf:RDF>
+  </metadata>
+  <g
+     transform="translate(-40.921303,-17.416526)"
+     id="layer1">
+    <circle
+       r="0"
+       style="fill:none;stroke:#000000;stroke-width:12;stroke-miterlimit:4;stroke-dasharray:none;stroke-opacity:1"
+       cy="92"
+       cx="75"
+       id="path3713" />
+    <circle
+       r="30"
+       cy="53.902557"
+       cx="75.921303"
+       id="path834"
+       style="fill:none;fill-opacity:1;stroke:#3050ff;stroke-width:10;stroke-miterlimit:4;stroke-dasharray:none;stroke-opacity:1" />
+    <path
+       d="m 67.514849,37.91524 a 18,18 0 0 1 21.051475,3.312407 18,18 0 0 1 3.137312,21.078282"
+       id="path852"
+       style="fill:none;fill-opacity:1;stroke:#3050ff;stroke-width:5;stroke-miterlimit:4;stroke-dasharray:none;stroke-opacity:1" />
+    <rect
+       transform="rotate(-46.234709)"
+       ry="1.8669105e-13"
+       y="122.08995"
+       x="3.7063529"
+       height="39.963303"
+       width="18.846331"
+       id="rect912"
+       style="opacity:1;fill:#3050ff;fill-opacity:1;stroke:none;stroke-width:8;stroke-miterlimit:4;stroke-dasharray:none;stroke-opacity:1" />
+  </g>
+</svg>

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

@@ -0,0 +1,25 @@
+from typing import Any
+
+from core.tools.errors import ToolProviderCredentialValidationError
+from core.tools.provider.builtin.searxng.tools.searxng_search import SearXNGSearchTool
+from core.tools.provider.builtin_tool_provider import BuiltinToolProviderController
+
+
+class SearXNGProvider(BuiltinToolProviderController):
+    def _validate_credentials(self, credentials: dict[str, Any]) -> None:
+        try:
+            SearXNGSearchTool().fork_tool_runtime(
+                meta={
+                    "credentials": credentials,
+                }
+            ).invoke(
+                user_id='',
+                tool_parameters={
+                    "query": "SearXNG",
+                    "limit": 1,
+                    "search_type": "page",
+                    "result_type": "link"
+                },
+            )
+        except Exception as e:
+            raise ToolProviderCredentialValidationError(str(e))

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

@@ -0,0 +1,24 @@
+identity:
+  author: Junytang
+  name: searxng
+  label:
+    en_US: SearXNG
+    zh_Hans: SearXNG
+  description:
+    en_US: A free internet metasearch engine.
+    zh_Hans: 开源互联网元搜索引擎
+  icon: icon.svg
+credentials_for_provider:
+  searxng_base_url:
+    type: secret-input
+    required: true
+    label:
+      en_US: SearXNG base URL
+      zh_Hans: SearXNG base URL
+    help:
+      en_US: Please input your SearXNG base URL
+      zh_Hans: 请输入您的 SearXNG base URL
+    placeholder:
+      en_US: Please input your SearXNG base URL
+      zh_Hans: 请输入您的 SearXNG base URL
+    url: https://docs.dify.ai/tutorials/tool-configuration/searxng

api/core/tools/provider/builtin/searxng/tools/searxng_search.py

@@ -0,0 +1,124 @@
+import json
+from typing import Any
+
+import requests
+
+from core.tools.entities.tool_entities import ToolInvokeMessage
+from core.tools.tool.builtin_tool import BuiltinTool
+
+
+class SearXNGSearchResults(dict):
+    """Wrapper for search results."""
+
+    def __init__(self, data: str):
+        super().__init__(json.loads(data))
+        self.__dict__ = self
+
+    @property
+    def results(self) -> Any:
+        return self.get("results", [])
+
+
+class SearXNGSearchTool(BuiltinTool):
+    """
+    Tool for performing a search using SearXNG engine.
+    """
+
+    SEARCH_TYPE = {
+        "page": "general",
+        "news": "news",
+        "image": "images",
+        # "video": "videos",
+        # "file": "files"
+    }
+    LINK_FILED = {
+        "page": "url",
+        "news": "url",
+        "image": "img_src",
+        # "video": "iframe_src",
+        # "file": "magnetlink"
+    }
+    TEXT_FILED = {
+        "page": "content",
+        "news": "content",
+        "image": "img_src",
+        # "video": "iframe_src",
+        # "file": "magnetlink"
+    }
+
+    def _invoke_query(self, user_id: str, host: str, query: str, search_type: str, result_type: str, topK: int = 5) -> list[dict]:
+        """Run query and return the results."""
+
+        search_type = search_type.lower()
+        if search_type not in self.SEARCH_TYPE.keys():
+            search_type= "page"
+
+        response = requests.get(host, params={
+            "q": query, 
+            "format": "json", 
+            "categories": self.SEARCH_TYPE[search_type]
+        })
+
+        if response.status_code != 200:
+            raise Exception(f'Error {response.status_code}: {response.text}')
+        
+        search_results = SearXNGSearchResults(response.text).results[:topK]
+
+        if result_type == 'link':
+            results = []
+            if search_type == "page" or search_type == "news":
+                for r in search_results:
+                    results.append(self.create_text_message(
+                        text=f'{r["title"]}: {r.get(self.LINK_FILED[search_type], "")}'
+                    ))
+            elif search_type == "image":
+                for r in search_results:
+                    results.append(self.create_image_message(
+                        image=r.get(self.LINK_FILED[search_type], "")
+                    ))
+            else:
+                for r in search_results:
+                    results.append(self.create_link_message(
+                        link=r.get(self.LINK_FILED[search_type], "")
+                    ))
+
+            return results
+        else:
+            text = ''
+            for i, r in enumerate(search_results):
+                text += f'{i+1}: {r["title"]} - {r.get(self.TEXT_FILED[search_type], "")}\n'
+
+            return self.create_text_message(text=self.summary(user_id=user_id, content=text))
+
+
+    def _invoke(self, user_id: str, tool_parameters: dict[str, Any]) -> ToolInvokeMessage | list[ToolInvokeMessage]:
+        """
+        Invoke the SearXNG search tool.
+
+        Args:
+            user_id (str): The ID of the user invoking the tool.
+            tool_parameters (dict[str, Any]): The parameters for the tool invocation.
+
+        Returns:
+            ToolInvokeMessage | list[ToolInvokeMessage]: The result of the tool invocation.
+        """
+
+        host = self.runtime.credentials.get('searxng_base_url', None)
+        if not host:
+            raise Exception('SearXNG api is required')
+                
+        query = tool_parameters.get('query', None)
+        if not query:
+            return self.create_text_message('Please input query')
+                
+        num_results = min(tool_parameters.get('num_results', 5), 20)
+        search_type = tool_parameters.get('search_type', 'page') or 'page'
+        result_type = tool_parameters.get('result_type', 'text') or 'text'
+
+        return self._invoke_query(
+            user_id=user_id, 
+            host=host, 
+            query=query, 
+            search_type=search_type, 
+            result_type=result_type, 
+            topK=num_results)

api/core/tools/provider/builtin/searxng/tools/searxng_search.yaml

@@ -0,0 +1,89 @@
+identity:
+  name: searxng_search
+  author: Tice
+  label:
+    en_US: SearXNG Search
+    zh_Hans: SearXNG 搜索
+description:
+  human:
+    en_US: Perform searches on SearXNG and get results.
+    zh_Hans: 在 SearXNG 上进行搜索并获取结果。
+  llm: Perform searches on SearXNG and get results.
+parameters:
+  - name: query
+    type: string
+    required: true
+    label:
+      en_US: Query string
+      zh_Hans: 查询语句
+    human_description:
+      en_US: The search query.
+      zh_Hans: 搜索查询语句。
+    llm_description: Key words for searching
+    form: llm
+  - name: search_type
+    type: select
+    required: true
+    label:
+      en_US: search type
+      zh_Hans: 搜索类型
+      pt_BR: search type
+    human_description:
+      en_US: search type for page, news or image.
+      zh_Hans: 选择搜索的类型：网页，新闻，图片。
+      pt_BR: search type for page, news or image.
+    default: Page
+    options:
+      - value: Page
+        label:
+          en_US: Page
+          zh_Hans: 网页
+          pt_BR: Page
+      - value: News
+        label:
+          en_US: News
+          zh_Hans: 新闻
+          pt_BR: News
+      - value: Image
+        label:
+          en_US: Image
+          zh_Hans: 图片
+          pt_BR: Image
+    form: form
+  - name: num_results
+    type: number
+    required: true
+    label:
+      en_US: Number of query results
+      zh_Hans: 返回查询数量
+    human_description:
+      en_US: The number of query results.
+      zh_Hans: 返回查询结果的数量。
+    form: form
+    default: 5
+    min: 1
+    max: 20
+  - name: result_type
+    type: select
+    required: true
+    label:
+      en_US: result type
+      zh_Hans: 结果类型
+      pt_BR: result type
+    human_description:
+      en_US: return a list of links or texts.
+      zh_Hans: 返回一个连接列表还是纯文本内容。
+      pt_BR: return a list of links or texts.
+    default: text
+    options:
+      - value: link
+        label:
+          en_US: Link
+          zh_Hans: 链接
+          pt_BR: Link
+      - value: text
+        label:
+          en_US: Text
+          zh_Hans: 文本
+          pt_BR: Text
+    form: form