feat(文件管理): 新增支持HTML文件类型解析并完善许可证说明 (#95)

README.md

@@ -142,6 +142,18 @@ pnpm dev
 
 目前该项目仍在持续更新中，更新日志会在我的微信公众号[我有一计]上发布，欢迎关注。
 
+## 📜  许可证与使用限制
+1. **本仓库基于AGPLv3许可证**  
+   由于包含第三方AGPLv3代码，本项目必须遵循AGPLv3的全部条款。这意味着：
+   - 任何**衍生作品**（包括修改或组合代码）必须继续使用AGPLv3并公开源代码。  
+   - 若通过**网络服务**提供本软件，用户有权获取对应源码。
+
+2. **商用限制说明**  
+   - 本项目**不允许闭源商用**，除非获得所有代码版权持有人的书面授权（包括上游AGPLv3代码作者）。
+
+3. **免责声明**  
+   本项目不提供任何担保，使用者需自行承担合规风险。若需法律建议，请咨询专业律师。
+   
 ## ✨ Star History
 
 ![Stargazers over time](https://starchart.cc/zstar1003/ragflow-plus.svg)

README_EN.md

@@ -109,6 +109,19 @@ This project is developed based on the following open-source projects:
 
 This project is under continuous development. Update logs will be posted on my WeChat public account **[我有一计]** — feel free to follow!
 
+## 📜 License and Usage Restrictions
+
+1. **This Repository is Licensed Under AGPLv3**  
+   As it incorporates third-party AGPLv3 code, this project must fully comply with AGPLv3 terms. This means:
+   - Any **derivative works** (including modifications or combined code) must remain under AGPLv3 with source code publicly available.  
+   - If provided as a **network service**, users are entitled to obtain the corresponding source code.
+
+2. **Commercial Use Restrictions**  
+   - **Closed-source commercial use is prohibited** unless obtaining written authorization from all copyright holders (including original AGPLv3 code authors).
+
+3. **Disclaimer**  
+   This project comes with no warranties. Users shall bear all compliance risks. Consult legal professionals for legal advice.
+
 ## ✨ Star History
 
 ![Stargazers over time](https://starchart.cc/zstar1003/ragflow-plus.svg)

management/server/services/files/file_service.py

@@ -1,4 +1,4 @@
-from peewee import *
+from peewee import *  # noqa: F403
 from .base_service import BaseService
 from .models import File
 from .utils import FileType, get_uuid

management/server/services/files/service.py

@@ -13,7 +13,7 @@ load_dotenv("../../docker/.env")
 
 temp_dir = tempfile.gettempdir()
 UPLOAD_FOLDER = os.path.join(temp_dir, "uploads")
-ALLOWED_EXTENSIONS = {"pdf", "doc", "docx", "ppt", "pptx", "xls", "xlsx", "jpg", "jpeg", "png", "txt", "md"}
+ALLOWED_EXTENSIONS = {"pdf", "doc", "docx", "ppt", "pptx", "xls", "xlsx", "jpg", "jpeg", "png", "bmp", "txt", "md", "html"}
 
 
 def allowed_file(filename):
@@ -37,6 +37,8 @@ def filename_type(filename):
         return FileType.PPT.value
     elif ext in [".txt", ".md"]:
         return FileType.TEXT.value
+    elif ext in [".html"]:
+        return FileType.HTML.value
 
     return FileType.OTHER.value
 
@@ -602,5 +604,7 @@ def upload_files_to_server(files, parent_id=None, user_id=None):
                 # 删除临时文件
                 if os.path.exists(filepath):
                     os.remove(filepath)
+        else:
+            raise RuntimeError({"name": filename, "error": "不支持的文件类型", "status": "failed"})
 
     return {"code": 0, "data": results, "message": f"成功上传 {len([r for r in results if r['status'] == 'success'])}/{len(files)} 个文件"}

management/server/services/files/utils.py

@@ -12,6 +12,7 @@ class FileType(StrEnum):
     PPT = "ppt"
     VISUAL = "visual"
     TEXT = "txt"
+    HTML = "html"
     OTHER = "other"
 
 class FileSource(StrEnum):

management/server/services/knowledgebases/document_parser.py

@@ -361,7 +361,7 @@ def perform_parse(doc_id, doc_info, file_info, embedding_config):
             middle_content = pipe_result.get_middle_json()
             middle_json_content = json.loads(middle_content)
 
-        elif file_type.endswith("word") or file_type.endswith("ppt") or file_type.endswith("txt") or file_type.endswith("md"):
+        elif file_type.endswith("word") or file_type.endswith("ppt") or file_type.endswith("txt") or file_type.endswith("md") or file_type.endswith("html"):
             update_progress(0.3, "使用MinerU解析器")
             # 创建临时文件保存文件内容
             temp_dir = tempfile.gettempdir()