Merge pull request #668 from LollipopsAndWine/dev

feat: 集成前端界面，配置一键启动

projects/web_api/web_api/api/__init__.py → projects/web_demo/web_demo/api/__init__.py

@@ -4,7 +4,7 @@ from common.web_hook import before_request
 from common.logger import setup_log
 
 root_dir = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
-print("root_dir", root_dir)
+
 
 def _register_db(flask_app):
     from common import import_models
@@ -30,6 +30,8 @@ def create_app(config):
     ma.init_app(app=app)
     from .analysis import analysis_blue
     app.register_blueprint(analysis_blue)
+    from .react_app import react_app_blue
+    app.register_blueprint(react_app_blue)
 
     app.before_request(before_request)
 

projects/web_api/web_api/api/analysis/__init__.py → projects/web_demo/web_demo/api/analysis/__init__.py

@@ -4,6 +4,7 @@ from .upload_view import UploadPdfView
 from .analysis_view import AnalysisTaskView, AnalysisTaskProgressView
 from .img_md_view import ImgView, MdView
 from .task_view import TaskView, HistoricalTasksView, DeleteTaskView
+from .markdown_view import MarkdownView
 
 analysis_blue = Blueprint('analysis', __name__)
 
@@ -15,4 +16,5 @@ api_v2.add_resource(ImgView, '/analysis/pdf_img')
 api_v2.add_resource(MdView, '/analysis/pdf_md')
 api_v2.add_resource(TaskView, '/extract/taskQueue')
 api_v2.add_resource(HistoricalTasksView, '/extract/list')
-api_v2.add_resource(DeleteTaskView, '/extract/task/<int:id>')
\ No newline at end of file
+api_v2.add_resource(DeleteTaskView, '/extract/task/<int:id>')
+api_v2.add_resource(MarkdownView, '/extract/markdown')
\ No newline at end of file

projects/web_api/web_api/api/analysis/analysis_view.py → projects/web_demo/web_demo/api/analysis/analysis_view.py

 import json
 import threading
+from multiprocessing import Process
 from pathlib import Path
 from flask import request, current_app, url_for
 from flask_restful import Resource
@@ -37,6 +38,7 @@ class AnalysisTaskProgressView(Resource):
                         "status": analysis_pdf.status,
                         "url": file_url,
                         "fileName": file_name,
+                        "file_key": analysis_task.file_key,
                         "content": [],
                         "markdownUrl": [],
                         "fullMdLink": "",
@@ -53,6 +55,7 @@ class AnalysisTaskProgressView(Resource):
                             "status": analysis_pdf.status,
                             "url": file_url,
                             "fileName": file_name,
+                            "file_key": analysis_task.file_key,
                             "content": bbox_info,
                             "markdownUrl": md_link_list,
                             "fullMdLink": full_md_link,
@@ -65,6 +68,7 @@ class AnalysisTaskProgressView(Resource):
                             "status": analysis_pdf.status,
                             "url": file_url,
                             "fileName": file_name,
+                            "file_key": analysis_task.file_key,
                             "content": [],
                             "markdownUrl": [],
                             "fullMdLink": "",
@@ -78,6 +82,7 @@ class AnalysisTaskProgressView(Resource):
                         "status": analysis_pdf.status,
                         "url": file_url,
                         "fileName": file_name,
+                        "file_key": analysis_task.file_key,
                         "content": [],
                         "markdownUrl": [],
                         "fullMdLink": "",
@@ -212,10 +217,10 @@ class AnalysisTaskView(Resource):
                     pdf_analysis_folder = current_app.config['PDF_ANALYSIS_FOLDER']
                     pdf_dir = f"{current_app.static_folder}/{pdf_analysis_folder}/{file_stem}"
                     image_dir = f"{pdf_dir}/images"
-                    t = threading.Thread(target=analysis_pdf_task,
-                                         args=(pdf_dir, image_dir, file_path, analysis_task.is_ocr,
-                                               analysis_task.analysis_pdf_id))
-                    t.start()
+                    process = Process(target=analysis_pdf_task,
+                                      args=(pdf_dir, image_dir, file_path, analysis_task.is_ocr,
+                                            analysis_task.analysis_pdf_id))
+                    process.start()
 
                 # 生成文件的URL路径
                 file_url = url_for('analysis.uploadpdfview', filename=analysis_task.file_name, as_attachment=False)

projects/web_demo/web_demo/api/analysis/markdown_view.py

+import json
+from pathlib import Path
+from flask import request, current_app
+from flask_restful import Resource
+from common.custom_response import generate_response
+
+
+class MarkdownView(Resource):
+
+    def put(self):
+        """
+        编辑markdown
+        """
+        params = json.loads(request.data)
+        file_key = params.get('file_key')
+        data = params.get('data', {})
+        if not data:
+            return generate_response(code=400, msg="empty data", msgZH="数据为空，无法更新markdown")
+
+        pdf_analysis_folder = current_app.config['PDF_ANALYSIS_FOLDER']
+        pdf_dir = f"{current_app.static_folder}/{pdf_analysis_folder}"
+        markdown_file_dir = ""
+        for path_obj in Path(pdf_dir).iterdir():
+            if path_obj.name.startswith(file_key):
+                markdown_file_dir = path_obj
+                break
+
+        if markdown_file_dir and Path(markdown_file_dir).exists():
+            for k, v in data.items():
+                md_path = f"{markdown_file_dir}/{k}.md"
+                if Path(md_path).exists():
+                    with open(md_path, 'w', encoding="utf-8") as f:
+                        f.write(v)
+
+            full_content = ""
+            for path_obj in Path(markdown_file_dir).iterdir():
+                if path_obj.is_file() and path_obj.suffix == ".md" and path_obj.stem != "full":
+                    with open(path_obj, 'r', encoding="utf-8") as f:
+                        full_content += f.read() + "\n"
+            with open(f"{markdown_file_dir}/full.md", 'w', encoding="utf-8") as f:
+                f.write(full_content)
+        else:
+            return generate_response(code=400, msg="Invalid file_key", msgZH="文件哈希错误")
+        return generate_response()

projects/web_api/web_api/api/analysis/pdf_ext.py → projects/web_demo/web_demo/api/analysis/pdf_ext.py

@@ -97,7 +97,7 @@ def analysis_pdf_task(pdf_dir, image_dir, pdf_path, is_ocr, analysis_pdf_id):
             full_md_content += item["md_content"] + "\n"
 
         full_md_name = "full.md"
-        with open(f"{pdf_dir}/{full_md_name}", "w") as file:
+        with open(f"{pdf_dir}/{full_md_name}", "w", encoding="utf-8") as file:
             file.write(full_md_content)
         with app.app_context():
             full_md_link = url_for('analysis.mdview', filename=full_md_name, as_attachment=False)
@@ -108,7 +108,7 @@ def analysis_pdf_task(pdf_dir, image_dir, pdf_path, is_ocr, analysis_pdf_id):
             for n, md in enumerate(json.loads(md_content)):
                 md_content = md["md_content"]
                 md_name = f"{md.get('page_no', n)}.md"
-                with open(f"{pdf_dir}/{md_name}", "w") as file:
+                with open(f"{pdf_dir}/{md_name}", "w", encoding="utf-8") as file:
                     file.write(md_content)
                 md_url = url_for('analysis.mdview', filename=md_name, as_attachment=False)
                 md_link_list.append(f"{md_url}&pdf={pdf_name}")

projects/web_api/web_api/api/extentions.py → projects/web_demo/web_demo/api/extentions.py

@@ -59,3 +59,4 @@ db = SQLAlchemy()
 migrate = Migrate()
 jwt = JWTManager()
 ma = Marshmallow()
+folder = app.config.get("REACT_APP_DIST")

projects/web_demo/web_demo/api/react_app/__init__.py

+from pathlib import Path
+from flask import Blueprint
+from ..extentions import app, Api
+from .react_app_view import ReactAppView
+from loguru import logger
+
+folder = Path(app.config.get("REACT_APP_DIST", "../../web/dist/")).resolve()
+logger.info(f"react_app folder: {folder}")
+react_app_blue = Blueprint('react_app', __name__,  static_folder=folder, static_url_path='', template_folder=folder)
+react_app_api = Api(react_app_blue, prefix='')
+react_app_api.add_resource(ReactAppView, '/')
\ No newline at end of file

projects/web_demo/web_demo/api/react_app/react_app_view.py

+from flask import render_template, Response
+from flask_restful import Resource
+
+
+class ReactAppView(Resource):
+    def get(self):
+        # 创建自定义的响应对象
+        rendered_template = render_template('index.html')
+        response = Response(rendered_template, mimetype='text/html')
+
+        return response