optimize recovery (#8346)

* optimize recovery

* update

ppstructure/predict_system.py

@@ -229,7 +229,9 @@ def main(args):
 
         if args.recovery and args.use_pdf2docx_api and flag_pdf:
             from pdf2docx.converter import Converter
-            docx_file = os.path.join(args.output, '{}.docx'.format(img_name))
+            os.makedirs(args.output, exist_ok=True)
+            docx_file = os.path.join(args.output,
+                                     '{}_api.docx'.format(img_name))
             cv = Converter(image_file)
             cv.convert(docx_file)
             cv.close()

ppstructure/recovery/recovery_to_doc.py

@@ -73,7 +73,7 @@ def convert_info_docx(img, res, save_folder, img_name):
                 text_run.font.size = shared.Pt(10)
 
     # save to docx
-    docx_path = os.path.join(save_folder, '{}.docx'.format(img_name))
+    docx_path = os.path.join(save_folder, '{}_ocr.docx'.format(img_name))
     doc.save(docx_path)
     logger.info('docx save to {}'.format(docx_path))
 

ppstructure/recovery/requirements.txt

 python-docx
-PyMuPDF==1.19.0
 beautifulsoup4
 fonttools>=4.24.0
 fire>=0.3.0

ppstructure/recovery/table_process.py

@@ -278,8 +278,6 @@ class HtmlToDocx(HTMLParser):
                 cell_col += colspan
             cell_row += 1
 
-        doc.save('1.docx')
-
     def handle_data(self, data):
         if self.skip:
             return

requirements.txt

@@ -14,4 +14,4 @@ lxml
 premailer
 openpyxl
 attrdict
-PyMuPDF==1.19.0
\ No newline at end of file
+PyMuPDF<1.21.0
\ No newline at end of file