TextStorage class

app/main/reports/text_storage/text_storage.py

@@ -1,4 +1,5 @@
 from app.main.reports.docx_uploader import DocxUploader
+import pickle
 
 
 class TextStorage:
@@ -23,11 +24,14 @@ def parse_headers_and_pages(self):
 
     def parse_chapters(self):
         for chapter in self.docx.chapters:
+            head = chapter["styled_text"]["text"]
+            if "ПРИЛОЖЕНИЕ" in head:
+                head = head.split(".")[0]
             if chapter["child"] != [] and "heading" in chapter["style"]:
                 temp_text = ""
                 for i in range(len(chapter["child"])):
                     temp_text += chapter["child"][i]["styled_text"]["text"]
-                self.chapters.append({"header": chapter["styled_text"]["text"], "start_page": 0, "text": temp_text})
+                self.chapters.append({"header": head, "start_page": 0, "text": temp_text})
 
 
 if __name__ == "__main__":
@@ -36,5 +40,14 @@ def parse_chapters(self):
     text_storage = TextStorage(docx_path, pdf_path)
     text_storage.parse_chapters()
     text_storage.parse_headers_and_pages()
-    for ch in text_storage.chapters:
-        print(ch)
+    # for ch in text_storage.chapters:
+    #     print(ch)
+
+    with open("pickle.bin", "wb") as outfile:
+        pickle.dump(text_storage.chapters, outfile)
+
+    with open("pickle.bin", "rb") as infile:
+        deserialized = pickle.load(infile)
+
+    if text_storage.chapters == deserialized:
+        print("Serialization is successful")