add parsed text to DB

app/db/db_types.py

@@ -104,6 +104,7 @@ def __init__(self, dictionary=None):
         self.is_failed = dictionary.get('is_failed', None)
         self.is_ended = dictionary.get('is_ended', True)
         self.is_passed = dictionary.get('is_passed', int(self.score) == 1)
+        self.parsed_chapters = dictionary.get('parsed_chapters', [])
 
     def calc_score(self):
         # check after implementation criterion pack

app/main/reports/parse_file/parse_file.py

@@ -0,0 +1,28 @@
+import re
+
+
+def parse_headers_and_pages(chapters, docx):
+    text_on_page = docx.pdf_file.get_text_on_page()
+    for page, text in text_on_page.items():
+        text = re.sub(r"(-\n)", "", text)
+        text = re.sub(r"\s\n", " ", text)
+        if "СОДЕРЖАНИЕ" in text:
+            continue
+        for chapter in chapters:
+            if chapter["header"] in text:
+                chapter["start_page"] = page
+    return chapters
+
+
+def parse_chapters(docx):
+    chapters = []
+    for chapter in 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"]
+            chapters.append({"header": head, "start_page": 0, "text": temp_text})
+    return chapters

app/server.py

@@ -18,6 +18,7 @@
 from flask_recaptcha import ReCaptcha
 
 import servants.user as user
+from app.main.reports.docx_uploader import DocxUploader
 from app.utils import format_check_for_table, check_file
 from db import db_methods
 from db.db_types import Check
@@ -29,6 +30,7 @@
 from servants import pre_luncher
 from tasks import create_task
 from utils import checklist_filter, decorator_assertion, get_file_len, format_check
+from main.reports.parse_file.parse_file import parse_chapters, parse_headers_and_pages
 
 logger = get_root_logger('web')
 UPLOAD_FOLDER = '/usr/src/project/files'
@@ -213,6 +215,15 @@ def run_task():
         converted_id = db_methods.add_file_to_db(filenamepdf, filepathpdf)
     else:
         converted_id = db_methods.write_pdf(filename, filepath)
+
+    parsed_file = DocxUploader()
+    parsed_file.upload(filepath)
+    parsed_file.parse()
+    parsed_file.make_chapters("VKR")
+    parsed_file.make_headers("VKR")
+    chapters = parse_chapters(parsed_file)
+    chapters_with_headers = parse_headers_and_pages(chapters, parsed_file)
+
     check = Check({
         '_id': file_id,
         'conv_pdf_fs_id': converted_id,
@@ -225,7 +236,8 @@ def run_task():
         'score': -1,  # score=-1 -> checking in progress
         'is_ended': False,
         'is_failed': False,
-        'params_for_passback': current_user.params_for_passback
+        'params_for_passback': current_user.params_for_passback,
+        'parsed_chapters': chapters_with_headers
     })
     db_methods.add_check(file_id, check)  # add check for parsed_file to db
     task = create_task.delay(check.pack(to_str=True))  # add check to queue