(self, content:bytearray, max_pages=2)
| 39 | return None |
| 40 | |
| 41 | def _read_pdf(self, content:bytearray, max_pages=2)->str: |
| 42 | text = "" |
| 43 | pdf = io.BytesIO(content) |
| 44 | reader = PdfReader(pdf) |
| 45 | total_pages = len(reader.pages) |
| 46 | pages = min(total_pages, max_pages) |
| 47 | for page in reader.pages: |
| 48 | text+=page.extract_text()+'\n' |
| 49 | |
| 50 | return text |
| 51 | |
| 52 | def _read_word(self, zip_file, document_name)->str: |
| 53 | text="" |