Spaces:
Build error
Build error
Commit ·
2e86579
1
Parent(s): df697c8
Delete scan_pdf_parser.py
Browse files- scan_pdf_parser.py +0 -10
scan_pdf_parser.py
DELETED
|
@@ -1,10 +0,0 @@
|
|
| 1 |
-
import pytesseract
|
| 2 |
-
from pdf2image import convert_from_path
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
def get_text_from_scanned_pdf(pdf_path):
|
| 6 |
-
text = ''
|
| 7 |
-
images = convert_from_path(pdf_path)
|
| 8 |
-
for img in images:
|
| 9 |
-
text += pytesseract.image_to_string(img)
|
| 10 |
-
return text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|