Spaces:
Runtime error
Runtime error
Update extractor.py
Browse files- extractor.py +4 -0
extractor.py
CHANGED
|
@@ -3,7 +3,9 @@ from pypdf import PdfReader
|
|
| 3 |
|
| 4 |
|
| 5 |
def extract_ppt_text(file):
|
|
|
|
| 6 |
prs = Presentation(file)
|
|
|
|
| 7 |
text = ""
|
| 8 |
|
| 9 |
for slide in prs.slides:
|
|
@@ -15,11 +17,13 @@ def extract_ppt_text(file):
|
|
| 15 |
|
| 16 |
|
| 17 |
def extract_pdf_text(file):
|
|
|
|
| 18 |
reader = PdfReader(file)
|
| 19 |
|
| 20 |
text = ""
|
| 21 |
|
| 22 |
for page in reader.pages:
|
|
|
|
| 23 |
page_text = page.extract_text()
|
| 24 |
|
| 25 |
if page_text:
|
|
|
|
| 3 |
|
| 4 |
|
| 5 |
def extract_ppt_text(file):
|
| 6 |
+
|
| 7 |
prs = Presentation(file)
|
| 8 |
+
|
| 9 |
text = ""
|
| 10 |
|
| 11 |
for slide in prs.slides:
|
|
|
|
| 17 |
|
| 18 |
|
| 19 |
def extract_pdf_text(file):
|
| 20 |
+
|
| 21 |
reader = PdfReader(file)
|
| 22 |
|
| 23 |
text = ""
|
| 24 |
|
| 25 |
for page in reader.pages:
|
| 26 |
+
|
| 27 |
page_text = page.extract_text()
|
| 28 |
|
| 29 |
if page_text:
|