hhk
Browse files
app.py
CHANGED
@@ -43,7 +43,7 @@ def extract_text(doc):
|
|
43 |
# Remove null bytes without decoding
|
44 |
raw = raw.replace(b'\x00', b'')
|
45 |
|
46 |
-
pdf = pdfplumber.open(BytesIO(raw))
|
47 |
pages = [page.extract_text() for page in pdf.pages]
|
48 |
return "\n".join(pages)
|
49 |
|
|
|
43 |
# Remove null bytes without decoding
|
44 |
raw = raw.replace(b'\x00', b'')
|
45 |
|
46 |
+
pdf = pdfplumber.open(io.BytesIO(raw))
|
47 |
pages = [page.extract_text() for page in pdf.pages]
|
48 |
return "\n".join(pages)
|
49 |
|