Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -12,12 +12,14 @@ import traceback
|
|
12 |
|
13 |
def save_image(element, images):
|
14 |
try:
|
15 |
-
if hasattr(element, 'stream'):
|
16 |
image_data = element.stream.get_rawdata()
|
17 |
image = Image.open(io.BytesIO(image_data))
|
18 |
image_filename = f"extracted_image_{len(images)}.png"
|
19 |
image.save(image_filename)
|
20 |
images.append({"filename": image_filename})
|
|
|
|
|
21 |
except Exception as e:
|
22 |
print(f"Error extracting image: {e}")
|
23 |
|
@@ -45,6 +47,7 @@ def parse_pdf(pdf_file, output_format, progress=gr.Progress()):
|
|
45 |
if isinstance(element, LTTextBoxHorizontal):
|
46 |
text += element.get_text()
|
47 |
elif isinstance(element, (LTFigure, LTImage)):
|
|
|
48 |
save_image(element, images)
|
49 |
|
50 |
with pdfplumber.open(pdf_file) as pdf:
|
@@ -115,4 +118,4 @@ iface = gr.Interface(
|
|
115 |
)
|
116 |
|
117 |
if __name__ == "__main__":
|
118 |
-
iface.launch() # Temporarily disable sharing for debugging
|
|
|
12 |
|
13 |
def save_image(element, images):
|
14 |
try:
|
15 |
+
if hasattr(element, 'stream') and element.stream:
|
16 |
image_data = element.stream.get_rawdata()
|
17 |
image = Image.open(io.BytesIO(image_data))
|
18 |
image_filename = f"extracted_image_{len(images)}.png"
|
19 |
image.save(image_filename)
|
20 |
images.append({"filename": image_filename})
|
21 |
+
else:
|
22 |
+
print("No stream data for image element")
|
23 |
except Exception as e:
|
24 |
print(f"Error extracting image: {e}")
|
25 |
|
|
|
47 |
if isinstance(element, LTTextBoxHorizontal):
|
48 |
text += element.get_text()
|
49 |
elif isinstance(element, (LTFigure, LTImage)):
|
50 |
+
print(f"Processing element: {type(element)}")
|
51 |
save_image(element, images)
|
52 |
|
53 |
with pdfplumber.open(pdf_file) as pdf:
|
|
|
118 |
)
|
119 |
|
120 |
if __name__ == "__main__":
|
121 |
+
iface.launch() # Temporarily disable sharing for debugging
|