rehan953 commited on
Commit
60c3854
·
verified ·
1 Parent(s): 28e752d

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +5 -1
app.py CHANGED
@@ -238,15 +238,18 @@ def run_ocr(uploaded_file):
238
  if is_pdf:
239
  doc = fitz.open(path)
240
  page_images = []
 
241
  for i in range(len(doc)):
242
  pix = doc[i].get_pixmap(matrix=fitz.Matrix(1.5, 1.5), alpha=False)
243
  img_path = os.path.join(tempfile.gettempdir(), f"maas_page_{i}.png")
244
  pix.save(img_path)
245
  page_images.append(img_path)
 
246
  doc.close()
247
  results = parser.parse(page_images)
248
  else:
249
  page_images = [path]
 
250
  results = parser.parse(path)
251
 
252
  if not isinstance(results, list):
@@ -255,7 +258,8 @@ def run_ocr(uploaded_file):
255
  all_pages = []
256
  for page_num, page_result in enumerate(results):
257
  page_md, regions = get_page_md_and_regions(page_result)
258
- header_end_frac, footer_start_frac = get_header_footer_zones(regions, 1000)
 
259
 
260
  # MaaS does not return header/footer regions; always use at least default bands
261
  he = header_end_frac if header_end_frac is not None else DEFAULT_ZONE_FRAC
 
238
  if is_pdf:
239
  doc = fitz.open(path)
240
  page_images = []
241
+ page_heights = []
242
  for i in range(len(doc)):
243
  pix = doc[i].get_pixmap(matrix=fitz.Matrix(1.5, 1.5), alpha=False)
244
  img_path = os.path.join(tempfile.gettempdir(), f"maas_page_{i}.png")
245
  pix.save(img_path)
246
  page_images.append(img_path)
247
+ page_heights.append(pix.height)
248
  doc.close()
249
  results = parser.parse(page_images)
250
  else:
251
  page_images = [path]
252
+ page_heights = []
253
  results = parser.parse(path)
254
 
255
  if not isinstance(results, list):
 
258
  all_pages = []
259
  for page_num, page_result in enumerate(results):
260
  page_md, regions = get_page_md_and_regions(page_result)
261
+ img_h = page_heights[page_num] if page_num < len(page_heights) else 1000
262
+ header_end_frac, footer_start_frac = get_header_footer_zones(regions, img_h)
263
 
264
  # MaaS does not return header/footer regions; always use at least default bands
265
  he = header_end_frac if header_end_frac is not None else DEFAULT_ZONE_FRAC