bstraehle commited on
Commit
8acf5f0
·
verified ·
1 Parent(s): 4352501

Update util.py

Browse files
Files changed (1) hide show
  1. util.py +5 -4
util.py CHANGED
@@ -39,17 +39,18 @@ def read_docx(file_path):
39
  for block in doc.element.body:
40
  if block.tag.endswith("p"):
41
  for paragraph in doc.paragraphs:
42
- if paragraph._element == block and paragraph.text:
43
- text.append(paragraph.text)
 
 
 
44
  elif block.tag.endswith("tbl"):
45
  for table in doc.tables:
46
  if table._element == block:
47
  for row in table.rows:
48
  row_text = []
49
-
50
  for cell in row.cells:
51
  row_text.append(cell.text.strip())
52
-
53
  text.append(" | ".join(row_text))
54
 
55
  return "\n".join(text)
 
39
  for block in doc.element.body:
40
  if block.tag.endswith("p"):
41
  for paragraph in doc.paragraphs:
42
+ if paragraph._element == block:
43
+ if paragraph.style.name.startswith("Heading"):
44
+ text.append("\n**" + paragraph.text + "**\n")
45
+ elif paragraph.text:
46
+ text.append(paragraph.text)
47
  elif block.tag.endswith("tbl"):
48
  for table in doc.tables:
49
  if table._element == block:
50
  for row in table.rows:
51
  row_text = []
 
52
  for cell in row.cells:
53
  row_text.append(cell.text.strip())
 
54
  text.append(" | ".join(row_text))
55
 
56
  return "\n".join(text)