Update util.py
Browse files
util.py
CHANGED
@@ -39,8 +39,11 @@ 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
|
43 |
-
|
|
|
|
|
|
|
44 |
elif block.tag.endswith("tbl"):
|
45 |
for table in doc.tables:
|
46 |
if table._element == block:
|
@@ -50,7 +53,7 @@ def read_docx(file_path):
|
|
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)
|
56 |
|
|
|
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:
|
|
|
53 |
for cell in row.cells:
|
54 |
row_text.append(cell.text.strip())
|
55 |
|
56 |
+
text.append(" | ".join(row_text))
|
57 |
|
58 |
return "\n".join(text)
|
59 |
|