Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
|
@@ -5,15 +5,22 @@ from utils import extract_text_from_pdf, parse_pdf_to_dict
|
|
| 5 |
|
| 6 |
from docx.shared import RGBColor
|
| 7 |
|
| 8 |
-
|
| 9 |
-
|
|
|
|
|
|
|
|
|
|
| 10 |
|
| 11 |
for para in doc.paragraphs:
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 17 |
|
| 18 |
return doc
|
| 19 |
|
|
|
|
| 5 |
|
| 6 |
from docx.shared import RGBColor
|
| 7 |
|
| 8 |
+
from docx import Document
|
| 9 |
+
from docx.shared import RGBColor
|
| 10 |
+
|
| 11 |
+
def replace_red_text_with_data(word_path, data_dict):
|
| 12 |
+
doc = Document(word_path)
|
| 13 |
|
| 14 |
for para in doc.paragraphs:
|
| 15 |
+
full_text = para.text
|
| 16 |
+
for i, run in enumerate(para.runs):
|
| 17 |
+
if run.font.color and run.font.color.rgb == RGBColor(255, 0, 0):
|
| 18 |
+
# Search for the key (label) before this red text
|
| 19 |
+
preceding_text = ''.join(r.text for r in para.runs[:i]).lower()
|
| 20 |
+
for key in data_dict:
|
| 21 |
+
if key.lower() in preceding_text:
|
| 22 |
+
run.text = data_dict[key]
|
| 23 |
+
break
|
| 24 |
|
| 25 |
return doc
|
| 26 |
|