Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -15,7 +15,7 @@ PRIOR_MESSAGE_MARKERS = [
|
|
| 15 |
re.compile(r'^Begin forwarded message:', re.IGNORECASE),
|
| 16 |
|
| 17 |
# Custom separators in email
|
| 18 |
-
re.compile(r'^-+.*-+$'),
|
| 19 |
|
| 20 |
# Portuguese patterns
|
| 21 |
re.compile(r'^Em .* escreveu:', re.IGNORECASE),
|
|
@@ -72,7 +72,7 @@ def remove_quoted_text(soup):
|
|
| 72 |
|
| 73 |
# Remove tables with dotted borders (a typical marker of a previous conversation)
|
| 74 |
for table in soup.find_all('table'):
|
| 75 |
-
if 'border-top:1px dotted' in
|
| 76 |
table.decompose()
|
| 77 |
|
| 78 |
return soup
|
|
@@ -83,6 +83,9 @@ def extract_latest_message_from_lines(lines):
|
|
| 83 |
for line in lines:
|
| 84 |
# Clean up the line
|
| 85 |
line = line.strip()
|
|
|
|
|
|
|
|
|
|
| 86 |
# Check if the line matches any prior message markers
|
| 87 |
if any(marker.match(line) for marker in PRIOR_MESSAGE_MARKERS):
|
| 88 |
break # Stop if a prior message marker is found
|
|
|
|
| 15 |
re.compile(r'^Begin forwarded message:', re.IGNORECASE),
|
| 16 |
|
| 17 |
# Custom separators in email
|
| 18 |
+
re.compile(r'^-+.*-+$'), # For lines like "--------------------------------------------------"
|
| 19 |
|
| 20 |
# Portuguese patterns
|
| 21 |
re.compile(r'^Em .* escreveu:', re.IGNORECASE),
|
|
|
|
| 72 |
|
| 73 |
# Remove tables with dotted borders (a typical marker of a previous conversation)
|
| 74 |
for table in soup.find_all('table'):
|
| 75 |
+
if table.get('style') and 'border-top:1px dotted' in table.get('style'):
|
| 76 |
table.decompose()
|
| 77 |
|
| 78 |
return soup
|
|
|
|
| 83 |
for line in lines:
|
| 84 |
# Clean up the line
|
| 85 |
line = line.strip()
|
| 86 |
+
# Ensure that we're handling None values safely
|
| 87 |
+
if line is None or line == '':
|
| 88 |
+
continue
|
| 89 |
# Check if the line matches any prior message markers
|
| 90 |
if any(marker.match(line) for marker in PRIOR_MESSAGE_MARKERS):
|
| 91 |
break # Stop if a prior message marker is found
|