Create diff_utils.py

diff_utils.py

@@ -0,0 +1,95 @@
+"""Diff visualization utilities."""
+
+import difflib
+from typing import Iterator
+
+
+def generate_html_diff(original: str, cleaned: str) -> str:
+    """Generate HTML diff between original and cleaned text."""
+    differ = difflib.HtmlDiff(tabsize=2, wrapcolumn=80)
+    
+    original_lines = original.splitlines(keepends=True)
+    cleaned_lines = cleaned.splitlines(keepends=True)
+    
+    html_diff = differ.make_file(
+        original_lines,
+        cleaned_lines,
+        "Original Text",
+        "Cleaned Text",
+        context=True,
+        numlines=3
+    )
+    
+    return _style_diff_html(html_diff)
+
+
+def _style_diff_html(html_diff: str) -> str:
+    """Add custom styling to the diff HTML."""
+    custom_style = """
+    <style>
+        .diff {
+            font-family: 'Monaco', 'Menlo', 'Ubuntu Mono', monospace;
+            font-size: 12px;
+            line-height: 1.4;
+            border: 1px solid #ddd;
+            border-radius: 8px;
+            overflow: auto;
+            max-height: 600px;
+        }
+        .diff table {
+            width: 100%;
+            border-collapse: collapse;
+            margin: 0;
+        }
+        .diff td {
+            padding: 2px 8px;
+            vertical-align: top;
+            white-space: pre-wrap;
+            word-wrap: break-word;
+        }
+        .diff_header {
+            background: #f8f9fa;
+            font-weight: bold;
+            text-align: center;
+            border-bottom: 1px solid #dee2e6;
+        }
+        .diff_next {
+            background: #e9ecef;
+            text-align: center;
+            font-size: 10px;
+        }
+        .diff_add {
+            background: #d4edda;
+            color: #155724;
+        }
+        .diff_chg {
+            background: #fff3cd;
+            color: #856404;
+        }
+        .diff_sub {
+            background: #f8d7da;
+            color: #721c24;
+        }
+    </style>
+    """
+    
+    # Insert custom style before closing </head> tag
+    styled_html = html_diff.replace('</head>', f'{custom_style}</head>')
+    
+    # Add diff class to the table
+    styled_html = styled_html.replace('<table class="diff"', '<div class="diff"><table class="diff"')
+    styled_html = styled_html.replace('</table>', '</table></div>')
+    
+    return styled_html
+
+
+def get_diff_stats(original: str, cleaned: str) -> dict[str, int]:
+    """Get statistics about the diff between original and cleaned text."""
+    differ = difflib.SequenceMatcher(None, original, cleaned)
+    
+    return {
+        'original_length': len(original),
+        'cleaned_length': len(cleaned),
+        'similarity_ratio': round(differ.ratio() * 100, 2),
+        'characters_removed': len(original) - len(cleaned)
+    }