refactor: Implement smart line matching for scroll position preservation

gtm

@@ -18,6 +18,44 @@ def get_file_at_commit(commit_hash, filename):
     result = subprocess.run(cmd, capture_output=True, text=True)
     return result.stdout.splitlines()
 
+def find_best_matching_line(reference_line, file_lines, max_lines=None):
+    """Find the best matching line in file_lines that matches reference_line.
+    Returns the line index or None if no good match is found."""
+    if not reference_line or not file_lines:
+        return None
+        
+    # First try exact match
+    for i, line in enumerate(file_lines):
+        if line == reference_line:
+            return i
+            
+    # If no exact match, try to find the most similar line
+    # Only search through a reasonable number of lines for performance
+    search_lines = file_lines[:max_lines] if max_lines else file_lines
+    
+    best_match = None
+    best_score = 0
+    
+    for i, line in enumerate(search_lines):
+        # Simple similarity score: count of common characters
+        score = sum(1 for a, b in zip(reference_line, line) if a == b)
+        
+        # Adjust score based on length difference
+        length_diff = abs(len(reference_line) - len(line))
+        adjusted_score = score - (length_diff * 0.5)
+        
+        if adjusted_score > best_score:
+            best_score = adjusted_score
+            best_match = i
+            
+    # Only return a match if it's reasonably good
+    # (at least 60% of the shorter string length)
+    min_length = min(len(reference_line), 1)  # Avoid division by zero
+    if best_score > (min_length * 0.6):
+        return best_match
+        
+    return None
+
 def get_diff_info(current_commit, prev_commit, filename):
     """Get diff information between two commits for a file"""
     if not prev_commit:
@@ -129,10 +167,15 @@ def main(stdscr, filename, show_whole_diff=False, show_additions=False, show_del
 
         # Calculate current scroll position as percentage before changing commits
         scroll_percentage = 0
+        reference_line = None
         if len(file_lines) > 0:
             max_scroll = max(0, len(file_lines) - (height - 1))
             if max_scroll > 0:
                 scroll_percentage = scroll_offset / max_scroll
+            
+            # Store the content of the top visible line as reference
+            if scroll_offset < len(file_lines):
+                reference_line = file_lines[scroll_offset]
         
         # Only fetch file content when commit changes
         if (key in [curses.KEY_DOWN, curses.KEY_UP, ord('j'), ord('k'), 
@@ -148,12 +191,26 @@ def main(stdscr, filename, show_whole_diff=False, show_additions=False, show_del
             if show_whole_diff or show_additions or show_deletions:
                 added_lines, deleted_lines = get_diff_info(commit_hash, prev_commit_hash, filename)
             
-            # Apply the saved scroll percentage to the new file
-            max_scroll = max(0, len(file_lines) - (height - 1))
-            if max_scroll > 0:
-                scroll_offset = int(scroll_percentage * max_scroll)
+            # Try to find the same line in the new file version
+            if reference_line:
+                # Limit search to first 1000 lines for performance
+                matching_line_idx = find_best_matching_line(reference_line, file_lines, 1000)
+                if matching_line_idx is not None:
+                    scroll_offset = matching_line_idx
+                else:
+                    # Fall back to percentage-based scrolling
+                    max_scroll = max(0, len(file_lines) - (height - 1))
+                    if max_scroll > 0:
+                        scroll_offset = int(scroll_percentage * max_scroll)
+                    else:
+                        scroll_offset = 0
             else:
-                scroll_offset = 0
+                # Fall back to percentage-based scrolling
+                max_scroll = max(0, len(file_lines) - (height - 1))
+                if max_scroll > 0:
+                    scroll_offset = int(scroll_percentage * max_scroll)
+                else:
+                    scroll_offset = 0
         
         # Recalculate max_scroll and ensure scroll_offset is within bounds
         max_scroll = max(0, len(file_lines) - (height - 1))