Merge pull request #1420 from Codium-ai/tr/review_fix

fix: improve line extraction from files with missing content

pr_agent/algo/git_patch_processing.py

@@ -364,48 +364,51 @@ __old hunk__
 
 
 def extract_hunk_lines_from_patch(patch: str, file_name, line_start, line_end, side) -> tuple[str, str]:
+    try:
+        patch_with_lines_str = f"\n\n## File: '{file_name.strip()}'\n\n"
+        selected_lines = ""
+        patch_lines = patch.splitlines()
+        RE_HUNK_HEADER = re.compile(
+            r"^@@ -(\d+)(?:,(\d+))? \+(\d+)(?:,(\d+))? @@[ ]?(.*)")
+        match = None
+        start1, size1, start2, size2 = -1, -1, -1, -1
+        skip_hunk = False
+        selected_lines_num = 0
+        for line in patch_lines:
+            if 'no newline at end of file' in line.lower():
+                continue
 
-    patch_with_lines_str = f"\n\n## File: '{file_name.strip()}'\n\n"
+            if line.startswith('@@'):
-    selected_lines = ""
+                skip_hunk = False
-    patch_lines = patch.splitlines()
+                selected_lines_num = 0
-    RE_HUNK_HEADER = re.compile(
+                header_line = line
-        r"^@@ -(\d+)(?:,(\d+))? \+(\d+)(?:,(\d+))? @@[ ]?(.*)")
-    match = None
-    start1, size1, start2, size2 = -1, -1, -1, -1
-    skip_hunk = False
-    selected_lines_num = 0
-    for line in patch_lines:
-        if 'no newline at end of file' in line.lower():
-            continue
 
-        if line.startswith('@@'):
+                match = RE_HUNK_HEADER.match(line)
-            skip_hunk = False
-            selected_lines_num = 0
-            header_line = line
 
-            match = RE_HUNK_HEADER.match(line)
+                section_header, size1, size2, start1, start2 = extract_hunk_headers(match)
 
-            section_header, size1, size2, start1, start2 = extract_hunk_headers(match)
-
-            # check if line range is in this hunk
-            if side.lower() == 'left':
                 # check if line range is in this hunk
-                if not (start1 <= line_start <= start1 + size1):
+                if side.lower() == 'left':
-                    skip_hunk = True
+                    # check if line range is in this hunk
-                    continue
+                    if not (start1 <= line_start <= start1 + size1):
-            elif side.lower() == 'right':
+                        skip_hunk = True
-                if not (start2 <= line_start <= start2 + size2):
+                        continue
-                    skip_hunk = True
+                elif side.lower() == 'right':
-                    continue
+                    if not (start2 <= line_start <= start2 + size2):
-            patch_with_lines_str += f'\n{header_line}\n'
+                        skip_hunk = True
+                        continue
+                patch_with_lines_str += f'\n{header_line}\n'
 
-        elif not skip_hunk:
+            elif not skip_hunk:
-            if side.lower() == 'right' and line_start <= start2 + selected_lines_num <= line_end:
+                if side.lower() == 'right' and line_start <= start2 + selected_lines_num <= line_end:
-                selected_lines += line + '\n'
+                    selected_lines += line + '\n'
-            if side.lower() == 'left' and start1 <= selected_lines_num + start1 <= line_end:
+                if side.lower() == 'left' and start1 <= selected_lines_num + start1 <= line_end:
-                selected_lines += line + '\n'
+                    selected_lines += line + '\n'
-            patch_with_lines_str += line + '\n'
+                patch_with_lines_str += line + '\n'
-            if not line.startswith('-'): # currently we don't support /ask line for deleted lines
+                if not line.startswith('-'): # currently we don't support /ask line for deleted lines
-                selected_lines_num += 1
+                    selected_lines_num += 1
+    except Exception as e:
+        get_logger().error(f"Failed to extract hunk lines from patch: {e}", artifact={"traceback": traceback.format_exc()})
+        return "", ""
 
     return patch_with_lines_str.rstrip(), selected_lines.rstrip()

pr_agent/algo/utils.py

@@ -23,6 +23,7 @@ from pydantic import BaseModel
 from starlette_context import context
 
 from pr_agent.algo import MAX_TOKENS
+from pr_agent.algo.git_patch_processing import extract_hunk_lines_from_patch
 from pr_agent.algo.token_handler import TokenEncoder
 from pr_agent.algo.types import FilePatchInfo
 from pr_agent.config_loader import get_settings, global_settings
@@ -272,7 +273,11 @@ def convert_to_markdown_v2(output_data: dict,
 
     return markdown_text
 
-def extract_relevant_lines_str(end_line, files, relevant_file, start_line, dedent=False):
+
+def extract_relevant_lines_str(end_line, files, relevant_file, start_line, dedent=False) -> str:
+    """
+    Finds 'relevant_file' in 'files', and extracts the lines from 'start_line' to 'end_line' string from the file content.
+    """
     try:
         relevant_lines_str = ""
         if files:
@@ -280,10 +285,23 @@ def extract_relevant_lines_str(end_line, files, relevant_file, start_line, deden
             for file in files:
                 if file.filename.strip() == relevant_file:
                     if not file.head_file:
-                        get_logger().warning(f"No content found in file: {file.filename}")
+                        # as a fallback, extract relevant lines directly from patch
-                        return ""
+                        patch = file.patch
-                    relevant_file_lines = file.head_file.splitlines()
+                        get_logger().info(f"No content found in file: '{file.filename}' for 'extract_relevant_lines_str'. Using patch instead")
-                    relevant_lines_str = "\n".join(relevant_file_lines[start_line - 1:end_line])
+                        _, selected_lines = extract_hunk_lines_from_patch(patch, file.filename, start_line, end_line,side='right')
+                        if not selected_lines:
+                            get_logger().error(f"Failed to extract relevant lines from patch: {file.filename}")
+                            return ""
+                        # filter out '-' lines
+                        relevant_lines_str = ""
+                        for line in selected_lines.splitlines():
+                            if line.startswith('-'):
+                                continue
+                            relevant_lines_str += line[1:] + '\n'
+                    else:
+                        relevant_file_lines = file.head_file.splitlines()
+                        relevant_lines_str = "\n".join(relevant_file_lines[start_line - 1:end_line])
+
                     if dedent and relevant_lines_str:
                         # Remove the longest leading string of spaces and tabs common to all lines.
                         relevant_lines_str = textwrap.dedent(relevant_lines_str)