gac 3.6.0__py3-none-any.whl → 3.10.10__py3-none-any.whl

gac/postprocess.py

@@ -0,0 +1,155 @@
+"""Post-processing for AI-generated commit messages.
+
+This module handles cleaning and normalization of commit messages generated
+by AI models, removing artifacts like think tags, code blocks, and XML tags.
+"""
+
+import re
+
+from gac.constants import CommitMessageConstants
+
+
+def _remove_think_tags(message: str) -> str:
+    """Remove AI reasoning <think> tags and their content from the message.
+
+    Args:
+        message: The message to clean
+
+    Returns:
+        Message with <think> tags removed
+    """
+    while re.search(r"<think>(?:(?!</think>)[^\n])*\n.*?</think>", message, flags=re.DOTALL | re.IGNORECASE):
+        message = re.sub(
+            r"<think>(?:(?!</think>)[^\n])*\n.*?</think>\s*", "", message, flags=re.DOTALL | re.IGNORECASE, count=1
+        )
+
+    message = re.sub(r"\n\n+\s*<think>.*?</think>\s*", "", message, flags=re.DOTALL | re.IGNORECASE)
+    message = re.sub(r"<think>.*?</think>\s*\n\n+", "", message, flags=re.DOTALL | re.IGNORECASE)
+
+    message = re.sub(r"<think>\s*\n.*$", "", message, flags=re.DOTALL | re.IGNORECASE)
+
+    conventional_prefixes_pattern = r"(" + "|".join(CommitMessageConstants.CONVENTIONAL_PREFIXES) + r")[\(:)]"
+    if re.search(r"^.*?</think>", message, flags=re.DOTALL | re.IGNORECASE):
+        prefix_match = re.search(conventional_prefixes_pattern, message, flags=re.IGNORECASE)
+        think_match = re.search(r"</think>", message, flags=re.IGNORECASE)
+
+        if not prefix_match or (think_match and think_match.start() < prefix_match.start()):
+            message = re.sub(r"^.*?</think>\s*", "", message, flags=re.DOTALL | re.IGNORECASE)
+
+    message = re.sub(r"</think>\s*$", "", message, flags=re.IGNORECASE)
+
+    return message
+
+
+def _remove_code_blocks(message: str) -> str:
+    """Remove markdown code blocks from the message.
+
+    Args:
+        message: The message to clean
+
+    Returns:
+        Message with code blocks removed
+    """
+    return re.sub(r"```[\w]*\n|```", "", message)
+
+
+def _extract_commit_from_reasoning(message: str) -> str:
+    """Extract the actual commit message from reasoning/preamble text.
+
+    Args:
+        message: The message potentially containing reasoning
+
+    Returns:
+        Extracted commit message
+    """
+    for indicator in CommitMessageConstants.COMMIT_INDICATORS:
+        if indicator.lower() in message.lower():
+            message = message.split(indicator, 1)[1].strip()
+            break
+
+    lines = message.split("\n")
+    for i, line in enumerate(lines):
+        if any(line.strip().startswith(f"{prefix}:") for prefix in CommitMessageConstants.CONVENTIONAL_PREFIXES):
+            message = "\n".join(lines[i:])
+            break
+
+    return message
+
+
+def _remove_xml_tags(message: str) -> str:
+    """Remove XML tags that might have leaked into the message.
+
+    Args:
+        message: The message to clean
+
+    Returns:
+        Message with XML tags removed
+    """
+    for tag in CommitMessageConstants.XML_TAGS_TO_REMOVE:
+        message = message.replace(tag, "")
+    return message
+
+
+def _fix_double_prefix(message: str) -> str:
+    """Fix double type prefix issues like 'chore: feat(scope):' to 'feat(scope):'.
+
+    Args:
+        message: The message to fix
+
+    Returns:
+        Message with double prefix corrected
+    """
+    double_prefix_pattern = re.compile(
+        r"^("
+        + r"|\s*".join(CommitMessageConstants.CONVENTIONAL_PREFIXES)
+        + r"):\s*("
+        + r"|\s*".join(CommitMessageConstants.CONVENTIONAL_PREFIXES)
+        + r")\(([^)]+)\):"
+    )
+    match = double_prefix_pattern.match(message)
+
+    if match:
+        second_type = match.group(2)
+        scope = match.group(3)
+        description = message[match.end() :].strip()
+        message = f"{second_type}({scope}): {description}"
+
+    return message
+
+
+def _normalize_whitespace(message: str) -> str:
+    """Normalize whitespace, ensuring no more than one blank line between paragraphs.
+
+    Args:
+        message: The message to normalize
+
+    Returns:
+        Message with normalized whitespace
+    """
+    return re.sub(r"\n(?:[ \t]*\n){2,}", "\n\n", message).strip()
+
+
+def clean_commit_message(message: str) -> str:
+    """Clean up a commit message generated by an AI model.
+
+    This function:
+    1. Removes any preamble or reasoning text
+    2. Removes code block markers and formatting
+    3. Removes XML tags that might have leaked into the response
+    4. Fixes double type prefix issues (e.g., "chore: feat(scope):")
+    5. Normalizes whitespace
+
+    Args:
+        message: Raw commit message from AI
+
+    Returns:
+        Cleaned commit message ready for use
+    """
+    message = message.strip()
+    message = _remove_think_tags(message)
+    message = _remove_code_blocks(message)
+    message = _extract_commit_from_reasoning(message)
+    message = _remove_xml_tags(message)
+    message = _fix_double_prefix(message)
+    message = _normalize_whitespace(message)
+    return message