Feature：添加代码行评审建议，通过MERGE_DETAIL_REVIEW_ENABLED开启

biz/github/webhook_handler.py

@@ -43,10 +43,7 @@ def filter_changes(changes: list):
     # 过滤 `new_path` 以支持的扩展名结尾的元素, 仅保留diff和new_path字段
     filtered_changes = [
         {
-            'diff': item.get('diff', ''),
-            'new_path': item['new_path'],
-            'additions': item.get('additions', 0),
-            'deletions': item.get('deletions', 0),
+            **item
         }
         for item in not_deleted_changes
         if any(item.get('new_path', '').endswith(ext) for ext in supported_extensions)
@@ -105,9 +102,11 @@ def get_pull_request_changes(self) -> list:
                     changes = []
                     for file in files:
                         change = {
-                            'old_path': file.get('filename'),
+                            'old_path': file.get('previous_filename', ''),
                             'new_path': file.get('filename'),
                             'diff': file.get('patch', ''),
+                            'status': file.get('status', ''),
+                            'renamed_file': file.get('status') == 'renamed',
                             'additions': file.get('additions', 0),
                             'deletions': file.get('deletions', 0)
                         }
@@ -192,6 +191,57 @@ def target_branch_protected(self) -> bool:
             logger.warn(f"Failed to get protected branches: {response.status_code}, {response.text}")
             return False
 
+    def add_pull_request_comment(self, review):
+        """向 GitHub Pull Request 的特定行添加评论"""
+        head_sha = self.webhook_data.get("pull_request", {}).get("head", {}).get("sha")
+        if not head_sha:
+            logger.error("无法添加评论，缺少 head_sha。")
+            return False
+
+        url = f"https://api.github.com/repos/{self.repo_full_name}/pulls/{self.pull_request_number}/comments"
+        headers = {
+            "Authorization": f"token {self.github_token}",
+            "Accept": "application/vnd.github.v3+json",
+            "Content-Type": "application/json"
+        }
+
+        body = f"""**AI Review [{review.get('severity', 'N/A').upper()}]**: {review.get('category', 'General')}
+
+**分析**: {review.get('analysis', 'N/A')}
+
+**建议**:
+```suggestion
+{review.get('suggestion', 'N/A')}
+```
+"""
+
+        lines_info = review.get("lines", {})
+        file_path = review.get("file")
+
+        if not file_path:
+            logger.warning("跳过评论，审查缺少 'file' 路径。")
+            return False
+        if not lines_info or not lines_info.get('new'):
+            logger.warning("跳过评论，审查缺少 'lines' 信息。")
+            return False
+
+        payload = {
+            "body": body,
+            "commit_id": head_sha,
+            "path": file_path,
+            "line": lines_info["new"]
+        }
+
+        target_desc = f"file {file_path} line {lines_info['new']}"
+        try:
+            response = requests.post(url, headers=headers, json=payload)
+            response.raise_for_status()
+            logger.info(f"成功向 GitHub PR #{self.pull_request_number} ({target_desc}) 添加评论")
+            return True
+        except Exception as e:
+            logger.exception(f"添加 GitHub 评论 ({target_desc}) 时发生意外错误: {e}")
+            return False
+
 
 class PushHandler:
     def __init__(self, webhook_data: dict, github_token: str, github_url: str):

biz/gitlab/webhook_handler.py

@@ -15,13 +15,13 @@ def filter_changes(changes: list):
     # 从环境变量中获取支持的文件扩展名
     supported_extensions = os.getenv('SUPPORTED_EXTENSIONS', '.java,.py,.php').split(',')
 
+    # 过滤删除的文件
     filter_deleted_files_changes = [change for change in changes if not change.get("deleted_file")]
 
-    # 过滤 `new_path` 以支持的扩展名结尾的元素, 仅保留diff和new_path字段
+    # 过滤 `new_path` 以支持的扩展名结尾的元素
     filtered_changes = [
         {
-            'diff': item.get('diff', ''),
-            'new_path': item['new_path'],
+            **item,
             'additions': len(re.findall(r'^\+(?!\+\+)', item.get('diff', ''), re.MULTILINE)),
             'deletions': len(re.findall(r'^-(?!--)', item.get('diff', ''), re.MULTILINE))
         }
@@ -73,6 +73,21 @@ def parse_merge_request_event(self):
         self.project_id = merge_request.get('target_project_id')
         self.action = merge_request.get('action')
 
+    def get_merge_request(self) -> dict:
+        # 调用 GitLab API 获取 Merge Request 的 changes
+        url = urljoin(f"{self.gitlab_url}/",
+                      f"api/v4/projects/{self.project_id}/merge_requests/{self.merge_request_iid}/changes")
+        headers = {
+            'Private-Token': self.gitlab_token
+        }
+        response = requests.get(url, headers=headers, verify=False)
+        # 检查请求是否成功
+        if response.status_code == 200:
+            return response.json()
+        else:
+            logger.warn(f"Failed to get changes from GitLab (URL: {url}): {response.status_code}, {response.text}")
+            return {}
+
     def get_merge_request_changes(self) -> list:
         # 检查是否为 Merge Request Hook 事件
         if self.event_type != 'merge_request':
@@ -83,28 +98,16 @@ def get_merge_request_changes(self) -> list:
         max_retries = 3  # 最大重试次数
         retry_delay = 10  # 重试间隔时间（秒）
         for attempt in range(max_retries):
-            # 调用 GitLab API 获取 Merge Request 的 changes
-            url = urljoin(f"{self.gitlab_url}/",
-                          f"api/v4/projects/{self.project_id}/merge_requests/{self.merge_request_iid}/changes")
-            headers = {
-                'Private-Token': self.gitlab_token
-            }
-            response = requests.get(url, headers=headers, verify=False)
+            data = self.get_merge_request()
             logger.debug(
-                f"Get changes response from GitLab (attempt {attempt + 1}): {response.status_code}, {response.text}, URL: {url}")
-
-            # 检查请求是否成功
-            if response.status_code == 200:
-                changes = response.json().get('changes', [])
-                if changes:
-                    return changes
-                else:
-                    logger.info(
-                        f"Changes is empty, retrying in {retry_delay} seconds... (attempt {attempt + 1}/{max_retries}), URL: {url}")
-                    time.sleep(retry_delay)
+                f"Get changes response from GitLab (attempt {attempt + 1}): {data}")
+            changes = data.get('changes', [])
+            if changes:
+                return changes
             else:
-                logger.warn(f"Failed to get changes from GitLab (URL: {url}): {response.status_code}, {response.text}")
-                return []
+                logger.info(
+                    f"Changes is empty, retrying in {retry_delay} seconds... (attempt {attempt + 1}/{max_retries})")
+                time.sleep(retry_delay)
 
         logger.warning(f"Max retries ({max_retries}) reached. Changes is still empty.")
         return []  # 达到最大重试次数后返回空列表
@@ -165,6 +168,66 @@ def target_branch_protected(self) -> bool:
             logger.warn(f"Failed to get protected branches: {response.status_code}, {response.text}")
             return False
 
+    def add_merge_request_comment(self, review, position_info):
+        """向 GitLab Merge Request 的特定行添加评论"""
+        if not position_info or not position_info.get("head_sha") or not position_info.get(
+                "base_sha") or not position_info.get("start_sha"):
+            logger.error(
+                f"错误: 无法添加评论，缺少必要的位置信息 (head_sha/base_sha/start_sha)。得到: {position_info}")
+            return False
+
+        url = f"{self.gitlab_url}/api/v4/projects/{self.project_id}/merge_requests/{self.merge_request_iid}/discussions"
+        headers = {"PRIVATE-TOKEN": self.gitlab_token, "Content-Type": "application/json"}
+
+        body = f"""**AI Review [{review.get('severity', 'N/A').upper()}]**: {review.get('category', 'General')}
+
+**分析**: {review.get('analysis', 'N/A')}
+
+**建议**:
+```suggestion
+{review.get('suggestion', 'N/A')}
+```
+"""
+        position_data = {
+            "base_sha": position_info.get("base_sha"),
+            "start_sha": position_info.get("start_sha"),
+            "head_sha": position_info.get("head_sha"),
+            "position_type": "text",
+        }
+
+        lines_info = review.get("lines", {})
+        file_path = review.get("file")
+        old_file_path = review.get("old_path")
+
+        if not file_path:
+            logger.warning("跳过评论，审查缺少 'file' 路径。")
+            return False
+
+        if lines_info and lines_info.get("new") is not None:
+            position_data["new_path"] = file_path
+            position_data["new_line"] = lines_info["new"]
+            position_data["old_path"] = old_file_path if old_file_path else file_path
+            target_desc = f"file {file_path} line {lines_info['new']}"
+        elif lines_info and lines_info.get("old") is not None:
+            position_data["old_path"] = old_file_path if old_file_path else file_path
+            position_data["old_line"] = lines_info["old"]
+            position_data["new_path"] = file_path
+            target_desc = f"文件 {position_data['old_path']} 旧行号 {lines_info['old']}"
+        else:
+            logger.warning("跳过评论，审查缺少 'lines' 信息。")
+            return False
+
+        payload = {"body": body, "position": position_data}
+        logger.info(f"尝试向 {target_desc} 添加带位置的评论")
+        try:
+            response_obj = requests.post(url, headers=headers, json=payload)
+            response_obj.raise_for_status()
+            logger.info(f"成功向 GitLab MR {self.merge_request_iid} ({target_desc}) 添加评论")
+            return True
+        except Exception as e:
+            logger.exception(f"添加 GitLab 评论 ({target_desc}) 时发生意外错误: {e}")
+            return False
+
 
 class PushHandler:
     def __init__(self, webhook_data: dict, gitlab_token: str, gitlab_url: str):

biz/queue/worker.py

@@ -11,7 +11,6 @@
 from biz.utils.log import logger
 
 
-
 def handle_push_event(webhook_data: dict, gitlab_token: str, gitlab_url: str, gitlab_url_slug: str):
     push_review_enabled = os.environ.get('PUSH_REVIEW_ENABLED', '0') == '1'
     try:
@@ -75,6 +74,7 @@ def handle_merge_request_event(webhook_data: dict, gitlab_token: str, gitlab_url
     :return:
     '''
     merge_review_only_protected_branches = os.environ.get('MERGE_REVIEW_ONLY_PROTECTED_BRANCHES_ENABLED', '0') == '1'
+    merge_detail_review = os.environ.get('MERGE_DETAIL_REVIEW_ENABLED', '0') == '1'
     try:
         # 解析Webhook数据
         handler = MergeRequestHandler(webhook_data, gitlab_token, gitlab_url)
@@ -111,8 +111,8 @@ def handle_merge_request_event(webhook_data: dict, gitlab_token: str, gitlab_url
 
         # review 代码
         commits_text = ';'.join(commit['title'] for commit in commits)
-        review_result = CodeReviewer().review_and_strip_code(str(changes), commits_text)
-
+        reviewer = CodeReviewer()
+        review_result = reviewer.review_and_strip_code(str(changes), commits_text)
         # 将review结果提交到Gitlab的 notes
         handler.add_merge_request_notes(f'Auto Review Result: \n{review_result}')
 
@@ -135,11 +135,26 @@ def handle_merge_request_event(webhook_data: dict, gitlab_token: str, gitlab_url
             )
         )
 
+        if merge_detail_review:
+            # 如果开启Merge请求详细Review，对每个变更的文件进行review
+            all_detail_review = reviewer.detail_review(changes)
+            merge_request = handler.get_merge_request()
+            success_count = 0
+            fail_count = 0
+            for review in all_detail_review:
+                if handler.add_merge_request_comment(review=review, position_info=merge_request.get("diff_refs")):
+                    success_count += 1
+                else:
+                    fail_count += 1
+            logger.info(f'Gitlab merge request detail review count: {len(all_detail_review)}, '
+                        f'Success count: {success_count}, Fail count: {fail_count}')
+
     except Exception as e:
         error_message = f'AI Code Review 服务出现未知错误: {str(e)}\n{traceback.format_exc()}'
         notifier.send_notification(content=error_message)
         logger.error('出现未知错误: %s', error_message)
 
+
 def handle_github_push_event(webhook_data: dict, github_token: str, github_url: str, github_url_slug: str):
     push_review_enabled = os.environ.get('PUSH_REVIEW_ENABLED', '0') == '1'
     try:
@@ -203,6 +218,7 @@ def handle_github_pull_request_event(webhook_data: dict, github_token: str, gith
     :return:
     '''
     merge_review_only_protected_branches = os.environ.get('MERGE_REVIEW_ONLY_PROTECTED_BRANCHES_ENABLED', '0') == '1'
+    merge_detail_review = os.environ.get('MERGE_DETAIL_REVIEW_ENABLED', '0') == '1'
     try:
         # 解析Webhook数据
         handler = GithubPullRequestHandler(webhook_data, github_token, github_url)
@@ -239,7 +255,8 @@ def handle_github_pull_request_event(webhook_data: dict, github_token: str, gith
 
         # review 代码
         commits_text = ';'.join(commit['title'] for commit in commits)
-        review_result = CodeReviewer().review_and_strip_code(str(changes), commits_text)
+        reviewer = CodeReviewer()
+        review_result = reviewer.review_and_strip_code(str(changes), commits_text)
 
         # 将review结果提交到GitHub的 notes
         handler.add_pull_request_notes(f'Auto Review Result: \n{review_result}')
@@ -262,6 +279,19 @@ def handle_github_pull_request_event(webhook_data: dict, github_token: str, gith
                 deletions=deletions,
             ))
 
+        if merge_detail_review:
+            # 如果开启Merge请求详细Review，对每个变更的文件进行review
+            all_detail_review = reviewer.detail_review(changes)
+            success_count = 0
+            fail_count = 0
+            for review in all_detail_review:
+                if handler.add_pull_request_comment(review=review):
+                    success_count += 1
+                else:
+                    fail_count += 1
+            logger.info(f'Github merge request detail review count: {len(all_detail_review)}, '
+                        f'Success count: {success_count}, Fail count: {fail_count}')
+
     except Exception as e:
         error_message = f'服务出现未知错误: {str(e)}\n{traceback.format_exc()}'
         notifier.send_notification(content=error_message)

biz/utils/code_parser.py

@@ -44,3 +44,75 @@ def get_new_code(self):
         if self.new_code is None:
             self.parse_diff()
         return self.new_code
+
+
+def parse_single_file_diff(diff_text, file_path, old_file_path=None):
+    """
+        解析单个文件的 unified diff 格式文本，提取变更信息。
+        返回包含该文件变更详情和上下文的字典。
+        """
+    file_changes = {
+        "path": file_path,
+        "old_path": old_file_path,
+        "changes": [],
+        "context": {"old": [], "new": []},
+        "lines_changed": 0
+    }
+
+    old_line_num_current = 0
+    new_line_num_current = 0
+    hunk_context_lines = []
+
+    lines = diff_text.splitlines()
+    i = 0
+    while i < len(lines):
+        line = lines[i]
+        if line.startswith('--- ') or line.startswith('+++ '):
+            i += 1
+            continue
+        elif line.startswith('@@ '):
+            match = re.match(r'@@ -(\d+)(?:,(\d+))? \+(\d+)(?:,(\d+))? @@', line)
+            if match:
+                old_line_num_start = int(match.group(1))
+                new_line_num_start = int(match.group(3))
+                old_line_num_current = old_line_num_start
+                new_line_num_current = new_line_num_start
+                if hunk_context_lines:  # 将上一个 hunk 的上下文添加到 file_changes
+                    file_changes["context"]["old"].extend(hunk_context_lines)
+                    file_changes["context"]["new"].extend(hunk_context_lines)
+                    hunk_context_lines = []  # 为新的 hunk 重置
+            else:
+                old_line_num_current = 0
+                new_line_num_current = 0
+        elif line.startswith('+'):
+            file_changes["changes"].append({
+                "type": "add",
+                "old_line": None,
+                "new_line": new_line_num_current,
+                "content": line[1:]
+            })
+            new_line_num_current += 1
+        elif line.startswith('-'):
+            file_changes["changes"].append({
+                "type": "delete",
+                "old_line": old_line_num_current,
+                "new_line": None,
+                "content": line[1:]
+            })
+            old_line_num_current += 1
+        elif line.startswith(' '):  # Context line
+            hunk_context_lines.append(f"{old_line_num_current} -> {new_line_num_current}: {line[1:]}")
+            old_line_num_current += 1
+            new_line_num_current += 1
+        i += 1
+
+    if hunk_context_lines:  # 添加最后一个 hunk 的上下文
+        file_changes["context"]["old"].extend(hunk_context_lines)
+        file_changes["context"]["new"].extend(hunk_context_lines)
+
+    limit = 20  # 限制上下文行数
+    file_changes["context"]["old"] = "\n".join(file_changes["context"]["old"][-limit:])
+    file_changes["context"]["new"] = "\n".join(file_changes["context"]["new"][-limit:])
+    file_changes["lines_changed"] = len([c for c in file_changes["changes"] if c['type'] in ['add', 'delete']])
+
+    return file_changes