add logs

apps/common/handle/impl/media/media_split_handle.py

@@ -2,6 +2,7 @@
 """
 音视频处理器 - MaxKB集成层
 """
+import traceback
 from typing import List
 from common.handle.base_split_handle import BaseSplitHandle
 from common.utils.logger import maxkb_logger
@@ -335,18 +336,30 @@ class MediaSplitHandle(BaseSplitHandle):
 
     def smart_split_transcription(self, text, limit=1000, overlap=100):
         """智能分割转录文本"""
+        maxkb_logger.info(f"🔧 smart_split_transcription called - text length: {len(text)}, limit: {limit}, overlap: {overlap}")
+
         if len(text) <= limit:
+            maxkb_logger.info(f"✅ Text length {len(text)} <= limit {limit}, no splitting needed")
             return [text]
 
         # 预处理：移除过多的空行
+        original_length = len(text)
         text = '\n'.join(line for line in text.split('\n') if line.strip())
+        cleaned_length = len(text)
+        maxkb_logger.info(f"🧹 Text preprocessing: {original_length} -> {cleaned_length} characters")
 
         chunks = []
         start = 0
+        iteration = 0
+        max_iterations = len(text) // (limit // 2) + 10  # 防止无限循环
+
+        while start < len(text) and iteration < max_iterations:
+            iteration += 1
+            maxkb_logger.info(f"🔄 Iteration {iteration}: start={start}, text_length={len(text)}")
 
-        while start < len(text):
             # 计算当前块的结束位置
             end = min(start + limit, len(text))
+            maxkb_logger.info(f"📍 Initial end position: {end}")
 
             # 如果不是最后一块，尝试在句号处断开
             if end < len(text):
@@ -355,39 +368,68 @@ class MediaSplitHandle(BaseSplitHandle):
 
                 # 优先在句号处断开
                 last_period = text.rfind('。', start, end)
+                maxkb_logger.info(f"🔍 Last period position: {last_period}")
+
                 if last_period > start + limit // 2:
                     best_end = last_period + 1
+                    maxkb_logger.info(f"✂️ Splitting at period: position {best_end}")
                 else:
                     # 其次在换行符处断开
                     last_newline = text.rfind('\n', start, end)
+                    maxkb_logger.info(f"🔍 Last newline position: {last_newline}")
+
                     if last_newline > start + limit // 3:
                         best_end = last_newline + 1
+                        maxkb_logger.info(f"✂️ Splitting at newline: position {best_end}")
+                    else:
+                        maxkb_logger.info(f"⚠️ No good split point found, using default position {end}")
 
                 end = best_end
 
             # 提取当前块
             chunk = text[start:end].strip()
+            chunk_length = len(chunk)
+            maxkb_logger.info(f"📦 Chunk {len(chunks)+1}: length={chunk_length}, start={start}, end={end}")
+            maxkb_logger.info(f"📝 Chunk preview: '{chunk[:100]}...'")
+
             if chunk:
                 chunks.append(chunk)
+            else:
+                maxkb_logger.warning(f"⚠️ Empty chunk created, skipping")
 
             # 计算下一块的开始位置（考虑重叠）
-            start = max(end - overlap, len(chunks[-1]) if chunks else 0)
+            # 修复：确保start总是增加的
+            old_start = start
+            start = max(end - overlap, end - chunk_length // 2)
+
+            # 确保至少前进1个字符
+            if start <= old_start:
+                start = old_start + 1
+                maxkb_logger.warning(f"⚠️ Adjusting start position to prevent infinite loop: {start}")
+
+            maxkb_logger.info(f"➡️ Next start position: {start}")
 
             # 避免无限循环
             if start >= len(text):
+                maxkb_logger.info(f"🏁 Reached end of text")
                 break
 
+        if iteration >= max_iterations:
+            maxkb_logger.error(f"❌ Max iterations ({max_iterations}) reached, potential infinite loop")
+
+        maxkb_logger.info(f"✅ Split completed: {len(chunks)} chunks created in {iteration} iterations")
         return chunks
 
     def _apply_smart_split(self, result: dict, limit: int = 1000, with_filter: bool = False):
         """应用智能分块到转录结果"""
-        overlap = 100  # 前后重叠字符数
+        try:
+            overlap = 100  # 前后重叠字符数
 
-        maxkb_logger.info(f"🔧 Starting smart split process - limit: {limit}, with_filter: {with_filter}")
+            maxkb_logger.info(f"🔧 Starting smart split process - limit: {limit}, with_filter: {with_filter}")
         original_paragraphs = result.get('content', [])
-        maxkb_logger.info(f"📊 Original paragraphs count: {len(original_paragraphs)}")
+            maxkb_logger.info(f"📊 Original paragraphs count: {len(original_paragraphs)}")
 
-        new_paragraphs = []
+            new_paragraphs = []
         total_chunks_created = 0
 
         for idx, paragraph in enumerate(original_paragraphs):
@@ -408,15 +450,17 @@ class MediaSplitHandle(BaseSplitHandle):
                     maxkb_logger.info(f"✂️ Paragraph {idx+1} needs splitting (length {content_length} > limit {limit})")
 
                     # 应用智能分块
-                    chunks = self.smart_split_transcription(content, limit, overlap)
-                    maxkb_logger.info(f"✂️ Split paragraph {idx+1} into {len(chunks)} chunks")
+                    try:
+                        maxkb_logger.info(f"🔄 Calling smart_split_transcription for paragraph {idx+1}")
+                        chunks = self.smart_split_transcription(content, limit, overlap)
+                        maxkb_logger.info(f"✂️ Split paragraph {idx+1} into {len(chunks)} chunks")
 
-                    # 记录每个chunk的详细信息
-                    for c_idx, chunk in enumerate(chunks):
-                        maxkb_logger.info(f"📦 Chunk {c_idx+1}/{len(chunks)}: length={len(chunk)}, preview='{chunk[:50]}...'")
+                        # 记录每个chunk的详细信息
+                        for c_idx, chunk in enumerate(chunks):
+                            maxkb_logger.info(f"📦 Chunk {c_idx+1}/{len(chunks)}: length={len(chunk)}, preview='{chunk[:50]}...'")
 
-                    # 创建新的段落
-                    for c_idx, chunk in enumerate(chunks):
+                        # 创建新的段落
+                        for c_idx, chunk in enumerate(chunks):
                         # 保留原始元数据，但更新分段相关信息
                         metadata = paragraph.get('metadata', {}).copy()
                         metadata.update({
@@ -437,6 +481,14 @@ class MediaSplitHandle(BaseSplitHandle):
                         }
                         new_paragraphs.append(new_paragraph)
                         total_chunks_created += 1
+
+                    except Exception as split_error:
+                        maxkb_logger.error(f"❌ Error splitting paragraph {idx+1}: {str(split_error)}")
+                        maxkb_logger.error(f"🔍 Split error traceback: {traceback.format_exc()}")
+                        # 如果分块失败，保留原始段落
+                        maxkb_logger.info(f"🔄 Keeping original paragraph due to split error")
+                        new_paragraphs.append(paragraph)
+                        total_chunks_created += 1
                 else:
                     maxkb_logger.info(f"📄 Paragraph {idx+1} does not need splitting (length {content_length} <= limit {limit})")
                     new_paragraphs.append(paragraph)
@@ -462,9 +514,16 @@ class MediaSplitHandle(BaseSplitHandle):
         result['metadata'] = metadata
 
         maxkb_logger.info(f"✅ Smart split completed - original: {len(original_paragraphs)} paragraphs, final: {len(new_paragraphs)} chunks")
-        maxkb_logger.info(f"📈 Total chunks created: {total_chunks_created}")
+            maxkb_logger.info(f"📈 Total chunks created: {total_chunks_created}")
 
-        return result
+            return result
+
+        except Exception as e:
+            maxkb_logger.error(f"❌ Fatal error in _apply_smart_split: {str(e)}")
+            maxkb_logger.error(f"🔍 Full error traceback: {traceback.format_exc()}")
+            # 返回原始结果，不做处理
+            maxkb_logger.info(f"🔄 Returning original result without splitting")
+            return result
 
     def _clean_text(self, text):
         """清理文本：去掉重复多余符号空格、空行、制表符"""