auto-coder 0.1.253__py3-none-any.whl → 0.1.255__py3-none-any.whl

autocoder/index/entry.py

@@ -107,7 +107,13 @@ def build_index_and_filter_files(
                 model_name = "unknown(without default model name)"    
             printer.print_in_terminal("quick_filter_start", style="blue", model_name=model_name)
             quick_filter = QuickFilter(index_manager,stats,sources)        
-            final_files = quick_filter.filter(index_manager.read_index(),args.query)
+            quick_filter_result = quick_filter.filter(index_manager.read_index(),args.query)
+            if quick_filter_result.has_error:                
+                raise KeyboardInterrupt(printer.get_message_from_key_with_format("quick_filter_failed",error=quick_filter_result.error_message))
+
+            # Merge quick filter results into final_files
+            final_files.update(quick_filter_result.files)    
+
         
         if not args.skip_filter_index and not args.index_filter_model:
             model_name = getattr(index_manager.llm, 'default_model_name', None)
@@ -115,7 +121,8 @@ def build_index_and_filter_files(
                 model_name = "unknown(without default model name)" 
             printer.print_in_terminal("normal_filter_start", style="blue",model_name=model_name)
             normal_filter = NormalFilter(index_manager,stats,sources)
-            final_files = normal_filter.filter(index_manager.read_index(),args.query)
+            # Merge normal filter results into final_files
+            final_files.update(normal_filter.filter(index_manager.read_index(),args.query))
         
 
     def display_table_and_get_selections(data):

autocoder/index/filter/quick_filter.py

@@ -1,4 +1,5 @@
-from typing import List, Union, Dict, Any
+from typing import List, Union, Dict, Any, Optional
+from pydantic import BaseModel
 from autocoder.utils.auto_coder_utils.chat_stream_out import stream_out
 from autocoder.common.utils_code_auto_generate import stream_chat_with_continue
 from byzerllm.utils.str2model import to_model
@@ -26,6 +27,11 @@ def get_file_path(file_path):
     return file_path
 
 
+class QuickFilterResult(BaseModel):
+    files: Dict[str, TargetFile]
+    has_error: bool
+    error_message: Optional[str] = None
+
 class QuickFilter():
     def __init__(self, index_manager: IndexManager,stats:Dict[str,Any],sources:List[SourceCode]):
         self.index_manager = index_manager
@@ -36,11 +42,7 @@ class QuickFilter():
         self.max_tokens = self.args.index_filter_model_max_input_length
 
 
-    def big_filter(self, index_items: List[IndexItem],) -> Dict[str, TargetFile]:
-        
-        final_files: Dict[str, TargetFile] = {}
-        final_files_lock = threading.Lock()
-        
+    def big_filter(self, index_items: List[IndexItem],) -> QuickFilterResult:
         chunks = []
         current_chunk = []
                 
@@ -70,9 +72,10 @@ class QuickFilter():
                 split_size=len(chunks)
             )    
 
-        def process_chunk(chunk_index: int, chunk: List[IndexItem]) -> None:
+        def process_chunk(chunk_index: int, chunk: List[IndexItem]) -> QuickFilterResult:
             try:
                 model_name = ",".join(get_llm_names(self.index_manager.index_filter_llm))
+                files: Dict[str, TargetFile] = {}
                 
                 if chunk_index == 0:
                     # 第一个chunk使用流式输出
@@ -93,13 +96,16 @@ class QuickFilter():
                     file_number_list = self.quick_filter_files.with_llm(self.index_manager.index_filter_llm).with_return_type(FileNumberList).run(chunk, self.args.query)
                 
                 if file_number_list:
-                    with final_files_lock:
-                        for file_number in file_number_list.file_list:
-                            file_path = get_file_path(chunk[file_number].module_name)
-                            final_files[file_path] = TargetFile(
-                                file_path=chunk[file_number].module_name,
-                                reason=self.printer.get_message_from_key("quick_filter_reason")
-                            )
+                    for file_number in file_number_list.file_list:
+                        file_path = get_file_path(chunk[file_number].module_name)
+                        files[file_path] = TargetFile(
+                            file_path=chunk[file_number].module_name,
+                            reason=self.printer.get_message_from_key("quick_filter_reason")
+                        )
+                return QuickFilterResult(
+                    files=files,
+                    has_error=False
+                )
                             
             except Exception as e:
                 self.printer.print_in_terminal(
@@ -107,18 +113,40 @@ class QuickFilter():
                     style="red",
                     error=str(e)
                 )
+                return QuickFilterResult(
+                    files={},
+                    has_error=True,
+                    error_message=str(e)
+                )
 
+        results: List[QuickFilterResult] = []
         if chunks:
             with ThreadPoolExecutor() as executor:
-                # 提交所有chunks到线程池
+                # 提交所有chunks到线程池并收集结果
                 futures = [executor.submit(process_chunk, i, chunk) 
                           for i, chunk in enumerate(chunks)]
                 
-                # 等待所有任务完成
+                # 等待所有任务完成并收集结果
                 for future in futures:
-                    future.result()
-                
-        return final_files    
+                    results.append(future.result())
+        
+        # 合并所有结果
+        final_files: Dict[str, TargetFile] = {}
+        has_error = False
+        error_messages: List[str] = []
+        
+        for result in results:
+            if result.has_error:
+                has_error = True
+                if result.error_message:
+                    error_messages.append(result.error_message)
+            final_files.update(result.files)
+        
+        return QuickFilterResult(
+            files=final_files,
+            has_error=has_error,
+            error_message="\n".join(error_messages) if error_messages else None
+        )
 
     @byzerllm.prompt()
     def quick_filter_files(self,file_meta_list:List[IndexItem],query:str) -> str:
@@ -161,8 +189,8 @@ class QuickFilter():
         }
         return context    
 
-    def filter(self, index_items: List[IndexItem], query: str) -> Dict[str, TargetFile]:
-        final_files: Dict[str, TargetFile] = {}                
+    def filter(self, index_items: List[IndexItem], query: str) -> QuickFilterResult:
+        final_files: Dict[str, TargetFile] = {}
         start_time = time.monotonic()        
 
         prompt_str = self.quick_filter_files.prompt(index_items,query)            
@@ -217,7 +245,11 @@ class QuickFilter():
                 style="red",
                 error=str(e)
             )
-            return final_files
+            return QuickFilterResult(
+                files=final_files,
+                has_error=True,
+                error_message=str(e)
+            )
         
         if file_number_list:
             for file_number in file_number_list.file_list:
@@ -227,4 +259,7 @@ class QuickFilter():
                 )
         end_time = time.monotonic()            
         self.stats["timings"]["quick_filter"] = end_time - start_time            
-        return final_files
+        return QuickFilterResult(
+            files=final_files,
+            has_error=False
+        )    

autocoder/models.py

@@ -180,10 +180,7 @@ def update_model_speed(name: str, speed: float) -> bool:
         
     Returns:
         bool: 是否更新成功
-    """
-    if speed <= 0:
-        raise ValueError("Speed must be positive")
-        
+    """            
     models = load_models()
     updated = False
     for model in models:

autocoder/rag/doc_filter.py

@@ -64,7 +64,7 @@ class DocFilter:
             self.recall_llm = self.llm
         
         self.args = args
-        self.relevant_score = self.args.rag_doc_filter_relevance or 5
+        self.relevant_score = self.args.rag_doc_filter_relevance
         self.on_ray = on_ray
         self.path = path        
 
@@ -140,6 +140,7 @@ class DocFilter:
                     f"  - File: {doc.module_name}\n"
                     f"  - Relevance: {'Relevant' if relevance and relevance.is_relevant else 'Not Relevant'}\n"
                     f"  - Score: {relevance.relevant_score if relevance else 'N/A'}\n"
+                    f"  - Score Threshold: {self.relevant_score}\n"
                     f"  - Raw Response: {v}\n"
                     f"  - Timing:\n"
                     f"    * Total Duration: {task_timing.duration:.2f}s\n"

autocoder/rag/long_context_rag.py

@@ -449,20 +449,7 @@ class LongContextRAG:
 
             if highly_relevant_docs:
                 relevant_docs = highly_relevant_docs
-                logger.info(f"Found {len(relevant_docs)} highly relevant documents")
-            else:
-                if relevant_docs:
-                    prefix_chunk = FilterDoc(
-                        source_code=SourceCode(
-                            module_name="特殊说明",
-                            source_code="没有找到特别相关的内容，下面的内容是一些不是很相关的文档。在根据后续文档回答问题前，你需要和用户先提前说一下。",
-                        ),
-                        relevance=DocRelevance(False, 0),
-                    )
-                    relevant_docs.insert(0, prefix_chunk)
-                    logger.info(
-                        "No highly relevant documents found. Added a prefix chunk to indicate this."
-                    )
+                logger.info(f"Found {len(relevant_docs)} highly relevant documents")            
 
             logger.info(
                 f"Filter time: {filter_time:.2f} seconds with {len(relevant_docs)} docs"

autocoder/version.py

@@ -1 +1 @@
-__version__ = "0.1.253"
+__version__ = "0.1.255"