auto-coder 0.1.235__py3-none-any.whl → 0.1.238__py3-none-any.whl

autocoder/common/code_modification_ranker.py

@@ -1,70 +1,79 @@
 import byzerllm
-from typing import List,Union
+from typing import List, Union
 from autocoder.common import AutoCoderArgs
 from autocoder.common.types import CodeGenerateResult
 from pydantic import BaseModel
-from loguru import logger
+from autocoder.common.printer import Printer
 from concurrent.futures import ThreadPoolExecutor, as_completed
 import traceback
-
+from autocoder.common.utils_code_auto_generate import chat_with_continue
+from byzerllm.utils.str2model import to_model
 class RankResult(BaseModel):
-    rank_result:List[int]
+    rank_result: List[int]
+
 
 class CodeModificationRanker:
     def __init__(self, llm: byzerllm.ByzerLLM, args: AutoCoderArgs):
         self.llm = llm
-        self.args = args     
-        self.llms = self.llm.get_sub_client("generate_rerank_model") or [self.llm]
+        self.args = args
+        self.llms = self.llm.get_sub_client(
+            "generate_rerank_model") or [self.llm]
         if not isinstance(self.llms, list):
             self.llms = [self.llms]
-    
+        self.printer = Printer()
+
     @byzerllm.prompt()
-    def _rank_modifications(self, s:CodeGenerateResult) -> str:
+    def _rank_modifications(self, s: CodeGenerateResult) -> str:
         '''
         对一组代码修改进行质量评估并排序。
 
         下面是修改需求：
-        
+
         <edit_requirement>
         {{ s.conversations[0][-2]["content"] }}
         </edit_requirement>
-        
+
         下面是相应的代码修改：
         {% for content in s.contents %}
         <edit_block id="{{ loop.index0 }}">
         {{content}}
         </edit_block>
         {% endfor %}
-        
+
         请输出如下格式的评估结果,只包含 JSON 数据:
 
         ```json
         {
-            "rank_result": [id1, id2, id3]  // id 为 edit_block 的 id,按质量从高到低排序
+            "rank_result": [id1, id2, id3] 
         }
         ```
 
-        注意：        
-        1. 只输出前面要求的 Json 格式就好，不要输出其他内容，Json 需要使用 ```json ```包裹        
+        注意：   
+        1. id 为 edit_block 的 id,按质量从高到低排序，并且 id 必须是数字        
+        2. 只输出前面要求的 Json 格式就好，不要输出其他内容，Json 需要使用 ```json ```包裹                
         '''
-        
 
     def rank_modifications(self, generate_result: CodeGenerateResult) -> CodeGenerateResult:
         import time
         from collections import defaultdict
 
         start_time = time.time()
-        
+
         # 如果只有一个候选，直接返回
         if len(generate_result.contents) == 1:
-            logger.info("Only 1 candidate, skip ranking")
+            self.printer.print_in_terminal("ranking_skip", style="blue")
             return generate_result
-            
-        logger.info(f"Start ranking {len(generate_result.contents)} candidates")
+
+        self.printer.print_in_terminal(
+            "ranking_start", style="blue", count=len(generate_result.contents))
         generate_times = self.args.generate_times_same_model
         total_tasks = len(self.llms) * generate_times
+
+        query = self._rank_modifications.prompt(generate_result)
+        input_tokens_count = 0
+        generated_tokens_count = 0
         try:
-            # Create a thread pool with (number of models * generate_times) workers            
+            # Create a thread pool with (number of models * generate_times) workers
             with ThreadPoolExecutor(max_workers=total_tasks) as executor:
                 # Submit tasks for each model and generate_times
                 futures = []
@@ -72,49 +81,70 @@ class CodeModificationRanker:
                     for _ in range(generate_times):
                         futures.append(
                             executor.submit(
-                                self._rank_modifications.with_llm(llm).with_return_type(RankResult).run,
-                                generate_result
+                                chat_with_continue,
+                                llm,
+                                [{"role": "user", "content": query}],
+                                {}
                             )
                         )
-                
+
                 # Collect all results
                 results = []
                 for future in as_completed(futures):
                     try:
-                        v = future.result()
+                        result = future.result()
+                        input_tokens_count += result.input_tokens_count
+                        generated_tokens_count += result.generated_tokens_count
+                        v = to_model(result.content,RankResult)                        
                         results.append(v.rank_result)
                     except Exception as e:
-                        logger.warning(f"Ranking request failed: {str(e)}")
-                        logger.debug(traceback.format_exc())
+                        self.printer.print_in_terminal(
+                            "ranking_failed_request", style="yellow", error=str(e))
+                        if self.args.debug:
+                            print(traceback.format_exc())
                         continue
-                
+
                 if not results:
-                    raise Exception("All ranking requests failed")
-                
+                    raise Exception(
+                        self.printer.get_message_from_key("ranking_all_failed"))
+
                 # Calculate scores for each candidate
                 candidate_scores = defaultdict(float)
                 for rank_result in results:
                     for idx, candidate_id in enumerate(rank_result):
                         # Score is 1/(position + 1) since position starts from 0
                         candidate_scores[candidate_id] += 1.0 / (idx + 1)
-                
+
                 # Sort candidates by score in descending order
-                sorted_candidates = sorted(candidate_scores.keys(), 
-                                         key=lambda x: candidate_scores[x], 
-                                         reverse=True)
-                
+                sorted_candidates = sorted(candidate_scores.keys(),
+                                           key=lambda x: candidate_scores[x],
+                                           reverse=True)
+
                 elapsed = time.time() - start_time
                 # Format scores for logging
-                score_details = ", ".join([f"candidate {i}: {candidate_scores[i]:.2f}" for i in sorted_candidates])
-                logger.info(f"Ranking completed in {elapsed:.2f}s, total voters: {total_tasks}, best candidate index: {sorted_candidates[0]}, scores: {score_details}")
-                
-                rerank_contents = [generate_result.contents[i] for i in sorted_candidates]
-                rerank_conversations = [generate_result.conversations[i] for i in sorted_candidates]
-                return CodeGenerateResult(contents=rerank_contents,conversations=rerank_conversations)
-                
+                score_details = ", ".join(
+                    [f"candidate {i}: {candidate_scores[i]:.2f}" for i in sorted_candidates])
+                self.printer.print_in_terminal(
+                    "ranking_complete",
+                    style="green",
+                    elapsed=f"{elapsed:.2f}",
+                    total_tasks=total_tasks,
+                    best_candidate=sorted_candidates[0],
+                    scores=score_details,
+                    input_tokens=input_tokens_count,
+                    output_tokens=generated_tokens_count
+                )
+
+                rerank_contents = [generate_result.contents[i]
+                                   for i in sorted_candidates]
+                rerank_conversations = [
+                    generate_result.conversations[i] for i in sorted_candidates]
+                return CodeGenerateResult(contents=rerank_contents, conversations=rerank_conversations)
+
         except Exception as e:
-            logger.error(f"Ranking process failed: {str(e)}")
-            logger.debug(traceback.format_exc())
+            self.printer.print_in_terminal(
+                "ranking_process_failed", style="red", error=str(e))            
             elapsed = time.time() - start_time
-            logger.warning(f"Ranking failed in {elapsed:.2f}s, using original order")
+            self.printer.print_in_terminal(
+                "ranking_failed", style="yellow", elapsed=f"{elapsed:.2f}")
             return generate_result

autocoder/common/printer.py

@@ -1,5 +1,7 @@
 from rich.console import Console
-from typing import Optional
+from rich.panel import Panel
+from rich.text import Text
+from typing import Optional,Dict,Any
 from byzerllm.utils import format_str_jinja2
 from autocoder.common.auto_coder_lang import get_message
 from autocoder.chat_auto_coder_lang import get_message as get_chat_message
@@ -10,20 +12,26 @@ class Printer:
         else:
             self.console = console
 
-    def get_message_from_key(self, key: str):
+    def get_message_from_key(self, msg_key: str):
         try:
-            return get_message(key)
+            return get_message(msg_key)
         except Exception as e:
-            return get_chat_message(key)
+            return get_chat_message(msg_key)
 
-    def print_in_terminal(self, key: str, style: str = None,**kwargs):     
+    def get_message_from_key_with_format(self, msg_key: str, **kwargs):
+        try:
+            return format_str_jinja2(self.get_message_from_key(msg_key), **kwargs)
+        except Exception as e:
+            return format_str_jinja2(self.get_chat_message_from_key(msg_key), **kwargs)
+
+    def print_in_terminal(self, msg_key: str, style: str = None,**kwargs):     
         try:
             if style:
-                self.console.print(format_str_jinja2(self.get_message_from_key(key),**kwargs), style=style)
+                self.console.print(format_str_jinja2(self.get_message_from_key(msg_key),**kwargs), style=style)
             else:
-                self.console.print(format_str_jinja2(self.get_message_from_key(key),**kwargs))
+                self.console.print(format_str_jinja2(self.get_message_from_key(msg_key),**kwargs))
         except Exception as e:
-            print(self.get_message_from_key(key))
+            print(self.get_message_from_key(msg_key))
 
     
     def print_str_in_terminal(self, content: str, style: str = None):     
@@ -33,4 +41,9 @@ class Printer:
             else:
                 self.console.print(content)
         except Exception as e:
-            print(content)        
+            print(content)   
+
+    def print_panel(self, content: str, text_options:Dict[str,Any], panel_options:Dict[str,Any]):
+        panel = Panel(Text(content, **text_options), **panel_options)
+        self.console.print(panel)    
+                 

autocoder/dispacher/actions/action.py

@@ -25,9 +25,10 @@ from autocoder.common.image_to_page import ImageToPage, ImageToPageDirectly
 from autocoder.utils.conversation_store import store_code_model_conversation
 from loguru import logger
 import time
+from autocoder.common.printer import Printer
 
 
-class BaseAction:
+class BaseAction:    
     def _get_content_length(self, content: str) -> int:
         try:
             tokenizer = BuildinTokenizer()
@@ -43,6 +44,7 @@ class ActionTSProject(BaseAction):
         self.args = args
         self.llm = llm
         self.pp = None
+        self.printer = Printer()
 
     def run(self):
         args = self.args
@@ -96,8 +98,8 @@ class ActionTSProject(BaseAction):
                     f"Content(send to model) is {content_length} tokens, which is larger than the maximum input length {self.args.model_max_input_length}"
                 )                
 
-        if args.execute:            
-            logger.info("Auto generate the code...")
+        if args.execute:             
+            self.printer.print_in_terminal("code_generation_start")
             start_time = time.time()
             if args.auto_merge == "diff":
                 generate = CodeAutoGenerateDiff(
@@ -121,10 +123,15 @@ class ActionTSProject(BaseAction):
                 generate_result = generate.single_round_run(
                     query=args.query, source_content=content
                 )
-            logger.info(f"Code generation completed in {time.time() - start_time:.2f} seconds, input_tokens_count: {generate_result.metadata.get('input_tokens_count', 0)}, generated_tokens_count: {generate_result.metadata.get('generated_tokens_count', 0)}")    
+            self.printer.print_in_terminal(
+                "code_generation_complete",
+                duration=time.time() - start_time,
+                input_tokens=generate_result.metadata.get('input_tokens_count', 0),
+                output_tokens=generate_result.metadata.get('generated_tokens_count', 0)
+            )
             merge_result = None
             if args.execute and args.auto_merge:
-                logger.info("Auto merge the code...")
+                self.printer.print_in_terminal("code_merge_start")
                 if args.auto_merge == "diff":
                     code_merge = CodeAutoMergeDiff(llm=self.llm, args=self.args)
                     merge_result = code_merge.merge_code(generate_result=generate_result)
@@ -165,6 +172,7 @@ class ActionPyScriptProject(BaseAction):
     ) -> None:
         self.args = args
         self.llm = llm
+        self.printer = Printer()
 
     def run(self) -> bool:
         args = self.args
@@ -180,7 +188,7 @@ class ActionPyScriptProject(BaseAction):
     def process_content(self, content: str):
         args = self.args
         if args.execute:
-            logger.info("Auto generate the code...")
+            self.printer.print_in_terminal("code_generation_start")
             start_time = time.time()
             if args.auto_merge == "diff":
                 generate = CodeAutoGenerateDiff(
@@ -205,10 +213,15 @@ class ActionPyScriptProject(BaseAction):
                     query=args.query, source_content=content
                 )
 
-            logger.info(f"Code generation completed in {time.time() - start_time:.2f} seconds, input_tokens_count: {generate_result.metadata.get('input_tokens_count', 0)}, generated_tokens_count: {generate_result.metadata.get('generated_tokens_count', 0)}")    
+            self.printer.print_in_terminal(
+                "code_generation_complete",
+                duration=time.time() - start_time,
+                input_tokens=generate_result.metadata.get('input_tokens_count', 0),
+                output_tokens=generate_result.metadata.get('generated_tokens_count', 0)
+            )
             merge_result = None
             if args.execute and args.auto_merge:
-                logger.info("Auto merge the code...")
+                self.printer.print_in_terminal("code_merge_start")
                 if args.auto_merge == "diff":
                     code_merge = CodeAutoMergeDiff(llm=self.llm, args=self.args)
                     merge_result = code_merge.merge_code(generate_result=generate_result)
@@ -241,7 +254,12 @@ class ActionPyScriptProject(BaseAction):
                 )
 
             end_time = time.time()
-            logger.info(f"Code generation completed in {end_time - start_time:.2f} seconds")
+            self.printer.print_in_terminal(
+                "code_generation_complete",
+                duration=end_time - start_time,
+                input_tokens=generate_result.metadata.get('input_tokens_count', 0),
+                output_tokens=generate_result.metadata.get('generated_tokens_count', 0)
+            )
             with open(self.args.target_file, "w") as file:
                 file.write(content)
 
@@ -253,6 +271,7 @@ class ActionPyProject(BaseAction):
         self.args = args
         self.llm = llm
         self.pp = None
+        self.printer = Printer()
 
     def run(self):
         args = self.args
@@ -282,12 +301,15 @@ class ActionPyProject(BaseAction):
         if args.execute and self.llm and not args.human_as_model:
             content_length = self._get_content_length(content)
             if content_length > self.args.model_max_input_length:
-                logger.warning(
-                    f'''Content(send to model) is {content_length} tokens (you may collect too much files), which is larger than the maximum input length {self.args.model_max_input_length}'''
-                )
+                self.printer.print_in_terminal(
+                "code_execution_warning",
+                style="yellow",
+                content_length=content_length,
+                max_length=self.args.model_max_input_length
+            )
 
         if args.execute:
-            logger.info("Auto generate the code...")
+            self.printer.print_in_terminal("code_generation_start")
             start_time = time.time()
             if args.auto_merge == "diff":
                 generate = CodeAutoGenerateDiff(
@@ -313,10 +335,15 @@ class ActionPyProject(BaseAction):
                 generate_result = generate.single_round_run(
                     query=args.query, source_content=content
                 )
-            logger.info(f"Code generation completed in {time.time() - start_time:.2f} seconds, input_tokens_count: {generate_result.metadata.get('input_tokens_count', 0)}, generated_tokens_count: {generate_result.metadata.get('generated_tokens_count', 0)}")    
+            self.printer.print_in_terminal(
+                "code_generation_complete",
+                duration=time.time() - start_time,
+                input_tokens=generate_result.metadata.get('input_tokens_count', 0),
+                output_tokens=generate_result.metadata.get('generated_tokens_count', 0)
+            )
             merge_result = None
             if args.execute and args.auto_merge:
-                logger.info("Auto merge the code...")
+                self.printer.print_in_terminal("code_merge_start")
                 if args.auto_merge == "diff":
                     code_merge = CodeAutoMergeDiff(llm=self.llm, args=self.args)
                     merge_result = code_merge.merge_code(generate_result=generate_result)
@@ -358,6 +385,7 @@ class ActionSuffixProject(BaseAction):
         self.args = args
         self.llm = llm
         self.pp = None
+        self.printer = Printer()
 
     def run(self):
         args = self.args
@@ -387,7 +415,7 @@ class ActionSuffixProject(BaseAction):
                 )                
 
         if args.execute:
-            logger.info("Auto generate the code...")
+            self.printer.print_in_terminal("code_generation_start")
             start_time = time.time()
             if args.auto_merge == "diff":
                 generate = CodeAutoGenerateDiff(
@@ -412,10 +440,15 @@ class ActionSuffixProject(BaseAction):
                     query=args.query, source_content=content
                 )
               
-        logger.info(f"Code generation completed in {time.time() - start_time:.2f} seconds, input_tokens_count: {generate_result.metadata.get('input_tokens_count', 0)}, generated_tokens_count: {generate_result.metadata.get('generated_tokens_count', 0)}")
+        self.printer.print_in_terminal(
+            "code_generation_complete",
+            duration=time.time() - start_time,
+            input_tokens=generate_result.metadata.get('input_tokens_count', 0),
+            output_tokens=generate_result.metadata.get('generated_tokens_count', 0)
+        )
         merge_result = None
         if args.execute and args.auto_merge:
-            logger.info("Auto merge the code...")
+            self.printer.print_in_terminal("code_merge_start")
             if args.auto_merge == "diff":
                 code_merge = CodeAutoMergeDiff(llm=self.llm, args=self.args)
                 merge_result = code_merge.merge_code(generate_result=generate_result)

autocoder/index/entry.py

@@ -22,6 +22,7 @@ from autocoder.index.types import (
 from autocoder.index.filter.quick_filter import QuickFilter
 from autocoder.index.filter.normal_filter import NormalFilter
 from autocoder.index.index import IndexManager
+from loguru import logger
 
 def build_index_and_filter_files(
     llm, args: AutoCoderArgs, sources: List[SourceCode]
@@ -99,11 +100,14 @@ def build_index_and_filter_files(
                     })
                 )
             )
-
-        quick_filter = QuickFilter(index_manager,stats,sources)
+        
+        #MARK
+        printer.print_in_terminal("quick_filter_start", style="blue")
+        quick_filter = QuickFilter(index_manager,stats,sources)        
         final_files = quick_filter.filter(index_manager.read_index(),args.query)
         
         if not final_files:
+            printer.print_in_terminal("normal_filter_start", style="blue")
             normal_filter = NormalFilter(index_manager,stats,sources)
             final_files = normal_filter.filter(index_manager.read_index(),args.query)
         
@@ -279,7 +283,7 @@ def build_index_and_filter_files(
 
     # Print final statistics in a more structured way
     summary = f"""
-=== Indexing and Filtering Summary ===
+=== File Stat ===    
 • Total files scanned: {stats['total_files']}
 • Files indexed: {stats['indexed_files']}
 • Files filtered:
@@ -288,7 +292,7 @@ def build_index_and_filter_files(
   - Relevance verified: {stats.get('verified_files', 0)}
 • Final files selected: {stats['final_files']}
 
-=== Time Breakdown ===
+=== Time Stat ===
 • Index build: {stats['timings'].get('build_index', 0):.2f}s
 • Quick filter: {stats['timings'].get('quick_filter', 0):.2f}s
 • Normal filter: 
@@ -297,9 +301,18 @@ def build_index_and_filter_files(
     - Relevance check: {stats['timings']["normal_filter"].get('relevance_verification', 0):.2f}s
 • File selection: {stats['timings'].get('file_selection', 0):.2f}s
 • Total time: {total_time:.2f}s
-====================================
+
 """
-    printer.print_str_in_terminal(summary)
+    printer.print_panel(
+        summary,
+        text_options={"justify": "left", "style": "bold white"},
+        panel_options={
+            "title": "Indexing and Filtering Summary", 
+            "border_style": "bold blue",
+            "padding": (1, 2),
+            "expand": False
+        }
+    )
 
     if args.request_id and not args.skip_events:
         queue_communicate.send_event(

autocoder/index/filter/quick_filter.py

@@ -58,7 +58,7 @@ class QuickFilter():
         特别注意
         1. 如果用户的query里 @文件 或者 @@符号，那么被@的文件或者@@的符号必须要返回，并且查看他们依赖的文件是否相关。        
         2. 如果 query 里是一段历史对话，那么对话里的内容提及的文件路径必须要返回。
-        3. json格式数据不允许有注释
+        3. json格式数据不允许有注释        
         '''
         file_meta_str = "\n".join([f"##[{index}]{item.module_name}\n{item.symbols}" for index,item in enumerate(file_meta_list)])
         context = {

autocoder/index/index.py

@@ -263,22 +263,21 @@ class IndexManager:
         else:
             index_data = {}
 
-        @byzerllm.prompt()
-        def error_message(source_dir: str, file_path: str):
-            """
-            The source_dir is different from the path in index file (e.g. file_path:{{ file_path }} source_dir:{{ source_dir }}).
-            You may need to replace the prefix with the source_dir in the index file or Just delete the index file to rebuild it.
-            """
-
-        for item in index_data.keys():
-            if not item.startswith(self.source_dir):
+        # 清理已不存在的文件索引
+        keys_to_remove = []
+        for file_path in index_data:
+            if not os.path.exists(file_path):
+                keys_to_remove.append(file_path)                
+        
+        # 删除无效条目并记录日志
+        for key in set(keys_to_remove):
+            if key in index_data:
+                del index_data[key]
                 self.printer.print_in_terminal(
-                    "index_source_dir_mismatch",
+                    "index_file_removed", 
                     style="yellow",
-                    source_dir=self.source_dir,
-                    file_path=item
+                    file_path=key
                 )
-                break
 
         updated_sources = []
 
@@ -336,9 +335,16 @@ class IndexManager:
                     index_data[module_name] = result
                     updated_sources.append(module_name)
 
-        if updated_sources:
+        # 如果 updated_sources 或 keys_to_remove 有值，则保存索引文件
+        if updated_sources or keys_to_remove:
             with open(self.index_file, "w") as file:
                 json.dump(index_data, file, ensure_ascii=False, indent=2)
+            self.printer.print_in_terminal(
+                "index_file_saved",
+                style="green",
+                updated_files=len(updated_sources),
+                removed_files=len(keys_to_remove)
+            )
 
         return index_data
 

autocoder/version.py

@@ -1 +1 @@
-__version__ = "0.1.235"
+__version__ = "0.1.238"