auto-coder 0.1.346__py3-none-any.whl → 0.1.347__py3-none-any.whl

{auto_coder-0.1.346.dist-info → auto_coder-0.1.347.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: auto-coder
-Version: 0.1.346
+Version: 0.1.347
 Summary: AutoCoder: AutoCoder
 Author: allwefantasy
 Classifier: Programming Language :: Python :: 3.10

{auto_coder-0.1.346.dist-info → auto_coder-0.1.347.dist-info}/RECORD

@@ -14,7 +14,7 @@ autocoder/command_parser.py,sha256=fx1g9E6GaM273lGTcJqaFQ-hoksS_Ik2glBMnVltPCE,1
 autocoder/lang.py,sha256=PFtATuOhHRnfpqHQkXr6p4C893JvpsgwTMif3l-GEi0,14321
 autocoder/models.py,sha256=_SCar82QIeBFTZZBdM2jPS6atKVhHnvE0gX3V0CsxD4,11590
 autocoder/run_context.py,sha256=IUfSO6_gp2Wt1blFWAmOpN0b0nDrTTk4LmtCYUBIoro,1643
-autocoder/version.py,sha256=FUAij2FADNMtAmQCx7M7pqiOpTP65zxKY6khe4EGQUs,23
+autocoder/version.py,sha256=R1KcSahF3DMgAzBREOFaR3Pjcqm9a3u4b36Hb2RFm50,23
 autocoder/agent/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 autocoder/agent/agentic_edit.py,sha256=XsfePZ-t6M-uBSdG1VLZXk1goqXk2HPeJ_A8IYyBuWQ,58896
 autocoder/agent/agentic_edit_types.py,sha256=oFcDd_cxJ2yH9Ed1uTpD3BipudgoIEWDMPb5pAkq4gI,3288
@@ -54,7 +54,7 @@ autocoder/common/action_yml_file_manager.py,sha256=DdF5P1R_B_chCnnqoA2IgogakWLZk
 autocoder/common/anything2images.py,sha256=0ILBbWzY02M-CiWB-vzuomb_J1hVdxRcenAfIrAXq9M,25283
 autocoder/common/anything2img.py,sha256=iZQmg8srXlD7N5uGl5b_ONKJMBjYoW8kPmokkG6ISF0,10118
 autocoder/common/audio.py,sha256=Kn9nWKQddWnUrAz0a_ZUgjcu4VUU_IcZBigT7n3N3qc,7439
-autocoder/common/auto_coder_lang.py,sha256=ozoGTy4ZFn3YsO5zWhvAGCu54mK4LtnRfC2yCvrMc_8,42462
+autocoder/common/auto_coder_lang.py,sha256=bqBoICLIvi9l8jRCwcNLWR6n5pI3ix7YDPGpmqQDmgc,42677
 autocoder/common/auto_configure.py,sha256=D4N-fl9v8bKM5-Ds-uhkC2uGDmHH_ZjLJ759F8KXMKs,13129
 autocoder/common/buildin_tokenizer.py,sha256=L7d5t39ZFvUd6EoMPXUhYK1toD0FHlRH1jtjKRGokWU,1236
 autocoder/common/chunk_validation.py,sha256=BrR_ZWavW8IANuueEE7hS8NFAwEvm8TX34WnPx_1hs8,3030
@@ -126,18 +126,19 @@ autocoder/common/v2/code_editblock_manager.py,sha256=G0CIuV9Ki0FqMLnpA8nBT4pnkCN
 autocoder/common/v2/code_manager.py,sha256=C403bS-f6urixwitlKHcml-J03hci-UyNwHJOqBiY6Q,9182
 autocoder/common/v2/code_strict_diff_manager.py,sha256=v-J1kDyLg7tLGg_6_lbO9S4fNkx7M_L8Xr2G7fPptiU,9347
 autocoder/common/v2/agent/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-autocoder/common/v2/agent/agentic_edit.py,sha256=Ghx6HTCdjIYj96zoB5xkL6aGLizftgCmy5ZcUSaFXtQ,92599
+autocoder/common/v2/agent/agentic_edit.py,sha256=pxgKD-FZaaRQkPgWsxvLleUXDRFC8qSlYfNUHU6ZENU,94794
 autocoder/common/v2/agent/agentic_edit_conversation.py,sha256=qLLhTegH619JQTp3s1bj5FVn2hAcoV-DlhGO3UyIOMc,7338
-autocoder/common/v2/agent/agentic_edit_types.py,sha256=6qBLLmvdlcsbzrpMHsYQVIHqbOWubMXOnmkqTs1pBWQ,4629
+autocoder/common/v2/agent/agentic_edit_types.py,sha256=VJMrictg6hJ3mC45VgQGRd43DyDUPDUvPV1Rf3z72NI,4776
 autocoder/common/v2/agent/agentic_tool_display.py,sha256=WKirt-2V346KLnbHgH3NVJiK3xvriD9oaCWj2IdvzLU,7309
 autocoder/common/v2/agent/ignore_utils.py,sha256=gnUchRzKMLbUm_jvnKL-r-K9MWKPtt-6iiuzijY7Es0,1717
-autocoder/common/v2/agent/agentic_edit_tools/__init__.py,sha256=wGICCc1dYh07osB21j62zOQ9Ws0PyyOQ12UYRHmHrtI,1229
+autocoder/common/v2/agent/agentic_edit_tools/__init__.py,sha256=RbPZZcZg_VnGssL577GxSyFrYrxQ_LopJ4G_-mY3z_Q,1337
 autocoder/common/v2/agent/agentic_edit_tools/ask_followup_question_tool_resolver.py,sha256=bwtf4m9N82TCP3piK5UglJk1FVFFm7ZX59XerA2qxko,3131
 autocoder/common/v2/agent/agentic_edit_tools/attempt_completion_tool_resolver.py,sha256=82ZGKeRBSDKeead_XVBW4FxpiE-5dS7tBOk_3RZ6B5s,1511
 autocoder/common/v2/agent/agentic_edit_tools/base_tool_resolver.py,sha256=Zid2m1uZd-2wVFGc_n_KAViXZyNjbdLSpI5n7ut1RUQ,1036
 autocoder/common/v2/agent/agentic_edit_tools/execute_command_tool_resolver.py,sha256=GueQfIY2hVu2R5j9R5rBtn2znl5MlmEdGtsa6snsMHs,4112
 autocoder/common/v2/agent/agentic_edit_tools/list_code_definition_names_tool_resolver.py,sha256=8QoMsADUDWliqiDt_dpguz31403syB8eeW0Pcw-qfb8,3842
 autocoder/common/v2/agent/agentic_edit_tools/list_files_tool_resolver.py,sha256=d0LzGPA3zsIHK5s1-arPry6ddWFSymRsMY3VbkV6v5A,5795
+autocoder/common/v2/agent/agentic_edit_tools/list_package_info_tool_resolver.py,sha256=dIdV12VuczHpHuHgx2B1j_3BZYc9PL0jfHCuBk9ryk8,2005
 autocoder/common/v2/agent/agentic_edit_tools/plan_mode_respond_tool_resolver.py,sha256=lGT4_QYJK6Fa9f6HVSGo0cSsGK7qCsDYgJGUowNxPzk,1499
 autocoder/common/v2/agent/agentic_edit_tools/read_file_tool_resolver.py,sha256=9Bh0KVbL0qiIqwChlb77biiBiETQ3zekxGe5Fj7hXAg,2800
 autocoder/common/v2/agent/agentic_edit_tools/replace_in_file_tool_resolver.py,sha256=lpD4fCbVR8GTrynqXON69IjM94nPy3nuUL62Ashm5O4,7988
@@ -211,11 +212,11 @@ autocoder/rag/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 autocoder/rag/api_server.py,sha256=StGyxrM-7-W2vYHJq-i_Fv-MHrl9UgVWY272Hd-6VJ4,13090
 autocoder/rag/conversation_to_queries.py,sha256=xwmErn4WbdADnhK1me-h_6fV3KYrl_y1qPNQl1aoI6o,4810
 autocoder/rag/doc_filter.py,sha256=UduVO2mlrngwJICrefjDJTYfdmQ4GcRXrfWDQ7xXksk,14206
-autocoder/rag/document_retriever.py,sha256=5BDqKVJqLPScEnua5S5suXhWuCaALIfPf5obXeJoWfs,8461
+autocoder/rag/document_retriever.py,sha256=rFwbAuHTvEFJq16HQNlmRLyJp2ddn2RNFslw_ncU7NI,8847
 autocoder/rag/lang.py,sha256=HvcMeu6jReEJOGxyLMn4rwBoD-myFwmykS3VLceBJLs,3364
 autocoder/rag/llm_wrapper.py,sha256=Ht5GF5yJtrztoliujsZzx_ooWZmHkd5xLZKcGEiicZw,4303
-autocoder/rag/long_context_rag.py,sha256=84haMjfbPd2Rw_8Zdj-Azf_v9tbYST_OUPW0NZlmeJY,42189
-autocoder/rag/qa_conversation_strategy.py,sha256=vv62JhmdZvLJ3U1gG4ZZ3m9thIlGvU1H0RXHtrRTJ6s,11788
+autocoder/rag/long_context_rag.py,sha256=syPIxO_TQJpBgjZ0taF-G7xVGvkNjKWL65KTI-sy4io,42234
+autocoder/rag/qa_conversation_strategy.py,sha256=N4zcLstmTEZH4iGTnmNhzlRSRPfdZnTlKOXPqnqSEUQ,11727
 autocoder/rag/rag_config.py,sha256=8LwFcTd8OJWWwi1_WY4IzjqgtT6RyE2j4PjxS5cCTDE,802
 autocoder/rag/rag_entry.py,sha256=6TKtErZ0Us9XSV6HgRKXA6yR3SiZGPHpynOKSaR1wgE,2463
 autocoder/rag/raw_rag.py,sha256=BOr0YGf3umjqXOIDVO1LXQ0bIHx8hzBdiubND2ezyxc,2946
@@ -232,13 +233,14 @@ autocoder/rag/utils.py,sha256=f21ybCAlYVgr3tJP9MkVoM9d82-uG5NHu2gsv2oaVBQ,4961
 autocoder/rag/variable_holder.py,sha256=PFvBjFcR7-fNDD4Vcsc8CpH2Te057vcpwJMxtrfUgKI,75
 autocoder/rag/cache/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 autocoder/rag/cache/base_cache.py,sha256=EaYYYbclMBvnlOUoM7qonnluwZX5oSvUjdvGvFun8_8,742
-autocoder/rag/cache/byzer_storage_cache.py,sha256=S_afj4_fFt88A0KJ23N-Gc3mMwPTDjzv9E3J9O_IiT4,29319
+autocoder/rag/cache/byzer_storage_cache.py,sha256=buUrsY7-gWf6BXtzhFVKRJ-ME4LGrr9EEzjoSVu9G6g,27937
 autocoder/rag/cache/cache_result_merge.py,sha256=VnTdbT2OMBmWl_83bqds97d9_M33IhPNX8tF7KH2GMM,10556
-autocoder/rag/cache/file_monitor_cache.py,sha256=OdSXTH3vo6inAzkN5d55I0RN03GUlSlnUEKmXpjFl78,9443
-autocoder/rag/cache/local_byzer_storage_cache.py,sha256=1xskK7X_hFEAsHHoT_F9lFYhQOTrpQtsFyFCIeI2Mvk,31964
-autocoder/rag/cache/local_duckdb_storage_cache.py,sha256=37WH2mlOgykBUrp6ow43kcaaKFZc1CRTkNLSFdELPcE,32714
+autocoder/rag/cache/failed_files_utils.py,sha256=kITguXANLC3EEJy5JoKzNXrtwvTkmZT-ANPwcno42Ck,1183
+autocoder/rag/cache/file_monitor_cache.py,sha256=TelBRzw2uD9QLUIvyrdD6XWmgvb3MyTgx63ZrpptaXI,9512
+autocoder/rag/cache/local_byzer_storage_cache.py,sha256=X7EVGa4xcmUWC7WLnyGPx17I1l9NvW3OHfkErO4Ww0M,30784
+autocoder/rag/cache/local_duckdb_storage_cache.py,sha256=6NsNlqA8-nGSD8ILXQgO-6QbgE8eUy4uT3eM1t69nic,34447
 autocoder/rag/cache/rag_file_meta.py,sha256=RQ3n4wfkHlB-1ljS3sFSi8ijbsUPeIqBSgjmmbRuwRI,20521
-autocoder/rag/cache/simple_cache.py,sha256=yGmt8iXY2ZVif4_2-_DqWGl_zG0fzgSI3tUOxvdqHZU,14230
+autocoder/rag/cache/simple_cache.py,sha256=yrGgRXGcMNrWSQOviPshm3Qlo2QDNVFoRZPLNTV_nQs,15997
 autocoder/rag/loaders/__init__.py,sha256=EQHEZ5Cmz-mGP2SllUTvcIbYCnF7W149dNpNItfs0yE,304
 autocoder/rag/loaders/docx_loader.py,sha256=ZswPqiiLngUEpzLhNNm1nmwEYV7ZHFEfIoXoG7c5GDU,614
 autocoder/rag/loaders/excel_loader.py,sha256=Ue8YB1z_kBs8SjIPuBskyM08Q1JiONs_BJZPrzi59oo,896
@@ -277,9 +279,9 @@ autocoder/utils/types.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 autocoder/utils/auto_coder_utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 autocoder/utils/auto_coder_utils/chat_stream_out.py,sha256=KW0mlmcHlStXi8-_6fXZ2-ifeJ5mgP0OV7DQFzCtIsw,14008
 autocoder/utils/chat_auto_coder_utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-auto_coder-0.1.346.dist-info/LICENSE,sha256=HrhfyXIkWY2tGFK11kg7vPCqhgh5DcxleloqdhrpyMY,11558
-auto_coder-0.1.346.dist-info/METADATA,sha256=pXxlRR0gvLNboAcWQtU9VzpKpiaCTj4evVC9Ztas7R4,2728
-auto_coder-0.1.346.dist-info/WHEEL,sha256=GV9aMThwP_4oNCtvEC2ec3qUYutgWeAzklro_0m4WJQ,91
-auto_coder-0.1.346.dist-info/entry_points.txt,sha256=0nzHtHH4pNcM7xq4EBA2toS28Qelrvcbrr59GqD_0Ak,350
-auto_coder-0.1.346.dist-info/top_level.txt,sha256=Jqc0_uJSw2GwoFQAa9iJxYns-2mWla-9ok_Y3Gcznjk,10
-auto_coder-0.1.346.dist-info/RECORD,,
+auto_coder-0.1.347.dist-info/LICENSE,sha256=HrhfyXIkWY2tGFK11kg7vPCqhgh5DcxleloqdhrpyMY,11558
+auto_coder-0.1.347.dist-info/METADATA,sha256=-wG5ilXMo-fYflOC9zJZY2N0JdHBluyyxYNqxhzNVwQ,2728
+auto_coder-0.1.347.dist-info/WHEEL,sha256=GV9aMThwP_4oNCtvEC2ec3qUYutgWeAzklro_0m4WJQ,91
+auto_coder-0.1.347.dist-info/entry_points.txt,sha256=0nzHtHH4pNcM7xq4EBA2toS28Qelrvcbrr59GqD_0Ak,350
+auto_coder-0.1.347.dist-info/top_level.txt,sha256=Jqc0_uJSw2GwoFQAa9iJxYns-2mWla-9ok_Y3Gcznjk,10
+auto_coder-0.1.347.dist-info/RECORD,,

autocoder/common/auto_coder_lang.py

@@ -833,6 +833,14 @@ MESSAGES = {
     "/agent/edit/user_query":{
         "en":"User Query",
         "zh":"用户查询"
+    },
+    "/agent/edit/apply_pre_changes":{
+        "en":"Commit user changes",
+        "zh":"提交用户修改"
+    },
+    "/agent/edit/apply_changes":{
+        "en":"Commit changes",
+        "zh":"提交修改"
     }
 }
 

autocoder/common/v2/agent/agentic_edit.py

@@ -52,8 +52,10 @@ from autocoder.common.v2.agent.agentic_edit_tools import (  # Import specific re
     ExecuteCommandToolResolver, ReadFileToolResolver, WriteToFileToolResolver,
     ReplaceInFileToolResolver, SearchFilesToolResolver, ListFilesToolResolver,
     ListCodeDefinitionNamesToolResolver, AskFollowupQuestionToolResolver,
-    AttemptCompletionToolResolver, PlanModeRespondToolResolver, UseMcpToolResolver
+    AttemptCompletionToolResolver, PlanModeRespondToolResolver, UseMcpToolResolver,
+    ListPackageInfoToolResolver
 )
+
 from autocoder.common.v2.agent.agentic_edit_types import (AgenticEditRequest, ToolResult,
                                                           MemoryConfig, CommandConfig, BaseTool,
                                                           ExecuteCommandTool, ReadFileTool,
@@ -63,6 +65,7 @@ from autocoder.common.v2.agent.agentic_edit_types import (AgenticEditRequest, To
                                                           ListFilesTool,
                                                           ListCodeDefinitionNamesTool, AskFollowupQuestionTool,
                                                           AttemptCompletionTool, PlanModeRespondTool, UseMcpTool,
+                                                          ListPackageInfoTool,
                                                           TOOL_MODEL_MAP,
                                                           # Event Types
                                                           LLMOutputEvent, LLMThinkingEvent, ToolCallEvent,
@@ -83,6 +86,7 @@ TOOL_RESOLVER_MAP: Dict[Type[BaseTool], Type[BaseToolResolver]] = {
     SearchFilesTool: SearchFilesToolResolver,
     ListFilesTool: ListFilesToolResolver,
     ListCodeDefinitionNamesTool: ListCodeDefinitionNamesToolResolver,
+    ListPackageInfoTool: ListPackageInfoToolResolver,
     AskFollowupQuestionTool: AskFollowupQuestionToolResolver,
     AttemptCompletionTool: AttemptCompletionToolResolver,  # Will stop the loop anyway
     PlanModeRespondTool: PlanModeRespondToolResolver,
@@ -245,6 +249,7 @@ class AgenticEdit:
 
         Always adhere to this format for the tool use to ensure proper parsing and execution.
 
+
         # Tools
 
         ## execute_command
@@ -258,6 +263,15 @@ class AgenticEdit:
         <requires_approval>true or false</requires_approval>
         </execute_command>
 
+        ## list_package_info
+        Description: Request to retrieve information about a source code package, such as recent changes or documentation summary, to better understand the code context. It accepts a directory path (absolute or relative to the current project).
+        Parameters:
+        - path: (required) The source code package directory path.
+        Usage:
+        <list_package_info>
+        <path>relative/or/absolute/package/path</path>
+        </list_package_info>
+
         ## read_file
         Description: Request to read the contents of a file at the specified path. Use this when you need to examine the contents of an existing file you do not know the contents of, for example to analyze code, review text files, or extract information from configuration files. Automatically extracts raw text from PDF and DOCX files. May not be suitable for other types of binary files, as it returns the raw content as a string.
         Parameters:
@@ -670,7 +684,7 @@ class AgenticEdit:
 
         {% if extra_docs %}
         ====
-        
+
         RULES PROVIDED BY USER
 
         The following rules are provided by the user, and you must follow them strictly.
@@ -683,7 +697,8 @@ class AgenticEdit:
         """
         import os
         extra_docs = {}
-        rules_dir = os.path.join(self.args.source_dir,".auto-coder", "autocoderrules")
+        rules_dir = os.path.join(self.args.source_dir,
+                                 ".auto-coder", "autocoderrules")
         if os.path.isdir(rules_dir):
             for fname in os.listdir(rules_dir):
                 if fname.endswith(".md"):
@@ -1137,6 +1152,7 @@ class AgenticEdit:
         standard event system format and writing them using the event manager.
         """
         event_manager = get_event_manager(self.args.event_file)
+        self.apply_pre_changes()
 
         try:
             event_stream = self.analyze(request)
@@ -1280,6 +1296,25 @@ class AgenticEdit:
             # Re-raise the exception if needed, or handle appropriately
             raise e
 
+    def apply_pre_changes(self):
+        # get the file name
+        file_name = os.path.basename(self.args.file)
+        if not self.args.skip_commit:
+            try:
+                get_event_manager(self.args.event_file).write_result(
+                    EventContentCreator.create_result(
+                        content=self.printer.get_message_from_key("/agent/edit/apply_pre_changes")), metadata=EventMetadata(
+                        action_file=self.args.file,
+                        is_streaming=False,
+                        path="/agent/edit/apply_pre_changes",
+                        stream_out_type="/agent/edit").to_dict())
+                git_utils.commit_changes(
+                    self.args.source_dir, f"auto_coder_pre_{file_name}")
+            except Exception as e:
+                self.printer.print_in_terminal("git_init_required",
+                                               source_dir=self.args.source_dir, error=str(e))
+                return
+
     def apply_changes(self):
         """
         Apply all tracked file changes to the original project directory.
@@ -1297,6 +1332,12 @@ class AgenticEdit:
                         f"{self.args.query}\nauto_coder_{file_name}",
                     )
 
+                    get_event_manager(self.args.event_file).write_result(
+                        EventContentCreator.create_result(
+                            content=self.printer.get_message_from_key("/agent/edit/apply_changes")), metadata=EventMetadata(
+                            action_file=self.args.file,
+                            is_streaming=False,
+                            stream_out_type="/agent/edit").to_dict())
                     action_yml_file_manager = ActionYmlFileManager(
                         self.args.source_dir)
                     action_file_name = os.path.basename(self.args.file)
@@ -1343,6 +1384,7 @@ class AgenticEdit:
             f"[bold]{get_message('/agent/edit/user_query')}:[/bold]\n{request.user_input}", title=get_message("/agent/edit/objective"), border_style="blue"))
 
         try:
+            self.apply_pre_changes()
             event_stream = self.analyze(request)
             for event in event_stream:
                 if isinstance(event, LLMThinkingEvent):
@@ -1434,7 +1476,8 @@ class AgenticEdit:
                             logger.warning(
                                 f"Error formatting tool result content: {e}")
                             panel_content.append(
-                                _format_content(str(result.content)))  # Fallback
+                                # Fallback
+                                _format_content(str(result.content)))
 
                     # Print the base info panel
                     console.print(Panel("\n".join(

autocoder/common/v2/agent/agentic_edit_tools/__init__.py

@@ -11,6 +11,7 @@ from .ask_followup_question_tool_resolver import AskFollowupQuestionToolResolver
 from .attempt_completion_tool_resolver import AttemptCompletionToolResolver
 from .plan_mode_respond_tool_resolver import PlanModeRespondToolResolver
 from .use_mcp_tool_resolver import UseMcpToolResolver
+from .list_package_info_tool_resolver import ListPackageInfoToolResolver
 
 __all__ = [
     "BaseToolResolver",
@@ -25,4 +26,5 @@ __all__ = [
     "AttemptCompletionToolResolver",
     "PlanModeRespondToolResolver",
     "UseMcpToolResolver",
+    "ListPackageInfoToolResolver",
 ]

autocoder/common/v2/agent/agentic_edit_tools/list_package_info_tool_resolver.py

@@ -0,0 +1,42 @@
+
+import os
+from typing import Optional
+from autocoder.common.v2.agent.agentic_edit_tools.base_tool_resolver import BaseToolResolver
+from autocoder.common.v2.agent.agentic_edit_types import ListPackageInfoTool, ToolResult
+from loguru import logger
+import typing
+
+if typing.TYPE_CHECKING:
+    from autocoder.common.v2.agent.agentic_edit import AgenticEdit
+
+class ListPackageInfoToolResolver(BaseToolResolver):
+    def __init__(self, agent: Optional['AgenticEdit'], tool: ListPackageInfoTool, args):
+        super().__init__(agent, tool, args)
+        self.tool: ListPackageInfoTool = tool
+
+    def resolve(self) -> ToolResult:
+        source_dir = self.args.source_dir or "."
+        abs_source_dir = os.path.abspath(source_dir)
+
+        input_path = self.tool.path.strip()
+        abs_input_path = os.path.abspath(os.path.join(source_dir, input_path)) if not os.path.isabs(input_path) else input_path
+
+        # 校验输入目录是否在项目目录内
+        if not abs_input_path.startswith(abs_source_dir):
+            return ToolResult(success=False, message=f"Error: Access denied. Path outside project: {self.tool.path}")
+
+        rel_package_path = os.path.relpath(abs_input_path, abs_source_dir)
+        active_md_path = os.path.join(abs_source_dir, ".auto-coder", "active-context", rel_package_path, "active.md")
+
+        logger.info(f"Looking for package info at: {active_md_path}")
+
+        if not os.path.exists(active_md_path):
+            return ToolResult(success=True, message="No package info found for this path.", content="没有相关包信息。")
+
+        try:
+            with open(active_md_path, 'r', encoding='utf-8', errors='replace') as f:
+                content = f.read()
+            return ToolResult(success=True, message="Successfully retrieved package info.", content=content)
+        except Exception as e:
+            logger.error(f"Error reading package info file: {e}")
+            return ToolResult(success=False, message=f"Error reading package info file: {e}")

autocoder/common/v2/agent/agentic_edit_types.py

@@ -57,6 +57,9 @@ class UseMcpTool(BaseTool):
     tool_name: str
     query:str
 
+class ListPackageInfoTool(BaseTool):
+    path: str  # 源码包目录，相对路径或绝对路径
+
 # Event Types for Rich Output Streaming
 class LLMOutputEvent(BaseModel):
     """Represents plain text output from the LLM."""
@@ -112,6 +115,7 @@ TOOL_MODEL_MAP: Dict[str, Type[BaseTool]] = {
     "attempt_completion": AttemptCompletionTool,
     "plan_mode_respond": PlanModeRespondTool,
     "use_mcp_tool": UseMcpTool,
+    "list_package_info": ListPackageInfoTool,
 }
 
 class FileChangeEntry(BaseModel):

autocoder/rag/cache/byzer_storage_cache.py

@@ -29,6 +29,7 @@ import hashlib
 from typing import Union
 from pydantic import BaseModel
 from autocoder.rag.cache.cache_result_merge import CacheResultMerger, MergeStrategy
+from .failed_files_utils import save_failed_files, load_failed_files
 import time
 
 if platform.system() != "Windows":
@@ -65,74 +66,18 @@ class ByzerStorageCache(BaseCacheManager):
         ignore_spec,
         required_exts,
         extra_params: Optional[AutoCoderArgs] = None,
+        args=None,
+        llm=None,
     ):
         """
         初始化基于云端 Byzer Storage 的 RAG 缓存管理器。
-        
-        参数:
-            path: 需要索引的代码库根目录
-            ignore_spec: 指定哪些文件/目录应被忽略的规则
-            required_exts: 需要处理的文件扩展名列表
-            extra_params: 额外的配置参数，包含向量索引相关设置
-            
-        缓存结构 (self.cache):
-            self.cache 是一个字典，键为文件路径，值为 CacheItem 对象:
-            {
-                "file_path1": CacheItem(
-                    file_path: str,              # 文件的绝对路径
-                    relative_path: str,          # 相对于项目根目录的路径
-                    content: List[Dict],         # 文件内容的结构化表示，每个元素是 SourceCode 对象的序列化
-                    modify_time: float,          # 文件最后修改时间的时间戳
-                    md5: str                     # 文件内容的 MD5 哈希值，用于检测变更
-                ),
-                "file_path2": CacheItem(...),
-                ...
-            }
-            
-            这个缓存有两层存储:
-            1. 本地文件缓存: 保存在项目根目录的 .cache/byzer_storage_speedup.jsonl 文件中
-               - 用于跟踪文件变更和快速加载
-               - 使用 JSONL 格式存储，每行是一个 CacheItem 的 JSON 表示
-            
-            2. 云端 Byzer Storage 向量数据库:
-               - 存储文件内容的分块和向量嵌入
-               - 每个文件被分割成大小为 chunk_size 的文本块
-               - 每个块都会生成向量嵌入，用于语义搜索
-               - 存储结构包含: 文件路径、内容块、原始内容、向量嵌入、修改时间
-        
-        源代码处理流程:
-            在缓存更新过程中使用了两个关键函数:
-            
-            1. process_file_in_multi_process: 在多进程环境中处理文件
-               - 参数: file_info (文件信息元组)
-               - 返回值: List[SourceCode] 或 None
-               - 用途: 在初始构建缓存时并行处理多个文件
-            
-            2. process_file_local: 在当前进程中处理单个文件
-               - 参数: file_path (文件路径)
-               - 返回值: List[SourceCode] 或 None
-               - 用途: 在检测到文件更新时处理单个文件
-            
-            文件处理后，会:
-            1. 更新内存中的缓存 (self.cache)
-            2. 将缓存持久化到本地文件
-            3. 将内容分块并更新到 Byzer Storage 向量数据库
-        
-        更新机制:
-            - 通过单独的线程异步处理文件变更
-            - 使用 MD5 哈希值检测文件是否发生变化
-            - 支持文件添加、更新和删除事件
-            - 使用向量数据库进行语义检索，支持相似度搜索
-            
-        与 LocalByzerStorageCache 的区别:
-            - 使用云端 ByzerStorage 而非本地存储
-            - 适用于需要远程访问和共享索引的场景
-            - 支持大规模分布式检索和更高级的查询功能
         """
         self.path = path
         self.ignore_spec = ignore_spec
         self.required_exts = required_exts
         self.extra_params = extra_params
+        self.args = args
+        self.llm = llm
         self.rag_build_name = extra_params.rag_build_name
         self.storage = ByzerStorage("byzerai_store", "rag", self.rag_build_name)
         self.queue = []
@@ -150,16 +95,20 @@ class ByzerStorageCache(BaseCacheManager):
         self.cache_file = os.path.join(self.cache_dir, "byzer_storage_speedup.jsonl")
         self.cache: Dict[str, CacheItem] = {}
 
+        # 创建缓存目录
+        if not os.path.exists(self.cache_dir):
+            os.makedirs(self.cache_dir)
+
+        # failed files support
+        self.failed_files_path = os.path.join(self.cache_dir, "failed_files.json")
+        self.failed_files = load_failed_files(self.failed_files_path)
+
         self.lock = threading.Lock()
         self.stop_event = threading.Event()
         self.thread = threading.Thread(target=self.process_queue)
         self.thread.daemon = True
         self.thread.start()
 
-        # 创建缓存目录
-        if not os.path.exists(self.cache_dir):
-            os.makedirs(self.cache_dir)
-
         # 加载缓存
         self.cache = self._load_cache()
 
@@ -455,6 +404,10 @@ class ByzerStorageCache(BaseCacheManager):
                 for item in file_list.file_paths:
                     logger.info(f"[QUEUE PROCESSING] Processing file deletion: {item}")
                     del self.cache[item]
+                    # remove from failed files if present
+                    if item in self.failed_files:
+                        self.failed_files.remove(item)
+                        save_failed_files(self.failed_files_path, self.failed_files)
                     # Create a temporary FileInfo object
                     file_info = FileInfo(file_path=item, relative_path="", modify_time=0, file_md5="")
                     self.update_storage(file_info, is_delete=True)
@@ -463,16 +416,29 @@ class ByzerStorageCache(BaseCacheManager):
                 logger.info(f"[QUEUE PROCESSING] Processing add/update event, total files: {len(file_list.file_infos)}")
                 for file_info in file_list.file_infos:
                     logger.info(f"[QUEUE PROCESSING] Processing file update: {file_info.file_path}")
-                    # Process file and create CacheItem
-                    content = process_file_local(self.fileinfo_to_tuple(file_info))
-                    self.cache[file_info.file_path] = CacheItem(
-                        file_path=file_info.file_path,
-                        relative_path=file_info.relative_path,
-                        content=[c.model_dump() for c in content],
-                        modify_time=file_info.modify_time,
-                        md5=file_info.file_md5,
-                    )
-                    self.update_storage(file_info, is_delete=False)
+                    try:
+                        content = process_file_local(self.fileinfo_to_tuple(file_info))
+                        if content:
+                            self.cache[file_info.file_path] = CacheItem(
+                                file_path=file_info.file_path,
+                                relative_path=file_info.relative_path,
+                                content=[c.model_dump() for c in content],
+                                modify_time=file_info.modify_time,
+                                md5=file_info.file_md5,
+                            )
+                            self.update_storage(file_info, is_delete=False)
+                            # remove from failed files if present
+                            if file_info.file_path in self.failed_files:
+                                self.failed_files.remove(file_info.file_path)
+                                save_failed_files(self.failed_files_path, self.failed_files)
+                        else:
+                            logger.warning(f"Empty result for file: {file_info.file_path}, treat as parse failed, skipping cache update")
+                            self.failed_files.add(file_info.file_path)
+                            save_failed_files(self.failed_files_path, self.failed_files)
+                    except Exception as e:
+                        logger.error(f"Error in process_queue: {e}")
+                        self.failed_files.add(file_info.file_path)
+                        save_failed_files(self.failed_files_path, self.failed_files)
             self.write_cache()
         
         elapsed = time.time() - start_time
@@ -486,6 +452,10 @@ class ByzerStorageCache(BaseCacheManager):
         current_files = set()
         for file_info in self.get_all_files():            
             current_files.add(file_info.file_path)
+            # skip failed files
+            if file_info.file_path in self.failed_files:
+                logger.info(f"文件 {file_info.file_path} 之前解析失败，跳过此次更新")
+                continue
             if (
                 file_info.file_path not in self.cache
                 or self.cache[file_info.file_path].md5 != file_info.file_md5

autocoder/rag/cache/failed_files_utils.py

@@ -0,0 +1,39 @@
+
+import os
+import json
+from loguru import logger
+
+def load_failed_files(failed_files_path: str) -> set:
+    """
+    Load the set of failed file paths from a JSON file.
+    
+    Args:
+        failed_files_path: Path to the JSON file storing failed files.
+    
+    Returns:
+        A set of failed file paths.
+    """
+    directory = os.path.dirname(failed_files_path)
+    if not os.path.exists(directory):
+        os.makedirs(directory, exist_ok=True)
+    if os.path.exists(failed_files_path):
+        try:
+            with open(failed_files_path, "r", encoding="utf-8") as f:
+                return set(json.load(f))
+        except Exception:
+            return set()
+    return set()
+
+def save_failed_files(failed_files_path: str, failed_files: set) -> None:
+    """
+    Save the set of failed file paths to a JSON file.
+    
+    Args:
+        failed_files_path: Path to the JSON file.
+        failed_files: A set of failed file paths.
+    """
+    try:
+        with open(failed_files_path, "w", encoding="utf-8") as f:
+            json.dump(list(failed_files), f, ensure_ascii=False, indent=2)
+    except Exception as e:
+        logger.error(f"Error saving failed files list: {e}")

autocoder/rag/cache/file_monitor_cache.py

@@ -48,7 +48,7 @@ class AutoCoderRAGDocListener(BaseCacheManager):
         r"^test.*$",
     ]
 
-    def __init__(self, path: str, ignore_spec, required_exts: List) -> None:
+    def __init__(self, path: str, ignore_spec, required_exts: List, args=None, llm=None) -> None:
         """
         初始化文件监控缓存管理器。
         
@@ -89,6 +89,8 @@ class AutoCoderRAGDocListener(BaseCacheManager):
         self.path = path
         self.ignore_spec = ignore_spec
         self.required_exts = required_exts
+        self.args = args
+        self.llm = llm
         self.stop_event = threading.Event()
 
         # connect list

autocoder/rag/cache/local_byzer_storage_cache.py

@@ -30,6 +30,7 @@ from typing import Union
 from byzerllm import SimpleByzerLLM, ByzerLLM
 from autocoder.rag.cache.cache_result_merge import CacheResultMerger, MergeStrategy
 import time
+from .failed_files_utils import save_failed_files, load_failed_files
 
 if platform.system() != "Windows":
     import fcntl
@@ -70,71 +71,17 @@ class LocalByzerStorageCache(BaseCacheManager):
         emb_llm: Union[ByzerLLM, SimpleByzerLLM] = None,
         host: str = "127.0.0.1",
         port: int = 33333,
+        args=None,
+        llm=None,
     ):
         """
         初始化基于 Byzer Storage 的 RAG 缓存管理器。
-        
-        参数:
-            path: 需要索引的代码库根目录
-            ignore_spec: 指定哪些文件/目录应被忽略的规则
-            required_exts: 需要处理的文件扩展名列表
-            extra_params: 额外的配置参数，包含向量索引相关设置
-            emb_llm: 用于生成文本向量嵌入的 ByzerLLM 实例
-            host: Byzer Storage 服务的主机地址
-            port: Byzer Storage 服务的端口
-            
-        缓存结构 (self.cache):
-            self.cache 是一个字典，键为文件路径，值为 CacheItem 对象:
-            {
-                "file_path1": CacheItem(
-                    file_path: str,              # 文件的绝对路径
-                    relative_path: str,          # 相对于项目根目录的路径
-                    content: List[Dict],         # 文件内容的结构化表示，每个元素是 SourceCode 对象的序列化
-                    modify_time: float,          # 文件最后修改时间的时间戳
-                    md5: str                     # 文件内容的 MD5 哈希值，用于检测变更
-                ),
-                "file_path2": CacheItem(...),
-                ...
-            }
-            
-            这个缓存有两层存储:
-            1. 本地文件缓存: 保存在项目根目录的 .cache/byzer_storage_speedup.jsonl 文件中
-               - 用于跟踪文件变更和快速加载
-               - 使用 JSONL 格式存储，每行是一个 CacheItem 的 JSON 表示
-            
-            2. Byzer Storage 向量数据库:
-               - 存储文件内容的分块和向量嵌入
-               - 每个文件被分割成大小为 chunk_size 的文本块
-               - 每个块都会生成向量嵌入，用于语义搜索
-               - 存储结构包含: 文件路径、内容块、原始内容、向量嵌入、修改时间
-        
-        源代码处理流程:
-            在缓存更新过程中使用了两个关键函数:
-            
-            1. process_file_in_multi_process: 在多进程环境中处理文件
-               - 参数: file_info (文件信息元组)
-               - 返回值: List[SourceCode] 或 None
-               - 用途: 在初始构建缓存时并行处理多个文件
-            
-            2. process_file_local: 在当前进程中处理单个文件
-               - 参数: file_path (文件路径)
-               - 返回值: List[SourceCode] 或 None
-               - 用途: 在检测到文件更新时处理单个文件
-            
-            文件处理后，会:
-            1. 更新内存中的缓存 (self.cache)
-            2. 将缓存持久化到本地文件
-            3. 将内容分块并更新到 Byzer Storage 向量数据库
-        
-        更新机制:
-            - 通过单独的线程异步处理文件变更
-            - 使用 MD5 哈希值检测文件是否发生变化
-            - 支持文件添加、更新和删除事件
-            - 使用向量数据库进行语义检索，支持相似度搜索
         """
         self.path = path
         self.ignore_spec = ignore_spec
         self.required_exts = required_exts
+        self.args = args
+        self.llm = llm
         self.rag_build_name = extra_params.rag_build_name
         self.storage = LocalByzerStorage("byzerai_store",
             "rag_test", self.rag_build_name, host=host, port=port,emb_llm=emb_llm)
@@ -153,16 +100,20 @@ class LocalByzerStorageCache(BaseCacheManager):
             self.cache_dir, "byzer_storage_speedup.jsonl")
         self.cache: Dict[str, CacheItem] = {}
 
+        # 创建缓存目录
+        if not os.path.exists(self.cache_dir):
+            os.makedirs(self.cache_dir)
+
+        # failed files support
+        self.failed_files_path = os.path.join(self.cache_dir, "failed_files.json")
+        self.failed_files = load_failed_files(self.failed_files_path)
+
         self.lock = threading.Lock()
         self.stop_event = threading.Event()
         self.thread = threading.Thread(target=self.process_queue)
         self.thread.daemon = True
         self.thread.start()
 
-        # 创建缓存目录
-        if not os.path.exists(self.cache_dir):
-            os.makedirs(self.cache_dir)
-
         # 加载缓存
         self.cache = self._load_cache()
 
@@ -485,6 +436,10 @@ class LocalByzerStorageCache(BaseCacheManager):
                 for item in file_list.file_paths:
                     logger.info(f"[QUEUE PROCESSING] Processing file deletion: {item}")
                     del self.cache[item]
+                    # remove from failed files if present
+                    if item in self.failed_files:
+                        self.failed_files.remove(item)                        
+                        save_failed_files(self.failed_files_path, self.failed_files)
                     # Create a temporary FileInfo object
                     file_info = FileInfo(
                         file_path=item, relative_path="", modify_time=0, file_md5="")
@@ -495,17 +450,30 @@ class LocalByzerStorageCache(BaseCacheManager):
                 for file_info in file_list.file_infos:
                     logger.info(
                         f"[QUEUE PROCESSING] Processing file update: {file_info.file_path}")
-                    # Process file and create CacheItem
-                    content = process_file_local(
-                        self.fileinfo_to_tuple(file_info))
-                    self.cache[file_info.file_path] = CacheItem(
-                        file_path=file_info.file_path,
-                        relative_path=file_info.relative_path,
-                        content=[c.model_dump() for c in content],
-                        modify_time=file_info.modify_time,
-                        md5=file_info.file_md5,
-                    )
-                    self.update_storage(file_info, is_delete=False)
+                    try:
+                        content = process_file_local(
+                            self.fileinfo_to_tuple(file_info))
+                        if content:
+                            self.cache[file_info.file_path] = CacheItem(
+                                file_path=file_info.file_path,
+                                relative_path=file_info.relative_path,
+                                content=[c.model_dump() for c in content],
+                                modify_time=file_info.modify_time,
+                                md5=file_info.file_md5,
+                            )
+                            self.update_storage(file_info, is_delete=False)
+                            # remove from failed files if present
+                            if file_info.file_path in self.failed_files:
+                                self.failed_files.remove(file_info.file_path)                                
+                                save_failed_files(self.failed_files_path, self.failed_files)
+                        else:
+                            logger.warning(f"Empty result for file: {file_info.file_path}, treat as parse failed, skipping cache update")
+                            self.failed_files.add(file_info.file_path)                            
+                            save_failed_files(self.failed_files_path, self.failed_files)
+                    except Exception as e:
+                        logger.error(f"Error in process_queue: {e}")
+                        self.failed_files.add(file_info.file_path)                        
+                        save_failed_files(self.failed_files_path, self.failed_files)
             self.write_cache()
         
         elapsed = time.time() - start_time
@@ -519,6 +487,10 @@ class LocalByzerStorageCache(BaseCacheManager):
         current_files = set()
         for file_info in self.get_all_files():
             current_files.add(file_info.file_path)
+            # skip failed files
+            if file_info.file_path in self.failed_files:
+                logger.info(f"文件 {file_info.file_path} 之前解析失败，跳过此次更新")
+                continue
             if (
                 file_info.file_path not in self.cache
                 or self.cache[file_info.file_path].md5 != file_info.file_md5

autocoder/rag/cache/local_duckdb_storage_cache.py

@@ -28,6 +28,7 @@ from autocoder.rag.cache.base_cache import (
 from autocoder.rag.utils import process_file_in_multi_process, process_file_local
 from autocoder.rag.variable_holder import VariableHolder
 from byzerllm import SimpleByzerLLM, ByzerLLM
+from .failed_files_utils import save_failed_files, load_failed_files
 
 if platform.system() != "Windows":
     import fcntl
@@ -300,12 +301,16 @@ class LocalDuckDBStorageCache(BaseCacheManager):
             ignore_spec,
             required_exts,
             extra_params: Optional[AutoCoderArgs] = None,
-            emb_llm: Union[ByzerLLM, SimpleByzerLLM] = None
+            emb_llm: Union[ByzerLLM, SimpleByzerLLM] = None,
+            args=None,
+            llm=None
     ):
         self.path = path
         self.ignore_spec = ignore_spec
         self.required_exts = required_exts
         self.extra_params = extra_params
+        self.args = args
+        self.llm = llm
 
         self.storage = LocalDuckdbStorage(
             llm=emb_llm,
@@ -325,6 +330,11 @@ class LocalDuckDBStorageCache(BaseCacheManager):
         if not os.path.exists(self.cache_dir):
             os.makedirs(self.cache_dir)
 
+        # failed files support
+        from .failed_files_utils import load_failed_files
+        self.failed_files_path = os.path.join(self.cache_dir, "failed_files.json")
+        self.failed_files = load_failed_files(self.failed_files_path)
+
         self.lock = threading.Lock()
         self.stop_event = threading.Event()
         self.thread = threading.Thread(target=self.process_queue)
@@ -569,6 +579,10 @@ class LocalDuckDBStorageCache(BaseCacheManager):
                 for item in file_list.file_paths:
                     logger.info(f"{item} is detected to be removed")
                     del self.cache[item]
+                    # remove from failed files if present
+                    if item in self.failed_files:
+                        self.failed_files.remove(item)
+                        save_failed_files(self.failed_files_path, self.failed_files)
                     # 创建一个临时的 FileInfo 对象
                     file_info = FileInfo(
                         file_path=item, relative_path="", modify_time=0, file_md5="")
@@ -578,18 +592,30 @@ class LocalDuckDBStorageCache(BaseCacheManager):
                 for file_info in file_list.file_infos:
                     logger.info(
                         f"{file_info.file_path} is detected to be updated")
-                    # 处理文件并创建 CacheItem
-                    # content = process_file_local(
-                    #     self.fileinfo_to_tuple(file_info))
-                    content = process_file_local(file_info.file_path)
-                    self.cache[file_info.file_path] = CacheItem(
-                        file_path=file_info.file_path,
-                        relative_path=file_info.relative_path,
-                        content=[c.model_dump() for c in content],
-                        modify_time=file_info.modify_time,
-                        md5=file_info.file_md5,
-                    )
-                    self.update_storage(file_info, is_delete=False)
+                    try:
+                        content = process_file_local(file_info.file_path)
+                        if content:
+                            self.cache[file_info.file_path] = CacheItem(
+                                file_path=file_info.file_path,
+                                relative_path=file_info.relative_path,
+                                content=[c.model_dump() for c in content],
+                                modify_time=file_info.modify_time,
+                                md5=file_info.file_md5,
+                            )
+                            self.update_storage(file_info, is_delete=False)
+                            # remove from failed files if present
+                            if file_info.file_path in self.failed_files:
+                                self.failed_files.remove(file_info.file_path)
+                                save_failed_files(self.failed_files_path, self.failed_files)
+                        else:
+                            logger.warning(f"Empty result for file: {file_info.file_path}, treat as parse failed, skipping cache update")
+                            self.failed_files.add(file_info.file_path)
+                            save_failed_files(self.failed_files_path, self.failed_files)
+                    except Exception as e:
+                        logger.error(f"Error in process_queue: {e}")
+                        self.failed_files.add(file_info.file_path)
+                        save_failed_files(self.failed_files_path, self.failed_files)
+
             self.write_cache()
 
     def trigger_update(self):
@@ -598,6 +624,10 @@ class LocalDuckDBStorageCache(BaseCacheManager):
         current_files = set()
         for file_info in self.get_all_files():
             current_files.add(file_info.file_path)
+            # skip failed files
+            if file_info.file_path in self.failed_files:
+                logger.info(f"文件 {file_info.file_path} 之前解析失败，跳过此次更新")
+                continue
             if (
                     file_info.file_path not in self.cache
                     or self.cache[file_info.file_path].md5 != file_info.file_md5

autocoder/rag/cache/simple_cache.py

@@ -19,6 +19,7 @@ from loguru import logger
 from autocoder.rag.utils import process_file_in_multi_process, process_file_local
 from autocoder.rag.variable_holder import VariableHolder
 import hashlib
+from .failed_files_utils import load_failed_files, save_failed_files
 
 
 default_ignore_dirs = [
@@ -45,7 +46,7 @@ def generate_content_md5(content: Union[str, bytes]) -> str:
 
 
 class AutoCoderRAGAsyncUpdateQueue(BaseCacheManager):
-    def __init__(self, path: str, ignore_spec, required_exts: list, update_interval: int = 5):
+    def __init__(self, path: str, ignore_spec, required_exts: list, update_interval: int = 5, args=None, llm=None):
         """
         初始化异步更新队列，用于管理代码文件的缓存。
         
@@ -91,24 +92,31 @@ class AutoCoderRAGAsyncUpdateQueue(BaseCacheManager):
         self.path = path
         self.ignore_spec = ignore_spec
         self.required_exts = required_exts
+        self.args = args
+        self.llm = llm
         self.update_interval = update_interval
         self.queue = []
         self.cache = {}  # 初始化为空字典，稍后通过 read_cache() 填充
         self.lock = threading.Lock()
         self.stop_event = threading.Event()
-        
+
+        # 用于存放解析失败的文件路径集合
+        self.failed_files_path = os.path.join(self.path, ".cache", "failed_files.json")
+        self.failed_files = load_failed_files(self.failed_files_path)
+
         # 启动处理队列的线程
         self.queue_thread = threading.Thread(target=self._process_queue)
         self.queue_thread.daemon = True
         self.queue_thread.start()
-        
+
         # 启动定时触发更新的线程
         self.update_thread = threading.Thread(target=self._periodic_update)
         self.update_thread.daemon = True
         self.update_thread.start()
-        
+
         self.cache = self.read_cache()
 
+
     def _process_queue(self):
         while not self.stop_event.is_set():
             try:
@@ -183,13 +191,18 @@ class AutoCoderRAGAsyncUpdateQueue(BaseCacheManager):
         files_to_process = []
         current_files = set()
         for file_info in self.get_all_files():
-            file_path, _, _, file_md5 = file_info
+            file_path, relative_path, modify_time, file_md5 = file_info
             current_files.add(file_path)
+            # 如果文件曾经解析失败，跳过本次增量更新
+            if file_path in self.failed_files:
+                logger.info(f"文件 {file_path} 之前解析失败，跳过此次更新")
+                continue
+            # 变更检测
             if (
                 file_path not in self.cache
-                or self.cache[file_path].get("md5","") != file_md5
+                or self.cache[file_path].get("md5", "") != file_md5
             ):
-                files_to_process.append(file_info)
+                files_to_process.append((file_path, relative_path, modify_time, file_md5))
 
         deleted_files = set(self.cache.keys()) - current_files
         logger.info(f"files_to_process: {files_to_process}")
@@ -213,19 +226,34 @@ class AutoCoderRAGAsyncUpdateQueue(BaseCacheManager):
             if isinstance(file_list, DeleteEvent):
                 for item in file_list.file_paths:
                     logger.info(f"{item} is detected to be removed")
-                    del self.cache[item]
+                    if item in self.cache:
+                        del self.cache[item]
+                    # 删除时也从失败列表中移除（防止文件已修复）
+                    if item in self.failed_files:
+                        self.failed_files.remove(item)
+                        save_failed_files(self.failed_files_path, self.failed_files)
             elif isinstance(file_list, AddOrUpdateEvent):
                 for file_info in file_list.file_infos:
                     logger.info(f"{file_info.file_path} is detected to be updated")
                     try:
                         result = process_file_local(file_info.file_path)
-                        if result:  # 只有当result不为空时才更新缓存
+                        if result:
+                            # 解析成功且非空
                             self.update_cache(self.fileinfo_to_tuple(file_info), result)
+                            # 如果之前失败过且本次成功，移除失败记录
+                            if file_info.file_path in self.failed_files:
+                                self.failed_files.remove(file_info.file_path)
+                                save_failed_files(self.failed_files_path, self.failed_files)
                         else:
-                            logger.warning(f"Empty result for file: {file_info.file_path}, skipping cache update")
+                            # 只要为空也认为解析失败，加入失败列表
+                            logger.warning(f"Empty result for file: {file_info.file_path}, treat as parse failed, skipping cache update")
+                            self.failed_files.add(file_info.file_path)
+                            save_failed_files(self.failed_files_path, self.failed_files)
                     except Exception as e:
-                        logger.error(
-                            f"SimpleCache Error in process_queue: {e}")
+                        logger.error(f"SimpleCache Error in process_queue: {e}")
+                        # 解析失败则加入失败列表
+                        self.failed_files.add(file_info.file_path)
+                        save_failed_files(self.failed_files_path, self.failed_files)
 
             self.write_cache()
 

autocoder/rag/document_retriever.py

@@ -2,7 +2,6 @@ import threading
 from typing import Dict, Generator, List, Tuple, Any, Optional,Union
 
 from byzerllm import ByzerLLM, SimpleByzerLLM
-
 from loguru import logger
 from autocoder.common import SourceCode
 from uuid import uuid4
@@ -37,6 +36,8 @@ class LocalDocumentRetriever(BaseDocumentRetriever):
 
     def __init__(
         self,
+        args: AutoCoderArgs,
+        llm: Union[ByzerLLM,SimpleByzerLLM],
         path: str,
         ignore_spec,
         required_exts: list,
@@ -45,9 +46,12 @@ class LocalDocumentRetriever(BaseDocumentRetriever):
         single_file_token_limit: int = 60000,
         disable_auto_window: bool = False,
         enable_hybrid_index: bool = False,
-        extra_params: Optional[AutoCoderArgs] = None,
-        emb_llm: Union[ByzerLLM, SimpleByzerLLM] = None,
+        extra_params: Optional['AutoCoderArgs'] = None,
+        emb_llm: Union['ByzerLLM', 'SimpleByzerLLM'] = None,
     ) -> None:
+        self.args = args
+        self.llm = llm
+
         self.path = path
         self.ignore_spec = ignore_spec
         self.required_exts = required_exts
@@ -65,27 +69,32 @@ class LocalDocumentRetriever(BaseDocumentRetriever):
         if self.enable_hybrid_index:
             if self.on_ray:
                 self.cacher = ByzerStorageCache(
-                    path, ignore_spec, required_exts, extra_params
+                    path, ignore_spec, required_exts, extra_params,
+                    args=self.args, llm=self.llm
                 )
             else:
                 if extra_params.rag_storage_type == "duckdb":
                     self.cacher = LocalDuckDBStorageCache(
                         path, ignore_spec, required_exts, extra_params,
-                        emb_llm=emb_llm
+                        emb_llm=emb_llm,
+                        args=self.args, llm=self.llm
                     )
                 elif extra_params.rag_storage_type in ["byzer-storage", "byzer_storage"]:
                     self.cacher = LocalByzerStorageCache(
                         path, ignore_spec, required_exts, extra_params,
-                        emb_llm=emb_llm
+                        emb_llm=emb_llm,
+                        args=self.args, llm=self.llm
                     )
         else:
             if self.monitor_mode:
                 self.cacher = AutoCoderRAGDocListener(
-                    path, ignore_spec, required_exts
+                    path, ignore_spec, required_exts,
+                    args=self.args, llm=self.llm
                 )
             else:
                 self.cacher = AutoCoderRAGAsyncUpdateQueue(
-                    path, ignore_spec, required_exts
+                    path, ignore_spec, required_exts,
+                    args=self.args, llm=self.llm
                 )
 
         logger.info(f"DocumentRetriever initialized with:")

autocoder/rag/long_context_rag.py

@@ -183,6 +183,8 @@ class LongContextRAG:
                     "emb_llm is required for local byzer storage cache")
 
         self.document_retriever = retriever_class(
+            self.args,
+            self.llm,
             self.path,
             self.ignore_spec,
             self.required_exts,

autocoder/rag/qa_conversation_strategy.py

@@ -92,18 +92,6 @@ class MultiRoundStrategy(QAConversationStrategy):
         {% endfor %}
         </documents>
 
-        ====
-        
-        {% if extra_docs %}
-        AUTO EXTENSION DOCS
-
-        The following extension documents are loaded dynamically to enhance your understanding or provide special instructions, rules, or context.
-
-        {% for key, value in extra_docs.items() %}
-        ### {{ key }}
-        {{ value }}
-        {% endfor %}
-
         ====
         {% endif %}
 
@@ -124,6 +112,19 @@ class MultiRoundStrategy(QAConversationStrategy):
         - Format your answer with Markdown for readability.                
         - Always use the language used by the user in their question.
 
+        {% if extra_docs %}
+        ====
+
+        RULES PROVIDED BY USER
+
+        The following rules are provided by the user, and you must follow them strictly.
+
+        {% for key, value in extra_docs.items() %}
+        ### {{ key }}
+        {{ value }}
+        {% endfor %}        
+        {% endif %}
+
         """
 
         import os
@@ -213,17 +214,6 @@ class SingleRoundStrategy(QAConversationStrategy):
         {% endfor %}
         </documents>
 
-        ====
-        {% if extra_docs %}
-        AUTO EXTENSION DOCS
-
-        The following extension documents are loaded dynamically to enhance your understanding or provide special instructions, rules, or context.
-
-        {% for key, value in extra_docs.items() %}
-        ### {{ key }}
-        {{ value }}
-        {% endfor %}
-
         ====
         {% endif %}
 
@@ -252,6 +242,19 @@ class SingleRoundStrategy(QAConversationStrategy):
         - Format your answer with Markdown for readability.        
         - Always use the language used by the user in their question.
 
+        {% if extra_docs %}
+        ====
+
+        RULES PROVIDED BY USER
+
+        The following rules are provided by the user, and you must follow them strictly.
+
+        {% for key, value in extra_docs.items() %}
+        ### {{ key }}
+        {{ value }}
+        {% endfor %}        
+        {% endif %}
+
         """
         import os
         extra_docs = {}

autocoder/version.py

@@ -1 +1 @@
-__version__ = "0.1.346"
+__version__ = "0.1.347"