auto-coder 0.1.355__py3-none-any.whl → 0.1.357__py3-none-any.whl

autocoder/common/command_completer_v2.py

@@ -41,6 +41,7 @@ COMMAND_HIERARCHY = {
     "/clear": {},
     "/cls": {},
     "/debug": {},
+    "/rules": {"/list", "/get", "/remove", "/analyze", "/commit", "/help"},
 }
 
 class CommandCompleterV2(Completer):
@@ -86,6 +87,7 @@ class CommandCompleterV2(Completer):
             "/ask": self._handle_text_with_symbols, # Treat like chat for @/@@
             "/summon": self._handle_text_with_symbols,
             "/design": self._handle_design,
+            "/rules": self._handle_rules,
             # Add handlers for other commands if they need specific logic beyond @/@@
             # Default handler for plain text or commands not explicitly handled
             "default": self._handle_text_with_symbols,
@@ -438,6 +440,43 @@ class CommandCompleterV2(Completer):
         elif word.startswith("<"): # Potential tag completion
              yield from self._handle_img_tag(document, complete_event, word, text)
 
+    def _handle_rules(self, document: Document, complete_event: CompleteEvent, word: str, text: str) -> Iterable[Completion]:
+        """处理 /rules 命令的补全，支持子命令和规则文件路径。同时支持 @ 和 @@ 符号。"""
+        args_text = text[len("/rules"):].lstrip()
+        parts = args_text.split()
+        last_part = parts[-1] if parts and not text.endswith(" ") else ""
+
+        # 补全子命令
+        if not args_text or (len(parts) == 1 and not text.endswith(" ") and parts[0].startswith("/")):
+            for sub_cmd in COMMAND_HIERARCHY["/rules"]:
+                if sub_cmd.startswith(last_part):
+                    yield Completion(sub_cmd, start_position=-len(last_part))
+            return
+
+        # 根据子命令补全参数
+        if parts and parts[0] == "/list" or parts[0] == "/get" or parts[0] == "/remove":
+            # 获取规则文件或目录补全，可以是通配符
+            # 这里可以简单地提供文件路径补全
+            yield from self._complete_file_paths(last_part, text)
+            # 也可以添加常用通配符补全
+            common_patterns = ["*.md", "*.rules", "*.txt"]
+            for pattern in common_patterns:
+                if pattern.startswith(last_part):
+                    yield Completion(pattern, start_position=-len(last_part))
+            return
+
+        # 对于 /commit 子命令，补全 /query
+        if parts and parts[0] == "/commit":
+            if "/query".startswith(last_part):
+                yield Completion("/query", start_position=-len(last_part))
+            return
+
+        # 支持 @ 和 @@ 符号的补全，不管当前命令是什么
+        if word.startswith("@") and not word.startswith("@@"):
+            yield from self._handle_at_completion(document, complete_event, word, text)
+        elif word.startswith("@@"):
+            yield from self._handle_double_at_completion(document, complete_event, word, text)
+
 
     # --- Symbol/Tag Handlers ---
     def _handle_at_completion(self, document: Document, complete_event: CompleteEvent, word: str, text: str) -> Iterable[Completion]:

autocoder/common/rulefiles/autocoderrules_utils.py

@@ -4,6 +4,9 @@ from threading import Lock
 import threading
 from typing import Dict, List, Optional
 from loguru import logger
+import re
+import yaml
+from pydantic import BaseModel, Field
 
 # 尝试导入 FileMonitor
 try:
@@ -15,6 +18,15 @@ except ImportError:
     Change = None
 
 
+class RuleFile(BaseModel):
+    """规则文件的Pydantic模型"""
+    description: str = Field(default="", description="规则的描述")
+    globs: List[str] = Field(default_factory=list, description="文件匹配模式列表")
+    always_apply: bool = Field(default=False, alias="alwaysApply", description="是否总是应用规则")
+    content: str = Field(default="", description="规则文件的正文内容")
+    file_path: str = Field(default="", description="规则文件的路径")
+
+
 class AutocoderRulesManager:
     """
     管理和监控 autocoderrules 目录中的规则文件。
@@ -157,9 +169,66 @@ class AutocoderRulesManager:
             self._load_rules()
             logger.info("已重新加载规则")
 
+    def parse_rule_file(self, file_path: str) -> RuleFile:
+        """
+        解析规则文件并返回结构化的Pydantic模型对象
+        
+        Args:
+            file_path: 规则文件的路径
+            
+        Returns:
+            RuleFile: 包含规则文件结构化内容的Pydantic模型
+        """
+        if not os.path.exists(file_path) or not file_path.endswith('.md'):
+            logger.warning(f"无效的规则文件路径: {file_path}")
+            return RuleFile(file_path=file_path)
+            
+        try:
+            with open(file_path, 'r', encoding='utf-8') as f:
+                content = f.read()
+                
+            # 解析YAML头部和Markdown内容
+            yaml_pattern = re.compile(r'^---\s*\n(.*?)\n---\s*\n', re.DOTALL)
+            yaml_match = yaml_pattern.search(content)
+            
+            metadata = {}
+            markdown_content = content
+            
+            if yaml_match:
+                yaml_content = yaml_match.group(1)
+                try:
+                    metadata = yaml.safe_load(yaml_content)
+                    # 移除YAML部分，仅保留Markdown内容
+                    markdown_content = content[yaml_match.end():]
+                except Exception as e:
+                    logger.warning(f"解析规则文件YAML头部时出错: {e}")
+            
+            # 创建并返回Pydantic模型
+            rule = RuleFile(
+                description=metadata.get('description', ''),
+                globs=metadata.get('globs', []),
+                always_apply=metadata.get('alwaysApply', False),
+                content=markdown_content.strip(),
+                file_path=file_path
+            )
+            
+            return rule
+            
+        except Exception as e:
+            logger.warning(f"解析规则文件时出错: {file_path}, 错误: {e}")
+            return RuleFile(file_path=file_path)
+
     def get_rules(self) -> Dict[str, str]:
         """获取所有规则文件内容"""
         return self._rules.copy()
+        
+    def get_parsed_rules(self) -> List[RuleFile]:
+        """获取所有解析后的规则文件"""
+        parsed_rules = []
+        for file_path in self._rules:
+            parsed_rule = self.parse_rule_file(file_path)
+            parsed_rules.append(parsed_rule)
+        return parsed_rules
 
 
 # 对外提供单例
@@ -171,3 +240,17 @@ def get_rules(project_root: Optional[str] = None) -> Dict[str, str]:
     if _rules_manager is None:
         _rules_manager = AutocoderRulesManager(project_root=project_root)
     return _rules_manager.get_rules()
+
+def get_parsed_rules(project_root: Optional[str] = None) -> List[RuleFile]:
+    """获取所有解析后的规则文件，可指定项目根目录"""
+    global _rules_manager
+    if _rules_manager is None:
+        _rules_manager = AutocoderRulesManager(project_root=project_root)
+    return _rules_manager.get_parsed_rules()
+
+def parse_rule_file(file_path: str, project_root: Optional[str] = None) -> RuleFile:
+    """解析指定的规则文件，可指定项目根目录"""
+    global _rules_manager
+    if _rules_manager is None:
+        _rules_manager = AutocoderRulesManager(project_root=project_root)
+    return _rules_manager.parse_rule_file(file_path)

autocoder/index/index.py

@@ -487,6 +487,7 @@ class IndexManager:
                 keys_to_remove.append(file_path)
 
         # 删除被排除的文件
+        exclude_patterns = []
         try:
             exclude_patterns = self.parse_exclude_files(
                 self.args.exclude_files)
@@ -498,7 +499,7 @@ class IndexManager:
                 "index_exclude_files_error",
                 style="red",
                 error=str(e)
-            )
+            )            
 
         # 删除无效条目并记录日志
         for key in set(keys_to_remove):
@@ -528,6 +529,9 @@ class IndexManager:
                 if self.should_skip(file_path):
                     continue
 
+                if self.filter_exclude_files(file_path, exclude_patterns):
+                    continue
+
                 source_code = source.source_code
                 if self.args.auto_merge == "strict_diff":
                     v = source.source_code.splitlines()

autocoder/rag/loaders/image_loader.py

@@ -33,6 +33,9 @@ class ImageLoader:
     and converting the content to markdown format.
     """
     
+    # 存储不同参数组合的PaddleOCR实例
+    _ocr_instances = {}
+    
     @staticmethod
     def parse_diff(diff_content: str) -> List[Tuple[str, str]]:
         """
@@ -106,19 +109,28 @@ class ImageLoader:
             print("paddleocr not installed")
             return ""
 
-        # 初始化 OCR
-        try:
-            ocr = PaddleOCR(
-                use_angle_cls=use_angle_cls,
-                lang=lang,
-                page_num=page_num,
-                det_model_dir=det_model_dir,
-                rec_model_dir=rec_model_dir,
-                **kwargs
-            )
-        except Exception:
-            traceback.print_exc()
-            return ""
+        # 创建一个参数的哈希键，用于在缓存中存储OCR实例
+        param_key = f"{lang}_{use_angle_cls}_{page_num}_{det_model_dir}_{rec_model_dir}_{hash(frozenset(kwargs.items()) if kwargs else 0)}"
+        
+        # 检查是否已经有对应参数的OCR实例
+        if param_key not in ImageLoader._ocr_instances:
+            try:
+                # 初始化OCR并缓存
+                ImageLoader._ocr_instances[param_key] = PaddleOCR(
+                    use_angle_cls=use_angle_cls,
+                    lang=lang,
+                    page_num=page_num,
+                    det_model_dir=det_model_dir,
+                    rec_model_dir=rec_model_dir,
+                    **kwargs
+                )
+                logger.info(f"初始化新的PaddleOCR实例，参数：{param_key}")
+            except Exception:
+                traceback.print_exc()
+                return ""
+        
+        # 使用缓存的OCR实例
+        ocr = ImageLoader._ocr_instances[param_key]
 
         try:
             ext = os.path.splitext(file_path)[1].lower()

autocoder/rag/long_context_rag.py

@@ -27,7 +27,7 @@ from autocoder.rag.relevant_utils import (
     DocFilterResult
 )
 from autocoder.rag.token_checker import check_token_limit
-from autocoder.rag.token_counter import RemoteTokenCounter, TokenCounter
+from autocoder.rag.token_counter import RemoteTokenCounter, TokenCounter,count_tokens
 from autocoder.rag.token_limiter import TokenLimiter
 from tokenizers import Tokenizer
 from autocoder.rag.variable_holder import VariableHolder
@@ -782,7 +782,7 @@ class LongContextRAG:
                 )
 
             # 记录令牌统计
-            request_tokens = sum([doc.tokens for doc in relevant_docs])
+            request_tokens = sum([count_tokens(doc.source_code) for doc in relevant_docs])
             target_model = target_llm.default_model_name
             logger.info(
                 f"=== LLM Request ===\n"

autocoder/utils/model_provider_selector.py

@@ -26,7 +26,7 @@ PROVIDER_INFO_LIST = [
         name="volcano",
         endpoint="https://ark.cn-beijing.volces.com/api/v3",   
         r1_model="deepseek-r1-250120",
-        v3_model="deepseek-v3-241226",
+        v3_model="deepseek-v3-250324",
         api_key="",
         r1_input_price=2.0,
         r1_output_price=8.0,
@@ -55,6 +55,17 @@ PROVIDER_INFO_LIST = [
         v3_input_price=2.0,
         v3_output_price=8.0,
     ),
+    ProviderInfo(
+        name="openrouter",
+        endpoint="https://openrouter.ai/api/v1",
+        r1_model="deepseek/deepseek-r1",
+        v3_model="deepseek/deepseek-chat-v3-0324",
+        api_key="",
+        r1_input_price=0.0,
+        r1_output_price=0.0,
+        v3_input_price=0.0,
+        v3_output_price=0.0,
+    )
 ]
 
 dialog_style = Style.from_dict({
@@ -147,7 +158,8 @@ class ModelProviderSelector:
             values=[
                 ("volcano", self.printer.get_message_from_key("model_provider_volcano")),
                 ("siliconflow", self.printer.get_message_from_key("model_provider_siliconflow")),
-                ("deepseek", self.printer.get_message_from_key("model_provider_deepseek"))
+                ("deepseek", self.printer.get_message_from_key("model_provider_deepseek")),
+                ("openrouter", self.printer.get_message_from_key("model_provider_openrouter"))
             ],
             style=dialog_style
         ).run()

autocoder/version.py

@@ -1 +1 @@
-__version__ = "0.1.355"
+__version__ = "0.1.357"