crawlo 1.4.2__py3-none-any.whl → 1.4.3__py3-none-any.whl

crawlo/initialization/phases.py

@@ -1,149 +1,149 @@
-#!/usr/bin/python
-# -*- coding: UTF-8 -*-
-"""
-初始化阶段定义
-"""
-
-from enum import Enum
-from dataclasses import dataclass
-from typing import List, Optional
-
-
-class InitializationPhase(Enum):
-    """初始化阶段枚举"""
-    
-    # 阶段0：准备阶段
-    PREPARING = "preparing"
-    
-    # 阶段1：日志系统初始化
-    LOGGING = "logging"
-    
-    # 阶段2：配置系统初始化  
-    SETTINGS = "settings"
-    
-    # 阶段3：核心组件初始化
-    CORE_COMPONENTS = "core_components"
-    
-    # 阶段4：扩展组件初始化
-    EXTENSIONS = "extensions"
-    
-    # 阶段5：框架启动日志记录
-    FRAMEWORK_STARTUP_LOG = "framework_startup_log"
-    
-    # 阶段6：完成
-    COMPLETED = "completed"
-    
-    # 错误状态
-    ERROR = "error"
-
-
-@dataclass
-class PhaseResult:
-    """阶段执行结果"""
-    phase: InitializationPhase
-    success: bool
-    duration: float = 0.0
-    error: Optional[Exception] = None
-    artifacts: dict = None  # 阶段产生的工件
-    
-    def __post_init__(self):
-        if self.artifacts is None:
-            self.artifacts = {}
-
-
-@dataclass 
-class PhaseDefinition:
-    """阶段定义"""
-    phase: InitializationPhase
-    name: str
-    description: str
-    dependencies: List[InitializationPhase] = None
-    optional: bool = False
-    timeout: float = 30.0  # 超时时间（秒）
-    
-    def __post_init__(self):
-        if self.dependencies is None:
-            self.dependencies = []
-
-
-# 预定义的初始化阶段
-PHASE_DEFINITIONS = [
-    PhaseDefinition(
-        phase=InitializationPhase.PREPARING,
-        name="准备阶段",
-        description="初始化基础环境和检查前置条件",
-        dependencies=[],
-        timeout=5.0
-    ),
-    PhaseDefinition(
-        phase=InitializationPhase.LOGGING,
-        name="日志系统",
-        description="配置和初始化日志系统",
-        dependencies=[],  # 移除对PREPARING的依赖
-        timeout=10.0
-    ),
-    PhaseDefinition(
-        phase=InitializationPhase.SETTINGS,
-        name="配置系统", 
-        description="加载和验证配置",
-        dependencies=[InitializationPhase.LOGGING],
-        timeout=15.0
-    ),
-    PhaseDefinition(
-        phase=InitializationPhase.CORE_COMPONENTS,
-        name="核心组件",
-        description="初始化框架核心组件",
-        dependencies=[InitializationPhase.SETTINGS],
-        timeout=20.0
-    ),
-    PhaseDefinition(
-        phase=InitializationPhase.EXTENSIONS,
-        name="扩展组件",
-        description="加载和初始化扩展组件",
-        dependencies=[InitializationPhase.CORE_COMPONENTS],
-        optional=True,
-        timeout=15.0
-    ),
-    PhaseDefinition(
-        phase=InitializationPhase.FRAMEWORK_STARTUP_LOG,
-        name="框架启动日志",
-        description="记录框架启动相关信息",
-        dependencies=[InitializationPhase.LOGGING, InitializationPhase.SETTINGS],
-        timeout=5.0
-    ),
-    PhaseDefinition(
-        phase=InitializationPhase.COMPLETED,
-        name="初始化完成",
-        description="框架初始化完成",
-        dependencies=[
-            InitializationPhase.CORE_COMPONENTS,
-            InitializationPhase.FRAMEWORK_STARTUP_LOG
-        ],  # Extensions是可选的
-        timeout=5.0
-    )
-]
-
-
-def get_phase_definition(phase: InitializationPhase) -> Optional[PhaseDefinition]:
-    """获取阶段定义"""
-    for definition in PHASE_DEFINITIONS:
-        if definition.phase == phase:
-            return definition
-    return None
-
-
-def get_execution_order() -> List[InitializationPhase]:
-    """获取执行顺序"""
-    return [definition.phase for definition in PHASE_DEFINITIONS]
-
-
-def validate_dependencies() -> bool:
-    """验证阶段依赖关系的正确性"""
-    phases = {definition.phase for definition in PHASE_DEFINITIONS}
-    
-    for definition in PHASE_DEFINITIONS:
-        for dependency in definition.dependencies:
-            if dependency not in phases:
-                return False
-    
+#!/usr/bin/python
+# -*- coding: UTF-8 -*-
+"""
+初始化阶段定义
+"""
+
+from enum import Enum
+from dataclasses import dataclass
+from typing import List, Optional
+
+
+class InitializationPhase(Enum):
+    """初始化阶段枚举"""
+    
+    # 阶段0：准备阶段
+    PREPARING = "preparing"
+    
+    # 阶段1：日志系统初始化
+    LOGGING = "logging"
+    
+    # 阶段2：配置系统初始化  
+    SETTINGS = "settings"
+    
+    # 阶段3：核心组件初始化
+    CORE_COMPONENTS = "core_components"
+    
+    # 阶段4：扩展组件初始化
+    EXTENSIONS = "extensions"
+    
+    # 阶段5：框架启动日志记录
+    FRAMEWORK_STARTUP_LOG = "framework_startup_log"
+    
+    # 阶段6：完成
+    COMPLETED = "completed"
+    
+    # 错误状态
+    ERROR = "error"
+
+
+@dataclass
+class PhaseResult:
+    """阶段执行结果"""
+    phase: InitializationPhase
+    success: bool
+    duration: float = 0.0
+    error: Optional[Exception] = None
+    artifacts: dict = None  # 阶段产生的工件
+    
+    def __post_init__(self):
+        if self.artifacts is None:
+            self.artifacts = {}
+
+
+@dataclass 
+class PhaseDefinition:
+    """阶段定义"""
+    phase: InitializationPhase
+    name: str
+    description: str
+    dependencies: List[InitializationPhase] = None
+    optional: bool = False
+    timeout: float = 30.0  # 超时时间（秒）
+    
+    def __post_init__(self):
+        if self.dependencies is None:
+            self.dependencies = []
+
+
+# 预定义的初始化阶段
+PHASE_DEFINITIONS = [
+    PhaseDefinition(
+        phase=InitializationPhase.PREPARING,
+        name="准备阶段",
+        description="初始化基础环境和检查前置条件",
+        dependencies=[],
+        timeout=5.0
+    ),
+    PhaseDefinition(
+        phase=InitializationPhase.LOGGING,
+        name="日志系统",
+        description="配置和初始化日志系统",
+        dependencies=[],  # 移除对PREPARING的依赖
+        timeout=10.0
+    ),
+    PhaseDefinition(
+        phase=InitializationPhase.SETTINGS,
+        name="配置系统", 
+        description="加载和验证配置",
+        dependencies=[InitializationPhase.LOGGING],
+        timeout=15.0
+    ),
+    PhaseDefinition(
+        phase=InitializationPhase.CORE_COMPONENTS,
+        name="核心组件",
+        description="初始化框架核心组件",
+        dependencies=[InitializationPhase.SETTINGS],
+        timeout=20.0
+    ),
+    PhaseDefinition(
+        phase=InitializationPhase.EXTENSIONS,
+        name="扩展组件",
+        description="加载和初始化扩展组件",
+        dependencies=[InitializationPhase.CORE_COMPONENTS],
+        optional=True,
+        timeout=15.0
+    ),
+    PhaseDefinition(
+        phase=InitializationPhase.FRAMEWORK_STARTUP_LOG,
+        name="框架启动日志",
+        description="记录框架启动相关信息",
+        dependencies=[InitializationPhase.LOGGING, InitializationPhase.SETTINGS],
+        timeout=5.0
+    ),
+    PhaseDefinition(
+        phase=InitializationPhase.COMPLETED,
+        name="初始化完成",
+        description="框架初始化完成",
+        dependencies=[
+            InitializationPhase.CORE_COMPONENTS,
+            InitializationPhase.FRAMEWORK_STARTUP_LOG
+        ],  # Extensions是可选的
+        timeout=5.0
+    )
+]
+
+
+def get_phase_definition(phase: InitializationPhase) -> Optional[PhaseDefinition]:
+    """获取阶段定义"""
+    for definition in PHASE_DEFINITIONS:
+        if definition.phase == phase:
+            return definition
+    return None
+
+
+def get_execution_order() -> List[InitializationPhase]:
+    """获取执行顺序"""
+    return [definition.phase for definition in PHASE_DEFINITIONS]
+
+
+def validate_dependencies() -> bool:
+    """验证阶段依赖关系的正确性"""
+    phases = {definition.phase for definition in PHASE_DEFINITIONS}
+    
+    for definition in PHASE_DEFINITIONS:
+        for dependency in definition.dependencies:
+            if dependency not in phases:
+                return False
+    
     return True

crawlo/initialization/registry.py

@@ -1,146 +1,146 @@
-#!/usr/bin/python
-# -*- coding: UTF-8 -*-
-"""
-初始化器注册表 - 管理所有初始化器的注册和执行
-"""
-
-import threading
-from typing import Dict, Optional, Callable, List
-from .context import InitializationContext
-from .phases import InitializationPhase, PhaseResult
-
-
-class Initializer:
-    """初始化器基类"""
-    
-    def __init__(self, phase: InitializationPhase):
-        self._phase = phase
-    
-    @property
-    def phase(self) -> InitializationPhase:
-        """获取初始化阶段"""
-        return self._phase
-    
-    def initialize(self, context: InitializationContext) -> PhaseResult:
-        """执行初始化 - 子类必须实现"""
-        raise NotImplementedError("Subclasses must implement initialize method")
-
-
-class BaseInitializer(Initializer):
-    """基础初始化器类 - 为向后兼容保留"""
-    
-    def __init__(self, phase: InitializationPhase):
-        super().__init__(phase)
-    
-    def _create_result(self, success: bool, duration: float = 0.0, 
-                      artifacts: Optional[Dict] = None, error: Optional[Exception] = None) -> PhaseResult:
-        """创建初始化结果"""
-        return PhaseResult(
-            phase=self.phase,
-            success=success,
-            duration=duration,
-            artifacts=artifacts or {},
-            error=error
-        )
-
-
-class InitializerRegistry:
-    """
-    初始化器注册表 - 管理所有初始化器的注册和执行
-    
-    特点：
-    1. 线程安全的注册和执行
-    2. 支持函数式和类式初始化器
-    3. 统一的结果处理
-    """
-    
-    def __init__(self):
-        self._initializers: Dict[InitializationPhase, Initializer] = {}
-        self._lock = threading.RLock()
-    
-    def register(self, initializer: Initializer):
-        """注册初始化器"""
-        with self._lock:
-            phase = initializer.phase
-            if phase in self._initializers:
-                raise ValueError(f"Initializer for phase {phase} already registered")
-            self._initializers[phase] = initializer
-    
-    def register_function(self, phase: InitializationPhase, 
-                         init_func: Callable[[InitializationContext], PhaseResult]):
-        """注册函数式初始化器"""
-        
-        class FunctionInitializer:
-            def __init__(self, phase: InitializationPhase, func: Callable):
-                self._phase = phase
-                self._func = func
-            
-            @property  
-            def phase(self) -> InitializationPhase:
-                return self._phase
-            
-            def initialize(self, context: InitializationContext) -> PhaseResult:
-                return self._func(context)
-        
-        self.register(FunctionInitializer(phase, init_func))
-    
-    def get_initializer(self, phase: InitializationPhase) -> Optional[Initializer]:
-        """获取指定阶段的初始化器"""
-        with self._lock:
-            return self._initializers.get(phase)
-    
-    def get_all_phases(self) -> List[InitializationPhase]:
-        """获取所有已注册的阶段"""
-        with self._lock:
-            return list(self._initializers.keys())
-    
-    def has_initializer(self, phase: InitializationPhase) -> bool:
-        """检查是否有指定阶段的初始化器"""
-        with self._lock:
-            return phase in self._initializers
-    
-    def clear(self):
-        """清空注册表"""
-        with self._lock:
-            self._initializers.clear()
-    
-    def execute_phase(self, phase: InitializationPhase, 
-                     context: InitializationContext) -> PhaseResult:
-        """执行指定阶段的初始化"""
-        initializer = self.get_initializer(phase)
-        if not initializer:
-            error = ValueError(f"No initializer registered for phase {phase}")
-            return PhaseResult(
-                phase=phase,
-                success=False,
-                error=error
-            )
-        
-        try:
-            return initializer.initialize(context)
-        except Exception as e:
-            return PhaseResult(
-                phase=phase,
-                success=False,
-                error=e
-            )
-
-
-# 全局注册表实例
-_global_registry = InitializerRegistry()
-
-
-def get_global_registry() -> InitializerRegistry:
-    """获取全局注册表"""
-    return _global_registry
-
-
-def register_initializer(initializer: Initializer):
-    """注册初始化器到全局注册表"""
-    _global_registry.register(initializer)
-
-
-def register_phase_function(phase: InitializationPhase, 
-                           init_func: Callable[[InitializationContext], PhaseResult]):
-    """注册函数式初始化器到全局注册表"""
+#!/usr/bin/python
+# -*- coding: UTF-8 -*-
+"""
+初始化器注册表 - 管理所有初始化器的注册和执行
+"""
+
+import threading
+from typing import Dict, Optional, Callable, List
+from .context import InitializationContext
+from .phases import InitializationPhase, PhaseResult
+
+
+class Initializer:
+    """初始化器基类"""
+    
+    def __init__(self, phase: InitializationPhase):
+        self._phase = phase
+    
+    @property
+    def phase(self) -> InitializationPhase:
+        """获取初始化阶段"""
+        return self._phase
+    
+    def initialize(self, context: InitializationContext) -> PhaseResult:
+        """执行初始化 - 子类必须实现"""
+        raise NotImplementedError("Subclasses must implement initialize method")
+
+
+class BaseInitializer(Initializer):
+    """基础初始化器类 - 为向后兼容保留"""
+    
+    def __init__(self, phase: InitializationPhase):
+        super().__init__(phase)
+    
+    def _create_result(self, success: bool, duration: float = 0.0, 
+                      artifacts: Optional[Dict] = None, error: Optional[Exception] = None) -> PhaseResult:
+        """创建初始化结果"""
+        return PhaseResult(
+            phase=self.phase,
+            success=success,
+            duration=duration,
+            artifacts=artifacts or {},
+            error=error
+        )
+
+
+class InitializerRegistry:
+    """
+    初始化器注册表 - 管理所有初始化器的注册和执行
+    
+    特点：
+    1. 线程安全的注册和执行
+    2. 支持函数式和类式初始化器
+    3. 统一的结果处理
+    """
+    
+    def __init__(self):
+        self._initializers: Dict[InitializationPhase, Initializer] = {}
+        self._lock = threading.RLock()
+    
+    def register(self, initializer: Initializer):
+        """注册初始化器"""
+        with self._lock:
+            phase = initializer.phase
+            if phase in self._initializers:
+                raise ValueError(f"Initializer for phase {phase} already registered")
+            self._initializers[phase] = initializer
+    
+    def register_function(self, phase: InitializationPhase, 
+                         init_func: Callable[[InitializationContext], PhaseResult]):
+        """注册函数式初始化器"""
+        
+        class FunctionInitializer:
+            def __init__(self, phase: InitializationPhase, func: Callable):
+                self._phase = phase
+                self._func = func
+            
+            @property  
+            def phase(self) -> InitializationPhase:
+                return self._phase
+            
+            def initialize(self, context: InitializationContext) -> PhaseResult:
+                return self._func(context)
+        
+        self.register(FunctionInitializer(phase, init_func))
+    
+    def get_initializer(self, phase: InitializationPhase) -> Optional[Initializer]:
+        """获取指定阶段的初始化器"""
+        with self._lock:
+            return self._initializers.get(phase)
+    
+    def get_all_phases(self) -> List[InitializationPhase]:
+        """获取所有已注册的阶段"""
+        with self._lock:
+            return list(self._initializers.keys())
+    
+    def has_initializer(self, phase: InitializationPhase) -> bool:
+        """检查是否有指定阶段的初始化器"""
+        with self._lock:
+            return phase in self._initializers
+    
+    def clear(self):
+        """清空注册表"""
+        with self._lock:
+            self._initializers.clear()
+    
+    def execute_phase(self, phase: InitializationPhase, 
+                     context: InitializationContext) -> PhaseResult:
+        """执行指定阶段的初始化"""
+        initializer = self.get_initializer(phase)
+        if not initializer:
+            error = ValueError(f"No initializer registered for phase {phase}")
+            return PhaseResult(
+                phase=phase,
+                success=False,
+                error=error
+            )
+        
+        try:
+            return initializer.initialize(context)
+        except Exception as e:
+            return PhaseResult(
+                phase=phase,
+                success=False,
+                error=e
+            )
+
+
+# 全局注册表实例
+_global_registry = InitializerRegistry()
+
+
+def get_global_registry() -> InitializerRegistry:
+    """获取全局注册表"""
+    return _global_registry
+
+
+def register_initializer(initializer: Initializer):
+    """注册初始化器到全局注册表"""
+    _global_registry.register(initializer)
+
+
+def register_phase_function(phase: InitializationPhase, 
+                           init_func: Callable[[InitializationContext], PhaseResult]):
+    """注册函数式初始化器到全局注册表"""
     _global_registry.register_function(phase, init_func)

crawlo/items/__init__.py

@@ -1,23 +1,23 @@
-#!/usr/bin/python
-# -*- coding: UTF-8 -*-
-"""
-crawlo.items 包
-===============
-提供 Item 和 Field 类用于数据定义和验证。
-"""
-from .items import Item
-from .fields import Field
-from .base import ItemMeta
-
-from crawlo.exceptions import ItemInitError, ItemAttributeError
-
-__all__ = [
-    'Item',
-    'Field',
-    'ItemMeta',
-    'ItemInitError',
-    'ItemAttributeError'
-]
-
-
-
+#!/usr/bin/python
+# -*- coding: UTF-8 -*-
+"""
+crawlo.items 包
+===============
+提供 Item 和 Field 类用于数据定义和验证。
+"""
+from .items import Item
+from .fields import Field
+from .base import ItemMeta
+
+from crawlo.exceptions import ItemInitError, ItemAttributeError
+
+__all__ = [
+    'Item',
+    'Field',
+    'ItemMeta',
+    'ItemInitError',
+    'ItemAttributeError'
+]
+
+
+