crackerjack 0.37.9__py3-none-any.whl → 0.45.2__py3-none-any.whl

crackerjack/config/hooks.py

@@ -1,7 +1,9 @@
-from dataclasses import dataclass
+from dataclasses import dataclass, field
 from enum import Enum
 from pathlib import Path
 
+from crackerjack.config.tool_commands import get_tool_command
+
 
 class HookStage(Enum):
     FAST = "fast"
@@ -27,15 +29,42 @@ class HookDefinition:
     command: list[str]
     timeout: int = 60
     stage: HookStage = HookStage.FAST
+    description: str | None = None
     retry_on_failure: bool = False
     is_formatting: bool = False
     manual_stage: bool = False
     config_path: Path | None = None
     security_level: SecurityLevel = SecurityLevel.MEDIUM
+    use_precommit_legacy: bool = True  # Phase 8.2: Backward compatibility flag
+    accepts_file_paths: bool = False  # Phase 10.4.4: Can tool process individual files?
+    _direct_cmd_cache: list[str] | None = field(default=None, init=False, repr=False)
 
     def get_command(self) -> list[str]:
+        """Get the command to execute this hook.
+
+        Returns the appropriate command based on use_precommit_legacy flag:
+        - If use_precommit_legacy=True: Returns pre-commit wrapper command (legacy mode)
+        - If use_precommit_legacy=False: Returns direct tool command (Phase 8+ mode)
+
+        Returns:
+            List of command arguments for subprocess execution
+        """
+        # Phase 8.2: Direct invocation mode (new behavior)
+        if not self.use_precommit_legacy:
+            if self._direct_cmd_cache is None:
+                try:
+                    self._direct_cmd_cache = get_tool_command(
+                        self.name, pkg_path=Path.cwd()
+                    )
+                except KeyError:
+                    # Fallback to pre-commit if tool not in registry
+                    # This ensures graceful degradation during migration
+                    self._direct_cmd_cache = None
+            if self._direct_cmd_cache is not None:
+                return self._direct_cmd_cache
+
+        # Legacy mode: Use pre-commit wrapper (Phase 1-7 behavior)
         import shutil
-        from pathlib import Path
 
         pre_commit_path = None
         current_dir = Path.cwd()
@@ -53,6 +82,34 @@ class HookDefinition:
         cmd.extend([self.name, "--all-files"])
         return cmd
 
+    def build_command(self, files: list[Path] | None = None) -> list[str]:
+        """Build command with optional file paths for targeted execution.
+
+        Phase 10.4.4: Enables incremental execution on specific files when supported.
+
+        Args:
+            files: Optional list of file paths to process. If None, processes all files.
+
+        Returns:
+            Command list with file paths appended if tool accepts them.
+
+        Example:
+            >>> hook = HookDefinition(
+            ...     name="ruff-check",
+            ...     command=["ruff", "check"],
+            ...     accepts_file_paths=True,
+            ... )
+            >>> hook.build_command([Path("foo.py"), Path("bar.py")])
+            ["ruff", "check", "foo.py", "bar.py"]
+        """
+        base_cmd = self.get_command().copy()
+
+        # Append file paths if tool accepts them and files are provided
+        if files and self.accepts_file_paths:
+            base_cmd.extend([str(f) for f in files])
+
+        return base_cmd
+
 
 @dataclass
 class HookStrategy:
@@ -69,74 +126,124 @@ FAST_HOOKS = [
         name="validate-regex-patterns",
         command=[],
         is_formatting=True,
-        timeout=30,
+        timeout=120,  # Increased from 60 to handle larger codebases and prevent timeout issues
         retry_on_failure=True,
         security_level=SecurityLevel.HIGH,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
     ),
     HookDefinition(
         name="trailing-whitespace",
         command=[],
         is_formatting=True,
+        timeout=120,  # Increased from 60 to handle larger codebases and prevent timeout issues
         retry_on_failure=True,
         security_level=SecurityLevel.LOW,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.4.4: File-level fixer
     ),
     HookDefinition(
         name="end-of-file-fixer",
         command=[],
         is_formatting=True,
+        timeout=120,  # Increased from 60 to handle larger codebases and prevent timeout issues
         retry_on_failure=True,
         security_level=SecurityLevel.LOW,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.4.4: File-level fixer
     ),
     HookDefinition(
         name="check-yaml",
         command=[],
+        timeout=60,  # Increased from 20 to reduce timeout issues with larger YAML files
         security_level=SecurityLevel.MEDIUM,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.4.4: File-level validator
     ),
     HookDefinition(
         name="check-toml",
         command=[],
+        timeout=150,  # Increased from 79 to reduce timeout issues with larger TOML files
         security_level=SecurityLevel.MEDIUM,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.4.4: File-level validator
     ),
     HookDefinition(
-        name="check-added-large-files",
+        name="check-json",
         command=[],
+        timeout=90,  # Increased from 30 to reduce timeout issues with larger JSON files
+        security_level=SecurityLevel.MEDIUM,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.4.4: File-level validator
+    ),
+    HookDefinition(
+        name="check-ast",
+        command=[],
+        timeout=90,  # Increased from 30 to reduce timeout issues with larger Python files
         security_level=SecurityLevel.HIGH,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.4.4: File-level validator
     ),
     HookDefinition(
-        name="uv-lock",
+        name="format-json",
+        command=[],
+        is_formatting=True,
+        timeout=120,  # Increased from 45 to reduce timeout issues with larger JSON files
+        retry_on_failure=True,
+        security_level=SecurityLevel.LOW,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.4.4: File-level formatter
+    ),
+    HookDefinition(
+        name="check-added-large-files",
         command=[],
+        timeout=90,  # Increased from 30 to reduce timeout issues with repositories with many files
         security_level=SecurityLevel.HIGH,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
     ),
     HookDefinition(
-        name="gitleaks",
+        name="uv-lock",
         command=[],
-        security_level=SecurityLevel.CRITICAL,
+        timeout=60,  # Increased from 20 to reduce timeout issues with complex dependency trees
+        security_level=SecurityLevel.HIGH,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
     ),
     HookDefinition(
         name="codespell",
         command=[],
+        timeout=150,  # Increased from 45 to reduce timeout issues with large codebases
         security_level=SecurityLevel.LOW,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.4.4: File-level spell checker
     ),
     HookDefinition(
         name="ruff-check",
         command=[],
         is_formatting=True,
+        timeout=240,  # Increased from 120 to reduce timeout issues with large codebases
         retry_on_failure=True,
         security_level=SecurityLevel.MEDIUM,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.4.4: File-level Python linter
     ),
     HookDefinition(
         name="ruff-format",
         command=[],
         is_formatting=True,
+        timeout=240,  # Increased from 120 to reduce timeout issues with large codebases
         retry_on_failure=True,
         security_level=SecurityLevel.LOW,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.4.4: File-level Python formatter
     ),
     HookDefinition(
         name="mdformat",
         command=[],
         is_formatting=True,
+        timeout=300,  # Increased from 120 to reduce timeout issues with larger markdown files
         retry_on_failure=True,
         security_level=SecurityLevel.LOW,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.4.4: File-level markdown formatter
     ),
 ]
 
@@ -144,50 +251,111 @@ COMPREHENSIVE_HOOKS = [
     HookDefinition(
         name="zuban",
         command=[],
-        timeout=30,
+        timeout=240,  # Increased from 80 to reduce timeout issues with larger codebases during type checking
+        stage=HookStage.COMPREHENSIVE,
+        manual_stage=True,
+        security_level=SecurityLevel.HIGH,  # Changed from CRITICAL to HIGH to allow other hooks to run
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.5: Allow incremental execution on changed files to avoid virtual environments
+    ),
+    HookDefinition(
+        name="semgrep",
+        command=[],
+        timeout=480,  # Increased from 240 to reduce timeout issues with comprehensive security scans
         stage=HookStage.COMPREHENSIVE,
         manual_stage=True,
         security_level=SecurityLevel.CRITICAL,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.4.4: File-level SAST scanner
     ),
     HookDefinition(
-        name="bandit",
+        name="pyscn",
         command=[],
-        timeout=300,
+        timeout=300,  # CFG analysis, clone detection, complexity metrics
+        stage=HookStage.COMPREHENSIVE,
+        manual_stage=True,
+        security_level=SecurityLevel.HIGH,  # Security + quality analysis
+        use_precommit_legacy=False,  # Direct invocation
+        accepts_file_paths=True,  # Can scan specific Python files
+    ),
+    # NOTE: Bandit replaced with Semgrep (using uvx for Python 3.13 isolation)
+    # HookDefinition(
+    #     name="bandit",
+    #     command=[],
+    #     timeout=180,  # 3 minutes for SAST scanning
+    #     stage=HookStage.COMPREHENSIVE,
+    #     manual_stage=True,
+    #     security_level=SecurityLevel.CRITICAL,
+    #     use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+    #     accepts_file_paths=True,  # Phase 10.4.4: File-level SAST scanner
+    # ),
+    HookDefinition(
+        name="gitleaks",
+        command=[],
+        timeout=180,  # Increased from 45 to reduce timeout issues with comprehensive security scans
         stage=HookStage.COMPREHENSIVE,
         manual_stage=True,
         security_level=SecurityLevel.CRITICAL,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+    ),
+    HookDefinition(
+        name="pip-audit",
+        command=[],
+        timeout=180,  # Network calls to vulnerability databases
+        stage=HookStage.COMPREHENSIVE,
+        manual_stage=True,
+        security_level=SecurityLevel.CRITICAL,  # CVE vulnerabilities are critical
+        use_precommit_legacy=False,  # Direct invocation
+        accepts_file_paths=False,  # Scans entire environment/requirements
     ),
     HookDefinition(
         name="skylos",
         command=[],
-        timeout=30,
+        timeout=180,  # Increased from 60 to reduce timeout issues with comprehensive dead code scanning
         stage=HookStage.COMPREHENSIVE,
         manual_stage=True,
         security_level=SecurityLevel.MEDIUM,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.5: Incremental execution on changed files
     ),
     HookDefinition(
         name="refurb",
         command=[],
-        timeout=300,
+        timeout=480,  # Increased from 240 to reduce timeout issues with comprehensive refactoring analysis
         stage=HookStage.COMPREHENSIVE,
         manual_stage=True,
         security_level=SecurityLevel.MEDIUM,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.5: Incremental execution on changed files (240s -> ~10s)
     ),
     HookDefinition(
         name="creosote",
         command=[],
-        timeout=300,
+        timeout=360,  # Increased from 180 to reduce timeout issues with comprehensive dependency analysis
         stage=HookStage.COMPREHENSIVE,
         manual_stage=True,
         security_level=SecurityLevel.HIGH,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
     ),
     HookDefinition(
         name="complexipy",
         command=[],
-        timeout=60,
+        timeout=300,  # Increased from 120 to reduce timeout issues with comprehensive complexity analysis
         stage=HookStage.COMPREHENSIVE,
         manual_stage=True,
         security_level=SecurityLevel.MEDIUM,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.5: Incremental execution on changed files
+    ),
+    HookDefinition(
+        name="check-jsonschema",
+        command=[],
+        timeout=180,  # Increased from 60 to reduce timeout issues with complex schema validation
+        stage=HookStage.COMPREHENSIVE,
+        manual_stage=True,
+        security_level=SecurityLevel.HIGH,
+        use_precommit_legacy=False,  # Phase 8.4: Direct invocation
+        accepts_file_paths=True,  # Phase 10.4.4: File-level schema validator
     ),
 ]
 
@@ -195,15 +363,19 @@ COMPREHENSIVE_HOOKS = [
 FAST_STRATEGY = HookStrategy(
     name="fast",
     hooks=FAST_HOOKS,
-    timeout=60,
+    timeout=300,  # Increased from 60 to accommodate all increased hook timeouts
     retry_policy=RetryPolicy.FORMATTING_ONLY,
+    parallel=True,  # Phase 6: Enable parallel execution for 2-3x speedup
+    max_workers=4,  # Optimal concurrency for fast hooks
 )
 
 COMPREHENSIVE_STRATEGY = HookStrategy(
     name="comprehensive",
     hooks=COMPREHENSIVE_HOOKS,
-    timeout=300,
+    timeout=1800,  # Increased from 300 to accommodate all increased hook timeouts
     retry_policy=RetryPolicy.NONE,
+    parallel=True,  # Phase 6: Enable parallel execution for 2x speedup
+    max_workers=4,  # Optimal concurrency for comprehensive hooks
 )
 
 

crackerjack/config/loader.py

@@ -0,0 +1,239 @@
+"""Settings loader for ACB configuration with YAML support.
+
+This module provides utilities for loading ACB Settings from YAML files
+with support for configuration layering and environment-specific overrides.
+
+Configuration Priority (highest to lowest):
+    1. settings/local.yaml (local overrides, gitignored)
+    2. settings/crackerjack.yaml (main configuration)
+    3. Default values from Settings class
+
+Example:
+    >>> from crackerjack.config import CrackerjackSettings
+    >>> settings = CrackerjackSettings.load()
+    >>> settings.verbose
+    False
+
+    >>> # Async loading with full ACB initialization
+    >>> settings = await CrackerjackSettings.load_async()
+"""
+
+from __future__ import annotations
+
+import logging
+import typing as t
+from pathlib import Path
+from typing import TypeVar
+
+import yaml
+from acb.config import Settings
+
+logger = logging.getLogger(__name__)
+
+T = TypeVar("T", bound=Settings)
+
+
+def _load_single_config_file(config_file: Path) -> dict[str, t.Any]:
+    """Load a single config file and return its data."""
+    if not config_file.exists():
+        logger.debug(f"Configuration file not found: {config_file}")
+        return {}
+
+    try:
+        with config_file.open() as f:
+            loaded_data: t.Any = yaml.safe_load(f)
+            if isinstance(loaded_data, dict):
+                logger.debug(f"Loaded configuration from {config_file}")
+                return loaded_data
+            else:
+                logger.warning(
+                    f"Invalid YAML format in {config_file}: expected dict, got {type(loaded_data).__name__}"
+                )
+                return {}
+    except yaml.YAMLError as e:
+        logger.error(f"Failed to parse YAML from {config_file}: {e}")
+        return {}
+    except OSError as e:
+        logger.error(f"Failed to read {config_file}: {e}")
+        return {}
+
+
+def _merge_config_data(config_files: list[Path]) -> dict[str, t.Any]:
+    """Merge data from all config files."""
+    merged_data = {}
+    for config_file in config_files:
+        file_data = _load_single_config_file(config_file)
+        merged_data.update(file_data)
+    return merged_data
+
+
+def load_settings[T: Settings](
+    settings_class: type[T],
+    settings_dir: Path | None = None,
+) -> T:
+    """Load settings from YAML files with layered configuration.
+
+    This function loads configuration from multiple YAML files and merges them
+    with priority-based overriding. Unknown YAML fields are silently ignored.
+
+    Args:
+        settings_class: The Settings class to instantiate
+        settings_dir: Directory containing YAML files (default: ./settings)
+
+    Returns:
+        Initialized Settings instance with merged configuration
+
+    Configuration Priority (highest to lowest):
+        1. settings/local.yaml (local overrides, gitignored)
+        2. settings/crackerjack.yaml (main configuration)
+        3. Default values from Settings class
+
+    Example:
+        >>> class MySettings(Settings):
+        ...     debug: bool = False
+        ...     max_workers: int = 4
+        >>> settings = load_settings(MySettings)
+        >>> settings.debug
+        False
+    """
+    if settings_dir is None:
+        settings_dir = Path.cwd() / "settings"
+
+    # Configuration files in priority order (lowest to highest)
+    config_files = [
+        settings_dir / "crackerjack.yaml",
+        settings_dir / "local.yaml",
+    ]
+
+    # Load and merge data from all config files
+    merged_data = _merge_config_data(config_files)
+
+    # Filter to only fields defined in the Settings class
+    # This prevents validation errors from unknown YAML keys
+    relevant_data = {
+        k: v for k, v in merged_data.items() if k in settings_class.model_fields
+    }
+
+    # Log filtered fields if any were excluded
+    excluded_fields = set(merged_data.keys()) - set(relevant_data.keys())
+    if excluded_fields:
+        logger.debug(
+            f"Ignored unknown configuration fields: {', '.join(sorted(excluded_fields))}"
+        )
+
+    logger.info(
+        f"Loaded {len(relevant_data)} configuration values for {settings_class.__name__}"
+    )
+
+    # Synchronous initialization (safe for module-level code)
+    return settings_class(**relevant_data)
+
+
+async def load_settings_async[T: Settings](
+    settings_class: type[T],
+    settings_dir: Path | None = None,
+) -> T:
+    """Load settings asynchronously with full ACB initialization.
+
+    This function provides the same configuration loading as load_settings()
+    but uses ACB's async initialization which includes secret loading and
+    other async setup operations.
+
+    Args:
+        settings_class: The Settings class to instantiate
+        settings_dir: Directory containing YAML files (default: ./settings)
+
+    Returns:
+        Initialized Settings instance with async initialization complete
+
+    Note:
+        Use this for application runtime when async context is available.
+        For module-level initialization, use synchronous load_settings().
+
+    Example:
+        >>> settings = await load_settings_async(CrackerjackSettings)
+        >>> settings.verbose
+        False
+    """
+    if settings_dir is None:
+        settings_dir = Path.cwd() / "settings"
+
+    # Configuration files in priority order (lowest to highest)
+    config_files = [
+        settings_dir / "crackerjack.yaml",
+        settings_dir / "local.yaml",
+    ]
+
+    # Load and merge YAML data from all files
+    merged_data = await _load_yaml_data(config_files)
+
+    # Process the loaded data
+    relevant_data = _filter_relevant_data(merged_data, settings_class)
+    _log_filtered_fields(merged_data, relevant_data)
+    _log_load_info(settings_class, relevant_data)
+
+    # Async initialization (loads secrets, performs async setup)
+    return await settings_class.create_async(**relevant_data)
+
+
+async def _load_yaml_data(config_files: list[Path]) -> dict[str, t.Any]:
+    """Load and merge YAML data from configuration files."""
+    merged_data: dict[str, t.Any] = {}
+    for config_file in config_files:
+        file_data = await _load_single_yaml_file(config_file)
+        if file_data is not None:
+            merged_data.update(file_data)
+        elif not config_file.exists():
+            logger.debug(f"Configuration file not found: {config_file}")
+    return merged_data
+
+
+async def _load_single_yaml_file(config_file: Path) -> dict[str, t.Any] | None:
+    """Load a single YAML file and return its content."""
+    if not config_file.exists():
+        return None
+
+    try:
+        with config_file.open() as f:
+            loaded_data: t.Any = yaml.safe_load(f)
+            if isinstance(loaded_data, dict):
+                logger.debug(f"Loaded configuration from {config_file}")
+                return loaded_data
+            else:
+                logger.warning(
+                    f"Invalid YAML format in {config_file}: expected dict, got {type(loaded_data).__name__}"
+                )
+                return {}
+    except yaml.YAMLError as e:
+        logger.error(f"Failed to parse YAML from {config_file}: {e}")
+        return None
+    except OSError as e:
+        logger.error(f"Failed to read {config_file}: {e}")
+        return None
+
+
+def _filter_relevant_data[T: Settings](
+    merged_data: dict[str, t.Any], settings_class: type[T]
+) -> dict[str, t.Any]:
+    """Filter the loaded data to only fields defined in the Settings class."""
+    return {k: v for k, v in merged_data.items() if k in settings_class.model_fields}
+
+
+def _log_filtered_fields(
+    merged_data: dict[str, t.Any], relevant_data: dict[str, t.Any]
+) -> None:
+    """Log any fields that were excluded due to not being in the settings class."""
+    excluded_fields = set(merged_data.keys()) - set(relevant_data.keys())
+    if excluded_fields:
+        logger.debug(
+            f"Ignored unknown configuration fields: {', '.join(sorted(excluded_fields))}"
+        )
+
+
+def _log_load_info[T: Settings](
+    settings_class: type[T], relevant_data: dict[str, t.Any]
+) -> None:
+    """Log information about the loaded configuration."""
+    logger.info(
+        f"Loaded {len(relevant_data)} configuration values for {settings_class.__name__} (async)"
+    )