oscura 0.0.1__py3-none-any.whl → 0.1.0__py3-none-any.whl

oscura/extensibility/extensions.py

@@ -0,0 +1,1125 @@
+"""Extension point registry and management system.
+
+This module implements a central registry for extension points that allows
+plugins and custom code to extend TraceKit functionality at well-defined
+integration points.
+"""
+
+from __future__ import annotations
+
+import logging
+from dataclasses import dataclass, field
+from enum import Enum, auto
+from typing import TYPE_CHECKING, Any, TypeVar
+
+if TYPE_CHECKING:
+    from collections.abc import Callable
+
+logger = logging.getLogger(__name__)
+
+T = TypeVar("T")
+
+
+class HookErrorPolicy(Enum):
+    """Policy for handling hook errors.
+
+    Attributes:
+        CONTINUE: Continue executing remaining hooks after error
+        ABORT: Stop execution immediately on error
+        IGNORE: Ignore error silently
+    """
+
+    CONTINUE = auto()
+    ABORT = auto()
+    IGNORE = auto()
+
+
+@dataclass
+class ExtensionPointSpec:
+    """Specification for an extension point.
+
+    Defines the contract that implementations must follow including
+    required and optional methods, version info, and documentation.
+
+    Attributes:
+        name: Unique name for the extension point
+        version: API version (semver format)
+        description: Human-readable description
+        required_methods: List of method names that must be implemented
+        optional_methods: List of optional method names
+        interface: Optional interface class that implementations should inherit from
+
+    Example:
+        >>> spec = ExtensionPointSpec(
+        ...     name="protocol_decoder",
+        ...     version="1.0.0",
+        ...     description="Decode protocol from waveform",
+        ...     required_methods=["decode", "get_metadata"],
+        ...     optional_methods=["configure", "reset"]
+        ... )
+
+    References:
+        EXT-001: Extension Point Registry
+    """
+
+    name: str
+    version: str = "1.0.0"
+    description: str = ""
+    required_methods: list[str] = field(default_factory=list)
+    optional_methods: list[str] = field(default_factory=list)
+    interface: type | None = None
+
+    def validate_implementation(self, impl: Any) -> tuple[bool, list[str]]:
+        """Validate that implementation matches interface.
+
+        Args:
+            impl: Implementation to validate
+
+        Returns:
+            Tuple of (is_valid, list of missing methods)
+
+        Example:
+            >>> is_valid, missing = spec.validate_implementation(MyDecoder())
+            >>> if not is_valid:
+            ...     print(f"Missing methods: {missing}")
+        """
+        missing = []
+        for method in self.required_methods:
+            if not hasattr(impl, method) or not callable(getattr(impl, method)):
+                missing.append(method)
+        return len(missing) == 0, missing
+
+
+@dataclass
+class RegisteredAlgorithm:
+    """Metadata for a registered algorithm.
+
+    Attributes:
+        name: Algorithm name
+        category: Algorithm category
+        func: Algorithm implementation
+        priority: Execution priority (higher = first)
+        performance: Performance characteristics
+        supports: Supported data types
+        description: Human-readable description
+        complexity: Time complexity string
+        capabilities: Algorithm capabilities
+        memory_usage: Memory usage characteristics
+        registration_order: Order in which algorithm was registered
+
+    References:
+        EXT-002: Algorithm Registration (capability queries, performance metadata)
+        EXT-004: Priority System (registration order for tie-breaking)
+    """
+
+    name: str
+    category: str
+    func: Callable[..., Any]
+    priority: int = 50
+    performance: dict[str, str] = field(default_factory=dict)
+    supports: list[str] = field(default_factory=list)
+    description: str = ""
+    complexity: str = "O(n)"
+    capabilities: dict[str, Any] = field(default_factory=dict)
+    memory_usage: str = "unknown"
+    registration_order: int = 0
+
+    def can(self, capability: str) -> bool:
+        """Check if algorithm has a specific capability.
+
+        Args:
+            capability: Capability name to check
+
+        Returns:
+            True if algorithm supports the capability
+
+        Example:
+            >>> algo.can("multi_channel")
+            True
+
+        References:
+            EXT-002: Algorithm Registration (capability queries)
+        """
+        return self.capabilities.get(capability, False)  # type: ignore[no-any-return]
+
+    def get_capabilities(self) -> dict[str, Any]:
+        """Get all capabilities of this algorithm.
+
+        Returns:
+            Dictionary of capability names to values
+
+        Example:
+            >>> caps = algo.get_capabilities()
+            >>> print(caps)
+            {'multi_channel': True, 'real_time': False, 'max_sample_rate': 1000000}
+
+        References:
+            EXT-002: Algorithm Registration (capability queries)
+        """
+        return self.capabilities.copy()
+
+
+@dataclass
+class HookContext:
+    """Context passed to hook functions.
+
+    Attributes:
+        data: Primary data being processed
+        metadata: Additional context metadata
+        abort: Set to True to abort operation
+        abort_reason: Reason for abort
+
+    Example:
+        >>> @tk.hooks.register("pre_decode")
+        >>> def validate_waveform(context):
+        ...     if context.data.sample_rate < 1000:
+        ...         context.abort = True
+        ...         context.abort_reason = "Sample rate too low"
+        ...     return context
+
+    References:
+        EXT-005: Hook System
+    """
+
+    data: Any = None
+    metadata: dict[str, Any] = field(default_factory=dict)
+    abort: bool = False
+    abort_reason: str = ""
+
+    def __post_init__(self):  # type: ignore[no-untyped-def]
+        """Initialize metadata if None."""
+        if self.metadata is None:
+            self.metadata = {}  # type: ignore[unreachable]
+
+
+@dataclass
+class RegisteredHook:
+    """Registered hook function.
+
+    Attributes:
+        hook_point: Name of hook point
+        func: Hook function
+        priority: Execution priority (higher = first)
+        name: Optional hook name
+    """
+
+    hook_point: str
+    func: Callable[[HookContext], HookContext]
+    priority: int = 50
+    name: str = ""
+
+
+class ExtensionPointRegistry:
+    """Central registry of all extension points in TraceKit.
+
+    Manages registration and lookup of extension points, algorithms,
+    and hooks throughout the system.
+
+    Example:
+        >>> # List all extension points
+        >>> extension_points = tk.extensions.list()
+        >>> for ep in extension_points:
+        ...     print(f"{ep.name} v{ep.version}")
+
+        >>> # Get specific extension point
+        >>> decoder_ep = tk.extensions.get("protocol_decoder")
+        >>> print(f"Required methods: {decoder_ep.required_methods}")
+
+    References:
+        EXT-001: Extension Point Registry
+        EXT-002: Algorithm Registration
+        EXT-003: Algorithm Selection
+        EXT-004: Priority System
+        EXT-005: Hook System
+        EXT-006: Custom Decoder Registration
+    """
+
+    _instance: ExtensionPointRegistry | None = None
+
+    def __new__(cls) -> ExtensionPointRegistry:
+        """Ensure singleton instance.
+
+        Returns:
+            Singleton ExtensionPointRegistry instance.
+        """
+        if cls._instance is None:
+            cls._instance = super().__new__(cls)
+            cls._instance._extension_points: dict[str, ExtensionPointSpec] = {}  # type: ignore[misc, attr-defined]
+            cls._instance._algorithms: dict[str, dict[str, RegisteredAlgorithm]] = {}  # type: ignore[misc, attr-defined]
+            cls._instance._hooks: dict[str, list[RegisteredHook]] = {}  # type: ignore[misc, attr-defined]
+            cls._instance._hook_error_policy = HookErrorPolicy.CONTINUE
+            cls._instance._log_hook_errors = True
+            cls._instance._initialized = False  # type: ignore[has-type]
+            cls._instance._registration_counter = 0  # type: ignore[misc, attr-defined]
+        return cls._instance
+
+    def initialize(self) -> None:
+        """Initialize built-in extension points.
+
+        Registers the standard extension points that come with TraceKit.
+        """
+        if self._initialized:  # type: ignore[has-type]
+            return
+
+        # Register standard extension points
+        self.register_point(
+            ExtensionPointSpec(
+                name="protocol_decoder",
+                version="1.0.0",
+                description="Decode protocol from waveform or digital trace",
+                required_methods=["decode", "get_metadata"],
+                optional_methods=["configure", "reset", "validate_config"],
+            )
+        )
+
+        self.register_point(
+            ExtensionPointSpec(
+                name="file_loader",
+                version="1.0.0",
+                description="Load trace data from file format",
+                required_methods=["load", "can_load"],
+                optional_methods=["get_metadata", "get_channels"],
+            )
+        )
+
+        self.register_point(
+            ExtensionPointSpec(
+                name="measurement",
+                version="1.0.0",
+                description="Compute measurement from trace",
+                required_methods=["measure"],
+                optional_methods=["validate_input", "get_units"],
+            )
+        )
+
+        self.register_point(
+            ExtensionPointSpec(
+                name="exporter",
+                version="1.0.0",
+                description="Export trace data to file format",
+                required_methods=["export"],
+                optional_methods=["get_supported_formats"],
+            )
+        )
+
+        self.register_point(
+            ExtensionPointSpec(
+                name="algorithm",
+                version="1.0.0",
+                description="Signal processing algorithm",
+                required_methods=["process"],
+                optional_methods=["configure", "get_parameters"],
+            )
+        )
+
+        self._initialized = True
+        logger.debug("Extension point registry initialized with built-in points")
+
+    # =========================================================================
+    # Extension Point Management (EXT-001)
+    # =========================================================================
+
+    def register_point(self, spec: ExtensionPointSpec) -> None:
+        """Register an extension point.
+
+        Args:
+            spec: Extension point specification
+
+        Raises:
+            ValueError: If extension point already exists
+
+        Example:
+            >>> spec = ExtensionPointSpec(
+            ...     name="my_extension",
+            ...     version="1.0.0",
+            ...     required_methods=["process"]
+            ... )
+            >>> registry.register_point(spec)
+        """
+        if spec.name in self._extension_points:  # type: ignore[attr-defined]
+            raise ValueError(f"Extension point '{spec.name}' already registered")
+        self._extension_points[spec.name] = spec  # type: ignore[attr-defined]
+        logger.debug(f"Registered extension point: {spec.name} v{spec.version}")
+
+    def get_point(self, name: str) -> ExtensionPointSpec:
+        """Get extension point specification.
+
+        Args:
+            name: Extension point name
+
+        Returns:
+            Extension point specification
+
+        Raises:
+            KeyError: If extension point not found
+        """
+        if name not in self._extension_points:  # type: ignore[attr-defined]
+            raise KeyError(
+                f"Extension point '{name}' not found. "
+                f"Available: {list(self._extension_points.keys())}"  # type: ignore[attr-defined]
+            )
+        return self._extension_points[name]  # type: ignore[no-any-return, attr-defined]
+
+    def list_points(self) -> list[ExtensionPointSpec]:
+        """List all registered extension points.
+
+        Returns:
+            List of extension point specifications
+        """
+        return list(self._extension_points.values())  # type: ignore[attr-defined]
+
+    def exists(self, name: str) -> bool:
+        """Check if extension point exists.
+
+        Args:
+            name: Extension point name
+
+        Returns:
+            True if exists
+        """
+        return name in self._extension_points  # type: ignore[attr-defined]
+
+    # =========================================================================
+    # Algorithm Management (EXT-002, EXT-003, EXT-004)
+    # =========================================================================
+
+    def register_algorithm(
+        self,
+        name: str,
+        func: Callable[..., Any],
+        category: str,
+        priority: int = 50,
+        performance: dict[str, str] | None = None,
+        supports: list[str] | None = None,
+        description: str = "",
+        complexity: str = "O(n)",
+        capabilities: dict[str, Any] | None = None,
+        memory_usage: str = "unknown",
+    ) -> None:
+        """Register a custom algorithm implementation.
+
+        Args:
+            name: Algorithm name
+            func: Algorithm function
+            category: Algorithm category
+            priority: Execution priority (0-100, higher = first)
+            performance: Performance characteristics dict (speed/accuracy/memory)
+            supports: List of supported data types
+            description: Human-readable description
+            complexity: Time complexity string (e.g., "O(n)", "O(n log n)")
+            capabilities: Algorithm capabilities dict (e.g., {'multi_channel': True})
+            memory_usage: Memory usage characteristics (low/medium/high/unknown)
+
+        Raises:
+            ValueError: If algorithm already registered
+            TypeError: If func is not callable
+
+        Example:
+            >>> def my_edge_detector(data, threshold=0.5):
+            ...     return find_edges(data, threshold)
+            >>> registry.register_algorithm(
+            ...     name="my_detector",
+            ...     func=my_edge_detector,
+            ...     category="edge_detection",
+            ...     priority=75,
+            ...     performance={"speed": "fast", "accuracy": "medium", "memory": "low"},
+            ...     capabilities={"multi_channel": True, "max_sample_rate": 1000000},
+            ...     memory_usage="low"
+            ... )
+
+        References:
+            EXT-002: Algorithm Registration (capability queries, performance metadata)
+        """
+        if not callable(func):
+            raise TypeError(f"Algorithm must be callable, got {type(func).__name__}")
+
+        if category not in self._algorithms:  # type: ignore[attr-defined]
+            self._algorithms[category] = {}  # type: ignore[attr-defined]
+
+        if name in self._algorithms[category]:  # type: ignore[attr-defined]
+            raise ValueError(f"Algorithm '{name}' already registered in category '{category}'")
+
+        # Increment registration counter
+        self._registration_counter += 1  # type: ignore[attr-defined]
+
+        algo = RegisteredAlgorithm(
+            name=name,
+            category=category,
+            func=func,
+            priority=priority,
+            performance=performance or {},
+            supports=supports or [],
+            description=description,
+            complexity=complexity,
+            capabilities=capabilities or {},
+            memory_usage=memory_usage,
+            registration_order=self._registration_counter,  # type: ignore[attr-defined]
+        )
+
+        self._algorithms[category][name] = algo  # type: ignore[attr-defined]
+        logger.debug(f"Registered algorithm: {name} in category {category}")
+
+    def get_algorithm(self, category: str, name: str) -> RegisteredAlgorithm:
+        """Get algorithm by category and name.
+
+        Args:
+            category: Algorithm category
+            name: Algorithm name
+
+        Returns:
+            Registered algorithm metadata
+
+        Raises:
+            KeyError: If not found
+        """
+        if category not in self._algorithms:  # type: ignore[attr-defined]
+            raise KeyError(f"Category '{category}' not found")
+        if name not in self._algorithms[category]:  # type: ignore[attr-defined]
+            raise KeyError(f"Algorithm '{name}' not found in category '{category}'")
+        return self._algorithms[category][name]  # type: ignore[no-any-return, attr-defined]
+
+    def select_algorithm(
+        self,
+        category: str,
+        name: str | None = None,
+        *,
+        optimize_for: str = "speed",
+        constraints: dict[str, Any] | None = None,
+        required_capabilities: list[str] | None = None,
+    ) -> RegisteredAlgorithm:
+        """Select algorithm implementation at runtime.
+
+        Selects by name if provided, otherwise auto-selects based on
+        optimization criteria and capability matching.
+
+        Args:
+            category: Algorithm category
+            name: Specific algorithm name (optional)
+            optimize_for: Optimization target: "speed", "accuracy", "memory"
+            constraints: Filter constraints on performance/supports
+            required_capabilities: List of required capabilities for auto-selection
+
+        Returns:
+            Selected algorithm
+
+        Raises:
+            KeyError: If category not found or no matching algorithm
+
+        Example:
+            >>> # Select by name
+            >>> algo = registry.select_algorithm("edge_detection", "fast_detector")
+
+            >>> # Auto-select for speed
+            >>> algo = registry.select_algorithm(
+            ...     "edge_detection",
+            ...     optimize_for="speed"
+            ... )
+
+            >>> # Auto-select by capability matching
+            >>> algo = registry.select_algorithm(
+            ...     "edge_detection",
+            ...     required_capabilities=["multi_channel", "real_time"]
+            ... )
+
+        References:
+            EXT-003: Algorithm Selection (auto-selection by capability matching)
+        """
+        if category not in self._algorithms:  # type: ignore[attr-defined]
+            raise KeyError(f"Category '{category}' not found")
+
+        if name:
+            return self.get_algorithm(category, name)
+
+        # Auto-select based on criteria
+        candidates = list(self._algorithms[category].values())  # type: ignore[attr-defined]
+
+        if not candidates:
+            raise KeyError(f"No algorithms registered in category '{category}'")
+
+        # Filter by required capabilities (EXT-003)
+        if required_capabilities:
+            filtered = []
+            for algo in candidates:
+                if all(algo.can(cap) for cap in required_capabilities):
+                    filtered.append(algo)
+            candidates = filtered
+
+        if not candidates:
+            raise KeyError(f"No algorithms match required capabilities in category '{category}'")
+
+        # Apply constraints
+        if constraints:
+            filtered = []
+            for algo in candidates:
+                match = True
+                for key, value in constraints.items():
+                    if key.startswith("performance."):
+                        perf_key = key.split(".", 1)[1]
+                        if algo.performance.get(perf_key) != value:
+                            match = False
+                            break
+                    elif key.startswith("capabilities."):
+                        cap_key = key.split(".", 1)[1]
+                        if algo.capabilities.get(cap_key) != value:
+                            match = False
+                            break
+                    elif key == "supports":
+                        if isinstance(value, list):
+                            if not any(s in algo.supports for s in value):
+                                match = False
+                                break
+                        elif value not in algo.supports:
+                            match = False
+                            break
+                    elif key == "memory_usage":
+                        if algo.memory_usage != value:
+                            match = False
+                            break
+                if match:
+                    filtered.append(algo)
+            candidates = filtered
+
+        if not candidates:
+            raise KeyError(f"No algorithms match constraints in category '{category}'")
+
+        # Sort by optimization criteria
+        if optimize_for == "speed":
+
+            def sort_key(a):  # type: ignore[no-untyped-def]
+                return (
+                    0
+                    if a.performance.get("speed") == "fast"
+                    else 1
+                    if a.performance.get("speed") == "medium"
+                    else 2,
+                    -a.priority,
+                )
+        elif optimize_for == "accuracy":
+
+            def sort_key(a):  # type: ignore[no-untyped-def]
+                return (
+                    0
+                    if a.performance.get("accuracy") == "high"
+                    else 1
+                    if a.performance.get("accuracy") == "medium"
+                    else 2,
+                    -a.priority,
+                )
+        elif optimize_for == "memory":
+
+            def sort_key(a):  # type: ignore[no-untyped-def]
+                return (
+                    0
+                    if a.performance.get("memory") == "low"
+                    else 1
+                    if a.performance.get("memory") == "medium"
+                    else 2,
+                    -a.priority,
+                )
+        else:
+            # Default to priority only
+            def sort_key(a):  # type: ignore[no-untyped-def]
+                return -a.priority
+
+        candidates.sort(key=sort_key)
+        return candidates[0]  # type: ignore[no-any-return]
+
+    def list_algorithms(
+        self,
+        category: str,
+        ordered: bool = False,
+        tie_break: str = "name",
+    ) -> list[RegisteredAlgorithm]:
+        """List all algorithms in a category.
+
+        Args:
+            category: Algorithm category
+            ordered: If True, sort by priority (highest first)
+            tie_break: Tie-breaking rule: "name" (alphabetical) or "registration" (order registered)
+
+        Returns:
+            List of registered algorithms
+
+        Raises:
+            KeyError: If category not found
+
+        Example:
+            >>> # Get algorithms sorted by priority, ties broken by name
+            >>> algos = registry.list_algorithms("edge_detection", ordered=True, tie_break="name")
+
+            >>> # Get algorithms sorted by priority, ties broken by registration order
+            >>> algos = registry.list_algorithms("edge_detection", ordered=True, tie_break="registration")
+
+        References:
+            EXT-004: Priority System (tie-breaking rules by name or registration order)
+        """
+        if category not in self._algorithms:  # type: ignore[attr-defined]
+            raise KeyError(f"Category '{category}' not found")
+
+        algos = list(self._algorithms[category].values())  # type: ignore[attr-defined]
+
+        if ordered:
+            if tie_break == "registration":
+                # Sort by priority (highest first), then by registration order for ties
+                algos.sort(key=lambda a: (-a.priority, a.registration_order))
+            else:
+                # Sort by priority (highest first), then by name for ties (default)
+                algos.sort(key=lambda a: (-a.priority, a.name))
+
+        return algos
+
+    def list_categories(self) -> list[str]:
+        """List all algorithm categories.
+
+        Returns:
+            List of category names
+        """
+        return list(self._algorithms.keys())  # type: ignore[attr-defined]
+
+    def benchmark_algorithms(
+        self,
+        category: str,
+        test_data: Any,
+        *,
+        metrics: list[str] | None = None,
+        iterations: int = 10,
+    ) -> dict[str, dict[str, float]]:
+        """Benchmark all algorithms in a category.
+
+        Runs performance tests on all registered algorithms and measures
+        execution time, memory usage, and optionally custom metrics.
+
+        Args:
+            category: Algorithm category to benchmark
+            test_data: Test data to pass to algorithms
+            metrics: List of metrics to measure (defaults to ["execution_time"])
+            iterations: Number of iterations to average over
+
+        Returns:
+            Dict mapping algorithm names to metric results
+
+        Raises:
+            KeyError: If category is not found.
+
+        Example:
+            >>> import numpy as np
+            >>> test_signal = np.random.randn(1000)
+            >>> results = registry.benchmark_algorithms(
+            ...     "edge_detection",
+            ...     test_signal,
+            ...     metrics=["execution_time", "memory_usage"],
+            ...     iterations=100
+            ... )
+            >>> for name, metrics in results.items():
+            ...     print(f"{name}: {metrics['execution_time']:.3f}s")
+
+        References:
+            EXT-003: Algorithm Selection (benchmarking support)
+        """
+        import time
+        import tracemalloc
+
+        if category not in self._algorithms:  # type: ignore[attr-defined]
+            raise KeyError(f"Category '{category}' not found")
+
+        if metrics is None:
+            metrics = ["execution_time"]
+
+        results = {}
+
+        for name, algo in self._algorithms[category].items():  # type: ignore[attr-defined]
+            algo_results = {}
+
+            if "execution_time" in metrics:
+                times = []
+                for _ in range(iterations):
+                    start = time.perf_counter()
+                    try:
+                        algo.func(test_data)
+                    except Exception as e:
+                        logger.warning(f"Algorithm {name} failed during benchmark: {e}")
+                        times.append(float("inf"))
+                        continue
+                    end = time.perf_counter()
+                    times.append(end - start)
+
+                algo_results["execution_time"] = sum(times) / len(times)
+                algo_results["min_time"] = min(times)
+                algo_results["max_time"] = max(times)
+
+            if "memory_usage" in metrics:
+                tracemalloc.start()
+                try:
+                    algo.func(test_data)
+                    current, peak = tracemalloc.get_traced_memory()
+                    algo_results["memory_current"] = current / 1024 / 1024  # MB
+                    algo_results["memory_peak"] = peak / 1024 / 1024  # MB
+                except Exception as e:
+                    logger.warning(f"Algorithm {name} failed during benchmark: {e}")
+                    algo_results["memory_current"] = float("inf")
+                    algo_results["memory_peak"] = float("inf")
+                finally:
+                    tracemalloc.stop()
+
+            results[name] = algo_results
+
+        return results
+
+    def configure_priorities(self, config: dict[str, dict[str, int]]) -> None:
+        """Override algorithm priorities via configuration.
+
+        Args:
+            config: Dict mapping category -> {algorithm_name: new_priority}
+
+        Example:
+            >>> registry.configure_priorities({
+            ...     "edge_detection": {
+            ...         "fast_detector": 100,
+            ...         "accurate_detector": 50
+            ...     }
+            ... })
+
+        References:
+            EXT-004: Priority System
+        """
+        for category, priorities in config.items():
+            if category not in self._algorithms:  # type: ignore[attr-defined]
+                continue
+            for name, priority in priorities.items():
+                if name in self._algorithms[category]:  # type: ignore[attr-defined]
+                    self._algorithms[category][name].priority = priority  # type: ignore[attr-defined]
+                    logger.debug(f"Set priority for {category}/{name} to {priority}")
+
+    # =========================================================================
+    # Hook System (EXT-005)
+    # =========================================================================
+
+    def register_hook(
+        self,
+        hook_point: str,
+        func: Callable[[HookContext], HookContext],
+        priority: int = 50,
+        name: str = "",
+    ) -> None:
+        """Register a hook function.
+
+        Args:
+            hook_point: Name of hook point (e.g., "pre_decode", "post_decode")
+            func: Hook function accepting and returning HookContext
+            priority: Execution priority (higher = first)
+            name: Optional hook name for identification
+
+        Example:
+            >>> @tk.hooks.register("pre_decode")
+            >>> def validate_waveform(context):
+            ...     if context.data.sample_rate < 1000:
+            ...         raise ValueError("Sample rate too low")
+            ...     return context
+
+        References:
+            EXT-005: Hook System
+        """
+        if hook_point not in self._hooks:  # type: ignore[attr-defined]
+            self._hooks[hook_point] = []  # type: ignore[attr-defined]
+
+        hook = RegisteredHook(
+            hook_point=hook_point,
+            func=func,
+            priority=priority,
+            name=name or func.__name__,
+        )
+
+        self._hooks[hook_point].append(hook)  # type: ignore[attr-defined]
+        # Sort by priority (highest first)
+        self._hooks[hook_point].sort(key=lambda h: -h.priority)  # type: ignore[attr-defined]
+
+        logger.debug(f"Registered hook '{hook.name}' at point '{hook_point}'")
+
+    def execute_hooks(self, hook_point: str, context: HookContext) -> HookContext:
+        """Execute all hooks at a hook point with chaining and error isolation.
+
+        Hooks are executed in priority order (highest first). Each hook receives
+        the context from the previous hook (chaining). If a hook fails, the error
+        is isolated based on the configured error policy, preventing one hook's
+        failure from stopping other hooks.
+
+        Args:
+            hook_point: Hook point name
+            context: Hook context to pass through
+
+        Returns:
+            Modified context after all hooks
+
+        Raises:
+            Exception: If error policy is ABORT and a hook fails.
+
+        Example:
+            >>> context = HookContext(data=trace)
+            >>> context = registry.execute_hooks("pre_decode", context)
+            >>> if context.abort:
+            ...     raise ValueError(context.abort_reason)
+
+        References:
+            EXT-005: Hook System (hook chaining, error isolation)
+        """
+        if hook_point not in self._hooks:  # type: ignore[attr-defined]
+            return context
+
+        # Execute hooks in priority order (hook chaining - EXT-005)
+        for hook in self._hooks[hook_point]:  # type: ignore[attr-defined]
+            try:
+                context = hook.func(context)
+                if context.abort:
+                    logger.info(f"Hook '{hook.name}' requested abort: {context.abort_reason}")
+                    break
+            except Exception as e:
+                # Error isolation - EXT-005: one hook failure doesn't stop others
+                if self._log_hook_errors:
+                    logger.error(f"Hook '{hook.name}' at '{hook_point}' failed: {e}")
+
+                if self._hook_error_policy == HookErrorPolicy.ABORT:
+                    raise
+                elif self._hook_error_policy == HookErrorPolicy.CONTINUE:
+                    continue  # Continue to next hook despite error
+                # IGNORE falls through
+
+        return context
+
+    def configure_hooks(self, on_error: str = "continue", log_errors: bool = True) -> None:
+        """Configure hook error handling behavior.
+
+        Args:
+            on_error: Error policy: "continue", "abort", "ignore"
+            log_errors: Whether to log hook errors
+
+        References:
+            EXT-005: Hook System
+        """
+        policy_map = {
+            "continue": HookErrorPolicy.CONTINUE,
+            "abort": HookErrorPolicy.ABORT,
+            "ignore": HookErrorPolicy.IGNORE,
+        }
+        self._hook_error_policy = policy_map.get(on_error, HookErrorPolicy.CONTINUE)
+        self._log_hook_errors = log_errors
+
+    def list_hooks(self, hook_point: str | None = None) -> dict[str, list[str]]:
+        """List registered hooks.
+
+        Args:
+            hook_point: Specific hook point, or None for all
+
+        Returns:
+            Dict mapping hook points to list of hook names
+        """
+        if hook_point:
+            if hook_point not in self._hooks:  # type: ignore[attr-defined]
+                return {hook_point: []}
+            return {hook_point: [h.name for h in self._hooks[hook_point]]}  # type: ignore[attr-defined]
+
+        return {point: [h.name for h in hooks] for point, hooks in self._hooks.items()}  # type: ignore[attr-defined]
+
+    def clear_hooks(self, hook_point: str | None = None) -> None:
+        """Clear registered hooks.
+
+        Args:
+            hook_point: Specific hook point to clear, or None for all
+        """
+        if hook_point:
+            self._hooks.pop(hook_point, None)  # type: ignore[attr-defined]
+        else:
+            self._hooks.clear()  # type: ignore[attr-defined]
+
+    # =========================================================================
+    # Custom Decoder Registration (EXT-006)
+    # =========================================================================
+
+    def register_decoder(self, protocol: str, decoder_class: type, priority: int = 50) -> None:
+        """Register a custom protocol decoder.
+
+        Args:
+            protocol: Protocol name (e.g., "uart", "spi", "my_custom")
+            decoder_class: Decoder class implementing ProtocolDecoder interface
+            priority: Registration priority
+
+        Raises:
+            ValueError: If decoder doesn't implement required interface or lacks documentation
+
+        Example:
+            >>> class MyDecoder:
+            ...     '''Custom decoder for my protocol.'''
+            ...     def decode(self, trace):
+            ...         return []
+            ...     def get_metadata(self):
+            ...         return {"name": "my_decoder"}
+            >>> registry.register_decoder("my_protocol", MyDecoder)
+
+        References:
+            EXT-006: Custom Decoder Registration (validation of decoder interface, documentation requirements)
+        """
+        # Validate decoder implements required interface
+        spec = self.get_point("protocol_decoder")
+        instance = decoder_class()
+        is_valid, missing = spec.validate_implementation(instance)
+
+        if not is_valid:
+            raise ValueError(f"Decoder '{protocol}' missing required methods: {missing}")
+
+        # Check documentation requirements (EXT-006)
+        if not decoder_class.__doc__ or not decoder_class.__doc__.strip():
+            raise ValueError(
+                f"Decoder '{protocol}' must have a docstring documenting its purpose and usage"
+            )
+
+        # Register as algorithm in protocol_decoder category
+        self.register_algorithm(
+            name=protocol,
+            func=decoder_class,
+            category="protocol_decoder",
+            priority=priority,
+            description=decoder_class.__doc__.strip().split("\n")[0]
+            if decoder_class.__doc__
+            else f"Protocol decoder for {protocol}",
+        )
+
+        logger.info(f"Registered custom decoder for protocol: {protocol}")
+
+    def get_decoder(self, protocol: str) -> type:
+        """Get decoder class for a protocol.
+
+        Args:
+            protocol: Protocol name
+
+        Returns:
+            Decoder class
+        """
+        algo = self.get_algorithm("protocol_decoder", protocol)
+        return algo.func  # type: ignore[return-value]
+
+    def list_decoders(self) -> list[str]:
+        """List all registered protocol decoders.
+
+        Returns:
+            List of protocol names
+        """
+        if "protocol_decoder" not in self._algorithms:  # type: ignore[attr-defined]
+            return []
+        return list(self._algorithms["protocol_decoder"].keys())  # type: ignore[attr-defined]
+
+
+# Global registry instance
+_registry = ExtensionPointRegistry()
+
+
+# =========================================================================
+# Module-Level Convenience Functions
+# =========================================================================
+
+
+def get_registry() -> ExtensionPointRegistry:
+    """Get the global extension point registry.
+
+    Returns:
+        Global ExtensionPointRegistry instance
+    """
+    _registry.initialize()
+    return _registry
+
+
+def list_extension_points() -> list[ExtensionPointSpec]:
+    """List all registered extension points.
+
+    Returns:
+        List of extension point specifications
+
+    References:
+        EXT-001: Extension Point Registry
+    """
+    return get_registry().list_points()
+
+
+def get_extension_point(name: str) -> ExtensionPointSpec:
+    """Get extension point by name.
+
+    Args:
+        name: Extension point name
+
+    Returns:
+        Extension point specification
+
+    References:
+        EXT-001: Extension Point Registry
+    """
+    return get_registry().get_point(name)
+
+
+def extension_point_exists(name: str) -> bool:
+    """Check if extension point exists.
+
+    Args:
+        name: Extension point name
+
+    Returns:
+        True if exists
+
+    References:
+        EXT-001: Extension Point Registry
+    """
+    return get_registry().exists(name)
+
+
+def register_extension_point(spec: ExtensionPointSpec) -> None:
+    """Register a new extension point.
+
+    Args:
+        spec: Extension point specification
+
+    References:
+        EXT-001: Extension Point Registry
+    """
+    get_registry().register_point(spec)
+
+
+# Hook decorator
+def hook(hook_point: str, priority: int = 50, name: str = ""):  # type: ignore[no-untyped-def]
+    """Decorator for registering hook functions.
+
+    Args:
+        hook_point: Hook point name
+        priority: Execution priority
+        name: Optional hook name
+
+    Returns:
+        Decorator function that registers the hook.
+
+    Example:
+        >>> @hook("pre_decode", priority=100)
+        >>> def validate_input(context):
+        ...     # validation logic
+        ...     return context
+
+    References:
+        EXT-005: Hook System
+    """
+
+    def decorator(func: Callable[[HookContext], HookContext]):  # type: ignore[no-untyped-def]
+        get_registry().register_hook(hook_point, func, priority, name or func.__name__)
+        return func
+
+    return decorator
+
+
+__all__ = [
+    "ExtensionPointRegistry",
+    "ExtensionPointSpec",
+    "HookContext",
+    "HookErrorPolicy",
+    "RegisteredAlgorithm",
+    "RegisteredHook",
+    "extension_point_exists",
+    "get_extension_point",
+    "get_registry",
+    "hook",
+    "list_extension_points",
+    "register_extension_point",
+]