aiecs 1.0.1__py3-none-any.whl → 1.7.6__py3-none-any.whl

aiecs/tools/apisource/providers/base.py

@@ -0,0 +1,671 @@
+"""
+Base API Provider Interface
+
+Abstract base class for all API data source providers in the API Source Tool.
+Provides common functionality for rate limiting, caching, error handling, and metadata.
+
+Enhanced with:
+- Detailed metrics and health monitoring
+- Smart error handling with retries
+- Data quality assessment
+- Comprehensive metadata with quality scores
+- Operation exposure for AI agent visibility
+"""
+
+import logging
+import time
+from abc import ABC, abstractmethod
+from datetime import datetime
+from threading import Lock
+from typing import Any, Dict, List, Optional, Tuple
+
+from aiecs.tools.apisource.monitoring.metrics import DetailedMetrics
+from aiecs.tools.apisource.reliability.error_handler import SmartErrorHandler
+from aiecs.tools.apisource.utils.validators import DataValidator
+
+logger = logging.getLogger(__name__)
+
+
+def expose_operation(operation_name: str, description: str):
+    """
+    Decorator: Mark provider operations that should be exposed to AI agents.
+
+    This decorator allows provider operations to be automatically discovered by the
+    LangChain adapter and exposed as individual tools to AI agents, providing
+    fine-grained visibility into provider capabilities.
+
+    Args:
+        operation_name: The name of the operation (e.g., 'get_series_observations')
+        description: Human-readable description of what the operation does
+
+    Returns:
+        Decorated function with metadata for operation discovery
+
+    Example:
+        @expose_operation(
+            operation_name='get_series_observations',
+            description='Get FRED economic time series data'
+        )
+        def get_series_observations(self, series_id: str, ...):
+            pass
+    """
+
+    def decorator(func):
+        func._exposed_operation = True
+        func._operation_name = operation_name
+        func._operation_description = description
+        return func
+
+    return decorator
+
+
+class RateLimiter:
+    """Token bucket rate limiter for API requests"""
+
+    def __init__(self, tokens_per_second: float = 1.0, max_tokens: int = 10):
+        """
+        Initialize rate limiter with token bucket algorithm.
+
+        Args:
+            tokens_per_second: Rate at which tokens are added to the bucket
+            max_tokens: Maximum number of tokens the bucket can hold
+        """
+        self.tokens_per_second = tokens_per_second
+        self.max_tokens = max_tokens
+        self.tokens = max_tokens
+        self.last_update = time.time()
+        self.lock = Lock()
+
+    def acquire(self, tokens: int = 1) -> bool:
+        """
+        Acquire tokens from the bucket.
+
+        Args:
+            tokens: Number of tokens to acquire
+
+        Returns:
+            True if tokens were acquired, False otherwise
+        """
+        with self.lock:
+            now = time.time()
+            elapsed = now - self.last_update
+
+            # Add new tokens based on elapsed time
+            self.tokens = int(min(self.max_tokens, self.tokens + elapsed * self.tokens_per_second))
+            self.last_update = now
+
+            if self.tokens >= tokens:
+                self.tokens -= tokens
+                return True
+            return False
+
+    def wait(self, tokens: int = 1, timeout: float = 30.0) -> bool:
+        """
+        Wait until tokens are available.
+
+        Args:
+            tokens: Number of tokens to acquire
+            timeout: Maximum time to wait in seconds
+
+        Returns:
+            True if tokens were acquired, False if timeout
+        """
+        start_time = time.time()
+        while time.time() - start_time < timeout:
+            if self.acquire(tokens):
+                return True
+            time.sleep(0.1)
+        return False
+
+
+class BaseAPIProvider(ABC):
+    """
+    Abstract base class for all API data source providers.
+
+    Provides:
+    - Rate limiting with token bucket algorithm
+    - Standardized error handling
+    - Metadata about provider capabilities
+    - Parameter validation
+    - Response formatting
+    """
+
+    def __init__(self, config: Optional[Dict[str, Any]] = None):
+        """
+        Initialize the API provider.
+
+        Args:
+            config: Configuration dictionary with API keys, rate limits, etc.
+        """
+        self.config = config or {}
+        self.logger = logging.getLogger(f"{__name__}.{self.__class__.__name__}")
+
+        # Initialize rate limiter
+        rate_limit = self.config.get("rate_limit", 10)  # requests per second
+        max_burst = self.config.get("max_burst", 20)
+        self.rate_limiter = RateLimiter(tokens_per_second=rate_limit, max_tokens=max_burst)
+
+        # Initialize detailed metrics
+        self.metrics = DetailedMetrics(max_response_times=100)
+
+        # Initialize smart error handler
+        self.error_handler = SmartErrorHandler(
+            max_retries=self.config.get("max_retries", 3),
+            backoff_factor=self.config.get("backoff_factor", 2.0),
+            initial_delay=self.config.get("initial_delay", 1.0),
+            max_delay=self.config.get("max_delay", 30.0),
+        )
+
+        # Initialize data validator
+        self.validator = DataValidator()
+
+        # Legacy stats for backwards compatibility
+        self.stats: Dict[str, Any] = {
+            "total_requests": 0,
+            "successful_requests": 0,
+            "failed_requests": 0,
+            "last_request_time": None,
+        }
+        self.stats_lock = Lock()
+
+    @property
+    @abstractmethod
+    def name(self) -> str:
+        """Provider name (e.g., 'fred', 'worldbank')"""
+
+    @property
+    @abstractmethod
+    def description(self) -> str:
+        """Human-readable description of the provider"""
+
+    @property
+    @abstractmethod
+    def supported_operations(self) -> List[str]:
+        """List of supported operation names"""
+
+    @abstractmethod
+    def validate_params(self, operation: str, params: Dict[str, Any]) -> Tuple[bool, Optional[str]]:
+        """
+        Validate parameters for a specific operation.
+
+        Args:
+            operation: Operation name
+            params: Parameters to validate
+
+        Returns:
+            Tuple of (is_valid, error_message)
+        """
+
+    @abstractmethod
+    def fetch(self, operation: str, params: Dict[str, Any]) -> Dict[str, Any]:
+        """
+        Fetch data from the API.
+
+        Args:
+            operation: Operation to perform
+            params: Operation parameters
+
+        Returns:
+            Response data in standardized format
+
+        Raises:
+            ValueError: If operation is not supported
+            Exception: If API request fails
+        """
+
+    def get_metadata(self) -> Dict[str, Any]:
+        """
+        Get provider metadata including health status and detailed metrics.
+
+        Returns:
+            Dictionary with comprehensive provider information
+        """
+        return {
+            "name": self.name,
+            "description": self.description,
+            "operations": self.supported_operations,
+            "stats": self.metrics.get_summary(),  # Use detailed metrics
+            "health": {
+                "score": self.metrics.get_health_score(),
+                "status": ("healthy" if self.metrics.get_health_score() > 0.7 else "degraded"),
+            },
+            "config": {
+                "rate_limit": self.config.get("rate_limit", 10),
+                "timeout": self.config.get("timeout", 30),
+                "max_retries": self.config.get("max_retries", 3),
+            },
+        }
+
+    def get_operation_schema(self, operation: str) -> Optional[Dict[str, Any]]:
+        """
+        Get schema for a specific operation.
+
+        Args:
+            operation: Operation name
+
+        Returns:
+            Schema dictionary or None if not available
+        """
+        # Override in subclass to provide operation-specific schemas
+        return None
+
+    @classmethod
+    def get_exposed_operations(cls) -> List[Dict[str, Any]]:
+        """
+        Get all operations that are exposed to AI agents via the @expose_operation decorator.
+
+        This method discovers all methods decorated with @expose_operation and returns
+        their metadata along with their schemas. This enables the LangChain adapter to
+        automatically create individual tools for each provider operation.
+
+        Returns:
+            List of operation dictionaries, each containing:
+                - name: Operation name
+                - description: Operation description
+                - schema: Operation schema (parameters, types, descriptions)
+                - method_name: The actual method name on the class
+
+        Example:
+            >>> FREDProvider.get_exposed_operations()
+            [
+                {
+                    'name': 'get_series_observations',
+                    'description': 'Get FRED economic time series data',
+                    'schema': {...},
+                    'method_name': 'get_series_observations'
+                },
+                ...
+            ]
+        """
+        operations = []
+
+        # Create a temporary instance to access get_operation_schema
+        # We need this because get_operation_schema might be an instance method
+        try:
+            # Try to get schema without instantiation first
+            for attr_name in dir(cls):
+                # Skip private and special methods
+                if attr_name.startswith("_"):
+                    continue
+
+                try:
+                    attr = getattr(cls, attr_name)
+                except AttributeError:
+                    continue
+
+                # Check if this is an exposed operation
+                if callable(attr) and hasattr(attr, "_exposed_operation"):
+                    operation_name = attr._operation_name
+                    operation_description = attr._operation_description
+
+                    # Schema retrieval requires an instance, so skip at class level
+                    # Schema will be available at runtime when provider instances are created
+                    schema = None
+
+                    operations.append(
+                        {
+                            "name": operation_name,
+                            "description": operation_description,
+                            "schema": schema,
+                            "method_name": attr_name,
+                        }
+                    )
+
+                    logger.debug(f"Discovered exposed operation: {operation_name} from {cls.__name__}")
+
+        except Exception as e:
+            logger.warning(f"Error discovering exposed operations for {cls.__name__}: {e}")
+
+        return operations
+
+    def validate_and_clean_data(self, operation: str, raw_data: Any) -> Dict[str, Any]:
+        """
+        Validate and clean data (optional, override in subclass).
+
+        Providers can implement custom validation logic for their specific data formats.
+
+        Args:
+            operation: Operation that produced the data
+            raw_data: Raw data from API
+
+        Returns:
+            Dictionary with:
+                - data: Cleaned data
+                - validation_warnings: List of warnings
+                - statistics: Data quality statistics
+        """
+        # Default implementation: no validation
+        return {"data": raw_data, "validation_warnings": [], "statistics": {}}
+
+    def calculate_data_quality(self, operation: str, data: Any, response_time_ms: float) -> Dict[str, Any]:
+        """
+        Calculate quality metadata for the response.
+
+        Can be overridden by providers for custom quality assessment.
+
+        Args:
+            operation: Operation performed
+            data: Response data
+            response_time_ms: Response time in milliseconds
+
+        Returns:
+            Quality metadata dictionary
+        """
+        quality: Dict[str, Any] = {
+            "score": 0.7,  # Default quality score
+            "completeness": 1.0,  # Assume complete unless validated otherwise
+            "freshness_hours": None,  # Unknown freshness
+            "confidence": 0.8,  # Default confidence
+            "authority_level": "verified",  # Provider is verified
+        }
+
+        # Adjust score based on response time
+        if response_time_ms < 500:
+            score = quality.get("score", 0.7)
+            if isinstance(score, (int, float)):
+                quality["score"] = min(score + 0.1, 1.0)
+        elif response_time_ms > 5000:
+            score = quality.get("score", 0.7)
+            if isinstance(score, (int, float)):
+                quality["score"] = max(score - 0.1, 0.0)
+
+        # Check if data is empty
+        if data is None:
+            quality["completeness"] = 0.0
+            quality["score"] = 0.0
+        elif isinstance(data, list) and len(data) == 0:
+            quality["completeness"] = 0.0
+            score = quality.get("score", 0.7)
+            if isinstance(score, (int, float)):
+                quality["score"] = max(score - 0.3, 0.0)
+
+        return quality
+
+    def _update_stats(self, success: bool):
+        """Update request statistics"""
+        with self.stats_lock:
+            total = self.stats.get("total_requests", 0)
+            if isinstance(total, (int, float)):
+                self.stats["total_requests"] = total + 1
+            if success:
+                successful = self.stats.get("successful_requests", 0)
+                if isinstance(successful, (int, float)):
+                    self.stats["successful_requests"] = successful + 1
+            else:
+                failed = self.stats.get("failed_requests", 0)
+                if isinstance(failed, (int, float)):
+                    self.stats["failed_requests"] = failed + 1
+            self.stats["last_request_time"] = datetime.utcnow().isoformat()
+
+    def _format_response(
+        self,
+        operation: str,
+        data: Any,
+        source: Optional[str] = None,
+        response_time_ms: Optional[float] = None,
+        validation_result: Optional[Dict[str, Any]] = None,
+    ) -> Dict[str, Any]:
+        """
+        Format response in standardized format with enhanced metadata.
+
+        Args:
+            operation: Operation that was performed
+            data: Response data
+            source: Data source URL or identifier
+            response_time_ms: Response time in milliseconds
+            validation_result: Optional validation result from validate_and_clean_data
+
+        Returns:
+            Standardized response dictionary with comprehensive metadata
+        """
+        # Calculate quality metadata
+        quality = self.calculate_data_quality(operation, data, response_time_ms or 0)
+
+        # Calculate coverage information
+        coverage = self._calculate_coverage(data)
+
+        # Build metadata
+        metadata = {
+            "timestamp": datetime.utcnow().isoformat(),
+            "source": source or f"{self.name} API",
+            "quality": quality,
+            "coverage": coverage,
+        }
+
+        # Add API info if response time provided
+        if response_time_ms is not None:
+            metadata["api_info"] = {
+                "response_time_ms": round(response_time_ms, 2),
+                "provider": self.name,
+            }
+
+        # Add validation warnings if present
+        if validation_result and validation_result.get("validation_warnings"):
+            metadata["validation_warnings"] = validation_result["validation_warnings"]
+
+        # Add statistics if present
+        if validation_result and validation_result.get("statistics"):
+            metadata["statistics"] = validation_result["statistics"]
+
+        return {
+            "provider": self.name,
+            "operation": operation,
+            "data": data,
+            "metadata": metadata,
+        }
+
+    def _calculate_coverage(self, data: Any) -> Dict[str, Any]:
+        """
+        Calculate data coverage information.
+
+        Args:
+            data: Response data
+
+        Returns:
+            Coverage information dictionary
+        """
+        coverage: Dict[str, Any] = {}
+
+        # Calculate record count
+        if isinstance(data, list):
+            coverage["total_records"] = len(data)
+
+            # Try to extract date range from time series data
+            if len(data) > 0 and isinstance(data[0], dict):
+                date_fields = ["date", "observation_date", "timestamp"]
+                for date_field in date_fields:
+                    if date_field in data[0]:
+                        dates = [item.get(date_field) for item in data if date_field in item and item.get(date_field)]
+                        if dates:
+                            try:
+                                # Sort to get earliest and latest
+                                dates_sorted = sorted(dates)
+                                coverage["start_date"] = dates_sorted[0]
+                                coverage["end_date"] = dates_sorted[-1]
+
+                                # Try to infer frequency
+                                frequency = self.validator.infer_data_frequency(data, date_field)
+                                if frequency:
+                                    coverage["frequency"] = frequency
+                            except Exception:
+                                pass
+                        break
+        elif isinstance(data, dict):
+            # For dict responses
+            if "articles" in data:
+                coverage["total_records"] = len(data["articles"])
+            elif "total_results" in data:
+                coverage["total_results"] = data["total_results"]
+            else:
+                coverage["total_records"] = 1
+        else:
+            coverage["total_records"] = 1 if data is not None else 0
+
+        return coverage
+
+    def _get_api_key(self, key_name: Optional[str] = None) -> Optional[str]:
+        """
+        Get API key from config or environment.
+
+        Args:
+            key_name: Specific key name to retrieve
+
+        Returns:
+            API key or None if not found
+        
+        Note: When used through APISourceTool, API keys are loaded from .env files
+        via BaseSettings and passed via config dict. This fallback is for backward
+        compatibility and independent provider usage.
+        """
+        import os
+
+        # Try config first (primary path - API keys come from APISourceTool's BaseSettings)
+        if "api_key" in self.config:
+            return self.config["api_key"]
+
+        # Fallback: Try environment variable (ensures .env files are loaded)
+        # Use ToolConfigLoader to ensure .env files are loaded if not already
+        try:
+            from aiecs.config.tool_config import get_tool_config_loader
+            
+            loader = get_tool_config_loader()
+            loader.load_env_config()  # Ensures .env files are loaded
+        except Exception:
+            # If loader is unavailable, try direct dotenv load as fallback
+            try:
+                from dotenv import load_dotenv
+                from pathlib import Path
+                
+                # Try to load .env files from common locations
+                for env_file in [".env", ".env.local"]:
+                    env_path = Path(env_file)
+                    if env_path.exists():
+                        load_dotenv(env_path, override=False)
+                        break
+            except Exception:
+                pass  # If dotenv is unavailable, continue with os.environ
+
+        # Try environment variable (now includes values from .env files)
+        env_var = key_name or f"{self.name.upper()}_API_KEY"
+        return os.environ.get(env_var)
+
+    def execute(self, operation: str, params: Dict[str, Any]) -> Dict[str, Any]:
+        """
+        Execute an operation with rate limiting, error handling, and metrics tracking.
+
+        Args:
+            operation: Operation to perform
+            params: Operation parameters
+
+        Returns:
+            Response data with enhanced metadata
+
+        Raises:
+            ValueError: If operation is invalid or parameters are invalid
+            Exception: If API request fails after all retries
+        """
+        # Validate operation
+        if operation not in self.supported_operations:
+            available_ops = ", ".join(self.supported_operations)
+            schema = self.get_operation_schema(operation)
+            operation_error_msg = f"Operation '{operation}' not supported by {self.name}.\n" f"Supported operations: {available_ops}"
+            if schema:
+                operation_error_msg += f"\nSee get_operation_schema('{operation}') for details"
+            raise ValueError(operation_error_msg)
+
+        # Validate parameters with enhanced error messages
+        validation_result = self.validate_params(operation, params)
+        is_valid: bool = validation_result[0]
+        error_msg: Optional[str] = validation_result[1]
+        if not is_valid:
+            schema = self.get_operation_schema(operation)
+            enhanced_error = f"Invalid parameters for {self.name}.{operation}: {error_msg or 'Unknown error'}"
+
+            if schema and "parameters" in schema:
+                # Add helpful parameter information
+                required_params = [name for name, info in schema["parameters"].items() if info.get("required", False)]
+                if required_params:
+                    enhanced_error += f"\nRequired parameters: {', '.join(required_params)}"
+
+                # Add examples if available
+                if "examples" in schema and schema["examples"]:
+                    example = schema["examples"][0]
+                    enhanced_error += f"\nExample: {example.get('params', {})}"
+
+            raise ValueError(enhanced_error)
+
+        # Apply rate limiting
+        wait_start = time.time()
+        if not self.rate_limiter.wait(tokens=1, timeout=30):
+            self.metrics.record_request(success=False, response_time_ms=0, error_type="rate_limit")
+            raise Exception(f"Rate limit exceeded for {self.name}. " "Please try again later or increase rate limits in config.")
+
+        # Track rate limit wait time
+        wait_time_ms = (time.time() - wait_start) * 1000
+        if wait_time_ms > 100:  # Only record significant waits
+            self.metrics.record_rate_limit_wait(wait_time_ms)
+
+        # Execute with smart retry logic
+        def fetch_operation():
+            """Wrapper for fetch with timing"""
+            start_time = time.time()
+            result = self.fetch(operation, params)
+            response_time_ms = (time.time() - start_time) * 1000
+            return result, response_time_ms
+
+        # Use error handler for retries
+        execution_result = self.error_handler.execute_with_retry(
+            operation_func=fetch_operation,
+            operation_name=operation,
+            provider_name=self.name,
+        )
+
+        if execution_result["success"]:
+            result, response_time_ms = execution_result["data"]
+
+            # Calculate data size for metrics
+            data = result.get("data") if isinstance(result, dict) else result
+            record_count = len(data) if isinstance(data, list) else (1 if data else 0)
+
+            # Record success metrics
+            self.metrics.record_request(
+                success=True,
+                response_time_ms=response_time_ms,
+                record_count=record_count,
+                cached=False,
+            )
+
+            # Update legacy stats
+            self._update_stats(success=True)
+
+            self.logger.info(f"Successfully executed {self.name}.{operation} " f"in {response_time_ms:.0f}ms ({record_count} records)")
+
+            return result
+        else:
+            # All retries failed
+            error_info = execution_result["error"]
+            retry_info = execution_result["retry_info"]
+
+            # Record failure metrics
+            self.metrics.record_request(
+                success=False,
+                response_time_ms=0,
+                error_type=error_info.get("type", "unknown"),
+                error_message=error_info.get("message"),
+            )
+
+            # Update legacy stats
+            self._update_stats(success=False)
+
+            # Build comprehensive error message
+            error_msg = f"Failed to execute {self.name}.{operation} after " f"{retry_info['attempts']} attempts.\n" f"Error: {error_info['message']}"
+
+            # Add recovery suggestions
+            if retry_info.get("recovery_suggestions"):
+                error_msg += "\n\nSuggestions:"
+                for suggestion in retry_info["recovery_suggestions"][:3]:
+                    error_msg += f"\n  - {suggestion}"
+
+            self.logger.error(error_msg)
+
+            raise Exception(error_msg)