abstractcore 2.4.4__py3-none-any.whl → 2.4.6__py3-none-any.whl

abstractcore/media/handlers/anthropic_handler.py

@@ -0,0 +1,326 @@
+"""
+Anthropic-specific media handler.
+
+This module provides media formatting capabilities specifically for Anthropic's API,
+including support for Claude Vision and document processing.
+"""
+
+from pathlib import Path
+from typing import Dict, Any, List, Optional, Union
+
+from ..base import BaseProviderMediaHandler, MediaProcessingError
+from ..types import MediaContent, MediaType, ContentFormat
+
+
+class AnthropicMediaHandler(BaseProviderMediaHandler):
+    """
+    Media handler for Anthropic API formatting.
+
+    Formats media content according to Anthropic's API specifications for
+    Claude Vision and other multimodal capabilities.
+    """
+
+    def __init__(self, model_capabilities: Optional[Dict[str, Any]] = None, **kwargs):
+        """
+        Initialize Anthropic media handler.
+
+        Args:
+            model_capabilities: Model capabilities from model_capabilities.json
+            **kwargs: Additional configuration
+        """
+        super().__init__("anthropic", model_capabilities, **kwargs)
+
+        # Anthropic-specific configuration
+        self.max_image_size = kwargs.get('max_image_size', 5 * 1024 * 1024)  # 5MB
+        self.max_images_per_message = kwargs.get('max_images_per_message', 20)  # Claude supports up to 20 images
+
+        self.logger.debug(f"Initialized Anthropic media handler with capabilities: {self.capabilities}")
+
+    def _process_internal(self, file_path: Path, media_type: MediaType, **kwargs) -> MediaContent:
+        """
+        Process file using appropriate processor and return Anthropic-formatted content.
+
+        Args:
+            file_path: Path to the file
+            media_type: Type of media
+            **kwargs: Processing options
+
+        Returns:
+            MediaContent formatted for Anthropic
+        """
+        # Use appropriate processor based on media type
+        if media_type == MediaType.IMAGE:
+            from ..processors import ImageProcessor
+            # Ensure maximum resolution for best Anthropic vision results
+            if 'prefer_max_resolution' not in kwargs:
+                kwargs['prefer_max_resolution'] = True
+            processor = ImageProcessor(**kwargs)
+        elif media_type == MediaType.DOCUMENT:
+            if file_path.suffix.lower() == '.pdf':
+                from ..processors import PDFProcessor
+                processor = PDFProcessor(**kwargs)
+            else:
+                from ..processors import TextProcessor
+                processor = TextProcessor(**kwargs)
+        else:
+            from ..processors import TextProcessor
+            processor = TextProcessor(**kwargs)
+
+        # Process the file
+        result = processor.process_file(file_path, **kwargs)
+
+        if not result.success:
+            raise MediaProcessingError(f"Failed to process {file_path}: {result.error_message}")
+
+        return result.media_content
+
+    def format_for_provider(self, media_content: MediaContent) -> Dict[str, Any]:
+        """
+        Format media content for Anthropic API.
+
+        Args:
+            media_content: MediaContent to format
+
+        Returns:
+            Dictionary formatted for Anthropic API
+        """
+        if media_content.media_type == MediaType.IMAGE:
+            return self._format_image_for_anthropic(media_content)
+        elif media_content.media_type in [MediaType.DOCUMENT, MediaType.TEXT]:
+            return self._format_text_for_anthropic(media_content)
+        else:
+            raise MediaProcessingError(f"Unsupported media type for Anthropic: {media_content.media_type}")
+
+    def _format_image_for_anthropic(self, media_content: MediaContent) -> Dict[str, Any]:
+        """
+        Format image content for Anthropic's image format.
+
+        Args:
+            media_content: Image MediaContent
+
+        Returns:
+            Anthropic-compatible image object
+        """
+        if media_content.content_format != ContentFormat.BASE64:
+            raise MediaProcessingError("Anthropic image formatting requires base64 content")
+
+        # Create Anthropic image object
+        return {
+            "type": "image",
+            "source": {
+                "type": "base64",
+                "media_type": media_content.mime_type,
+                "data": media_content.content
+            }
+        }
+
+    def _format_text_for_anthropic(self, media_content: MediaContent) -> Dict[str, Any]:
+        """
+        Format text/document content for Anthropic API.
+
+        Args:
+            media_content: Text/Document MediaContent
+
+        Returns:
+            Anthropic-compatible text object
+        """
+        if isinstance(media_content.content, bytes):
+            content = media_content.content.decode('utf-8')
+        else:
+            content = str(media_content.content)
+
+        return {
+            "type": "text",
+            "text": content
+        }
+
+    def create_multimodal_message(self, text: str, media_contents: List[MediaContent]) -> Dict[str, Any]:
+        """
+        Create a multimodal message for Anthropic API.
+
+        Args:
+            text: Text content
+            media_contents: List of media contents
+
+        Returns:
+            Anthropic-compatible message object
+        """
+        content = []
+
+        # Add text content first (Anthropic prefers text before images)
+        if text.strip():
+            content.append({
+                "type": "text",
+                "text": text
+            })
+
+        # Add media contents
+        image_count = 0
+        for media_content in media_contents:
+            if self.can_handle_media(media_content):
+                # Check image limits
+                if media_content.media_type == MediaType.IMAGE:
+                    if image_count >= self.max_images_per_message:
+                        self.logger.warning(f"Skipping image - exceeded max images per message ({self.max_images_per_message})")
+                        continue
+                    image_count += 1
+
+                formatted_content = self.format_for_provider(media_content)
+                content.append(formatted_content)
+            else:
+                self.logger.warning(f"Skipping unsupported media type: {media_content.media_type}")
+
+        return {
+            "role": "user",
+            "content": content
+        }
+
+    def validate_media_for_model(self, media_content: MediaContent, model: str) -> bool:
+        """
+        Validate if media content is compatible with specific Anthropic model.
+
+        Args:
+            media_content: MediaContent to validate
+            model: Anthropic model name
+
+        Returns:
+            True if compatible, False otherwise
+        """
+        model_lower = model.lower()
+
+        # Vision model validation
+        if media_content.media_type == MediaType.IMAGE:
+            # Check if model supports vision
+            if not self.model_capabilities.get('vision_support', False):
+                return False
+
+            # Check image size
+            if hasattr(media_content, 'metadata'):
+                file_size = media_content.metadata.get('file_size', 0)
+                if file_size > self.max_image_size:
+                    return False
+
+            # Model-specific checks
+            if 'claude-3' in model_lower:
+                return True  # All Claude 3 models support vision
+            elif 'claude-3.5' in model_lower:
+                return True  # All Claude 3.5 models support vision
+            elif 'claude-4' in model_lower:
+                return True  # Future Claude 4 models
+
+        # Text/document validation
+        elif media_content.media_type in [MediaType.TEXT, MediaType.DOCUMENT]:
+            # All Anthropic models support text
+            return True
+
+        # Audio validation (future support)
+        elif media_content.media_type == MediaType.AUDIO:
+            return self.model_capabilities.get('audio_support', False)
+
+        return False
+
+    def estimate_tokens_for_media(self, media_content: MediaContent) -> int:
+        """
+        Estimate token usage for media content.
+
+        Args:
+            media_content: MediaContent to estimate
+
+        Returns:
+            Estimated token count
+        """
+        if media_content.media_type == MediaType.IMAGE:
+            # Anthropic image token estimation
+            # Roughly ~1600 tokens per image for most cases
+            # This varies based on image content and complexity
+            return 1600
+
+        elif media_content.media_type in [MediaType.TEXT, MediaType.DOCUMENT]:
+            # Rough estimation: 3.5 characters per token (slightly better than GPT)
+            content_length = len(str(media_content.content))
+            return int(content_length / 3.5)
+
+        return 0
+
+    def get_model_media_limits(self, model: str) -> Dict[str, Any]:
+        """
+        Get media-specific limits for Anthropic model.
+
+        Args:
+            model: Anthropic model name
+
+        Returns:
+            Dictionary of limits
+        """
+        limits = {
+            'max_images_per_message': self.max_images_per_message,
+            'max_image_size_bytes': self.max_image_size,
+            'supported_image_formats': ['png', 'jpeg', 'jpg', 'gif', 'webp'],
+            'max_image_resolution': '1568x1568'  # Anthropic's documented limit
+        }
+
+        model_lower = model.lower()
+
+        # Model-specific adjustments
+        if 'claude-3.5' in model_lower:
+            limits.update({
+                'parallel_tool_use': self.model_capabilities.get('parallel_tools', False),
+                'disable_parallel_tool_use_option': True
+            })
+        elif 'claude-3' in model_lower and 'opus' in model_lower:
+            limits.update({
+                'parallel_tool_use': False,  # Claude 3 Opus doesn't support parallel tools
+                'max_tools_per_request': 1
+            })
+
+        return limits
+
+    def create_document_analysis_prompt(self, media_contents: List[MediaContent],
+                                      analysis_type: str = "general") -> str:
+        """
+        Create a specialized prompt for document analysis with Claude.
+
+        Args:
+            media_contents: List of media contents (documents/images)
+            analysis_type: Type of analysis ('general', 'summary', 'extract', 'qa')
+
+        Returns:
+            Optimized prompt for document analysis
+        """
+        document_count = len([mc for mc in media_contents if mc.media_type in [MediaType.DOCUMENT, MediaType.TEXT]])
+        image_count = len([mc for mc in media_contents if mc.media_type == MediaType.IMAGE])
+
+        if analysis_type == "summary":
+            return f"""Please analyze and summarize the provided {'documents' if document_count > 0 else 'images'}.
+            Provide a clear, concise summary that captures the key information, main points, and important details.
+
+            {"Documents provided: " + str(document_count) if document_count > 0 else ""}
+            {"Images provided: " + str(image_count) if image_count > 0 else ""}"""
+
+        elif analysis_type == "extract":
+            return f"""Please extract all relevant information from the provided {'documents' if document_count > 0 else 'images'}.
+            Focus on:
+            - Key facts and data points
+            - Important names, dates, and numbers
+            - Structured information like tables or lists
+            - Any actionable items or conclusions
+
+            Present the information in a clear, organized format."""
+
+        elif analysis_type == "qa":
+            return f"""I will ask questions about the provided {'documents' if document_count > 0 else 'images'}.
+            Please read through all the content carefully and be prepared to answer detailed questions about:
+            - Specific information contained in the materials
+            - Relationships between different pieces of information
+            - Analysis and interpretation of the content
+
+            Please confirm that you have reviewed all the provided materials."""
+
+        else:  # general
+            return f"""Please analyze the provided {'documents' if document_count > 0 else 'images'} and provide:
+            1. A brief overview of what is contained in each item
+            2. Key insights or important information
+            3. Any notable patterns, relationships, or conclusions
+            4. Suggestions for how this information might be used or what actions might be taken
+
+            Be thorough but concise in your analysis."""