isa-model 0.3.5__py3-none-any.whl → 0.3.7__py3-none-any.whl

isa_model/scripts/register_models_with_embeddings.py

@@ -0,0 +1,510 @@
+#!/usr/bin/env python3
+"""
+增强的模型注册脚本
+从 YAML 配置文件中读取模型定义，注册到 Supabase 数据库，并生成向量嵌入
+
+Features:
+- 支持注册到 Supabase 的 models 和 model_capabilities 表
+- 自动生成模型描述的向量嵌入并存储到 model_embedding 表
+- 支持批量注册、更新和验证
+- 支持新的 omni 模型类型
+
+Usage:
+    python register_models_with_embeddings.py --all                    # 注册所有providers的模型
+    python register_models_with_embeddings.py --provider openai        # 只注册openai的模型
+    python register_models_with_embeddings.py --dry-run               # 仅验证配置，不实际注册
+    python register_models_with_embeddings.py --update                # 更新已存在的模型
+    python register_models_with_embeddings.py --embeddings-only       # 只生成嵌入，不注册模型
+"""
+
+import argparse
+import asyncio
+import logging
+import sys
+import json
+from pathlib import Path
+from typing import Dict, List, Any, Optional
+import yaml
+
+# 添加项目根目录到Python路径
+project_root = Path(__file__).parent.parent
+sys.path.insert(0, str(project_root))
+
+# 配置日志
+logging.basicConfig(
+    level=logging.INFO,
+    format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
+)
+logger = logging.getLogger(__name__)
+
+try:
+    from isa_model.core.storage.supabase_storage import SupabaseModelRegistry
+    from isa_model.inference.ai_factory import AIFactory
+    SUPABASE_AVAILABLE = True
+except ImportError as e:
+    logger.error(f"Required dependencies not available: {e}")
+    SUPABASE_AVAILABLE = False
+
+
+class EnhancedModelRegistrationScript:
+    """增强的模型注册脚本，支持 Supabase 和向量嵌入"""
+    
+    def __init__(self):
+        if not SUPABASE_AVAILABLE:
+            raise ImportError("Supabase and other required dependencies not available")
+        
+        self.registry = SupabaseModelRegistry()
+        self.ai_factory = AIFactory.get_instance()
+        self.embedding_service = None
+        self.config_dir = project_root / "config" / "models"
+        
+        # 支持的模型类型（包含新的 omni）
+        self.supported_model_types = {
+            "vision", "image", "audio", "text", "embedding", "omni"
+        }
+        
+        # 支持的能力
+        self.supported_capabilities = {
+            "text_generation", "chat", "reasoning", "code_generation",
+            "image_analysis", "image_understanding", "ocr", "ui_detection", 
+            "table_detection", "image_generation", "style_transfer",
+            "text_to_speech", "speech_to_text", "audio_transcription",
+            "text_embedding", "image_enhancement"
+        }
+    
+    async def initialize_embedding_service(self):
+        """初始化嵌入服务"""
+        try:
+            self.embedding_service = self.ai_factory.get_embed("text-embedding-3-small", "openai")
+            logger.info("Embedding service initialized")
+        except Exception as e:
+            logger.warning(f"Failed to initialize embedding service: {e}")
+            self.embedding_service = None
+    
+    def load_provider_config(self, provider: str) -> Dict[str, Any]:
+        """加载 provider 的 YAML 配置"""
+        config_file = self.config_dir / f"{provider}_models.yaml"
+        
+        if not config_file.exists():
+            raise FileNotFoundError(f"Configuration file not found: {config_file}")
+        
+        try:
+            with open(config_file, 'r', encoding='utf-8') as f:
+                config = yaml.safe_load(f)
+            
+            logger.info(f"Loaded configuration for provider '{provider}' with {len(config.get('models', []))} models")
+            return config
+            
+        except yaml.YAMLError as e:
+            raise ValueError(f"Invalid YAML in {config_file}: {e}")
+    
+    def validate_model_config(self, model_config: Dict[str, Any], provider: str) -> bool:
+        """验证单个模型配置"""
+        required_fields = ["model_id", "model_type", "capabilities", "metadata"]
+        
+        for field in required_fields:
+            if field not in model_config:
+                logger.error(f"Missing required field '{field}' in model config: {model_config}")
+                return False
+        
+        # 验证 model_type
+        model_type = model_config["model_type"]
+        if model_type not in self.supported_model_types:
+            logger.error(f"Invalid model_type '{model_type}' for model {model_config['model_id']}")
+            logger.info(f"Supported types: {', '.join(self.supported_model_types)}")
+            return False
+        
+        # 验证 capabilities
+        capabilities = model_config["capabilities"]
+        if not isinstance(capabilities, list) or not capabilities:
+            logger.error(f"Capabilities must be a non-empty list for model {model_config['model_id']}")
+            return False
+        
+        for cap in capabilities:
+            if cap not in self.supported_capabilities:
+                logger.warning(f"Unknown capability '{cap}' for model {model_config['model_id']}")
+                # 不失败，只是警告
+        
+        # 验证 metadata
+        metadata = model_config["metadata"]
+        required_metadata = ["description", "performance_tier", "provider_model_name"]
+        
+        for field in required_metadata:
+            if field not in metadata:
+                logger.error(f"Missing required metadata field '{field}' for model {model_config['model_id']}")
+                return False
+        
+        return True
+    
+    def validate_provider_config(self, config: Dict[str, Any], provider: str) -> bool:
+        """验证 provider 的完整配置"""
+        if "provider" not in config:
+            logger.error(f"Missing 'provider' field in config")
+            return False
+        
+        if config["provider"] != provider:
+            logger.error(f"Provider mismatch: expected '{provider}', got '{config['provider']}'")
+            return False
+        
+        if "models" not in config:
+            logger.error(f"Missing 'models' field in config")
+            return False
+        
+        models = config["models"]
+        if not isinstance(models, list) or not models:
+            logger.error(f"Models must be a non-empty list")
+            return False
+        
+        # 验证每个模型
+        valid_count = 0
+        for i, model_config in enumerate(models):
+            if self.validate_model_config(model_config, provider):
+                valid_count += 1
+            else:
+                logger.error(f"Invalid configuration for model #{i+1}")
+        
+        if valid_count != len(models):
+            logger.error(f"Only {valid_count}/{len(models)} models have valid configuration")
+            return False
+        
+        logger.info(f"All {len(models)} models in {provider} configuration are valid")
+        return True
+    
+    def create_search_text(self, model_config: Dict[str, Any]) -> str:
+        """创建用于嵌入的搜索文本"""
+        metadata = model_config.get("metadata", {})
+        
+        # 组合描述、能力和专业任务
+        parts = []
+        
+        # 添加描述
+        description = metadata.get("description", "")
+        if description:
+            parts.append(description)
+        
+        # 添加能力
+        capabilities = model_config.get("capabilities", [])
+        if capabilities:
+            parts.append(f"Capabilities: {', '.join(capabilities)}")
+        
+        # 添加专业任务
+        specialized_tasks = metadata.get("specialized_tasks", [])
+        if specialized_tasks:
+            parts.append(f"Tasks: {', '.join(specialized_tasks)}")
+        
+        # 添加性能层级
+        performance_tier = metadata.get("performance_tier", "")
+        if performance_tier:
+            parts.append(f"Performance: {performance_tier}")
+        
+        return " ".join(parts)
+    
+    async def create_embedding(self, search_text: str) -> Optional[List[float]]:
+        """创建文本的向量嵌入"""
+        if not self.embedding_service:
+            logger.warning("Embedding service not available")
+            return None
+        
+        try:
+            embedding = await self.embedding_service.create_text_embedding(search_text)
+            return embedding
+        except Exception as e:
+            logger.error(f"Failed to create embedding: {e}")
+            return None
+    
+    async def register_model(self, model_config: Dict[str, Any], provider: str, update: bool = False, embeddings_only: bool = False) -> bool:
+        """注册单个模型到 Supabase"""
+        try:
+            model_id = model_config["model_id"]
+            model_type = model_config["model_type"]
+            capabilities = model_config["capabilities"]
+            metadata = model_config["metadata"].copy()
+            metadata["provider"] = provider
+            
+            # 检查模型是否已存在
+            existing_model = self.registry.get_model_info(model_id)
+            
+            if existing_model and not update and not embeddings_only:
+                logger.warning(f"Model {model_id} already exists. Use --update to overwrite.")
+                return False
+            
+            success = True
+            
+            # 注册模型到基础表（除非只生成嵌入）
+            if not embeddings_only:
+                if existing_model and update:
+                    logger.info(f"Updating existing model {model_id}")
+                else:
+                    logger.info(f"Registering new model {model_id}")
+                
+                reg_success = self.registry.register_model(
+                    model_id=model_id,
+                    model_type=model_type,
+                    capabilities=capabilities,
+                    metadata=metadata
+                )
+                
+                if not reg_success:
+                    logger.error(f"❌ Failed to register model {model_id} to basic tables")
+                    success = False
+            
+            # 生成并存储嵌入
+            if self.embedding_service:
+                try:
+                    search_text = self.create_search_text(model_config)
+                    embedding = await self.create_embedding(search_text)
+                    
+                    if embedding:
+                        # 存储到 model_embedding 表
+                        embed_success = await self._store_model_embedding(
+                            model_id, provider, model_type, 
+                            metadata.get("description", ""),
+                            search_text, metadata, embedding
+                        )
+                        
+                        if embed_success:
+                            logger.info(f"✅ Created embedding for {model_id}")
+                        else:
+                            logger.warning(f"⚠️ Failed to store embedding for {model_id}")
+                    else:
+                        logger.warning(f"⚠️ Failed to create embedding for {model_id}")
+                        
+                except Exception as e:
+                    logger.error(f"❌ Error creating embedding for {model_id}: {e}")
+            
+            if success:
+                action = "embedding generated" if embeddings_only else "registered"
+                logger.info(f"✅ Successfully {action} for {model_id}")
+            
+            return success
+            
+        except Exception as e:
+            logger.error(f"❌ Error processing model {model_config.get('model_id', 'unknown')}: {e}")
+            return False
+    
+    async def _store_model_embedding(
+        self, 
+        model_id: str, 
+        provider: str, 
+        model_type: str,
+        description: str,
+        search_text: str, 
+        metadata: Dict[str, Any], 
+        embedding: List[float]
+    ) -> bool:
+        """存储模型嵌入到数据库"""
+        try:
+            # 这里需要直接访问 Supabase 客户端
+            supabase = self.registry.supabase
+            
+            embedding_data = {
+                'model_id': model_id,
+                'provider': provider,
+                'model_type': model_type,
+                'description': description,
+                'search_text': search_text,
+                'metadata': json.dumps(metadata),
+                'embedding': embedding
+            }
+            
+            # 插入或更新
+            result = supabase.table('model_embedding').upsert(
+                embedding_data, 
+                on_conflict='model_id'
+            ).execute()
+            
+            return bool(result.data)
+            
+        except Exception as e:
+            logger.error(f"Failed to store embedding for {model_id}: {e}")
+            return False
+    
+    async def register_provider_models(self, provider: str, dry_run: bool = False, update: bool = False, embeddings_only: bool = False) -> bool:
+        """注册 provider 的所有模型"""
+        try:
+            # 加载配置
+            config = self.load_provider_config(provider)
+            
+            # 验证配置
+            if not self.validate_provider_config(config, provider):
+                logger.error(f"❌ Invalid configuration for provider {provider}")
+                return False
+            
+            if dry_run:
+                logger.info(f"✅ Dry run successful for provider {provider} - configuration is valid")
+                return True
+            
+            # 初始化嵌入服务
+            if not embeddings_only:  # 注册模式需要检查嵌入服务
+                await self.initialize_embedding_service()
+            else:  # 只生成嵌入模式，必须有嵌入服务
+                await self.initialize_embedding_service()
+                if not self.embedding_service:
+                    logger.error("❌ Embedding service required for embeddings-only mode")
+                    return False
+            
+            # 注册所有模型
+            models = config["models"]
+            success_count = 0
+            
+            action = "generating embeddings for" if embeddings_only else "registering"
+            logger.info(f"🚀 Starting {action} {len(models)} models for provider {provider}")
+            
+            for model_config in models:
+                if await self.register_model(model_config, provider, update, embeddings_only):
+                    success_count += 1
+            
+            # 汇总结果
+            if success_count == len(models):
+                logger.info(f"🎉 All {len(models)} models processed successfully for provider {provider}")
+            else:
+                logger.warning(f"⚠️ Only {success_count}/{len(models)} models processed successfully for provider {provider}")
+            
+            return success_count == len(models)
+            
+        except Exception as e:
+            logger.error(f"❌ Failed to process models for provider {provider}: {e}")
+            return False
+    
+    async def register_all_providers(self, dry_run: bool = False, update: bool = False, embeddings_only: bool = False) -> bool:
+        """注册所有 providers 的模型"""
+        providers = ["openai", "replicate", "yyds", "ollama"]
+        
+        action = "embedding generation" if embeddings_only else "registration"
+        logger.info(f"🚀 Starting {action} for all providers: {', '.join(providers)}")
+        
+        overall_success = True
+        results = {}
+        
+        for provider in providers:
+            logger.info(f"\n📁 Processing provider: {provider}")
+            try:
+                success = await self.register_provider_models(provider, dry_run, update, embeddings_only)
+                results[provider] = success
+                if not success:
+                    overall_success = False
+            except FileNotFoundError:
+                logger.warning(f"⚠️ Configuration file not found for provider {provider}, skipping")
+                results[provider] = None
+            except Exception as e:
+                logger.error(f"❌ Unexpected error with provider {provider}: {e}")
+                results[provider] = False
+                overall_success = False
+        
+        # 打印汇总报告
+        logger.info("\n" + "="*60)
+        logger.info("📊 PROCESSING SUMMARY")
+        logger.info("="*60)
+        
+        for provider, success in results.items():
+            if success is True:
+                logger.info(f"✅ {provider}: SUCCESS")
+            elif success is False:
+                logger.error(f"❌ {provider}: FAILED")
+            else:
+                logger.warning(f"⚠️ {provider}: SKIPPED (no config)")
+        
+        if overall_success:
+            action = "embedding generation" if embeddings_only else "model registration"
+            logger.info(f"\n🎉 All {action} completed successfully!")
+        else:
+            logger.error("\n❌ Some operations failed. Check logs for details.")
+        
+        return overall_success
+    
+    def list_available_providers(self) -> List[str]:
+        """列出可用的 provider 配置文件"""
+        providers = []
+        for config_file in self.config_dir.glob("*_models.yaml"):
+            provider = config_file.stem.replace("_models", "")
+            providers.append(provider)
+        return sorted(providers)
+
+
+async def main():
+    """主函数"""
+    parser = argparse.ArgumentParser(
+        description="Register models from YAML configuration files to Supabase with vector embeddings",
+        formatter_class=argparse.RawDescriptionHelpFormatter,
+        epilog="""
+Examples:
+  python register_models_with_embeddings.py --all                    # Register all providers
+  python register_models_with_embeddings.py --provider openai        # Register only OpenAI models
+  python register_models_with_embeddings.py --dry-run --all         # Validate all configs
+  python register_models_with_embeddings.py --update --provider openai  # Update existing models
+  python register_models_with_embeddings.py --embeddings-only --all  # Only generate embeddings
+        """
+    )
+    
+    parser.add_argument("--all", action="store_true", 
+                       help="Process models from all provider config files")
+    parser.add_argument("--provider", type=str,
+                       help="Process models from specific provider (e.g., openai, replicate)")
+    parser.add_argument("--dry-run", action="store_true",
+                       help="Validate configuration without actually processing models")
+    parser.add_argument("--update", action="store_true",
+                       help="Update existing models if they already exist")
+    parser.add_argument("--embeddings-only", action="store_true",
+                       help="Only generate and store embeddings, don't register models")
+    parser.add_argument("--list-providers", action="store_true",
+                       help="List available provider configuration files")
+    
+    args = parser.parse_args()
+    
+    # 检查依赖
+    if not SUPABASE_AVAILABLE:
+        logger.error("❌ Required dependencies not available. Please install supabase-py and other dependencies.")
+        sys.exit(1)
+    
+    # 创建注册脚本实例
+    try:
+        script = EnhancedModelRegistrationScript()
+    except Exception as e:
+        logger.error(f"❌ Failed to initialize script: {e}")
+        sys.exit(1)
+    
+    # 处理 list-providers
+    if args.list_providers:
+        providers = script.list_available_providers()
+        logger.info("Available provider configurations:")
+        for provider in providers:
+            logger.info(f"  - {provider}")
+        return
+    
+    # 验证参数
+    if not args.all and not args.provider:
+        parser.error("Must specify either --all or --provider")
+    
+    if args.all and args.provider:
+        parser.error("Cannot specify both --all and --provider")
+    
+    # 执行处理
+    try:
+        if args.dry_run:
+            logger.info("🔍 Running in DRY-RUN mode - no actual processing will occur")
+        
+        if args.embeddings_only:
+            logger.info("🎯 Running in EMBEDDINGS-ONLY mode - only generating vector embeddings")
+        
+        if args.all:
+            success = await script.register_all_providers(args.dry_run, args.update, args.embeddings_only)
+        else:
+            success = await script.register_provider_models(args.provider, args.dry_run, args.update, args.embeddings_only)
+        
+        if success:
+            logger.info("✅ Script completed successfully")
+            sys.exit(0)
+        else:
+            logger.error("❌ Script completed with errors")
+            sys.exit(1)
+            
+    except KeyboardInterrupt:
+        logger.info("\n⏹️ Processing cancelled by user")
+        sys.exit(1)
+    except Exception as e:
+        logger.error(f"❌ Unexpected error: {e}")
+        sys.exit(1)
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

isa_model/serving/api/fastapi_server.py

@@ -11,7 +11,7 @@ import time
 import logging
 from typing import Dict, Any
 
-from .routes import ui_analysis, vision, llm, health
+from .routes import ui_analysis, vision, llm, health, unified
 from .middleware.request_logger import RequestLoggerMiddleware
 
 logger = logging.getLogger(__name__)
@@ -60,6 +60,11 @@ def create_app(config: Dict[str, Any] = None) -> FastAPI:
     
     # Include routers
     app.include_router(health.router, prefix="/health", tags=["health"])
+    
+    # MAIN UNIFIED API - Single endpoint for all AI services
+    app.include_router(unified.router, prefix="/api/v1", tags=["unified-api"])
+    
+    # Legacy specific endpoints (kept for backward compatibility)
     app.include_router(ui_analysis.router, prefix="/ui-analysis", tags=["ui-analysis"])
     app.include_router(vision.router, prefix="/vision", tags=["vision"])
     app.include_router(llm.router, prefix="/llm", tags=["llm"])