isa-model 0.3.6__py3-none-any.whl → 0.3.8__py3-none-any.whl

isa_model/serving/api/routes/unified.py

@@ -10,10 +10,12 @@ This is the main API that handles all types of AI requests:
 """
 
 from fastapi import APIRouter, HTTPException, UploadFile, File, Form
+from fastapi.responses import StreamingResponse
 from pydantic import BaseModel, Field
 from typing import Optional, Dict, Any, Union, List
 import logging
 import asyncio
+import json
 from pathlib import Path
 
 from isa_model.client import ISAModelClient
@@ -28,6 +30,7 @@ class UnifiedRequest(BaseModel):
     service_type: str = Field(..., description="Service type (text, vision, audio, image, embedding)")
     model_hint: Optional[str] = Field(None, description="Optional model preference")
     provider_hint: Optional[str] = Field(None, description="Optional provider preference")
+    stream: Optional[bool] = Field(False, description="Enable streaming for text services")
     parameters: Optional[Dict[str, Any]] = Field(default_factory=dict, description="Additional task parameters")
 
 class UnifiedResponse(BaseModel):
@@ -108,6 +111,75 @@ async def unified_invoke(request: UnifiedRequest) -> UnifiedResponse:
             }
         )
 
+@router.post("/stream")
+async def unified_stream(request: UnifiedRequest):
+    """
+    **Unified streaming endpoint for text services**
+    
+    Returns Server-Sent Events (SSE) stream for real-time token generation.
+    Only supports text service types.
+    """
+    try:
+        # Validate streaming request
+        if request.service_type != "text":
+            raise HTTPException(status_code=400, detail="Streaming only supported for text services")
+        
+        # Get ISA client instance (local mode)
+        client = get_isa_client()
+        
+        async def generate_stream():
+            """Generator for SSE streaming"""
+            try:
+                # Use client's streaming method
+                stream_gen = await client.invoke(
+                    input_data=request.input_data,
+                    task=request.task,
+                    service_type=request.service_type,
+                    model_hint=request.model_hint,
+                    provider_hint=request.provider_hint,
+                    stream=True,
+                    **request.parameters
+                )
+                
+                # Stream tokens as SSE format
+                async for token in stream_gen:
+                    # SSE format: "data: {json}\n\n"
+                    token_data = {
+                        "token": token,
+                        "type": "token"
+                    }
+                    yield f"data: {json.dumps(token_data)}\n\n"
+                
+                # Send completion signal
+                completion_data = {
+                    "type": "completion",
+                    "status": "finished"
+                }
+                yield f"data: {json.dumps(completion_data)}\n\n"
+                
+            except Exception as e:
+                logger.error(f"Streaming error: {e}")
+                error_data = {
+                    "type": "error",
+                    "error": str(e)
+                }
+                yield f"data: {json.dumps(error_data)}\n\n"
+        
+        # Return SSE stream response
+        return StreamingResponse(
+            generate_stream(),
+            media_type="text/plain",
+            headers={
+                "Cache-Control": "no-cache",
+                "Connection": "keep-alive",
+                "Content-Type": "text/plain; charset=utf-8"
+            }
+        )
+        
+    except Exception as e:
+        logger.error(f"Streaming setup failed: {e}")
+        raise HTTPException(status_code=500, detail=str(e))
+
 @router.post("/invoke-file", response_model=UnifiedResponse)
 async def unified_invoke_file(
     task: str = Form(...),

{isa_model-0.3.6.dist-info → isa_model-0.3.8.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: isa_model
-Version: 0.3.6
+Version: 0.3.8
 Summary: Unified AI model serving framework
 Author: isA_Model Contributors
 Classifier: Development Status :: 3 - Alpha

{isa_model-0.3.6.dist-info → isa_model-0.3.8.dist-info}/RECORD

@@ -1,5 +1,5 @@
 isa_model/__init__.py,sha256=bAbHdrDVQ-mySC_GJtgfLNI8KPcs2LfReBkIOOtpaQQ,867
-isa_model/client.py,sha256=5u_hqGB1hcFX8MQdVYlCLqzs4ylQFY3rq91h3iTI24c,27500
+isa_model/client.py,sha256=oo2WcxElcD9WMtw6STj3vS9KwjI-DvS_ldqC1X-HRBs,35456
 isa_model/core/config.py,sha256=h9GVTEEMlaJYSCDd0W9q1KtaWTV5V5TawMsKtGuphds,15686
 isa_model/core/pricing_manager.py,sha256=b7HcZsBQ8ZSCzMepOhqN-J9kU43vhTxX5NabQS0aM70,17125
 isa_model/core/types.py,sha256=XLUs442WGNc8E0gF2M-nb6dutD_s-XCfpr2BfGBCA2M,8445
@@ -30,10 +30,11 @@ isa_model/deployment/gpu_int8_ds8/scripts/test_client.py,sha256=aCULgRYzEQj_ELUK
 isa_model/deployment/gpu_int8_ds8/scripts/test_client_os.py,sha256=XXrneTCHUeh1LNRcu-YtZQ5B4pNawlrxC-cTWmJU2A8,936
 isa_model/deployment/runtime/deployed_service.py,sha256=0Z_Hg42oXasEVvuKjwBylJPazcmJYXhS-L9uPainaIg,13400
 isa_model/deployment/services/__init__.py,sha256=JrLlmBlLb6RfiqGMzVVxKZfF5tAKliQqpon_rPoNoeA,216
-isa_model/deployment/services/auto_deploy_vision_service.py,sha256=Luo8FaXsEBoKjGw5HQ7veOnv9Eh0e7U0claXaGa3-1o,19624
+isa_model/deployment/services/auto_deploy_vision_service.py,sha256=bZmkNG2DWvG6DdHfHvUuf8fonygic4vI_A4aogrXzvU,19670
 isa_model/deployment/services/model_service.py,sha256=_ncC--8hr5BUwzCWh59yRXPKIPVLapx_31TorB2DIr8,13492
 isa_model/deployment/services/service_monitor.py,sha256=P1zGoeqkNEJwt9AXZF2qTjfSLRm5PKUa80GJVNDSIdA,15223
 isa_model/deployment/services/service_registry.py,sha256=LQgWQOvoP0lb7mC6WTS6shEt6WuX6xc8rRmcixrKwTc,22765
+isa_model/deployment/services/simple_auto_deploy_vision_service.py,sha256=rfXsv9mh_w5cXHVYxA4fBD5ppyNY4HplsH34xp4WpY8,9882
 isa_model/eval/__init__.py,sha256=CRbxC5SN7ow4ymdALSNTHawqw4f82DEdAb7twNT_Pw0,2447
 isa_model/eval/benchmarks.py,sha256=_L4Vwj2hwf2yhqoleIASO9z5e3LRCClCVEVCQbGt0I8,16885
 isa_model/eval/factory.py,sha256=bm5OVY7HIxdBgjlH1n7e5K1YO4ytv8e4KB7z_JS9HVQ,20737
@@ -46,19 +47,9 @@ isa_model/eval/evaluators/llm_evaluator.py,sha256=yfFJFdxwGV2F3mzEWjZ-0fr9u8SR3A
 isa_model/eval/infrastructure/__init__.py,sha256=fxTdtwAFtjCDOV9MJ3GbhY0A-DqKeTwr_u9WTBnGI_U,648
 isa_model/eval/infrastructure/experiment_tracker.py,sha256=yfMWIAk6oA8Lfer3AtmKg0OEZiGhczmsCD5gmp--uew,15283
 isa_model/inference/__init__.py,sha256=usfuQJ4zYY2RRtHkE-V6LuJ5aN7WJogtPUj9Qmy4Wvw,318
-isa_model/inference/ai_factory.py,sha256=IN-q3jNmcraZ-PWNTHyhdABoxxhIweZYcXO2fr_uXoM,16478
+isa_model/inference/ai_factory.py,sha256=oGtRd4wp6IZOTyI3GVKBNN4AtlnrLS7yFZuq2wvkaUg,19784
 isa_model/inference/base.py,sha256=qwOddnSGI0GUdD6qIdGBPQpkW7UjU3Y-zaZvu70B4WA,1278
 isa_model/inference/adapter/unified_api.py,sha256=67_Ok8W20m6Otf6r9WyOEVpnxondP4UAxOASk9ozDk4,8668
-isa_model/inference/providers/__init__.py,sha256=a83q-LMFv8u47wf0XtxvqOw_mlVgA_90wtuwy02qdDE,581
-isa_model/inference/providers/base_provider.py,sha256=PT-YnGwBu-Jn_4T3iAphkAJw_mYmKVLjUID62vf2_Ow,2711
-isa_model/inference/providers/ml_provider.py,sha256=4oGGF7lVWQ91Qh3h7olyPFoACLxCROaMxUZlDiZrRL4,1661
-isa_model/inference/providers/modal_provider.py,sha256=klRYXECD5TJicodHIElsGNGMAsAWRFhvn4yfCRcqdVs,3654
-isa_model/inference/providers/model_cache_manager.py,sha256=dLRpx7OJweQ5LcSAkU7D0DQRfLtIhG6nGvg4W_gau80,15315
-isa_model/inference/providers/ollama_provider.py,sha256=IfM9XhdzfE1faguzS2-4GfhK30v5kDPecD3l4z2eB1w,3620
-isa_model/inference/providers/openai_provider.py,sha256=tB8FMsMivlRx0cfPJ0Yrxh1DCvuXyyjNFXrO4lMkkhA,5366
-isa_model/inference/providers/replicate_provider.py,sha256=0oi_BglIE6-HYgzLau9ifP8OdpAMO-QkwYk0OXRUzPk,4490
-isa_model/inference/providers/triton_provider.py,sha256=GKlth7cTOx6ERbsXXJ0gDNby3kVGQNULBDt098BXBSU,15258
-isa_model/inference/providers/yyds_provider.py,sha256=KbDsopShs11_G9oX3b2i2NgHIqkZV7HYWe9K9uZLccc,4284
 isa_model/inference/services/__init__.py,sha256=yfLz0YGl8ixk6LfTRL6cRTvZMb9F_Pv1QRgGyNc9xYM,386
 isa_model/inference/services/base_service.py,sha256=fVaSx0CogHK71UEsNJeSyM8mhqmq5_9ePbbSZVi3Al8,5085
 isa_model/inference/services/audio/base_stt_service.py,sha256=sfzAfreFdvEOBHtphoTrQSjb-gCoCOW4WCj6iIe51oU,5804
@@ -117,7 +108,7 @@ isa_model/serving/api/routes/__init__.py,sha256=RIaG9OPg0AjAIVbtMzwnqGyNU-tuQXbd
 isa_model/serving/api/routes/health.py,sha256=NwQcC_bpcaI4YZHTIKbGtg82yQ6QLdp0TwcqbEiqbWs,2208
 isa_model/serving/api/routes/llm.py,sha256=5ZVxWugff0i6VBKz_Nv5CqacMZJsPZEKyoSB6XDrW34,385
 isa_model/serving/api/routes/ui_analysis.py,sha256=-WxLaRKQNHnRh4okB85cWA4blTegpEPZtzHTsF3yeeU,6848
-isa_model/serving/api/routes/unified.py,sha256=rSoHPtMWkGlzFwzzoZeFlCmFGWj2r3q-5QH9VeNQQxA,7074
+isa_model/serving/api/routes/unified.py,sha256=r6O_X9ql2EDqiTWaWz_anPERwfzNnF9ZvSdjqht8WxE,9727
 isa_model/serving/api/routes/vision.py,sha256=U9jxssQYe6igtayUW0C2fcYwqmLRIE15__X-5Ru9J4c,396
 isa_model/serving/api/schemas/__init__.py,sha256=Tu_hzxoKW1ZHpww3-5ER4A2hNuDByZ0rAfrgaJ7Bs-M,275
 isa_model/serving/api/schemas/common.py,sha256=HVaAS7wlvqrwC1gMZ2Cvo0vzHB053x2uOTAwUoY2vsE,696
@@ -141,7 +132,7 @@ isa_model/training/core/config.py,sha256=oqgKpBvtzrN6jwLIQYQ2707lH6nmjrktRiSxp9i
 isa_model/training/core/dataset.py,sha256=XCFsnf0NUMU1dJpdvo_CAMyvXB-9_RCUEiy8TU50e20,7802
 isa_model/training/core/trainer.py,sha256=h5TjqjdFr0Fsv5y4-0siy1KmOlqLfliVaUXybvuoeXU,26932
 isa_model/training/core/utils.py,sha256=Nik0M2ssfNbWqP6fKO0Kfyhzr_H6Q19ioxB-qCYbn5E,8387
-isa_model-0.3.6.dist-info/METADATA,sha256=TMGcK76gGTCDWcXfCp17JuAoWxVN4TfVv5Nu-8mN8JE,12326
-isa_model-0.3.6.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-isa_model-0.3.6.dist-info/top_level.txt,sha256=eHSy_Xb3kNkh2kK11mi1mZh0Wz91AQ5b8k2KFYO-rE8,10
-isa_model-0.3.6.dist-info/RECORD,,
+isa_model-0.3.8.dist-info/METADATA,sha256=rQ6RLtoPX1xoT2BFeRQZTx9gRSeCjRxuTbH3xseKbdE,12326
+isa_model-0.3.8.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
+isa_model-0.3.8.dist-info/top_level.txt,sha256=eHSy_Xb3kNkh2kK11mi1mZh0Wz91AQ5b8k2KFYO-rE8,10
+isa_model-0.3.8.dist-info/RECORD,,

isa_model/inference/providers/__init__.py

@@ -1,19 +0,0 @@
-"""
-Providers - Components for integrating with different model providers
-
-File: isa_model/inference/providers/__init__.py
-This module contains provider implementations for different AI model backends.
-"""
-
-from .base_provider import BaseProvider
-
-__all__ = [
-    "BaseProvider",
-]
-
-# Provider implementations can be imported individually as needed
-# from .triton_provider import TritonProvider
-# from .ollama_provider import OllamaProvider
-# from .yyds_provider import YYDSProvider
-# from .openai_provider import OpenAIProvider
-# from .replicate_provider import ReplicateProvider 

isa_model/inference/providers/base_provider.py

@@ -1,77 +0,0 @@
-from abc import ABC, abstractmethod
-from typing import Dict, List, Any, Optional
-import os
-import logging
-from pathlib import Path
-import dotenv
-
-from isa_model.inference.base import ModelType, Capability
-
-logger = logging.getLogger(__name__)
-
-class BaseProvider(ABC):
-    """Base class for all AI providers - handles API key management"""
-    
-    def __init__(self, config: Optional[Dict[str, Any]] = None):
-        self.config = config or {}
-        self._load_environment_config()
-        self._validate_config()
-    
-    def _load_environment_config(self):
-        """Load configuration from environment variables"""
-        # Load .env file if it exists
-        project_root = Path(__file__).parent.parent.parent.parent
-        env_path = project_root / ".env"
-        
-        if env_path.exists():
-            dotenv.load_dotenv(env_path)
-        
-        # Subclasses should override this to load provider-specific env vars
-        self._load_provider_env_vars()
-    
-    @abstractmethod
-    def _load_provider_env_vars(self):
-        """Load provider-specific environment variables"""
-        pass
-    
-    def _validate_config(self):
-        """Validate that required configuration is present"""
-        # Subclasses should override this to validate provider-specific config
-        pass
-    
-    def get_api_key(self) -> Optional[str]:
-        """Get the API key for this provider"""
-        return self.config.get("api_key")
-    
-    def has_valid_credentials(self) -> bool:
-        """Check if provider has valid credentials"""
-        return bool(self.get_api_key())
-    
-    @abstractmethod
-    def get_capabilities(self) -> Dict[ModelType, List[Capability]]:
-        """Get provider capabilities by model type"""
-        pass
-    
-    @abstractmethod
-    def get_models(self, model_type: ModelType) -> List[str]:
-        """Get available models for given type"""
-        pass
-    
-    def get_config(self) -> Dict[str, Any]:
-        """Get provider configuration (without sensitive data)"""
-        # Return a copy without sensitive information
-        config_copy = self.config.copy()
-        if "api_key" in config_copy:
-            config_copy["api_key"] = "***" if config_copy["api_key"] else ""
-        if "api_token" in config_copy:
-            config_copy["api_token"] = "***" if config_copy["api_token"] else ""
-        return config_copy
-    
-    def get_full_config(self) -> Dict[str, Any]:
-        """Get full provider configuration (including sensitive data) - for internal use only"""
-        return self.config.copy()
-    
-    @abstractmethod
-    def is_reasoning_model(self, model_name: str) -> bool:
-        """Check if the model is optimized for reasoning tasks"""
-        pass

isa_model/inference/providers/ml_provider.py

@@ -1,50 +0,0 @@
-from isa_model.inference.providers.base_provider import BaseProvider
-from isa_model.inference.base import ModelType, Capability
-from typing import Dict, List, Any
-import logging
-
-logger = logging.getLogger(__name__)
-
-class MLProvider(BaseProvider):
-    """Provider for traditional ML models"""
-    
-    def __init__(self, config=None):
-        default_config = {
-            "model_directory": "./models/ml",
-            "cache_models": True,
-            "max_cache_size": 5
-        }
-        
-        merged_config = {**default_config, **(config or {})}
-        super().__init__(config=merged_config)
-        self.name = "ml"
-        
-        logger.info(f"Initialized MLProvider with model directory: {self.config['model_directory']}")
-    
-    def get_capabilities(self) -> Dict[ModelType, List[Capability]]:
-        """Get provider capabilities"""
-        return {
-            ModelType.LLM: [],  # ML models are not LLMs
-            ModelType.EMBEDDING: [],
-            ModelType.VISION: [],
-            "ML": [  # Custom model type for traditional ML
-                "CLASSIFICATION",
-                "REGRESSION", 
-                "CLUSTERING",
-                "FEATURE_EXTRACTION"
-            ]
-        }
-    
-    def get_models(self, model_type: str = "ML") -> List[str]:
-        """Get available ML models"""
-        # In practice, this would scan the model directory
-        return [
-            "fraud_detection_rf",
-            "customer_churn_xgb", 
-            "price_prediction_lr",
-            "recommendation_kmeans"
-        ]
-    
-    def get_config(self) -> Dict[str, Any]:
-        """Get provider configuration"""
-        return self.config

isa_model/inference/providers/modal_provider.py

@@ -1,109 +0,0 @@
-"""
-Modal Provider
-
-Provider for ISA self-hosted Modal services
-No API keys needed since we deploy our own services
-"""
-
-import os
-import logging
-from typing import Dict, Any, Optional, List
-from .base_provider import BaseProvider
-from isa_model.inference.base import ModelType, Capability
-
-logger = logging.getLogger(__name__)
-
-class ModalProvider(BaseProvider):
-    """Provider for ISA Modal services"""
-    
-    def __init__(self, config: Optional[Dict[str, Any]] = None):
-        super().__init__(config)
-        self.name = "modal"
-        self.base_url = "https://modal.com"  # Not used directly
-    
-    def _load_provider_env_vars(self):
-        """Load Modal-specific environment variables"""
-        # Modal doesn't need API keys for deployed services
-        # But we can load Modal token if available
-        modal_token = os.getenv("MODAL_TOKEN_ID") or os.getenv("MODAL_TOKEN_SECRET")
-        if modal_token:
-            self.config["modal_token"] = modal_token
-        
-        # Set default config
-        if "timeout" not in self.config:
-            self.config["timeout"] = 300
-        if "deployment_region" not in self.config:
-            self.config["deployment_region"] = "us-east-1"
-        if "gpu_type" not in self.config:
-            self.config["gpu_type"] = "T4"
-        
-    def get_api_key(self) -> str:
-        """Modal services don't need API keys for deployed apps"""
-        return "modal-deployed-service"  # Placeholder
-    
-    def get_base_url(self) -> str:
-        """Get base URL for Modal services"""
-        return self.base_url
-    
-    def validate_credentials(self) -> bool:
-        """
-        Validate Modal credentials
-        For deployed services, we assume they're accessible
-        """
-        try:
-            # Check if Modal is available
-            import modal
-            return True
-        except ImportError:
-            logger.warning("Modal package not available")
-            return False
-    
-    def get_capabilities(self) -> Dict[ModelType, List[Capability]]:
-        """Get Modal provider capabilities"""
-        return {
-            ModelType.VISION: [
-                Capability.OBJECT_DETECTION,
-                Capability.IMAGE_ANALYSIS,
-                Capability.UI_DETECTION,
-                Capability.OCR,
-                Capability.DOCUMENT_ANALYSIS
-            ]
-        }
-    
-    def get_models(self, model_type: ModelType) -> List[str]:
-        """Get available models for given type"""
-        if model_type == ModelType.VISION:
-            return [
-                "omniparser-v2.0",
-                "table-transformer-detection",
-                "table-transformer-structure-v1.1",
-                "paddleocr-3.0",
-                "yolov8"
-            ]
-        return []
-    
-    def is_reasoning_model(self, model_name: str) -> bool:
-        """Check if the model is optimized for reasoning tasks"""
-        # Vision models are not reasoning models
-        return False
-    
-    def get_default_config(self) -> Dict[str, Any]:
-        """Get default configuration for Modal services"""
-        return {
-            "timeout": 300,  # 5 minutes
-            "max_retries": 3,
-            "deployment_region": "us-east-1",
-            "gpu_type": "T4"
-        }
-    
-    def get_billing_info(self) -> Dict[str, Any]:
-        """Get billing information for Modal services"""
-        return {
-            "provider": "modal",
-            "billing_model": "compute_usage",
-            "cost_per_hour": {
-                "T4": 0.60,
-                "A100": 4.00
-            },
-            "note": "Costs depend on actual usage time, scales to zero when not in use"
-        }