Merge pull request #24 from offx-zinth/master

Master

Author: offx-zinth

ERROR

@@ -6,8 +6,13 @@ services:
       NEO4J_dbms_security_procedures_unrestricted: apoc.*
       NEO4J_dbms_security_procedures_allowlist: apoc.*
     ports:
+<<<<<<< HEAD
+      - "7475:7474" # Host 7475 maps to Container 7474
+      - "7688:7687" # Host 7688 maps to Container 7687
+=======
       - "7474:7474"
       - "7687:7687"
+>>>>>>> 87cfd9650622e51c4c94d43d490450a82a87ad3d
     volumes:
       - neo4j_data:/data
       - neo4j_logs:/logs
@@ -49,4 +54,8 @@ volumes:
   neo4j_data:
   neo4j_logs:
   chroma_data:
-  smp_data:
+<<<<<<< HEAD
+  smp_data:
+=======
+  smp_data:
+>>>>>>> 87cfd9650622e51c4c94d43d490450a82a87ad3d

docker-compose.yml

@@ -21,6 +21,10 @@ dependencies = [
     "tree-sitter-typescript>=0.23",
     "python-dotenv>=1.0",
     "structlog>=24.0",
+<<<<<<< HEAD
+    "chromadb", 
+=======
+>>>>>>> 87cfd9650622e51c4c94d43d490450a82a87ad3d
 ]
 
 [project.optional-dependencies]

pyproject.toml

@@ -0,0 +1,124 @@
+"""Embedding service using NVIDIA NIM or OpenAI."""
+
+from __future__ import annotations
+
+import os
+from typing import Any
+
+import httpx
+
+from smp.logging import get_logger
+
+log = get_logger(__name__)
+
+
+class EmbeddingService:
+    """Generate embeddings via NVIDIA NIM or OpenAI."""
+
+    def __init__(
+        self,
+        provider: str = "nvidia",
+        api_key: str | None = None,
+        model: str | None = None,
+        base_url: str | None = None,
+        dimension: int = 768,
+    ) -> None:
+        self._provider = provider
+        self._api_key = api_key or os.environ.get("NVIDIA_NIM_API_KEY") or os.environ.get("OPENAI_API_KEY", "")
+        self._model = model or os.environ.get("EMBEDDING_MODEL", "nvidia/nv-embed-qa-4")
+        self._base_url = base_url or os.environ.get(
+            "EMBEDDING_BASE_URL", "https://integrate.api.nvidia.com/v1"
+        )
+        self._dimension = dimension
+        self._client: httpx.AsyncClient | None = None
+
+    async def connect(self) -> None:
+        self._client = httpx.AsyncClient(
+            base_url=self._base_url,
+            headers={"Authorization": f"Bearer {self._api_key}"},
+            timeout=60.0,
+        )
+        log.info("embedding_service_connected", provider=self._provider, model=self._model)
+
+    async def close(self) -> None:
+        if self._client:
+            await self._client.aclose()
+            self._client = None
+
+    @property
+    def dimension(self) -> int:
+        return self._dimension
+
+    async def embed(self, text: str) -> list[float]:
+        """Generate embedding for a single text."""
+        if self._client is None:
+            raise RuntimeError("EmbeddingService not connected")
+
+        if self._provider == "nvidia":
+            return await self._embed_nvidia(text)
+        elif self._provider == "openai":
+            return await self._embed_openai(text)
+        else:
+            raise ValueError(f"Unknown provider: {self._provider}")
+
+    async def embed_batch(self, texts: list[str]) -> list[list[float]]:
+        """Generate embeddings for multiple texts."""
+        if self._client is None:
+            raise RuntimeError("EmbeddingService not connected")
+
+        if self._provider == "nvidia":
+            return await self._embed_batch_nvidia(texts)
+        elif self._provider == "openai":
+            return await self._embed_batch_openai(texts)
+        else:
+            raise ValueError(f"Unknown provider: {self._provider}")
+
+    async def _embed_nvidia(self, text: str) -> list[float]:
+        payload = {
+            "input": text,
+            "model": self._model,
+        }
+        response = await self._client.post("/embeddings", json=payload)
+        response.raise_for_status()
+        data = response.json()
+        return data["data"][0]["embedding"]
+
+    async def _embed_batch_nvidia(self, texts: list[str]) -> list[list[float]]:
+        payload = {
+            "input": texts,
+            "model": self._model,
+        }
+        response = await self._client.post("/embeddings", json=payload)
+        response.raise_for_status()
+        data = response.json()
+        return [item["embedding"] for item in data["data"]]
+
+    async def _embed_openai(self, text: str) -> list[float]:
+        payload = {
+            "input": text,
+            "model": self._model,
+        }
+        response = await self._client.post("/embeddings", json=payload)
+        response.raise_for_status()
+        data = response.json()
+        return data["data"][0]["embedding"]
+
+    async def _embed_batch_openai(self, texts: list[str]) -> list[list[float]]:
+        payload = {
+            "input": texts,
+            "model": self._model,
+        }
+        response = await self._client.post("/embeddings", json=payload)
+        response.raise_for_status()
+        data = response.json()
+        return [item["embedding"] for item in data["data"]]
+
+
+def create_embedding_service() -> EmbeddingService:
+    """Create embedding service from environment variables."""
+    provider = os.getenv("EMBEDDING_PROVIDER", "nvidia")
+    api_key = os.getenv("NVIDIA_NIM_API_KEY") or os.getenv("OPENAI_API_KEY")
+    model = os.getenv("EMBEDDING_MODEL")
+    base_url = os.getenv("EMBEDDING_BASE_URL")
+    dimension = int(os.getenv("EMBEDDING_DIMENSION", "768"))
+    return EmbeddingService(provider=provider, api_key=api_key, model=model, base_url=base_url, dimension=dimension)

smp/engine/embedding.py

@@ -1,19 +1,33 @@
+<<<<<<< HEAD
+"""Static semantic enricher with optional LLM-based embedding."""
+=======
 """Static semantic enricher — AST-based extraction.
 
 Extracts docstrings, inline comments, decorators, type annotations,
 and computes source hashes purely from the AST.
 No LLM or embedding generation.
 """
+>>>>>>> 87cfd9650622e51c4c94d43d490450a82a87ad3d
 
 from __future__ import annotations
 
 import hashlib
 from datetime import UTC, datetime
+<<<<<<< HEAD
+from typing import TYPE_CHECKING
+=======
+>>>>>>> 87cfd9650622e51c4c94d43d490450a82a87ad3d
 
 from smp.core.models import GraphNode
 from smp.engine.interfaces import SemanticEnricher as SemanticEnricherInterface
 from smp.logging import get_logger
 
+<<<<<<< HEAD
+if TYPE_CHECKING:
+    from smp.engine.embedding import EmbeddingService
+
+=======
+>>>>>>> 87cfd9650622e51c4c94d43d490450a82a87ad3d
 log = get_logger(__name__)
 
 
@@ -24,15 +38,28 @@ def _compute_source_hash(name: str, file_path: str, start: int, end: int, signat
 
 
 class StaticSemanticEnricher(SemanticEnricherInterface):
+<<<<<<< HEAD
+    """Static AST-based semantic enricher with optional embedding support."""
+
+    def __init__(self, embedding_service: EmbeddingService | None = None) -> None:
+=======
     """Static AST-based semantic enricher. No LLM, no embeddings."""
 
     def __init__(self) -> None:
+>>>>>>> 87cfd9650622e51c4c94d43d490450a82a87ad3d
         self._enrichment_counts: dict[str, int] = {
             "enriched": 0,
             "skipped": 0,
             "no_metadata": 0,
             "failed": 0,
         }
+<<<<<<< HEAD
+        self._embedding_service = embedding_service
+
+    def set_embedding_service(self, service: EmbeddingService) -> None:
+        self._embedding_service = service
+=======
+>>>>>>> 87cfd9650622e51c4c94d43d490450a82a87ad3d
 
     async def enrich_node(
         self,
@@ -83,9 +110,28 @@ async def enrich_batch(
             enriched.append(result)
         return enriched
 
+<<<<<<< HEAD
+    @property
+    def has_llm(self) -> bool:
+        """Check if LLM-based embedding is available."""
+        return self._embedding_service is not None
+
+    async def embed(self, text: str) -> list[float]:
+        """Generate embedding using the embedding service if available."""
+        if self._embedding_service is None:
+            return []
+        return await self._embedding_service.embed(text)
+
+    async def embed_batch(self, texts: list[str]) -> list[list[float]]:
+        """Generate embeddings for multiple texts."""
+        if self._embedding_service is None:
+            return [[] for _ in texts]
+        return await self._embedding_service.embed_batch(texts)
+=======
     async def embed(self, text: str) -> list[float]:
         """No-op embedding — static enricher does not use vectors."""
         return []
+>>>>>>> 87cfd9650622e51c4c94d43d490450a82a87ad3d
 
     def get_counts(self) -> dict[str, int]:
         """Return enrichment statistics."""
@@ -94,4 +140,8 @@ def get_counts(self) -> dict[str, int]:
     def reset_counts(self) -> None:
         """Reset enrichment counters."""
         for key in self._enrichment_counts:
+<<<<<<< HEAD
+            self._enrichment_counts[key] = 0
+=======
             self._enrichment_counts[key] = 0
+>>>>>>> 87cfd9650622e51c4c94d43d490450a82a87ad3d

smp/engine/enricher.py

@@ -55,6 +55,13 @@ async def enrich_batch(self, nodes: list[GraphNode], force: bool = False) -> lis
     async def embed(self, text: str) -> list[float]:
         """No-op for static enricher."""
 
+<<<<<<< HEAD
+    @abc.abstractmethod
+    async def embed_batch(self, texts: list[str]) -> list[list[float]]:
+        """Generate embeddings for multiple texts."""
+
+=======
+>>>>>>> 87cfd9650622e51c4c94d43d490450a82a87ad3d
 
 class QueryEngine(abc.ABC):
     """High-level query interface over the memory store."""