Add PostgreSQL vector store support with pgvector and Add config to Setting

README.md

@@ -202,6 +202,7 @@ For embeddings models, you can use these providers :
 For your vector store, you can use :
 
 - Chroma (`Settings.CHROMA`)
+- Postgres (`Settings.POSTGRES`)
 
 ## Quick Start 🚀
 

pyproject.toml

@@ -33,6 +33,7 @@ dependencies = [
     "python-dotenv>=1.1.1",
     "pydantic>=2.11.7",
     "google-genai>=1.2.0",
+    "langchain-postgres==0.0.16",
 ]
 [project.scripts]
 raglight = "raglight.cli.main:app"

src/raglight/config/settings.py

@@ -18,6 +18,13 @@ def setup_logging() -> None:
         )
 
     CHROMA = "Chroma"
+    POSTGRES = "Postgres"
+    POSTGRES_HOST = os.environ.get("POSTGRES_HOST", "localhost")
+    POSTGRES_PORT = os.environ.get("POSTGRES_PORT", 5432)
+    POSTGRES_DATABASE = os.environ.get("POSTGRES_DATABASE", "")
+    POSTGRES_USER = os.environ.get("POSTGRES_USER", "postgres")
+    POSTGRES_PASSWORD = os.environ.get("POSTGRES_PASSWORD", "")
+
     OLLAMA = "Ollama"
     MISTRAL = "Mistral"
     VLLM = "vLLM"

src/raglight/rag/builder.py

@@ -13,6 +13,7 @@
 from ..llm.openai_model import OpenAIModel
 from ..vectorstore.vector_store import VectorStore
 from ..vectorstore.chroma import ChromaVS
+from ..vectorstore.postgres import PostgresVS
 from ..config.settings import Settings
 from .rag import RAG
 from ..rat.rat import RAT
@@ -119,6 +120,8 @@ def with_vector_store(self, type: str, **kwargs) -> Builder:
             )
         elif type == Settings.CHROMA:
             self.vector_store = ChromaVS(embeddings_model=self.embeddings, **kwargs)
+        elif type == Settings.POSTGRES:
+            self.vector_store = PostgresVS(embeddings_model=self.embeddings, **kwargs)
         else:
             raise ValueError(f"Unknown VectorStore type: {type}")
         logging.info("✅ VectorStore created")

src/raglight/vectorstore/postgres.py

@@ -0,0 +1,126 @@
+import logging
+from typing import List, Dict
+from typing_extensions import override
+from langchain_core.documents import Document
+
+from ..config.settings import Settings
+
+from .vector_store import VectorStore
+from ..embeddings.embeddings_model import EmbeddingsModel
+
+from enum import auto
+from langchain_postgres import PGVector
+
+class PostgresVS(VectorStore):
+    """
+    Concrete implementation for PostgresVDB.
+
+    It inherits the main ingestion logic from the base VectorStore class and
+    only implements the Postgres-specific methods for adding documents and
+    performing searches.
+    """
+
+    def __init__(
+        self,
+        collection_name: str,
+        embeddings_model: EmbeddingsModel,
+        persist_directory: str = None,
+        host: str = Settings.POSTGRES_HOST,
+        port: int = Settings.POSTGRES_PORT,
+        database: str = Settings.POSTGRES_DATABASE,
+        user: str = Settings.POSTGRES_USER,
+        password: str = Settings.POSTGRES_PASSWORD
+
+    ) -> None:
+        """
+        Initializes a PostgresVS instance.
+        """
+        super().__init__(persist_directory, embeddings_model)
+
+        self.persist_directory = persist_directory
+        self.host = host
+        self.port = port
+        self.database = database
+        connection_string = f"postgresql://{user}:{password}@{host}:{port}/{database}"
+
+        if host and port:
+
+            self.vector_store = PGVector(
+                collection_name=collection_name,
+                connection=connection_string,
+                embeddings=embeddings_model,
+                use_jsonb=True,
+            )
+
+            self.vector_store_classes = PGVector(
+                collection_name=f"{collection_name}_classes",
+                connection=connection_string,
+                embeddings=self.embeddings_model,
+                use_jsonb=True,
+            )
+        else:
+            raise ValueError(
+                "Invalid configuration for PostgresVS: "
+                "You must either:\n"
+                "  • Provide both host and port (for PostgresDB), OR\n"
+                "  • Provide both user and password  (for PostgreDsB).\n"
+                "  • Provide exist database   (for PostgresDB).\n"
+
+                f"Received -> host={host}, port={port}, user={user}, password={'*' * len(password) if password else None}, database={database}"
+            )
+
+    @override
+    def add_documents(self, documents: List[Document]) -> None:
+        """
+        Implements the logic to add documents specifically to the main PostgresDB collection,
+        using batching for efficiency.
+        """
+        if not documents:
+            return
+
+        # logging.info(
+        #     f"⏳ Adding {len(documents)} document chunks to PostgresDB collection '{self.vector_store._collection_name}'..."
+        # )
+        self.vector_store.add_documents(documents=documents)
+        logging.info("✅ Documents successfully added to the main collection.")
+
+    @override
+    def add_class_documents(self, documents: List[Document]) -> None:
+        """
+        Implements the logic to add class signature documents to the dedicated PostgresDB
+        collection for classes.
+        """
+        if not documents:
+            return
+
+        logging.info(
+            f"⏳ Adding {len(documents)} class documents to PostgresDB collection '{self.vector_store_classes._collection_name}'..."
+        )
+        self.vector_store_classes.add_documents(documents=documents)
+        logging.info("✅ Class documents successfully added to the class collection.")
+
+    @override
+    def similarity_search(
+        self,
+        question: str,
+        k: int = 5,
+        filter: Dict[str, str] = None,
+        collection_name: str = None,
+    ) -> List[Document]:
+        """
+        Implements similarity search using the main PostgresDB client.
+        """
+        return self.vector_store.similarity_search(question, k=k, filter=filter)
+
+    @override
+    def similarity_search_class(
+        self,
+        question: str,
+        k: int = 5,
+        filter: Dict[str, str] = None,
+        collection_name: str = None,
+    ) -> List[Document]:
+        """
+        Implements similarity search using the dedicated class PostgresDB client.
+        """
+        pass

tests/test_config.py

@@ -8,3 +8,9 @@ class TestsConfig:
     SYSTEM_PROMPT = "/Users/labess40/dev/RAGLight/systemPrompt.txt"
     COLLECTION_NAME = "grece"
     DATA_PATH = "/Users/labess40/Desktop/nutrition/grece"
+    OLLAMA_EMBEDDING_MODEL="nomic-embed-text:v1.5"
+    POSTGRES_HOST = "192.168.1.103"
+    POSTGRES_PORT=5432
+    POSTGRES_DATABASE="ai_db"
+    POSTGRES_USER=""
+    POSTGRES_PASSWORD=""

tests/tests_vector_store/test_postgres_vector_store.py

@@ -0,0 +1,25 @@
+import unittest
+from ..test_config import TestsConfig
+
+from ...src.raglight.embeddings.ollama_embeddings import OllamaEmbeddingsModel
+from ...src.raglight.vectorstore.postgres import PostgresVS
+
+
+class TestVectorStore(unittest.TestCase):
+    def setUp(self):
+        model_embeddings = TestsConfig.OLLAMA_EMBEDDING_MODEL
+        collection_name = TestsConfig.COLLECTION_NAME
+        self.data_path = TestsConfig.DATA_PATH
+        embeddings = OllamaEmbeddingsModel(model_embeddings)
+        self.store = PostgresVS(
+            embeddings_model=embeddings,
+            collection_name=collection_name,
+        )
+
+    def test_ingest(self):
+        self.store.ingest(data_path=self.data_path)
+        self.assertEqual(True, True, "Embedding should be added to the store.")
+
+
+if __name__ == "__main__":
+    unittest.main()