Vector Stores API Reference

The empire_chain.vector_stores module provides interfaces to various vector databases for efficient similarity search.

QdrantVectorStore

from empire_chain.vector_stores import QdrantVectorStore

Class for interacting with Qdrant vector database.

Constructor

QdrantVectorStore(location: str = ":memory:", collection_name: str = "default")

Parameters: - location (str): Location of the Qdrant database - :memory: for in-memory storage - URL or path for persistent storage - collection_name (str): Name of the collection to use

Methods

add()

def add(self, text: str, embedding: List[float]) -> None

Add text and its embedding to the store.

Parameters: - text (str): The text to store - embedding (List[float]): Vector representation of the text

Example:

store = QdrantVectorStore(":memory:")
store.add("Sample text", embedding_vector)

query()

def query(self, embedding: List[float], k: int = 3) -> List[str]

Retrieve similar texts based on embedding.

Parameters: - embedding (List[float]): Query vector - k (int): Number of results to return

Returns: - List[str]: List of similar texts

Example:

similar_texts = store.query(query_embedding, k=5)

delete()

def delete(self, ids: List[str]) -> None

Delete entries by their IDs.

Parameters: - ids (List[str]): List of IDs to delete

clear()

def clear(self) -> None

Clear all entries from the store.

ChromaVectorStore

from empire_chain.vector_stores import ChromaVectorStore

Class for interacting with ChromaDB.

Constructor

ChromaVectorStore(path: Optional[str] = None, collection_name: str = "default")

Parameters: - path (Optional[str]): Path for persistent storage - collection_name (str): Name of the collection

Methods

add()

def add(self, text: str, embedding: List[float]) -> None

Add text and its embedding to ChromaDB.

Parameters: - text (str): The text to store - embedding (List[float]): Vector representation of the text

Example:

store = ChromaVectorStore()
store.add("Sample text", embedding_vector)

query()

def query(self, embedding: List[float], k: int = 3) -> List[str]

Retrieve similar texts from ChromaDB.

Parameters: - embedding (List[float]): Query vector - k (int): Number of results to return

Returns: - List[str]: List of similar texts

Example:

similar_texts = store.query(query_embedding, k=5)

Common Usage Patterns

Basic RAG Setup

from empire_chain.vector_stores import QdrantVectorStore
from empire_chain.embeddings import OpenAIEmbeddings

# Initialize components
store = QdrantVectorStore(":memory:")
embeddings = OpenAIEmbeddings("text-embedding-3-small")

# Add documents
text = "Your document text here"
embedding = embeddings.embed(text)
store.add(text, embedding)

# Query
query = "Your query here"
query_embedding = embeddings.embed(query)
results = store.query(query_embedding, k=3)

Persistent Storage

# Qdrant with persistent storage
qdrant_store = QdrantVectorStore(
    location="path/to/storage",
    collection_name="my_documents"
)

# ChromaDB with persistent storage
chroma_store = ChromaVectorStore(
    path="path/to/storage",
    collection_name="my_documents"
)

Error Handling

try:
    store = QdrantVectorStore(":memory:")
    store.add(text, embedding)
except Exception as e:
    print(f"Error: {e}")

Best Practices

1. Memory Management

   # For development/testing
   store = QdrantVectorStore(":memory:")
   
   # For production
   store = QdrantVectorStore("path/to/persistent/storage")
   

2. Batch Operations

   # Add multiple documents efficiently
   for text, embedding in zip(texts, embeddings):
       store.add(text, embedding)
   

3. Collection Management

   # Use separate collections for different purposes
   docs_store = QdrantVectorStore(collection_name="documents")
   qa_store = QdrantVectorStore(collection_name="qa_pairs")