Tatry
Docs
Integrations API

Integrations API

The Integrations API provides details on how to integrate the Tatry Python SDK with other frameworks and libraries.

LangChain

The Tatry Python SDK includes components that work out-of-the-box with LangChain to enhance retrieval capabilities.

TatryRetrieverLangchain

The TatryRetrieverLangchain class allows you to use Tatry as a retriever in your LangChain applications.

from content_retriever.integrations.langchain import TatryRetrieverLangchain
from langchain.chains import RetrievalQA
from langchain.chat_models import ChatOpenAI
 
# Initialize LLM
llm = ChatOpenAI(temperature=0)
 
# Initialize Tatry retriever
retriever = TatryRetrieverLangchain(api_key="your-api-key")
 
# Create a simple QA chain
qa_chain = RetrievalQA.from_chain_type(
    llm=llm,
    retriever=retriever,
    chain_type="stuff"  # Simple document concatenation
)
 
# Use the chain
response = qa_chain.run("What are the key features of quantum computing?")
print(response)

Initialization

TatryRetrieverLangchain(api_key: str, base_url: Optional[str] = None, top_k: int = 5)
  • api_key (str): Your Tatry API key.
  • base_url (Optional[str]): The base URL for the Tatry API. Defaults to the production API.
  • top_k (int): The number of documents to retrieve. Defaults to 5.

Usage

The TatryRetrieverLangchain class implements the BaseRetriever interface from LangChain, so it can be used in any LangChain component that expects a retriever.

documents = retriever.get_relevant_documents(query="What are quantum computers?")
  • query (str): The search query.

Returns:

  • A list of Document objects from LangChain.

LlamaIndex

The Tatry Python SDK can be integrated with LlamaIndex to enhance retrieval capabilities in your RAG applications.

TatryRetrieverLlamaIndex

To use Tatry as a retriever in your LlamaIndex applications, you can create a custom Retriever class that uses the Tatry API.

from llama_index.core import VectorStoreIndex, SimpleDirectoryReader, ServiceContext
from llama_index.core.retrievers import BaseRetriever
from llama_index.core.schema import NodeWithScore
from content_retriever.integrations.tatry import TatryRetriever
 
class TatryRetrieverLlamaIndex(BaseRetriever):
    def __init__(self, api_key: str, top_k: int = 5):
        self._tatry_retriever = TatryRetriever(api_key=api_key)
        self._top_k = top_k
 
    def _retrieve(self, query: str, **kwargs):
        documents = self._tatry_retriever.retrieve(query=query, max_results=self._top_k)
        nodes_with_scores = []
        for doc in documents:
            node = NodeWithScore(node=TextNode(text=doc["content"], extra_info=doc["metadata"]), score=doc["relevance_score"])
            nodes_with_scores.append(node)
        return nodes_with_scores
 
# Usage
api_key = "your-api-key"
tatry_retriever = TatryRetrieverLlamaIndex(api_key=api_key, top_k=5)
 
# Build basic LlamaIndex
index = VectorStoreIndex.from_documents([], service_context=service_context)
query_engine = index.as_query_engine(retriever=tatry_retriever)
 
response = query_engine.query("What are the key features of quantum computing?")
print(response)

Usage

The TatryRetrieverLlamaIndex class implements the BaseRetriever interface from LlamaIndex, so it can be used in any LlamaIndex component that expects a retriever.

from content_retriever.integrations.tatry import TatryRetriever
 
tatry_retriever = TatryRetriever(api_key="your-api-key")
documents = tatry_retriever.retrieve(query="What are quantum computers?", max_results=5)
  • query (str): The search query.
  • max_results (int): The number of documents to retrieve. Defaults to 5.

Returns:

  • A list of Document objects.

Next Steps

SourcesUsage