LlamaIndex Integration

FraiseQLReader is a LlamaIndex BaseReader that executes GraphQL queries and returns the results as Document objects, making FraiseQL a first-class data source for RAG pipelines and knowledge-base agents.

Installation

pip install "fraiseql[llamaindex]"

Basic Usage

import asyncio
from fraiseql.client import FraiseQLClient
from fraiseql.integrations.llamaindex import FraiseQLReader

async def main() -> None:
    client = FraiseQLClient(
        "http://localhost:8080/graphql",
        auth_token="Bearer eyJ..."
    )
    reader = FraiseQLReader(client=client)

    documents = await reader.aload_data(
        query="{ products { id name description category price } }",
        text_template="{name}: {description}",
        metadata_fields=["id", "category", "price"],
    )

    for doc in documents:
        print(doc.text)        # "Widget Pro: The best widget for everyday use"
        print(doc.metadata)    # {"id": "p-123", "category": "tools", "price": 29.99}

asyncio.run(main())

Parameters

Parameter	Type	Description
`query`	`str`	GraphQL query string
`variables`	`dict \| None`	Query variables
`text_template`	`str \| None`	Python format string for `Document.text`. Uses field names as keys. If `None`, the full JSON is used.
`metadata_fields`	`list[str] \| None`	Fields to include in `Document.metadata`. If `None`, all fields are included.

With LlamaIndex VectorStore

Load FraiseQL results into a vector index for semantic search:

from llama_index.core import VectorStoreIndex
from fraiseql.client import FraiseQLClient
from fraiseql.integrations.llamaindex import FraiseQLReader

client = FraiseQLClient("http://localhost:8080/graphql")
reader = FraiseQLReader(client=client)

# Load product catalog as documents
documents = await reader.aload_data(
    query="{ products { id name description } }",
    text_template="{name}: {description}",
)

# Build vector index for semantic search
index = VectorStoreIndex.from_documents(documents)
query_engine = index.as_query_engine()

response = await query_engine.aquery(
    "Which products are suitable for outdoor use?"
)
print(response)