Document Import¶

Vector Graph RAG includes a lightweight document import layer for turning files, URLs, and text into LangChain Document chunks. The importer is intentionally pluggable: simple workflows can use the default converter, while more demanding document pipelines can provide a stronger parser adapter.

Installation¶

Install loader dependencies when you need URL fetching or local file conversion:

uv add "vector-graph-rag[loaders]"

Optional parser adapters have their own extras:

uv add "vector-graph-rag[docling]"
uv add "vector-graph-rag[mineru]"

Import URLs And Files¶

from vector_graph_rag import VectorGraphRAG
from vector_graph_rag.loaders import DocumentImporter

importer = DocumentImporter(chunk_size=1000, chunk_overlap=200)
result = importer.import_sources(
    [
        "https://en.wikipedia.org/wiki/Albert_Einstein",
        "/path/to/document.pdf",
        "/path/to/report.docx",
    ]
)

rag = VectorGraphRAG(collection_prefix="demo")
rag.rebuild_documents(result.documents, extract_triplets=True)

The default local-file converter uses MarkItDown for common document formats. URL content is fetched and converted to text before chunking.

Use Docling¶

Use Docling when you want a document parser with richer PDF layout handling:

from vector_graph_rag.loaders import DoclingConverter, DocumentImporter

importer = DocumentImporter(
    converter=DoclingConverter(),
    chunk_size=1000,
    chunk_overlap=200,
)

result = importer.import_sources(["/path/to/report.pdf"])

Docling may download parser models on first use. In offline or restricted environments, pre-download those assets and pass a preconfigured Docling converter through converter_kwargs.

Use MinerU¶

Use MinerU when your parsing pipeline already depends on MinerU or when you need its document processing accuracy:

from vector_graph_rag.loaders import DocumentImporter, MinerUConverter

importer = DocumentImporter(
    converter=MinerUConverter(
        output_dir="./mineru-output",
        backend="pipeline",
    ),
    chunk_size=1000,
    chunk_overlap=200,
)

result = importer.import_sources(["/path/to/report.pdf"])

Follow MinerU's setup guide for parser models and runtime backends. Vector Graph RAG reads the Markdown output and returns standard LangChain Document chunks.

Connect Import To Incremental Updates¶

For source-level updates, every chunk produced from the same external object should carry the same stable source value.

result = importer.import_sources(["/path/to/report.pdf"])

rag.upsert_documents_by_source(
    result.documents,
    source="file:report-123",
    extract_triplets=True,
)

If your parser extracts triplets itself, store them in each chunk's metadata["triplets"] and disable LLM extraction:

rag.upsert_documents_by_source(
    result.documents,
    source="file:report-123",
    extract_triplets=False,
)

See Incremental Updates for source key design, retry behavior, and delete flows.

API Reference¶

The importer and converters are documented in the Python API Reference: