graphrag/tests/verbs/test_create_final_documents.py

# Copyright (c) 2024 Microsoft Corporation.
# Licensed under the MIT License

from graphrag.config.models.graph_rag_config import GraphRagConfig
from graphrag.data_model.schemas import DOCUMENTS_FINAL_COLUMNS
from graphrag.index.workflows.create_final_documents import (
    run_workflow,
)
from graphrag.utils.storage import load_table_from_storage

from .util import (
    DEFAULT_MODEL_CONFIG,
    compare_outputs,
    create_test_context,
    load_test_table,
    update_document_metadata,
)


async def test_create_final_documents():
    expected = load_test_table("documents")

    context = await create_test_context(
        storage=["text_units"],
    )

    config = GraphRagConfig(models=DEFAULT_MODEL_CONFIG)  # type: ignore

    await run_workflow(config, context)

    actual = await load_table_from_storage("documents", context.output_storage)

    compare_outputs(actual, expected)

    for column in DOCUMENTS_FINAL_COLUMNS:
        assert column in actual.columns


async def test_create_final_documents_with_metadata_column():
    context = await create_test_context(
        storage=["text_units"],
    )

    config = GraphRagConfig(models=DEFAULT_MODEL_CONFIG)  # type: ignore
    config.input.metadata = ["title"]

    # simulate the metadata construction during initial input loading
    await update_document_metadata(config.input.metadata, context)

    expected = await load_table_from_storage("documents", context.output_storage)

    await run_workflow(config, context)

    actual = await load_table_from_storage("documents", context.output_storage)

    compare_outputs(actual, expected)

    for column in DOCUMENTS_FINAL_COLUMNS:
        assert column in actual.columns