dify/api/tasks/remove_document_from_index_task.py

import logging
import time

import click
from celery import shared_task
from sqlalchemy import select

from core.db.session_factory import session_factory
from core.rag.index_processor.index_processor_factory import IndexProcessorFactory
from extensions.ext_redis import redis_client
from libs.datetime_utils import naive_utc_now
from models.dataset import Document, DocumentSegment

logger = logging.getLogger(__name__)


@shared_task(queue="dataset")
def remove_document_from_index_task(document_id: str):
    """
    Async Remove document from index
    :param document_id: document id

    Usage: remove_document_from_index.delay(document_id)
    """
    logger.info(click.style(f"Start remove document segments from index: {document_id}", fg="green"))
    start_at = time.perf_counter()

    with session_factory.create_session() as session:
        document = session.query(Document).where(Document.id == document_id).first()
        if not document:
            logger.info(click.style(f"Document not found: {document_id}", fg="red"))
            return

        if document.indexing_status != "completed":
            logger.info(click.style(f"Document is not completed, remove is not allowed: {document_id}", fg="red"))
            return

        indexing_cache_key = f"document_{document.id}_indexing"

        try:
            dataset = document.dataset

            if not dataset:
                raise Exception("Document has no dataset")

            index_processor = IndexProcessorFactory(document.doc_form).init_index_processor()

            segments = session.scalars(select(DocumentSegment).where(DocumentSegment.document_id == document.id)).all()

            # Disable summary indexes for all segments in this document
            from services.summary_index_service import SummaryIndexService

            segment_ids_list = [segment.id for segment in segments]
            if segment_ids_list:
                try:
                    SummaryIndexService.disable_summaries_for_segments(
                        dataset=dataset,
                        segment_ids=segment_ids_list,
                        disabled_by=document.disabled_by,
                    )
                except Exception as e:
                    logger.warning("Failed to disable summaries for document %s: %s", document.id, str(e))

            index_node_ids = [segment.index_node_id for segment in segments]
            if index_node_ids:
                try:
                    index_processor.clean(dataset, index_node_ids, with_keywords=True, delete_child_chunks=False)
                except Exception:
                    logger.exception("clean dataset %s from index failed", dataset.id)
            # update segment to disable
            session.query(DocumentSegment).where(DocumentSegment.document_id == document.id).update(
                {
                    DocumentSegment.enabled: False,
                    DocumentSegment.disabled_at: naive_utc_now(),
                    DocumentSegment.disabled_by: document.disabled_by,
                    DocumentSegment.updated_at: naive_utc_now(),
                }
            )
            session.commit()

            end_at = time.perf_counter()
            logger.info(
                click.style(
                    f"Document removed from index: {document.id} latency: {end_at - start_at}",
                    fg="green",
                )
            )
        except Exception:
            logger.exception("remove document from index failed")
            if not document.archived:
                document.enabled = True
                session.commit()
        finally:
            redis_client.delete(indexing_cache_key)