chore: 纳入本地 Dify 1.19.0 源码及品牌与免登改动
演示目录只保留可重建的源码。运行密钥、数据卷、依赖和带口令的本地配置不进入仓库。 Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
1 parent
60158935f4
commit
6519682885
15831 files changed
+3063428
No files matched your search
@@ -0,0 +1,140 @@
|
||||
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.constant.doc_type import DocType
|
||||
from core.rag.index_processor.constant.index_type import IndexStructureType
|
||||
from core.rag.index_processor.index_processor import IndexProcessorFactory
|
||||
from core.rag.models.document import AttachmentDocument, ChildDocument, Document
|
||||
from extensions.ext_redis import redis_client
|
||||
from libs.datetime_utils import naive_utc_now
|
||||
from models.dataset import DocumentSegment
|
||||
from models.enums import IndexingStatus, SegmentStatus
|
||||
from repositories.knowledge.dataset_read_repository import (
|
||||
get_segment_child_chunks,
|
||||
get_segment_dataset,
|
||||
get_segment_document,
|
||||
)
|
||||
from repositories.knowledge.segment_read_adapter import get_segment_attachments
|
||||
|
||||
logger = logging.getLogger(__name__)
|
||||
|
||||
|
||||
@shared_task(queue="dataset")
|
||||
def enable_segment_to_index_task(segment_id: str):
|
||||
"""
|
||||
Async enable segment to index
|
||||
:param segment_id:
|
||||
|
||||
Usage: enable_segment_to_index_task.delay(segment_id)
|
||||
"""
|
||||
logger.info(click.style(f"Start enable segment to index: {segment_id}", fg="green"))
|
||||
start_at = time.perf_counter()
|
||||
|
||||
with session_factory.create_session() as session:
|
||||
segment = session.scalar(select(DocumentSegment).where(DocumentSegment.id == segment_id).limit(1))
|
||||
if not segment:
|
||||
logger.info(click.style(f"Segment not found: {segment_id}", fg="red"))
|
||||
return
|
||||
|
||||
if segment.status != SegmentStatus.COMPLETED:
|
||||
logger.info(click.style(f"Segment is not completed, enable is not allowed: {segment_id}", fg="red"))
|
||||
return
|
||||
|
||||
indexing_cache_key = f"segment_{segment.id}_indexing"
|
||||
|
||||
try:
|
||||
document = Document(
|
||||
page_content=segment.content,
|
||||
metadata={
|
||||
"doc_id": segment.index_node_id,
|
||||
"doc_hash": segment.index_node_hash,
|
||||
"document_id": segment.document_id,
|
||||
"dataset_id": segment.dataset_id,
|
||||
},
|
||||
)
|
||||
|
||||
dataset = get_segment_dataset(segment, session=session)
|
||||
|
||||
if not dataset:
|
||||
logger.info(click.style(f"Segment {segment.id} has no dataset, pass.", fg="cyan"))
|
||||
return
|
||||
|
||||
dataset_document = get_segment_document(segment, session=session)
|
||||
|
||||
if not dataset_document:
|
||||
logger.info(click.style(f"Segment {segment.id} has no document, pass.", fg="cyan"))
|
||||
return
|
||||
|
||||
if (
|
||||
not dataset_document.enabled
|
||||
or dataset_document.archived
|
||||
or dataset_document.indexing_status != IndexingStatus.COMPLETED
|
||||
):
|
||||
logger.info(click.style(f"Segment {segment.id} document status is invalid, pass.", fg="cyan"))
|
||||
return
|
||||
|
||||
index_processor = IndexProcessorFactory(dataset_document.doc_form).init_index_processor()
|
||||
if dataset_document.doc_form == IndexStructureType.PARENT_CHILD_INDEX:
|
||||
child_chunks = get_segment_child_chunks(segment, session=session)
|
||||
if child_chunks:
|
||||
child_documents = []
|
||||
for child_chunk in child_chunks:
|
||||
child_document = ChildDocument(
|
||||
page_content=child_chunk.content,
|
||||
metadata={
|
||||
"doc_id": child_chunk.index_node_id,
|
||||
"doc_hash": child_chunk.index_node_hash,
|
||||
"document_id": segment.document_id,
|
||||
"dataset_id": segment.dataset_id,
|
||||
},
|
||||
)
|
||||
child_documents.append(child_document)
|
||||
document.children = child_documents
|
||||
multimodel_documents = []
|
||||
if dataset.is_multimodal:
|
||||
for attachment in get_segment_attachments(segment, session=session):
|
||||
multimodel_documents.append(
|
||||
AttachmentDocument(
|
||||
page_content=attachment["name"],
|
||||
metadata={
|
||||
"doc_id": attachment["id"],
|
||||
"doc_hash": "",
|
||||
"document_id": segment.document_id,
|
||||
"dataset_id": segment.dataset_id,
|
||||
"doc_type": DocType.IMAGE,
|
||||
},
|
||||
)
|
||||
)
|
||||
|
||||
# save vector index
|
||||
index_processor.load(dataset, [document], multimodal_documents=multimodel_documents, session=session)
|
||||
session.commit()
|
||||
|
||||
# Enable summary index for this segment
|
||||
from services.knowledge.summaries.adapters import SummaryIndexAdapter
|
||||
|
||||
try:
|
||||
SummaryIndexAdapter.enable_summaries_for_segments(
|
||||
dataset=dataset,
|
||||
segment_ids=[segment.id],
|
||||
)
|
||||
except Exception:
|
||||
logger.warning("Failed to enable summary for segment %s", segment.id, exc_info=True)
|
||||
|
||||
end_at = time.perf_counter()
|
||||
logger.info(click.style(f"Segment enabled to index: {segment.id} latency: {end_at - start_at}", fg="green"))
|
||||
except Exception as e:
|
||||
logger.exception("enable segment to index failed")
|
||||
session.rollback()
|
||||
segment.enabled = False
|
||||
segment.disabled_at = naive_utc_now()
|
||||
segment.status = SegmentStatus.ERROR
|
||||
segment.error = str(e)
|
||||
session.commit()
|
||||
finally:
|
||||
redis_client.delete(indexing_cache_key)
|
||||
Reference in new issue
Block a user