dify/api/tasks/batch_create_segment_to_index_task.py

98 lines
3.9 KiB
Python
Raw Normal View History

import datetime
import logging
import time
import uuid
from typing import Optional, List
import click
from celery import shared_task
from sqlalchemy import func
from werkzeug.exceptions import NotFound
from core.index.index import IndexBuilder
from core.indexing_runner import IndexingRunner
from core.model_providers.model_factory import ModelFactory
from extensions.ext_database import db
from extensions.ext_redis import redis_client
from libs import helper
from models.dataset import DocumentSegment, Dataset, Document
@shared_task(queue='dataset')
def batch_create_segment_to_index_task(job_id: str, content: List, dataset_id: str, document_id: str,
tenant_id: str, user_id: str):
"""
Async batch create segment to index
:param job_id:
:param content:
:param dataset_id:
:param document_id:
:param tenant_id:
:param user_id:
Usage: batch_create_segment_to_index_task.delay(segment_id)
"""
logging.info(click.style('Start batch create segment jobId: {}'.format(job_id), fg='green'))
start_at = time.perf_counter()
indexing_cache_key = 'segment_batch_import_{}'.format(job_id)
try:
dataset = db.session.query(Dataset).filter(Dataset.id == dataset_id).first()
if not dataset:
raise ValueError('Dataset not exist.')
dataset_document = db.session.query(Document).filter(Document.id == document_id).first()
if not dataset_document:
raise ValueError('Document not exist.')
if not dataset_document.enabled or dataset_document.archived or dataset_document.indexing_status != 'completed':
raise ValueError('Document is not available.')
document_segments = []
embedding_model = None
if dataset.indexing_technique == 'high_quality':
embedding_model = ModelFactory.get_embedding_model(
tenant_id=dataset.tenant_id,
model_provider_name=dataset.embedding_model_provider,
model_name=dataset.embedding_model
)
for segment in content:
content = segment['content']
doc_id = str(uuid.uuid4())
segment_hash = helper.generate_text_hash(content)
# calc embedding use tokens
tokens = embedding_model.get_num_tokens(content) if embedding_model else 0
max_position = db.session.query(func.max(DocumentSegment.position)).filter(
DocumentSegment.document_id == dataset_document.id
).scalar()
segment_document = DocumentSegment(
tenant_id=tenant_id,
dataset_id=dataset_id,
document_id=document_id,
index_node_id=doc_id,
index_node_hash=segment_hash,
position=max_position + 1 if max_position else 1,
content=content,
word_count=len(content),
tokens=tokens,
created_by=user_id,
indexing_at=datetime.datetime.utcnow(),
status='completed',
completed_at=datetime.datetime.utcnow()
)
if dataset_document.doc_form == 'qa_model':
segment_document.answer = segment['answer']
db.session.add(segment_document)
document_segments.append(segment_document)
# add index to db
indexing_runner = IndexingRunner()
indexing_runner.batch_add_segments(document_segments, dataset)
db.session.commit()
redis_client.setex(indexing_cache_key, 600, 'completed')
end_at = time.perf_counter()
logging.info(click.style('Segment batch created job: {} latency: {}'.format(job_id, end_at - start_at), fg='green'))
except Exception as e:
logging.exception("Segments batch created index failed:{}".format(str(e)))
redis_client.setex(indexing_cache_key, 600, 'error')