Feat/dataset notion import (#392)

Co-authored-by: StyleZhang <jasonapring2015@outlook.com>
Co-authored-by: JzoNg <jzongcode@gmail.com>
This commit is contained in:
Jyong
2023-06-16 21:47:51 +08:00
committed by GitHub
parent f350948bde
commit 9253f72dea
96 changed files with 4479 additions and 367 deletions

View File

@@ -13,32 +13,36 @@ from models.dataset import Document
@shared_task
def document_indexing_task(dataset_id: str, document_id: str):
def document_indexing_task(dataset_id: str, document_ids: list):
"""
Async process document
:param dataset_id:
:param document_id:
:param document_ids:
Usage: document_indexing_task.delay(dataset_id, document_id)
"""
logging.info(click.style('Start process document: {}'.format(document_id), fg='green'))
start_at = time.perf_counter()
documents = []
for document_id in document_ids:
logging.info(click.style('Start process document: {}'.format(document_id), fg='green'))
start_at = time.perf_counter()
document = db.session.query(Document).filter(
Document.id == document_id,
Document.dataset_id == dataset_id
).first()
document = db.session.query(Document).filter(
Document.id == document_id,
Document.dataset_id == dataset_id
).first()
if not document:
raise NotFound('Document not found')
if not document:
raise NotFound('Document not found')
document.indexing_status = 'parsing'
document.processing_started_at = datetime.datetime.utcnow()
document.indexing_status = 'parsing'
document.processing_started_at = datetime.datetime.utcnow()
documents.append(document)
db.session.add(document)
db.session.commit()
try:
indexing_runner = IndexingRunner()
indexing_runner.run(document)
indexing_runner.run(documents)
end_at = time.perf_counter()
logging.info(click.style('Processed document: {} latency: {}'.format(document.id, end_at - start_at), fg='green'))
except DocumentIsPausedException: