KMS/dify
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge branch 'feat/support-knowledge-metadata' into deploy/dev

Browse Source
This commit is contained in:
jyong 2025-03-13 17:19:08 +08:00
commit 729730923e
3 changed files with 31 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
api/core/rag/retrieval/dataset_retrieval.py
View File

@ -878,7 +878,7 @@ class DatasetRetrieval:
document_query = document_query.filter(and_(*filters))
documents = document_query.all()
# group by dataset_id
metadata_filter_document_ids = defaultdict(list)
metadata_filter_document_ids = defaultdict(list) if documents else None
for document in documents:
metadata_filter_document_ids[document.dataset_id].append(document.id)
return metadata_filter_document_ids, metadata_condition

29
api/core/workflow/nodes/knowledge_retrieval/knowledge_retrieval_node.py
View File

@ -4,7 +4,8 @@ import time
from collections.abc import Mapping, Sequence
from typing import Any, Optional, cast
from sqlalchemy import and_, func, or_, text
from sqlalchemy import Integer, and_, func, or_, text
from sqlalchemy import cast as sqlalchemy_cast
from core.app.app_config.entities import DatasetRetrieveConfigEntity
from core.app.entities.app_invoke_entities import ModelConfigWithCredentialsEntity
@ -340,10 +341,12 @@ class KnowledgeRetrievalNode(LLMNode):
)
)
metadata_condition = MetadataCondition(
logical_operator="or",
logical_operator=node_data.metadata_filtering_conditions.logical_operator,
conditions=conditions,
)
elif node_data.metadata_filtering_mode == "manual":
if node_data.metadata_filtering_conditions:
metadata_condition = MetadataCondition(**node_data.metadata_filtering_conditions.model_dump())
if node_data.metadata_filtering_conditions:
for condition in node_data.metadata_filtering_conditions.conditions:
metadata_name = condition.name
@ -364,10 +367,10 @@ class KnowledgeRetrievalNode(LLMNode):
document_query = document_query.filter(and_(*filters))
else:
document_query = document_query.filter(or_(*filters))
documnents = document_query.all()
documents = document_query.all()
# group by dataset_id
metadata_filter_document_ids = defaultdict(list)
for document in documnents:
metadata_filter_document_ids = defaultdict(list) if documents else None
for document in documents:
metadata_filter_document_ids[document.dataset_id].append(document.id)
return metadata_filter_document_ids, metadata_condition
@ -458,24 +461,28 @@ class KnowledgeRetrievalNode(LLMNode):
if isinstance(value, str):
filters.append(Document.doc_metadata[metadata_name] == f'"{value}"')
else:
filters.append(Document.doc_metadata[metadata_name] == value)
filters.append(
sqlalchemy_cast(Document.doc_metadata[metadata_name].astext, Integer) == value
)
case "is not" | "≠":
if isinstance(value, str):
filters.append(Document.doc_metadata[metadata_name] != f'"{value}"')
else:
filters.append(Document.doc_metadata[metadata_name] != value)
filters.append(
sqlalchemy_cast(Document.doc_metadata[metadata_name].astext, Integer) != value
)
case "empty":
filters.append(Document.doc_metadata[metadata_name].is_(None))
case "not empty":
filters.append(Document.doc_metadata[metadata_name].isnot(None))
case "before" | "<":
filters.append(Document.doc_metadata[metadata_name] < value)
filters.append(sqlalchemy_cast(Document.doc_metadata[metadata_name].astext, Integer) < value)
case "after" | ">":
filters.append(Document.doc_metadata[metadata_name] > value)
filters.append(sqlalchemy_cast(Document.doc_metadata[metadata_name].astext, Integer) > value)
case "≤" | ">=":
filters.append(Document.doc_metadata[metadata_name] <= value)
filters.append(sqlalchemy_cast(Document.doc_metadata[metadata_name].astext, Integer) <= value)
case "≥" | ">=":
filters.append(Document.doc_metadata[metadata_name] >= value)
filters.append(sqlalchemy_cast(Document.doc_metadata[metadata_name].astext, Integer) >= value)
case _:
pass
return filters

2
api/services/tag_service.py
View File

@ -20,7 +20,7 @@ class TagService:
)
if keyword:
query = query.filter(db.and_(Tag.name.ilike(f"%{keyword}%")))
query = query.group_by(Tag.id)
query = query.group_by(Tag.id, Tag.type, Tag.name)
results: list = query.order_by(Tag.created_at.desc()).all()
return results