zlei6/ragflow
1
0
Fork 0
mirror of https://github.com/infiniflow/ragflow.git synced 2026-07-25 01:43:27 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix: preserve compilation template kinds and names (#17151)

Browse Source
This commit is contained in:
buua436
2026-07-21 13:40:08 +08:00
committed by GitHub
parent bc07b18b88
commit a730ce7aa6
3 changed files with 130 additions and 8 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
api/apps/restful_apis/chunk_api.py
View File

@@ -166,10 +166,7 @@ def _get_dataset_tenant_id(dataset_id):
def _compilation_template_kind(kind) -> str:
if not isinstance(kind, str):
return ""
normalized = kind.strip().lower().replace("-", "_")
if normalized in {"pageindex", "page_index", "knowledge_graph"}:
return "timeline"
return normalized
return kind.strip().lower().replace("-", "_")
def _resolve_reference_metadata(req: dict, search_config: dict | None = None):
@@ -581,6 +578,7 @@ async def get_document_structure_graph(tenant_id, dataset_id, document_id):
"""
from rag.nlp import search
from api.db.services.compilation_template_group_service import CompilationTemplateGroupService
from api.db.services.compilation_template_service import CompilationTemplateService
if not KnowledgebaseService.accessible(kb_id=dataset_id, user_id=tenant_id):
return get_error_data_result(message=f"You don't own the dataset {dataset_id}.")
@@ -624,6 +622,7 @@ async def get_document_structure_graph(tenant_id, dataset_id, document_id):
seen_configured_ids: set[str] = set()
template_meta: dict[str, dict] = {}
template_meta_by_kind: dict[str, list[dict]] = {}
template_meta_by_id: dict[str, dict] = {}
for group_id in group_ids:
group = CompilationTemplateGroupService.get_saved(group_id, tenant_id)
if not group:
@@ -648,6 +647,7 @@ async def get_document_structure_graph(tenant_id, dataset_id, document_id):
"kind_norm": kind_norm,
}
template_meta[template_id] = meta
template_meta_by_id[template_id] = meta
template_meta_by_kind.setdefault(kind_norm, []).append(meta)
# Load every graph row for this doc in one shot. Each row corresponds
@@ -740,6 +740,24 @@ async def get_document_structure_graph(tenant_id, dataset_id, document_id):
bucket_id = tid
row_kind_norm = _compilation_template_kind(kind_val)
meta = template_meta.get(bucket_id)
if not meta:
# Pipeline Compiler receives template groups as component
# parameters and does not persist those group ids in the
# document parser_config. Resolve the exact template id
# directly so Pipeline-produced rows do not fall back to
# exposing the opaque id as the display name.
if bucket_id not in template_meta_by_id:
saved_template = CompilationTemplateService.get_saved(bucket_id, tenant_id)
if saved_template:
template_meta_by_id[bucket_id] = {
"template_id": bucket_id,
"template_name": saved_template.get("name") or bucket_id,
"kind": saved_template.get("kind") or kind_val,
"kind_norm": _compilation_template_kind(saved_template.get("kind")),
}
else:
template_meta_by_id[bucket_id] = {}
meta = template_meta_by_id[bucket_id] or None
if not meta:
kind_matches = template_meta_by_kind.get(row_kind_norm) or []
if len(kind_matches) == 1: