update to external knowledge api

feat/external-knowledge-api
jyong 2 years ago
parent a69dcb8bee
commit c927c97310

@ -111,7 +111,7 @@ class DatasetListApi(Resource):
help="Invalid indexing technique.", help="Invalid indexing technique.",
) )
parser.add_argument( parser.add_argument(
"external_api_template_id", "external_knowledge_api_id",
type=str, type=str,
nullable=True, nullable=True,
required=False, required=False,
@ -144,7 +144,7 @@ class DatasetListApi(Resource):
account=current_user, account=current_user,
permission=DatasetPermissionEnum.ONLY_ME, permission=DatasetPermissionEnum.ONLY_ME,
provider=args["provider"], provider=args["provider"],
external_api_template_id=args["external_api_template_id"], external_knowledge_api_id=args["external_knowledge_api_id"],
external_knowledge_id=args["external_knowledge_id"], external_knowledge_id=args["external_knowledge_id"],
) )
except services.errors.dataset.DatasetNameDuplicateError: except services.errors.dataset.DatasetNameDuplicateError:

@ -37,12 +37,12 @@ class ExternalApiTemplateListApi(Resource):
limit = request.args.get("limit", default=20, type=int) limit = request.args.get("limit", default=20, type=int)
search = request.args.get("keyword", default=None, type=str) search = request.args.get("keyword", default=None, type=str)
api_templates, total = ExternalDatasetService.get_external_api_templates( external_knowledge_apis, total = ExternalDatasetService.get_external_knowledge_apis(
page, limit, current_user.current_tenant_id, search page, limit, current_user.current_tenant_id, search
) )
response = { response = {
"data": [item.to_dict() for item in api_templates], "data": [item.to_dict() for item in external_knowledge_apis],
"has_more": len(api_templates) == limit, "has_more": len(external_knowledge_apis) == limit,
"limit": limit, "limit": limit,
"total": total, "total": total,
"page": page, "page": page,
@ -77,13 +77,13 @@ class ExternalApiTemplateListApi(Resource):
raise Forbidden() raise Forbidden()
try: try:
api_template = ExternalDatasetService.create_api_template( external_knowledge_api = ExternalDatasetService.create_external_knowledge_api(
tenant_id=current_user.current_tenant_id, user_id=current_user.id, args=args tenant_id=current_user.current_tenant_id, user_id=current_user.id, args=args
) )
except services.errors.dataset.DatasetNameDuplicateError: except services.errors.dataset.DatasetNameDuplicateError:
raise DatasetNameDuplicateError() raise DatasetNameDuplicateError()
return api_template.to_dict(), 201 return external_knowledge_api.to_dict(), 201
class ExternalApiTemplateApi(Resource): class ExternalApiTemplateApi(Resource):
@ -92,11 +92,11 @@ class ExternalApiTemplateApi(Resource):
@account_initialization_required @account_initialization_required
def get(self, external_knowledge_api_id): def get(self, external_knowledge_api_id):
external_knowledge_api_id = str(external_knowledge_api_id) external_knowledge_api_id = str(external_knowledge_api_id)
api_template = ExternalDatasetService.get_api_template(external_knowledge_api_id) external_knowledge_api = ExternalDatasetService.get_external_knowledge_api(external_knowledge_api_id)
if api_template is None: if external_knowledge_api is None:
raise NotFound("API template not found.") raise NotFound("API template not found.")
return api_template.to_dict(), 200 return external_knowledge_api.to_dict(), 200
@setup_required @setup_required
@login_required @login_required
@ -122,14 +122,14 @@ class ExternalApiTemplateApi(Resource):
args = parser.parse_args() args = parser.parse_args()
ExternalDatasetService.validate_api_list(args["settings"]) ExternalDatasetService.validate_api_list(args["settings"])
api_template = ExternalDatasetService.update_api_template( external_knowledge_api = ExternalDatasetService.update_external_knowledge_api(
tenant_id=current_user.current_tenant_id, tenant_id=current_user.current_tenant_id,
user_id=current_user.id, user_id=current_user.id,
external_knowledge_api_id=external_knowledge_api_id, external_knowledge_api_id=external_knowledge_api_id,
args=args, args=args,
) )
return api_template.to_dict(), 200 return external_knowledge_api.to_dict(), 200
@setup_required @setup_required
@login_required @login_required
@ -141,7 +141,7 @@ class ExternalApiTemplateApi(Resource):
if not current_user.is_editor or current_user.is_dataset_operator: if not current_user.is_editor or current_user.is_dataset_operator:
raise Forbidden() raise Forbidden()
ExternalDatasetService.delete_api_template(current_user.current_tenant_id, external_knowledge_api_id) ExternalDatasetService.delete_external_knowledge_api(current_user.current_tenant_id, external_knowledge_api_id)
return {"result": "success"}, 200 return {"result": "success"}, 200
@ -152,10 +152,10 @@ class ExternalApiUseCheckApi(Resource):
def get(self, external_knowledge_api_id): def get(self, external_knowledge_api_id):
external_knowledge_api_id = str(external_knowledge_api_id) external_knowledge_api_id = str(external_knowledge_api_id)
external_api_template_is_using, count = ExternalDatasetService.external_api_template_use_check( external_knowledge_api_is_using, count = ExternalDatasetService.external_knowledge_api_use_check(
external_knowledge_api_id external_knowledge_api_id
) )
return {"is_using": external_api_template_is_using, "count": count}, 200 return {"is_using": external_knowledge_api_is_using, "count": count}, 200
class ExternalDatasetInitApi(Resource): class ExternalDatasetInitApi(Resource):
@ -168,7 +168,7 @@ class ExternalDatasetInitApi(Resource):
raise Forbidden() raise Forbidden()
parser = reqparse.RequestParser() parser = reqparse.RequestParser()
parser.add_argument("api_template_id", type=str, required=True, nullable=True, location="json") parser.add_argument("external_knowledge_api_id", type=str, required=True, nullable=True, location="json")
# parser.add_argument('name', nullable=False, required=True, # parser.add_argument('name', nullable=False, required=True,
# help='name is required. Name must be between 1 to 100 characters.', # help='name is required. Name must be between 1 to 100 characters.',
# type=_validate_name) # type=_validate_name)
@ -184,7 +184,7 @@ class ExternalDatasetInitApi(Resource):
# validate args # validate args
ExternalDatasetService.document_create_args_validate( ExternalDatasetService.document_create_args_validate(
current_user.current_tenant_id, args["api_template_id"], args["process_parameter"] current_user.current_tenant_id, args["external_knowledge_api_id"], args["process_parameter"]
) )
try: try:
@ -210,7 +210,7 @@ class ExternalDatasetCreateApi(Resource):
raise Forbidden() raise Forbidden()
parser = reqparse.RequestParser() parser = reqparse.RequestParser()
parser.add_argument("external_api_template_id", type=str, required=True, nullable=False, location="json") parser.add_argument("external_knowledge_api_id", type=str, required=True, nullable=False, location="json")
parser.add_argument("external_knowledge_id", type=str, required=True, nullable=False, location="json") parser.add_argument("external_knowledge_id", type=str, required=True, nullable=False, location="json")
parser.add_argument( parser.add_argument(
"name", "name",

@ -83,7 +83,7 @@ class DatasetListApi(DatasetApiResource):
nullable=False, nullable=False,
) )
parser.add_argument( parser.add_argument(
"external_api_template_id", "external_knowledge_api_id",
type=str, type=str,
nullable=True, nullable=True,
required=False, required=False,
@ -112,7 +112,7 @@ class DatasetListApi(DatasetApiResource):
account=current_user, account=current_user,
permission=args["permission"], permission=args["permission"],
provider=args["provider"], provider=args["provider"],
external_api_template_id=args["external_api_template_id"], external_knowledge_api_id=args["external_knowledge_api_id"],
external_knowledge_id=args["external_knowledge_id"], external_knowledge_id=args["external_knowledge_id"],
) )
except services.errors.dataset.DatasetNameDuplicateError: except services.errors.dataset.DatasetNameDuplicateError:

@ -2,7 +2,7 @@ from flask_restful import fields
from libs.helper import TimestampField from libs.helper import TimestampField
api_template_query_detail_fields = { external_knowledge_api_query_detail_fields = {
"id": fields.String, "id": fields.String,
"name": fields.String, "name": fields.String,
"setting": fields.String, "setting": fields.String,

@ -12,7 +12,7 @@ from sqlalchemy.dialects import postgresql
# revision identifiers, used by Alembic. # revision identifiers, used by Alembic.
revision = '6af6a521a53e' revision = '6af6a521a53e'
down_revision = 'ec3df697ebbb' down_revision = 'd57ba9ebb251'
branch_labels = None branch_labels = None
depends_on = None depends_on = None

@ -1,8 +1,8 @@
"""external_knowledge """external_knowledge_api
Revision ID: ec3df697ebbb Revision ID: 33f5fac87f29
Revises: 675b5321501b Revises: 6af6a521a53e
Create Date: 2024-09-18 06:59:54.048478 Create Date: 2024-09-25 04:34:57.249436
""" """
from alembic import op from alembic import op
@ -11,15 +11,15 @@ import sqlalchemy as sa
from sqlalchemy.dialects import postgresql from sqlalchemy.dialects import postgresql
# revision identifiers, used by Alembic. # revision identifiers, used by Alembic.
revision = 'ec3df697ebbb' revision = '33f5fac87f29'
down_revision = '675b5321501b' down_revision = '6af6a521a53e'
branch_labels = None branch_labels = None
depends_on = None depends_on = None
def upgrade(): def upgrade():
# ### commands auto generated by Alembic - please adjust! ### # ### commands auto generated by Alembic - please adjust! ###
op.create_table('external_api_templates', op.create_table('external_knowledge_apis',
sa.Column('id', models.types.StringUUID(), server_default=sa.text('uuid_generate_v4()'), nullable=False), sa.Column('id', models.types.StringUUID(), server_default=sa.text('uuid_generate_v4()'), nullable=False),
sa.Column('name', sa.String(length=255), nullable=False), sa.Column('name', sa.String(length=255), nullable=False),
sa.Column('description', sa.String(length=255), nullable=False), sa.Column('description', sa.String(length=255), nullable=False),
@ -29,16 +29,16 @@ def upgrade():
sa.Column('created_at', sa.DateTime(), server_default=sa.text('CURRENT_TIMESTAMP(0)'), nullable=False), sa.Column('created_at', sa.DateTime(), server_default=sa.text('CURRENT_TIMESTAMP(0)'), nullable=False),
sa.Column('updated_by', models.types.StringUUID(), nullable=True), sa.Column('updated_by', models.types.StringUUID(), nullable=True),
sa.Column('updated_at', sa.DateTime(), server_default=sa.text('CURRENT_TIMESTAMP(0)'), nullable=False), sa.Column('updated_at', sa.DateTime(), server_default=sa.text('CURRENT_TIMESTAMP(0)'), nullable=False),
sa.PrimaryKeyConstraint('id', name='external_api_template_pkey') sa.PrimaryKeyConstraint('id', name='external_knowledge_apis_pkey')
) )
with op.batch_alter_table('external_api_templates', schema=None) as batch_op: with op.batch_alter_table('external_knowledge_apis', schema=None) as batch_op:
batch_op.create_index('external_api_templates_name_idx', ['name'], unique=False) batch_op.create_index('external_knowledge_apis_name_idx', ['name'], unique=False)
batch_op.create_index('external_api_templates_tenant_idx', ['tenant_id'], unique=False) batch_op.create_index('external_knowledge_apis_tenant_idx', ['tenant_id'], unique=False)
op.create_table('external_knowledge_bindings', op.create_table('external_knowledge_bindings',
sa.Column('id', models.types.StringUUID(), server_default=sa.text('uuid_generate_v4()'), nullable=False), sa.Column('id', models.types.StringUUID(), server_default=sa.text('uuid_generate_v4()'), nullable=False),
sa.Column('tenant_id', models.types.StringUUID(), nullable=False), sa.Column('tenant_id', models.types.StringUUID(), nullable=False),
sa.Column('external_api_template_id', models.types.StringUUID(), nullable=False), sa.Column('external_knowledge_api_id', models.types.StringUUID(), nullable=False),
sa.Column('dataset_id', models.types.StringUUID(), nullable=False), sa.Column('dataset_id', models.types.StringUUID(), nullable=False),
sa.Column('external_knowledge_id', sa.Text(), nullable=False), sa.Column('external_knowledge_id', sa.Text(), nullable=False),
sa.Column('created_by', models.types.StringUUID(), nullable=False), sa.Column('created_by', models.types.StringUUID(), nullable=False),
@ -49,7 +49,7 @@ def upgrade():
) )
with op.batch_alter_table('external_knowledge_bindings', schema=None) as batch_op: with op.batch_alter_table('external_knowledge_bindings', schema=None) as batch_op:
batch_op.create_index('external_knowledge_bindings_dataset_idx', ['dataset_id'], unique=False) batch_op.create_index('external_knowledge_bindings_dataset_idx', ['dataset_id'], unique=False)
batch_op.create_index('external_knowledge_bindings_external_api_template_idx', ['external_api_template_id'], unique=False) batch_op.create_index('external_knowledge_bindings_external_knowledge_api_idx', ['external_knowledge_api_id'], unique=False)
batch_op.create_index('external_knowledge_bindings_external_knowledge_idx', ['external_knowledge_id'], unique=False) batch_op.create_index('external_knowledge_bindings_external_knowledge_idx', ['external_knowledge_id'], unique=False)
batch_op.create_index('external_knowledge_bindings_tenant_idx', ['tenant_id'], unique=False) batch_op.create_index('external_knowledge_bindings_tenant_idx', ['tenant_id'], unique=False)
@ -58,17 +58,16 @@ def upgrade():
def downgrade(): def downgrade():
# ### commands auto generated by Alembic - please adjust! ### # ### commands auto generated by Alembic - please adjust! ###
with op.batch_alter_table('external_knowledge_bindings', schema=None) as batch_op: with op.batch_alter_table('external_knowledge_bindings', schema=None) as batch_op:
batch_op.drop_index('external_knowledge_bindings_tenant_idx') batch_op.drop_index('external_knowledge_bindings_tenant_idx')
batch_op.drop_index('external_knowledge_bindings_external_knowledge_idx') batch_op.drop_index('external_knowledge_bindings_external_knowledge_idx')
batch_op.drop_index('external_knowledge_bindings_external_api_template_idx') batch_op.drop_index('external_knowledge_bindings_external_knowledge_api_idx')
batch_op.drop_index('external_knowledge_bindings_dataset_idx') batch_op.drop_index('external_knowledge_bindings_dataset_idx')
op.drop_table('external_knowledge_bindings') op.drop_table('external_knowledge_bindings')
with op.batch_alter_table('external_api_templates', schema=None) as batch_op: with op.batch_alter_table('external_knowledge_apis', schema=None) as batch_op:
batch_op.drop_index('external_api_templates_tenant_idx') batch_op.drop_index('external_knowledge_apis_tenant_idx')
batch_op.drop_index('external_api_templates_name_idx') batch_op.drop_index('external_knowledge_apis_name_idx')
op.drop_table('external_api_templates') op.drop_table('external_knowledge_apis')
# ### end Alembic commands ### # ### end Alembic commands ###

@ -180,18 +180,18 @@ class Dataset(db.Model):
) )
if not external_knowledge_binding: if not external_knowledge_binding:
return None return None
external_api_template = ( external_knowledge_api = (
db.session.query(ExternalApiTemplates) db.session.query(ExternalKnowledgeApis)
.filter(ExternalApiTemplates.id == external_knowledge_binding.external_api_template_id) .filter(ExternalKnowledgeApis.id == external_knowledge_binding.external_knowledge_api_id)
.first() .first()
) )
if not external_api_template: if not external_knowledge_api:
return None return None
return { return {
"external_knowledge_id": external_knowledge_binding.external_knowledge_id, "external_knowledge_id": external_knowledge_binding.external_knowledge_id,
"external_knowledge_api_id": external_api_template.id, "external_knowledge_api_id": external_knowledge_api.id,
"external_knowledge_api_name": external_api_template.name, "external_knowledge_api_name": external_knowledge_api.name,
"external_knowledge_api_endpoint": json.loads(external_api_template.settings).get("endpoint", ""), "external_knowledge_api_endpoint": json.loads(external_knowledge_api.settings).get("endpoint", ""),
} }
@staticmethod @staticmethod
@ -721,12 +721,12 @@ class DatasetPermission(db.Model):
created_at = db.Column(db.DateTime, nullable=False, server_default=db.text("CURRENT_TIMESTAMP(0)")) created_at = db.Column(db.DateTime, nullable=False, server_default=db.text("CURRENT_TIMESTAMP(0)"))
class ExternalApiTemplates(db.Model): class ExternalKnowledgeApis(db.Model):
__tablename__ = "external_api_templates" __tablename__ = "external_knowledge_apis"
__table_args__ = ( __table_args__ = (
db.PrimaryKeyConstraint("id", name="external_api_template_pkey"), db.PrimaryKeyConstraint("id", name="external_knowledge_apis_pkey"),
db.Index("external_api_templates_tenant_idx", "tenant_id"), db.Index("external_knowledge_apis_tenant_idx", "tenant_id"),
db.Index("external_api_templates_name_idx", "name"), db.Index("external_knowledge_apis_name_idx", "name"),
) )
id = db.Column(StringUUID, nullable=False, server_default=db.text("uuid_generate_v4()")) id = db.Column(StringUUID, nullable=False, server_default=db.text("uuid_generate_v4()"))
@ -762,7 +762,7 @@ class ExternalApiTemplates(db.Model):
def dataset_bindings(self): def dataset_bindings(self):
external_knowledge_bindings = ( external_knowledge_bindings = (
db.session.query(ExternalKnowledgeBindings) db.session.query(ExternalKnowledgeBindings)
.filter(ExternalKnowledgeBindings.external_api_template_id == self.id) .filter(ExternalKnowledgeBindings.external_knowledge_api_id == self.id)
.all() .all()
) )
dataset_ids = [binding.dataset_id for binding in external_knowledge_bindings] dataset_ids = [binding.dataset_id for binding in external_knowledge_bindings]
@ -781,12 +781,12 @@ class ExternalKnowledgeBindings(db.Model):
db.Index("external_knowledge_bindings_tenant_idx", "tenant_id"), db.Index("external_knowledge_bindings_tenant_idx", "tenant_id"),
db.Index("external_knowledge_bindings_dataset_idx", "dataset_id"), db.Index("external_knowledge_bindings_dataset_idx", "dataset_id"),
db.Index("external_knowledge_bindings_external_knowledge_idx", "external_knowledge_id"), db.Index("external_knowledge_bindings_external_knowledge_idx", "external_knowledge_id"),
db.Index("external_knowledge_bindings_external_api_template_idx", "external_api_template_id"), db.Index("external_knowledge_bindings_external_knowledge_api_idx", "external_knowledge_api_id"),
) )
id = db.Column(StringUUID, nullable=False, server_default=db.text("uuid_generate_v4()")) id = db.Column(StringUUID, nullable=False, server_default=db.text("uuid_generate_v4()"))
tenant_id = db.Column(StringUUID, nullable=False) tenant_id = db.Column(StringUUID, nullable=False)
external_api_template_id = db.Column(StringUUID, nullable=False) external_knowledge_api_id = db.Column(StringUUID, nullable=False)
dataset_id = db.Column(StringUUID, nullable=False) dataset_id = db.Column(StringUUID, nullable=False)
external_knowledge_id = db.Column(db.Text, nullable=False) external_knowledge_id = db.Column(db.Text, nullable=False)
created_by = db.Column(StringUUID, nullable=False) created_by = db.Column(StringUUID, nullable=False)

@ -143,7 +143,7 @@ class DatasetService:
account: Account, account: Account,
permission: Optional[str] = None, permission: Optional[str] = None,
provider: str = "vendor", provider: str = "vendor",
external_api_template_id: Optional[str] = None, external_knowledge_api_id: Optional[str] = None,
external_knowledge_id: Optional[str] = None, external_knowledge_id: Optional[str] = None,
): ):
# check if dataset name already exists # check if dataset name already exists
@ -167,14 +167,14 @@ class DatasetService:
db.session.add(dataset) db.session.add(dataset)
db.session.flush() db.session.flush()
if provider == "external" and external_api_template_id: if provider == "external" and external_knowledge_api_id:
external_api_template = ExternalDatasetService.get_api_template(external_api_template_id) external_knowledge_api = ExternalDatasetService.get_external_knowledge_api(external_knowledge_api_id)
if not external_api_template: if not external_knowledge_api:
raise ValueError("External API template not found.") raise ValueError("External API template not found.")
external_knowledge_binding = ExternalKnowledgeBindings( external_knowledge_binding = ExternalKnowledgeBindings(
tenant_id=tenant_id, tenant_id=tenant_id,
dataset_id=dataset.id, dataset_id=dataset.id,
external_api_template_id=external_api_template_id, external_knowledge_api_id=external_knowledge_api_id,
external_knowledge_id=external_knowledge_id, external_knowledge_id=external_knowledge_id,
created_by=account.id, created_by=account.id,
) )

@ -19,7 +19,7 @@ class ProcessStatusSetting(BaseModel):
url: str url: str
class ApiTemplateSetting(BaseModel): class ExternalKnowledgeApiSetting(BaseModel):
url: str url: str
request_method: str request_method: str
headers: Optional[dict] = None headers: Optional[dict] = None

@ -15,26 +15,26 @@ from extensions.ext_database import db
from models.dataset import ( from models.dataset import (
Dataset, Dataset,
Document, Document,
ExternalApiTemplates, ExternalKnowledgeApis,
ExternalKnowledgeBindings, ExternalKnowledgeBindings,
) )
from models.model import UploadFile from models.model import UploadFile
from services.entities.external_knowledge_entities.external_knowledge_entities import ApiTemplateSetting, Authorization from services.entities.external_knowledge_entities.external_knowledge_entities import ExternalKnowledgeApiSetting, Authorization
from services.errors.dataset import DatasetNameDuplicateError from services.errors.dataset import DatasetNameDuplicateError
class ExternalDatasetService: class ExternalDatasetService:
@staticmethod @staticmethod
def get_external_api_templates(page, per_page, tenant_id, search=None) -> tuple[list[ExternalApiTemplates], int]: def get_external_knowledge_api(page, per_page, tenant_id, search=None) -> tuple[list[ExternalKnowledgeApis], int]:
query = ExternalApiTemplates.query.filter(ExternalApiTemplates.tenant_id == tenant_id).order_by( query = ExternalKnowledgeApis.query.filter(ExternalKnowledgeApis.tenant_id == tenant_id).order_by(
ExternalApiTemplates.created_at.desc() ExternalKnowledgeApis.created_at.desc()
) )
if search: if search:
query = query.filter(ExternalApiTemplates.name.ilike(f"%{search}%")) query = query.filter(ExternalKnowledgeApis.name.ilike(f"%{search}%"))
api_templates = query.paginate(page=page, per_page=per_page, max_per_page=100, error_out=False) external_knowledge_apis = query.paginate(page=page, per_page=per_page, max_per_page=100, error_out=False)
return api_templates.items, api_templates.total return external_knowledge_apis.items, external_knowledge_apis.total
@classmethod @classmethod
def validate_api_list(cls, api_settings: dict): def validate_api_list(cls, api_settings: dict):
@ -46,8 +46,8 @@ class ExternalDatasetService:
raise ValueError("api_key is required") raise ValueError("api_key is required")
@staticmethod @staticmethod
def create_api_template(tenant_id: str, user_id: str, args: dict) -> ExternalApiTemplates: def create_external_knowledge_api(tenant_id: str, user_id: str, args: dict) -> ExternalKnowledgeApis:
api_template = ExternalApiTemplates( external_knowledge_api = ExternalKnowledgeApis(
tenant_id=tenant_id, tenant_id=tenant_id,
created_by=user_id, created_by=user_id,
updated_by=user_id, updated_by=user_id,
@ -56,41 +56,41 @@ class ExternalDatasetService:
settings=json.dumps(args.get("settings"), ensure_ascii=False), settings=json.dumps(args.get("settings"), ensure_ascii=False),
) )
db.session.add(api_template) db.session.add(external_knowledge_api)
db.session.commit() db.session.commit()
return api_template return external_knowledge_api
@staticmethod @staticmethod
def get_api_template(external_knowledge_api_id: str) -> ExternalApiTemplates: def get_external_knowledge_api(external_knowledge_api_id: str) -> ExternalKnowledgeApis:
return ExternalApiTemplates.query.filter_by(id=external_knowledge_api_id).first() return ExternalKnowledgeApis.query.filter_by(id=external_knowledge_api_id).first()
@staticmethod @staticmethod
def update_api_template(tenant_id, user_id, external_knowledge_api_id, args) -> ExternalApiTemplates: def update_external_knowledge_api(tenant_id, user_id, external_knowledge_api_id, args) -> ExternalKnowledgeApis:
api_template = ExternalApiTemplates.query.filter_by(id=external_knowledge_api_id, tenant_id=tenant_id).first() external_knowledge_api = ExternalKnowledgeApis.query.filter_by(id=external_knowledge_api_id, tenant_id=tenant_id).first()
if api_template is None: if external_knowledge_api is None:
raise ValueError("api template not found") raise ValueError("api template not found")
api_template.name = args.get("name") external_knowledge_api.name = args.get("name")
api_template.description = args.get("description", "") external_knowledge_api.description = args.get("description", "")
api_template.settings = json.dumps(args.get("settings"), ensure_ascii=False) external_knowledge_api.settings = json.dumps(args.get("settings"), ensure_ascii=False)
api_template.updated_by = user_id external_knowledge_api.updated_by = user_id
api_template.updated_at = datetime.now(timezone.utc).replace(tzinfo=None) external_knowledge_api.updated_at = datetime.now(timezone.utc).replace(tzinfo=None)
db.session.commit() db.session.commit()
return api_template return external_knowledge_api
@staticmethod @staticmethod
def delete_api_template(tenant_id: str, external_knowledge_api_id: str): def delete_external_knowledge_api(tenant_id: str, external_knowledge_api_id: str):
api_template = ExternalApiTemplates.query.filter_by(id=external_knowledge_api_id, tenant_id=tenant_id).first() external_knowledge_api = ExternalKnowledgeApis.query.filter_by(id=external_knowledge_api_id, tenant_id=tenant_id).first()
if api_template is None: if external_knowledge_api is None:
raise ValueError("api template not found") raise ValueError("api template not found")
db.session.delete(api_template) db.session.delete(external_knowledge_api)
db.session.commit() db.session.commit()
@staticmethod @staticmethod
def external_api_template_use_check(external_knowledge_api_id: str) -> tuple[bool, int]: def external_knowledge_api_use_check(external_knowledge_api_id: str) -> tuple[bool, int]:
count = ExternalKnowledgeBindings.query.filter_by(external_api_template_id=external_knowledge_api_id).count() count = ExternalKnowledgeBindings.query.filter_by(external_knowledge_api_id=external_knowledge_api_id).count()
if count > 0: if count > 0:
return True, count return True, count
return False, 0 return False, 0
@ -105,11 +105,11 @@ class ExternalDatasetService:
return external_knowledge_binding return external_knowledge_binding
@staticmethod @staticmethod
def document_create_args_validate(tenant_id: str, api_template_id: str, process_parameter: dict): def document_create_args_validate(tenant_id: str, external_knowledge_api_id: str, process_parameter: dict):
api_template = ExternalApiTemplates.query.filter_by(id=api_template_id, tenant_id=tenant_id).first() external_knowledge_api = ExternalKnowledgeApis.query.filter_by(id=external_knowledge_api_id, tenant_id=tenant_id).first()
if api_template is None: if external_knowledge_api is None:
raise ValueError("api template not found") raise ValueError("api template not found")
settings = json.loads(api_template.settings) settings = json.loads(external_knowledge_api.settings)
for setting in settings: for setting in settings:
custom_parameters = setting.get("document_process_setting") custom_parameters = setting.get("document_process_setting")
if custom_parameters: if custom_parameters:
@ -119,15 +119,15 @@ class ExternalDatasetService:
@staticmethod @staticmethod
def init_external_dataset(tenant_id: str, user_id: str, args: dict, created_from: str = "web"): def init_external_dataset(tenant_id: str, user_id: str, args: dict, created_from: str = "web"):
api_template_id = args.get("api_template_id") external_knowledge_api_id = args.get("external_knowledge_api_id")
data_source = args.get("data_source") data_source = args.get("data_source")
if data_source is None: if data_source is None:
raise ValueError("data source is required") raise ValueError("data source is required")
process_parameter = args.get("process_parameter") process_parameter = args.get("process_parameter")
api_template = ExternalApiTemplates.query.filter_by(id=api_template_id, tenant_id=tenant_id).first() external_knowledge_api = ExternalKnowledgeApis.query.filter_by(id=external_knowledge_api_id, tenant_id=tenant_id).first()
if api_template is None: if external_knowledge_api is None:
raise ValueError("api template not found") raise ValueError("api template not found")
dataset = Dataset( dataset = Dataset(
@ -175,12 +175,12 @@ class ExternalDatasetService:
db.session.flush() db.session.flush()
document_ids.append(document.id) document_ids.append(document.id)
db.session.commit() db.session.commit()
# external_document_indexing_task.delay(dataset.id, api_template_id, data_source, process_parameter) # external_document_indexing_task.delay(dataset.id, external_knowledge_api_id, data_source, process_parameter)
return dataset return dataset
@staticmethod @staticmethod
def process_external_api(settings: ApiTemplateSetting, files: Union[None, dict[str, Any]]) -> httpx.Response: def process_external_api(settings: ExternalKnowledgeApiSetting, files: Union[None, dict[str, Any]]) -> httpx.Response:
""" """
do http request depending on api bundle do http request depending on api bundle
""" """
@ -222,19 +222,19 @@ class ExternalDatasetService:
return headers return headers
@staticmethod @staticmethod
def get_api_template_settings(settings: dict) -> ApiTemplateSetting: def get_external_knowledge_api_settings(settings: dict) -> ExternalKnowledgeApiSetting:
return ApiTemplateSetting.parse_obj(settings) return ExternalKnowledgeApiSetting.parse_obj(settings)
@staticmethod @staticmethod
def create_external_dataset(tenant_id: str, user_id: str, args: dict) -> Dataset: def create_external_dataset(tenant_id: str, user_id: str, args: dict) -> Dataset:
# check if dataset name already exists # check if dataset name already exists
if Dataset.query.filter_by(name=args.get("name"), tenant_id=tenant_id).first(): if Dataset.query.filter_by(name=args.get("name"), tenant_id=tenant_id).first():
raise DatasetNameDuplicateError(f"Dataset with name {args.get('name')} already exists.") raise DatasetNameDuplicateError(f"Dataset with name {args.get('name')} already exists.")
api_template = ExternalApiTemplates.query.filter_by( external_knowledge_api = ExternalKnowledgeApis.query.filter_by(
id=args.get("external_api_template_id"), tenant_id=tenant_id id=args.get("external_knowledge_api_id"), tenant_id=tenant_id
).first() ).first()
if api_template is None: if external_knowledge_api is None:
raise ValueError("api template not found") raise ValueError("api template not found")
dataset = Dataset( dataset = Dataset(
@ -252,7 +252,7 @@ class ExternalDatasetService:
external_knowledge_binding = ExternalKnowledgeBindings( external_knowledge_binding = ExternalKnowledgeBindings(
tenant_id=tenant_id, tenant_id=tenant_id,
dataset_id=dataset.id, dataset_id=dataset.id,
external_api_template_id=args.get("external_api_template_id"), external_knowledge_api_id=args.get("external_knowledge_api_id"),
external_knowledge_id=args.get("external_knowledge_id"), external_knowledge_id=args.get("external_knowledge_id"),
created_by=user_id, created_by=user_id,
) )
@ -272,13 +272,13 @@ class ExternalDatasetService:
if not external_knowledge_binding: if not external_knowledge_binding:
raise ValueError("external knowledge binding not found") raise ValueError("external knowledge binding not found")
external_api_template = ExternalApiTemplates.query.filter_by( external_knowledge_api = ExternalKnowledgeApis.query.filter_by(
id=external_knowledge_binding.external_api_template_id id=external_knowledge_binding.external_knowledge_api_id
).first() ).first()
if not external_api_template: if not external_knowledge_api:
raise ValueError("external api template not found") raise ValueError("external api template not found")
settings = json.loads(external_api_template.settings) settings = json.loads(external_knowledge_api.settings)
headers = {"Content-Type": "application/json"} headers = {"Content-Type": "application/json"}
if settings.get("api_key"): if settings.get("api_key"):
headers["Authorization"] = f"Bearer {settings.get('api_key')}" headers["Authorization"] = f"Bearer {settings.get('api_key')}"
@ -286,13 +286,13 @@ class ExternalDatasetService:
external_retrieval_parameters["query"] = query external_retrieval_parameters["query"] = query
external_retrieval_parameters["external_knowledge_id"] = external_knowledge_binding.external_knowledge_id external_retrieval_parameters["external_knowledge_id"] = external_knowledge_binding.external_knowledge_id
api_template_setting = { external_knowledge_api_setting = {
"url": f"{settings.get('endpoint')}/dify/external-knowledge/retrieval-documents", "url": f"{settings.get('endpoint')}/dify/external-knowledge/retrieval-documents",
"request_method": "post", "request_method": "post",
"headers": headers, "headers": headers,
"params": external_retrieval_parameters, "params": external_retrieval_parameters,
} }
response = ExternalDatasetService.process_external_api(ApiTemplateSetting(**api_template_setting), None) response = ExternalDatasetService.process_external_api(ExternalKnowledgeApiSetting(**external_knowledge_api_setting), None)
if response.status_code == 200: if response.status_code == 200:
return response.json() return response.json()
return [] return []

@ -8,17 +8,17 @@ from celery import shared_task
from core.indexing_runner import DocumentIsPausedException from core.indexing_runner import DocumentIsPausedException
from extensions.ext_database import db from extensions.ext_database import db
from extensions.ext_storage import storage from extensions.ext_storage import storage
from models.dataset import Dataset, ExternalApiTemplates from models.dataset import Dataset, ExternalKnowledgeApis
from models.model import UploadFile from models.model import UploadFile
from services.external_knowledge_service import ExternalDatasetService from services.external_knowledge_service import ExternalDatasetService
@shared_task(queue="dataset") @shared_task(queue="dataset")
def external_document_indexing_task(dataset_id: str, api_template_id: str, data_source: dict, process_parameter: dict): def external_document_indexing_task(dataset_id: str, external_knowledge_api_id: str, data_source: dict, process_parameter: dict):
""" """
Async process document Async process document
:param dataset_id: :param dataset_id:
:param api_template_id: :param external_knowledge_api_id:
:param data_source: :param data_source:
:param process_parameter: :param process_parameter:
Usage: external_document_indexing_task.delay(dataset_id, document_id) Usage: external_document_indexing_task.delay(dataset_id, document_id)
@ -33,16 +33,16 @@ def external_document_indexing_task(dataset_id: str, api_template_id: str, data_
return return
# get external api template # get external api template
api_template = ( external_knowledge_api = (
db.session.query(ExternalApiTemplates) db.session.query(ExternalKnowledgeApis)
.filter(ExternalApiTemplates.id == api_template_id, ExternalApiTemplates.tenant_id == dataset.tenant_id) .filter(ExternalKnowledgeApis.id == external_knowledge_api_id, ExternalKnowledgeApis.tenant_id == dataset.tenant_id)
.first() .first()
) )
if not api_template: if not external_knowledge_api:
logging.info( logging.info(
click.style( click.style(
"Processed external dataset: {} failed, api template: {} not exit.".format(dataset_id, api_template_id), "Processed external dataset: {} failed, api template: {} not exit.".format(dataset_id, external_knowledge_api_id),
fg="red", fg="red",
) )
) )
@ -59,7 +59,7 @@ def external_document_indexing_task(dataset_id: str, api_template_id: str, data_
if file: if file:
files[file.id] = (file.name, storage.load_once(file.key), file.mime_type) files[file.id] = (file.name, storage.load_once(file.key), file.mime_type)
try: try:
settings = ExternalDatasetService.get_api_template_settings(json.loads(api_template.settings)) settings = ExternalDatasetService.get_external_knowledge_api_settings(json.loads(external_knowledge_api.settings))
# assemble headers # assemble headers
headers = ExternalDatasetService.assembling_headers(settings.authorization, settings.headers) headers = ExternalDatasetService.assembling_headers(settings.authorization, settings.headers)

Loading…
Cancel
Save