mirror of
https://github.com/infiniflow/ragflow.git
synced 2025-12-20 04:39:00 +08:00
Change knowledge base to dataset (#11976)
### What problem does this PR solve? As title ### Type of change - [x] Refactoring --------- Signed-off-by: Jin Hai <haijin.chn@gmail.com>
This commit is contained in:
Jin Hai
@ -30,9 +30,9 @@ from api.utils.api_utils import get_parser_config, get_data_error_result
|
||||
|
||||
|
||||
class KnowledgebaseService(CommonService):
|
||||
"""Service class for managing knowledge base operations.
|
||||
"""Service class for managing dataset operations.
|
||||
|
||||
This class extends CommonService to provide specialized functionality for knowledge base
|
||||
This class extends CommonService to provide specialized functionality for dataset
|
||||
management, including document parsing status tracking, access control, and configuration
|
||||
management. It handles operations such as listing, creating, updating, and deleting
|
||||
knowledge bases, as well as managing their associated documents and permissions.
|
||||
@ -41,7 +41,7 @@ class KnowledgebaseService(CommonService):
|
||||
- Document parsing status verification
|
||||
- Knowledge base access control
|
||||
- Parser configuration management
|
||||
- Tenant-based knowledge base organization
|
||||
- Tenant-based dataset organization
|
||||
|
||||
Attributes:
|
||||
model: The Knowledgebase model class for database operations.
|
||||
@ -51,18 +51,18 @@ class KnowledgebaseService(CommonService):
|
||||
@classmethod
|
||||
@DB.connection_context()
|
||||
def accessible4deletion(cls, kb_id, user_id):
|
||||
"""Check if a knowledge base can be deleted by a specific user.
|
||||
"""Check if a dataset can be deleted by a specific user.
|
||||
|
||||
This method verifies whether a user has permission to delete a knowledge base
|
||||
by checking if they are the creator of that knowledge base.
|
||||
This method verifies whether a user has permission to delete a dataset
|
||||
by checking if they are the creator of that dataset.
|
||||
|
||||
Args:
|
||||
kb_id (str): The unique identifier of the knowledge base to check.
|
||||
kb_id (str): The unique identifier of the dataset to check.
|
||||
user_id (str): The unique identifier of the user attempting the deletion.
|
||||
|
||||
Returns:
|
||||
bool: True if the user has permission to delete the knowledge base,
|
||||
False if the user doesn't have permission or the knowledge base doesn't exist.
|
||||
bool: True if the user has permission to delete the dataset,
|
||||
False if the user doesn't have permission or the dataset doesn't exist.
|
||||
|
||||
Example:
|
||||
>>> KnowledgebaseService.accessible4deletion("kb123", "user456")
|
||||
@ -71,10 +71,10 @@ class KnowledgebaseService(CommonService):
|
||||
Note:
|
||||
- This method only checks creator permissions
|
||||
- A return value of False can mean either:
|
||||
1. The knowledge base doesn't exist
|
||||
2. The user is not the creator of the knowledge base
|
||||
1. The dataset doesn't exist
|
||||
2. The user is not the creator of the dataset
|
||||
"""
|
||||
# Check if a knowledge base can be deleted by a user
|
||||
# Check if a dataset can be deleted by a user
|
||||
docs = cls.model.select(
|
||||
cls.model.id).where(cls.model.id == kb_id, cls.model.created_by == user_id).paginate(0, 1)
|
||||
docs = docs.dicts()
|
||||
@ -85,7 +85,7 @@ class KnowledgebaseService(CommonService):
|
||||
@classmethod
|
||||
@DB.connection_context()
|
||||
def is_parsed_done(cls, kb_id):
|
||||
# Check if all documents in the knowledge base have completed parsing
|
||||
# Check if all documents in the dataset have completed parsing
|
||||
#
|
||||
# Args:
|
||||
# kb_id: Knowledge base ID
|
||||
@ -96,13 +96,13 @@ class KnowledgebaseService(CommonService):
|
||||
from common.constants import TaskStatus
|
||||
from api.db.services.document_service import DocumentService
|
||||
|
||||
# Get knowledge base information
|
||||
# Get dataset information
|
||||
kbs = cls.query(id=kb_id)
|
||||
if not kbs:
|
||||
return False, "Knowledge base not found"
|
||||
kb = kbs[0]
|
||||
|
||||
# Get all documents in the knowledge base
|
||||
# Get all documents in the dataset
|
||||
docs, _ = DocumentService.get_by_kb_id(kb_id, 1, 1000, "create_time", True, "", [], [])
|
||||
|
||||
# Check parsing status of each document
|
||||
@ -119,9 +119,9 @@ class KnowledgebaseService(CommonService):
|
||||
@classmethod
|
||||
@DB.connection_context()
|
||||
def list_documents_by_ids(cls, kb_ids):
|
||||
# Get document IDs associated with given knowledge base IDs
|
||||
# Get document IDs associated with given dataset IDs
|
||||
# Args:
|
||||
# kb_ids: List of knowledge base IDs
|
||||
# kb_ids: List of dataset IDs
|
||||
# Returns:
|
||||
# List of document IDs
|
||||
doc_ids = cls.model.select(Document.id.alias("document_id")).join(Document, on=(cls.model.id == Document.kb_id)).where(
|
||||
@ -235,11 +235,11 @@ class KnowledgebaseService(CommonService):
|
||||
@classmethod
|
||||
@DB.connection_context()
|
||||
def get_kb_ids(cls, tenant_id):
|
||||
# Get all knowledge base IDs for a tenant
|
||||
# Get all dataset IDs for a tenant
|
||||
# Args:
|
||||
# tenant_id: Tenant ID
|
||||
# Returns:
|
||||
# List of knowledge base IDs
|
||||
# List of dataset IDs
|
||||
fields = [
|
||||
cls.model.id,
|
||||
]
|
||||
@ -250,11 +250,11 @@ class KnowledgebaseService(CommonService):
|
||||
@classmethod
|
||||
@DB.connection_context()
|
||||
def get_detail(cls, kb_id):
|
||||
# Get detailed information about a knowledge base
|
||||
# Get detailed information about a dataset
|
||||
# Args:
|
||||
# kb_id: Knowledge base ID
|
||||
# Returns:
|
||||
# Dictionary containing knowledge base details
|
||||
# Dictionary containing dataset details
|
||||
fields = [
|
||||
cls.model.id,
|
||||
cls.model.embd_id,
|
||||
@ -294,13 +294,13 @@ class KnowledgebaseService(CommonService):
|
||||
@classmethod
|
||||
@DB.connection_context()
|
||||
def update_parser_config(cls, id, config):
|
||||
# Update parser configuration for a knowledge base
|
||||
# Update parser configuration for a dataset
|
||||
# Args:
|
||||
# id: Knowledge base ID
|
||||
# config: New parser configuration
|
||||
e, m = cls.get_by_id(id)
|
||||
if not e:
|
||||
raise LookupError(f"knowledgebase({id}) not found.")
|
||||
raise LookupError(f"dataset({id}) not found.")
|
||||
|
||||
def dfs_update(old, new):
|
||||
# Deep update of nested configuration
|
||||
@ -325,7 +325,7 @@ class KnowledgebaseService(CommonService):
|
||||
def delete_field_map(cls, id):
|
||||
e, m = cls.get_by_id(id)
|
||||
if not e:
|
||||
raise LookupError(f"knowledgebase({id}) not found.")
|
||||
raise LookupError(f"dataset({id}) not found.")
|
||||
|
||||
m.parser_config.pop("field_map", None)
|
||||
cls.update_by_id(id, {"parser_config": m.parser_config})
|
||||
@ -335,7 +335,7 @@ class KnowledgebaseService(CommonService):
|
||||
def get_field_map(cls, ids):
|
||||
# Get field mappings for knowledge bases
|
||||
# Args:
|
||||
# ids: List of knowledge base IDs
|
||||
# ids: List of dataset IDs
|
||||
# Returns:
|
||||
# Dictionary of field mappings
|
||||
conf = {}
|
||||
@ -347,7 +347,7 @@ class KnowledgebaseService(CommonService):
|
||||
@classmethod
|
||||
@DB.connection_context()
|
||||
def get_by_name(cls, kb_name, tenant_id):
|
||||
# Get knowledge base by name and tenant ID
|
||||
# Get dataset by name and tenant ID
|
||||
# Args:
|
||||
# kb_name: Knowledge base name
|
||||
# tenant_id: Tenant ID
|
||||
@ -365,9 +365,9 @@ class KnowledgebaseService(CommonService):
|
||||
@classmethod
|
||||
@DB.connection_context()
|
||||
def get_all_ids(cls):
|
||||
# Get all knowledge base IDs
|
||||
# Get all dataset IDs
|
||||
# Returns:
|
||||
# List of all knowledge base IDs
|
||||
# List of all dataset IDs
|
||||
return [m["id"] for m in cls.model.select(cls.model.id).dicts()]
|
||||
|
||||
|
||||
@ -471,7 +471,7 @@ class KnowledgebaseService(CommonService):
|
||||
@classmethod
|
||||
@DB.connection_context()
|
||||
def accessible(cls, kb_id, user_id):
|
||||
# Check if a knowledge base is accessible by a user
|
||||
# Check if a dataset is accessible by a user
|
||||
# Args:
|
||||
# kb_id: Knowledge base ID
|
||||
# user_id: User ID
|
||||
@ -488,12 +488,12 @@ class KnowledgebaseService(CommonService):
|
||||
@classmethod
|
||||
@DB.connection_context()
|
||||
def get_kb_by_id(cls, kb_id, user_id):
|
||||
# Get knowledge base by ID and user ID
|
||||
# Get dataset by ID and user ID
|
||||
# Args:
|
||||
# kb_id: Knowledge base ID
|
||||
# user_id: User ID
|
||||
# Returns:
|
||||
# List containing knowledge base information
|
||||
# List containing dataset information
|
||||
kbs = cls.model.select().join(UserTenant, on=(UserTenant.tenant_id == Knowledgebase.tenant_id)
|
||||
).where(cls.model.id == kb_id, UserTenant.user_id == user_id).paginate(0, 1)
|
||||
kbs = kbs.dicts()
|
||||
@ -502,12 +502,12 @@ class KnowledgebaseService(CommonService):
|
||||
@classmethod
|
||||
@DB.connection_context()
|
||||
def get_kb_by_name(cls, kb_name, user_id):
|
||||
# Get knowledge base by name and user ID
|
||||
# Get dataset by name and user ID
|
||||
# Args:
|
||||
# kb_name: Knowledge base name
|
||||
# user_id: User ID
|
||||
# Returns:
|
||||
# List containing knowledge base information
|
||||
# List containing dataset information
|
||||
kbs = cls.model.select().join(UserTenant, on=(UserTenant.tenant_id == Knowledgebase.tenant_id)
|
||||
).where(cls.model.name == kb_name, UserTenant.user_id == user_id).paginate(0, 1)
|
||||
kbs = kbs.dicts()
|
||||
|
||||
Reference in New Issue
Block a user