"""Response schemas for knowledge graph operations.
This module defines the response formats for all knowledge graph operations.
Each response includes complete information about the affected entities,
including IDs that can be used in subsequent operations.
Key Features:
1. Every created/updated object gets an ID
2. Relations are included with their parent entities
3. Responses include everything needed for next operations
4. Bulk operations return all affected items
"""
from datetime import datetime
from typing import List, Optional, Dict
from pydantic import BaseModel, ConfigDict, Field, model_validator
from basic_memory.schemas.base import Relation, Permalink, EntityType, ContentType, Observation
class SQLAlchemyModel(BaseModel):
"""Base class for models that read from SQLAlchemy attributes.
This base class handles conversion of SQLAlchemy model attributes
to Pydantic model fields. All response models extend this to ensure
proper handling of database results.
"""
model_config = ConfigDict(from_attributes=True)
class ObservationResponse(Observation, SQLAlchemyModel):
"""Schema for observation data returned from the service.
Each observation gets a unique ID that can be used for later
reference or deletion.
Example Response:
{
"category": "feature",
"content": "Added support for async operations",
"context": "Initial database design meeting"
}
"""
permalink: Permalink
class RelationResponse(Relation, SQLAlchemyModel):
"""Response schema for relation operations.
Extends the base Relation model with a unique ID that can be
used for later modification or deletion.
Example Response:
{
"from_id": "test/memory_test",
"to_id": "component/memory-service",
"relation_type": "validates",
"context": "Comprehensive test suite"
}
"""
permalink: Permalink
# Override base Relation fields to allow Optional values
from_id: Optional[Permalink] = Field(default=None) # pyright: ignore[reportIncompatibleVariableOverride]
to_id: Optional[Permalink] = Field(default=None) # pyright: ignore[reportIncompatibleVariableOverride]
to_name: Optional[str] = Field(default=None)
@model_validator(mode="before")
@classmethod
def resolve_entity_references(cls, data):
"""Resolve from_id and to_id from joined entities, falling back to file_path.
When loading from SQLAlchemy models, the from_entity and to_entity relationships
are joined. We extract the permalink from these entities, falling back to
file_path when permalink is None.
We use file_path directly (not converted to permalink format) because if the
entity doesn't have a permalink, the system won't be able to find it by a
generated one anyway. Using the actual file_path preserves the real identifier.
"""
# Handle dict input (e.g., from API or tests)
if isinstance(data, dict):
from_entity = data.get("from_entity")
to_entity = data.get("to_entity")
# Resolve from_id: prefer permalink, fall back to file_path
if from_entity and isinstance(from_entity, dict):
permalink = from_entity.get("permalink")
if permalink:
data["from_id"] = permalink
elif from_entity.get("file_path"):
data["from_id"] = from_entity["file_path"]
# Resolve to_id: prefer permalink, fall back to file_path
if to_entity and isinstance(to_entity, dict):
permalink = to_entity.get("permalink")
if permalink:
data["to_id"] = permalink
elif to_entity.get("file_path"):
data["to_id"] = to_entity["file_path"]
# Also resolve to_name from entity title
if to_entity.get("title") and not data.get("to_name"):
data["to_name"] = to_entity["title"]
return data
# Handle SQLAlchemy model input (from_attributes=True)
# Access attributes directly from the ORM model
from_entity = getattr(data, "from_entity", None)
to_entity = getattr(data, "to_entity", None)
# Build a dict from the model's attributes
result = {}
# Copy base fields
for field in ["permalink", "relation_type", "context", "to_name"]:
if hasattr(data, field):
result[field] = getattr(data, field)
# Resolve from_id: prefer permalink, fall back to file_path
if from_entity:
permalink = getattr(from_entity, "permalink", None)
file_path = getattr(from_entity, "file_path", None)
if permalink:
result["from_id"] = permalink
elif file_path:
result["from_id"] = file_path
# Resolve to_id: prefer permalink, fall back to file_path
if to_entity:
permalink = getattr(to_entity, "permalink", None)
file_path = getattr(to_entity, "file_path", None)
if permalink:
result["to_id"] = permalink
elif file_path:
result["to_id"] = file_path
# Also resolve to_name from entity title if not set
if not result.get("to_name"):
title = getattr(to_entity, "title", None)
if title:
result["to_name"] = title
return result
class EntityResponse(SQLAlchemyModel):
"""Complete entity data returned from the service.
This is the most comprehensive entity view, including:
1. Basic entity details (id, name, type)
2. All observations with their IDs
3. All relations with their IDs
4. Optional description
Example Response:
{
"permalink": "component/memory-service",
"file_path": "MemoryService",
"entity_type": "component",
"entity_metadata": {}
"content_type: "text/markdown"
"observations": [
{
"category": "feature",
"content": "Uses SQLite storage"
"context": "Initial design"
},
{
"category": "feature",
"content": "Implements async operations"
"context": "Initial design"
}
],
"relations": [
{
"from_id": "test/memory-test",
"to_id": "component/memory-service",
"relation_type": "validates",
"context": "Main test suite"
}
]
}
"""
permalink: Optional[Permalink]
title: str
file_path: str
entity_type: EntityType
entity_metadata: Optional[Dict] = None
checksum: Optional[str] = None
content_type: ContentType
observations: List[ObservationResponse] = []
relations: List[RelationResponse] = []
created_at: datetime
updated_at: datetime
class EntityListResponse(SQLAlchemyModel):
"""Response for create_entities operation.
Returns complete information about entities returned from the service,
including their permalinks, observations,
and any established relations.
Example Response:
{
"entities": [
{
"permalink": "component/search_service",
"title": "SearchService",
"entity_type": "component",
"description": "Knowledge graph search",
"observations": [
{
"content": "Implements full-text search"
}
],
"relations": []
},
{
"permalink": "document/api_docs",
"title": "API_Documentation",
"entity_type": "document",
"description": "API Reference",
"observations": [
{
"content": "Documents REST endpoints"
}
],
"relations": []
}
]
}
"""
entities: List[EntityResponse]
class SearchNodesResponse(SQLAlchemyModel):
"""Response for search operation.
Returns matching entities with their complete information,
plus the original query for reference.
Example Response:
{
"matches": [
{
"permalink": "component/memory-service",
"title": "MemoryService",
"entity_type": "component",
"description": "Core service",
"observations": [...],
"relations": [...]
}
],
"query": "memory"
}
Note: Each entity in matches includes full details
just like EntityResponse.
"""
matches: List[EntityResponse]
query: str
class DeleteEntitiesResponse(SQLAlchemyModel):
"""Response indicating successful entity deletion.
A simple boolean response confirming the delete operation
completed successfully.
Example Response:
{
"deleted": true
}
"""
deleted: bool
