Files
letta-server/letta/services/block_manager.py
2025-07-25 13:06:12 -07:00

691 lines
30 KiB
Python
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
import asyncio
from datetime import datetime
from typing import Dict, List, Optional
from sqlalchemy import delete, or_, select
from sqlalchemy.orm import Session
from letta.log import get_logger
from letta.orm.agent import Agent as AgentModel
from letta.orm.block import Block as BlockModel
from letta.orm.block_history import BlockHistory
from letta.orm.blocks_agents import BlocksAgents
from letta.orm.errors import NoResultFound
from letta.otel.tracing import trace_method
from letta.schemas.agent import AgentState as PydanticAgentState
from letta.schemas.block import Block as PydanticBlock
from letta.schemas.block import BlockUpdate
from letta.schemas.enums import ActorType
from letta.schemas.user import User as PydanticUser
from letta.server.db import db_registry
from letta.settings import DatabaseChoice, settings
from letta.utils import enforce_types
logger = get_logger(__name__)
class BlockManager:
"""Manager class to handle business logic related to Blocks."""
@enforce_types
@trace_method
def create_or_update_block(self, block: PydanticBlock, actor: PydanticUser) -> PydanticBlock:
"""Create a new block based on the Block schema."""
db_block = self.get_block_by_id(block.id, actor)
if db_block:
update_data = BlockUpdate(**block.model_dump(to_orm=True, exclude_none=True))
self.update_block(block.id, update_data, actor)
else:
with db_registry.session() as session:
data = block.model_dump(to_orm=True, exclude_none=True)
block = BlockModel(**data, organization_id=actor.organization_id)
block.create(session, actor=actor)
return block.to_pydantic()
@enforce_types
@trace_method
async def create_or_update_block_async(self, block: PydanticBlock, actor: PydanticUser) -> PydanticBlock:
"""Create a new block based on the Block schema."""
db_block = await self.get_block_by_id_async(block.id, actor)
if db_block:
update_data = BlockUpdate(**block.model_dump(to_orm=True, exclude_none=True))
return await self.update_block_async(block.id, update_data, actor)
else:
async with db_registry.async_session() as session:
data = block.model_dump(to_orm=True, exclude_none=True)
block = BlockModel(**data, organization_id=actor.organization_id)
await block.create_async(session, actor=actor, no_commit=True, no_refresh=True)
pydantic_block = block.to_pydantic()
await session.commit()
return pydantic_block
@enforce_types
@trace_method
def batch_create_blocks(self, blocks: List[PydanticBlock], actor: PydanticUser) -> List[PydanticBlock]:
"""
Batch-create multiple Blocks in one transaction for better performance.
Args:
blocks: List of PydanticBlock schemas to create
actor: The user performing the operation
Returns:
List of created PydanticBlock instances (with IDs, timestamps, etc.)
"""
if not blocks:
return []
with db_registry.session() as session:
block_models = [
BlockModel(**block.model_dump(to_orm=True, exclude_none=True), organization_id=actor.organization_id) for block in blocks
]
created_models = BlockModel.batch_create(items=block_models, db_session=session, actor=actor)
# Convert back to Pydantic
return [m.to_pydantic() for m in created_models]
@enforce_types
@trace_method
async def batch_create_blocks_async(self, blocks: List[PydanticBlock], actor: PydanticUser) -> List[PydanticBlock]:
"""
Batch-create multiple Blocks in one transaction for better performance.
Args:
blocks: List of PydanticBlock schemas to create
actor: The user performing the operation
Returns:
List of created PydanticBlock instances (with IDs, timestamps, etc.)
"""
if not blocks:
return []
async with db_registry.async_session() as session:
block_models = [
BlockModel(**block.model_dump(to_orm=True, exclude_none=True), organization_id=actor.organization_id) for block in blocks
]
created_models = await BlockModel.batch_create_async(
items=block_models, db_session=session, actor=actor, no_commit=True, no_refresh=True
)
result = [m.to_pydantic() for m in created_models]
await session.commit()
return result
@enforce_types
@trace_method
def update_block(self, block_id: str, block_update: BlockUpdate, actor: PydanticUser) -> PydanticBlock:
"""Update a block by its ID with the given BlockUpdate object."""
# Safety check for block
with db_registry.session() as session:
block = BlockModel.read(db_session=session, identifier=block_id, actor=actor)
update_data = block_update.model_dump(to_orm=True, exclude_unset=True, exclude_none=True)
for key, value in update_data.items():
setattr(block, key, value)
block.update(db_session=session, actor=actor)
return block.to_pydantic()
@enforce_types
@trace_method
async def update_block_async(self, block_id: str, block_update: BlockUpdate, actor: PydanticUser) -> PydanticBlock:
"""Update a block by its ID with the given BlockUpdate object."""
# Safety check for block
async with db_registry.async_session() as session:
block = await BlockModel.read_async(db_session=session, identifier=block_id, actor=actor)
update_data = block_update.model_dump(to_orm=True, exclude_unset=True, exclude_none=True)
for key, value in update_data.items():
setattr(block, key, value)
await block.update_async(db_session=session, actor=actor, no_commit=True, no_refresh=True)
pydantic_block = block.to_pydantic()
await session.commit()
return pydantic_block
@enforce_types
@trace_method
def delete_block(self, block_id: str, actor: PydanticUser) -> None:
"""Delete a block by its ID."""
with db_registry.session() as session:
# First, delete all references in blocks_agents table
session.execute(delete(BlocksAgents).where(BlocksAgents.block_id == block_id))
session.flush()
# Then delete the block itself
block = BlockModel.read(db_session=session, identifier=block_id)
block.hard_delete(db_session=session, actor=actor)
@enforce_types
@trace_method
async def delete_block_async(self, block_id: str, actor: PydanticUser) -> None:
"""Delete a block by its ID."""
async with db_registry.async_session() as session:
# First, delete all references in blocks_agents table
await session.execute(delete(BlocksAgents).where(BlocksAgents.block_id == block_id))
await session.flush()
# Then delete the block itself
block = await BlockModel.read_async(db_session=session, identifier=block_id, actor=actor)
await block.hard_delete_async(db_session=session, actor=actor)
@enforce_types
@trace_method
async def get_blocks_async(
self,
actor: PydanticUser,
label: Optional[str] = None,
is_template: Optional[bool] = None,
template_name: Optional[str] = None,
identity_id: Optional[str] = None,
identifier_keys: Optional[List[str]] = None,
project_id: Optional[str] = None,
before: Optional[str] = None,
after: Optional[str] = None,
limit: Optional[int] = 50,
ascending: bool = True,
) -> List[PydanticBlock]:
"""Async version of get_blocks method. Retrieve blocks based on various optional filters."""
from sqlalchemy import select
from sqlalchemy.orm import noload
from letta.orm.sqlalchemy_base import AccessType
async with db_registry.async_session() as session:
# Start with a basic query
query = select(BlockModel)
# Explicitly avoid loading relationships
query = query.options(noload(BlockModel.agents), noload(BlockModel.identities), noload(BlockModel.groups))
# Apply access control
query = BlockModel.apply_access_predicate(query, actor, ["read"], AccessType.ORGANIZATION)
# Add filters
query = query.where(BlockModel.organization_id == actor.organization_id)
if label:
query = query.where(BlockModel.label == label)
if is_template is not None:
query = query.where(BlockModel.is_template == is_template)
if template_name:
query = query.where(BlockModel.template_name == template_name)
if project_id:
query = query.where(BlockModel.project_id == project_id)
needs_distinct = False
if identifier_keys:
query = query.join(BlockModel.identities).filter(
BlockModel.identities.property.mapper.class_.identifier_key.in_(identifier_keys)
)
needs_distinct = True
if identity_id:
query = query.join(BlockModel.identities).filter(BlockModel.identities.property.mapper.class_.id == identity_id)
needs_distinct = True
if after:
result = (await session.execute(select(BlockModel.created_at, BlockModel.id).where(BlockModel.id == after))).first()
if result:
after_sort_value, after_id = result
# SQLite does not support as granular timestamping, so we need to round the timestamp
if settings.database_engine is DatabaseChoice.SQLITE and isinstance(after_sort_value, datetime):
after_sort_value = after_sort_value.strftime("%Y-%m-%d %H:%M:%S")
if ascending:
query = query.where(
BlockModel.created_at > after_sort_value,
or_(BlockModel.created_at == after_sort_value, BlockModel.id > after_id),
)
else:
query = query.where(
BlockModel.created_at < after_sort_value,
or_(BlockModel.created_at == after_sort_value, BlockModel.id < after_id),
)
if before:
result = (await session.execute(select(BlockModel.created_at, BlockModel.id).where(BlockModel.id == before))).first()
if result:
before_sort_value, before_id = result
# SQLite does not support as granular timestamping, so we need to round the timestamp
if settings.database_engine is DatabaseChoice.SQLITE and isinstance(before_sort_value, datetime):
before_sort_value = before_sort_value.strftime("%Y-%m-%d %H:%M:%S")
if ascending:
query = query.where(
BlockModel.created_at < before_sort_value,
or_(BlockModel.created_at == before_sort_value, BlockModel.id < before_id),
)
else:
query = query.where(
BlockModel.created_at > before_sort_value,
or_(BlockModel.created_at == before_sort_value, BlockModel.id > before_id),
)
# Apply ordering and handle distinct if needed
if needs_distinct:
if ascending:
query = query.distinct(BlockModel.id).order_by(BlockModel.id.asc(), BlockModel.created_at.asc())
else:
query = query.distinct(BlockModel.id).order_by(BlockModel.id.desc(), BlockModel.created_at.desc())
else:
if ascending:
query = query.order_by(BlockModel.created_at.asc(), BlockModel.id.asc())
else:
query = query.order_by(BlockModel.created_at.desc(), BlockModel.id.desc())
# Add limit
if limit:
query = query.limit(limit)
# Execute the query
result = await session.execute(query)
blocks = result.scalars().all()
return [block.to_pydantic() for block in blocks]
@enforce_types
@trace_method
def get_block_by_id(self, block_id: str, actor: Optional[PydanticUser] = None) -> Optional[PydanticBlock]:
"""Retrieve a block by its name."""
with db_registry.session() as session:
try:
block = BlockModel.read(db_session=session, identifier=block_id, actor=actor)
return block.to_pydantic()
except NoResultFound:
return None
@enforce_types
@trace_method
async def get_block_by_id_async(self, block_id: str, actor: Optional[PydanticUser] = None) -> Optional[PydanticBlock]:
"""Retrieve a block by its name."""
async with db_registry.async_session() as session:
try:
block = await BlockModel.read_async(db_session=session, identifier=block_id, actor=actor)
return block.to_pydantic()
except NoResultFound:
return None
@enforce_types
@trace_method
async def get_all_blocks_by_ids_async(self, block_ids: List[str], actor: Optional[PydanticUser] = None) -> List[PydanticBlock]:
"""Retrieve blocks by their ids without loading unnecessary relationships. Async implementation."""
from sqlalchemy import select
from sqlalchemy.orm import noload
from letta.orm.sqlalchemy_base import AccessType
if not block_ids:
return []
async with db_registry.async_session() as session:
# Start with a basic query
query = select(BlockModel)
# Add ID filter
query = query.where(BlockModel.id.in_(block_ids))
# Explicitly avoid loading relationships
query = query.options(noload(BlockModel.agents), noload(BlockModel.identities), noload(BlockModel.groups))
# Apply access control if actor is provided
if actor:
query = BlockModel.apply_access_predicate(query, actor, ["read"], AccessType.ORGANIZATION)
# TODO: Add soft delete filter if applicable
# if hasattr(BlockModel, "is_deleted"):
# query = query.where(BlockModel.is_deleted == False)
# Execute the query
result = await session.execute(query)
blocks = result.scalars().all()
# Convert to Pydantic models
pydantic_blocks = [block.to_pydantic() for block in blocks]
# For backward compatibility, add None for missing blocks
if len(pydantic_blocks) < len(block_ids):
{block.id for block in pydantic_blocks}
result_blocks = []
for block_id in block_ids:
block = next((b for b in pydantic_blocks if b.id == block_id), None)
result_blocks.append(block)
return result_blocks
return pydantic_blocks
@enforce_types
@trace_method
async def get_agents_for_block_async(
self,
block_id: str,
actor: PydanticUser,
include_relationships: Optional[List[str]] = None,
before: Optional[str] = None,
after: Optional[str] = None,
limit: Optional[int] = 50,
ascending: bool = True,
) -> List[PydanticAgentState]:
"""
Retrieve all agents associated with a given block with pagination support.
Args:
block_id: ID of the block to get agents for
actor: User performing the operation
include_relationships: List of relationships to include in the response
before: Cursor for pagination (get items before this ID)
after: Cursor for pagination (get items after this ID)
limit: Maximum number of items to return
ascending: Sort order (True for ascending, False for descending)
Returns:
List of agent states associated with the block
"""
async with db_registry.async_session() as session:
# Start with a basic query
query = (
select(AgentModel)
.where(AgentModel.id.in_(select(BlocksAgents.agent_id).where(BlocksAgents.block_id == block_id)))
.where(AgentModel.organization_id == actor.organization_id)
)
# Apply pagination using cursor-based approach
if after:
result = (await session.execute(select(AgentModel.created_at, AgentModel.id).where(AgentModel.id == after))).first()
if result:
after_sort_value, after_id = result
# SQLite does not support as granular timestamping, so we need to round the timestamp
if settings.database_engine is DatabaseChoice.SQLITE and isinstance(after_sort_value, datetime):
after_sort_value = after_sort_value.strftime("%Y-%m-%d %H:%M:%S")
if ascending:
query = query.where(
AgentModel.created_at > after_sort_value,
or_(AgentModel.created_at == after_sort_value, AgentModel.id > after_id),
)
else:
query = query.where(
AgentModel.created_at < after_sort_value,
or_(AgentModel.created_at == after_sort_value, AgentModel.id < after_id),
)
if before:
result = (await session.execute(select(AgentModel.created_at, AgentModel.id).where(AgentModel.id == before))).first()
if result:
before_sort_value, before_id = result
# SQLite does not support as granular timestamping, so we need to round the timestamp
if settings.database_engine is DatabaseChoice.SQLITE and isinstance(before_sort_value, datetime):
before_sort_value = before_sort_value.strftime("%Y-%m-%d %H:%M:%S")
if ascending:
query = query.where(
AgentModel.created_at < before_sort_value,
or_(AgentModel.created_at == before_sort_value, AgentModel.id < before_id),
)
else:
query = query.where(
AgentModel.created_at > before_sort_value,
or_(AgentModel.created_at == before_sort_value, AgentModel.id > before_id),
)
# Apply sorting
if ascending:
query = query.order_by(AgentModel.created_at.asc(), AgentModel.id.asc())
else:
query = query.order_by(AgentModel.created_at.desc(), AgentModel.id.desc())
# Apply limit
if limit:
query = query.limit(limit)
# Execute the query
result = await session.execute(query)
agents_orm = result.scalars().all()
agents = await asyncio.gather(*[agent.to_pydantic_async(include_relationships=include_relationships) for agent in agents_orm])
return agents
@enforce_types
@trace_method
async def size_async(self, actor: PydanticUser) -> int:
"""
Get the total count of blocks for the given user.
"""
async with db_registry.async_session() as session:
return await BlockModel.size_async(db_session=session, actor=actor)
# Block History Functions
@enforce_types
@trace_method
def checkpoint_block(
self,
block_id: str,
actor: PydanticUser,
agent_id: Optional[str] = None,
use_preloaded_block: Optional[BlockModel] = None, # For concurrency tests
) -> PydanticBlock:
"""
Create a new checkpoint for the given Block by copying its
current state into BlockHistory, using SQLAlchemy's built-in
version_id_col for concurrency checks.
- If the block was undone to an earlier checkpoint, we remove
any "future" checkpoints beyond the current state to keep a
strictly linear history.
- A single commit at the end ensures atomicity.
"""
with db_registry.session() as session:
# 1) Load the Block
if use_preloaded_block is not None:
block = session.merge(use_preloaded_block)
else:
block = BlockModel.read(db_session=session, identifier=block_id, actor=actor)
# 2) Identify the block's current checkpoint (if any)
current_entry = None
if block.current_history_entry_id:
current_entry = session.get(BlockHistory, block.current_history_entry_id)
# The current sequence, or 0 if no checkpoints exist
current_seq = current_entry.sequence_number if current_entry else 0
# 3) Truncate any future checkpoints
# If we are at seq=2, but there's a seq=3 or higher from a prior "redo chain",
# remove those, so we maintain a strictly linear undo/redo stack.
session.query(BlockHistory).filter(BlockHistory.block_id == block.id, BlockHistory.sequence_number > current_seq).delete()
# 4) Determine the next sequence number
next_seq = current_seq + 1
# 5) Create a new BlockHistory row reflecting the block's current state
history_entry = BlockHistory(
organization_id=actor.organization_id,
block_id=block.id,
sequence_number=next_seq,
description=block.description,
label=block.label,
value=block.value,
limit=block.limit,
metadata_=block.metadata_,
actor_type=ActorType.LETTA_AGENT if agent_id else ActorType.LETTA_USER,
actor_id=agent_id if agent_id else actor.id,
)
history_entry.create(session, actor=actor, no_commit=True)
# 6) Update the blocks pointer to the new checkpoint
block.current_history_entry_id = history_entry.id
# 7) Flush changes, then commit once
block = block.update(db_session=session, actor=actor, no_commit=True)
session.commit()
return block.to_pydantic()
@enforce_types
def _move_block_to_sequence(self, session: Session, block: BlockModel, target_seq: int, actor: PydanticUser) -> BlockModel:
"""
Internal helper that moves the 'block' to the specified 'target_seq' within BlockHistory.
1) Find the BlockHistory row at sequence_number=target_seq
2) Copy fields into the block
3) Update and flush (no_commit=True) - the caller is responsible for final commit
Raises:
NoResultFound: if no BlockHistory row for (block_id, target_seq)
"""
if not block.id:
raise ValueError("Block is missing an ID. Cannot move sequence.")
target_entry = (
session.query(BlockHistory)
.filter(
BlockHistory.block_id == block.id,
BlockHistory.sequence_number == target_seq,
)
.one_or_none()
)
if not target_entry:
raise NoResultFound(f"No BlockHistory row found for block_id={block.id} at sequence={target_seq}")
# Copy fields from target_entry to block
block.description = target_entry.description # type: ignore
block.label = target_entry.label # type: ignore
block.value = target_entry.value # type: ignore
block.limit = target_entry.limit # type: ignore
block.metadata_ = target_entry.metadata_ # type: ignore
block.current_history_entry_id = target_entry.id # type: ignore
# Update in DB (optimistic locking).
# We'll do a flush now; the caller does final commit.
updated_block = block.update(db_session=session, actor=actor, no_commit=True)
return updated_block
@enforce_types
@trace_method
def undo_checkpoint_block(self, block_id: str, actor: PydanticUser, use_preloaded_block: Optional[BlockModel] = None) -> PydanticBlock:
"""
Move the block to the immediately previous checkpoint in BlockHistory.
If older sequences have been pruned, we jump to the largest sequence
number that is still < current_seq.
"""
with db_registry.session() as session:
# 1) Load the current block
block = (
session.merge(use_preloaded_block)
if use_preloaded_block
else BlockModel.read(db_session=session, identifier=block_id, actor=actor)
)
if not block.current_history_entry_id:
raise ValueError(f"Block {block_id} has no history entry - cannot undo.")
current_entry = session.get(BlockHistory, block.current_history_entry_id)
if not current_entry:
raise NoResultFound(f"BlockHistory row not found for id={block.current_history_entry_id}")
current_seq = current_entry.sequence_number
# 2) Find the largest sequence < current_seq
previous_entry = (
session.query(BlockHistory)
.filter(BlockHistory.block_id == block.id, BlockHistory.sequence_number < current_seq)
.order_by(BlockHistory.sequence_number.desc())
.first()
)
if not previous_entry:
# No earlier checkpoint available
raise ValueError(f"Block {block_id} is already at the earliest checkpoint (seq={current_seq}). Cannot undo further.")
# 3) Move to that sequence
block = self._move_block_to_sequence(session, block, previous_entry.sequence_number, actor)
# 4) Commit
session.commit()
return block.to_pydantic()
@enforce_types
@trace_method
def redo_checkpoint_block(self, block_id: str, actor: PydanticUser, use_preloaded_block: Optional[BlockModel] = None) -> PydanticBlock:
"""
Move the block to the next checkpoint if it exists.
If some middle checkpoints have been pruned, we jump to the smallest
sequence > current_seq that remains.
"""
with db_registry.session() as session:
block = (
session.merge(use_preloaded_block)
if use_preloaded_block
else BlockModel.read(db_session=session, identifier=block_id, actor=actor)
)
if not block.current_history_entry_id:
raise ValueError(f"Block {block_id} has no history entry - cannot redo.")
current_entry = session.get(BlockHistory, block.current_history_entry_id)
if not current_entry:
raise NoResultFound(f"BlockHistory row not found for id={block.current_history_entry_id}")
current_seq = current_entry.sequence_number
# Find the smallest sequence that is > current_seq
next_entry = (
session.query(BlockHistory)
.filter(BlockHistory.block_id == block.id, BlockHistory.sequence_number > current_seq)
.order_by(BlockHistory.sequence_number.asc())
.first()
)
if not next_entry:
raise ValueError(f"Block {block_id} is at the highest checkpoint (seq={current_seq}). Cannot redo further.")
block = self._move_block_to_sequence(session, block, next_entry.sequence_number, actor)
session.commit()
return block.to_pydantic()
@enforce_types
@trace_method
async def bulk_update_block_values_async(
self, updates: Dict[str, str], actor: PydanticUser, return_hydrated: bool = False
) -> Optional[List[PydanticBlock]]:
"""
Bulk-update the `value` field for multiple blocks in one transaction.
Args:
updates: mapping of block_id -> new value
actor: the user performing the update (for org scoping, permissions, audit)
return_hydrated: whether to return the pydantic Block objects that were updated
Returns:
the updated Block objects as Pydantic schemas
Raises:
NoResultFound if any block_id doesn't exist or isn't visible to this actor
ValueError if any new value exceeds its block's limit
"""
async with db_registry.async_session() as session:
query = select(BlockModel).where(BlockModel.id.in_(updates.keys()), BlockModel.organization_id == actor.organization_id)
result = await session.execute(query)
blocks = result.scalars().all()
found_ids = {b.id for b in blocks}
missing = set(updates.keys()) - found_ids
if missing:
logger.warning(f"Block IDs not found or inaccessible, skipping during bulk update: {missing!r}")
for block in blocks:
new_val = updates[block.id]
if len(new_val) > block.limit:
logger.warning(f"Value length ({len(new_val)}) exceeds limit ({block.limit}) for block {block.id!r}, truncating...")
new_val = new_val[: block.limit]
block.value = new_val
await session.commit()
if return_hydrated:
# TODO: implement for async
pass
return None