chore: import upstream snapshot with attribution
CI / changes (push) Has been cancelled
CI / cd libs/checkpoint (push) Has been cancelled
CI / cd libs/checkpoint-conformance (push) Has been cancelled
CI / cd libs/checkpoint-postgres (push) Has been cancelled
CI / cd libs/checkpoint-sqlite (push) Has been cancelled
CI / cd libs/cli (push) Has been cancelled
CI / cd libs/prebuilt (push) Has been cancelled
CI / cd libs/sdk-py (push) Has been cancelled
CI / cd libs/langgraph (push) Has been cancelled
CI / Check SDK methods matching (push) Has been cancelled
CI / Check CLI schema hasn't changed #3.13 (push) Has been cancelled
CI / CLI integration test (push) Has been cancelled
CI / sdk-py integration test (push) Has been cancelled
CI / CI Success (push) Has been cancelled
baseline / benchmark (push) Has been cancelled
Deploy Redirects to GitHub Pages / deploy (push) Has been cancelled

This commit is contained in:
wehub-resource-sync
2026-07-13 12:37:18 +08:00
commit a7d6d88f6f
667 changed files with 232986 additions and 0 deletions
@@ -0,0 +1,595 @@
from __future__ import annotations
import threading
from collections import defaultdict
from collections.abc import Iterator, Mapping, Sequence
from contextlib import contextmanager
from typing import Any, cast
from langchain_core.runnables import RunnableConfig
from langgraph.checkpoint.base import (
WRITES_IDX_MAP,
ChannelVersions,
Checkpoint,
CheckpointMetadata,
CheckpointTuple,
DeltaChannelHistory,
get_checkpoint_id,
get_serializable_checkpoint_metadata,
)
from langgraph.checkpoint.serde.base import SerializerProtocol
from langgraph.checkpoint.serde.types import _DeltaSnapshot
from psycopg import Capabilities, Connection, Cursor, Pipeline
from psycopg.rows import DictRow, dict_row
from psycopg.types.json import Jsonb
from psycopg_pool import ConnectionPool
from langgraph.checkpoint.postgres import _internal
from langgraph.checkpoint.postgres.base import (
_DELTA_PAGE_SIZE,
BasePostgresSaver,
_build_delta_stage1_sql,
_build_delta_stage2_sql,
_DeltaStage2Row,
)
from langgraph.checkpoint.postgres.shallow import ShallowPostgresSaver
Conn = _internal.Conn # For backward compatibility
class PostgresSaver(BasePostgresSaver):
"""Checkpointer that stores checkpoints in a Postgres database."""
lock: threading.Lock
def __init__(
self,
conn: _internal.Conn,
pipe: Pipeline | None = None,
serde: SerializerProtocol | None = None,
) -> None:
super().__init__(serde=serde)
if isinstance(conn, ConnectionPool) and pipe is not None:
raise ValueError(
"Pipeline should be used only with a single Connection, not ConnectionPool."
)
self.conn = conn
self.pipe = pipe
self.lock = threading.Lock()
self.supports_pipeline = Capabilities().has_pipeline()
@classmethod
@contextmanager
def from_conn_string(
cls, conn_string: str, *, pipeline: bool = False
) -> Iterator[PostgresSaver]:
"""Create a new PostgresSaver instance from a connection string.
Args:
conn_string: The Postgres connection info string.
pipeline: whether to use Pipeline
Returns:
PostgresSaver: A new PostgresSaver instance.
"""
with Connection.connect(
conn_string, autocommit=True, prepare_threshold=0, row_factory=dict_row
) as conn:
if pipeline:
with conn.pipeline() as pipe:
yield cls(conn, pipe)
else:
yield cls(conn)
def setup(self) -> None:
"""Set up the checkpoint database asynchronously.
This method creates the necessary tables in the Postgres database if they don't
already exist and runs database migrations. It MUST be called directly by the user
the first time checkpointer is used.
"""
with self._cursor() as cur:
cur.execute(self.MIGRATIONS[0])
results = cur.execute(
"SELECT v FROM checkpoint_migrations ORDER BY v DESC LIMIT 1"
)
row = results.fetchone()
if row is None:
version = -1
else:
version = row["v"]
for v, migration in zip(
range(version + 1, len(self.MIGRATIONS)),
self.MIGRATIONS[version + 1 :],
strict=False,
):
cur.execute(migration)
cur.execute("INSERT INTO checkpoint_migrations (v) VALUES (%s)", (v,))
if self.pipe:
self.pipe.sync()
def list(
self,
config: RunnableConfig | None,
*,
filter: dict[str, Any] | None = None,
before: RunnableConfig | None = None,
limit: int | None = None,
) -> Iterator[CheckpointTuple]:
"""List checkpoints from the database.
This method retrieves a list of checkpoint tuples from the Postgres database based
on the provided config. The checkpoints are ordered by checkpoint ID in descending order (newest first).
Args:
config: The config to use for listing the checkpoints.
filter: Additional filtering criteria for metadata.
before: If provided, only checkpoints before the specified checkpoint ID are returned.
limit: The maximum number of checkpoints to return.
Yields:
An iterator of checkpoint tuples.
Examples:
>>> from langgraph.checkpoint.postgres import PostgresSaver
>>> DB_URI = "postgres://postgres:postgres@localhost:5432/postgres?sslmode=disable"
>>> with PostgresSaver.from_conn_string(DB_URI) as memory:
... # Run a graph, then list the checkpoints
>>> config = {"configurable": {"thread_id": "1"}}
>>> checkpoints = list(memory.list(config, limit=2))
>>> print(checkpoints)
[CheckpointTuple(...), CheckpointTuple(...)]
>>> config = {"configurable": {"thread_id": "1"}}
>>> before = {"configurable": {"checkpoint_id": "1ef4f797-8335-6428-8001-8a1503f9b875"}}
>>> with PostgresSaver.from_conn_string(DB_URI) as memory:
... # Run a graph, then list the checkpoints
>>> checkpoints = list(memory.list(config, before=before))
>>> print(checkpoints)
[CheckpointTuple(...), ...]
"""
where, args = self._search_where(config, filter, before)
query = self.SELECT_SQL + where + " ORDER BY checkpoint_id DESC"
params = list(args)
if limit is not None:
query += " LIMIT %s"
params.append(int(limit))
# if we change this to use .stream() we need to make sure to close the cursor
with self._cursor() as cur:
cur.execute(query, params)
values = cur.fetchall()
if not values:
return
# migrate pending sends if necessary
if to_migrate := [
v
for v in values
if v["checkpoint"]["v"] < 4 and v["parent_checkpoint_id"]
]:
cur.execute(
self.SELECT_PENDING_SENDS_SQL,
(
values[0]["thread_id"],
[v["parent_checkpoint_id"] for v in to_migrate],
),
)
grouped_by_parent = defaultdict(list)
for value in to_migrate:
grouped_by_parent[value["parent_checkpoint_id"]].append(value)
for sends in cur:
for value in grouped_by_parent[sends["checkpoint_id"]]:
if value["channel_values"] is None:
value["channel_values"] = []
self._migrate_pending_sends(
sends["sends"],
value["checkpoint"],
value["channel_values"],
)
for value in values:
yield self._load_checkpoint_tuple(value)
def get_tuple(self, config: RunnableConfig) -> CheckpointTuple | None:
"""Get a checkpoint tuple from the database.
This method retrieves a checkpoint tuple from the Postgres database based on the
provided config. If the config contains a `checkpoint_id` key, the checkpoint with
the matching thread ID and timestamp is retrieved. Otherwise, the latest checkpoint
for the given thread ID is retrieved.
Args:
config: The config to use for retrieving the checkpoint.
Returns:
The retrieved checkpoint tuple, or None if no matching checkpoint was found.
Examples:
Basic:
>>> config = {"configurable": {"thread_id": "1"}}
>>> checkpoint_tuple = memory.get_tuple(config)
>>> print(checkpoint_tuple)
CheckpointTuple(...)
With timestamp:
>>> config = {
... "configurable": {
... "thread_id": "1",
... "checkpoint_ns": "",
... "checkpoint_id": "1ef4f797-8335-6428-8001-8a1503f9b875",
... }
... }
>>> checkpoint_tuple = memory.get_tuple(config)
>>> print(checkpoint_tuple)
CheckpointTuple(...)
""" # noqa
thread_id = config["configurable"]["thread_id"]
checkpoint_id = get_checkpoint_id(config)
checkpoint_ns = config["configurable"].get("checkpoint_ns", "")
if checkpoint_id:
args: tuple[Any, ...] = (thread_id, checkpoint_ns, checkpoint_id)
where = "WHERE thread_id = %s AND checkpoint_ns = %s AND checkpoint_id = %s"
else:
args = (thread_id, checkpoint_ns)
where = "WHERE thread_id = %s AND checkpoint_ns = %s ORDER BY checkpoint_id DESC LIMIT 1"
with self._cursor() as cur:
cur.execute(
self.SELECT_SQL + where,
args,
)
value = cur.fetchone()
if value is None:
return None
# migrate pending sends if necessary
if value["checkpoint"]["v"] < 4 and value["parent_checkpoint_id"]:
cur.execute(
self.SELECT_PENDING_SENDS_SQL,
(thread_id, [value["parent_checkpoint_id"]]),
)
if sends := cur.fetchone():
if value["channel_values"] is None:
value["channel_values"] = []
self._migrate_pending_sends(
sends["sends"],
value["checkpoint"],
value["channel_values"],
)
return self._load_checkpoint_tuple(value)
def put(
self,
config: RunnableConfig,
checkpoint: Checkpoint,
metadata: CheckpointMetadata,
new_versions: ChannelVersions,
) -> RunnableConfig:
"""Save a checkpoint to the database.
This method saves a checkpoint to the Postgres database. The checkpoint is associated
with the provided config and its parent config (if any).
Args:
config: The config to associate with the checkpoint.
checkpoint: The checkpoint to save.
metadata: Additional metadata to save with the checkpoint.
new_versions: New channel versions as of this write.
Returns:
RunnableConfig: Updated configuration after storing the checkpoint.
Examples:
>>> from langgraph.checkpoint.postgres import PostgresSaver
>>> DB_URI = "postgres://postgres:postgres@localhost:5432/postgres?sslmode=disable"
>>> with PostgresSaver.from_conn_string(DB_URI) as memory:
>>> config = {"configurable": {"thread_id": "1", "checkpoint_ns": ""}}
>>> checkpoint = {"ts": "2024-05-04T06:32:42.235444+00:00", "id": "1ef4f797-8335-6428-8001-8a1503f9b875", "channel_values": {"key": "value"}}
>>> saved_config = memory.put(config, checkpoint, {"source": "input", "step": 1, "writes": {"key": "value"}}, {})
>>> print(saved_config)
{'configurable': {'thread_id': '1', 'checkpoint_ns': '', 'checkpoint_id': '1ef4f797-8335-6428-8001-8a1503f9b875'}}
"""
configurable = config["configurable"].copy()
thread_id = configurable.pop("thread_id")
checkpoint_ns = configurable.pop("checkpoint_ns")
checkpoint_id = configurable.pop("checkpoint_id", None)
copy = checkpoint.copy()
copy["channel_values"] = copy["channel_values"].copy()
next_config = {
"configurable": {
"thread_id": thread_id,
"checkpoint_ns": checkpoint_ns,
"checkpoint_id": checkpoint["id"],
}
}
# inline primitive values in checkpoint table
# others are stored in blobs table
blob_values = {}
for k, v in checkpoint["channel_values"].items():
if isinstance(v, _DeltaSnapshot):
blob_values[k] = copy["channel_values"].pop(k)
copy["channel_values"][k] = True
elif v is None or isinstance(v, (str, int, float, bool)):
pass
else:
blob_values[k] = copy["channel_values"].pop(k)
with self._cursor(pipeline=True) as cur:
if blob_versions := {
k: v for k, v in new_versions.items() if k in blob_values
}:
cur.executemany(
self.UPSERT_CHECKPOINT_BLOBS_SQL,
self._dump_blobs(
thread_id,
checkpoint_ns,
blob_values,
blob_versions,
),
)
cur.execute(
self.UPSERT_CHECKPOINTS_SQL,
(
thread_id,
checkpoint_ns,
checkpoint["id"],
checkpoint_id,
Jsonb(copy),
Jsonb(get_serializable_checkpoint_metadata(config, metadata)),
),
)
return next_config
def put_writes(
self,
config: RunnableConfig,
writes: Sequence[tuple[str, Any]],
task_id: str,
task_path: str = "",
) -> None:
"""Store intermediate writes linked to a checkpoint.
This method saves intermediate writes associated with a checkpoint to the Postgres database.
Args:
config: Configuration of the related checkpoint.
writes: List of writes to store.
task_id: Identifier for the task creating the writes.
"""
query = (
self.UPSERT_CHECKPOINT_WRITES_SQL
if all(w[0] in WRITES_IDX_MAP for w in writes)
else self.INSERT_CHECKPOINT_WRITES_SQL
)
with self._cursor(pipeline=True) as cur:
cur.executemany(
query,
self._dump_writes(
config["configurable"]["thread_id"],
config["configurable"]["checkpoint_ns"],
config["configurable"]["checkpoint_id"],
task_id,
task_path,
writes,
),
)
def delete_thread(self, thread_id: str) -> None:
"""Delete all checkpoints and writes associated with a thread ID.
Args:
thread_id: The thread ID to delete.
Returns:
None
"""
with self._cursor(pipeline=True) as cur:
cur.execute(
"DELETE FROM checkpoints WHERE thread_id = %s",
(str(thread_id),),
)
cur.execute(
"DELETE FROM checkpoint_blobs WHERE thread_id = %s",
(str(thread_id),),
)
cur.execute(
"DELETE FROM checkpoint_writes WHERE thread_id = %s",
(str(thread_id),),
)
@contextmanager
def _cursor(self, *, pipeline: bool = False) -> Iterator[Cursor[DictRow]]:
"""Create a database cursor as a context manager.
Args:
pipeline: whether to use pipeline for the DB operations inside the context manager.
Will be applied regardless of whether the PostgresSaver instance was initialized with a pipeline.
If pipeline mode is not supported, will fall back to using transaction context manager.
"""
with self.lock, _internal.get_connection(self.conn) as conn:
if self.pipe:
# a connection in pipeline mode can be used concurrently
# in multiple threads/coroutines, but only one cursor can be
# used at a time
try:
with conn.cursor(binary=True, row_factory=dict_row) as cur:
yield cur
finally:
if pipeline:
self.pipe.sync()
elif pipeline:
# a connection not in pipeline mode can only be used by one
# thread/coroutine at a time, so we acquire a lock
if self.supports_pipeline:
with (
conn.pipeline(),
conn.cursor(binary=True, row_factory=dict_row) as cur,
):
yield cur
else:
# Use connection's transaction context manager when pipeline mode not supported
with (
conn.transaction(),
conn.cursor(binary=True, row_factory=dict_row) as cur,
):
yield cur
else:
with conn.cursor(binary=True, row_factory=dict_row) as cur:
yield cur
def get_delta_channel_history(
self, *, config: RunnableConfig, channels: Sequence[str]
) -> Mapping[str, DeltaChannelHistory]:
"""Fast-path override of `BaseCheckpointSaver.get_delta_channel_history`.
Two-stage query, both stages cover ALL requested channels:
* Stage 1 (paged): dynamic SELECT over `checkpoints` with K parallel
JSONB key lookups (one column pair per channel) — no subquery, no
aggregation. Pages newest-first by `checkpoint_id` with a cursor;
page size is `_DELTA_PAGE_SIZE`. Stops paging when every channel
has found its seed or the chain is exhausted.
* Stage 2 (per-channel UNION ALL): one branch per channel reading
`checkpoint_writes` filtered to that channel's specific
`chain_cids`, plus one branch per channel that has a seed reading
`checkpoint_blobs` for that channel + version. Avoids the
over-fetch of a single `channel = ANY(channels)` filter when
channels have different chain depths.
"""
if not channels:
return {}
channels = list(channels)
thread_id = config["configurable"]["thread_id"]
checkpoint_ns = config["configurable"].get("checkpoint_ns", "")
checkpoint_id = get_checkpoint_id(config)
if checkpoint_id is None:
target = self.get_tuple(config)
if target is None:
return {ch: {"writes": []} for ch in channels}
checkpoint_id = target.config["configurable"]["checkpoint_id"]
# Stage 1: paged K-JSONB-lookup scan, walking the parent chain in
# Python after each page. Stops as soon as every channel has its seed.
stage1_sql = _build_delta_stage1_sql(channels, paged=True)
parent_of: dict[str, str | None] = {}
ver_by_i_by_cid: list[dict[str, str | None]] = [{} for _ in channels]
hs_by_i_by_cid: list[dict[str, bool]] = [{} for _ in channels]
chain_by_ch: dict[str, list[str]] = {ch: [] for ch in channels}
seed_ver_by_ch: dict[str, str | None] = {ch: None for ch in channels}
walk_cursor_by_ch: dict[str, str | None] = {}
seeded: set[str] = set()
cursor: str | None = None
with self._cursor() as cur:
while True:
stage1_params: list[Any] = []
for ch in channels:
stage1_params.extend([ch, ch])
stage1_params.extend(
[thread_id, checkpoint_ns, cursor, cursor, _DELTA_PAGE_SIZE]
)
cur.execute(stage1_sql, stage1_params)
page = cur.fetchall()
if not page:
break
oldest = self._ingest_stage1_page(
cast("list[Mapping[str, Any]]", page),
channels,
parent_of,
ver_by_i_by_cid,
hs_by_i_by_cid,
)
self._try_advance_walks(
checkpoint_id,
channels,
parent_of,
ver_by_i_by_cid,
hs_by_i_by_cid,
chain_by_ch,
seed_ver_by_ch,
walk_cursor_by_ch,
seeded,
)
# Stop if every channel is seeded, or the page was short
# (chain exhausted — no more rows to fetch).
if len(seeded) == len(channels) or len(page) < _DELTA_PAGE_SIZE:
break
cursor = oldest
# Stage 2: per-channel UNION ALL — one writes branch per channel
# with non-empty chain, plus one blob branch per seeded channel.
channels_with_chain = [ch for ch in channels if chain_by_ch[ch]]
channels_with_seed = [ch for ch in channels if seed_ver_by_ch[ch] is not None]
stage2_sql = _build_delta_stage2_sql(
channels_with_chain=channels_with_chain,
channels_with_seed=channels_with_seed,
)
if stage2_sql:
stage2_params: list[Any] = []
for ch in channels_with_chain:
stage2_params.extend([thread_id, checkpoint_ns, ch, chain_by_ch[ch]])
for ch in channels_with_seed:
stage2_params.extend([thread_id, checkpoint_ns, ch, seed_ver_by_ch[ch]])
with self._cursor() as cur:
cur.execute(stage2_sql, stage2_params)
stage2_rows = cur.fetchall()
else:
stage2_rows = []
return self._build_delta_channels_writes_history(
channels=channels,
chain_by_ch=chain_by_ch,
seed_ver_by_ch=seed_ver_by_ch,
stage2_rows=cast("list[_DeltaStage2Row]", stage2_rows),
)
def _load_checkpoint_tuple(self, value: DictRow) -> CheckpointTuple:
"""
Convert a database row into a CheckpointTuple object.
Args:
value: A row from the database containing checkpoint data.
Returns:
CheckpointTuple: A structured representation of the checkpoint,
including its configuration, metadata, parent checkpoint (if any),
and pending writes.
"""
return CheckpointTuple(
{
"configurable": {
"thread_id": value["thread_id"],
"checkpoint_ns": value["checkpoint_ns"],
"checkpoint_id": value["checkpoint_id"],
}
},
{
**value["checkpoint"],
"channel_values": {
**(value["checkpoint"].get("channel_values") or {}),
**self._load_blobs(value["channel_values"]),
},
},
value["metadata"],
(
{
"configurable": {
"thread_id": value["thread_id"],
"checkpoint_ns": value["checkpoint_ns"],
"checkpoint_id": value["parent_checkpoint_id"],
}
}
if value["parent_checkpoint_id"]
else None
),
self._load_writes(value["pending_writes"]),
)
__all__ = ["PostgresSaver", "BasePostgresSaver", "ShallowPostgresSaver", "Conn"]
@@ -0,0 +1,23 @@
"""Shared async utility functions for the Postgres checkpoint & storage classes."""
from collections.abc import AsyncIterator
from contextlib import asynccontextmanager
from psycopg import AsyncConnection
from psycopg.rows import DictRow
from psycopg_pool import AsyncConnectionPool
Conn = AsyncConnection[DictRow] | AsyncConnectionPool[AsyncConnection[DictRow]]
@asynccontextmanager
async def get_connection(
conn: Conn,
) -> AsyncIterator[AsyncConnection[DictRow]]:
if isinstance(conn, AsyncConnection):
yield conn
elif isinstance(conn, AsyncConnectionPool):
async with conn.connection() as conn:
yield conn
else:
raise TypeError(f"Invalid connection type: {type(conn)}")
@@ -0,0 +1,21 @@
"""Shared utility functions for the Postgres checkpoint & storage classes."""
from collections.abc import Iterator
from contextlib import contextmanager
from psycopg import Connection
from psycopg.rows import DictRow
from psycopg_pool import ConnectionPool
Conn = Connection[DictRow] | ConnectionPool[Connection[DictRow]]
@contextmanager
def get_connection(conn: Conn) -> Iterator[Connection[DictRow]]:
if isinstance(conn, Connection):
yield conn
elif isinstance(conn, ConnectionPool):
with conn.connection() as conn:
yield conn
else:
raise TypeError(f"Invalid connection type: {type(conn)}")
@@ -0,0 +1,684 @@
from __future__ import annotations
import asyncio
from collections import defaultdict
from collections.abc import AsyncIterator, Iterator, Mapping, Sequence
from contextlib import asynccontextmanager
from typing import Any, cast
from langchain_core.runnables import RunnableConfig
from langgraph.checkpoint.base import (
WRITES_IDX_MAP,
ChannelVersions,
Checkpoint,
CheckpointMetadata,
CheckpointTuple,
DeltaChannelHistory,
get_checkpoint_id,
get_serializable_checkpoint_metadata,
)
from langgraph.checkpoint.serde.base import SerializerProtocol
from langgraph.checkpoint.serde.types import _DeltaSnapshot
from psycopg import AsyncConnection, AsyncCursor, AsyncPipeline, Capabilities
from psycopg.rows import DictRow, dict_row
from psycopg.types.json import Jsonb
from psycopg_pool import AsyncConnectionPool
from langgraph.checkpoint.postgres import _ainternal
from langgraph.checkpoint.postgres.base import (
_DELTA_PAGE_SIZE,
BasePostgresSaver,
_build_delta_stage1_sql,
_build_delta_stage2_sql,
_DeltaStage2Row,
)
from langgraph.checkpoint.postgres.shallow import AsyncShallowPostgresSaver
Conn = _ainternal.Conn # For backward compatibility
class AsyncPostgresSaver(BasePostgresSaver):
"""Asynchronous checkpointer that stores checkpoints in a Postgres database."""
lock: asyncio.Lock
def __init__(
self,
conn: _ainternal.Conn,
pipe: AsyncPipeline | None = None,
serde: SerializerProtocol | None = None,
) -> None:
super().__init__(serde=serde)
if isinstance(conn, AsyncConnectionPool) and pipe is not None:
raise ValueError(
"Pipeline should be used only with a single AsyncConnection, not AsyncConnectionPool."
)
self.conn = conn
self.pipe = pipe
self.lock = asyncio.Lock()
self.loop = asyncio.get_running_loop()
self.supports_pipeline = Capabilities().has_pipeline()
@classmethod
@asynccontextmanager
async def from_conn_string(
cls,
conn_string: str,
*,
pipeline: bool = False,
serde: SerializerProtocol | None = None,
) -> AsyncIterator[AsyncPostgresSaver]:
"""Create a new AsyncPostgresSaver instance from a connection string.
Args:
conn_string: The Postgres connection info string.
pipeline: whether to use AsyncPipeline
Returns:
AsyncPostgresSaver: A new AsyncPostgresSaver instance.
"""
async with await AsyncConnection.connect(
conn_string, autocommit=True, prepare_threshold=0, row_factory=dict_row
) as conn:
if pipeline:
async with conn.pipeline() as pipe:
yield cls(conn=conn, pipe=pipe, serde=serde)
else:
yield cls(conn=conn, serde=serde)
async def setup(self) -> None:
"""Set up the checkpoint database asynchronously.
This method creates the necessary tables in the Postgres database if they don't
already exist and runs database migrations. It MUST be called directly by the user
the first time checkpointer is used.
"""
async with self._cursor() as cur:
await cur.execute(self.MIGRATIONS[0])
results = await cur.execute(
"SELECT v FROM checkpoint_migrations ORDER BY v DESC LIMIT 1"
)
row = await results.fetchone()
if row is None:
version = -1
else:
version = row["v"]
for v, migration in zip(
range(version + 1, len(self.MIGRATIONS)),
self.MIGRATIONS[version + 1 :],
strict=False,
):
await cur.execute(migration)
await cur.execute(
"INSERT INTO checkpoint_migrations (v) VALUES (%s)", (v,)
)
if self.pipe:
await self.pipe.sync()
async def alist(
self,
config: RunnableConfig | None,
*,
filter: dict[str, Any] | None = None,
before: RunnableConfig | None = None,
limit: int | None = None,
) -> AsyncIterator[CheckpointTuple]:
"""List checkpoints from the database asynchronously.
This method retrieves a list of checkpoint tuples from the Postgres database based
on the provided config. The checkpoints are ordered by checkpoint ID in descending order (newest first).
Args:
config: Base configuration for filtering checkpoints.
filter: Additional filtering criteria for metadata.
before: If provided, only checkpoints before the specified checkpoint ID are returned.
limit: Maximum number of checkpoints to return.
Yields:
An asynchronous iterator of matching checkpoint tuples.
"""
where, args = self._search_where(config, filter, before)
query = self.SELECT_SQL + where + " ORDER BY checkpoint_id DESC"
params = list(args)
if limit is not None:
query += " LIMIT %s"
params.append(int(limit))
# if we change this to use .stream() we need to make sure to close the cursor
async with self._cursor() as cur:
await cur.execute(query, params, binary=True)
values = await cur.fetchall()
if not values:
return
# migrate pending sends if necessary
if to_migrate := [
v
for v in values
if v["checkpoint"]["v"] < 4 and v["parent_checkpoint_id"]
]:
await cur.execute(
self.SELECT_PENDING_SENDS_SQL,
(
values[0]["thread_id"],
[v["parent_checkpoint_id"] for v in to_migrate],
),
)
grouped_by_parent = defaultdict(list)
for value in to_migrate:
grouped_by_parent[value["parent_checkpoint_id"]].append(value)
async for sends in cur:
for value in grouped_by_parent[sends["checkpoint_id"]]:
if value["channel_values"] is None:
value["channel_values"] = []
self._migrate_pending_sends(
sends["sends"],
value["checkpoint"],
value["channel_values"],
)
for value in values:
yield await self._load_checkpoint_tuple(value)
async def aget_tuple(self, config: RunnableConfig) -> CheckpointTuple | None:
"""Get a checkpoint tuple from the database asynchronously.
This method retrieves a checkpoint tuple from the Postgres database based on the
provided config. If the config contains a `checkpoint_id` key, the checkpoint with
the matching thread ID and "checkpoint_id" is retrieved. Otherwise, the latest checkpoint
for the given thread ID is retrieved.
Args:
config: The config to use for retrieving the checkpoint.
Returns:
The retrieved checkpoint tuple, or None if no matching checkpoint was found.
"""
thread_id = config["configurable"]["thread_id"]
checkpoint_id = get_checkpoint_id(config)
checkpoint_ns = config["configurable"].get("checkpoint_ns", "")
if checkpoint_id:
args: tuple[Any, ...] = (thread_id, checkpoint_ns, checkpoint_id)
where = "WHERE thread_id = %s AND checkpoint_ns = %s AND checkpoint_id = %s"
else:
args = (thread_id, checkpoint_ns)
where = "WHERE thread_id = %s AND checkpoint_ns = %s ORDER BY checkpoint_id DESC LIMIT 1"
async with self._cursor() as cur:
await cur.execute(
self.SELECT_SQL + where,
args,
binary=True,
)
value = await cur.fetchone()
if value is None:
return None
# migrate pending sends if necessary
if value["checkpoint"]["v"] < 4 and value["parent_checkpoint_id"]:
await cur.execute(
self.SELECT_PENDING_SENDS_SQL,
(thread_id, [value["parent_checkpoint_id"]]),
)
if sends := await cur.fetchone():
if value["channel_values"] is None:
value["channel_values"] = []
self._migrate_pending_sends(
sends["sends"],
value["checkpoint"],
value["channel_values"],
)
return await self._load_checkpoint_tuple(value)
async def aput(
self,
config: RunnableConfig,
checkpoint: Checkpoint,
metadata: CheckpointMetadata,
new_versions: ChannelVersions,
) -> RunnableConfig:
"""Save a checkpoint to the database asynchronously.
This method saves a checkpoint to the Postgres database. The checkpoint is associated
with the provided config and its parent config (if any).
Args:
config: The config to associate with the checkpoint.
checkpoint: The checkpoint to save.
metadata: Additional metadata to save with the checkpoint.
new_versions: New channel versions as of this write.
Returns:
RunnableConfig: Updated configuration after storing the checkpoint.
"""
configurable = config["configurable"].copy()
thread_id = configurable.pop("thread_id")
checkpoint_ns = configurable.pop("checkpoint_ns")
checkpoint_id = configurable.pop("checkpoint_id", None)
copy = checkpoint.copy()
copy["channel_values"] = copy["channel_values"].copy()
next_config = {
"configurable": {
"thread_id": thread_id,
"checkpoint_ns": checkpoint_ns,
"checkpoint_id": checkpoint["id"],
}
}
# inline primitive values in checkpoint table
# others are stored in blobs table
blob_values = {}
for k, v in checkpoint["channel_values"].items():
if isinstance(v, _DeltaSnapshot):
blob_values[k] = copy["channel_values"].pop(k)
copy["channel_values"][k] = True
elif v is None or isinstance(v, (str, int, float, bool)):
pass
else:
blob_values[k] = copy["channel_values"].pop(k)
async with self._cursor(pipeline=True) as cur:
if blob_versions := {
k: v for k, v in new_versions.items() if k in blob_values
}:
await cur.executemany(
self.UPSERT_CHECKPOINT_BLOBS_SQL,
await asyncio.to_thread(
self._dump_blobs,
thread_id,
checkpoint_ns,
blob_values,
blob_versions,
),
)
await cur.execute(
self.UPSERT_CHECKPOINTS_SQL,
(
thread_id,
checkpoint_ns,
checkpoint["id"],
checkpoint_id,
Jsonb(copy),
Jsonb(get_serializable_checkpoint_metadata(config, metadata)),
),
)
return next_config
async def aput_writes(
self,
config: RunnableConfig,
writes: Sequence[tuple[str, Any]],
task_id: str,
task_path: str = "",
) -> None:
"""Store intermediate writes linked to a checkpoint asynchronously.
This method saves intermediate writes associated with a checkpoint to the database.
Args:
config: Configuration of the related checkpoint.
writes: List of writes to store, each as (channel, value) pair.
task_id: Identifier for the task creating the writes.
"""
query = (
self.UPSERT_CHECKPOINT_WRITES_SQL
if all(w[0] in WRITES_IDX_MAP for w in writes)
else self.INSERT_CHECKPOINT_WRITES_SQL
)
params = await asyncio.to_thread(
self._dump_writes,
config["configurable"]["thread_id"],
config["configurable"]["checkpoint_ns"],
config["configurable"]["checkpoint_id"],
task_id,
task_path,
writes,
)
async with self._cursor(pipeline=True) as cur:
await cur.executemany(query, params)
async def adelete_thread(self, thread_id: str) -> None:
"""Delete all checkpoints and writes associated with a thread ID.
Args:
thread_id: The thread ID to delete.
Returns:
None
"""
async with self._cursor(pipeline=True) as cur:
await cur.execute(
"DELETE FROM checkpoints WHERE thread_id = %s",
(str(thread_id),),
)
await cur.execute(
"DELETE FROM checkpoint_blobs WHERE thread_id = %s",
(str(thread_id),),
)
await cur.execute(
"DELETE FROM checkpoint_writes WHERE thread_id = %s",
(str(thread_id),),
)
@asynccontextmanager
async def _cursor(
self, *, pipeline: bool = False
) -> AsyncIterator[AsyncCursor[DictRow]]:
"""Create a database cursor as a context manager.
Args:
pipeline: whether to use pipeline for the DB operations inside the context manager.
Will be applied regardless of whether the AsyncPostgresSaver instance was initialized with a pipeline.
If pipeline mode is not supported, will fall back to using transaction context manager.
"""
async with self.lock, _ainternal.get_connection(self.conn) as conn:
if self.pipe:
# a connection in pipeline mode can be used concurrently
# in multiple threads/coroutines, but only one cursor can be
# used at a time
try:
async with conn.cursor(binary=True, row_factory=dict_row) as cur:
yield cur
finally:
if pipeline:
await self.pipe.sync()
elif pipeline:
# a connection not in pipeline mode can only be used by one
# thread/coroutine at a time, so we acquire a lock
if self.supports_pipeline:
async with (
conn.pipeline(),
conn.cursor(binary=True, row_factory=dict_row) as cur,
):
yield cur
else:
# Use connection's transaction context manager when pipeline mode not supported
async with (
conn.transaction(),
conn.cursor(binary=True, row_factory=dict_row) as cur,
):
yield cur
else:
async with conn.cursor(binary=True, row_factory=dict_row) as cur:
yield cur
async def aget_delta_channel_history(
self, *, config: RunnableConfig, channels: Sequence[str]
) -> Mapping[str, DeltaChannelHistory]:
"""Fast-path override of `BaseCheckpointSaver.aget_delta_channel_history`.
See `PostgresSaver.get_delta_channel_history` for design notes; this is
the async equivalent with internal stage-1 paging and per-channel
UNION ALL stage-2.
"""
if not channels:
return {}
channels = list(channels)
thread_id = config["configurable"]["thread_id"]
checkpoint_ns = config["configurable"].get("checkpoint_ns", "")
checkpoint_id = get_checkpoint_id(config)
if checkpoint_id is None:
target = await self.aget_tuple(config)
if target is None:
return {ch: {"writes": []} for ch in channels}
checkpoint_id = target.config["configurable"]["checkpoint_id"]
stage1_sql = _build_delta_stage1_sql(channels, paged=True)
parent_of: dict[str, str | None] = {}
ver_by_i_by_cid: list[dict[str, str | None]] = [{} for _ in channels]
hs_by_i_by_cid: list[dict[str, bool]] = [{} for _ in channels]
chain_by_ch: dict[str, list[str]] = {ch: [] for ch in channels}
seed_ver_by_ch: dict[str, str | None] = {ch: None for ch in channels}
walk_cursor_by_ch: dict[str, str | None] = {}
seeded: set[str] = set()
cursor: str | None = None
async with self._cursor() as cur:
while True:
stage1_params: list[Any] = []
for ch in channels:
stage1_params.extend([ch, ch])
stage1_params.extend(
[thread_id, checkpoint_ns, cursor, cursor, _DELTA_PAGE_SIZE]
)
await cur.execute(stage1_sql, stage1_params)
page = await cur.fetchall()
if not page:
break
oldest = self._ingest_stage1_page(
cast("list[Mapping[str, Any]]", page),
channels,
parent_of,
ver_by_i_by_cid,
hs_by_i_by_cid,
)
self._try_advance_walks(
checkpoint_id,
channels,
parent_of,
ver_by_i_by_cid,
hs_by_i_by_cid,
chain_by_ch,
seed_ver_by_ch,
walk_cursor_by_ch,
seeded,
)
if len(seeded) == len(channels) or len(page) < _DELTA_PAGE_SIZE:
break
cursor = oldest
channels_with_chain = [ch for ch in channels if chain_by_ch[ch]]
channels_with_seed = [ch for ch in channels if seed_ver_by_ch[ch] is not None]
stage2_sql = _build_delta_stage2_sql(
channels_with_chain=channels_with_chain,
channels_with_seed=channels_with_seed,
)
if stage2_sql:
stage2_params: list[Any] = []
for ch in channels_with_chain:
stage2_params.extend([thread_id, checkpoint_ns, ch, chain_by_ch[ch]])
for ch in channels_with_seed:
stage2_params.extend([thread_id, checkpoint_ns, ch, seed_ver_by_ch[ch]])
async with self._cursor() as cur:
await cur.execute(stage2_sql, stage2_params)
stage2_rows = await cur.fetchall()
else:
stage2_rows = []
return self._build_delta_channels_writes_history(
channels=channels,
chain_by_ch=chain_by_ch,
seed_ver_by_ch=seed_ver_by_ch,
stage2_rows=cast("list[_DeltaStage2Row]", stage2_rows),
)
async def _load_checkpoint_tuple(self, value: DictRow) -> CheckpointTuple:
"""
Convert a database row into a CheckpointTuple object.
Args:
value: A row from the database containing checkpoint data.
Returns:
CheckpointTuple: A structured representation of the checkpoint,
including its configuration, metadata, parent checkpoint (if any),
and pending writes.
"""
return CheckpointTuple(
{
"configurable": {
"thread_id": value["thread_id"],
"checkpoint_ns": value["checkpoint_ns"],
"checkpoint_id": value["checkpoint_id"],
}
},
{
**value["checkpoint"],
"channel_values": {
**(value["checkpoint"].get("channel_values") or {}),
**self._load_blobs(value["channel_values"]),
},
},
value["metadata"],
(
{
"configurable": {
"thread_id": value["thread_id"],
"checkpoint_ns": value["checkpoint_ns"],
"checkpoint_id": value["parent_checkpoint_id"],
}
}
if value["parent_checkpoint_id"]
else None
),
await asyncio.to_thread(self._load_writes, value["pending_writes"]),
)
def list(
self,
config: RunnableConfig | None,
*,
filter: dict[str, Any] | None = None,
before: RunnableConfig | None = None,
limit: int | None = None,
) -> Iterator[CheckpointTuple]:
"""List checkpoints from the database.
This method retrieves a list of checkpoint tuples from the Postgres database based
on the provided config. The checkpoints are ordered by checkpoint ID in descending order (newest first).
Args:
config: Base configuration for filtering checkpoints.
filter: Additional filtering criteria for metadata.
before: If provided, only checkpoints before the specified checkpoint ID are returned.
limit: Maximum number of checkpoints to return.
Yields:
An iterator of matching checkpoint tuples.
"""
try:
# check if we are in the main thread, only bg threads can block
# we don't check in other methods to avoid the overhead
if asyncio.get_running_loop() is self.loop:
raise asyncio.InvalidStateError(
"Synchronous calls to AsyncPostgresSaver are only allowed from a "
"different thread. From the main thread, use the async interface. "
"For example, use `checkpointer.alist(...)` or `await "
"graph.ainvoke(...)`."
)
except RuntimeError:
pass
aiter_ = self.alist(config, filter=filter, before=before, limit=limit)
while True:
try:
yield asyncio.run_coroutine_threadsafe(
anext(aiter_), # type: ignore[arg-type] # noqa: F821
self.loop,
).result()
except StopAsyncIteration:
break
def get_tuple(self, config: RunnableConfig) -> CheckpointTuple | None:
"""Get a checkpoint tuple from the database.
This method retrieves a checkpoint tuple from the Postgres database based on the
provided config. If the config contains a `checkpoint_id` key, the checkpoint with
the matching thread ID and "checkpoint_id" is retrieved. Otherwise, the latest checkpoint
for the given thread ID is retrieved.
Args:
config: The config to use for retrieving the checkpoint.
Returns:
The retrieved checkpoint tuple, or None if no matching checkpoint was found.
"""
try:
# check if we are in the main thread, only bg threads can block
# we don't check in other methods to avoid the overhead
if asyncio.get_running_loop() is self.loop:
raise asyncio.InvalidStateError(
"Synchronous calls to AsyncPostgresSaver are only allowed from a "
"different thread. From the main thread, use the async interface. "
"For example, use `await checkpointer.aget_tuple(...)` or `await "
"graph.ainvoke(...)`."
)
except RuntimeError:
pass
return asyncio.run_coroutine_threadsafe(
self.aget_tuple(config), self.loop
).result()
def put(
self,
config: RunnableConfig,
checkpoint: Checkpoint,
metadata: CheckpointMetadata,
new_versions: ChannelVersions,
) -> RunnableConfig:
"""Save a checkpoint to the database.
This method saves a checkpoint to the Postgres database. The checkpoint is associated
with the provided config and its parent config (if any).
Args:
config: The config to associate with the checkpoint.
checkpoint: The checkpoint to save.
metadata: Additional metadata to save with the checkpoint.
new_versions: New channel versions as of this write.
Returns:
RunnableConfig: Updated configuration after storing the checkpoint.
"""
return asyncio.run_coroutine_threadsafe(
self.aput(config, checkpoint, metadata, new_versions), self.loop
).result()
def put_writes(
self,
config: RunnableConfig,
writes: Sequence[tuple[str, Any]],
task_id: str,
task_path: str = "",
) -> None:
"""Store intermediate writes linked to a checkpoint.
This method saves intermediate writes associated with a checkpoint to the database.
Args:
config: Configuration of the related checkpoint.
writes: List of writes to store, each as (channel, value) pair.
task_id: Identifier for the task creating the writes.
task_path: Path of the task creating the writes.
"""
return asyncio.run_coroutine_threadsafe(
self.aput_writes(config, writes, task_id, task_path), self.loop
).result()
def delete_thread(self, thread_id: str) -> None:
"""Delete all checkpoints and writes associated with a thread ID.
Args:
thread_id: The thread ID to delete.
Returns:
None
"""
try:
# check if we are in the main thread, only bg threads can block
# we don't check in other methods to avoid the overhead
if asyncio.get_running_loop() is self.loop:
raise asyncio.InvalidStateError(
"Synchronous calls to AsyncPostgresSaver are only allowed from a "
"different thread. From the main thread, use the async interface. "
"For example, use `await checkpointer.aget_tuple(...)` or `await "
"graph.ainvoke(...)`."
)
except RuntimeError:
pass
return asyncio.run_coroutine_threadsafe(
self.adelete_thread(thread_id), self.loop
).result()
__all__ = ["AsyncPostgresSaver", "AsyncShallowPostgresSaver", "Conn"]
@@ -0,0 +1,596 @@
from __future__ import annotations
import random
import warnings
from collections.abc import Mapping, Sequence
from importlib.metadata import version as get_version
from typing import Any, TypedDict, cast
from langchain_core.runnables import RunnableConfig
from langgraph.checkpoint.base import (
WRITES_IDX_MAP,
BaseCheckpointSaver,
ChannelVersions,
DeltaChannelHistory,
PendingWrite,
get_checkpoint_id,
)
from langgraph.checkpoint.serde.types import TASKS
from psycopg.types.json import Jsonb
# Page size for stage-1 paged scan in `get_delta_channel_history`. Internal
# constant — exposing this as a kwarg is left as a follow-up.
_DELTA_PAGE_SIZE = 1024
MetadataInput = dict[str, Any] | None
try:
major, minor = get_version("langgraph").split(".")[:2]
if int(major) == 0 and int(minor) < 5:
warnings.warn(
"You're using incompatible versions of langgraph and checkpoint-postgres. Please upgrade langgraph to avoid unexpected behavior.",
DeprecationWarning,
stacklevel=2,
)
except Exception:
# skip version check if running from source
pass
"""
To add a new migration, add a new string to the MIGRATIONS list.
The position of the migration in the list is the version number.
"""
MIGRATIONS = [
"""CREATE TABLE IF NOT EXISTS checkpoint_migrations (
v INTEGER PRIMARY KEY
);""",
"""CREATE TABLE IF NOT EXISTS checkpoints (
thread_id TEXT NOT NULL,
checkpoint_ns TEXT NOT NULL DEFAULT '',
checkpoint_id TEXT NOT NULL,
parent_checkpoint_id TEXT,
type TEXT,
checkpoint JSONB NOT NULL,
metadata JSONB NOT NULL DEFAULT '{}',
PRIMARY KEY (thread_id, checkpoint_ns, checkpoint_id)
);""",
"""CREATE TABLE IF NOT EXISTS checkpoint_blobs (
thread_id TEXT NOT NULL,
checkpoint_ns TEXT NOT NULL DEFAULT '',
channel TEXT NOT NULL,
version TEXT NOT NULL,
type TEXT NOT NULL,
blob BYTEA,
PRIMARY KEY (thread_id, checkpoint_ns, channel, version)
);""",
"""CREATE TABLE IF NOT EXISTS checkpoint_writes (
thread_id TEXT NOT NULL,
checkpoint_ns TEXT NOT NULL DEFAULT '',
checkpoint_id TEXT NOT NULL,
task_id TEXT NOT NULL,
idx INTEGER NOT NULL,
channel TEXT NOT NULL,
type TEXT,
blob BYTEA NOT NULL,
PRIMARY KEY (thread_id, checkpoint_ns, checkpoint_id, task_id, idx)
);""",
"ALTER TABLE checkpoint_blobs ALTER COLUMN blob DROP not null;",
# NOTE: this is a no-op migration to ensure that the versions in the migrations table are correct.
# This is necessary due to an empty migration previously added to the list.
"SELECT 1;",
"""
CREATE INDEX CONCURRENTLY IF NOT EXISTS checkpoints_thread_id_idx ON checkpoints(thread_id);
""",
"""
CREATE INDEX CONCURRENTLY IF NOT EXISTS checkpoint_blobs_thread_id_idx ON checkpoint_blobs(thread_id);
""",
"""
CREATE INDEX CONCURRENTLY IF NOT EXISTS checkpoint_writes_thread_id_idx ON checkpoint_writes(thread_id);
""",
"""ALTER TABLE checkpoint_writes ADD COLUMN IF NOT EXISTS task_path TEXT NOT NULL DEFAULT '';""",
]
SELECT_SQL = """
select
thread_id,
checkpoint,
checkpoint_ns,
checkpoint_id,
parent_checkpoint_id,
metadata,
(
select array_agg(array[bl.channel::bytea, bl.type::bytea, bl.blob])
from jsonb_each_text(checkpoint -> 'channel_versions')
inner join checkpoint_blobs bl
on bl.thread_id = checkpoints.thread_id
and bl.checkpoint_ns = checkpoints.checkpoint_ns
and bl.channel = jsonb_each_text.key
and bl.version = jsonb_each_text.value
) as channel_values,
(
select
array_agg(array[cw.task_id::text::bytea, cw.channel::bytea, cw.type::bytea, cw.blob] order by cw.task_id, cw.idx)
from checkpoint_writes cw
where cw.thread_id = checkpoints.thread_id
and cw.checkpoint_ns = checkpoints.checkpoint_ns
and cw.checkpoint_id = checkpoints.checkpoint_id
) as pending_writes
from checkpoints """
SELECT_PENDING_SENDS_SQL = f"""
select
checkpoint_id,
array_agg(array[type::bytea, blob] order by task_path, task_id, idx) as sends
from checkpoint_writes
where thread_id = %s
and checkpoint_id = any(%s)
and channel = '{TASKS}'
group by checkpoint_id
"""
UPSERT_CHECKPOINT_BLOBS_SQL = """
INSERT INTO checkpoint_blobs (thread_id, checkpoint_ns, channel, version, type, blob)
VALUES (%s, %s, %s, %s, %s, %s)
ON CONFLICT (thread_id, checkpoint_ns, channel, version) DO NOTHING
"""
UPSERT_CHECKPOINTS_SQL = """
INSERT INTO checkpoints (thread_id, checkpoint_ns, checkpoint_id, parent_checkpoint_id, checkpoint, metadata)
VALUES (%s, %s, %s, %s, %s, %s)
ON CONFLICT (thread_id, checkpoint_ns, checkpoint_id)
DO UPDATE SET
checkpoint = EXCLUDED.checkpoint,
metadata = EXCLUDED.metadata;
"""
UPSERT_CHECKPOINT_WRITES_SQL = """
INSERT INTO checkpoint_writes (thread_id, checkpoint_ns, checkpoint_id, task_id, task_path, idx, channel, type, blob)
VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s)
ON CONFLICT (thread_id, checkpoint_ns, checkpoint_id, task_id, idx) DO UPDATE SET
channel = EXCLUDED.channel,
type = EXCLUDED.type,
blob = EXCLUDED.blob;
"""
INSERT_CHECKPOINT_WRITES_SQL = """
INSERT INTO checkpoint_writes (thread_id, checkpoint_ns, checkpoint_id, task_id, task_path, idx, channel, type, blob)
VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s)
ON CONFLICT (thread_id, checkpoint_ns, checkpoint_id, task_id, idx) DO NOTHING
"""
class _DeltaStage2Row(TypedDict, total=False):
"""One row from `_build_delta_stage2_sql` (a UNION ALL of writes and blobs)."""
_kind: str # "w" or "b"
checkpoint_id: str | None # "w" rows only
channel: str | None # set on both "w" and "b" rows
type: str | None
blob: bytes | None
task_id: str | None # "w" rows only
idx: int | None # "w" rows only
version: str | None # "b" rows only
# Multi-channel two-stage DeltaChannel reconstruction.
#
# Stage 1 scans checkpoint metadata (no blob bytes) and emits one row per
# checkpoint with K parallel JSONB key lookups (one column pair per
# requested delta channel: ver_i / hs_i). No subqueries, no aggregation.
# Python walks the parent chain once across all channels.
#
# Stage 2 fetches all writes and the seed blobs for ALL channels in a
# single roundtrip via `channel = ANY(%s)` and chain/seed-version
# filtering.
#
# Empirical comparison vs an alternative "ship full channel_versions /
# channel_values JSONB and let Python pick" form (1000 checkpoints,
# 8 total channels in graph, 3 delta channels requested):
#
# Postgres execution: A=0.24ms vs B=0.38ms (both negligible)
# End-to-end latency: A=6.83ms vs B=2.28ms (B is 3.0x faster)
# Wire payload: A=836KB vs B=330KB (61% smaller)
# Buffer hits: identical (167 blocks)
#
# B (this dynamic-columns design) wins because it avoids JSONB
# serialization on the wire and JSONB-to-dict deserialization in
# psycopg. Even at K=8 (8 delta channels = 16 dynamic columns), B
# still beats A end-to-end (4.2ms vs 6.8ms).
def _build_delta_stage1_sql(channels: Sequence[str], *, paged: bool) -> str:
"""Build stage 1 SQL with 2K parallel JSONB key lookups.
For channels=["messages", "files"] (with `paged=True`) the result is::
SELECT checkpoint_id, parent_checkpoint_id,
checkpoint -> 'channel_versions' ->> %s AS ver_0,
(checkpoint -> 'channel_values' -> %s) IS NOT NULL AS hs_0,
checkpoint -> 'channel_versions' ->> %s AS ver_1,
(checkpoint -> 'channel_values' -> %s) IS NOT NULL AS hs_1
FROM checkpoints
WHERE thread_id = %s AND checkpoint_ns = %s
AND (%s::text IS NULL OR checkpoint_id < %s)
ORDER BY checkpoint_id DESC
LIMIT %s
Channel names are passed as `%s` parameters (safe from SQL injection).
Only the column aliases `ver_i` / `hs_i` are interpolated into the
SQL string (i is bounded by len(channels) and uses safe identifiers).
Caller must extend params with `[ch_0, ch_0, ch_1, ch_1, ...,
thread_id, ns, cursor, cursor, page_size]` when `paged=True`.
When `paged=False`, the WHERE has no cursor predicate and there's no
LIMIT/ORDER BY — kept as a non-public helper for tests/diagnostics.
"""
cols = []
for i in range(len(channels)):
cols.append(
f"checkpoint -> 'channel_versions' ->> %s AS ver_{i}, "
f"(checkpoint -> 'channel_values' -> %s) IS NOT NULL AS hs_{i}"
)
sql = (
"SELECT checkpoint_id, parent_checkpoint_id, "
+ ", ".join(cols)
+ " FROM checkpoints WHERE thread_id = %s AND checkpoint_ns = %s"
)
if paged:
sql += (
" AND (%s::text IS NULL OR checkpoint_id < %s)"
" ORDER BY checkpoint_id DESC LIMIT %s"
)
return sql
def _build_delta_stage2_sql(
*,
channels_with_chain: Sequence[str],
channels_with_seed: Sequence[str],
) -> str:
"""Build stage 2 SQL as a per-channel UNION ALL.
For each channel with a non-empty chain, emit one branch reading
`checkpoint_writes` for that specific channel + chain_cids. For each
channel with a seed_version, emit one branch reading `checkpoint_blobs`
for that channel + version. This avoids the over-fetch of the prior
`channel = ANY(channels) AND checkpoint_id = ANY(union)` form when
channels have different chain depths.
The caller must pass parameters in matching order:
for ch in channels_with_chain:
params += [thread_id, checkpoint_ns, ch, chain_cids[ch]]
for ch in channels_with_seed:
params += [thread_id, checkpoint_ns, ch, seed_version[ch]]
Returns an empty SQL string if both channel lists are empty (caller
must skip executing in that case).
"""
branches: list[str] = []
for _ in channels_with_chain:
branches.append(
"SELECT 'w'::text AS _kind, "
"checkpoint_id, channel, "
"type, blob, task_id, idx, NULL::text AS version "
"FROM checkpoint_writes "
"WHERE thread_id = %s AND checkpoint_ns = %s AND channel = %s "
"AND checkpoint_id = ANY(%s)"
)
for _ in channels_with_seed:
branches.append(
"SELECT 'b'::text AS _kind, NULL::text AS checkpoint_id, channel, "
"type, blob, NULL::text AS task_id, NULL::int AS idx, version "
"FROM checkpoint_blobs "
"WHERE thread_id = %s AND checkpoint_ns = %s AND channel = %s "
"AND version = %s"
)
return " UNION ALL ".join(branches)
# Stage 1 rows are dynamic-shape dicts: {checkpoint_id, parent_checkpoint_id,
# ver_0, hs_0, ver_1, hs_1, ...}. Walking is parameterized by the channel
# list to map indices back to channel names — no static TypedDict here.
# `dict[str, Any]` is the practical signature.
class BasePostgresSaver(BaseCheckpointSaver[str]):
SELECT_SQL = SELECT_SQL
SELECT_PENDING_SENDS_SQL = SELECT_PENDING_SENDS_SQL
MIGRATIONS = MIGRATIONS
UPSERT_CHECKPOINT_BLOBS_SQL = UPSERT_CHECKPOINT_BLOBS_SQL
UPSERT_CHECKPOINTS_SQL = UPSERT_CHECKPOINTS_SQL
UPSERT_CHECKPOINT_WRITES_SQL = UPSERT_CHECKPOINT_WRITES_SQL
INSERT_CHECKPOINT_WRITES_SQL = INSERT_CHECKPOINT_WRITES_SQL
supports_pipeline: bool
def _migrate_pending_sends(
self,
pending_sends: list[tuple[bytes, bytes]],
checkpoint: dict[str, Any],
channel_values: list[tuple[bytes, bytes, bytes]],
) -> None:
if not pending_sends:
return
# add to values
enc, blob = self.serde.dumps_typed(
[self.serde.loads_typed((c.decode(), b)) for c, b in pending_sends],
)
channel_values.append((TASKS.encode(), enc.encode(), blob))
# add to versions
checkpoint["channel_versions"][TASKS] = (
max(checkpoint["channel_versions"].values())
if checkpoint["channel_versions"]
else self.get_next_version(None, None)
)
def _load_blobs(
self, blob_values: list[tuple[bytes, bytes, bytes]]
) -> dict[str, Any]:
if not blob_values:
return {}
return {
k.decode(): self.serde.loads_typed((t.decode(), v))
for k, t, v in blob_values
if t.decode() != "empty"
}
@staticmethod
def _ingest_stage1_page(
stage1_rows: Sequence[Mapping[str, Any]],
channels: Sequence[str],
parent_of: dict[str, str | None],
ver_by_i_by_cid: list[dict[str, str | None]],
hs_by_i_by_cid: list[dict[str, bool]],
) -> str | None:
"""Fold one stage-1 page into the running walk-state mappings.
Returns the oldest checkpoint_id seen on this page (smallest, since
pages come back DESC). Caller uses it as the cursor for the next
page (`AND checkpoint_id < cursor`).
"""
oldest: str | None = None
for r in stage1_rows:
cid = cast(str, r["checkpoint_id"])
parent_of[cid] = cast("str | None", r["parent_checkpoint_id"])
for i in range(len(channels)):
ver_by_i_by_cid[i][cid] = cast("str | None", r.get(f"ver_{i}"))
hs_by_i_by_cid[i][cid] = bool(r.get(f"hs_{i}"))
# Rows are DESC; the last one is the smallest cid in the page.
oldest = cid
return oldest
@staticmethod
def _try_advance_walks(
target_id: str,
channels: Sequence[str],
parent_of: Mapping[str, str | None],
ver_by_i_by_cid: Sequence[Mapping[str, str | None]],
hs_by_i_by_cid: Sequence[Mapping[str, bool]],
chain_by_ch: dict[str, list[str]],
seed_ver_by_ch: dict[str, str | None],
walk_cursor_by_ch: dict[str, str | None],
seeded: set[str],
) -> None:
"""Advance each not-yet-seeded channel's walk as far as possible.
Uses the partial `parent_of` map accumulated so far. A walk stops
either because:
(a) it found a snapshot for its channel (channel becomes seeded),
(b) it reached a real root (parent_of[cid] is None — fully
materialized at this point), or
(c) the next ancestor cid isn't in `parent_of` yet (waiting for
a later page; the cursor stays put).
Mutates `chain_by_ch`, `seed_ver_by_ch`, `walk_cursor_by_ch`, and
`seeded` in place.
"""
for i, ch in enumerate(channels):
if ch in seeded:
continue
# First-time entry: cursor starts at the target's parent.
if ch not in walk_cursor_by_ch:
walk_cursor_by_ch[ch] = parent_of.get(target_id)
cur_cid = walk_cursor_by_ch[ch]
ch_chain = chain_by_ch[ch]
hs_i = hs_by_i_by_cid[i]
ver_i = ver_by_i_by_cid[i]
while cur_cid is not None:
if cur_cid not in parent_of:
# Need more pages to continue this walk.
break
ch_chain.append(cur_cid)
if hs_i.get(cur_cid, False):
seed_ver_by_ch[ch] = ver_i.get(cur_cid)
seeded.add(ch)
cur_cid = None
break
cur_cid = parent_of[cur_cid]
walk_cursor_by_ch[ch] = cur_cid
def _build_delta_channels_writes_history(
self,
*,
channels: Sequence[str],
chain_by_ch: Mapping[str, list[str]],
seed_ver_by_ch: Mapping[str, str | None],
stage2_rows: Sequence[_DeltaStage2Row],
) -> dict[str, DeltaChannelHistory]:
"""Demux stage 2 rows per channel; produce per-channel histories.
stage2_rows carry `channel` on every row. We build per-channel
`writes_by_cid` and per-channel `seed_blob` dicts, then assemble
a `DeltaChannelHistory` per requested channel. The `seed` key is omitted
when the walk reached root with no snapshot found, or when the
seed blob is sentinel "empty" — in both cases the consumer treats
absence as "start empty".
"""
# writes_by_ch_by_cid[channel][cid] = list of (type, blob, task_id, idx)
writes_by_ch_by_cid: dict[str, dict[str, list[tuple[str, bytes, str, int]]]] = {
ch: {} for ch in channels
}
# seed_blob_by_ver[(channel, version)] = (type, blob)
seed_blob_by_ver: dict[tuple[str, str], tuple[str, bytes]] = {}
for r in stage2_rows:
ch = cast(str, r["channel"])
kind = r["_kind"]
if kind == "w":
cid = cast(str, r["checkpoint_id"])
writes_by_ch_by_cid.setdefault(ch, {}).setdefault(cid, []).append(
cast(
"tuple[str, bytes, str, int]",
(r["type"], r["blob"], r["task_id"], r["idx"]),
)
)
else: # kind == "b"
ver = cast(str, r["version"])
seed_blob_by_ver[(ch, ver)] = cast(
"tuple[str, bytes]", (r["type"], r["blob"])
)
# Sort writes per (channel, cid) newest-first by (task_id, idx)
for cid_map in writes_by_ch_by_cid.values():
for ws in cid_map.values():
ws.sort(key=lambda w: (w[2], w[3]), reverse=True)
result: dict[str, DeltaChannelHistory] = {}
for ch in channels:
chain_cids = chain_by_ch.get(ch, [])
seed_version = seed_ver_by_ch.get(ch)
collected: list[PendingWrite] = []
cid_writes = writes_by_ch_by_cid.get(ch, {})
for cid in chain_cids:
for type_tag, write_blob, task_id, _idx in cid_writes.get(cid, []):
val = self.serde.loads_typed((type_tag, write_blob))
collected.append((task_id, ch, val))
collected.reverse()
entry: DeltaChannelHistory = {"writes": collected}
if seed_version is not None:
blob = seed_blob_by_ver.get((ch, seed_version))
if blob is not None and blob[0] != "empty":
entry["seed"] = self.serde.loads_typed(blob)
result[ch] = entry
return result
def _dump_blobs(
self,
thread_id: str,
checkpoint_ns: str,
values: dict[str, Any],
versions: ChannelVersions,
) -> list[tuple[str, str, str, str, str, bytes | None]]:
if not versions:
return []
return [
(
thread_id,
checkpoint_ns,
k,
cast(str, ver),
*(
self.serde.dumps_typed(values[k])
if k in values
else ("empty", None)
),
)
for k, ver in versions.items()
]
def _load_writes(
self, writes: list[tuple[bytes, bytes, bytes, bytes]]
) -> list[tuple[str, str, Any]]:
return (
[
(
tid.decode(),
channel.decode(),
self.serde.loads_typed((t.decode(), v)),
)
for tid, channel, t, v in writes
]
if writes
else []
)
def _dump_writes(
self,
thread_id: str,
checkpoint_ns: str,
checkpoint_id: str,
task_id: str,
task_path: str,
writes: Sequence[tuple[str, Any]],
) -> list[tuple[str, str, str, str, str, int, str, str, bytes]]:
return [
(
thread_id,
checkpoint_ns,
checkpoint_id,
task_id,
task_path,
WRITES_IDX_MAP.get(channel, idx),
channel,
*self.serde.dumps_typed(value),
)
for idx, (channel, value) in enumerate(writes)
]
def get_next_version(self, current: str | None, channel: None) -> str:
if current is None:
current_v = 0
elif isinstance(current, int):
current_v = current
else:
current_v = int(current.split(".")[0])
next_v = current_v + 1
next_h = random.random()
return f"{next_v:032}.{next_h:016}"
def _search_where(
self,
config: RunnableConfig | None,
filter: MetadataInput,
before: RunnableConfig | None = None,
) -> tuple[str, list[Any]]:
"""Return WHERE clause predicates for alist() given config, filter, before.
This method returns a tuple of a string and a tuple of values. The string
is the parametered WHERE clause predicate (including the WHERE keyword):
"WHERE column1 = $1 AND column2 IS $2". The list of values contains the
values for each of the corresponding parameters.
"""
wheres = []
param_values = []
# construct predicate for config filter
if config:
wheres.append("thread_id = %s ")
param_values.append(config["configurable"]["thread_id"])
checkpoint_ns = config["configurable"].get("checkpoint_ns")
if checkpoint_ns is not None:
wheres.append("checkpoint_ns = %s")
param_values.append(checkpoint_ns)
if checkpoint_id := get_checkpoint_id(config):
wheres.append("checkpoint_id = %s ")
param_values.append(checkpoint_id)
# construct predicate for metadata filter
if filter:
wheres.append("metadata @> %s ")
param_values.append(Jsonb(filter))
# construct predicate for `before`
if before is not None:
wheres.append("checkpoint_id < %s ")
param_values.append(get_checkpoint_id(before))
return (
"WHERE " + " AND ".join(wheres) if wheres else "",
param_values,
)
@@ -0,0 +1,967 @@
import asyncio
import threading
import warnings
from collections.abc import AsyncIterator, Iterator, Sequence
from contextlib import asynccontextmanager, contextmanager
from typing import Any
from langchain_core.runnables import RunnableConfig
from langgraph.checkpoint.base import (
WRITES_IDX_MAP,
ChannelVersions,
Checkpoint,
CheckpointMetadata,
CheckpointTuple,
get_serializable_checkpoint_metadata,
)
from langgraph.checkpoint.serde.base import SerializerProtocol
from langgraph.checkpoint.serde.types import TASKS
from psycopg import (
AsyncConnection,
AsyncCursor,
AsyncPipeline,
Capabilities,
Connection,
Cursor,
Pipeline,
)
from psycopg.rows import DictRow, dict_row
from psycopg.types.json import Jsonb
from psycopg_pool import AsyncConnectionPool, ConnectionPool
from langgraph.checkpoint.postgres import _ainternal, _internal
from langgraph.checkpoint.postgres.base import BasePostgresSaver
"""
To add a new migration, add a new string to the MIGRATIONS list.
The position of the migration in the list is the version number.
"""
MIGRATIONS = [
"""CREATE TABLE IF NOT EXISTS checkpoint_migrations (
v INTEGER PRIMARY KEY
);""",
"""CREATE TABLE IF NOT EXISTS checkpoints (
thread_id TEXT NOT NULL,
checkpoint_ns TEXT NOT NULL DEFAULT '',
type TEXT,
checkpoint JSONB NOT NULL,
metadata JSONB NOT NULL DEFAULT '{}',
PRIMARY KEY (thread_id, checkpoint_ns)
);""",
"""CREATE TABLE IF NOT EXISTS checkpoint_blobs (
thread_id TEXT NOT NULL,
checkpoint_ns TEXT NOT NULL DEFAULT '',
channel TEXT NOT NULL,
type TEXT NOT NULL,
blob BYTEA,
PRIMARY KEY (thread_id, checkpoint_ns, channel)
);""",
"""CREATE TABLE IF NOT EXISTS checkpoint_writes (
thread_id TEXT NOT NULL,
checkpoint_ns TEXT NOT NULL DEFAULT '',
checkpoint_id TEXT NOT NULL,
task_id TEXT NOT NULL,
idx INTEGER NOT NULL,
channel TEXT NOT NULL,
type TEXT,
blob BYTEA NOT NULL,
PRIMARY KEY (thread_id, checkpoint_ns, checkpoint_id, task_id, idx)
);""",
"""
CREATE INDEX CONCURRENTLY IF NOT EXISTS checkpoints_thread_id_idx ON checkpoints(thread_id);
""",
"""
CREATE INDEX CONCURRENTLY IF NOT EXISTS checkpoint_blobs_thread_id_idx ON checkpoint_blobs(thread_id);
""",
"""
CREATE INDEX CONCURRENTLY IF NOT EXISTS checkpoint_writes_thread_id_idx ON checkpoint_writes(thread_id);
""",
"""
ALTER TABLE checkpoint_writes ADD COLUMN IF NOT EXISTS task_path TEXT NOT NULL DEFAULT '';
""",
]
SELECT_SQL = f"""
select
thread_id,
checkpoint,
checkpoint_ns,
metadata,
(
select array_agg(array[bl.channel::bytea, bl.type::bytea, bl.blob])
from jsonb_each_text(checkpoint -> 'channel_versions')
inner join checkpoint_blobs bl
on bl.thread_id = checkpoints.thread_id
and bl.checkpoint_ns = checkpoints.checkpoint_ns
and bl.channel = jsonb_each_text.key
) as channel_values,
(
select
array_agg(array[cw.task_id::text::bytea, cw.channel::bytea, cw.type::bytea, cw.blob] order by cw.task_id, cw.idx)
from checkpoint_writes cw
where cw.thread_id = checkpoints.thread_id
and cw.checkpoint_ns = checkpoints.checkpoint_ns
and cw.checkpoint_id = (checkpoint->>'id')
) as pending_writes,
(
select array_agg(array[cw.type::bytea, cw.blob] order by cw.task_path, cw.task_id, cw.idx)
from checkpoint_writes cw
where cw.thread_id = checkpoints.thread_id
and cw.checkpoint_ns = checkpoints.checkpoint_ns
and cw.channel = '{TASKS}'
) as pending_sends
from checkpoints """
UPSERT_CHECKPOINT_BLOBS_SQL = """
INSERT INTO checkpoint_blobs (thread_id, checkpoint_ns, channel, type, blob)
VALUES (%s, %s, %s, %s, %s)
ON CONFLICT (thread_id, checkpoint_ns, channel) DO UPDATE SET
type = EXCLUDED.type,
blob = EXCLUDED.blob;
"""
UPSERT_CHECKPOINTS_SQL = """
INSERT INTO checkpoints (thread_id, checkpoint_ns, checkpoint, metadata)
VALUES (%s, %s, %s, %s)
ON CONFLICT (thread_id, checkpoint_ns)
DO UPDATE SET
checkpoint = EXCLUDED.checkpoint,
metadata = EXCLUDED.metadata;
"""
UPSERT_CHECKPOINT_WRITES_SQL = """
INSERT INTO checkpoint_writes (thread_id, checkpoint_ns, checkpoint_id, task_id, task_path, idx, channel, type, blob)
VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s)
ON CONFLICT (thread_id, checkpoint_ns, checkpoint_id, task_id, idx) DO UPDATE SET
channel = EXCLUDED.channel,
type = EXCLUDED.type,
blob = EXCLUDED.blob;
"""
INSERT_CHECKPOINT_WRITES_SQL = """
INSERT INTO checkpoint_writes (thread_id, checkpoint_ns, checkpoint_id, task_id, task_path, idx, channel, type, blob)
VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s)
ON CONFLICT (thread_id, checkpoint_ns, checkpoint_id, task_id, idx) DO NOTHING
"""
def _dump_blobs(
serde: SerializerProtocol,
thread_id: str,
checkpoint_ns: str,
values: dict[str, Any],
versions: ChannelVersions,
) -> list[tuple[str, str, str, str, bytes | None]]:
if not versions:
return []
return [
(
thread_id,
checkpoint_ns,
k,
*(serde.dumps_typed(values[k]) if k in values else ("empty", None)),
)
for k in versions
]
class ShallowPostgresSaver(BasePostgresSaver):
"""A checkpoint saver that uses Postgres to store checkpoints.
This checkpointer ONLY stores the most recent checkpoint and does NOT retain any history.
It is meant to be a light-weight drop-in replacement for the PostgresSaver that
supports most of the LangGraph persistence functionality with the exception of time travel.
"""
SELECT_SQL = SELECT_SQL
MIGRATIONS = MIGRATIONS
UPSERT_CHECKPOINT_BLOBS_SQL = UPSERT_CHECKPOINT_BLOBS_SQL
UPSERT_CHECKPOINTS_SQL = UPSERT_CHECKPOINTS_SQL
UPSERT_CHECKPOINT_WRITES_SQL = UPSERT_CHECKPOINT_WRITES_SQL
INSERT_CHECKPOINT_WRITES_SQL = INSERT_CHECKPOINT_WRITES_SQL
lock: threading.Lock
def __init__(
self,
conn: _internal.Conn,
pipe: Pipeline | None = None,
serde: SerializerProtocol | None = None,
) -> None:
warnings.warn(
"ShallowPostgresSaver is deprecated as of version 2.0.20 and will be removed in 3.0.0. "
"Use PostgresSaver instead, and invoke the graph with `graph.invoke(..., durability='exit')`.",
DeprecationWarning,
stacklevel=2,
)
super().__init__(serde=serde)
if isinstance(conn, ConnectionPool) and pipe is not None:
raise ValueError(
"Pipeline should be used only with a single Connection, not ConnectionPool."
)
self.conn = conn
self.pipe = pipe
self.lock = threading.Lock()
self.supports_pipeline = Capabilities().has_pipeline()
@classmethod
@contextmanager
def from_conn_string(
cls, conn_string: str, *, pipeline: bool = False
) -> Iterator["ShallowPostgresSaver"]:
"""Create a new ShallowPostgresSaver instance from a connection string.
Args:
conn_string: The Postgres connection info string.
pipeline: whether to use Pipeline
Returns:
ShallowPostgresSaver: A new ShallowPostgresSaver instance.
"""
with Connection.connect(
conn_string, autocommit=True, prepare_threshold=0, row_factory=dict_row
) as conn:
if pipeline:
with conn.pipeline() as pipe:
yield cls(conn, pipe)
else:
yield cls(conn)
def setup(self) -> None:
"""Set up the checkpoint database asynchronously.
This method creates the necessary tables in the Postgres database if they don't
already exist and runs database migrations. It MUST be called directly by the user
the first time checkpointer is used.
"""
with self._cursor() as cur:
cur.execute(self.MIGRATIONS[0])
results = cur.execute(
"SELECT v FROM checkpoint_migrations ORDER BY v DESC LIMIT 1"
)
row = results.fetchone()
if row is None:
version = -1
else:
version = row["v"]
for v, migration in zip(
range(version + 1, len(self.MIGRATIONS)),
self.MIGRATIONS[version + 1 :],
strict=False,
):
cur.execute(migration)
cur.execute("INSERT INTO checkpoint_migrations (v) VALUES (%s)", (v,))
if self.pipe:
self.pipe.sync()
def list(
self,
config: RunnableConfig | None,
*,
filter: dict[str, Any] | None = None,
before: RunnableConfig | None = None,
limit: int | None = None,
) -> Iterator[CheckpointTuple]:
"""List checkpoints from the database.
This method retrieves a list of checkpoint tuples from the Postgres database based
on the provided config. For ShallowPostgresSaver, this method returns a list with
ONLY the most recent checkpoint.
"""
where, args = self._search_where(config, filter, before)
query = self.SELECT_SQL + where
params = list(args)
if limit is not None:
query += " LIMIT %s"
params.append(int(limit))
with self._cursor() as cur:
cur.execute(query, params, binary=True)
for value in cur:
checkpoint: Checkpoint = {
**value["checkpoint"],
"channel_values": self._load_blobs(value["channel_values"]),
"pending_sends": [
self.serde.loads_typed((t.decode(), v))
for t, v in value["pending_sends"]
]
if value["pending_sends"]
else [],
}
yield CheckpointTuple(
config={
"configurable": {
"thread_id": value["thread_id"],
"checkpoint_ns": value["checkpoint_ns"],
"checkpoint_id": checkpoint["id"],
}
},
checkpoint=checkpoint,
metadata=value["metadata"],
pending_writes=self._load_writes(value["pending_writes"]),
)
def get_tuple(self, config: RunnableConfig) -> CheckpointTuple | None:
"""Get a checkpoint tuple from the database.
This method retrieves a checkpoint tuple from the Postgres database based on the
provided config (matching the thread ID in the config).
Args:
config: The config to use for retrieving the checkpoint.
Returns:
The retrieved checkpoint tuple, or None if no matching checkpoint was found.
Examples:
Basic:
>>> config = {"configurable": {"thread_id": "1"}}
>>> checkpoint_tuple = memory.get_tuple(config)
>>> print(checkpoint_tuple)
CheckpointTuple(...)
With timestamp:
>>> config = {
... "configurable": {
... "thread_id": "1",
... "checkpoint_ns": "",
... "checkpoint_id": "1ef4f797-8335-6428-8001-8a1503f9b875",
... }
... }
>>> checkpoint_tuple = memory.get_tuple(config)
>>> print(checkpoint_tuple)
CheckpointTuple(...)
""" # noqa
thread_id = config["configurable"]["thread_id"]
checkpoint_ns = config["configurable"].get("checkpoint_ns", "")
args = (thread_id, checkpoint_ns)
where = "WHERE thread_id = %s AND checkpoint_ns = %s"
with self._cursor() as cur:
cur.execute(
self.SELECT_SQL + where,
args,
binary=True,
)
for value in cur:
checkpoint: Checkpoint = {
**value["checkpoint"],
"channel_values": self._load_blobs(value["channel_values"]),
"pending_sends": [
self.serde.loads_typed((t.decode(), v))
for t, v in value["pending_sends"]
]
if value["pending_sends"]
else [],
}
return CheckpointTuple(
config={
"configurable": {
"thread_id": thread_id,
"checkpoint_ns": checkpoint_ns,
"checkpoint_id": checkpoint["id"],
}
},
checkpoint=checkpoint,
metadata=value["metadata"],
pending_writes=self._load_writes(value["pending_writes"]),
)
def put(
self,
config: RunnableConfig,
checkpoint: Checkpoint,
metadata: CheckpointMetadata,
new_versions: ChannelVersions,
) -> RunnableConfig:
"""Save a checkpoint to the database.
This method saves a checkpoint to the Postgres database. The checkpoint is associated
with the provided config. For ShallowPostgresSaver, this method saves ONLY the most recent
checkpoint and overwrites a previous checkpoint, if it exists.
Args:
config: The config to associate with the checkpoint.
checkpoint: The checkpoint to save.
metadata: Additional metadata to save with the checkpoint.
new_versions: New channel versions as of this write.
Returns:
RunnableConfig: Updated configuration after storing the checkpoint.
Examples:
>>> from langgraph.checkpoint.postgres import ShallowPostgresSaver
>>> DB_URI = "postgres://postgres:postgres@localhost:5432/postgres?sslmode=disable"
>>> with ShallowPostgresSaver.from_conn_string(DB_URI) as memory:
>>> config = {"configurable": {"thread_id": "1", "checkpoint_ns": ""}}
>>> checkpoint = {"ts": "2024-05-04T06:32:42.235444+00:00", "id": "1ef4f797-8335-6428-8001-8a1503f9b875", "channel_values": {"key": "value"}}
>>> saved_config = memory.put(config, checkpoint, {"source": "input", "step": 1, "writes": {"key": "value"}}, {})
>>> print(saved_config)
{'configurable': {'thread_id': '1', 'checkpoint_ns': '', 'checkpoint_id': '1ef4f797-8335-6428-8001-8a1503f9b875'}}
"""
configurable = config["configurable"].copy()
thread_id = configurable.pop("thread_id")
checkpoint_ns = configurable.pop("checkpoint_ns")
copy = checkpoint.copy()
next_config = {
"configurable": {
"thread_id": thread_id,
"checkpoint_ns": checkpoint_ns,
"checkpoint_id": checkpoint["id"],
}
}
with self._cursor(pipeline=True) as cur:
cur.execute(
"""DELETE FROM checkpoint_writes
WHERE thread_id = %s AND checkpoint_ns = %s AND checkpoint_id NOT IN (%s, %s)""",
(
thread_id,
checkpoint_ns,
checkpoint["id"],
configurable.get("checkpoint_id", ""),
),
)
cur.executemany(
self.UPSERT_CHECKPOINT_BLOBS_SQL,
_dump_blobs(
self.serde,
thread_id,
checkpoint_ns,
copy.pop("channel_values"), # type: ignore[misc]
new_versions,
),
)
cur.execute(
self.UPSERT_CHECKPOINTS_SQL,
(
thread_id,
checkpoint_ns,
Jsonb(copy),
Jsonb(get_serializable_checkpoint_metadata(config, metadata)),
),
)
return next_config
def put_writes(
self,
config: RunnableConfig,
writes: Sequence[tuple[str, Any]],
task_id: str,
task_path: str = "",
) -> None:
"""Store intermediate writes linked to a checkpoint.
This method saves intermediate writes associated with a checkpoint to the Postgres database.
Args:
config: Configuration of the related checkpoint.
writes: List of writes to store.
task_id: Identifier for the task creating the writes.
"""
query = (
self.UPSERT_CHECKPOINT_WRITES_SQL
if all(w[0] in WRITES_IDX_MAP for w in writes)
else self.INSERT_CHECKPOINT_WRITES_SQL
)
with self._cursor(pipeline=True) as cur:
cur.executemany(
query,
self._dump_writes(
config["configurable"]["thread_id"],
config["configurable"]["checkpoint_ns"],
config["configurable"]["checkpoint_id"],
task_id,
task_path,
writes,
),
)
@contextmanager
def _cursor(self, *, pipeline: bool = False) -> Iterator[Cursor[DictRow]]:
"""Create a database cursor as a context manager.
Args:
pipeline: whether to use pipeline for the DB operations inside the context manager.
Will be applied regardless of whether the ShallowPostgresSaver instance was initialized with a pipeline.
If pipeline mode is not supported, will fall back to using transaction context manager.
"""
with _internal.get_connection(self.conn) as conn:
if self.pipe:
# a connection in pipeline mode can be used concurrently
# in multiple threads/coroutines, but only one cursor can be
# used at a time
try:
with conn.cursor(binary=True, row_factory=dict_row) as cur:
yield cur
finally:
if pipeline:
self.pipe.sync()
elif pipeline:
# a connection not in pipeline mode can only be used by one
# thread/coroutine at a time, so we acquire a lock
if self.supports_pipeline:
with (
self.lock,
conn.pipeline(),
conn.cursor(binary=True, row_factory=dict_row) as cur,
):
yield cur
else:
# Use connection's transaction context manager when pipeline mode not supported
with (
self.lock,
conn.transaction(),
conn.cursor(binary=True, row_factory=dict_row) as cur,
):
yield cur
else:
with self.lock, conn.cursor(binary=True, row_factory=dict_row) as cur:
yield cur
class AsyncShallowPostgresSaver(BasePostgresSaver):
"""A checkpoint saver that uses Postgres to store checkpoints asynchronously.
This checkpointer ONLY stores the most recent checkpoint and does NOT retain any history.
It is meant to be a light-weight drop-in replacement for the AsyncPostgresSaver that
supports most of the LangGraph persistence functionality with the exception of time travel.
"""
SELECT_SQL = SELECT_SQL
MIGRATIONS = MIGRATIONS
UPSERT_CHECKPOINT_BLOBS_SQL = UPSERT_CHECKPOINT_BLOBS_SQL
UPSERT_CHECKPOINTS_SQL = UPSERT_CHECKPOINTS_SQL
UPSERT_CHECKPOINT_WRITES_SQL = UPSERT_CHECKPOINT_WRITES_SQL
INSERT_CHECKPOINT_WRITES_SQL = INSERT_CHECKPOINT_WRITES_SQL
lock: asyncio.Lock
def __init__(
self,
conn: _ainternal.Conn,
pipe: AsyncPipeline | None = None,
serde: SerializerProtocol | None = None,
) -> None:
warnings.warn(
"AsyncShallowPostgresSaver is deprecated as of version 2.0.20 and will be removed in 3.0.0. "
"Use AsyncPostgresSaver instead, and invoke the graph with `await graph.ainvoke(..., durability='exit')`.",
DeprecationWarning,
stacklevel=2,
)
super().__init__(serde=serde)
if isinstance(conn, AsyncConnectionPool) and pipe is not None:
raise ValueError(
"Pipeline should be used only with a single AsyncConnection, not AsyncConnectionPool."
)
self.conn = conn
self.pipe = pipe
self.lock = asyncio.Lock()
self.loop = asyncio.get_running_loop()
self.supports_pipeline = Capabilities().has_pipeline()
@classmethod
@asynccontextmanager
async def from_conn_string(
cls,
conn_string: str,
*,
pipeline: bool = False,
serde: SerializerProtocol | None = None,
) -> AsyncIterator["AsyncShallowPostgresSaver"]:
"""Create a new AsyncShallowPostgresSaver instance from a connection string.
Args:
conn_string: The Postgres connection info string.
pipeline: whether to use AsyncPipeline
Returns:
AsyncShallowPostgresSaver: A new AsyncShallowPostgresSaver instance.
"""
async with await AsyncConnection.connect(
conn_string, autocommit=True, prepare_threshold=0, row_factory=dict_row
) as conn:
if pipeline:
async with conn.pipeline() as pipe:
yield cls(conn=conn, pipe=pipe, serde=serde)
else:
yield cls(conn=conn, serde=serde)
async def setup(self) -> None:
"""Set up the checkpoint database asynchronously.
This method creates the necessary tables in the Postgres database if they don't
already exist and runs database migrations. It MUST be called directly by the user
the first time checkpointer is used.
"""
async with self._cursor() as cur:
await cur.execute(self.MIGRATIONS[0])
results = await cur.execute(
"SELECT v FROM checkpoint_migrations ORDER BY v DESC LIMIT 1"
)
row = await results.fetchone()
if row is None:
version = -1
else:
version = row["v"]
for v, migration in zip(
range(version + 1, len(self.MIGRATIONS)),
self.MIGRATIONS[version + 1 :],
strict=False,
):
await cur.execute(migration)
await cur.execute(
"INSERT INTO checkpoint_migrations (v) VALUES (%s)", (v,)
)
if self.pipe:
await self.pipe.sync()
async def alist(
self,
config: RunnableConfig | None,
*,
filter: dict[str, Any] | None = None,
before: RunnableConfig | None = None,
limit: int | None = None,
) -> AsyncIterator[CheckpointTuple]:
"""List checkpoints from the database asynchronously.
This method retrieves a list of checkpoint tuples from the Postgres database based
on the provided config. For ShallowPostgresSaver, this method returns a list with
ONLY the most recent checkpoint.
"""
where, args = self._search_where(config, filter, before)
query = self.SELECT_SQL + where
params = list(args)
if limit is not None:
query += " LIMIT %s"
params.append(int(limit))
async with self._cursor() as cur:
await cur.execute(query, params, binary=True)
async for value in cur:
checkpoint: Checkpoint = {
**value["checkpoint"],
"channel_values": self._load_blobs(value["channel_values"]),
"pending_sends": [
self.serde.loads_typed((t.decode(), v))
for t, v in value["pending_sends"]
]
if value["pending_sends"]
else [],
}
yield CheckpointTuple(
config={
"configurable": {
"thread_id": value["thread_id"],
"checkpoint_ns": value["checkpoint_ns"],
"checkpoint_id": checkpoint["id"],
}
},
checkpoint=checkpoint,
metadata=value["metadata"],
pending_writes=await asyncio.to_thread(
self._load_writes, value["pending_writes"]
),
)
async def aget_tuple(self, config: RunnableConfig) -> CheckpointTuple | None:
"""Get a checkpoint tuple from the database asynchronously.
This method retrieves a checkpoint tuple from the Postgres database based on the
provided config (matching the thread ID in the config).
Args:
config: The config to use for retrieving the checkpoint.
Returns:
The retrieved checkpoint tuple, or None if no matching checkpoint was found.
"""
thread_id = config["configurable"]["thread_id"]
checkpoint_ns = config["configurable"].get("checkpoint_ns", "")
args = (thread_id, checkpoint_ns)
where = "WHERE thread_id = %s AND checkpoint_ns = %s"
async with self._cursor() as cur:
await cur.execute(
self.SELECT_SQL + where,
args,
binary=True,
)
async for value in cur:
checkpoint: Checkpoint = {
**value["checkpoint"],
"channel_values": self._load_blobs(value["channel_values"]),
"pending_sends": [
self.serde.loads_typed((t.decode(), v))
for t, v in value["pending_sends"]
]
if value["pending_sends"]
else [],
}
return CheckpointTuple(
config={
"configurable": {
"thread_id": thread_id,
"checkpoint_ns": checkpoint_ns,
"checkpoint_id": checkpoint["id"],
}
},
checkpoint=checkpoint,
metadata=value["metadata"],
pending_writes=await asyncio.to_thread(
self._load_writes, value["pending_writes"]
),
)
async def aput(
self,
config: RunnableConfig,
checkpoint: Checkpoint,
metadata: CheckpointMetadata,
new_versions: ChannelVersions,
) -> RunnableConfig:
"""Save a checkpoint to the database asynchronously.
This method saves a checkpoint to the Postgres database. The checkpoint is associated
with the provided config. For AsyncShallowPostgresSaver, this method saves ONLY the most recent
checkpoint and overwrites a previous checkpoint, if it exists.
Args:
config: The config to associate with the checkpoint.
checkpoint: The checkpoint to save.
metadata: Additional metadata to save with the checkpoint.
new_versions: New channel versions as of this write.
Returns:
RunnableConfig: Updated configuration after storing the checkpoint.
"""
configurable = config["configurable"].copy()
thread_id = configurable.pop("thread_id")
checkpoint_ns = configurable.pop("checkpoint_ns")
copy = checkpoint.copy()
next_config = {
"configurable": {
"thread_id": thread_id,
"checkpoint_ns": checkpoint_ns,
"checkpoint_id": checkpoint["id"],
}
}
async with self._cursor(pipeline=True) as cur:
await cur.execute(
"""DELETE FROM checkpoint_writes
WHERE thread_id = %s AND checkpoint_ns = %s AND checkpoint_id NOT IN (%s, %s)""",
(
thread_id,
checkpoint_ns,
checkpoint["id"],
configurable.get("checkpoint_id", ""),
),
)
await cur.executemany(
self.UPSERT_CHECKPOINT_BLOBS_SQL,
_dump_blobs(
self.serde,
thread_id,
checkpoint_ns,
copy.pop("channel_values"), # type: ignore[misc]
new_versions,
),
)
await cur.execute(
self.UPSERT_CHECKPOINTS_SQL,
(
thread_id,
checkpoint_ns,
Jsonb(copy),
Jsonb(get_serializable_checkpoint_metadata(config, metadata)),
),
)
return next_config
async def aput_writes(
self,
config: RunnableConfig,
writes: Sequence[tuple[str, Any]],
task_id: str,
task_path: str = "",
) -> None:
"""Store intermediate writes linked to a checkpoint asynchronously.
This method saves intermediate writes associated with a checkpoint to the database.
Args:
config: Configuration of the related checkpoint.
writes: List of writes to store, each as (channel, value) pair.
task_id: Identifier for the task creating the writes.
"""
query = (
self.UPSERT_CHECKPOINT_WRITES_SQL
if all(w[0] in WRITES_IDX_MAP for w in writes)
else self.INSERT_CHECKPOINT_WRITES_SQL
)
params = await asyncio.to_thread(
self._dump_writes,
config["configurable"]["thread_id"],
config["configurable"]["checkpoint_ns"],
config["configurable"]["checkpoint_id"],
task_id,
task_path,
writes,
)
async with self._cursor(pipeline=True) as cur:
await cur.executemany(query, params)
@asynccontextmanager
async def _cursor(
self, *, pipeline: bool = False
) -> AsyncIterator[AsyncCursor[DictRow]]:
"""Create a database cursor as a context manager.
Args:
pipeline: whether to use pipeline for the DB operations inside the context manager.
Will be applied regardless of whether the AsyncShallowPostgresSaver instance was initialized with a pipeline.
If pipeline mode is not supported, will fall back to using transaction context manager.
"""
async with _ainternal.get_connection(self.conn) as conn:
if self.pipe:
# a connection in pipeline mode can be used concurrently
# in multiple threads/coroutines, but only one cursor can be
# used at a time
try:
async with conn.cursor(binary=True, row_factory=dict_row) as cur:
yield cur
finally:
if pipeline:
await self.pipe.sync()
elif pipeline:
# a connection not in pipeline mode can only be used by one
# thread/coroutine at a time, so we acquire a lock
if self.supports_pipeline:
async with (
self.lock,
conn.pipeline(),
conn.cursor(binary=True, row_factory=dict_row) as cur,
):
yield cur
else:
# Use connection's transaction context manager when pipeline mode not supported
async with (
self.lock,
conn.transaction(),
conn.cursor(binary=True, row_factory=dict_row) as cur,
):
yield cur
else:
async with (
self.lock,
conn.cursor(binary=True, row_factory=dict_row) as cur,
):
yield cur
def list(
self,
config: RunnableConfig | None,
*,
filter: dict[str, Any] | None = None,
before: RunnableConfig | None = None,
limit: int | None = None,
) -> Iterator[CheckpointTuple]:
"""List checkpoints from the database.
This method retrieves a list of checkpoint tuples from the Postgres database based
on the provided config. For ShallowPostgresSaver, this method returns a list with
ONLY the most recent checkpoint.
"""
aiter_ = self.alist(config, filter=filter, before=before, limit=limit)
while True:
try:
yield asyncio.run_coroutine_threadsafe(
anext(aiter_), # type: ignore[arg-type] # noqa: F821
self.loop,
).result()
except StopAsyncIteration:
break
def get_tuple(self, config: RunnableConfig) -> CheckpointTuple | None:
"""Get a checkpoint tuple from the database.
This method retrieves a checkpoint tuple from the Postgres database based on the
provided config (matching the thread ID in the config).
Args:
config: The config to use for retrieving the checkpoint.
Returns:
The retrieved checkpoint tuple, or None if no matching checkpoint was found.
"""
try:
# check if we are in the main thread, only bg threads can block
# we don't check in other methods to avoid the overhead
if asyncio.get_running_loop() is self.loop:
raise asyncio.InvalidStateError(
"Synchronous calls to AsyncShallowPostgresSaver are only allowed from a "
"different thread. From the main thread, use the async interface."
"For example, use `await checkpointer.aget_tuple(...)` or `await "
"graph.ainvoke(...)`."
)
except RuntimeError:
pass
return asyncio.run_coroutine_threadsafe(
self.aget_tuple(config), self.loop
).result()
def put(
self,
config: RunnableConfig,
checkpoint: Checkpoint,
metadata: CheckpointMetadata,
new_versions: ChannelVersions,
) -> RunnableConfig:
"""Save a checkpoint to the database.
This method saves a checkpoint to the Postgres database. The checkpoint is associated
with the provided config. For AsyncShallowPostgresSaver, this method saves ONLY the most recent
checkpoint and overwrites a previous checkpoint, if it exists.
Args:
config: The config to associate with the checkpoint.
checkpoint: The checkpoint to save.
metadata: Additional metadata to save with the checkpoint.
new_versions: New channel versions as of this write.
Returns:
RunnableConfig: Updated configuration after storing the checkpoint.
"""
return asyncio.run_coroutine_threadsafe(
self.aput(config, checkpoint, metadata, new_versions), self.loop
).result()
def put_writes(
self,
config: RunnableConfig,
writes: Sequence[tuple[str, Any]],
task_id: str,
task_path: str = "",
) -> None:
"""Store intermediate writes linked to a checkpoint.
This method saves intermediate writes associated with a checkpoint to the database.
Args:
config: Configuration of the related checkpoint.
writes: List of writes to store, each as (channel, value) pair.
task_id: Identifier for the task creating the writes.
task_path: Path of the task creating the writes.
"""
return asyncio.run_coroutine_threadsafe(
self.aput_writes(config, writes, task_id, task_path), self.loop
).result()