LangGraph Oracle Persistence (Checkpoint + Store)
⭐ If this package is useful to you, consider starring the repository — it helps other developers discover these integrations.
Oracle-backed implementations for:
- Checkpoints: OracleSaver (sync) and AsyncOracleSaver (async)
- Key/Value Store with optional vector search: OracleStore (sync) and AsyncOracleStore (async)
Supports:
- Oracle Database for AI Vector Search
- Python 3.10+ and
oracledbdriver
Quickstart
Checkpoints (Async)
import os
import asyncio
from dotenv import load_dotenv
from langgraph_oracledb.checkpoint.oracle import AsyncOracleSaver
load_dotenv()
async def main():
conn_string = f"{os.environ['ORACLE_USERNAME']}/{os.environ['ORACLE_PASSWORD']}@{os.environ['ORACLE_DSN']}"
async with AsyncOracleSaver.from_conn_string(conn_string) as checkpointer:
await checkpointer.setup() # Create tables & apply migrations once
# Then pass to your graph compile (example)
# graph = app.compile(checkpointer=checkpointer)
# await graph.ainvoke(...)
if __name__ == "__main__":
asyncio.run(main())
Sync variant:
from langgraph_oracledb.checkpoint.oracle import OracleSaver
conn_string = "user/password@localhost:1521/FREEPDB1"
with OracleSaver.from_conn_string(conn_string) as checkpointer:
checkpointer.setup()
# graph = app.compile(checkpointer=checkpointer)
# graph.invoke(...)
Store (Async, basic key/value)
import asyncio
from langgraph_oracledb.store.oracle import AsyncOracleStore
async def main():
conn_string = "user/password@localhost:1521/FREEPDB1"
async with AsyncOracleStore.from_conn_string(conn_string) as store:
await store.setup() # Create tables & apply migrations once
ns = ("readme", "example")
await store.aput(ns, "doc1", {"text": "hello"})
item = await store.aget(ns, "doc1")
print(item.value) # {"text": "hello"}
# Non-vector search (lists items by namespace)
results = await store.asearch(ns, limit=10)
print(len(results) >= 1)
if __name__ == "__main__":
asyncio.run(main())
Sync variant:
from langgraph_oracledb.store.oracle import OracleStore
conn_string = "user/password@localhost:1521/FREEPDB1"
with OracleStore.from_conn_string(conn_string) as store:
store.setup()
ns = ("readme", "example")
store.put(ns, "doc1", {"text": "hello"})
item = store.get(ns, "doc1")
print(item.value)
results = store.search(ns, limit=10)
Vector Search (optional)
Vector search is enabled by passing an index configuration with:
dims: embedding dimensionembed: a LangChain Embeddings implementation (e.g., OpenAI, HF, or your own)- optional
fields: which JSON fields to embed (default: whole document) - optional
index_type: HNSW/IVF and parameters
Example (async):
# Pseudo-embeddings for illustration; use any LangChain Embeddings implementation
from langchain_core.embeddings import Embeddings
from langgraph_oracledb.store.oracle import AsyncOracleStore
class FakeEmbeddings(Embeddings):
def embed_documents(self, texts): return [[0.0]*8 for _ in texts]
def embed_query(self, text): return [0.0]*8
async with AsyncOracleStore.from_conn_string(
"user/password@localhost:1521/FREEPDB1",
index={
"dims": 8,
"embed": FakeEmbeddings(),
"fields": ["text"], # embed only the 'text' field
"index_type": {"type": "hnsw", "neighbors": 16, "efconstruction": 200, "distance_metric": "COSINE"},
},
) as store:
await store.setup()
ns = ("docs",)
await store.aput(ns, "a", {"text": "alpha"})
await store.aput(ns, "b", {"text": "beta"})
results = await store.asearch(ns, query="alphabet", limit=2)
Notes:
- Call
setup()once per database/schema to create/upgrade tables. - Vector search requires Oracle 23c/23ai+ with AI Vector Search enabled.
Testing the Examples
The repository's tests will skip automatically if an Oracle instance is not reachable.
- Place your Oracle credentials in
.env(see Configuration) - Run:
pytest -q
This repository includes tests that validate the examples above:
- Async checkpoint setup works
- Async store put/get/search works
Metadata
Release files for langgraph-oracledb 1.0.2
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| langgraph_oracledb-1.0.2.tar.gz | 40.9 kB | Without provenance | Details for langgraph_oracledb-1.0.2.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| langgraph_oracledb-1.0.2-py3-none-any.whl | Python 3 | none | any | Details for langgraph_oracledb-1.0.2-py3-none-any.whl |
Total release size: 86.1 kB
Release files / langgraph_oracledb-1.0.2.tar.gz
| Download URL | langgraph_oracledb-1.0.2.tar.gz |
|---|---|
| Size | 40.9 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
99b6b610b217cd162fd20b221704c7ab43ee700eef43523af378b4a02d278b4d
|
|
BLAKE2b-256 checksum How to use checksums |
a138da2fd16480c3ef072faaa76538293fc8fb6f4936c9215e6b3c7f04dd3ed0
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.14.8
|
Release files / langgraph_oracledb-1.0.2-py3-none-any.whl
| Download URL | langgraph_oracledb-1.0.2-py3-none-any.whl |
|---|---|
| Size | 45.3 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
687c88d034fa0a394779c80cce6dc02042a80916411bfe8ab6c3b03c88a04d80
|
|
BLAKE2b-256 checksum How to use checksums |
3876d5096078d1689f244ff719d9936471d072e2c866015e09fac2ab1e5f6f21
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.14.8
|
Security for langgraph-oracledb 1.0.2
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.