Skip to main content

PyPI version Downloads

🛰️ deepseek-chat-sdk

A thin, battle-tested Python wrapper around the DeepSeek /chat/completions API
with built-in conversation-tree persistence.

  • Drop-in call style – mirrors openai-python (messages=[…])
  • Two storage back-ends
    MongoDB (default) or ultra-light JSON file – same API, switch in one flag
  • Tree-structured memory – each reply is a node; fork, branch & query whole tree
  • Token-budget guard – automatic context trimming (or raise)
  • 100 % sync code – keep dependencies minimal, trivially pluggable into any code-base
pip install deepseek_chat          # from PyPI (or pip install -e .)
export DEEPSEEK_API_KEY=sk-...

⭐ Quick Start

from deepseek_chat import create_session, continue_chat, get_tree

# --- root turn --------------------------------------------------------
answer, sess_id, root_id = create_session(
    system_prompt="你好,我是助手。",
    question="帮我推荐一部电影。",
    api_key="sk-...",                     # or rely on $DEEPSEEK_API_KEY
)
print(answer)

# --- branch off the same parent --------------------------------------
continue_chat(session_id=sess_id, parent_node_id=root_id,
              question="有没有同类型但评分更高的?")

# --- show the whole tree --------------------------------------------
print(get_tree(sess_id))

Switch to local file storage (no Mongo)

answer, sid, nid = create_session(
    system_prompt="你好!",
    question="今天杭州天气如何?",
    api_key="sk-...",
    store_backend="file",              # ← key flag
    json_path="./history.json",        # optional; default ~/.deepseek_chat/history.json
)

Custom Mongo URI (user/pass, replica set …)

import os
os.environ["DEEPSEEK_API_KEY"] = "sk-..."
answer, sid, nid = create_session(
    system_prompt="You are helpful.",
    question="Tell me a joke.",
    mongo_uri="mongodb://dbUser:dbPwd@mongo-1:27017,mongo-2:27017/?replicaSet=rs0",
)

🔧 Installation

# bare minimum
pip install deepseek_chat

# dev / tests / ruff
pip install "deepseek_chat"
pytest -q
Dependency Used for Optional?
requests REST calls ❌
pymongo Mongo back-end ✅ (not imported if you choose file)
tiktoken token counting ✅ (falls back to word count)

🗺️ API Surface

Functional helpers (one-liners)

function returns must pass
create_session(system_prompt, question, …) (answer, session_id, node_id) system_prompt, question
continue_chat(session_id, parent_node_id, question, …) (answer, new_node_id) session_id, parent_node_id, question
get_tree(session_id, …) nested dict session_id

All take the same optional kwargs:

api_key        DeepSeek API key (fallback $DEEPSEEK_API_KEY)
base_url       custom gateway (fallback $DEEPSEEK_BASE_URL)
model          deepseek-chat / deepseek-coder / …
store_backend  "mongo" | "file"   (default "mongo")
mongo_uri      mongodb connection string
json_path      path for file store

Class-level control

Need multiple clients or stores side-by-side? Use the OOP facade:

from deepseek_chat import ChatService, DeepSeekClient, FileStore

svc = ChatService(
    client=DeepSeekClient(api_key="sk-..."),
    store=FileStore("/tmp/ds_history.json"),
)

ans, sid, nid = svc.create_session(
    system_prompt="You are GPT-5",
    question="What's new in 2025?",
)

⚙️ Configuration matrix

config source DeepSeek Mongo File
ENV var (default singleton) DEEPSEEK_API_KEY DEEPSEEK_BASE_URL MONGO_URI MONGO_DB —
per-call kwargs api_key= base_url= mongo_uri= json_path=
programmatic build your own DeepSeekClient + ChatStore/FileStore and pass into ChatService

📚 Advanced features

JSON Output & Function Calling

Simply forward the native OpenAI-style params:

client = DeepSeekClient(api_key="sk-...")

resp = client.chat(
    messages=[...],
    response_format={"type": "json_object"},
)

resp_tool = client.chat(
    messages=[...],
    tools=[{ "type": "function", "function": {...} }],
)

The higher-level helpers (create_session/continue_chat) accept any extra kwargs and pass them straight to DeepSeekClient.chat().

Token-budget enforcement

ensure_token_budget() truncates eldest user/assistant pairs once total tokens exceed the limit (default 32 k). Raise TokenLimitExceeded if only the system prompt already exceeds the model context.

Set a custom limit:

export TOKEN_LIMIT=16000

🛠️ Project layout

deepseek_chat/
├── client.py        # DeepSeek REST wrapper
├── service.py       # high-level orchestration
├── store.py         # Mongo persistence
├── filestore.py     # JSON persistence
├── utils.py         # token helpers
└── __init__.py      # facade + functional helpers

Unit tests live under tests/, demo scripts under examples/.


🪄 Roadmap

  • async variant (httpx.AsyncClient, motor)
  • automatic summarisation for ultra-long histories
  • plug-in hooks (on_new_message, on_error)

Contributions welcome – just open an issue or PR.


Metadata

Release files for deepseek-chat 0.1.1

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for deepseek-chat 0.1.1
File Size Uploaded Provenance Actions
deepseek_chat-0.1.1.tar.gz 16.2 kB Without provenance Details for deepseek_chat-0.1.1.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for deepseek-chat 0.1.1
File Interpreter ABI Platform Actions
deepseek_chat-0.1.1-py3-none-any.whl Python 3 none any Details for deepseek_chat-0.1.1-py3-none-any.whl

Total release size: 33.1 kB

Release files / deepseek_chat-0.1.1.tar.gz

Download URL deepseek_chat-0.1.1.tar.gz
Size 16.2 kB
Tags Source
SHA-256 checksum
How to use checksums
22f03ac7f6c66aa67b9a154aff26a1b40b3b319e678d2ea47576eea42b52c63b
BLAKE2b-256 checksum
How to use checksums
0f767b69d250ab0703cac908c1f679574a91f486d2cb389fe0568d167fd2ac0d
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.1.0 CPython/3.10.10

Release files / deepseek_chat-0.1.1-py3-none-any.whl

Download URL deepseek_chat-0.1.1-py3-none-any.whl
Size 16.9 kB
Tags Python 3
SHA-256 checksum
How to use checksums
d3d6a0a00cee4c0fdca008df68e7b5de09cf64b1dc165faee23ad06aebb27591
BLAKE2b-256 checksum
How to use checksums
c555abb704e160f010c66e178be0eaf32fa1c72afc08d0ff60589ee80eae2c69
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/6.1.0 CPython/3.10.10

Security for deepseek-chat 0.1.1

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.

Release history Release notifications | RSS feed

This release

0.1.1 This release

2 release files No provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page