pyth-pandas

pandas-style client for the Pyth Pro Router (Pyth Lazer) REST + WebSocket API

Every list endpoint returns a pandas.DataFrame, every dict endpoint returns a TypedDict, and every shape is enforced by a pandera schema. Sync + async, plus optional WebSocket, Streamlit explorer, and FastMCP server extras.

This client targets Pyth Pro Router API v1.0.0 (per the published OpenAPI spec). The package version (0.1.0) is independent of the upstream API version — see the API_VERSION / SUPPORTED_API_VERSIONS class attributes on PythPandas.

Install

uv pip install pyth-pandas              # core
uv pip install "pyth-pandas[ws]"        # + WebSocket
uv pip install "pyth-pandas[explorer]"  # + Streamlit dashboard
uv pip install "pyth-pandas[mcp]"       # + FastMCP server
uv pip install "pyth-pandas[dev]"       # + test/lint toolchain

Quickstart

from pyth_pandas import PythPandas

with PythPandas() as client:
    df = client.fetch_latest_prices(
        symbols=["Crypto.BTC/USD", "Crypto.ETH/USD"],
        properties=["price", "confidence", "exponent", "publisherCount"],
        formats=[],
    )
    df["humanPrice"] = df["price"] * 10 ** df["exponent"]
    print(df[["priceFeedId", "humanPrice", "publisherCount"]])

The DataFrame has one row per feed. The on-chain payloads (when requested via formats=["evm", ...]) and the update timestamp are attached on df.attrs. For the raw JsonUpdate dict, use fetch_latest_prices_raw(...).

Async

import asyncio
from pyth_pandas import AsyncPythPandas

async def main():
    async with AsyncPythPandas() as client:
        df = await client.fetch_latest_prices(
            symbols=["Crypto.BTC/USD"],
            properties=["price", "exponent"],
            formats=[],
        )
        print(df)

asyncio.run(main())

Authentication

Pyth Pro requires a bearer access token. Set it via env var (recommended) or constructor arg. Copy the shipped .env.example to .env and fill in your key — examples and tests load it via python-dotenv:

cp .env.example .env
$EDITOR .env       # set PYTH_API_KEY=...

Or set it in your shell:

export PYTH_API_KEY=...

Or pass it directly:

PythPandas(api_key="...")

Endpoints

WebSocket: PythWebSocket / AsyncPythWebSocket (extra: ws) → connects to wss://pyth-lazer.dourolabs.app/v1/stream.

Compatibility

Error handling

from pyth_pandas import PythAuthError, PythRateLimitError, PythAPIError

try:
    df = client.fetch_latest_prices(symbols=["Crypto.BTC/USD"], properties=["price"])
except PythAuthError:
    ...  # 401/403 or missing PYTH_API_KEY
except PythRateLimitError:
    ...  # 429
except PythAPIError as e:
    print(e.status_code, e.url, e.detail)

Development

uv pip install -e ".[dev,ws,explorer,mcp]"
uv run pytest tests/test_unit.py tests/test_async_unit.py -v
uv run ruff check pyth_pandas tests
uv run mypy pyth_pandas

CI runs the same checks on every push and PR via .github/workflows/ci.yml across Python 3.11, 3.12, 3.13.

Releasing

Releases are cut by pushing a v* tag. .github/workflows/release.yml:

1. Runs ruff + mypy + mocked pytest, then live integration tests against the real upstream API. If the test job fails, build / publish / release are all skipped.
2. Builds sdist + wheel via uv build.
3. Publishes to PyPI via trusted publishing.
4. Creates a GitHub release.

One-time setup:

1. Reserve pyth-pandas on PyPI.
2. At https://pypi.org/manage/account/publishing/, add a pending publisher:
   • Project: pyth-pandas
   • Owner: your GitHub user/org
   • Repo: pyth-pandas
   • Workflow: release.yml
   • Environment: pypi
3. Create the pypi environment in GitHub repo settings.
4. Add PYTH_API_KEY as an environment secret on the pypi environment so live integration tests can run before publishing.
5. Enable GitHub Pages with source = "GitHub Actions" so docs.yml can deploy the Sphinx site.

To force a release if the test infra is broken:

gh workflow run release.yml -f tag=v0.1.1 -f skip_tests=true

License

Apache-2.0