CoinbasePro: A Python API

Features

• Full support of Coinbase Pro/Coinbase Exchange REST API

• Rate-limiting - no more 429 error responses!

• Pythonic abstractions for a clean interface

  • Return values are returned as Python data types instead of all string values:

  >>> import coinbasepro as cbp
  >>> client = cbp.PublicClient()
  # datetime and Decimal are among the return types in the dict returned
  # by this call:
  >>> client.get_product_ticker('BTC-USD')
  {'trade_id': 2845680,
  'price': Decimal('2496.69000000'),
  'size': Decimal('0.00100000'),
  'time': datetime.datetime(2019, 3, 20, 23, 53, 59, 596000),
  'bid': Decimal('2496.69'), 'ask': Decimal('2496.7'),
  'volume': Decimal('771.51495215')}

  • Paginated endpoints are abstracted as generators:

  >>> import itertools
  # get_product_trades is a generator
  >>> client.get_product_trades('BTC-USD')
  <generator object PublicClient.get_product_trades.<locals>.<genexpr> at 0x1098d6f68>
  
  # Get 2 most recent trades. For many trade requests (>100), multiple
  # HTTP requests will be made under the hood.
  >>> list(itertools.islice(client.get_product_trades('BTC-USD'), 2))
  [{'time': datetime.datetime(2019, 3, 21, 0, 2, 45, 609000),
  'trade_id': 2845779,
  'price': Decimal('2463.38000000'),
  'size': Decimal('0.00100000'),
  'side': 'buy'},
  {'time': datetime.datetime(2019, 3, 21, 0, 2, 39, 877000),
  'trade_id': 2845778,
  'price': Decimal('2463.39000000'),
  'size': Decimal('0.00100000'),
  'side': 'sell'}]

  • Warts in the Coinbase REST API are smoothed out:

  # CBPro API returns raw candles from this call as tuples, which would
  # require user to look up value meaning in API docs. This python API
  # returns candles as a list of dicts, similar to other API endpoints.
  
  # Get first candle:
  >>> client.get_product_historic_rates('BTC-USD')[0]
  {'time': datetime.datetime(2019, 3, 21, 0, 6),
  'low': Decimal('2463.3'),
  'high': Decimal('2463.31'),
  'open': Decimal('2463.3'),
  'close': Decimal('2463.31'),
  'volume': Decimal('0.006')}

  • Python API uses typing available in Python3:

  # Example function prototype in API
  def get_product_ticker(self, product_id: str) -> Dict[str, Any]:

• Exceptions to enable easy handling of Coinbase error responses

>>> client.get_product_ticker(product_id='fake_product')
coinbasepro.exceptions.CoinbaseAPIError: NotFound

>>> auth_client = cbp.AuthenticatedClient(key='fake',
                                          secret='fake',
                                          passphrase='fake')
>>> auth_client.get_accounts()
coinbasepro.exceptions.BadRequest: Invalid API Key

# Authenticated client using API key which doesn't have withdrawal
# privileges:
>>> auth_client.withdraw_to_coinbase(0.01, 'BTC', 'fake_acct_id')
coinbasepro.exceptions.InvalidAuthorization: Forbidden

# This call throws a BadRequest exception
>>> auth_client.get_order('invalid_order_num')
coinbasepro.exceptions.BadRequest: Invalid order id

# CoinbaseAPIError is the parent exception for all exceptions the API
# throws, so catching this will catch anything
>>> try:
>>>     auth_client.get_order('invalid_order_num')
>>> except cbp.exceptions.CoinbaseAPIError as e:
>>>     print('Caught error: {}'.format(e))
Caught error: Invalid order id

Installation

$ pip install coinbasepro

Development

Environment Setup

1. Create virtual environment using preferred tool
2. Bootstrap pip-tools by installing dev requirements directly:

$ pip install -r requirements.txt

Once pip-tools is installed in your environment, you can update requirements by:

$ make install-requirements

3. (Optional) Install pre-commit git hooks. This will run pre-commit with every commit, which should fix any lint issues before you push changes to your remote branch.