Skip to content

Latest commit

 

History

History
520 lines (422 loc) · 27.7 KB

README.md

File metadata and controls

520 lines (422 loc) · 27.7 KB
Raw

Coin Metrics Python API v4 client library

This is an official Python API client for Coin Metrics API v4.

Installation and Updates

To install the client you can run the following command:

pip install coinmetrics-api-client

Note that the client is updated regularly to reflect the changes made in API v4. Ensure that your latest version matches with what's in pyPI

To update your version, run the following command:

pip install coinmetrics-api-client -U

Introduction

You can use this client for querying all kinds of data with your API.

To initialize the client you should use your API key, and the CoinMetricsClient class like the following.

from coinmetrics.api_client import CoinMetricsClient

client = CoinMetricsClient("<cm_api_key>")

# or to use community API:
client = CoinMetricsClient()

After that you can use the client object for getting information such as available market trades as a list of dictionaries:

print(client.catalog_market_trades_v2().to_list())

or to iterate over each page of data:

for data in client.catalog_market_trades_v2():
    print(data)

you can also use filters for the catalog endpoints like this:

print(client.catalog_market_trades_v2(exchange="binance").to_list())

All the catalog V2 endpoints meant to help access the historical data served by other endpoints. For example, you can get all the BTC market trades for a certain day from binance like this:

import os
from coinmetrics.api_client import CoinMetricsClient
client = CoinMetricsClient(os.environ['CM_API_KEY'])
btc_binance_markets = [market['market'] for market in client.catalog_market_trades_v2(exchange="binance", asset="btc").to_list()]
start_time = "2023-01-01"
end_time = "2023-01-02"
binance_market_trades = client.get_market_trades(markets=btc_binance_markets, start_time=start_time, end_time=end_time, page_size=1000).export_to_csv("binance_trades.csv")

in this case you would get all the information markets that trade on binance only.

You can use this client to connect to our API v4 and get catalog or timeseries data from python environment. It natively supports paging over the data so you can use it to iterate over timeseries entries seamlessly.

The client can be used to query both pro and community data.

The full list of methods can be found in the API Client Spec.

If you'd like a more holistic view of what is offered from an API endpoint you can use the to_dataframe() function associated with our catalog endpoints. The code snippet below shows getting a dataframe of information on all the assets that data is provided for:

print(client.catalog_market_metrics_v2(exchange="binance", page_size=1000).to_dataframe())

Output:

                             market                                            metrics
0       binance-1000BTTCUSDT-future  [{'metric': 'liquidity_depth_0_1_percent_ask_v...
1      binance-1000FLOKIUSDT-future  [{'metric': 'liquidations_reported_future_buy_...
2       binance-1000LUNCBUSD-future  [{'metric': 'liquidations_reported_future_buy_...
3       binance-1000LUNCUSDT-future  [{'metric': 'liquidations_reported_future_buy_...
4       binance-1000PEPEUSDT-future  [{'metric': 'liquidations_reported_future_buy_...

Now you can use the pandas Dataframe functionality to do useful transformations, such as filtering out the assets without metrics available, then saving that data to a csv file:

import pandas as pd
import os
from coinmetrics.api_client import CoinMetricsClient
from datetime import timedelta
client = CoinMetricsClient(os.environ['CM_API_KEY'])
binance_markets = client.catalog_market_trades_v2(exchange="binance", page_size=1000).to_dataframe()
binance_markets['max_time'] = pd.to_datetime(binance_markets['max_time'], utc=True)
current_utc_time = pd.Timestamp.now(tz='UTC')
one_day_ago = current_utc_time - timedelta(days=1)
filtered_binance_markets = binance_markets[binance_markets['max_time'] > one_day_ago]

Parallel execution for faster data export

There are times when it may be useful to pull in large amounts of data at once. The most effective way to do this when calling the CoinMetrics API is to split your request into many different queries. This functionality is now built into the API Client directly to allow for faster data export:

import os
from coinmetrics.api_client import CoinMetricsClient


if __name__ == '__main__':
    client = CoinMetricsClient(os.environ['CM_API_KEY'])
    binance_eth_markets = [market['market'] for market in client.catalog_market_candles(exchange="binance", base="eth")]
    start_time = "2022-03-01"
    end_time = "2023-05-01"
    client.get_market_candles(markets=binance_eth_markets, start_time=start_time, end_time=end_time, page_size=1000).parallel().export_to_json_files()

What this feature does is rather request all the data in one thread, it will split into many threads or processes and either store them in separate files in the case of .parallel().export_to_csv_files() and .parallel().export_to_json_files or combine them all into one file or data structure in the case of .parallel().to_list(), .parallel().to_dataframe(), .parallel().export_to_json(). It's important to know that in order to send more requests per second to the CoinMetrics this uses the parallel tasks features in Python's concurrent.futures package. This means when using this feature, the API Client will use significantly more resources and may approach the Coin Metrics rate limits.

In terms of resource usage and speed, these usages are in order from most performant to least:

  • .export_to_json_files()
  • .export_to_csv_files()
  • .to_list()
  • .export_to_json()
  • .to_dataframe()

Splitting single parameter queries into many requests for increased performance

There is a feature time_increment that can be used to split a single query into many based on time range, and then combine them later. Consider this example where we speed up getting a 2 months worth of BTC ReferenceRateUSD data into many parallel threads to create a dataframe faster:

import datetime
import os
from coinmetrics.api_client import CoinMetricsClient
from dateutil.relativedelta import relativedelta
client = CoinMetricsClient(os.environ.get("CM_API_KEY"))
start_time = datetime.datetime.now()
assets = ["btc", "eth", "algo"]
if __name__ == '__main__':
    client.get_asset_metrics(
        assets=assets,
        metrics="ReferenceRateUSD",
        frequency="1m",
        start_time="2022-03-01",
        end_time="2023-03-01",
        page_size=1000,
        end_inclusive=False).parallel(
        time_increment=relativedelta(months=1)).export_to_csv("btcRRs.csv")
    print(f"Time taken parallel: {datetime.datetime.now() - start_time}")
    start_time = datetime.datetime.now()
    client.get_asset_metrics(
        assets=assets,
        metrics="ReferenceRateUSD",
        frequency="1m",
        start_time="2022-03-01",
        end_time="2023-03-01",
        page_size=1000,
        end_inclusive=False).export_to_csv("btcRRsNormal.csv")

Notice we pass in the time_increment=relativedelta(months=1) so that means we will split the threads up by month, in addition to by asset. So this will run a total 36 separate threads, 12 threads for each month x 3 threads for each asset. The difference it takes in time is dramatic:

Exporting to dataframe type: 100%|██████████| 36/36 [00:00<00:00, 54.62it/s]
Time taken parallel: 0:00:36.654147
Time taken normal: 0:05:20.073826

Please note that for short time periods you can pass in a time_increment with datetime.timedelta to specify up to several weeks, for larger time frames you can use dateutil.relativedelta.relativedelta in order to split requests up by increments of months or years.

To keep in mind when using using parallel feature or generally writing high performance code using API Client:

  • If you are using a small page_size and trying to export a large number amount of, this will be your biggest bottleneck. Generally the fastest page_size is 1000 to 10000
  • If you are unsure why an action is taking a long time, running the CoinMetricsClient using verbose=True or debug=True can give better insight into what is happening under the hood
  • The parallel feature is best used when you are exporting a large amount of data, that can be split by query params into many smaller requests. A good example of this is market candles over a long time frame - if you are querying hundreds of markets and are sure there will be data, using .parallel().export_to_csv_files("...") can have a huge performance increase, if you are just querying a single market you will not see a difference
  • The parallel feature is highly configurable, there is several configuration options that may be suitable for advanced users like tweaking the max_workers parameter, or changing the default ProcessPoolExecutor to a ThreadPoolExectuor
  • Using multithreaded code is inherently more complex, it will be harder to debug issues with long running queries when running parallel exports compared to normal single threaded code
  • For that reason, this tool is best suited for exporting historical data rather than supporting a real time production system
  • The methods that create separate files for each thread will be the safest and most performant to use - .export_to_csv_files() and .export_to_json_files(). Using the methods that return a single output - .export_to_csv(), export_to_list(), and .export_to_dataframe() need to join the data from many threads before it can be returned, this may use a lot of memory if you are accessing data types like market orderbooks or market trades and could fail altogether
  • If using export_to_csv/json_files() functions, note that by default they will be saved in the directory format /{endpoint}/{parallelize_on}. For example, in export_to_json_files(), client.get_market_trades("coinbase-eth-btc-spot,coinbase-eth-usdc-spot").parallel("markets") will create a file each like ./market-trades/coinbase-eth-btc-spot.json, ./market-trades/coinbase-eth-usdc-spot.json client.get_asset_metrics('btc,eth', 'ReferenceRateUSD', start_time='2024-01-01', limit_per_asset=1).parallel("assets,metrics", time_increment=timedelta(days=1)) will create a file each like ./asset-metrics/btc/ReferenceRateUSD/start_time=2024-01-01T00-00-00Z.json, ./asset-metrics/eth/ReferenceRateUSD/start_time=2024-01-01T00-00-00Z.json
  • If you get the error BrokenProcessPool it might be because you're missing a main() function

Examples

The API Client allows you to chain together workflows for importing, transforming, then exporting Coin Metrics data. Below are examples of common use-cases that can be altered to tailor your specific needs. In addition to the examples listed below, there's examples covering all the API methods, found in the examples directory.

Example Notebooks

  • walkthrough_community.ipynb: Walks through the basic functionality available using the community client.

Asset Metrics

  • bbb_metrics_csv_exporter_using_plain_requests.py: Queries block-by-block metrics using the requests library and exports the output into a CSV file.
  • bbb_metrics_json_exporter.py: Queries block-by-block metrics and exports the output into a JSON file.
  • eod_metrics_csv_exporter.py: Exports a set of user-defined metrics and assets published at end-of-day and exports the output into a CSV file.
  • reference_rates_json_exporter.py: Queries Coin Metrics Reference Rates at a user-defined frequency for a set of assets, then exports the output into a JSON file.

Market Data

  • books_json_exporter.py: Queries market orderbook data then exports the output into a JSON file.
  • candles_json_exporter.py: Queries market candles data then exports the output into a JSON file.
  • funding_rates_json_exporter.py: Queries market funding rates data then exports the output into a JSON file.
  • trades_csv_exporter.py: Queries market trades data then exports the output into a CSV file.
  • trades_json_exporter.py: Queries market trades data then exports the output into a JSON file.

**Parallel processing exports

  • candles_csv_export.py: Exports market candles in parallel to many separate csv files
  • candles_json_export.py: Exports market candles in parallel to many separate json files
  • market_trades_list.py: Creates a list of market trades, using .parallel() feature to improve performance
  • market_orderbooks.py: Exports market orderbooks to many csv files
  • candles_csv_export_manual.py: Example of parallelism using the API Client without using the .parallel() feature
  • btc_1m_metrics_export.py: Example of splitting a large request for asset metrics by metric to improve performance, exporting a single csv and also separate csv.
  • market_orderbooks_csv_exporter_by_day.py: Example of splitting a market orderbook export up by day, to increase export performance

Getting timeseries data

For getting timeseries data you want to use methods of the client class that start with get_. It's important to note that the timeseries endpoints return data of a parent class type DataCollection. The DataCollection class is meant to support a variety of different data export and data manipulation use cases, so just calling one of the client methods such as data = client.get_market_trades(markets="coinbase-btc-usd-spot") will not actually retrieve the data related to this API call. You must then call a function on this DataCollection such as data.export_to_csv("coinbase_btc_usd_spot_trades.csv) or data.to_dataframe() in order to access the data. There is more explicit examples below.If you are curious to see how the API calls are being made and with what parameters, instantiating the client with the verbose argument like CoinMetricsClient(api_key=<YOUR_API_KEY>, verbose=True) will print the API calls as well as information on performance to console.

For example if you want to get a bunch of market data trades for coinbase btc-usd pair you can run something similar to the following:

for trade in client.get_market_trades(
    markets='coinbase-btc-usd-spot', 
    start_time='2020-01-01', 
    end_time='2020-01-03',
    limit_per_market=10
):
    print(trade)

This example uses the DataCollection as a Python iterator, so with each iteration of the Python for loop it will call the Coin Metrics API and return data. The default page_size for calls to the API is 100, so each call will return 100 trades until it reaches the end of the query. To get more trades in each API call, you can add the parameter page_size to the .get_market_trades() method call, up to a maximum of 10000.

Or if you want to see daily btc asset metrics you can use something like this:

for metric_data in client.get_asset_metrics(assets='btc', 
                                            metrics=['ReferenceRateUSD', 'BlkHgt', 'AdrActCnt',  
                                                     'AdrActRecCnt', 'FlowOutBFXUSD'], 
                                            frequency='1d',
                                            limit_per_asset=10):
    print(metric_data)

This will print you the requested metrics for all the days where we have any of the metrics present.

DataFrames

(New in >=2021.9.30.14.30)

Timeseries data can be transformed into a pandas dataframe by using the to_dataframe() method. The code snippet below shows how:

import pandas as pd
from coinmetrics.api_client import CoinMetricsClient
from os import environ

client = CoinMetricsClient()
trades = client.get_market_trades(
    markets='coinbase-btc-usd-spot', 
    start_time='2021-09-19T00:00:00Z', 
    limit_per_market=10
)
trades_df = trades.to_dataframe()
print(trades_df.head())

If you want to use dataframes, then you will need to install pandas

Notes

  • This only works with requests that return the type DataCollection. Thus, catalog requests, which return lists cannot be returned as dataframes. Please see the API Client Spec for a full list of requests and their return types.
  • API restrictions apply. Some requests may return empty results due to limited access to the API from you API key.

Type Conversion

(New in >=2021.12.17.18.00)

As of version 2021.12.17.18.00 or later, outputs from the to_dataframe function automatically convert the dtypes for a dataframe to the optimal pandas types.

metrics_list = ['volume_trusted_spot_usd_1d', 'SplyFF', 'AdrBalUSD1Cnt']
asset_list = ['btc','xmr']
start_time='2021-12-01'
df_metrics = client.get_asset_metrics(
  assets=asset_list, metrics=metrics_list, start_time=start_time, limit_per_asset=3
).to_dataframe()
print(df_metrics.dtypes)
asset                          string
time                           datetime64[ns, tzutc()]
AdrBalUSD1Cnt                   Int64
SplyFF                        Float64
volume_trusted_spot_usd_1d    Float64
dtype: object

This can be turned off by setting optimize_pandas_types=False

Alternatively, you can manually enter your own type conversion by passing in a dictionary for dtype_mapper. This can be done in conjunction with pandas' built in type optimizations.

mapper = {
  'SplyFF': 'Float64',
  'AdrBalUSD1Cnt': 'Int64',
}
df_mapped = client.get_asset_metrics(
  assets=asset_list, metrics=metrics_list, start_time=start_time, limit_per_asset=3
).to_dataframe(dtype_mapper=mapper, optimize_pandas_types=True)
print(df_mapped.dtypes)
asset                                          object
time                          datetime64[ns, tzutc()]
AdrBalUSD1Cnt                                   Int64
SplyFF                                        Float64
volume_trusted_spot_usd_1d                    float64
dtype: object

Or as strictly the only types in the dataframe

dtype_mapper = {
    'ReferenceRateUSD': np.float64,
    'time': np.datetime64
}
df = client.get_asset_metrics(
  assets='btc', metrics='ReferenceRateUSD', start_time='2022-06-15', limit_per_asset=1
).to_dataframe(dtype_mapper=dtype_mapper, optimize_pandas_types=False)
df.info()
RangeIndex: 1 entries, 0 to 0
Data columns (total 3 columns):
 #   Column            Non-Null Count  Dtype         
---  ------            --------------  -----         
 0   asset             1 non-null      object        
 1   time              1 non-null      datetime64[ns]
 2   ReferenceRateUSD  1 non-null      float64       
dtypes: datetime64[ns](1), float64(1), object(1)
memory usage: 152.0+ bytes

Note that in order to pass a custom datetime object, setting a dtype_mapper is mandatory.

Pandas type conversion tends to be more performant. But if there are custom operations that must be done using numpy datatypes, this option will let you perform them.

Exporting to csv and json files:

You can also easily export timeseries data to csv and json files with builtin functions on the DataCollection type. For example this script will export Coinbase btc and eth trades for a date to csv and json files respectively:

    start_date = datetime.date(year=2022, month=1, day=1)
    end_date = datetime.datetime(year=2022, month=1, day=1)
    market_trades_btc = client.get_market_trades(page_size=1000, markets="coinbase-btc-usd-spot", start_time=start_date, end_time=end_date)
    market_trades_btc.export_to_csv("jan_1_2022_coinbase_btc_trades.csv")
    market_trades_eth = client.get_market_trades(page_size=1000, markets="coinbase-eth-usd-spot", start_time=start_date, end_time=end_date)
    market_trades_eth.export_to_json("jan_1_2022_coinbase_eth.json")

Paging

You can make the datapoints to iterate from start (default) or from end.

for that you should use a paging_from argument like the following:

from coinmetrics.api_client import CoinMetricsClient
from coinmetrics.constants import PagingFrom

client = CoinMetricsClient()

for metric_data in client.get_asset_metrics(assets='btc', metrics=['ReferenceRateUSD'],
                                            paging_from=PagingFrom.START):
    print(metric_data)

PagingFrom.END: is available but by default it will page from the start.

Debugging the API Client

There are two additional options for the API Client - debug_mode and verbose. These two options log network calls to the console, and in the case of debug_mode it will generate a log file of all the network requests and the time it takes to call them. These tools can be used to diagnose issues in your code and also to get a better understanding of request times so that users can write more performant code. For example, running the below code:

import os

from coinmetrics.api_client import CoinMetricsClient

api_key = os.environ['CM_API_KEY']

if __name__ == '__main__':
    client = CoinMetricsClient(api_key=api_key, debug_mode=True)
    reference_rates_example = client.get_asset_metrics(assets=['btc', 'algo', 'eth'], metrics=['ReferenceRateUSD'])
    for data in reference_rates_example:
        continue

The console output will look like:

[DEBUG] 2023-01-09 11:01:02,044 - Starting API Client debugging session. logging to stdout and cm_api_client_debug_2023_01_09_11_01_02.txt
[DEBUG] 2023-01-09 11:01:02,044 - Using coinmetrics version 2022.11.14.16
[DEBUG] 2023-01-09 11:01:02,044 - Current state of API Client, excluding API KEY: {'_verify_ssl_certs': True, '_api_base_url': 'https://api.coinmetrics.io/v4', '_ws_api_base_url': 'wss://api.coinmetrics.io/v4', '_http_header': {'Api-Client-Version': '2022.11.14.16'}, '_proxies': {'http': None, 'https': None}, 'debug_mode': True, 'verbose': False}
[DEBUG] 2023-01-09 11:01:02,044 - Attempting to call url: timeseries/asset-metrics with params: {'assets': ['btc', 'algo', 'eth'], 'metrics': ['ReferenceRateUSD'], 'frequency': None, 'page_size': None, 'paging_from': 'start', 'start_time': None, 'end_time': None, 'start_height': None, 'end_height': None, 'start_inclusive': None, 'end_inclusive': None, 'timezone': None, 'sort': None, 'limit_per_asset': None}
[DEBUG] 2023-01-09 11:01:02,387 - Response status code: 200 for url: https://api.coinmetrics.io/v4/timeseries/asset-metrics?api_key=[REDACTED]&assets=btc%2Calgo%2Ceth&metrics=ReferenceRateUSD&paging_from=start took: 0:00:00.342874 response body size (bytes): 9832
[DEBUG] 2023-01-09 11:01:02,388 - Attempting to call url: timeseries/asset-metrics with params: {'assets': ['btc', 'algo', 'eth'], 'metrics': ['ReferenceRateUSD'], 'frequency': None, 'page_size': None, 'paging_from': 'start', 'start_time': None, 'end_time': None, 'start_height': None, 'end_height': None, 'start_inclusive': None, 'end_inclusive': None, 'timezone': None, 'sort': None, 'limit_per_asset': None, 'next_page_token': '0.MjAxOS0wOS0zMFQwMDowMDowMFo'}
[DEBUG] 2023-01-09 11:01:02,559 - Response status code: 200 for url: https://api.coinmetrics.io/v4/timeseries/asset-metrics?api_key=[REDACTED]&assets=btc%2Calgo%2Ceth&metrics=ReferenceRateUSD&paging_from=start&next_page_token=0.MjAxOS0wOS0zMFQwMDowMDowMFo took: 0:00:00.171487 response body size (bytes): 9857

Then it can be easier to understand what network calls the API Client is making, and where any issues may exist. If you wish to dig even deeper, you may consider modifying the _send_request() method of the API Client to log additional data about the state of your environment, or anything else that would help diagnose issues. You will notice a log file generated in the format cm_api_client_debug_2023_01_09_11_01_02.txt. This log file might be helpful for your own use or to give more context if you are working with Coin Metrics customer success.

SSL Certs verification

Sometimes your organization network have special rules on SSL certs verification and in this case you might face the following error when running the script:

SSLError: HTTPSConnectionPool(host='api.coinmetrics.io', port=443): Max retries exceeded with url: <some_url_path> (Caused by SSLError(SSLCertVerificationError(1, '[SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed: self signed certificate in certificate chain (_ssl.c:1123)')))

In this case, you can pass an option during client initialization to disable ssl verification for requests like this:

client = CoinMetricsClient(verify_ssl_certs=False)

We don't recommend setting it to False by default and you should make sure you understand the security risks of disabling SSL certs verification.

Additionally, you may choose to specify the path to the SSL certificates on your machine. This may cause errors where Python is unable to locate the certificates on your machine, particularly when using Python virtual environments.

from coinmetrics.api_client import CoinMetricsClient
SSL_CERT_LOCATION = '/Users/<USER_NAME>/Library/Python/3.8/lib/python/site-packages/certifi/cacert.pem'
client = CoinMetricsClient(verify_ssl_certs=SSL_CERT_LOCATION)

A quick way to find the certs on your machine is:
python3 -c "import requests; print(requests.certs.where())"
And note that this will change based on whether or not you are using a Python virtual environment or not

Installing and running coinmetrics package and other python packages behind a secure python network

Related to SSL Certs verification, you may have trouble installing and updating PyPi packages to your local environment. So you may need to choose the best solution for your company and environment - either using package managers or installing offline.

Installing using package managers

Full instructions for setting up your environment to use conda, pip, yarn, npm, etc. can be found here. Additionally, a workaround to disable SSL verification when installing a trusted Python package is this:

pip install --trusted-host pypi.python.org <packagename>

Although it is important to make sure you understand the risks associated with disabling SSL verification and ensure compliance with company policies.

Installing Python packages locally/ offline

It may be easier to download and install the package locally. Steps:

  1. Download the files for the Coin Metrics API Client from PyPi
  2. Install it locally

Requests Proxy

Sometimes your organization has special rules on making requests to third parties and you have to use proxies in order to comply with the rules.

For proxies that don't require auth you can specify them similar to this example:

client = CoinMetricsClient(proxy_url=f'http://<hostname>:<port>')

For proxies that require auth, you should be able to specify username and password similar to this example:

client = CoinMetricsClient(proxy_url=f'http://<username>:<password>@<hostname>:<port>')

Extended documentation

For more information about the available methods in the client please reference API Client Spec