DataCollection.parallel

coinmetrics._data_collection.DataCollection.parallel(
    parallelize_on=None,
    executor=None,
    max_workers=None,
    progress_bar=None,
    time_increment=None,
    height_increment=None,
)

Convert this DataCollection into a ParallelDataCollection, splitting a single HTTP request into many parallel requests for faster data export.

By default the request is split on the primary query parameter (for example, get_asset_metrics(assets=...) is split into one request per asset).

• Parameters:

  • parallelize_on (Optional *[*Union [str , List [str ] ] ]) -- Parameter(s) to parallelize on. Must be one of the list-type parameters of the underlying endpoint (assets, markets, metrics, ...).

  • executor (Optional *[*Callable [ *[*Any ] , concurrent.futures.Executor ] ]) -- Executor class used for concurrency. Defaults to concurrent.futures.ThreadPoolExecutor; pass a ProcessPoolExecutor (or any other Executor subclass) to swap it out.

  • max_workers (Optional [int ]) -- Number of parallel workers. Defaults to 10 and is capped at 10 to respect API rate limits.

  • progress_bar (Optional [bool ]) -- Whether to display a tqdm progress bar while the workers run. Defaults to True.

  • time_increment (Optional *[*Union *[*relativedelta , timedelta , DateOffset ] ]) -- Optionally split the request along the time axis as well. Use datetime.timedelta for sub-month windows and dateutil.relativedelta.relativedelta for month / year windows. Requires start_time to be set on the original request.

  • height_increment (Optional [int ]) -- Optionally split the request along the block-height axis. Requires start_height to be set on the original request.

• Returns: A ParallelDataCollection that mirrors the original request but executes its workload across multiple workers.

Returns:

• ParallelDataCollection