pyarrow#
PyArrow contrib functionality.
Class ParquetFormatter#
Class ParquetTableFormatter#
Functions#
- merge_parquet_files(src_paths, dst_path, force=True, callback=None, writer_opts=None, copy_single=False, skip_empty=True, target_row_group_size=0)#
Merges parquet files in src_paths into a new file at dst_path. Intermediate directories are created automatically. When dst_path exists and force is True, the file is removed first. Otherwise, an exception is thrown.
callback can refer to a callable accepting a single integer argument representing the index of the file after it was merged. writer_opts can be a dictionary of keyword arguments that are passed to the ParquetWriter instance. When src_paths contains only a single file and copy_single is True, the file is copied to dst_path and no merging takes place. Files containing empty tables are skipped unless skip_empty is False.
When target_row_group_size is a positive number, the merging is done on the level of particular row groups. These groups are merged in-memory such that each resulting group stored on disk, potentially except for the last one, will target_row_group_size rows.
The absolute, expanded dst_path is returned.
- merge_parquet_task(task, inputs, output, local=False, cwd=None, force=True, **kwargs)#
This method is intended to be used by tasks that are supposed to merge parquet files, e.g. when inheriting from
law.contrib.tasks.MergeCascade. inputs should be a sequence of targets that represent the files to merge into output.When local is False and files need to be copied from remote first, cwd can be a set as the dowload directory. When empty, a temporary directory is used. The task itself is used to print and publish messages via its
law.Task.publish_message()andlaw.Task.publish_step()methods. When force is True, any existing output file is overwritten.All additional kwargs are forwarded to
merge_parquet_files()which is used internally for the actual merging.