Creates an XML file with a given number of prefixed XML attributes on a single XML tag.

write_many_atts.py

# Copyright (c) 2021 Sebastian Pipping <sebastian@pipping.org>
# Licensed under the Apache license version 2.0
#
# Creates an XML file with a given number of prefixed XML attributes
# on a single XML tag.
# Needs Python >=3.6 and PyPI package "base58"
#
# 2021-12-31 00:47 UTC+1

import argparse
import math

import base58

_BATCH_SIZE = 2**13


def _to_name(i: int) -> str:
    bites = []
    while i >= 256:
        bite = i % 256
        i //= 256
        bites.append(bite)
    bites.append(i)
    return base58.b58encode(bytes(bites))


def _validate_attributes_count(candidate: str) -> int:
    res = int(candidate)  # may throw
    if res <= 0 or res % _BATCH_SIZE != 0:
        raise ValueError(f'not a multiple of batch size {_BATCH_SIZE}')
    return res

_validate_attributes_count.__name__ = 'attributes count'   # for argparse


if __name__ == '__main__':
    parser = argparse.ArgumentParser()
    parser.add_argument('attributes_count', metavar='COUNT',
                        type=_validate_attributes_count,
                        help=('number of XML attributes to write'
                            ' excluding the namespace declaration'
                            ', must be a multiple of 2^13'
                            '; e.g. try $((2**15)) in Bash'))
    parser.add_argument('output_filename', metavar='FILENAME',
                        help='filename of the file to write to')
    config = parser.parse_args()

    runs_count = config.attributes_count // _BATCH_SIZE

    with open(config.output_filename, 'bw') as f:
        f.write(b'<r xmlns:a="http://whatever.invalid/"')
        for run in range(runs_count):
            f.write(b'\n')
            print(f'Run {run+1} of {runs_count}'
                  f': {run * _BATCH_SIZE + 1} to {(run + 1) * _BATCH_SIZE}')
            f.write(b'\n'.join(
                b'a:a' + _to_name(run * _BATCH_SIZE + i) + b'=""'
                for i in range(_BATCH_SIZE)))
            f.flush()
        f.write(b'\n/>')

    print('Done.')