Chris@0: #! /usr/bin/env python3
Chris@0: import sys
Chris@0: from os.path import abspath, expanduser, dirname, join
Chris@0: from itertools import chain
Chris@0: import json
Chris@0: import argparse
Chris@0: 
Chris@0: from vis import vis, unvis, VIS_WHITE
Chris@0: 
Chris@0: 
Chris@0: __dir__ = dirname(abspath(__file__))
Chris@0: 
Chris@0: OUTPUT_FILE = join(__dir__, '..', 'fixtures', 'unvis_fixtures.json')
Chris@0: 
Chris@0: # Add custom fixtures here
Chris@0: CUSTOM_FIXTURES = [
Chris@0:     # test long multibyte string
Chris@0:     ''.join(chr(cp) for cp in range(1024)),
Chris@0:     'foo bar',
Chris@0:     'foo\nbar',
Chris@0:     "$bar = 'baz';",
Chris@0:     r'$foo = "\x20\\x20\\\x20\\\\x20"',
Chris@0:     '$foo = function($bar) use($baz) {\n\treturn $baz->getFoo()\n};'
Chris@0: ]
Chris@0: 
Chris@0: RANGES = {
Chris@0:     # All valid codepoints in the BMP
Chris@0:     'bmp': chain(range(0x0000, 0xD800), range(0xE000, 0xFFFF)),
Chris@0:     # Smaller set of pertinent? codepoints inside BMP
Chris@0:     # see: http://en.wikipedia.org/wiki/Plane_(Unicode)#Basic_Multilingual_Plane
Chris@0:     'small': chain(
Chris@0:         # latin blocks
Chris@0:         range(0x0000, 0x0250),
Chris@0:         # Greek, Cyrillic
Chris@0:         range(0x0370, 0x0530),
Chris@0:         # Hebrew, Arabic
Chris@0:         range(0x590, 0x0700),
Chris@0:         # CJK radicals
Chris@0:         range(0x2E80, 0x2F00),
Chris@0:         # Hiragana, Katakana
Chris@0:         range(0x3040, 0x3100)
Chris@0:     )
Chris@0: }
Chris@0: 
Chris@0: 
Chris@0: if __name__ == '__main__':
Chris@0: 
Chris@0:     argp = argparse.ArgumentParser(
Chris@0:         description='Generates test data for Psy\\Test\\Util\\StrTest')
Chris@0:     argp.add_argument('-f', '--format-output', action='store_true',
Chris@0:                       help='Indent JSON output to ease debugging')
Chris@0:     argp.add_argument('-a', '--all', action='store_true',
Chris@0:                       help="""Generates test data for all codepoints of the BMP.
Chris@0:                       (same as --range=bmp). WARNING: You will need quite
Chris@0:                       a lot of RAM to run the testsuite !
Chris@0:                       """)
Chris@0:     argp.add_argument('-r', '--range',
Chris@0:                       help="""Choose the range of codepoints used to generate
Chris@0:                       test data.""",
Chris@0:                       choices=list(RANGES.keys()),
Chris@0:                       default='small')
Chris@0:     argp.add_argument('-o', '--output-file',
Chris@0:                       help="""Write test data to OUTPUT_FILE
Chris@0:                       (defaults to PSYSH_DIR/test/fixtures)""")
Chris@0:     args = argp.parse_args()
Chris@0: 
Chris@0:     cp_range = RANGES['bmp'] if args.all else RANGES[args.range]
Chris@0:     indent = 2 if args.format_output else None
Chris@0:     if args.output_file:
Chris@0:         OUTPUT_FILE = abspath(expanduser(args.output_file))
Chris@0: 
Chris@0:     fixtures = []
Chris@0: 
Chris@0:     # use SMALL_RANGE by default, it should be enough.
Chris@0:     # use BMP_RANGE for a more complete smoke test
Chris@0:     for codepoint in cp_range:
Chris@0:         char = chr(codepoint)
Chris@0:         encoded = vis(char, VIS_WHITE)
Chris@0:         decoded = unvis(encoded)
Chris@0:         fixtures.append((encoded, decoded))
Chris@0: 
Chris@0:     # Add our own custom fixtures at the end,
Chris@0:     # since they would fail anyway if one of the previous did.
Chris@0:     for fixture in CUSTOM_FIXTURES:
Chris@0:         encoded = vis(fixture, VIS_WHITE)
Chris@0:         decoded = unvis(encoded)
Chris@0:         fixtures.append((encoded, decoded))
Chris@0: 
Chris@0:     with open(OUTPUT_FILE, 'w') as fp:
Chris@0:         # dump as json to avoid backslashin and quotin nightmare
Chris@0:         # between php and python
Chris@0:         json.dump(fixtures, fp, indent=indent)
Chris@0: 
Chris@0:     sys.exit(0)