michael@0: #!/usr/bin/python -B
michael@0: 
michael@0: """ Usage: make-normalize-generateddata-input.py PATH_TO_MOZILLA_CENTRAL
michael@0: 
michael@0:     This script generates test input data for String.prototype.normalize
michael@0:     from intl/icu/source/data/unidata/NormalizationTest.txt
michael@0:     to js/src/tests/ecma_6/String/normalize-generateddata-input.js
michael@0: """
michael@0: 
michael@0: from __future__ import print_function
michael@0: import re, sys
michael@0: 
michael@0: sep_pat = re.compile(' +')
michael@0: def to_code_list(codes):
michael@0:     return '[' + ', '.join(map(lambda x: '0x{0}'.format(x), re.split(sep_pat, codes))) + ']'
michael@0: 
michael@0: def convert(dir):
michael@0:     ver_pat = re.compile('NormalizationTest-([0-9\.]+)\.txt')
michael@0:     part_pat = re.compile('^@(Part([0-9]+) .+)$')
michael@0:     test_pat = re.compile('^([0-9A-Fa-f ]+);([0-9A-Fa-f ]+);([0-9A-Fa-f ]+);([0-9A-Fa-f ]+);([0-9A-Fa-f ]+);$')
michael@0:     ignore_pat = re.compile('^#|^$')
michael@0:     js_path = 'js/src/tests/ecma_6/String/normalize-generateddata-input.js'
michael@0:     txt_path = 'intl/icu/source/data/unidata/NormalizationTest.txt'
michael@0: 
michael@0:     part_opened = False
michael@0:     not_empty = False
michael@0:     with open('{dir}/{path}'.format(dir=dir, path=txt_path), 'r') as f:
michael@0:         with open('{dir}/{path}'.format(dir=dir, path=js_path), 'w') as outf:
michael@0:             for line in f:
michael@0:                 m = test_pat.search(line)
michael@0:                 if m:
michael@0:                     if not_empty:
michael@0:                         outf.write(',')
michael@0:                     outf.write('\n')
michael@0:                     pat = '{{ source: {source}, NFC: {NFC}, NFD: {NFD}, NFKC: {NFKC}, NFKD: {NFKD} }}'
michael@0:                     outf.write(pat.format(source=to_code_list(m.group(1)),
michael@0:                                           NFC=to_code_list(m.group(2)),
michael@0:                                           NFD=to_code_list(m.group(3)),
michael@0:                                           NFKC=to_code_list(m.group(4)),
michael@0:                                           NFKD=to_code_list(m.group(5))))
michael@0:                     not_empty = True
michael@0:                     continue
michael@0:                 m = part_pat.search(line)
michael@0:                 if m:
michael@0:                     desc = m.group(1)
michael@0:                     part = m.group(2)
michael@0:                     if part_opened:
michael@0:                         outf.write('\n];\n')
michael@0:                     outf.write('/* {desc} */\n'.format(desc=desc))
michael@0:                     outf.write('var tests_part{part} = ['.format(part=part))
michael@0:                     part_opened = True
michael@0:                     not_empty = False
michael@0:                     continue
michael@0:                 m = ver_pat.search(line)
michael@0:                 if m:
michael@0:                     ver = m.group(1)
michael@0:                     outf.write('/* created from NormalizationTest-{ver}.txt */\n'.format(ver=ver))
michael@0:                     continue
michael@0:                 m = ignore_pat.search(line)
michael@0:                 if m:
michael@0:                     continue
michael@0:                 print("Unknown line: {0}".format(line), file=sys.stderr)
michael@0:             if part_opened:
michael@0:                 outf.write('\n];\n')
michael@0: 
michael@0: if __name__ == '__main__':
michael@0:     if len(sys.argv) < 2:
michael@0:         print("Usage: make-normalize-generateddata-input.py PATH_TO_MOZILLA_CENTRAL", file=sys.stderr)
michael@0:         sys.exit(1)
michael@0:     convert(sys.argv[1])