|
Packit |
6c4009 |
#!/usr/bin/python3
|
|
Packit |
6c4009 |
#
|
|
Packit |
6c4009 |
# Generate a Unicode conforming LC_CTYPE category from a UnicodeData file.
|
|
Packit |
6c4009 |
# Copyright (C) 2014-2018 Free Software Foundation, Inc.
|
|
Packit |
6c4009 |
# This file is part of the GNU C Library.
|
|
Packit |
6c4009 |
# Based on gen-unicode-ctype.c by Bruno Haible <haible@clisp.cons.org>, 2000.
|
|
Packit |
6c4009 |
#
|
|
Packit |
6c4009 |
# The GNU C Library is free software; you can redistribute it and/or
|
|
Packit |
6c4009 |
# modify it under the terms of the GNU Lesser General Public
|
|
Packit |
6c4009 |
# License as published by the Free Software Foundation; either
|
|
Packit |
6c4009 |
# version 2.1 of the License, or (at your option) any later version.
|
|
Packit |
6c4009 |
#
|
|
Packit |
6c4009 |
# The GNU C Library is distributed in the hope that it will be useful,
|
|
Packit |
6c4009 |
# but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
Packit |
6c4009 |
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
Packit |
6c4009 |
# Lesser General Public License for more details.
|
|
Packit |
6c4009 |
#
|
|
Packit |
6c4009 |
# You should have received a copy of the GNU Lesser General Public
|
|
Packit |
6c4009 |
# License along with the GNU C Library; if not, see
|
|
Packit |
6c4009 |
# <http://www.gnu.org/licenses/>.
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
'''
|
|
Packit |
6c4009 |
Generate a Unicode conforming LC_CTYPE category from UnicodeData.txt and
|
|
Packit |
6c4009 |
DerivedCoreProperties.txt files.
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
To see how this script is used, call it with the “-h” option:
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
$ ./gen_unicode_ctype.py -h
|
|
Packit |
6c4009 |
… prints usage message …
|
|
Packit |
6c4009 |
'''
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
import argparse
|
|
Packit |
6c4009 |
import time
|
|
Packit |
6c4009 |
import re
|
|
Packit |
6c4009 |
import unicode_utils
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
def code_point_ranges(is_class_function):
|
|
Packit |
6c4009 |
'''Returns a list of ranges of code points for which is_class_function
|
|
Packit |
6c4009 |
returns True.
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
Example:
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
[[65, 90], [192, 214], [216, 222], [256], … ]
|
|
Packit |
6c4009 |
'''
|
|
Packit |
6c4009 |
cp_ranges = []
|
|
Packit |
6c4009 |
for code_point in sorted(unicode_utils.UNICODE_ATTRIBUTES):
|
|
Packit |
6c4009 |
if is_class_function(code_point):
|
|
Packit |
6c4009 |
if (cp_ranges
|
|
Packit |
6c4009 |
and cp_ranges[-1][-1] == code_point - 1):
|
|
Packit |
6c4009 |
if len(cp_ranges[-1]) == 1:
|
|
Packit |
6c4009 |
cp_ranges[-1].append(code_point)
|
|
Packit |
6c4009 |
else:
|
|
Packit |
6c4009 |
cp_ranges[-1][-1] = code_point
|
|
Packit |
6c4009 |
else:
|
|
Packit |
6c4009 |
cp_ranges.append([code_point])
|
|
Packit |
6c4009 |
return cp_ranges
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
def output_charclass(i18n_file, class_name, is_class_function):
|
|
Packit |
6c4009 |
'''Output a LC_CTYPE character class section
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
Example:
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
upper /
|
|
Packit |
6c4009 |
<U0041>..<U005A>;<U00C0>..<U00D6>;<U00D8>..<U00DE>;<U0100>;<U0102>;/
|
|
Packit |
6c4009 |
…
|
|
Packit |
6c4009 |
<U0001D790>..<U0001D7A8>;<U0001D7CA>;<U0001F130>..<U0001F149>;/
|
|
Packit |
6c4009 |
<U0001F150>..<U0001F169>;<U0001F170>..<U0001F189>
|
|
Packit |
6c4009 |
'''
|
|
Packit |
6c4009 |
cp_ranges = code_point_ranges(is_class_function)
|
|
Packit |
6c4009 |
if cp_ranges:
|
|
Packit |
6c4009 |
i18n_file.write('%s /\n' %class_name)
|
|
Packit |
6c4009 |
max_column = 75
|
|
Packit |
6c4009 |
prefix = ' '
|
|
Packit |
6c4009 |
line = prefix
|
|
Packit |
6c4009 |
range_string = ''
|
|
Packit |
6c4009 |
for code_point_range in cp_ranges:
|
|
Packit |
6c4009 |
if line.strip():
|
|
Packit |
6c4009 |
line += ';'
|
|
Packit |
6c4009 |
if len(code_point_range) == 1:
|
|
Packit |
6c4009 |
range_string = unicode_utils.ucs_symbol(code_point_range[0])
|
|
Packit |
6c4009 |
else:
|
|
Packit |
6c4009 |
range_string = unicode_utils.ucs_symbol_range(
|
|
Packit |
6c4009 |
code_point_range[0], code_point_range[-1])
|
|
Packit |
6c4009 |
if len(line+range_string) > max_column:
|
|
Packit |
6c4009 |
i18n_file.write(line+'/\n')
|
|
Packit |
6c4009 |
line = prefix
|
|
Packit |
6c4009 |
line += range_string
|
|
Packit |
6c4009 |
if line.strip():
|
|
Packit |
6c4009 |
i18n_file.write(line+'\n')
|
|
Packit |
6c4009 |
i18n_file.write('\n')
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
def output_charmap(i18n_file, map_name, map_function):
|
|
Packit |
6c4009 |
'''Output a LC_CTYPE character map section
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
Example:
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
toupper /
|
|
Packit |
6c4009 |
(<U0061>,<U0041>);(<U0062>,<U0042>);(<U0063>,<U0043>);(<U0064>,<U0044>);/
|
|
Packit |
6c4009 |
…
|
|
Packit |
6c4009 |
(<U000118DC>,<U000118BC>);(<U000118DD>,<U000118BD>);/
|
|
Packit |
6c4009 |
(<U000118DE>,<U000118BE>);(<U000118DF>,<U000118BF>)
|
|
Packit |
6c4009 |
'''
|
|
Packit |
6c4009 |
max_column = 75
|
|
Packit |
6c4009 |
prefix = ' '
|
|
Packit |
6c4009 |
line = prefix
|
|
Packit |
6c4009 |
map_string = ''
|
|
Packit |
6c4009 |
i18n_file.write('%s /\n' %map_name)
|
|
Packit |
6c4009 |
for code_point in sorted(unicode_utils.UNICODE_ATTRIBUTES):
|
|
Packit |
6c4009 |
mapped = map_function(code_point)
|
|
Packit |
6c4009 |
if code_point != mapped:
|
|
Packit |
6c4009 |
if line.strip():
|
|
Packit |
6c4009 |
line += ';'
|
|
Packit |
6c4009 |
map_string = '(' \
|
|
Packit |
6c4009 |
+ unicode_utils.ucs_symbol(code_point) \
|
|
Packit |
6c4009 |
+ ',' \
|
|
Packit |
6c4009 |
+ unicode_utils.ucs_symbol(mapped) \
|
|
Packit |
6c4009 |
+ ')'
|
|
Packit |
6c4009 |
if len(line+map_string) > max_column:
|
|
Packit |
6c4009 |
i18n_file.write(line+'/\n')
|
|
Packit |
6c4009 |
line = prefix
|
|
Packit |
6c4009 |
line += map_string
|
|
Packit |
6c4009 |
if line.strip():
|
|
Packit |
6c4009 |
i18n_file.write(line+'\n')
|
|
Packit |
6c4009 |
i18n_file.write('\n')
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
def read_input_file(filename):
|
|
Packit |
6c4009 |
'''Reads the original glibc i18n file to get the original head
|
|
Packit |
6c4009 |
and tail.
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
We want to replace only the character classes in LC_CTYPE, and the
|
|
Packit |
6c4009 |
date stamp. All the rest of the i18n file should stay unchanged.
|
|
Packit |
6c4009 |
To avoid having to cut and paste the generated data into the
|
|
Packit |
6c4009 |
original file, it is helpful to read the original file here
|
|
Packit |
6c4009 |
to be able to generate a complete result file.
|
|
Packit |
6c4009 |
'''
|
|
Packit |
6c4009 |
head = tail = ''
|
|
Packit |
6c4009 |
with open(filename, mode='r') as i18n_file:
|
|
Packit |
6c4009 |
for line in i18n_file:
|
|
Packit |
6c4009 |
match = re.match(
|
|
Packit |
6c4009 |
r'^(?P<key>date\s+)(?P<value>"[0-9]{4}-[0-9]{2}-[0-9]{2}")',
|
|
Packit |
6c4009 |
line)
|
|
Packit |
6c4009 |
if match:
|
|
Packit |
6c4009 |
line = match.group('key') \
|
|
Packit |
6c4009 |
+ '"{:s}"\n'.format(time.strftime('%Y-%m-%d'))
|
|
Packit |
6c4009 |
head = head + line
|
|
Packit |
6c4009 |
if line.startswith('LC_CTYPE'):
|
|
Packit |
6c4009 |
break
|
|
Packit |
6c4009 |
for line in i18n_file:
|
|
Packit |
6c4009 |
if line.startswith('translit_start'):
|
|
Packit |
6c4009 |
tail = line
|
|
Packit |
6c4009 |
break
|
|
Packit |
6c4009 |
for line in i18n_file:
|
|
Packit |
6c4009 |
tail = tail + line
|
|
Packit |
6c4009 |
return (head, tail)
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
def output_head(i18n_file, unicode_version, head=''):
|
|
Packit |
6c4009 |
'''Write the header of the output file, i.e. the part of the file
|
|
Packit |
6c4009 |
before the “LC_CTYPE” line.
|
|
Packit |
6c4009 |
'''
|
|
Packit |
6c4009 |
if ARGS.input_file and head:
|
|
Packit |
6c4009 |
i18n_file.write(head)
|
|
Packit |
6c4009 |
else:
|
|
Packit |
6c4009 |
i18n_file.write('escape_char /\n')
|
|
Packit |
6c4009 |
i18n_file.write('comment_char %\n')
|
|
Packit |
6c4009 |
i18n_file.write('\n')
|
|
Packit |
6c4009 |
i18n_file.write('% Generated automatically by '
|
|
Packit |
6c4009 |
+ 'gen_unicode_ctype.py '
|
|
Packit |
6c4009 |
+ 'for Unicode {:s}.\n'.format(unicode_version))
|
|
Packit |
6c4009 |
i18n_file.write('\n')
|
|
Packit |
6c4009 |
i18n_file.write('LC_IDENTIFICATION\n')
|
|
Packit |
6c4009 |
i18n_file.write('title "Unicode {:s} FDCC-set"\n'.format(
|
|
Packit |
6c4009 |
unicode_version))
|
|
Packit |
6c4009 |
i18n_file.write('source "UnicodeData.txt, '
|
|
Packit |
6c4009 |
+ 'DerivedCoreProperties.txt"\n')
|
|
Packit |
6c4009 |
i18n_file.write('address ""\n')
|
|
Packit |
6c4009 |
i18n_file.write('contact ""\n')
|
|
Packit |
6c4009 |
i18n_file.write('email "bug-glibc-locales@gnu.org"\n')
|
|
Packit |
6c4009 |
i18n_file.write('tel ""\n')
|
|
Packit |
6c4009 |
i18n_file.write('fax ""\n')
|
|
Packit |
6c4009 |
i18n_file.write('language ""\n')
|
|
Packit |
6c4009 |
i18n_file.write('territory "Earth"\n')
|
|
Packit |
6c4009 |
i18n_file.write('revision "{:s}"\n'.format(unicode_version))
|
|
Packit |
6c4009 |
i18n_file.write('date "{:s}"\n'.format(
|
|
Packit |
6c4009 |
time.strftime('%Y-%m-%d')))
|
|
Packit |
6c4009 |
i18n_file.write('category "i18n:2012";LC_CTYPE\n')
|
|
Packit |
6c4009 |
i18n_file.write('END LC_IDENTIFICATION\n')
|
|
Packit |
6c4009 |
i18n_file.write('\n')
|
|
Packit |
6c4009 |
i18n_file.write('LC_CTYPE\n')
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
def output_tail(i18n_file, tail=''):
|
|
Packit |
6c4009 |
'''Write the tail of the output file, i.e. the part of the file
|
|
Packit |
6c4009 |
after the last “LC_CTYPE” character class.
|
|
Packit |
6c4009 |
'''
|
|
Packit |
6c4009 |
if ARGS.input_file and tail:
|
|
Packit |
6c4009 |
i18n_file.write(tail)
|
|
Packit |
6c4009 |
else:
|
|
Packit |
6c4009 |
i18n_file.write('END LC_CTYPE\n')
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
def output_tables(i18n_file, unicode_version, turkish):
|
|
Packit |
6c4009 |
'''Write the new LC_CTYPE character classes to the output file'''
|
|
Packit |
6c4009 |
i18n_file.write('% The following is the 14652 i18n fdcc-set '
|
|
Packit |
6c4009 |
+ 'LC_CTYPE category.\n')
|
|
Packit |
6c4009 |
i18n_file.write('% It covers Unicode version {:s}.\n'.format(
|
|
Packit |
6c4009 |
unicode_version))
|
|
Packit |
6c4009 |
i18n_file.write('% The character classes and mapping tables were '
|
|
Packit |
6c4009 |
+ 'automatically\n')
|
|
Packit |
6c4009 |
i18n_file.write('% generated using the gen_unicode_ctype.py '
|
|
Packit |
6c4009 |
+ 'program.\n\n')
|
|
Packit |
6c4009 |
i18n_file.write('% The "upper" class reflects the uppercase '
|
|
Packit |
6c4009 |
+ 'characters of class "alpha"\n')
|
|
Packit |
6c4009 |
output_charclass(i18n_file, 'upper', unicode_utils.is_upper)
|
|
Packit |
6c4009 |
i18n_file.write('% The "lower" class reflects the lowercase '
|
|
Packit |
6c4009 |
+ 'characters of class "alpha"\n')
|
|
Packit |
6c4009 |
output_charclass(i18n_file, 'lower', unicode_utils.is_lower)
|
|
Packit |
6c4009 |
i18n_file.write('% The "alpha" class of the "i18n" FDCC-set is '
|
|
Packit |
6c4009 |
+ 'reflecting\n')
|
|
Packit |
6c4009 |
i18n_file.write('% the recommendations in TR 10176 annex A\n')
|
|
Packit |
6c4009 |
output_charclass(i18n_file, 'alpha', unicode_utils.is_alpha)
|
|
Packit |
6c4009 |
i18n_file.write('% The "digit" class must only contain the '
|
|
Packit |
6c4009 |
+ 'BASIC LATIN digits, says ISO C 99\n')
|
|
Packit |
6c4009 |
i18n_file.write('% (sections 7.25.2.1.5 and 5.2.1).\n')
|
|
Packit |
6c4009 |
output_charclass(i18n_file, 'digit', unicode_utils.is_digit)
|
|
Packit |
6c4009 |
i18n_file.write('% The "outdigit" information is by default '
|
|
Packit |
6c4009 |
+ '"0" to "9". We don\'t have to\n')
|
|
Packit |
6c4009 |
i18n_file.write('% provide it here since localedef will fill '
|
|
Packit |
6c4009 |
+ 'in the bits and it would\n')
|
|
Packit |
6c4009 |
i18n_file.write('% prevent locales copying this file define '
|
|
Packit |
6c4009 |
+ 'their own values.\n')
|
|
Packit |
6c4009 |
i18n_file.write('% outdigit /\n')
|
|
Packit |
6c4009 |
i18n_file.write('% <U0030>..<U0039>\n\n')
|
|
Packit |
6c4009 |
# output_charclass(i18n_file, 'outdigit', is_outdigit)
|
|
Packit |
6c4009 |
output_charclass(i18n_file, 'space', unicode_utils.is_space)
|
|
Packit |
6c4009 |
output_charclass(i18n_file, 'cntrl', unicode_utils.is_cntrl)
|
|
Packit |
6c4009 |
output_charclass(i18n_file, 'punct', unicode_utils.is_punct)
|
|
Packit |
6c4009 |
output_charclass(i18n_file, 'graph', unicode_utils.is_graph)
|
|
Packit |
6c4009 |
output_charclass(i18n_file, 'print', unicode_utils.is_print)
|
|
Packit |
6c4009 |
i18n_file.write('% The "xdigit" class must only contain the '
|
|
Packit |
6c4009 |
+ 'BASIC LATIN digits and A-F, a-f,\n')
|
|
Packit |
6c4009 |
i18n_file.write('% says ISO C 99 '
|
|
Packit |
6c4009 |
+ '(sections 7.25.2.1.12 and 6.4.4.1).\n')
|
|
Packit |
6c4009 |
output_charclass(i18n_file, 'xdigit', unicode_utils.is_xdigit)
|
|
Packit |
6c4009 |
output_charclass(i18n_file, 'blank', unicode_utils.is_blank)
|
|
Packit |
6c4009 |
if turkish:
|
|
Packit |
6c4009 |
i18n_file.write('% The case conversions reflect '
|
|
Packit |
6c4009 |
+ 'Turkish conventions.\n')
|
|
Packit |
6c4009 |
output_charmap(i18n_file, 'toupper', unicode_utils.to_upper_turkish)
|
|
Packit |
6c4009 |
output_charmap(i18n_file, 'tolower', unicode_utils.to_lower_turkish)
|
|
Packit |
6c4009 |
else:
|
|
Packit |
6c4009 |
output_charmap(i18n_file, 'toupper', unicode_utils.to_upper)
|
|
Packit |
6c4009 |
output_charmap(i18n_file, 'tolower', unicode_utils.to_lower)
|
|
Packit |
6c4009 |
output_charmap(i18n_file, 'map "totitle";', unicode_utils.to_title)
|
|
Packit |
6c4009 |
i18n_file.write('% The "combining" class reflects ISO/IEC 10646-1 '
|
|
Packit |
6c4009 |
+ 'annex B.1\n')
|
|
Packit |
6c4009 |
i18n_file.write('% That is, all combining characters (level 2+3).\n')
|
|
Packit |
6c4009 |
output_charclass(i18n_file, 'class "combining";',
|
|
Packit |
6c4009 |
unicode_utils.is_combining)
|
|
Packit |
6c4009 |
i18n_file.write('% The "combining_level3" class reflects '
|
|
Packit |
6c4009 |
+ 'ISO/IEC 10646-1 annex B.2\n')
|
|
Packit |
6c4009 |
i18n_file.write('% That is, combining characters of level 3.\n')
|
|
Packit |
6c4009 |
output_charclass(i18n_file, 'class "combining_level3";',
|
|
Packit |
6c4009 |
unicode_utils.is_combining_level3)
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
if __name__ == "__main__":
|
|
Packit |
6c4009 |
PARSER = argparse.ArgumentParser(
|
|
Packit |
6c4009 |
description='''
|
|
Packit |
6c4009 |
Generate a Unicode conforming LC_CTYPE category from
|
|
Packit |
6c4009 |
UnicodeData.txt and DerivedCoreProperties.txt files.
|
|
Packit |
6c4009 |
''')
|
|
Packit |
6c4009 |
PARSER.add_argument(
|
|
Packit |
6c4009 |
'-u', '--unicode_data_file',
|
|
Packit |
6c4009 |
nargs='?',
|
|
Packit |
6c4009 |
type=str,
|
|
Packit |
6c4009 |
default='UnicodeData.txt',
|
|
Packit |
6c4009 |
help=('The UnicodeData.txt file to read, '
|
|
Packit |
6c4009 |
+ 'default: %(default)s'))
|
|
Packit |
6c4009 |
PARSER.add_argument(
|
|
Packit |
6c4009 |
'-d', '--derived_core_properties_file',
|
|
Packit |
6c4009 |
nargs='?',
|
|
Packit |
6c4009 |
type=str,
|
|
Packit |
6c4009 |
default='DerivedCoreProperties.txt',
|
|
Packit |
6c4009 |
help=('The DerivedCoreProperties.txt file to read, '
|
|
Packit |
6c4009 |
+ 'default: %(default)s'))
|
|
Packit |
6c4009 |
PARSER.add_argument(
|
|
Packit |
6c4009 |
'-i', '--input_file',
|
|
Packit |
6c4009 |
nargs='?',
|
|
Packit |
6c4009 |
type=str,
|
|
Packit |
6c4009 |
help='''The original glibc/localedata/locales/i18n file.''')
|
|
Packit |
6c4009 |
PARSER.add_argument(
|
|
Packit |
6c4009 |
'-o', '--output_file',
|
|
Packit |
6c4009 |
nargs='?',
|
|
Packit |
6c4009 |
type=str,
|
|
Packit |
6c4009 |
default='i18n.new',
|
|
Packit |
6c4009 |
help='''The file which shall contain the generated LC_CTYPE category,
|
|
Packit |
6c4009 |
default: %(default)s. If the original
|
|
Packit |
6c4009 |
glibc/localedata/locales/i18n has been given
|
|
Packit |
6c4009 |
as an option, all data from the original file
|
|
Packit |
6c4009 |
except the newly generated LC_CTYPE character
|
|
Packit |
6c4009 |
classes and the date stamp in
|
|
Packit |
6c4009 |
LC_IDENTIFICATION will be copied unchanged
|
|
Packit |
6c4009 |
into the output file. ''')
|
|
Packit |
6c4009 |
PARSER.add_argument(
|
|
Packit |
6c4009 |
'--unicode_version',
|
|
Packit |
6c4009 |
nargs='?',
|
|
Packit |
6c4009 |
required=True,
|
|
Packit |
6c4009 |
type=str,
|
|
Packit |
6c4009 |
help='The Unicode version of the input files used.')
|
|
Packit |
6c4009 |
PARSER.add_argument(
|
|
Packit |
6c4009 |
'--turkish',
|
|
Packit |
6c4009 |
action='store_true',
|
|
Packit |
6c4009 |
help='Use Turkish case conversions.')
|
|
Packit |
6c4009 |
ARGS = PARSER.parse_args()
|
|
Packit |
6c4009 |
|
|
Packit |
6c4009 |
unicode_utils.fill_attributes(
|
|
Packit |
6c4009 |
ARGS.unicode_data_file)
|
|
Packit |
6c4009 |
unicode_utils.fill_derived_core_properties(
|
|
Packit |
6c4009 |
ARGS.derived_core_properties_file)
|
|
Packit |
6c4009 |
unicode_utils.verifications()
|
|
Packit |
6c4009 |
HEAD = TAIL = ''
|
|
Packit |
6c4009 |
if ARGS.input_file:
|
|
Packit |
6c4009 |
(HEAD, TAIL) = read_input_file(ARGS.input_file)
|
|
Packit |
6c4009 |
with open(ARGS.output_file, mode='w') as I18N_FILE:
|
|
Packit |
6c4009 |
output_head(I18N_FILE, ARGS.unicode_version, head=HEAD)
|
|
Packit |
6c4009 |
output_tables(I18N_FILE, ARGS.unicode_version, ARGS.turkish)
|
|
Packit |
6c4009 |
output_tail(I18N_FILE, tail=TAIL)
|