Create Softbank and Unified Emoji mapping for iOS.

parse_full.py

#! /usr/bin/env python3

'''Create Softbank and Unified Emoji mapping for iOS.'''

# download full.html to same directory first
# curl -O http://www.unicode.org/%7Escherer/emoji4unicode/snapshot/full.html

import re
import sys

def convert(code):
    if code.startswith('U+'):
        hex = code.replace('U+', '')
        integer = int(hex, 16)
        if integer <= 127:
            return '\\x%02X' % integer
        elif integer <= 65535:
            return '\\u%04X' % integer
        return '\\U%08X' % integer
    return code

# 1:1 mapping
all_unified = []
all_softbank = []

# 2:1 mapping
multi_unified = []
multi_softbank = []

with open('full.html') as html:
    for line in html.readlines():
        id = re.findall('^<tr id=(e-...)', line)
        if id:
            id, rep, name_anno, *rest = re.findall('<td.*?</td>', line)
            unified = re.findall("U\+[0-9A-F]{4,5}(?: U\+[0-9A-F]{4,5})?", rep)

            if unified:
                maps = []
                for i in rest:
                     found = re.findall("class='round_trip'.*?(U\+[0-9A-F]{4,5})", i)
                     maps.append(found)

                unified = unified[0]
                softbank = maps[2]
                if softbank:
                    softbank = convert(softbank[0])

                    if ' ' in unified:
                        a, b = unified.split(' ')
                        unified = convert(a) + convert(b)
                        multi_unified.append(unified)
                        multi_softbank.append(softbank)
                    else:
                        unified = convert(unified)
                        all_unified.append(unified)
                        all_softbank.append(softbank)

def assert_unique(array):
    assert(len(set(array)) == len(array))

def print_array(name, array):
    assert_unique(array)
    print('static NSString *const %s[] = {' % name, end='')
    for u in array:
        print('@"' + u + '",', end=' ')
    print('};')

def print_string(name, array):
    print('static NSString *const %s = @"' % name, end='')
    for i in array:
        print(i, end='')
    print('";')

print_array('all_unified', all_unified)
print_array('all_softbank', all_softbank)

print_array('multi_unified', multi_unified)
print_array('multi_softbank', multi_softbank)

print('// just all_unified')
print_string('all_unified_set', all_unified)

print('// all_softbank + multi_softbank')
print_string('softbank_set', all_softbank + multi_softbank)

print('static const NSInteger kNumOfEmoji = %d;' % len(all_unified))
print('static const NSInteger kNumOfMultiUnicodeEmoji = %d;' % len(multi_unified))