hexchat/po/validate-textevent-translations

#!/usr/bin/env python3

import os
import re
import sys

ret = 0


def log(prefix, suffix, fallback):
    try:
        print(prefix, suffix)  # Non-utf8 output...
    except UnicodeEncodeError:
        print(prefix, fallback)


def validate_translation(input, translation):
    if not translation:
        return True

    if re.findall(r'(?:(?<!%)%[^%OCRUHBIH]|\$[^at1234])', translation):
        log('Translation includes invalid formatting:', translation, input)
        return False

    in_vars = re.findall(r'(\$(?:\d|t))', input)
    if not all(var in translation for var in in_vars):
        log('Translation does not contain all variables:', translation, input)
        return False

    in_ascii = re.findall(r'\$a(\d{3})', translation)
    if any(int(i) > 256 for i in in_ascii):
        log('Translation contains invalid ascii value:', translation, input)
        return False

    # We could try to validate colors but that is pretty flexible
    return True


def validate_language(path):
    global ret

    print('Validating', path)

    with open(path, 'r', encoding='utf-8') as f:
        in_event = False
        event_input = ''

        for line in f:
            if 'textevents.h' in line:
                in_event = True
            elif in_event is False:
                continue
            elif line.startswith('msgid'):
                event_input = line[7:-2]
            elif line.startswith('msgstr'):
                if not validate_translation(event_input, line[8:-2]):
                    ret = 1
                in_event = False
            elif line == '\n':
                print('Failed to find translation for', event_input)
                in_event = False


with open(sys.argv[1], 'r') as linguas:
    for lang in linguas:
        path = os.path.join(sys.argv[2], lang.strip() + '.po')
        validate_language(path)

sys.exit(ret)