Find unused translations in BlogoText.

find-translations.py

#!/usr/bin/env python
# coding: utf-8
#
# Find unused translations.
# Python 3.
#

from collections import OrderedDict
from glob import glob
import re


def find_used_translations(path_files):
    ''' Find uniq translations. '''

    regexp = re.compile(r"\$GLOBALS\['lang']\['([\w]+)']")
    translations = OrderedDict()
    files = []

    # Retrieve PHP files
    for file in path_files:
        if file.endswith('/'):
            # A directory
            for file in glob('{}/*.php'.format(file)):
                files.append(file)
        else:
            files.append(file)

    # Find used translations
    for file in files:
        with open(file, 'r') as fileh:
            for trans in re.findall(regexp, fileh.read()):
                if trans not in translations:
                    translations[trans] = 1

    return (files, translations)


def find_translations(file):
    ''' Find defined translations. '''

    regexp = re.compile(r"^'([\w]+)'")
    translations = OrderedDict()

    with open(file, 'r') as fileh:
        for line in fileh.read().splitlines():
            for trans in re.findall(regexp, line):
                if trans not in translations:
                    translations[trans] = 1

    return translations


def main():

    paths = ['addons/*/', 'admin/', 'inc/', 'index.php', 'atom.php', 'rss.php']
    files, translations = find_used_translations(paths)
    print('Fichiers traités :', len(files))
    print('Traductions utilisées :', len(translations))

    print()
    translations_fr = find_translations('inc/lang/fr_FR.php')
    print('Traductions FR :', len(translations_fr))

    translations_en = find_translations('inc/lang/en_EN.php')
    print('Traductions EN :', len(translations_en))

    # Find unused translations
    print()
    for translation in translations:
        if translation in translations_fr:
            del translations_fr[translation]
        if translation in translations_en:
            del translations_en[translation]
    print('Traductions inutilisée FR :', len(translations_fr))
    print('Traductions inutilisée EN :', len(translations_en))

    # Print unused translations
    num = 1
    for translation in translations_fr:
        if translation in translations_en:
            del translations_en[translation]
            print(num, 'FR|EN', translation)
        else:
            print(num, 'FR', translation)
        num += 1
    for translation in translations_en:
        print(num, 'EN', translation)
        num += 1

    return 0

if __name__ == '__main__':
    exit(main())