MediaWiki:Gadget-translation editor.js/Statistiques/code/total traductions

import re
import time

def count_all_trads():
  input_file = 'C:\\Users\\automatik\\Downloads\\frwiktionary-20180901-pages-articles.xml'
  
  counter_pages = 0
  counter_trads = 0
  
  with open(input_file, encoding='utf-8') as f:
    pattern = re.compile(r'{{trad{0,2}\|')
    ns = -1
    for line in f:
      if '<ns>' in line:
        ns = int(line)
        counter_pages += 1
        if counter_pages % 100000 == 0:
          print(str(counter_trads) + ' traductions dénombrées pour ' + str(counter_pages) + ' pages scannees')
      if ns != 0:
        continue
      if ('{{trad|' in line or '{{trad+|' in line or '{{trad-|' in line or '{{trad--|' in line) and '{{T|' in line and not '<comment>' in line:
        counter_trads += len(re.findall(pattern, line))
  
  print('Total : {} traductions dans {} pages'.format(counter_trads, counter_pages))

if __name__ == '__main__':
  start_time = time.time()
  count_all_trads()
  print('Done in {} sec'.format(str(time.time() - start_time))) # last runtime: 90 sec
  # Dump du 1/12/2023 : 1 259 193 traductions dans 5 378 163 pages