User:Şêr/rhyme-not-found.py

From Wiktionary, the free dictionary
Jump to navigation Jump to search
import pywikibot

site = pywikibot.Site("en", "wiktionary")

def read_titles_from_file(file_path):
    with open(file_path, 'r', encoding='utf-8') as file:
        titles = file.readlines()
    titles = [title.strip() for title in titles]
    return titles

file_path = 'page_titles.txt'

page_titles = read_titles_from_file(file_path)

not_found_pages = []

for title in page_titles:
    page = pywikibot.Page(site, title)
    if "rhyme list begin" not in page.text:
        not_found_pages.append(title)

not_found_file_path = 'not_found_pages.txt'
with open(not_found_file_path, 'w', encoding='utf-8') as not_found_file:
    for page_title in not_found_pages:
        not_found_file.write(page_title + '\n')

print(f"The titles of the pages where the text was not found have been saved in '{not_found_file_path}'.")