[T261595] Add link to English website about Arnberg to Musikverket's music files

add_english_arnberg.py

# -*- coding: utf-8 -*-
"""Add link to English site about Arnberg

https://phabricator.wikimedia.org/T261595
Correct descriptions of files from Musikverket's
Arnberg collection by adding links to their
English website, not only Swedish.
"""
import pywikibot
from pywikibot import pagegenerators as pg
import mwparserfromhell as parser

site = pywikibot.Site("commons", 'commons')
catname = "Audio files from the Swedish Performing Arts Agency"
cat = pywikibot.Category(site, catname)
gen = pg.CategorizedPageGenerator(cat)


def add_missing_english(page):
    edit_summary = "Adding link to glam's english page"
    arnberg_desc = ("[https://musikverket.se/"
                    "svensktvisarkiv/arnbergs-resa-i-folkmusiken/ "
                    "Matts Arnberg]")
    new_arnberg = ("Matts Arnberg "
                   "[https://musikverket.se/"
                   "svensktvisarkiv/arnbergs-resa-i-folkmusiken/ SV] "
                   "[https://musikverket.se/"
                   "svensktvisarkiv/arnbergs-resa-i-folkmusiken/?lang=en EN]")
    parsed = parser.parse(page.text)
    templates = parsed.filter_templates()
    for t in templates:
        if t.name.matches("Musikverket-audio"):
            old_desc = str(t.get("description").value)
            if arnberg_desc in old_desc:
                new_desc = old_desc.replace(arnberg_desc, new_arnberg)
                new_desc = new_desc.replace("{{sv|", "{{sv|1=")
                t.add("description", new_desc)
                page.text = str(parsed)
                page.save(edit_summary)


if __name__ == "__main__":
    for g in gen:
        add_missing_english(g)