IWBot.py

import time
from  urllib.parse import unquote
import re
import mw_api_client as mwc

try:
    with open(".lastpage-iw", "r", encoding="utf-8") as f:
        lastpage = f.read()
except OSError:
    lastpage = ""

def make_from_pairs(pairs):
    s = ""
    for p in pairs:
        s += f"[[{p[0]}:{p[1]}]]"
    return s

ALLOWED_IWLINKS = "en", "de", "ru", "fr", "hu", "id", "nl", "tw"

IW_REGEX = re.compile(r"\[\[({IWPREFIXES}):([^\]]*)\]\]".format(IWPREFIXES="|".join(ALLOWED_IWLINKS)))

def get_pair(content):
    return [m.groups() for m in IW_REGEX.finditer(content)]

jaw = mwc.Wiki("https://ja.scratch-wiki.info/w/api.php", "IWBot/mw-api-client")
jaw.login("Username@IWBot","Enter your bot password here...")

enw = mwc.Wiki("https://en.scratch-wiki.info/w/api.php", "IWBot/mw-api-client")

for page in jaw.allpages(getinfo=True, apfrom=lastpage, apfilterredir="nonredirects", apfilterlanglinks="withlanglinks", **{"assert":"bot"}):
    print(f"-------------------- NOW DOING {page.title} --------------------")
    if page.protection:
        continue
    content = page.read()
    pairs = get_pair(content)
    
    enw_pair = tuple(filter(lambda a:a[0]=="en", pairs))
    if not enw_pair:
        continue # enwiki not found!
    
    enpgname = unquote(enw_pair[0][1])
    print(f"------------------- FOUND EN {enpgname} --------------")
    enpage = enw.page(enpgname)
    enpcont = enpage.read()
    enp_pairs = get_pair(content)
    new_pair = pairs.copy()
    for index, np in enumerate(pairs):
        temp = tuple(filter(lambda a:a[0]==np[0], enp_pairs))
        if temp:
            new_pair[index] = temp[0]
    temp2 = list(filter(lambda a:a[0] not in [b[0] for b in pairs], enp_pairs))
    new_pair += temp2
    
    content = IW_REGEX.sub("", content)
    content += make_from_pairs(new_pair)
    
    print("-------------------- CONFIRM ---------------------------")
    print(f"Old: {pairs}\nNew: {new_pair}")
    
    stamp = time.time()
    if pairs != new_pair and input("Save? > "):
        page.edit(content, "Interwiki (Automated)")
        print(f"-------------------- SAVED {page.title} --------------------")
    else:
        print(f"-------------------- DONE {page.title}  --------------------")
    with open(".lastpage-iw", "w", encoding="utf-8") as f:
        f.write(page.title)
    
    slpt = stamp + 10 - time.time()
    if slpt > 0:
        print(f"----------------- WAITING FOR {round(slpt*10)/10} SECS --------------")
        time.sleep(slpt)