Interwiki for CX

amire80.py

  GNU nano 2.2.6                                File: scripts/amire80.py                                                                        

# License: MIT
import pywikibot, codecs, json
site = pywikibot.Site('ca')
offset = 0
_base_dir = '/data/project/dexbot/pywikibot-core/'
cases = []
while True:
    req = pywikibot.data.api.Request(site=site, action='query', list='cxpublishedtranslations', limit=500, offset=offset)
    offset += 500
    res = req.submit()
    if not res['result']['translations']:
        break
    for case in res['result']['translations']:
        if '/wiki/User:' in case['targetURL']:
            continue
        new_case = {
            'sourceTitle': case['sourceTitle'],
            'targetTitle': case['targetTitle'],
            'sourceLanguage': case['sourceLanguage'],
            'targetLanguage': case['targetLanguage'],
        }
        cases.append(new_case)

with codecs.open('res.txt', 'w', 'utf-8') as f:
    f.write(json.dumps(cases))
with codecs.open('%sres.txt' % _base_dir, 'r', 'utf-8') as f:
    cases = json.loads(f.read())
issues = []
for case in cases:
#    if not 'Baebele' in str(case):
#        continue
    try:
        source_site = pywikibot.Site(case['sourceLanguage'])
        target_site = pywikibot.Site(case['targetLanguage'])
        source_page = pywikibot.Page(source_site, case['sourceTitle'])
        target_page = pywikibot.Page(target_site, case['targetTitle'])
        target_page.exists()
        source_page.exists()
    except:
        continue
        case['notes'] = 'Unknown Error'
        issues.append(case)
        with codecs.open('%serrors.txt' % _base_dir, 'w', 'utf-8') as f:
            f.write(json.dumps(cases))
        continue
    if not target_page.exists() or not source_page.exists():
        continue
        case['notes'] = 'Source or target does not exist'
        issues.append(case)
        with codecs.open('%serrors.txt' % _base_dir, 'w', 'utf-8') as f:
            f.write(json.dumps(issues))
        continue
    if target_page.namespace() != source_page.namespace():
        print("Namespaces doesn't match")
        continue
    if source_page.isRedirectPage():
        source_page = source_page.getRedirectTarget()
    if target_page.isRedirectPage():
        target_page = target_page.getRedirectTarget()
    source_item = None
    target_item = None
    try:
        source_item = pywikibot.ItemPage.fromPage(source_page)
    except:
        pass
    try:
        target_item = pywikibot.ItemPage.fromPage(target_page)
    except:
        pass
    if not target_item:
        if not source_item:
            case['notes'] = 'None of them are in Wikidata'
            issues.append(case)
            with codecs.open('%serrors.txt' % _base_dir, 'w', 'utf-8') as f:
                f.write(json.dumps(issues))
        elif source_item.exists():
            try:
                if source_item.getSitelink(target_site):
                    print source_item.getID()
                else:
                    source_item.setSitelink(target_page)
            except pywikibot.exceptions.NoPage:
                try:
                    source_item.setSitelink(target_page)
                except:
                    pass
    elif not source_item and target_item.exists():
        try:
            target_item.getSitelink(source_site)
        except:
            continue
        if target_item.getSitelink(source_site):
            print target_item.getID()
        else:
            target_item.setSitelink(source_page)
    else:
        if target_item.getID() != source_item.getID():
            case['notes'] = 'Items need merge'
            issues.append(case)
            with codecs.open('%serrors.txt' % _base_dir, 'w', 'utf-8') as f:
                f.write(json.dumps(issues))