def authors(report=False, dryrun=False, force=False):
index = set_hosts_index()
logprint('debug', '------------------------------------------------------------------------')
logprint('debug', 'getting mw_authors...')
mw_author_titles = Proxy().authors(cached_ok=False)
mw_articles = Proxy().articles_lastmod()
logprint('debug', 'getting es_authors...')
es_authors = Author.authors()
if force:
logprint('debug', 'forcibly update all authors')
authors_new = [page['title'] for page in es_authors]
authors_delete = []
else:
logprint('debug', 'determining new,delete...')
authors_new,authors_delete = Elasticsearch.authors_to_update(
mw_author_titles, mw_articles, es_authors)
logprint('debug', 'mediawiki authors: %s' % len(mw_author_titles))
logprint('debug', 'authors to add: %s' % len(authors_new))
logprint('debug', 'authors to delete: %s' % len(authors_delete))
if report:
return
logprint('debug', 'deleting...')
for n,title in enumerate(authors_delete):
logprint('debug', '--------------------')
logprint('debug', '%s/%s %s' % (n, len(authors_delete), title))
author = Author.get(title)
if not dryrun:
author.delete()
logprint('debug', 'adding...')
for n,title in enumerate(authors_new):
logprint('debug', '--------------------')
logprint('debug', '%s/%s %s' % (n, len(authors_new), title))
logprint('debug', 'getting from mediawiki')
mwauthor = Proxy().page(title)
try:
existing_author = Author.get(title)
logprint('debug', 'exists in elasticsearch')
except:
existing_author = None
logprint('debug', 'creating author')
author = Author.from_mw(mwauthor, author=existing_author)
if not dryrun:
logprint('debug', 'saving')
author.save()
try:
a = Author.get(title)
except NotFoundError:
logprint('error', 'ERROR: Author(%s) NOT SAVED!' % title)
logprint('debug', 'DONE')
