util.py 文件源码

def correct_orphan_records(self, provider='europeana', end=None):
        """[#185] Delete records from the search engine which aren't found in the database"""
        s = Search()
        q = Q('term', provider=provider)
        s = s.query(q)
        response = s.execute()
        total = response.hits.total
        # A file extracted from the production database listing all of the europeana identifiers
        identifier_file = '/tmp/europeana-identifiers.json'
        db_identifiers = set(json.load(open(identifier_file)))
        total_in_db = len(db_identifiers)
        log.info("Using search engine instance %s", settings.ELASTICSEARCH_URL)
        log.info("Total records: %d (search engine), %d (database) [diff=%d]", total, total_in_db, total - total_in_db)
        deleted_count = 0
        for r in s.scan():
            if r.identifier not in db_identifiers:
                img = search.Image.get(id=r.identifier)
                log.debug("Going to delete image %s", img)
                deleted_count += 1
        log.info("Deleted %d from search engine", deleted_count)