pandora_cms/ontology/update_keywords.py

#!/usr/bin/python3
import collections
import json

import ox
import ox.api

site = 'pandora.cinemusespace.com'
api = ox.api.signin('https://%s/api/' % site)

keywords = collections.Counter()
for annotation in api.findAnnotations({
    'query': {
        'conditions': [{
            'key': 'layer',
            'value': 'keywords',
            'operator': '=='
        }],
        'operator': '&'
    },
    'keys': ['id', 'in', 'out', 'value', 'user', 'created'],
    'range': [0, 500000]
})['data']['items']:
    if annotatoin['id'].startswith('BA/'):
        continue
    keyword = annotation['value']
    if ': ' not in keyword:
        keyword = 'other: ' + keyword
    keywords[keyword] += 1

with open('keywords.json', 'w') as fd:
    json.dump(keywords, fd, indent=4, ensure_ascii=False, sort_keys=True)