prepare.py 文件源码

def get_thingpedia(input_words, workdir, snapshot):
    thingpedia_url = os.getenv('THINGPEDIA_URL', 'https://thingpedia.stanford.edu/thingpedia')

    output = dict()
    with urllib.request.urlopen(thingpedia_url + '/api/snapshot/' + str(snapshot) + '?meta=1', context=ssl_context) as res:
        output['devices'] = json.load(res)['data']
        for device in output['devices']:
            if device['kind_type'] == 'global':
                continue
            if device['kind_canonical']:
                add_words(input_words, device['kind_canonical'])
            else:
                print('WARNING: missing canonical for tt-device:%s' % (device['kind'],))
            for function_type in ('triggers', 'queries', 'actions'):
                for function_name, function in device[function_type].items():
                    if not function['canonical']:
                        print('WARNING: missing canonical for tt:%s.%s' % (device['kind'], function_name))
                    else:
                        add_words(input_words, function['canonical'])
                    for argname, argcanonical in zip(function['args'], function['argcanonicals']):
                        if argcanonical:
                            add_words(input_words, argcanonical)
                        else:
                            add_words(input_words, clean(argname))
                    for argtype in function['schema']:
                        if not argtype.startswith('Enum('):
                            continue
                        enum_entries = argtype[len('Enum('):-1].split(',')
                        for enum_value in enum_entries:
                            add_words(input_words, clean(enum_value))

    with urllib.request.urlopen(thingpedia_url + '/api/entities?snapshot=' + str(snapshot), context=ssl_context) as res:
        output['entities'] = json.load(res)['data']
        for entity in output['entities']:
            if entity['is_well_known'] == 1:
                continue
            add_words(input_words, tokenize(entity['name']))

    with open(os.path.join(workdir, 'thingpedia.json'), 'w') as fp:
        json.dump(output, fp, indent=2)