WLE_stats_generation.py 文件源码

def get_global_usage (api_url, query_string_dict, title_list) :
    usage_dict_ = dict()
    usage_dict_["image"] = dict()
    usage_dict_["article"] = dict()
    raw_api_query_string = unicode(u'|'.join(title_list)).encode('utf-8')
    #print raw_api_query_string
    API_QUERY_STRING["titles"] = raw_api_query_string
    f = urlopen(API_BASE_URL, urlencode(API_QUERY_STRING))
    response = f.read()
    response_dict = json.loads(response)
    for key, value in response_dict["query"]["pages"].iteritems():
        if len(value[u'globalusage']) > 0:
            #print value
            found_dict = dict()
            for item in value[u'globalusage']:
                if (item[u'ns'] == u'0') or (item[u'ns'] == u'104'):
                    if item[u'wiki'] in usage_dict_["article"]:
                        usage_dict_["article"][item[u'wiki']] += 1
                    else:
                        usage_dict_["article"][item[u'wiki']] = 1
                    found_dict[item[u'wiki']] = True
            for key, value in found_dict.iteritems():
                if key in usage_dict_["image"]:
                    usage_dict_["image"][key] += 1
                else:
                    usage_dict_["image"][key] = 1
    #print usage_dict_
    return usage_dict_