class FeatureCollection(object):
def __init__(self,geodatabase_path,ianadatabase_path,wfbdatabase_path):
self.ip_location_extraction = IPLocationExtraction(geodatabase_path)
self.tld_location_extraction = TLDLocationExtraction(ianadatabase_path, wfbdatabase_path)
self.website_language_extraction = WebsiteLanguageExtraction()
def get_features(self, article_name, url):
#generate json from article_name and url
article_url_dictionary = {}
urls = []
urls.append(url)
article_url_dictionary[article_name] = urls
json_data = json.loads(json.dumps(article_url_dictionary))
ip_locations = self.ip_location_extraction.get_ip_locations(json_data)
tld_locations = self.tld_location_extraction.get_tld_locations(json_data)
website_languages = self.website_language_extraction.get_website_languages(json_data)
features = {}
if url in ip_locations:
features["ip-location"] = ip_locations[url]
if url in tld_locations:
features["tld-location"] = tld_locations[url]
if url in website_languages:
features["website-language"] = website_languages[url]
return features
def __init__(self,geodatabase_path,ianadatabase_path,wfbdatabase_path):
self.ip_location_extraction = IPLocationExtraction(geodatabase_path)
self.tld_location_extraction = TLDLocationExtraction(ianadatabase_path, wfbdatabase_path)
self.website_language_extraction = WebsiteLanguageExtraction()