def transform_open_format(x): ''' Original format: (u'NutritionalAnarchy.com', {u'2nd type': u'', u'3rd type': u'', u'Source Notes (things to know?)': u'', u'type': u'unreliable'}) ''' urls = mongo_driver.get_url('opensources') if x[0] in urls: return template = { 'Category': 'conspiracy', 'Reference': 'http://mediabiasfactcheck.com/zero-hedge/', 'Truthiness': 'MIXED', 'url': 'http://www.zerohedge.com/' } out_dict = dict().fromkeys(template) out_dict['url'] = x[0] out_dict['Category'] = ', '.join( list(set([x[1][_] for _ in x[1].keys() if 'type' in _ and x[1][_]]))) out_dict['Reference'] = 'http://www.opensources.co' mongo_driver.insert('opensources', out_dict)
def transform_open_format(x): ''' Original format: (u'NutritionalAnarchy.com', {u'2nd type': u'', u'3rd type': u'', u'Source Notes (things to know?)': u'', u'type': u'unreliable'}) ''' urls = mongo_driver.get_url('opensources') if x[0] in urls: return template = { 'Category': 'conspiracy', 'Reference': 'http://mediabiasfactcheck.com/conspiracy-times/', 'Truthiness': 'MIXED', 'url': 'http://www.conspiracy-times.com/' } out_dict = dict().fromkeys(template) out_dict['url'] = x[0] out_dict['Category'] = ', '.join( list(set([x[1][_] for _ in x[1].keys() if 'type' in _ and x[1][_]]))) out_dict['Reference'] = 'http://www.opensources.co' mongo_driver.insert('opensources', out_dict)