Merge pull request #84 from soxoj/yelp

Yelp added

Author: soxoj

CHANGELOG.md

@@ -2,6 +2,7 @@
 
 ## [Unreleased]
 * added username-from-email extractor
+* added Yelp
 
 ## [0.0.18] - 2021-05-04
 * fixed and improved Behance

socid_extractor/postprocessor.py

@@ -50,8 +50,6 @@ def process(self):
 
 class Email:
     def __init__(self, data):
-        import logging
-        logging.error(data)
         self.data = data
 
     def process(self):

socid_extractor/schemes.py

@@ -1333,5 +1333,15 @@
             'location': lambda x: x['location']['city'] + ', ' + x['location']['country']['name'],
             'created_at': lambda x: timestamp_to_datetime(x['registration_date']),
         }
+    },
+    'Yelp': {
+        'flags': ['yelp.www.init.user_details'],
+        'bs': True,
+        'fields': {
+            'yelp_userid': lambda x: x.find('meta', {'property': 'og:url'}).get('content').split('=')[-1],
+            'fullname': lambda x: x.find('div', {'class': 'user-profile_info'}).find('h1').contents[0],
+            'location': lambda x: x.find('div', {'class': 'user-profile_info'}).find('h3').contents[0].split(' ', 1)[1],
+            'image': lambda x: x.find('div', {'class': 'user-profile_avatar'}).find('img').get('src'),
+        }
     }
 }

tests/test_e2e.py

@@ -954,3 +954,21 @@ def test_freelancer():
     assert info.get('role') == 'freelancer'
     assert info.get('location') == 'Islamabad, Pakistan'
     assert info.get('created_at').startswith('2012-12-09')
+
+
+def test_yelp_username():
+    info = extract(parse('http://dima.yelp.com')[0])
+
+    assert info.get('yelp_userid') == 'b_a5icXGK-AXVYZKehgKLw'
+    assert info.get('fullname') == 'Dima "ZOMG" M.'
+    assert info.get('location') == 'Brooklyn, NY'
+    assert info.get('image') == 'https://s3-media0.fl.yelpcdn.com/photo/bGiNMDL6FZAtPpMfljRGtg/ls.jpg'
+
+
+def test_yelp_userid():
+    info = extract(parse('https://www.yelp.com/user_details?userid=b_a5icXGK-AXVYZKehgKLw')[0])
+
+    assert info.get('yelp_userid') == 'b_a5icXGK-AXVYZKehgKLw'
+    assert info.get('fullname') == 'Dima "ZOMG" M.'
+    assert info.get('location') == 'Brooklyn, NY'
+    assert info.get('image') == 'https://s3-media0.fl.yelpcdn.com/photo/bGiNMDL6FZAtPpMfljRGtg/ls.jpg'