emails

api/server.py

@@ -61,14 +61,16 @@ def split():
             first_name, last_name = extract_names_from_email(nd, q)
             if first_name is not None:
                 result_first_name = nd.search(first_name)['first_name']
+                if result_first_name is not None:
+                    result_first_name['name'] = first_name
             else:
                 result_first_name = None
             if last_name is not None:
                 result_last_name = nd.search(last_name)['last_name']
+                if result_last_name is not None:
+                    result_last_name['name'] = last_name
             else:
                 result_last_name = None
-            result_first_name['name'] = first_name
-            result_last_name['name'] = last_name
             result = {
                 'first_name': result_first_name,
                 'last_name': result_last_name

names_dataset/emails.py

@@ -53,20 +53,25 @@ def _infer_best_split(nd: NameDataset, full_name: str):
 
 def _general_score(nd: NameDataset, candidate: str):
     c = nd.search(candidate)
-    s1 = _compute_score(c['first_name'])
-    s2 = _compute_score(c['last_name'])
-    return max(s1, s2)
+    if c['first_name'] is not None and c['last_name'] is not None:
+        s1 = _compute_score(c['first_name'])
+        s2 = _compute_score(c['last_name'])
+        return max(s1, s2)
+    return float('-inf')
 
 
 def extract_names_from_email(nd: NameDataset, email: str):
+    email = email.strip()
+    if '' in email:
+        email = email.split(' ')[0]
     if '@' not in email:
         email += '@gmail.com'
 
     email = ''.join([e for e in list(email) if not e.isnumeric()])
 
     prefix, suffix = email.split('@')
 
-    no_names = ['contact', 'sales', 'info', 'hello', 'reply']
+    no_names = ['contact', 'sales', 'info', 'hello', 'reply', 'service', 'client']
     for no_name in no_names:
         if no_name in prefix:
             return None, None