Use dict from data/nicknames_dict for nicknames

data/.gitignore

@@ -0,0 +1 @@
+nicknames_dict

profiles/models.py

@@ -12,12 +12,36 @@ from django.db import models
 
 BASE_DIR = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
 
-NICKNAMES = open("/usr/share/dict/american-english").read().splitlines()
+NICKNAMES = None
 LASTNAMES = open(BASE_DIR + "/data/lastnames.txt").read().splitlines()
 FIRSTNAMES = open(BASE_DIR + "/data/firstnames.txt").read().splitlines()
 EMAIL_DOMAINS = open(BASE_DIR + "/data/email_domains.txt").read().splitlines()
 
 
+def require_nicknames(fct):
+    def read_file(path):
+        global NICKNAMES
+        print("Trying {}".format(path))
+        with open(path, 'r') as handle:
+            NICKNAMES = handle.read().splitlines()
+
+    nicknames_files = [
+        os.path.join(BASE_DIR, 'data/nicknames_dict'),
+        "/usr/share/dict/american-english",
+    ]
+    if NICKNAMES is None:
+        for nick_file in nicknames_files:
+            try:
+                read_file(nick_file)
+                break
+            except FileNotFoundError:
+                pass
+        if NICKNAMES is None:
+            raise FileNotFoundError
+
+    return fct
+
+
 class InvalidData(Exception):
     ''' Thrown when the DB contains invalid data, and cannot perform
     something '''
@@ -72,7 +96,7 @@ class Website(models.Model):
         elif rand <= 0.1:
             url = random.choice(self.notable_pages).url
         elif rand <= 0.8:
-            search_term_text = self.name +  " " + \
+            search_term_text = self.name + " " + \
                     random.choice(self.keywords)
             url = user.search_engine.search_url(search_term_text)
         else:
@@ -122,7 +146,6 @@ class Event(models.Model):
         return user.search_engine.search_url(" ".join(possibilities))
 
 
-
 class BrowserFingerprint(models.Model):
     ''' A browser fingerprint, containing things like a user agent '''
 
@@ -147,11 +170,11 @@ class BrowserFingerprint(models.Model):
 
     def serialize_headers(self):
         return {
-            "Description" : str(self.description),
-            "User-Agent" : str(self.useragent),
-            "Accept-Encoding" : str(self.accept_encoding),
-            "Accept" : str(self.accept_default),
-            "Accept-Language" : str(self.accept_lang),
+            "Description": str(self.description),
+            "User-Agent": str(self.useragent),
+            "Accept-Encoding": str(self.accept_encoding),
+            "Accept": str(self.accept_default),
+            "Accept-Language": str(self.accept_lang),
         }
 
 
@@ -214,11 +237,13 @@ def generate_email(nick, first_name, last_name):
     if random.random() < 0.3:
         email = first_name + "." + last_name + "@" + domain
     else:
-        email =  nick + "@" + domain
+        email = nick + "@" + domain
     return email
 
+
+@require_nicknames
 def create_profile(nick=None):
-    nick = "".join(random.sample(NICKNAMES, random.randrange(2,5)))
+    nick = "".join(random.sample(NICKNAMES, random.randrange(2, 5)))
     first_name = random.choice(FIRSTNAMES)
     last_name = random.choice(LASTNAMES)
     email = generate_email(nick, first_name, last_name)
@@ -231,4 +256,7 @@ def create_profile(nick=None):
     )
     profile.search_engine = random.choice(SearchEngine.objects.all())
     profile.browser_fingerprint = random.choice(BrowserFingerprint.objects.all())
+
+    profile.full_clean()
     profile.save()
+    return profile