Created a new ArchivedURLSet class for lists of ArchivedURL objects. …

…Right now it's nothing more than a standard list that prevents duplicates and only allows ArchivedURL objects as members

storytracker/analysis.py

@@ -1,5 +1,6 @@
 import os
 import gzip
+import copy
 import storytracker
 from bs4 import BeautifulSoup
 
@@ -44,6 +45,23 @@ def hyperlinks(self):
         return link_list
 
 
+class ArchivedURLSet(list):
+    """
+    A list of archived URLs sorted by their timestamp
+    """
+    def append(self, obj):
+        # Verify that the user is trying to add an ArchivedURL object
+        if not isinstance(obj, ArchivedURL):
+            raise TypeError("Only ArchivedURL objects can be added")
+
+        # Check if the object is already in the list
+        if obj in [o for o in list(self.__iter__())]:
+            raise ValueError("This object is already in the list")
+
+        # If it's all true, append it.
+        super(ArchivedURLSet, self).append(copy.copy(obj))
+
+
 class Hyperlink(object):
     """
     A hyperlink extracted from an archived URL with tools for analysis

test.py

@@ -7,7 +7,9 @@
 import storytracker
 from datetime import datetime
 from bs4 import BeautifulSoup
-from storytracker.analysis import ArchivedURL, Hyperlink
+from storytracker.analysis import ArchivedURL
+from storytracker.analysis import ArchivedURLSet
+from storytracker.analysis import Hyperlink
 
 
 class NullDevice():
@@ -100,6 +102,20 @@ def test_url_hyperlinks(self):
         self.assertTrue(isinstance(obj.hyperlinks, list))
         [self.assertTrue(isinstance(a, Hyperlink)) for a in obj.hyperlinks]
 
+    def test_urlset_creation(self):
+        obj = ArchivedURL(self.url, datetime.now(), "foobar")
+        obj2 = ArchivedURL(self.url, datetime.now(), "foobar")
+        obj3 = ArchivedURL(self.url, datetime.now(), "foobar")
+        urlset = ArchivedURLSet([obj, obj2])
+        self.assertEqual(len(urlset), 2)
+        with self.assertRaises(TypeError):
+            urlset.append(1)
+            urlset.append([1,2,3])
+        with self.assertRaises(ValueError):
+            urlset.append(obj)
+        urlset.append(obj3)
+        self.assertEqual(len(urlset), 3)
+
 
 if __name__ == '__main__':
     if six.PY3: