feat: add tests for pastebin url analyzer (#109)

* feat: add tests for pastebin url analyzer

* fix: remove www from valid url

* feat(url-analyzer): make pastbin www tests to positive

* FIX: Analyzer should only catch certain links

The pastebinurlanalyzer should only match on certain strings in the URL. I specified it a bit more precisely

pastepwn/analyzers/pastebinurlanalyzer.py

@@ -7,5 +7,5 @@ class PastebinURLAnalyzer(URLAnalyzer):
     name = "PastebinURLAnalyzer"
 
     def __init__(self, actions, resolve=False):
-        regex = r"((?:https?:\/\/)?pastebin.com\/\S+)"
+        regex = r"((?:https?:\/\/)?pastebin.com\/[a-zA-Z0-9]{3,})"
         super().__init__(actions, regex, resolve)

pastepwn/analyzers/tests/pastebinurlanalyzer_test.py

@@ -0,0 +1,43 @@
+# -*- coding: utf-8 -*-
+import unittest
+from unittest import mock
+
+from pastepwn.analyzers.pastebinurlanalyzer import PastebinURLAnalyzer
+
+
+class TestPastebinURLAnalyzer(unittest.TestCase):
+    def setUp(self):
+        self.analyzer = PastebinURLAnalyzer(None)
+        self.paste = mock.Mock()
+
+    def test_match_positive(self):
+        self.paste.body = "https://pastebin.com/xyz"
+        self.assertTrue(self.analyzer.match(self.paste))
+        self.paste.body = "https://pastebin.com/xyz/"
+        self.assertTrue(self.analyzer.match(self.paste))
+        self.paste.body = "http://pastebin.com/xyz"
+        self.assertTrue(self.analyzer.match(self.paste))
+        self.paste.body = "http://pastebin.com/xyz/"
+        self.assertTrue(self.analyzer.match(self.paste))
+        self.paste.body = "https://pastebin.com/xyz  "
+        self.assertTrue(self.analyzer.match(self.paste))
+        self.paste.body = "www.pastebin.com/xyz"
+        self.assertTrue(self.analyzer.match(self.paste))
+        self.paste.body = "pastebin.com/xyx"
+        self.assertTrue(self.analyzer.match(self.paste))
+        self.paste.body = "This is a pastebin URL: pastebin.com/ya249asd - and this is a test"
+
+    def test_match_negative(self):
+        self.paste.body = ""
+        self.assertFalse(self.analyzer.match(self.paste))
+        self.paste.body = None
+        self.assertFalse(self.analyzer.match(self.paste))
+        self.paste.body = "https://google.com/xyz"
+        self.assertFalse(self.analyzer.match(self.paste))
+        self.paste.body = "xyzpastebin.com/k"
+        self.assertFalse(self.analyzer.match(self.paste))
+        self.paste.body = "https://pastebin.com/"
+        self.assertFalse(self.analyzer.match(self.paste))
+
+    if __name__ == '__main__':
+        unittest.main()