Implement Mega Link Analyzer (#141)

* Implement Mega Link Analyzer

* Fixed failed test

changed regex to match url boundary

pastepwn/analyzers/__init__.py

@@ -33,6 +33,7 @@
 from .awsaccesskeyanalyzer import AWSAccessKeyAnalyzer
 from .googleoauthkeyanalyzer import GoogleOAuthKeyAnalyzer
 from .slackwebhookanalyzer import SlackWebhookAnalyzer
+from .megalinkanalyzer import MegaLinkAnalyzer
 
 
 __all__ = (
@@ -64,9 +65,10 @@
     'EmailPasswordPairAnalyzer',
     'FacebookAccessTokenAnalyzer',
     'Base64Analyzer',
-    'StripeApiKeyAnalyzer',
+    'SlackWebhookAnalyzer',
+    'MegaLinkAnalyzer',
     'AWSSecretKeyAnalyzer',
     'AWSAccessKeyAnalyzer',
     'GoogleOAuthKeyAnalyzer',
-    'SlackWebhookAnalyzer'
+    'StripeApiKeyAnalyzer'
 )

pastepwn/analyzers/megalinkanalyzer.py

@@ -0,0 +1,13 @@
+# -*- coding: utf-8 -*-
+from .regexanalyzer import RegexAnalyzer
+
+
+class MegaLinkAnalyzer(RegexAnalyzer):
+    """
+    Analyzer to match mega links via regex
+    """
+    name = "MegaLinkAnalyzer"
+
+    def __init__(self, actions):
+        regex = r"\b(https?:\/\/)?mega\.nz\/#F![\w-]{8}(![\w-]{22}(![\w-]{8})?)?\b"
+        super().__init__(actions, regex)

pastepwn/analyzers/tests/megalinkanalyzer_test.py

@@ -0,0 +1,53 @@
+# -*- coding: utf-8 -*-
+import unittest
+from unittest import mock
+
+from pastepwn.analyzers.megalinkanalyzer import MegaLinkAnalyzer
+
+
+class TestMegaLinkAnalyzer(unittest.TestCase):
+    def setUp(self):
+        self.analyzer = MegaLinkAnalyzer(None)
+        self.paste = mock.Mock()
+
+    def test_match_positive(self):
+        """Test if positives are recognized"""
+        # working mega link (short version)
+        self.paste.body = "https://mega.nz/#F!XTQVEAZZ"
+        self.assertTrue(self.analyzer.match(self.paste))
+
+        # working mega link (medium version)
+        self.paste.body = "https://mega.nz/#F!XTQVEAZZ!eqxlOvTxJKnvAkYvjC0O8g"
+        self.assertTrue(self.analyzer.match(self.paste))
+
+        # working mega link (long version)
+        self.paste.body = "https://mega.nz/#F!PB8SSawR!SUokSlF2Zy8CR004DNFfNw!LQtniCoa"
+        self.assertTrue(self.analyzer.match(self.paste))
+
+        # without https header
+        self.paste.body = "mega.nz/#F!XTQVEAZZ!eqxlOvTxJKnvAkYvjC0O8g"
+        self.assertTrue(self.analyzer.match(self.paste))
+
+        # http header
+        self.paste.body = "http://mega.nz/#F!XTQVEAZZ!eqxlOvTxJKnvAkYvjC0O8g"
+        self.assertTrue(self.analyzer.match(self.paste))
+
+        # in a sentence
+        self.paste.body = "check out this file:https://mega.nz/#F!PB8SSawR!SUokSlF2Zy8CR004DNFfNw!LQtniCoa"
+        self.assertTrue(self.analyzer.match(self.paste))
+
+    def test_match_negative(self):
+        """Test if negatives are not recognized"""
+        self.paste.body = ""
+        self.assertFalse(self.analyzer.match(self.paste))
+
+        self.paste.body = None
+        self.assertFalse(self.analyzer.match(self.paste))
+
+        # Invalid segment length
+        self.paste.body = "https://mega.nz/#F!XTQVEAZZ1!eqxlOvTxJKnvAkYvjC0O8g"
+        self.assertFalse(self.analyzer.match(self.paste))
+
+
+if __name__ == '__main__':
+    unittest.main()