Merge pull request #66 from d-Rickyy-b/dev

Multiple fixes and few new features

pastepwn/analyzers/bcrypthashanalyzer.py

@@ -7,5 +7,5 @@ class BcryptHashAnalyzer(RegexAnalyzer):
     name = "BcryptHashAnalyzer"
 
     def __init__(self, actions):
-        regex = r"\$2[ayb]\$.{56}"
+        regex = r"\$2[ayb]\$[\d]{2}\$[./A-Za-z0-9]{53}"
         super().__init__(actions, regex)

pastepwn/analyzers/genericanalyzer.py

@@ -13,6 +13,8 @@ def __init__(self, actions, match_func):
 
         if match_func is None:
             raise ValueError("Function to be called cannot be None")
+        elif not callable(match_func):
+            raise ValueError("Function you provided isn't callable")
 
         self.match_func = match_func
 
@@ -22,6 +24,6 @@ def match(self, paste):
             result = self.match_func(paste)
         except Exception as e:
             result = False
-            logging.getLogger(__name__).warning("Executing custom match function raised an exception! {}".format(e))
+            logging.getLogger(__name__).warning("Executing custom match function '{}' raised an exception! {}".format(self.match_func.__name__, e))
 
         return result

pastepwn/analyzers/regexanalyzer.py

@@ -14,5 +14,8 @@ def __init__(self, actions, regex):
 
     def match(self, paste):
         """Match the content of a paste via regex. Return true if regex matches"""
+        if paste is None:
+            return False
+
         paste_content = paste.body or ""
         return self.regex.search(paste_content) is not None

pastepwn/analyzers/tests/bcrypthashanalyzer_test.py

@@ -0,0 +1,59 @@
+# -*- coding: utf-8 -*-
+import unittest
+from unittest import mock
+
+from pastepwn.analyzers.bcrypthashanalyzer import BcryptHashAnalyzer
+
+
+class TestBcryptHashAnalyzer(unittest.TestCase):
+    def setUp(self):
+        self.analyzer = BcryptHashAnalyzer(None)
+        self.obj = mock.Mock()
+
+    def test_match(self):
+        valid_hashes = ["$2a$10$BIgnlSmYE8qYiONM0NQ53eRWBw5G4HIJEbXKzcsRVt.08IDnqH/V.",
+                        "$2a$11$EppiRqR0kG9EKy56edDWTOnsv/oGW0dqAJB9ucmn3augbmcm8v/iy",
+                        "$2b$10$FpOpno43SIE8e1hWnlOdR.9hG2J8dd5FD1kQq8hn4zLdKa5eIiFUO",
+                        "$2a$10$SVy7GlMnWsemiZByHSnV0O3WoEHGImFt8v07uH.K3ZXwH5j9o/DP.",
+                        "$2a$10$59SNkcZ0rdC2VgeWaavVyea9PFget/xmtbV7.9IeJl3CUq.Q954i2",
+                        "$2b$10$Y8CJ9YIwrxt1YYgMqqU1delCYoTpIl18SRtYYI2kyM3jduKPHvWMC",
+                        "$2a$10$2tNCUb.FUpSutyhkbqmMBuNnLzhqI4q9Miqurnj6eu.XsiIjww7I6",
+                        "$2a$10$OyrADUFmj9QEqsd8frkEDOEYSPQalW5qoI1s2z6taCWwgUsjKzk5m"]
+
+        invalid_hashes = ["7168D46050573DDA4CE409FA1515638BD28E86346D45F686310ED0678172BABCD4117FD15DD380B964352FE879FB745B573A730D526BB1188B2790FBA06E8ACA",
+                          "5FD924625F6AB16A19CC9807C7C506AE1813490E4BA675F843D5A10E0BAACDB8",
+                          "522F02FEA11E70C03C90C247C50410443246BFCB",
+                          "8433FD5A3B0ED71D21CFB9F291BD89B9",
+                          "$2a$124$SVy7GlMnWsemiZByHSnV0O3WoEHGImFt8v07uH.K3ZXwH5j9o/DP.a",
+                          "$2a$12$SVy7GlMnWsemiZByHSnV0O3WoEHGImFt8v{07uH.K3ZXwH5j9o/DP.a",
+                          "$2a$14$SVy7GlMnWsemiZByHSnV0O3WGImFt8v07uH.K3ZXwH5j9o/DP.a",
+                          "@x,Y8q+jnYeZr$;",
+                          "This is a test",
+                          "$2a$10$   asdf 1234 how are you?"]
+
+        for test_hash in valid_hashes:
+            self.obj.body = test_hash
+            self.assertTrue(self.analyzer.match(self.obj), test_hash)
+
+        for test_hash in invalid_hashes:
+            self.obj.body = test_hash
+            self.assertFalse(self.analyzer.match(self.obj), test_hash)
+
+    def test_match_none(self):
+        self.obj.body = None
+        self.assertFalse(self.analyzer.match(self.obj))
+
+        self.obj = None
+        self.assertFalse(self.analyzer.match(self.obj))
+
+    def test_match_empty(self):
+        self.obj.body = ""
+        self.assertFalse(self.analyzer.match(self.obj))
+
+    def test_actions_present(self):
+        analyzer = BcryptHashAnalyzer(self.obj)
+        self.assertEqual([self.obj], analyzer.actions)
+
+
+if __name__ == '__main__':
+    unittest.main()

pastepwn/analyzers/tests/wordanalyzer_test.py

@@ -0,0 +1,136 @@
+# -*- coding: utf-8 -*-
+import unittest
+from unittest import mock
+
+from pastepwn.analyzers.wordanalyzer import WordAnalyzer
+
+
+class TestWordAnalyzer(unittest.TestCase):
+    def setUp(self):
+        self.obj = mock.Mock()
+
+    def test_match(self):
+        analyzer = WordAnalyzer(None, "Test")
+        self.obj.body = "This is a Test"
+        self.assertTrue(analyzer.match(self.obj))
+
+        analyzer = WordAnalyzer(None, "Test")
+        self.obj.body = "There are tests for everything"
+        self.assertTrue(analyzer.match(self.obj))
+
+        self.obj.body = "This is a Test for a longer sentence"
+        self.assertTrue(analyzer.match(self.obj))
+
+        self.obj.body = "Completely unrelated"
+        self.assertFalse(analyzer.match(self.obj))
+
+    def test_blacklist(self):
+        blacklist = ["fake", "bad"]
+        analyzer = WordAnalyzer(None, "Test", blacklist=blacklist)
+
+        self.obj.body = "This is a Test"
+        self.assertTrue(analyzer.match(self.obj))
+
+        self.obj.body = "This is a fake Test"
+        self.assertFalse(analyzer.match(self.obj))
+
+        analyzer = WordAnalyzer(None, "Test", blacklist=blacklist, case_sensitive=True)
+
+        self.obj.body = "This is a Test"
+        self.assertTrue(analyzer.match(self.obj))
+
+        self.obj.body = "This is a Fake Test"
+        self.assertTrue(analyzer.match(self.obj))
+
+    def test_multiple_words(self):
+        analyzer = WordAnalyzer(None, None)
+        self.assertEqual(analyzer.words, [])
+
+        analyzer = WordAnalyzer(None, ["My", "first", "Test"])
+        self.obj.body = "This is a little test for something"
+        self.assertTrue(analyzer.match(self.obj))
+
+        self.obj.body = "You are my best friend so far!"
+        self.assertTrue(analyzer.match(self.obj))
+
+        self.obj.body = "This is the first time I try this"
+        self.assertTrue(analyzer.match(self.obj))
+
+        self.obj.body = "This time we try to match multiple words/tests for the first time."
+        self.assertTrue(analyzer.match(self.obj))
+
+        self.obj.body = "None of the words are contained!"
+        self.assertFalse(analyzer.match(self.obj))
+
+        # Check for case sensitivity for multiple words
+        analyzer2 = WordAnalyzer(None, ["My", "first", "Test"], case_sensitive=True)
+
+        self.obj.body = "That's not my issue!"
+        self.assertFalse(analyzer2.match(self.obj))
+
+        self.obj.body = "That's not My issue!"
+        self.assertTrue(analyzer2.match(self.obj))
+
+    def test_add_word(self):
+        analyzer = WordAnalyzer(None, "Test")
+        self.assertEqual(len(analyzer.words), 1)
+        self.assertEqual(analyzer.words, ["Test"])
+
+        analyzer.add_word("second")
+        self.assertEqual(len(analyzer.words), 2)
+        self.assertEqual(analyzer.words, ["Test", "second"])
+
+    def test_case_sensitive(self):
+        analyzer = WordAnalyzer(None, "Test", case_sensitive=True)
+        self.obj.body = "This is a Test for case sensitivity"
+        self.assertTrue(analyzer.match(self.obj))
+
+        self.obj.body = "This is a test for case sensitivity"
+        self.assertFalse(analyzer.match(self.obj))
+
+        self.obj.body = "This is a tESt for case sensitivity"
+        self.assertFalse(analyzer.match(self.obj))
+
+        analyzer2 = WordAnalyzer(None, "Te1st")
+        self.obj.body = "This is a te1st for case sensitivity"
+        self.assertTrue(analyzer2.match(self.obj))
+
+        analyzer2 = WordAnalyzer(None, "Te1st")
+        self.obj.body = "This is a tE1st for case sensitivity"
+        self.assertTrue(analyzer2.match(self.obj))
+
+    def test_multiple_case_sensitive(self):
+        """Test if it's possible to match any of multiple words in a wordanalyzer when case sensitivty is activated"""
+        analyzer = WordAnalyzer(None, ["My", "first", "Test"], case_sensitive=True)
+        self.obj.body = "This is a little test for something"
+        self.assertFalse(analyzer.match(self.obj))
+
+        self.obj.body = "You are my best friend so far!"
+        self.assertFalse(analyzer.match(self.obj))
+
+        self.obj.body = "This is a Test for case sensitivity"
+        self.assertTrue(analyzer.match(self.obj))
+
+        self.obj.body = "This is a test for case sensitivity. It's the first of its kind."
+        self.assertTrue(analyzer.match(self.obj))
+
+    def test_match_none(self):
+        analyzer = WordAnalyzer(None, "Test")
+        self.obj.body = None
+        self.assertFalse(analyzer.match(self.obj))
+
+        self.obj = None
+        self.assertFalse(analyzer.match(self.obj))
+
+    def test_match_empty(self):
+        analyzer = WordAnalyzer(None, "Test")
+        self.obj.body = ""
+        self.assertFalse(analyzer.match(self.obj))
+
+    def test_actions_present(self):
+        analyzer = WordAnalyzer(self.obj, "Test")
+        self.assertEqual([self.obj], analyzer.actions)
+
+
+if __name__ == '__main__':
+    unittest.main()

pastepwn/analyzers/wordanalyzer.py

@@ -3,17 +3,24 @@
 
 
 class WordAnalyzer(BasicAnalyzer):
-    """Analyzer to match the content of a paste via regular expressions"""
+    """Analyzer to match the content of a paste by words"""
     name = "WordAnalyzer"
 
-    def __init__(self, actions, word, blacklist=None, case_sensitive=False):
-        super().__init__(actions, "{0} ({1})".format(self.name, word))
-        self.word = word
+    def __init__(self, actions, words, blacklist=None, case_sensitive=False):
+        super().__init__(actions, "{0} ({1})".format(self.name, words))
+
+        if words is None:
+            self.words = []
+        elif isinstance(words, list):
+            self.words = words
+        else:
+            self.words = [words]
+
         self.blacklist = blacklist or []
         self.case_sensitive = case_sensitive
 
     def _blacklist_word_found(self, text):
-        if self.case_sensitive:
+        if not self.case_sensitive:
             text = text.lower()
             self.blacklist = [x.lower() for x in self.blacklist]
 
@@ -23,14 +30,29 @@ def _blacklist_word_found(self, text):
 
         return False
 
+    def add_word(self, word):
+        """Add a word to the analyzer"""
+        self.words.append(word)
+
     def match(self, paste):
-        """Check if the specified word is part of the paste text"""
-        paste_content = paste.body
+        """Check if the specified words are part of the paste text"""
+        if paste is None:
+            return False
+
+        paste_content = paste.body or ""
 
         if self._blacklist_word_found(paste_content):
             return False
 
         if self.case_sensitive:
-            return self.word in paste_content
+            for word in self.words:
+                # Never use 'return word in paste_content' - otherwise you will
+                # return false before all words have been checked
+                if word in paste_content:
+                    return True
         else:
-            return self.word.lower() in paste_content.lower()
+            for word in self.words:
+                if word.lower() in paste_content.lower():
+                    return True
+
+        return False

pastepwn/core/pastedispatcher.py

@@ -9,6 +9,7 @@
 
 
 class PasteDispatcher(object):
+    """The PasteDispatcher dispatches the downloaded pastes to the analyzers"""
 
     def __init__(self, paste_queue, action_queue=None, exception_event=None):
         self.logger = logging.getLogger(__name__)
@@ -28,7 +29,9 @@ def _pool_thread(self):
             pass
 
     def add_analyzer(self, analyzer):
-        self.analyzers.append(analyzer)
+        """Adds an analyzer to the list of analyzers"""
+        with self.__lock:
+            self.analyzers.append(analyzer)
 
     def start(self, workers=4, ready=None):
         """Starts dispatching the downloaded pastes to the list of analyzers"""
@@ -81,6 +84,7 @@ def _process_paste(self, paste):
                 self.action_queue.put((actions, paste, analyzer))
 
     def stop(self):
+        """Stops dispatching pastes to the analyzers"""
         self.__stop_event.set()
         while self.running:
             sleep(0.1)