1 files changed, 98 insertions, 0 deletions

diff --git a/modules/libfuzzy.py b/modules/libfuzzy.py
new file mode 100644
index 0000000..a0d3f15
--- /dev/null
+++ b/modules/libfuzzy.py
@@ -0,0 +1,98 @@
+''' Ugly-pseudo-bindings to libfuzzy (used by ssdeep) to check
+if a file is similar to a given list of samples
+'''
+
+import os
+import ctypes
+import pickle
+import sys
+import logging
+logging.basicConfig(level=logging.DEBUG)
+
+import scanmodule
+
+
+SPAMSUM_LENGTH = 64
+FUZZY_MAX_RESULT = SPAMSUM_LENGTH + SPAMSUM_LENGTH // 2 + 20
+
+def main():
+    return FuzzyMatcher()
+
+class FuzzyMatcher(scanmodule.ScanModule):
+    name = 'libfuzzy'
+    def __init__(self, samples_path=None, persistence_path=None):
+        self.__initialize_libfuzzy()
+        super(FuzzyMatcher, self).__init__()
+
+    def __initialize_libfuzzy(self):
+        ''' Bind to libfuzzy thanks to ctypes.
+        This will create the "fuzzy_hash_buf" and
+        the "fuzzy_compare" methods
+        '''
+        try:
+            fuzzy = ctypes.CDLL('libfuzzy.so')
+        except OSError:
+            print('[-] Please check that you installed libfuzzy')
+            sys.exit(1)
+
+        self.__fuzzy_hash_buf = fuzzy.fuzzy_hash_buf
+        self.__fuzzy_hash_buf.restype = ctypes.c_int
+        self.__fuzzy_hash_buf.argtypes = [
+            ctypes.c_char_p, #buf
+            ctypes.c_uint32, #buf_len
+            ctypes.c_char_p, #result
+        ]
+        self.__fuzzy_compare = fuzzy.fuzzy_compare
+        self.__fuzzy_compare.restype = ctypes.c_int
+        self.__fuzzy_compare.argtypes = [
+            ctypes.c_char_p, #sig1
+            ctypes.c_char_p, #sig2
+        ]
+
+    def populate(self, path):
+        ''' Computes fuzzy hashes of files under the given path,
+        and store them in the dict self.samples with the form dict {name: fuzzy_hash}
+        @param path Path containing the samples
+        '''
+        for root, _, filenames in os.walk(path):
+            for filename in filenames:
+                full_path = os.path.join(root, filename)
+                self.samples[full_path] = self.__hash_from_file(full_path)
+
+    def __hash_from_file(self, path):
+        ''' Return the hash of the given file
+        @param path Path to the file to hash
+        @ret Fuzzy hash of the given file
+        '''
+        with open(path, 'r') as f:
+            out = ctypes.create_string_buffer('\x00' * FUZZY_MAX_RESULT)
+            content = f.read()
+            self.__fuzzy_hash_buf(content, len(content), out)
+            return out.value
+
+    def evaluate(self, path):
+        ''' Compare the hash of the given path to every samples one.
+        @ret A sorted list of the form [name, match_in_percent_superior_to_zero]
+        '''
+        fuzzy_hash = self.__hash_from_file(path)
+
+        lst = list()
+        for f in self.samples:
+            score = self.__fuzzy_compare(fuzzy_hash, self.samples[f])
+            if score:
+                lst.append([f, score])
+        return sorted(lst, key=lambda lst: lst[1], reverse=True)
+
+    def is_malware(self, path):
+        max_score = 0
+        fuzzy_hash = self.__hash_from_file(path)
+
+        for f in self.samples:
+            score = self.__fuzzy_compare(fuzzy_hash, self.samples[f])
+            if score > max_score:
+                score = max_score
+                logging.info('fuzzy score for ' + path + ' matches ' + f + ' at ' + str(score) + '%%')
+
+        return max_score > 90
+
+