Fix asmsearch in Python 3

lib/nasm.py

@@ -37,7 +37,7 @@ class Nasm(object):
         asmcode = decode_string_escape(asmcode)
         asmcode = re.sub("PTR|ptr|ds:|DS:", "", asmcode)
         infd = tmpfile()
-        outfd = tmpfile()
+        outfd = tmpfile(binary_file=True)
         infd.write(asmcode)
         infd.flush()
         execute_external_command("%s -f bin -o %s %s" % (config.NASM, outfd.name, infd.name))

lib/utils.py

@@ -726,7 +726,7 @@ def _decode_string_escape_py3(str_):
     """
 
     # Based on: http://stackoverflow.com/a/4020824
-    return str_.decode("unicode_escape")
+    return codecs.decode(str_, "unicode_escape")
 
 
 def decode_string_escape(str_):
@@ -793,15 +793,52 @@ def _bytes_chr_py3(i):
     return bytes([i])
 
 
+def to_binary_string(text):
+    """
+    Converts a string to a binary string if it is not already one. Returns a str
+    in Python 2 and a bytes in Python3.
+
+    Use this instead of six.b when the text may already be a binary type
+    """
+    raise Exception('Should be overriden')
+
+
+def _to_binary_string_py2(text):
+    """
+    Converts a string to a binary string if it is not already one. Returns a str
+    in Python 2 and a bytes in Python3.
+
+    Do not use directly, use to_binary_string instead.
+    """
+    return str(text)
+
+
+def _to_binary_string_py3(text):
+    """
+    Converts a string to a binary string if it is not already one. Returns a str
+    in Python 2 and a bytes in Python3.
+
+    Do not use directly, use to_binary_string instead.
+    """
+    if isinstance(text, six.binary_type):
+        return text
+    elif isinstance(text, six.string_types):
+        return six.b(text)
+    else:
+        raise Exception('only takes string types')
+
+
 # Select functions based on Python version
 if six.PY2:
     decode_string_escape = _decode_string_escape_py2
     bytes_iterator = _bytes_iterator_py2
     bytes_chr = _bytes_chr_py2
+    to_binary_string = _to_binary_string_py2
 elif six.PY3:
     decode_string_escape = _decode_string_escape_py3
     bytes_iterator = _bytes_iterator_py3
     bytes_chr = _bytes_chr_py3
+    to_binary_string = _to_binary_string_py3
 else:
     raise Exception("Could not identify Python major version")
 

peda.py

@@ -2593,7 +2593,7 @@ class PEDA(object):
         magic_bytes = ["0x00", "0xff", "0xdead", "0xdeadbeef", "0xdeadbeefdeadbeef"]
 
         ops = [x for x in asmcode.split(';') if x]
-        def buildcode(code="", pos=0, depth=0):
+        def buildcode(code=b"", pos=0, depth=0):
             if depth == wildcard and pos == len(ops):
                 yield code
                 return
@@ -2603,6 +2603,7 @@ class PEDA(object):
             elif c == 0:
                 asm = self.assemble(ops[pos])
                 if asm:
+                    print(repr(code) + '; ' + repr(asm))
                     for code in buildcode(code + asm, pos+1, depth):
                         yield code
             else:
@@ -2626,16 +2627,16 @@ class PEDA(object):
 
         for machine_code in buildcode():
             search = re.escape(machine_code)
-            search = search.replace(decode_hex_escape(b"dead"),"..")\
-                .replace(decode_hex_escape(b"beef"),"..")\
-                .replace(decode_hex_escape(b"00"),".")\
-                .replace(decode_hex_escape(b"ff"),".")
+            search = search.replace(decode_hex_escape(b"dead"), b"..")\
+                .replace(decode_hex_escape(b"beef"), b"..")\
+                .replace(decode_hex_escape(b"00"), b".")\
+                .replace(decode_hex_escape(b"ff"), b".")
 
             if rop and 'ret' not in asmcode:
-                search = search + ".{0,24}\\xc3"
-            searches.append("%s" % (search))
+                search += b".{0,24}\\xc3"
+            searches.append(search)
 
-        search = "(?=(%s))" % ("|".join(searches))
+        search = b"(?=(" + b"|".join(searches) + b"))"
         candidates = self.searchmem(start, end, search)
 
         if rop:
@@ -5056,7 +5057,7 @@ class PEDACmd(object):
         if result:
             text = ""
             for (addr, (byte, code)) in result:
-                text += "%s : (%s)\t%s\n" % (to_address(addr), byte, code)
+                text += "%s : (%s)\t%s\n" % (to_address(addr), byte.decode('utf-8'), code)
         pager(text)
 
         return