Python3 support and linting

egghatch/block.py

@@ -8,6 +8,7 @@
     "Instruction", ("addr", "size", "mnemonic", "operands")
 )
 
+
 class Block(object):
     stop_insns = (
         "jmp", "jecxz", "ret", "loop", "loope", "loopne",

egghatch/main.py

@@ -7,16 +7,19 @@
 from egghatch.misc import str_as_db
 from egghatch.shellcode import Shellcode
 
+
 def main():
     if len(sys.argv) != 2:
-        print "Usage: python %s <sc.bin>" % sys.argv[0]
+        print("Usage: python %s <sc.bin>" % sys.argv[0])
         exit(1)
 
-    print Shellcode(open(sys.argv[1], "rb").read()).to_json()
+    print(Shellcode(open(sys.argv[1], "rb").read()).to_json())
+
 
 def parse(payload):
     return Shellcode(payload).to_dict()
 
+
 def as_text(payload):
     ret, sc = [], Shellcode(payload).to_dict()
 

egghatch/misc.py

@@ -14,12 +14,12 @@ def str_as_db(s):
 
     r2, idx = [], 0
     while idx < len(r1):
-        if isinstance(r1[idx], (int, long)):
+        if isinstance(r1[idx], (int, float)):
             r2.append("%s" % r1[idx])
             idx += 1
             continue
         jdx = idx
-        while idx < len(r1) and isinstance(r1[idx], basestring):
+        while idx < len(r1) and isinstance(r1[idx], str):
             idx += 1
         r2.append("'%s'" % "".join(r1[jdx:idx]))
 

egghatch/shellcode.py

@@ -6,6 +6,7 @@
 
 from egghatch.block import Block
 
+
 class Shellcode(object):
     def __init__(self, payload):
         import capstone
@@ -63,14 +64,14 @@ def handle_relative_call(self, block):
     def add_bbl(self, start, end):
         bbl_r = dict((v, k) for k, v in self.bbl.items())
         for start_, end_ in self.bbl.items():
-            if start >= start_ and start < end_:
+            if start_ <= start < end_:
                 self.bbl[start_] = start
                 self.bbl[start] = end_
                 break
             if start < start_ and end == end_:
                 self.bbl[start] = bbl_r.get(start_, start_)
                 break
-            if end and end > start_ and end <= end_:
+            if end and start_ < end <= end_:
                 self.bbl[start] = start_
                 self.bbl[start_] = end
                 if end != end_:
@@ -135,7 +136,7 @@ def basic_taint(self):
                 continue
 
             op = insn2.operands
-            for _ in xrange(64):
+            for _ in range(64):
                 if insn2.addr + insn2.size not in insns:
                     break
 
@@ -156,7 +157,7 @@ def extract_data(self):
         parsed[len(self.payload)] = len(self.payload)
 
         chunks = sorted(parsed.items())
-        for idx in xrange(1, len(chunks)):
+        for idx in range(1, len(chunks)):
             _, start = chunks[idx-1]
             end, _ = chunks[idx]
             if start != end and start < end:

setup.py

@@ -22,13 +22,13 @@
         "Natural Language :: English",
         "License :: OSI Approved :: GNU General Public License v3 (GPLv3)",
         "Operating System :: POSIX :: Linux",
-        "Programming Language :: Python :: 2.7",
+        "Programming Language :: Python :: 3.8",
         "Topic :: Security",
     ],
     url="https://cuckoosandbox.org/",
     license="GPLv3",
     description="Cuckoo Sandbox Shellcode Identification & Formatting",
-    long_description=open("README.rst", "rb").read(),
+    long_description=str(open("README.rst", "rb").read()),
     include_package_data=True,
     entry_points={
         "console_scripts": [
@@ -42,10 +42,10 @@
             "capstone-windows==3.0.4",
         ],
         ":sys_platform == 'darwin'": [
-            "capstone==3.0.5rc2",
+            "capstone==4.0.2",
         ],
         ":sys_platform == 'linux2'": [
-            "capstone==3.0.5rc2",
+            "capstone==4.0.2",
         ],
     },
 )

tests/test_blocks.py

@@ -4,8 +4,9 @@
 
 from egghatch.shellcode import Shellcode
 
+
 def test_parse():
-    sc = Shellcode("\x90\x75\x02\x90\x90\x90")
+    sc = Shellcode(b"\x90\x75\x02\x90\x90\x90")
     assert sc.to_dict() == {
         "text": [
             (0, 1, "nop", ""),
@@ -22,6 +23,7 @@ def test_parse():
         "data": [],
     }
 
+
 def test_add_bbl1():
     sc = Shellcode("")
     sc.parsed[97] = False
@@ -39,6 +41,7 @@ def test_add_bbl1():
         136: 192,
     }
 
+
 def test_add_bbl2():
     sc = Shellcode("")
     sc.parsed[209] = False
@@ -58,6 +61,7 @@ def test_add_bbl2():
         290: 308,
     }
 
+
 def test_sd():
     sc = Shellcode(open("tests/files/plain/sd.bin", "rb").read())
     assert sc.to_dict()["bbl"] == [
@@ -66,6 +70,7 @@ def test_sd():
         (0x39, 0x45),
     ]
 
+
 def test_bin1():
     sc = Shellcode(open("tests/files/plain/1.bin", "rb").read())
     assert sc.to_dict()["bbl"] == [
@@ -100,6 +105,7 @@ def test_bin1():
         (0x14b, "www.service.chrome-up.date\x00"),
     ]
 
+
 def test_bin2():
     sc = Shellcode(open("tests/files/plain/2.bin", "rb").read())
     assert sc.to_dict()["bbl"] == [
@@ -131,6 +137,7 @@ def test_bin2():
         (192, "ddos400.ddns.net\x00"),
     ]
 
+
 def test_bin3():
     sc = Shellcode(open("tests/files/plain/3.bin", "rb").read())
     assert sc.to_dict()["bbl"] == [

tests/test_json.py

@@ -7,6 +7,7 @@
 
 from egghatch.shellcode import Shellcode
 
+
 def test_sd():
     sc = Shellcode(open("tests/files/plain/sd.bin", "rb").read())
     assert json.loads(sc.to_json()) == {
@@ -49,6 +50,7 @@ def test_sd():
         "data": [],
     }
 
+
 def test_bin1():
     sc = Shellcode(open("tests/files/plain/1.bin", "rb").read())
     assert json.loads(sc.to_json()) == {
@@ -60,6 +62,7 @@ def test_bin1():
         ],
     }
 
+
 def test_bin2():
     sc = Shellcode(open("tests/files/plain/1.bin", "rb").read())
     assert json.loads(sc.to_json()) == {

tests/test_main.py

@@ -4,6 +4,7 @@
 
 from egghatch import parse, as_text
 
+
 def test_parse():
     assert parse("\xfc\xeb\xfe") == {
         "bbl": [
@@ -17,6 +18,7 @@ def test_parse():
         "data": [],
     }
 
+
 def test_as_text_cld_jmpinf():
     assert as_text("\xfc\xeb\xfe") == (
         "bbl_0x0000:\n"
@@ -25,6 +27,7 @@ def test_as_text_cld_jmpinf():
         "    0x0001: jmp 1\n"
     )
 
+
 def test_as_text_sc():
     def f(filename):
         return open("tests/files/plain/%s" % filename, "rb").read()

tests/test_misc.py

@@ -4,6 +4,7 @@
 
 from egghatch.misc import str_as_db
 
+
 def test_str_as_db():
     assert str_as_db("\x00") == "0"
     assert str_as_db("foo") == "'foo'"