Spaces:
Sleeping
Sleeping
| import editdistance | |
| from hexdump2 import hexdump | |
| import gradio as gr | |
| import shlex | |
| import subprocess | |
| import tempfile | |
| description = """This is a space testing a method for evaluating the quality of decompilation. | |
| Currently unhandled features: | |
| * PIC stuff | |
| * Global references | |
| * Function calls | |
| * Wildcards in target function? | |
| * How to extract compilable decompilation from decompilers? | |
| """ | |
| def trim(str, n): | |
| return "\n".join(str.splitlines()[n:]) | |
| def trim_objdump(str): | |
| return trim(str, 7) | |
| def disassemble_bytes(byte_data, architecture): | |
| with tempfile.NamedTemporaryFile(suffix=".bin", delete=False) as temp_bin_file: | |
| temp_bin_file.write(byte_data) | |
| temp_bin_file_name = temp_bin_file.name | |
| disassembly = subprocess.run( | |
| ["objdump", "-D", "-b", "binary", "-m", architecture, temp_bin_file_name], | |
| capture_output=True, | |
| text=True | |
| ).stdout | |
| disassembly = trim_objdump(disassembly) | |
| return disassembly | |
| def compile(compiler, flags, source): | |
| # Create a temporary file for the C source code | |
| with tempfile.NamedTemporaryFile(suffix=".c", delete=False) as temp_c_file: | |
| temp_c_file.write(source.encode()) | |
| temp_c_file_name = temp_c_file.name | |
| # Create a temporary file for the object file | |
| with tempfile.NamedTemporaryFile(suffix=".o", delete=False) as temp_o_file: | |
| temp_o_file_name = temp_o_file.name | |
| # Compile the C file to an object file | |
| result = subprocess.run( | |
| [compiler, "-c", temp_c_file_name] | |
| + shlex.split(flags) | |
| + ["-o", temp_o_file_name], | |
| capture_output=True, | |
| text=True, | |
| ) | |
| compile_output = result.stdout + result.stderr | |
| # Create a temporary file for the raw bytes | |
| with tempfile.NamedTemporaryFile(suffix=".raw", delete=True) as raw_bytes_file: | |
| subprocess.run( | |
| [ | |
| "objcopy", | |
| "--only-section", | |
| ".text", | |
| # XXX in reality we should probably look at the sections | |
| "--only-section", | |
| ".text.*", | |
| "-O", | |
| "binary", | |
| temp_o_file_name, | |
| raw_bytes_file.name, | |
| ] | |
| ) | |
| compiled_bytes = raw_bytes_file.read() | |
| # Disassemble the object file | |
| disassembly = subprocess.run( | |
| ["objdump", "-d", temp_o_file_name], | |
| capture_output=True, | |
| text=True | |
| ).stdout | |
| disassembly = trim_objdump(disassembly) | |
| if result.returncode == 0: | |
| return compiled_bytes, compile_output, disassembly | |
| else: | |
| return None, compile_output, disassembly | |
| def predict(target_bytes, source, compiler, flags, architecture): | |
| target_bytes = bytes.fromhex(target_bytes) | |
| compiled_bytes, compile_output, compiled_disassembly = compile(compiler, flags, source) | |
| target_disassembly = disassemble_bytes(target_bytes, architecture) | |
| if compiled_bytes is not None: | |
| return ( | |
| hexdump(compiled_bytes, result="return"), | |
| hexdump(target_bytes, result="return"), | |
| editdistance.eval(compiled_bytes, target_bytes), | |
| compile_output, | |
| compiled_disassembly, | |
| target_disassembly | |
| ) | |
| else: | |
| return ( | |
| "Compilation failed", | |
| hexdump(target_bytes, result="return"), | |
| -1, | |
| compile_output, | |
| compiled_disassembly, | |
| target_disassembly | |
| ) | |
| def run(): | |
| demo = gr.Interface( | |
| fn=predict, | |
| description=description, | |
| inputs=[ | |
| gr.Textbox( | |
| lines=10, | |
| label="Bytes of Target Function (in hex)", | |
| value="b8 2a 00 00 00 c3", | |
| ), | |
| gr.Textbox( | |
| lines=10, | |
| label="Decompiled C Source Code", | |
| value="int foo() { return 0; }", | |
| ), | |
| gr.Textbox(label="Compiler", value="g++"), | |
| gr.Textbox(label="Compiler Flags", value="-O2"), | |
| gr.Textbox(label="Architecture (for disassembler)", value="i386"), | |
| ], | |
| outputs=[ | |
| gr.Textbox(label="Compiled bytes"), | |
| gr.Textbox(label="Target bytes"), | |
| gr.Number(label="Edit distance (lower is better)"), | |
| gr.Textbox(label="Compiler Output"), | |
| gr.Textbox(label="Compiled Disassembly"), | |
| gr.Textbox(label="Target Disassembly"), | |
| ], | |
| ) | |
| demo.launch(server_name="0.0.0.0", server_port=7860, show_error=True) | |
| run() | |