zack3d/pylingual
1
0
Fork 0
mirror of https://github.com/syssec-utd/pylingual.git synced 2026-05-10 18:39:03 -07:00
Code Issues Packages Projects Releases Wiki Activity

eval script type nits

Browse Source
This commit is contained in:
Josh Wiedemeier
2025-12-15 11:18:19 -06:00
parent a8e5a0aa44
commit 599c63a3c1
1 changed files with 36 additions and 33 deletions
Download Patch File Download Diff File Expand all files Collapse all files
dev_scripts/evaluate_decompiler.py
+36 -33
View File
@@ -2,75 +2,78 @@ import pathlib
import tqdm
import click
import csv
from typing import TextIO
from datetime import datetime
import signal
from pylingual.decompiler import decompile
@click.command(help= "Evaluation script for pylingual")
@click.argument("pyc_list")
@click.argument("out_dir")
def main(pyc_list, out_dir):
@click.command(help="Evaluation script for pylingual")
@click.argument("pyc_list", type=click.File("r"))
@click.argument("out_dir", type=click.Path(file_okay=False, dir_okay=True, writable=True, path_type=pathlib.Path))
def main(pyc_list: TextIO, out_dir: pathlib.Path):
start_time = datetime.now()
def timeout_handler(signum, frame):
raise TimeoutError()
signal.signal(signal.SIGALRM, timeout_handler)
pyc_list = pathlib.Path(pyc_list)
out_dir = pathlib.Path(out_dir) / f"pylingual-{datetime.now().strftime('%Y-%m-%d_%H-%M-%S')}"
pyc_files = [pathlib.Path(pyc_path_line.strip()) for pyc_path_line in pyc_list.read_text().splitlines()]
out_dir = out_dir / f"pylingual-{datetime.now().strftime('%Y-%m-%d_%H-%M-%S')}"
pyc_files = [pathlib.Path(pyc_path_line.strip()) for pyc_path_line in pyc_list.readlines()]
out_dir.mkdir(parents=True, exist_ok=True)
evaluation_results_file = out_dir / 'evaluation_results.csv'
evaluation_results_stream = evaluation_results_file.open('w', newline='')
evaluation_writer = csv.DictWriter(evaluation_results_stream, fieldnames=['pyc_file', 'py_file', 'identifier', 'success', 'category', 'notes'])
evaluation_results_file = out_dir / "evaluation_results.csv"
evaluation_results_stream = evaluation_results_file.open("w", newline="")
evaluation_writer = csv.DictWriter(evaluation_results_stream, fieldnames=["pyc_file", "py_file", "identifier", "success", "category", "notes"])
evaluation_writer.writeheader()
# decompile all the pyc files
total_files_succeeded = 0
total_files_attempted = 0
total_files_attempted = 0
for pyc_file in (evaluation_progress := tqdm.tqdm(pyc_files)):
decompiler_results_dir = out_dir / 'decompilation_results'
decompiler_results_dir = out_dir / "decompilation_results"
decompiler_results_dir.mkdir(parents=True, exist_ok=True)
target_out_dir = decompiler_results_dir / pyc_file.parent.name
target_out_dir = decompiler_results_dir / pyc_file.parent.name
identifier = str(pyc_file).split("/")[-2]
# update progress bar
if total_files_attempted > 0:
evaluation_progress.set_postfix({
'file_success': f'{total_files_succeeded}/{total_files_attempted} ({total_files_succeeded / total_files_attempted:.2%})',
})
evaluation_progress.set_postfix(
{
"file_success": f"{total_files_succeeded}/{total_files_attempted} ({total_files_succeeded / total_files_attempted:.2%})",
}
)
total_files_attempted += 1
# decompile the file
try:
signal.alarm(300) # 5-minute timeout for decompiling one file
signal.alarm(300) # 5-minute timeout for decompiling one file
py_file = decompile(pyc_file, target_out_dir)
signal.alarm(0) # success; disable timer
signal.alarm(0) # success; disable timer
except Exception as err:
signal.alarm(0)
evaluation_writer.writerow({'pyc_file': pyc_file, 'py_file': '', 'identifier': 'FILE', 'success': False, 'category': 'DECOMPILER ERROR', 'notes': repr(err)})
signal.alarm(0)
evaluation_writer.writerow({"pyc_file": pyc_file, "py_file": "", "identifier": "FILE", "success": False, "category": "DECOMPILER ERROR", "notes": repr(err)})
continue
if all([result.success for result in py_file.equivalence_results]):
evaluation_writer.writerow({'pyc_file': pyc_file, 'py_file': pyc_file.parent.name, 'identifier': 'FILE', 'success': True, 'category': 'Equal', 'notes': ''})
evaluation_writer.writerow({"pyc_file": pyc_file, "py_file": pyc_file.parent.name, "identifier": "FILE", "success": True, "category": "Equal", "notes": ""})
total_files_succeeded += 1
else:
evaluation_writer.writerow({'pyc_file': pyc_file, 'py_file': pyc_file.parent.name, 'identifier': 'FILE', 'success': False, 'category': 'Different', 'notes': ''})
evaluation_writer.writerows({'pyc_file': pyc_file, 'py_file': pyc_file.parent.name, 'identifier': identifier, 'success': result.success, 'notes': ''} for result in py_file.equivalence_results)
evaluation_writer.writerow({"pyc_file": pyc_file, "py_file": pyc_file.parent.name, "identifier": "FILE", "success": False, "category": "Different", "notes": ""})
evaluation_writer.writerows({"pyc_file": pyc_file, "py_file": pyc_file.parent.name, "identifier": identifier, "success": result.success, "notes": ""} for result in py_file.equivalence_results)
evaluation_results_stream.close()
elapsed_time = datetime.now() - start_time
with open(out_dir / 'elapsed_time.txt', 'w') as time_file:
with open(out_dir / "elapsed_time.txt", "w") as time_file:
time_file.write(f"Elapsed Time: {str(elapsed_time)}\n")
time_file.write(f"File success: {total_files_succeeded}/{total_files_attempted} {total_files_succeeded/total_files_attempted :.2%}")
time_file.write(f"File success: {total_files_succeeded}/{total_files_attempted} {total_files_succeeded / total_files_attempted:.2%}")
if __name__ == "__main__":
main()