You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
AppInfoScanner/libs/task/web_task.py

46 lines
1.8 KiB

#! /usr/bin/python3
# -*- coding: utf-8 -*-
# Author: kelvinBen
# Github: https://github.com/kelvinBen/AppInfoScanner
import os
import config
import hashlib
from queue import Queue
class WebTask(object):
thread_list =[]
value_list = []
result_dict = {}
def __init__(self, path):
self.path = path
self.file_queue = Queue()
self.file_identifier = []
def start(self):
if len(config.web_file_suffix) <=0:
scanner_file_suffix = ["html","js","html","xml"]
scanner_file_suffix = config.web_file_suffix
if os.path.isdir(self.path):
self.__get_scanner_file__(self.path,scanner_file_suffix)
else:
if not (self.path.split(".")[-1] in scanner_file_suffix):
err_info = ("Retrieval of this file type is not supported. Select a file or directory with a suffix of %s" % ",".join(scanner_file_suffix))
raise Exception(err_info)
self.file_queue.put(self.path)
return {"comp_list":[],"shell_flag":False,"file_queue":self.file_queue,"packagename":None,"file_identifier":self.file_identifier}
def __get_scanner_file__(self,scanner_dir,file_suffix):
dir_or_files = os.listdir(scanner_dir)
for dir_file in dir_or_files:
dir_file_path = os.path.join(scanner_dir,dir_file)
if os.path.isdir(dir_file_path):
self.__get_scanner_file__(dir_file_path,file_suffix)
else:
if len(dir_file.split("."))>1:
if dir_file.split(".")[-1] in file_suffix:
with open(dir_file_path,'rb') as f:
dex_md5 = str(hashlib.md5().update(f.read()).hexdigest()).upper()
self.file_identifier.append(dex_md5)
self.file_queue.put(dir_file_path)