apkminer

Simple program to mine through APKs at high speed. It uses a modular method of calling specific analyzers on each apk provided

Setup

git submodule init
git submodule update

Standard CPython works fine but I highly recommend pypy, I have seen 70% faster runs using pypy.

Usage

usage: apkminer.py [-h] [-i IN_DIR] [-o LOG_FILE] [-c CORES] [-a ANALYZER]
                   [-l]

analyzer of APKs

optional arguments:
  -h, --help            show this help message and exit
  -i IN_DIR, --in_dir IN_DIR
                        directory of apk files to analyze
  -o LOG_FILE, --log_file LOG_FILE
                        log file to write to
  -c CORES, --cores CORES
                        force a number of cores to use
  -a ANALYZER, --analyzer ANALYZER
                        Select the analyzer you want to use.
  -l, --list_analyzers  List the possible analyzers

Analyzers

private_keys  -  Find private keys in files or dex strings
elf_files     -  Report string data from specific sections of elf files
aws_finder    -  Find AWS key pairs in files and dex strings
so_census     -  Report on data about .so's in APKs
silverpush    -  Finds apks that contain the silverpush library

Dependencies

• pyelftools

Writing an analyzer

Below I will layout the steps for writing an analyzer and the components of apkminer that a analyzer developer should understand.

Analyzer template

# import the utils.py file for helper functions and Logger object
from utils import *

# Define the analyzer() function, this function name needs to be the same for each analyzer
# because apkminer searches for this function name.
def analyze(args, apk_queue, res_queue, output_data):
	# The Logger class uses a multiprocessing Queue to perform atomic writes to the defined log file
	# this is helpful for debugging data and logging and errors that might occur during the run.
	log = Logger(args.log_file, res_queue)

	# Continually check the input 'apk_queue' for new file names
	while True:
		# break the loop if the queue is empty
		if apk_queue.empty():
			return
		else:
			# fetch the file off the queue
			apk_file = apk_queue.get()

			# Logging works similar to stdout / stderr,
			# the log() function writes to an internal buffer (new line delimited)
			# then flush() pushes the data the actually logging process
			log.log(apk_file)
			log.flush()

			# write analyzer here.

In order to register a analyzer inside of apkminer, save the analyzer as a .py in the analyzers/ directory and then edit analyzers/init.py to include the name of your analyzer.

For example:

analyzers/test_analyzer.py

Then add "test_analyzer" to the line import list in init.py

Check out the aws_finder.py or other analyzers for examples. Also spend some time looking at the helper functions inside of utils.py.

Optional features for analyzers

In order to enable structured output that is separate from the log file a analyzer writer can define two other methods in their .py file:

1. output_results - Used for bulk writes after completion of all input apk's.
2. stream_results - Used for streaming results as they are generated.

output_results example

import pickle

def output_results(output_data):
	fd = open("output.pick", "wb")
	pickle.dump(output_data, fd)
	fd.close()

stream_results example

import csv
import Queue

def stream_results(output_queue, end_event):
	csv_fd = open('test.csv', 'wb')
	datawriter = csv.writer(csv_fd)

	while not end_event.is_set():
		try:
			data = output_queue.get(True, 1)
			datawriter.writerow(data)

		except Queue.Empty:
			continue