PyJFuzz is a small, extensible and ready-to-use framework used to fuzz JSON inputs, such as mobile endpoint REST API, JSON implementation, Browsers, cli executable and much more.

Installation

Dependencies

In order to work PyJFuzz need some dependency, bottle,netifaces,GitPython and gramfuzz, you can install them from automatic setup.py installation.

Installation

You can install PyJFuzz with the following command

git clone https://github.com/mseclab/PyJFuzz.git && cd PyJFuzz && sudo python setup.py install

Documentation and Examples

CLI tool

Once installed PyJFuzz will create both a python library and a command-line utility called pjf (screenshot below)

Library

PyJFuzz could also work as a library, you can import in your project like following

from pyjfuzz.lib import *

Classes

The available object/class are the following:

• PJFServer - User to start and stop built-in HTTP and HTTPS servers
• PJFProcessMonitor - Used to monitor process crash, it will automatically restart proccess each time it crash
• PJFTestcaseServer - The testcase server is used in conjunction with PJFProcessMonitor, whenever a process crash the testcase server will register and store the JSON which cause the crash
• PJFFactory - It's the main object used to do the real fuzz of JSON objects
• PJFConfiguration - It's the configuration file for each of the available objects
• PJFExternalFuzzer - Used by PJFactory is a auxiliary class which provide an interface to other command line fuzzer such as radamsa
• PJFMutation - Used by PJFFactory provide all the mutation used during fuzzing session
• PJFExecutor - Provides an interface to interact with external process

Examples

Below some trivial example of how-to implement PyJFuzz powered program

simple_fuzzer.py

from argparse import Namespace
from pyjfuzz.lib import *

config = PJFConfiguration(Namespace(json={"test": ["1", 2, True]}, nologo=True, level=6))
fuzzer = PJFFactory(config)
while True:
    print fuzzer.fuzzed

custom_techniques.py

from argparse import Namespace
from pyjfuzz.lib import *

# Techniques may be defined by group , or by technique number
# groups are CHTPRSX , to understand what they are , please run pyjfuzz with -h switch or look at the command line screenshot
# This below will initalizate a config object which use only the P group attacks where P stay for Path Traversal
config = PJFConfiguration(Namespace(json={"test": ["1", 2, True]}, nologo=True, level=6, techniques="P"))
# once a config object is defined you can access to config.techniques to view the selected techniques for your group
print("Techniques IDs: {0}".format(str(config.techniques)))
# you can eventually modify them!
config.techniques = [2]
# This way only attack number 2 (LFI Attack) will be performed!
fuzzer = PJFFactory(config)
while True:
    print fuzzer.fuzzed

simple_server.py

from argparse import Namespace
from pyjfuzz.lib import *

config = PJFConfiguration(Namespace(json={"test": ["1", 2, True]}, nologo=True, level=6, debug=True, indent=True))
PJFServer(config).run()

Sometimes you may need to modify standard non customizable settings such as HTTPS or HTTP server port, this can be done in the following way

from argparse import Namespace
from pyjfuzz.lib import *

config = PJFConfiguration(Namespace(json={"test": ["1", 2, True]}, nologo=True, level=6, indent=True))
print config.ports["servers"]["HTTP_PORT"]   # 8080
print config.ports["servers"]["HTTPS_PORT"]  # 8443
print config.ports["servers"]["TCASE_PORT"]  # 8888
config.ports["servers"]["HTTPS_PORT"] = 443  # Change HTTPS port to 443

Remember: When changing default ports, you should always handle exception due to needed privileges!

Below a comprehensive list of all available settings / customization of PJFConfiguration object:

Configuration table

Techniques table

Screenshots

Below some screenshot just to let you know what you should expect from PyJFuzz

Built-in tool

PyJFuzz is shipped with a built-in tool called PyJFuzz Web Fuzzer, this tool will provide an automatic fuzzing console via HTTP and HTTPS server, it can be used to easly fuzz almost any web browser even when you can't control the process state!

There are two switch used to launch this tool (--browser-auto and --fuzz-web), the first one perform automatic browser restart when a crash occur, the other one try to catch when a browser doesn't make requests anymore. Both of them always save the testcases, below some screenshots.

Issue

Please send any issue here via GitHub I'll provide a fix as soon as possible.

Result

Below a list of know issue found by PyJFuzz, the list will be updated weekly

• Double free in cJSON (https://github.com/DaveGamble/cJSON/issues/105)
• Unhandled exception in picojson (https://github.com/kazuho/picojson/issues/94)
• Memory leak in simpleJSON (https://github.com/nbsdx/SimpleJSON/issues/8)
• Stack base buffer overflow in frozen (https://github.com/cesanta/frozen/issues/14)
• Memory corruption with custom EIP (https://github.com/cesanta/frozen/issues/15)

End

Thanks for using PyJFuzz!

Happy Fuzzing from mseclab