Attack Surface Meter
Attack Surface Meter is a Python package for collecting attack surface metrics from a software system. In its current version, Attack Surface Meter is capable of analyzing software systems written in the C programming language with skeletal support for analyzing software systems written in the Java programming language.
The attack surface metrics collected are:
- Proximity to Entry/Exit/Dangerous - The mean of shortest unweighted path length from a function to Entry Points/Exit Points/Dangerous Functions.
- Risky Walk - The probability that a function will be invoked on a random execution path starting at the attack surface.
Installation
PyPI
pip install attacksurfacemeter
Source
python setup.py install
Usage
API
The Attack Surface Meter works off of the call graph representation of a software system. A call graph is parsed by the correponding loader to generate an internal representation. In this version, the Attack Surface Meter is capable of parsing the call graph generated by one of the following utilities:
Extending the Attack Surface Meter to analyze a software system written in a programming language other than C or Java would require defining a new loader to parse a call graph generated by a particular language-specific utility.
Example
The code snippet that follows depicts using the Attack Surface Meter API to analyze the a C program for which a call graph generated by GNU cflow is available.
import os
from attacksurfacemeter.call_graph import CallGraph
from attacksurfacemeter.loaders.cflow_loader import CflowLoader
loader = CflowLoader(os.path.expanduser('~/cflow.callgraph.txt'))
call_graph = CallGraph.from_loader(loader)
The call_graph object is an instance of the attacksurfacemeter.call_graph.CallGraph class and supports several methods to collect the proximity and risky metrics for a given function. For more information on these methods, please refer to the call_graph.py file which has all the methods extensively documented using Python documentation comments.
Command Line
usage: attack_surface_meter.py [-h] [-c CFLOW] [--reverse] [-g GPROF]
[-p PROCESSES] [-j JAVACG] [-a [P [P ...]]]
[--output OUTPUT] [--verbose] [--showerrors]
Collect attack surface metrics from the call graph representation of a
software system.
optional arguments:
-h, --help show this help message and exit
-c CFLOW Absolute path of the file containing the textual
representation of the call graph generated by GNU cflow or
of the directory containing the source code of the software
system to be analyzed.
--reverse cflow call graph was generated with the -r option.
-g GPROF Absolute path of the file containing the textual
representation of the call graph generated by GNU gprof or
of a directory containing multiple such text files.
-p PROCESSES Number of processes to spawn when loaded multiple gprof
call graph files. Default is 2.
-j JAVACG Absolute path of the file containing the textual
representation of the call graph generated by java-
callgraph.
-a [P [P ...]] When using java-callgraph for call graph generation of
android apps, specify the fully qualified package name of
the method calls that will be included in the call graph.
This is generally the name of the java package inside which
the app's classes are defined.
--output OUTPUT Absolute path of the file to which the output should be
written to. The format of output is inferred from the file
extension. txt, html, and xml are currently supported. In
cases when the output format cannot be inferred, txt is
used. When an output path is not specified, standard output
is used.
--verbose Output itemized report including metric values collected
for each function.
--showerrors Display errors encountered when parsing call graph (if
any).
