bandit 0.17.0

Overview

Bandit is a tool designed to find common security issues in Python code. To do this Bandit processes each file, builds an AST from it, and runs appropriate plugins against the AST nodes. Once Bandit has finished scanning all the files it generates a report.

Installation

Bandit is distributed on PyPI. The best way to install it is with pip:

Create a virtual environment (optional):

virtualenv bandit-env

Install Bandit:

pip install bandit
# Or, if you're working with a Python 3 project
pip3.4 install bandit

Run Bandit:

bandit -r path/to/your/code

Bandit can also be installed from source. To do so, download the source tarball from PyPI, then install it:

python setup.py install

Usage

Example usage across a code tree:

bandit -r ~/openstack-repo/keystone

Example usage across the examples/ directory, showing three lines of context and only reporting on the high-severity issues:

bandit examples/*.py -n 3 -lll

Bandit can be run with profiles. To run Bandit against the examples directory using only the plugins listed in the ShellInjection profile:

bandit examples/*.py -p ShellInjection

Usage:

$ bandit -h
usage: bandit [-h] [-r] [-a {file,vuln}] [-n CONTEXT_LINES] [-c CONFIG_FILE]
              [-p PROFILE] [-l] [-i] [-f {csv,html,json,txt,xml}]
              [-o OUTPUT_FILE] [-v] [-d] [--ignore-nosec] [-x EXCLUDED_PATHS]
              [-b BASELINE]
              targets [targets ...]

Bandit - a Python source code analyzer.

positional arguments:
  targets               source file(s) or directory(s) to be tested

optional arguments:
  -h, --help            show this help message and exit
  -r, --recursive       process files in subdirectories
  -a {file,vuln}, --aggregate {file,vuln}
                        group results by vulnerability type or file it occurs
                        in
  -n CONTEXT_LINES, --number CONTEXT_LINES
                        max number of code lines to display for each issue
                        identified
  -c CONFIG_FILE, --configfile CONFIG_FILE
                        if omitted default locations are checked. Check
                        documentation for searched paths
  -p PROFILE, --profile PROFILE
                        test set profile in config to use (defaults to all
                        tests)
  -l, --level           results severity filter. Show only issues of a given
                        severity level or higher. -l for LOW, -ll for MEDIUM,
                        -lll for HIGH
  -i, --confidence      confidence results filter, show only issues of this
                        level or higher. -i for LOW, -ii for MEDIUM, -iii for
                        HIGH
  -f {csv,html,json,txt,xml}, --format {csv,html,json,txt,xml}
                        specify output format
  -o OUTPUT_FILE, --output OUTPUT_FILE
                        write report to filename
  -v, --verbose         show extra information like excluded and included
                        files
  -d, --debug           turn on debug mode
  --ignore-nosec        do not skip lines with # nosec comments
  -x, --exclude EXCLUDED_PATHS
                        Comma separated list of paths to exclude from scan.
                        Note that these are in addition to the excluded paths
                        provided in the config file.
  -b BASELINE, --baseline BASELINE
                        Path to a baseline report, in JSON format. Note:
                        baseline reports must be output in one of the
                        following formats: ['txt', 'html']

The following plugin suites were discovered and loaded:
  any_other_function_with_shell_equals_true
  assert_used
  blacklist_calls
  blacklist_import_func
  blacklist_imports
  exec_used
  execute_with_run_as_root_equals_true
  flask_debug_true
  hardcoded_bind_all_interfaces
  hardcoded_password_default
  hardcoded_password_funcarg
  hardcoded_password_string
  hardcoded_sql_expressions
  hardcoded_tmp_directory
  jinja2_autoescape_false
  linux_commands_wildcard_injection
  paramiko_calls
  password_config_option_not_marked_secret
  request_with_no_cert_validation
  set_bad_file_permissions
  ssl_with_bad_defaults
  ssl_with_bad_version
  ssl_with_no_version
  start_process_with_a_shell
  start_process_with_no_shell
  start_process_with_partial_path
  subprocess_popen_with_shell_equals_true
  subprocess_without_shell_equals_true
  try_except_pass
  use_of_mako_templates
  weak_cryptographic_key

Configuration

The Bandit config file is used to set several things, including:

• profiles - defines group of tests which should or shouldn’t be run

• exclude_dirs - sections of the path, that if matched, will be excluded from scanning

• plugin configs - used to tune plugins, for example: by tuning blacklist_imports, you can set which imports should be flagged

• other - plugins directory, included file types, shell display colors, etc.

Bandit requires a config file which can be specified on the command line via -c/–configfile. If this is not provided Bandit will search for a default config file (bandit.yaml) in the following preference order:

GNU/Linux:

• ./bandit.yaml

• ~/.config/bandit/bandit.yaml

• /etc/bandit/bandit.yaml

• /usr/local/etc/bandit/bandit.yaml

• <path to venv>/etc/bandit/bandit.yaml (if running within virtualenv)

Mac OSX:

• ./bandit.yaml

• /Users/${USER}/Library/Application Support/bandit/bandit.yaml

• /Library/Application Support/bandit/bandit.yaml

• /usr/local/etc/bandit/bandit.yaml

• <path to venv>/bandit/config/bandit.yaml (if running within virtualenv)

Exclusions

In the event that a line of code triggers a Bandit issue, but that the line has been reviewed and the issue is a false positive or acceptable for some other reason, the line can be marked with a # nosec and any results associated with it will not be reported.

For example, although this line may cause Bandit to report a potential security issue, it will not be reported:

self.process = subprocess.Popen('/bin/echo', shell=True)  # nosec

Vulnerability Tests

Vulnerability tests or “plugins” are defined in files in the plugins directory.

Tests are written in Python and are autodiscovered from the plugins directory. Each test can examine one or more type of Python statements. Tests are marked with the types of Python statements they examine (for example: function call, string, import, etc).

Tests are executed by the BanditNodeVisitor object as it visits each node in the AST.

Test results are maintained in the BanditResultStore and aggregated for output at the completion of a test run.

Writing Tests

To write a test:

• Identify a vulnerability to build a test for, and create a new file in examples/ that contains one or more cases of that vulnerability.

• Consider the vulnerability you’re testing for, mark the function with one or more of the appropriate decorators: - @checks(‘Call’) - @checks(‘Import’, ‘ImportFrom’) - @checks(‘Str’)

• Create a new Python source file to contain your test, you can reference existing tests for examples.

• The function that you create should take a parameter “context” which is an instance of the context class you can query for information about the current element being examined. You can also get the raw AST node for more advanced use cases. Please see the context.py file for more.

• Extend your Bandit configuration file as needed to support your new test.

• Execute Bandit against the test file you defined in examples/ and ensure that it detects the vulnerability. Consider variations on how this vulnerability might present itself and extend the example file and the test function accordingly.

Extending Bandit

Bandit allows users to write and register extensions for checks and formatters. Bandit will load plugins from two entry-points:

• bandit.formatters

• bandit.plugins

Formatters need to accept 4 things:

• result_store: An instance of bandit.core.BanditResultStore

• file_list: The list of files which were inspected in the scope

• scores: The scores awarded to each file in the scope

• excluded_files: The list of files that were excluded from the scope

Plugins tend to take advantage of the bandit.checks decorator which allows the author to register a check for a particular type of AST node. For example,

@bandit.checks('Call')
def prohibit_unsafe_deserialization(context):
    if 'unsafe_load' in context.call_function_name_qual:
        return bandit.Issue(
            severity=bandit.HIGH,
            confidence=bandit.HIGH,
            text="Unsafe deserialization detected."
        )

To register your plugin, you have two options:

1. If you’re using setuptools directly, add something like the following to your setup call:

   # If you have an imaginary bson formatter in the bandit_bson module
   # and a function called `formatter`.
   entry_points={'bandit.formatters': ['bson = bandit_bson:formatter']}
   # Or a check for using mako templates in bandit_mako that
   entry_points={'bandit.plugins': ['mako = bandit_mako']}

2. If you’re using pbr, add something like the following to your setup.cfg file:

   [entry_points]
   bandit.formatters =
       bson = bandit_bson:formatter
   bandit.plugins =
       mako = bandit_mako

Contributing

Contributions to Bandit are always welcome! We can be found on #openstack-security on Freenode IRC.

The best way to get started with Bandit is to grab the source:

git clone https://git.openstack.org/openstack/bandit.git

You can test any changes with tox:

pip install tox
tox -e pep8
tox -e py27
tox -e py34
tox -e cover

Reporting Bugs

Bugs should be reported on Launchpad. To file a bug against Bandit, visit: https://bugs.launchpad.net/bandit/+filebug

Under Which Version of Python Should I Install Bandit?

The answer to this question depends on the project(s) you will be running Bandit against. If your project is only compatible with Python 2.7, you should install Bandit to run under Python 2.7. If your project is only compatible with Python 3.4, then use 3.4. If your project supports both, you could run Bandit with both versions but you don’t have to.

Bandit uses the ast module from Python’s standard library in order to analyze your Python code. The ast module is only able to parse Python code that is valid in the version of the interpreter from which it is imported. In other words, if you try to use Python 2.7’s ast module to parse code written for 3.4 that uses, for example, yield from with asyncio, then you’ll have syntax errors that will prevent Bandit from working properly. Alternatively, if you are relying on 2.7’s octal notation of 0777 then you’ll have a syntax error if you run Bandit on 3.4.