Using YARA from Python

YARA can be also used from Python through the yara-python library. Once the library is built and installed as described in Compiling and installing YARA you'll have access to the full potential of YARA from your Python scripts.

The first step is importing the YARA library:

import yara

Then you will need to compile your YARA rules before applying them to your data, the rules can be compiled from a file path:

rules = yara.compile(filepath='/foo/bar/myrules')

The default argument is filepath, so you don't need to explicitly specify its name:

rules = yara.compile('/foo/bar/myrules')

You can also compile your rules from a file object:

fh = open('/foo/bar/myrules')
rules = yara.compile(file=fh)
fh.close()

Or you can compile them directly from a Python string:

rules = yara.compile(source='rule dummy { condition: true }')

If you want to compile a group of files or strings at the same time you can do it by using the filepaths or sources named arguments:

rules = yara.compile(filepaths={

  'namespace1':'/my/path/rules1',
  'namespace2':'/my/path/rules2'
})

rules = yara.compile(sources={

  'namespace1':'rule dummy { condition: true }',
  'namespace2':'rule dummy { condition: false }'
})

Notice that both filepaths and sources must be dictionaries with keys of string type. The dictionary keys are used as a namespace identifier, allowing to differentiate between rules with the same name in different sources, as occurs in the second example with the dummy name.

The compile method also has an optional boolean parameter named includes which allows you to control whether or not the include directive should be accepted in the source files, for example:

rules = yara.compile('/foo/bar/my_rules', includes=False)

If the source file contains include directives the previous line would raise an exception.

If includes are used, a python callback can be set to define a custom source for the imported files (by default they are read from disk). This callback function is set through the include_callback optional parameter. It receives the following parameters:

  • requested_filename: file requested with 'include'
  • filename: file containing the 'include' directive if applicable, else None
  • namespace: namespace

And returns the requested rules sources as a single string.

If you are using external variables in your rules you must define those external variables either while compiling the rules, or while applying the rules to some file. To define your variables at the moment of compilation you should pass the externals parameter to the compile method. For example:

rules = yara.compile('/foo/bar/my_rules’,
  externals= {'var1': 'some string’, 'var2': 4, 'var3': True})

The externals parameter must be a dictionary with the names of the variables as keys and an associated value of either string, integer or boolean type.

The compile method also accepts the optional boolean argument error_on_warning. This arguments tells YARA to raise an exception when a warning is issued during compilation. Such warnings are typically issued when your rules contains some construct that could be slowing down the scanning. The default value for the error_on_warning argument is False.

In all cases compile returns an instance of the class yara.Rules Rules. This class has a save method that can be used to save the compiled rules to a file:

rules.save('/foo/bar/my_compiled_rules')

The compiled rules can be loaded later by using the load method:

rules = yara.load('/foo/bar/my_compiled_rules')

Starting with YARA 3.4 both save and load accept file objects. For example, you can save your rules to a memory buffer with this code:

import StringIO

buff = StringIO.StringIO()
rules.save(file=buff)

The saved rules can be loaded from the memory buffer:

buff.seek(0)
rule = yara.load(file=buff)

The result of load is also an instance of the class yara.Rules.

Starting with YARA 4.3.0, Rules have a warning member which contains a list of warnings generated by the compiler. This allows you to know if the compiler generated warnings without them being hard errors using the error_on_warning argument.

Instances of Rules also have a match method, which allows you to apply the rules to a file:

matches = rules.match('/foo/bar/my_file')

But you can also apply the rules to a Python string:

with open('/foo/bar/my_file', 'rb') as f:
  matches = rules.match(data=f.read())

Or to a running process:

matches = rules.match(pid=1234)

As in the case of compile, the match method can receive definitions for external variables in the externals argument.

matches = rules.match('/foo/bar/my_file',
  externals= {'var1': 'some other string', 'var2': 100})

External variables defined during compile-time don’t need to be defined again in subsequent calls to the match method. However you can redefine any variable as needed, or provide additional definitions that weren’t provided during compilation.

In some situations involving a very large set of rules or huge files the match method can take too much time to run. In those situations you may find useful the timeout argument:

matches = rules.match('/foo/bar/my_huge_file', timeout=60)

If the match function does not finish before the specified number of seconds elapsed, a TimeoutError exception is raised.

You can also specify a callback function when invoking the match method. By default, the provided function will be called for every rule, no matter if matching or not. You can choose when your callback function is called by setting the which_callbacks parameter to one of yara.CALLBACK_MATCHES, yara.CALLBACK_NON_MATCHES or yara.CALLBACK_ALL. The default is to use yara.CALLBACK_ALL. Your callback function should expect a single parameter of dictionary type, and should return CALLBACK_CONTINUE to proceed to the next rule or CALLBACK_ABORT to stop applying rules to your data.

Here is an example:

import yara

def mycallback(data):
  print(data)
  return yara.CALLBACK_CONTINUE

matches = rules.match('/foo/bar/my_file', callback=mycallback, which_callbacks=yara.CALLBACK_MATCHES)

The passed dictionary will be something like this:

{
  'tags': ['foo', 'bar'],
  'matches': True,
  'namespace': 'default',
  'rule': 'my_rule',
  'meta': {},
  'strings': [StringMatch, StringMatch]
}

The matches field indicates if the rule matches the data or not. The strings field is a list of yara.StringMatch objects.

The match method returns a list of instances of the class yara.Match. Instances of this class have the same attributes as the dictionary passed to the callback function.

You can also specify a module callback function when invoking the match method. The provided function will be called for every imported module that scanned a file. Your callback function should expect a single parameter of dictionary type, and should return CALLBACK_CONTINUE to proceed to the next rule or CALLBACK_ABORT to stop applying rules to your data.

Here is an example:

import yara

def modules_callback(data):
  print(data)
  return yara.CALLBACK_CONTINUE

matches = rules.match('/foo/bar/my_file', modules_callback=modules_callback)

The passed dictionary will contain the information from the module.

You can also specify a warning callback function when invoking the match method. The provided function will be called for every runtime warning. Your callback function should expect two parameters. The first is an integer which contains the type of warning and the second is a string with the warning message. Your callback should return CALLBACK_CONTINUE to proceed with the scan or CALLBACK_ABORT to stop.

Possible values for the type are:

CALLBACK_TOO_MANY_MATCHES

Contents of the callback message depend on the type of the callback.

For CALLBACK_TOO_MANY_MATCHES, the message is a named tuple containing 3 items: namespace, rule and string. All contain string identifiers.

Here is an example:

import yara

def warnings_callback(warning_type, message):
  if warning_type == yara.CALLBACK_TOO_MANY_MATCHES:
      print(f"namespace:'{message.namespace}' rule:'{message.rule}' string:'{message.string}'")
  return yara.CALLBACK_CONTINUE

matches = rules.match('/foo/bar/my_file', warnings_callback=warnings_callback)

If you do not use a warning callback a warning message will be sent to the normal python warning system for you and scanning will continue.

With YARA 4.2.0 a new console module was introduced which allows you to send log messages within YARA. These are, by default, printed to stdout in yara-python, but you can handle them in your own callback using the console_callback parameter.

Here is an example:

import yara

r = """
import "console"

rule a { condition: console.log("Hello from Python!") }
"""

def console(message):
    print(f"Callback: {message}")

rules = yara.compile(source=r)
rules.match("/bin/ls", console_callback=console)
rules.match("/bin/ls")

The type of the message parameter is a string.

You may also find that the default sizes for the stack for the matching engine in yara or the default size for the maximum number of strings per rule is too low. In the C libyara API, you can modify these using the YR_CONFIG_STACK_SIZE and YR_CONFIG_MAX_STRINGS_PER_RULE variables via the yr_set_configuration_uint32 function in libyara. The command-line tool exposes these as the --stack-size (-k) and --max-strings-per-rule command-line arguments. In order to set these values via the Python API, you can use yara.set_config with either or both stack_size and max_strings_per_rule provided as kwargs. At the time of this writing, the default stack size was 16384 and the default maximum strings per rule was 10000.

Also, yara.set_config accepts the max_match_data argument for controlling the maximum number of bytes that will be returned for each matching string. This is equivalent to using YR_CONFIG_MAX_MATCH_DATA with the yr_set_configuration_uint32 in the C API. By the default this is set to 512.

Here are a few example calls:

yara.set_config(stack_size=65536)
yara.set_config(max_strings_per_rule=50000, stack_size=65536)
yara.set_config(max_strings_per_rule=20000)
yara.set_config(max_match_data=128)

Reference

yara.compile(...)

Compile YARA sources.

Either filepath, source, file, filepaths or sources must be provided. The remaining arguments are optional.

Parameters:
  • filepath (str) -- Path to the source file.
  • source (str) -- String containing the rules code.
  • file (file-object) -- Source file as a file object.
  • filepaths (dict) -- Dictionary where keys are namespaces and values are paths to source files.
  • sources (dict) -- Dictionary where keys are namespaces and values are strings containing rules code.
  • externals (dict) -- Dictionary with external variables. Keys are variable names and values are variable values.
  • includes (boolean) -- True if include directives are allowed or False otherwise. Default value: True.
  • error_on_warning (boolean) -- If true warnings are treated as errors, raising an exception.
Returns:

Compiled rules object.

Return type:

yara.Rules

Raises:
  • yara.SyntaxError -- If a syntax error was found.
  • yara.Error -- If an error occurred.
yara.load(...)

Changed in version 3.4.0.

Load compiled rules from a path or file object. Either filepath or file must be provided.

Parameters:
  • filepath (str) -- Path to a compiled rules file
  • file (file-object) -- A file object supporting the read method.
Returns:

Compiled rules object.

Return type:

yara.Rules

Raises:

yara.Error: If an error occurred while loading the file.

yara.set_config(...)

Set the configuration variables accessible through the yr_set_configuration API.

Provide either stack_size, max_strings_per_rule, or max_match_data. These kwargs take unsigned integer values as input and will assign the provided value to the yr_set_configuration(...) variables YR_CONFIG_STACK_SIZE, YR_CONFIG_MAX_STRINGS_PER_RULE, and YR_CONFIG_MAX_MATCH_DATA respectively.

Parameters:
  • stack_size (int) -- Stack size to use for YR_CONFIG_STACK_SIZE
  • max_strings_per_rule (int) -- Maximum number of strings to allow per yara rule. Will be mapped to YR_CONFIG_MAX_STRINGS_PER_RULE.
  • max_match_data (int) -- Maximum number of bytes to allow per yara match. Will be mapped to YR_CONFIG_MAX_MATCH_DATA.
Returns:

None

Return type:

NoneType

Raises:

yara.Error: If an error occurred.

class yara.Rules

Instances of this class are returned by yara.compile() and represents a set of compiled rules.

match(filepath, pid, data, externals=None, callback=None, fast=False, timeout=None, modules_data=None, modules_callback=None, warnings_callback=None, which_callbacks=CALLBACK_ALL, console_callback=None)

Scan a file, process memory or data string.

Either filepath, pid or data must be provided. The remaining arguments are optional.

Parameters:
  • filepath (str) -- Path to the file to be scanned.
  • pid (int) -- Process id to be scanned.
  • data (str/bytes) -- Data to be scanned.
  • externals (dict) -- Dictionary with external variables. Keys are variable names and values are variable values.
  • callback (function) -- Callback function invoked for each rule.
  • fast (bool) -- If true performs a fast mode scan.
  • timeout (int) -- Aborts the scanning when the number of specified seconds have elapsed.
  • modules_data (dict) -- Dictionary with additional data to modules. Keys are module names and values are bytes objects containing the additional data.
  • modules_callback (function) -- Callback function invoked for each module.
  • warnings_callback (function) -- Callback function invoked for warning, like yara.CALLBACK_TOO_MANY_MATCHES.
  • which_callbacks (int) -- An integer that indicates in which cases the callback function must be called. Possible values are yara.CALLBACK_ALL, yara.CALLBACK_MATCHES and yara.CALLBACK_NON_MATCHES.
  • console_callback (function) -- Callback function invoked for each console module call.
Raises:
  • yara.TimeoutError -- If the timeout was reached.
  • yara.Error -- If an error occurred during the scan.
save(...)

Changed in version 3.4.0: Save compiled rules to a file. Either filepath or file must be provided.

param str filepath:
 Path to the file.
param file-object file:
 A file object supporting the write method.
raises:yara.Error: If an error occurred while saving the file.
class yara.Match

New in version 4.3.0.

Objects returned by yara.Rules.match(), representing a match.

rule

Name of the matching rule.

namespace

Namespace associated to the matching rule.

tags

Array of strings containing the tags associated to the matching rule.

meta

Dictionary containing metadata associated to the matching rule.

strings

List of StringMatch objects.

class yara.StringMatch

New in version 3.4.0.

Objects which represent string matches.

identifier

Name of the matching string.

instances

List of StringMatchInstance objects.

is_xor()

Returns a boolean if the string is using the xor modifier.

class yara.StringMatchInstance

New in version 4.3.0.

Objects which represent instances of matched strings.

matched_data

Bytes of the matched data.

matched_length

Length of the matched data.

offset

Offset of the matched data.

xor_key

XOR key found for the string.

plaintext()

Returns the plaintext version of the string after xor key is applied. If the string is not an xor string then no modification is done.