A schema and validator for YAML.





GitHub Stars



Last Commit

2mos ago










Yamale (ya·ma·lē)


A schema and validator for YAML.

What's YAML? See the current spec here and an introduction to the syntax here.

Build Status PyPI


  • Python 3.6+
  • PyYAML
  • ruamel.yaml (optional)



$ pip install yamale

NOTE: Some platforms, e.g., Mac OS, may ship with only Python 2 and may not have pip installed. Installation of Python 3 should also install pip. To preserve any system dependencies on default software, consider installing Python 3 as a local package. Please note replacing system-provided Python may disrupt other software. Mac OS users may wish to investigate MacPorts, homebrew, or building Python 3 from source; in all three cases, Apple's Command Line Tools (CLT) for Xcode may be required. See also developers, below.


  1. Download Yamale from: https://github.com/23andMe/Yamale/archive/master.zip
  2. Unzip somewhere temporary
  3. Run python setup.py install (may have to prepend sudo)


Command line

Yamale can be run from the command line to validate one or many YAML files. Yamale will search the directory you supply (current directory is default) for YAML files. Each YAML file it finds it will look in the same directory as that file for its schema, if there is no schema Yamale will keep looking up the directory tree until it finds one. If Yamale can not find a schema it will tell you.


usage: yamale [-h] [-s SCHEMA] [-n CPU_NUM] [-p PARSER] [--no-strict] [PATH]

Validate yaml files.

positional arguments:
  PATH                  folder to validate. Default is current directory.

optional arguments:
  -h, --help            show this help message and exit
  -s SCHEMA, --schema SCHEMA
                        filename of schema. Default is schema.yaml.
  -n CPU_NUM, --cpu-num CPU_NUM
                        number of CPUs to use. Default is 4.
  -p PARSER, --parser PARSER
                        YAML library to load files. Choices are "ruamel" or
                        "pyyaml" (default).
  --no-strict           Disable strict mode, unexpected elements in the data
                        will be accepted.


There are several ways to feed Yamale schema and data files. The simplest way is to let Yamale take care of reading and parsing your YAML files.

All you need to do is supply the files' path:

# Import Yamale and make a schema object:
import yamale
schema = yamale.make_schema('./schema.yaml')

# Create a Data object
data = yamale.make_data('./data.yaml')

# Validate data against the schema. Throws a ValueError if data is invalid.
yamale.validate(schema, data)

You can pass a string of YAML to make_schema() and make_data() instead of passing a file path by using the content= parameter:

data = yamale.make_data(content="""
name: Bill
age: 26
height: 6.2
awesome: True

If data is valid, nothing will happen. However, if data is invalid Yamale will throw a YamaleError with a message containing all the invalid nodes:

    yamale.validate(schema, data)
    print('Validation success! 👍')
except ValueError as e:
    print('Validation failed!\n%s' % str(e))

and an array of ValidationResult.

    yamale.validate(schema, data)
    print('Validation success! 👍')
except YamaleError as e:
    print('Validation failed!\n')
    for result in e.value.results:
        print("Error validating data '%s' with '%s'\n\t" % (result.data, result.schema))
        for error in result.errors:
            print('\t%s' % error)

You can also specify an optional parser if you'd like to use the ruamel.yaml (YAML 1.2 support) instead:

# Import Yamale and make a schema object, make sure ruamel.yaml is installed already.
import yamale
schema = yamale.make_schema('./schema.yaml', parser='ruamel')

# Create a Data object
data = yamale.make_data('./data.yaml', parser='ruamel')

# Validate data against the schema same as before.
yamale.validate(schema, data)


To use Yamale you must make a schema. A schema is a valid YAML file with one or more documents inside. Each node terminates in a string which contains valid Yamale syntax. For example, str() represents a String validator.

A basic schema:

name: str()
age: int(max=200)
height: num()
awesome: bool()

And some YAML that validates:

name: Bill
age: 26
height: 6.2
awesome: True

Take a look at the Examples section for more complex schema ideas.


Schema files may contain more than one YAML document (nodes separated by ---). The first document found will be the base schema. Any additional documents will be treated as Includes. Includes allow you to define a valid structure once and use it several times. They also allow you to do recursion.

A schema with an Include validator:

person1: include('person')
person2: include('person')
    name: str()
    age: int()

Some valid YAML:

    name: Bill
    age: 70

    name: Jill
    age: 20

Every root node not in the first YAML document will be treated like an include:

person: include('friend')
group: include('family')
    name: str()
    name: str()

Is equivalent to:

person: include('friend')
group: include('family')
    name: str()
    name: str()

You can get recursion using the Include validator.

This schema:

person: include('human')
    name: str()
    age: int()
    friend: include('human', required=False)

Will validate this data:

    name: Bill
    age: 50
        name: Jill
        age: 20
            name: Will
            age: 10
Adding external includes

After you construct a schema you can add extra, external include definitions by calling schema.add_include(dict). This method takes a dictionary and adds each key as another include.

Strict mode

By default Yamale will provide errors for extra elements present in lists and maps that are not covered by the schema. With strict mode disabled (using the --no-strict command line option), additional elements will not cause any errors. In the API, strict mode can be toggled by passing the strict=True/False flag to the validate function.

It is possible to mix strict and non-strict mode by setting the strict=True/False flag in the include validator, setting the option only for the included validators.


Here are all the validators Yamale knows about. Every validator takes a required keyword telling Yamale whether or not that node must exist. By default every node is required. Example: str(required=False)

You can also require that an optional value is not None by using the none keyword. By default Yamale will accept None as a valid value for a key that's not required. Reject None values with none=False in any validator. Example: str(required=False, none=False).

Some validators take keywords and some take arguments, some take both. For instance the enum() validator takes one or more constants as arguments and the required keyword: enum('a string', 1, False, required=False)

String - str(min=int, max=int, exclude=string)

Validates strings.

  • keywords
    • min: len(string) >= min
    • max: len(string) <= max
    • exclude: Rejects strings that contains any character in the excluded value.


  • str(max=10, exclude='?!'): Allows only strings less than 11 characters that don't contain ? or !.

Regex - regex([patterns], name=string, ignore_case=False, multiline=False, dotall=False)

Validates strings against one or more regular expressions.

  • arguments: one or more Python regular expression patterns
  • keywords:
    • name: A friendly description for the patterns.
    • ignore_case: Validates strings in a case-insensitive manner.
    • multiline: ^ and $ in a pattern match at the beginning and end of each line in a string in addition to matching at the beginning and end of the entire string. (A pattern matches at the beginning of a string even in multiline mode; see below for a workaround.)
    • dotall: . in a pattern matches newline characters in a validated string in addition to matching every character that isn't a newline.


  • regex('^[^?!]{,10}$'): Allows only strings less than 11 characters that don't contain ? or !.
  • regex(r'^(\d+)(\s\1)+$', name='repeated natural'): Allows only strings that contain two or more identical digit sequences, each separated by a whitespace character. Non-matching strings like sugar are rejected with a message like 'sugar' is not a repeated natural.
  • regex('.*^apples$', multiline=True, dotall=True): Allows the string apples as well as multiline strings that contain the line apples.

Integer - int(min=int, max=int)

Validates integers.

  • keywords
    • min: int >= min
    • max: int <= max

Number - num(min=float, max=float)

Validates integers and floats.

  • keywords
    • min: num >= min
    • max: num <= max

Boolean - bool()

Validates booleans.

Null - null()

Validates null values.

Enum - enum([primitives])

Validates from a list of constants.

  • arguments: constants to test equality with


  • enum('a string', 1, False): a value can be either 'a string', 1 or False

Day - day(min=date, max=date)

Validates a date in the form of YYYY-MM-DD.

  • keywords
    • min: date >= min
    • max: date <= max


  • day(min='2001-01-01', max='2100-01-01'): Only allows dates between 2001-01-01 and 2100-01-01.

Timestamp - timestamp(min=time, max=time)

Validates a timestamp in the form of YYYY-MM-DD HH:MM:SS.

  • keywords
    • min: time >= min
    • max: time <= max


  • timestamp(min='2001-01-01 01:00:00', max='2100-01-01 23:00:00'): Only allows times between 2001-01-01 01:00:00 and 2100-01-01 23:00:00.

List - list([validators], min=int, max=int)

Validates lists. If one or more validators are passed to list() only nodes that pass at least one of those validators will be accepted.

  • arguments: one or more validators to test values with
  • keywords
    • min: len(list) >= min
    • max: len(list) <= max


  • list(): Validates any list
  • list(include('custom'), int(), min=4): Only validates lists that contain the custom include or integers and contains a minimum of 4 items.

Map - map([validators], key=validator, min=int, max=int)

Validates maps. Use when you want a node to contain freeform data. Similar to List, Map takes one or more validators to run against the values of its nodes, and only nodes that pass at least one of those validators will be accepted. By default, only the values of nodes are validated and the keys aren't checked.

  • arguments: one or more validators to test values with
  • keywords
    • key: A validator for the keys of the map.
    • min: len(map) >= min
    • max: len(map) <= max


  • map(): Validates any map
  • map(str(), int()): Only validates maps whose values are strings or integers.
  • map(str(), key=int()): Only validates maps whose keys are integers and values are strings. 1: one would be valid but '1': one would not.
  • map(str(), min=1): Only validates a non-empty map.

IP Address - ip()

Validates IPv4 and IPv6 addresses.

  • keywords
    • version: 4 or 6; explicitly force IPv4 or IPv6 validation


  • ip(): Allows any valid IPv4 or IPv6 address
  • ip(version=4): Allows any valid IPv4 address
  • ip(version=6): Allows any valid IPv6 address

MAC Address - mac()

Validates MAC addresses.


  • mac(): Allows any valid MAC address

Any - any([validators])

Validates against a union of types. Use when a node must contain one and only one of several types. It is valid if at least one of the listed validators is valid. If no validators are given, accept any value.

  • arguments: validators to test values with (if none is given, allow any value; if one or more are given, one must be present)


  • any(int(), null()): Validates either an integer or a null value.
  • any(num(), include('vector')): Validates either a number or an included 'vector' type.
  • any(str(min=3, max=3),str(min=5, max=5),str(min=7, max=7)): validates to a string that is exactly 3, 5, or 7 characters long
  • any(): Allows any value.

Subset - subset([validators], allow_empty=False)

Validates against a subset of types. Unlike the Any validator, this validators allows one or more of several types. As such, it automatically validates against a list. It is valid if all values can be validated against at least one validator.

  • arguments: validators to test with (at least one; if none is given, a ValueError exception will be raised)
  • keywords:
    - `allow_empty`: Allow the subset to be empty (and is, therefore, also optional). This overrides the `required`


  • subset(int(), str()): Validators against an integer, a string, or a list of either.
  • subset(int(), str(), allow_empty=True): Same as above, but allows the empty set and makes the subset optional.

Include - include(include_name)

Validates included structures. Must supply the name of a valid include.

  • arguments: single name of a defined include, surrounded by quotes.


  • include('person')

Custom validators

It is also possible to add your own custom validators. This is an advanced topic, but here is an example of adding a Date validator and using it in a schema as date()

import yamale
import datetime
from yamale.validators import DefaultValidators, Validator

class Date(Validator):
    """ Custom Date validator """
    tag = 'date'

    def _is_valid(self, value):
        return isinstance(value, datetime.date)

validators = DefaultValidators.copy()  # This is a dictionary
validators[Date.tag] = Date
schema = yamale.make_schema('./schema.yaml', validators=validators)
# Then use `schema` as normal


Using keywords


optional: str(required=False)
optional_min: int(min=1, required=False)
min: num(min=1.5)
max: int(max=100)

Valid Data:

optional_min: 10
min: 1.6
max: 100

Includes and recursion


customerA: include('customer')
customerB: include('customer')
recursion: include('recurse')
    name: str()
    age: int()
    custom: include('custom_type')

    integer: int()

    level: int()
    again: include('recurse', required=False)

Valid Data:

    name: bob
    age: 900
        integer: 1
    name: jill
    age: 1
        integer: 3
    level: 1
        level: 2
            level: 3
                level: 4



list_with_two_types: list(str(), include('variant'))
questions: list(include('question'))
  rsid: str()
  name: str()

  choices: list(include('choices'))
  questions: list(include('question'), required=False)

  id: str()

Valid Data:

  - 'some'
  - rsid: 'rs123'
    name: 'some SNP'
  - 'thing'
  - rsid: 'rs312'
    name: 'another SNP'
  - choices:
      - id: 'id_str'
      - id: 'id_str1'
      - choices:
        - id: 'id_str'
        - id: 'id_str1'

The data is a list of items without a keyword at the top level


list(include('human'), min=2, max=2)

  name: str()
  age: int(max=200)
  height: num()
  awesome: bool()

Valid Data:

- name: Bill
  age: 26
  height: 6.2
  awesome: True

- name: Adrian
  age: 23
  height: 6.3
  awesome: True

Writing Tests

To validate YAML files when you run your program's tests use Yamale's YamaleTestCase


class TestYaml(YamaleTestCase):
    base_dir = os.path.dirname(os.path.realpath(__file__))
    schema = 'schema.yaml'
    yaml = 'data.yaml'
    # or yaml = ['data-*.yaml', 'some_data.yaml']

    def runTest(self):

base_dir: String path to prepend to all other paths. This is optional.

schema: String of path to the schema file to use. One schema file per test case.

yaml: String or list of yaml files to validate. Accepts globs.



Yamale uses Tox to run its tests against multiple Python versions. To run tests, first checkout Yamale, install Tox, then run make test in Yamale's root directory. You may also have to install the correct Python versions to test with as well.

NOTE on Python versions: tox.ini specifies the lowest and highest versions of Python supported by Yamale. Unless your development environment is configured to support testing against multiple Python versions, one or more of the test branches may fail. One method of enabling testing against multiple versions of Python is to install pyenv and tox-pyenv and to use pyenv install and pyenv local to ensure that tox is able to locate appropriate Pythons.


Yamale uses Github Actions to upload new tags to PyPi. To release a new version:

  1. Make a commit with the new version in setup.py.
  2. Run tests for good luck.
  3. Run make release.

Github Actions will take care of the rest.

Rate & Review

Great Documentation0
Easy to Use0
Highly Customizable0
Bleeding Edge0
Responsive Maintainers0
Poor Documentation0
Hard to Use0
Unwelcoming Community0