SAML-Toolkits / python3-saml

MIT License
704 stars 309 forks source link

SAML Python3 Toolkit

Python package PyPI Downloads Coverage Status PyPi Version Python versions

Add SAML support to your Python software using this library. Forget those complicated libraries and use the open source library provided by the SAML tool community.

This version supports Python3. Python 2 support was deprecated on Jan 1st, 2020: python-saml

Warning

Version 1.16.X is the latest version supporting Python2, consider its use deprecated. 1.17 won't be Python2 and old Python3 compatible.

Version 1.13.0 sets sha256 and rsa-sha256 as default algorithms

Version 1.8.0 sets strict mode active by default

Update python3-saml to 1.5.0, this version includes security improvements for preventing XEE and Xpath Injections.

Update python3-saml to 1.4.0, this version includes a fix for the CVE-2017-11427 vulnerability.

This version also changes how the calculate fingerprint method works, and will expect as input a formatted X.509 certificate.

Update python3-saml to 1.2.6 that adds the use defusedxml that will prevent XEE and other attacks based on the abuse of XML. (CVE-2017-9672)

Update python3-saml to >= 1.2.1, 1.2.0 had a bug on signature validation process (when using wantAssertionsSigned and wantMessagesSigned). CVE-2016-1000251

1.2.0 version includes a security patch that contains extra validations that will prevent signature wrapping attacks.

python3-saml < v1.2.0 is vulnerable and allows signature wrapping!

Security Guidelines

If you believe you have discovered a security vulnerability in this toolkit, please report it by mail to the maintainer: sixto.martin.garcia+security@gmail.com

Sponsors

Thanks to the following sponsors for their support:

84codes

Why add SAML support to my software?

SAML is an XML-based standard for web browser single sign-on and is defined by the OASIS Security Services Technical Committee. The standard has been around since 2002, but lately it is becoming popular due its advantages:

General Description

SAML Python toolkit lets you turn your Python application into a SP (Service Provider) that can be connected to an IdP (Identity Provider).

Supports:

Key Features:

Installation

Dependencies

Review the pyproject.toml file to know the version of the library that python3-saml is using

Code

Option 1. Download from GitHub

The toolkit is hosted on GitHub. You can download it from:

Find the core of the library at src/onelogin/saml2 folder.

Option 2. Download from pypi

The toolkit is hosted in pypi, you can find the python3-saml package at https://pypi.python.org/pypi/python3-saml

You can install it executing:

$ pip install python3-saml

If you want to know how a project can handle python packages review this guide and review this sampleproject

NOTE

To avoid libxml2 library version incompatibilities between xmlsec and lxml it is recommended that lxml is not installed from binary.

This can be ensured by executing:

$ pip install --force-reinstall --no-binary lxml lxml

Security Warning

In production, the strict parameter MUST be set as "true". Otherwise your environment is not secure and will be exposed to attacks.

In production also we highly recommend to register on the settings the IdP certificate instead of using the fingerprint method. The fingerprint, is a hash, so at the end is open to a collision attack that can end on a signature validation bypass. Other SAML toolkits deprecated that mechanism, we maintain it for compatibility and also to be used on test environment.

Avoiding Open Redirect attacks

Some implementations uses the RelayState parameter as a way to control the flow when SSO and SLO succeeded. So basically the user is redirected to the value of the RelayState.

If you are using Signature Validation on the HTTP-Redirect binding, you will have the RelayState value integrity covered, otherwise, and on HTTP-POST binding, you can't trust the RelayState so before executing the validation, you need to verify that its value belong a trusted and expected URL.

Read more about Open Redirect CWE-601.

Avoiding Replay attacks

A replay attack is basically try to reuse an intercepted valid SAML Message in order to impersonate a SAML action (SSO or SLO).

SAML Messages have a limited timelife (NotBefore, NotOnOrAfter) that make harder this kind of attacks, but they are still possible.

In order to avoid them, the SP can keep a list of SAML Messages or Assertion IDs already validated and processed. Those values only need to be stored the amount of time of the SAML Message life time, so we don't need to store all processed message/assertion Ids, but the most recent ones.

The OneLogin_Saml2_Auth class contains the get_last_request_id, get_last_message_id and get_last_assertion_id methods to retrieve the IDs

Checking that the ID of the current Message/Assertion does not exists in the list of the ones already processed will prevent replay attacks.

Getting Started

Knowing the toolkit

The new SAML Toolkit contains different folders (certs, lib, demo-django, demo-flask and tests) and some files.

Let's start describing them:

src

This folder contains the heart of the toolkit, onelogin/saml2 folder contains the new version of the classes and methods that are described in a later section.

demo-django

This folder contains a Django project that will be used as demo to show how to add SAML support to the Django Framework. demo is the main folder of the Django project (with its settings.py, views.py, urls.py), templates is the Django templates of the project and saml is a folder that contains the certs folder that could be used to store the X.509 public and private key, and the SAML toolkit settings (settings.json and advanced_settings.json).

Notice about certs

SAML requires a X.509 cert to sign and encrypt elements like NameID, Message, Assertion, Metadata.

If our environment requires sign or encrypt support, the certs folder may contain the X.509 cert and the private key that the SP will use:

Or also we can provide those data in the setting file at the x509cert and the privateKey JSON parameters of the sp element.

Sometimes we could need a signature on the metadata published by the SP, in this case we could use the X.509 cert previously mentioned or use a new X.509 cert: metadata.crt and metadata.key.

Use sp_new.crt if you are in a key rollover process and you want to publish that X.509 certificate on Service Provider metadata.

If you want to create self-signed certs, you can do it at the https://www.samltool.com/self_signed_certs.php service, or using the command:

openssl req -new -x509 -days 3652 -nodes -out sp.crt -keyout sp.key

demo-flask

This folder contains a Flask project that will be used as demo to show how to add SAML support to the Flask Framework. index.py is the main Flask file that has all the code, this file uses the templates stored at the templates folder. In the saml folder we found the certs folder to store the X.509 public and private key, and the SAML toolkit settings (settings.json and advanced_settings.json).

demo_pyramid

This folder contains a Pyramid project that will be used as demo to show how to add SAML support to the Pyramid Web Framework. \_\_init__.py is the main file that configures the app and its routes, views.py is where all the logic and SAML handling takes place, and the templates are stored in the templates folder. The saml folder is the same as in the other two demos.

demo-tornado

This folder contains a Tornado project that will be used as demo to show how to add SAML support to the Tornado Framework. views.py (with its settings.py) is the main Flask file that has all the code, this file uses the templates stored at the templates folder. In the saml folder we found the certs folder to store the X.509 public and private key, and the SAML toolkit settings (settings.json and advanced_settings.json).

It requires python3.8 (it's using tornado 6.4.1)

tests

Contains the unit test of the toolkit.

In order to execute the test you only need to load the virtualenv with the toolkit installed on it properly:

make install-test

and execute:

make pytest

The previous line will run the tests for the whole toolkit. You can also run the tests for a specific module. To do so for the auth module you would have to execute this:

pytest tests/src/OneLogin/saml2_tests/auth_test.py::OneLogin_Saml2_Auth_Test

Or for an specific method:

pytest tests/src/OneLogin/saml2_tests/auth_test.py::OneLogin_Saml2_Auth_Test::testBuildRequestSignature

How It Works

Settings

First of all we need to configure the toolkit. The SP's info, the IdP's info, and in some cases, configure advanced security issues like signatures and encryption.

There are two ways to provide the settings information:

In the demo-django and in the demo-flask folders you will find a saml folder, inside there is a certs folder and a settings.json and advanced_settings.json file. Those files contain the settings for the SAML toolkit. Copy them in your project and set the correct values.

This is the settings.json file:

{
    // If strict is True, then the Python Toolkit will reject unsigned
    // or unencrypted messages if it expects them to be signed or encrypted.
    // Also it will reject the messages if the SAML standard is not strictly
    // followed. Destination, NameId, Conditions ... are validated too.
    "strict": true,

    // Enable debug mode (outputs errors).
    "debug": true,

    // Service Provider Data that we are deploying.
    "sp": {
        // Identifier of the SP entity  (must be a URI)
        "entityId": "https://<sp_domain>/metadata/",
        // Specifies info about where and how the <AuthnResponse> message MUST be
        // returned to the requester, in this case our SP.
        "assertionConsumerService": {
            // URL Location where the <Response> from the IdP will be returned
            "url": "https://<sp_domain>/?acs",
            // SAML protocol binding to be used when returning the <Response>
            // message. SAML Toolkit supports this endpoint for the
            // HTTP-POST binding only.
            "binding": "urn:oasis:names:tc:SAML:2.0:bindings:HTTP-POST"
        },
        // Specifies info about where and how the <Logout Request/Response> message MUST be sent.
        "singleLogoutService": {
            // URL Location where the <LogoutRequest> from the IdP will be sent (IdP-initiated logout)
            "url": "https://<sp_domain>/?sls",
            // URL Location where the <LogoutResponse> from the IdP will sent (SP-initiated logout, reply)
            // OPTIONAL: only specify if different from url parameter
            //"responseUrl": "https://<sp_domain>/?sls",
            // SAML protocol binding to be used when returning the <Response>
            // message. SAML Toolkit supports the HTTP-Redirect binding
            // only for this endpoint.
            "binding": "urn:oasis:names:tc:SAML:2.0:bindings:HTTP-Redirect"
        },
        // If you need to specify requested attributes, set a
        // attributeConsumingService. nameFormat, attributeValue and
        // friendlyName can be omitted
        "attributeConsumingService": {
                // OPTIONAL: only specify if SP requires this.
                // index is an integer which identifies the attributeConsumingService used
                // to the SP. SAML toolkit supports configuring only one attributeConsumingService
                // but in certain cases the SP requires a different value.  Defaults to '1'.
                // "index": '1',
                "serviceName": "SP test",
                "serviceDescription": "Test Service",
                "requestedAttributes": [
                    {
                        "name": "",
                        "isRequired": false,
                        "nameFormat": "",
                        "friendlyName": "",
                        "attributeValue": []
                    }
                ]
        },
        // Specifies the constraints on the name identifier to be used to
        // represent the requested subject.
        // Take a look on src/onelogin/saml2/constants.py to see the NameIdFormat that are supported.
        "NameIDFormat": "urn:oasis:names:tc:SAML:1.1:nameid-format:unspecified",
        // Usually X.509 cert and privateKey of the SP are provided by files placed at
        // the certs folder. But we can also provide them with the following parameters
        "x509cert": "",
        "privateKey": ""

        /*
         * Key rollover
         * If you plan to update the SP X.509cert and privateKey
         * you can define here the new X.509cert and it will be
         * published on the SP metadata so Identity Providers can
         * read them and get ready for rollover.
         */
        // 'x509certNew': '',
    },

    // Identity Provider Data that we want connected with our SP.
    "idp": {
        // Identifier of the IdP entity  (must be a URI)
        "entityId": "https://app.onelogin.com/saml/metadata/<onelogin_connector_id>",
        // SSO endpoint info of the IdP. (Authentication Request protocol)
        "singleSignOnService": {
            // URL Target of the IdP where the Authentication Request Message
            // will be sent.
            "url": "https://app.onelogin.com/trust/saml2/http-post/sso/<onelogin_connector_id>",
            // SAML protocol binding to be used when returning the <Response>
            // message. SAML Toolkit supports the HTTP-Redirect binding
            // only for this endpoint.
            "binding": "urn:oasis:names:tc:SAML:2.0:bindings:HTTP-Redirect"
        },
        // SLO endpoint info of the IdP.
        "singleLogoutService": {
            // URL Location where the <LogoutRequest> from the IdP will be sent (IdP-initiated logout)
            "url": "https://app.onelogin.com/trust/saml2/http-redirect/slo/<onelogin_connector_id>",
            // URL Location where the <LogoutResponse> from the IdP will sent (SP-initiated logout, reply)
            // OPTIONAL: only specify if different from url parameter
            "responseUrl": "https://app.onelogin.com/trust/saml2/http-redirect/slo_return/<onelogin_connector_id>",
            // SAML protocol binding to be used when returning the <Response>
            // message. SAML Toolkit supports the HTTP-Redirect binding
            // only for this endpoint.
            "binding": "urn:oasis:names:tc:SAML:2.0:bindings:HTTP-Redirect"
        },
        // Public X.509 certificate of the IdP
        "x509cert": "<onelogin_connector_cert>"
        /*
         *  Instead of using the whole X.509cert you can use a fingerprint in order to
         *  validate a SAMLResponse (but you still need the X.509cert to validate LogoutRequest and LogoutResponse using the HTTP-Redirect binding).
         *  But take in mind that the algorithm for the fingerprint should be as strong as the algorithm in a normal certificate signature
     *  (e.g. SHA256 or strong)
         *
         *  (openssl x509 -noout -fingerprint -in "idp.crt" to generate it,
         *  or add for example the -sha256 , -sha384 or -sha512 parameter)
         *
         *  If a fingerprint is provided, then the certFingerprintAlgorithm is required in order to
         *  let the toolkit know which algorithm was used.
         Possible values: sha1, sha256, sha384 or sha512
         *  'sha1' is the default value.
         *
         *  Notice that if you want to validate any SAML Message sent by the HTTP-Redirect binding, you
         *  will need to provide the whole X.509cert.
         */
        // "certFingerprint": "",
        // "certFingerprintAlgorithm": "sha1",

        /* In some scenarios the IdP uses different certificates for
         * signing/encryption, or is under key rollover phase and
         * more than one certificate is published on IdP metadata.
         * In order to handle that the toolkit offers that parameter.
         * (when used, 'X.509cert' and 'certFingerprint' values are
         * ignored).
         */
        // 'x509certMulti': {
        //      'signing': [
        //          '<cert1-string>'
        //      ],
        //      'encryption': [
        //          '<cert2-string>'
        //      ]
        // }
    }
}

In addition to the required settings data (idp, sp), extra settings can be defined in advanced_settings.json:

{
    // Security settings
    "security": {

        /** signatures and encryptions offered **/

        // Indicates that the nameID of the <samlp:logoutRequest> sent by this SP
        // will be encrypted.
        "nameIdEncrypted": false,

        // Indicates whether the <samlp:AuthnRequest> messages sent by this SP
        // will be signed.  [Metadata of the SP will offer this info]
        "authnRequestsSigned": false,

        // Indicates whether the <samlp:logoutRequest> messages sent by this SP
        // will be signed.
        "logoutRequestSigned": false,

        // Indicates whether the <samlp:logoutResponse> messages sent by this SP
        // will be signed.
        "logoutResponseSigned": false,

        /* Sign the Metadata
         false || true (use sp certs) || {
                                            "keyFileName": "metadata.key",
                                            "certFileName": "metadata.crt"
                                         }
        */
        "signMetadata": false,

        /** signatures and encryptions required **/

        // Indicates a requirement for the <samlp:Response>, <samlp:LogoutRequest>
        // and <samlp:LogoutResponse> elements received by this SP to be signed.
        "wantMessagesSigned": false,

        // Indicates a requirement for the <saml:Assertion> elements received by
        // this SP to be signed. [Metadata of the SP will offer this info]
        "wantAssertionsSigned": false,

        // Indicates a requirement for the <saml:Assertion>
        // elements received by this SP to be encrypted.
        "wantAssertionsEncrypted": false,

        // Indicates a requirement for the NameID element on the SAMLResponse
        // received by this SP to be present.
        "wantNameId": true,

        // Indicates a requirement for the NameID received by
        // this SP to be encrypted.
        "wantNameIdEncrypted": false,

        // Indicates a requirement for the AttributeStatement element
        "wantAttributeStatement": true,

        // Authentication context.
        // Set to false and no AuthContext will be sent in the AuthNRequest,
        // Set true or don't present this parameter and you will get an AuthContext 'exact' 'urn:oasis:names:tc:SAML:2.0:ac:classes:PasswordProtectedTransport'
        // Set an array with the possible auth context values: array ('urn:oasis:names:tc:SAML:2.0:ac:classes:Password', 'urn:oasis:names:tc:SAML:2.0:ac:classes:X509'),
        "requestedAuthnContext": true,
    // Allows the authn comparison parameter to be set, defaults to 'exact' if the setting is not present.
        "requestedAuthnContextComparison": "exact",
        // Set to true to check that the AuthnContext(s) received match(es) the requested.
        "failOnAuthnContextMismatch": false,

        // In some environment you will need to set how long the published metadata of the Service Provider gonna be valid.
        // is possible to not set the 2 following parameters (or set to null) and default values will be set (2 days, 1 week)
        // Provide the desire TimeStamp, for example 2015-06-26T20:00:00Z
        "metadataValidUntil": null,
        // Provide the desire Duration, for example PT518400S (6 days)
        "metadataCacheDuration": null,

        // If enabled, URLs with single-label-domains will
        // be allowed and not rejected by the settings validator (Enable it under Docker/Kubernetes/testing env, not recommended on production)
        "allowSingleLabelDomains": false,

        // Algorithm that the toolkit will use on signing process. Options:
        //    'http://www.w3.org/2000/09/xmldsig#rsa-sha1'
        //    'http://www.w3.org/2000/09/xmldsig#dsa-sha1'
        //    'http://www.w3.org/2001/04/xmldsig-more#rsa-sha256'
        //    'http://www.w3.org/2001/04/xmldsig-more#rsa-sha384'
        //    'http://www.w3.org/2001/04/xmldsig-more#rsa-sha512'
        "signatureAlgorithm": "http://www.w3.org/2001/04/xmldsig-more#rsa-sha256",

        // Algorithm that the toolkit will use on digest process. Options:
        //    'http://www.w3.org/2000/09/xmldsig#sha1'
        //    'http://www.w3.org/2001/04/xmlenc#sha256'
        //    'http://www.w3.org/2001/04/xmldsig-more#sha384'
        //    'http://www.w3.org/2001/04/xmlenc#sha512'
        'digestAlgorithm': "http://www.w3.org/2001/04/xmlenc#sha256",

        // Specify if you want the SP to view assertions with duplicated Name or FriendlyName attributes to be valid
        // Defaults to false if not specified
        'allowRepeatAttributeName': false,

        // If the toolkit receive a message signed with a
        // deprecated algorithm (defined at the constant class)
        // will raise an error and reject the message
        "rejectDeprecatedAlgorithm": true
    },

    // Contact information template, it is recommended to suply a
    // technical and support contacts.
    "contactPerson": {
        "technical": {
            "givenName": "technical_name",
            "emailAddress": "technical@example.com"
        },
        "support": {
            "givenName": "support_name",
            "emailAddress": "support@example.com"
        }
    },

    // Organization information template, the info in en_US lang is
    // recommended, add more if required.
    "organization": {
        "en-US": {
            "name": "sp_test",
            "displayname": "SP test",
            "url": "http://sp.example.com"
        }
    }
}

In the security section, you can set the way that the SP will handle the messages and assertions. Contact the admin of the IdP and ask them what the IdP expects, and decide what validations will handle the SP and what requirements the SP will have and communicate them to the IdP's admin too.

Once we know what kind of data could be configured, let's talk about the way settings are handled within the toolkit.

The settings files described (settings.json and advanced_settings.json) are loaded by the toolkit if not other dict with settings info is provided in the constructors of the toolkit. Let's see some examples.

# Initializes toolkit with settings.json & advanced_settings.json files.
auth = OneLogin_Saml2_Auth(req)
# or
settings = OneLogin_Saml2_Settings()

# Initializes toolkit with settings.json & advanced_settings.json files from a custom base path.
custom_folder = '/var/www/django-project'
auth = OneLogin_Saml2_Auth(req, custom_base_path=custom_folder)
# or
settings = OneLogin_Saml2_Settings(custom_base_path=custom_folder)

# Initializes toolkit with the dict provided.
auth = OneLogin_Saml2_Auth(req, settings_data)
# or
settings = OneLogin_Saml2_Settings(settings_data)

You can declare the settings_data in the file that contains the constructor execution or locate them in any file and load the file in order to get the dict available as we see in the following example:

filename = "/var/www/django-project/custom_settings.json" # The custom_settings.json contains a
json_data_file = open(filename, 'r')                      # settings_data dict.
settings_data = json.load(json_data_file)
json_data_file.close()

auth = OneLogin_Saml2_Auth(req, settings_data)

Metadata Based Configuration

The method above requires a little extra work to manually specify attributes about the IdP. (And your SP application)

There's an easier method -- use a metadata exchange. Metadata is just an XML file that defines the capabilities of both the IdP and the SP application. It also contains the X.509 public key certificates which add to the trusted relationship. The IdP administrator can also configure custom settings for an SP based on the metadata.

Using parse_remote IdP metadata can be obtained and added to the settings without further ado.

Take in mind that the OneLogin_Saml2_IdPMetadataParser class does not validate in any way the URL that is introduced in order to be parsed.

Usually the same administrator that handles the Service Provider also sets the URL to the IdP, which should be a trusted resource.

But there are other scenarios, like a SAAS app where the administrator of the app delegates this functionality to other users. In this case, extra precaution should be taken in order to validate such URL inputs and avoid attacks like SSRF.

idp_data = OneLogin_Saml2_IdPMetadataParser.parse_remote('https://example.com/auth/saml2/idp/metadata')

You can specify a timeout in seconds for metadata retrieval, without it is not guaranteed that the request will complete

idp_data = OneLogin_Saml2_IdPMetadataParser.parse_remote('https://example.com/auth/saml2/idp/metadata', timeout=5)

If the Metadata contains several entities, the relevant EntityDescriptor can be specified when retrieving the settings from the IdpMetadataParser by its entityId value:

idp_data = OneLogin_Saml2_IdPMetadataParser.parse_remote(https://example.com/metadatas, entity_id='idp_entity_id')

How load the library

In order to use the toolkit library you need to import the file that contains the class that you will need on the top of your python file.

from onelogin.saml2.auth import OneLogin_Saml2_Auth
from onelogin.saml2.settings import OneLogin_Saml2_Settings
from onelogin.saml2.utils import OneLogin_Saml2_Utils

The Request

Building a OneLogin\_Saml2\_Auth object requires a request parameter:

auth = OneLogin_Saml2_Auth(req)

This parameter has the following scheme:

req = {
    "http_host": "",
    "script_name": "",
    "get_data": "",
    "post_data": "",

    # Advanced request options
    "https": "",
    "request_uri": "",
    "query_string": "",
    "validate_signature_from_qs": False,
    "lowercase_urlencoding": False
}

Each Python framework builds its own request object, you may map its data to match what the SAML toolkit expects. Let`s see some examples:

def prepare_from_django_request(request):
    return {
        'http_host': request.META['HTTP_HOST'],
        'script_name': request.META['PATH_INFO'],
        'get_data': request.GET.copy(),
        'post_data': request.POST.copy()
    }

def prepare_from_flask_request(request):
    url_data = urlparse(request.url)
    return {
        'http_host': request.netloc,
        'script_name': request.path,
        'get_data': request.args.copy(),
        'post_data': request.form.copy()
    }

An explanation of some advanced request parameters:

Initiate SSO

In order to send an AuthNRequest to the IdP:

from onelogin.saml2.auth import OneLogin_Saml2_Auth

req = prepare_request_for_toolkit(request)
auth = OneLogin_Saml2_Auth(req)   # Constructor of the SP, loads settings.json
                                  # and advanced_settings.json

auth.login()      # This method will build and return a AuthNRequest URL that can be
                  # either redirected to, or printed out onto the screen as a hyperlink

The AuthNRequest will be sent signed or unsigned based on the security info of the advanced_settings.json file (i.e. authnRequestsSigned).

The IdP will then return the SAML Response to the user's client. The client is then forwarded to the Assertion Consumer Service (ACS) of the SP with this information.

We can set a return_to url parameter to the login function and that will be converted as a RelayState parameter:

target_url = 'https://example.com'
auth.login(return_to=target_url)

The login method can receive 3 more optional parameters:

If a match on the future SAMLResponse ID and the AuthNRequest ID to be sent is required, that AuthNRequest ID must to be extracted and stored for future validation, we can get that ID by

auth.get_last_request_id()

The SP Endpoints

Related to the SP there are 3 important endpoints: The metadata view, the ACS view and the SLS view. The toolkit provides examples of those views in the demos, but let's see an example.

SP Metadata

This code will provide the XML metadata file of our SP, based on the info that we provided in the settings files.

req = prepare_request_for_toolkit(request)
auth = OneLogin_Saml2_Auth(req)
saml_settings = auth.get_settings()
metadata = saml_settings.get_sp_metadata()
errors = saml_settings.validate_metadata(metadata)
if len(errors) == 0:
    print(metadata)
else:
    print("Error found on Metadata: %s" % (', '.join(errors)))

The get_sp_metadata will return the metadata signed or not based on the security info of the advanced_settings.json (signMetadata).

Before the XML metadata is exposed, a check takes place to ensure that the info to be provided is valid.

Instead of using the Auth object, you can directly use

saml_settings = OneLogin_Saml2_Settings(settings=None, custom_base_path=None, sp_validation_only=True)

to get the settings object and with the sp_validation_only=True parameter we will avoid the IdP settings validation.

Assertion Consumer Service (ACS)

This code handles the SAML response that the IdP forwards to the SP through the user's client.

req = prepare_request_for_toolkit(request)
auth = OneLogin_Saml2_Auth(req)
auth.process_response()
errors = auth.get_errors()
if not errors:
    if auth.is_authenticated():
        request.session['samlUserdata'] = auth.get_attributes()
        if 'RelayState' in req['post_data'] and
          OneLogin_Saml2_Utils.get_self_url(req) != req['post_data']['RelayState']:
            # To avoid 'Open Redirect' attacks, before execute the redirection confirm
                # the value of the req['post_data']['RelayState'] is a trusted URL.
            auth.redirect_to(req['post_data']['RelayState'])
        else:
            for attr_name in request.session['samlUserdata'].keys():
                print('%s ==> %s' % (attr_name, '|| '.join(request.session['samlUserdata'][attr_name])))
    else:
      print('Not authenticated')
else:
    print("Error when processing SAML Response: %s %s" % (', '.join(errors), auth.get_last_error_reason()))

The SAML response is processed and then checked that there are no errors. It also verifies that the user is authenticated and stored the userdata in session.

At that point there are 2 possible alternatives:

Notice that we saved the user data in the session before the redirection to have the user data available at the RelayState view.

In order to retrieve attributes we use:

attributes = auth.get_attributes()

With this method we get a dict with all the user data provided by the IdP in the assertion of the SAML response.

If we execute print attributes we could get:

{
    "cn": ["Jhon"],
    "sn": ["Doe"],
    "mail": ["Doe"],
    "groups": ["users", "members"]
}

Each attribute name can be used as a key to obtain the value. Every attribute is a list of values. A single-valued attribute is a list of a single element.

The following code is equivalent:

attributes = auth.get_attributes()
print(attributes['cn'])

print(auth.get_attribute('cn'))

Before trying to get an attribute, check that the user is authenticated. If the user isn't authenticated, an empty dict will be returned. For example, if we call to get_attributes before a auth.process_response, the get_attributes() will return an empty dict.

Single Logout Service (SLS)

This code handles the Logout Request and the Logout Responses.

delete_session_callback = lambda: request.session.flush()
url = auth.process_slo(delete_session_cb=delete_session_callback)
errors = auth.get_errors()
if len(errors) == 0:
    if url is not None:
        # To avoid 'Open Redirect' attacks, before execute the redirection confirm
        # the value of the url is a trusted URL.
        return redirect(url)
    else:
        print("Successfully Logged out")
else:
    print("Error when processing SLO: %s %s" % (', '.join(errors), auth.get_last_error_reason()))

If the SLS endpoints receives a Logout Response, the response is validated and the session could be closed, using the callback.

# Part of the process_slo method
logout_response = OneLogin_Saml2_Logout_Response(self.__settings, self.__request_data['get_data']['SAMLResponse'])
if not logout_response.is_valid(self.__request_data, request_id):
    self.__errors.append('invalid_logout_response')
elif logout_response.get_status() != OneLogin_Saml2_Constants.STATUS_SUCCESS:
    self.__errors.append('logout_not_success')
elif not keep_local_session:
    OneLogin_Saml2_Utils.delete_local_session(delete_session_cb)

If the SLS endpoints receives an Logout Request, the request is validated, the session is closed and a Logout Response is sent to the SLS endpoint of the IdP.

# Part of the process_slo method
request = OneLogin_Saml2_Utils.decode_base64_and_inflate(self.__request_data['get_data']['SAMLRequest'])
if not OneLogin_Saml2_Logout_Request.is_valid(self.__settings, request, self.__request_data):
    self.__errors.append('invalid_logout_request')
else:
    if not keep_local_session:
        OneLogin_Saml2_Utils.delete_local_session(delete_session_cb)

    in_response_to = request.id
    response_builder = OneLogin_Saml2_Logout_Response(self.__settings)
    response_builder.build(in_response_to)
    logout_response = response_builder.get_response()

    parameters = {'SAMLResponse': logout_response}
    if 'RelayState' in self.__request_data['get_data']:
        parameters['RelayState'] = self.__request_data['get_data']['RelayState']

    security = self.__settings.get_security_data()
    if 'logoutResponseSigned' in security and security['logoutResponseSigned']:
        parameters['SigAlg'] = OneLogin_Saml2_Constants.RSA_SHA1
        parameters['Signature'] = self.build_response_signature(logout_response, parameters.get('RelayState', None))

    return self.redirect_to(self.get_slo_url(), parameters)

If we don't want that process_slo to destroy the session, pass a true parameter to the process_slo method:

keepLocalSession = true
auth.process_slo(keep_local_session=keepLocalSession);

Initiate SLO

In order to send a Logout Request to the IdP:

The Logout Request will be sent signed or unsigned based on the security info of the advanced_settings.json (logoutRequestSigned).

The IdP will return the Logout Response through the user's client to the Single Logout Service (SLS) of the SP.

We can set a return_to url parameter to the logout function and that will be converted as a RelayState parameter:

target_url = 'https://example.com'
auth.logout(return_to=target_url)

Also there are another 5 optional parameters that can be set:

If no name_id is provided, the LogoutRequest will contain a NameID with the entity Format. If name_id is provided and no name_id_format is provided, the NameIDFormat of the settings will be used.

If a match on the LogoutResponse ID and the LogoutRequest ID to be sent is required, that LogoutRequest ID must to be extracted and stored for future validation, we can get that ID by:

auth.get_last_request_id()

Example of a view that initiates the SSO request and handles the response (is the acs target)

We can code a unique file that initiates the SSO process, handle the response, get the attributes, initiate the SLO and processes the logout response.

Note: Review the demos, in a later section we explain the demo use case further in detail.

req = prepare_request_for_toolkit(request)  # Process the request and build the request dict that
                                            # the toolkit expects

auth = OneLogin_Saml2_Auth(req)             # Initialize the SP SAML instance

if 'sso' in request.args:                   # SSO action (SP-SSO initited).  Will send an AuthNRequest to the IdP
    return redirect(auth.login())
elif 'sso2' in request.args:                       # Another SSO init action
    return_to = '%sattrs/' % request.host_url      # but set a custom RelayState URL
    return redirect(auth.login(return_to))
elif 'slo' in request.args:                     # SLO action. Will sent a Logout Request to IdP
    nameid = request.session['samlNameId']
    nameid_format = request.session['samlNameIdFormat']
    nameid_nq = request.session['samlNameIdNameQualifier']
    nameid_spnq = request.session['samlNameIdSPNameQualifier']
    session_index = request.session['samlSessionIndex']
    return redirect(auth.logout(None, nameid, session_index, nameid_nq, nameid_format, nameid_spnq))
elif 'acs' in request.args:                 # Assertion Consumer Service
    auth.process_response()                     # Process the Response of the IdP
    errors = auth.get_errors()              # This method receives an array with the errors
    if len(errors) == 0:                    # that could took place during the process
        if not auth.is_authenticated():         # This check if the response was ok and the user
            msg = "Not authenticated"           # data retrieved or not (user authenticated)
        else:
            request.session['samlUserdata'] = auth.get_attributes()     # Retrieves user data
            request.session['samlNameId'] = auth.get_nameid()
            request.session['samlNameIdFormat'] = auth.get_nameid_format()
            request.session['samlNameIdNameQualifier'] = auth.get_nameid_nq()
            request.session['samlNameIdSPNameQualifier'] = auth.get_nameid_spnq()
            request.session['samlSessionIndex'] = auth.get_session_index()
            self_url = OneLogin_Saml2_Utils.get_self_url(req)
            if 'RelayState' in request.form and self_url != request.form['RelayState']:
                # To avoid 'Open Redirect' attacks, before execute the redirection confirm
                # the value of the request.form['RelayState'] is a trusted URL.
                return redirect(auth.redirect_to(request.form['RelayState']))   # Redirect if there is a relayState
            else:                           # If there is user data we save that to print it later.
                msg = ''
                for attr_name in request.session['samlUserdata'].keys():
                    msg += '%s ==> %s' % (attr_name, '|| '.join(request.session['samlUserdata'][attr_name]))
elif 'sls' in request.args:                                             # Single Logout Service
    delete_session_callback = lambda: session.clear()           # Obtain session clear callback
    url = auth.process_slo(delete_session_cb=delete_session_callback)   # Process the Logout Request & Logout Response
    errors = auth.get_errors()              #  Retrieves possible validation errors
    if len(errors) == 0:
        if url is not None:
            # To avoid 'Open Redirect' attacks, before execute the redirection confirm
            # the value of the url is a trusted URL.
            return redirect(url)
        else:
            msg = "Successfully logged out"

if len(errors) == 0:
  print(msg)
else:
  print(', '.join(errors))

SP Key rollover

If you plan to update the SP x509cert and privateKey you can define the new x509cert as settings['sp']['x509certNew'] and it will be published on the SP metadata so Identity Providers can read them and get ready for rollover.

IdP with multiple certificates

In some scenarios the IdP uses different certificates for signing/encryption, or is under key rollover phase and more than one certificate is published on IdP metadata.

In order to handle that the toolkit offers the settings['idp']['x509certMulti'] parameter.

When that parameter is used, x509cert and certFingerprint values will be ignored by the toolkit.

The x509certMulti is an array with 2 keys:

Replay attacks

In order to avoid replay attacks, you can store the ID of the SAML messages already processed, to avoid processing them twice. Since the Messages expires and will be invalidated due that fact, you don't need to store those IDs longer than the time frame that you currently accepting.

Get the ID of the last processed message/assertion with the get_last_message_id/get_last_assertion_id method of the Auth object.

Main classes and methods

Described below are the main classes and methods that can be invoked from the SAML2 library.

OneLogin_Saml2_Auth - auth.py

Main class of SAML Python Toolkit

OneLogin_Saml2_Auth - authn_request.py

SAML 2 Authentication Request class

OneLogin_Saml2_Response - response.py

SAML 2 Authentication Response class

OneLogin_Saml2_LogoutRequest - logout_request.py

SAML 2 Logout Request class

OneLogin_Saml2_LogoutResponse - logout_response.py

SAML 2 Logout Response class

OneLogin_Saml2_Settings - settings.py

Configuration of the SAML Python Toolkit

OneLogin_Saml2_Metadata - metadata.py

A class that contains functionality related to the metadata of the SP

OneLogin_Saml2_Utils - utils.py

Auxiliary class that contains several methods

OneLogin_Saml2_XML- xml_utils.py

A class that contains methods to handle XMLs

OneLogin_Saml2_IdPMetadataParser - idp_metadata_parser.py

A class that contains methods to obtain and parse metadata from IdP

For more info, look at the source code. Each method is documented and details about what does and how to use it are provided. Make sure to also check the doc folder where HTML documentation about the classes and methods is provided.

Demos included in the toolkit

The toolkit includes 3 demos to teach how use the toolkit (A Django, Flask and a Tornado project), take a look on it. Demos require that SP and IdP are well configured before test it, so edit the settings files.

Notice that each python framework has it own way to handle routes/urls and process request, so focus on how it deployed. New demos using other python frameworks are welcome as a contribution.

Getting Started

We said that this toolkit includes a Django application demo and a Flask application demo, let's see how fast is it to deploy them.

Virtualenv

The use of a virtualenv is highly recommended.

Virtualenv helps isolating the python environment used to run the toolkit. You can find more details and an installation guide in the official documentation.

Once you have your virtualenv ready and loaded, then you can install the toolkit executing this:

 make install-req

Demo Flask

You'll need a virtualenv with the toolkit installed on it.

To run the demo you need to install the requirements first. Load your virtualenv and execute:

 pip install -r demo-flask/requirements.txt

This will install flask and its dependencies. Once it has finished, you have to complete the configuration of the toolkit. You'll find it at demo-flask/settings.json

Now, with the virtualenv loaded, you can run the demo like this:

 cd demo-flask
 python index.py

You'll have the demo running at http://localhost:8000

Content

The flask project contains:

SP setup

The SAML Python Toolkit allows you to provide the settings info in 2 ways: Settings files or define a setting dict. In the demo-flask, it uses the first method.

In the index.py file we define the app.config['SAML_PATH'], that will target to the saml folder. We require it in order to load the settings files.

First we need to edit the saml/settings.json file, configure the SP part and review the metadata of the IdP and complete the IdP info. Later edit the saml/advanced_settings.json files and configure the how the toolkit will work. Check the settings section of this document if you have any doubt.

IdP setup

Once the SP is configured, the metadata of the SP is published at the /metadata url. Based on that info, configure the IdP.

How it works

  1. First time you access to the main view (http://localhost:8000), you can select to login and return to the same view or login and be redirected to /?attrs (attrs view).

  2. When you click:

    2.1 in the first link, we access to /?sso (index view). An AuthNRequest is sent to the IdP, we authenticate at the IdP and then a Response is sent through the user's client to the SP, specifically the Assertion Consumer Service view: /?acs. Notice that a RelayState parameter is set to the url that initiated the process, the index view.

    2.2 in the second link we access to /?attrs (attrs view), we will expetience have the same process described at 2.1 with the diference that as RelayState is set the attrs url.

  3. The SAML Response is processed in the ACS /?acs, if the Response is not valid, the process stops here and a message is shown. Otherwise we are redirected to the RelayState view. a) / or b) /?attrs

  4. We are logged in the app and the user attributes are showed. At this point, we can test the single log out functionality.

    The single log out functionality could be tested by 2 ways.

    5.1 SLO Initiated by SP. Click on the logout link at the SP, after that a Logout Request is sent to the IdP, the session at the IdP is closed and replies through the client to the SP with a Logout Response (sent to the Single Logout Service endpoint). The SLS endpoint /?sls of the SP process the Logout Response and if is valid, close the user session of the local app. Notice that the SLO Workflow starts and ends at the SP.

    5.2 SLO Initiated by IdP. In this case, the action takes place on the IdP side, the logout process is initiated at the IdP, sends a Logout Request to the SP (SLS endpoint, /?sls). The SLS endpoint of the SP process the Logout Request and if is valid, close the session of the user at the local app and send a Logout Response to the IdP (to the SLS endpoint of the IdP). The IdP receives the Logout Response, process it and close the session at of the IdP. Notice that the SLO Workflow starts and ends at the IdP.

Notice that all the SAML Requests and Responses are handled at a unique view (index) and how GET parameters are used to know the action that must be done.

Demo Tornado

You'll need a virtualenv with the toolkit installed on it.

First of all you need some packages, execute:

apt-get install libxml2-dev libxmlsec1-dev libxmlsec1-openssl

To run the demo you need to install the requirements first. Load your virtualenv and execute:

 pip install -r demo-tornado/requirements.txt

This will install tornado and its dependencies. Once it has finished, you have to complete the configuration of the toolkit. You'll find it at demo-tornado/saml/settings.json

Now, with the virtualenv loaded, you can run the demo like this:

 cd demo-tornado
 python views.py

You'll have the demo running at http://localhost:8000

Content

The tornado project contains:

SP setup

The SAML Python Toolkit allows you to provide the settings info in 2 ways: Settings files or define a setting dict. In the demo-tornado, it uses the first method.

In the settings.py file we define the SAML_PATH, that will target to the saml folder. We require it in order to load the settings files.

First we need to edit the saml/settings.json file, configure the SP part and review the metadata of the IdP and complete the IdP info. Later edit the saml/advanced_settings.json files and configure the how the toolkit will work. Check the settings section of this document if you have any doubt.

IdP setup

Once the SP is configured, the metadata of the SP is published at the /metadata url. Based on that info, configure the IdP.

How it works

  1. First time you access to the main view (http://localhost:8000), you can select to login and return to the same view or login and be redirected to /?attrs (attrs view).

    1. When you click:

    2.1 in the first link, we access to /?sso (index view). An AuthNRequest is sent to the IdP, we authenticate at the IdP and then a Response is sent through the user's client to the SP, specifically the Assertion Consumer Service view: /?acs. Notice that a RelayState parameter is set to the url that initiated the process, the index view.

    2.2 in the second link we access to /?attrs (attrs view), we will expetience have the same process described at 2.1 with the diference that as RelayState is set the attrs url.

    1. The SAML Response is processed in the ACS /?acs, if the Response is not valid, the process stops here and a message is shown. Otherwise we are redirected to the RelayState view. a) / or b) /?attrs

    2. We are logged in the app and the user attributes are showed. At this point, we can test the single log out functionality.

    The single log out functionality could be tested by 2 ways.

    5.1 SLO Initiated by SP. Click on the logout link at the SP, after that a Logout Request is sent to the IdP, the session at the IdP is closed and replies through the client to the SP with a Logout Response (sent to the Single Logout Service endpoint). The SLS endpoint /?sls of the SP process the Logout Response and if is valid, close the user session of the local app. Notice that the SLO Workflow starts and ends at the SP.

    5.2 SLO Initiated by IdP. In this case, the action takes place on the IdP side, the logout process is initiated at the IdP, sends a Logout Request to the SP (SLS endpoint, /?sls). The SLS endpoint of the SP process the Logout Request and if is valid, close the session of the user at the local app and send a Logout Response to the IdP (to the SLS endpoint of the IdP). The IdP receives the Logout Response, process it and close the session at of the IdP. Notice that the SLO Workflow starts and ends at the IdP.

Notice that all the SAML Requests and Responses are handled at a unique view (index) and how GET parameters are used to know the action that must be done.

Demo Django

You'll need a virtualenv with the toolkit installed on it.

To run the demo you need to install the requirements first. Load your virtualenv and execute:

 pip install -r demo-django/requirements.txt

This will install django and its dependencies. Once it has finished, you have to complete the configuration of the toolkit.

Later, with the virtualenv loaded, you can run the demo like this:

 cd demo-django
 python manage.py runserver 0.0.0.0:8000

You'll have the demo running at http://localhost:8000.

Note that many of the configuration files expect HTTPS. This is not required by the demo, as replacing these SP URLs with HTTP will work just fine. HTTPS is however highly encouraged, and left as an exercise for the reader for their specific needs.

If you want to integrate a production django application, take a look on this SAMLServiceProviderBackend that uses our toolkit to add SAML support: https://github.com/KristianOellegaard/django-saml-service-provider

Content

The django project contains:

SP setup

The SAML Python Toolkit allows you to provide the settings info in 2 ways: settings files or define a setting dict. In the demo-django it used the first method.

After set the SAML_FOLDER in the demo/settings.py, the settings of the Python toolkit will be loaded on the Django web.

First we need to edit the saml/settings.json, configure the SP part and review the metadata of the IdP and complete the IdP info. Later edit the saml/advanced_settings.json files and configure the how the toolkit will work. Check the settings section of this document if you have any doubt.

IdP setup

Once the SP is configured, the metadata of the SP is published at the /metadata url. Based on that info, configure the IdP.

How it works

This demo works very similar to the flask-demo (We did it intentionally).

Getting up and running on Heroku

Getting python3-saml up and running on Heroku will require some extra legwork: python3-saml depends on python-xmlsec which depends on headers from the xmlsec1-dev Linux package to install correctly.

First you will need to add the apt buildpack to your build server:

heroku buildpacks:add --index=1 -a your-app heroku-community/apt
heroku buildpacks:add --index=2 -a your-app heroku/python

You can confirm the buildpacks have been added in the correct order with heroku buildpacks -a your-app, you should see the apt buildpack first followed by the Python buildpack.

Then add an Aptfile into the root of your repository containing the libxmlsec1-dev package, the file should look like:

libxmlsec1-dev

Finally, add python3-saml to your requirements.txt and git push to trigger a build.

Demo Pyramid

Unlike the other two projects, you don't need a pre-existing virtualenv to get up and running here, since Pyramid comes from the buildout school of thought.

To run the demo you need to install Pyramid, the requirements, etc.:

 cd demo_pyramid
 python3 -m venv env
 env/bin/pip install --upgrade pip setuptools
 env/bin/pip install -e ".[testing]"

If you want to make sure the tests pass, run:

 env/bin/pytest

Next, edit the settings in demo_pyramid/saml/settings.json. (Pyramid runs on port 6543 by default.)

Now you can run the demo like this:

 env/bin/pserve development.ini

If that worked, the demo is now running at http://localhost:6543.

Content

The Pyramid project contains:

SP setup

The SAML Python Toolkit allows you to provide the settings info in 2 ways: settings files or define a setting dict. In demo_pyramid the first method is used.

In the views.py file we define the SAML_PATH, which will target the saml folder. We require it in order to load the settings files.

First we need to edit the saml/settings.json, configure the SP part and review the metadata of the IdP and complete the IdP info. Later edit the saml/advanced_settings.json files and configure the how the toolkit will work. Check the settings section of this document if you have any doubt.

IdP setup

Once the SP is configured, the metadata of the SP is published at the /metadata url. Based on that info, configure the IdP.

How it works

  1. First time you access to the main view (http://localhost:6543), you can select to login and return to the same view or login and be redirected to /?attrs (attrs view).

    1. When you click:

    2.1 in the first link, we access to /?sso (index view). An AuthNRequest is sent to the IdP, we authenticate at the IdP and then a Response is sent through the user's client to the SP, specifically the Assertion Consumer Service view: /?acs. Notice that a RelayState parameter is set to the url that initiated the process, the index view.

    2.2 in the second link we access to /?attrs (attrs view), we will experience the same process described at 2.1 with the diference that as RelayState is set the attrs url.

    1. The SAML Response is processed in the ACS /?acs, if the Response is not valid, the process stops here and a message is shown. Otherwise we are redirected to the RelayState view. a) / or b) /?attrs

    2. We are logged in the app and the user attributes are showed. At this point, we can test the single log out functionality.

    The single log out functionality could be tested by 2 ways.

    5.1 SLO Initiated by SP. Click on the "logout" link at the SP, after that a Logout Request is sent to the IdP, the session at the IdP is closed and replies through the client to the SP with a Logout Response (sent to the Single Logout Service endpoint). The SLS endpoint /?sls of the SP process the Logout Response and if is valid, close the user session of the local app. Notice that the SLO Workflow starts and ends at the SP.

    5.2 SLO Initiated by IdP. In this case, the action takes place on the IdP side, the logout process is initiated at the IdP, sends a Logout Request to the SP (SLS endpoint, /?sls). The SLS endpoint of the SP process the Logout Request and if is valid, close the session of the user at the local app and send a Logout Response to the IdP (to the SLS endpoint of the IdP). The IdP receives the Logout Response, process it and close the session at of the IdP. Notice that the SLO Workflow starts and ends at the IdP.

Notice that all the SAML Requests and Responses are handled at a unique view (index) and how GET parameters are used to know the action that must be done.