path: root/testing/web-platform/tests/common/security-features/tools/generate.py

#!/usr/bin/env python3

from __future__ import print_function

import argparse
import collections
import copy
import json
import os
import sys

import spec_validator
import util


def expand_pattern(expansion_pattern, test_expansion_schema):
    expansion = {}
    for artifact_key in expansion_pattern:
        artifact_value = expansion_pattern[artifact_key]
        if artifact_value == '*':
            expansion[artifact_key] = test_expansion_schema[artifact_key]
        elif isinstance(artifact_value, list):
            expansion[artifact_key] = artifact_value
        elif isinstance(artifact_value, dict):
            # Flattened expansion.
            expansion[artifact_key] = []
            values_dict = expand_pattern(artifact_value,
                                         test_expansion_schema[artifact_key])
            for sub_key in values_dict.keys():
                expansion[artifact_key] += values_dict[sub_key]
        else:
            expansion[artifact_key] = [artifact_value]

    return expansion


def permute_expansion(expansion,
                      artifact_order,
                      selection={},
                      artifact_index=0):
    assert isinstance(artifact_order, list), "artifact_order should be a list"

    if artifact_index >= len(artifact_order):
        yield selection
        return

    artifact_key = artifact_order[artifact_index]

    for artifact_value in expansion[artifact_key]:
        selection[artifact_key] = artifact_value
        for next_selection in permute_expansion(expansion, artifact_order,
                                                selection, artifact_index + 1):
            yield next_selection


# Dumps the test config `selection` into a serialized JSON string.
def dump_test_parameters(selection):
    return json.dumps(
        selection,
        indent=2,
        separators=(',', ': '),
        sort_keys=True,
        cls=util.CustomEncoder)


def get_test_filename(spec_directory, spec_json, selection):
    '''Returns the filname for the main test HTML file'''

    selection_for_filename = copy.deepcopy(selection)
    # Use 'unset' rather than 'None' in test filenames.
    if selection_for_filename['delivery_value'] is None:
        selection_for_filename['delivery_value'] = 'unset'

    return os.path.join(
        spec_directory,
        spec_json['test_file_path_pattern'] % selection_for_filename)


def get_csp_value(value):
    '''
    Returns actual CSP header values (e.g. "worker-src 'self'") for the
    given string used in PolicyDelivery's value (e.g. "worker-src-self").
    '''

    # script-src
    # Test-related scripts like testharness.js and inline scripts containing
    # test bodies.
    # 'unsafe-inline' is added as a workaround here. This is probably not so
    # bad, as it shouldn't intefere non-inline-script requests that we want to
    # test.
    if value == 'script-src-wildcard':
        return "script-src * 'unsafe-inline'"
    if value == 'script-src-self':
        return "script-src 'self' 'unsafe-inline'"
    # Workaround for "script-src 'none'" would be more complicated, because
    # - "script-src 'none' 'unsafe-inline'" is handled somehow differently from
    #   "script-src 'none'", i.e.
    #   https://w3c.github.io/webappsec-csp/#match-url-to-source-list Step 3
    #   handles the latter but not the former.
    # - We need nonce- or path-based additional values to allow same-origin
    #   test scripts like testharness.js.
    # Therefore, we disable 'script-src-none' tests for now in
    # `/content-security-policy/spec.src.json`.
    if value == 'script-src-none':
        return "script-src 'none'"

    # worker-src
    if value == 'worker-src-wildcard':
        return 'worker-src *'
    if value == 'worker-src-self':
        return "worker-src 'self'"
    if value == 'worker-src-none':
        return "worker-src 'none'"
    raise Exception('Invalid delivery_value: %s' % value)

def handle_deliveries(policy_deliveries):
    '''
    Generate <meta> elements and HTTP headers for the given list of
    PolicyDelivery.
    TODO(hiroshige): Merge duplicated code here, scope/document.py, etc.
    '''

    meta = ''
    headers = {}

    for delivery in policy_deliveries:
        if delivery.value is None:
            continue
        if delivery.key == 'referrerPolicy':
            if delivery.delivery_type == 'meta':
                meta += \
                    '<meta name="referrer" content="%s">' % delivery.value
            elif delivery.delivery_type == 'http-rp':
                headers['Referrer-Policy'] = delivery.value
                # TODO(kristijanburnik): Limit to WPT origins.
                headers['Access-Control-Allow-Origin'] = '*'
            else:
                raise Exception(
                    'Invalid delivery_type: %s' % delivery.delivery_type)
        elif delivery.key == 'mixedContent':
            assert (delivery.value == 'opt-in')
            if delivery.delivery_type == 'meta':
                meta += '<meta http-equiv="Content-Security-Policy" ' + \
                       'content="block-all-mixed-content">'
            elif delivery.delivery_type == 'http-rp':
                headers['Content-Security-Policy'] = 'block-all-mixed-content'
            else:
                raise Exception(
                    'Invalid delivery_type: %s' % delivery.delivery_type)
        elif delivery.key == 'contentSecurityPolicy':
            csp_value = get_csp_value(delivery.value)
            if delivery.delivery_type == 'meta':
                meta += '<meta http-equiv="Content-Security-Policy" ' + \
                       'content="' + csp_value + '">'
            elif delivery.delivery_type == 'http-rp':
                headers['Content-Security-Policy'] = csp_value
            else:
                raise Exception(
                    'Invalid delivery_type: %s' % delivery.delivery_type)
        elif delivery.key == 'upgradeInsecureRequests':
            # https://w3c.github.io/webappsec-upgrade-insecure-requests/#delivery
            assert (delivery.value == 'upgrade')
            if delivery.delivery_type == 'meta':
                meta += '<meta http-equiv="Content-Security-Policy" ' + \
                       'content="upgrade-insecure-requests">'
            elif delivery.delivery_type == 'http-rp':
                headers[
                    'Content-Security-Policy'] = 'upgrade-insecure-requests'
            else:
                raise Exception(
                    'Invalid delivery_type: %s' % delivery.delivery_type)
        else:
            raise Exception('Invalid delivery_key: %s' % delivery.key)
    return {"meta": meta, "headers": headers}


def generate_selection(spec_json, selection):
    '''
    Returns a scenario object (with a top-level source_context_list entry,
    which will be removed in generate_test_file() later).
    '''

    target_policy_delivery = util.PolicyDelivery(selection['delivery_type'],
                                                 selection['delivery_key'],
                                                 selection['delivery_value'])
    del selection['delivery_type']
    del selection['delivery_key']
    del selection['delivery_value']

    # Parse source context list and policy deliveries of source contexts.
    # `util.ShouldSkip()` exceptions are raised if e.g. unsuppported
    # combinations of source contexts and policy deliveries are used.
    source_context_list_scheme = spec_json['source_context_list_schema'][
        selection['source_context_list']]
    selection['source_context_list'] = [
        util.SourceContext.from_json(source_context, target_policy_delivery,
                                     spec_json['source_context_schema'])
        for source_context in source_context_list_scheme['sourceContextList']
    ]

    # Check if the subresource is supported by the innermost source context.
    innermost_source_context = selection['source_context_list'][-1]
    supported_subresource = spec_json['source_context_schema'][
        'supported_subresource'][innermost_source_context.source_context_type]
    if supported_subresource != '*':
        if selection['subresource'] not in supported_subresource:
            raise util.ShouldSkip()

    # Parse subresource policy deliveries.
    selection[
        'subresource_policy_deliveries'] = util.PolicyDelivery.list_from_json(
            source_context_list_scheme['subresourcePolicyDeliveries'],
            target_policy_delivery, spec_json['subresource_schema']
            ['supported_delivery_type'][selection['subresource']])

    # Generate per-scenario test description.
    selection['test_description'] = spec_json[
        'test_description_template'] % selection

    return selection


def generate_test_file(spec_directory, test_helper_filenames,
                       test_html_template_basename, test_filename, scenarios):
    '''
    Generates a test HTML file (and possibly its associated .headers file)
    from `scenarios`.
    '''

    # Scenarios for the same file should have the same `source_context_list`,
    # including the top-level one.
    # Note: currently, non-top-level source contexts aren't necessarily required
    # to be the same, but we set this requirement as it will be useful e.g. when
    # we e.g. reuse a worker among multiple scenarios.
    for scenario in scenarios:
        assert (scenario['source_context_list'] == scenarios[0]
                ['source_context_list'])

    # We process the top source context below, and do not include it in
    # the JSON objects (i.e. `scenarios`) in generated HTML files.
    top_source_context = scenarios[0]['source_context_list'].pop(0)
    assert (top_source_context.source_context_type == 'top')
    for scenario in scenarios[1:]:
        assert (scenario['source_context_list'].pop(0) == top_source_context)

    parameters = {}

    # Sort scenarios, to avoid unnecessary diffs due to different orders in
    # `scenarios`.
    serialized_scenarios = sorted(
        [dump_test_parameters(scenario) for scenario in scenarios])

    parameters['scenarios'] = ",\n".join(serialized_scenarios).replace(
        "\n", "\n" + " " * 10)

    test_directory = os.path.dirname(test_filename)

    parameters['helper_js'] = ""
    for test_helper_filename in test_helper_filenames:
        parameters['helper_js'] += '    <script src="%s"></script>\n' % (
            os.path.relpath(test_helper_filename, test_directory))
    parameters['sanity_checker_js'] = os.path.relpath(
        os.path.join(spec_directory, 'generic', 'sanity-checker.js'),
        test_directory)
    parameters['spec_json_js'] = os.path.relpath(
        os.path.join(spec_directory, 'generic', 'spec_json.js'),
        test_directory)

    test_headers_filename = test_filename + ".headers"

    test_html_template = util.get_template(test_html_template_basename)
    disclaimer_template = util.get_template('disclaimer.template')

    html_template_filename = os.path.join(util.template_directory,
                                          test_html_template_basename)
    generated_disclaimer = disclaimer_template \
        % {'generating_script_filename': os.path.relpath(sys.argv[0],
                                                         util.test_root_directory),
           'spec_directory': os.path.relpath(spec_directory,
                                             util.test_root_directory)}

    # Adjust the template for the test invoking JS. Indent it to look nice.
    parameters['generated_disclaimer'] = generated_disclaimer.rstrip()

    # Directory for the test files.
    try:
        os.makedirs(test_directory)
    except:
        pass

    delivery = handle_deliveries(top_source_context.policy_deliveries)

    if len(delivery['headers']) > 0:
        with open(test_headers_filename, "w") as f:
            for header in delivery['headers']:
                f.write('%s: %s\n' % (header, delivery['headers'][header]))

    parameters['meta_delivery_method'] = delivery['meta']
    # Obey the lint and pretty format.
    if len(parameters['meta_delivery_method']) > 0:
        parameters['meta_delivery_method'] = "\n    " + \
                                            parameters['meta_delivery_method']

    # Write out the generated HTML file.
    util.write_file(test_filename, test_html_template % parameters)


def generate_test_source_files(spec_directory, test_helper_filenames,
                               spec_json, target):
    test_expansion_schema = spec_json['test_expansion_schema']
    specification = spec_json['specification']

    if target == "debug":
        spec_json_js_template = util.get_template('spec_json.js.template')
        util.write_file(
            os.path.join(spec_directory, "generic", "spec_json.js"),
            spec_json_js_template % {'spec_json': json.dumps(spec_json)})
        util.write_file(
            os.path.join(spec_directory, "generic",
                         "debug-output.spec.src.json"),
            json.dumps(spec_json, indent=2, separators=(',', ': ')))

    # Choose a debug/release template depending on the target.
    html_template = "test.%s.html.template" % target

    artifact_order = test_expansion_schema.keys()
    artifact_order.remove('expansion')

    excluded_selection_pattern = ''
    for key in artifact_order:
        excluded_selection_pattern += '%(' + key + ')s/'

    # Create list of excluded tests.
    exclusion_dict = set()
    for excluded_pattern in spec_json['excluded_tests']:
        excluded_expansion = \
            expand_pattern(excluded_pattern, test_expansion_schema)
        for excluded_selection in permute_expansion(excluded_expansion,
                                                    artifact_order):
            excluded_selection['delivery_key'] = spec_json['delivery_key']
            exclusion_dict.add(excluded_selection_pattern % excluded_selection)

    # `scenarios[filename]` represents the list of scenario objects to be
    # generated into `filename`.
    scenarios = {}

    for spec in specification:
        # Used to make entries with expansion="override" override preceding
        # entries with the same |selection_path|.
        output_dict = {}

        for expansion_pattern in spec['test_expansion']:
            expansion = expand_pattern(expansion_pattern,
                                       test_expansion_schema)
            for selection in permute_expansion(expansion, artifact_order):
                selection['delivery_key'] = spec_json['delivery_key']
                selection_path = spec_json['selection_pattern'] % selection
                if selection_path in output_dict:
                    if expansion_pattern['expansion'] != 'override':
                        print("Error: expansion is default in:")
                        print(dump_test_parameters(selection))
                        print("but overrides:")
                        print(dump_test_parameters(
                            output_dict[selection_path]))
                        sys.exit(1)
                output_dict[selection_path] = copy.deepcopy(selection)

        for selection_path in output_dict:
            selection = output_dict[selection_path]
            if (excluded_selection_pattern % selection) in exclusion_dict:
                print('Excluding selection:', selection_path)
                continue
            try:
                test_filename = get_test_filename(spec_directory, spec_json,
                                                  selection)
                scenario = generate_selection(spec_json, selection)
                scenarios[test_filename] = scenarios.get(test_filename,
                                                         []) + [scenario]
            except util.ShouldSkip:
                continue

    for filename in scenarios:
        generate_test_file(spec_directory, test_helper_filenames,
                           html_template, filename, scenarios[filename])


def merge_json(base, child):
    for key in child:
        if key not in base:
            base[key] = child[key]
            continue
        # `base[key]` and `child[key]` both exists.
        if isinstance(base[key], list) and isinstance(child[key], list):
            base[key].extend(child[key])
        elif isinstance(base[key], dict) and isinstance(child[key], dict):
            merge_json(base[key], child[key])
        else:
            base[key] = child[key]


def main():
    parser = argparse.ArgumentParser(
        description='Test suite generator utility')
    parser.add_argument(
        '-t',
        '--target',
        type=str,
        choices=("release", "debug"),
        default="release",
        help='Sets the appropriate template for generating tests')
    parser.add_argument(
        '-s',
        '--spec',
        type=str,
        default=os.getcwd(),
        help='Specify a file used for describing and generating the tests')
    # TODO(kristijanburnik): Add option for the spec_json file.
    args = parser.parse_args()

    spec_directory = os.path.abspath(args.spec)

    # Read `spec.src.json` files, starting from `spec_directory`, and
    # continuing to parent directories as long as `spec.src.json` exists.
    spec_filenames = []
    test_helper_filenames = []
    spec_src_directory = spec_directory
    while len(spec_src_directory) >= len(util.test_root_directory):
        spec_filename = os.path.join(spec_src_directory, "spec.src.json")
        if not os.path.exists(spec_filename):
            break
        spec_filenames.append(spec_filename)
        test_filename = os.path.join(spec_src_directory, 'generic',
                                     'test-case.sub.js')
        assert (os.path.exists(test_filename))
        test_helper_filenames.append(test_filename)
        spec_src_directory = os.path.abspath(
            os.path.join(spec_src_directory, ".."))

    spec_filenames = list(reversed(spec_filenames))
    test_helper_filenames = list(reversed(test_helper_filenames))

    if len(spec_filenames) == 0:
        print('Error: No spec.src.json is found at %s.' % spec_directory)
        return

    # Load the default spec JSON file, ...
    default_spec_filename = os.path.join(util.script_directory,
                                         'spec.src.json')
    spec_json = collections.OrderedDict()
    if os.path.exists(default_spec_filename):
        spec_json = util.load_spec_json(default_spec_filename)

    # ... and then make spec JSON files in subdirectories override the default.
    for spec_filename in spec_filenames:
        child_spec_json = util.load_spec_json(spec_filename)
        merge_json(spec_json, child_spec_json)

    spec_validator.assert_valid_spec_json(spec_json)
    generate_test_source_files(spec_directory, test_helper_filenames,
                               spec_json, args.target)


if __name__ == '__main__':
    main()