path: root/tests/test_build.py

# SPDX-License-Identifier: CC0-1.0

# Copyright (C) 2022 Wojtek Kosior <koszko@koszko.org>
#
# Available under the terms of Creative Commons Zero v1.0 Universal.

# Enable using with Python 3.7.
from __future__ import annotations

import pytest
import json
import shutil

from tempfile import TemporaryDirectory
from pathlib import Path, PurePosixPath
from hashlib import sha256
from zipfile import ZipFile
from contextlib import contextmanager

from jsonschema import ValidationError

from hydrilla import util as hydrilla_util
from hydrilla.builder import build, _version, local_apt
from hydrilla.builder.common_errors import *

from .helpers import *

here = Path(__file__).resolve().parent

expected_generated_by = {
    'name': 'hydrilla.builder',
    'version': _version.version
}

orig_srcdir = here / 'source-package-example'

index_text = (orig_srcdir / 'index.json').read_text()
index_obj = json.loads(hydrilla_util.strip_json_comments(index_text))

def read_files(*file_list):
    """
    Take names of files under srcdir and return a dict that maps them to their
    contents (as bytes).
    """
    return dict((name, (orig_srcdir / name).read_bytes()) for name in file_list)

dist_files = {
    **read_files('LICENSES/CC0-1.0.txt', 'bye.js', 'hello.js', 'message.js'),
    'report.spdx': b'dummy spdx output'
}
src_files = {
    **dist_files,
    **read_files('README.txt', 'README.txt.license', '.reuse/dep5',
                 'index.json')
}
extra_archive_files = {
}

sha256_hashes = dict((name, sha256(contents).digest().hex())
                     for name, contents in src_files.items())

del src_files['report.spdx']

expected_resources = [{
    '$schema': 'https://hydrilla.koszko.org/schemas/api_resource_description-1.schema.json',
    'source_name': 'hello',
    'source_copyright': [{
        'file': 'report.spdx',
        'sha256': sha256_hashes['report.spdx']
    }, {
        'file': 'LICENSES/CC0-1.0.txt',
        'sha256': sha256_hashes['LICENSES/CC0-1.0.txt']
    }],
    'type': 'resource',
    'identifier': 'helloapple',
    'long_name': 'Hello Apple',
    'uuid': 'a6754dcb-58d8-4b7a-a245-24fd7ad4cd68',
    'version': [2021, 11, 10],
    'revision': 1,
    'description': 'greets an apple',
    'dependencies': [{'identifier': 'hello-message'}],
    'scripts': [{
        'file': 'hello.js',
        'sha256': sha256_hashes['hello.js']
    }, {
        'file': 'bye.js',
        'sha256': sha256_hashes['bye.js']
    }],
    'generated_by': expected_generated_by
}, {
    '$schema': 'https://hydrilla.koszko.org/schemas/api_resource_description-1.schema.json',
    'source_name': 'hello',
    'source_copyright': [{
        'file': 'report.spdx',
        'sha256': sha256_hashes['report.spdx']
    }, {
        'file': 'LICENSES/CC0-1.0.txt',
        'sha256': sha256_hashes['LICENSES/CC0-1.0.txt']
    }],
    'type': 'resource',
    'identifier': 'hello-message',
    'long_name': 'Hello Message',
    'uuid': '1ec36229-298c-4b35-8105-c4f2e1b9811e',
    'version': [2021, 11, 10],
    'revision': 2,
    'description': 'define messages for saying hello and bye',
    'dependencies': [],
    'scripts': [{
        'file': 'message.js',
        'sha256': sha256_hashes['message.js']
    }],
    'generated_by': expected_generated_by
}]

expected_mapping = {
    '$schema': 'https://hydrilla.koszko.org/schemas/api_mapping_description-1.schema.json',
    'source_name': 'hello',
    'source_copyright': [{
        'file': 'report.spdx',
        'sha256': sha256_hashes['report.spdx']
    }, {
        'file': 'LICENSES/CC0-1.0.txt',
        'sha256': sha256_hashes['LICENSES/CC0-1.0.txt']
    }],
    'type': 'mapping',
    'identifier': 'helloapple',
    'long_name': 'Hello Apple',
    'uuid': '54d23bba-472e-42f5-9194-eaa24c0e3ee7',
    'version': [2021, 11, 10],
    'description': 'causes apple to get greeted on Hydrillabugs issue tracker',
    'payloads': {
	'https://hydrillabugs.koszko.org/***': {
	    'identifier': 'helloapple'
	},
	'https://hachettebugs.koszko.org/***': {
	    'identifier': 'helloapple'
        }
    },
    'generated_by': expected_generated_by
}

expected_source_description = {
    '$schema': 'https://hydrilla.koszko.org/schemas/api_source_description-1.schema.json',
    'source_name': 'hello',
    'source_copyright': [{
        'file': 'report.spdx',
        'sha256': sha256_hashes['report.spdx']
    }, {
        'file': 'LICENSES/CC0-1.0.txt',
        'sha256': sha256_hashes['LICENSES/CC0-1.0.txt']
    }],
    'source_archives': {
        'zip': {
            'sha256': '!!!!value to fill during test!!!!',
        }
    },
    'upstream_url': 'https://git.koszko.org/hydrilla-source-package-example',
    'definitions': [{
        'type': 'resource',
        'identifier': 'helloapple',
        'long_name': 'Hello Apple',
        'version': [2021, 11, 10],
    }, {
        'type':       'resource',
        'identifier': 'hello-message',
        'long_name': 'Hello Message',
        'version':     [2021, 11, 10],
    }, {
        'type': 'mapping',
        'identifier': 'helloapple',
	'long_name': 'Hello Apple',
        'version': [2021, 11, 10],
    }],
    'generated_by': expected_generated_by
}

expected = [*expected_resources, expected_mapping, expected_source_description]

@pytest.fixture
def tmpdir() -> Iterable[str]:
    """
    Provide test case with a temporary directory that will be automatically
    deleted after the test.
    """
    with TemporaryDirectory() as tmpdir:
        yield Path(tmpdir)

def run_reuse(command, **kwargs):
    """
    Instead of running a 'reuse' command, check if 'mock_reuse_missing' file
    exists under root directory. If yes, raise FileNotFoundError as if 'reuse'
    command was missing. If not, check if 'README.txt.license' file exists
    in the requested directory and return zero if it does.
    """
    expected = ['reuse', '--root', '<root>',
                'lint' if 'lint' in command else 'spdx']

    root_path = Path(process_command(command, expected)['root'])

    if (root_path / 'mock_reuse_missing').exists():
        raise FileNotFoundError('dummy')

    is_reuse_compliant = (root_path / 'README.txt.license').exists()

    return MockedCompletedProcess(command, 1 - is_reuse_compliant,
                                  stdout=f'dummy {expected[-1]} output',
                                  text_output=kwargs.get('text'))

mocked_piggybacked_archives = [
    PurePosixPath('apt/something.deb'),
    PurePosixPath('apt/something.orig.tar.gz'),
    PurePosixPath('apt/something.debian.tar.xz'),
    PurePosixPath('othersystem/other-something.tar.gz')
]

@pytest.fixture
def mock_piggybacked_apt_system(monkeypatch):
    """Make local_apt.piggybacked_system() return a mocked result."""
    # We set 'td' to a temporary dir path further below.
    td = None

    class MockedPiggybacked:
        """Minimal mock of Piggybacked object."""
        package_license_files = [PurePosixPath('.apt-root/.../copyright')]
        package_must_depend = [{'identifier': 'apt-common-licenses'}]

        def resolve_file(path):
            """
            For each path that starts with '.apt-root' return a valid
            dummy file path.
            """
            if path.parts[0] != '.apt-root':
                return None

            (td / path.name).write_text(f'dummy {path.name}')

            return (td / path.name)

        def archive_files():
            """Yield some valid dummy file path tuples."""
            for desired_path in mocked_piggybacked_archives:
                real_path = td / desired_path.name
                real_path.write_text(f'dummy {desired_path.name}')

                yield desired_path, real_path

    @contextmanager
    def mocked_piggybacked_system(piggyback_def, piggyback_files):
        """Mock the execution of local_apt.piggybacked_system()."""
        assert piggyback_def == {
	    'system': 'apt',
	    'distribution': 'nabia',
	    'packages': ['somelib=1.0'],
	    'dependencies': False
        }
        if piggyback_files is not None:
            assert {str(path) for path in mocked_piggybacked_archives} == \
                {path.relative_to(piggyback_files).as_posix()
                 for path in piggyback_files.rglob('*') if path.is_file()}

        yield MockedPiggybacked

    monkeypatch.setattr(local_apt, 'piggybacked_system',
                        mocked_piggybacked_system)

    with TemporaryDirectory() as td:
        td = Path(td)
        yield

@pytest.fixture
def sample_source():
    """Prepare a directory with sample Haketilo source package."""
    with TemporaryDirectory() as td:
        sample_source = Path(td) / 'hello'
        for name, contents in src_files.items():
            path = sample_source / name
            path.parent.mkdir(parents=True, exist_ok=True)
            path.write_bytes(contents)

        yield sample_source

variant_makers = []
def variant_maker(function):
    """Decorate function by placing it in variant_makers array."""
    variant_makers.append(function)
    return function

@variant_maker
def sample_source_change_index_json(monkeypatch, sample_source):
    """
    Return a non-standard path for index.json. Ensure parent directories exist.
    """
    # Use a path under sample_source so that it gets auto-deleted after the
    # test. Use a file under .git because .git is ignored by REUSE.
    path = sample_source / '.git' / 'replacement.json'
    path.parent.mkdir()
    return path

@variant_maker
def sample_source_add_comments(monkeypatch, sample_source):
    """Add index.json comments that should be preserved."""
    for dictionary in (index_obj, expected_source_description):
        monkeypatch.setitem(dictionary, 'comment', 'index.json comment')

    for i, dicts in enumerate(zip(index_obj['definitions'], expected)):
        for dictionary in dicts:
            monkeypatch.setitem(dictionary, 'comment', 'index.json comment')

@variant_maker
def sample_source_remove_spdx(monkeypatch, sample_source):
    """Remove spdx report generation."""
    monkeypatch.delitem(index_obj, 'reuse_generate_spdx_report')

    for obj, key in [
            (index_obj, 'copyright'),
            *((definition, 'source_copyright') for definition in expected)
    ]:
        new_list = [r for r in obj[key] if r['file'] != 'report.spdx']
        monkeypatch.setitem(obj, key, new_list)

    monkeypatch.delitem(dist_files, 'report.spdx')

    # To verify that reuse does not get called now, make mocked subprocess.run()
    # raise an error if called.
    (sample_source / 'mock_reuse_missing').touch()

@variant_maker
def sample_source_remove_additional_files(monkeypatch, sample_source):
    """Use default value ([]) for 'additionall_files' property."""
    monkeypatch.delitem(index_obj, 'additional_files')

    for name in 'README.txt', 'README.txt.license', '.reuse/dep5':
        monkeypatch.delitem(src_files, name)

@variant_maker
def sample_source_remove_script(monkeypatch, sample_source):
    """Use default value ([]) for 'scripts' property in one of the resources."""
    monkeypatch.delitem(index_obj['definitions'][1], 'scripts')

    monkeypatch.setitem(expected_resources[1], 'scripts', [])

    for files in dist_files, src_files:
        monkeypatch.delitem(files, 'message.js')

@variant_maker
def sample_source_remove_payloads(monkeypatch, sample_source):
    """Use default value ({}) for 'payloads' property in mapping."""
    monkeypatch.delitem(index_obj['definitions'][2], 'payloads')

    monkeypatch.setitem(expected_mapping, 'payloads', {})

@variant_maker
def sample_source_remove_uuids(monkeypatch, sample_source):
    """Don't use UUIDs (they are optional)."""
    for definition in index_obj['definitions']:
        monkeypatch.delitem(definition, 'uuid')

    for description in expected:
        if 'uuid' in description:
            monkeypatch.delitem(description, 'uuid')

@variant_maker
def sample_source_add_extra_props(monkeypatch, sample_source):
    """Add some unrecognized properties that should be stripped."""
    to_process = [index_obj]
    while to_process:
        processed = to_process.pop()

        if type(processed) is list:
            to_process.extend(processed)
        elif type(processed) is dict and 'spurious_property' not in processed:
            to_process.extend(v for k, v in processed.items()
                              if k != 'payloads')
            monkeypatch.setitem(processed, 'spurious_property', 'some_value')

piggyback_archive_names = [
    'apt/something.deb',
    'apt/something.orig.tar.gz',
    'apt/something.debian.tar.xz',
    'othersystem/other-something.tar.gz'
]

@variant_maker
def sample_source_add_piggyback(monkeypatch, sample_source,
                                extra_build_args={}):
    """Add piggybacked foreign system packages."""
    old_build = build.Build
    new_build = lambda *a, **kwa: old_build(*a, **kwa, **extra_build_args)
    monkeypatch.setattr(build, 'Build', new_build)

    monkeypatch.setitem(index_obj, 'piggyback_on', {
	'system': 'apt',
	'distribution': 'nabia',
	'packages': ['somelib=1.0'],
	'dependencies': False
    })
    schema = 'https://hydrilla.koszko.org/schemas/package_source-2.schema.json'
    monkeypatch.setitem(index_obj, '$schema', schema)

    new_refs = {}
    for name in '.apt-root/.../copyright', '.apt-root/.../script.js':
        contents = f'dummy {PurePosixPath(name).name}'.encode()
        digest = sha256(contents).digest().hex()
        monkeypatch.setitem(dist_files, name, contents)
        monkeypatch.setitem(sha256_hashes, name, digest)
        new_refs[PurePosixPath(name).name] = {'file': name, 'sha256': digest}

    for obj in expected:
        new_list = [*obj['source_copyright'], new_refs['copyright']]
        monkeypatch.setitem(obj, 'source_copyright', new_list)

    for obj in expected_resources:
        new_list = [{'identifier': 'apt-common-licenses'}, *obj['dependencies']]
        monkeypatch.setitem(obj, 'dependencies', new_list)

    for obj in index_obj['definitions'][0], expected_resources[0]:
        new_list = [new_refs['script.js'], *obj['scripts']]
        monkeypatch.setitem(obj, 'scripts', new_list)

    for name in piggyback_archive_names:
        path = PurePosixPath('hello.foreign-packages') / name
        monkeypatch.setitem(extra_archive_files, str(path),
                            f'dummy {path.name}'.encode())

def prepare_foreign_packages_dir(path):
    """
    Put some dummy archive in the directory so that it can be passed to
    piggybacked_system().
    """
    for name in piggyback_archive_names:
        archive_path = path / name
        archive_path.parent.mkdir(parents=True, exist_ok=True)
        archive_path.write_text(f'dummy {archive_path.name}')

@variant_maker
def sample_source_add_piggyback_pass_archives(monkeypatch, sample_source):
    """
    Add piggybacked foreign system packages, use pre-downloaded foreign package
    archives (have Build() find them in their default directory).
    """
    # Dir next to 'sample_source' will also be gc'd by sample_source() fixture.
    foreign_packages_dir = sample_source.parent / 'arbitrary-name'

    prepare_foreign_packages_dir(foreign_packages_dir)

    sample_source_add_piggyback(monkeypatch, sample_source,
                                {'piggyback_files': foreign_packages_dir})

@variant_maker
def sample_source_add_piggyback_find_archives(monkeypatch, sample_source):
    """
    Add piggybacked foreign system packages, use pre-downloaded foreign package
    archives (specify their directory as argument to Build()).
    """
    # Dir next to 'sample_source' will also be gc'd by sample_source() fixture.
    foreign_packages_dir = sample_source.parent / 'hello.foreign-packages'

    prepare_foreign_packages_dir(foreign_packages_dir)

    sample_source_add_piggyback(monkeypatch, sample_source)

@variant_maker
def sample_source_add_piggyback_no_download(monkeypatch, sample_source,
                                            pass_directory_to_build=False):
    """
    Add piggybacked foreign system packages, use pre-downloaded foreign package
    archives.
    """
    # Use a dir next to 'sample_source'; have it gc'd by sample_source fixture.
    if pass_directory_to_build:
        foreign_packages_dir = sample_source.parent / 'arbitrary-name'
    else:
        foreign_packages_dir = sample_source.parent / 'hello.foreign-packages'

    prepare_foreign_packages_dir(foreign_packages_dir)

    sample_source_add_piggyback(monkeypatch, sample_source)

@pytest.fixture(params=[lambda m, s: None, *variant_makers])
def sample_source_make_variants(request, monkeypatch, sample_source,
                                mock_piggybacked_apt_system):
    """
    Prepare a directory with sample Haketilo source package in multiple slightly
    different versions (all correct). Return an index.json path that should be
    used when performing test build.
    """
    index_path = request.param(monkeypatch, sample_source) or Path('index.json')

    index_text = json.dumps(index_obj)

    (sample_source / index_path).write_text(index_text)

    monkeypatch.setitem(src_files, 'index.json', index_text.encode())

    return index_path

@pytest.mark.subprocess_run(build, run_reuse)
@pytest.mark.usefixtures('mock_subprocess_run')
def test_build(sample_source, sample_source_make_variants, tmpdir):
    """Build the sample source package and verify the produced files."""
    index_json_path = sample_source_make_variants

    # First, build the package
    build.Build(sample_source, index_json_path).write_package_files(tmpdir)

    # Verify directories under destination directory
    assert {'file', 'resource', 'mapping', 'source'} == \
        set([path.name for path in tmpdir.iterdir()])

    # Verify files under 'file/'
    file_dir = tmpdir / 'file' / 'sha256'

    for name, contents in dist_files.items():
        dist_file_path = file_dir / sha256_hashes[name]
        assert dist_file_path.is_file()
        assert dist_file_path.read_bytes() == contents

    assert {p.name for p in file_dir.iterdir()} == \
        {sha256_hashes[name] for name in dist_files.keys()}

    # Verify files under 'resource/'
    resource_dir = tmpdir / 'resource'

    assert {rj['identifier'] for rj in expected_resources} == \
        {path.name for path in resource_dir.iterdir()}

    for resource_json in expected_resources:
        subdir = resource_dir / resource_json['identifier']
        assert ['2021.11.10'] == [path.name for path in subdir.iterdir()]

        assert json.loads((subdir / '2021.11.10').read_text()) == resource_json

        hydrilla_util.validator_for('api_resource_description-1.0.1.schema.json')\
                     .validate(resource_json)

    # Verify files under 'mapping/'
    mapping_dir = tmpdir / 'mapping'
    assert ['helloapple'] == [path.name for path in mapping_dir.iterdir()]

    subdir = mapping_dir / 'helloapple'
    assert ['2021.11.10'] == [path.name for path in subdir.iterdir()]

    assert json.loads((subdir / '2021.11.10').read_text()) == expected_mapping

    hydrilla_util.validator_for('api_mapping_description-1.0.1.schema.json')\
                 .validate(expected_mapping)

    # Verify files under 'source/'
    source_dir = tmpdir / 'source'
    assert {'hello.json', 'hello.zip'} == \
        {path.name for path in source_dir.iterdir()}

    archive_files = {**dict((f'hello/{name}', contents)
                            for name, contents in src_files.items()),
                     **extra_archive_files}

    with ZipFile(source_dir / 'hello.zip', 'r') as archive:
        print(archive.namelist())
        assert len(archive.namelist()) == len(archive_files)

        for name, contents in archive_files.items():
            assert archive.read(name) == contents

    zip_ref = expected_source_description['source_archives']['zip']
    zip_contents = (source_dir / 'hello.zip').read_bytes()
    zip_ref['sha256'] = sha256(zip_contents).digest().hex()

    assert json.loads((source_dir / 'hello.json').read_text()) == \
        expected_source_description

    hydrilla_util.validator_for('api_source_description-1.0.1.schema.json')\
                 .validate(expected_source_description)

error_makers = []
def error_maker(function):
    """Decorate function by placing it in error_makers array."""
    error_makers.append(function)

@error_maker
def sample_source_error_missing_file(monkeypatch, sample_source):
    """
    Modify index.json to expect missing report.spdx file and cause an error.
    """
    monkeypatch.delitem(index_obj, 'reuse_generate_spdx_report')
    return FileNotFoundError

@error_maker
def sample_source_error_index_schema(monkeypatch, sample_source):
    """Modify index.json to be incompliant with the schema."""
    monkeypatch.delitem(index_obj, 'definitions')
    return ValidationError

@error_maker
def sample_source_error_bad_comment(monkeypatch, sample_source):
    """Modify index.json to have an invalid '/' in it."""
    return json.JSONDecodeError, json.dumps(index_obj) + '/something\n'

@error_maker
def sample_source_error_bad_json(monkeypatch, sample_source):
    """Modify index.json to not be valid json even after comment stripping."""
    return json.JSONDecodeError, json.dumps(index_obj) + '???/\n'

@error_maker
def sample_source_error_missing_reuse(monkeypatch, sample_source):
    """Cause mocked reuse process invocation to fail with FileNotFoundError."""
    (sample_source / 'mock_reuse_missing').touch()
    return build.ReuseError

@error_maker
def sample_source_error_missing_license(monkeypatch, sample_source):
    """Remove a file to make package REUSE-incompliant."""
    (sample_source / 'README.txt.license').unlink()
    return build.ReuseError

@error_maker
def sample_source_error_file_outside(monkeypatch, sample_source):
    """Make index.json illegally reference a file outside srcdir."""
    new_list = [*index_obj['copyright'], {'file': '../abc'}]
    monkeypatch.setitem(index_obj, 'copyright', new_list)
    return FileReferenceError

@error_maker
def sample_source_error_reference_itself(monkeypatch, sample_source):
    """Make index.json illegally reference index.json."""
    new_list = [*index_obj['copyright'], {'file': 'index.json'}]
    monkeypatch.setitem(index_obj, 'copyright', new_list)
    return FileReferenceError

@error_maker
def sample_source_error_report_excluded(monkeypatch, sample_source):
    """
    Make index.json require generation of report.spdx but don't include it among
    copyright files.
    """
    new_list = [file_ref for file_ref in index_obj['copyright']
                if file_ref['file'] != 'report.spdx']
    monkeypatch.setitem(index_obj, 'copyright', new_list)
    return FileReferenceError

@pytest.fixture(params=error_makers)
def sample_source_make_errors(request, monkeypatch, sample_source):
    """
    Prepare a directory with sample Haketilo source package in multiple slightly
    broken versions. Return an error type that should be raised when running
    test build.
    """
    index_text = None
    error_type = request.param(monkeypatch, sample_source)
    if type(error_type) is tuple:
        error_type, index_text = error_type

    index_text = index_text or json.dumps(index_obj)

    (sample_source / 'index.json').write_text(index_text)

    monkeypatch.setitem(src_files, 'index.json', index_text.encode())

    return error_type

@pytest.mark.subprocess_run(build, run_reuse)
@pytest.mark.usefixtures('mock_subprocess_run')
def test_build_error(tmpdir, sample_source, sample_source_make_errors):
    """Try building the sample source package and verify generated errors."""
    error_type = sample_source_make_errors

    dstdir = Path(tmpdir) / 'dstdir'
    tmpdir = Path(tmpdir) / 'example'

    dstdir.mkdir(exist_ok=True)
    tmpdir.mkdir(exist_ok=True)

    with pytest.raises(error_type):
        build.Build(sample_source, Path('index.json'))\
             .write_package_files(dstdir)