class SourceScanTraitTransformer(TraitTransformer):
name = 'scan_sources'
def __init__(self, trait, *args, **kwargs):
self.trait = trait
super().__init__(*args, **kwargs)
def inject_steps(self):
self.source_scan_step = PipelineStep(
name='scan_sources',
raw_dict={},
is_synthetic=True,
notification_policy=StepNotificationPolicy.NO_NOTIFICATION,
script_type=ScriptType.PYTHON3)
self.source_scan_step.add_input(
name=concourse.model.traits.component_descriptor.DIR_NAME,
variable_name=concourse.model.traits.component_descriptor.
ENV_VAR_NAME,
)
self.source_scan_step.set_timeout(duration_string='18h')
yield self.source_scan_step
def process_pipeline_args(self, pipeline_args: JobVariant):
# our step depends on dependency descriptor step
component_descriptor_step = pipeline_args.step('component_descriptor')
self.source_scan_step._add_dependency(component_descriptor_step)
@classmethod
def dependencies(cls):
return {'component_descriptor'}
class ImageScanTraitTransformer(TraitTransformer):
name = 'image_scan'
def __init__(self, trait, *args, **kwargs):
self.trait = trait
super().__init__(*args, **kwargs)
def inject_steps(self):
self.image_scan_step = PipelineStep(
name='scan_container_images',
raw_dict={},
is_synthetic=True,
notification_policy=StepNotificationPolicy.NO_NOTIFICATION,
script_type=ScriptType.PYTHON3)
self.image_scan_step.add_input(*COMPONENT_DESCRIPTOR_DIR_INPUT)
self.image_scan_step.set_timeout(duration_string='12h')
yield self.image_scan_step
def process_pipeline_args(self, pipeline_args: 'JobVariant'):
# our step depends on dependency descriptor step
component_descriptor_step = pipeline_args.step('component_descriptor')
self.image_scan_step._add_dependency(component_descriptor_step)
@classmethod
def dependencies(cls):
return {'component_descriptor'}
class ImageScanTraitTransformer(TraitTransformer):
name = 'image_scan'
def __init__(self, trait, *args, **kwargs):
self.trait = trait
super().__init__(*args, **kwargs)
def inject_steps(self):
self.image_scan_step = PipelineStep(
name='scan_container_images',
raw_dict={},
is_synthetic=True,
notification_policy=StepNotificationPolicy.NO_NOTIFICATION,
script_type=ScriptType.PYTHON3)
self.image_scan_step.add_input(
name=concourse.model.traits.component_descriptor.DIR_NAME,
variable_name=concourse.model.traits.component_descriptor.
ENV_VAR_NAME,
)
self.image_scan_step.set_timeout(duration_string='12h')
yield self.image_scan_step
def process_pipeline_args(self, pipeline_args: JobVariant):
# our step depends on dependency descriptor step
component_descriptor_step = pipeline_args.step('component_descriptor')
self.image_scan_step._add_dependency(component_descriptor_step)
for trait_name in self.trait.trait_depends():
if not pipeline_args.has_trait(trait_name):
raise ModelValidationError(
f'dependency towards absent trait: {trait_name}')
depended_on_trait = pipeline_args.trait(trait_name)
# XXX refactor Trait/TraitTransformer
transformer = depended_on_trait.transformer()
# XXX step-injection may have (unintended) side-effects :-/
depended_on_step_names = {
step.name
for step in transformer.inject_steps()
}
for step in pipeline_args.steps():
if not step.name in depended_on_step_names:
continue
self.image_scan_step._add_dependency(step)
# prevent cyclic dependencies (from auto-injected depends)
if self.image_scan_step.name in step.depends():
step._remove_dependency(self.image_scan_step)
@classmethod
def dependencies(cls):
return {'component_descriptor'}
@classmethod
def order_dependencies(cls):
# required in case image-scanning should be done after publish
# (-> auto-injected dependency for "prepare"-step towards _all_ steps)
return {'publish'}
class UpdateComponentDependenciesTraitTransformer(TraitTransformer):
name = 'update_component_deps'
def __init__(self, trait, *args, **kwargs):
self.trait = trait
super().__init__(*args, **kwargs)
@classmethod
def order_dependencies(cls):
return {'component_descriptor'}
@classmethod
def dependencies(cls):
return {'component_descriptor'}
def inject_steps(self):
if self.trait.set_dependency_version_script_container_image():
privilege_mode = PrivilegeMode.PRIVILEGED
else:
privilege_mode = PrivilegeMode.UNPRIVILEGED
# declare no dependencies --> run asap, but do not block other steps
self.update_component_deps_step = PipelineStep(
name='update_component_dependencies',
raw_dict={
'privilege_mode': privilege_mode,
},
is_synthetic=True,
pull_request_notification_policy=PullRequestNotificationPolicy.
NO_NOTIFICATION,
injecting_trait_name=self.name,
script_type=ScriptType.PYTHON3)
self.update_component_deps_step.add_input(
name=concourse.model.traits.component_descriptor.DIR_NAME,
variable_name=concourse.model.traits.component_descriptor.
ENV_VAR_NAME,
)
self.update_component_deps_step.set_timeout(duration_string='30m')
for name, value in self.trait.vars().items():
self.update_component_deps_step.variables()[name] = value
yield self.update_component_deps_step
def process_pipeline_args(self, pipeline_args: JobVariant):
# our step depends on dependendency descriptor step
component_descriptor_step = pipeline_args.step(
concourse.model.traits.component_descriptor.
DEFAULT_COMPONENT_DESCRIPTOR_STEP_NAME)
self.update_component_deps_step._add_dependency(
component_descriptor_step)
upstream_component_name = self.trait.upstream_component_name()
if upstream_component_name:
self.update_component_deps_step.variables(
)['UPSTREAM_COMPONENT_NAME'] = '"{cn}"'.format(
cn=upstream_component_name, )
class UpdateComponentDependenciesTraitTransformer(TraitTransformer):
name = 'update_component_deps'
def __init__(self, trait, *args, **kwargs):
self.trait = trait
super().__init__(*args, **kwargs)
@classmethod
def order_dependencies(cls):
return {'component_descriptor'}
@classmethod
def dependencies(cls):
return {'component_descriptor'}
def inject_steps(self):
# declare no dependencies --> run asap, but do not block other steps
self.update_component_deps_step = PipelineStep(
name='update_component_dependencies',
raw_dict={},
is_synthetic=True,
notification_policy=StepNotificationPolicy.NO_NOTIFICATION,
script_type=ScriptType.PYTHON3
)
self.update_component_deps_step.add_input(
name=concourse.model.traits.component_descriptor.DIR_NAME,
variable_name=concourse.model.traits.component_descriptor.ENV_VAR_NAME,
)
self.update_component_deps_step.set_timeout(duration_string='30m')
for name, value in self.trait.vars().items():
self.update_component_deps_step.variables()[name] = value
yield self.update_component_deps_step
def process_pipeline_args(self, pipeline_args: JobVariant):
# our step depends on dependendency descriptor step
component_descriptor_step = pipeline_args.step('component_descriptor')
self.update_component_deps_step._add_dependency(component_descriptor_step)
upstream_component_name = self.trait.upstream_component_name()
if upstream_component_name:
self.update_component_deps_step.variables()['UPSTREAM_COMPONENT_NAME'] = '"{cn}"'.format(
cn=upstream_component_name,
)
class ComponentDescriptorTraitTransformer(TraitTransformer):
name = 'component_descriptor'
def __init__(self, trait: ComponentDescriptorTrait, *args, **kwargs):
super().__init__(*args, **kwargs)
self.trait = not_none(trait)
def inject_steps(self):
self.descriptor_step = PipelineStep(
name=self.trait.step_name(),
raw_dict={},
is_synthetic=True,
notification_policy=StepNotificationPolicy.NO_NOTIFICATION,
injected_by_trait=self.name,
script_type=ScriptType.PYTHON3,
)
self.descriptor_step.add_output(
name=DIR_NAME,
variable_name=ENV_VAR_NAME,
)
self.descriptor_step.set_timeout(duration_string='45m')
yield self.descriptor_step
def process_pipeline_args(self, pipeline_args: 'JobVariant'):
if pipeline_args.has_step('release'):
release_step = pipeline_args.step('release')
release_step.add_input(
name=DIR_NAME,
variable_name=ENV_VAR_NAME,
)
if pipeline_args.has_trait('draft_release'):
draft_release_step = pipeline_args.step(
'create_draft_release_notes')
draft_release_step.add_input(
name=DIR_NAME,
variable_name=ENV_VAR_NAME,
)
# inject component_name if not configured
if not self.trait.raw.get('component_name'):
main_repo = pipeline_args.main_repository()
component_name = '/'.join((
main_repo.repo_hostname(),
main_repo.repo_path(),
))
self.trait.raw['component_name'] = component_name
# add configured (step-)inputs
for step_input in self.trait.inputs():
if not step_input.type == 'step':
raise NotImplementedError(step_input.type)
try:
step: PipelineStep = pipeline_args.step(step_input.step_name)
except KeyError as ke:
raise ValueError(
f'no such step: {step_input.step_name=}') from ke
self.descriptor_step._add_dependency(step)
if step_input.output_name:
output_name = step_input.output_name
else:
# choose only output if omitted
outputs = {
name: v
for name, v in step.outputs().items()
if not name == 'on_error_dir' # XXX hack hack hack
}
if len(outputs) < 1:
raise ValueError(f'{step.name=} does not have any outputs')
elif len(outputs) > 1:
raise ValueError(
f'{step.name=} has more than one output (need to tell step_name)'
)
output_name = next(outputs.keys().__iter__())
self.descriptor_step.add_input(
name=output_name,
variable_name=output_name,
)
@classmethod
def dependencies(cls):
return {'version'}
@classmethod
def order_dependencies(cls):
# dependency is required, as we need to patch the 'release' step
return {'release'}
