def _initialize_metadata_inputs(dataset,
path_for_part,
tmp_dir,
kwds,
real_metadata_object=True):
filename_out = path_for_part("out")
filename_results_code = path_for_part("results")
filename_kwds = path_for_part("kwds")
filename_override_metadata = path_for_part("override")
open(
filename_out, 'wt+'
) # create the file on disk, so it cannot be reused by tempfile (unlikely, but possible)
# create the file on disk, so it cannot be reused by tempfile (unlikely, but possible)
json.dump((False, 'External set_meta() not called'),
open(filename_results_code, 'wt+'))
json.dump(kwds, open(filename_kwds, 'wt+'), ensure_ascii=True)
override_metadata = []
for meta_key, spec_value in dataset.metadata.spec.items():
if isinstance(
spec_value.param, FileParameter) and dataset.metadata.get(
meta_key, None) is not None:
if not real_metadata_object:
metadata_temp = MetadataTempFile()
metadata_temp.tmp_dir = tmp_dir
shutil.copy(
dataset.metadata.get(meta_key, None).file_name,
metadata_temp.file_name)
override_metadata.append((meta_key, metadata_temp.to_JSON()))
json.dump(override_metadata, open(filename_override_metadata, 'wt+'))
return filename_out, filename_results_code, filename_kwds, filename_override_metadata
def test_edit_metadata_files():
app = _mock_app(store_by="uuid")
sa_session = app.model.context
u = model.User(email="*****@*****.**", password="******")
h = model.History(name="Test History", user=u)
d1 = _create_datasets(sa_session, h, 1, extension="bam")[0]
sa_session.add_all((h, d1))
sa_session.flush()
index = NamedTemporaryFile("w")
index.write("cool bam index")
metadata_dict = {
"bam_index":
MetadataTempFile.from_JSON({
"kwds": {},
"filename": index.name
})
}
d1.metadata.from_JSON_dict(json_dict=metadata_dict)
assert d1.metadata.bam_index
assert isinstance(d1.metadata.bam_index, model.MetadataFile)
temp_directory = mkdtemp()
with store.DirectoryModelExportStore(
temp_directory, app=app, for_edit=True,
strip_metadata_files=False) as export_store:
export_store.add_dataset(d1)
import_history = model.History(name="Test History for Import", user=u)
sa_session.add(import_history)
sa_session.flush()
_perform_import_from_directory(temp_directory, app, u, import_history,
store.ImportOptions(allow_edit=True))
def _create_hda(model,
object_store,
history,
path,
visible=True,
include_metadata_file=False):
hda = HistoryDatasetAssociation(extension="bam",
create_dataset=True,
sa_session=model.context)
hda.visible = visible
model.context.add(hda)
model.context.flush([hda])
object_store.update_from_file(hda, file_name=path, create=True)
if include_metadata_file:
hda.metadata.from_JSON_dict(
json_dict={
"bam_index":
MetadataTempFile.from_JSON({
"kwds": {},
"filename": path
})
})
_check_metadata_file(hda)
hda.set_size()
history.add_dataset(hda)
hda.add_item_annotation(model.context, history.user, hda,
"annotation #%d" % hda.hid)
return hda
def cleanup_external_metadata(self, sa_session):
log.debug('Cleaning up external metadata files')
for metadata_files in sa_session.query(galaxy.model.Job).get(self.job_id).external_output_metadata:
# we need to confirm that any MetadataTempFile files were removed, if not we need to remove them
# can occur if the job was stopped before completion, but a MetadataTempFile is used in the set_meta
MetadataTempFile.cleanup_from_JSON_dict_filename(metadata_files.filename_out)
dataset_key = self._get_dataset_metadata_key(metadata_files.dataset)
for key, fname in [('filename_in', metadata_files.filename_in),
('filename_out', metadata_files.filename_out),
('filename_results_code', metadata_files.filename_results_code),
('filename_kwds', metadata_files.filename_kwds),
('filename_override_metadata', metadata_files.filename_override_metadata)]:
try:
os.remove(fname)
except Exception as e:
log.debug('Failed to cleanup external metadata file (%s) for %s: %s' % (key, dataset_key, e))
def _create_hda(model, object_store, history, path, visible=True, include_metadata_file=False):
hda = model.HistoryDatasetAssociation(extension="bam", create_dataset=True, sa_session=model.context)
hda.visible = visible
model.context.add(hda)
model.context.flush([hda])
object_store.update_from_file(hda, file_name=path, create=True)
if include_metadata_file:
hda.metadata.from_JSON_dict(json_dict={"bam_index": MetadataTempFile.from_JSON({"kwds": {}, "filename": path})})
_check_metadata_file(hda)
hda.set_size()
history.add_dataset(hda)
hda.add_item_annotation(model.context, history.user, hda, "annotation #%d" % hda.hid)
return hda
def test_nested_collection_attributes(self):
model = self.model
u = model.User(email="*****@*****.**", password="******")
h1 = model.History(name="History 1", user=u)
d1 = model.HistoryDatasetAssociation(extension="bam", history=h1, create_dataset=True, sa_session=model.session)
index = NamedTemporaryFile("w")
index.write("cool bam index")
index2 = NamedTemporaryFile("w")
index2.write("cool bam index 2")
metadata_dict = {"bam_index": MetadataTempFile.from_JSON({"kwds": {}, "filename": index.name}), "bam_csi_index": MetadataTempFile.from_JSON({"kwds": {}, "filename": index2.name})}
d1.metadata.from_JSON_dict(json_dict=metadata_dict)
assert d1.metadata.bam_index
assert d1.metadata.bam_csi_index
assert isinstance(d1.metadata.bam_index, model.MetadataFile)
assert isinstance(d1.metadata.bam_csi_index, model.MetadataFile)
d2 = model.HistoryDatasetAssociation(extension="txt", history=h1, create_dataset=True, sa_session=model.session)
c1 = model.DatasetCollection(collection_type='paired')
dce1 = model.DatasetCollectionElement(collection=c1, element=d1, element_identifier="forward", element_index=0)
dce2 = model.DatasetCollectionElement(collection=c1, element=d2, element_identifier="reverse", element_index=1)
c2 = model.DatasetCollection(collection_type="list:paired")
dce3 = model.DatasetCollectionElement(collection=c2, element=c1, element_identifier="inner_list", element_index=0)
c3 = model.DatasetCollection(collection_type="list:list")
c4 = model.DatasetCollection(collection_type="list:list:paired")
dce4 = model.DatasetCollectionElement(collection=c4, element=c2, element_identifier="outer_list", element_index=0)
model.session.add_all([d1, d2, c1, dce1, dce2, c2, dce3, c3, c4, dce4])
model.session.flush()
q = c2._get_nested_collection_attributes(element_attributes=('element_identifier',), hda_attributes=('extension',), dataset_attributes=('state',))
assert [(r.keys()) for r in q] == [['element_identifier_0', 'element_identifier_1', 'extension', 'state'], ['element_identifier_0', 'element_identifier_1', 'extension', 'state']]
assert q.all() == [('inner_list', 'forward', 'bam', 'new'), ('inner_list', 'reverse', 'txt', 'new')]
q = c2._get_nested_collection_attributes(return_entities=(model.HistoryDatasetAssociation,))
assert q.all() == [d1, d2]
q = c2._get_nested_collection_attributes(return_entities=(model.HistoryDatasetAssociation, model.Dataset))
assert q.all() == [(d1, d1.dataset), (d2, d2.dataset)]
# Assert properties that use _get_nested_collection_attributes return correct content
assert c2.dataset_instances == [d1, d2]
assert c2.dataset_elements == [dce1, dce2]
assert c2.dataset_action_tuples == []
assert c2.populated_optimized
assert c2.dataset_states_and_extensions_summary == ({'new'}, {'txt', 'bam'})
assert c2.element_identifiers_extensions_paths_and_metadata_files == [[('inner_list', 'forward'), 'bam', 'mock_dataset_14.dat', [('bai', 'mock_dataset_14.dat'), ('bam.csi', 'mock_dataset_14.dat')]], [('inner_list', 'reverse'), 'txt', 'mock_dataset_14.dat', []]]
assert c3.dataset_instances == []
assert c3.dataset_elements == []
assert c3.dataset_states_and_extensions_summary == (set(), set())
q = c4._get_nested_collection_attributes(element_attributes=('element_identifier',))
assert q.all() == [('outer_list', 'inner_list', 'forward'), ('outer_list', 'inner_list', 'reverse')]
assert c4.dataset_elements == [dce1, dce2]
assert c4.element_identifiers_extensions_and_paths == [(('outer_list', 'inner_list', 'forward'), 'bam', 'mock_dataset_14.dat'), (('outer_list', 'inner_list', 'reverse'), 'txt', 'mock_dataset_14.dat')]
def set_metadata_portable():
tool_job_working_directory = os.path.abspath(os.getcwd())
metadata_tmp_files_dir = os.path.join(tool_job_working_directory,
"metadata")
MetadataTempFile.tmp_dir = metadata_tmp_files_dir
metadata_params_path = os.path.join("metadata", "params.json")
try:
with open(metadata_params_path) as f:
metadata_params = json.load(f)
except OSError:
raise Exception(
f"Failed to find metadata/params.json from cwd [{tool_job_working_directory}]"
)
datatypes_config = metadata_params["datatypes_config"]
job_metadata = metadata_params["job_metadata"]
provided_metadata_style = metadata_params.get("provided_metadata_style")
max_metadata_value_size = metadata_params.get(
"max_metadata_value_size") or 0
outputs = metadata_params["outputs"]
datatypes_registry = validate_and_load_datatypes_config(datatypes_config)
tool_provided_metadata = load_job_metadata(job_metadata,
provided_metadata_style)
def set_meta(new_dataset_instance, file_dict):
set_meta_with_tool_provided(new_dataset_instance, file_dict,
set_meta_kwds, datatypes_registry,
max_metadata_value_size)
object_store_conf_path = os.path.join("metadata", "object_store_conf.json")
extended_metadata_collection = os.path.exists(object_store_conf_path)
object_store = None
job_context = None
version_string = ""
export_store = None
final_job_state = Job.states.OK
if extended_metadata_collection:
tool_dict = metadata_params["tool"]
stdio_exit_code_dicts, stdio_regex_dicts = tool_dict[
"stdio_exit_codes"], tool_dict["stdio_regexes"]
stdio_exit_codes = list(map(ToolStdioExitCode, stdio_exit_code_dicts))
stdio_regexes = list(map(ToolStdioRegex, stdio_regex_dicts))
with open(object_store_conf_path) as f:
config_dict = json.load(f)
assert config_dict is not None
object_store = build_object_store_from_config(None,
config_dict=config_dict)
Dataset.object_store = object_store
outputs_directory = os.path.join(tool_job_working_directory, "outputs")
if not os.path.exists(outputs_directory):
outputs_directory = tool_job_working_directory
# TODO: constants...
if os.path.exists(os.path.join(outputs_directory, "tool_stdout")):
with open(os.path.join(outputs_directory, "tool_stdout"),
"rb") as f:
tool_stdout = f.read()
with open(os.path.join(outputs_directory, "tool_stderr"),
"rb") as f:
tool_stderr = f.read()
elif os.path.exists(os.path.join(tool_job_working_directory,
"stdout")):
with open(os.path.join(tool_job_working_directory, "stdout"),
"rb") as f:
tool_stdout = f.read()
with open(os.path.join(tool_job_working_directory, "stderr"),
"rb") as f:
tool_stderr = f.read()
elif os.path.exists(os.path.join(outputs_directory, "stdout")):
# Puslar style output directory? Was this ever used - did this ever work?
with open(os.path.join(outputs_directory, "stdout"), "rb") as f:
tool_stdout = f.read()
with open(os.path.join(outputs_directory, "stderr"), "rb") as f:
tool_stderr = f.read()
else:
wdc = os.listdir(tool_job_working_directory)
odc = os.listdir(outputs_directory)
error_desc = "Failed to find tool_stdout or tool_stderr for this job, cannot collect metadata"
error_extra = f"Working dir contents [{wdc}], output directory contents [{odc}]"
log.warn(f"{error_desc}. {error_extra}")
raise Exception(error_desc)
job_id_tag = metadata_params["job_id_tag"]
exit_code_file = default_exit_code_file(".", job_id_tag)
tool_exit_code = read_exit_code_from(exit_code_file, job_id_tag)
check_output_detected_state, tool_stdout, tool_stderr, job_messages = check_output(
stdio_regexes, stdio_exit_codes, tool_stdout, tool_stderr,
tool_exit_code, job_id_tag)
if check_output_detected_state == DETECTED_JOB_STATE.OK and not tool_provided_metadata.has_failed_outputs(
):
final_job_state = Job.states.OK
else:
final_job_state = Job.states.ERROR
version_string = ""
if os.path.exists(COMMAND_VERSION_FILENAME):
version_string = open(COMMAND_VERSION_FILENAME).read()
expression_context = ExpressionContext(
dict(stdout=tool_stdout, stderr=tool_stderr))
# Load outputs.
export_store = store.DirectoryModelExportStore(
'metadata/outputs_populated',
serialize_dataset_objects=True,
for_edit=True,
strip_metadata_files=False,
serialize_jobs=False)
try:
import_model_store = store.imported_store_for_metadata(
'metadata/outputs_new', object_store=object_store)
except AssertionError:
# Remove in 21.09, this should only happen for jobs that started on <= 20.09 and finish now
import_model_store = None
job_context = SessionlessJobContext(
metadata_params,
tool_provided_metadata,
object_store,
export_store,
import_model_store,
os.path.join(tool_job_working_directory, "working"),
final_job_state=final_job_state,
)
unnamed_id_to_path = {}
for unnamed_output_dict in job_context.tool_provided_metadata.get_unnamed_outputs(
):
destination = unnamed_output_dict["destination"]
elements = unnamed_output_dict["elements"]
destination_type = destination["type"]
if destination_type == 'hdas':
for element in elements:
filename = element.get('filename')
if filename:
unnamed_id_to_path[element['object_id']] = os.path.join(
job_context.job_working_directory, filename)
for output_name, output_dict in outputs.items():
dataset_instance_id = output_dict["id"]
klass = getattr(
galaxy.model,
output_dict.get('model_class', 'HistoryDatasetAssociation'))
dataset = None
if import_model_store:
dataset = import_model_store.sa_session.query(klass).find(
dataset_instance_id)
if dataset is None:
# legacy check for jobs that started before 21.01, remove on 21.05
filename_in = os.path.join(f"metadata/metadata_in_{output_name}")
import pickle
dataset = pickle.load(open(filename_in,
'rb')) # load DatasetInstance
assert dataset is not None
filename_kwds = os.path.join(f"metadata/metadata_kwds_{output_name}")
filename_out = os.path.join(f"metadata/metadata_out_{output_name}")
filename_results_code = os.path.join(
f"metadata/metadata_results_{output_name}")
override_metadata = os.path.join(
f"metadata/metadata_override_{output_name}")
dataset_filename_override = output_dict["filename_override"]
# pre-20.05 this was a per job parameter and not a per dataset parameter, drop in 21.XX
legacy_object_store_store_by = metadata_params.get(
"object_store_store_by", "id")
# Same block as below...
set_meta_kwds = stringify_dictionary_keys(
json.load(open(filename_kwds))
) # load kwds; need to ensure our keywords are not unicode
try:
dataset.dataset.external_filename = unnamed_id_to_path.get(
dataset_instance_id, dataset_filename_override)
store_by = output_dict.get("object_store_store_by",
legacy_object_store_store_by)
extra_files_dir_name = f"dataset_{getattr(dataset.dataset, store_by)}_files"
files_path = os.path.abspath(
os.path.join(tool_job_working_directory, "working",
extra_files_dir_name))
dataset.dataset.external_extra_files_path = files_path
file_dict = tool_provided_metadata.get_dataset_meta(
output_name, dataset.dataset.id, dataset.dataset.uuid)
if 'ext' in file_dict:
dataset.extension = file_dict['ext']
# Metadata FileParameter types may not be writable on a cluster node, and are therefore temporarily substituted with MetadataTempFiles
override_metadata = json.load(open(override_metadata))
for metadata_name, metadata_file_override in override_metadata:
if MetadataTempFile.is_JSONified_value(metadata_file_override):
metadata_file_override = MetadataTempFile.from_JSON(
metadata_file_override)
setattr(dataset.metadata, metadata_name,
metadata_file_override)
if output_dict.get("validate", False):
set_validated_state(dataset)
if dataset_instance_id not in unnamed_id_to_path:
# We're going to run through set_metadata in collect_dynamic_outputs with more contextual metadata,
# so skip set_meta here.
set_meta(dataset, file_dict)
if extended_metadata_collection:
meta = tool_provided_metadata.get_dataset_meta(
output_name, dataset.dataset.id, dataset.dataset.uuid)
if meta:
context = ExpressionContext(meta, expression_context)
else:
context = expression_context
# Lazy and unattached
# if getattr(dataset, "hidden_beneath_collection_instance", None):
# dataset.visible = False
dataset.blurb = 'done'
dataset.peek = 'no peek'
dataset.info = (dataset.info or '')
if context['stdout'].strip():
# Ensure white space between entries
dataset.info = f"{dataset.info.rstrip()}\n{context['stdout'].strip()}"
if context['stderr'].strip():
# Ensure white space between entries
dataset.info = f"{dataset.info.rstrip()}\n{context['stderr'].strip()}"
dataset.tool_version = version_string
dataset.set_size()
if 'uuid' in context:
dataset.dataset.uuid = context['uuid']
if dataset_filename_override and dataset_filename_override != dataset.file_name:
# This has to be a job with outputs_to_working_directory set.
# We update the object store with the created output file.
object_store.update_from_file(
dataset.dataset,
file_name=dataset_filename_override,
create=True)
collect_extra_files(object_store, dataset, ".")
if Job.states.ERROR == final_job_state:
dataset.blurb = "error"
dataset.mark_unhidden()
else:
# If the tool was expected to set the extension, attempt to retrieve it
if dataset.ext == 'auto':
dataset.extension = context.get('ext', 'data')
dataset.init_meta(copy_from=dataset)
# This has already been done:
# else:
# self.external_output_metadata.load_metadata(dataset, output_name, self.sa_session, working_directory=self.working_directory, remote_metadata_directory=remote_metadata_directory)
line_count = context.get('line_count', None)
try:
# Certain datatype's set_peek methods contain a line_count argument
dataset.set_peek(line_count=line_count)
except TypeError:
# ... and others don't
dataset.set_peek()
for context_key in TOOL_PROVIDED_JOB_METADATA_KEYS:
if context_key in context:
context_value = context[context_key]
setattr(dataset, context_key, context_value)
# We never want to persist the external_filename.
dataset.dataset.external_filename = None
export_store.add_dataset(dataset)
else:
dataset.metadata.to_JSON_dict(
filename_out) # write out results of set_meta
json.dump((True, 'Metadata has been set successfully'),
open(filename_results_code,
'wt+')) # setting metadata has succeeded
except Exception:
json.dump((False, traceback.format_exc()),
open(filename_results_code,
'wt+')) # setting metadata has failed somehow
if extended_metadata_collection:
# discover extra outputs...
output_collections = {}
for name, output_collection in metadata_params[
"output_collections"].items():
output_collections[name] = import_model_store.sa_session.query(
HistoryDatasetCollectionAssociation).find(
output_collection["id"])
outputs = {}
for name, output in metadata_params["outputs"].items():
klass = getattr(
galaxy.model,
output.get('model_class', 'HistoryDatasetAssociation'))
outputs[name] = import_model_store.sa_session.query(klass).find(
output["id"])
input_ext = json.loads(metadata_params["job_params"].get(
"__input_ext", '"data"'))
collect_primary_datasets(
job_context,
outputs,
input_ext=input_ext,
)
collect_dynamic_outputs(job_context, output_collections)
if export_store:
export_store._finalize()
write_job_metadata(tool_job_working_directory, job_metadata, set_meta,
tool_provided_metadata)
def set_metadata_portable():
tool_job_working_directory = os.path.abspath(os.getcwd())
metadata_tmp_files_dir = os.path.join(tool_job_working_directory, "metadata")
MetadataTempFile.tmp_dir = metadata_tmp_files_dir
metadata_params = get_metadata_params(tool_job_working_directory)
datatypes_config = metadata_params["datatypes_config"]
job_metadata = metadata_params["job_metadata"]
provided_metadata_style = metadata_params.get("provided_metadata_style")
max_metadata_value_size = metadata_params.get("max_metadata_value_size") or 0
max_discovered_files = metadata_params.get("max_discovered_files")
outputs = metadata_params["outputs"]
datatypes_registry = validate_and_load_datatypes_config(datatypes_config)
tool_provided_metadata = load_job_metadata(job_metadata, provided_metadata_style)
def set_meta(new_dataset_instance, file_dict):
set_meta_with_tool_provided(new_dataset_instance, file_dict, set_meta_kwds, datatypes_registry, max_metadata_value_size)
try:
object_store = get_object_store(tool_job_working_directory=tool_job_working_directory)
except (FileNotFoundError, AssertionError):
object_store = None
extended_metadata_collection = bool(object_store)
job_context = None
version_string = None
export_store = None
final_job_state = Job.states.OK
job_messages = []
if extended_metadata_collection:
tool_dict = metadata_params["tool"]
stdio_exit_code_dicts, stdio_regex_dicts = tool_dict["stdio_exit_codes"], tool_dict["stdio_regexes"]
stdio_exit_codes = list(map(ToolStdioExitCode, stdio_exit_code_dicts))
stdio_regexes = list(map(ToolStdioRegex, stdio_regex_dicts))
outputs_directory = os.path.join(tool_job_working_directory, "outputs")
if not os.path.exists(outputs_directory):
outputs_directory = tool_job_working_directory
# TODO: constants...
locations = [
(outputs_directory, 'tool_'),
(tool_job_working_directory, ''),
(outputs_directory, ''), # # Pulsar style output directory? Was this ever used - did this ever work?
]
for directory, prefix in locations:
if os.path.exists(os.path.join(directory, f"{prefix}stdout")):
with open(os.path.join(directory, f"{prefix}stdout"), 'rb') as f:
tool_stdout = f.read(MAX_STDIO_READ_BYTES)
with open(os.path.join(directory, f"{prefix}stderr"), 'rb') as f:
tool_stderr = f.read(MAX_STDIO_READ_BYTES)
break
else:
if os.path.exists(os.path.join(tool_job_working_directory, 'task_0')):
# We have a task splitting job
tool_stdout = b''
tool_stderr = b''
paths = Path(tool_job_working_directory).glob('task_*')
for path in paths:
with open(path / 'outputs' / 'tool_stdout', 'rb') as f:
task_stdout = f.read(MAX_STDIO_READ_BYTES)
if task_stdout:
tool_stdout = b"%s[%s stdout]\n%s\n" % (tool_stdout, path.name.encode(), task_stdout)
with open(path / 'outputs' / 'tool_stderr', 'rb') as f:
task_stderr = f.read(MAX_STDIO_READ_BYTES)
if task_stderr:
tool_stderr = b"%s[%s stdout]\n%s\n" % (tool_stderr, path.name.encode(), task_stderr)
else:
wdc = os.listdir(tool_job_working_directory)
odc = os.listdir(outputs_directory)
error_desc = "Failed to find tool_stdout or tool_stderr for this job, cannot collect metadata"
error_extra = f"Working dir contents [{wdc}], output directory contents [{odc}]"
log.warn(f"{error_desc}. {error_extra}")
raise Exception(error_desc)
job_id_tag = metadata_params["job_id_tag"]
exit_code_file = default_exit_code_file(".", job_id_tag)
tool_exit_code = read_exit_code_from(exit_code_file, job_id_tag)
check_output_detected_state, tool_stdout, tool_stderr, job_messages = check_output(stdio_regexes, stdio_exit_codes, tool_stdout, tool_stderr, tool_exit_code, job_id_tag)
if check_output_detected_state == DETECTED_JOB_STATE.OK and not tool_provided_metadata.has_failed_outputs():
final_job_state = Job.states.OK
else:
final_job_state = Job.states.ERROR
version_string_path = os.path.join('outputs', COMMAND_VERSION_FILENAME)
version_string = collect_shrinked_content_from_path(version_string_path)
expression_context = ExpressionContext(dict(stdout=tool_stdout[:255], stderr=tool_stderr[:255]))
# Load outputs.
export_store = store.DirectoryModelExportStore('metadata/outputs_populated', serialize_dataset_objects=True, for_edit=True, strip_metadata_files=False, serialize_jobs=True)
try:
import_model_store = store.imported_store_for_metadata('metadata/outputs_new', object_store=object_store)
except AssertionError:
# Remove in 21.09, this should only happen for jobs that started on <= 20.09 and finish now
import_model_store = None
tool_script_file = os.path.join(tool_job_working_directory, 'tool_script.sh')
job = None
if import_model_store and export_store:
job = next(iter(import_model_store.sa_session.objects[Job].values()))
job_context = SessionlessJobContext(
metadata_params,
tool_provided_metadata,
object_store,
export_store,
import_model_store,
os.path.join(tool_job_working_directory, "working"),
final_job_state=final_job_state,
max_discovered_files=max_discovered_files,
)
if extended_metadata_collection:
# discover extra outputs...
output_collections = {}
for name, output_collection in metadata_params["output_collections"].items():
# TODO: remove HistoryDatasetCollectionAssociation fallback on 22.01, model_class used to not be serialized prior to 21.09
model_class = output_collection.get('model_class', 'HistoryDatasetCollectionAssociation')
collection = import_model_store.sa_session.query(getattr(galaxy.model, model_class)).find(output_collection["id"])
output_collections[name] = collection
output_instances = {}
for name, output in metadata_params["outputs"].items():
klass = getattr(galaxy.model, output.get('model_class', 'HistoryDatasetAssociation'))
output_instances[name] = import_model_store.sa_session.query(klass).find(output["id"])
input_ext = json.loads(metadata_params["job_params"].get("__input_ext") or '"data"')
try:
collect_primary_datasets(
job_context,
output_instances,
input_ext=input_ext,
)
collect_dynamic_outputs(job_context, output_collections)
except MaxDiscoveredFilesExceededError as e:
final_job_state = Job.states.ERROR
job_messages.append(str(e))
if job:
job.job_messages = job_messages
job.state = final_job_state
if os.path.exists(tool_script_file):
with open(tool_script_file) as command_fh:
command_line_lines = []
for i, line in enumerate(command_fh):
if i == 0 and line.endswith('COMMAND_VERSION 2>&1;'):
# Don't record version command as part of command line
continue
command_line_lines.append(line)
job.command_line = "".join(command_line_lines).strip()
export_store.export_job(job, include_job_data=False)
unnamed_id_to_path = {}
for unnamed_output_dict in job_context.tool_provided_metadata.get_unnamed_outputs():
destination = unnamed_output_dict["destination"]
elements = unnamed_output_dict["elements"]
destination_type = destination["type"]
if destination_type == 'hdas':
for element in elements:
filename = element.get('filename')
object_id = element.get('object_id')
if filename and object_id:
unnamed_id_to_path[object_id] = os.path.join(job_context.job_working_directory, filename)
for output_name, output_dict in outputs.items():
dataset_instance_id = output_dict["id"]
klass = getattr(galaxy.model, output_dict.get('model_class', 'HistoryDatasetAssociation'))
dataset = None
if import_model_store:
dataset = import_model_store.sa_session.query(klass).find(dataset_instance_id)
if dataset is None:
# legacy check for jobs that started before 21.01, remove on 21.05
filename_in = os.path.join(f"metadata/metadata_in_{output_name}")
import pickle
dataset = pickle.load(open(filename_in, 'rb')) # load DatasetInstance
assert dataset is not None
filename_kwds = os.path.join(f"metadata/metadata_kwds_{output_name}")
filename_out = os.path.join(f"metadata/metadata_out_{output_name}")
filename_results_code = os.path.join(f"metadata/metadata_results_{output_name}")
override_metadata = os.path.join(f"metadata/metadata_override_{output_name}")
dataset_filename_override = output_dict["filename_override"]
# pre-20.05 this was a per job parameter and not a per dataset parameter, drop in 21.XX
legacy_object_store_store_by = metadata_params.get("object_store_store_by", "id")
# Same block as below...
set_meta_kwds = stringify_dictionary_keys(json.load(open(filename_kwds))) # load kwds; need to ensure our keywords are not unicode
try:
external_filename = unnamed_id_to_path.get(dataset_instance_id, dataset_filename_override)
if not os.path.exists(external_filename):
matches = glob.glob(external_filename)
assert len(matches) == 1, f"More than one file matched by output glob '{external_filename}'"
external_filename = matches[0]
assert safe_contains(tool_job_working_directory, external_filename), f"Cannot collect output '{external_filename}' from outside of working directory"
created_from_basename = os.path.relpath(external_filename, os.path.join(tool_job_working_directory, 'working'))
dataset.dataset.created_from_basename = created_from_basename
# override filename if we're dealing with outputs to working directory and dataset is not linked to
link_data_only = metadata_params.get("link_data_only")
if not link_data_only:
# Only set external filename if we're dealing with files in job working directory.
# Fixes link_data_only uploads
dataset.dataset.external_filename = external_filename
store_by = output_dict.get("object_store_store_by", legacy_object_store_store_by)
extra_files_dir_name = f"dataset_{getattr(dataset.dataset, store_by)}_files"
files_path = os.path.abspath(os.path.join(tool_job_working_directory, "working", extra_files_dir_name))
dataset.dataset.external_extra_files_path = files_path
file_dict = tool_provided_metadata.get_dataset_meta(output_name, dataset.dataset.id, dataset.dataset.uuid)
if 'ext' in file_dict:
dataset.extension = file_dict['ext']
# Metadata FileParameter types may not be writable on a cluster node, and are therefore temporarily substituted with MetadataTempFiles
override_metadata = json.load(open(override_metadata))
for metadata_name, metadata_file_override in override_metadata:
if MetadataTempFile.is_JSONified_value(metadata_file_override):
metadata_file_override = MetadataTempFile.from_JSON(metadata_file_override)
setattr(dataset.metadata, metadata_name, metadata_file_override)
if output_dict.get("validate", False):
set_validated_state(dataset)
if dataset_instance_id not in unnamed_id_to_path:
# We're going to run through set_metadata in collect_dynamic_outputs with more contextual metadata,
# so skip set_meta here.
set_meta(dataset, file_dict)
if extended_metadata_collection:
collect_extra_files(object_store, dataset, ".")
dataset.state = dataset.dataset.state = final_job_state
if extended_metadata_collection:
if not link_data_only and os.path.getsize(external_filename):
# Here we might be updating a disk based objectstore when outputs_to_working_directory is used,
# or a remote object store from its cache path.
object_store.update_from_file(dataset.dataset, file_name=external_filename, create=True)
# TODO: merge expression_context into tool_provided_metadata so we don't have to special case this (here and in _finish_dataset)
meta = tool_provided_metadata.get_dataset_meta(output_name, dataset.dataset.id, dataset.dataset.uuid)
if meta:
context = ExpressionContext(meta, expression_context)
else:
context = expression_context
dataset.blurb = 'done'
dataset.peek = 'no peek'
dataset.info = (dataset.info or '')
if context['stdout'].strip():
# Ensure white space between entries
dataset.info = f"{dataset.info.rstrip()}\n{context['stdout'].strip()}"
if context['stderr'].strip():
# Ensure white space between entries
dataset.info = f"{dataset.info.rstrip()}\n{context['stderr'].strip()}"
dataset.tool_version = version_string
if 'uuid' in context:
dataset.dataset.uuid = context['uuid']
if not final_job_state == Job.states.ERROR:
line_count = context.get('line_count', None)
try:
# Certain datatype's set_peek methods contain a line_count argument
dataset.set_peek(line_count=line_count)
except TypeError:
# ... and others don't
dataset.set_peek()
for context_key in TOOL_PROVIDED_JOB_METADATA_KEYS:
if context_key in context:
context_value = context[context_key]
setattr(dataset, context_key, context_value)
# We only want to persist the external_filename if the dataset has been linked in.
if not link_data_only:
dataset.dataset.external_filename = None
dataset.dataset.extra_files_path = None
export_store.add_dataset(dataset)
else:
dataset.metadata.to_JSON_dict(filename_out) # write out results of set_meta
json.dump((True, 'Metadata has been set successfully'), open(filename_results_code, 'wt+')) # setting metadata has succeeded
except Exception:
json.dump((False, traceback.format_exc()), open(filename_results_code, 'wt+')) # setting metadata has failed somehow
if export_store:
export_store._finalize()
write_job_metadata(tool_job_working_directory, job_metadata, set_meta, tool_provided_metadata)