def make_workflow_rocrate(workflow_path,
wf_type,
include_files=[],
fetch_remote=False,
cwl=None,
diagram=None):
# Properties
# missing?
# input
# output
# programmingLanguage
# url
# version
# sdPublisher - current set to the person that provided the metadata,
# decision to change to the Workflow Hub itself - Done
# publisher - where it came came from, e.g. Galaxy, github, or WF Hub
# if uploaded - Done
# producer - to describe the Project or Team - Done
# creator - the creators / authors - Done
# maintainer - new recommended property to describe the uploader +
# additional people with manage rights - Done
# funder - example of cordis reference
# https://cordis.europa.eu/project/id/730976
# https://schema.org/FundingScheme linked to funder
# Examples at the bottom of https://schema.org/Grant - funding looks
# ideal but not currently legal
# Is needed to fulfill the OpenAire “Funding Reference” property
# datePublished - becomes an optional property, and we use the date a
# DOI was minted (this property is needed for dataCite) - Done
# creativeWorkStatus - Maturity level, to be added to BioSchemas - Done
# Identifier - can be DOI if this function is enabled in WorkflowHub - Done
# returns a complete ROCrate object corresponding to a Workflow template
# file
# wf_type: Galaxy, CWL, Nextflow, ...
# cwl: CWL/CWL-Abstract representation of the workflow.
# diagram: an image/graphical workflow representation.
# If a CWL/CWLAbstract file is provided, this is generated using cwltool
wf_crate = roc.ROCrate()
workflow_path = Path(workflow_path)
wf_file = wf_crate.add_workflow(str(workflow_path),
workflow_path.name,
fetch_remote=fetch_remote,
main=True,
lang=wf_type,
gen_cwl=(cwl is None))
# if the source is a remote URL then add https://schema.org/codeRepository
# property to it this can be checked by checking if the source is a URL
# instead of a local path
if 'url' in wf_file.properties():
wf_file['codeRepository'] = wf_file['url']
# add extra files
for file_entry in include_files:
wf_crate.add_file(file_entry)
return wf_crate
def make_workflow_rocrate(workflow_path,
wf_type,
include_files=[],
fetch_remote=False,
cwl=None,
diagram=None):
# Properties
# missing?
# input
# output
# programmingLanguage
# url
# version
# sdPublisher - current set to the person that provided the metadata,
# decision to change to the Workflow Hub itself - Done
# publisher - where it came came from, e.g. Galaxy, github, or WF Hub
# if uploaded - Done
# producer - to describe the Project or Team - Done
# creator - the creators / authors - Done
# maintainer - new recommended property to describe the uploader +
# additional people with manage rights - Done
# funder - example of cordis reference
# https://cordis.europa.eu/project/id/730976
# https://schema.org/FundingScheme linked to funder
# Examples at the bottom of https://schema.org/Grant - funding looks
# ideal but not currently legal
# Is needed to fulfill the OpenAire “Funding Reference” property
# datePublished - becomes an optional property, and we use the date a
# DOI was minted (this property is needed for dataCite) - Done
# creativeWorkStatus - Maturity level, to be added to BioSchemas - Done
# Identifier - can be DOI if this function is enabled in WorkflowHub - Done
# returns a complete ROCrate object corresponding to a Workflow template
# file
# wf_type: Galaxy, CWL, Nextflow, ...
# cwl: CWL/CWL-Abstract representation of the workflow.
# diagram: an image/graphical workflow representation.
# If a CWL/CWLAbstract file is provided, this is generated using cwltool
wf_crate = roc.ROCrate()
wf_path = Path(workflow_path)
# should this be added in a special path within the crate?
wf_file = Workflow(wf_crate, str(wf_path), wf_path.name)
wf_crate._add_data_entity(wf_file)
wf_crate.set_main_entity(wf_file)
if wf_type == 'CWL':
programming_language_entity = entity.Entity(
wf_crate,
'https://www.commonwl.org/v1.1/',
properties={
"@type": ["ComputerLanguage", "SoftwareApplication"],
'name': 'CWL',
'url': 'https://www.commonwl.org/v1.1/',
'version': '1.1'
})
if wf_type == 'Galaxy':
if not cwl:
# create cwl_abstract
with tempfile.NamedTemporaryFile(mode='w',
delete=False,
suffix=".cwl") as f:
with redirect_stdout(f):
get_cwl_interface.main(['1', workflow_path])
atexit.register(os.unlink, f.name)
abstract_wf_id = wf_path.with_suffix(".cwl").name
abstract_wf_file = Workflow(wf_crate, f.name, abstract_wf_id)
wf_crate._add_data_entity(abstract_wf_file)
wf_file["subjectOf"] = abstract_wf_file
programming_language_entity = entity.Entity(
wf_crate, 'https://galaxyproject.org/')
if programming_language_entity:
wf_file['programmingLanguage'] = programming_language_entity
# if the source is a remote URL then add https://schema.org/codeRepository
# property to it this can be checked by checking if the source is a URL
# instead of a local path
if 'url' in wf_file.properties():
wf_file['codeRepository'] = wf_file['url']
# add extra files
for file_entry in include_files:
wf_crate.add_file(file_entry)
return wf_crate
def createResearchObject(self, wf_url, input_files, execution_path,
wf_yaml):
""""
Create RO-crate from execution provenance.
:param wf_url: Remote workflow location.
:type wf_url: str
:param input_files: Dictionary of input files locations.
:type input_files: dict
:param execution_path: Working directory.
:type execution_path: str
:param wf_yaml: YAML filename.
:type wf_yaml: str
"""
try:
# Create RO-Crate
wf_crate = rocrate.ROCrate()
wf_file = wf_crate.add_workflow(wf_url,
fetch_remote=True,
main=True)
# Add url, codeRepository and isBasedOn to RO-crate
parsed_wf_url = parse.urlparse(wf_url)
wf_path = parsed_wf_url.path.split("/")
if parsed_wf_url.netloc == "raw.githubusercontent.com":
repoURL = None
repoTag = None
repoRelPath = None
if len(wf_path) >= 3:
repoGitPath = wf_path[:3]
repoURL = parse.urlunparse(
("https", "github.com", "/".join(repoGitPath), "", "",
""))
if len(wf_path) >= 4:
repoTag = wf_path[3]
if len(wf_path) >= 5:
repoRelPath = "/".join(wf_path[4:])
repoGit = repoURL + "/tree/" + repoTag + "/" + repoRelPath
wf_file.properties()['url'] = repoGit.replace("tree", "blob")
wf_file.properties()['codeRepository'] = os.path.dirname(
repoGit)
wf_crate.isBasedOn = os.path.dirname(repoGit)
else:
logger.error(
"FIXME: Unsupported http(s) GitHub repository {}".format(
parsed_wf_url))
# Add inputs provenance data to RO-crate
for in_id, in_value in input_files.items():
if isinstance(in_value, list):
for elem in in_value:
self.addInputToResearchObject(wf_crate, in_id, elem)
else:
self.addInputToResearchObject(wf_crate, in_id, in_value)
# Add outputs provenance data to RO-crate
# TODO
wf_crate_path = os.path.join(execution_path, self.provenance_path)
wf_crate.writeCrate(wf_crate_path)
# Add YAML file to RO-Crate
shutil.move(wf_yaml, self.provenance_path)
# Compress RO-crate to zip
shutil.make_archive(self.provenance_path, "zip", wf_crate_path)
shutil.rmtree(wf_crate_path)
except:
errstr = "Cannot create RO-Crate. See logs."
logger.error(errstr)
raise Exception(errstr)
def make_workflow_rocrate(workflow_path,
wf_type,
include_files=[],
fetch_remote=False,
cwl=None,
diagram=None):
# Properties
# missing?
# input
# output
# programmingLanguage
# url
#version
# sdPublisher - current set to the person that provided the metadata, decision to change to the Workflow Hub itself - Done
# publisher - where it came came from, e.g. Galaxy, github, or WF Hub if uploaded - Done
# producer - to describe the Project or Team Done
# creator - the creators/ authors Done
# maintainer - new recommended property to describe the uploader + additional people with manage rights Done
# funder - example of cordis reference - https://cordis.europa.eu/project/id/730976
# https://schema.org/FundingScheme linked to funder
# Examples at the bottom of https://schema.org/Grant - funding looks ideal but not currently legal
# Is needed to fulfill the OpenAire “Funding Reference” property
# datePublished - becomes an optional property, and we use the date a DOI was minted (this property is needed for dataCite) Done
# creativeWorkStatus - Maturity level, to be added to BioSchemas Done
# Identifier - can be DOI if this function is enabled in WorkflowHub Done
#returns a complete ROCrate object corresponding to a Workflow template file
# wf_type: Galaxy, CWL , Nextflow..
# cwl: CWL/CWL-Abstract representation of the workflow. If the
# diagram: an image/graphical workflow representation.
# If a CWL/CWLAbstract file is provided then this is generated using cwltool
#abs_path = os.path.abspath(workflow_path)
wf_crate = roc.ROCrate()
# add main workflow file
file_name = os.path.basename(workflow_path)
wf_file = wf_crate.add_file(
workflow_path,
file_name) # should I add it in a special path within the crate?
wf_crate.set_main_entity(wf_file)
if wf_type == 'CWL':
programming_language_entity = entity.Entity(
wf_crate,
'https://www.commonwl.org/v1.1/',
properties={
"@type": ["ComputerLanguage", "SoftwareApplication"],
'name': 'CWL',
'url': 'https://www.commonwl.org/v1.1/',
'version': '1.1'
})
if wf_type == 'Galaxy':
if not cwl:
#create cwl_abstract
with tempfile.NamedTemporaryFile(mode='w',
delete=False) as cwl_abstract_out:
with redirect_stdout(cwl_abstract_out):
get_cwl_interface.main(['1', workflow_path])
wf_file = wf_crate.add_file(
cwl_abstract_out.name,
'abstract_wf.cwl',
properties={
"@type": ["ComputerLanguage", "SoftwareApplication"]
})
programming_language_entity = entity.Entity(
wf_crate, 'https://galaxyproject.org/')
### SET PROPERTIES
# A contextual entity representing a SoftwareApplication or ComputerLanguage MUST have a name, url and version,
# which should indicate a known version the workflow/script was developed or tested with
if programming_language_entity:
wf_file['programmingLanguage'] = programming_language_entity
# based on ro-crate specification. for workflows: @type is an array with at least File and Workflow as values.
wf_type = wf_file['@type']
if not isinstance(wf_type, list):
wf_type = [wf_type]
if 'Workflow' not in wf_type:
wf_type.append('Workflow')
if 'SoftwareSourceCode' not in wf_type:
wf_type.append('SoftwareSourceCode')
wf_file['@type'] = wf_type
# if the source is a remote URL then add https://schema.org/codeRepository property to it
# this can be checked by checking if the source is a URL instead of a local path
if 'url' in wf_file.properties().keys():
wf_file['codeRepository'] = wf_file['url']
# add extra files
for file_entry in include_files:
wf_crate.add_file(file_entry)
return wf_crate
with open(argsFile) as f:
data = json.load(f)
wf_path = data['location']
files_list = []
# Workflow.TYPES=["File", "SoftwareSourceCode"]
#The commented command below created an additional html preview file in previous versions. In the current version
#the preview file is not created, therefore we call the commands below to include it.
# wf_crate = rocrate_api.make_workflow_rocrate(workflow_path=wf_path,wf_type="CWL",include_files=files_list)
cwl = None
wf_crate = roc.ROCrate(gen_preview=True)
workflow_path = Path(wf_path)
wf_file = wf_crate.add_workflow(str(workflow_path),
workflow_path.name,
fetch_remote=False,
main=True,
lang="CWL",
gen_cwl=(cwl is None))
# if the source is a remote URL then add https://schema.org/codeRepository
# property to it this can be checked by checking if the source is a URL
# instead of a local path
if 'url' in wf_file.properties():
wf_file['codeRepository'] = wf_file['url']
# add extra files