def createWorkflowStep(workflow,
position,
id,
type,
language="KNIME",
extension=None):
file_binding = cwlgen.CommandLineBinding()
# Individual step input
workflow_step = cwlgen.WorkflowStep(str(position), id + ".cwl")
workflow_step.inputs.append(
cwlgen.WorkflowStepInput("inputModule", "inputModule" + str(position)))
if (not "external" in type):
if (position == 1):
workflow_step.inputs.append(
cwlgen.WorkflowStepInput("potentialCases", "potentialCases"))
else:
workflow_step.inputs.append(
cwlgen.WorkflowStepInput("potentialCases",
source=str(position - 1) + "/output"))
# Individual step output
workflow_step.out.append(cwlgen.WorkflowStepOutput("output"))
workflow.steps = workflow.steps + [workflow_step]
# Overall workflow input
if (position == 1 and (not "external" in type)):
workflow_input = cwlgen.InputParameter(
"potentialCases",
param_type='File',
input_binding=file_binding,
doc="Input of potential cases for processing")
workflow.inputs.append(workflow_input)
workflow_input = cwlgen.InputParameter(
"inputModule" + str(position),
param_type='File',
input_binding=file_binding,
doc=language[0].upper() + language[1:] + " implementation unit")
workflow.inputs.append(workflow_input)
# Overall workflow output
if (extension):
workflow_output = cwlgen.WorkflowOutputParameter(
param_id='cases',
param_type="File",
output_source=str(position) + "/output",
output_binding=cwlgen.CommandOutputBinding(glob="*." + extension))
workflow.outputs.append(workflow_output)
return workflow
cwl_tool.inputs.append(no_of_cores)
# Path to fetch outputs from
output = cwlgen.workflow.WorkflowOutputParameter('output_file',
param_type='File[]',
output_source='west_tool/west_output_file',
label='Output File generated with west code')
cwl_tool.outputs.append(output)
# Toggle between pw and wstat for pw.in and wstat.in
workflow_west = cwlgen.workflow.WorkflowStep('west_tool',
'west.cwl',
doc='runs west.cwl',
scatter=['script_type','input_file'],
scatter_method='dotproduct')
# Add workflow step inputs
workflow_west_script_file = cwlgen.WorkflowStepInput('script_file','script_file')
workflow_west_input_file = cwlgen.WorkflowStepInput('input_file','input_file_array')
workflow_west_URLs = cwlgen.WorkflowStepInput('URLs','URLs')
workflow_west_script_type = cwlgen.WorkflowStepInput('script_type','script_type_array')
workflow_west_cores = cwlgen.WorkflowStepInput('no_of_cores','no_of_cores')
workflow_west.inputs.extend((workflow_west_script_file,workflow_west_input_file,workflow_west_URLs,workflow_west_script_type,workflow_west_cores))
workflow_west_output = cwlgen.WorkflowStepOutput('west_output_file')
workflow_west.out.append(workflow_west_output)
#Add first step
cwl_tool.steps.append(workflow_west)
#Export cwl
cwl_tool.export("west_workflow.cwl")
def _convert_composite(run, tmpdir, basedir, filename=None):
"""Converts a workflow made up of several steps."""
inputs = {}
arguments = {}
outputs = {}
consumed_outputs = set()
steps = []
input_index = 1
argument_index = 1
subprocesses, _ = _recurse_subprocesses(run, 1)
# preprocess to add dummy outputs in case of output directories
previous_output_dirs = defaultdict(list)
for _, subprocess in subprocesses:
for input in subprocess.inputs:
entity = input.consumes
key = (entity.commit.hexsha, entity.path)
if key not in previous_output_dirs:
continue
for previous_process in previous_output_dirs[key]:
previous_process.outputs.append(
CommandOutput(produces=entity, create_folder=False))
for output in subprocess.outputs:
entity = output.produces
if not isinstance(entity, Collection):
continue
for e in entity.entities:
if e.commit.hexsha != entity.commit.hexsha:
continue
key = (e.commit.hexsha, e.path)
previous_output_dirs[key].append(subprocess)
# Convert workflow steps
for i, subprocess in subprocesses:
tool, path = CWLConverter._convert_step(subprocess, tmpdir,
basedir)
step = WorkflowStep('step_{}'.format(i), path)
for input in subprocess.inputs:
input_path = input.consumes.path
sanitized_id = input.sanitized_id
if input.mapped_to:
sanitized_id = 'input_stdin'
if input_path in inputs:
# already used as top-level input elsewhere, reuse
step.inputs.append(
cwlgen.WorkflowStepInput(input.sanitized_id,
source=inputs[input_path]))
elif input_path in outputs:
# output of a previous step, refer to it
consumed_outputs.add(outputs[input_path][0])
step.inputs.append(
cwlgen.WorkflowStepInput(input.sanitized_id,
source='{}/{}'.format(
outputs[input_path][1],
outputs[input_path][0])))
else:
# input isn't output and doesn't exist yet, add new
inputs[input_path] = 'input_{}'.format(input_index)
step.inputs.append(
cwlgen.WorkflowStepInput(input.sanitized_id,
source=inputs[input_path]))
input_index += 1
for argument in subprocess.arguments:
argument_id = 'argument_{}'.format(argument_index)
arguments[argument_id] = argument.value
step.inputs.append(
cwlgen.WorkflowStepInput(argument.sanitized_id,
source=argument_id))
argument_index += 1
for output in subprocess.outputs:
sanitized_id = output.sanitized_id
if output.mapped_to:
sanitized_id = 'output_{}'.format(
output.mapped_to.stream_type)
outputs[output.produces.path] = (sanitized_id, step.id)
step.out.append(cwlgen.WorkflowStepOutput(sanitized_id))
steps.append(step)
workflow_object = cwlgen.Workflow(str(uuid4()), cwl_version='v1.0')
workflow_object.hints = []
workflow_object.requirements = []
# check types of paths and add as top level inputs/outputs
for path, id_ in inputs.items():
type_ = 'Directory' if os.path.isdir(path) else 'File'
workflow_object.inputs.append(
cwlgen.InputParameter(id_,
param_type=type_,
default={
'path':
os.path.abspath(
os.path.join(basedir, path)),
'class':
type_
}))
for id_, value in arguments.items():
value, type_ = _get_argument_type(value)
workflow_object.inputs.append(
cwlgen.InputParameter(id_, param_type=type_, default=value))
for index, (path, (id_, step_id)) in enumerate(outputs.items(), 1):
type_ = 'Directory' if os.path.isdir(path) else 'File'
workflow_object.outputs.append(
cwlgen.WorkflowOutputParameter('output_{}'.format(index),
output_source='{}/{}'.format(
step_id, id_),
param_type=type_))
workflow_object.steps.extend(steps)
if not filename:
filename = 'parent_{}.cwl'.format(uuid4())
path = os.path.join(tmpdir, filename)
workflow_object.export(path)
return workflow_object, path
input = cwlgen.InputParameter('message_array', param_type='string[]')
cwl_workflow.inputs.append(input)
# add outputs
output = cwlgen.WorkflowOutputParameter('output',
output_source='cat_2/output',
param_type='File')
cwl_workflow.outputs.append(output)
# add step 1 (echo)
step1 = cwlgen.WorkflowStep('echo', run='1st-tool.cwl', scatter='message')
#print(dir(step1))
# add inputs to step1
step1_input = cwlgen.WorkflowStepInput('message', source='message_array')
step1.inputs.append(step1_input)
# add outputs to step1
step1_output = cwlgen.WorkflowStepOutput('echo_out')
step1.out.append(step1_output)
# add step 2 (cat)
step2 = cwlgen.WorkflowStep('cat', run='./cat.cwl')
# add inputs to step2
step2_input = cwlgen.WorkflowStepInput('files', source='echo/echo_out')
step2.inputs.append(step2_input)
# add outputs to step2
step2_output = cwlgen.WorkflowStepOutput('output')
def main():
workflowName = sys.argv[1]
listOfNames = sys.argv[2:]
# Get information about workflow from the user
questions = [
inquirer.Text(
'label',
message=
"Please provide some information about the workflow you are creating"
)
]
answers = inquirer.prompt(questions)
# Initialize the tool we want to build
cwl_tool = cwlgen.Workflow(workflow_id=os.path.splitext(workflowName)[0],
label=answers['label'],
cwl_version='v1.1')
# Parse CLT tools which were provided on the command line to get a list of inputs and outputs
CLT_Inputs = {}
CLT_Outputs = {}
for i in listOfNames:
with open(os.path.abspath(i), 'r') as cwl_file:
cwl_dict = yaml.safe_load(cwl_file)
try:
if not isinstance(cwl_dict.get('inputs'),
dict) or not isinstance(
cwl_dict.get('outputs'), dict):
print(
"Your CWL files are not all of the same format. Please use ToolJig to make sure they all"
" have the same format.")
sys.exit()
else:
CLT_Inputs[i] = cwl_dict['inputs']
CLT_Outputs[i] = cwl_dict['outputs']
except AttributeError:
pass
# Declare first step of our Workflow (cwl_tool)
step = cwlgen.workflow.WorkflowStep(step_id=os.path.splitext(
listOfNames[0])[0],
run=listOfNames[0])
# Parse the inputs of the first file to save as Workflow inputs
workflowInputs = []
for item in CLT_Inputs[listOfNames[0]]:
input_Info = cwlgen.workflow.InputParameter(
param_id=item,
label=CLT_Inputs[listOfNames[0]][item].get('label'),
doc=CLT_Inputs[listOfNames[0]][item].get('doc'),
param_type=CLT_Inputs[listOfNames[0]][item].get('type'))
cwl_tool.inputs.append(input_Info)
idToShow = {
'ID': item,
'Label': CLT_Inputs[listOfNames[0]][item].get('label'),
'Type': CLT_Inputs[listOfNames[0]][item].get('type')
}
workflowInputs.append(idToShow)
step_inputs = cwlgen.WorkflowStepInput(input_id=item, source=item)
step.inputs.append(step_inputs)
# Get outputs of first step and append it to the whole workflow
for y in CLT_Outputs[listOfNames[0]]:
step_outputs = cwlgen.WorkflowStepOutput(output_id=y)
step.out.append(step_outputs)
cwl_tool.steps.append(step)
# LARGE LOOP: Make the steps and designate how inputs and outputs fit together -------------------------------------
for i in range(0, len(listOfNames)):
# Get outputs from "i" step that are of the type Directory or File
prevStepOutputs = CLT_Outputs[listOfNames[i]]
importantOutputs = []
for j in prevStepOutputs:
idToAdd = {'id': j}
idToAdd.update(prevStepOutputs[j])
importantOutputs.append(idToAdd)
# Get inputs from the "i+1" step that are of type Directory or File
nextInputs = []
importantInputs = []
try:
nextInputs = CLT_Inputs[listOfNames[i + 1]]
step = cwlgen.workflow.WorkflowStep(step_id=os.path.splitext(
listOfNames[i + 1])[0],
run=listOfNames[i + 1])
except:
# This is at the end, when the last outputs are workflow outputs, designate them as such
for x in importantOutputs:
output = cwlgen.workflow.WorkflowOutputParameter(
param_id=x.get('id'),
doc=x.get('doc'),
param_type=x.get('doc'),
output_source=os.path.splitext(listOfNames[i])[0])
cwl_tool.outputs.append(output)
for k in nextInputs:
if nextInputs[k]['type'] == 'File' or nextInputs[k][
'type'] == 'Directory':
idToAdd = {'id': k}
idToAdd.update(nextInputs[k])
importantInputs.append(idToAdd)
# Logic for matching inputs and outputs
if len(importantInputs) == len(importantOutputs) and len(
importantInputs) == 1:
step_inputs = cwlgen.WorkflowStepInput(
input_id=importantOutputs[0].get('id'),
source=listOfNames[i] + "/" + importantOutputs[0].get('id'))
step.inputs.append(step_inputs)
elif len(importantInputs) != len(importantOutputs) or len(
importantInputs) != 1 or len(importantOutputs) != 1:
for m in importantInputs:
# Declare variables ----------------------------------------------
first_index = 0
externalInputToName = 'It is an external input that has yet to be referenced'
previousOutput = 'It is the output of the workflow, but not the most recently previous step'
# Provide options ----------------------------------------------
print("Input ",
importantInputs.index(m) + 1, "/", len(importantInputs),
"of Command Line File ", i + 1, "/", len(listOfNames))
print(
"Your inputs and outputs don't match. Please specify where this input should be retrieved from:",
m)
print("")
options = ['It is the output of the previous step:']
for t in importantOutputs:
options.append(t)
first_index = first_index + 1
if cwl_tool.inputs:
options.append(
'It is an external input that already exists:')
for y in workflowInputs:
options.append(y)
captions = [
0, first_index + 1,
first_index + len(cwl_tool.inputs) + 2
]
else:
captions = [0, first_index + 1
] # This gets the first line and "other"
options.append('Other')
options.append(externalInputToName)
options.append(previousOutput)
selection = options[cutie.select(options,
caption_indices=captions)]
# Logic for selection ----------------------------------------------
if selection == externalInputToName:
questions = [
inquirer.Text(
'newID',
message="What is the ID of the new input?"),
inquirer.Text(
'newLabel',
message="What is the label of the new input?")
]
answers = inquirer.prompt(questions)
# add it as a master input
input_Info = cwlgen.workflow.InputParameter(
param_id=answers.get('newID'),
label=answers.get('newLabel'),
param_type=m.get('type'))
cwl_tool.inputs.append(input_Info)
idToShow = {
'ID': answers.get('newID'),
'Label': answers.get('newLabel'),
'Type': m.get('type')
}
workflowInputs.append(idToShow)
# add it as a step input
step_inputs = cwlgen.WorkflowStepInput(
input_id=answers.get('newID'),
source=answers.get('newID'))
elif selection == previousOutput:
print(
"\nPlease select which previous output corresponds to your input:"
)
listOfAllOutputs = []
for o in range(0, i + 1):
for output in CLT_Outputs.get(listOfNames[o]):
toAdd = {'ID': output, 'From step': listOfNames[o]}
toAdd.update(
CLT_Outputs.get(listOfNames[o])[output])
listOfAllOutputs.append(toAdd)
selection = listOfAllOutputs[cutie.select(
listOfAllOutputs)]
step_inputs = cwlgen.WorkflowStepInput(
input_id=selection['ID'],
source=selection['From step'] + "/" + selection['ID'])
elif selection in workflowInputs:
print(selection)
step_inputs = cwlgen.WorkflowStepInput(
input_id=selection.get('ID'),
source=selection.get('ID'))
else:
step_inputs = cwlgen.WorkflowStepInput(
input_id=m.get('id'),
source=listOfNames[i] + "/" + selection.get('id'))
step.inputs.append(step_inputs)
try:
for y in CLT_Outputs[listOfNames[i + 1]]:
step_outputs = cwlgen.WorkflowStepOutput(output_id=y)
step.out.append(step_outputs)
except:
pass
cwl_tool.steps.append(step)
cwl_tool.export(workflowName)
def translate_step(
step: StepNode,
is_nested_tool=False,
resource_overrides=Dict[str, str],
use_run_ref=True,
):
tool = step.tool
if use_run_ref:
run_ref = ("{tool}.cwl" if is_nested_tool else
"tools/{tool}.cwl").format(tool=tool.id())
else:
from janis_core.workflow.workflow import Workflow
has_resources_overrides = len(resource_overrides) > 0
if isinstance(tool, Workflow):
run_ref = CwlTranslator.translate_workflow_to_all_in_one(
tool, with_resource_overrides=has_resources_overrides)
else:
run_ref = CwlTranslator.translate_tool_internal(
tool, True, with_resource_overrides=has_resources_overrides)
cwlstep = cwlgen.WorkflowStep(
step_id=step.id(),
run=run_ref,
# label=step.step.label,
doc=step.doc,
scatter=None, # Filled by StepNode
scatter_method=None, # Filled by StepNode
)
cwlstep.out = [
cwlgen.WorkflowStepOutput(output_id=o.tag)
for o in step.tool.outputs()
]
ins = step.inputs()
for k in ins:
inp = ins[k]
if k not in step.sources:
if inp.input_type.optional or inp.default:
continue
else:
raise Exception(
f"Error when building connections for cwlstep '{step.id()}', "
f"could not find required connection: '{k}'")
edge = step.sources[k]
ss = edge.slashed_source()
link_merge = None
if (ss is not None and not isinstance(ss, list)
and isinstance(inp.input_type, Array)):
start = edge.source().start
outssval = start.outputs()
source_type = (first_value(outssval) if len(outssval) == 1 else
outssval[edge.source().stag]).output_type
# has scattered = isinstance(start, StepNode) and start.scatter
if not isinstance(source_type, Array) and not (isinstance(
start, StepNode) and start.scatter):
ss = [ss]
link_merge = "merge_nested"
d = cwlgen.WorkflowStepInput(
input_id=inp.tag,
source=ss,
link_merge=
link_merge, # this will need to change when edges have multiple source_map
value_from=None,
)
cwlstep.inputs.append(d)
for r in resource_overrides:
cwlstep.inputs.append(
cwlgen.WorkflowStepInput(input_id=r, source=resource_overrides[r]))
if step.scatter:
if len(step.scatter.fields) > 1:
Logger.info(
"Discovered more than one scatterable field on cwlstep '{step_id}', "
"deciding scatterMethod to be '{method}".format(
step_id=step.id(), method=step.scatter.method))
cwlstep.scatterMethod = step.scatter.method.cwl()
cwlstep.scatter = step.scatter.fields
return cwlstep
def create_workflow(drops, cwl_filename, buffer):
"""
Create a CWL workflow from a given Physical Graph Template
A CWL workflow consists of multiple files. A single file describing the
workflow, and multiple files each describing one step in the workflow. All
the files are combined into one zip file, so that a single file can be
downloaded by the user.
NOTE: CWL only supports workflow steps that are bash shell applications
Non-BashShellApp nodes are unable to be implemented in CWL
"""
# search the drops for non-BashShellApp drops,
# if found, the graph cannot be translated into CWL
for index, node in enumerate(drops):
dataType = node.get('dt', '')
if dataType not in SUPPORTED_CATEGORIES:
raise Exception('Node {0} has an unsupported category: {1}'.format(
index, dataType))
# create list for command line tool description files
step_files = []
# create the workflow
cwl_workflow = cwlgen.Workflow('', label='', doc='', cwl_version='v1.0')
# create files dictionary
files = {}
# look for input and output files in the pg_spec
for index, node in enumerate(drops):
command = node.get('command', None)
dataType = node.get('dt', None)
outputId = node.get('oid', None)
outputs = node.get('outputs', [])
if len(outputs) > 0:
files[outputs[0]] = "step" + str(index) + "/output_file_0"
# add steps to the workflow
for index, node in enumerate(drops):
dataType = node.get('dt', '')
if dataType == 'BashShellApp':
name = node.get('nm', '')
inputs = node.get('inputs', [])
outputs = node.get('outputs', [])
# create command line tool description
filename = "step" + str(index) + ".cwl"
contents = create_command_line_tool(node)
# add contents of command line tool description to list of step files
step_files.append({"filename": filename, "contents": contents})
# create step
step = cwlgen.WorkflowStep("step" + str(index), run=filename)
# add input to step
for index, input in enumerate(inputs):
step.inputs.append(
cwlgen.WorkflowStepInput('input_file_' + str(index),
source=files[input]))
# add output to step
for index, output in enumerate(outputs):
step.out.append(
cwlgen.WorkflowStepOutput('output_file_' + str(index)))
# add step to workflow
cwl_workflow.steps.append(step)
# put workflow and command line tool description files all together in a zip
zipObj = ZipFile(buffer, 'w')
for step_file in step_files:
zipObj.writestr(step_file["filename"], six.b(step_file["contents"]))
zipObj.writestr(cwl_filename, six.b(cwl_workflow.export_string()))
zipObj.close()
