Merge branch '10172-crunch2-container-output' closes #10172
[arvados.git] / sdk / cwl / arvados_cwl / arvworkflow.py
1 import os
2 import json
3 import copy
4 import logging
5
6 from cwltool.pack import pack
7 from cwltool.load_tool import fetch_document
8 from cwltool.process import shortname
9 from cwltool.workflow import Workflow, WorkflowException
10 from cwltool.pathmapper import adjustFileObjs, adjustDirObjs
11
12 import ruamel.yaml as yaml
13
14 from .runner import upload_docker, upload_dependencies, trim_listing
15 from .arvtool import ArvadosCommandTool
16 from .perf import Perf
17
18 logger = logging.getLogger('arvados.cwl-runner')
19 metrics = logging.getLogger('arvados.cwl-runner.metrics')
20
21 def upload_workflow(arvRunner, tool, job_order, project_uuid, update_uuid):
22     upload_docker(arvRunner, tool)
23
24     document_loader, workflowobj, uri = (tool.doc_loader, tool.doc_loader.fetch(tool.tool["id"]), tool.tool["id"])
25
26     packed = pack(document_loader, workflowobj, uri, tool.metadata)
27
28     adjustDirObjs(job_order, trim_listing)
29
30     main = [p for p in packed["$graph"] if p["id"] == "#main"][0]
31     for inp in main["inputs"]:
32         sn = shortname(inp["id"])
33         if sn in job_order:
34             inp["default"] = job_order[sn]
35
36     name = os.path.basename(tool.tool["id"])
37     upload_dependencies(arvRunner, name, document_loader,
38                         packed, uri, False)
39
40     body = {
41         "workflow": {
42             "owner_uuid": project_uuid,
43             "name": tool.tool.get("label", name),
44             "description": tool.tool.get("doc", ""),
45             "definition":yaml.safe_dump(packed)
46         }}
47
48     if update_uuid:
49         return arvRunner.api.workflows().update(uuid=update_uuid, body=body).execute(num_retries=arvRunner.num_retries)["uuid"]
50     else:
51         return arvRunner.api.workflows().create(body=body).execute(num_retries=arvRunner.num_retries)["uuid"]
52
53 class ArvadosWorkflow(Workflow):
54     """Wrap cwltool Workflow to override selected methods."""
55
56     def __init__(self, arvrunner, toolpath_object, **kwargs):
57         super(ArvadosWorkflow, self).__init__(toolpath_object, **kwargs)
58         self.arvrunner = arvrunner
59         self.work_api = kwargs["work_api"]
60
61     def job(self, joborder, output_callback, **kwargs):
62         kwargs["work_api"] = self.work_api
63         req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
64         if req:
65             document_loader, workflowobj, uri = (self.doc_loader, self.doc_loader.fetch(self.tool["id"]), self.tool["id"])
66
67             with Perf(metrics, "subworkflow upload_deps"):
68                 workflowobj["requirements"] = self.requirements + workflowobj.get("requirements", [])
69                 workflowobj["hints"] = self.hints + workflowobj.get("hints", [])
70                 packed = pack(document_loader, workflowobj, uri, self.metadata)
71
72                 upload_dependencies(self.arvrunner,
73                                     kwargs.get("name", ""),
74                                     document_loader,
75                                     packed,
76                                     uri,
77                                     False)
78
79                 upload_dependencies(self.arvrunner,
80                                     os.path.basename(joborder.get("id", "#")),
81                                     document_loader,
82                                     joborder,
83                                     joborder.get("id", "#"),
84                                     False)
85
86             with Perf(metrics, "subworkflow adjust"):
87                 joborder_keepmount = copy.deepcopy(joborder)
88
89                 def keepmount(obj):
90                     if obj["location"].startswith("keep:"):
91                         obj["location"] = "/keep/" + obj["location"][5:]
92                         if "listing" in obj:
93                             del obj["listing"]
94                     elif obj["location"].startswith("_:"):
95                         del obj["location"]
96                     else:
97                         raise WorkflowException("Location is not a keep reference or a literal: '%s'" % obj["location"])
98
99                 adjustFileObjs(joborder_keepmount, keepmount)
100                 adjustDirObjs(joborder_keepmount, keepmount)
101                 adjustFileObjs(packed, keepmount)
102                 adjustDirObjs(packed, keepmount)
103
104             wf_runner = {
105                 "class": "CommandLineTool",
106                 "baseCommand": "cwltool",
107                 "inputs": self.tool["inputs"],
108                 "outputs": self.tool["outputs"],
109                 "stdout": "cwl.output.json",
110                 "requirements": workflowobj["requirements"]+[
111                     {
112                     "class": "InitialWorkDirRequirement",
113                     "listing": [{
114                             "entryname": "workflow.cwl",
115                             "entry": yaml.safe_dump(packed).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
116                         }, {
117                             "entryname": "cwl.input.yml",
118                             "entry": yaml.safe_dump(joborder_keepmount).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
119                         }]
120                 }],
121                 "hints": workflowobj["hints"],
122                 "arguments": ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl#main", "cwl.input.yml"]
123             }
124             kwargs["loader"] = self.doc_loader
125             kwargs["avsc_names"] = self.doc_schema
126             return ArvadosCommandTool(self.arvrunner, wf_runner, **kwargs).job(joborder, output_callback, **kwargs)
127         else:
128             return super(ArvadosWorkflow, self).job(joborder, output_callback, **kwargs)