9e70a6e66f2b36f5393cd2f32039a38ec2489dd4
[arvados.git] / sdk / cwl / arvados_cwl / arvworkflow.py
1 import os
2 import json
3 import copy
4 import logging
5
6 from schema_salad.sourceline import SourceLine, cmap
7
8 from cwltool.pack import pack
9 from cwltool.load_tool import fetch_document
10 from cwltool.process import shortname
11 from cwltool.workflow import Workflow, WorkflowException
12 from cwltool.pathmapper import adjustFileObjs, adjustDirObjs
13
14 import ruamel.yaml as yaml
15
16 from .runner import upload_docker, upload_dependencies, trim_listing
17 from .arvtool import ArvadosCommandTool
18 from .perf import Perf
19
20 logger = logging.getLogger('arvados.cwl-runner')
21 metrics = logging.getLogger('arvados.cwl-runner.metrics')
22
23 def upload_workflow(arvRunner, tool, job_order, project_uuid, uuid=None,
24                     submit_runner_ram=0, name=None):
25     upload_docker(arvRunner, tool)
26
27     document_loader, workflowobj, uri = (tool.doc_loader, tool.doc_loader.fetch(tool.tool["id"]), tool.tool["id"])
28
29     packed = pack(document_loader, workflowobj, uri, tool.metadata)
30
31     adjustDirObjs(job_order, trim_listing)
32
33     main = [p for p in packed["$graph"] if p["id"] == "#main"][0]
34     for inp in main["inputs"]:
35         sn = shortname(inp["id"])
36         if sn in job_order:
37             inp["default"] = job_order[sn]
38
39     if not name:
40         name = tool.tool.get("label", os.path.basename(tool.tool["id"]))
41
42     upload_dependencies(arvRunner, name, document_loader,
43                         packed, uri, False)
44
45     # TODO nowhere for submit_runner_ram to go.
46
47     body = {
48         "workflow": {
49             "name": name,
50             "description": tool.tool.get("doc", ""),
51             "definition":yaml.round_trip_dump(packed)
52         }}
53     if project_uuid:
54         body["workflow"]["owner_uuid"] = project_uuid
55
56     if uuid:
57         call = arvRunner.api.workflows().update(uuid=uuid, body=body)
58     else:
59         call = arvRunner.api.workflows().create(body=body)
60     return call.execute(num_retries=arvRunner.num_retries)["uuid"]
61
62 class ArvadosWorkflow(Workflow):
63     """Wrap cwltool Workflow to override selected methods."""
64
65     def __init__(self, arvrunner, toolpath_object, **kwargs):
66         super(ArvadosWorkflow, self).__init__(toolpath_object, **kwargs)
67         self.arvrunner = arvrunner
68         self.work_api = kwargs["work_api"]
69
70     def job(self, joborder, output_callback, **kwargs):
71         kwargs["work_api"] = self.work_api
72         req, _ = self.get_requirement("http://arvados.org/cwl#RunInSingleContainer")
73         if req:
74             with SourceLine(self.tool, None, WorkflowException):
75                 if "id" not in self.tool:
76                     raise WorkflowException("%s object must have 'id'" % (self.tool["class"]))
77             document_loader, workflowobj, uri = (self.doc_loader, self.doc_loader.fetch(self.tool["id"]), self.tool["id"])
78
79             with Perf(metrics, "subworkflow upload_deps"):
80                 workflowobj["requirements"] = self.requirements + workflowobj.get("requirements", [])
81                 workflowobj["hints"] = self.hints + workflowobj.get("hints", [])
82                 packed = pack(document_loader, workflowobj, uri, self.metadata)
83
84                 upload_dependencies(self.arvrunner,
85                                     kwargs.get("name", ""),
86                                     document_loader,
87                                     packed,
88                                     uri,
89                                     False)
90
91                 upload_dependencies(self.arvrunner,
92                                     os.path.basename(joborder.get("id", "#")),
93                                     document_loader,
94                                     joborder,
95                                     joborder.get("id", "#"),
96                                     False)
97
98             with Perf(metrics, "subworkflow adjust"):
99                 joborder_keepmount = copy.deepcopy(joborder)
100
101                 def keepmount(obj):
102                     with SourceLine(obj, None, WorkflowException):
103                         if "location" not in obj:
104                             raise WorkflowException("%s object is missing required 'location' field: %s" % (obj["class"], obj))
105                     with SourceLine(obj, "location", WorkflowException):
106                         if obj["location"].startswith("keep:"):
107                             obj["location"] = "/keep/" + obj["location"][5:]
108                             if "listing" in obj:
109                                 del obj["listing"]
110                         elif obj["location"].startswith("_:"):
111                             del obj["location"]
112                         else:
113                             raise WorkflowException("Location is not a keep reference or a literal: '%s'" % obj["location"])
114
115                 adjustFileObjs(joborder_keepmount, keepmount)
116                 adjustDirObjs(joborder_keepmount, keepmount)
117                 adjustFileObjs(packed, keepmount)
118                 adjustDirObjs(packed, keepmount)
119
120             wf_runner = cmap({
121                 "class": "CommandLineTool",
122                 "baseCommand": "cwltool",
123                 "inputs": self.tool["inputs"],
124                 "outputs": self.tool["outputs"],
125                 "stdout": "cwl.output.json",
126                 "requirements": workflowobj["requirements"]+[
127                     {
128                     "class": "InitialWorkDirRequirement",
129                     "listing": [{
130                             "entryname": "workflow.cwl",
131                             "entry": yaml.round_trip_dump(packed).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
132                         }, {
133                             "entryname": "cwl.input.yml",
134                             "entry": yaml.round_trip_dump(joborder_keepmount).replace("\\", "\\\\").replace('$(', '\$(').replace('${', '\${')
135                         }]
136                 }],
137                 "hints": workflowobj["hints"],
138                 "arguments": ["--no-container", "--move-outputs", "--preserve-entire-environment", "workflow.cwl#main", "cwl.input.yml"]
139             })
140             kwargs["loader"] = self.doc_loader
141             kwargs["avsc_names"] = self.doc_schema
142             return ArvadosCommandTool(self.arvrunner, wf_runner, **kwargs).job(joborder, output_callback, **kwargs)
143         else:
144             return super(ArvadosWorkflow, self).job(joborder, output_callback, **kwargs)