Merge pull request #70 from simleo/output_encoding_format

Convert encoding format for output files

Author: simleo

cwl/grepsort/grepsort.cwl

@@ -19,6 +19,7 @@ inputs:
 outputs:
   grepsort_out:
     type: File
+    format: "https://www.iana.org/assignments/media-types/text/plain"
     outputSource: sorted/sort_out
 
 steps:

cwl/grepsort/sorttool.cwl

@@ -20,6 +20,7 @@ inputs:
 outputs:
   sort_out:
     type: File
+    format: "https://www.iana.org/assignments/media-types/text/plain"
     outputBinding:
       glob: sort_out.txt
 stdout: sort_out.txt

src/runcrate/convert.py

@@ -40,6 +40,7 @@
 
 WORKFLOW_BASENAME = "packed.cwl"
 INPUTS_FILE_BASENAME = "primary-job.json"
+OUTPUTS_FILE_BASENAME = "primary-output.json"
 MANIFEST_FILE = "manifest-sha1.txt"
 
 CWL_TYPE_MAP = {
@@ -294,6 +295,7 @@ def build(self):
         self.add_action(crate, self.workflow_run)
         self.patch_workflow_input_collection(crate)
         self.add_inputs_file(crate)
+        self.add_output_formats(crate)
         return crate
 
     def add_root_metadata(self, crate):
@@ -777,3 +779,10 @@ def add_inputs_file(self, crate):
                 "name": "input object document",
                 "encodingFormat": "application/json",
             })
+
+    def add_output_formats(self, crate):
+        path = self.root / "workflow" / OUTPUTS_FILE_BASENAME
+        if path.is_file():
+            with open(path) as f:
+                data = json.load(f)
+            self._map_input_data(crate, data)

tests/data/grepsort-run-1/bag-info.txt

@@ -1,6 +1,6 @@
-Bag-Software-Agent: cwltool 3.1.20230513155734
+Bag-Software-Agent: cwltool 3.1.20231020140205
 BagIt-Profile-Identifier: https://w3id.org/ro/bagit/profile
-Bagging-Date: 2023-05-26
+Bagging-Date: 2023-11-09
 External-Description: Research Object of CWL workflow run
-External-Identifier: arcp://uuid,c54a7289-731a-498b-9f08-3cc2bb801ba1/
+External-Identifier: arcp://uuid,8b3eb82c-de04-4dd0-b519-0354c0ce7cf5/
 Payload-Oxum: 192.4

tests/data/grepsort-run-1/metadata/logs/engine.158cac0a-3a3b-4716-b023-2f9d48de837a.txt

@@ -0,0 +1,20 @@
+[2023-11-09T13:19:38,436.000000Z] [cwltool] /home/simleo/git/runcrate/venv/bin/cwltool --provenance ro grepsort.cwl params.yml
+[2023-11-09T13:19:38,439.000000Z] Resolved 'grepsort.cwl' to 'file:///home/simleo/git/runcrate/cwl/grepsort/grepsort.cwl'
+[2023-11-09T13:19:38,940.000000Z] [provenance] Adding to RO file:///home/simleo/git/runcrate/cwl/grepsort/lines.txt
+[2023-11-09T13:19:38,940.000000Z] [provenance] Adding to RO file:///home/simleo/git/runcrate/cwl/grepsort/lines.aux
+[2023-11-09T13:19:38,941.000000Z] [workflow ] start
+[2023-11-09T13:19:38,942.000000Z] [workflow ] starting step grep
+[2023-11-09T13:19:38,942.000000Z] [step grep] start
+[2023-11-09T13:19:38,943.000000Z] [job grep] /tmp/mruwmsr2$ bash \
+    -c \
+    'grep -f /tmp/b6vih64i/stg5a0c19bf-8d22-4f74-b973-641b81910030/lines.aux /tmp/b6vih64i/stg5a0c19bf-8d22-4f74-b973-641b81910030/lines.txt >grep_out.txt'
+[2023-11-09T13:19:38,953.000000Z] [job grep] completed success
+[2023-11-09T13:19:38,953.000000Z] [step grep] completed success
+[2023-11-09T13:19:38,953.000000Z] [workflow ] starting step sorted
+[2023-11-09T13:19:38,953.000000Z] [step sorted] start
+[2023-11-09T13:19:38,954.000000Z] [job sorted] /tmp/w0c9z4je$ sort \
+    /tmp/epamolr9/stgfc1b1b27-6041-4eb9-be0b-5ae94c510df3/grep_out.txt > /tmp/w0c9z4je/sort_out.txt
+[2023-11-09T13:19:38,957.000000Z] [job sorted] completed success
+[2023-11-09T13:19:38,957.000000Z] [step sorted] completed success
+[2023-11-09T13:19:38,957.000000Z] [workflow ] completed success
+[2023-11-09T13:19:38,987.000000Z] Final process status is success