Merge pull request #1458 from elichad/check-creator-identifier

Workflow best practices: check that creator identifiers are URIs

planemo/workflow_lint.py

@@ -13,6 +13,7 @@
     Tuple,
     TYPE_CHECKING,
 )
+from urllib.parse import urlparse
 
 import requests
 import yaml
@@ -226,8 +227,22 @@ def check_json_for_untyped_params(j):
         lint_context.warn("Workflow is not annotated.")
 
     # creator
-    if not len(workflow_dict.get("creator", [])) > 0:
+    creators = workflow_dict.get("creator", [])
+    if not len(creators) > 0:
         lint_context.warn("Workflow does not specify a creator.")
+    else:
+        if not isinstance(creators, list):
+            # Don't know if this can happen, if we implement schema validation on the Galaxy side
+            # this won't be needed.
+            creators = [creators]
+        for creator in creators:
+            if creator.get("class", "").lower() == "person" and "identifier" in creator:
+                identifier = creator["identifier"]
+                parsed_url = urlparse(identifier)
+                if not parsed_url.scheme:
+                    lint_context.warn(
+                        f'Creator identifier "{identifier}" should be a fully qualified URI, for example "https://orcid.org/0000-0002-1825-0097".'
+                    )
 
     # license
     if not workflow_dict.get("license"):

tests/data/wf19-unlinted-author-identifier-best-practices.ga

@@ -0,0 +1,104 @@
+{
+    "a_galaxy_workflow": "true",
+    "annotation": "",
+    "format-version": "0.1",
+    "name": "bp (imported from uploaded file)",
+    "creator": [
+        {
+            "class": "Person",
+            "identifier": "0000-0002-1825-0097",
+            "name": "Josiah Carberry",
+            "url": "https://orcid.org/0000-0002-1825-0097"
+        }
+    ],
+    "steps": {
+        "0": {
+            "annotation": "",
+            "content_id": null,
+            "errors": null,
+            "id": 0,
+            "input_connections": {},
+            "inputs": [],
+            "label": null,
+            "name": "Input dataset",
+            "outputs": [],
+            "position": {
+                "bottom": 730.3166656494141,
+                "height": 82.55000305175781,
+                "left": 1155,
+                "right": 1355,
+                "top": 647.7666625976562,
+                "width": 200,
+                "x": 1155,
+                "y": 647.7666625976562
+            },
+            "tool_id": null,
+            "tool_state": "{\"optional\": false}",
+            "tool_version": null,
+            "type": "data_input",
+            "uuid": "4219d43a-e182-49c0-bee3-8422e6344911",
+            "workflow_outputs": []
+        },
+        "1": {
+            "annotation": "",
+            "content_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_replace_in_column/1.1.3",
+            "errors": null,
+            "id": 1,
+            "input_connections": {},
+            "inputs": [
+                {
+                    "description": "runtime parameter for tool Replace Text",
+                    "name": "infile"
+                }
+            ],
+            "label": null,
+            "name": "Replace Text",
+            "outputs": [
+                {
+                    "name": "outfile",
+                    "type": "input"
+                }
+            ],
+            "position": {
+                "bottom": 744.8333282470703,
+                "height": 114.06666564941406,
+                "left": 1465,
+                "right": 1665,
+                "top": 630.7666625976562,
+                "width": 200,
+                "x": 1465,
+                "y": 630.7666625976562
+            },
+            "post_job_actions": {
+                "TagDatasetActionoutfile": {
+                    "action_arguments": {
+                        "tags": "${report_name}"
+                    },
+                    "action_type": "TagDatasetAction",
+                    "output_name": "outfile"
+                }
+            },
+            "tool_id": "toolshed.g2.bx.psu.edu/repos/bgruening/text_processing/tp_replace_in_column/1.1.3",
+            "tool_shed_repository": {
+                "changeset_revision": "ddf54b12c295",
+                "name": "text_processing",
+                "owner": "bgruening",
+                "tool_shed": "toolshed.g2.bx.psu.edu"
+            },
+            "tool_state": "{\"infile\": {\"__class__\": \"RuntimeValue\"}, \"replacements\": [{\"__index__\": 0, \"column\": \"1\", \"find_pattern\": \"${report_name}\", \"replace_pattern\": \"\"}], \"__page__\": null, \"__rerun_remap_job_id__\": null}",
+            "tool_version": "1.1.3",
+            "type": "tool",
+            "uuid": "e429e21f-f01e-4efb-b665-56f454cbe38b",
+            "workflow_outputs": [
+                {
+                    "label": null,
+                    "output_name": "outfile",
+                    "uuid": "e4066fe7-c9a2-43af-a9f5-30a75a5b2107"
+                }
+            ]
+        }
+    },
+    "tags": [],
+    "uuid": "8c13f49c-c0c2-4f3a-85d7-5bad6be1776f",
+    "version": 2
+}

tests/test_cmd_workflow_lint.py

@@ -171,6 +171,18 @@ def test_best_practices_linting_ga(self):
         for warning in warnings:
             assert warning in result.output
 
+    def test_author_identifier_best_practices_linting_ga(self):
+        workflow_path = "/".join((TEST_DATA_DIR, "wf19-unlinted-author-identifier-best-practices.ga"))
+        lint_cmd = ["workflow_lint", workflow_path]
+        result = self._runner.invoke(self._cli.planemo, lint_cmd)
+
+        warnings = [
+            'Creator identifier "0000-0002-1825-0097" should be a fully qualified URI, for example "https://orcid.org/0000-0002-1825-0097".',
+        ]
+
+        for warning in warnings:
+            assert warning in result.output
+
     def test_assertion_linting(self):
         workflow_path = "/".join((TEST_DATA_DIR, "wf15-test-assertions.yml"))
         lint_cmd = ["workflow_lint", workflow_path]