Merge pull request #92 from simleo/rde_properties

Root Data Entity required properties

Author: simleo

src/runcrate/cli.py

@@ -56,15 +56,27 @@ def cli():
     type=click.Path(exists=True, dir_okay=False, readable=True, path_type=Path),
     help="path to a README file (should be README.md in Markdown format)",
 )
-def convert(root, output, license, workflow_name, readme):
+@click.option(
+    "-n",
+    "--crate-name",
+    metavar="STRING",
+    help="crate name",
+)
+@click.option(
+    "-d",
+    "--crate-desc",
+    metavar="STRING",
+    help="crate description",
+)
+def convert(root, output, license, workflow_name, readme, crate_name, crate_desc):
     """\
     Convert a CWLProv RO bundle into a Workflow Run RO-Crate.
 
     RO_DIR: top-level directory of the CWLProv RO
     """
     if not output:
         output = Path(f"{root.name}.crate.zip")
-    builder = ProvCrateBuilder(root, workflow_name, license, readme)
+    builder = ProvCrateBuilder(root, workflow_name, license, readme, crate_name, crate_desc)
     crate = builder.build()
     if output.suffix == ".zip":
         crate.write_zip(output)

src/runcrate/convert.py

@@ -61,7 +61,9 @@
 
 CWLPROV_NONE = "https://w3id.org/cwl/prov#None"
 
+# https://w3id.org/workflowhub/workflow-ro-crate/1.0
 WROC_PROFILE_VERSION = "1.0"
+DEFAULT_LICENSE = "notspecified"
 
 DOCKER_IMG_TYPE = "https://w3id.org/ro/terms/workflow-run#DockerImage"
 
@@ -197,11 +199,14 @@ def get_workflow(wf_path):
 
 class ProvCrateBuilder:
 
-    def __init__(self, root, workflow_name=None, license=None, readme=None):
+    def __init__(self, root, workflow_name=None, license=None, readme=None,
+                 crate_name=None, crate_desc=None):
         self.root = Path(root)
         self.workflow_name = workflow_name
-        self.license = license
+        self.license = license or DEFAULT_LICENSE
         self.readme = Path(readme) if readme else readme
+        self.crate_name = crate_name
+        self.crate_desc = crate_desc
         self.wf_path = self.root / "workflow" / WORKFLOW_BASENAME
         self.cwl_defs = get_workflow(self.wf_path)
         self.step_maps = self._get_step_maps(self.cwl_defs)
@@ -290,9 +295,9 @@ def get_dict(self, entity):
     def build(self):
         crate = ROCrate(gen_preview=False)
         crate.metadata.extra_contexts.append(TERMS_NAMESPACE)
-        self.add_root_metadata(crate)
         self.add_profiles(crate)
         self.add_workflow(crate)
+        self.add_root_metadata(crate)
         self.add_engine_run(crate)
         self.add_action(crate, self.workflow_run)
         self.patch_workflow_input_collection(crate)
@@ -301,13 +306,16 @@ def build(self):
         return crate
 
     def add_root_metadata(self, crate):
-        if self.license:
-            crate.root_dataset["license"] = self.license
+        crate.root_dataset["license"] = self.license
         if self.readme:
             readme = crate.add_file(self.readme)
             readme["about"] = crate.root_dataset
             if self.readme.suffix.lower() == ".md":
                 readme["encodingFormat"] = "text/markdown"
+        if not self.crate_name:
+            self.crate_name = f"run of {self.workflow_name}"
+        crate.root_dataset["name"] = self.crate_name
+        crate.root_dataset["description"] = self.crate_desc or self.crate_name
 
     def add_profiles(self, crate):
         profiles = []
@@ -338,10 +346,11 @@ def add_workflow(self, crate):
             lang_version=lang_version, gen_cwl=False, properties=properties
         )
         cwl_workflow = self.cwl_defs[workflow.id]
-        wf_name = self.wf_path.name
+        if not self.workflow_name:
+            self.workflow_name = self.wf_path.name
         if hasattr(cwl_workflow, "label") and cwl_workflow.label:
-            wf_name = cwl_workflow.label
-        workflow["name"] = self.workflow_name or wf_name
+            self.workflow_name = cwl_workflow.label
+        workflow["name"] = self.workflow_name
         if hasattr(cwl_workflow, "doc") and cwl_workflow.doc:
             workflow["description"] = cwl_workflow.doc
         # cannot convert "intent" to featureList: workflow is not a SoftwareApplication

tests/test_cli.py

@@ -32,7 +32,7 @@ def test_cli_convert(data_dir, tmpdir, monkeypatch):
     crate_zip = tmpdir / f"{root.name}.crate.zip"
     assert crate_zip.is_file()
     crate = ROCrate(crate_zip)
-    assert not crate.root_dataset.get("license")
+    assert crate.root_dataset.get("license") == "notspecified"
     workflow = crate.mainEntity
     assert workflow["name"] == "packed.cwl"
 
@@ -47,7 +47,7 @@ def test_cli_convert_output(data_dir, tmpdir):
     assert result.exit_code == 0, result.exception
     assert crate_zip.is_file()
     crate = ROCrate(crate_zip)
-    assert not crate.root_dataset.get("license")
+    assert crate.root_dataset.get("license") == "notspecified"
     workflow = crate.mainEntity
     assert workflow["name"] == "packed.cwl"
     crate_zip.unlink()
@@ -59,7 +59,7 @@ def test_cli_convert_output(data_dir, tmpdir):
     assert result.exit_code == 0, result.exception
     assert crate_dir.is_dir()
     crate = ROCrate(crate_dir)
-    assert not crate.root_dataset.get("license")
+    assert crate.root_dataset.get("license") == "notspecified"
     workflow = crate.mainEntity
     assert workflow["name"] == "packed.cwl"
 
@@ -102,6 +102,34 @@ def test_cli_convert_workflow_name(data_dir, tmpdir):
     assert workflow["name"] == workflow_name
 
 
+def test_cli_convert_crate_name(data_dir, tmpdir):
+    root = data_dir / "revsort-run-1"
+    crate_dir = tmpdir / "revsort-run-1-crate"
+    workflow_name = "RevSort"
+    crate_name = "name of the crate"
+    runner = CliRunner()
+    args = ["convert", str(root), "-o", str(crate_dir), "-w", workflow_name, "-n", crate_name]
+    result = runner.invoke(cli, args)
+    assert result.exit_code == 0, result.exception
+    crate = ROCrate(crate_dir)
+    assert crate.root_dataset["name"] == crate_name
+    assert crate.root_dataset["description"] == crate_name
+
+
+def test_cli_convert_crate_desc(data_dir, tmpdir):
+    root = data_dir / "revsort-run-1"
+    crate_dir = tmpdir / "revsort-run-1-crate"
+    workflow_name = "RevSort"
+    crate_desc = "description of the crate"
+    runner = CliRunner()
+    args = ["convert", str(root), "-o", str(crate_dir), "-w", workflow_name, "-d", crate_desc]
+    result = runner.invoke(cli, args)
+    assert result.exit_code == 0, result.exception
+    crate = ROCrate(crate_dir)
+    assert crate.root_dataset["name"] == f"run of {workflow_name}"
+    assert crate.root_dataset["description"] == crate_desc
+
+
 def test_cli_convert_readme(data_dir, tmpdir):
     root = data_dir / "revsort-run-1"
     crate_dir = tmpdir / "revsort-run-1-crate"

tests/test_cwlprov_crate_builder.py

@@ -47,7 +47,9 @@ def test_revsort(data_dir, tmpdir):
     assert readme_f["about"] is crate.root_dataset
     workflow = crate.mainEntity
     assert workflow.id == "packed.cwl"
-    assert workflow["name"] == "RevSort"
+    assert workflow["name"] == workflow_name
+    assert crate.root_dataset["name"] == f"run of {workflow_name}"
+    assert crate.root_dataset["description"] == f"run of {workflow_name}"
     tools = workflow["hasPart"]
     assert len(tools) == 2
     for entity in tools:
@@ -1191,3 +1193,24 @@ def test_agent_no_name(data_dir, tmpdir):
     assert agent.type == "Person"
     assert agent.id == "https://orcid.org/0000-0001-9447-460X"
     assert "name" not in agent
+
+
+def test_no_license(data_dir):
+    root = data_dir / "revsort-run-1"
+    builder = ProvCrateBuilder(root)
+    crate = builder.build()
+    assert crate.root_dataset["license"] == "notspecified"
+
+
+def test_name_desc(data_dir):
+    root = data_dir / "revsort-run-1"
+    workflow_name = "RevSort"
+    crate_name = "name of the crate"
+    builder = ProvCrateBuilder(root, workflow_name=workflow_name, crate_name=crate_name)
+    crate = builder.build()
+    assert crate.root_dataset["name"] == crate_name
+    assert crate.root_dataset["description"] == crate_name
+    crate_desc = "description of the crate"
+    builder = ProvCrateBuilder(root, workflow_name=workflow_name, crate_name=crate_name, crate_desc=crate_desc)
+    crate = builder.build()
+    assert crate.root_dataset["description"] == crate_desc