new option to add a README file

src/runcrate/__init__.py

@@ -180,10 +180,11 @@ def get_workflow(wf_path):
 
 class ProvCrateBuilder:
 
-    def __init__(self, root, workflow_name=None, license=None):
+    def __init__(self, root, workflow_name=None, license=None, readme=None):
         self.root = Path(root)
         self.workflow_name = workflow_name
         self.license = license
+        self.readme = Path(readme) if readme else readme
         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)
@@ -261,13 +262,23 @@ def get_dict(self, entity):
     def build(self):
         crate = ROCrate(gen_preview=False)
         crate.metadata.extra_terms.update(EXTRA_TERMS)
+        self.add_root_metadata(crate)
         self.add_profiles(crate)
         self.add_workflow(crate)
         self.add_engine_run(crate)
         self.add_action(crate, self.workflow_run)
         self.patch_workflow_input_collection(crate)
         return crate
 
+    def add_root_metadata(self, crate):
+        if 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"
+
     def add_profiles(self, crate):
         profiles = []
         for p in "process", "workflow", "provenance":
@@ -297,8 +308,6 @@ def add_workflow(self, crate):
             self.wf_path, self.wf_path.name, main=True, lang="cwl",
             lang_version=lang_version, gen_cwl=False, properties=properties
         )
-        if self.license:
-            crate.root_dataset["license"] = self.license
         cwl_workflow = self.cwl_defs[workflow.id]
         workflow["input"] = self.add_params(crate, cwl_workflow.inputs)
         workflow["output"] = self.add_params(crate, cwl_workflow.outputs)

src/runcrate/cli.py

@@ -47,13 +47,18 @@ def cli():
     metavar="STRING",
     help="original workflow name",
 )
-def convert(root, output, license, workflow_name):
+@click.option(
+    "--readme",
+    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):
     """\
     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)
+    builder = ProvCrateBuilder(root, workflow_name, license, readme)
     crate = builder.build()
     if output.suffix == ".zip":
         crate.write_zip(output)

tests/test_cli.py

@@ -78,3 +78,14 @@ def test_cli_convert_workflow_name(data_dir, tmpdir):
     crate = ROCrate(crate_dir)
     workflow = crate.mainEntity
     assert workflow["name"] == workflow_name
+
+
+def test_cli_convert_readme(data_dir, tmpdir):
+    root = data_dir / "revsort-run-1"
+    crate_dir = tmpdir / "revsort-run-1-crate"
+    readme = data_dir / "README.md"
+    runner = CliRunner()
+    args = ["convert", str(root), "-o", str(crate_dir), "--readme", readme]
+    assert runner.invoke(cli, args).exit_code == 0
+    crate = ROCrate(crate_dir)
+    assert crate.get(readme.name)

tests/test_cwlprov_crate_builder.py

@@ -29,11 +29,17 @@ def test_revsort(data_dir, tmpdir):
     root = data_dir / "revsort-run-1"
     output = tmpdir / "revsort-run-1-crate"
     license = "Apache-2.0"
+    readme = data_dir / "README.md"
     workflow_name = "RevSort"
-    builder = ProvCrateBuilder(root, workflow_name=workflow_name, license=license)
+    builder = ProvCrateBuilder(root, workflow_name=workflow_name, license=license, readme=readme)
     crate = builder.build()
     crate.write(output)
     assert crate.root_dataset["license"] == "Apache-2.0"
+    readme_f = crate.get(readme.name)
+    assert readme_f
+    assert readme_f.type == "File"
+    assert readme_f["encodingFormat"] == "text/markdown"
+    assert readme_f["about"] is crate.root_dataset
     workflow = crate.mainEntity
     assert workflow.id == "packed.cwl"
     assert workflow["name"] == "RevSort"
@@ -136,6 +142,7 @@ def test_revsort(data_dir, tmpdir):
     assert (output / rev_output_file.id).read_text() == rev_out_text
     out_text = (root / "data/b9/b9214658cc453331b62c2282b772a5c063dbd284").read_text()
     assert (output / wf_output_file.id).read_text() == out_text
+    assert (output / readme.name).read_text() == readme.read_text()
     # declared profile conformance
     proc_prof = crate.get("https://w3id.org/ro/wfrun/process/0.1")
     wf_prof = crate.get("https://w3id.org/ro/wfrun/workflow/0.1")