add option to remap data entity names to the original ones

Author: simleo

src/runcrate/cli.py

@@ -56,15 +56,20 @@ 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(
+    "--remap-names",
+    help="remap file/dir names to the original ones (MAY LEAD TO CLASHES!)",
+    is_flag=True
+)
+def convert(root, output, license, workflow_name, readme, remap_names):
     """\
     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, remap_names=remap_names)
     crate = builder.build()
     if output.suffix == ".zip":
         crate.write_zip(output)

src/runcrate/convert.py

@@ -192,7 +192,8 @@ 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,
+                 remap_names=False):
         self.root = Path(root)
         self.workflow_name = workflow_name
         self.license = license
@@ -213,6 +214,7 @@ def __init__(self, root, workflow_name=None, license=None, readme=None):
         # map source files to destination files
         self.file_map = {}
         self.manifest = self._get_manifest()
+        self.remap_names = remap_names
 
     @staticmethod
     def _get_step_maps(cwl_defs):
@@ -583,14 +585,19 @@ def convert_param(self, prov_param, crate, convert_secondary=True, parent=None):
             return action_p
         if "wf4ever:File" in type_names:
             hash_ = self.hashes[prov_param.id.localpart]
-            dest = Path(parent.id if parent else "") / hash_
+            if self.remap_names:
+                basename = getattr(prov_param, "basename", hash_)
+            else:
+                basename = hash_
+            dest = Path(parent.id if parent else "") / basename
             action_p = crate.dereference(dest.as_posix())
             if not action_p:
                 source = self.manifest[hash_]
                 action_p = crate.add_file(source, dest, properties={
                     "sha1": hash_,
                 })
-                self._set_alternate_name(prov_param, action_p, parent=parent)
+                if not self.remap_names:
+                    self._set_alternate_name(prov_param, action_p, parent=parent)
                 try:
                     source_k = str(source.resolve(strict=False))
                 except RuntimeError:
@@ -599,11 +606,16 @@ def convert_param(self, prov_param, crate, convert_secondary=True, parent=None):
             return action_p
         if "ro:Folder" in type_names:
             hash_ = self.hashes[prov_param.id.localpart]
-            dest = Path(parent.id if parent else "") / hash_
+            if self.remap_names:
+                basename = getattr(prov_param, "basename", hash_)
+            else:
+                basename = hash_
+            dest = Path(parent.id if parent else "") / basename
             action_p = crate.dereference(dest.as_posix())
             if not action_p:
                 action_p = crate.add_directory(dest_path=dest)
-                self._set_alternate_name(prov_param, action_p, parent=parent)
+                if not self.remap_names:
+                    self._set_alternate_name(prov_param, action_p, parent=parent)
                 for child in self.get_dict(prov_param).values():
                     part = self.convert_param(child, crate, parent=action_p)
                     action_p.append_to("hasPart", part)

tests/test_cwlprov_crate_builder.py

@@ -1153,3 +1153,49 @@ def test_revsort_inline(data_dir, tmpdir, cwl_version):
         (reverse_sort_param.id, reverse_param.id),
     }
     assert set(_connected(workflow)) == {(sort_out_param.id, out_file_param.id)}
+
+
+def test_remap_names(data_dir, tmpdir):
+    root = data_dir / "grepucase-run-1"
+    output = tmpdir / "grepucase-run-1-crate"
+    license = "Apache-2.0"
+    builder = ProvCrateBuilder(root, license=license, remap_names=True)
+    crate = builder.build()
+    crate.write(output)
+    crate = ROCrate(output)
+    workflow = crate.mainEntity
+    action_map = {_["instrument"].id: _ for _ in crate.contextual_entities
+                  if "CreateAction" in _.type}
+    assert len(action_map) == 3
+    wf_action = action_map["packed.cwl"]
+    assert wf_action["instrument"] is workflow
+    wf_objects = wf_action["object"]
+    wf_results = wf_action["result"]
+    assert len(wf_objects) == 2
+    assert len(wf_results) == 1
+    wf_objects_map = {_.id: _ for _ in wf_objects}
+    wf_input_dir = wf_objects_map.get("grepucase_in/")
+    assert wf_input_dir
+    wf_output_dir = wf_results[0]
+    assert wf_output_dir.id == "ucase_out/"
+    assert set(_.id for _ in wf_input_dir["hasPart"]) == {
+        "grepucase_in/bar", "grepucase_in/foo"
+    }
+    assert set(_.id for _ in wf_output_dir["hasPart"]) == {
+        "ucase_out/bar.out/", "ucase_out/foo.out/"
+    }
+    for d in wf_output_dir["hasPart"]:
+        if d.id == "ucase_out/bar.out/":
+            assert d["hasPart"][0].id == "ucase_out/bar.out/bar.out.out"
+        else:
+            assert d["hasPart"][0].id == "ucase_out/foo.out/foo.out.out"
+    greptool_action = action_map["packed.cwl#greptool.cwl"]
+    greptool_objects = greptool_action["object"]
+    greptool_results = greptool_action["result"]
+    assert len(greptool_objects) == 2
+    assert len(greptool_results) == 1
+    greptool_objects_map = {_.id: _ for _ in greptool_objects}
+    greptool_input_dir = greptool_objects_map.get("grepucase_in/")
+    assert greptool_input_dir is wf_input_dir
+    greptool_output_dir = greptool_results[0]
+    assert greptool_output_dir.id == "grep_out/"