Merge pull request #69 from simleo/more_file_metadata

More file metadata

Author: simleo

src/runcrate/convert.py

@@ -566,6 +566,8 @@ def add_action_params(self, crate, activity, to_wf_p, ptype="usage"):
             ))
             if len(action_p["exampleOfWork"]) == 1:
                 action_p["exampleOfWork"] = action_p["exampleOfWork"][0]
+            if ptype == "generation":
+                action_p["dateCreated"] = rel.time.isoformat()
             action_params.append(action_p)
         return action_params
 
@@ -606,6 +608,7 @@ def convert_param(self, prov_param, crate, convert_secondary=True, parent=None):
                 source = self.manifest[hash_]
                 action_p = crate.add_file(source, dest, properties={
                     "sha1": hash_,
+                    "contentSize": str(Path(source).stat().st_size)
                 })
                 self._set_alternate_name(prov_param, action_p, parent=parent)
                 try:
@@ -739,9 +742,9 @@ def patch_workflow_input_collection(self, crate, wf=None):
             if "ComputationalWorkflow" in as_list(tool.type):
                 self.patch_workflow_input_collection(crate, wf=tool)
 
-    def _map_input_data(self, data):
+    def _map_input_data(self, crate, data):
         if isinstance(data, list):
-            return [self._map_input_data(_) for _ in data]
+            return [self._map_input_data(crate, _) for _ in data]
         if isinstance(data, dict):
             rval = {}
             for k, v in data.items():
@@ -753,8 +756,13 @@ def _map_input_data(self, data):
                         source_k = str(source)
                     dest = self.file_map.get(source_k)
                     rval[k] = str(dest) if dest else v
+                    fmt = data.get("format")
+                    if fmt:
+                        entity = crate.get(str(dest))
+                        if entity:
+                            entity["encodingFormat"] = fmt
                 else:
-                    rval[k] = self._map_input_data(v)
+                    rval[k] = self._map_input_data(crate, v)
             return rval
         return data
 
@@ -763,7 +771,7 @@ def add_inputs_file(self, crate):
         if path.is_file():
             with open(path) as f:
                 data = json.load(f)
-            data = self._map_input_data(data)
+            data = self._map_input_data(crate, data)
             source = StringIO(json.dumps(data, indent=4))
             crate.add_file(source, path.name, properties={
                 "name": "input object document",

tests/test_cwlprov_crate_builder.py

@@ -95,10 +95,14 @@ def test_revsort(data_dir, tmpdir):
             assert "File" in entity.type
             assert entity["alternateName"] == "whale.txt"
             assert entity["sha1"] == entity.id.rsplit("/")[-1]
+            assert entity["contentSize"] == "1111"
+            assert "encodingFormat" in entity
             wf_input_file = entity
     wf_output_file = wf_results[0]
     assert wf_output_file["alternateName"] == "output.txt"
     assert wf_output_file["sha1"] == wf_output_file.id.rsplit("/")[-1]
+    assert wf_output_file["dateCreated"] == "2018-10-25T15:46:38.058365"
+    assert wf_output_file["contentSize"] == "1111"
     assert "File" in wf_output_file.type
     steps = workflow["step"]
     assert len(steps) == 2
@@ -118,6 +122,8 @@ def test_revsort(data_dir, tmpdir):
             assert rev_input_file is wf_input_file
             rev_output_file = results[0]
             assert "File" in rev_output_file.type
+            assert rev_output_file["dateCreated"] == "2018-10-25T15:46:36.963254"
+            assert rev_output_file["contentSize"] == "1111"
             assert step["position"] == "0"
             assert set(_connected(step)) == set([
                 ("packed.cwl#main/input", "packed.cwl#revtool.cwl/input"),
@@ -357,6 +363,7 @@ def test_dir_io(data_dir, tmpdir):
             assert "Dataset" in entity.type
             wf_input_dir = entity
     wf_output_dir = wf_results[0]
+    assert wf_output_dir["dateCreated"] == "2023-02-17T16:20:30.288242"
     assert wf_input_dir.type == wf_output_dir.type == "Dataset"
     assert wf_input_dir["alternateName"] == "grepucase_in"
     assert len(wf_input_dir["hasPart"]) == 2
@@ -395,6 +402,7 @@ def test_dir_io(data_dir, tmpdir):
     assert greptool_input_dir is wf_input_dir
     greptool_output_dir = greptool_results[0]
     assert "Dataset" in greptool_output_dir.type
+    assert greptool_output_dir["dateCreated"] == "2023-02-17T16:20:30.262141"
     ucasetool_action = action_map["packed.cwl#ucasetool.cwl"]
     ucasetool_objects = ucasetool_action["object"]
     ucasetool_results = ucasetool_action["result"]