label change in prov

Author: pauldg

rocrate/provenance_profile.py

@@ -121,7 +121,17 @@ def __init__(
         metadata_export = load_ga_history_export(ga_export)
         self.generate_prov_doc()
         self.jobs = []
+        self.datasets = []
+
         # print(metadata_export["jobs_attrs"][0]["params"])
+        for dataset in metadata_export["datasets_attrs"]:
+            ds_attrs = GalaxyDataset()
+            ds_attrs.parse_ga_dataset_attrs(dataset)
+            self.datasets.append(ds_attrs.attributes)
+            # self.declare_process(ds_attrs.attributes)
+
+        print(self.datasets[0])
+
         for job in metadata_export["jobs_attrs"]:
             job_attrs = GalaxyJob()
             job_attrs.parse_ga_jobs_attrs(job)
@@ -234,7 +244,7 @@ def declare_process(
         cmd = ga_export_jobs_attrs["command_line"]
         process_name = ga_export_jobs_attrs["tool_id"]
         tool_version = ga_export_jobs_attrs["tool_version"]
-        prov_label = "Run of ga_export/jobs_attrs.txt#" + process_name
+        prov_label = "Run of " + process_name
         start_time = ga_export_jobs_attrs["create_time"]
         end_time = ga_export_jobs_attrs["update_time"]
 
@@ -294,7 +304,7 @@ def used_artefacts(
                 if isinstance(key, str):
                     key = key.replace("|", "_")
                 if isinstance(value, str):    
-                    val = value.replace("|", "_")
+                    value = value.replace("|", "_")
 
                 prov_role = self.wf_ns[f"{base}/{key}"]
 
@@ -353,6 +363,7 @@ def declare_artefact(self, value: Any) -> ProvEntity:
             )
 
         if isinstance(value, Dict):
+            # print(value, end ='\n')
             if "@id" in value:
                 # Already processed this value, but it might not be in this PROV
                 entities = self.document.get_record(value["@id"])

test/test_parse_ga_export.py

@@ -31,7 +31,7 @@ def test_ga_history_loading(test_data_dir, tmpdir, helpers):
 def test_ga_history_parsing(test_data_dir, tmpdir, helpers):
     export_dir = "test_ga_history_export"
     export_path = test_data_dir / export_dir / "history_export"
-    prov_path = "/home/padge/Elixir/workflow-export/ro-crate-py/test/test-data/test_ga_history_export/history_export/provenance"
+    prov_path = export_path / "provenance"
     # prov_name = "ga_export.cwlprov"
     # crate_path = test_data_dir / export_dir / "history_export_crate"
 

tools/load_ga_export.py

@@ -58,21 +58,21 @@ def parse_ga_jobs_attrs(self, job_attrs):
                         self.attributes["parameters"].update(job_attrs[key])
 
 class GalaxyDataset(Dict):
-    def __init__(self, ga_export_dataset_attrs):
+    def __init__(self):
         """
         Initialize the GalaxyDataset object.
         """
         self.attributes = {}
         self.attributes["metadata"] = {}
 
-    def parse_ga_dataset_attrs(self, job_attrs):
+    def parse_ga_dataset_attrs(self, ds_attrs):
 
-        for key, value in job_attrs.items():
+        for key, value in ds_attrs.items():
             if not isinstance(value, dict):
                 self.attributes[key] = value
             else:
                 if len(value) == 0:
                     pass
                 else:
                     if "metadata" in key:
-                        self.attributes["metadata"].update(job_attrs[key])
+                        self.attributes["metadata"].update(ds_attrs[key])