Merge pull request #136 from MaRDI4NFDI/new_fixes

fix format and pull description
MaRDI4NFDI · Apr 12, 2024 · 6eeef1b · 6eeef1b
2 parents 74a8ef8 + 97ee19e
commit 6eeef1b
Show file tree

Hide file tree

Showing 4 changed files with 14 additions and 2 deletions.
diff --git a/mardi_importer/mardi_importer/openml/OpenMLDataset.py b/mardi_importer/mardi_importer/openml/OpenMLDataset.py
@@ -41,6 +41,7 @@ def __init__(
             integrator, 
             name,
             dataset_id,
+            description,
             version,
             creators,
             contributors,
@@ -67,6 +68,7 @@ def __init__(
         self.api = integrator
         self.name = name #done
         self.dataset_id = str(dataset_id) #done
+        self.description=description
         self.version = version #done
         self.creators = creators
         self.contributors = contributors
@@ -110,6 +112,9 @@ def insert_claims(self):
         if self.version is not None and self.version != "None":
             prop_nr = self.api.get_local_id_by_label("dataset version identifier", "property")
             self.item.add_claim(prop_nr, str(self.version))
+        if self.description is not None:
+            prop_nr = self.api.get_local_id_by_label("description", "property")
+            self.item.add_claim(prop_nr, str(self.description))
         if self.creators and self.creators != "None":
             #object has role 
             qualifier = [self.api.get_claim("wdt:P3831", "wd:Q59275219")]

diff --git a/mardi_importer/mardi_importer/openml/OpenMLSource.py b/mardi_importer/mardi_importer/openml/OpenMLSource.py
@@ -48,7 +48,7 @@ def create_local_entities(self):
                 item.write()
 
     def pull(self):
-        dataset_dict = {"name": [], "dataset_id": [], "version": [], "creators": [],
+        dataset_dict = {"name": [], "dataset_id": [], "description":[], "version": [], "creators": [],
                         "contributors" : [], "collection_date": [], "upload_date": [], 
                         "license": [], "url":[], "default_target_attribute":[], "row_id_attribute":[],
                         "tags":[], "original_data_url":[], "paper_url":[],
@@ -65,6 +65,7 @@ def pull(self):
                 ds = openml.datasets.get_dataset(int(did), download_data=False, download_qualities=False, download_features_meta_data=False)
             dataset_dict["name"].append(ds.name)
             dataset_dict["dataset_id"].append(did)
+            dataset_dict["description"].append(ds.description)
             dataset_dict["version"].append(ds.version)
             dataset_dict["creators"].append(ds.creator)
             dataset_dict["contributors"].append(ds.contributor)
@@ -78,6 +79,7 @@ def pull(self):
             dataset_dict["original_data_url"].append(ds.original_data_url)
             dataset_dict["paper_url"].append(ds.paper_url)
             dataset_dict["md5_checksum"].append(ds.md5_checksum)
+            dataset_dict["format"].append(ds.format)
             try:
                 qualities = ds.qualities
             except:

diff --git a/mardi_importer/mardi_importer/openml/new_entities.json b/mardi_importer/mardi_importer/openml/new_entities.json
@@ -5,6 +5,11 @@
             "description": "Version of a dataset",
             "datatype": "string"
         },
+        {
+            "label": "description",
+            "description": "long description of an item",
+            "datatype": "string"
+        },
         {
             "label": "collection date",
             "description": "date as a string",

diff --git a/mardi_importer/mardi_importer/openml/wikidata_entities.txt b/mardi_importer/mardi_importer/openml/wikidata_entities.txt
@@ -14,4 +14,4 @@ P459
 P11238
 P3831
 Q59275219
-Q20204892
+Q20204892