Added more tests

tests/dataset/test_dataset.py

@@ -24,6 +24,10 @@ def test_get_context():
     assert "@version" in context
     assert len(context) > 20
 
+    # Check for consistency between context online and on disk
+    online_context = get_context(fromfile=False)
+    assert online_context == context
+
 
 def test_get_prefixes():
     """Test get_prefixes()."""
@@ -223,16 +227,24 @@ def test_save_and_load():
     assert newimage2.distribution["@type"] == DCAT.Distribution
     assert newimage2.distribution.downloadURL == f"file:{newfile2}"
 
+    # Test save anonymous dataset with existing distribution
+    newfile2.unlink(missing_ok=True)
+    save(ts, buf, distribution=SEMDATA.newdistr2)
+    assert newfile2.exists()
+    assert newfile2.stat().st_size == len(buf)
+
     # Test searching the triplestore
     SAMPLE = ts.namespaces["sample"]
-    assert set(list_dataset_iris(ts)) == {
+    datasets = list_dataset_iris(ts)
+    named_datasets = {
         SEMDATA["SEM_cement_batch2/77600-23-001/77600-23-001_5kV_400x_m001"],
         SEMDATA["SEM_cement_batch2/77600-23-001"],
         SEMDATA["SEM_cement_batch2"],
         SAMPLE["SEM_cement_batch2/77600-23-001"],
         SEMDATA.newimage,
         SEMDATA.newimage2,
     }
+    assert not named_datasets.difference(datasets)
     assert set(list_dataset_iris(ts, creator="Sigurd Wenner")) == {
         SEMDATA["SEM_cement_batch2/77600-23-001/77600-23-001_5kV_400x_m001"],
         SEMDATA["SEM_cement_batch2/77600-23-001"],

tests/input/semdata.yaml

@@ -77,3 +77,4 @@ generators:
     generatorType: application/vnd.dlite-generate
     configuration:
       driver: hitachi
+      options: mode=w

tripper/context/0.2/context.json

@@ -75,6 +75,15 @@
     "format": "dcterms:format",
     "checksum": "spdx:checksum",
 
+    "abstract": "dcterms:abstract",
+    "bibliographicCitation": "dcterms:bibliographicCitation",
+    "source": "dcterms:source",
+    "deprecated": "owl:deprecated",
+    "comment": "rdfs:comment",
+    "isDefinedBy": "rdfs:isDefinedBy",
+    "label": "rdfs:label",
+    "seeAlso": "rdfs:seeAlso",
+
     "parser": {
       "@id": "oteio:parser",
       "@type": "@id"

tripper/dataset/dataset.py

@@ -104,6 +104,7 @@ def save(
     class_iri: "Optional[str]" = None,
     dataset: "Optional[Union[str, dict]]" = None,
     distribution: "Optional[Union[str, dict]]" = None,
+    generator: "Optional[str]" = None,
     prefixes: "Optional[dict]" = None,
     use_sparql: "Optional[bool]" = None,
 ) -> None:
@@ -123,6 +124,8 @@ def save(
         distribution: IRI of distribution for the data to be saved.
             Or a dict additional documentation of the distribution,
             like media type, parsers, generators etc...
+        generator: Name of generator to use in case the distribution has
+            several generators.
         prefixes: Dict with prefixes in addition to those included in the
             context.  Should map namespace prefixes to IRIs.
         use_sparql: Whether to access the triplestore with SPARQL.
@@ -194,6 +197,19 @@ def save(
         )
     distribution: dict  # Tell mypy that this now is a dict
 
+    if isinstance(generator, str):
+        for gen in get(distribution, "generator"):
+            if gen.get("@id") == generator:
+                break
+        else:
+            raise ValueError(
+                f"dataset '{dataset}' has no such generator: {generator}"
+            )
+    elif "generator" in distribution:
+        gen = get(distribution, "generator")[0]
+    else:
+        gen = None
+
     # __TODO__: Check if `class_iri` already has the value restriction.
     # If not, add it to triples
 
@@ -210,12 +226,16 @@ def save(
     location = (
         f"{scheme}://{p.netloc}{p.path}" if p.netloc else f"{scheme}:{p.path}"
     )
+    options = [p.query] if p.query else []
+    if gen and "configuration" in gen and "options" in gen.configuration:
+        # __TODO__: allow options to also be a dict
+        options.append(gen.configuration["options"])
     id = (
         distribution["accessService"].get("identifier")
         if "accessService" in distribution
         else None
     )
-    with Protocol(scheme, location, options=p.query) as pr:
+    with Protocol(scheme, location, options=";".join(options)) as pr:
         pr.save(data, id)
 
     # Update triplestore