feat(api): list datasets from a commit

Author: jirikuncar

renku/api/datasets.py

@@ -51,6 +51,26 @@ def renku_datasets_path(self):
         """Return a ``Path`` instance of Renku dataset metadata folder."""
         return self.renku_path.joinpath(self.DATASETS)
 
+    def datasets_from_commit(self, commit=None):
+        """Return datasets defined in a commit."""
+        commit = commit or self.repo.head.commit
+
+        try:
+            datasets = commit.tree / self.renku_home / self.DATASETS
+        except KeyError:
+            return
+
+        for tree in datasets:
+            try:
+                blob = tree / self.METADATA
+            except KeyError:
+                continue
+
+            yield Dataset.from_jsonld(
+                yaml.safe_load(blob.data_stream.read()),
+                __reference__=Path(blob.path),
+            )
+
     @property
     def datasets(self):
         """Return mapping from path to dataset."""
@@ -90,6 +110,7 @@ def with_dataset(self, name=None):
                     self.METADATA
                 )
                 path.parent.mkdir(parents=True, exist_ok=True)
+                setattr(dataset, '__reference__', path)
 
                 if name:
                     LinkReference.create(

renku/cli/_format/datasets.py

@@ -23,15 +23,13 @@
 import click
 
 
-def tabular(client, datasets=None):
+def tabular(client, datasets):
     """Format datasets with a tabular output."""
     from renku.models._tabulate import tabulate
 
-    datasets = datasets or client.datasets
-
     click.echo(
         tabulate(
-            datasets.values(),
+            datasets,
             headers=OrderedDict((
                 ('short_id', 'id'),
                 ('name', None),
@@ -42,16 +40,18 @@ def tabular(client, datasets=None):
     )
 
 
-def jsonld(client, datasets=None):
+def jsonld(client, datasets):
     """Format datasets as JSON-LD."""
     from renku.models._json import dumps
     from renku.models._jsonld import asjsonld
 
-    datasets = datasets or client.datasets
     data = [
         asjsonld(
-            dataset, basedir=os.path.relpath('.', start=str(path.parent))
-        ) for path, dataset in datasets.items()
+            dataset,
+            basedir=os.path.relpath(
+                '.', start=str(dataset.__reference__.parent)
+            )
+        ) for dataset in datasets
     ]
     click.echo(dumps(data, indent=2))
 

renku/cli/dataset.py

@@ -88,6 +88,7 @@
 
 
 @click.group(invoke_without_command=True)
+@click.option('--revision', default=None)
 @click.option('--datadir', default='data', type=click.Path(dir_okay=True))
 @click.option(
     '--format',
@@ -97,14 +98,19 @@
 )
 @pass_local_client(clean=False, commit=False)
 @click.pass_context
-def dataset(ctx, client, datadir, format):
+def dataset(ctx, client, revision, datadir, format):
     """Handle datasets."""
     ctx.meta['renku.datasets.datadir'] = datadir
 
     if ctx.invoked_subcommand is not None:
         return
 
-    DATASETS_FORMATS[format](client)
+    if revision is None:
+        datasets = client.datasets.values()
+    else:
+        datasets = client.datasets_from_commit(client.repo.commit(revision))
+
+    DATASETS_FORMATS[format](client, datasets)
 
 
 @dataset.command()

tests/cli/test_datasets.py

@@ -93,6 +93,12 @@ def test_datasets_list_non_empty(output_format, runner, project):
     assert 0 == result.exit_code
     assert 'dataset' in result.output
 
+    result = runner.invoke(
+        cli.cli, ['dataset', '--revision=HEAD~1', format_option]
+    )
+    assert result.exit_code == 0
+    assert 'dataset' not in result.output
+
 
 def test_multiple_file_to_dataset(
     tmpdir, data_repository, runner, project, client