feat: metadata on dataset creation

renku/cli/dataset.py

@@ -385,9 +385,34 @@ def dataset(ctx, revision, datadir, format):
 
 @dataset.command()
 @click.argument('name')
-def create(name):
+@click.option(
+    '--short-name', default='', help='A convenient name for dataset.'
+)
+@click.option(
+    '-d', '--description', default='', help='Dataset\'s description.'
+)
+@click.option(
+    '-c',
+    '--creator',
+    default=None,
+    multiple=True,
+    help='Creator\'s name and email ("Name <email>").'
+)
+def create(name, short_name, description, creator):
     """Create an empty dataset in the current repo."""
-    create_dataset(name)
+    creators = creator or ()
+
+    dataset = create_dataset(
+        name=name,
+        short_name=short_name,
+        description=description,
+        creators=creators
+    )
+    click.echo(
+        'Use the name "{}" to refer to this dataset.'.format(
+            dataset.short_name
+        )
+    )
     click.secho('OK', fg='green')
 
 
@@ -606,14 +631,16 @@ def export_(id, provider, publish, tag):
 
 @dataset.command('import')
 @click.argument('uri')
-@click.option('-n', '--name', help='Dataset name.')
+@click.option(
+    '--short-name', default='', help='A convenient name for dataset.'
+)
 @click.option(
     '-x',
     '--extract',
     is_flag=True,
     help='Extract files before importing to dataset.'
 )
-def import_(uri, name, extract):
+def import_(uri, short_name, extract):
     """Import data from a 3rd party provider.
 
     Supported providers: [Zenodo, Dataverse]
@@ -638,9 +665,9 @@ def _init(lock, id_queue):
         tqdm.set_lock(lock)
 
     import_dataset(
-        uri,
-        name,
-        extract,
+        uri=uri,
+        short_name=short_name,
+        extract=extract,
         with_prompt=True,
         pool_init_fn=_init,
         pool_init_args=(mp.RLock(), id_queue),

renku/core/commands/checks/migration.py

@@ -162,7 +162,7 @@ def migrate_broken_dataset_paths(client):
         # migrate the refs
         ref = LinkReference.create(
             client=client,
-            name='datasets/{0}'.format(dataset.display_name),
+            name='datasets/{0}'.format(dataset.short_name),
             force=True,
         )
         ref.set_reference(expected_path / client.METADATA)

renku/core/commands/dataset.py

@@ -42,7 +42,7 @@
     MigrationRequired, ParameterError, UsageError
 from renku.core.management.datasets import DATASET_METADATA_PATHS
 from renku.core.management.git import COMMIT_DIFF_STRATEGY
-from renku.core.models.datasets import Dataset
+from renku.core.models.datasets import Dataset, generate_default_short_name
 from renku.core.models.provenance.agents import Person
 from renku.core.models.refs import LinkReference
 from renku.core.models.tabulate import tabulate
@@ -101,15 +101,26 @@ def dataset_parent(client, revision, datadir, format, ctx=None):
 @pass_local_client(
     clean=False, commit=True, commit_only=DATASET_METADATA_PATHS
 )
-def create_dataset(client, name, commit_message=None):
+def create_dataset(
+    client, name, short_name, description, creators, commit_message=None
+):
     """Create an empty dataset in the current repo.
 
     :raises: ``renku.core.errors.ParameterError``
     """
-    with client.with_dataset(name=name, create=True) as dataset:
-        creator = Person.from_git(client.repo)
-        if creator not in dataset.creator:
-            dataset.creator.append(creator)
+    if not creators:
+        creators = [Person.from_git(client.repo)]
+    else:
+        creators = [Person.from_string(c) for c in creators]
+
+    dataset, _, __ = client.create_dataset(
+        name=name,
+        short_name=short_name,
+        description=description,
+        creators=creators
+    )
+
+    return dataset
 
 
 @pass_local_client(
@@ -284,7 +295,7 @@ def dataset_remove(
     commit_message=None
 ):
     """Delete a dataset."""
-    datasets = {name: client.dataset_path(name) for name in names}
+    datasets = {name: client.get_dataset_path(name) for name in names}
 
     if not datasets:
         raise ParameterError(
@@ -422,8 +433,8 @@ def export_dataset(
 def import_dataset(
     client,
     uri,
-    name,
-    extract,
+    short_name='',
+    extract=False,
     with_prompt=False,
     pool_init_fn=None,
     pool_init_args=None,
@@ -474,6 +485,15 @@ def import_dataset(
         )
 
     if files:
+        if not short_name:
+            short_name = generate_default_short_name(
+                dataset.name, dataset.version
+            )
+
+        dataset.short_name = short_name
+
+        client.create_dataset(name=dataset.name, short_name=short_name)
+
         data_folder = tempfile.mkdtemp()
 
         pool_size = min(
@@ -511,20 +531,18 @@ def import_dataset(
             ))
         pool.close()
 
-        dataset_name = name or dataset.display_name
         dataset.url = remove_credentials(dataset.url)
         add_to_dataset(
             client,
             urls=[str(p) for p in Path(data_folder).glob('*')],
-            name=dataset_name,
-            with_metadata=dataset,
-            create=True
+            name=short_name,
+            with_metadata=dataset
         )
 
         if dataset.version:
             tag_name = re.sub('[^a-zA-Z0-9.-_]', '_', dataset.version)
             tag_dataset(
-                client, dataset_name, tag_name,
+                client, short_name, tag_name,
                 'Tag {} created by renku import'.format(dataset.version)
             )
 
@@ -633,7 +651,7 @@ def _filter(client, names=None, creators=None, include=None, exclude=None):
 
     records = []
     for path_, dataset in client.datasets.items():
-        if not names or dataset.name in names:
+        if not names or dataset.short_name in names:
             for file_ in dataset.files:
                 file_.dataset = dataset.name
                 path_ = file_.full_path.relative_to(client.path)

renku/core/commands/format/datasets.py

@@ -31,7 +31,7 @@ def tabular(client, datasets):
         datasets,
         headers=OrderedDict((
             ('uid', 'id'),
-            ('display_name', None),
+            ('short_name', None),
             ('version', None),
             ('created', None),
             ('creators_csv', 'creators'),

renku/core/errors.py

@@ -128,7 +128,7 @@ def __init__(self, message=None):
             'Please use the "git config" command to configure it.\n\n'
             '\tgit config --set user.email "john.doe@example.com"\n'
         )
-        super(MissingUsername, self).__init__(message)
+        super().__init__(message)
 
 
 class AuthenticationError(RenkuException):

renku/core/management/datasets.py

@@ -36,7 +36,8 @@
 from renku.core import errors
 from renku.core.management.clone import clone
 from renku.core.management.config import RENKU_HOME
-from renku.core.models.datasets import Dataset, DatasetFile, DatasetTag
+from renku.core.models.datasets import Dataset, DatasetFile, DatasetTag, \
+    generate_default_short_name, is_dataset_name_valid
 from renku.core.models.git import GitURL
 from renku.core.models.locals import with_reference
 from renku.core.models.provenance.agents import Person
@@ -85,31 +86,35 @@ def datasets(self):
         result = {}
         paths = (self.path / self.renku_datasets_path).rglob(self.METADATA)
         for path in paths:
-            result[path] = self.get_dataset(path)
+            result[path] = self.load_dataset_from_path(path)
         return result
 
-    def get_dataset(self, path, commit=None):
+    def load_dataset_from_path(self, path, commit=None):
         """Return a dataset from a given path."""
+        path = Path(path)
         if not path.is_absolute():
             path = self.path / path
         return Dataset.from_yaml(path, client=self, commit=commit)
 
-    def dataset_path(self, name):
+    def get_dataset_path(self, name):
         """Get dataset path from name."""
         path = self.renku_datasets_path / name / self.METADATA
         if not path.exists():
-            path = LinkReference(
-                client=self, name='datasets/' + name
-            ).reference
+            try:
+                path = LinkReference(
+                    client=self, name='datasets/' + name
+                ).reference
+            except errors.ParameterError:
+                return None
 
         return path
 
     def load_dataset(self, name=None):
         """Load dataset reference file."""
         if name:
-            path = self.dataset_path(name)
-            if path.exists():
-                return self.get_dataset(path)
+            path = self.get_dataset_path(name)
+            if path and path.exists():
+                return self.load_dataset_from_path(path)
 
     @contextmanager
     def with_dataset(self, name=None, identifier=None, create=False):
@@ -118,50 +123,25 @@ def with_dataset(self, name=None, identifier=None, create=False):
         clean_up_required = False
 
         if dataset is None:
-            # Avoid nested datasets: name mustn't have '/' in it
-            if len(Path(name).parts) > 1:
-                raise errors.ParameterError(
-                    'Dataset name {} is not valid.'.format(name)
-                )
-
             if not create:
                 raise errors.DatasetNotFound
-            clean_up_required = True
-            dataset_ref = None
-            identifier = str(uuid.uuid4())
-            path = (self.renku_datasets_path / identifier / self.METADATA)
-            try:
-                path.parent.mkdir(parents=True, exist_ok=False)
-            except FileExistsError:
-                raise errors.DatasetExistsError(
-                    'Dataset with reference {} exists'.format(path.parent)
-                )
-
-            with with_reference(path):
-                dataset = Dataset(
-                    identifier=identifier, name=name, client=self
-                )
-
-            if name:
-                dataset_ref = LinkReference.create(
-                    client=self, name='datasets/' + name
-                )
-                dataset_ref.set_reference(path)
 
+            clean_up_required = True
+            dataset, path, dataset_ref = self.create_dataset(name)
         elif create:
             raise errors.DatasetExistsError(
                 'Dataset exists: "{}".'.format(name)
             )
 
-        dataset_path = self.path / self.datadir / dataset.name
+        dataset_path = self.path / self.datadir / dataset.short_name
         dataset_path.mkdir(parents=True, exist_ok=True)
 
         try:
             yield dataset
         except Exception:
             # TODO use a general clean-up strategy
             # https://github.com/SwissDataScienceCenter/renku-python/issues/736
-            if clean_up_required and dataset_ref:
+            if clean_up_required:
                 dataset_ref.delete()
                 shutil.rmtree(path.parent, ignore_errors=True)
             raise
@@ -174,6 +154,54 @@ def with_dataset(self, name=None, identifier=None, create=False):
 
         dataset.to_yaml()
 
+    def create_dataset(
+        self, name, short_name=None, description='', creators=()
+    ):
+        """Create a dataset."""
+        if not name:
+            raise errors.ParameterError('Dataset name must be provided.')
+
+        if not short_name:
+            short_name = generate_default_short_name(name, None)
+
+        if not is_dataset_name_valid(short_name):
+            raise errors.ParameterError(
+                'Dataset name "{}" is not valid.'.format(short_name)
+            )
+
+        if self.load_dataset(name=short_name):
+            raise errors.DatasetExistsError(
+                'Dataset exists: "{}".'.format(short_name)
+            )
+
+        identifier = str(uuid.uuid4())
+        path = (self.renku_datasets_path / identifier / self.METADATA)
+        try:
+            path.parent.mkdir(parents=True, exist_ok=False)
+        except FileExistsError:
+            raise errors.DatasetExistsError(
+                'Dataset with reference {} exists'.format(path.parent)
+            )
+
+        with with_reference(path):
+            dataset = Dataset(
+                client=self,
+                identifier=identifier,
+                name=name,
+                short_name=short_name,
+                description=description,
+                creator=creators
+            )
+
+        dataset_ref = LinkReference.create(
+            client=self, name='datasets/' + short_name
+        )
+        dataset_ref.set_reference(path)
+
+        dataset.to_yaml()
+
+        return dataset, path, dataset_ref
+
     def add_data_to_dataset(
         self,
         dataset,
@@ -186,7 +214,7 @@ def add_data_to_dataset(
     ):
         """Import the data into the data directory."""
         warning_message = ''
-        dataset_path = self.path / self.datadir / dataset.name
+        dataset_path = self.path / self.datadir / dataset.short_name
 
         destination = destination or Path('.')
         destination = self._resolve_path(dataset_path, destination)