feat(dataset): add existing file from current repo

(addresses #99)

Author: jirikuncar

renku/api/datasets.py

@@ -141,7 +141,9 @@ def add_data_to_dataset(
         ignored = self.find_ignored_paths(
             *[
                 os.path.relpath(
-                    str(self.renku_datasets_path / dataset.name / key),
+                    str(
+                        self.renku_datasets_path / dataset.identifier.hex / key
+                    ),
                     start=str(self.path),
                 ) for key in files.keys()
             ]
@@ -252,6 +254,30 @@ def _add_from_git(self, dataset, path, url, target, **kwargs):
         relative_to = kwargs.get('relative_to', None)
 
         if u.scheme in ('', 'file'):
+            try:
+                relative_url = Path(url).resolve().relative_to(self.path)
+            except Exception:
+                relative_url = None
+
+            if relative_url:
+                result = str(
+                    os.path.relpath(
+                        str(relative_url),
+                        start=str(
+                            self.renku_datasets_path / dataset.identifier.hex
+                        ),
+                    )
+                )
+                return {
+                    result:
+                        DatasetFile(
+                            path=result,
+                            url=url,
+                            authors=dataset.authors,
+                            dataset=dataset.name,
+                        )
+                }
+
             warnings.warn('Importing local git repository, use HTTPS')
             # determine where is the base repo path
             r = Repo(url, search_parent_directories=True)

tests/cli/test_dataset.py renamed to tests/cli/test_datasets.py

@@ -123,6 +123,31 @@ def test_multiple_file_to_dataset(
     assert result.exit_code == 0
 
 
+def test_repository_file_to_dataset(runner, project, client):
+    """Test adding a file from the repository into a dataset."""
+    # create a dataset
+    result = runner.invoke(cli.cli, ['dataset', 'create', 'dataset'])
+    assert result.exit_code == 0
+
+    with (client.path / 'a').open('w') as fp:
+        fp.write('a')
+
+    client.repo.git.add('a')
+    client.repo.git.commit(message='Added file a')
+
+    # add data
+    result = runner.invoke(
+        cli.cli,
+        ['dataset', 'add', 'dataset', 'a'],
+        catch_exceptions=False,
+    )
+    assert result.exit_code == 0
+
+    with client.with_dataset('dataset') as dataset:
+        assert dataset.name == 'dataset'
+        assert '../../../a' in dataset.files
+
+
 def test_relative_import_to_dataset(
     tmpdir, data_repository, runner, project, client
 ):