fix: always add commit to dataset if possible (#648)

* fix: always add commit to dataset if possible

addresses #646

* Update renku/models/_jsonld.py

Co-Authored-By: Sam <contact@justsam.io>

* expand kwargs

Author: rokroskar

renku/api/datasets.py

@@ -50,7 +50,7 @@ class DatasetsApiMixin(object):
     @property
     def renku_datasets_path(self):
         """Return a ``Path`` instance of Renku dataset metadata folder."""
-        return self.renku_path.joinpath(self.DATASETS)
+        return Path(self.renku_home).joinpath(self.DATASETS)
 
     def datasets_from_commit(self, commit=None):
         """Return datasets defined in a commit."""
@@ -80,11 +80,11 @@ def datasets(self):
             result[path] = self.get_dataset(path)
         return result
 
-    def get_dataset(self, path):
+    def get_dataset(self, path, commit=None):
         """Return a dataset from a given path."""
         if not path.is_absolute():
             path = self.path / path
-        return Dataset.from_yaml(path, client=self)
+        return Dataset.from_yaml(path, client=self, commit=commit)
 
     def dataset_path(self, name):
         """Get dataset path from name."""

renku/models/_jsonld.py

@@ -314,6 +314,7 @@ def from_jsonld(
         cls,
         data,
         client=None,
+        commit=None,
         __reference__=None,
         __source__=None,
     ):
@@ -331,7 +332,9 @@ def from_jsonld(
             ) != type_:
                 new_cls = cls.__type_registry__[type_]
                 if cls != new_cls:
-                    return new_cls.from_jsonld(data, client=client)
+                    return new_cls.from_jsonld(
+                        data, client=client, commit=commit
+                    )
 
         if cls._jsonld_translate:
             data = ld.compact(data, {'@context': cls._jsonld_translate})
@@ -355,6 +358,7 @@ def from_jsonld(
         data_ = {}
         if client:
             data_['client'] = client
+            data_['commit'] = commit
 
         for k, v in compacted.items():
             if k in fields:
@@ -372,7 +376,7 @@ def from_jsonld(
         return self
 
     @classmethod
-    def from_yaml(cls, path, client=None):
+    def from_yaml(cls, path, client=None, commit=None):
         """Return an instance from a YAML file."""
         import yaml
 
@@ -381,10 +385,10 @@ def from_yaml(cls, path, client=None):
             self = cls.from_jsonld(
                 source,
                 client=client,
+                commit=commit,
                 __reference__=path,
-                __source__=deepcopy(source),
+                __source__=deepcopy(source)
             )
-
         return self
 
     def asjsonld(self):

renku/models/datasets.py

@@ -467,7 +467,7 @@ def unlink_file(self, file_path):
         return self.files.pop(index)
 
     def __attrs_post_init__(self):
-        """Post-Init hook to set _id field."""
+        """Post-Init hook."""
         self._id = self.identifier
 
         if not self._label:
@@ -487,3 +487,11 @@ def __attrs_post_init__(self):
                     )
 
                     datasetfile.client = client
+
+        try:
+            self.commit = self.client.find_previous_commit(
+                self.path, revision=self.commit or 'HEAD'
+            )
+        except KeyError:
+            # if with_dataset is used, the dataset is not committed yet
+            pass

renku/models/provenance/entities.py

@@ -18,6 +18,7 @@
 """Represent provenance entities."""
 
 import weakref
+from pathlib import Path
 
 import attr
 
@@ -77,6 +78,13 @@ def default_project(self):
         if self.client:
             return self.client.project
 
+    def __attrs_post_init__(self):
+        """Post-init hook."""
+        if self.path:
+            path = Path(self.path)
+            if path.is_absolute():
+                self.path = str(path.relative_to(self.client.path))
+
 
 @jsonld.s(
     type=[

tests/test_dataset.py

@@ -88,7 +88,6 @@ def test_data_add(
         assert not os.access(
             'data/dataset/file', stat.S_IXUSR | stat.S_IXGRP | stat.S_IXOTH
         )
-        # assert os.stat('data/dataset/file/metadata.yml')
 
         # check the linking
         if scheme in ('', 'file://'):