Fixes #29 - ensure that code resource for notebook is specified

dataworkspaces/__init__.py

@@ -1,3 +1,3 @@
 # Copyright 2018,2019 by MPI-SWS and Data-ken Research. Licensed under Apache 2.0. See LICENSE.txt.
 
-__version__ = '1.0.0'
+__version__ = '1.0.1'

dataworkspaces/kits/jupyter.py

@@ -16,7 +16,7 @@
 import sys
 
 from dataworkspaces.lineage import LineageBuilder
-from dataworkspaces.utils.lineage_utils import infer_script_path
+from dataworkspaces.utils.lineage_utils import infer_script_path, ResourceRef
 
 
 def _get_notebook_name() -> Optional[str]:
@@ -79,6 +79,14 @@ def is_notebook() -> bool:
         return False      # Probably standard Python interpreter or a script
 
 
+def get_notebook_directory():
+    notebook_path = _get_notebook_name()
+    if notebook_path is not None:
+        return dirname(notebook_path)
+    else:
+        return curdir
+
+
 class NotebookLineageBuilder(LineageBuilder):
     """Notebooks are the final step in a pipeline
     (and potentially the only step). We customize

dataworkspaces/kits/scikit_learn.py

@@ -19,7 +19,7 @@
 from os.path import join, abspath, expanduser
 
 from dataworkspaces.lineage import LineageBuilder
-from .jupyter import is_notebook, get_step_name_for_notebook
+from .jupyter import is_notebook, get_step_name_for_notebook, get_notebook_directory
 
 
 class Metrics:
@@ -206,7 +206,9 @@ def train_and_predict_with_cv(classifier_class:ClassifierMixin,
     lb = LineageBuilder().with_parameters(lineage_params)\
                          .as_results_step(results_dir, run_description)\
                          .with_input_path(input_dir)
-    lb = lb.with_step_name(get_step_name_for_notebook()) if is_notebook()\
+    lb = lb.with_step_name(get_step_name_for_notebook())\
+           .with_code_path(get_notebook_directory()) \
+         if is_notebook() \
          else lb.as_script_step()
 
     with lb.eval() as lineage:

dataworkspaces/lineage.py

@@ -396,6 +396,10 @@ def with_no_inputs(self) -> 'LineageBuilder':
         self.no_inputs = True
         return self
 
+    def with_code_path(self, path:str) -> 'LineageBuilder':
+        self.code.append(path)
+        return self
+
     def with_code_ref(self, ref:ResourceRef) -> 'LineageBuilder':
         self.code.append(ref)
         return self