fixes doc building

docs/source/hooks.py

@@ -1,8 +1,50 @@
 from os import environ
 from pathlib import Path
+
+import jupytext
+import nbformat
 from ploomber import DAG
 from ploomber.tasks import NotebookRunner
 from ploomber.products import File
+from ploomber.constants import TaskStatus
+
+
+def make_task(dag, path_in, path_out):
+    nb = jupytext.read(path_in)
+
+    fmt = nbformat.versions[nbformat.current_nbformat]
+    nb.cells.append(fmt.new_code_cell(metadata=dict(tags=['parameters'])))
+
+    name = Path(path_in).name.split('.')[0]
+    path_preprocessed = Path(path_in).parent / (name + '-preprocessed.ipynb')
+    nbformat.write(nb, path_preprocessed)
+
+    NotebookRunner(Path(path_preprocessed),
+                   File(path_out),
+                   dag,
+                   kernelspec_name='python3',
+                   name=name,
+                   local_execution=True)
+
+
+def remove_with_tag(nb, tag):
+    idx = None
+
+    for i, cell in enumerate(nb.cells):
+        if tag in cell.metadata.tags:
+            idx = i
+            break
+
+    nb.cells.pop(idx)
+
+
+def post_process_nb(path):
+    nb = jupytext.read(path)
+
+    remove_with_tag(nb, 'injected-parameters')
+    remove_with_tag(nb, 'parameters')
+
+    jupytext.write(nb, path)
 
 
 def config_init(app, config):
@@ -15,15 +57,16 @@ def config_init(app, config):
 
     dag = DAG()
 
-    NotebookRunner(base_path / 'nbs/SQLiteTracker.md',
-                   File(base_path / 'user_guide/SQLiteTracker.ipynb'),
-                   dag=dag,
-                   kernelspec_name='python3')
+    make_task(dag, base_path / 'nbs/SQLiteTracker.md',
+              base_path / 'user_guide/SQLiteTracker.ipynb')
 
-    NotebookRunner(base_path / 'nbs/NotebookCollection.py',
-                   File(base_path / 'user_guide/NotebookCollection.ipynb'),
-                   dag=dag,
-                   kernelspec_name='python3',
-                   local_execution=True)
+    make_task(dag, base_path / 'nbs/NotebookCollection.py',
+              base_path / 'user_guide/NotebookCollection.ipynb')
 
     dag.build()
+
+    for task_name in dag:
+        task = dag[task_name]
+
+        if task.exec_status == TaskStatus.Executed:
+            post_process_nb(str(task.product))

docs/source/nbs/train.py

@@ -25,7 +25,7 @@
 
 # + tags=["parameters"]
 model = 'sklearn.ensemble.RandomForestRegressor'
-params = {'min_samples_leaf': 1,  'n_estimators': 50}
+params = {'min_samples_leaf': 1, 'n_estimators': 50}
 
 # + tags=["model_name"]
 model
@@ -40,7 +40,7 @@
 class_ = getattr(importlib.import_module(module), name)
 
 # + tags=["feature_names"]
-d['feature_names']
+list(d['feature_names'])
 # -
 
 X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33)
@@ -85,9 +85,11 @@
 error_river = df.groupby('CHAS')[['error_abs', 'error_sq']].mean()
 error_river.columns = ['mae', 'mse']
 
+
 def r2_score(df):
     return metrics.r2_score(df.y_true, df.y_pred)
 
+
 r2 = pd.DataFrame(df.groupby('CHAS').apply(r2_score))
 r2.columns = ['r2']
 

src/sklearn_evaluation/nb/NotebookIntrospector.py

@@ -16,7 +16,7 @@ def _safe_literal_eval(source, to_df=False, none_if_error=False):
 
         return result
 
-    except SyntaxError:
+    except (SyntaxError, ValueError):
         return None if none_if_error else source