downloading chromium if using webpdf during exporter init

setup.py

@@ -95,6 +95,9 @@ def read(name):
     # optional dependencies for @serializer and @unserializer
     'joblib',
     'cloudpickle',
+
+    # for testing the webpdf converter
+    'nbconvert[webpdf]',
 ]
 
 # packages needed for development

src/ploomber/tasks/notebook.py

@@ -5,34 +5,25 @@
 import subprocess
 from subprocess import PIPE
 from pathlib import Path
-import importlib
+from importlib.util import find_spec
 import warnings
 
+import jupytext
+import nbformat
+import nbconvert
 from nbconvert import ExporterNameError
 from nbconvert.exporters.webpdf import WebPDFExporter
 
-try:
-    # papermill is importing a deprecated module from pyarrow
-    with warnings.catch_warnings():
-        warnings.simplefilter('ignore', FutureWarning)
-        import papermill as pm
-except ImportError:
-    pm = None
-
-try:
-    import jupytext
-except ImportError:
-    jupytext = None
-
-try:
-    import nbformat
-except ImportError:
-    nbformat = None
+# papermill is importing a deprecated module from pyarrow
+with warnings.catch_warnings():
+    warnings.simplefilter('ignore', FutureWarning)
+    import papermill as pm
 
 try:
-    import nbconvert
+    # this is an optional dependency
+    from pyppeteer import chromium_downloader
 except ImportError:
-    nbconvert = None
+    chromium_downloader = None
 
 from ploomber.exceptions import TaskBuildError, TaskInitializationError
 from ploomber.sources import NotebookSource
@@ -91,22 +82,30 @@ def _suggest_passing_product_dictionary():
                 'and any other output paths in other keys)')
 
 
-def _check_can_use_exporter(exporter):
+def _check_exporter(exporter, path_to_output):
+    """
+    Validate if the user can use the selected exporter
+    """
 
     if isinstance(exporter, WebPDFExporter):
-        pyppeteer_installed = importlib.util.find_spec('pyppeteer') is not None
+        pyppeteer_installed = find_spec('pyppeteer') is not None
 
         if not pyppeteer_installed:
             raise TaskInitializationError(
                 'pyppeteer is required to use '
                 'webpdf, install it '
                 'with:\npip install "nbconvert[webpdf]"')
         else:
-            from pyppeteer import chromium_downloader
-
             if not chromium_downloader.check_chromium():
                 chromium_downloader.download_chromium()
 
+        if Path(path_to_output).suffix != '.pdf':
+            raise TaskInitializationError(
+                'Expected output to have '
+                'extension .pdf when using the webpdf '
+                f'exporter, got: {path_to_output}. Change the extension '
+                'and try again')
+
 
 class NotebookConverter:
     """
@@ -169,8 +168,6 @@ def __init__(self,
 
         self._exporter = self._get_exporter(exporter_name, path_to_output)
 
-        _check_can_use_exporter(self._exporter)
-
         self._path_to_output = path_to_output
         self._nbconvert_export_kwargs = nbconvert_export_kwargs or {}
 
@@ -201,7 +198,6 @@ def _get_exporter(exporter_name, path_to_output):
         if an exporter can't be located
         """
         extension2exporter_name = {'md': 'markdown'}
-        exporter_params = {'webpdf': {'allow_chromium_download': True}}
 
         # sometimes extension does not match with the exporter name, fix
         # if needed
@@ -213,9 +209,7 @@ def _get_exporter(exporter_name, path_to_output):
         else:
             try:
                 exporter = nbconvert.get_exporter(exporter_name)
-                params = exporter_params.get(exporter_name)
-                if params:
-                    exporter = exporter(**params)
+
             # nbconvert 5.6.1 raises ValueError, beginning in version 6,
             # it raises ExporterNameError. However the exception is defined
             # since 5.6.1 so we can safely import it
@@ -232,6 +226,8 @@ def _get_exporter(exporter_name, path_to_output):
                     "Choose one from: "
                     f'{pretty_print.iterable(names)}')
 
+        _check_exporter(exporter, path_to_output)
+
         return exporter
 
     @staticmethod

tests/tasks/test_notebook.py

@@ -20,6 +20,27 @@ def fake_from_notebook_node(self, nb, resources):
     return bytes(42), None
 
 
+def test_checks_exporter(monkeypatch):
+    # simulate pyppeteer not installed
+    monkeypatch.setattr(notebook, 'find_spec', lambda _: None)
+
+    with pytest.raises(TaskInitializationError) as excinfo:
+        notebook.NotebookConverter('out.pdf', 'webpdf')
+
+    expected = 'pip install "nbconvert[webpdf]"'
+    assert expected in str(excinfo.value)
+
+
+def test_downloads_chromium_if_needed(monkeypatch):
+    mock = Mock()
+    mock.check_chromium.return_value = False
+    monkeypatch.setattr(notebook, 'chromium_downloader', mock)
+
+    notebook.NotebookConverter('out.pdf', 'webpdf')
+
+    mock.download_chromium.assert_called_once_with()
+
+
 def test_error_when_path_has_no_extension():
     with pytest.raises(TaskInitializationError) as excinfo:
         notebook.NotebookConverter('a')
@@ -52,12 +73,13 @@ def test_notebook_converter_get_exporter_from_name(exporter_name, exporter):
     assert converter._exporter == exporter
 
 
-def test_notebook_convertor_get_exporter_webpdf():
-    exporter_name = 'webpdf'
-    exporter = nbconvert.exporters.webpdf.WebPDFExporter
-    converter = notebook.NotebookConverter('file.ext', exporter_name)
-    assert isinstance(converter._exporter, exporter)
-    assert converter._exporter.allow_chromium_download is True
+def test_notebook_converter_validates_extension():
+    with pytest.raises(TaskInitializationError) as excinfo:
+        notebook.NotebookConverter('file.not_a_pdf', 'webpdf')
+
+    expected = ('Expected output to have extension .pdf when using the '
+                'webpdf exporter, got: file.not_a_pdf')
+    assert expected in str(excinfo.value)
 
 
 @pytest.mark.parametrize('output', [
@@ -620,7 +642,6 @@ def tmp_dag(tmp_directory):
 
 
 def test_develop_saves_changes(tmp_dag, monkeypatch):
-
     def mock_jupyter_notebook(args, check):
         nb = jupytext.reads('2 + 2', fmt='py')
         # args: "jupyter" {app} {path} {other args, ...}