Test jupytext/nbconvert composition

#321

tests/notebooks/md/plain_markdown.md

@@ -0,0 +1,45 @@
+---
+title: A sample document
+---
+
+This document is a plain Markdown document that was not created from a notebook.
+We use this document to test that inputing a Markdown file into Jupytext, and then converting the
+resulting notebook to a Markdown file using nbconvert, is the identity
+
+Another paragraph
+
+# A header
+
+Indented code
+
+    def f(x):
+        return 1
+
+
+    def h(x):
+        return f(x)+2
+
+
+A Python code snippet
+
+```python
+"""This code cell starts with ` ```python`"""
+1 + 1
+```
+
+Code snippet without an explicit language
+```
+echo 'Hello world'
+```
+
+Markdown comments
+<!-- #comment -->
+
+VS Code region markers
+
+<!-- #region -->
+This Markdown cell has two consecutive blank lines
+
+
+And continues here
+<!-- #endregion -->

tests/test_jupytext_nbconvert_round_trip.py

@@ -0,0 +1,79 @@
+import pytest
+import jupytext
+from jupytext.header import header_to_metadata_and_cell
+from .utils import list_notebooks, requires_nbconvert
+
+
+@requires_nbconvert
+@pytest.mark.parametrize('md_file', list_notebooks('md', skip='jupytext'))
+def test_markdown_jupytext_nbconvert_is_identity(md_file):
+    """Test that a Markdown file, converted to a notebook, then
+    exported back to Markdown with nbconvert, yields the original file"""
+
+    with open(md_file) as fp:
+        md_org = fp.read()
+
+    nb = jupytext.reads(md_org, 'md')
+    import nbconvert
+    md_nbconvert, _ = nbconvert.export(nbconvert.MarkdownExporter, nb)
+
+    # Our expectations
+    md_expected = md_org.splitlines()
+    # #region and #endregion comments are removed
+    md_expected = [line for line in md_expected if line not in ['<!-- #region -->', '<!-- #endregion -->']]
+    # language is not inserted by nbconvert
+    md_expected = ['```' if line.startswith('```') else line for line in md_expected]
+    # nbconvert inserts no empty line after the YAML header (which is in a Raw cell)
+    md_expected = '\n'.join(md_expected).replace('---\n\n', '---\n') + '\n'
+
+    # an extra blank line is inserted before code cells
+    md_nbconvert = md_nbconvert.replace('\n\n```', '\n```')
+
+    jupytext.compare.compare(md_nbconvert, md_expected)
+
+
+@requires_nbconvert
+@pytest.mark.parametrize('nb_file', list_notebooks(skip='(html|magic)'))
+def test_jupytext_markdown_similar_to_nbconvert(nb_file):
+    """Test that the nbconvert export for a notebook matches Jupytext's one"""
+
+    nb = jupytext.read(nb_file)
+
+    # Remove cell outputs and metadata
+    for cell in nb.cells:
+        cell.outputs = []
+        cell.execution_count = None
+        cell.metadata = {}
+
+    md_jupytext = jupytext.writes(nb, fmt='md')
+
+    import nbconvert
+    md_nbconvert, _ = nbconvert.export(nbconvert.MarkdownExporter, nb)
+
+    # our expectations
+
+    # nbconvert file has no YAML header
+    md_jupytext_lines = md_jupytext.splitlines()
+    _, _, raw_cell, pos = header_to_metadata_and_cell(md_jupytext_lines, '')
+    md_jupytext = '\n'.join(md_jupytext_lines[pos:]) + '\n'
+    if raw_cell is not None:
+        md_jupytext = raw_cell.source + '\n\n' + md_jupytext
+
+    # region comments are not in nbconvert
+    md_jupytext = md_jupytext.replace('<!-- #region -->\n', '').replace('<!-- #endregion -->\n', '')
+
+    # Jupytext uses HTML comments to keep track of raw cells
+    md_jupytext = (md_jupytext
+                   .replace('\n<!-- #raw -->\n', '')
+                   .replace('<!-- #raw -->\n', '')
+                   .replace('\n<!-- #endraw -->\n', ''))
+
+    # nbconvert file may start with an empty line
+    md_jupytext = md_jupytext.lstrip('\n')
+    md_nbconvert = md_nbconvert.lstrip('\n')
+
+    # Jupytext may not always have two blank lines between cells like Jupyter nbconvert
+    md_jupytext = md_jupytext.replace('\n\n\n', '\n\n')
+    md_nbconvert = md_nbconvert.replace('\n\n\n', '\n\n')
+
+    jupytext.compare.compare(md_nbconvert, md_jupytext)