def __init__(self, filename=None, markdown=None):
"""Perform validation on a Markdown document.
Validator accepts either the path to a file containing Markdown,
OR a valid Markdown string. The latter is useful for unit testing."""
self.filename = filename
if filename:
# Expect Markdown files to be in same directory as the input file
self.markdown_dir = os.path.dirname(filename)
self.lesson_dir = self.markdown_dir
with open(filename, 'rU') as f:
self.markdown = f.read()
else:
# Look for linked content in ../pages (relative to this file)
self.lesson_dir = os.path.abspath(
os.path.join(os.path.dirname(__file__), os.pardir))
self.markdown_dir = self.lesson_dir
self.markdown = markdown
ast = self._parse_markdown(self.markdown)
self.ast = vh.CommonMarkHelper(ast)
# Keep track of how many times callout box styles are used
self._callout_counts = collections.Counter()
def _validate_one_html_link(self, link_node, check_text=False):
"""
Any local html file being linked was generated as part of the lesson.
Therefore, file links (.html) must have a Markdown file
in the expected folder.
The title of the linked Markdown document should match the link text.
"""
dest, link_text = self.ast.get_link_info(link_node)
# HTML files in same folder are made from Markdown; special tests
fn = dest.split("#")[0] # Split anchor name from filename
expected_md_fn = os.path.splitext(fn)[0] + os.extsep + "md"
expected_md_path = os.path.join(self.markdown_dir,
expected_md_fn)
if not os.path.isfile(expected_md_path):
logging.error(
"In {0}: "
"The document links to {1}, but could not find "
"the expected markdown file {2}".format(
self.filename, fn, expected_md_path))
return False
if check_text is True:
# If file exists, parse and validate link text = node title
with open(expected_md_path, 'rU') as link_dest_file:
dest_contents = link_dest_file.read()
dest_ast = self._parse_markdown(dest_contents)
dest_ast = vh.CommonMarkHelper(dest_ast)
dest_page_title = dest_ast.get_doc_header_subtitle()
if dest_page_title != link_text:
logging.error(
"In {0}: "
"The linked page {1} exists, but "
"the link text '{2}' does not match the "
"(sub)title of that page, '{3}'.".format(
self.filename, dest,
link_text, dest_page_title))
return False
return True
