def test_site_root_directive_href():
"""Test links with $ssServerRelativeSiteRoot directives."""
result = html_to_markdown(
'<a href="[!--$ssServerRelativeSiteRoot--]A/Bunch/Of/Things/'
'ABC123?useSecondary=&showpage=456">'
"Page 456 of ABC 123</a>")
assert "[Page 456 of ABC 123](/content/ABC123_456)" in result
# Test with PDF content.
md_converter = learning_importer.MHRAMarkdownConverter(
content_prefix="/content/", asset_prefix="/asset/")
result = md_converter.convert(
'<a href="[!--$ssServerRelativeSiteRoot--]Opendocuments/OpenPDFdocuments/'
'ABC123">ABC 123 document</a>')
assert "[ABC 123 document](/asset/abc123.pdf)" in result
assert md_converter.stellent_assets_to_download == set(["ABC123"])
assert md_converter.assets_with_unknown_type == set()
# Test with known content redirection.
result = html_to_markdown('<a href="[!--$ssServerRelativeSiteRoot--]'
'Safetyinformation/DrugSafetyUpdate/CON123123">'
"Known content redirection</a>")
assert ("[Known content redirection]"
"(https://www.gov.uk/drug-safety-update/"
"addiction-to-benzodiazepines-and-codeine)") in result
# Test with unknown content.
md_converter = learning_importer.MHRAMarkdownConverter(
content_prefix="/content/", asset_prefix="/asset/")
result = md_converter.convert(
'<a href="[!--$ssServerRelativeSiteRoot--]A/Bunch/Of/Things/'
'ABC123">ABC 123 thing</a>')
assert "[ABC 123 thing](/asset/abc123.unknown)" in result
assert md_converter.stellent_assets_to_download == set(["ABC123"])
assert md_converter.assets_with_unknown_type == set(["ABC123"])
def test_http_relative_web_root_directive_href():
"""Test links with $HttpRelativeWebRoot directives."""
md_converter = learning_importer.MHRAMarkdownConverter(
content_prefix="/content/", asset_prefix="/asset/")
result = md_converter.convert(
"<a href='[!--$HttpRelativeWebRoot--]/something/abc123.pdf'>"
"ABC 123 document</a>")
assert "[ABC 123 document](/asset/abc123.pdf)" in result
assert md_converter.stellent_assets_to_download == set(["abc123"])
def test_web_layout_url_src():
"""Test image URLs with $ssWeblayoutUrl directives."""
md_converter = learning_importer.MHRAMarkdownConverter(
content_prefix="/content/", asset_prefix="/asset/")
result = md_converter.convert(
"<img src=\"[!--$ssWeblayoutUrl('ab/cd/abc123.jpg')--]\" "
"alt='ABC 123' title='Image for ABC 123' />"
"<img src=\"[!--$ssWeblayoutUrl('ab/cd/abc123.jpg')--]\" />"
"<img src=\"[!--$ssWeblayoutUrl('ab/cd/xyz789.jpg')--]\" />")
assert '' in result
assert md_converter.stellent_assets_to_download == set(
["abc123", "xyz789"])
def html_to_markdown(html):
"""Convert HTML to Markdown."""
md_converter = learning_importer.MHRAMarkdownConverter(
content_prefix="/content/", asset_prefix="/asset/"
)
return md_converter.convert(html)