Performance

OK. This is a better way: define a custom TreeWalker which converts named entities to utf-8 as it goes. This avoids having to do an extra tree traversal in sanitize_rexml, AND avoids the trainwreck that is html5/inputstream.rb.
This commit is contained in:
Jacques Distler 2007-10-14 21:07:46 -05:00
parent 198d7847bd
commit 1911d18f65
3 changed files with 81 additions and 12 deletions

View file

@ -164,14 +164,14 @@ class PageRendererTest < Test::Unit::TestCase
# wikiwords are invalid as styles, must be in "name: value" form
def test_content_with_wikiword_in_style_tag
assert_markup_parsed_as(
"<p>That is some <em style=''>Stylish Emphasis</em></p>",
'<p>That is some <em style="">Stylish Emphasis</em></p>',
'That is some <em style="WikiWord">Stylish Emphasis</em>')
end
# validates format of style..
def test_content_with_valid_style_in_style_tag
assert_markup_parsed_as(
"<p>That is some <em style='text-align: right;'>Stylish Emphasis</em></p>",
'<p>That is some <em style="text-align: right;">Stylish Emphasis</em></p>',
'That is some <em style="text-align: right">Stylish Emphasis</em>')
end
@ -199,24 +199,24 @@ class PageRendererTest < Test::Unit::TestCase
def test_content_with_link_in_parentheses
assert_markup_parsed_as(
"<p>(<a href='http://wiki.org/wiki.cgi?WhatIsWiki'>What is a wiki?</a>)</p>",
'<p>(<a href="http://wiki.org/wiki.cgi?WhatIsWiki">What is a wiki?</a>)</p>',
'([What is a wiki?](http://wiki.org/wiki.cgi?WhatIsWiki))')
end
def test_content_with_image_link
assert_markup_parsed_as(
"<p>This <img src='http://hobix.com/sample.jpg' alt=''/> is a Markdown image link.</p>",
'<p>This <img alt="" src="http://hobix.com/sample.jpg" /> is a Markdown image link.</p>',
'This ![](http://hobix.com/sample.jpg) is a Markdown image link.')
end
def test_content_with_inlined_img_tag
assert_markup_parsed_as(
"<p>This <img src='http://hobix.com/sample.jpg' alt=''/> is an inline image link.</p>",
'<p>This <img alt="" src="http://hobix.com/sample.jpg" /> is an inline image link.</p>',
'This <img src="http://hobix.com/sample.jpg" alt="" /> is an inline image link.')
# currently, upper case HTML elements are not allowed
assert_markup_parsed_as(
'<p>This &lt;IMG SRC="http://hobix.com/sample.jpg" alt=""/&gt; is an inline image link.</p>',
'<p>This &lt;IMG SRC="http://hobix.com/sample.jpg" alt=""&gt;&lt;/IMG&gt; is an inline image link.</p>',
'This <IMG SRC="http://hobix.com/sample.jpg" alt="" /> is an inline image link.')
end
@ -361,7 +361,7 @@ class PageRendererTest < Test::Unit::TestCase
EOL
assert_markup_parsed_as(
"<ul>\n<li><a href='~b'>a</a></li>\n\n<li>c~ d</li>\n</ul>",
"<ul>\n<li><a href=\"~b\">a</a></li>\n\n<li>c~ d</li>\n</ul>",
list_with_tildas)
end