Item13134: HTML::Tidy fails to validate html5

HTMLValtidation tests currently fail mostly due to HTML::Tidy's incapability to deal with HTML5.

Example:

<meta charset="utf-8" />

... produces

line 5 column 19 - Warning: <meta> proprietary attribute "charset"
line 5 column 19 - Warning: <meta> lacks "content" attribute

... which is obviously wrong.

Similarly any custom data-* as introduced by HTML5 will cause validation warnings.

See also Item10739. Paul's approach was to filter tidy's output and ignore those warnings.

Validating HTML5 (under perl) doesn't seem to be so easy. Alternatives are:

• https://github.com/w3c/tidy-html5 ... unmaintained for a couple of years now ... lots of github forks ... none of them merged back
• https://github.com/geoffmcl/tidy-fork ... seems to be the latest tidy5 tool
• HTML::HTML5::Parser / Writer ... no validation
• ??? ... what else is there

Note that there are more reasons to look into replacing HTML::Tidy with another HTML validator. For example, building HTML::Tidy using cpanm requires tidyp, yet another fork of the original tidy project. However tidyp isn't shipped by distributions. So you have to download and compile tidyp yourself before cpanm-ing HTML::Tidy.

-- MichaelDaum - 05 Dec 2014

Best approach atm is to filter out these false alarms. I have downloaded the html5 aware tidy version to check that no problems persist.

-- MichaelDaum - 13 Dec 2014

Task::HTML5 could be an option

-- MichaelDaum - 26 Feb 2016

Task::HTML5 has seen its latest release in 2011, so I assume it is not maintained anymore. The latest tool is called HTML::Tidy5 as per https://github.com/petdance/html-tidy5 and http://www.html-tidy.org/

However it does not comple ootb: https://github.com/petdance/html-tidy5/issues/3 as it requires tidy-5.6.0 ...yet debian and ubuntu only have 5.2.0.

The tidy commandline tool seems to be fine though.

-- MichaelDaum - 02 Jun 2019

HTML::Tidy5 is probably the correct route. it is from the same author as HTML::Tidy and depends on htmltidy which both appear to be actively maintained

-- TimothyLegge - 23 Feb 2020

Commenting is disabled while not logged in