Image title and alt attribute content should be texturized.

[stephdau]

gallery_shortcode() texturizes the caption shown underneath images in galleries.

For consistency, alt and title tags content should also be texturized.
This is also valuable for developers extending the gallery shortcode or output, such as with the WordPress.com (and Jetpack) ​Gallery Carousel feature, as it provides i18n'd texturization, for EG.

See attached patch, which:

• uses wptexturize() in wp_get_attachment_image() directly (/wp-includes/media.php), which makes it work with gallries, attachment pages, etc.
• also uses wptexturize() in get_image_tag() (/wp-includes/media.php), for consistency.
• uses wptexturize() in wp_get_attachment_link() (/wp-includespost-template.php), for consistency

[toscho]

Do not hard code the function, use a apply_filters() instead. See #19550.

[ericlewis]

Why is esc_attr() not enough already? What's the benefit of wp_texturize()?

[adamsilverstein]

Replying to toscho:

Do not hard code the function, use a apply_filters() instead. See #19550.

Based on #19550 and r28715, wp_texturize already has a filter to allow you to turn it off entirely, that seems sufficient.

[adamsilverstein]

Replying to ericlewis:

Why is esc_attr() not enough already? What's the benefit of wp_texturize()?

these do different things entirely - esc_attr is security related, stripping scripts, etc. wptexturize transforms text - "quotes to smart quotes, apostrophes, dashes, ellipses, the trademark symbol, and the multiplication symbol. "

I think would be nice to see this improved typography in alt and title tags, although I'm a bit wary of potential pitfalls or unexpected results.

@brianhogg is working on an updated patch here at #wcto. The current patch no longer applies cleanly.

[brianhogg]

patch refresh

[wonderboymusic]

This should have tests to show what is not happening now and what should happen when the patch is applied.

[welcher]

First pass unit test

[welcher]

I've added some initial unit tests. They are my first core tests so any feedback is appreciated. I am re-using some of the data providers already in the class and if that approach makes sense, I'll go back and add the rest.

[adamsilverstein]

Looks good @welcher, I haven't seen that @dataProvider technique before, excellent.

I verified tests failed before patch and pass after...

In 21221.4.diff​ I refreshed @brianhogg's patch and combined with your tests, plus made some tiny code standards fixes (missing space near closing bracket of unit test functions).

Replying to welcher:

I've added some initial unit tests. They are my first core tests so any feedback is appreciated. I am re-using some of the data providers already in the class and if that approach makes sense, I'll go back and add the rest.

[miqrogroove]

21221.4.diff seems safe, and I have some big picture questions about it:

• Why does an alt attribute need to be texturized?
• Why use an HTML parser in an attribute context?
• Is it okay to not texturize title attributes when they contain angle braces?

[joedolson]

I don't have answers for the second two questions; but for the first, the problem was for internationalization. Without texturizing, multibyte characters in attributes weren't expressed correctly.

[miqrogroove]

wptexturize is an ASCII only search and replace algorithm. What does this have to do with multibyte characters?

[joedolson]

Never mind then! I thought that was @stephdau's original intent from my prior review of this ticket, but if it's not a valid reason, then I'm not sure. I've never looked at this ticket in significant depth.

[adamsilverstein]

@stephdau: can you clarify this part: it provides i18n'd texturization, for EG

[DrewAPicture]

Looks like the feedback from @miqrogroove hasn't yet been fully addressed. Pushing to a future release.

[joedolson]

Just did some testing with screen readers, and this doesn't have any bearing on accessibility. The characters are read correctly whether they're texturized or not. Removing accessibility focus.