Test that searching for HTML entities works as expected

Really? I thought keywords had to be at least 3 characters to be considered anyway, and that punctuation was filtered out... We'll have to look into this...

I just tested this in Drupal 8. If I make a page with an HTML entity in the body, and verify that the entity does not display on the page itself, then search is also not displaying the entity. So this is not a bug in 8.x.

I also tested in Drupal 7.23. It is not a bug there either.

My guess is that your content does not really have an entity in it?

I also tested with your suggested content -- note that two & were missing from your original -- here is the corrected HTML source:

 Saigyo
<strong>&</strong>
&#039;s Poems
<strong>&</strong>
&quot;Upon Seeing

which renders as

Saigyo
&
's Poems
&
"Upon Seeing

This worked fine too (see screenshot).

sorry, wrong version/status

I am able to recreate this bug both in standard installs of D7 and D8 using these steps:

1. Using a standard install, create an article or page with the title "Aliquam & Qui"
2. Add some text to the body containing HTML entities in the first few lines. I added
   

   Dignissim Aliquam & Quieligo meus &lquot;natu quae quia te&rquot;. Damnum© erat— neo pneum. Facilisi feugiat ibidem ratis.

3. Check to make sure there are indeed HTML entities in the source code of the body field (I'm not sure how the above example will render here).
4. Index your content in search either by running cron or re-indexing at admin/config/search/settings
5. In search box, enter "Aliquam" and notice search result snippet renders correctly
6. In search box, enter "Aliquam & Quieligo" (type the ampersand key like a user would; do not use the HTML entity) and notice the search result snippet renders HTML entities
7. In search box, enter "Aliquam & Quieligo" (type the HTML entity) and notice that the search results render correctly again.

I think this may be related to how the search string renders the ampersand in the URL.

Hm. I tried this in Drupal 8... With the suggested text, even with the CKEditor module uninstalled and Full HTML text format, your text didn't work in the node body -- some of the entities didn't turn into characters at rendering. But I was able to get this text to work:

Dignissim Aliquam & Quieligo meus natu quae quia te. Damnum© erat— neo pneum. Facilisi feugiat ibidem ratis.

So I made a node with that in it, and ran cron. With that text, I was able to reproduce your result: if you put & in the keywords when you search, you'll get something like

<strong>&</strong>amp;

in the output.

I think there are two things going on:

a) When you search for "&", Drupal is actually searching for "& amp;" (without the space) in the indexed content, so it is more than 3 characters and is treated as an actual keyword.

b) When the excerpt is generated, it is now looking for just & and for some reason it thinks it is OK to break up an entity into pieces and wrap part of it in a STRONG tag.

So, thanks! This does seem to be a real bug and we should fix it.

This was on the wrong version.

Added summary.

Well, something changed in the last 3 months -- we've been working on some things in Search lately. I am unable to reproduce this issue in Drupal 8 now, and the test I wrote for it passed too. Now if you search for "&", you get the "You must include at least one positive keyword" message, which is appropriate. I even checked it with the search keywords minimum length set to 1 and got the same message -- I think & is punctuation, which I also think is appropriate.

So how about if we commit this small test change for 8, to make sure this stays fixed, and then move this issue to 7 for possible bug fixing (although without an active D7 search module maintainer, fixing it there is unlikely... maybe the Drupal 8 fix that fixed this will get into 7, but that is also unlikely).

I will review

Still applies and test still passes, and this is just a test added so it should be a quick review (hint hint!)

regression test looks useful for 8, and opens the door to fix 7

Updating summary so committers do not get confused about D8 status.

Weird, totally unrelated failure in Drupal\views\Tests\Plugin\RowRenderCacheTest.

Committed 9ac248b and pushed to 8.0.x. Thanks!

Thanks for adding the beta to the issue summary.

Patch attached for Drupal 7.

Interesting. I just tested on simplytest.me and the bug has indeed also vanished in D7. So let's just go ahead and commit the D7 test, which looks like a good backport. Thanks very much for the patch!

Committed to 7.x - thanks!