Add filter to ent2ncr

[GaryJ]

There's been several tickets regarding the use of hexadecimal numerical entities instead of decimal numerical entities - it's part of a W3C standard that ​Content SHOULD use the hexadecimal form of character escapes rather than the decimal form when there are both.

This patch adds hexadecimal entity support to ent2ncr() for the existing named entities. It adds an optional second argument, and defaults to returning the decimal entities so is fully backwards compatible.

[GaryJ]

Adds hexadecimal entity support to ent2ncr()

[GaryJ]

If there's a better performing way to pull out the entities from the $to_ncr array that having to loop through and pull them into a new array first, then please let me know!

I thought about creating two arrays ($to_ncr and $to_hncr), but then figured the sets of named entities may get out of sync. If there's not much change in this function it may be fine to do that.

Alternatively, an ent2hncr() function may be better, which ent2ncr defers to if the 2nd argument is correctly set?

[westi]

Should is not Must.

I think a better solution is just to add a filter to this function so a plugin can implement the alternative behaviour.

[GaryJ]

Add filter on return value of ent2ncr

[GaryJ]

Add filter to allow plugin to short circuit the function

[GaryJ]

Two patches - the first is a simple filter on the return value.

The second patch approaches it a slightly different way. The issue with the first, is that for a block of text that is being running through ent2ncr(), all of the string replacements would still be done, even if they weren't then used in the plugin filter function. By letting the plugin short circuit the function (if something truthy is being returned from the plugin filter function, return that from ent2ncr() immediately), it improves performance.

[nacin]

Short-circuit is fine by me. Let's use null instead.

[GaryJ]

Check against null in short circuit.

[GaryJ]

Replying to nacin:

Short-circuit is fine by me. Let's use null instead.

Like so?

[GaryJ]

Is it too late for this filter to go in to 3.2?

[nacin]

Yes. Freeze for enhancements was about 6 weeks ago.

[nacin]

By null I meant if ( null !== $filtered = ...

[GaryJ]

Proper identity check against null in short circuit

[GaryJ]

Wasn't aware you could do a comparison against an assignment like that - handy! Patched.

[nacin]

In [18456]:

Add pre_ent2ncr filter. props GaryJ, fixes #16432.

[westi]

Replying to nacin:

In [18456]:

Add pre_ent2ncr filter. props GaryJ, fixes #16432.

This change breaks our "no clever code" coding standard. Fixing.

[westi]

In [18465]:

Make [18456] more obvious. Fixes #16432.

[scribu]

+1 for [18465]. Just because you can doesn't mean you should.