[Solved] Failure with a REGEX

Pat64 · 2023-08-21 03:49:20

Could you tell me please why this REGEX fails (tests: https://regex101.com/r/p3q1wv/1)?

<txp:body trim="/\s?([?|:|;])/g" escape=' $1' />

Thanks in advance.

Last edited by Pat64 (2023-08-22 03:07:13)

jakob · 2023-08-21 22:32:39

I can’t test it at the moment but could it be that you need the replace attribute in place of escape?

phiw13 · 2023-08-22 01:18:52

This works for me (basic test, and after mucho trial and error till I went to double check the docs…):

trim="/\s?([?|:|;])/"

From the docs (look for “trim”)

Use /value/ to trigger regular expression matching

ps – and as jakob says: replace instead of escape

Last edited by phiw13 (2023-08-22 01:24:04)

Pat64 · 2023-08-22 03:06:46

You are absolutely right! How stupid I am… replace instead of escape

phiw13 · 2023-08-22 05:02:11

Patrick,

Do you have your (trim="/\s?([?|:|;])/g") regex actually working, that is, the replacement is actually happening? When I test it, it silently fails – check the source code, no   is inserted.

etc · 2023-08-22 08:19:17

phiw13 wrote #335675:

Do you have your (trim="/\s?([?|:|;])/g") regex actually working, that is, the replacement is actually happening? When I test it, it silently fails – check the source code, no   is inserted.

There is no g modifier in PHP regex, hence the pattern /.../g is considered invalid by txp. This should work:

<txp:body trim="/\s?([?:;])/" replace="&#160;$1" />

phiw13 · 2023-08-22 08:47:48

etc wrote #335676:

There is no g modifier in PHP regex, hence the pattern /.../g is considered invalid by txp.

That is what I had somewhere deeply buried in memory, but having seen Patrick playing more with regex, I thought I was wrong. Thanks for confirming.

<txp:body trim="/\s?([?:;])/" replace=" $1" />...

Interesting… this also catches ? but not ! and seems to do something odd with this:   – inserted in a code block. Output for that is:

&amp&#160;;#160X&#160;;

My variant above on Patricks regex works mostly I think, and even extended to deal with ? and !. It has a weakness in that it also seem to catch the character when not preceded by a standard white-space (U+0020).

Edited for clarity.

Last edited by phiw13 (2023-08-22 12:13:59)

Pat64 · 2023-08-22 12:31:24

Thanks all!

The solution provided by Oleg is perfect… Except when the text contains hyphens followed by a parenthesis before a semicolon. That’s annoying (and I can’t figure out the reason):

…) ; changed by:

… ;) ;

Edit: Nope! A strange behavior from my local server (MAMP)! All is perfect. Thanks again for your kindly help ;)

Last edited by Pat64 (2023-08-22 13:26:48)

phiw13 · 2023-08-23 00:32:58

Pat64 wrote #335678:

All is perfect. Thanks again for your kindly help ;)

Hmm, not really… Try inserting a (Textile) link, like this:

link to "forum post":https://forum.textpattern.com/viewtopic.php?pid=335677#p335677

The actual output:

<a href="https&#160;://forum.textpattern.com/viewtopic.php&#160;?pid=335677#p335677">forum</a>

Which will not be a very useful link… I think ? Inserting images has the same issue.

In both cases the regex not only finds the combination of “space” + punctuation (: or ?), but also the punctuation preceded by any character.

Pat64 · 2023-08-23 04:09:18

phiw13 wrote #335680:

Hmm, not really… Try inserting a (Textile) link, like this:

link to "forum post":https://forum.textpattern.com/viewtopic.php?pid=335677#p335677...

The actual output:

<a href="https ://forum.textpattern.com/viewtopic.php ?pid=335677#p335677">forum</a>...

Which will not be a very useful link… I think ? Inserting images has the same issue.

In both cases the regex not only finds the combination of “space” + punctuation (: or ?), but also the punctuation preceded by any character.

So the solution could be this one:

<txp:body trim="/\s([?!:;])/" replace='&#160;$1' />

Last edited by Pat64 (2023-08-23 04:09:42)

phiw13 · 2023-08-23 06:22:42

So the solution could be this one:

Hmm… I can’t break that one at the moment… :-( Looks like a winner!

Textpattern CMS

Textpattern CMS support forum

#1 2023-08-21 03:49:20