[SOLVED] trim bug in shortcode?

colak · 2024-01-06 03:07:18

I have the following shortcode for external links

<a<txp:if_yield name="nf"> rel="nofollow noopener"
<txp:else /> 
rel="external noopener"</txp:if_yield>
<txp:if_yield name="lang"> lang="<txp:yield name="lang" />"</txp:if_yield> 
href="<txp:yield name="url" />"<txp:if_yield name="email"> style="color:#ba0000;text-decoration:none;"</txp:if_yield>
<txp:if_yield name="class"> class="<txp:yield name="class" />"</txp:if_yield>>
<txp:if_yield name="txt"><txp:yield name="txt" /><txp:else />
<txp:yield name="url" trim="https?://www.?" /></txp:if_yield></a>

It all works just fine except, there appears to be a strange bug with the trim function.

When I use the shortcode like <txp::ext url="https://domain.tld" /> it all works fine and parses as it should: <a rel="external noopener" href="https://domain.tld">domain.tld</a>

If I use it the same way but the site domain starts with a t, the following happens

shortcode: <txp::ext url="https://tomain.tld" />

Parses to <a rel="external noopener" href="https://tomain.tld">omain.tld</a>, trimming the first t.

Any ideas why this is happening?

etc · 2024-01-06 10:56:29

It’s not really a bug, I’m surprised it works at all. To trigger regex mode, trim must be a valid delimited regex pattern, like #^https?://(www\.)?#i. Otherwise, it will just progressively strip all provided characters (including t) at url extremities.

colak · 2024-01-06 16:14:48

etc wrote #336297:

It’s not really a bug, I’m surprised it works at all. To trigger regex mode, trim must be a valid delimited regex pattern, like #^https?://(www\.)?#i. Otherwise, it will just progressively strip all provided characters (including t) at url extremities.

#^https?://(www\.)?#i did indeed fix it! I had the trim function for some time but I had no urls that started or finished the a t so I never noticed my error.

Thanks so much Oleg!

colak · 2024-11-12 05:08:46

I’m returning to this as some of the urls create long horizontal scrolls on mobile devices, so more trimming is needed.

The current simplified tag structure is:

<a rel="external noopener" href="<txp:yield name="url" />">
<txp:if_yield name="txt">
<txp:yield name="txt" />
<txp:else />
<txp:yield name="url" trim="#^https?://(www\.)?#i" />
</txp:if_yield>
</a>

Is there a way to limit the characters of the urls to 30 characters when they appear as linked text and replace those trimmed characters with replace="…?

etc · 2024-11-12 09:14:43

Something like this?

<txp:yield name="url" trim="#^https?://(?:www\.)?(.{5,25})\b.*$#is" replace="$1&hellip;" />

colak · 2024-11-12 13:52:36

etc wrote #338195:

Something like this?

<txp:yield name="url" trim="#^https?://(?:www\.)?(.{5,25})\b.*$#is" replace="$1…" />...

Indeed!!! Thanks!! I can understand the 25 in the structure, but what is the first 5 for?

etc · 2024-11-12 17:21:11

colak wrote #338198:

I can understand the 25 in the structure, but what is the first 5 for?

I guess for avoiding short urls truncation, but it does not seem very reliable. Maybe this?

/^https?:\/\/(?:www\.)?(.{25}(?:(?<=\w)\w)*).{5,}$/s

It is expected to skip 25 characters and stop at the first non-word character after it, if the remaining chunk is at least 5 characters long. But ~~Google~~ ChatGPT might be able to provide a better pattern :-)

jakob · 2024-11-12 18:31:28

Could it be that you’re missing a { at .{25}, e.g.

/^https?:\/\/(?:www\.)?(.{25}(?:(?<=\w)\w*|\b)).{5,}$/s

It’s an interesting idea, but for long domain names, 25 characters is quite limiting making the links essentially indistinguishable.

You can play around with it here on regex101.

etc · 2024-11-12 18:38:22

Oops, thank you, corrected.

You can add a ‘tail’:

<txp:yield name="url"
    trim="/^https?:\/\/(?:www\.)?(.{25}(?:(?<=\w)\w)*).{5,}(.\b.{5,})$/s"
    replace="$1&hellip;$2"
/>

But make sure first that url is long enough to match the pattern. It makes me think that <txp:if_yield /> needs match attribute, with pattern among other options.

jakob · 2024-11-12 23:27:38

Excellent. That’s a good idea for differentiating links that begin similarly and would otherwise look identical when truncated. I also learned some more regex from that, too.

colak · 2024-11-13 04:15:35

etc wrote #338202:

Oops, thank you, corrected.

You can add a ‘tail’:

<txp:yield name="url"...

But make sure first that url is long enough to match the pattern. It makes me think that <txp:if_yield /> needs match attribute, with pattern among other options.

This is indeed excellent! I particularly like how the end of the url is preserved, which is very important for links that direct to pdfs. The only minor issue that I have detected is that short urls do not get their first part trimmed. ie. https://www.sitesitesite.tld remains the same.

etc · 2024-11-13 11:08:32

colak wrote #338206:

The only minor issue that I have detected is that short urls do not get their first part trimmed. ie. https://www.sitesitesite.tld remains the same.

Yep, that’s what I’m alluding to by saying ‘long enough’. You can separate two cases:

<txp:if_request name='<txp:yield name="url" />' type="name" value="^.{50}" match="pattern">
    <txp:yield name="url" trim="/^https?:\/\/(?:www\.)?(.{25}(?:(?<=\w)\w)*).{5,}(.\b.{5,})$/s" replace="$1&hellip;$2" />
<txp:else />
    <txp:yield name="url" trim="/^https?:\/\/(?:www\.)?/i" />
</txp:if_request>

Or trim twice:

<txp:evaluate trim="/^\s*(.{25}(?:(?<=\w)\w)*).{5,}(.\b\S{5,})\s*$/s" replace="$1&hellip;$2">
    <txp:yield name="url" trim="/^https?:\/\/(?:www\.)?/i" />
</txp:evaluate>

And if one day conditional replacements are finally implemented in php, a universal pattern will be possible.

Textpattern CMS

Textpattern CMS support forum

#1 2024-01-06 03:07:18