Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.lang.javascript > #18886 > unrolled thread

RegExp problem

Back to article view | Back to comp.lang.javascript

Contents

  RegExp problem John English <john.foreign@gmail.com> - 2013-03-14 04:56 -0700
    Re: RegExp problem "Jukka K. Korpela" <jkorpela@cs.tut.fi> - 2013-03-14 14:22 +0200
      Re: RegExp problem "Evertjan." <exxjxw.hannivoort@inter.nl.net> - 2013-03-14 14:52 +0100
        Re: RegExp problem "Jukka K. Korpela" <jkorpela@cs.tut.fi> - 2013-03-14 16:08 +0200
        Re: RegExp problem Stefan Weiss <krewecherl@gmail.com> - 2013-03-14 15:25 +0100
          Re: RegExp problem Thomas 'PointedEars' Lahn <PointedEars@web.de> - 2013-03-14 16:47 +0100
            Re: RegExp problem Thomas 'PointedEars' Lahn <PointedEars@web.de> - 2013-03-14 21:11 +0100
          Re: RegExp problem "Evertjan." <exxjxw.hannivoort@inter.nl.net> - 2013-03-14 19:54 +0100
    Re: RegExp problem "Evertjan." <exxjxw.hannivoort@inter.nl.net> - 2013-03-14 13:26 +0100
    Re: RegExp problem Thomas 'PointedEars' Lahn <PointedEars@web.de> - 2013-03-14 17:31 +0100

#18886 — RegExp problem

I have this string in JavaScript:

  var s = "</p><ol><li>First\n</li><li>Second\n</li></ol><p>";

Then I do this (to remove the outer "</p>...<p>"):

  s = s.replace(/^<\/([^> ]+)[^>]*>(.*)<\1>$/,"$2");

Nothing happens (s is unchanged, and if I use match() instead of replace it returns false), but if I go try it at http://www.regular-expressions.info/javascriptexample.html, it works!

I've tried all sorts of things (creating a separate regExp object, using //g, taking out the ^$, replacing [^> ]+ with [a-z0-9]*...) but nothing makes any difference.

It's driving me nuts. Can anyone tell me what I'm doing wrong here?

TIA...

[toc] | [next] | [standalone]

#18887

2013-03-14 13:56, John English wrote:

> I have this string in JavaScript:
>
> var s = "</p><ol><li>First\n</li><li>Second\n</li></ol><p>";
>
> Then I do this (to remove the outer "</p>...<p>"):
>
> s = s.replace(/^<\/([^> ]+)[^>]*>(.*)<\1>$/,"$2");
>
> Nothing happens

The period "." does not match newline characters, and the data string 
contains two newlines (\n). To work around this, replace (.*) by
((.|\n)*)

Using regular expressions to process HTML fragments is generally clumsy 
and fragile. It might be better to operate on the DOM tree rather than 
serialized HTML. However, I don't quite see the point of this particular 
operation. Removing the optional </p> tag does not change document 
structure but may affect rendering due to browser bugs.

-- 
Yucca, http://www.cs.tut.fi/~jkorpela/

[toc] | [prev] | [next] | [standalone]

#18889

Jukka K. Korpela wrote on 14 mrt 2013 in comp.lang.javascript:

> The period "." does not match newline characters, and the data string 
> contains two newlines (\n). To work around this, replace (.*) by
> ((.|\n)*)

or:

[.\n]*

or my personal favorite for "any character": 

[\s\S]*


/..../m should work, but it doesn't, I believe.

-- 
Evertjan.
The Netherlands.
(Please change the x'es to dots in my emailaddress)

[toc] | [prev] | [next] | [standalone]

#18890

2013-03-14 15:52, "Evertjan." wrote:

> /..../m should work, but it doesn't, I believe.

That's what I though too, but having found out that it does not work, I 
checked this out and noticed that neither should it work. The effect of 
"m" flag is just that "^" and "$" match start of line and end of line, 
respectively, in addition to matching start and end of the string. It 
does not affect the meaning of ".".

-- 
Yucca, http://www.cs.tut.fi/~jkorpela/

[toc] | [prev] | [next] | [standalone]

#18891

On 2013-03-14 14:52, Evertjan. wrote:
> Jukka K. Korpela wrote on 14 mrt 2013 in comp.lang.javascript:
> 
>> The period "." does not match newline characters, and the data string 
>> contains two newlines (\n). To work around this, replace (.*) by
>> ((.|\n)*)
> 
> or:
> 
> [.\n]*

That's a literal dot character.
It does nothing special inside a character class.

> or my personal favorite for "any character": 
> 
> [\s\S]*

Shortest way to match "any character including \n":
[^]


- stefan

[toc] | [prev] | [next] | [standalone]

#18892

Stefan Weiss wrote:

> On 2013-03-14 14:52, Evertjan. wrote:
>> or my personal favorite for "any character":
>> 
>> [\s\S]*
> 
> Shortest way to match "any character including \n":
> [^]

My research showed that this is a syntax error in Microsoft JScript in 
version 5.8.23141 (IE/MSHTML 8.0) and before (and probably in version 
9.0.16440 in Browser Compatibility Mode, the same as with trailing commas in 
Object and Array initialisers), and in Opera ECMAScript in version 9.64 and 
before.  [\s\S] is the safe approach.

-- 
PointedEars

Twitter: @PointedEars2
Please do not Cc: me. / Bitte keine Kopien per E-Mail.

[toc] | [prev] | [next] | [standalone]

#18896

Thomas 'PointedEars' Lahn wrote:

> Stefan Weiss wrote:
>> On 2013-03-14 14:52, Evertjan. wrote:
>>> or my personal favorite for "any character":
>>> 
>>> [\s\S]*
>> 
>> Shortest way to match "any character including \n":
>> [^]
> 
> My research showed that this is a syntax error in Microsoft JScript in
> version 5.8.23141 (IE/MSHTML 8.0) and before (and probably in version
> 9.0.16440 in Browser Compatibility Mode, the same as with trailing commas
> in Object and Array initialisers), and in Opera ECMAScript in version 9.64
> and before.  [\s\S] is the safe approach.

However, it should be noted that JSX:regexp.js [1, 2] supports PCRE_DOTALL 
since revision 275 (2012-09-19 CE) with

  var rx = new jsx.regexp.RegExp(/./, "s");

or 

  var rx = new jsx.regexp.RegExp(".", "s");

being equivalent to

  var rx = /[\S\s]/;

The pattern-match modifiers “(?s)” and “(?-s)" are also supported with a 
first String argument.  (But they apply to the rest of the expression, not 
only the current group, as of yet.)

This is particularly useful if you have many any-character occurrences in 
the expression.  The “.” will _not_ be expanded in a character class, so as 
to retain its original meaning:

  var rx = new jsx.regexp.RegExp(/.[.]./, "s");

is equivalent to

  var rx = /[\S\s][.][\S\s]/;

jsx.regexp.RegExp() is a factory/constructor that returns a reference to an 
augmented RegExp instance.  So from here everything in your script can stay 
the same (global, case-insensitive, multi-line, and other flags are 
supported by the function as well).  You may carefully import it into the 
global namespace, like

  jsx._import(jsx.regexp, ["RegExp"], null, ["RegExp2"]);  // [3]

or a variable declaration if you want to keep it efficient but local.

________
[1] <http://PointedEars.de/scripts/test/regexp>
[2] 
<http://pointedears.de/websvn/filedetails.php?repname=JSX&path=%2Ftrunk%2Fregexp.js>
[3] 
<http://pointedears.de/websvn/filedetails.php?repname=JSX&path=%2Ftrunk%2Fobject.js>
-- 
PointedEars

Twitter: @PointedEars2
Please do not Cc: me. / Bitte keine Kopien per E-Mail.

[toc] | [prev] | [next] | [standalone]

#18894

Stefan Weiss wrote on 14 mrt 2013 in comp.lang.javascript:

> On 2013-03-14 14:52, Evertjan. wrote:
>> Jukka K. Korpela wrote on 14 mrt 2013 in comp.lang.javascript:
>> 
>>> The period "." does not match newline characters, and the data string 
>>> contains two newlines (\n). To work around this, replace (.*) by
>>> ((.|\n)*)
>> 
>> or:
>> 
>> [.\n]*

right, sorry.
 
> That's a literal dot character.
> It does nothing special inside a character class.
> 
>> or my personal favorite for "any character": 
>> 
>> [\s\S]*
> 
> Shortest way to match "any character including \n":
> [^]
> 
> 
> - stefan
> 



-- 
Evertjan.
The Netherlands.
(Please change the x'es to dots in my emailaddress)

[toc] | [prev] | [next] | [standalone]

#18888

John English wrote on 14 mrt 2013 in comp.lang.javascript:

> I have this string in JavaScript:
> 
>   var s = "</p><ol><li>First\n</li><li>Second\n</li></ol><p>";

A javascript variable containing a html string.

> 
> Then I do this (to remove the outer "</p>...<p>"):
> 
>   s = s.replace(/^<\/([^> ]+)[^>]*>(.*)<\1>$/,"$2");


s = s.replace(/^<\/p>|<p>$/g,'');

[..]

> It's driving me nuts. Can anyone tell me what I'm doing wrong here?

Never drive your nuts, javascript is not golf, PE should agree.


-- 
Evertjan.
The Netherlands.
(Please change the x'es to dots in my emailaddress)

[toc] | [prev] | [next] | [standalone]

#18893

John English wrote:

> I have this string in JavaScript:
> 
>   var s = "</p><ol><li>First\n</li><li>Second\n</li></ol><p>";
> 
> Then I do this (to remove the outer "</p>...<p>"):

One wonders how it got there in the first place, since it is not Valid.
One also wonders why you think you need to (or could, trivially) fix that.

>   s = s.replace(/^<\/([^> ]+)[^>]*>(.*)<\1>$/,"$2");

  <strong title="You did not expect that x > y, did you?"
          style="color: red; font-weight: bold; text-transform: uppercase"
          >wrong</strong>

This expression does _not_ accurately match HTML tags or elements, or XML 
tags or elements for that matter.  See HTML5 (CR) and XHTML 1.0/1.1 (REC) 
for details.
 
> Nothing happens (s is unchanged, and if I use match() instead of replace
> it returns false), but if I go try it at
> http://www.regular-expressions.info/javascriptexample.html, it works!
> 
> I've tried all sorts of things (creating a separate regExp object, using
> //g, taking out the ^$, replacing [^> ]+ with [a-z0-9]*...) but nothing
> makes any difference.
> 
> It's driving me nuts. Can anyone tell me what I'm doing wrong here?

You are using “.” which does not match "\n", so that expression matches as 
far as 

   </p><ol><li>First\n</li><li>Second\n</li></ol><p>  
  `---'`-----------'
    :        :
    :        '-------.
    :                :
  ,----------------.,--.   
  ^<\/([^> ]+)[^>]*>(.*)<\1>$

(use a fixed-width font).  Then it fails, because "\n" is not matched by 
“<”.

That said, like so many other people before you, you are trying to apply an 
expression describing a Chomsky type-3 language to match a sentence of a 
text in a Chomsky type-2 language, which usually *must* fail.

-- 
PointedEars

Twitter: @PointedEars2
Please do not Cc: me. / Bitte keine Kopien per E-Mail.

[toc] | [prev] | [standalone]

Back to top | Article view | comp.lang.javascript

csiph-web