Regexp Difficulties

douceur

Hi. I want to do something along the lines of:

preg_match('/X(not foo|bar)X/', $string);

Basically, it would match anything except 'XfooX' and 'XbarX'.

The closest pattern I can come up with is '/X[^{(foo|bar)]X/'.}

This, however, doesn't work because ()'s lose their "special powers" within []'s. Is there any simple way to do this? And if there's no simple way, what would be the complex way?

EDIT:

Wait a minute... I guess I actually posted incorrectly. Sorry about that... I would want it to match strings such as 'XasdfX' and 'XffffffffffffffffX' and 'XzxbcvasdfaX'. I just don't want it to match the string if it has 'foo' or 'bar' in the middle of the X's. In fact, I would also like it to match 'foo XasdfX bar'.

Weedpacket

Well, you're on the right track with preg_match - this would be a serious pain if you used ereg. You're going a bit off-course using character classes, since these only match single characters, not whole strings.

What preg_match() offers are assertions. In your case you want to match "X.*?X", while demanding that the bit between the Xs is neither "foo" nor "bar". In other words, the bit that follows the first X must not be "fooX" or "barX" (where the Xs are there so that "Xfoolish barristerX" still matches)

The syntax for making such a demand (a "forward negative assertion" in techspeak) is
code[/code]

The full regexp would be (note that I haven't tested this at all)

/X(?!fooX)(?!barX).*?X/

"'X', (which must not be followed by 'fooX', and must not be followed by 'barX'), followed by as few characters as possible before the next occurrence of 'X'".

Weedpacket

Oh, okay, you don't want foo or bar anywhere between the X's.

/X(?![^X]*foo)(?![^X]*bar).*?X/

The first assertion says "there is no foo before the next X"

douceur

Haha. It seems that for some reason today, I cannot correctly twist the English language to my liking. After reading your second post, I realized that my edit was not correct either. Your first reply, however, DID solve my problem, and I thank you. It was exactly what I wanted.

douceur

Hm... Well it seems that by oversimplifying my question, my problem wasn't really solved.... So let me run-down the entire problem. Basically, I'm taking (text) values from a form. Certain HTML tags are not allowed within the values, but others are.

Basically, I allow , , and tags, but nothing else. My initial solution to this was this:

$text = preg_replace('/<\\/?([^(br|b|i)])*[^>]*\\/?>/', '&lt;$1&gt;', $text)

Of course there are problems with that... There is the initial one which I originally posted to try to get fixed. Any help would definitely be appreciated.

P.S. The last '\' before the > is for XHTML compatibility.

douceur

Well, I seem to have finally solved my problem, minus one small difficulty. The code I ended up using is this:

$text = preg_replace('/<(\\/)?(?!(foo|bar) +[^>]*\\/?>)([^>]*?\\/?)>/', '&lt;$1$3&gt;', $text);

However, for some reason, if the tag is </bar> or </foo>, it still replaces the <>'s with < and >. It's as if it doesn't actually use the first '(\/)?' at all. $1 comes up blank, and $3 comes up as /bar or '/foo'

Weedpacket

Y'know, if you'd said that in the first place I'd've pointed you in the direction of [man]strip_tags[/man]. But, if you're wanting to escape all other tags...
One thing you can do with preg_* function is use your own delimiter instead of just the traditional /. The advantage of that is that you don't then have to escape the / when you use it inside the regexp. So that instead of

/<\/b>/

you can have, say

#</b>#

douceur

Good lord... Well, heh, thanks again for all the help. At least I learned something. I never knew before about the (?!) thing. lol. Sorry to cause you the trouble. 🙂

snaack

Hi!

I've got an annoying problem with a replace pattern to use with preg_replace.

$pattern = "'<A href=\"*(.)*\">*(.)*</A>'i";
$replacement = "<A href=\"\\1\"><u>\\2</u></A>";

$tmp = preg_replace($pattern, $replacement, $tmp);

If $tmp is a string like

<A href="http://www.foo.com">[url]www.foo.com[/url]</A>

$tmp after the replacement is

<A href=""><u></u></A>

What the hell am I doing wrong?

EDIT --> I love the replace-slash function:rolleyes:

Weedpacket

It's a bit hard to say, what with vBulletin mangling backslashes at random, but my guess would be that the \1 and \2 in $replacement ought to be escaped again (i.e., \1 and \2). The first backslash is to escape the second. Without that, \1 would be getting interpreted as ASCII 1. If they're already double-backslashed - try triple-backslashing them.

PS. Instead of (.), I think you're wanting (.).

snaack

Thnx for the reply, Weedpacket!

I already tried that, but unfortunately that also doesn't do the trick.

It a very easy thing I want to do, why does the solution has to be so damn complicated 🙁

Weedpacket

Well, neither of

$replacement = '<A href="\\1"><u>\\2</u></A>';
$replacement = "<A href=\"\\1\"><u>\\2</u></A>";

make weird characters. Problem is, I don't understand what the $pattern ought to be doing.

snaack

Basically I want to add and to the string between the <a> and the </a> tags.

All the tags that I need to handle have the format <a href="foo">foo</a>, I want to change it to <a href="foo">foo</a>

snaack

anybody?? it still doesn't work 🙁

douceur

Does this not work?

$link = preg_replace('#<a href="(.*)">(.*)</a>#i', '<a href="$1"><u>$2</u></a>', $link);

snaack

argh!

no, that also doesn't work 🙁

I feel so stupid as if I'm some PHP-n00b :o

How hard can this be? But I can't figure it out..

Anybody any ideas?

Weedpacket

Hm, a regexp I've got for grabbing the href= attribute of tags...

Modify that to remember all the bits involved, and to make sure that the
bits between the <a> and </a> tags matches the href attribute itself... retains any of the other attributes of the <a> tag as well (while rearranging them a bit so that href is the first...)

preg_replace(
	"#<a(.+?)href\s*=\s*([\"']?)(.+?)\\2(.*?)>\\3</a>#is"
	,"<a href=\"$3\" $1 $4><u>$3</u></a>"
	,$string);

Though of course these days you should be using a style of text_decoration:underline instead of tags 🙂

snaack

Weedpacket, thnx for the help!

I finally was able to fix the regular expression with this code:

preg_replace("
	/<a.+?href\\s*=\\s*([\\"']?)(.+?)\\\\1.*?>(.*?)<\\/a>/is", 
	"<a href=\\"$2\\"><u>$3</u></a>",
	 $strHTML
);

The reason why I needed to do this is because I cannot apply a CSS-style, the script writes a XML-document which is loaded into a Flash-file. In order to underline the links in Flash I had to add the tags!