regx help please.......

SicChild

hey, I'm having a bit of a problem with regular expressions. I'm trying to open a file and grab the urls of images, and the url that images are linked to. So for example:

I want to match & return /mypath/mypage.html & myimages/myimage.jpg

And, if possible it should also match the following things and return the urls:

I'm trying to use preg_match_all....any help would be appreciated...Thanks.

Weedpacket

http://www.phpbuilder.com/board/showthread.php?s=&threadid=10223623

Matches hrefs - do a second to find <img> src= attributes and merge the two.

SicChild

Wow, thanks a lot. I was having a little trouble understanding some of the syntax of regx, but your posted cleared it up. Great job. Thanks.

Also, for anyone looking to do this, this is what I came up with:

"/<a.+?href\s*=\s*([\"']?)(.+?)\\1.*?><img.+?src\s*=\s*([\"']?)(.+?[\.jpg|\.gif|\.png])\\3.*?>/is"

SicChild

okay, now I'm running into a problem:

"/<a.+?href\s*=\s*([\"']?)(.+?)\\1[b].*?>[/b]<img.+?src\s*=\s*([\"']?)(.+?[\.jpg|\.gif|\.png])\\3.*?><\/a>/is"

Specifically, at the point I bolded. I want to allow more attributes after the "href=", hence the .*?. However, it should stop after it hits the first >. Right now, it just keeps matching and allows a lot of junk, including more html tags, to be matched before it finds ><img ....any ideas?

Weedpacket

What's happening is that it won't stop until it finds ><img. Are you only wanting the hrefs and srcs of <a> and <img> tags that immediately follow each other, or all <a> hrefs and all <img> srcs in the page?

In the latter case, you'd use two separate preg_match_all()s, one for <a> tags and one for <img> tags (this is what I had assumed you were after originally, but I see better what you're after now.)

In the former case, replace the problematic .? with [^>] - this should ensure it stops matching that subexpression as soon as it sees any > (and keep looking until it finds one or hits the end of the string) - of course if that > is not then immediately followed by a <img then the whole expression will fail.

SicChild

ok thanks for the help. I have one last problem I think 🙂

/<a.+?href\s*\=\s*([\"']?)(.+?)\\1[^>]*>\s*<img.+?src\s*\=\s*([\"']?)(.+?[\.jpg|\.gif|\.png])\\3.*?>\s*<\/a>/ism

That is the regex I have now. I added \s* to match some whitespace if there is. For some reason now it's matching:

<a href="http://www.somesite.com/main.htm?id=asdf"> s</a><a href="http://www.somesite.com/main.htm?id=asdf">ome other text</a> <a href="http://www.somesite.com/main.htm?id=asdf"> Some Damn Text</a>   <a target="_top" href="http://somesite/?login=asdf"> <img src="http://somesite/i.gif" height=38 border=0 width=41 alt=""></a>

Any ideas?

Matches:

matches[16][0] = <a href="http://www.somesite.com/main.htm?id=asdf"> <font size="6" color="#0000FF"><b>s...[i][size=1]cut[/size][/i]...</p> <a target="_top" href="http://somesite.com/?login=asdf"> <img src="http://somesite/i.gif" height=38 border=0 width=41 alt=""></a>
matches[16][1] = "
matches[16][2] = http://www.somesite.com/main.htm?id=asdf"> <fon...[i][size=1]cut[/size][/i]...</p> <a target=
matches[16][3] = "
matches[16][4] = http://somesite.com.com/i.gif

Mod edit: Extremely long lines elided for your reading convenience

Weedpacket

Yeah, I reckon the problem would be right at the beginning:

<a.+?href\s...

The problem is that there's nothing to stop .+? from matching href - so if the first <a> tag isn't being followed by <img, then the regexp engine will try again by having .+? match everything up to the next href. (The original expression would match only single tags, without regard for what they might be followed by.)

This can be solved in a manner that allows preg_match to shine out over ereg - because PCRE regexps allow you to assert that certain patterns must not be matched (href in this case) - but there is a much simpler solution in this particular instance...

The problem with the overenthusiastic match is that it runs on past the end of the <a> tag, and on into the following text in pursuit of that <img. Now the crucial feature of any tag is that it starts with < and ends with >. So if we match a > we can pretty much safely assume we got carried away. In other words, we're happy with that part of the expression matching anything else, but not that.

So if we were to replace .+? with [^>]+, it will happily match characters only so long as it doesn't go past the end of the tag. There are other places where this would be needed (near the start immediately after <img, for instance). There's actually already an instance of this near the ends of both tags (where it says [^>]* as short for "the rest of the tag". The bit in the img src= attribute would be [^.]+ (is there any particular reason why you only want to match tag pairs with certain file extensions?) and ... um, reckon that might be about it.