Regular Expressions... I have no idea :)

darren_wheatley

Heya.

I am trying to grab a URL from a line of HTML code. Example:

Currently I am looping through the amn-day thing and getting it character by character.

Is there someone out there who can use a preg_grep or ereg_replace style command to extract the /blah/blah.html?searchId=104191981846759133&listingId=99262163135 from out of that string? Yes, I don't know what I am talking about with regex.

I found some threads, eg: PHP Builder Community Forums > PHP Help > General Help > Regular Expression for matching a date, but they didn't really help me.

A huge help would be an answer in the form of (syntax may vary 😉 ) :
$strURL = <insert function name>($strHTMLLine, <insert rexex pattern for the above example>);

Thanks in advance,

Darren

Weedpacket

Well, if the URLs you want to match are the href attributes of <a> tags, and you want all of them, the appropriate function to use would be [man]preg_match_all[/man]

preg_match_all(
    "/<a.+?href\s*=\s*([\"']?)(.+?)\\1.*?>/is", 
    $string,
    $matches
);

The URLs will be in $matches[2].

The expression reads:
/
Start of the regexp

<a
"<a"

.+?
some characters between <a and the next bit

href\s=\s
we expect to see "href=" at some point, maybe with space around the = sign.

b[/b]
Attribute values should be quoted with either " or '. We can't assume this, unfortunately. Whichever of the three possibilities occurs, though, we need to remember it.

b[/b]
The real engine room of the match. The ? is so's we don't get carried away and match all the way to the end of the string..

\1
That remembered (possible lack of) quote, to end the attribute value.

.*?
Maybe some more stuff before the final

/is
The end of the expression. The modifier "i" is used to make the match case-insensitive (so that <A HREF will match as well as <a href), and "s" is so that . will match \n (in case the tag wraps across more than one line).

There are more efficient solutions.

darren_wheatley

Weedpacket,

It looks as though you nailed that answer.

I'll give it a go 🙂 Thanks for explaining it.

Darren

superwormy

As a note, you can also do this with simple string matching, sometimes it's faster than a complicated regular expression.

http://phpbuilder.net/board/showthread.php?s=&threadid=10237538&highlight=links

Weedpacket

Is that faster? I'd hesitate to say it's simpler... all you're doing is building the pattern matcher by hand.

nightowl

I have tried your solution here but it didn't return the expected results (yet?). When using it for example on www.google.be it doesn't return those links which are not surrounded by " ".

Do you have any idea how to fix this ?

in $matches[0], I see the url's correctly (without the "), but in $matches[2], they don't show up (only a /).

<?php

// load page

if (empty($url)) {
$url = "http://www.google.be";
}

echo "Please specify an url to grep via the parameter 'url'. Please include [url]http://[/url] ";

echo "The links on the page $url can be found here... ";

$string = implode ("", file($url));

// search for href tags
preg_match_all("/<a.+?href\s=\s([\"']?)(.+?)\1.*?>/is", $string, $matches);

echo "<pre>";
print_r($matches);
echo "</pre>";

nightowl

Seemed to me the problem was solved when I left out a dot ...

preg_match_all("/<a.+?href\s=\s([\"']?)(.+?)\1*?>/is", $string, $matches);

Mind the missing dot between \1 and * ...

Indeed the correct url's of www.google.be were returned but the function also grabs everything between end of href and > . Not what we need 🙁

nightowl

I found a solution to the problem I guess. Found a sample on php.net which I altered a little bit. Guess I took the best of the two ...

preg_match_all("/<a.+?href\s=\s[\"']?([^\"' >]+)/i", $string, $matches);

Seems to work fine.

On to the next challenge - making it useable for <img> tags too 🙂