Extracting mutiple hunks of text from a string

Anon

Hello,
Does anyone know of a way to extract hunks of text that partially match a string?

Example:
You have an html page with links in it and a bunch of text before, after, and inbetween different links and you want to extract just the links.
asdfasdf <a href="abc.com"> abc.com</a> asdfdlkfjieoj <a href="asdf.com">asdf.com</a> lskdjflokwjeoifjweoilkdjlkj and so on....

Is there a function that just extracts hunks that include wildcards?
example(ficticious):
str_extract("str1", "str2", $string)
where str1 and str2 are the beginning and end qualifiers and $string is the string to examine. It would probably return an array of substring "hunks".

I have tried several methods, like ereg_replace to replace text inbetween the </a> and <a href tags to leave just the beginning and end junk to be cut off (pretty easy), but for some reason I can't get wild cards to work for the junk inbetween (could be any characters). I want to just capture the stuff between <a href and </a> and dump the stuff between </a> and <a href if that helps clarify my situation.
Any info or techinques would be appriciated.

Thanks!

Anon

You can use an ereg_match call with the regex:

"(<a href[ a-zA-z0-9/\"._].*</a>)"

that should get you all of the text of every link in the page. You might have to use preg_match, I seem to remember that ereg_match will always return an array with 10 elements. Also you'll probably have to add some other characters inside the [] to fill out the entire character set that could be found in the link.

Anon

Doesn't ereg() use a "greedy" algorithm, so that in your regex the
.*</a> would match match everything up to the last </a> on
that line?

From the brief description, it wasn't clear to me whether Marc
was wanting to match just the URL or the enter <a>..</a> and
whether the URL and the descriptive text were supposed to
be matched together or separately.

Anon

Thanks for replying,
I would like to capture everything from <a to </a>. If it's not possible to get the actual </a> in the capture, that's ok, I'll just add it back in with the code.

Any suggestions?

thanks,
Marc

Anon

To help clarify a little better, I would like the equivalent of this:
ereg("<a href*</a>", $page_capture, $filtered_capture);
But, of course, this won't work... It's not a regular expression.

Any Ideas?

Thanks,
Marc

Anon

You are right ereg is greedy, but the regex that I put up should work, execpt for one flaw.

This:
[ a-zA-z0-9/\"._].*

Should be:
[ a-zA-z0-9/\"._]*

The * after the [] is telling the regex to search for 0 to many or the characters that are inside the []. I specifically didn't use the < character to try and combat the greedy nature of the function. This section should stop processing as soon as it finds a < and that should be the start of the </a> tag.

Ofcourse I haven't tested this regex, so it could have many other flaws. Right away I can see that it doesn't handle the tags having different case letters.

"(<a href[ a-zA-z0-9/\"._]*</a>)"

I hope this helps. Maybe I'll even have time to test it out today.

Anon

Ok, here is code that works (mostly). I tried this out on a copy of slashdot's index.html. It found 137 links, and looks like thats all of them but the ones that have embedded tags in them. Like the banner add at the top.

<?php
error_reporting(15);
print "Opening test.html: \n\n";
$fp = fopen("test.html", "r");

$text = "";
$x = 0;

print "Parsing test.html \n";
while ($line = fgets($fp, 2000)) {
print "Parsing line ".++$x." \n";
$text .= $line;
}

print " Pattern Matching: \n\n";
$pattern = "|(<[Aa] [HhRrEeFf][ a-zA-Z0-9/\"'.?&=:,-]>";
$pattern .= "[ a-zA-Z0-9/\"'.?&=:,-]</[Aa]>)|";

preg_match_all($pattern, $text, $matches);

print "".count($matches[1])." Matches Found \n";

for ($i = 0; $i < count($matches[1]); $i++) {
print "Link Found: ".$matches[1][$i]." \n";
}
?>

I hope this helps out.

Anon

Thank you very much,
It works partially, probably just have some more characters that I have to include (including /n). One interesting thing though, preg_match_all returns a 2 dimesional array (an array of arrays) where one array @ $matches[0] is the same as $matches[1]. Wierd... so I have to reference the inside array with $matches[0][$x].

any idea why? Just wondering, I can live with what I have right now, but would be nice to know why it stored the array of results in 2 sets of arrays that are exactly identical.

Thanks again,
Marc

Anon

I answered my own question,
preg_match_all sticks 2 types of matches in it's array. 1st is an array for the full pattern match and all subsequent arrays contain results for pattern matching on individual subpatterns, pretty neat actually.

Thanks for your help, so I didn't have to play around with it for 2 weeks until I figured it out.

-Marc

Anon

That's because the first array [0] is the complete set of mached data. The subsiquent arrays [1-n] are the () sections of the match that you are performing. Since this regex only has () section the array's end up being the same.

But you could have made it 2 sections like this:

pattern = "|(<[Aa] [HhRrEeFf][ a-zA-Z0-9/\"'.?&=:,-]>)";
$pattern .= "([ a-zA-Z0-9/\"'.?&=:,-])</[Aa]>)|";

Which would have given you [1] holding the <a href=*****> and [2] holding the text in between the > < of the link.