Posted by Toby A Inkster on 05/01/07 10:45

Fabri wrote:

> "Le'ts go to <a href="my.htm">my car</a>. Tomorrow I'll have to buy a
> new car. My new car is <em>red</em>! Please don't think to be in Nascar!!"

Try the following... [Credits due: Brad Choate, John Gruber, Matthew
McGlynn and Alex Rosenberg for the _tokenize() function.]

<?php

// how many times do you want to allow the tokenizer to loop?
// The higher the value, the longer your system could churn
// given an infinite-loop bug (or really really really long text string).
define('MAX_TOKENIZER_LOOPS', 2000);

// print error on tokenizer loop problem?
define('ADVISE_TOKENIZER_FAILURE', FALSE);

// keys for $tokens hash
define('TOKENS_TYPE_TEXT', 'text');
define('TOKENS_TYPE_TAG', 'tag');

function _tokenize(&$str, &$tokens) {
#
# Parameter: Pointer to string containing HTML markup,
# pointer to array to store results.
#
# Output array contains tokens comprising the input
# string. Each token is either a tag (possibly with nested,
# tags contained therein, such as <a href="<MTFoo>">, or a
# run of text between tags. Each element of the array is a
# two-element array; the first is either 'tag' or 'text';
# the second is the actual value.
#
# Based on the _tokenize() subroutine from Brad Choate's MTRegex plugin.
# <http://www.bradchoate.com/past/mtregex.php>


$len = strlen($str);

$depth = 6;
$nested_tags = str_repeat('(?:<(?:[^<>]|', $depth);
$nested_tags = substr($nested_tags, 0, -1);
$nested_tags .= str_repeat(')*>)', $depth);

$match = "/(?s: <! ( -- .*? -- \s* )+ > ) |
(?s: <\? .*? \?> ) |
$nested_tags/x";

$last_tag_end = -1;
$loops = $offset = 0;

//433 PHP 4.3.3 is required for this
//433 while (preg_match($match, $str, $hits, PREG_OFFSET_CAPTURE, $offset)) {
while (preg_match($match, substr($str, $offset), $hits, PREG_OFFSET_CAPTURE)) {

$extracted_tag = $hits[0][0]; // contains the full HTML tag
//433 $tag_start = (int)$hits[0][1]; // position of captured in string
$tag_start = $offset + (int)$hits[0][1]; // position of captured in string
$offset = $tag_start + 1; // tells preg_match where to start on next iteration

// if this tag isn't next to the previous one, store the interstitial text
if ($tag_start > $last_tag_end) {
$tokens[] = array('type' => TOKENS_TYPE_TEXT,
'body' => substr($str, $last_tag_end+1, $tag_start-$last_tag_end-1));
}

$tokens[] = array('type' => TOKENS_TYPE_TAG,
'body' => $extracted_tag);

$last_tag_end = $tag_start + strlen($extracted_tag) - 1;

if ($loops++ > MAX_TOKENIZER_LOOPS) {

if (ADVISE_TOKENIZER_FAILURE) {
print "SmartyPants _tokenize failure.";
}
return;
}
}


// if text remains after the close of the last tag, grab it
if ($offset < $len) {
$tokens[] = array('type' => TOKENS_TYPE_TEXT,
'body' => substr($str, $last_tag_end + 1));
}

return;

}

/**
* Make a particular word in an HTML string into a link.
*
* @copyright Copyright (C) 2007 Toby A Inkster
* @param string $haystack HTML string to search through.
* @param string $needle Word or phrase to find.
* @param string $link Link to add to this word. Opt; default Wikipedia.
* @param boolean $case_sensitive Matching sensitivity. Opt; FALSE.
*/
function linkity ($haystack, $needle, $link='', $case_sensitive=FALSE)
{
if ($link=='')
$link = 'http://en.wikipedia.org/wiki/'.ucfirst($word);

$regexp = '#\b('.$word.')\b#'.($case_senitive?'':'i');
$inlink = FALSE;
$out = '';

$tokens = array();
_tokenize($string, $tokens);

foreach ($tokens as $t)
{
if ($t['type']==TOKENS_TYPE_TAG)
{
if (preg_match('#<a#i', $t['body']))
$inlink = TRUE;
elseif (preg_match('#</a#i', $t['body']))
$inlink = FALSE;
$out .= $t['body'];
}
else
{
if ($inlink)
$out .= $t['body'];
else
$out .= preg_replace($regexp,
"<a href=\"{$link}\">$1</a>",
$t['body']);
}
}
return $out;
}

# Test -- should only link the second and third occurances of the word 'car'.
$str = 'Le\'ts go to <a href="my.htm">my car</a>. Tomorrow I\'ll have to buy
a new car. My new car is <em>red</em>! Please don\'t think to be in Nascar!!';
print linkity($str, 'car')."\n";

?>


--
Toby A Inkster BSc (Hons) ARCS
http://tobyinkster.co.uk/
Geek of ~ HTML/SQL/Perl/PHP/Python/Apache/Linux

[Back to original message]