A complicated preg_replace() problem....

[karnaf]

I've been trying to find a way through this, but it's just getting frustrating... The closest info I've found about this was in thread434-523609 but the problem there is different from mine.

I need to go over an html file and highlight words or part of words in it (kind of what you get at the end of a search engine result like google). The problem is, I don't want words within html tags, any html tag, to be highlighted.

For example,
$text=&quot;<br>let's bring the house down<br><center>Here we are in the center of town</center>&quot;
key words for searching - br AND center
What I want to get is
$text=&quot;<br>let's <b>br</b>ing the house down<br><center>Here we are in the <b>center</b> of town</center>&quot;
and not
$text=&quot;<<b>br</b>>let's <b>br</b>ing the house down<<b>br</b>><<b>center</b>>Here we are in the <b>center</b> of town</<b>center</b>>&quot;


The basic idea is that I have an array of the keywords to search for $keys_arr, where all the values are of the form &quot;/$word/si&quot;. The replace array $replace_arr has values of the form &quot;<b>$word</b>&quot;. The action line is
$text = preg_replace($keys_arr,$replace_arr,$text);

I'm pretty sure there should be some regular expression solution, but I just can't seem to find it :

Thanks

 

[davidshields]

Do you HAVE to preserve the original HTML ? If not you could have a regexp that strips out HTML, giving

let's bring the house down Here we are in the center of town

then one to highlight br and center.

Just a thought.

David.

 

[piti]

if you use &quot;/(?<!\<|\<\/)$word/si&quot; in the $keys_arr, it will not touch the html tags, but only if the searched $word is identical to the tag beginning
e.g.
searching for &quot;center&quot; or &quot;cent&quot;, will not match <center> or </center>, but search for &quot;enter&quot; will modify these html tags
if you prepare some algorithm to test if $word is a part of any html tag but not the starting part (e.g. &quot;enter&quot; is part of <center>) then &quot;/(?<!\<|\<\/)([a-z]+)$word/si&quot; in $keys_arr and &quot;\${1}<b>$word</b>&quot; in $replace_arr will do the job
hope this helps

 

[piti]

well i forgot to mention that, as you can see from the code, it has more limitations than mentioned
like searching some tag params (align, height, style, ...) modifies the tag too

 

[Westbury]

The only thing I can think of is what davidshields suggested. However, if you need to keep all of the HTML tag then I suggest that you go through the page character by character. Add the character to a buffer and set a flag to true if it encounters a &quot;<&quot; and false is it encounters a &quot;>&quot;. Then if your flag is set to false then check the buffer for the key words. Finally if you encounter either the end of the file or a space, output the buffer.
Its a long winded way of doing it but at least you could customise (or customize, since this seems to be an american site) how the search is performed.

 

[karnaf]

Thanks, but the problem is greater than this..

A. I have to preserve the html page, just modify it.
B. what if we have something like &quot;please check if a<5 and b>6&quot; as the text? how can we tell that &quot;<5 and b>&quot; is not an html tag when using preg? we cant use the rule of avoiding anything within <>....

The only solution I see is to code my own function that will parse the text and look explicitly for all known html tags... I just hate to think there is no easier solution, something more elegant...

thanks!

 

[Westbury]

In html the < and > character are coded as &gt; and &lt; so it shouldn't pick up anything that isn't a html tag.
My character by character search would preserve the HTML. maybe I didn't explain it very well. I will try and put it into a sort of pseudo code:
1. add character to buffer
2. if character is &quot;<&quot; set flag to true
3. if character is &quot;>&quot; set flag to false and output buffer
4. if flag is false then perform string replace
5. if character is &quot; &quot; or EOF then output buffer and clear buffer
6.loop until EOF

This way the str_replace is only used when it outside of a HTML tag, and everything is output.

Hope that clears it up a little.

 

[karnaf]

Thanks...

I've found a solution that seems to be working fine, though it hasn't yet been tested in the real workd

I didn't feel like making Turing machine, that will go char by char, so I used other stuff, but in general, used your idea Westbury. The idea is to cut the text into pieces acording to the locations of < and >. Then preg_replace the necessary parts.


Here you have the code.
------------------------------------------------------------
$old_gt = 0; // Last > position in the text
$old_lt = 0; // Last < position in the text
$text = ' '.$text;
$res_text=&quot;&quot;;
while((($cur_lt = strpos($text, '<', $old_gt+1)) !== false)&&
(($cur_gt = strpos($text, '>', $cur_lt+1)) !== false))
{
$tmp_lt = strpos($text, '<', $cur_lt+1);
while (($tmp_lt !== false) && ($cur_gt > $tmp_lt))
{
$cur_lt = $tmp_lt;
$tmp_lt = strpos($text, '<', $cur_lt+1);
}
$res_text.= preg_replace($keys_arr,$replace_arr,substr($text, $old_gt, ($cur_lt-$old_gt)));
$res_text.= substr($text, $cur_lt, ($cur_gt-$cur_lt));
$old_gt = $cur_gt;
$old_lt = $cur_lt;
}
$res_text.= preg_replace($keys_arr,$replace_arr,substr($text, $cur_gt));
$text = trim($res_text);
------------------------------------------------------------
$keys_arr = the words that needs to be highlighted
$replace_arr = the words highlighted.

Thanks all.
I was just looking for a more elegant solution I and will still love to get one better if any of you can think of one.

karnaf