preg_replace_callback highlight pattern not match in result - php

Question

I have this code:
$string = 'The quick brown fox jumped over the lazy dog and lived to tell about it to his crazy moped.';
$text = explode("#", str_replace(" ", " #", $string)); //ugly trick to preserve space when exploding, but it works (faster than preg_split)
foreach ($text as $value) {
echo preg_replace_callback("/(.*p.*e.*d.*|.*a.*y.*)/", function ($matches) {
return " <strong>".$matches[0]."</strong> ";
}, $value);
}
The point of it is to be able to enter a sequence of characters (in the code above it's a fixed pattern), and it finds and highlights those characters in the matched word. The code I have now highlights the entire word. I'm looking for the most efficient way of highlighting the characters.
The result of the current code:
The quick brown fox jumped over the lazy dog and lived to tell about it to his crazy moped.
What I would like to have:
The quick brown fox jumped over the lazy dog and lived to tell about it to his crazy moped.
Did I take the wrong approach? It would be awesome if someone could point me in the right way, I've been searching for hours and didn't find what I was looking for.
EDIT 2:
Divaka's been a great help. Almost there.. I apologize if I haven't been clear enough on what my goal is. I will try to explain further.
- Part A -
One of the things I will be using this code for is a phone book. A simple example:
When following characters are entered:
Jan
I need it to match following examples:
Jan Verhoeven
Arjan Peters
Raj Naren
Jered Von Tran
The problem is that I will be iterating over the entire phone book, person-record per person-record. Each person also has email-addresses, a postal address, maybe a website, a extra note, ect.. This means that the text I'm actually search can contain anything from letters, numbers, special characters(&#()%_- etc..), newlines, and most importantly spaces. So an entire record (csv) might contain the following info:
Name;Address;Email address;Website;Note
Jan Verhoeven;Veldstraat 2a, 3209 Herkstad;jan#werk.be;www.janophetwerk.be,jan#telemet.be;Jan die ik ontmoet heb op de bouwbeurs.\n Zelfstandige vertegenwoordiger van bouwmaterialen.
Raj Naren;Kerklaan 334, 5873 Biep;raj#werk.be;;Rechtstreekse contactpersoon bij Werk.be (#654 intern)
The \n is meant to be an actual newline. So if I search for #werk.be, I'd like to see both these records as a result.
- Part B -
Something else I want to use this for is searching song-texts. When I'm looking for a song and I can only remember it had to do something with ducks or docks and a circle, I would enter dckcircle and get the following result:
... and the ducks were all dancing in a great big circle, around the great big bonfire ...
To be able to fine-tune the searching I'd like to be able to limit the number of spaces (or any other character), because I would imagine it finding a simple pattern like eve in every song while I'm only looking for a song that has the exact word eve in it.
- Conclusion -
If I summarize this in pseudo-regex, for a search pattern abc with a max of 3 spaces in-between it would be something like this: (I might be totally off here)
(a)(any character, max 3 spaces)(b)(any character, max 3 spaces)(c)
Or more generic:
(a)({any character}{these characters with a limit of 3})(b)({any character}{these characters with a limit of 3})(c)
This can even be extended to this fairly easily I'm guessing:
(a)({any character}{these characters with a limit of 3}{not these characters})(b)({any character}{these characters with a limit of 3}{not these characters})(c)
(I know the ´{}´ brackets are not to be used that way in a regular expression, but I don't know how else to put it without using a character that has a meaning in regular expressions.)
If anyone wonders, I know the sql like statement would be able to do 80% (I'm guessing, might even be more) of what I'm trying to do, but I'm trying to avoid using a database to make this as portable as possible.
When the correct answer has been found, I'll clean this question (and the code) up and post the resulting php-class here (maybe I'll even put it up on github if that would be useful), so anyone looking for the same will have a fully working class to work with :).

Answer 1

I've came up with this. Tell me if it's what you want!
//$string = "The quick brown fox jumped over the lazy dog and lived to tell about it to his crazy moped.";
$string = "abcdefo";
//$pattern_array1 = array(a,y);
//$pattern_array2 = array(p,e,d);
$pattern_array1 = array(e,f);
$pattern_array2 = array(o);
$pattern_array2 = array(a,f);
$number_of_patterns = 2;
$regexp1 = generate_regexp($pattern_array1, 1);
$regexp2 = generate_regexp($pattern_array2, 2);
$string = preg_replace($regexp1["pattern"], $regexp1["replacement"], $string);
$string = preg_replace($regexp2["pattern"], $regexp2["replacement"], $string);
$string = transform_multimatched_chars($string);
// transforming other chars after transforming the multimatched ones
for($i = 1; $i <= $number_of_patterns; $i++) {
$string = str_replace("#{$i}", "<strong>", $string);
$string = str_replace("#/{$i}", "</strong>", $string);
}
echo $string;
function generate_regexp($pattern_array, $pattern_num) {
$regexp["pattern"] = "/";
$regexp["replacement"] = "";
$i = 0;
foreach($pattern_array as $key => $char) {
$regexp["pattern"] .= "({$char})";
$regexp["replacement"] .= "#{$pattern_num}\$". ($key + $i+1) . "#/{$pattern_num}";
if($key < count($pattern_array) - 1) {
$regexp["pattern"] .= "(?s)((?:(?!{$pattern_array[$key + 1]})(?!\s).)*)";
$regexp["replacement"] .= "\$".($key + $i+2) . "";
}
$i = $key + 1;
}
$regexp["pattern"] .= "/";
return $regexp;
}
function transform_multimatched_chars($string)
{
preg_match_all("/((#[0-9]){2,})(.*)((#\/[0-9]){2,})/", $string, $matches);
// change this for your purposes
$start_replacement = '<span style="color:red;">';
$end_replacement = '</span>';
foreach($matches[1] as $key => $match)
{
$string = str_replace($match, $start_replacement, $string);
$string = str_replace($matches[4][$key], $end_replacement, $string);
}
return $string;
}