How can I replace multiple strings within a string without overlapping results?

Question

I'm trying to create common masks from a string like so:

012abc.d+e_fg~hijk => 012{start}.d+{middle}_fg~{end}jk

replace:

$arrFromTo = array(
              'st' => '{pre}',
              'abc' => '{start}',
              'e' => '{middle}',
              'hi' => '{end}',
              'dd' => '{post}'
             );

Instead I keep overlapping replacements and get something like this instead (using a loop of str_replace's):

012{{pre}art}.d+{mi{post}le}_fg~{end}jk

Because the st is found in the already replaced {start} and dd is found in {middle}.

How would you replace the following?

$str = 'abc.d+e_fg~hijk';

echo replace_vars($str); // Desired output: 012{start}.d+{middle}_fg~{end}kJ

I was thinking rearrange the array so 'dd' => '{post}' was in a different position, but that raised other issues. — Tim Lewis
– Tim Lewis, Commented Oct 28, 2014 at 20:28
It's not often I suggest regex, but in this case it might be prudent so you perform lookahead/behinds to check for what you are expecting before you replace — scrowler
– scrowler, Commented Oct 28, 2014 at 20:34
@scrowler Unfortunately there's a bug in my PCRE library with variable length negative look behinds. I'm unable to upgrade PCRE library so I stopped going down that path at this point. — Ryan
– Ryan, Commented Oct 28, 2014 at 20:48
The issue you'll have with regular expressions is that php doesn't support variable-length negative lookbehinds. If it did, you could use a regular expression like: /(?<!{\w*)(e)(?!\w*})/i This would mean give me any "e" where it is not between curly braces and any other word character. The \w* makes it a variable-length lookahead/behind. Apparently pcre handles the variable-length lookahead well, but not the lookbehind. — OneCleverMonkey
– OneCleverMonkey, Commented Oct 28, 2014 at 20:50
You could use a state machine. As you loop through the string, keep a list of the last n characters that match the first n characters of something in your replace list. Once you have a full match, replace the last n characters with your replacement string. — Scott Harris
– Scott Harris, Commented Oct 28, 2014 at 21:09

Rudie · Accepted Answer · 2014-10-28 21:32:27Z

6

I might misunderstand, but you don't seem to need regex for the replacing. They're simple, literal replacements.

$from = '012abc.d+e_fg~hijk';
$arrFromTo = array(
              'st' => '{pre}',
              'abc' => '{start}',
              'e' => '{middle}',
              'hi' => '{end}',
              'dd' => '{post}'
             );
$to = strtr($from, $arrFromTo); // 012{start}.d+{middle}_fg~{end}jk

strtr() is awesome. It takes a very readable input and it doesn't re-replace like your problem in the loop.

answered Oct 28, 2014 at 21:32

Rudie

54.2k42 gold badges137 silver badges176 bronze badges

Sign up to request clarification or add additional context in comments.

Comments

anubhava · Accepted Answer · 2014-10-28 21:22:21Z

2

You can use preg_replace like this:

$str = '012abc.d+e_fg~hijk';
$arrFromTo = array(
              'st' => '{pre}',
              'abc' => '{start}',
              'e' => '{middle}',
              'hi' => '{end}',
              'dd' => '{post}'
             );

$reArr=array();
foreach($arrFromTo as $k=>$v){
   $reArr['/' . $k . '(?![^{}]*})/'] = $v;
}

echo preg_replace(array_keys($reArr), array_values($reArr), $str);
//=> 012{start}.d+{middle}_fg~{end}jk

Core of this regex is this negative lookaead: (?![^{}]*})

Which avoid matching keys of array if it is enclosed in {...} since all the replacements are enclosed in {...}.

answered Oct 28, 2014 at 21:22

anubhava

790k67 gold badges603 silver badges671 bronze badges

7 Comments

Rudie Over a year ago

Why preg_replace? Why not str_replace or strtr? No regexing necessary, is there?

anubhava Over a year ago

How can you use a lookahead in str_replace?

Rudie Over a year ago

Why do you need lookahead?? They're literal replacements... There are no { and } in the source string.

Rudie Over a year ago

Oh, okay, so str_replace isn't as smart as strtr. Too bad. Now we know!

anubhava Over a year ago

Yes right but still you were quite right about strtr +1 to your answer.

|

KevBot · Accepted Answer · 2014-10-28 21:40:23Z

0

This will search the string for each replacement in order. If it finds one, it will split the string, and search the remainder of the string for any other replacements.

$str = '012abc.d+e_fg~hijk';

$rep = array(
    'st' => '{pre}',
    'abc' => '{start}',
    'e' => '{middle}',
    'hi' => '{end}',
    'dd' => '{post}'
);

$searched = '';

foreach ($rep as $key => $r) {
    if (strpos($str, $key) !== false) {

        $searched .= substr($str, 0, strpos($str, $key)) . $r;
        $str = substr($str, strpos($str, $key) + strlen($key));

    }
}

$searched .= $str;

echo $searched; //012{start}.d+{middle}_fg~{end}jk

It will search and find them in the order that you have specified.

answered Oct 28, 2014 at 21:40

KevBot

19k5 gold badges54 silver badges71 bronze badges

Collectives™ on Stack Overflow

How can I replace multiple strings within a string without overlapping results?

3 Answers 3

Comments

7 Comments

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

3 Answers 3

Comments

7 Comments

Comments

Your Answer

Sign up or log in

Post as a guest

Related