substr_replace

It's worth noting that when start and length are both negative -and- the length is less than or equal to start, the length will have the effect of being set as 0.
<?php
substr_replace('eggs','x',-1,-1); //eggxs
substr_replace('eggs','x',-1,-2); //eggxs
substr_replace('eggs','x',-1,-2); //eggxs
?>

Same as: 
<?php
substr_replace('eggs','x',-1,0); //eggxs
?>

<?php
substr_replace('huevos','x',-2,-2); //huevxos
substr_replace('huevos','x',-2,-3); //huevxos
substr_replace('huevos','x',-2,-3); //huevxos
?>

Same as: 
<?php
substr_replace('huevos','x',-2,0); //huevxos
?>

Another note, if length is negative and start offsets the same position as length, length (yet again) will have the effect as being set as 0. (Of course, as mentioned in the manual, when length is negative it actually represents the position before it)
<?php
substr_replace('abcd', 'x', 0, -4); //xabcd
?>

Same as: 
<?php
substr_replace('abcd','x',0,0); //xabcd
?>

<?php
substr_replace('abcd', 'x', 1, -3); //axbcd
?>

Same as:
<?php
substr_replace('abcd', 'x', 1, 0); //axbcd
?>

Add prefix to strings:
<?php
substr_replace($strings, '_prefix', 0, 0);
?>
Add suffix/postfix to strings:
<?php
substr_replace($strings, '_suffix', array_map('strlen', $strings), 0);
?>

Forget all of the mb_substr_replace() implementations mentioned in this page, they're all buggy.
Here is a version that mimics the behavior of substr_replace() exactly:
<?php
if (function_exists('mb_substr_replace') === false)
{
 function mb_substr_replace($string, $replacement, $start, $length = null, $encoding = null)
 {
 if (extension_loaded('mbstring') === true)
 {
 $string_length = (is_null($encoding) === true) ? mb_strlen($string) : mb_strlen($string, $encoding);
 
 if ($start < 0)
 {
 $start = max(0, $string_length + $start);
 }
 
 else if ($start > $string_length)
 {
 $start = $string_length;
 }
 
 if ($length < 0)
 {
 $length = max(0, $string_length - $start + $length);
 }
 
 else if ((is_null($length) === true) || ($length > $string_length))
 {
 $length = $string_length;
 }
 
 if (($start + $length) > $string_length)
 {
 $length = $string_length - $start;
 }
 
 if (is_null($encoding) === true)
 {
 return mb_substr($string, 0, $start) . $replacement . mb_substr($string, $start + $length, $string_length - $start - $length);
 }
 
 return mb_substr($string, 0, $start, $encoding) . $replacement . mb_substr($string, $start + $length, $string_length - $start - $length, $encoding);
 }
 
 return (is_null($length) === true) ? substr_replace($string, $replacement, $start) : substr_replace($string, $replacement, $start, $length);
 }
}
?>

Using substr_replace() can be avoided by using substr() instead:
<?
$string = substr($string, 0, $position_needle).$replace.substr($string, $position_needle+$length_needle);
?>
This can be useful when you need to replace parts of multibyte strings like strings encoded with utf-8. There isn't a multibute variant for substr_replace(), but for php substr() there is mb_substr(). For more information on multibyte strings see http://nl3.php.net/manual/en/ref.mbstring.php 

PHP version of Java's removeCharAt() function:
<?php
function removeCharAt($str, $int){
 return substr_replace($str,"",$int,1);
}
?>

I have a little function that works like substr_replace () what I use for some purpose. Maybe someone needs it.
<?php
function putinplace($string=NULL, $put=NULL, $position=false)
{
 $d1=$d2=$i=false;
 $d=array(strlen($string), strlen($put));
 if($position > $d[0]) $position=$d[0];
 for($i=$d[0]; $i >= $position; $i--) $string[$i+$d[1]]=$string[$i];
 for($i=0; $i<$d[1]; $i++) $string[$position+$i]=$put[$i];
 return $string;
}
// Explanation
$string='My dog dont love postman'; // string
$put="'"; // put ' on position
$position=10; // number of characters (position)
print_r( putinplace($string, $put, $position) ); 
?>

RESULT: My dog don't love postman
This is a small powerful function that performs its job flawlessly.

See array_splice if you want to do this sort of thing to an array.

This will truncate a longer string to a smaller string of specified length while replacing the middle portion with a separator exactly in the middle.
<?php
$longString = 'abcdefghijklmnopqrstuvwxyz0123456789z.jpg';
$separator = '/.../';
$separatorlength = strlen($separator) ;
$maxlength = 25 - $separatorlength;
$start = $maxlength / 2 ;
$trunc = strlen($longString) - $maxlength;
echo substr_replace($longString, $separator, $start, $trunc);
//prints "abcdefghij/.../56789z.jpg"
?>

I've just taken a look at the post by ntoniazzi and I have a very small correction to make.
In the second if statement, it should be a triple equals, so:
<?php if ($length === null) ?>

It requires the triple equals, for the case of pure insertion, where $length = 0, the double equals, will catch this, causing the string to be cut short. I hope this helps someone.

I recently ran across a situation where I need to strip a heavily nested html list such that only the top level was preserved. I started with a regular expression solution, but found that I kept matching the wrong closing ul with an outer opening ul.
This was my alternative solution, and it seems to work well:
<?php
function stripNestedLists($str)
{
 $str2 = $str;
 $lastStr = $str2;
 
 do
 {
 // Find the first closing ul
 $cul = strpos($str2, '</ul>');
 $ul = 0;
 $lastUL = 0;
 do
 {
 // Find the next opening ul
 $lastUL = $ul;
 $ul = strpos($str2, '<ul', $ul+1);
 }
 while ($ul !== false && $ul < $cul);
 
 $lastStr = $str2;
 $str2 = substr_replace($str2, '', $lastUL, $cul-$lastUL+5);
 $str2 = trim($str2);
 }
 while (strlen($str2) > 0);
 
 return $lastStr;
}
?>

Hope this helps someone.

THE DOT DOT DOT ISSUE
PROBLEM:
You want to abbreviate a string.
E.g. You want "BritneySpears" to show as "BritneySpe...", being only the ten first characters followed by "..."
SOLUTION:
<?
$oRIGINAL = "BritneySpears";
$sHORTER = substr_replace($oRIGINAL, '...', 10);
echo ($sHORTER);
?>
This will result in BritneySpe...

I like the truncate function below...however, I found a few issues. Particularly if you have content that may have any kind of punctuation in it (?, !, ?!?, --, ..., .., ;, etc.)
The older function would end up looking like "blah blah?..." or "blah blah,..." which doesn't look so nice to me...
Here's my fix. It removes all trailing punctuation (that you include in the $punctuation string below) and then adds an ellipse. So even if it has an ellipse with 3 dots, 2 dots, 4 dots, it'll be removed, then re-added.
<?php
function truncate($text,$numb,$etc = "...") {
$text = html_entity_decode($text, ENT_QUOTES);
if (strlen($text) > $numb) {
$text = substr($text, 0, $numb);
$text = substr($text,0,strrpos($text," "));
$punctuation = ".!?:;,-"; //punctuation you want removed
$text = (strspn(strrev($text), $punctuation)!=0)
 ?
 substr($text, 0, -strspn(strrev($text), $punctuation))
 :
$text;
$text = $text.$etc;
}
$text = htmlentities($text, ENT_QUOTES);
return $text;
}
?>

I also needed a sort of "middle" truncate. The above function truncates around the end, but if you want to truncate around the middle (ie "Hello this is a long string." --> "Hello this ... long string.") you can use this (requires the truncate function):
<?php
function mtruncate($text, $numb, $etc = " ... ") {
 $first_part = truncate(truncate($text, strlen($text)/2, ""), $numb/2, "");
 $second_part = truncate(strrev(truncate(strrev($text), strlen($text)/2, "")), $numb/2, "");
 return $first_part.$etc.$second_part;
}
?>

I imagine the description of the parameters really means "number of bytes" where it says "number of characters" (confirmed by testing).

First Example can be simplified =>
 $input = array('A: XXX', 'B: XXX', 'C: XXX');
 substr_replace($input, 'YYY', -3);
output: Array ( [0] => A: YYY [1] => B: YYY [2] => C: YYY )

I suggest changing the function suggested by Guru Evi slightly. I found that it doesn't work as written here.
Original:
function add_3dots($string,$repl,$start,$limit) {
 if(strlen($string) > $limit) {
 return substr_replace(strip_tags($string),$repl,$start,$limit);
 } else {
 return $string;
 };
};
I suggest:
function add_3dots($string,$repl,$limit) {
 if(strlen($string) > $limit) {
 return substr_replace(strip_tags($string),$repl,$limit-strlen($repl));
 } else {
 return $string;
 }
 }
Usage:
$max_length=10;//the max number of characters you want to display
$too_long_string="BLAH BLAH BLAH BLAH BLAH etc.";//the string you want to shorten (if it's longer than the $limit)
$shorter_string=add_3_dots($too_long_string,"...",$max_length);

If your string is not long enough to meet what you specify in start and length then the replacement string is added towards the end of the string.
I wanted to replace the end of the string with ... if the string was too long to display (for instance article preview on a website). The problem was that my string was sometimes not that long and it still added the replacement string. So I wrote a function to replace substr_replace in my website:
function add_3dots($string,$repl,$start,$limit) {
 if(strlen($string) > $limit) {
 return substr_replace(strip_tags($string),$repl,$start,$limit);
 } else {
 return $string;
 };
};
I use strip_tags to strip out the HTML otherwise you might get a screwed up HTML (when a tags open in the string, but because you cut-off it doesn't)

I don't know if this function is multibyte safe but I've written a function that will do the same in multibyte mode.
<?php
//Check to see if it exists in case PHP has this function later
if (!function_exists("mb_substr_replace")){
 //Same parameters as substr_replace with the extra encoding parameter.
 function mb_substr_replace($string,$replacement,$start,$length=null,$encoding = null){
 if ($encoding == null){
 if ($length == null){
 return mb_substr($string,0,$start).$replacement;
 }
 else{
 return mb_substr($string,0,$start).$replacement.mb_substr($string,$start + $length);
 }
 }
 else{
 if ($length == null){
 return mb_substr($string,0,$start,$encoding).$replacement;
 }
 else{
 return mb_substr($string,0,$start,$encoding). $replacement. mb_substr($string,$start + $length,mb_strlen($string,$encoding),$encoding);
 }
 }
 }
}
?>

The preemptive test to see if $string is "too long" shouldn't add strlen($replacement) to $max. $max should represent the absolute maximum length of string returned. The size of the $replacement is irrelevant in that determination.
The rest of the function (unchanged below) operates as defined above. Meaning, the size of the $replacement is subtracted from the $max, so that the returned string is exactly the length of $max.
<?php
function truncate($string, $max = 20, $replacement = '')
{
 if (strlen($string) <= $max)
 {
 return $string;
 }
 $leave = $max - strlen ($replacement);
 return substr_replace($string, $replacement, $leave);
}
?>

<?php
// shortens a long string to a max length while inserting a string into the exact middle
function strShorten($str, $maxlen = 10, $insert = '/.../') {
 if ($str && !is_array($str)) { // valid string
 if ($maxlen && is_numeric($maxlen) && $maxlen < strlen($str)) { // string needs shortening
 if ($insert && ($ilen = strlen($insert))) { // insert string and length
 if ($ilen >= $maxlen) { // insert string too long so use default insert
 $insert = '**'; // short default so works even when a very small $maxlen
 $ilen = 2;
 }
 }
 $chars = $maxlen - $ilen; // number of $str chars to keep
 $start = ceil($chars/2); // position to start cutting
 $end = floor($chars/2); // position from end to stop cutting
 return substr_replace($str, $insert, $start, -$end); // first.insert.last
 } else { // string already short enough
 return $str; // return original string
 }
 }
}
echo strShorten('123456789', 6, ''); // outputs 123789
echo strShorten('123456789', 6, '-'); // outputs 123-89
echo strShorten('123456789', 6, 'longstring'); // outputs 12**89
echo strShorten('abcdefghijklmnopqrstuvwxyz', 10, '..'); // outputs abcd..wxyz
echo strShorten('abcdefghijklmnopqrstuvwxyz'); // outputs abc/.../yz
?>

I recently needed a routine that would remove the characters in one string from another, like the regex
<?php
 $result = preg_replace("/[$chars]/", "", $string);
?>

and I needed it to be fast, and accept pretty much all input. The regex above won't work when strlen($chars) == 0. I came up with this, admittedly pretty horrible-looking code, that is quite fast:
<?php
function RemoveChars($string, $chars)
{
 return isset($chars{0}) ? str_replace($chars{0}, "", strtr($string, $chars, str_pad($chars{0}, strlen($chars), $chars{0}))) : $string;
}
?>

According to my own measurements, the regex in ONLY faster for when strlen($chars) == 1; for longer strings, my routine is faster. What does it do? Let's say you want to remove the period, the comma and the exclamation mark from a string, like so:
$result = RemoveChars("Isn't this, like, totally neat..!?", ".?!");
The str_pad function creates a string equal in length to the string that contains the character to be removed, but consisting only of the first character of that string:
The input is ".,!"
The output is "..."
The strtr function translates all characters in the string-to-be-processed ("Isn't this...") that also occur in the input (".,!") to the characters in the same position in the output ("..."). In other words:
Isn't this, like, totally neat..!?
becomes
Isn't this. like. totally neat....
Finally, the first character from the input (".,!") which happens to be, again, the period, is removed from that string by the str_replace call:
Isn't this like totally neat?
The function needs to check is $chars has at least one character, or else the str_pad function will fail. If it's empty, then the unprocessed string is returned.

Here is a simple function to shorten a string and add an ellipsis
<?php
/**
 * truncate() Simple function to shorten a string and add an ellipsis
 * 
 * @param string $string Origonal string
 * @param integer $max Maximum length
 * @param string $rep Replace with... (Default = '' - No elipsis -)
 * @return string
 * @author David Duong
 **/
function truncate ($string, $max = 50, $rep = '') {
 $leave = $max - strlen ($rep);
 return substr_replace($string, $rep, $leave);
} 
echo truncate ('akfhslakdhglksjdgh', 10, '...');
// Returns akfhsla... (10 chrs)
?>

Just to add to the examples, if replacement is longer than length, only the length number of chars are removed from string and all of replacement is put in its place, and therefor strlen($string) is inreased.
$var = 'ABCDEFGH:/MNRPQR/';
/* Should return ABCDEFGH:/testingRPQR/ */
echo substr_replace ($var, 'testing', 10, 2);

This is my version of making dotted strings:
<?php
function dot($str, $len, $dots = "...") {
 if (strlen($str) > $len) {
 $dotlen = strlen($dots);
 substr_replace($str, $dots, $len - $dotlen);
 }
}
?>

Regarding "...", even the short functions are too long and complicated, and there's no need to use substr_replace. substr() works better and is way faster prior to 4.3.5 as the below poster stated.
function shorten( $str, $num = 100 ) {
 if( strlen( $str ) > $num ) $str = substr( $str, 0, $num ) . "...";
 return $str;
}

My problem was that substr_replace() always added $replacement, so i wrote my own function.
This function only adds $replacement, if substr() took action.
The parameter $length is optional - like substr()'s.
Or I was too stupid using $start and $length...
<?php
function substr_replace_provided($string,$replacement,$start,$length=NULL)
{
 $tmp=substr($string,$start,$length);
 if($string!==$tmp) {
 $string = $tmp.$replacement;
 }
 return $string;
}
?>

I wrote a function that you can use for example in combination with a search script to cut off the articles that are too long.
<?php
function substr_index($text, $maxChars = 20, $splitter
= '...') {
$theReturn = $text;
$lastSpace = false;
if (strlen($text) > $maxChars) {
$theReturn = substr($text, 0, $maxChars - 1);
if (in_array(substr($text, $maxChars - 1, 1),
array(' ', '.', '!', '?'))) {
$theReturn .= substr($text, $maxChars, 1);
} else {
$theReturn = substr($theReturn, 0, $maxChars -
strlen($splitter));
$lastSpace = strrpos($theReturn, ' ');
if ($lastSpace !== false) {
$theReturn = substr($theReturn, 0, $lastSpace);
}
if (in_array(substr($theReturn, -1, 1), array(','))) {
$theReturn = substr($theReturn, 0, -1);
}
$theReturn .= $splitter;
}
}
return $theReturn;
}
?>

the version of my predecessor will add $rep even if the string is shorter than max. fixed version:
<?php
function truncate($string, $max = 20, $rep = '')
{
 if (strlen($string) <= ($max + strlen($rep)))
 {
 return $string;
 }
 $leave = $max - strlen ($rep);
 return substr_replace($string, $rep, $leave);
}
?>

To preserve the filename extension you can call it like this:
truncate([filename], 30, '...' . end(explode('.', [filename])))

This may be obvious to others, but I just spent hours and my feeble brain only caught up to it after a long break. 
If you are looping through a string which has multiple substrings that need to be replaced, you have to add an offset factor to each original offset before you replaced any strings. Here is a real world example: 
From draft.js we get paragraphs with multiple links designated only with offset, anchor text length, url, target. So each anchor text must be wrapped in the <a href="url" target="target">anchortext</a> to save proper content in the database. 
Here is the implementation of offset factor:
 $offset_factor = 0;
 foreach($content->links->links as $index=>$link){
 $replacement = '<a href="'.$link->href.'" target="$link->target">'.$link->anchorText.'</a>';
 $new_offset = $link->offset + $offset_factor;
 $newtext = \substr_replace($content->text, $replacement, $new_offset, $link->length);
 //now we reset the original paragraph text with newtext
 $content->text = $newtext;
 
 //calculate the new offset by calculating the difference in replacement length and original length and add that to the offset_factor
 $additional_characters = strlen($replacement) - $link->length;
 $offset_factor = $offset_factor + $additional_characters;
 }
I hope this helps a noobie :) If there is another easier way, I would love to hear about it.

If you would like to remove characters from the start or end of a string, try the substr() function.
For example, to remove the last three characters from a string:
$string = "To be or not to be.";
$string = substr ($string, 0, -3);

Please note that the function array_slice(), which has a similar functionality but for arrays rather than for strings, has its parameters in a different order.