When to use strtr vs str_replace?

Question

I'm having a hard time understanding when strtr would be preferable to str_replace or vice versa. It seems that it's possible to achieve the exact same results using either function, although the order in which substrings are replaced is reversed. For example:

echo strtr('test string', 'st', 'XY')."\n";
echo strtr('test string', array( 's' => 'X', 't' => 'Y', 'st' => 'Z' ))."\n";
echo str_replace(array('s', 't', 'st'), array('X', 'Y', 'Z'), 'test string')."\n";
echo str_replace(array('st', 't', 's'), array('Z', 'Y', 'X'), 'test string');

This outputs

YeXY XYring
YeZ Zring
YeXY XYring
YeZ Zring

Aside from syntax, is there any benefit to using one over the other? Any cases where one would not be sufficient to achieve a desired result?

Answer 1

First difference:

An interesting example of a different behaviour between strtr and str_replace is in the comments section of the PHP Manual:

<?php
$arrFrom = array("1","2","3","B");
$arrTo = array("A","B","C","D");
$word = "ZBB2";
echo str_replace($arrFrom, $arrTo, $word);
?>

• I would expect as result: "ZDDB"
• However, this return: "ZDDD" (Because B = D according to our array)

To make this work, use "strtr" instead:

<?php
$arr = array("1" => "A","2" => "B","3" => "C","B" => "D");
$word = "ZBB2";
echo strtr($word,$arr);
?>

• This returns: "ZDDB"

This means that str_replace is a more global approach to replacements, while strtr simply translates the chars one by one.

---

Another difference:

Given the following code (taken from PHP String Replacement Speed Comparison):

<?php
$text = "PHP: Hypertext Preprocessor";

$text_strtr = strtr($text
    , array("PHP" => "PHP: Hypertext Preprocessor"
        , "PHP: Hypertext Preprocessor" => "PHP"));
$text_str_replace = str_replace(array("PHP", "PHP: Hypertext Preprocessor")
    , array("PHP: Hypertext Preprocessor", "PHP")
    , $text);
var_dump($text_strtr);
var_dump($text_str_replace);
?>

The resulting lines of text will be:

string(3) "PHP"
string(27) "PHP: Hypertext Preprocessor"

---

The main explanation:

This happens because:

• strtr: it sorts its parameters by length, in descending order, so:

  1. it will give "more importance" to the largest one, and then, as the subject text is itself the largest key of the replacement array, it gets translated.
  2. because all the chars of the subject text have been replaced, the process ends there.

• str_replace: it works in the order the keys are defined, so:

  1. it finds the key “PHP” in the subject text and replaces it with: “PHP: Hypertext Preprocessor”, what gives as result:

     “PHP: Hypertext Preprocessor: Hypertext Preprocessor”.

  2. then it finds the next key: “PHP: Hypertext Preprocessor” in the resulting text of the former step, so it gets replaced by "PHP", which gives as result:

     “PHP: Hypertext Preprocessor”.

  3. there are no more keys to look for, so the replacement ends there.

Answer 2

It seems that it's possible to achieve the exact same results using either function

That's not always true and depends on the search and replace data you provide. For example where the two function differ see: Does PHP str_replace have a greater than 13 character limit?

• strtr will not replace in parts of the string that already have been replaced - str_replace will replace inside replaces.
• strtr will start with the longest key first in case you call it with two parameters - str_replace will replace from left to right.
• str_replace can return the number of replacements done - strtr does not offer such a count value.