levenshtein()

The levenshtein() function calculates the Levenshtein distance between two strings — the minimum number of single-character edits (insertions, deletions, or substitutions) needed to turn one string into the other. A smaller distance means the strings are more alike, so levenshtein() is the go-to tool for fuzzy matching: spell-checkers, "did you mean…?" suggestions, deduplicating near-identical records, and ranking search results by closeness.

This chapter covers the syntax, the optional cost weights, how it differs from related functions, common gotchas, and runnable examples.

Syntax

levenshtein(string $string1, string $string2): int

Or, with custom edit costs:

levenshtein(
    string $string1,
    string $string2,
    int $insertion_cost,
    int $replacement_cost,
    int $deletion_cost
): int

Parameters

• $string1 — the first string to compare.
• $string2 — the second string to compare.
• $insertion_cost (optional) — cost of inserting a character. Default 1.
• $replacement_cost (optional) — cost of replacing a character. Default 1.
• $deletion_cost (optional) — cost of deleting a character. Default 1.

The function returns the Levenshtein distance as an int. With the default weights, that distance is symmetric — levenshtein($a, $b) equals levenshtein($b, $a).

Basic example

The output of this code is:

4

The function returns 4: turning "Hello" into "World" takes four substitutions (H→W, e→o, l→r, o→d); only the second l stays in place.

How the distance is built

Each edit operation counts as one step (with default weights). The classic textbook pair "kitten" → "sitting" needs three edits:

<?php
echo levenshtein("kitten", "sitting"); // 3
// k → s   (substitution)
// e → i   (substitution)
// (append) g   (insertion)
?>

Output:

3

levenshtein() is case-sensitive

Different letter casing counts as an edit, which often surprises people:

<?php
echo levenshtein("Hello", "hello"), "\n"; // 1  (H vs h)
echo levenshtein(strtolower("Hello"), strtolower("hello")), "\n"; // 0
?>

Output:

1
0

For case-insensitive comparison, normalize both strings with strtolower() (or mb_strtolower() for multibyte text) before calling levenshtein().

Weighting edits differently

When insertions, deletions, and replacements should not all cost the same, pass the three cost arguments. Here a deletion is made expensive:

<?php
// $insertion_cost = 1, $replacement_cost = 1, $deletion_cost = 5
echo levenshtein("cats", "cat", 1, 1, 5); // 5
?>

Output:

5

Removing the trailing s is a single deletion, but at a cost of 5 the reported distance is 5. This is handy when one kind of typo should be penalized more than another.

Practical use: "did you mean?" suggestions

A common real-world job is finding the closest known word to a user's input:

<?php
$input = "comit";
$dictionary = ["commit", "command", "comment", "compile"];

$best = null;
$bestDistance = PHP_INT_MAX;

foreach ($dictionary as $word) {
    $d = levenshtein($input, $word);
    if ($d < $bestDistance) {
        $bestDistance = $d;
        $best = $word;
    }
}

echo "Did you mean: {$best}? (distance {$bestDistance})";
?>

Output:

Did you mean: commit? (distance 1)

Gotchas

• Bytes, not characters. levenshtein() works on single bytes, so multibyte UTF-8 characters (accents, emoji, non-Latin scripts) can be miscounted. For accurate results with such text, transliterate or compare normalized ASCII.
• Long strings cost memory/time. Complexity is roughly proportional to the product of the two string lengths, so avoid it on very large inputs.
• Case and whitespace count. Trim and lowercase first if those differences should be ignored.

Related functions

• similar_text() — returns the number of matching characters (and an optional percentage) instead of an edit distance.
• soundex() and metaphone() — compare strings by how they sound rather than how they are spelled.
• strcmp() — a strict, binary-safe comparison that only tells you whether strings are equal or which sorts first.

Practice