Finding the 3 most occurring substrings in a string in PHP - php

Question

I wanted to go over the thought process of these since I am not sure how to improve this. I have a string that are separated by commas and they have reoccurring substrings and I want to find the 3 most occurring substrings.
I was going to explode the string by commas into an array.
Perform a substr_count in the original string for each element in the array and store it in a separate array to store the counts? (Not sure how to improve this since that would create duplicate counts for the same substring)
Perform a max on the array to find the first, second, and third most occurring substrings.
Return an array with the first, second, and third most occurring substrings.
I am guessing after I perform an explode, I can do a quick sort and go from there?
This is what I have tried so far:
$result = findThreeMostOccuringStrings("apple, apple, berry, cherry, cherry, cherry, dog, dog, dog");
var_dump($result);
function findThreeMostOccuringStrings($str){
$first = PHP_INT_MIN;
$second = PHP_INT_MIN;
$third = PHP_INT_MIN;
$arr = explode(",", $str);
for ($i = 0; $i < count($str); $i++){
$arrIdx[] = substr_count($arr[$i]);
}
$first = max($arrIdx);
$arrIdx[$first] = -1;
$second = max($arrIdx);
$arrIdx[$first] = -1;
$third = max($arrIdx);
$arrIdx[$first] = -1;
$threeMostOccuringStrings = array($first, $second, $third);
return $threeMostOccuringStrings;
}

Answer 1

If by substring you mean only the strings separated by commas and not substrings of these, use array_count_values after explode

Answer 2

If you're looking for an efficient way to solve substring search, the answer is a Trie, or prefix tree. This basically reduces the look up time to search for a substring as it creates a deterministic path for all prefixes (i.e. a prefix tree).
Consider the string "A cat does not categorize its food while among other cats." Here the substrings categorize and cats all share the same prefix of cat. So finding the most frequented substring in a prefix tree is easy if you count the number of EOW nodes stemming from each branch in the root.
Building the tree is also quite trivial.
function insertString($str, $trie) {
$str = strtolower($str); // normalize case
$node = $trie;
foreach(str_split($str) as $letter) {
if (!isset($node->$letter)) {
$node->$letter = new stdClass;
}
$node = $node->$letter;
}
$node->EOW = true; // place EOL node
}
function countNodes($branch) {
$n = 0;
foreach($branch as $e => $node) {
if ($node instanceof stdClass) {
$n += countNodes($node);
} elseif ($e === 'EOW') {
$n++;
}
}
return $n;
}
$trie = new stdClass;
$str = "A cat does not categorize its food while among other cats";
foreach(explode(' ', $str) as $word) {
insertString($word, $trie);
}
$s = [];
foreach($trie as $n => $rootNodes) {
$s[$n] = countNodes($rootNodes);
}
var_dump($s);
Which should give you...
array(8) {
["a"]=>
int(2)
["c"]=>
int(3)
["d"]=>
int(1)
["n"]=>
int(1)
["i"]=>
int(1)
["f"]=>
int(1)
["w"]=>
int(1)
["o"]=>
int(1)
}
From there you can see the root branch c has the highest number of substrings (which if walked match cat, cats, and categorize.

Answer 3

Based on your post and the comments, you actually want to tally terms in a comma delimited list of terms, and then rank them, so: let's just do that, using an associative array for tallying and arsort to sort that associative array by value, in reverse order (so that the highest counts are at the start of the array):
function rank($input) {
$terms = explode(',', $input);
$ranked = array();
foreach($terms as $word) {
$word = trim($word);
if (!isset($ranked[$word])) {
$ranked[$word] = 0;
}
$ranked[$word]++;
}
arsort($ranked);
return $ranked;
}
So if we run that through print_r(rank("apple, apple, berry, cherry, cherry, cherry, dog, dog, dog")) we get:
Array
(
[dog] => 3
[cherry] => 3
[apple] => 2
[berry] => 1
)
Splendid.