ru_fonetics.php

<?php

$str = "к скалам бурым";
print "ru_soundex($str) = " . ru_soundex($str) . "<br />\r\n";

$str = "с каламбуром";
print "ru_soundex($str) = " . ru_soundex($str) . "<br />\r\n";

$str = "с калом бурым";
print "ru_soundex($str) = " . ru_soundex($str) . "<br />\r\n";

function str_to_translit($str)
{
    return strtr(
        $str,
        "abcdefghijklmnopqrstuvwxyz" .
        "абвгдеёжзийклмнопрстуфхцчшщъыьэюя",
        "abcdefghijklmnopqrstuvwxyz" .
        "abvgde" .
        "?*ziik" .
        "lmnopr" .
        "stufhc" .
        "4ww\"y`" .
        "eua"
    );
}

function ru_soundex($source)
{
    $literal = [
        'А' => ['status' => 'гласный', 'sound' => 'а', 'stressed' => 'а'],
        'Е' => ['status' => 'гласный', 'sound' => 'и', 'stressed' => 'э', 'АаЕеЁёИиОоУуЭэЮюЯяЬьЫыЪъ' => 'йэ'],
        'Ё' => ['status' => 'гласный', 'sound' => 'о', 'stressed' => 'о', 'АаЕеЁёИиОоУуЭэЮюЯяЬьЫыЪъ' => 'йо'],
        'И' => ['status' => 'гласный', 'sound' => 'и', 'stressed' => 'и'],
        'О' => ['status' => 'гласный', 'sound' => 'а', 'stressed' => 'о'],
        'У' => ['status' => 'гласный', 'sound' => 'у', 'stressed' => 'у'],
        'Ы' => ['status' => 'гласный', 'sound' => 'ы', 'stressed' => 'ы'],
        'Э' => ['status' => 'гласный', 'sound' => 'э', 'stressed' => 'э'],
        'Ю' => ['status' => 'гласный', 'sound' => 'у', 'stressed' => 'у', 'АаЕеЁёИиОоУуЭэЮюЯяЬьЫыЪъ' => 'йу'],
        'Я' => ['status' => 'гласный', 'sound' => 'а', 'stressed' => 'а', 'АаЕеЁёИиОоУуЭэЮюЯяЬьЫыЪъ' => 'йа'],
        'Б' => ['status' => 'согласный', 'sound' => 'б', 'КкПпСсТтФфХхЦцЧчШшЩщ' => 'п'],
        'В' => ['status' => 'согласный', 'sound' => 'в', 'КкПпСсТтФфХхЦцЧчШшЩщ' => 'ф'],
        'Г' => ['status' => 'согласный', 'sound' => 'Г', 'КкПпСсТтФфХхЦцЧчШшЩщ' => 'к'],
        'Д' => ['status' => 'согласный', 'sound' => 'д', 'КкПпСсТтФфХхЦцЧчШшЩщ' => 'т'],
        'Ж' => ['status' => 'согласный', 'sound' => 'ж', 'КкПпСсТтФфХхЦцЧчШшЩщ' => 'ш'],
        'З' => ['status' => 'согласный', 'sound' => 'з', 'КкПпСсТтФфХхЦцЧчШшЩщ' => 'с'],
        'Й' => ['status' => 'согласный', 'sound' => 'й'],
        'К' => ['status' => 'согласный', 'sound' => 'к', 'КкПпСсТтФфХхЦцЧчШшЩщ' => ''],
        'Л' => ['status' => 'согласный', 'sound' => 'л'],
        'М' => ['status' => 'согласный', 'sound' => 'м'],
        'Н' => ['status' => 'согласный', 'sound' => 'н'],
        'П' => ['status' => 'согласный', 'sound' => 'п', 'КкПпСсТтФфХхЦцЧчШшЩщ' => ''],
        'Р' => ['status' => 'согласный', 'sound' => 'р'],
        'С' => ['status' => 'согласный', 'sound' => 'с'],
        'Т' => ['status' => 'согласный', 'sound' => 'т', 'КкПпСсТтФфХхЦцЧчШшЩщ' => ''],
        'Ф' => ['status' => 'согласный', 'sound' => 'ф', 'КкПпСсТтФфХхЦцЧчШшЩщ' => ''],
        'Х' => ['status' => 'согласный', 'sound' => 'х'],
        'Ц' => ['status' => 'согласный', 'sound' => 'ц'],
        'Ч' => ['status' => 'согласный', 'sound' => 'чь'],
        'Ш' => ['status' => 'согласный', 'sound' => 'ш'],
        'Щ' => ['status' => 'согласный', 'sound' => 'щь'],
        'Ъ' => ['status' => 'знак', 'sound' => ' '],
        'Ь' => ['status' => 'знак', 'sound' => 'ь'],
        'ТС' => ['status' => 'сочетание', 'sound' => 'ц'],
        'ТЬС' => ['status' => 'сочетание', 'sound' => 'ц'],
        'ШЬ' => ['status' => 'сочетание', 'sound' => 'ш'],
        'СОЛНЦ' => ['status' => 'сочетание', 'sound' => 'сонц'],
        'ЯИЧНИЦ' => ['status' => 'сочетание', 'sound' => 'еишниц'],
        'КОНЕЧНО' => ['status' => 'сочетание', 'sound' => 'канешно'],
        'ЧТО' => ['status' => 'сочетание', 'sound' => 'што'],
        'ЗАЯ' => ['status' => 'сочетание', 'sound' => 'зайэ']
    ];

    $v_pattern = 'АаЕеЁёИиОоУуЭэЮюЯяЬьЫыЪъ';
    $sound = mb_strtoupper($source);

    foreach (array_filter($literal, fn($item) => $item["status"] === "сочетание") as $sign => $translate) {
        $sound = str_replace($sign, $translate["sound"], $sound);
    }

    foreach (array_filter($literal, fn($item) => $item["status"] === "знак") as $sign => $translate) {
        $sound = str_replace($sign, $translate["sound"], $sound);
    }

    $words = preg_split('~[,.\~`1234567890-=\~!@#$%^&*()_+|{}\]\];:\'"<>/? ]~', $sound, -1, PREG_SPLIT_NO_EMPTY);

    foreach (array_filter($literal, fn($item) => $item["status"] === "гласный") as $sign => $translate) {
        foreach ($words as &$word) {
            $vowel = preg_match_all("~[$v_pattern]~", $word, $del_me);
            $cur_pos = 0;
            $cur_vowel = 0;
            while (false !== $cur_pos = strpos($word, $sign, $cur_pos)) {
                $cur_vowel++;
                if (sizeof($translate) == 4 && ($cur_pos === 0 || strpos($v_pattern, $word[$cur_pos - 1]))) {
                    $word = substr_replace($word, $translate[$v_pattern], $cur_pos, 1);
                } elseif ($vowel == 1) {
                    $word = substr_replace($word, $translate["stressed"], $cur_pos, 1);
                } elseif ($vowel == 2 && $cur_vowel == 1) {
                    $word = substr_replace($word, $translate["stressed"], $cur_pos, 1);
                } elseif ($vowel >= 3 && $cur_vowel == $vowel - 2) {
                    $word = substr_replace($word, $translate["stressed"], $cur_pos, 1);
                } else {
                    $word = substr_replace($word, $translate["sound"], $cur_pos, 1);
                }
                $cur_pos++;
            }
        }
    }

    $sound = implode(' ', $words);

    foreach (array_filter($literal, fn($item) => $item["status"] === "согласный") as $sign => $translate) {
        $cur_pos = 0;
        while (false !== $cur_pos = strpos($sound, $sign, $cur_pos)) {
            if (sizeof($translate) == 3) {
                $keys = array_keys($translate);
                $x = array_pop($keys);
                if (strpos($x, $sound[$cur_pos + 1]) || $cur_pos === strlen($sound)) {
                    $sound = substr_replace($sound, $translate[$x], $cur_pos, 1);
                } elseif ($sound[$cur_pos] === $sound[$cur_pos + 1]) {
                    $sound = substr_replace($sound, $translate["sound"], $cur_pos, 2);
                } else {
                    $sound = substr_replace($sound, $translate["sound"], $cur_pos, 1);
                }
            } else {
                $sound = substr_replace($sound, $translate["sound"], $cur_pos, 1);
            }
            $cur_pos++;
        }
    }

    $sound = preg_replace('~[,.\~`1234567890-=\~!@#$%^&*()_+|{}\]\];:\'"<>/? ]~', '', $sound);
    $res = mb_strtoupper($source[0]) . mb_substr(soundex(str_to_translit($sound)), 1);

    return $res; 
}
?>