Fisharebest\Webtrees\Soundex::daitchMokotoffWord()

Complexity

Conditions	24
Paths	1344

Size

Total Lines	110
Code Lines	61

Duplication

Lines	0
Ratio	0 %

Importance

Changes

0

<?php
    private static function daitchMokotoffWord(string $name): array
    {
        // Apply special transformation rules to the input string
        $name = I18N::strtoupper($name);
        foreach (self::TRANSFORM_NAMES as $transformRule) {
            $name = str_replace($transformRule[0], $transformRule[1], $name);
        }

        // Initialize
        $name_script = I18N::textScript($name);
        $noVowels    = $name_script === 'Hebr' || $name_script === 'Arab';

        $lastPos         = strlen($name) - 1;
        $currPos         = 0;
        $state           = 1; // 1: start of input string, 2: before vowel, 3: other
        $result          = []; // accumulate complete 6-digit D-M codes here
        $partialResult   = []; // accumulate incomplete D-M codes here
        $partialResult[] = ['!']; // initialize 1st partial result  ('!' stops "duplicate sound" check)

        // Loop through the input string.
        // Stop when the string is exhausted or when no more partial results remain
        while ($partialResult !== [] && $currPos <= $lastPos) {
            // Find the DM coding table entry for the chunk at the current position
            $thisEntry = substr($name, $currPos, self::MAXCHAR); // Get maximum length chunk
            while ($thisEntry !== '') {
                if (isset(self::DM_SOUNDS[$thisEntry])) {
                    break;
                }
                $thisEntry = substr($thisEntry, 0, -1); // Not in table: try a shorter chunk
            }
            if ($thisEntry === '') {
                $currPos++; // Not in table: advance pointer to next byte
                continue; // and try again
            }

            $soundTableEntry = self::DM_SOUNDS[$thisEntry];
            $workingResult   = $partialResult;
            $partialResult   = [];
            $currPos += strlen($thisEntry);

            // Not at beginning of input string
            if ($state !== 1) {
                if ($currPos <= $lastPos) {
                    // Determine whether the next chunk is a vowel
                    $nextEntry = substr($name, $currPos, self::MAXCHAR); // Get maximum length chunk
                    while ($nextEntry !== '') {
                        if (isset(self::DM_SOUNDS[$nextEntry])) {
                            break;
                        }
                        $nextEntry = substr($nextEntry, 0, -1); // Not in table: try a shorter chunk
                    }
                } else {
                    $nextEntry = '';
                }
                if ($nextEntry !== '' && self::DM_SOUNDS[$nextEntry][0] !== '0') {
                    $state = 2;
                } else {
                    // Next chunk is a vowel
                    $state = 3;
                }
            }

            while ($state < count($soundTableEntry)) {
                // empty means 'ignore this sound in this state'
                if ($soundTableEntry[$state] === '') {
                    foreach ($workingResult as $workingEntry) {
                        $tempEntry                        = $workingEntry;
                        $tempEntry[count($tempEntry) - 1] .= '!'; // Prevent false 'doubles'
                        $partialResult[]                  = $tempEntry;
                    }
                } else {
                    foreach ($workingResult as $workingEntry) {
                        if ($soundTableEntry[$state] !== $workingEntry[count($workingEntry) - 1]) {
                            // Incoming sound isn't a duplicate of the previous sound
                            $workingEntry[] = $soundTableEntry[$state];
                        } elseif ($noVowels) {
                            // Incoming sound is a duplicate of the previous sound
                            // For Hebrew and Arabic, we need to create a pair of D-M sound codes,
                            // one of the pair with only a single occurrence of the duplicate sound,
                            // the other with both occurrences
                            $workingEntry[] = $soundTableEntry[$state];
                        }

                        if (count($workingEntry) < 7) {
                            $partialResult[] = $workingEntry;
                        } else {
                            // This is the 6th code in the sequence
                            // We're looking for 7 entries because the first is '!' and doesn't count
                            $tempResult = str_replace('!', '', implode('', $workingEntry));
                            // Only return codes from recognisable sounds
                            if ($tempResult !== '') {
                                $result[] = substr($tempResult . '000000', 0, 6);
                            }
                        }
                    }
                }
                $state += 3; // Advance to next triplet while keeping the same basic state
            }
        }

        // Zero-fill and copy all remaining partial results
        foreach ($partialResult as $workingEntry) {
            $tempResult = str_replace('!', '', implode('', $workingEntry));
            // Only return codes from recognisable sounds
            if ($tempResult !== '') {
                $result[] = substr($tempResult . '000000', 0, 6);
            }
        }

        return $result;
    }