Find exact and approximate duplicates of source translations.

find-i18n-duplicates.php

<?php

declare(strict_types=1);

namespace I18nDuplicateFinder;

use RuntimeException;
use SplFileObject;

use function array_key_exists;
use function fwrite;
use function iconv;
use function implode;
use function in_array;
use function is_array;
use function is_string;
use function preg_replace;
use function sprintf;
use function strtolower;
use function substr_replace;
use function trim;

use const PHP_EOL;
use const STDERR;
use const STDOUT;

class Application
{
    /** @throws RuntimeException */
    public function run(string $newTranslationsFilePath, string $referenceTranslationsFilePath): void
    {
        $referenceSources = [];
        foreach ($this->getCsvFileReader($referenceTranslationsFilePath) as $referenceLine) {
            if (is_array($referenceLine) && array_key_exists(0, $referenceLine) && is_string($referenceLine[0])) {
                $referenceSourceKey = $this->slugify($referenceLine[0]);
                if ($referenceSourceKey === null) {
                    continue;
                }
                if (!array_key_exists($referenceSourceKey, $referenceSources)) {
                    $referenceSources[$referenceSourceKey] = [];
                }
                if (!in_array($referenceLine[0], $referenceSources[$referenceSourceKey], true)) {
                    $referenceSources[$referenceSourceKey][] = $referenceLine[0];
                }
            }
        }
        foreach ($this->getCsvFileReader($newTranslationsFilePath) as $newLine) {
            if (is_array($newLine) && array_key_exists(0, $newLine) && is_string($newLine[0])) {
                $newSourceKey = $this->slugify($newLine[0]);
                if ($newSourceKey !== null && array_key_exists($newSourceKey, $referenceSources)) {
                    $this->printDuplicate($newLine[0], $referenceSources[$newSourceKey]);
                }
            }
        }
    }

    /** @throws RuntimeException */
    private function getCsvFileReader(string $filePath): SplFileObject
    {
        $splFileObject = new SplFileObject($filePath);
        $splFileObject->setFlags(SplFileObject::READ_CSV);
        return $splFileObject;
    }

    private function slugify(string $value): ?string
    {
        $result = preg_replace('~[^\pL\d]+~u', '-', $value);
        if (!is_string($result)) {
            return null;
        }
        $result = iconv('utf-8', 'us-ascii//TRANSLIT', $result);
        if ($result === false) {
            return null;
        }
        $result = preg_replace('~[^-\w]+~', '', $result);
        if (!is_string($result)) {
            return null;
        }
        $result = trim($result, '-');
        $result = preg_replace('~-+~', '-', $result);
        if (!is_string($result)) {
            return null;
        }
        $result = strtolower($result);
        if ($result === '') {
            return null;
        }
        return $result;
    }

    /** @param array<string> $existingSources */
    private function printDuplicate(string $newSource, array $existingSources): void
    {
        if (in_array($newSource, $existingSources, true)) {
            fwrite(STDOUT, $newSource . PHP_EOL);
        } else {
            $approximateDuplicatesOutput = implode(', ', substr_replace($existingSources, '≈ ', 0, 0));
            fwrite(STDOUT, sprintf('%s (%s)', $newSource, $approximateDuplicatesOutput) . PHP_EOL);
        }
    }
}

if ($argc !== 3) {
    fwrite(STDERR, sprintf(<<<'HELP_TEXT'
Usage: php %s NEW_TRANSLATIONS_FILE REFERENCE_TRANSLATIONS_FILE
Find exact and approximate duplicates of source translations from NEW_TRANSLATIONS_FILE in REFERENCE_TRANSLATIONS_FILE.
Assumes files are in CSV format and the first column contains source translations.

HELP_TEXT, $argv[0]));
    exit(1);
}
(new Application())->run($argv[1], $argv[2]);