Clean XLIFF (php test)

gistfile1.php

#!/usr/bin/env php
<?php

$command_help = "clean_xliff - Clean localized XLIFF file\n" .
                "Usage: clean_xliff [reference_file] [localized file]\n\n" .
                "Example: clean_xliff en-US/firefox-ios.xliff it/firefox-ios.xliff\n" .
                "(remove obsolete strings from it/firefox-ios.xliff)\n";
$missing_parameter = "This command needs more parameters, please check mark_active --help.\n";

if (php_sapi_name() != 'cli') {
    die('This command can only be used in CLI mode.');
}

if (isset($argv[1])) {
    if (in_array($argv[1], ['-h', '--help']) && isset($command_help)) {
        die($command_help);
    }
}

if (count($argv) < 3) {
    die($missing_parameter);
}


$ref_root = simplexml_load_file($argv[1]);
$ref_root->registerXPathNamespace('x', $ref_root->getNamespaces()['']);
$l10n_root = simplexml_load_file($argv[2]);
$l10n_root->registerXPathNamespace('x', $l10n_root->getNamespaces()['']);
$file_changed = false;

// Check file elements
$ref_files = [];
$ref_strings = [];
if (count($ref_root)) {
    $files = $ref_root->xpath('//x:file');
    foreach ($files as $file_element) {
        // Store filename
        $filename = (string) $file_element['original'];
        $ref_files[] = $filename;

        // Store strings. Multidimensional array, because the same ID can be
        // in multiple files.
        $ref_strings[$filename] = [];
        foreach ($file_element->body->{'trans-unit'} as $trans_unit) {
            $ref_strings[$filename][] = (string) $trans_unit['id'];
        }
    }
}

$l10n_files = [];
$l10n_strings = [];
if (count($l10n_root)) {
    $files = $l10n_root->xpath('//x:file');
    foreach ($files as $file_element) {
        $filename = (string) $file_element['original'];
        $l10n_files[] = $filename;
        if (! in_array($filename, $ref_files)) {
            echo "Removing obsolete <file> element ({$filename})\n";
            unset($file_element[0]);
            $file_changed = true;
        } else {
            // File is not obsolete, check strings inside files
            $l10n_strings[$filename] = [];

            // I use a clone, can't unset nodes directly without breaking the iterator
            $iterator_strings = clone $file_element->body->{'trans-unit'};
            foreach ($iterator_strings as $trans_unit) {
                $string_id = (string) $trans_unit['id'];
                $l10n_strings[$filename][] = $string_id;
                if (! in_array($string_id, $ref_strings[$filename])) {
                    echo "Removing obsolete string from {$filename}. String ID: {$string_id}.\n";
                    $f = $l10n_root->xpath("//x:file[@original='{$filename}']/x:body/x:trans-unit[@id='{$string_id}']");
                    unset($f[0][0]);
                    $file_changed = true;
                }
            }
        }
    }
}

if ($file_changed) {
    // Saving back the XML file, removing empty lines
    $new_output = preg_split('/$\R?^/m', $l10n_root->asXml());
    foreach ($new_output as $key => $line) {
        if (trim($line) == '') {
            unset($new_output[$key]);
        }
    }
    $new_output = implode($new_output, "\n");
    file_put_contents($argv[2], $new_output);
} else {
    echo "\nNo obsolete strings found.\n";
}

// Warn about missing files or strings
$missing_files = array_diff($ref_files, $l10n_files);
if (count($missing_files) > 0) {
    echo "\nMissing <file> elements in localized file:\n";
    foreach ($missing_files as $filename) {
        echo "* {$filename}\n";
    }
}

foreach ($ref_strings as $filename => $strings) {
    $missing_strings = array_diff($strings, $l10n_strings[$filename]);
    if (count($missing_strings) > 0) {
        echo "\nMissing strings in localized file {$filename}:\n";
        foreach ($missing_strings as $filename) {
            echo "* {$filename}\n";
        }
    }
}