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

set_error_handler(function ($errno, $errstr, $errfile = null, $errline = null) {
    $message = $errstr ? $errstr : "Error {$errno}";
    if ($errfile) {
        $message .= "\nFile: {$errfile}";
        if ($errline) {
            $message .= "\nLine: {$errline}";
        }
    }
    throw new RuntimeException($message);
});

error_reporting(-1);

/**
 * @param string[] $argv
 *
 * @throws RuntimeException
 *
 * @return void
 */
function main(array $argv)
{
    $options = new Options($argv);
    if (!is_dir($options->outputDir) && !mkdir($options->outputDir, 0777, true)) {
        throw new RuntimeException("Cannot create output directory: {$options->outputDir}\n");
    }
    $options->outputDir = str_replace(DIRECTORY_SEPARATOR, '/', realpath($options->outputDir));
    $documentStorage = new DocumentStorage($options);
    echo 'Processing languages... ';
    $languages = new Languages($options, $documentStorage);
    echo "done.\n";
    echo 'Processing scripts... ';
    $scripts = new Scripts($options, $documentStorage);
    echo "done.\n";
    echo 'Processing territories... ';
    $territories = new Territories($options, $documentStorage);
    echo "done.\n";
    echo 'Processing plural rules... ';
    $plurals = new Plurals($options, $documentStorage, $languages);
    echo "done.\n";
    echo 'Saving... ';

    $languages->save();
    $scripts->save();
    $territories->save();
    $plurals->save();

    echo "done.\n";
}

class Options
{
    /**
     * @var string
     */
    public $cldrVersion;

    /**
     * @var string
     */
    public $outputDir;

    public function __construct(array $argv)
    {
        if (array_intersect($argv, array('-h', '--help'))) {
            $this->showSyntax($argv[0], 0);
        }
        $this->outputDir = $this->getDefaultOutputDir();
        switch (count($argv)) {
            case 3:
                $this->outputDir = str_replace(DIRECTORY_SEPARATOR, '/', $argv[2]);
                // no break
            case 2:
                $this->cldrVersion = $argv[1];
                if (!preg_match('/^\d+(\.\d+)?(-(alpha|beta)\d+)?$/', $this->cldrVersion)) {
                    throw new RuntimeException("{$this->cldrVersion} is not a valid CLDR version identifier");
                }
                break;
            default:
                $this->showSyntax($argv[0], 1);
        }
    }

    /**
     * @param string $programName
     * @param int $exitCode
     *
     * @return never
     */
    private function showSyntax($programName, $exitCode)
    {
        $programName = str_replace('/', DIRECTORY_SEPARATOR, $programName);
        $defaultOutputDir = str_replace('/', DIRECTORY_SEPARATOR, $this->getDefaultOutputDir());

        echo <<<EOT
Syntax: {$programName} <cldr-version> [output-dir]

Arguments:
  cldr-version: the version of the CLDR data.
    Examples:
      47
      47-beta2
      47-alpha1
      46.1
      46.1-beta1
  output-dir: the directory where the data will be written to
    Default: {$defaultOutputDir}

EOT;
        exit($exitCode);
    }

    /**
     * @return string
     */
    private function getDefaultOutputDir()
    {
        return str_replace(DIRECTORY_SEPARATOR, '/', dirname(__DIR__)) . '/src/cldr-data';
    }
}

class DocumentStorage
{
    /**
     * @var string
     */
    private $baseUrl;

    private $context;

    private $cache;

    public function __construct(Options $options)
    {
        $this->baseUrl = 'https://raw.githubusercontent.com/unicode-org/cldr/refs/tags/release-' . str_replace('.', '-', $options->cldrVersion);
        $this->context = stream_context_create(array(
            'http' => array(
                'follow_location' => 1,
                'ignore_errors' => false,
            ),
        ));
        $this->cache = array();
    }

    /**
     * @param string $path
     *
     * @throws RuntimeException
     *
     * @return DOMDocument
     */
    public function get($path)
    {
        if (!isset($this->cache[$path])) {
            $xml = $this->fetch($path);
            $doc = $this->loadXml($xml);
            $this->cache[$path] = $doc;
        }

        return $this->cache[$path];
    }

    /**
     * @param string $path
     *
     * @throws RuntimeException
     *
     * @return string
     */
    private function fetch($path)
    {
        $url = $this->baseUrl . '/' . ltrim($path, '/');
        set_error_handler(function () {}, -1);
        $content = file_get_contents($url, false, $this->context);
        restore_error_handler();
        if ($content === false) {
            $details = '';
            /** @var array $http_response_header */
            if (!empty($http_response_header)) {
                $details = " - {$http_response_header[0]}";
            }
            throw new RuntimeException("Failed to download from {$url}{$details}");
        }

        return $content;
    }

    /**
     * @param string $xml
     *
     * @throws RuntimeException
     *
     * @return DOMDocument
     */
    private function loadXml($xml)
    {
        $doc = new DOMDocument();
        libxml_clear_errors();
        $restore = libxml_use_internal_errors(true);
        $loaded = $doc->loadXML($xml);
        $errors = libxml_get_errors();
        libxml_use_internal_errors($restore);
        $lines = array();
        foreach ($errors as $error) {
            $lines[] = "{$error->message} at line {$error->line}";
        }
        if (!$loaded || $errors !== array()) {
            throw new RuntimeException("Failed to parse XML:\n" . implode("\n", $lines));
        }

        return $doc;
    }
}

abstract class Processor
{
    /**
     * @var Options
     */
    protected $options;

    /**
     * @var array
     */
    protected $data;

    /**
     * @var DocumentStorage
     */
    private $documentStorage;

    /**
     * @var string
     */
    private $path;

    protected function __construct(Options $options, DocumentStorage $documentStorage, $path)
    {
        $this->options = $options;
        $this->documentStorage = $documentStorage;
        $this->path = ltrim($path, '/');
        $doc = $this->documentStorage->get($this->path);
        $this->data = $this->parse($doc);
    }

    /**
     * @return void
     */
    public function save()
    {
        $file = $this->getOutputFile();
        $dir = dirname($file);
        if (!is_dir($dir) && !mkdir($dir, 0777, true)) {
            throw new RuntimeException("Cannot create directory: {$dir}");
        }
        $flags = 0;
        if (defined('JSON_UNESCAPED_SLASHES')) {
            $flags |= JSON_UNESCAPED_SLASHES;
        }
        if (defined('JSON_UNESCAPED_UNICODE')) {
            $flags |= JSON_UNESCAPED_UNICODE;
        }
        if (defined('JSON_PRETTY_PRINT')) {
            $flags |= JSON_PRETTY_PRINT;
        }
        if (defined('JSON_THROW_ON_ERROR')) {
            $flags |= JSON_THROW_ON_ERROR;
        }
        $json = json_encode($this->data, $flags);
        if (!file_put_contents($file, $json)) {
            throw new RuntimeException("Failed to write to file: {$file}");
        }
    }

    /**
     * @return array
     */
    abstract protected function parse(DOMDocument $doc);

    /**
     * @return void
     */
    protected function sortByKeyWithPossiblyAlt(array &$data)
    {
        uksort($data, function ($a, $b) {
            $aAlt = strpos($a, '-alt-') !== false;
            $bAlt = strpos($b, '-alt-') !== false;
            if ($aAlt !== $bAlt) {
                if (strpos("{$a}-alt-", $b) === 0) {
                    return 0;
                }
                if (strpos($a, "{$b}-alt-") === 0) {
                    return -1;
                }
            }

            return strcasecmp($a, $b);
        });
    }

    /**
     * @return string
     */
    abstract protected function getOutputRelativeFileName();

    /**
     * @param string $xml
     *
     * @return DOMDocument
     */
    private function loadXml($xml)
    {
        $doc = new DOMDocument();
        libxml_clear_errors();
        $restore = libxml_use_internal_errors(true);
        $loaded = $doc->loadXML($xml);
        $errors = libxml_get_errors();
        libxml_use_internal_errors($restore);
        $lines = array();
        foreach ($errors as $error) {
            $lines[] = "{$error->message} at line {$error->line}";
        }
        if (!$loaded || $errors !== array()) {
            throw new RuntimeException("Failed to parse XML:\n" . implode("\n", $lines));
        }
        return $doc;
    }

    /**
     * @return string
     */
    private function getOutputFile()
    {
        return $this->options->outputDir . '/' . ltrim($this->getOutputRelativeFileName(), '/');
    }
}

class Plurals extends Processor
{
    /**
     * @var Languages
     */
    private $languages;

    public function __construct(Options $options, DocumentStorage $documentStorage, Languages $languages)
    {
        $this->languages = $languages;
        parent::__construct($options, $documentStorage, 'common/supplemental/plurals.xml');
    }

    /**
     * {@inheritdoc}
     *
     * @see Processor::parse()
     */
    protected function parse(DOMDocument $doc)
    {
        $data = array();
        $xpath = new DOMXPath($doc);
        $xPluralRulesList = $xpath->query('/supplementalData/plurals[@type="cardinal"]/pluralRules');
        $definedLanguageIDs = $this->languages->getDefinedLanguageIDs();
        $knownMissingLanguages = array(
            'guw', // Gun
            'lld', //  Dolomitic Ladin
            'hnj', // Hmong Njua
            'nah', // Nahuatl
            'smi', // Sami
        );
        $replacements = array(
            'in' => 'id', // Former Indonesian
            'iw' => 'he', // Former Hebrew
            'jw' => 'jv', // Former Javanese
            'ji' => 'yi', // Former Yiddish
            'mo' => 'ro-MD', // former Moldavian
            'bh' => '', // Former Bihari: dismissed because it can be 'bho', 'mai' or 'mag'
            // Just a CLDR placeholder
            'root' => '',
        );
        $unrecognizedLocaleCodes = array();
        foreach ($xPluralRulesList as $xPluralRules) {
            $locales = preg_split('/\s+/', (string) $xPluralRules->getAttribute('locales'), -1, PREG_SPLIT_NO_EMPTY);
            if ($locales === array()) {
                throw new RuntimeException('No locales found in pluralRules element');
            }
            $elements = array(
                'pluralRule-count-zero' => null,
                'pluralRule-count-one' => null,
                'pluralRule-count-two' => null,
                'pluralRule-count-few' => null,
                'pluralRule-count-many' => null,
                'pluralRule-count-other' => null,
            );
            foreach ($xPluralRules->childNodes as $xPluralRule) {
                if (!$xPluralRule instanceof DOMElement) {
                    continue;
                }
                if ($xPluralRule->tagName !== 'pluralRule') {
                    throw new RuntimeException("Unexpected element: {$xPluralRule->tagName}");
                }
                $count = (string) $xPluralRule->getAttribute('count');
                if ($count === '') {
                    throw new RuntimeException('Missing count attribute');
                }
                $key = "pluralRule-count-{$count}";
                if (!array_key_exists($key, $elements)) {
                    throw new RuntimeException("Unknown count: {$count}");
                }
                if ($elements[$key] !== null) {
                    throw new RuntimeException("Duplicate count: {$count}");
                }
                $elements[$key] = $xPluralRule->textContent;
            }
            $elements = array_filter($elements, function ($value) {
                return $value !== null;
            });
            if ($elements === array()) {
                throw new RuntimeException('No plural rules found');
            }
            foreach ($locales as $locale) {
                $locale = str_replace('_', '-', $locale);
                $overwrite = true;
                if (isset($data[$locale]) && array_search($locale, $replacements, true) === false) {
                    throw new RuntimeException("Duplicate locale: {$locale}");
                }
                if (!in_array($locale, $definedLanguageIDs, true) && !in_array($locale, $knownMissingLanguages, true)) {
                    if (!isset($replacements[$locale])) {
                        $unrecognizedLocaleCodes[] = $locale;
                        continue;
                    }
                    $locale = $replacements[$locale];
                    if ($locale === '') {
                        continue;
                    }
                    $overwrite = false;
                }
                if ($overwrite || !isset($data[$locale])) {
                    $data[$locale] = $elements;
                }
            }
        }
        if ($unrecognizedLocaleCodes !== array()) {
            throw new RuntimeException("The following locales are not defined:\n- " . implode("\n- ", $unrecognizedLocaleCodes));
        }
        if ($data === array()) {
            throw new RuntimeException('No plural rules found');
        }
        $this->sortByKeyWithPossiblyAlt($data);

        return array(
            'supplemental' => array(
                'version' => array(
                    '_cldrVersion' => $this->options->cldrVersion,
                ),
                'plurals-type-cardinal' => $data,
            ),
        );
    }

    /**
     * {@inheritdoc}
     *
     * @see Processor::getOutputRelativeFileName()
     */
    protected function getOutputRelativeFileName()
    {
        return 'supplemental/plurals.json';
    }
}

abstract class LocaleDisplayName extends Processor
{
    public function __construct(Options $options, DocumentStorage $documentStorage)
    {
        parent::__construct($options, $documentStorage, 'common/main/en.xml');
    }

    /**
     * {@inheritdoc}
     *
     * @see Processor::parse()
     */
    protected function parse(DOMDocument $doc)
    {
        $data = array();
        $xpath = new DOMXPath($doc);
        $xElementList = $xpath->query($this->getXPathSelector());
        foreach ($xElementList as $xElement) {
            $type = (string) $xElement->getAttribute('type');
            if ($type === '') {
                throw new RuntimeException('Missing type attribute');
            }
            $key = str_replace('_', '-', $type);
            $alt = (string) $xElement->getAttribute('alt');
            if ($alt !== '') {
                $key = "{$key}-alt-{$alt}";
            }
            if (isset($data[$key])) {
                throw new RuntimeException("Duplicate key: {$key}");
            }
            $data[$key] = (string) $xElement->textContent;
        }
        if ($data === array()) {
            throw new RuntimeException('No elements found');
        }
        $this->sortByKeyWithPossiblyAlt($data);

        return array(
            'main' => array(
                'en-US' => array(
                    'identity' => array(
                        'version' => array(
                            '_cldrVersion' => $this->options->cldrVersion,
                        ),
                        'language' => 'en',
                        'territory' => 'US',
                    ),
                    'localeDisplayNames' => array(
                        $this->getExportedNodeName() => $data,
                    ),
                ),
            ),
        );
    }

    /**
     * @return string
     */
    abstract protected function getXPathSelector();

    /**
     * @return string
     */
    abstract protected function getExportedNodeName();
}

class Languages extends LocaleDisplayName
{
    /**
     * @return string[]
     */
    public function getDefinedLanguageIDs()
    {
        return array_values(array_filter(
            array_keys($this->data['main']['en-US']['localeDisplayNames'][$this->getExportedNodeName()]),
            function ($key) {
                return strpos((string) $key, '-alt-') === false;
            }
        ));
    }

    /**
     * {@inheritdoc}
     *
     * @see LocaleDisplayName::getXPathSelector()
     */
    protected function getXPathSelector()
    {
        return '/ldml/localeDisplayNames/languages/language';
    }

    /**
     * {@inheritdoc}
     *
     * @see LocaleDisplayName::getExportedNodeName()
     */
    protected function getExportedNodeName()
    {
        return 'languages';
    }

    /**
     * {@inheritdoc}
     *
     * @see Processor::getOutputRelativeFileName()
     */
    protected function getOutputRelativeFileName()
    {
        return 'main/en-US/languages.json';
    }
}

class Scripts extends LocaleDisplayName
{
    /**
     * {@inheritdoc}
     *
     * @see LocaleDisplayName::getXPathSelector()
     */
    protected function getXPathSelector()
    {
        return '/ldml/localeDisplayNames/scripts/script';
    }

    /**
     * {@inheritdoc}
     *
     * @see LocaleDisplayName::getExportedNodeName()
     */
    protected function getExportedNodeName()
    {
        return 'scripts';
    }

    /**
     * {@inheritdoc}
     *
     * @see Processor::getOutputRelativeFileName()
     */
    protected function getOutputRelativeFileName()
    {
        return 'main/en-US/scripts.json';
    }
}

class Territories extends LocaleDisplayName
{
    /**
     * {@inheritdoc}
     *
     * @see LocaleDisplayName::getXPathSelector()
     */
    protected function getXPathSelector()
    {
        return '/ldml/localeDisplayNames/territories/territory';
    }

    /**
     * {@inheritdoc}
     *
     * @see LocaleDisplayName::getExportedNodeName()
     */
    protected function getExportedNodeName()
    {
        return 'territories';
    }

    /**
     * {@inheritdoc}
     *
     * @see Processor::getOutputRelativeFileName()
     */
    protected function getOutputRelativeFileName()
    {
        return 'main/en-US/territories.json';
    }
}

try {
    main($argv);
} catch (RuntimeException $e) {
    fwrite(STDERR, $e->getMessage() . "\n");
    exit(1);
}