<?php
/**
 * Wikidata-backed acronym overrides for NaturalTitle.
 *
 * Filtering pipeline before querying Wikidata:
 *   1. A common-English word list (google-10000-english). If the token is a
 *      common English word, it is never treated as an acronym.
 * Only tokens that survive that filter are queried against Wikidata's
 * P1813 "short name" property. Results are cached locally for 30 days.
 *
 * Word list source: https://github.com/first20hours/google-10000-english
 */

if (!defined('DOKU_INC')) die();

use RJT\TitleCase\Overrides;

class NaturalTitleAcronyms {

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

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

    /** @var array<string,bool> */
    private $words = [];

    /** @var int Cache TTL in seconds (30 days) */
    private $ttl = 2592000;

    /** @var string */
    private $userAgent = 'DokuWiki-NaturalTitle/1.0 (https://grimore.org)';

    public function __construct() {
        $this->cacheFile = DOKU_INC . 'data/meta/naturaltitle_acronyms.json';
        $this->wordsFile = __DIR__ . '/english-words.txt';
        $this->loadWordList($this->wordsFile, $this->words);
    }

    /**
     * Build an Overrides object for the tokens in a title.
     *
     * @param string $title A pre-normalized title (result of titleCase())
     * @return Overrides
     */
    public function buildOverrides(string $title): Overrides {
        $tokens   = preg_split('/\s+/', $title, -1, PREG_SPLIT_NO_EMPTY);
        $cache    = $this->loadCache();
        $titleMap = [];

        foreach ($tokens as $token) {
            // Strip trailing punctuation so "API," still matches
            $bare = rtrim($token, '.,;:!?');
            if (!preg_match('/^[A-Za-z]{2,8}$/', $bare)) continue;

            $lower = strtolower($bare);

            // Common English word? Not an acronym.
            if (isset($this->words[$lower])) continue;

            $upper = strtoupper($bare);

            if (array_key_exists($upper, $cache)) {
                if ($cache[$upper] === true) {
                    $titleMap[$lower] = $upper;
                }
                continue;
            }

            $isAcronym = $this->lookupWikidata($upper);
            $cache[$upper] = $isAcronym;
            if ($isAcronym) {
                $titleMap[$lower] = $upper;
            }
        }

        $this->saveCache($cache);
        return new Overrides($titleMap, [], []);
    }

    /**
     * Load a plain-text word list (one word per line) into the target set.
     */
    private function loadWordList(string $path, array &$target): void {
        if (!file_exists($path)) return;
        $raw = file($path, FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES);
        if ($raw === false) return;
        foreach ($raw as $line) {
            $line = trim(strtolower($line));
            if ($line === '' || $line[0] === '#') continue;
            $target[$line] = true;
        }
    }

    /**
     * Query Wikidata for whether a term is a known acronym (P1813 short name).
     */
    private function lookupWikidata(string $term): bool {
        $query = sprintf(
            'SELECT ?item WHERE { ?item wdt:P1813 "%s"@en . } LIMIT 1',
            addslashes($term)
        );

        $url = 'https://query.wikidata.org/sparql?format=json&query=' . urlencode($query);
        $ctx = stream_context_create([
            'http' => [
                'timeout' => 5,
                'header'  => 'User-Agent: ' . $this->userAgent . "\r\n",
            ],
        ]);

        $response = @file_get_contents($url, false, $ctx);
        if ($response === false) return false;

        $data = json_decode($response, true);
        return !empty($data['results']['bindings']);
    }

    private function loadCache(): array {
        if (!file_exists($this->cacheFile)) return [];
        $raw = @file_get_contents($this->cacheFile);
        if ($raw === false) return [];
        $data = json_decode($raw, true);
        if (!is_array($data)) return [];

        $now = time();
        $out = [];
        foreach ($data as $key => $entry) {
            if (isset($entry['t'], $entry['v']) && ($now - $entry['t']) <= $this->ttl) {
                $out[$key] = (bool) $entry['v'];
            }
        }
        return $out;
    }

    private function saveCache(array $cache): void {
        $dir = dirname($this->cacheFile);
        if (!is_dir($dir)) @mkdir($dir, 0755, true);

        $now = time();
        $out = [];
        foreach ($cache as $k => $v) {
            $out[$k] = ['v' => (bool) $v, 't' => $now];
        }
        @file_put_contents($this->cacheFile, json_encode($out), LOCK_EX);
    }
}