First commit

index.php

@@ -0,0 +1,291 @@
+<?php
+// index.php
+// Gère génération du cache, redirection classique et gestion des bangs (!code)
+
+// Constantes
+define('CACHE_DIR', __DIR__ . '/cache');
+define('INSTANCES_JSON', CACHE_DIR . '/instances.json');
+define('URLS_TXT',      CACHE_DIR . '/urls.txt');
+define('INSTANCES_URL', 'https://searx.space/data/instances.json');
+define('CACHE_MAX_AGE', 3600); // 1 heure
+
+// Headers de sécurité
+header('X-Frame-Options: SAMEORIGIN');
+header('X-Content-Type-Options: nosniff');
+header('Referrer-Policy: no-referrer-when-downgrade');
+header("Content-Security-Policy: default-src 'none'; frame-ancestors 'none'; sandbox");
+
+// Création dossier cache si nécessaire
+if (!is_dir(CACHE_DIR) && !mkdir(CACHE_DIR, 0755, true)) {
+    error_log('Erreur création dossier cache');
+    http_response_code(500);
+    exit('Erreur interne');
+}
+
+// Mise à jour du cache si trop vieux ou absent
+if (cache_expired()) {
+    if (!download_instances_json() || !extract_valid_urls()) {
+        error_log('Erreur lors de la génération du cache');
+        http_response_code(500);
+        exit('Erreur interne');
+    }
+}
+
+// Récupérer la requête
+$query = $_GET['q'] ?? null;
+
+if ($query !== null) {
+    if (!is_valid_query($query)) {
+        http_response_code(400);
+        exit('Paramètre q invalide');
+    }
+
+    // Gestion bang : si retourne true, redirection faite et script termine
+    if (!try_redirect_bang($query)) {
+        redirect_search($query);
+    }
+} else {
+    redirect_random_url();
+}
+
+// — Fonctions — //
+
+function cache_expired(): bool
+{
+    if (!file_exists(URLS_TXT)) {
+        return true;
+    }
+    $age = time() - filemtime(URLS_TXT);
+    return $age > CACHE_MAX_AGE;
+}
+
+function download_instances_json(): bool
+{
+    $ch = curl_init(INSTANCES_URL);
+    curl_setopt_array($ch, [
+        CURLOPT_RETURNTRANSFER    => true,
+        CURLOPT_FOLLOWLOCATION    => true,
+        CURLOPT_TIMEOUT           => 10,
+        CURLOPT_SSL_VERIFYPEER    => true,
+    ]);
+    $data = curl_exec($ch);
+    $code = curl_getinfo($ch, CURLINFO_HTTP_CODE);
+    if ($data === false) {
+        error_log('curl_exec failed: ' . curl_error($ch));
+    }
+    curl_close($ch);
+
+    if ($code === 200 && $data !== false) {
+        return safe_file_put_contents(INSTANCES_JSON, $data);
+    }
+    error_log("Erreur téléchargement instances.json, HTTP $code");
+    return false;
+}
+
+function extract_valid_urls(): bool
+{
+    if (!file_exists(INSTANCES_JSON)) {
+        error_log("instances.json introuvable");
+        return false;
+    }
+
+    $content = file_get_contents(INSTANCES_JSON);
+    if ($content === false) {
+        error_log("Impossible de lire instances.json");
+        return false;
+    }
+
+    $j = json_decode($content, true);
+    if (!isset($j['instances']) || !is_array($j['instances'])) {
+        error_log("JSON mal formé");
+        return false;
+    }
+
+    $valid = [];
+    foreach ($j['instances'] as $url => $data) {
+        if (
+            ($data['network_type'] ?? '') === 'normal'
+            && ($data['http']['status_code'] ?? 0) === 200
+            && ($data['timing']['search']['success_percentage'] ?? 0) === 100.0
+            && ($data['timing']['initial']['success_percentage'] ?? 0) === 100.0
+            && filter_var($url, FILTER_VALIDATE_URL)
+            && in_array(parse_url($url, PHP_URL_SCHEME), ['http', 'https'], true)
+        ) {
+            $valid[] = rtrim($url, '/');
+        }
+    }
+
+    if (empty($valid)) {
+        error_log("Aucune URL valide");
+        return false;
+    }
+
+    return safe_file_put_contents(URLS_TXT, implode("\n", $valid));
+}
+
+function safe_file_put_contents(string $filename, string $data): bool
+{
+    $fp = fopen($filename, 'c');
+    if (!$fp) {
+        error_log("Impossible d'ouvrir $filename en écriture");
+        return false;
+    }
+    if (!flock($fp, LOCK_EX)) {
+        fclose($fp);
+        error_log("Impossible de verrouiller $filename");
+        return false;
+    }
+    ftruncate($fp, 0);
+    $written = fwrite($fp, $data);
+    fflush($fp);
+    flock($fp, LOCK_UN);
+    fclose($fp);
+
+    return $written === strlen($data);
+}
+
+function get_all_urls(): array
+{
+    if (!file_exists(URLS_TXT)) {
+        return [];
+    }
+
+    $fp = fopen(URLS_TXT, 'r');
+    if (!$fp) {
+        error_log("Impossible d'ouvrir urls.txt");
+        return [];
+    }
+
+    if (!flock($fp, LOCK_SH)) {
+        fclose($fp);
+        error_log("Impossible de verrouiller urls.txt en lecture");
+        return [];
+    }
+
+    $urls = [];
+    while (($line = fgets($fp)) !== false) {
+        $line = trim($line);
+        if ($line !== '' && filter_var($line, FILTER_VALIDATE_URL)) {
+            $urls[] = $line;
+        }
+    }
+    flock($fp, LOCK_UN);
+    fclose($fp);
+
+    return $urls;
+}
+
+function redirect_random_url(): void
+{
+    $urls = get_all_urls();
+    if (empty($urls)) {
+        http_response_code(500);
+        exit('Instances indisponibles');
+    }
+    $target = $urls[array_rand($urls)];
+    header("Location: $target");
+    exit;
+}
+
+function redirect_search(string $query): void
+{
+    $urls = get_all_urls();
+    if (empty($urls)) {
+        http_response_code(500);
+        exit('Instances indisponibles pour recherche');
+    }
+    $base = $urls[array_rand($urls)];
+
+    $parsed = parse_url($base);
+    if ($parsed === false) {
+        error_log("Base URL invalide dans urls.txt");
+        http_response_code(500);
+        exit('Erreur interne');
+    }
+
+    $path = $parsed['path'] ?? '';
+    $path = rtrim($path, '/') . '/search';
+
+    $query_string = http_build_query(['q' => $query]);
+
+    $target = $parsed['scheme'] . '://' . $parsed['host'];
+    if (isset($parsed['port'])) {
+        $target .= ':' . $parsed['port'];
+    }
+    $target .= $path . '?' . $query_string;
+
+    header("Location: $target");
+    exit;
+}
+
+function is_valid_query(string $query): bool
+{
+    $query = trim($query);
+    if ($query === '') {
+        return false;
+    }
+    if (mb_strlen($query) > 200) {
+        return false;
+    }
+
+    // Vérifier que la chaîne est en UTF-8 valide
+    if (!mb_check_encoding($query, 'UTF-8')) {
+        return false;
+    }
+
+    return preg_match('/^[^\x00-\x1F\x7F]+$/u', $query) === 1;
+}
+
+// --- Gestion bangs --- //
+
+function load_bangs(): array
+{
+    $file = __DIR__ . '/bang.json';
+    if (!file_exists($file)) {
+        error_log("bang.json introuvable");
+        return [];
+    }
+    $content = file_get_contents($file);
+    if ($content === false) {
+        error_log("Impossible de lire bang.json");
+        return [];
+    }
+    $bangs = json_decode($content, true);
+    if (!is_array($bangs)) {
+        error_log("bang.json mal formé");
+        return [];
+    }
+    return $bangs;
+}
+
+function try_redirect_bang(string $query): bool
+{
+    if (substr($query, 0, 1) !== '!') {
+        return false;
+    }
+
+    if (!preg_match('/^!(\w+)\s?(.*)$/u', $query, $matches)) {
+        return false;
+    }
+    $bang_code = $matches[1];
+    $search_term = trim($matches[2]);
+
+    if ($search_term === '') {
+        return false;
+    }
+
+    $bangs = load_bangs();
+    if (empty($bangs)) {
+        return false;
+    }
+
+    foreach ($bangs as $bang) {
+        if (($bang['bang'] ?? '') === $bang_code) {
+            $base_url = rtrim($bang['url'], '/');
+            $search_url = $base_url . '?q=' . rawurlencode($search_term);
+            header("Location: $search_url");
+            exit;
+        }
+    }
+    return false;
+}