php-func-lib/link-meta.php

<?php
declare(strict_types=1);
require_once __DIR__ . '/http-limits.php';

function httpContext(?int $timeout = null) {
    $limits = httpLimits();
    $resolvedTimeout = $timeout === null ? $limits['timeout'] : max(1, $timeout);
    return stream_context_create([
        'http' => [
            'timeout' => $resolvedTimeout,
            'follow_location' => 1,
            'max_redirects' => $limits['max_redirects'],
            'user_agent' => $limits['user_agent'],
            'ignore_errors' => true
        ],
        'ssl' => [
            'verify_peer' => true,
            'verify_peer_name' => true
        ]
    ]);
}

function normalizeUrl(string $url): ?string {
    $url = trim($url);
    if (!filter_var($url, FILTER_VALIDATE_URL)) {
        return null;
    }

    $parts = parse_url($url);
    if (!$parts || !isset($parts['scheme'])) {
        return null;
    }

    $scheme = strtolower($parts['scheme']);
    if ($scheme !== 'http' && $scheme !== 'https') {
        return null;
    }

    return $url;
}

function resolveUrl(string $url, string $baseUrl): ?string {
    $url = trim($url);
    if (filter_var($url, FILTER_VALIDATE_URL)) {
        return $url;
    }

    $baseParts = parse_url($baseUrl);
    if (!$baseParts || !isset($baseParts['scheme']) || !isset($baseParts['host'])) {
        return null;
    }

    if (strpos($url, '//') === 0) {
        return $baseParts['scheme'] . ':' . $url;
    }

    $path = '/';
    if (!empty($baseParts['path'])) {
        $path = preg_replace('#/[^/]*$#', '/', $baseParts['path']);
        if ($path === null || $path === '') {
            $path = '/';
        }
    }

    if (strlen($url) && $url[0] === '/') {
        return $baseParts['scheme'] . '://' . $baseParts['host'] . $url;
    }

    return $baseParts['scheme'] . '://' . $baseParts['host'] . $path . $url;
}

function safeFetch(string $url, ?int $timeout = null): ?string {
    $limits = httpLimits();
    $ctx = httpContext($timeout);
    $content = @file_get_contents($url, false, $ctx);
    if ($content === false || strlen($content) > $limits['max_bytes']) {
        return null;
    }
    return $content;
}

function downloadImageFromUrl(string $url, string $baseUrl, string $destinationFolder = 'upl/'): ?string {
    $resolved = resolveUrl($url, $baseUrl);
    if ($resolved === null) {
        return null;
    }

    if (!preg_match('/\.(jpg|jpeg|png|gif|bmp|webp)(?:\?|#|$)/i', $resolved)) {
        return null;
    }

    $imageContent = safeFetch($resolved);
    if ($imageContent === null || strlen($imageContent) === 0) {
        return null;
    }

    $path = parse_url($resolved, PHP_URL_PATH) ?? '';
    $ext = strtolower(pathinfo($path, PATHINFO_EXTENSION));
    if ($ext === '' || !preg_match('/^(jpg|jpeg|png|gif|bmp|webp)$/', $ext)) {
        $ext = 'png';
    }

    if (!is_dir($destinationFolder)) {
        @mkdir($destinationFolder, 0775, true);
    }

    $md5Hash = md5($imageContent);
    $filePath = rtrim($destinationFolder, '/\\') . '/' . $md5Hash . '.' . $ext;
    $written = @file_put_contents($filePath, $imageContent);
    if ($written === false) {
        return null;
    }

    return $filePath;
}

function parseMetaContent(string $html, string $attr, string $name): ?string {
    $pattern = '/<meta[^>]*' . $attr . '=["\']' . preg_quote($name, '/') . '["\'][^>]*content=["\']([^"\']+)["\'][^>]*>/i';
    if (preg_match($pattern, $html, $matches) && isset($matches[1])) {
        return trim(html_entity_decode($matches[1], ENT_QUOTES | ENT_HTML5, 'UTF-8'));
    }
    return null;
}

function getPageInfo(string $url): array {
    $ret = [
        'ok' => false,
        'title' => '',
        'description' => '',
        'logo' => null,
        'error' => null
    ];

    $normalized = normalizeUrl($url);
    if ($normalized === null) {
        $ret['error'] = 'ungueltige_url';
        return $ret;
    }

    $html = safeFetch($normalized);
    if ($html === null) {
        $ret['error'] = 'seite_nicht_erreichbar';
        return $ret;
    }

    $title = parseMetaContent($html, 'property', 'og:title') ?? parseMetaContent($html, 'name', 'title');
    $description = parseMetaContent($html, 'property', 'og:description') ?? parseMetaContent($html, 'name', 'description');
    $image = parseMetaContent($html, 'property', 'og:image') ?? parseMetaContent($html, 'name', 'image');

    if ($title === null && preg_match('/<title>\s*(.*?)\s*<\/title>/is', $html, $matchTitle)) {
        $title = trim(html_entity_decode($matchTitle[1], ENT_QUOTES | ENT_HTML5, 'UTF-8'));
    }

    $logo = null;
    if ($image !== null && $image !== '') {
        $img = downloadImageFromUrl($image, $normalized);
        if ($img !== null) {
            $logo = '/' . ltrim($img, '/');
        }
    }

    $ret['ok'] = true;
    $ret['title'] = $title ?? '';
    $ret['description'] = $description ?? '';
    $ret['logo'] = $logo;
    return $ret;
}

function sanitizeTags(array $input): array {
    $ret = [];
    foreach ($input as $tag) {
        if (!is_string($tag)) {
            continue;
        }
        $clean = onlyAlpha(trim($tag), '_\-');
        if ($clean === '') {
            continue;
        }
        $clean = ucfirst(substr($clean, 0, 35));
        $ret[$clean] = true;
        if (count($ret) >= 20) {
            break;
        }
    }
    return array_keys($ret);
}