fourth-wall-embed-wp/libs/shortcode.php

<?php

/**
 * How long fetched content is considered fresh, in seconds.
 *
 * @return int Seconds
 */
function fwembed_cache_ttl() {
    $options = get_option('fourthwall_settings_name');
    $minutes = isset($options['cache_ttl']) ? intval($options['cache_ttl']) : 60;

    if ($minutes < 1) {
        $minutes = 60;
    }

    return $minutes * MINUTE_IN_SECONDS;
}

/**
 * How long a cached entry is kept after it goes stale.
 *
 * Stale content is still served while a refresh runs in the background, so it
 * has to outlive the freshness window by a wide margin. Once this elapses the
 * entry is gone and the next visitor pays for a synchronous fetch.
 *
 * @return int Seconds
 */
function fwembed_cache_retention() {
    return max(fwembed_cache_ttl() * 24, DAY_IN_SECONDS);
}

/**
 * Transient key for a URL.
 *
 * @param string $url The URL
 * @return string Cache key
 */
function fwembed_cache_key($url) {
    return 'fwembed_' . md5($url);
}

/**
 * Read a cached response.
 *
 * @param string $url The URL
 * @return array|null Cached entry, or null if nothing usable is stored
 */
function fwembed_cache_read($url) {
    $cached = get_transient(fwembed_cache_key($url));

    if (!is_array($cached) || !array_key_exists('content', $cached)) {
        return null;
    }

    // Entries written by older versions have no freshness stamp; treat them as
    // stale so they get served once and then refreshed.
    $cached['expires_at'] = isset($cached['expires_at']) ? (int) $cached['expires_at'] : 0;

    return $cached;
}

/**
 * Store a response and stamp its freshness window.
 *
 * @param string $url The URL
 * @param array $result The response to cache
 * @return array The stamped result
 */
function fwembed_cache_write($url, $result) {
    $result['cached_at'] = time();
    $result['expires_at'] = time() + fwembed_cache_ttl();

    set_transient(fwembed_cache_key($url), $result, fwembed_cache_retention());

    return $result;
}

/**
 * Whether a cached entry is still fresh.
 *
 * @param array $entry A cache entry
 * @return bool
 */
function fwembed_cache_is_fresh($entry) {
    return isset($entry['expires_at']) && $entry['expires_at'] > time();
}

/**
 * Queue a background refresh for a stale URL.
 *
 * WP-Cron fires on a later request, so the visitor who found the stale entry is
 * never the one who waits for the network.
 *
 * @param string $url The URL to refresh
 * @return void
 */
function fwembed_schedule_refresh($url) {
    if (!function_exists('wp_next_scheduled') || !function_exists('wp_schedule_single_event')) {
        return;
    }

    // Collapses a stampede: many visitors hitting the same stale entry queue
    // one job between them.
    if (wp_next_scheduled('fwembed_refresh_url', array($url))) {
        return;
    }

    wp_schedule_single_event(time() + 1, 'fwembed_refresh_url', array($url));
}

/**
 * WP-Cron handler: re-fetch a URL and replace its cache entry.
 *
 * @param string $url The URL to refresh
 * @return void
 */
function fwembed_do_refresh($url) {
    $lock_key = fwembed_cache_key($url) . '_lock';

    // Guards against two overlapping cron runs fetching the same URL.
    if (get_transient($lock_key) !== false) {
        return;
    }
    set_transient($lock_key, 1, 2 * MINUTE_IN_SECONDS);

    $result = fwembed_fetch($url);
    if (!$result['error'] && $result['status_code'] == 200) {
        fwembed_cache_write($url, $result);
    }

    delete_transient($lock_key);
}
add_action('fwembed_refresh_url', 'fwembed_do_refresh');

/**
 * Build a configured cURL handle.
 *
 * @param string $url The URL to fetch
 * @return resource|CurlHandle
 */
function fwembed_curl_handle($url) {
    // More complete browser-like headers
    $headers = array(
        'Accept: text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8',
        'Accept-Language: en-US,en;q=0.5',
        'Connection: keep-alive',
        'Upgrade-Insecure-Requests: 1',
        'Cache-Control: max-age=0'
    );

    $ch = curl_init();
    curl_setopt($ch, CURLOPT_URL, $url);
    curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
    curl_setopt($ch, CURLOPT_FOLLOWLOCATION, true);
    curl_setopt($ch, CURLOPT_ENCODING, "");
    curl_setopt($ch, CURLOPT_HTTPHEADER, $headers);
    curl_setopt($ch, CURLOPT_USERAGENT, 'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:109.0) Gecko/20100101 Firefox/115.0');
    curl_setopt($ch, CURLOPT_TIMEOUT, 30);
    curl_setopt($ch, CURLOPT_CONNECTTIMEOUT, 10);
    curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, true);
    curl_setopt($ch, CURLOPT_SSL_VERIFYHOST, 2);

    return $ch;
}

/**
 * Turn a finished cURL handle into a response array.
 *
 * @param resource|CurlHandle $ch The handle
 * @param string|bool $content The transferred body
 * @return array Response with 'content', 'error' and 'status_code'
 */
function fwembed_curl_result($ch, $content) {
    $error = null;
    $status_code = 0;

    if (curl_errno($ch)) {
        $error = curl_error($ch);
    } else {
        $status_code = (int) curl_getinfo($ch, CURLINFO_HTTP_CODE);
        if ($status_code == 403) {
            $error = "Access forbidden (403). The website may be blocking automated requests.";
        } elseif ($status_code >= 400) {
            $error = "Request failed with HTTP status " . $status_code . ".";
        }
    }

    return array(
        'content'     => $content,
        'error'       => $error,
        'status_code' => $status_code,
    );
}

/**
 * Fetch a single URL, bypassing the cache.
 *
 * @param string $url The URL to fetch
 * @return array Response array
 */
function fwembed_fetch($url) {
    $ch = fwembed_curl_handle($url);
    $content = curl_exec($ch);
    $result = fwembed_curl_result($ch, $content);
    curl_close($ch);

    return $result;
}

/**
 * Fetch several URLs concurrently, bypassing the cache.
 *
 * @param array $urls URLs to fetch
 * @param int $concurrency Maximum simultaneous connections
 * @return array Map of URL => response array
 */
function fwembed_fetch_many($urls, $concurrency = 8) {
    $urls = array_values(array_unique($urls));
    $results = array();

    if (empty($urls)) {
        return $results;
    }
    if (count($urls) === 1 || !function_exists('curl_multi_init')) {
        foreach ($urls as $url) {
            $results[$url] = fwembed_fetch($url);
        }
        return $results;
    }

    foreach (array_chunk($urls, max(1, (int) $concurrency)) as $batch) {
        $multi = curl_multi_init();
        $handles = array();

        foreach ($batch as $url) {
            $ch = fwembed_curl_handle($url);
            curl_multi_add_handle($multi, $ch);
            $handles[$url] = $ch;
        }

        $running = null;
        do {
            $status = curl_multi_exec($multi, $running);
            if ($running) {
                curl_multi_select($multi, 1.0);
            }
        } while ($running && $status === CURLM_OK);

        foreach ($handles as $url => $ch) {
            $results[$url] = fwembed_curl_result($ch, curl_multi_getcontent($ch));
            curl_multi_remove_handle($multi, $ch);
            curl_close($ch);
        }

        curl_multi_close($multi);
    }

    return $results;
}

/**
 * Common function to make HTTP requests to Fourthwall
 *
 * Serves cached content immediately whenever any is stored. A stale entry is
 * returned as-is and refreshed by WP-Cron, so expiry never lands the cost of a
 * network round trip on a visitor.
 *
 * @param string $url The URL to fetch
 * @return array Array containing 'content' and 'error' keys
 */
function fwembed_make_request($url) {
    $cached = fwembed_cache_read($url);

    if ($cached !== null) {
        if (!fwembed_cache_is_fresh($cached)) {
            fwembed_schedule_refresh($url);
        }
        return $cached;
    }

    $result = fwembed_fetch($url);

    if (!$result['error'] && $result['status_code'] == 200) {
        $result = fwembed_cache_write($url, $result);
    }

    return $result;
}

/**
 * Make several requests at once, using the cache where possible.
 *
 * Anything not already cached is fetched in parallel, turning N sequential
 * round trips into one batch.
 *
 * @param array $urls URLs to fetch
 * @return array Map of URL => response array
 */
function fwembed_make_requests($urls) {
    $results = array();
    $pending = array();

    foreach (array_unique($urls) as $url) {
        $cached = fwembed_cache_read($url);

        if ($cached !== null) {
            if (!fwembed_cache_is_fresh($cached)) {
                fwembed_schedule_refresh($url);
            }
            $results[$url] = $cached;
            continue;
        }

        $pending[] = $url;
    }

    foreach (fwembed_fetch_many($pending) as $url => $result) {
        if (!$result['error'] && $result['status_code'] == 200) {
            $result = fwembed_cache_write($url, $result);
        }
        $results[$url] = $result;
    }

    return $results;
}

/**
 * Build an XPath predicate matching a single CSS class name.
 *
 * Fourthwall reorders and appends classes between releases, so never match
 * @class exactly - pad the attribute and look for the padded class name.
 *
 * @param string $class The class name to match
 * @return string XPath predicate fragment
 */
function fwembed_class_predicate($class) {
    return 'contains(concat(" ", normalize-space(@class), " "), " ' . $class . ' ")';
}

/**
 * Run XPath queries in priority order and return the first one that matches.
 *
 * Lets each selector degrade to a looser fallback when Fourthwall changes
 * their markup, instead of the whole product silently disappearing.
 *
 * @param DOMXPath $xpath The XPath engine
 * @param array $queries XPath expressions, most specific first
 * @param DOMNode|null $context Optional context node for relative queries
 * @return DOMNodeList|null First non-empty result, or null if none matched
 */
function fwembed_query_first(DOMXPath $xpath, array $queries, $context = null) {
    foreach ($queries as $query) {
        $nodes = $context === null ? $xpath->query($query) : $xpath->query($query, $context);
        if ($nodes !== false && $nodes->length > 0) {
            return $nodes;
        }
    }
    return null;
}

/**
 * Resolve a possibly-relative href against the page it was found on.
 *
 * Product hrefs are root-relative ("/products/foo"), so naive concatenation
 * breaks whenever the store URL has a trailing slash or a path such as
 * /collections/all.
 *
 * @param string $href The href to resolve
 * @param string $page_url The URL of the page the href was found on
 * @return string Absolute URL
 */
function fwembed_resolve_url($href, $page_url) {
    $href = trim($href);
    if ($href === '') {
        return '';
    }
    if (preg_match('#^https?://#i', $href)) {
        return $href;
    }

    $parts = parse_url($page_url);
    if (empty($parts['scheme']) || empty($parts['host'])) {
        return $href;
    }

    $origin = $parts['scheme'] . '://' . $parts['host'];
    if (!empty($parts['port'])) {
        $origin .= ':' . $parts['port'];
    }

    if (strpos($href, '//') === 0) {
        return $parts['scheme'] . ':' . $href;
    }
    if (strpos($href, '/') === 0) {
        return $origin . $href;
    }

    $path = isset($parts['path']) ? $parts['path'] : '/';
    $dir = (substr($path, -1) === '/') ? $path : str_replace('\\', '/', dirname($path)) . '/';

    return $origin . $dir . $href;
}

/**
 * Parse HTML content and extract product tiles
 *
 * @param string $html_content The HTML content to parse
 * @param string $base_url The base URL for constructing links
 * @return string Parsed HTML content
 */
function fwembed_parse_product_tiles($html_content, $base_url) {
    $html = '';
    libxml_use_internal_errors(true);
    $dom = new DOMDocument();
    @$dom->loadHTML(loadHTML5($html_content), LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
    $dom->documentURI = $base_url;
    $xpath = new DOMXPath($dom);

    $products = $xpath->query('//*[@data-testid="product"]');
    if ($products === false || $products->length === 0) {
        $products = $xpath->query('//*[' . fwembed_class_predicate('product-tile') . ']');
    }

    foreach ($products as $product) {
        $tileLink = fwembed_query_first($xpath, array(
            './/a[' . fwembed_class_predicate('tile') . ']',
            './/a[@href]',
        ), $product);

        $tileItem = fwembed_query_first($xpath, array(
            './/img[' . fwembed_class_predicate('tile__item--1') . ']',
            './/*[' . fwembed_class_predicate('image__object') . ']//img',
            './/img',
        ), $product);

        $tileDesc = fwembed_query_first($xpath, array(
            './/*[' . fwembed_class_predicate('tile__description') . ']',
        ), $product);

        $productHTML = '';
        $linkHref = '';
        if ($tileLink !== null) {
            $linkHref = fwembed_resolve_url($tileLink->item(0)->getAttribute('href'), $base_url);
        }
        if ($tileItem !== null) {
            $productHTML .= $dom->saveHTML($tileItem->item(0));
        }
        if ($tileDesc !== null) {
            $productHTML .= $dom->saveHTML($tileDesc->item(0));
        }

        if ($productHTML === '') {
            continue;
        }

        $html .= '<div class="product-tile"><a class="product-link" target="_blank" href="' . esc_url($linkHref) . '">' . $productHTML . '</a></div>';
    }
    libxml_clear_errors();
    return $html;
}

function fwembed_parse_html($url = null) {
    if ($url === null) {
        throw new ValueError("Missing URL");
    }

    $result = fwembed_make_request($url);
    
    if ($result['error']) {
        return "Error fetching URL: " . $result['error'];
    }

    return fwembed_parse_product_tiles($result['content'], $url);
}

function loadHTML5($html) {
    return '<!DOCTYPE html><html><body>' . $html . '</body></html>';
}

function fwembed_shortcode( $atts ) {
    $options = get_option( 'fourthwall_settings_name' );
    $value = isset( $options['fourth_url'] ) ? $options['fourth_url'] : 'https://fourthwall.com';
    $store_html = fwembed_parse_html($value);
    $store_render = '<div class="fw-store-parent">' . PHP_EOL . $store_html . PHP_EOL . '</div>';
    return $store_render;
}

/**
 * Decode every JSON-LD block on a page.
 *
 * @param DOMXPath $xpath The XPath engine
 * @return array List of decoded JSON-LD structures
 */
function fwembed_jsonld_blocks(DOMXPath $xpath) {
    $blocks = array();
    $scripts = $xpath->query('//script[contains(@type, "ld+json")]');
    if ($scripts === false) {
        return $blocks;
    }

    foreach ($scripts as $script) {
        $decoded = json_decode(trim($script->textContent), true);
        if (json_last_error() === JSON_ERROR_NONE && is_array($decoded)) {
            $blocks[] = $decoded;
        }
    }
    return $blocks;
}

/**
 * Walk a decoded JSON-LD structure looking for a schema.org Product node.
 *
 * Handles the shapes publishers actually emit: a bare object, a list, or a
 * node wrapped in @graph / itemListElement / mainEntity.
 *
 * @param mixed $data Decoded JSON-LD fragment
 * @param int $depth Current recursion depth
 * @return array|null The Product node, or null if this branch has none
 */
function fwembed_find_jsonld_product($data, $depth = 0) {
    if (!is_array($data) || $depth > 6) {
        return null;
    }

    if (isset($data['@type'])) {
        foreach ((array) $data['@type'] as $type) {
            if (is_string($type) && strcasecmp($type, 'Product') === 0) {
                return $data;
            }
        }
    }

    foreach (array('@graph', 'itemListElement', 'mainEntity', 'item') as $key) {
        if (isset($data[$key])) {
            $found = fwembed_find_jsonld_product($data[$key], $depth + 1);
            if ($found !== null) {
                return $found;
            }
        }
    }

    // A plain list of nodes.
    foreach ($data as $key => $item) {
        if (is_int($key) && is_array($item)) {
            $found = fwembed_find_jsonld_product($item, $depth + 1);
            if ($found !== null) {
                return $found;
            }
        }
    }

    return null;
}

/**
 * Pick the cheapest offer from a JSON-LD Product node.
 *
 * Variants each get their own Offer, so the lowest price is what a storefront
 * shows as the headline figure. AggregateOffer states it directly as lowPrice.
 *
 * @param array $product A schema.org Product node
 * @return array Keys 'price', 'currency', 'availability'; empty if no offer
 */
function fwembed_jsonld_offer($product) {
    if (empty($product['offers'])) {
        return array();
    }

    $offers = $product['offers'];
    if (isset($offers['@type']) || isset($offers['price']) || isset($offers['lowPrice'])) {
        $offers = array($offers);
    }

    $best = array();
    foreach ((array) $offers as $offer) {
        if (!is_array($offer)) {
            continue;
        }
        foreach (array('lowPrice', 'price') as $field) {
            if (!isset($offer[$field]) || !is_numeric($offer[$field])) {
                continue;
            }
            $price = (float) $offer[$field];
            if (empty($best) || $price < $best['price']) {
                $availability = isset($offer['availability']) ? (string) $offer['availability'] : '';
                $best = array(
                    'price'        => $price,
                    'currency'     => isset($offer['priceCurrency']) ? (string) $offer['priceCurrency'] : '',
                    // "https://schema.org/InStock" -> "InStock"
                    'availability' => $availability === '' ? '' : substr($availability, strrpos($availability, '/') + 1),
                );
            }
            break;
        }
    }
    return $best;
}

/**
 * Format a numeric price for display.
 *
 * Only used when the storefront's own formatted price could not be scraped.
 *
 * @param float $amount The price
 * @param string $currency ISO 4217 currency code
 * @return string Display-ready price
 */
function fwembed_format_price($amount, $currency) {
    $currency = strtoupper(trim($currency));
    $symbols = array(
        'USD' => '$', 'CAD' => '$', 'AUD' => '$', 'NZD' => '$',
        'EUR' => "\xE2\x82\xAC", 'GBP' => "\xC2\xA3", 'JPY' => "\xC2\xA5",
    );

    $formatted = number_format($amount, $currency === 'JPY' ? 0 : 2);

    if (isset($symbols[$currency])) {
        return $symbols[$currency] . $formatted;
    }
    return $currency === '' ? $formatted : $currency . ' ' . $formatted;
}

/**
 * Extract product fields from a page's JSON-LD.
 *
 * Preferred over scraping: it is a published contract rather than styling, so
 * it survives the theme changes that repeatedly break CSS-class selectors.
 *
 * @param DOMXPath $xpath The XPath engine
 * @return array Normalised product fields (any may be absent)
 */
function fwembed_product_from_jsonld(DOMXPath $xpath) {
    $product = null;
    foreach (fwembed_jsonld_blocks($xpath) as $block) {
        $product = fwembed_find_jsonld_product($block);
        if ($product !== null) {
            break;
        }
    }
    if ($product === null) {
        return array();
    }

    $fields = array();

    if (!empty($product['name']) && is_string($product['name'])) {
        $fields['title'] = $product['name'];
    }

    if (!empty($product['image'])) {
        $image = $product['image'];
        // May be a string, a list, or an ImageObject.
        if (is_array($image)) {
            $image = isset($image[0]) ? $image[0] : (isset($image['url']) ? $image['url'] : null);
        }
        if (is_array($image) && isset($image['url'])) {
            $image = $image['url'];
        }
        if (is_string($image) && $image !== '') {
            $fields['image'] = $image;
        }
    }

    if (!empty($product['description']) && is_string($product['description'])) {
        $fields['description_text'] = $product['description'];
    }

    if (!empty($product['sku']) && is_string($product['sku'])) {
        $fields['sku'] = $product['sku'];
    }

    $offer = fwembed_jsonld_offer($product);
    if (!empty($offer)) {
        $fields['price'] = fwembed_format_price($offer['price'], $offer['currency']);
        $fields['currency'] = $offer['currency'];
        $fields['availability'] = $offer['availability'];
    }

    return $fields;
}

/**
 * Extract product fields by scraping the rendered markup.
 *
 * Fallback for stores whose JSON-LD is missing or incomplete, and the source of
 * the store's own price formatting and rich HTML description.
 *
 * @param DOMDocument $dom The parsed page
 * @param DOMXPath $xpath The XPath engine
 * @param string $url The product URL
 * @return array Normalised product fields (any may be absent)
 */
function fwembed_product_from_html(DOMDocument $dom, DOMXPath $xpath, $url) {
    $fields = array();

    // Fourthwall moved the title from <h2> to <h1>, so match on the class /
    // data-testid rather than the tag, and fall back to the OpenGraph tags
    // (which have stayed stable across their redesigns).
    $productTitle = fwembed_query_first($xpath, array(
        '//*[@data-testid="product.name"]',
        '//*[' . fwembed_class_predicate('product-info__title') . ']',
        '//meta[@property="og:title"]',
    ));
    $productPrice = fwembed_query_first($xpath, array(
        '//*[@data-testid="product.price"]',
        '//*[' . fwembed_class_predicate('product-info__price--original') . ']',
        '//*[' . fwembed_class_predicate('product-info__price') . ']',
    ));
    $productImage = fwembed_query_first($xpath, array(
        '//*[@data-testid="product.image"]//img[' . fwembed_class_predicate('gallery__image-object') . ']',
        '//*[@data-gallery="gallery-slide"]//img[' . fwembed_class_predicate('gallery__image-object') . ']',
        '//img[' . fwembed_class_predicate('gallery__image-object') . ']',
        '//meta[@property="og:image"]',
    ));
    $productDesc = fwembed_query_first($xpath, array(
        '//*[' . fwembed_class_predicate('product-info__description') . ']//*[' . fwembed_class_predicate('html-formatter') . ']',
        '//*[' . fwembed_class_predicate('product-info__description') . ']',
    ));

    if ($productTitle !== null) {
        $node = $productTitle->item(0);
        $fields['title'] = $node->nodeName === 'meta' ? $node->getAttribute('content') : $node->textContent;
    }

    if ($productPrice !== null) {
        $fields['price'] = $productPrice->item(0)->textContent;
    }

    if ($productImage !== null) {
        $node = $productImage->item(0);
        if ($node->nodeName === 'meta') {
            $src = $node->getAttribute('content');
        } else {
            // Lazy-loaded galleries keep the real URL in data-src.
            $src = $node->getAttribute('src');
            if ($src === '' || strpos($src, 'data:') === 0) {
                $src = $node->getAttribute('data-src');
            }
            $alt = $node->getAttribute('alt');
            if ($alt !== '') {
                $fields['image_alt'] = $alt;
            }
        }
        if (trim($src) !== '') {
            $fields['image'] = fwembed_resolve_url($src, $url);
        }
    }

    if ($productDesc !== null) {
        $fields['description_html'] = $dom->saveHTML($productDesc->item(0));
    }

    return $fields;
}

/**
 * Parse HTML content and extract single product information
 *
 * Reads JSON-LD first and falls back to scraping field by field, so a markup
 * change can only degrade whatever it actually touched.
 *
 * @param string $html_content The HTML content to parse
 * @param string $url The product URL
 * @param bool $show_description Whether to show product description
 * @return string Parsed HTML content
 */
function fwembed_parse_single_product($html_content, $url, $show_description = false) {
    $html = '';
    libxml_use_internal_errors(true);
    $dom = new DOMDocument();
    @$dom->loadHTML(loadHTML5($html_content), LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
    $dom->documentURI = $url;
    $xpath = new DOMXPath($dom);

    $structured = fwembed_product_from_jsonld($xpath);
    $scraped = fwembed_product_from_html($dom, $xpath, $url);

    // JSON-LD wins where it is the more stable source. The scraped price wins
    // because it carries the store's own formatting ("from $20", locale rules),
    // and the scraped description wins because JSON-LD flattens it to plain text.
    $title = fwembed_first_value(array(
        isset($structured['title']) ? $structured['title'] : null,
        isset($scraped['title']) ? $scraped['title'] : null,
    ));
    $image = fwembed_first_value(array(
        isset($structured['image']) ? $structured['image'] : null,
        isset($scraped['image']) ? $scraped['image'] : null,
    ));
    $price = fwembed_first_value(array(
        isset($scraped['price']) ? $scraped['price'] : null,
        isset($structured['price']) ? $structured['price'] : null,
    ));

    libxml_clear_errors();

    if ($title === '' || $image === '') {
        return '';
    }

    $image = fwembed_resolve_url($image, $url);
    $imageAlt = isset($scraped['image_alt']) ? $scraped['image_alt'] : $title;

    $availability = isset($structured['availability']) ? $structured['availability'] : '';
    $classes = 'product-tile';
    if (strcasecmp($availability, 'OutOfStock') === 0 || strcasecmp($availability, 'SoldOut') === 0) {
        $classes .= ' product-tile--sold-out';
    }

    $attributes = '';
    if ($availability !== '') {
        $attributes .= ' data-availability="' . esc_attr($availability) . '"';
    }
    if (!empty($structured['sku'])) {
        $attributes .= ' data-sku="' . esc_attr($structured['sku']) . '"';
    }

    $html = '<div class="' . esc_attr($classes) . '"' . $attributes . '>';
    $html .= '<a class="product-link" target="_blank" rel="noopener noreferrer" href="' . esc_url($url) . '">';
    $html .= '<img class="tile__item tile__item--1" loading="lazy" src="' . esc_url($image) . '" alt="' . esc_attr($imageAlt) . '">';

    $html .= '<div class="tile__description">';
    $html .= '<h3 class="tile__heading">' . esc_html(trim($title)) . '</h3>';
    $html .= '<div class="tile__prices">';
    $html .= '<span class="tile__price tile__price--original">' . esc_html(trim($price)) . '</span>';
    $html .= '</div>';
    $html .= '</div>';
    $html .= '</a>';

    // Add description if show_description is true
    if ($show_description) {
        if (isset($scraped['description_html'])) {
            $html .= '<div class="product-description">' . $scraped['description_html'] . '</div>';
        } elseif (isset($structured['description_text'])) {
            $html .= '<div class="product-description">' . wpautop(esc_html($structured['description_text'])) . '</div>';
        }
    }

    $html .= '</div>';

    return $html;
}

/**
 * Return the first non-empty string from a list of candidates.
 *
 * @param array $candidates Values in priority order
 * @return string First usable value, or an empty string
 */
function fwembed_first_value($candidates) {
    foreach ($candidates as $candidate) {
        if (is_string($candidate) && trim($candidate) !== '') {
            return $candidate;
        }
    }
    return '';
}

function fwembed_parse_html_single($url = null, $show_description = false) {
    if ($url === null) {
        throw new ValueError("Missing URL");
    }

    $result = fwembed_make_request($url);
    
    if ($result['error']) {
        return "Error fetching URL: " . $result['error'];
    }

    return fwembed_parse_single_product($result['content'], $url, $show_description);
}

function fwembed_single_shortcode($atts) {
    $atts = shortcode_atts(
        array(
            'url' => '',
            'show_description' => 'false',
        ),
        $atts
    );
    
    if (empty($atts['url'])) {
        return '<p>Error: URL is required for [fourthwall_single] shortcode</p>';
    }
    
    // Convert string 'true'/'false' to boolean
    $show_description = filter_var($atts['show_description'], FILTER_VALIDATE_BOOLEAN);
    
    $product_html = fwembed_parse_html_single($atts['url'], $show_description);
    return '<div class="fw-single-product">' . PHP_EOL . $product_html . PHP_EOL . '</div>';
}
add_shortcode('fourthwall_single', 'fwembed_single_shortcode');
add_shortcode( 'fourthwall', 'fwembed_shortcode' );
add_shortcode( 'fourthwall_random', 'fwembed_random_shortcode' );

/**
 * Extract all product URLs from a store page
 * 
 * @param string $html_content The HTML content to parse
 * @param string $base_url The base URL for constructing links
 * @return array Array of product URLs
 */
function fwembed_extract_product_urls($html_content, $base_url) {
    $product_urls = array();
    libxml_use_internal_errors(true);
    $dom = new DOMDocument();
    @$dom->loadHTML(loadHTML5($html_content), LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
    $dom->documentURI = $base_url;
    $xpath = new DOMXPath($dom);

    $products = $xpath->query('//*[@data-testid="product"]');
    if ($products === false || $products->length === 0) {
        $products = $xpath->query('//*[' . fwembed_class_predicate('product-tile') . ']');
    }

    foreach ($products as $product) {
        $tileLink = fwembed_query_first($xpath, array(
            './/a[' . fwembed_class_predicate('tile') . ']',
            './/a[@href]',
        ), $product);

        if ($tileLink !== null) {
            $full_url = fwembed_resolve_url($tileLink->item(0)->getAttribute('href'), $base_url);
            if ($full_url !== '') {
                $product_urls[] = $full_url;
            }
        }
    }
    libxml_clear_errors();

    // The same product can appear in several sections of one page.
    return array_values(array_unique($product_urls));
}

/**
 * Read product URLs from a sitemap, following one level of sitemap index.
 *
 * @param string $sitemap_url The sitemap to read
 * @param string $store_url The store URL, used to reject off-site entries
 * @param int $depth Current recursion depth
 * @return array Product URLs
 */
function fwembed_read_sitemap($sitemap_url, $store_url, $depth = 0) {
    if ($depth > 1) {
        return array();
    }

    $result = fwembed_make_request($sitemap_url);
    if ($result['error'] || empty($result['content'])) {
        return array();
    }

    libxml_use_internal_errors(true);
    $xml = new DOMDocument();
    $loaded = @$xml->loadXML($result['content']);
    if (!$loaded || $xml->documentElement === null) {
        libxml_clear_errors();
        return array();
    }

    $store_host = parse_url($store_url, PHP_URL_HOST);
    $is_index = strcasecmp($xml->documentElement->localName, 'sitemapindex') === 0;
    $urls = array();
    $nested = 0;

    foreach ($xml->getElementsByTagName('loc') as $loc) {
        $value = trim($loc->textContent);
        if ($value === '') {
            continue;
        }

        // Never follow a <loc> pointing off the configured store.
        if (parse_url($value, PHP_URL_HOST) !== $store_host) {
            continue;
        }

        if ($is_index) {
            if ($nested >= 10) {
                break;
            }
            $nested++;
            $urls = array_merge($urls, fwembed_read_sitemap($value, $store_url, $depth + 1));
        } elseif (strpos($value, '/products/') !== false) {
            $urls[] = $value;
        }
    }

    libxml_clear_errors();
    return array_values(array_unique($urls));
}

/**
 * List every product in a store via its sitemap.
 *
 * The store page only renders whatever collection it is configured to feature,
 * so scraping it under-reports the catalogue. The sitemap is the full list.
 *
 * @param string $store_url The store URL
 * @return array Product URLs
 */
function fwembed_extract_product_urls_from_sitemap($store_url) {
    $sitemap_url = fwembed_resolve_url('/sitemap.xml', $store_url);
    if ($sitemap_url === '') {
        return array();
    }
    return fwembed_read_sitemap($sitemap_url, $store_url);
}

/**
 * Collect candidate product URLs for a store.
 *
 * @param string $store_url The store URL
 * @param string $source One of 'auto', 'sitemap', 'page'
 * @return array|string Product URLs, or an error string
 */
function fwembed_collect_product_urls($store_url, $source = 'auto') {
    if ($source !== 'page') {
        $urls = fwembed_extract_product_urls_from_sitemap($store_url);
        if (!empty($urls) || $source === 'sitemap') {
            return $urls;
        }
    }

    $result = fwembed_make_request($store_url);
    if ($result['error']) {
        return "Error fetching store URL: " . $result['error'];
    }

    return fwembed_extract_product_urls($result['content'], $store_url);
}

/**
 * Get random products from store or specified URLs
 *
 * @param string $store_url The store URL to fetch products from (optional if urls provided)
 * @param array $urls Array of specific product URLs to randomize from
 * @param int $count Number of products to display
 * @param string $source Where to discover products: 'auto', 'sitemap' or 'page'
 * @return string HTML content of random products
 */
function fwembed_get_random_products($store_url = null, $urls = array(), $count = 3, $source = 'auto') {
    $product_urls = array();

    // If specific URLs are provided, use those
    if (!empty($urls)) {
        $product_urls = $urls;
    }
    // Otherwise, discover all products in the store
    elseif ($store_url) {
        $product_urls = fwembed_collect_product_urls($store_url, $source);
        if (is_string($product_urls)) {
            return $product_urls;
        }
    } else {
        return "Error: Either store URL or product URLs must be provided";
    }

    // Shuffle the URLs to randomize
    shuffle($product_urls);
    
    // Limit to requested count
    $selected_urls = array_slice($product_urls, 0, $count);
    
    // Fetch every uncached product page in one parallel batch rather than
    // serially, so a cold cache costs one round trip instead of N.
    $responses = fwembed_make_requests($selected_urls);

    $html = '';
    foreach ($selected_urls as $url) {
        if (empty($responses[$url]) || $responses[$url]['error'] || empty($responses[$url]['content'])) {
            continue;
        }

        $html .= fwembed_parse_single_product($responses[$url]['content'], $url, false);
    }

    return $html;
}

function fwembed_random_shortcode($atts) {
    $atts = shortcode_atts(
        array(
            'count' => '3',
            'urls' => '',
            'store_url' => '',
            'source' => 'auto',
        ),
        $atts
    );

    $source = strtolower(trim($atts['source']));
    if (!in_array($source, array('auto', 'sitemap', 'page'), true)) {
        $source = 'auto';
    }

    $count = intval($atts['count']);
    if ($count <= 0) {
        $count = 3;
    }
    
    $urls = array();
    if (!empty($atts['urls'])) {
        // Split URLs by comma and clean them up
        $url_array = explode(',', $atts['urls']);
        foreach ($url_array as $url) {
            $clean_url = trim($url);
            if (!empty($clean_url)) {
                $urls[] = $clean_url;
            }
        }
    }
    
    $store_url = '';
    if (!empty($atts['store_url'])) {
        $store_url = trim($atts['store_url']);
    } else {
        // Use default store URL from settings if no store_url provided
        $options = get_option('fourthwall_settings_name');
        $store_url = isset($options['fourth_url']) ? $options['fourth_url'] : '';
    }
    
    $products_html = fwembed_get_random_products($store_url, $urls, $count, $source);
    
    if (empty($products_html)) {
        return '<p>No products found to display.</p>';
    }
    
    return '<div class="fw-random-products">' . PHP_EOL . $products_html . PHP_EOL . '</div>';
}