* - The whole site, without touching any page: load this file before every PHP script. * Apache with mod_php, in .htaccess: php_value auto_prepend_file "/full/path/to/bizgenius-ssr.php" * PHP-FPM, in .user.ini: auto_prepend_file = "/full/path/to/bizgenius-ssr.php" * If your framework throws away output buffers, pass the finished HTML through bizgenius_ssr_fill($html). * * Optional settings: define them anywhere before the page finishes, or uncomment them here. */ // define('BIZGENIUS_SSR_KEY', 'YOUR_WIDGET_KEY'); // widget key, if it isn't in the page's meta tag or script URL // define('BIZGENIUS_SSR_CACHE_DIR', '/path/to/cache'); // a writable folder (default: the system temp folder) // define('BIZGENIUS_SSR_DISABLE', true); // switch it off /* * Each box is cached for 12 hours, also when there is nothing to show; a failed request is retried after 30 minutes. If BizGenius * doesn't answer within 3 seconds, the page is sent unchanged and the widget loads in the browser as before. * Only HTML pages are touched: JSON, images and downloads pass straight through. * * Requires PHP 7.2+ with the curl extension (or allow_url_fopen). */ // Everything below is inside this "if", so a second copy of this file (say, auto_prepend_file plus a // require of a copy elsewhere) is skipped instead of stopping the page with "Cannot redeclare function". if (!function_exists('bizgenius_ssr_fill')): if (!defined('BIZGENIUS_SSR_ENDPOINT')) { define('BIZGENIUS_SSR_ENDPOINT', 'https://www.bizgeniusapp.com/api/render.php'); } define('BIZGENIUS_SSR_TTL', 12 * 3600); // seconds a rendered box is reused define('BIZGENIUS_SSR_MISS_TTL', 30 * 60); // seconds before retrying a box whose request failed define('BIZGENIUS_SSR_TIMEOUT', 3); // seconds for all requests of one page (they run in parallel) // What counts as an "empty" box: whitespace or non-breaking spaces (some editors save
as
 
) define('BIZGENIUS_SSR_EMPTY', '(?:\s| | |\xC2\xA0)*'); function bizgenius_ssr_buffer($chunk, $phase) { if (($phase & PHP_OUTPUT_HANDLER_CLEAN) || (defined('BIZGENIUS_SSR_DISABLE') && BIZGENIUS_SSR_DISABLE) || !bizgenius_ssr_is_html()) { return $chunk; } try { return bizgenius_ssr_fill($chunk); } catch (Throwable $e) { return $chunk; // never break the page } } /** The page with every empty, supported widget box filled. Safe to call more than once. */ function bizgenius_ssr_fill($html) { // A page sent in parts (ob_flush(), or a page over 4 MB) has its key in the first part static $key = ''; static $sourcekeyword = ''; if (!is_string($html) || stripos($html, 'bizgenius') === false) { return $html; } $found = bizgenius_ssr_find_key($html); if ($found !== '') { $key = $found; } $found = bizgenius_ssr_find_sourcekeyword($html); if ($found !== null) { $sourcekeyword = $found; } return bizgenius_ssr_fill_page($html, bizgenius_ssr_host(), $key, 'bizgenius_ssr_fetch_all', $sourcekeyword); } /** Whether this response is an HTML page (the Content-Type we are sending, or PHP's default). */ function bizgenius_ssr_is_html() { if (isset($_SERVER['REQUEST_METHOD']) && $_SERVER['REQUEST_METHOD'] === 'HEAD') { return false; } foreach (headers_list() as $header) { if (stripos($header, 'content-type:') === 0) { return stripos($header, 'text/html') !== false; } } $default = (string)ini_get('default_mimetype'); return $default === '' || stripos($default, 'text/html') !== false; } function bizgenius_ssr_host() { $host = isset($_SERVER['HTTP_HOST']) ? strtolower((string)$_SERVER['HTTP_HOST']) : ''; return preg_replace('/:\d+$/', '', $host); } /** * Fill every empty, supported widget box in $html. Pure function apart from $fetcher, which receives * a list of render.php URLs and returns url => html. Same page scan as the WordPress plugin's * bgssr_fill() (integrations/wordpress/bizgenius-ssr). * * The page's own URL is deliberately not sent (render.php's href is optional). Nothing in the response * depends on it, and the request URL is the cache key, so including it would give every page - and every * ?utm_source=... variant of it - its own cache entry and its own call to BizGenius. * * $sourcekeyword is the widget script URL's sourcekeyword=... (one account on several brand sites, each * showing only its own papers); it goes with every request, as citation.js sends it. */ function bizgenius_ssr_fill_page($html, $host, $key, $fetcher, $sourcekeyword = '') { if (!is_string($html) || $html === '' || stripos($html, 'bizgenius') === false || $host === '' || $key === '') { return $html; } // Page-wide context that citation.js sends with the counts/form request. $form = bizgenius_ssr_container_attrs($html, 'bizgenius-citation-widget-container-form'); $map = bizgenius_ssr_container_attrs($html, 'bizgenius-citation-widget-container-map'); $labels = array(); foreach (array('count-total', 'count-year', 'count-month', 'count-total-year-month') as $c) { $attrs = bizgenius_ssr_container_attrs($html, 'bizgenius-citation-widget-container-' . $c); if ($attrs !== null) { $labels[$c] = isset($attrs['label']) ? $attrs['label'] : ''; } } $jobs = array(); // each: url, offset, length, and for product badges the product name // Empty boxes only:
(or , as the AI Summary box may be) with nothing but whitespace // or   inside. if (preg_match_all('~<(div|span)\b([^>]*)>(' . BIZGENIUS_SSR_EMPTY . ')~i', $html, $hits, PREG_SET_ORDER | PREG_OFFSET_CAPTURE)) { foreach ($hits as $hit) { $attrs = bizgenius_ssr_parse_attrs($hit[2][0]); $part = bizgenius_ssr_part($attrs); if ($part === '') { continue; } $query = array('key' => $key, 'domain' => $host, 'part' => $part, 'a' => bizgenius_ssr_data_attrs($attrs)); if (strpos($part, 'count-') === 0 || $part === 'form') { // same page context for all of them, so they make one cached getData request // and show the same numbers $query['f'] = $form === null ? array() : $form; $query['m'] = $map === null ? array() : $map; $query['l'] = $labels; } if ($sourcekeyword !== '') { $query['sourcekeyword'] = $sourcekeyword; } $jobs[] = array('url' => BIZGENIUS_SSR_ENDPOINT . '?' . http_build_query($query), 'offset' => $hit[3][1], 'length' => strlen($hit[3][0]), 'product' => null); } } // Product citation-count badges ( or
, often dozens on a listing page): one request // per 100 products that share the same badge settings, not one per badge. $groups = array(); if (preg_match_all('~<(span|div)\b([^>]*\bbizgenius-citation-widget-container-count-product\b[^>]*)>(' . BIZGENIUS_SSR_EMPTY . ')~i', $html, $hits, PREG_SET_ORDER | PREG_OFFSET_CAPTURE)) { foreach ($hits as $hit) { $data = bizgenius_ssr_data_attrs(bizgenius_ssr_parse_attrs($hit[2][0])); $product = isset($data['product-name']) ? trim($data['product-name']) : ''; if ($product === '') { continue; } unset($data['product-name']); ksort($data); $groups[serialize($data)]['settings'] = $data; $groups[serialize($data)]['badges'][] = array('product' => $product, 'offset' => $hit[3][1], 'length' => strlen($hit[3][0])); } } foreach ($groups as $group) { $products = array_values(array_unique(array_map(function ($b) { return $b['product']; }, $group['badges']))); foreach (array_chunk($products, 100) as $chunk) { $url = BIZGENIUS_SSR_ENDPOINT . '?' . http_build_query(array('key' => $key, 'domain' => $host, 'part' => 'count-product', 'products' => $chunk, 'a' => $group['settings']) + ($sourcekeyword !== '' ? array('sourcekeyword' => $sourcekeyword) : array())); foreach ($group['badges'] as $b) { if (in_array($b['product'], $chunk, true)) { $jobs[] = array('url' => $url, 'offset' => $b['offset'], 'length' => $b['length'], 'product' => $b['product']); } } } } if (!$jobs) { return $html; } $urls = array(); foreach ($jobs as $job) { $urls[$job['url']] = true; } $fragments = call_user_func($fetcher, array_keys($urls)); // Badge responses are JSON: product => html. $badgeHtml = array(); foreach ($jobs as $job) { if ($job['product'] !== null && !isset($badgeHtml[$job['url']])) { $badgeHtml[$job['url']] = array(); $decoded = isset($fragments[$job['url']]) ? json_decode((string)$fragments[$job['url']], true) : null; foreach ((is_array($decoded) && isset($decoded['results']) && is_array($decoded['results'])) ? $decoded['results'] : array() as $r) { if (isset($r['product'], $r['html']) && is_string($r['html'])) { $badgeHtml[$job['url']][(string)$r['product']] = $r['html']; } } } } // Insert from the end of the page backwards so earlier offsets stay valid. usort($jobs, function ($x, $y) { return $y['offset'] - $x['offset']; }); foreach ($jobs as $job) { if ($job['product'] === null) { $fragment = isset($fragments[$job['url']]) ? $fragments[$job['url']] : ''; } else { $fragment = isset($badgeHtml[$job['url']][$job['product']]) ? $badgeHtml[$job['url']][$job['product']] : ''; } if (is_string($fragment) && strpos($fragment, 'bizgenius-ssr') !== false) { $html = substr_replace($html, $fragment, $job['offset'], $job['length']); } } return $html; } /** Which render.php part a box is, or '' if we don't server-render it. */ function bizgenius_ssr_part($attrs) { $id = isset($attrs['id']) ? strtolower($attrs['id']) : ''; $prefix = 'bizgenius-citation-widget-container-'; $byId = array('list', 'list-top', 'list-new-papers-pubmed', 'list-new-papers-pmc', 'list-figure', 'count-total', 'count-year', 'count-month', 'count-total-year-month', 'form', 'summary'); if (strpos($id, $prefix) === 0 && in_array(substr($id, strlen($prefix)), $byId, true)) { $part = substr($id, strlen($prefix)); return $part === 'summary' ? 'aisummary' : $part; } $class = isset($attrs['class']) ? ' ' . strtolower($attrs['class']) . ' ' : ''; if (strpos($class, ' bizgenius-top-journal-widget-container ') !== false) { return 'top-journal'; } if (strpos($class, ' bizgenius-top-university-widget-container ') !== false) { return 'top-university'; } if (strpos($class, ' bizgenius-top-company-widget-container ') !== false) { return 'top-company'; } if (strpos($class, ' bizgenius-top-university-and-company-widget-container ') !== false) { return 'top-university-and-company'; } if (strpos($class, ' bizgenius-citation-widget-container-aisummary ') !== false || strpos($class, ' bizgenius-citation-widget-container-summary ') !== false) { return 'aisummary'; // the summary box is the AI Summary under its short-lived earlier name } return ''; } /** Widget key: BIZGENIUS_SSR_KEY, , or citation.min.js?key=... */ function bizgenius_ssr_find_key($html) { if (defined('BIZGENIUS_SSR_KEY') && BIZGENIUS_SSR_KEY) { return (string)BIZGENIUS_SSR_KEY; } if (preg_match_all('~]*)>~i', $html, $metas)) { foreach ($metas[1] as $raw) { $attrs = bizgenius_ssr_parse_attrs($raw); if (isset($attrs['name'], $attrs['content']) && strtolower($attrs['name']) === 'bizgeniusapp-key') { return trim($attrs['content']); } } } // www.biz-genius.com/api/citation.min.js is the widget's old address, still on some pages if (preg_match('~(?:bizgeniusapp\.(?:com|cn)|biz-genius\.com)/api/citation[\w.]*\.js\?[^"\'>]*?\bkey=([^&"\'>\s]+)~i', $html, $m)) { return trim(rawurldecode(html_entity_decode($m[1], ENT_QUOTES))); } return ''; } /** * The sourcekeyword=... of the widget's citation.min.js URL ('' if it has none), or null when the * script isn't in $html. Like citation.js, the first widget script on the page counts. */ function bizgenius_ssr_find_sourcekeyword($html) { if (!preg_match('~]*\bsrc\s*=\s*["\']?[^"\'>\s]*(?:bizgeniusapp\.(?:com|cn)|biz-genius\.com)/api/citation[\w.]*\.js(\?[^"\'>\s]*)?~i', $html, $m)) { return null; } parse_str(isset($m[1]) ? html_entity_decode(substr($m[1], 1), ENT_QUOTES) : '', $query); return isset($query['sourcekeyword']) && is_string($query['sourcekeyword']) ? trim($query['sourcekeyword']) : ''; } /** Attributes of the element with this id, or null if it is not on the page. */ function bizgenius_ssr_container_attrs($html, $id) { if (!preg_match('~]*\bid\s*=\s*["\']' . preg_quote($id, '~') . '["\'][^>]*)>~i', $html, $m)) { return null; } return bizgenius_ssr_data_attrs(bizgenius_ssr_parse_attrs($m[1])); } function bizgenius_ssr_parse_attrs($raw) { $out = array(); if (preg_match_all('~([a-zA-Z_:][-a-zA-Z0-9_:.]*)\s*=\s*(?:"([^"]*)"|\'([^\']*)\'|([^\s"\'=<>`]+))~', $raw, $m, PREG_SET_ORDER)) { foreach ($m as $a) { $value = isset($a[4]) && $a[4] !== '' ? $a[4] : (isset($a[3]) && $a[3] !== '' ? $a[3] : $a[2]); $out[strtolower($a[1])] = html_entity_decode($value, ENT_QUOTES, 'UTF-8'); } } return $out; } /** data-x attributes as x => value. */ function bizgenius_ssr_data_attrs($attrs) { $out = array(); foreach ($attrs as $name => $value) { if (strpos($name, 'data-') === 0) { $out[substr($name, 5)] = $value; } } return $out; } /** Fetch render.php URLs: cached boxes first, the rest in parallel. Returns url => html ('' = nothing). */ function bizgenius_ssr_fetch_all($urls) { $out = array(); $missing = array(); foreach ($urls as $url) { $file = bizgenius_ssr_cache_file($url); // a file's mtime is when it expires minus BIZGENIUS_SSR_TTL (see bizgenius_ssr_cache_put) if ($file !== '' && is_file($file) && filemtime($file) > time() - BIZGENIUS_SSR_TTL) { $out[$url] = (string)@file_get_contents($file); } else { $missing[] = $url; } } if (!$missing) { return $out; } $fetched = function_exists('curl_multi_init') ? bizgenius_ssr_fetch_parallel($missing) : bizgenius_ssr_fetch_serial($missing); foreach ($missing as $url) { $body = isset($fetched[$url]) ? $fetched[$url] : null; // null = request failed, '' = nothing to show (204) $ok = is_string($body) && strpos($body, 'bizgenius-ssr') !== false; // "nothing to show" is an answer, so keep it as long as a box: a 204 (e.g. an uncited product's list), // or product badges that are all hidden at 0 ({"results":[...]} with no badge HTML in it) $answered = $ok || $body === '' || (is_string($body) && strncmp($body, '{"results":', 11) === 0); bizgenius_ssr_cache_put(bizgenius_ssr_cache_file($url), $ok ? $body : '', $answered ? BIZGENIUS_SSR_TTL : BIZGENIUS_SSR_MISS_TTL); $out[$url] = $ok ? $body : ''; } return $out; } /** Cache file for one render.php URL, or '' if there is no writable cache folder. */ function bizgenius_ssr_cache_file($url) { $dir = defined('BIZGENIUS_SSR_CACHE_DIR') ? (string)BIZGENIUS_SSR_CACHE_DIR : sys_get_temp_dir() . '/bizgenius-ssr'; $dir = rtrim($dir, '/\\'); if (!is_dir($dir) && !@mkdir($dir, 0700, true) && !is_dir($dir)) { return ''; } return $dir . '/' . md5($url) . '.html'; } /** Write a cache file that expires $ttl seconds from now, without readers ever seeing half a file. */ function bizgenius_ssr_cache_put($file, $body, $ttl) { if ($file === '') { return; } $tmp = $file . '.' . getmypid() . '.tmp'; if (@file_put_contents($tmp, $body) !== false && @rename($tmp, $file)) { @touch($file, time() - BIZGENIUS_SSR_TTL + $ttl); } else { @unlink($tmp); } } function bizgenius_ssr_user_agent() { return 'BizGenius-SSR/1.0 (PHP; ' . bizgenius_ssr_host() . ')'; } function bizgenius_ssr_fetch_parallel($urls) { $multi = curl_multi_init(); $handles = array(); foreach ($urls as $url) { $h = curl_init($url); curl_setopt_array($h, array( CURLOPT_RETURNTRANSFER => true, CURLOPT_FOLLOWLOCATION => false, CURLOPT_CONNECTTIMEOUT => BIZGENIUS_SSR_TIMEOUT, CURLOPT_TIMEOUT => BIZGENIUS_SSR_TIMEOUT, CURLOPT_USERAGENT => bizgenius_ssr_user_agent(), CURLOPT_ENCODING => '', )); curl_multi_add_handle($multi, $h); $handles[$url] = $h; } do { $status = curl_multi_exec($multi, $running); if ($running) { curl_multi_select($multi, 0.2); } } while ($running && $status === CURLM_OK); $out = array(); foreach ($handles as $url => $h) { $code = (int)curl_getinfo($h, CURLINFO_HTTP_CODE); $body = curl_multi_getcontent($h); $out[$url] = ($code === 200 && is_string($body)) ? $body : ($code === 204 ? '' : null); curl_multi_remove_handle($multi, $h); if (PHP_VERSION_ID < 80000) { curl_close($h); // a no-op since PHP 8.0, deprecated in 8.5 } } curl_multi_close($multi); return $out; } /** Without curl: one request at a time, all within BIZGENIUS_SSR_TIMEOUT seconds. */ function bizgenius_ssr_fetch_serial($urls) { $out = array(); $deadline = microtime(true) + BIZGENIUS_SSR_TIMEOUT; foreach ($urls as $url) { $left = $deadline - microtime(true); if ($left <= 0) { $out[$url] = null; continue; } $context = stream_context_create(array('http' => array( 'timeout' => $left, 'follow_location' => 0, 'ignore_errors' => true, 'user_agent' => bizgenius_ssr_user_agent(), ))); $fp = @fopen($url, 'rb', false, $context); if ($fp === false) { $out[$url] = null; continue; } $meta = stream_get_meta_data($fp); $body = stream_get_contents($fp); fclose($fp); $code = (isset($meta['wrapper_data'][0]) && preg_match('~^HTTP/\S+\s+(\d{3})~', (string)$meta['wrapper_data'][0], $m)) ? (int)$m[1] : 0; $out[$url] = ($code === 200 && is_string($body)) ? $body : ($code === 204 ? '' : null); } return $out; } // Start collecting the page (after the functions above exist: inside the "if" they are declared in order). // The chunk size keeps a large download from being held in memory: past 4 MB the buffer is sent on in // parts, and bizgenius_ssr_buffer() leaves anything that isn't HTML alone. if (PHP_SAPI !== 'cli' && !(defined('BIZGENIUS_SSR_DISABLE') && BIZGENIUS_SSR_DISABLE)) { ob_start('bizgenius_ssr_buffer', 4 * 1024 * 1024); } endif;