使用Sockets获取网页内容

socket.php

/**
 * Returns the contents of a web page
 *
 * Opens a remote web document, and returns the contents.  This method
 * uses sockets to get the document.  Returns the contents of the document with
 * the headers being stored in the $headers variable.
 * This method can handle 301 and 302 redirects, but will return false
 * if anything other than those status codes, or a 200, is returned by
 * the web server.
 *
 * @param       string $url          The URL to the web page
 * @param       string $headers The HTTP headers will be stored in this variable
 * @param       int        $port       Optional port number to use for grabbing the document
 * @param       int        $timeout    The number of seconds to try before giving up
 * @return      mixed
 */
function get_http_document($url, &$headers = null, $port = 80, $timeout = 8) {
        $pURL = parse_url($url);
       
        if (empty($pURL['host'])) {
                return false;
        }
 
        $remotePath =     (isset($pURL['path'])) ? $pURL['path'] : '/';
        $remoteDocument =       (empty($pURL['query'])) ? $remotePath : $remotePath . '?' . $pURL['query'];
       
        if (!$fp = fsockopen($pURL['host'], $port, $errno, $errstr, $timeout)) {
                return false;
        }
       
        $out =  "GET $remoteDocument HTTP/1.0\r\n";     
        $out .=         "Host: {$pURL['host']}\r\n";
        $out .=         "Connection: Close\r\n\r\n";
        fwrite($fp, $out);
        unset($out);
       
        $received = '';
        while (!feof($fp)) {
                $received .= fread($fp, 128);
        }
        fclose($fp);
 
        // Seperate the headers from the content
        $parts = explode("\r\n\r\n", $received, 2);
        $headers = $parts[0];
        $content = $parts[1];
        unset($parts);
       
 
        $headerParts = explode("\r\n", $headers);
        if (!preg_match('~HTTP/1\.\d ([\d]+)~i', $headerParts[0], $matches)) {
                return false;
        }
        $statusCode = $matches[1];
       
        if ($statusCode == 200) {
                return $content;
        } else if ($statusCode != 301 && $statusCode != 302) {
                return false;
        }
       
        if (!preg_match('~^Location:(.*)$~im', $headers, $matches)) {
                return false;
        }
        $newLocation = trim($matches[1]);
        return get_http_document($newLocation, $headers, $port, $timeout);
}