Home  >  Article  >  php教程  >  PHP supports breakpoint resume downloading and chunked downloading classes

PHP supports breakpoint resume downloading and chunked downloading classes

大家讲道理
大家讲道理Original
2016-11-08 13:28:061100browse

<?php
 
/**
 * 支持断点下载的类
 */
class downloader {
 
    /**
     * download file to local path
     *
     * @param       $url
     * @param       $save_file
     * @param int   $speed
     * @param array $headers
     * @param int   $timeout
     * @return bool
     * @throws Exception
     */
    static function get($url, $save_file, $speed = 10240, $headers = array(), $timeout = 10) {
        $url_info = self::parse_url($url);
        if (!$url_info[&#39;host&#39;]) {
            throw new Exception(&#39;Url is Invalid&#39;);
        }
 
        // default header
        $def_headers = array(
            &#39;Accept&#39;          => &#39;*/*&#39;,
            &#39;User-Agent&#39;      => &#39;Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0)&#39;,
            &#39;Accept-Encoding&#39; => &#39;gzip, deflate&#39;,
            &#39;Host&#39;            => $url_info[&#39;host&#39;],
            &#39;Connection&#39;      => &#39;Close&#39;,
            &#39;Accept-Language&#39; => &#39;zh-cn&#39;,
        );
 
        // merge heade
        $headers = array_merge($def_headers, $headers);
        // get content length
        $content_length = self::get_content_size($url_info[&#39;host&#39;], $url_info[&#39;port&#39;], $url_info[&#39;request&#39;], $headers, $timeout);
 
        // content length not exist
        if (!$content_length) {
            throw new Exception(&#39;Content-Length is Not Exists&#39;);
        }
        // get exists length
        $exists_length = is_file($save_file) ? filesize($save_file) : 0;
        // get tmp data file
        $data_file = $save_file . &#39;.data&#39;;
        // get tmp data
        $exists_data = is_file($data_file) ? json_decode(file_get_contents($data_file), 1) : array();
        // check file is valid
        if ($exists_length == $content_length) {
            $exists_data && @unlink($data_file);
            return true;
        }
        // check file is expire
        if ($exists_data[&#39;length&#39;] != $content_length || $exists_length > $content_length) {
            $exists_data = array(
                &#39;length&#39; => $content_length,
            );
        }
        // write exists data
        file_put_contents($data_file, json_encode($exists_data));
 
        try {
            $download_status = self::download_content($url_info[&#39;host&#39;], $url_info[&#39;port&#39;], $url_info[&#39;request&#39;], $save_file, $content_length, $exists_length, $speed, $headers, $timeout);
            if ($download_status) {
                @unlink($data_file);
            }
        } catch (Exception $e) {
            throw new Exception($e->getMessage());
        }
        return true;
    }
 
    /**
     * parse url
     *
     * @param $url
     * @return bool|mixed
     */
    static function parse_url($url) {
        $url_info = parse_url($url);
        if (!$url_info[&#39;host&#39;]) {
            return false;
        }
        $url_info[&#39;port&#39;]    = $url_info[&#39;port&#39;] ? $url_info[&#39;host&#39;] : 80;
        $url_info[&#39;request&#39;] = $url_info[&#39;path&#39;] . ($url_info[&#39;query&#39;] ? &#39;?&#39; . $url_info[&#39;query&#39;] : &#39;&#39;);
        return $url_info;
    }
 
    /**
     * download content by chunk
     *
     * @param $host
     * @param $port
     * @param $url_path
     * @param $headers
     * @param $timeout
     */
    static function download_content($host, $port, $url_path, $save_file, $content_length, $range_start, $speed, &$headers, $timeout) {
        $request = self::build_header(&#39;GET&#39;, $url_path, $headers, $range_start);
        $fsocket = @fsockopen($host, $port, $errno, $errstr, $timeout);
        stream_set_blocking($fsocket, TRUE);
        stream_set_timeout($fsocket, $timeout);
        fwrite($fsocket, $request);
        $status = stream_get_meta_data($fsocket);
        if ($status[&#39;timed_out&#39;]) {
            throw new Exception(&#39;Socket Connect Timeout&#39;);
        }
        $is_header_end = 0;
        $total_size    = $range_start;
        $file_fp       = fopen($save_file, &#39;a+&#39;);
        while (!feof($fsocket)) {
            if (!$is_header_end) {
                $line = @fgets($fsocket);
                if (in_array($line, array("\n", "\r\n"))) {
                    $is_header_end = 1;
                }
                continue;
            }
            $resp        = fread($fsocket, $speed);
            $read_length = strlen($resp);
            if ($resp === false || $content_length < $total_size + $read_length) {
                fclose($fsocket);
                fclose($file_fp);
                throw new Exception(&#39;Socket I/O Error Or File Was Changed&#39;);
            }
            $total_size += $read_length;
            fputs($file_fp, $resp);
            // check file end
            if ($content_length == $total_size) {
                break;
            }
            sleep(1);
            // for test
            //break;
        }
        fclose($fsocket);
        fclose($file_fp);
        return true;
 
    }
 
    /**
     * get content length
     *
     * @param $host
     * @param $port
     * @param $url_path
     * @param $headers
     * @param $timeout
     * @return int
     */
    static function get_content_size($host, $port, $url_path, &$headers, $timeout) {
        $request = self::build_header(&#39;HEAD&#39;, $url_path, $headers);
        $fsocket = @fsockopen($host, $port, $errno, $errstr, $timeout);
        stream_set_blocking($fsocket, TRUE);
        stream_set_timeout($fsocket, $timeout);
        fwrite($fsocket, $request);
        $status = stream_get_meta_data($fsocket);
        $length = 0;
        if ($status[&#39;timed_out&#39;]) {
            return 0;
        }
        while (!feof($fsocket)) {
            $line = @fgets($fsocket);
            if (in_array($line, array("\n", "\r\n"))) {
                break;
            }
            $line = strtolower($line);
            // get location
            if (substr($line, 0, 9) == &#39;location:&#39;) {
                $location = trim(substr($line, 9));
                $url_info = self::parse_url($location);
                if (!$url_info[&#39;host&#39;]) {
                    return 0;
                }
                fclose($fsocket);
                return self::get_content_size($url_info[&#39;host&#39;], $url_info[&#39;port&#39;], $url_info[&#39;request&#39;], $headers, $timeout);
            }
            // get content length
            if (strpos($line, &#39;content-length:&#39;) !== false) {
                list(, $length) = explode(&#39;content-length:&#39;, $line);
                $length = (int)trim($length);
            }
        }
        fclose($fsocket);
        return $length;
 
    }
 
    /**
     * build header for socket
     *
     * @param     $action
     * @param     $url_path
     * @param     $headers
     * @param int $range_start
     * @return string
     */
    static function build_header($action, $url_path, &$headers, $range_start = -1) {
        $out = $action . " {$url_path} HTTP/1.0\r\n";
        foreach ($headers as $hkey => $hval) {
            $out .= $hkey . &#39;: &#39; . $hval . "\r\n";
        }
        if ($range_start > -1) {
            $out .= "Accept-Ranges: bytes\r\n";
            $out .= "Range: bytes={$range_start}-\r\n";
        }
        $out .= "\r\n";
 
        return $out;
    }
}
 
 
#use age
/*
try {
    if (downloader::get(&#39;http://dzs.aqtxt.com/files/11/23636/201604230358308081.rar&#39;, &#39;test.rar&#39;)) {
        //todo
        echo &#39;Download Succ&#39;;
    }
} catch (Exception $e) {
    echo &#39;Download Failed&#39;;
}
*/
?>

Statement:
The content of this article is voluntarily contributed by netizens, and the copyright belongs to the original author. This site does not assume corresponding legal responsibility. If you find any content suspected of plagiarism or infringement, please contact admin@php.cn
Previous article:curl class encapsulationNext article:curl class encapsulation