php抓取网站图片的程序

<p><?php</p>/*完成网页内容捕获功能*/<br />function get_img_url($site_name){<br />    $site_fd = fopen($site_name, "r");<br />    $site_content = "";<br />    while (!feof($site_fd)) {<br />        $site_content .= fread($site_fd, 1024);<br />    }<br />   /*利用正则表达式得到图片链接*/<br />    $reg_tag = '/<img .*?\"([^\"]*(jpg|bmp|jpeg|gif)).*? alt="php抓取网站图片的程序" >/';<br />    $ret = preg_match_all($reg_tag, $site_content, $match_result);<br />    fclose($site_fd);<br />    return $match_result[1];<br />}<br /><br />/* 对图片链接进行修正 */<br />function revise_site($site_list, $base_site){<br />    foreach($site_list as $site_item) {<br />        if (preg_match('/^http/', $site_item)) {<br />            $return_list[] = $site_item;<br />        }else{<br />            $return_list[] = $base_site."/".$site_item;<br />    }<br />    }<br />    return $return_list;<br />}<br /><br />/*得到图片名字，并将其保存在指定位置*/<br />function get_pic_file($pic_url_array, $pos){<br />    $reg_tag = '/.*\/(.*?)$/';<br />    $count = 0;<br />    foreach($pic_url_array as $pic_item){<br />        $ret = preg_match_all($reg_tag,$pic_item,$t_pic_name);<br />        $pic_name = $pos.$t_pic_name[1][0];<br />        $pic_url = $pic_item;<br />    print("Downloading ".$pic_url." ");<br />        $img_read_fd = fopen($pic_url,"r");<br />        $img_write_fd = fopen($pic_name,"w");<br />        $img_content = "";<br />        while(!feof($img_read_fd)){<br />            $img_content .= fread($img_read_fd,1024);<br />          <br />        }<br />        fwrite($img_write_fd,$img_content);<br />        fclose($img_read_fd);<br />        fclose($img_write_fd);<br />        print("[OK] ");<br />    }<br />    return 0;<br />}<br /><br />function main(){<br />/* 待抓取图片的网页地址 */<br />    $site_name = "http://image.cn.yahoo.com";<br />    $img_url = get_img_url($site_name);<br />    $img_url_revised = revise_site($img_url, $site_name);<br />    $img_url_unique = array_unique($img_url_revised); //unique array<br />    get_pic_file($img_url_unique,"./"); <br />}<br /><br />main();<br /><p>?></p>