php模仿百度spider蜘蛛爬虫程序例子

<?php

#加载页面
function curl_get($url){
        $ch=curl_init();
        curl_setopt($ch,CURLOPT_URL,$url);
        curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
        curl_setopt($ch,CURLOPT_HEADER,1);
        $result=curl_exec($ch);
        $code=curl_getinfo($ch,CURLINFO_HTTP_CODE);
 if($code!='404' && $result){
  return $result;
 }
        curl_close($ch);
}

#获取页面url链接
function get_page_urls($spider_page_result,$base_url){
  $get_url_result=preg_match_all("/<[a|A].*?href=[＼'＼"]{0,1}([^>＼'＼"＼ ]*).*?>/",$spider_page_result,$out);
  if($get_url_result){
   return $out[1];
  }else{
   return;
  }
}

#相对路径转绝对路径
function xdtojd($base_url,$url_list){
 if(is_array($url_list)){
  foreach($url_list as $url_item){
   if(preg_match("/^(http:＼/＼/|https:＼/＼/|javascript:)/",$url_item)){
    $result_url_list[]=$url_item;
   }else {
    if(preg_match("/^＼//",$url_item)){
     $real_url = $base_url.$url_item;
    }else{
     $real_url = $base_url."/".$url_item;
    }
    #$real_url = 'http://www.sumpay.cn/'.$url_item; 
                  $result_url_list[] = $real_url; 
   }
  }
  return $result_url_list;
 }else{
  return;
 }
}

#删除其他站点url
function other_site_url_del($jd_url_list,$url_base){
 
 if(is_array($jd_url_list)){
  foreach($jd_url_list as $all_url){
   echo $all_url;
   if(strpos($all_url,$url_base)===0){
    $all_url_list[]=$all_url;
   }
 

  }
  return $all_url_list;
 }else{
  return;
 }

}

#删除相同URL
function url_same_del($array_url){
 
 if(is_array($array_url)){
  $insert_url=array();
  $pizza=file_get_contents("/tmp/url.txt");
                if($pizza){
   $pizza=explode("＼r＼n",$pizza);
   foreach($array_url as $array_value_url){
    if(!in_array($array_value_url,$pizza)){
     $insert_url[]=$array_value_url;
    }
   }
   if($insert_url){
                                foreach($insert_url as $key => $insert_url_value){
                                        #这里只做了参数相同去重处理
     $update_insert_url=preg_replace('/=[^&]*/','=leesec',$insert_url_value);
                                        foreach($pizza as $pizza_value){
                                         $update_pizza_value=preg_replace('/=[^&]*/','=leesec',$pizza_value);
                                                if($update_insert_url==$update_pizza_value){
                                                 unset($insert_url[$key]);
                                                        continue;
                                                }

                                        }

    }
   }
 
  }else{
   $insert_url=array();
   $insert_new_url=array();
   $insert_url=$array_url;
   foreach($insert_url as $insert_url_value){
    $update_insert_url=preg_replace('/=[^&]*/','=leesec',$insert_url_value);
    $insert_new_url[]=$update_insert_url;  
   }
   $insert_new_url=array_unique($insert_new_url);
   foreach($insert_new_url as $key => $insert_new_url_val){
    $insert_url_bf[]=$insert_url[$key];
   }
   $insert_url=$insert_url_bf;
  }
  return $insert_url;
 }else{
  return;
 }
}

$current_url=$argv[1];
$fp_puts = fopen("/tmp/url.txt","ab");//记录url列表 
$fp_gets = fopen("/tmp/url.txt","r");//保存url列表 
$url_base_url=parse_url($current_url);
if($url_base_url['scheme']==""){
 $url_base="http://".$url_base_url['host'];
}else{
 $url_base=$url_base_url['scheme']."://".$url_base_url['host'];
}

do{
 $spider_page_result=curl_get($current_url);
 #var_dump($spider_page_result);
 $url_list=get_page_urls($spider_page_result,$url_base);
 #var_dump($url_list);
 if(!$url_list){
  continue;
 }
 $jd_url_list=xdtojd($url_base,$url_list);
 #var_dump($jd_url_list);
 $result_url_arr=other_site_url_del($jd_url_list,$url_base);
 var_dump($result_url_arr);
 $result_url_arr=url_same_del($result_url_arr);
 #var_dump($result_url_arr);
 if(is_array($result_url_arr)){ 
     $result_url_arr=array_unique($result_url_arr);
            foreach($result_url_arr as $new_url) { 
                fputs($fp_puts,$new_url."＼r＼n"); 
            }
 } 
 
}

while ($current_url = fgets($fp_gets,1024));//不断获得url 
preg_match_all("/<a[^>]+href=[＼"']([^＼"']+)[＼"'][^>]+>/",$spider_page_result,$out);
# echo a href
#var_dump($out[1]);

?>

<?php
header('content-type:text/html charset:utf-8');
$dir_base = "./files/"; //文件上传根目录
//没有成功上传文件，报错并退出。
if(empty($_FILES)) {
 echo "<textarea><img src='{$dir_base}error.jpg'/></textarea>";
 exit(0);
}
$output = "<textarea>";
$index = 0; //$_FILES 以文件name为数组下标，不适用foreach($_FILES as $index=>$file)
foreach($_FILES as $file){
 $upload_file_name = 'upload_file' . $index;//对应index.html FomData中的文件命名
 $filename = $_FILES[$upload_file_name]['name'];
 $gb_filename = iconv('utf-8','gb2312',$filename); //名字转换成gb2312处理
 //文件不存在才上传
 if(!file_exists($dir_base.$gb_filename)) {
  $isMoved = false;  //默认上传失败
  $MAXIMUM_FILESIZE = 1 * 1024 * 1024;  //文件大小限制 1M = 1 * 1024 * 1024 B;
  $rEFileTypes = "/^＼.(jpg|jpeg|gif|png){1}$/i";
  if ($_FILES[$upload_file_name]['size'] <= $MAXIMUM_FILESIZE &&
   preg_match($rEFileTypes, strrchr($gb_filename, '.'))) { 
   $isMoved = @move_uploaded_file ( $_FILES[$upload_file_name]['tmp_name'], $dir_base.$gb_filename);  //上传文件
  }
 }else{
  $isMoved = true;//已存在文件设置为上传成功
 }
 if($isMoved){
  //输出图片文件<img>标签
  //注：在一些系统src可能需要urlencode处理，发现图片无法显示，
  //    请尝试 urlencode($gb_filename) 或 urlencode($filename)，不行请查看HTML中显示的src并酌情解决。
  $output .= "<img src='{$dir_base}{$filename}' title='{$filename}' alt='{$filename}'/>";
 }else {
  $output .= "<img src='{$dir_base}error.jpg' title='{$filename}' alt='{$filename}'/>";
 }
 $index++;
}
echo $output."</textarea>";

/*PHP图片加文字水印类库
该类库暂时只支持文字水印，位置为右下角，颜色随机

调用方法：
1、在需要加水印的文件顶部引入类库：
include_once 'imageClass.php';
2、声明新类：
$tpl=new image_fu;
3、给图片水印提供参数：
$tpl->img(图片路径,水印文字,字体路径,字体大小,字体角度);
比如：$tpl->img('abc.jpg','这是水印文字','ziti.ttf',30,0)
*/

代码如下：

watermark.php
<?php
class image_fu{

private $image;
private $img_info;
private $img_width;
private $img_height;
private $img_im;
private $img_text;
private $img_ttf='';
private $img_new;
private $img_text_size;
private $img_jd;

function img($img='',$txt='',$ttf='',$size=12,$jiaodu=0){
if(isset($img)&&file_exists($img)){//检测图片是否存在
$this->image   =$img;
$this->img_text=$txt;
$this->img_text_size=$size;
$this->img_jd=$jiaodu;
if(file_exists($ttf)){
$this->img_ttf=$ttf;
}else{
exit('字体文件：'.$ttf.'不存在！');
}
$this->imgyesno();
}else{
exit('图片文件:'.$img.'不存在');
}
}

private function imgyesno(){

$this->img_info  =getimagesize($this->image);
$this->img_width =$this->img_info[0];//图片宽
$this->img_height=$this->img_info[1];//图片高

//检测图片类型
switch($this->img_info[2]){
case 1:$this->img_im = imagecreatefromgif($this->image);break;
case 2:$this->img_im = imagecreatefromjpeg($this->image);break;
case 3:$this->img_im = imagecreatefrompng($this->image);break;
default:exit('图片格式不支持水印');
}

$this->img_text();
}

private function img_text(){

imagealphablending($this->img_im,true);

//设定颜色
$color=imagecolorallocate($this->img_im,rand(0,255),rand(0,255),rand(0,255));
$txt_height=$this->img_text_size;
$txt_jiaodu=$this->img_jd;
$ttf_im=imagettfbbox($txt_height,$txt_jiaodu,$this->img_ttf,$this->img_text);
$w = $ttf_im[2] - $ttf_im[6];
$h = $ttf_im[3] - $ttf_im[7];
//$w = $ttf_im[7];
//$h = $ttf_im[8];

unset($ttf_im);

$txt_y     =$this->img_height-$h;
$txt_x     =$this->img_width-$w;
//$txt_y     =0;
//$txt_x     =0;

$this->img_new=@imagettftext($this->img_im,$txt_height,$txt_jiaodu,$txt_x,$txt_y,$color,

$this->img_ttf,$this->img_text);

@unlink($this->image);//删除图片
switch($this->img_info[2]) {//取得背景图片的格式
case 1:imagegif($this->img_im,$this->image);break;
case 2:imagejpeg($this->img_im,$this->image);break;
case 3:imagepng($this->img_im,$this->image);break;
default: exit('水印图片失败');
}

}

//显示图片
function img_show(){echo '<img src="'.$this->image.'" border="0" alt="'.$this->img_text.'" />';}

//释放内存
private function img_nothing(){
unset($this->img_info);
imagedestroy($this->img_im);
}

}

$addwk    =    new image_fu();
$addwk->img('2.jpg','testwatermark','华康少女字体.ttf',33,0);
$addwk->img_show();   //调用显示图片

?>

//重新设置session(防止新传头像无法显示)
$this->session->set_userdata('user_info',$user_info);
 
//项目决对路径
$base_path = rtrim(str_replace('＼＼','/', FCPATH),'/');
 
require APPPATH.'libraries/avatar.class.php';
 
$au = new avatar($base_path.'/uploads/user/'.$user_info['mid'].'/',$base_path.'/uploads/user/temp/','/uploads/user/','/faceapi/');
if(!$user_info['face']){
    $uid = 'uface_'.$user_info['mid'];
}else{
    $uid = $user_info['face'];
}
 
$this->data['urlAvatarBig']   = $user_info['face'] ? $au->getAvatarUrl($uid,'big') : (($this->user_info['sex']==0 )? static_url().'/uploads/user/dfgirl.png' : static_url().'/uploads/user/dfboy.png' );
$this->data['urlCameraFlash'] = $au->renderHtml($uid);
$this->data['uid'] = $uid;//头像标示
$this->load->view('center/member/edit_face',$this->data);

<div style="float:left;width:455px;height:253px;overflow:hidden;margin-left:10px;"><?php echo $urlCameraFlash ?></div>
 
<script type="text/javascript">
function updateavatar(){
    $('#userface').attr('src','/uploads/user/<?php echo $user_info['mid'].'/'.$uid?>_big.jpg?aid='+Math.random());
}
</script>

<?php
class avatar{
 public $savePath;
 public $tempPath;
 public $viewPath;
 public $urlPath;
 public $mid;
 public function __construct($savePath,$tempPath,$viewPath,$urlPath,$mid){
  $this->savePath=$savePath;
  $this->tempPath=$tempPath;
  $this->viewPath='http://'.$_SERVER['HTTP_HOST'].$viewPath;
  $this->urlPath='http://'.$_SERVER['HTTP_HOST'].$urlPath;
  $this->mid = $mid;

 }
 // 第一步：上传原始图片文件
 private function uploadAvatar($uid){
  // 检查上传文件的有效性
  if(empty($_FILES['Filedata'])){
   return -3; // No photograph be upload!
  }
  // 本地临时存储位置
  $tmpPath = $this->tempPath."{$uid}.jpg";
  // 如果临时存储的文件夹不存在，先创建它
  $dir=dirname($tmpPath);
  if(!file_exists($dir)){
   @mkdir($dir,0777, true );
  }
  // 如果同名的临时文件已经存在，先删除它
  if(file_exists($tmpPath)){
   @unlink($tmpPath);
  }
  // 把上传的图片文件保存到预定位置
  if ( @copy($_FILES['Filedata']['tmp_name'], $tmpPath) || @move_uploaded_file($_FILES['Filedata']['tmp_name'], $tmpPath)) {
   @unlink($_FILES['Filedata']['tmp_name']);
   list($width, $height, $type, $attr) = getimagesize($tmpPath);
   if ( $width < 10 || $height < 10 || $width > 3000 || $height > 3000 || $type == 4 ) {
    @unlink($tmpPath);
    return -2; // Invalid photograph!
   }
  } else {
   @unlink($_FILES['Filedata']['tmp_name']);
   return -4; // Can not write to the data/tmp folder!
  }

  // 用于访问临时图片文件的 url
  $tmpUrl =$this->viewPath."temp/{$uid}.jpg";
  return $tmpUrl;
 }
 private function flashdata_decode($s) {
  $r = '';
  $l = strlen($s);
  for($i=0; $i<$l; $i=$i+2) {
   $k1 = ord($s[$i]) - 48;
   $k1 -= $k1 > 9 ? 7 : 0;
   $k2 = ord($s[$i+1]) - 48;
   $k2 -= $k2 > 9 ? 7 : 0;
   $r .= chr($k1 << 4 | $k2);
  }
  return $r;
 }

 // 第二步：上传分割后的三个图片数据流
 private function rectAvatar( $uid ){
  // 从 $_POST 中提取出三个图片数据流
  $bigavatar    = $this->flashdata_decode( $_POST['avatar1'] );
  $middleavatar = $this->flashdata_decode( $_POST['avatar2'] );
  if ( !$bigavatar || !$middleavatar) {
   return '<root><message type="error" value="-2" /></root>';
  }
  
  //不存在目录，则创建
  if(!file_exists($this->savePath)){
   @mkdir($this->savePath,0777, true );
  }

  // 保存为图片文件
  $bigavatarfile    = $this->savePath."{$uid}_big.jpg";
  $middleavatarfile = $this->savePath."{$uid}_middle.jpg";

  $success = 1;
  $fp = @fopen($bigavatarfile, 'wb');
  @fwrite($fp, $bigavatar);
  @fclose($fp);

  $fp = @fopen($middleavatarfile, 'wb');
  @fwrite($fp, $middleavatar);
  @fclose($fp);

  // 验证图片文件的正确性
  $biginfo    = @getimagesize($bigavatarfile);
  $middleinfo = @getimagesize($middleavatarfile);
  if ( !$biginfo || !$middleinfo || $biginfo[2] == 4 || $middleinfo[2] == 4 ) {
   file_exists($bigavatarfile) && unlink($bigavatarfile);
   file_exists($middleavatarfile) && unlink($middleavatarfile);
   $success = 0;
  }
  // 删除临时存储的图片
  $tmpPath = $this->tempPath."{$uid}.jpg";
  @unlink($tmpPath);
  //临时保存头像
  $con=mysql_connect('localhost','root','root');
  mysql_select_db('zenyue');
  mysql_query("set names utf8");
  $sql="update zen_user set `face`='".$uid."' where mid='".$this->mid."'";
  mysql_query($sql);
  return '<?xml version="1.0" ?><root><face success="' . $success . '"/></root>';
 }
 // 从客户端访问头像图片的 url
 public function getAvatarUrl( $uid, $size='middle' ){
  $ci = &get_instance();
  $user_info = $ci->session->userdata('user_info');
  return $this->viewPath."{$user_info['mid']}/{$uid}_{$size}.jpg";
 }
 // 处理 HTTP Request
 // 返回值：如果是可识别的 request，处理后返回 true；否则返回 false。
 public function processRequest(){
  // 从 input 参数里拆解出自定义参数
  $arr = array();
  parse_str( $_GET['input'], $arr );
  $uid = $arr['uid'];
  if ( $_GET['a'] == 'uploadavatar') {
   // 第一步：上传原始图片文件
   echo $this->uploadAvatar( $uid );
   return true;
  } else if ( $_GET['a'] == 'rectavatar') {
   // 第二步：上传分割后的三个图片数据流
   echo $this->rectAvatar( $uid );
   return true;
  }
  return false;
 }
 // 编辑页面中包含 camera.swf 的 HTML 代码
 public function renderHtml( $uid ){
  // 把需要回传的自定义参数都组装在 input 里
  $ci = &get_instance();
  $user_info = $ci->session->userdata('user_info');
  $input = urlencode("uid={$uid}&FSESSIONID=".session_id().'&mid='.$user_info['mid']);
  $baseUrl = '/public/zen/js/avatar/';
  $uc_api = urlencode( $this->urlPath.'avatar.php');
  $urlCameraFlash = "{$baseUrl}camera.swf?m=user&inajax=1&appid=1&ucapi={$uc_api}&input={$input}&uploadSize=2048";
  $urlCameraFlash = '<script src="'.$baseUrl.'common.js?B6k" type="text/javascript"></script><script type="text/javascript">document.write(AC_FL_RunContent("width","455","height","253","scale","exactfit","src","'.$urlCameraFlash.'","id","mycamera","name","mycamera","quality","high","bgcolor","#ffffff","wmode","transparent","menu","false","swLiveConnect","true","allowScriptAccess","always"));</script>';
  return $urlCameraFlash;
 }
}

<?php
$num = 0; //初始化计数变量
$_REQUEST['val'.$num]; $num++; //插入到需要获取值的地方，每个需要获取传值的地方重复使用这串代码即可
?>