Capture the 'IC Trading Network' supplier program_PHP tutorial

/**
* Capture the "IC Trading Network" supplier main program
* author Lee.
* Last modify $Date: 2012-2-6 10:44:32$
* Note: This program is executed according to the encoding GB2312, because the "IC Trading Network" website is encoded in GB2312, and the database must be consistent
*/ 
class ic { 
    private $key; // 型号  
    private $pageNum; // 页码  
 
    /**
* Entry program
​​*/ 
    public function go($key) { 
        $this->key = $key; 
        $this->pageNum = $this->getPageNum(); 
        $this->getInfo(); 
    } 
 
    /**
* Get the supplier url link array
* @return ArrayObject
​​*/ 
    private function getInfo() { 
        if ($this->pageNum==1) { # 处理只有一页的情况 
            $arr = $this->shopUrlMatchReArr($this->getContent()); 
            $this->isAddSuccess($arr); 
        } elseif ($this->pageNum>1) { # 多页 
            for ($i=1; $i<=$this->pageNum; $i++) { 
                $arr = $this->shopUrlMatchReArr($this->getContent($i)); 
                $this->isAddSuccess($arr); 
            } 
        }    
    } 
     
    /**
* Print whether the addition was successful
* @param ArrayObject $arr
* @return string
​​*/ 
    private function isAddSuccess($arr) { 
        foreach ($arr as $k=>$v) { 
            if ($this->execAdd($this->getInfoByShopUrl($v))) { 
                echo 'Add Success!!'; 
            } else { 
                echo 'Add Faild!!'; 
            } 
        } 
    } 
 
    /**
* Execute adding to database
* @param ArrayObject $infoArr
* @return Number The number of affected rows
​​*/ 
    private function execAdd($infoArr) { 
        $mysqli = $this->getDb(); 
        if (!emptyempty($infoArr['company'])) { 
            if (!$this->isExists($mysqli, $infoArr)) { 
                $num = $mysqli->query("INSERT INTO ic(company,address,phone,mobile,fax,zip,person,qq,msn,email,website,regDate,shopUrl) VALUES ('{$infoArr['company']}','{$infoArr['address']}','{$infoArr['phone']}','{$infoArr['mobile']}','{$infoArr['fax']}','{$infoArr['zip']}','{$infoArr['person']}','{$infoArr['qq']}','{$infoArr['msn']}','{$infoArr['email']}','{$infoArr['website']}','{$infoArr['regDate']}','{$infoArr['shopUrl']}')"); 
                return $num; 
            } else { 
                return false; # 表示数据已经存在 
            } 
        } else { 
            return false; 
        } 
    } 
 
    /**
* Connect to database
​​*/ 
    private function getDb() { 
        $mysqli = new mysqli('localhost', 'root', '1715544', 'weiku'); 
        $mysqli->query('SET NAMES GB2312'); 
        return $mysqli; 
    } 
 
    /**
* Check if the company already exists
* @param Resource $mysqli
* @param ArrayObject $infoArr
* @return bool
​​*/ 
    private function isExists($mysqli, $infoArr) { 
        $mysqli->query("SELECT company FROM ic WHERE company = '{$infoArr['company']}'"); 
        if ($mysqli->affected_rows) { 
            return true; 
        } else { 
            return false; 
        } 
    } 
 
    /**
* Format string
* @param string $str
* @return string
​​*/ 
    private function formatString($str) { 
        return trim($str); 
    } 
 
    /**
* Grab information
* @param $url
* @return ArrayObject
​​*/ 
    private function getInfoByShopUrl($url) { 
        $re = $this->getUrlInfo($url); 
        if (stristr($re, '')) $re = preg_replace('/.*/Usi', '', $re); 
        preg_match_all('//Usi', $re, $companyArr); 
        preg_match_all('/地址:(.*)/Usi', $re, $addressArr); 
        preg_match_all('/电话：(.*)/Usi', $re, $phoneArr); 
        preg_match_all('/手机：(.*)/Usi', $re, $mobileArr); 
        preg_match_all('/传真：(.*)/Usi', $re, $faxArr); 
        preg_match_all('/邮编：(.*)/Usi', $re, $zipArr); 
        preg_match_all('/联系人：(.*)/Usi', $re, $personArr); 
        preg_match_all('/QQ：(.*)/Usi', $re, $qqArr); 
        preg_match_all('/MSN：(.*)/Usi', $re, $msnArr); 
        preg_match_all('/Email：(.*)/Usi', $re, $emailArr); 
        preg_match_all('/网址：(.*)/Usi', $re, $websiteArr); 
        preg_match_all('/注册日期：(.*)/Usi', $re, $regDateArr); 
        $infoArr = array( 
            'company'=>$this->formatString($companyArr[1][0]), 
            'address'=>$this->formatString($addressArr[1][0]), 
            'phone'=>$this->formatString($phoneArr[1][0]), 
            'mobile'=>$this->formatString($mobileArr[1][0]), 
            'fax'=>$this->formatString($faxArr[1][0]), 
            'zip'=>$this->formatString($zipArr[1][0]), 
            'person'=>$this->formatString($personArr[1][0]), 
            'qq'=>$this->formatString($qqArr[1][0]), 
            'msn'=>$this->formatString($msnArr[1][0]), 
            'email'=>$this->formatString($emailArr[1][0]), 
            'website'=>$this->stripATags($this->formatString($websiteArr[1][0])), 
            'regDate'=>$this->formatString($regDateArr[1][0]), 
            'shopUrl'=>$url 
        ); 
        return $infoArr; 
    } 
 
    /**
* Get the supplier url array based on the page
* @param string $re
* @return ArrayObject
​​*/ 
    private function shopUrlMatchReArr($re) { 
        preg_match_all('/.+/Usi', $re, $arr); 
        $arr = $this->formatUrlArr(array_unique($arr[1])); 
        return $arr; 
    } 
     
    /**
* * Format array
* @param Array $arr
* @return ArrayObject
​​*/ 
    private function formatUrlArr($arr) { 
        $newArr = array(); 
        foreach ($arr as $key=>$value) { 
            if ($this->isExistsHttp($value)) { 
                $newArr[$key] = $value; 
            } 
        } 
        return $newArr; 
    } 
     
    /**
* Format QQ
* @param string $str
* @return string
​​*/ 
    private function formatQqMsn($str, $e='QQ') { 
        if (emptyempty($str)) return ''; 
        preg_match_all('/alt="'.$e.':(.+)"/Usi', $str, $arr); 
        if (count($arr[1])==1) return $arr[1][0]; 
        $newStr = null; 
        foreach ($arr[1] as $value) { 
            $newStr .= $value . ' '; 
        } 
        return rtrim($newStr, ' '); 
    } 
 
    /**
     * 去掉网址的 A 标签
     * @param string $site
     * @return string
     */ 
    private function stripATags($site) { 
        $site = preg_replace('/(.+)/', '1', $site); 
        return $site; 
    } 
 
    /**
* Check if the url has http
* @param string $url
* @return bool
​​*/ 
    private function isExistsHttp($url) { 
        if (stristr($url, 'http://')) { 
            return true; 
        } else { 
            return false; 
        } 
    } 
     
    /**
* Get page content
* @param Number $page
* @return string
​​*/ 
    private function getContent($page=1) { 
        $re = file_get_contents($this->getUrl($this->key, $page)); 
        return $re; 
    } 
     
    /**
* Get page number
* @return Number
​​*/ 
    private function getPageNum() { 
        $i = null; 
        $re = $this->getContent(); 
        preg_match_all('/共(.+)页/Usi', $re, $arr); 
        $i = $arr[1][0]; 
        return $i; 
    } 
 
    /**
* Get URL link
* @param string $str
* @param int $page page number
* @return string
​​*/ 
    private function getUrl($str, $page=1) { 
        return "http://www.ic.net.cn/partsearch/searchinstock.asp?newtype=1&area=&Page={$page}&partnumber={$str}&mfg=&DateCode=&QTY=&PRICE=&Exact=&orderby=inputdate&qty_filter=50&usertype2=1&pack="; 
    } 
 
    /**
* Get page content
* @param string $url
* @return string
​​*/ 
    private function getUrlInfo($url) { 
        $re = file_get_contents($url); 
        return $re; 
    } 
} 
 
/*
程序运行思路：根据“IC 交易网”的IC搜索功能，输入型号进行搜索，然后抓取供应商信息
 
数据库结构
CREATE TABLE `ic` (
    `id` mediumint(8) unsigned NOT NULL auto_increment,
    `company` varchar(500) NOT NULL,
    `address` varchar(500) default NULL,
    `phone` varchar(500) default NULL,
    `mobile` varchar(500) default NULL,
    `fax` varchar(300) default NULL,
    `zip` varchar(300) default NULL,
    `person` varchar(500) default NULL,
    `qq` varchar(300) default NULL,
    `msn` varchar(300) default NULL,
    `email` varchar(500) default NULL,
    `website` varchar(300) default NULL,
    `regDate` varchar(500) default NULL,
    PRIMARY KEY  (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=gb2312
*/ 
 
$i = new ic(); 
$arr = array_unique(array('MAX3232', 'AML8613', 'MT6225A', 'OM8373PS/N3/A', 'PT7313', 'MAX8212ESA', 'TL431', 'S3C2440', 'TMS320F2812PGFA', 'PCM1704', 'AN6717', 'CA3162E', 'CA3161E', 'LM393N', 'DS18B20', 'SHT10', 'AML8613', 'AN6717', 'LM393N', 'CA3161E', 'CA3162E', 'PCM1704', 'STK392-040', 'K1667', 'MAX232', 'STM32F103', 'LM358')); 
foreach ($arr as $v) { 
    $i->go($v); 
} 
?> 
/**
* Capture the "IC Trading Network" supplier main program
* author Lee.
* Last modify $Date: 2012-2-6 10:44:32$
* Note: This program is executed according to the encoding GB2312, because the "IC Trading Network" website is encoded in GB2312, and the database must be consistent
*/
class ic {
 private $key; // 型号
 private $pageNum; // 页码

 /**
* Entry program
​*/
 public function go($key) {
  $this->key = $key;
  $this->pageNum = $this->getPageNum();
  $this->getInfo();
 }

 /**
* Get supplier url link array
* @return ArrayObject
​*/
 private function getInfo() {
  if ($this->pageNum==1) { # 处理只有一页的情况
   $arr = $this->shopUrlMatchReArr($this->getContent());
   $this->isAddSuccess($arr);
  } elseif ($this->pageNum>1) { # 多页
   for ($i=1; $i<=$this->pageNum; $i++) {
    $arr = $this->shopUrlMatchReArr($this->getContent($i));
    $this->isAddSuccess($arr);
   }
  } 
 }
 
 /**
* Print whether the addition was successful
* @param ArrayObject $arr
* @return string
​*/
 private function isAddSuccess($arr) {
  foreach ($arr as $k=>$v) {
   if ($this->execAdd($this->getInfoByShopUrl($v))) {
    echo 'Add Success!!';
   } else {
    echo 'Add Faild!!';
   }
  }
 }

 /**
* Execute adding to database
* @param ArrayObject $infoArr
* @return Number Number of affected rows
​*/
 private function execAdd($infoArr) {
  $mysqli = $this->getDb();
  if (!empty($infoArr['company'])) {
   if (!$this->isExists($mysqli, $infoArr)) {
    $num = $mysqli->query("INSERT INTO ic(company,address,phone,mobile,fax,zip,person,qq,msn,email,website,regDate,shopUrl) VALUES ('{$infoArr['company']}','{$infoArr['address']}','{$infoArr['phone']}','{$infoArr['mobile']}','{$infoArr['fax']}','{$infoArr['zip']}','{$infoArr['person']}','{$infoArr['qq']}','{$infoArr['msn']}','{$infoArr['email']}','{$infoArr['website']}','{$infoArr['regDate']}','{$infoArr['shopUrl']}')");
    return $num;
   } else {
    return false; # 表示数据已经存在
   }
  } else {
   return false;
  }
 }

 /**
* Connect to database
​*/
 private function getDb() {
  $mysqli = new mysqli('localhost', 'root', '1715544', 'weiku');
  $mysqli->query('SET NAMES GB2312');
  return $mysqli;
 }

 /**
* Check if the company already exists
* @param Resource $mysqli
* @param ArrayObject $infoArr
* @return bool
​*/
 private function isExists($mysqli, $infoArr) {
  $mysqli->query("SELECT company FROM ic WHERE company = '{$infoArr['company']}'");
  if ($mysqli->affected_rows) {
   return true;
  } else {
   return false;
  }
 }

 /**
* Format string
* @param string $str
* @return string
​*/
 private function formatString($str) {
  return trim($str);
 }

 /**
* Grab information
* @param $url
* @return ArrayObject
​*/
 private function getInfoByShopUrl($url) {
  $re = $this->getUrlInfo($url);
  if (stristr($re, '')) $re = preg_replace('/.*/Usi', '', $re);
  preg_match_all('//Usi', $re, $companyArr);
  preg_match_all('/地址:(.*)/Usi', $re, $addressArr);
  preg_match_all('/电话：(.*)/Usi', $re, $phoneArr);
  preg_match_all('/手机：(.*)/Usi', $re, $mobileArr);
  preg_match_all('/传真：(.*)/Usi', $re, $faxArr);
  preg_match_all('/邮编：(.*)/Usi', $re, $zipArr);
  preg_match_all('/联系人：(.*)/Usi', $re, $personArr);
  preg_match_all('/QQ：(.*)/Usi', $re, $qqArr);
  preg_match_all('/MSN：(.*)/Usi', $re, $msnArr);
  preg_match_all('/Email：(.*)/Usi', $re, $emailArr);
  preg_match_all('/网址：(.*)/Usi', $re, $websiteArr);
  preg_match_all('/注册日期：(.*)/Usi', $re, $regDateArr);
  $infoArr = array(
   'company'=>$this->formatString($companyArr[1][0]),
   'address'=>$this->formatString($addressArr[1][0]),
   'phone'=>$this->formatString($phoneArr[1][0]),
   'mobile'=>$this->formatString($mobileArr[1][0]),
   'fax'=>$this->formatString($faxArr[1][0]),
   'zip'=>$this->formatString($zipArr[1][0]),
   'person'=>$this->formatString($personArr[1][0]),
   'qq'=>$this->formatString($qqArr[1][0]),
   'msn'=>$this->formatString($msnArr[1][0]),
   'email'=>$this->formatString($emailArr[1][0]),
   'website'=>$this->stripATags($this->formatString($websiteArr[1][0])),
   'regDate'=>$this->formatString($regDateArr[1][0]),
   'shopUrl'=>$url
  );
  return $infoArr;
 }

 /**
* Get the supplier url array based on the page
* @param string $re
* @return ArrayObject
​*/
 private function shopUrlMatchReArr($re) {
  preg_match_all('/.+/Usi', $re, $arr);
  $arr = $this->formatUrlArr(array_unique($arr[1]));
  return $arr;
 }
 
 /**
* Format array
* @param Array $arr
* @return ArrayObject
​*/
 private function formatUrlArr($arr) {
  $newArr = array();
  foreach ($arr as $key=>$value) {
   if ($this->isExistsHttp($value)) {
    $newArr[$key] = $value;
   }
  }
  return $newArr;
 }
 
 /**
* Format QQ
* @param string $str
* @return string
​*/
 private function formatQqMsn($str, $e='QQ') {
  if (empty($str)) return '';
  preg_match_all('/alt="'.$e.':(.+)"/Usi', $str, $arr);
  if (count($arr[1])==1) return $arr[1][0];
  $newStr = null;
  foreach ($arr[1] as $value) {
   $newStr .= $value . ' ';
  }
  return rtrim($newStr, ' ');
 }

 /**
* Remove the A tag from the URL
* @param string $site
* @return string
​*/
 private function stripATags($site) {
  $site = preg_replace('/(.+)/', '1', $site);
  return $site;
 }

 /**
* Check if the url has http
* @param string $url
* @return bool
​*/
 private function isExistsHttp($url) {
  if (stristr($url, 'http://')) {
   return true;
  } else {
   return false;
  }
 }
 
 /**
* Get page content
* @param Number $page
* @return string
​*/
 private function getContent($page=1) {
  $re = file_get_contents($this->getUrl($this->key, $page));
  return $re;
 }
 
 /**
* Get page number
* @return Number
​*/
 private function getPageNum() {
  $i = null;
  $re = $this->getContent();
  preg_match_all('/共(.+)页/Usi', $re, $arr);
  $i = $arr[1][0];
  return $i;
 }

 /**
* Get URL link
* @param string $str
* @param int $page page number
* @return string
​*/
 private function getUrl($str, $page=1) {
  return "http://www.ic.net.cn/partsearch/searchinstock.asp?newtype=1&area=&Page={$page}&partnumber={$str}&mfg=&DateCode=&QTY=&PRICE=&Exact=&orderby=inputdate&qty_filter=50&usertype2=1&pack=";
 }

 /**
* Get page content
* @param string $url
* @return string
​*/
 private function getUrlInfo($url) {
  $re = file_get_contents($url);
  return $re;
 }
}

/*
程序运行思路：根据“IC 交易网”的IC搜索功能，输入型号进行搜索，然后抓取供应商信息

数据库结构
CREATE TABLE `ic` (
 `id` mediumint(8) unsigned NOT NULL auto_increment,
 `company` varchar(500) NOT NULL,
 `address` varchar(500) default NULL,
 `phone` varchar(500) default NULL,
 `mobile` varchar(500) default NULL,
 `fax` varchar(300) default NULL,
 `zip` varchar(300) default NULL,
 `person` varchar(500) default NULL,
 `qq` varchar(300) default NULL,
 `msn` varchar(300) default NULL,
 `email` varchar(500) default NULL,
 `website` varchar(300) default NULL,
 `regDate` varchar(500) default NULL,
 PRIMARY KEY  (`id`)
) ENGINE=InnoDB DEFAULT CHARSET=gb2312
*/

$i = new ic();
$arr = array_unique(array('MAX3232', 'AML8613', 'MT6225A', 'OM8373PS/N3/A', 'PT7313', 'MAX8212ESA', 'TL431', 'S3C2440', 'TMS320F2812PGFA', 'PCM1704', 'AN6717', 'CA3162E', 'CA3161E', 'LM393N', 'DS18B20', 'SHT10', 'AML8613', 'AN6717', 'LM393N', 'CA3161E', 'CA3162E', 'PCM1704', 'STK392-040', 'K1667', 'MAX232', 'STM32F103', 'LM358'));
foreach ($arr as $v) {
 $i->go($v);
}
?>

摘自 Lee.的专栏

http://www.bkjia.com/PHPjc/478401.htmlwww.bkjia.comtruehttp://www.bkjia.com/PHPjc/478401.htmlTechArticle?php /** * 抓取IC 交易网供应商主程序 * author Lee. * Last modify $Date: 2012-2-6 10:44:32$ * 注：本程序按照编码 GB2312 执行，因为IC 交易网网站是GB231...