php 采集常用函数_PHP采集程序中常用的函数

//获得当前的脚本网址

function get_php_url(){

if(!empty($_SERVER["REQUEST_URI"])){

$scriptName = $_SERVER["REQUEST_URI"];

$nowurl = $scriptName;

}else{

$scriptName = $_SERVER["PHP_SELF"];

if(empty($_SERVER["QUERY_STRING"])) $nowurl = $scriptName;

else $nowurl = $scriptName."?".$_SERVER["QUERY_STRING"];

}

return $nowurl;

}

//把全角数字转为半角数字

function GetAlabNum($fnum){

$nums = array("０","１","２","３","４","５","６","７","８","９");

$fnums = "0123456789";

for($i=0;$i<=9;$i++) $fnum = str_replace($nums[$i],$fnums[$i],$fnum);

$fnum = ereg_replace("[^0-9\.]|^0{1,}","",$fnum);

if($fnum=="") $fnum=0;

return $fnum;

}

//去除HTML标记

function Text2Html($txt){

$txt = str_replace("  ","　",$txt);

$txt = str_replace("

$txt = str_replace(">",">",$txt);

$txt = preg_replace("/[\r\n]{1,}/isU","
\r\n",$txt);

return $txt;

}

//清除HTML标记

function ClearHtml($str){

$str = str_replace('

$str = str_replace('>','>',$str);

return $str;

}

//相对路径转化成绝对路径

function relative_to_absolute($content, $feed_url) {

preg_match('/(http|https|ftp):\/\//', $feed_url, $protocol);

$server_url = preg_replace("/(http|https|ftp|news):\/\//", "", $feed_url);

$server_url = preg_replace("/\/.*/", "", $server_url);

if ($server_url == '') {

return $content;

}

if (isset($protocol[0])) {

$new_content = preg_replace('/href="\//', 'href="'.$protocol[0].$server_url.'/', $content);

$new_content = preg_replace('/src="\//', 'src="'.$protocol[0].$server_url.'/', $new_content);

} else {

$new_content = $content;

}

return $new_content;

}

//取得所有链接

function get_all_url($code){

preg_match_all('/"\' ]+)["|\']?\s*[^>]*>([^>]+)/i',$code,$arr);

return array('name'=>$arr[2],'url'=>$arr[1]);

}

//获取指定标记中的内容

function get_tag_data($str, $start, $end){

if ( $start == '' || $end == '' ){

return;

}

$str = explode($start, $str);

$str = explode($end, $str[1]);

return $str[0];

}

//HTML表格的每行转为CSV格式数组

function get_tr_array($table) {

$table = preg_replace("'

]*?>'si",'"',$table);

$table = str_replace("

",'",',$table);(zehoa)