文件批量转换器（从ANSI 到 UTF-8）（PHP）
其中几个函数很好用，我花了一天时间直接创建多层目录mymkdir($dirname)，
批处理目录下的文件function listDirTree( $dirName = null ,$callFunction) 可自定义处理函数。
如：
$callFunc="backupFile(\"$backupPath\",\$file,\$filePath,\$dirName);";
listDirTree($operateDir,$callFunc);
将文本中 x=y
u=a
提取为数组的函数
function myExplode($originalStr,$firstTag,$secondTag)
将文件转换为utf-8编码的函数。function gb2utf8($file,$filePath)
判断文件是否为utf-8编码的函数 function utf8_probability(&$rawtextstr)
<?php
//header('Content-Type: text/html; charset=utf-8');
$operateDir="d:/usr/www/html/mambog"; //需要备份的目录
$backupPath="d:/usr/www/html/bkup_mambog/"; //需要备份到的目录
echo "<PRE>";
$currentname="fileconv.php";
$currentpath=realpath($currentname);
$pos = strrpos ($currentpath,"\\" );
$currentpath=substr($currentpath,0,$pos);
$currentpath=str_replace("\\","/",$currentpath);
echo "当前执行文件所在目录:$currentpath <BR>";
$tm1=microtime();
$extname=".php;.xml;html;.htm;.css;.txt;.js;.ini;";
$utf8File=0;
$convertedFile=0;
$notTextFile=0;
$operateDirLen=strlen($operateDir);
mainsub(&$operateDir,&$backupPath);
function mainsub(&$operateDir,&$backupPath)
{
if(strpos($backupPath,$operateDir)===false)
{
if(!is_dir($backupPath))mymkdir($backupPath);
backup(&$operateDir,$backupPath);
$callFunc="gb2utf8(\$file,\$filePath);";
listDirTree($operateDir,$callFunc);
}
else
{
echo "Fetal Error You must change your backupdir ,<BR>It can't be under the directory you want to backup";
die;
}
}
function backup($operateDir,$backupPath)
{
global $currentpath;
if(readtag($backupPath)==0)//如果没有备份过,那么,执行备份
{
$callFunc="backupFile(\"$backupPath\",\$file,\$filePath,\$dirName);";
listDirTree($operateDir,$callFunc);
setTag($backupPath); //设置备份标记
}
else
{//如果备份过了，那么，exit;
echo "backuped already";die;
}
}//function backup end
function backupFile($backupPath,$file,$filePath,$dirName)
{ global $operateDirLen;
$subdir = substr($dirName."/",$operateDirLen+1);
mymkdir($backupPath.$subdir); //创建子目录
$bkfile=$backupPath.$subdir.$file;
//echo "备份文件为: $bkfile 原始文件为：$filePath<BR>";
$fstr=file_get_contents($filePath);
$fhandle=fopen($bkfile,"wb");
fwrite($fhandle,$fstr);
fclose($fhandle);
}
echo (microtime()-$tm1)."秒<BR>";
echo "<BR>utf8文件：".$utf8File;
echo "<BR>被转换的文件总数:".$convertedFile;
echo "<BR>非文本文件数目:".$notTextFile;
function readtag($backupDir)
{
if(!is_dir($backupDir)){
mymkdir($backupDir);
}
if(!is_file($backupDir."tag.ini"))//如果tag.ini不存在，则建立并初始化
{ if(!$fp=fopen($backupDir."tag.ini","w")){
echo "error while reading or creating the tag.ini ";
}
else
{
$str="backuped=0";
if(!fwrite($fp,$str)){echo "error while init the tag.ini ";}
fclose($fp);
}
gb2utf8("tag.ini",$backupDir."tag.ini"); //将新建的tag.ini转为utf-8编码
return 0;
}
else//如果tag.ini已经存在，读取其中的配置
{
$backupTag=file_get_contents($backupDir."tag.ini");
$tagTree=myExplode($backupTag,"/\n/","=");
//print_r($tagTree); //tagtree存储所有tag.ini里的 x=y 的信息
return $tagTree['backuped'];
}
}
function setTag($backupDir)
{
$tagini=file_get_contents($backupDir."tag.ini");
if(!$fp=fopen($backupDir."tag.ini","w")){
echo "error while reading or creating the tag.ini ";
}
else
{
$str="backuped=1";
$pattern="/backuped=0/";
if(!strpos($tagini,"backuped=",0))
{
$tagini=$str;
}
else
{
$tagini=preg_replace($pattern,$str,$tagini);
}
if(!fwrite($fp,$tagini))
{
echo "error while set the tag.ini ";
}
fclose($fp);
}
}
function myExplode($originalStr,$firstTag,$secondTag)
{
$firstArray=preg_split($firstTag,$originalStr);
foreach($firstArray as $first)
{
$secondArray=explode($secondTag,$first);
$tagTree[$secondArray]=$secondArray;
}
return $tagTree;
}
function mymkdir($mainDir)
{
if(!is_dir($mainDir))
{
global $validFatherDir;
getFatherDir($mainDir);
$fullLen=strlen($mainDir); //控制循环
//echo "需要创建的目录".$mainDir."长度为 $fullLen <BR>";
//echo "最高有效的父目录".$validFatherDir."<BR>";
$validDirLen=strlen($validFatherDir);
while(($i<10 )&&($fullLen>$validDirLen+1))
{
//echo "父目录长度：$validDirLen<BR>";
$pos=strpos($mainDir,"/",$validDirLen+1);
//echo "/所在位置：$pos<BR>";
$dirname=substr($mainDir,$validDirLen,$pos-$validDirLen);
//echo "正在创建目录".$validFatherDir.$dirname."<BR>";
$validFatherDir.=$dirname;
if(!mkdir($validFatherDir))
{
//echo "Fetal Error accoured while create directory";die;
}
$validDirLen=strlen($validFatherDir);
$i++;
}
}
}//mymkdir function end
function getFatherDir($mainDir)
{ global $validFatherDir;
$pos = strrpos($mainDir,"/");
if ($pos===false)
{
// not found
echo "Fetal Error While Get Valid Father Dir";die;
return false;
}
else
{
$fatherDir=substr($mainDir,0,$pos);
if(!is_dir($fatherDir))
{ echo "<font color=red >$fatherDir is not exists<BR></font>";
getFatherDir($fatherDir);
}
else
{
$validFatherDir= $fatherDir;
//mkdir($mainDir);echo "$mainDir is created<BR>";
}
}
}
/** 函数 listDirTree( $dirName = null )
* 功能 列出目录下所有文件及子目录
* 参数 $dirName 目录名称
* 返回 目录结构数组 false为失败
*/
function listDirTree( $dirName = null ,$callFunction)
{//global $tree;
if( empty( $dirName ) )
exit( "IBFileSystem: directory is empty." );
if( is_dir( $dirName ) )
{
if( $dh = opendir( $dirName ) )
{
//$tree = array();
while( ( $file = readdir( $dh ) ) !== false )
{
if( $file != "." && $file != ".." )
{
$filePath = $dirName . "/" . $file;
if( is_dir( $filePath ) )//为目录,递归
{
//$tree[$file] = listDirTree( $filePath );
listDirTree( $filePath ,$callFunction);
}
else//为文件,进行处理
{ //echo "文件处理函数为 $callFunction <BR>";
eval($callFunction);
//gb2utf8($file,$filePath);
//$tree[] = $file;
} //文件处理结束
}
}
closedir( $dh );
}
else
{
exit( "IBFileSystem: can not open directory $dirName.");
}
//返回当前的$tree
//return $tree;
}
else
{
exit( "IBFileSystem: $dirName is not a directory.");
}
}
function test($arga,$argb)
{
echo $arga."<BR>".$argb."<BR>";
}
function gb2utf8($file,$filePath)
{ global $extname,$utf8File,$convertedFile,$notTextFile;
$fileext=substr($file,-4,4);
preg_match("/".$fileext."/i",$extname, $matches);
if ($matches) //是文本文件扩展名为:.php,.xml,.css,.js.由数组$matches定义
{
$outfilename=$filePath;
$fstr=file_get_contents($outfilename);
$utf8score=utf8_probability(&$fstr);
if( 90<=$utf8score && $utf8score<=100)
{ $utf8File++;
//echo "<font color=red> $filePathis encoded with utf-8 already </font><BR>";//die;
}
else
{
$fp=fopen($outfilename,'wb');
$foutstr=iconv("GB2312","utf-8",$fstr); //如果碰到 此文件中不存在双字节字符，则这个函数并不起作用
//$foutstr.="\n <!--脚注-->";//这一句确保不存在双字节字符的文件也能得到转换,js文件经测试也可以。
//<!-- -->形式的注释可以使用于php,css,html,css等文件中
//本脚本（fileconv.php已经是utf-8编码保存的，所以加到转换后的字串里）
//然而，这样不行，搞的配套的代码都不能运行了header语句大多就失效了。
//echo $filePath."<BR>";
fwrite($fp,$foutstr);
$convertedFile++;
//echo $filePath . "<<<<<<<<< $matches<BR>";
fclose($fp);
}
}//是文本文件处理结束
else{
$notTextFile++;
//echo $file."=========<BR>";
}
}
function showdir($dir){
while ($file_name = readdir($dir)) {
if (($file_name != ".") && ($file_name != "..")) {
$file_list .= "<li>$file_name";
if(is_dir($file_name))
showdir($file_name);
}
}
}
function utf8_probability(&$rawtextstr) {
$score = 0;
$i = 0;
$rawtextlen = 0;
$goodbytes = 0;
$asciibytes = 0;
$rawtextarray = preg_split("//",$rawtextstr,-1, PREG_SPLIT_NO_EMPTY); //转换成char数组，如果是php5，则可使用str_split
$rawtext = array();
//var_dump($rawtextarray);die;
for($i=0;$i<count($rawtextarray);$i++)
$rawtext[] = ord($rawtextarray[$i]); //ord(char)
// Maybe also use UTF8 Byte Order Mark(BOM): EF BB BF
//BOM，某些utf8文件流的首3个字节，可以表示这个文件的编码方式
// Check to see if characters fit into acceptable ranges
//print_r($rawtext);
$rawtextlen = strlen($rawtextstr);
for ($i = 0; $i < $rawtextlen; $i++) {
if ($rawtext[$i] < 0x80) { // One byte
$asciibytes++; // Ignore ASCII, can throw off count
} else if (0xC0 <= $rawtext[$i] && $rawtext[$i] <= 0xDF && // Two bytes
$i+1 < $rawtextlen && 0x80 <= $rawtext[$i+1] && $rawtext[$i+1] <= 0xBF) {
$goodbytes += 2; $i++;
} else if (0xE0 <= $rawtext[$i] && $rawtext[$i] <= 0xEF && // Three bytes
$i+2 < $rawtextlen && 0x80 <= $rawtext[$i+1] && $rawtext[$i+1] <= 0xBF &&
0x80 <= $rawtext[$i+2] && $rawtext[$i+2] <= 0xBF) {
$goodbytes += 3; $i+=2;
}
}
//ascii is sub of utf8
if ($asciibytes == $rawtextlen) { return 0; }
$score = (int)(100 * ($goodbytes/($rawtextlen-$asciibytes)));
// If not above 98, reduce to zero to prevent coincidental matches
if ($score > 98) {
return $score;
} else if ($score > 95 && $goodbytes > 30) {
// Allows for some (few) bad formed sequences
return $score;
} else {
return 0;
}
}
?>
页:
[1]