php解析mht文件转换成html的实例

<?php

/**
 * 针对mht格式的文件进行解析
* 使用例子:
* 
* function mhtmlparsebody($filename) {

	if (file_exists ( $filename )) {
		if (is_dir ( $filename )) return false;
		
		$filename = strtolower ( $filename );
		if (strpos ( $filename, '.mht', 1 ) == false) return false;
			
		
		$o_mhtml = new mhtml ();
		$o_mhtml->set_file ( $filename );
		$o_mhtml->extract ();
		return $o_mhtml->get_part_to_file(0);

	}
	return null;
}

function mhtmlparseall($filename) {

	if (file_exists ( $filename )) {
		if (is_dir ( $filename )) return false;

		$filename = strtolower ( $filename );
		if (strpos ( $filename, '.mht', 1 ) == false) return false;
			

		$o_mhtml = new mhtml ();
		$o_mhtml->set_file ( $filename );
		$o_mhtml->extract ();
		return $o_mhtml->get_all_part_file();

	}
	return null;
}
*/

class mhtparse {

	var $file = '';
	var $boundary = '';
	var $filedata = '';
	var $countparts = 1;
	var $log = '';
	
	function extract() {
		$this->read_filedata ();
		$this->file_parts ();

		return 1;
	}
	
	function set_file($p) {
		$this->file = $p;
	}
	
	function get_log() {
		return $this->log;
	}
	
	function file_parts() {
		$lines = explode ( "\n", substr ( $this->filedata, 0, 8192 ) );
		foreach ( $lines as $line ) {
			$line = trim ( $line );
			if (strpos ( $line, '=' ) !== false) {
				if (strpos ( $line, 'boundary', 0 ) !== false) {
					$range = $this->getrange ( $line, '"', '"', 0 );
					$this->boundary = "--" . $range ['range'];
					$this->filedata = str_replace ( $line, '', $this->filedata );
					break;
				}
			}
		}
		if ($this->boundary != '') {
			$this->filedata = explode ( $this->boundary, $this->filedata );
			unset ( $this->filedata [0] );
			$this->filedata = array_values ( $this->filedata );
			$this->countparts = count ( $this->filedata );
		} else {
			$tmp = $this->filedata;
			$this->filedata = array (
					$tmp 
			);
		}
	}
	
	function get_all_part_file() {
		return $this->filedata;
	}
	
	function get_part_to_file($i) {
		$line_data_start = 0;
		$encoding = '';
		$part_lines = explode ( "\n", ltrim ( $this->filedata [$i] ) );
		foreach ( $part_lines as $line_id => $line ) {
			$line = trim ( $line );
			if ($line == '') {
				if (trim ( $part_lines [0] ) == '--')
					return 1;
				$line_data_start = $line_id;
				break;
			}
			if (strpos ( $line, ':' ) !== false) {
				$pos = strpos ( $line, ':' );
				$k = strtolower ( trim ( substr ( $line, 0, $pos ) ) );
				$v = trim ( substr ( $line, $pos + 1, strlen ( $line ) ) );
				if ($k == 'content-transfer-encoding') {
					$encoding = $v;
				}
				if ($k == 'content-location') {
					$location = $v;
				}
				if ($k == 'content-type') {
					$contenttype = $v;
				}
			}
		}
		
		foreach ( $part_lines as $line_id => $line ) {
			if ($line_id <= $line_data_start)
				$part_lines [$line_id] = '';
		}
		
		$part_lines = implode ( '', $part_lines );
		if ($encoding == 'base64')
			$part_lines = base64_decode ( $part_lines );
		elseif ($encoding == 'quoted-printable')
			$part_lines = imap_qprint ( $part_lines );
		
		return $part_lines;
	}
	
	function read_filedata() {
		$handle = fopen ( $this->file, 'r' );
		$this->filedata = fread ( $handle, filesize ( $this->file ) );
		fclose ( $handle );
	}
	
	function getrange(&$subject, $beginmark_str = '{', $endmark_str = '}', $start_pos = 0) {
		/*
		 * $str="sssss { x { xx } {xx{xx } x} x} sssss"; $range=string::getrange($str,'{','}',0); echo $range['range']; //tulem: " x { xx } {xx{xx } x} x" echo $range['behin']; //tulem: 6 echo $range['end']; //tulem: 30 (' ') -- l5pumã¤rgist jã¤rgnev out: array('range'=>$range,'begin'=>$begin_firstoccurence_pos,'end'=>$end_sequel_pos) | false v1.1 2004-2006,uku-kaarel j5esaar,ukjoesaar@hot.ee,http://www.hot.ee/ukjoesaar,+3725110693
		 */
		if (empty ( $beginmark_str ))
			$beginmark_str = '{';
		$beginmark_str_len = strlen ( $beginmark_str );
		
		if (empty ( $endmark_str ))
			$endmark_str = '}';
		$endmark_str_len = strlen ( $endmark_str );
		
		/* $start_pos_cache = 0; */
		do {
			/* !algus */
			if (! is_int ( $begin_firstoccurence_pos ))
				$start_pos_cache = $start_pos;
				
				/* ?algus-test */
			$start_pos_cache = @strpos ( $subject, $beginmark_str, $start_pos_cache );
			
			/* this is possible start for range */
			if (is_int ( $start_pos_cache )) {
				/* skip */
				$start_pos_cache = ($start_pos_cache + $beginmark_str_len);
				/* test possible range start pos */
				if (is_int ( $begin_firstoccurence_pos )) {
					if ($start_pos_cache < $range_end_pos)
						$rangeclean = 0;
					elseif ($start_pos_cache > $range_end_pos)
						$rangeclean = 1;
				}
				/* here it is */
				if (! is_int ( $begin_firstoccurence_pos ))
					$begin_firstoccurence_pos = $start_pos_cache;
			} /* viga nr 0 algust ei ole */
			
			if (! is_int ( $start_pos_cache )) {
				/* !algus */
	/* viga nr 1 algusmarki ei leitud : viimane voimalik algus */
	if (is_int ( $begin_firstoccurence_pos ) and ($start_pos_cache < $range_end_pos))
					$rangeclean = 1;
				else
					return false;
			}
			if (is_int ( $begin_firstoccurence_pos ) and ($rangeclean != 1)) {
				if (! is_int ( $end_pos_cache ))
					$end_sequel_pos = $begin_firstoccurence_pos;
				
				$end_pos_cache = strpos ( $subject, $endmark_str, $end_sequel_pos );
				
				/* ok */
				if (is_int ( $end_pos_cache ) and ($rangeclean != 1)) {
					$range_current_lenght = ($end_pos_cache - $begin_firstoccurence_pos);
					$end_sequel_pos = ($end_pos_cache + $endmark_str_len);
					$range_end_pos = $end_pos_cache;
				}
				/* viga nr 2 loppu ei leitud */
				if (! is_int ( $end_pos_cache ))
					if ($end_pos_cache == false)
						return false;
			}
		} while ( $rangeclean < 1 );
		
		if (is_int ( $begin_firstoccurence_pos ) and is_int ( $range_current_lenght ))
			$range = substr ( $subject, $begin_firstoccurence_pos, $range_current_lenght );
		else
			return false;
		
		return array (
				'range' => $range,
				'begin' => $begin_firstoccurence_pos,
				'end' => $end_sequel_pos 
		);
	} // end getrange()
} // class
?>