python解析mht文件_php解析mht文件转换成html

php解析mht文件，使用编辑器打开可以看到base64编码所以，mht是可以转换成html的。

/**

* 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 $range['end']; //tulem: 30 (' ') -- l5pumärgist järgnev out: array('range'=>$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

?>