将RTF格式的文件转成HTML并在网页中显示的代码
程序员文章站
2022-06-15 08:16:47
将rtf格式的文件转成html并在网页中显示的代码 它是这样工作的，将一个rtf文件上传，然后转成html显示出来，代码有点复杂,teaman还要好好研究,好象中文版有点问...
将rtf格式的文件转成html并在网页中显示的代码
它是这样工作的，将一个rtf文件上传，然后转成html显示出来，代码有点复杂,teaman还要好好研究,好象中文版有点问题。
<html>
<body>
<?
if(!($userfile)) {
?>
<form enctype="multipart/form-data" action="<?print($php_self);?>" method=post>
<input type=hidden name="max_file_size" value=2000>
send this file: <input name=userfile type=file>
<input type=submit value=upload>
</form>
</body>
</html>
<?
exit;
}
function processtags($tags, $line) {
$html = "";
global $color;
global $size;
global $bullets;
// remove spaces.
$tags = trim($tags);
// found the beginning of the bulleted l
// ist.
if(ereg("\\\pnindent", $tags)) {
$html .= "<ul><li>";
$bullets += $line;
$tags = ereg_replace("\\\par", "", $tags);
$tags = ereg_replace("\\\(tab)", "", $tags);
}
if($line - $bullets == 0) {
$tags = ereg_replace("\\\par", "", $tags);
}
elseif($line - $bullets == 1) {
if(ereg("\\\pntext", $tags)) {
$html .= "<li>";
$tags = ereg_replace("\\\par", "", $tags);
$tags = ereg_replace("\\\(tab)", "", $tags);
$bullets++;
}
else {
$html .= "</ul>";
$bullets = 0;
}
}
// convert bold.
if(ereg("\\\b0", $tags)){ $html .= "</b>"; }
elseif(ereg("\\\b", $tags)) { $html .= "<b>"; }
// convert italic.
if(ereg("\\\i0", $tags)){ $html .= "</i>"; }
elseif(ereg("\\\i", $tags)) { $html .= "<i>"; }
// convert underline.
if(ereg("\\\ulnone", $tags)){ $html .= "</u>"; }
elseif(ereg("\\\ul", $tags)){ $html .= "<u>"; }
// convert alignments.
if(ereg("\\\pard\\\qc", $tags)) { $html .= "<div align=center>"; }
elseif(ereg("\\\pard\\\qr", $tags)) { $html .= "<div align=right>"; }
elseif(ereg("\\\pard", $tags)){ $html .= "<div align=left>"; }
// remove \pard from the tags so it does
// n't get confused with \par.
$tags = ereg_replace("\\\pard", "", $tags);
// convert line breaks.
if(ereg("\\\par", $tags)){ $html .= "<br>"; }
// use the color table to capture the fo
// nt color changes.
if(ereg("\\\cf[0-9]", $tags)) {
global $fcolor;
$numcolors = count($fcolor);
for($i = 0; $i < $numcolors; $i++) {
$test = "\\\cf" . ($i + 1);
if(ereg($test, $tags)) {
$color = $fcolor[$i];
}
}
}
// capture font size changes.
if(ereg("\\\fs[0-9][0-9]", $tags, $temp)) {
$size = ereg_replace("\\\fs", "", $temp[0]);
$size /= 2;
if($size <= 10) { $size = 1; }
elseif($size <= 12) { $size = 2; }
elseif($size <= 14) { $size = 3; }
elseif($size <= 16) { $size = 4; }
elseif($size <= 18) { $size = 5; }
elseif($size <= 20) { $size = 6; }
elseif($size <= 22) { $size = 7; }
else{ $size = 8; }
}
// if there was a font color or size cha
// nge, change the font tag now.
if(ereg("(\\\cf[0-9])||(\\\fs[0-9][0-9])", $tags)) {
$html .= "</font><font size=$size color=$color>";
}
// replace \tab with alternating spaces
// and nonbreakingwhitespaces.
if(ereg("\\\(tab)", $tags)) { $html .= " "; }
return $html;
}
function processword($word) {
// replace \\ with \
$word = ereg_replace("[\\]{2,}", "\\", $word);
// replace \{ with {
$word = ereg_replace("[\\][\{]", "\{", $word);
// replace \} with }
$word = ereg_replace("[\\][\}]", "\}", $word);
// replace 2 spaces with one space.
$word = ereg_replace(" ", " ", $word);
return $word;
}
$color = "000000";
$size = 1;
$bullets = 0;
// read the uploaded file into an array.
//
$rtfile = file($userfile);
$filelength = count($rtfile);
// loop through the rest of the array
for($i = 1; $i < $filelength; $i++) {
/*
** if the line contains "\colortbl" then we found the color table.
** we'll have to split it up into each individual red, green, and blue
** convert it to hex and then put the red, green, and blue back together.
** then store each into an array called fcolor.
*/
if(ereg("^\{\\\colortbl", $rtfile[$i])) {
// split the line by the backslash.
$colors = explode("\\", $rtfile[$i]);
$numofcolors = count($colors);
for($k = 2; $k < $numofcolors; $k++) {
// find out how many different colors th
// ere are.
if(ereg("[0-9]+", $colors[$k], $matches)) {
$match[] = $matches[0];
}
}
// for each color, convert it to hex.
$numofcolors = count($match);
for($k = 0; $k < $numofcolors; $k += 3) {
$red = dechex($match[$k]);
$red = $match[$k] < 16 ? "0$red" : $red;
$green = dechex($match[$k + 1]);
$green = $match[$k +1] < 16 ? "0$green" : $green;
$blue = dechex($match[$k + 2]);
$blue = $match[$k + 2] < 16 ? "0$blue" : $blue;
$fcolor[] = "$red$green$blue";
}
$numofcolors = count($fcolor);
}
// or else, we parse the line, pulling o
// ff words and tags.
else {
$token = "";
$start = 0;
$linelength = strlen($rtfile[$i]);
for($k = 0; $k < $linelength; $k++) {
if($rtfile[$i][$start] == "\\" && $rtfile[$i][$start + 1] != "\\") {
// we are now dealing with a tag.
$token .= $rtfile[$i][$k];
if($rtfile[$i][$k] == " ") {
$newfile[$i] .= processtags($token, $i);
$token = "";
$start = $k + 1;
}
elseif($rtfile[$i][$k] == "\n") {
$newfile[$i] .= processtags($token, $i);
$token = "";
}
}
elseif($rtfile[$i][$start] == "{") {
// we are now dealing with a tag.
$token .= $rtfile[$i][$k];
if($rtfile[$i][$k] == "}") {
$newfile[$i] .= processtags($token, $i);
$token = "";
$start = $k + 1;
}
}
else {
// we are now dealing with a word.
if($rtfile[$i][$k] == "\\" && $rtfile[$i][$k + 1] != "\\" && $rtfile[$i][$k - 1] != "\\") {
$newfile[$i] .= processword($token);
$token = $rtfile[$i][$k];
$start = $k;
}
else {
$token .= $rtfile[$i][$k];
}
}
}
}
}
$limit = sizeof($newfile);
for($i = 0; $i < $limit; $i++) {
print("$newfile[$i]\n");
}
?>
</body>
</html>
它是这样工作的，将一个rtf文件上传，然后转成html显示出来，代码有点复杂,teaman还要好好研究,好象中文版有点问题。
<html>
<body>
<?
if(!($userfile)) {
?>
<form enctype="multipart/form-data" action="<?print($php_self);?>" method=post>
<input type=hidden name="max_file_size" value=2000>
send this file: <input name=userfile type=file>
<input type=submit value=upload>
</form>
</body>
</html>
<?
exit;
}
function processtags($tags, $line) {
$html = "";
global $color;
global $size;
global $bullets;
// remove spaces.
$tags = trim($tags);
// found the beginning of the bulleted l
// ist.
if(ereg("\\\pnindent", $tags)) {
$html .= "<ul><li>";
$bullets += $line;
$tags = ereg_replace("\\\par", "", $tags);
$tags = ereg_replace("\\\(tab)", "", $tags);
}
if($line - $bullets == 0) {
$tags = ereg_replace("\\\par", "", $tags);
}
elseif($line - $bullets == 1) {
if(ereg("\\\pntext", $tags)) {
$html .= "<li>";
$tags = ereg_replace("\\\par", "", $tags);
$tags = ereg_replace("\\\(tab)", "", $tags);
$bullets++;
}
else {
$html .= "</ul>";
$bullets = 0;
}
}
// convert bold.
if(ereg("\\\b0", $tags)){ $html .= "</b>"; }
elseif(ereg("\\\b", $tags)) { $html .= "<b>"; }
// convert italic.
if(ereg("\\\i0", $tags)){ $html .= "</i>"; }
elseif(ereg("\\\i", $tags)) { $html .= "<i>"; }
// convert underline.
if(ereg("\\\ulnone", $tags)){ $html .= "</u>"; }
elseif(ereg("\\\ul", $tags)){ $html .= "<u>"; }
// convert alignments.
if(ereg("\\\pard\\\qc", $tags)) { $html .= "<div align=center>"; }
elseif(ereg("\\\pard\\\qr", $tags)) { $html .= "<div align=right>"; }
elseif(ereg("\\\pard", $tags)){ $html .= "<div align=left>"; }
// remove \pard from the tags so it does
// n't get confused with \par.
$tags = ereg_replace("\\\pard", "", $tags);
// convert line breaks.
if(ereg("\\\par", $tags)){ $html .= "<br>"; }
// use the color table to capture the fo
// nt color changes.
if(ereg("\\\cf[0-9]", $tags)) {
global $fcolor;
$numcolors = count($fcolor);
for($i = 0; $i < $numcolors; $i++) {
$test = "\\\cf" . ($i + 1);
if(ereg($test, $tags)) {
$color = $fcolor[$i];
}
}
}
// capture font size changes.
if(ereg("\\\fs[0-9][0-9]", $tags, $temp)) {
$size = ereg_replace("\\\fs", "", $temp[0]);
$size /= 2;
if($size <= 10) { $size = 1; }
elseif($size <= 12) { $size = 2; }
elseif($size <= 14) { $size = 3; }
elseif($size <= 16) { $size = 4; }
elseif($size <= 18) { $size = 5; }
elseif($size <= 20) { $size = 6; }
elseif($size <= 22) { $size = 7; }
else{ $size = 8; }
}
// if there was a font color or size cha
// nge, change the font tag now.
if(ereg("(\\\cf[0-9])||(\\\fs[0-9][0-9])", $tags)) {
$html .= "</font><font size=$size color=$color>";
}
// replace \tab with alternating spaces
// and nonbreakingwhitespaces.
if(ereg("\\\(tab)", $tags)) { $html .= " "; }
return $html;
}
function processword($word) {
// replace \\ with \
$word = ereg_replace("[\\]{2,}", "\\", $word);
// replace \{ with {
$word = ereg_replace("[\\][\{]", "\{", $word);
// replace \} with }
$word = ereg_replace("[\\][\}]", "\}", $word);
// replace 2 spaces with one space.
$word = ereg_replace(" ", " ", $word);
return $word;
}
$color = "000000";
$size = 1;
$bullets = 0;
// read the uploaded file into an array.
//
$rtfile = file($userfile);
$filelength = count($rtfile);
// loop through the rest of the array
for($i = 1; $i < $filelength; $i++) {
/*
** if the line contains "\colortbl" then we found the color table.
** we'll have to split it up into each individual red, green, and blue
** convert it to hex and then put the red, green, and blue back together.
** then store each into an array called fcolor.
*/
if(ereg("^\{\\\colortbl", $rtfile[$i])) {
// split the line by the backslash.
$colors = explode("\\", $rtfile[$i]);
$numofcolors = count($colors);
for($k = 2; $k < $numofcolors; $k++) {
// find out how many different colors th
// ere are.
if(ereg("[0-9]+", $colors[$k], $matches)) {
$match[] = $matches[0];
}
}
// for each color, convert it to hex.
$numofcolors = count($match);
for($k = 0; $k < $numofcolors; $k += 3) {
$red = dechex($match[$k]);
$red = $match[$k] < 16 ? "0$red" : $red;
$green = dechex($match[$k + 1]);
$green = $match[$k +1] < 16 ? "0$green" : $green;
$blue = dechex($match[$k + 2]);
$blue = $match[$k + 2] < 16 ? "0$blue" : $blue;
$fcolor[] = "$red$green$blue";
}
$numofcolors = count($fcolor);
}
// or else, we parse the line, pulling o
// ff words and tags.
else {
$token = "";
$start = 0;
$linelength = strlen($rtfile[$i]);
for($k = 0; $k < $linelength; $k++) {
if($rtfile[$i][$start] == "\\" && $rtfile[$i][$start + 1] != "\\") {
// we are now dealing with a tag.
$token .= $rtfile[$i][$k];
if($rtfile[$i][$k] == " ") {
$newfile[$i] .= processtags($token, $i);
$token = "";
$start = $k + 1;
}
elseif($rtfile[$i][$k] == "\n") {
$newfile[$i] .= processtags($token, $i);
$token = "";
}
}
elseif($rtfile[$i][$start] == "{") {
// we are now dealing with a tag.
$token .= $rtfile[$i][$k];
if($rtfile[$i][$k] == "}") {
$newfile[$i] .= processtags($token, $i);
$token = "";
$start = $k + 1;
}
}
else {
// we are now dealing with a word.
if($rtfile[$i][$k] == "\\" && $rtfile[$i][$k + 1] != "\\" && $rtfile[$i][$k - 1] != "\\") {
$newfile[$i] .= processword($token);
$token = $rtfile[$i][$k];
$start = $k;
}
else {
$token .= $rtfile[$i][$k];
}
}
}
}
}
$limit = sizeof($newfile);
for($i = 0; $i < $limit; $i++) {
print("$newfile[$i]\n");
}
?>
</body>
</html>