Fia
Member |
# Posted: 12 Mar 2007 14:39
Hi!
i'm trying to clean up all the shit that ms word does to you code and I found this
function cleanUp(){ //clean up crud inserted by Micro$oft Orifice
oW.document.execCommand("removeformat",false,null) ;
var h=oW.document.body.innerHTML;
var vicious=false;
if (gentleClean=="ask") vicious=confirm(t('Remove all styles and classes?'));
if (!gentleClean || vicious){
h=h.replace(/<\/?(SPAN|DEL|INS|U|DIR)[^>]*>/gi, "")
.replace(/\b(CLASS|STYLE)=\"[^\"]*\"/gi, "")
.replace(/\b(CLASS|STYLE)=\w+/gi, "");
} //.replace(/(<FONT([^>])*>/gi, "$1SPAN style='$2'")
h=h.replace(/<\/?(\?XML|ST1|SHAPE|V:|O:|F:|F |PATH|LOCK|IMAGEDATA|STROKE|FORMULAS)[^>]*>/gi, "")
.replace(/\bCLASS=\"?MSO\w*\"?/gi, "")
.replace(/[–]/g,'-') //long –
.replace(/[‘’]/g, "'") //single smartquotes ‘’
.replace(/[“”]/g, '"') //double smartquotes “”
.replace(/align="?justify"?/gi, "") //justify sends some browsers mad
.replace(/<(TABLE|TD|TH|COL)(.*)(WIDTH|HEIGHT)["'0 -9]*/gi, "<$1$2") //no fixed size tables (%OK) [^A-Za-z>]
.replace(/<([^>]+)>\s*<\/\1>/gi, ""); //empty tag
oW.document.body.innerHTML=h;
syncTextarea();
}
I tryed to implement that into the code with wyzz but couldn't get it to work. Is anybody good at this js coding and could help me???
Fia
|
RGaba
Member |
# Posted: 9 Jan 2010 18:21
Well, I wasn't able to get this exact script to work. Instead, I have a function that gets called onclick of the save button. This then cleans all the Word garbage and leaves the tags created by Wyzz. I had to make a lot of changes to the original script, but it can be found at:
www_1stclassmedia.co.uk/developers/clean-ms-word-f ormatting.php
The function I'm using (which works with Wyzz) is....
function CleanWordHTML( fieldId ){
var str = document.getElementById("wysiwyg" + fieldId).contentWindow.document.body.innerHTML;
str = str.replace(/<!(?:--[\s\S]*?--\s*)?>\s*/g,''); // Removes HTML Comments and their contents, looking for <!-- -->
str = str.replace(/<o:p>\s*<\/o:p>/g, "") ;
str = str.replace(/<o:p>.*?<\/o:p>/g, " ") ;
str = str.replace( /\s*mso-[^:]+:[^;"]+;?/gi, "" ) ;
str = str.replace( /\s*MARGIN: 0cm 0cm 0pt\s*;/gi, "" ) ;
str = str.replace( /\s*MARGIN: 0cm 0cm 0pt\s*"/gi, "\"" ) ;
str = str.replace( /\s*TEXT-INDENT: 0cm\s*;/gi, "" ) ;
str = str.replace( /\s*TEXT-INDENT: 0cm\s*"/gi, "\"" ) ;
str = str.replace( /\s*TEXT-ALIGN: [^\s;]+;?"/gi, "\"" ) ;
str = str.replace( /\s*PAGE-BREAK-BEFORE: [^\s;]+;?"/gi, "\"" ) ;
str = str.replace( /\s*FONT-VARIANT: [^\s;]+;?"/gi, "\"" ) ;
str = str.replace( /\s*tab-stops:[^;"]*;?/gi, "" ) ;
str = str.replace( /\s*tab-stops:[^"]*/gi, "" ) ;
str = str.replace( /\s*face="[^"]*"/gi, "" ) ;
str = str.replace( /\s*face=[^ >]*/gi, "" ) ;
str = str.replace( /\s*FONT-FAMILY:[^;"]*;?/gi, "" ) ;
str = str.replace(/<(\w[^>]*) class=([^ |>]*)([^>]*)/gi, "<$1$3") ;
str = str.replace(/<(\w[^>]*) lang=([^ |>]*)([^>]*)/gi, "<$1$3") ;
str = str.replace (/<\\?\?xml[^>]*>/gi, '') ;
str = str.replace(/<\/?\w+:[^>]*>/gi, "") ;
str = str.replace( /<H\d>\s*<\/H\d>/gi, '' ) ;
str = str.replace( /<H1([^>]*)>/gi, '' ) ;
str = str.replace( /<H2([^>]*)>/gi, '' ) ;
str = str.replace( /<H3([^>]*)>/gi, '' ) ;
str = str.replace( /<H4([^>]*)>/gi, '' ) ;
str = str.replace( /<H5([^>]*)>/gi, '' ) ;
str = str.replace( /<H6([^>]*)>/gi, '' ) ;
str = str.replace( /<\/H\d>/gi, '<br>' ) ; //remove this to take out breaks where Heading tags were
str = str.replace( /<link [^>]+>/g, '' ) ;
str = str.replace( /<([^\s>]+)[^>]*>\s*<\/\1>/g, '' ) ;
str = str.replace( /<([^\s>]+)[^>]*>\s*<\/\1>/g, '' ) ;
str = str.replace( /<([^\s>]+)[^>]*>\s*<\/\1>/g, '' ) ;
//some RegEx code for the picky browsers
var re = new RegExp("(<P)([^>]*>.*?)(<\/P>)","gi") ;
str = str.replace( re, "<div$2</div>" ) ;
document.getElementById("wysiwyg" + fieldId).contentWindow.document.body.innerHTML = str; // Sets Wyzz editor to new string
document.getElementById(fieldId).value = str; // Sets hidden field for Wyzz to new string
}
|
