From 27fd67142f499df89f5d23ab70ecec02bc6ffcc7 Mon Sep 17 00:00:00 2001
From: Alan <alan@roojs.com>
Date: Mon, 19 Dec 2022 14:57:47 +0800
Subject: [PATCH] Fix #7504 - include rtf parser in bootstrap editor

---
 buildSDK/dependancy_bootstrap.txt |  11 +
 roojs-bootstrap-debug.js          | 494 ++++++++++++++++++++++++++++++
 roojs-bootstrap.js                |  37 +++
 3 files changed, 542 insertions(+)

diff --git a/buildSDK/dependancy_bootstrap.txt b/buildSDK/dependancy_bootstrap.txt
index dc19c70ca4..cfa194648d 100644
--- a/buildSDK/dependancy_bootstrap.txt
+++ b/buildSDK/dependancy_bootstrap.txt
@@ -127,6 +127,16 @@ Roo.bootstrap.form.CheckBox
 Roo.bootstrap.form.Radio
 Roo.bootstrap.form.SecurePass
 
+
+Roo.rtf.namespace
+Roo.rtf.Hex
+Roo.rtf.Paragraph
+Roo.rtf.Span
+Roo.rtf.Group
+Roo.rtf.Document
+Roo.rtf.Ctrl
+Roo.rtf.Parser
+
 Roo.htmleditor.namespace
 Roo.htmleditor.Filter
 Roo.htmleditor.FilterAttributes
@@ -145,6 +155,7 @@ Roo.htmleditor.TidyWriter
 Roo.htmleditor.TidyEntities
 Roo.htmleditor.KeyEnter
 
+
 Roo.htmleditor.Block
 Roo.htmleditor.BlockFigure
 Roo.htmleditor.BlockTable
diff --git a/roojs-bootstrap-debug.js b/roojs-bootstrap-debug.js
index c639897318..add16b4037 100644
--- a/roojs-bootstrap-debug.js
+++ b/roojs-bootstrap-debug.js
@@ -26030,7 +26030,501 @@ Roo.extend(Roo.bootstrap.form.SecurePass, Roo.bootstrap.form.Input, {
         return this.IsLongEnough(pwd, 6) || !this.IsLongEnough(pwd, 0);
     }
           
+});Roo.rtf = {}; // namespace
+Roo.rtf.Hex = function(hex)
+{
+    this.hexstr = hex;
+};
+Roo.rtf.Paragraph = function(opts)
+{
+    this.content = []; ///??? is that used?
+};Roo.rtf.Span = function(opts)
+{
+    this.value = opts.value;
+};
+
+Roo.rtf.Group = function(parent)
+{
+    // we dont want to acutally store parent - it will make debug a nightmare..
+    this.content = [];
+    this.cn  = [];
+     
+       
+    
+};
+
+Roo.rtf.Group.prototype = {
+    ignorable : false,
+    content: false,
+    cn: false,
+    addContent : function(node) {
+        // could set styles...
+        this.content.push(node);
+    },
+    addChild : function(cn)
+    {
+        this.cn.push(cn);
+    },
+    // only for images really...
+    toDataURL : function()
+    {
+        var mimetype = false;
+        switch(true) {
+            case this.content.filter(function(a) { return a.value == 'pngblip' } ).length > 0: 
+                mimetype = "image/png";
+                break;
+             case this.content.filter(function(a) { return a.value == 'jpegblip' } ).length > 0:
+                mimetype = "image/jpeg";
+                break;
+            default :
+                return 'about:blank'; // ?? error?
+        }
+        
+        
+        var hexstring = this.content[this.content.length-1].value;
+        
+        return 'data:' + mimetype + ';base64,' + btoa(hexstring.match(/\w{2}/g).map(function(a) {
+            return String.fromCharCode(parseInt(a, 16));
+        }).join(""));
+    }
+    
+};
+// this looks like it's normally the {rtf{ .... }}
+Roo.rtf.Document = function()
+{
+    // we dont want to acutally store parent - it will make debug a nightmare..
+    this.rtlch  = [];
+    this.content = [];
+    this.cn = [];
+    
+};
+Roo.extend(Roo.rtf.Document, Roo.rtf.Group, { 
+    addChild : function(cn)
+    {
+        this.cn.push(cn);
+        switch(cn.type) {
+            case 'rtlch': // most content seems to be inside this??
+            case 'listtext':
+            case 'shpinst':
+                this.rtlch.push(cn);
+                return;
+            default:
+                this[cn.type] = cn;
+        }
+        
+    },
+    
+    getElementsByType : function(type)
+    {
+        var ret =  [];
+        this._getElementsByType(type, ret, this.cn, 'rtf');
+        return ret;
+    },
+    _getElementsByType : function (type, ret, search_array, path)
+    {
+        search_array.forEach(function(n,i) {
+            if (n.type == type) {
+                n.path = path + '/' + n.type + ':' + i;
+                ret.push(n);
+            }
+            if (n.cn.length > 0) {
+                this._getElementsByType(type, ret, n.cn, path + '/' + n.type+':'+i);
+            }
+        },this);
+    }
+    
 });
+ 
+Roo.rtf.Ctrl = function(opts)
+{
+    this.value = opts.value;
+    this.param = opts.param;
+};
+/**
+ *
+ *
+ * based on this https://github.com/iarna/rtf-parser
+ * it's really only designed to extract pict from pasted RTF 
+ *
+ * usage:
+ *
+ *  var images = new Roo.rtf.Parser().parse(a_string).filter(function(g) { return g.type == 'pict'; });
+ *  
+ *
+ */
+
+ 
+
+
+
+Roo.rtf.Parser = function(text) {
+    //super({objectMode: true})
+    this.text = '';
+    this.parserState = this.parseText;
+    
+    // these are for interpeter...
+    this.doc = {};
+    ///this.parserState = this.parseTop
+    this.groupStack = [];
+    this.hexStore = [];
+    this.doc = false;
+    
+    this.groups = []; // where we put the return.
+    
+    for (var ii = 0; ii < text.length; ++ii) {
+        ++this.cpos;
+        
+        if (text[ii] === '\n') {
+            ++this.row;
+            this.col = 1;
+        } else {
+            ++this.col;
+        }
+        this.parserState(text[ii]);
+    }
+    
+    
+    
+};
+Roo.rtf.Parser.prototype = {
+    text : '', // string being parsed..
+    controlWord : '',
+    controlWordParam :  '',
+    hexChar : '',
+    doc : false,
+    group: false,
+    groupStack : false,
+    hexStore : false,
+    
+    
+    cpos : 0, 
+    row : 1, // reportin?
+    col : 1, //
+
+     
+    push : function (el)
+    {
+        var m = 'cmd'+ el.type;
+        if (typeof(this[m]) == 'undefined') {
+            Roo.log('invalid cmd:' + el.type);
+            return;
+        }
+        this[m](el);
+        //Roo.log(el);
+    },
+    flushHexStore : function()
+    {
+        if (this.hexStore.length < 1) {
+            return;
+        }
+        var hexstr = this.hexStore.map(
+            function(cmd) {
+                return cmd.value;
+        }).join('');
+        
+        this.group.addContent( new Roo.rtf.Hex( hexstr ));
+              
+            
+        this.hexStore.splice(0)
+        
+    },
+    
+    cmdgroupstart : function()
+    {
+        this.flushHexStore();
+        if (this.group) {
+            this.groupStack.push(this.group);
+        }
+         // parent..
+        if (this.doc === false) {
+            this.group = this.doc = new Roo.rtf.Document();
+            return;
+            
+        }
+        this.group = new Roo.rtf.Group(this.group);
+    },
+    cmdignorable : function()
+    {
+        this.flushHexStore();
+        this.group.ignorable = true;
+    },
+    cmdendparagraph : function()
+    {
+        this.flushHexStore();
+        this.group.addContent(new Roo.rtf.Paragraph());
+    },
+    cmdgroupend : function ()
+    {
+        this.flushHexStore();
+        var endingGroup = this.group;
+        
+        
+        this.group = this.groupStack.pop();
+        if (this.group) {
+            this.group.addChild(endingGroup);
+        }
+        
+        
+        
+        var doc = this.group || this.doc;
+        //if (endingGroup instanceof FontTable) {
+        //  doc.fonts = endingGroup.table
+        //} else if (endingGroup instanceof ColorTable) {
+        //  doc.colors = endingGroup.table
+        //} else if (endingGroup !== this.doc && !endingGroup.get('ignorable')) {
+        if (endingGroup.ignorable === false) {
+            //code
+            this.groups.push(endingGroup);
+           // Roo.log( endingGroup );
+        }
+            //Roo.each(endingGroup.content, function(item)) {
+            //    doc.addContent(item);
+            //}
+            //process.emit('debug', 'GROUP END', endingGroup.type, endingGroup.get('ignorable'))
+        //}
+    },
+    cmdtext : function (cmd)
+    {
+        this.flushHexStore();
+        if (!this.group) { // an RTF fragment, missing the {\rtf1 header
+            //this.group = this.doc
+            return;  // we really don't care about stray text...
+        }
+        this.group.addContent(new Roo.rtf.Span(cmd));
+    },
+    cmdcontrolword : function (cmd)
+    {
+        this.flushHexStore();
+        if (!this.group.type) {
+            this.group.type = cmd.value;
+            return;
+        }
+        this.group.addContent(new Roo.rtf.Ctrl(cmd));
+        // we actually don't care about ctrl words...
+        return ;
+        /*
+        var method = 'ctrl$' + cmd.value.replace(/-(.)/g, (_, char) => char.toUpperCase())
+        if (this[method]) {
+            this[method](cmd.param)
+        } else {
+            if (!this.group.get('ignorable')) process.emit('debug', method, cmd.param)
+        }
+        */
+    },
+    cmdhexchar : function(cmd) {
+        this.hexStore.push(cmd);
+    },
+    cmderror : function(cmd) {
+        throw cmd.value;
+    },
+    
+    /*
+      _flush (done) {
+        if (this.text !== '\u0000') this.emitText()
+        done()
+      }
+      */
+      
+      
+    parseText : function(c)
+    {
+        if (c === '\\') {
+            this.parserState = this.parseEscapes;
+        } else if (c === '{') {
+            this.emitStartGroup();
+        } else if (c === '}') {
+            this.emitEndGroup();
+        } else if (c === '\x0A' || c === '\x0D') {
+            // cr/lf are noise chars
+        } else {
+            this.text += c;
+        }
+    },
+    
+    parseEscapes: function (c)
+    {
+        if (c === '\\' || c === '{' || c === '}') {
+            this.text += c;
+            this.parserState = this.parseText;
+        } else {
+            this.parserState = this.parseControlSymbol;
+            this.parseControlSymbol(c);
+        }
+    },
+    parseControlSymbol: function(c)
+    {
+        if (c === '~') {
+            this.text += '\u00a0'; // nbsp
+            this.parserState = this.parseText
+        } else if (c === '-') {
+             this.text += '\u00ad'; // soft hyphen
+        } else if (c === '_') {
+            this.text += '\u2011'; // non-breaking hyphen
+        } else if (c === '*') {
+            this.emitIgnorable();
+            this.parserState = this.parseText;
+        } else if (c === "'") {
+            this.parserState = this.parseHexChar;
+        } else if (c === '|') { // formula cacter
+            this.emitFormula();
+            this.parserState = this.parseText;
+        } else if (c === ':') { // subentry in an index entry
+            this.emitIndexSubEntry();
+            this.parserState = this.parseText;
+        } else if (c === '\x0a') {
+            this.emitEndParagraph();
+            this.parserState = this.parseText;
+        } else if (c === '\x0d') {
+            this.emitEndParagraph();
+            this.parserState = this.parseText;
+        } else {
+            this.parserState = this.parseControlWord;
+            this.parseControlWord(c);
+        }
+    },
+    parseHexChar: function (c)
+    {
+        if (/^[A-Fa-f0-9]$/.test(c)) {
+            this.hexChar += c;
+            if (this.hexChar.length >= 2) {
+              this.emitHexChar();
+              this.parserState = this.parseText;
+            }
+            return;
+        }
+        this.emitError("Invalid character \"" + c + "\" in hex literal.");
+        this.parserState = this.parseText;
+        
+    },
+    parseControlWord : function(c)
+    {
+        if (c === ' ') {
+            this.emitControlWord();
+            this.parserState = this.parseText;
+        } else if (/^[-\d]$/.test(c)) {
+            this.parserState = this.parseControlWordParam;
+            this.controlWordParam += c;
+        } else if (/^[A-Za-z]$/.test(c)) {
+          this.controlWord += c;
+        } else {
+          this.emitControlWord();
+          this.parserState = this.parseText;
+          this.parseText(c);
+        }
+    },
+    parseControlWordParam : function (c) {
+        if (/^\d$/.test(c)) {
+          this.controlWordParam += c;
+        } else if (c === ' ') {
+          this.emitControlWord();
+          this.parserState = this.parseText;
+        } else {
+          this.emitControlWord();
+          this.parserState = this.parseText;
+          this.parseText(c);
+        }
+    },
+    
+    
+    
+    
+    emitText : function () {
+        if (this.text === '') {
+            return;
+        }
+        this.push({
+            type: 'text',
+            value: this.text,
+            pos: this.cpos,
+            row: this.row,
+            col: this.col
+        });
+        this.text = ''
+    },
+    emitControlWord : function ()
+    {
+        this.emitText();
+        if (this.controlWord === '') {
+            // do we want to track this - it seems just to cause problems.
+            //this.emitError('empty control word');
+        } else {
+            this.push({
+                  type: 'controlword',
+                  value: this.controlWord,
+                  param: this.controlWordParam !== '' && Number(this.controlWordParam),
+                  pos: this.cpos,
+                  row: this.row,
+                  col: this.col
+            });
+        }
+        this.controlWord = '';
+        this.controlWordParam = '';
+    },
+    emitStartGroup : function ()
+    {
+        this.emitText();
+        this.push({
+            type: 'groupstart',
+            pos: this.cpos,
+            row: this.row,
+            col: this.col
+        });
+    },
+    emitEndGroup : function ()
+    {
+        this.emitText();
+        this.push({
+            type: 'groupend',
+            pos: this.cpos,
+            row: this.row,
+            col: this.col
+        });
+    },
+    emitIgnorable : function ()
+    {
+        this.emitText();
+        this.push({
+            type: 'ignorable',
+            pos: this.cpos,
+            row: this.row,
+            col: this.col
+        });
+    },
+    emitHexChar : function ()
+    {
+        this.emitText();
+        this.push({
+            type: 'hexchar',
+            value: this.hexChar,
+            pos: this.cpos,
+            row: this.row,
+            col: this.col
+        });
+        this.hexChar = ''
+    },
+    emitError : function (message)
+    {
+      this.emitText();
+      this.push({
+            type: 'error',
+            value: message,
+            row: this.row,
+            col: this.col,
+            char: this.cpos //,
+            //stack: new Error().stack
+        });
+    },
+    emitEndParagraph : function () {
+        this.emitText();
+        this.push({
+            type: 'endparagraph',
+            pos: this.cpos,
+            row: this.row,
+            col: this.col
+        });
+    }
+     
+} ;
 Roo.htmleditor = {};
  
 /**
diff --git a/roojs-bootstrap.js b/roojs-bootstrap.js
index d8f444087f..2524b1a4d5 100644
--- a/roojs-bootstrap.js
+++ b/roojs-bootstrap.js
@@ -1110,6 +1110,43 @@ pt.innerHTML=this.meterLabel+'&nbsp;'+this.pwdStrengths[B];this.errorMsg='';retu
 }var B=new Array(new this.CharacterSetChecks(this.kCapitalLetter),new this.CharacterSetChecks(this.kSmallLetter),new this.CharacterSetChecks(this.kDigit),new this.CharacterSetChecks(this.kPunctuation));for(var C=0;C<A.length;++C){for(var D=0;D<B.length;++D){if(!B[D].fResult&&this.isctype(A.charAt(C),B[D].type)){B[D].fResult=true;
 break;}}}var E=0;for(var D=0;D<B.length;++D){if(B[D].fResult){++E;}}if(E<nb){return false;}return true;},ClientSideStrongPassword:function(A){return this.IsLongEnough(A,8)&&this.SpansEnoughCharacterSets(A,3);},ClientSideMediumPassword:function(A){return this.IsLongEnough(A,7)&&this.SpansEnoughCharacterSets(A,2);
 },ClientSideWeakPassword:function(A){return this.IsLongEnough(A,6)||!this.IsLongEnough(A,0);}});
+// Roo/rtf/namespace.js
+Roo.rtf={};
+// Roo/rtf/Hex.js
+Roo.rtf.Hex=function(A){this.hexstr=A;};
+// Roo/rtf/Paragraph.js
+Roo.rtf.Paragraph=function(A){this.content=[];};
+// Roo/rtf/Span.js
+Roo.rtf.Span=function(A){this.value=A.value;};
+// Roo/rtf/Group.js
+Roo.rtf.Group=function(A){this.content=[];this.cn=[];};Roo.rtf.Group.prototype={ignorable:false,content:false,cn:false,addContent:function(A){this.content.push(A);},addChild:function(cn){this.cn.push(cn);},toDataURL:function(){var A=false;switch(true){case this.content.filter(function(a){return a.value=='pngblip'}
+).length>0:A="image/png";break;case this.content.filter(function(a){return a.value=='jpegblip'}).length>0:A="image/jpeg";break;default:return 'about:blank';}var B=this.content[this.content.length-1].value;return 'data:'+A+';base64,'+btoa(B.match(/\w{2}/g).map(function(a){return String.fromCharCode(parseInt(a,16));
+}).join(""));}};
+// Roo/rtf/Document.js
+Roo.rtf.Document=function(){this.rtlch=[];this.content=[];this.cn=[];};Roo.extend(Roo.rtf.Document,Roo.rtf.Group,{addChild:function(cn){this.cn.push(cn);switch(cn.type){case 'rtlch':case 'listtext':case 'shpinst':this.rtlch.push(cn);return;default:this[cn.type]=cn;
+}},getElementsByType:function(A){var B=[];this._getElementsByType(A,B,this.cn,'rtf');return B;},_getElementsByType:function(A,B,C,D){C.forEach(function(n,i){if(n.type==A){n.path=D+'/'+n.type+':'+i;B.push(n);}if(n.cn.length>0){this._getElementsByType(A,B,n.cn,D+'/'+n.type+':'+i);
+}},this);}});
+// Roo/rtf/Ctrl.js
+Roo.rtf.Ctrl=function(A){this.value=A.value;this.param=A.param;};
+// Roo/rtf/Parser.js
+Roo.rtf.Parser=function(A){this.text='';this.parserState=this.parseText;this.doc={};this.groupStack=[];this.hexStore=[];this.doc=false;this.groups=[];for(var ii=0;ii<A.length;++ii){++this.cpos;if(A[ii]==='\n'){++this.row;this.col=1;}else{++this.col;}this.parserState(A[ii]);
+}};Roo.rtf.Parser.prototype={text:'',controlWord:'',controlWordParam:'',hexChar:'',doc:false,group:false,groupStack:false,hexStore:false,cpos:0,row:1,col:1,push:function(el){var m='cmd'+el.type;if(typeof(this[m])=='undefined'){Roo.log('invalid cmd:'+el.type);
+return;}this[m](el);},flushHexStore:function(){if(this.hexStore.length<1){return;}var A=this.hexStore.map(function(B){return B.value;}).join('');this.group.addContent(new Roo.rtf.Hex(A));this.hexStore.splice(0)},cmdgroupstart:function(){this.flushHexStore();
+if(this.group){this.groupStack.push(this.group);}if(this.doc===false){this.group=this.doc=new Roo.rtf.Document();return;}this.group=new Roo.rtf.Group(this.group);},cmdignorable:function(){this.flushHexStore();this.group.ignorable=true;},cmdendparagraph:function(){this.flushHexStore();
+this.group.addContent(new Roo.rtf.Paragraph());},cmdgroupend:function(){this.flushHexStore();var A=this.group;this.group=this.groupStack.pop();if(this.group){this.group.addChild(A);}var B=this.group||this.doc;if(A.ignorable===false){this.groups.push(A);}}
+,cmdtext:function(A){this.flushHexStore();if(!this.group){return;}this.group.addContent(new Roo.rtf.Span(A));},cmdcontrolword:function(A){this.flushHexStore();if(!this.group.type){this.group.type=A.value;return;}this.group.addContent(new Roo.rtf.Ctrl(A));
+return;},cmdhexchar:function(A){this.hexStore.push(A);},cmderror:function(A){throw A.value;},parseText:function(c){if(c==='\\'){this.parserState=this.parseEscapes;}else if(c==='{'){this.emitStartGroup();}else if(c==='}'){this.emitEndGroup();}else if(c==='\x0A'||c==='\x0D'){}
+else{this.text+=c;}},parseEscapes:function(c){if(c==='\\'||c==='{'||c==='}'){this.text+=c;this.parserState=this.parseText;}else{this.parserState=this.parseControlSymbol;this.parseControlSymbol(c);}},parseControlSymbol:function(c){if(c==='~'){this.text+='\u00a0';
+this.parserState=this.parseText}else if(c==='-'){this.text+='\u00ad';}else if(c==='_'){this.text+='\u2011';}else if(c==='*'){this.emitIgnorable();this.parserState=this.parseText;}else if(c==="'"){this.parserState=this.parseHexChar;}else if(c==='|'){this.emitFormula();
+this.parserState=this.parseText;}else if(c===':'){this.emitIndexSubEntry();this.parserState=this.parseText;}else if(c==='\x0a'){this.emitEndParagraph();this.parserState=this.parseText;}else if(c==='\x0d'){this.emitEndParagraph();this.parserState=this.parseText;
+}else{this.parserState=this.parseControlWord;this.parseControlWord(c);}},parseHexChar:function(c){if(/^[A-Fa-f0-9]$/.test(c)){this.hexChar+=c;if(this.hexChar.length>=2){this.emitHexChar();this.parserState=this.parseText;}return;}this.emitError("Invalid character \""+c+"\" in hex literal.");
+this.parserState=this.parseText;},parseControlWord:function(c){if(c===' '){this.emitControlWord();this.parserState=this.parseText;}else if(/^[-\d]$/.test(c)){this.parserState=this.parseControlWordParam;this.controlWordParam+=c;}else if(/^[A-Za-z]$/.test(c)){this.controlWord+=c;
+}else{this.emitControlWord();this.parserState=this.parseText;this.parseText(c);}},parseControlWordParam:function(c){if(/^\d$/.test(c)){this.controlWordParam+=c;}else if(c===' '){this.emitControlWord();this.parserState=this.parseText;}else{this.emitControlWord();
+this.parserState=this.parseText;this.parseText(c);}},emitText:function(){if(this.text===''){return;}this.push({type:'text',value:this.text,pos:this.cpos,row:this.row,col:this.col});this.text=''},emitControlWord:function(){this.emitText();if(this.controlWord===''){}
+else{this.push({type:'controlword',value:this.controlWord,param:this.controlWordParam!==''&&Number(this.controlWordParam),pos:this.cpos,row:this.row,col:this.col});}this.controlWord='';this.controlWordParam='';},emitStartGroup:function(){this.emitText();this.push({type:'groupstart',pos:this.cpos,row:this.row,col:this.col}
+);},emitEndGroup:function(){this.emitText();this.push({type:'groupend',pos:this.cpos,row:this.row,col:this.col});},emitIgnorable:function(){this.emitText();this.push({type:'ignorable',pos:this.cpos,row:this.row,col:this.col});},emitHexChar:function(){this.emitText();
+this.push({type:'hexchar',value:this.hexChar,pos:this.cpos,row:this.row,col:this.col});this.hexChar=''},emitError:function(A){this.emitText();this.push({type:'error',value:A,row:this.row,col:this.col,char:this.cpos});},emitEndParagraph:function(){this.emitText();
+this.push({type:'endparagraph',pos:this.cpos,row:this.row,col:this.col});}};
 // Roo/htmleditor/namespace.js
 Roo.htmleditor={};
 // Roo/htmleditor/Filter.js
-- 
2.39.2