svx/source/editeng/eehtml.cxx

   1 /*************************************************************************
   2  *
   3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   4  *
   5  * Copyright 2008 by Sun Microsystems, Inc.
   6  *
   7  * OpenOffice.org - a multi-platform office productivity suite
   8  *
   9  * $RCSfile: eehtml.cxx,v $
  10  * $Revision: 1.20 $
  11  *
  12  * This file is part of OpenOffice.org.
  13  *
  14  * OpenOffice.org is free software: you can redistribute it and/or modify
  15  * it under the terms of the GNU Lesser General Public License version 3
  16  * only, as published by the Free Software Foundation.
  17  *
  18  * OpenOffice.org is distributed in the hope that it will be useful,
  19  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  20  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  21  * GNU Lesser General Public License version 3 for more details
  22  * (a copy is included in the LICENSE file that accompanied this code).
  23  *
  24  * You should have received a copy of the GNU Lesser General Public License
  25  * version 3 along with OpenOffice.org.  If not, see
  26  * <http://www.openoffice.org/license.html>
  27  * for a copy of the LGPLv3 License.
  28  *
  29  ************************************************************************/
  30
  31 // MARKER(update_precomp.py): autogen include statement, do not remove
  32 #include "precompiled_svx.hxx"
  33
  34 #include <eeng_pch.hxx>
  35
  36 #include <eehtml.hxx>
  37 #include <impedit.hxx>
  38 #include <svx/adjitem.hxx>
  39 #include <svx/flditem.hxx>
  40 #include <tools/urlobj.hxx>
  41 #include <svx/fhgtitem.hxx>
  42 #include "fontitem.hxx"
  43 #include <svx/ulspitem.hxx>
  44 #include <svx/wghtitem.hxx>
  45 #include <svtools/htmltokn.h>
  46 #include <svtools/htmlkywd.hxx>
  47
  48
  49 #define ACTION_INSERTTEXT                 1
  50 #define ACTION_INSERTPARABRK      2
  51
  52 #define STYLE_PRE                               101
  53
  54 EditHTMLParser::EditHTMLParser( SvStream& rIn, const String& rBaseURL, SvKeyValueIterator* pHTTPHeaderAttrs )
  55     : SfxHTMLParser( rIn, true )
  56     , aBaseURL( rBaseURL )
  57 {
  58     pImpEditEngine = 0;
  59     pCurAnchor = 0;
  60     bInPara = FALSE;
  61     bWasInPara = FALSE;
  62     nInTable = 0;
  63     nInCell = 0;
  64     nDefListLevel = 0;
  65     nBulletLevel = 0;
  66     nNumberingLevel = 0;
  67     bFieldsInserted = FALSE;
  68
  69     if ( pHTTPHeaderAttrs )
  70         SetEncodingByHTTPHeader( pHTTPHeaderAttrs );
  71 }
  72
  73 EditHTMLParser::~EditHTMLParser()
  74 {
  75     delete pCurAnchor;
  76 }
  77
  78 SvParserState EditHTMLParser::CallParser( ImpEditEngine* pImpEE, const EditPaM& rPaM )
  79 {
  80     DBG_ASSERT( pImpEE, "CallParser: ImpEditEngine ?!" );
  81     pImpEditEngine = pImpEE;
  82     SvParserState _eState = SVPAR_NOTSTARTED;
  83     if ( pImpEditEngine )
  84     {
  85         // Umbrechmimik vom RTF-Import einbauen?
  86         aCurSel = EditSelection( rPaM, rPaM );
  87
  88         if ( pImpEditEngine->aImportHdl.IsSet() )
  89         {
  90             ImportInfo aImportInfo( HTMLIMP_START, this, pImpEditEngine->CreateESel( aCurSel ) );
  91             pImpEditEngine->aImportHdl.Call( &aImportInfo );
  92         }
  93
  94         ImpSetStyleSheet( 0 );
  95         _eState = HTMLParser::CallParser();
  96
  97         if ( pImpEditEngine->aImportHdl.IsSet() )
  98         {
  99             ImportInfo aImportInfo( HTMLIMP_END, this, pImpEditEngine->CreateESel( aCurSel ) );
 100             pImpEditEngine->aImportHdl.Call( &aImportInfo );
 101         }
 102
 103         if ( bFieldsInserted )
 104             pImpEditEngine->UpdateFields();
 105     }
 106     return _eState;
 107 }
 108
 109 void EditHTMLParser::NextToken( int nToken )
 110 {
 111     #ifdef DBG_UTIL
 112         HTML_TOKEN_IDS xID = (HTML_TOKEN_IDS)nToken;
 113         (void)xID;
 114     #endif
 115
 116     switch( nToken )
 117     {
 118     case HTML_META:
 119     {
 120         const HTMLOptions *_pOptions = GetOptions();
 121         USHORT nArrLen = _pOptions->Count();
 122         BOOL bEquiv = FALSE;
 123         for ( USHORT i = 0; i < nArrLen; i++ )
 124         {
 125             const HTMLOption *pOption = (*_pOptions)[i];
 126             switch( pOption->GetToken() )
 127             {
 128                 case HTML_O_HTTPEQUIV:
 129                 {
 130                     bEquiv = TRUE;
 131                 }
 132                 break;
 133                 case HTML_O_CONTENT:
 134                 {
 135                     if ( bEquiv )
 136                     {
 137                         rtl_TextEncoding eEnc = GetEncodingByMIME( pOption->GetString() );
 138                         if ( eEnc != RTL_TEXTENCODING_DONTKNOW )
 139                             SetSrcEncoding( eEnc );
 140                     }
 141                 }
 142                 break;
 143             }
 144         }
 145
 146     }
 147     break;
 148     case HTML_PLAINTEXT_ON:
 149     case HTML_PLAINTEXT2_ON:
 150         bInPara = TRUE;
 151     break;
 152     case HTML_PLAINTEXT_OFF:
 153     case HTML_PLAINTEXT2_OFF:
 154         bInPara = FALSE;
 155     break;
 156
 157     case HTML_LINEBREAK:
 158     case HTML_NEWPARA:
 159     {
 160         if ( ( bInPara || nInTable ) &&
 161             ( ( nToken == HTML_LINEBREAK ) || HasTextInCurrentPara() ) )
 162         {
 163             ImpInsertParaBreak();
 164         }
 165     }
 166     break;
 167     case HTML_HORZRULE:
 168     {
 169         if ( HasTextInCurrentPara() )
 170             ImpInsertParaBreak();
 171         ImpInsertParaBreak();
 172     }
 173     case HTML_NONBREAKSPACE:
 174     {
 175         if ( bInPara )
 176         {
 177             ImpInsertText( String( RTL_CONSTASCII_USTRINGPARAM( " " ) ) );
 178         }
 179     }
 180     break;
 181     case HTML_TEXTTOKEN:
 182     {
 183         if ( !bInPara )
 184             StartPara( FALSE );
 185
 186 //        if ( bInPara || pCurAnchor )
 187         {
 188             String aText = aToken;
 189             if ( aText.Len() && ( aText.GetChar( 0 ) == ' ' )
 190                     && ThrowAwayBlank() && !IsReadPRE() )
 191                 aText.Erase( 0, 1 );
 192
 193             if ( pCurAnchor )
 194             {
 195                 pCurAnchor->aText += aText;
 196             }
 197             else
 198             {
 199                 // Nur bis HTML mit 319 geschrieben ?!
 200                 if ( IsReadPRE() )
 201                 {
 202                     USHORT nTabPos = aText.Search( '\t', 0 );
 203                     while ( nTabPos != STRING_NOTFOUND )
 204                     {
 205                         aText.Erase( nTabPos, 1 );
 206                         aText.Insert( String( RTL_CONSTASCII_USTRINGPARAM( "        " ) ), nTabPos );
 207                         nTabPos = aText.Search( '\t', nTabPos+8 );
 208                     }
 209                 }
 210                 ImpInsertText( aText );
 211             }
 212         }
 213     }
 214     break;
 215
 216     case HTML_CENTER_ON:
 217     case HTML_CENTER_OFF:       // if ( bInPara )
 218                             {
 219                                 USHORT nNode = pImpEditEngine->GetEditDoc().GetPos( aCurSel.Max().GetNode() );
 220                                 SfxItemSet aItems( aCurSel.Max().GetNode()->GetContentAttribs().GetItems() );
 221                                 aItems.ClearItem( EE_PARA_JUST );
 222                                 if ( nToken == HTML_CENTER_ON )
 223                                     aItems.Put( SvxAdjustItem( SVX_ADJUST_CENTER, EE_PARA_JUST ) );
 224                                 pImpEditEngine->SetParaAttribs( nNode, aItems );
 225                             }
 226                             break;
 227
 228     case HTML_ANCHOR_ON:        AnchorStart();
 229                             break;
 230     case HTML_ANCHOR_OFF:       AnchorEnd();
 231                             break;
 232
 233     case HTML_PARABREAK_ON:
 234         if( bInPara && HasTextInCurrentPara() )
 235             EndPara( TRUE );
 236         StartPara( TRUE );
 237         break;
 238
 239     case HTML_PARABREAK_OFF:
 240         if( bInPara )
 241             EndPara( TRUE );
 242         break;
 243
 244     case HTML_HEAD1_ON:
 245     case HTML_HEAD2_ON:
 246     case HTML_HEAD3_ON:
 247     case HTML_HEAD4_ON:
 248     case HTML_HEAD5_ON:
 249     case HTML_HEAD6_ON:
 250     {
 251         HeadingStart( nToken );
 252     }
 253     break;
 254
 255     case HTML_HEAD1_OFF:
 256     case HTML_HEAD2_OFF:
 257     case HTML_HEAD3_OFF:
 258     case HTML_HEAD4_OFF:
 259     case HTML_HEAD5_OFF:
 260     case HTML_HEAD6_OFF:
 261     {
 262         HeadingEnd( nToken );
 263     }
 264     break;
 265
 266     case HTML_PREFORMTXT_ON:
 267     case HTML_XMP_ON:
 268     case HTML_LISTING_ON:
 269     {
 270         StartPara( TRUE );
 271         ImpSetStyleSheet( STYLE_PRE );
 272     }
 273     break;
 274
 275     case HTML_DEFLIST_ON:
 276     {
 277         nDefListLevel++;
 278     }
 279     break;
 280
 281     case HTML_DEFLIST_OFF:
 282     {
 283         if( nDefListLevel )
 284             nDefListLevel--;
 285     }
 286     break;
 287
 288     case HTML_TABLE_ON:         nInTable++;
 289                             break;
 290     case HTML_TABLE_OFF:        DBG_ASSERT( nInTable, "Nicht in Table, aber TABLE_OFF?" );
 291                             nInTable--;
 292                             break;
 293
 294     case HTML_TABLEHEADER_ON:
 295     case HTML_TABLEDATA_ON:
 296         nInCell++;
 297     // fallthru
 298     case HTML_BLOCKQUOTE_ON:
 299     case HTML_BLOCKQUOTE_OFF:
 300     case HTML_BLOCKQUOTE30_ON:
 301     case HTML_BLOCKQUOTE30_OFF:
 302     case HTML_LISTHEADER_ON:
 303     case HTML_LI_ON:
 304     case HTML_DD_ON:
 305     case HTML_DT_ON:
 306     case HTML_ORDERLIST_ON:
 307     case HTML_UNORDERLIST_ON:
 308     {
 309         BOOL bHasText = HasTextInCurrentPara();
 310         if ( bHasText )
 311             ImpInsertParaBreak();
 312         StartPara( FALSE );
 313     }
 314     break;
 315
 316     case HTML_TABLEHEADER_OFF:
 317     case HTML_TABLEDATA_OFF:
 318     {
 319         if ( nInCell )
 320             nInCell--;
 321     }
 322     // fallthru
 323     case HTML_LISTHEADER_OFF:
 324     case HTML_LI_OFF:
 325     case HTML_DD_OFF:
 326     case HTML_DT_OFF:
 327     case HTML_ORDERLIST_OFF:
 328     case HTML_UNORDERLIST_OFF:  EndPara( FALSE );
 329                                 break;
 330
 331     case HTML_TABLEROW_ON:
 332     case HTML_TABLEROW_OFF:     // Nur nach einem CELL ein RETURN, fuer Calc
 333
 334     case HTML_COL_ON:
 335     case HTML_COLGROUP_ON:
 336     case HTML_COLGROUP_OFF: break;
 337
 338     case HTML_FONT_ON:          // ...
 339                             break;
 340     case HTML_FONT_OFF:         // ...
 341                             break;
 342
 343
 344     // #58335# kein SkipGroup on/off auf inline markup etc.
 345
 346     // globals
 347     case HTML_HTML_ON:
 348     case HTML_HTML_OFF:
 349     case HTML_BODY_ON:
 350     case HTML_BODY_OFF:
 351     case HTML_HEAD_ON:
 352     case HTML_HEAD_OFF:
 353     case HTML_FORM_ON:
 354     case HTML_FORM_OFF:
 355     case HTML_THEAD_ON:
 356     case HTML_THEAD_OFF:
 357     case HTML_TBODY_ON:
 358     case HTML_TBODY_OFF:
 359     case HTML_TITLE_ON:
 360     case HTML_TITLE_OFF:
 361     // inline elements, structural markup
 362     // HTML 3.0
 363     case HTML_BANNER_ON:
 364     case HTML_BANNER_OFF:
 365     case HTML_DIVISION_ON:
 366     case HTML_DIVISION_OFF:
 367 //      case HTML_LISTHEADER_ON:                //! special handling
 368 //      case HTML_LISTHEADER_OFF:
 369     case HTML_NOTE_ON:
 370     case HTML_NOTE_OFF:
 371     // inline elements, logical markup
 372     // HTML 2.0
 373     case HTML_ADDRESS_ON:
 374     case HTML_ADDRESS_OFF:
 375 //      case HTML_BLOCKQUOTE_ON:                //! extra Behandlung
 376 //      case HTML_BLOCKQUOTE_OFF:
 377     case HTML_CITIATION_ON:
 378     case HTML_CITIATION_OFF:
 379     case HTML_CODE_ON:
 380     case HTML_CODE_OFF:
 381     case HTML_DEFINSTANCE_ON:
 382     case HTML_DEFINSTANCE_OFF:
 383     case HTML_EMPHASIS_ON:
 384     case HTML_EMPHASIS_OFF:
 385     case HTML_KEYBOARD_ON:
 386     case HTML_KEYBOARD_OFF:
 387     case HTML_SAMPLE_ON:
 388     case HTML_SAMPLE_OFF:
 389     case HTML_STRIKE_ON:
 390     case HTML_STRIKE_OFF:
 391     case HTML_STRONG_ON:
 392     case HTML_STRONG_OFF:
 393     case HTML_VARIABLE_ON:
 394     case HTML_VARIABLE_OFF:
 395     // HTML 3.0
 396     case HTML_ABBREVIATION_ON:
 397     case HTML_ABBREVIATION_OFF:
 398     case HTML_ACRONYM_ON:
 399     case HTML_ACRONYM_OFF:
 400     case HTML_AUTHOR_ON:
 401     case HTML_AUTHOR_OFF:
 402 //      case HTML_BLOCKQUOTE30_ON:              //! extra Behandlung
 403 //      case HTML_BLOCKQUOTE30_OFF:
 404     case HTML_DELETEDTEXT_ON:
 405     case HTML_DELETEDTEXT_OFF:
 406     case HTML_INSERTEDTEXT_ON:
 407     case HTML_INSERTEDTEXT_OFF:
 408     case HTML_LANGUAGE_ON:
 409     case HTML_LANGUAGE_OFF:
 410     case HTML_PERSON_ON:
 411     case HTML_PERSON_OFF:
 412     case HTML_SHORTQUOTE_ON:
 413     case HTML_SHORTQUOTE_OFF:
 414     case HTML_SUBSCRIPT_ON:
 415     case HTML_SUBSCRIPT_OFF:
 416     case HTML_SUPERSCRIPT_ON:
 417     case HTML_SUPERSCRIPT_OFF:
 418     // inline elements, visual markup
 419     // HTML 2.0
 420     case HTML_BOLD_ON:
 421     case HTML_BOLD_OFF:
 422     case HTML_ITALIC_ON:
 423     case HTML_ITALIC_OFF:
 424     case HTML_TELETYPE_ON:
 425     case HTML_TELETYPE_OFF:
 426     case HTML_UNDERLINE_ON:
 427     case HTML_UNDERLINE_OFF:
 428     // HTML 3.0
 429     case HTML_BIGPRINT_ON:
 430     case HTML_BIGPRINT_OFF:
 431     case HTML_STRIKETHROUGH_ON:
 432     case HTML_STRIKETHROUGH_OFF:
 433     case HTML_SMALLPRINT_ON:
 434     case HTML_SMALLPRINT_OFF:
 435     // figures
 436     case HTML_FIGURE_ON:
 437     case HTML_FIGURE_OFF:
 438     case HTML_CAPTION_ON:
 439     case HTML_CAPTION_OFF:
 440     case HTML_CREDIT_ON:
 441     case HTML_CREDIT_OFF:
 442     // misc
 443     case HTML_DIRLIST_ON:
 444     case HTML_DIRLIST_OFF:
 445     case HTML_FOOTNOTE_ON:                      //! landen so im Text
 446     case HTML_FOOTNOTE_OFF:
 447     case HTML_MENULIST_ON:
 448     case HTML_MENULIST_OFF:
 449 //      case HTML_PLAINTEXT_ON:                 //! extra Behandlung
 450 //      case HTML_PLAINTEXT_OFF:
 451 //      case HTML_PREFORMTXT_ON:                //! extra Behandlung
 452 //      case HTML_PREFORMTXT_OFF:
 453     case HTML_SPAN_ON:
 454     case HTML_SPAN_OFF:
 455     // obsolete
 456 //      case HTML_XMP_ON:                               //! extra Behandlung
 457 //      case HTML_XMP_OFF:
 458 //      case HTML_LISTING_ON:                   //! extra Behandlung
 459 //      case HTML_LISTING_OFF:
 460     // Netscape
 461     case HTML_BLINK_ON:
 462     case HTML_BLINK_OFF:
 463     case HTML_NOBR_ON:
 464     case HTML_NOBR_OFF:
 465     case HTML_NOEMBED_ON:
 466     case HTML_NOEMBED_OFF:
 467     case HTML_NOFRAMES_ON:
 468     case HTML_NOFRAMES_OFF:
 469     // Internet Explorer
 470     case HTML_MARQUEE_ON:
 471     case HTML_MARQUEE_OFF:
 472 //      case HTML_PLAINTEXT2_ON:                //! extra Behandlung
 473 //      case HTML_PLAINTEXT2_OFF:
 474     break;
 475
 476     default:
 477     {
 478         if ( nToken & HTML_TOKEN_ONOFF )
 479         {
 480             if ( ( nToken == HTML_UNKNOWNCONTROL_ON ) || ( nToken == HTML_UNKNOWNCONTROL_OFF ) )
 481             {
 482                 ;
 483             }
 484             else if ( !(nToken & 1) )
 485             {
 486                 DBG_ASSERT( !( nToken & 1 ), "Kein Start-Token ?!" );
 487                 SkipGroup( nToken + 1 );
 488             }
 489         }
 490     }
 491     }   // SWITCH
 492
 493     if ( pImpEditEngine->aImportHdl.IsSet() )
 494     {
 495         ImportInfo aImportInfo( HTMLIMP_NEXTTOKEN, this, pImpEditEngine->CreateESel( aCurSel ) );
 496         aImportInfo.nToken = nToken;
 497         aImportInfo.nTokenValue = (short)nTokenValue;
 498         if ( nToken == HTML_TEXTTOKEN )
 499             aImportInfo.aText = aToken;
 500         pImpEditEngine->aImportHdl.Call( &aImportInfo );
 501     }
 502
 503 }
 504
 505 void EditHTMLParser::ImpInsertParaBreak()
 506 {
 507     if ( pImpEditEngine->aImportHdl.IsSet() )
 508     {
 509         ImportInfo aImportInfo( HTMLIMP_INSERTPARA, this, pImpEditEngine->CreateESel( aCurSel ) );
 510         pImpEditEngine->aImportHdl.Call( &aImportInfo );
 511     }
 512     aCurSel = pImpEditEngine->ImpInsertParaBreak( aCurSel );
 513     nLastAction = ACTION_INSERTPARABRK;
 514 }
 515
 516 void EditHTMLParser::ImpSetAttribs( const SfxItemSet& rItems, EditSelection* pSel )
 517 {
 518     // pSel, wenn Zeichenattribute, sonst Absatzattribute fuer den
 519     // aktuellen Absatz.
 520     DBG_ASSERT( pSel || ( aCurSel.Min().GetNode() == aCurSel.Max().GetNode() ), "ImpInsertAttribs: Selektion?" );
 521
 522     EditPaM aStartPaM( pSel ? pSel->Min() : aCurSel.Min() );
 523     EditPaM aEndPaM( pSel ? pSel->Max() : aCurSel.Max() );
 524
 525     if ( !pSel )
 526     {
 527         aStartPaM.SetIndex( 0 );
 528         aEndPaM.SetIndex( aEndPaM.GetNode()->Len() );
 529     }
 530
 531     if ( pImpEditEngine->aImportHdl.IsSet() )
 532     {
 533         EditSelection aSel( aStartPaM, aEndPaM );
 534         ImportInfo aImportInfo( HTMLIMP_SETATTR, this, pImpEditEngine->CreateESel( aSel ) );
 535         aImportInfo.pAttrs = (void*)&rItems;
 536         pImpEditEngine->aImportHdl.Call( &aImportInfo );
 537     }
 538
 539     ContentNode* pSN = aStartPaM.GetNode();
 540     USHORT nStartNode = pImpEditEngine->GetEditDoc().GetPos( pSN );
 541
 542     // Wenn ein Attribut von 0 bis aktuelle Absatzlaenge geht,
 543     // soll es ein Absatz-Attribut sein!
 544
 545     // Achtung: Selektion kann ueber mehrere Absaetze gehen.
 546     // Alle vollstaendigen Absaetze sind Absatzattribute...
 547
 548     // HTML eigentlich nicht:
 549 #ifdef DBG_UTIL
 550     ContentNode* pEN = aEndPaM.GetNode();
 551     USHORT nEndNode = pImpEditEngine->GetEditDoc().GetPos( pEN );
 552     DBG_ASSERT( nStartNode == nEndNode, "ImpSetAttribs: Mehrere Absaetze?" );
 553 #endif
 554
 555 /*
 556     for ( USHORT z = nStartNode+1; z < nEndNode; z++ )
 557     {
 558         DBG_ASSERT( pImpEditEngine->GetEditDoc().SaveGetObject( z ), "Node existiert noch nicht(RTF)" );
 559         pImpEditEngine->SetParaAttribs( z, rSet.GetAttrSet() );
 560     }
 561
 562     if ( aStartPaM.GetNode() != aEndPaM.GetNode() )
 563     {
 564         // Den Rest des StartNodes...
 565         if ( aStartPaM.GetIndex() == 0 )
 566             pImpEditEngine->SetParaAttribs( nStartNode, rSet.GetAttrSet() );
 567         else
 568             pImpEditEngine->SetAttribs( EditSelection( aStartPaM, EditPaM( aStartPaM.GetNode(), aStartPaM.GetNode()->Len() ) ), rSet.GetAttrSet() );
 569
 570         // Den Anfang des EndNodes....
 571         if ( aEndPaM.GetIndex() == aEndPaM.GetNode()->Len() )
 572             pImpEditEngine->SetParaAttribs( nEndNode, rSet.GetAttrSet() );
 573         else
 574             pImpEditEngine->SetAttribs( EditSelection( EditPaM( aEndPaM.GetNode(), 0 ), aEndPaM ), rSet.GetAttrSet() );
 575     }
 576     else
 577 */
 578     {
 579         if ( ( aStartPaM.GetIndex() == 0 ) && ( aEndPaM.GetIndex() == aEndPaM.GetNode()->Len() ) )
 580         {
 581             // Muesse gemergt werden:
 582             SfxItemSet aItems( pImpEditEngine->GetParaAttribs( nStartNode ) );
 583             aItems.Put( rItems );
 584             pImpEditEngine->SetParaAttribs( nStartNode, aItems );
 585         }
 586         else
 587             pImpEditEngine->SetAttribs( EditSelection( aStartPaM, aEndPaM ), rItems );
 588     }
 589 }
 590
 591 void EditHTMLParser::ImpSetStyleSheet( USHORT nHLevel )
 592 {
 593     /*
 594         nHLevel:        0:                      Ausschalten
 595                     1-6:                Heading
 596                     STYLE_PRE:  Preformatted
 597     */
 598
 599 //              if ( pImpEditEngine->GetStatus().DoImportRTFStyleSheets() )
 600 //              {
 601 //                      SvxRTFStyleType* pS = GetStyleTbl().Get( rSet.StyleNo() );
 602 //                      DBG_ASSERT( pS, "Vorlage in RTF nicht definiert!" );
 603 //                      if ( pS )
 604 //                              pImpEditEngine->SetStyleSheet( EditSelection( aStartPaM, aEndPaM ), pS->sName, SFX_STYLE_FAMILY_ALL );
 605 //              }
 606 //              else
 607         {
 608             // Harte Attribute erzeugen...
 609             // Reicht fuer Calc, bei StyleSheets muesste noch geklaert werden,
 610             // dass diese auch in der App liegen sollten, damit sie beim
 611             // fuettern in eine andere Engine auch noch da sind...
 612
 613             USHORT nNode = pImpEditEngine->GetEditDoc().GetPos( aCurSel.Max().GetNode() );
 614 //                      SfxItemSet aItems( pImpEditEngine->GetEmptyItemSet() );
 615             SfxItemSet aItems( aCurSel.Max().GetNode()->GetContentAttribs().GetItems() );
 616
 617             aItems.ClearItem( EE_PARA_ULSPACE );
 618             aItems.ClearItem( EE_CHAR_FONTHEIGHT );
 619             aItems.ClearItem( EE_CHAR_FONTINFO );
 620             aItems.ClearItem( EE_CHAR_WEIGHT );
 621
 622             // Fett in den ersten 3 Headings
 623             if ( ( nHLevel >= 1 ) && ( nHLevel <= 3 ) )
 624             {
 625                 SvxWeightItem aWeightItem( WEIGHT_BOLD, EE_CHAR_WEIGHT );
 626                 aItems.Put( aWeightItem );
 627             }
 628
 629             // Fonthoehe und Abstaende, wenn LogicToLogic moeglich:
 630             MapUnit eUnit = pImpEditEngine->GetRefMapMode().GetMapUnit();
 631             if ( ( eUnit != MAP_PIXEL ) && ( eUnit != MAP_SYSFONT ) &&
 632                  ( eUnit != MAP_APPFONT ) && ( eUnit != MAP_RELATIVE ) )
 633             {
 634                 long nPoints = 10;
 635                 if ( nHLevel == 1 )
 636                     nPoints = 22;
 637                 else if ( nHLevel == 2 )
 638                     nPoints = 16;
 639                 else if ( nHLevel == 3 )
 640                     nPoints = 12;
 641                 else if ( nHLevel == 4 )
 642                     nPoints = 11;
 643
 644                 nPoints = OutputDevice::LogicToLogic( nPoints, MAP_POINT, eUnit );
 645                 SvxFontHeightItem aHeightItem( nPoints, 100, EE_CHAR_FONTHEIGHT );
 646                 aItems.Put( aHeightItem );
 647
 648                 // Absatzabstaende, wenn Heading:
 649                 if ( !nHLevel || ((nHLevel >= 1) && (nHLevel <= 6)) )
 650                 {
 651                     SvxULSpaceItem aULSpaceItem( EE_PARA_ULSPACE );
 652                     aULSpaceItem.SetUpper( (USHORT)OutputDevice::LogicToLogic( 42, MAP_10TH_MM, eUnit ) );
 653                     aULSpaceItem.SetLower( (USHORT)OutputDevice::LogicToLogic( 35, MAP_10TH_MM, eUnit ) );
 654                     aItems.Put( aULSpaceItem );
 655                 }
 656             }
 657
 658             // Bei Pre einen proportionalen Font waehlen
 659             if ( nHLevel == STYLE_PRE )
 660             {
 661                 Font aFont = OutputDevice::GetDefaultFont( DEFAULTFONT_FIXED, LANGUAGE_SYSTEM, 0 );
 662                 SvxFontItem aFontItem( aFont.GetFamily(), aFont.GetName(), XubString(), aFont.GetPitch(), aFont.GetCharSet(), EE_CHAR_FONTINFO );
 663                 aItems.Put( aFontItem );
 664             }
 665
 666             pImpEditEngine->SetParaAttribs( nNode, aItems );
 667         }
 668 }
 669
 670 void EditHTMLParser::ImpInsertText( const String& rText )
 671 {
 672     String aText( rText );
 673     if ( pImpEditEngine->aImportHdl.IsSet() )
 674     {
 675         ImportInfo aImportInfo( HTMLIMP_INSERTTEXT, this, pImpEditEngine->CreateESel( aCurSel ) );
 676         aImportInfo.aText = aText;
 677         pImpEditEngine->aImportHdl.Call( &aImportInfo );
 678     }
 679
 680     aCurSel = pImpEditEngine->ImpInsertText( aCurSel, aText );
 681     nLastAction = ACTION_INSERTTEXT;
 682 }
 683
 684 void EditHTMLParser::SkipGroup( int nEndToken )
 685 {
 686     // #69109# groups in cells are closed upon leaving the cell, because those
 687     // ******* web authors don't know their job
 688     // for example: <td><form></td>   lacks a closing </form>
 689     BYTE nCellLevel = nInCell;
 690     int nToken;
 691     while( nCellLevel <= nInCell && ( (nToken = GetNextToken() ) != nEndToken ) && nToken )
 692     {
 693         switch ( nToken )
 694         {
 695             case HTML_TABLEHEADER_ON:
 696             case HTML_TABLEDATA_ON:
 697                 nInCell++;
 698             break;
 699             case HTML_TABLEHEADER_OFF:
 700             case HTML_TABLEDATA_OFF:
 701                 if ( nInCell )
 702                     nInCell--;
 703             break;
 704         }
 705     }
 706 }
 707
 708 void EditHTMLParser::StartPara( BOOL bReal )
 709 {
 710     if ( bReal )
 711     {
 712         const HTMLOptions *_pOptions = GetOptions();
 713         USHORT nArrLen = _pOptions->Count();
 714         SvxAdjust eAdjust = SVX_ADJUST_LEFT;
 715         for ( USHORT i = 0; i < nArrLen; i++ )
 716         {
 717             const HTMLOption *pOption = (*_pOptions)[i];
 718             switch( pOption->GetToken() )
 719             {
 720                 case HTML_O_ALIGN:
 721                 {
 722                     if ( pOption->GetString().CompareIgnoreCaseToAscii( OOO_STRING_SVTOOLS_HTML_AL_right ) == COMPARE_EQUAL )
 723                         eAdjust = SVX_ADJUST_RIGHT;
 724                     else if ( pOption->GetString().CompareIgnoreCaseToAscii( OOO_STRING_SVTOOLS_HTML_AL_middle ) == COMPARE_EQUAL )
 725                         eAdjust = SVX_ADJUST_CENTER;
 726                     else if ( pOption->GetString().CompareIgnoreCaseToAscii( OOO_STRING_SVTOOLS_HTML_AL_center ) == COMPARE_EQUAL )
 727                         eAdjust = SVX_ADJUST_CENTER;
 728                     else
 729                         eAdjust = SVX_ADJUST_LEFT;
 730                 }
 731                 break;
 732             }
 733         }
 734         SfxItemSet aItemSet( pImpEditEngine->GetEmptyItemSet() );
 735         aItemSet.Put( SvxAdjustItem( eAdjust, EE_PARA_JUST ) );
 736         ImpSetAttribs( aItemSet );
 737     }
 738     bInPara = TRUE;
 739 }
 740
 741 void EditHTMLParser::EndPara( BOOL )
 742 {
 743     if ( bInPara )
 744     {
 745         BOOL bHasText = HasTextInCurrentPara();
 746         if ( bHasText )
 747             ImpInsertParaBreak();
 748         // Nur, wenn ohne Absatzabstaende gearbeitet wird...
 749 //              if ( !nInTable && bReal && (nNumberingLevel<=1) && (nBulletLevel<=1) )
 750 //                      ImpInsertParaBreak();
 751     }
 752     bInPara = FALSE;
 753 }
 754
 755 BOOL EditHTMLParser::ThrowAwayBlank()
 756 {
 757     // Ein Blank muss weggeschmissen werden, wenn der neue Text mit einem
 758     // Blank beginnt und der aktuelle Absatz leer ist oder mit einem
 759     // Blank endet...
 760     ContentNode* pNode = aCurSel.Max().GetNode();
 761     if ( pNode->Len() && ( pNode->GetChar( pNode->Len()-1 ) != ' ' ) )
 762         return FALSE;
 763     return TRUE;
 764 }
 765
 766 BOOL EditHTMLParser::HasTextInCurrentPara()
 767 {
 768     return aCurSel.Max().GetNode()->Len() ? TRUE : FALSE;
 769 }
 770
 771 void EditHTMLParser::AnchorStart()
 772 {
 773     // Anker im Anker ignoriern
 774     if ( !pCurAnchor )
 775     {
 776         const HTMLOptions* _pOptions = GetOptions();
 777         USHORT nArrLen = _pOptions->Count();
 778
 779         String aRef;
 780
 781         for ( USHORT i = 0; i < nArrLen; i++ )
 782         {
 783             const HTMLOption* pOption = (*_pOptions)[i];
 784             switch( pOption->GetToken() )
 785             {
 786                 case HTML_O_HREF:
 787                     aRef = pOption->GetString();
 788                 break;
 789             }
 790         }
 791
 792         if ( aRef.Len() )
 793         {
 794             String aURL = aRef;
 795             if ( aURL.Len() && ( aURL.GetChar( 0 ) != '#' ) )
 796             {
 797                 INetURLObject aTargetURL;
 798                 INetURLObject aRootURL( aBaseURL );
 799                 aRootURL.GetNewAbsURL( aRef, &aTargetURL );
 800                 aURL = aTargetURL.GetMainURL( INetURLObject::DECODE_TO_IURI );
 801             }
 802             pCurAnchor = new AnchorInfo;
 803             pCurAnchor->aHRef = aURL;
 804         }
 805     }
 806 }
 807
 808 void EditHTMLParser::AnchorEnd()
 809 {
 810     if ( pCurAnchor )
 811     {
 812         // Als URL-Feld einfuegen...
 813         SvxFieldItem aFld( SvxURLField( pCurAnchor->aHRef, pCurAnchor->aText, SVXURLFORMAT_REPR ), EE_FEATURE_FIELD  );
 814         aCurSel = pImpEditEngine->InsertField( aCurSel, aFld );
 815         bFieldsInserted = TRUE;
 816         delete pCurAnchor;
 817         pCurAnchor = 0;
 818
 819         if ( pImpEditEngine->aImportHdl.IsSet() )
 820         {
 821             ImportInfo aImportInfo( HTMLIMP_INSERTFIELD, this, pImpEditEngine->CreateESel( aCurSel ) );
 822             pImpEditEngine->aImportHdl.Call( &aImportInfo );
 823         }
 824     }
 825 }
 826
 827 void EditHTMLParser::HeadingStart( int nToken )
 828 {
 829     bWasInPara = bInPara;
 830     StartPara( FALSE );
 831
 832     if ( bWasInPara && HasTextInCurrentPara() )
 833         ImpInsertParaBreak();
 834
 835     USHORT nId = sal::static_int_cast< USHORT >(
 836         1 + ( ( nToken - HTML_HEAD1_ON ) / 2 ) );
 837     DBG_ASSERT( (nId >= 1) && (nId <= 9), "HeadingStart: ID kann nicht stimmen!" );
 838     ImpSetStyleSheet( nId );
 839 }
 840
 841 void EditHTMLParser::HeadingEnd( int )
 842 {
 843     EndPara( FALSE );
 844     ImpSetStyleSheet( 0 );
 845
 846     if ( bWasInPara )
 847     {
 848         bInPara = TRUE;
 849         bWasInPara = FALSE;
 850     }
 851 }