sfx2/source/bastyp/sfxhtml.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: sfxhtml.cxx,v $
  10  * $Revision: 1.23 $
  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_sfx2.hxx"
  33
  34 #include <tools/urlobj.hxx>
  35
  36 #include <sfx2/objsh.hxx>
  37 #include <sfx2/docfile.hxx>
  38 #include "openflag.hxx"
  39
  40 #include <svtools/htmlkywd.hxx>
  41 #include <svtools/htmltokn.h>
  42 #include <svtools/imap.hxx>
  43 #include <svtools/imapcirc.hxx>
  44 #include <svtools/imapobj.hxx>
  45 #include <svtools/imappoly.hxx>
  46 #include <svtools/imaprect.hxx>
  47 #ifndef _SVSTDARR_ULONGS_DECL
  48 #define _SVSTDARR_ULONGS
  49 #include <svtools/svstdarr.hxx>
  50 #endif
  51 #include <svtools/zforlist.hxx>
  52 #include <rtl/tencinfo.h>
  53 #include <tools/tenccvt.hxx>
  54
  55 #include <sfx2/sfxhtml.hxx>
  56
  57 #include <com/sun/star/beans/XPropertyContainer.hpp>
  58
  59
  60 using namespace ::com::sun::star;
  61
  62
  63 sal_Char __FAR_DATA sHTML_MIME_text[] = "text/";
  64 sal_Char __FAR_DATA sHTML_MIME_application[] = "application/";
  65 sal_Char __FAR_DATA sHTML_MIME_experimental[] = "x-";
  66
  67 // <INPUT TYPE=xxx>
  68 #ifdef __MINGW32__ // for runtime pseudo reloc
  69 static HTMLOptionEnum aAreaShapeOptEnums[] =
  70 #else
  71 static HTMLOptionEnum __READONLY_DATA aAreaShapeOptEnums[] =
  72 #endif
  73 {
  74     { OOO_STRING_SVTOOLS_HTML_SH_rect,          IMAP_OBJ_RECTANGLE      },
  75     { OOO_STRING_SVTOOLS_HTML_SH_rectangle,     IMAP_OBJ_RECTANGLE      },
  76     { OOO_STRING_SVTOOLS_HTML_SH_circ,          IMAP_OBJ_CIRCLE         },
  77     { OOO_STRING_SVTOOLS_HTML_SH_circle,        IMAP_OBJ_CIRCLE         },
  78     { OOO_STRING_SVTOOLS_HTML_SH_poly,                  IMAP_OBJ_POLYGON        },
  79     { OOO_STRING_SVTOOLS_HTML_SH_polygon,               IMAP_OBJ_POLYGON        },
  80     { 0,                                        0                                       }
  81 };
  82
  83 SfxHTMLParser::SfxHTMLParser( SvStream& rStream, BOOL bIsNewDoc,
  84                               SfxMedium *pMed ) :
  85     HTMLParser( rStream, bIsNewDoc ),
  86     pMedium( pMed ), pDLMedium( 0 ),
  87     nMetaTags( 0 )
  88 {
  89     DBG_ASSERT( RTL_TEXTENCODING_DONTKNOW == GetSrcEncoding( ),
  90                 "SfxHTMLParser::SfxHTMLParser: Wo kommt der ZS her?" );
  91     DBG_ASSERT( !IsSwitchToUCS2(),
  92                 "SfxHTMLParser::SfxHTMLParser: Switch to UCS2?" );
  93
  94     // Altough the real default encoding is ISO8859-1, we use MS-1252
  95     // als default encoding.
  96     SetSrcEncoding( GetExtendedCompatibilityTextEncoding(  RTL_TEXTENCODING_ISO_8859_1 ) );
  97
  98     // If the file starts with a BOM, switch to UCS2.
  99     SetSwitchToUCS2( TRUE );
 100 }
 101
 102 __EXPORT SfxHTMLParser::~SfxHTMLParser()
 103 {
 104     DBG_ASSERT( !pDLMedium, "Da ist ein File-Download stehengeblieben" );
 105     delete pDLMedium;
 106 }
 107
 108 BOOL SfxHTMLParser::ParseMapOptions(ImageMap * pImageMap,
 109                                     const HTMLOptions * pOptions)
 110 {
 111     DBG_ASSERT( pImageMap, "ParseMapOptions: keine Image-Map" );
 112     DBG_ASSERT( pOptions, "ParseMapOptions: keine Optionen" );
 113
 114     String aName;
 115
 116     for( USHORT i=pOptions->Count(); i; )
 117     {
 118         const HTMLOption *pOption = (*pOptions)[--i];
 119         switch( pOption->GetToken() )
 120         {
 121         case HTML_O_NAME:
 122             aName = pOption->GetString();
 123             break;
 124         }
 125     }
 126
 127     if( aName.Len() )
 128         pImageMap->SetName( aName );
 129
 130     return aName.Len() > 0;
 131 }
 132
 133 BOOL SfxHTMLParser::ParseAreaOptions(ImageMap * pImageMap, const String& rBaseURL,
 134                                      const HTMLOptions * pOptions,
 135                                      USHORT nEventMouseOver,
 136                                      USHORT nEventMouseOut )
 137 {
 138     DBG_ASSERT( pImageMap, "ParseAreaOptions: keine Image-Map" );
 139     DBG_ASSERT( pOptions, "ParseAreaOptions: keine Optionen" );
 140
 141     USHORT nShape = IMAP_OBJ_RECTANGLE;
 142     SvULongs aCoords;
 143     String aName, aHRef, aAlt, aTarget, sEmpty;
 144     BOOL bNoHRef = FALSE;
 145     SvxMacroTableDtor aMacroTbl;
 146
 147     for( USHORT i=pOptions->Count(); i; )
 148     {
 149         USHORT nEvent = 0;
 150         ScriptType eScrpType = STARBASIC;
 151         const HTMLOption *pOption = (*pOptions)[--i];
 152         switch( pOption->GetToken() )
 153         {
 154         case HTML_O_NAME:
 155             aName = pOption->GetString();
 156             break;
 157         case HTML_O_SHAPE:
 158             pOption->GetEnum( nShape, aAreaShapeOptEnums );
 159             break;
 160         case HTML_O_COORDS:
 161             pOption->GetNumbers( aCoords, TRUE );
 162             break;
 163         case HTML_O_HREF:
 164             aHRef = INetURLObject::GetAbsURL( rBaseURL, pOption->GetString() );
 165             break;
 166         case HTML_O_NOHREF:
 167             bNoHRef = TRUE;
 168             break;
 169         case HTML_O_ALT:
 170             aAlt = pOption->GetString();
 171             break;
 172         case HTML_O_TARGET:
 173             aTarget = pOption->GetString();
 174             break;
 175
 176         case HTML_O_ONMOUSEOVER:
 177             eScrpType = JAVASCRIPT;
 178         case HTML_O_SDONMOUSEOVER:
 179             nEvent = nEventMouseOver;
 180             goto IMAPOBJ_SETEVENT;
 181
 182         case HTML_O_ONMOUSEOUT:
 183             eScrpType = JAVASCRIPT;
 184         case HTML_O_SDONMOUSEOUT:
 185             nEvent = nEventMouseOut;
 186             goto IMAPOBJ_SETEVENT;
 187 IMAPOBJ_SETEVENT:
 188             if( nEvent )
 189             {
 190                 String sTmp( pOption->GetString() );
 191                 if( sTmp.Len() )
 192                 {
 193                     sTmp.ConvertLineEnd();
 194                     aMacroTbl.Insert( nEvent,
 195                         new SvxMacro( sTmp, sEmpty, eScrpType ));
 196                 }
 197             }
 198             break;
 199         }
 200     }
 201
 202     if( bNoHRef )
 203         aHRef.Erase();
 204
 205     BOOL bNewArea = TRUE;
 206     switch( nShape )
 207     {
 208     case IMAP_OBJ_RECTANGLE:
 209         if( aCoords.Count() >=4 )
 210         {
 211             Rectangle aRec( aCoords[0], aCoords[1],
 212                             aCoords[2], aCoords[3] );
 213             IMapRectangleObject aMapRObj( aRec, aHRef, aAlt, String(), aTarget, aName,
 214                                           !bNoHRef );
 215             if( aMacroTbl.Count() )
 216                 aMapRObj.SetMacroTable( aMacroTbl );
 217             pImageMap->InsertIMapObject( aMapRObj );
 218         }
 219         break;
 220     case IMAP_OBJ_CIRCLE:
 221         if( aCoords.Count() >=3 )
 222         {
 223             Point aPoint( aCoords[0], aCoords[1] );
 224             IMapCircleObject aMapCObj( aPoint, aCoords[2],aHRef, aAlt, String(),
 225                                        aTarget, aName, !bNoHRef );
 226             if( aMacroTbl.Count() )
 227                 aMapCObj.SetMacroTable( aMacroTbl );
 228             pImageMap->InsertIMapObject( aMapCObj );
 229         }
 230         break;
 231     case IMAP_OBJ_POLYGON:
 232         if( aCoords.Count() >=6 )
 233         {
 234             USHORT nCount = aCoords.Count() / 2;
 235             Polygon aPoly( nCount );
 236             for( USHORT i=0; i<nCount; i++ )
 237                 aPoly[i] = Point( aCoords[2*i], aCoords[2*i+1] );
 238             IMapPolygonObject aMapPObj( aPoly, aHRef, aAlt, String(), aTarget, aName,
 239                                         !bNoHRef );
 240             if( aMacroTbl.Count() )
 241                 aMapPObj.SetMacroTable( aMacroTbl );
 242             pImageMap->InsertIMapObject( aMapPObj );
 243         }
 244         break;
 245     default:
 246         bNewArea = FALSE;
 247     }
 248
 249     return bNewArea;
 250 }
 251
 252
 253 void SfxHTMLParser::StartFileDownload( const String& rURL, int nToken,
 254                                        SfxObjectShell *pSh )
 255 {
 256     DBG_ASSERT( !pDLMedium, "StartFileDwonload bei aktivem Download" );
 257     if( pDLMedium )
 258         return;
 259
 260     pDLMedium = new SfxMedium( rURL, SFX_STREAM_READONLY, FALSE );
 261     if( pSh )
 262     {
 263         // Medium registrieren, damit abgebrochen werden kann
 264         pSh->RegisterTransfer( *pDLMedium );
 265
 266         // Target-Frame uebertragen, damit auch javascript:-URLs
 267         // "geladen" werden koennen.
 268         //const SfxMedium *pShMedium = pSh->GetMedium();
 269         //if( pShMedium )
 270         //      pDLMedium->SetLoadTargetFrame( pShMedium->GetLoadTargetFrame() );
 271     }
 272
 273     // Download anstossen (Achtung: Kann auch synchron sein).
 274     if ( TRUE /*pMedium->GetDoneLink() == Link()*/ )
 275         pDLMedium->DownLoad();
 276     else
 277     {
 278         // Downloading-Flag auf TRUE setzen. Es werden dann auch
 279         // Data-Available-Links, wenn wir in den Pending-Staus gelangen.
 280         SetDownloadingFile( TRUE );
 281         pDLMedium->DownLoad( STATIC_LINK( this, SfxHTMLParser, FileDownloadDone ) );
 282
 283         // Wenn das Dowsnloading-Flag noch gesetzt ist erfolgt der Download
 284         // asynchron. Wir gehen dann in den Pedning-Staus und warten dort.
 285         // Solange sind alle Aufrufe des Data-Avaialble-Link gesperrt.
 286         if( IsDownloadingFile() )
 287         {
 288             // Den aktuellen Zustand einfrieren und in den Pending-Status gehen.
 289             // Wenn der Download beendet oder abgebrochen wurde, wird ueber
 290             // NewDataRead ein Continue mit dem uebergeben Token angesteossen.
 291             SaveState( nToken );
 292             eState = SVPAR_PENDING;
 293         }
 294     }
 295 }
 296
 297 BOOL SfxHTMLParser::GetFileDownloadMIME( String& rMIME )
 298 {
 299     return pDLMedium && pDLMedium->GetErrorCode()==0 &&
 300            pDLMedium->GetMIMEAndRedirect(rMIME)==0;
 301 }
 302
 303 BOOL SfxHTMLParser::FinishFileDownload( String& rStr )
 304 {
 305     String aStr;
 306
 307     BOOL bOK = pDLMedium && pDLMedium->GetErrorCode()==0;
 308     if( bOK )
 309     {
 310         SvStream* pStream = pDLMedium->GetInStream();
 311         DBG_ASSERT( pStream, "Kein In-Stream vom Medium erhalten" );
 312
 313         SvMemoryStream aStream;
 314         if( pStream )   // HACK wegen #65563#
 315             aStream << *pStream;
 316
 317         aStream.Seek( STREAM_SEEK_TO_END );
 318         DBG_ASSERT( aStream.Tell() < STRING_MAXLEN,
 319                     "File zu lang fuer einen String, Ende abgeschnitten" );
 320         xub_StrLen nLen = aStream.Tell() < STRING_MAXLEN
 321                         ? (xub_StrLen)aStream.Tell()
 322                         : STRING_MAXLEN;
 323
 324         // TODO: untested!!!
 325         rtl_TextEncoding eEnc =
 326             GetExtendedCompatibilityTextEncoding( RTL_TEXTENCODING_ISO_8859_1 );
 327         String sMime;
 328         if( pDLMedium->GetMIMEAndRedirect( sMime ) == 0 )
 329         {
 330             rtl_TextEncoding eMimeEnc = GetEncodingByMIME( sMime );
 331             if( RTL_TEXTENCODING_DONTKNOW != eMimeEnc )
 332                 eEnc = eMimeEnc;
 333         }
 334
 335         ByteString sBuffer;
 336         sal_Char* pBuffer = sBuffer.AllocBuffer(nLen);
 337         aStream.Seek( 0 );
 338         aStream.Read((void*)pBuffer, nLen);
 339         rStr = String( S2U(pBuffer) );
 340     }
 341
 342     delete pDLMedium;
 343     pDLMedium = 0;
 344
 345     return bOK;
 346 }
 347
 348 IMPL_STATIC_LINK( SfxHTMLParser, FileDownloadDone, void*, EMPTYARG )
 349 {
 350     // Der Download ist jetzt abgeschlossen. Ausserdem muss/darf der
 351     // Data-Available-Link wieder durchgelassen werden.
 352     pThis->SetDownloadingFile( FALSE );
 353
 354     // ... und einmal aufrufen, damit weitergelesen wird.
 355     pThis->CallAsyncCallLink();
 356
 357     return 0;
 358 }
 359
 360 rtl_TextEncoding SfxHTMLParser::GetEncodingByHttpHeader( SvKeyValueIterator *pHTTPHeader )
 361 {
 362     rtl_TextEncoding eRet = RTL_TEXTENCODING_DONTKNOW;
 363     if( pHTTPHeader )
 364     {
 365         SvKeyValue aKV;
 366         for( BOOL bCont = pHTTPHeader->GetFirst( aKV ); bCont;
 367              bCont = pHTTPHeader->GetNext( aKV ) )
 368         {
 369             if( aKV.GetKey().EqualsIgnoreCaseAscii( OOO_STRING_SVTOOLS_HTML_META_content_type ) )
 370             {
 371                 if( aKV.GetValue().Len() )
 372                 {
 373                     eRet = SfxHTMLParser::GetEncodingByMIME( aKV.GetValue() );
 374                 }
 375             }
 376         }
 377     }
 378     return eRet;
 379 }
 380
 381 BOOL SfxHTMLParser::SetEncodingByHTTPHeader(
 382                                 SvKeyValueIterator *pHTTPHeader )
 383 {
 384     BOOL bRet = FALSE;
 385     rtl_TextEncoding eEnc = SfxHTMLParser::GetEncodingByHttpHeader( pHTTPHeader );
 386     if(RTL_TEXTENCODING_DONTKNOW != eEnc)
 387     {
 388         SetSrcEncoding( eEnc );
 389         bRet = TRUE;
 390     }
 391     return bRet;
 392 }
 393
 394 void SfxHTMLParser::GetScriptType_Impl( SvKeyValueIterator *pHTTPHeader )
 395 {
 396     aScriptType = DEFINE_CONST_UNICODE(SVX_MACRO_LANGUAGE_JAVASCRIPT);
 397     eScriptType = JAVASCRIPT;
 398     if( pHTTPHeader )
 399     {
 400         SvKeyValue aKV;
 401         for( BOOL bCont = pHTTPHeader->GetFirst( aKV ); bCont;
 402              bCont = pHTTPHeader->GetNext( aKV ) )
 403         {
 404             if( aKV.GetKey().EqualsIgnoreCaseAscii(
 405                                     OOO_STRING_SVTOOLS_HTML_META_content_script_type ) )
 406             {
 407                 if( aKV.GetValue().Len() )
 408                 {
 409                     String aTmp( aKV.GetValue() );
 410                     if( aTmp.EqualsIgnoreCaseAscii( sHTML_MIME_text, 0, 5 ) )
 411                         aTmp.Erase( 0, 5 );
 412                     else if( aTmp.EqualsIgnoreCaseAscii( sHTML_MIME_application,
 413                                                          0, 12 ) )
 414                         aTmp.Erase( 0, 12 );
 415                     else
 416                         break;
 417
 418                     if( aTmp.EqualsIgnoreCaseAscii( sHTML_MIME_experimental, 0,
 419                                                     2 ) )
 420                     {
 421                         aTmp.Erase( 0, 2 );
 422                     }
 423
 424                     if( aTmp.EqualsIgnoreCaseAscii( OOO_STRING_SVTOOLS_HTML_LG_starbasic ) )
 425                     {
 426                         eScriptType = STARBASIC;
 427                         aScriptType = DEFINE_CONST_UNICODE(SVX_MACRO_LANGUAGE_STARBASIC);
 428                     }
 429                     if( !aTmp.EqualsIgnoreCaseAscii( OOO_STRING_SVTOOLS_HTML_LG_javascript ) )
 430                     {
 431                         eScriptType = EXTENDED_STYPE;
 432                         aScriptType = aTmp;
 433                     }
 434                 }
 435                 break;
 436             }
 437         }
 438     }
 439 }
 440
 441 ScriptType SfxHTMLParser::GetScriptType( SvKeyValueIterator *pHTTPHeader ) const
 442 {
 443     if( !aScriptType.Len() )
 444         ((SfxHTMLParser *)this)->GetScriptType_Impl( pHTTPHeader );
 445
 446     return eScriptType;
 447 }
 448
 449 const String& SfxHTMLParser::GetScriptTypeString(
 450                                     SvKeyValueIterator *pHTTPHeader ) const
 451 {
 452     if( !aScriptType.Len() )
 453         ((SfxHTMLParser *)this)->GetScriptType_Impl( pHTTPHeader );
 454
 455     return aScriptType;
 456 }
 457
 458 double SfxHTMLParser::GetTableDataOptionsValNum( sal_uInt32& nNumForm,
 459         LanguageType& eNumLang, const String& aValStr, const String& aNumStr,
 460         SvNumberFormatter& rFormatter )
 461 {
 462     LanguageType eParseLang = (LanguageType )aNumStr.ToInt32();
 463     sal_uInt32 nParseForm =
 464         rFormatter.GetFormatForLanguageIfBuiltIn( 0, eParseLang );
 465     double fVal;
 466     rFormatter.IsNumberFormat( aValStr, nParseForm, fVal );
 467     if ( aNumStr.GetTokenCount( ';' ) > 2 )
 468     {
 469         eNumLang = (LanguageType)aNumStr.GetToken( 1, ';' ).ToInt32();
 470         xub_StrLen nPos = aNumStr.Search( ';' );
 471         nPos = aNumStr.Search( ';', nPos + 1 );
 472         String aFormat( aNumStr.Copy( nPos + 1 ) );
 473         xub_StrLen nCheckPos;
 474         short nType;
 475         if ( eNumLang != LANGUAGE_SYSTEM )
 476             rFormatter.PutEntry( aFormat, nCheckPos, nType, nNumForm, eNumLang );
 477         else
 478             rFormatter.PutandConvertEntry( aFormat, nCheckPos, nType, nNumForm,
 479                 eParseLang, eNumLang );
 480     }
 481     else
 482     {
 483         eNumLang = LANGUAGE_SYSTEM;
 484         nNumForm = rFormatter.GetFormatForLanguageIfBuiltIn( 0, eNumLang );
 485     }
 486     return fVal;
 487 }
 488