1 /************************************************************************* 2 * 3 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. 4 * 5 * Copyright 2000, 2010 Oracle and/or its affiliates. 6 * 7 * OpenOffice.org - a multi-platform office productivity suite 8 * 9 * This file is part of OpenOffice.org. 10 * 11 * OpenOffice.org is free software: you can redistribute it and/or modify 12 * it under the terms of the GNU Lesser General Public License version 3 13 * only, as published by the Free Software Foundation. 14 * 15 * OpenOffice.org is distributed in the hope that it will be useful, 16 * but WITHOUT ANY WARRANTY; without even the implied warranty of 17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 18 * GNU Lesser General Public License version 3 for more details 19 * (a copy is included in the LICENSE file that accompanied this code). 20 * 21 * You should have received a copy of the GNU Lesser General Public License 22 * version 3 along with OpenOffice.org. If not, see 23 * <http://www.openoffice.org/license.html> 24 * for a copy of the LGPLv3 License. 25 * 26 ************************************************************************/ 27 28 // MARKER(update_precomp.py): autogen include statement, do not remove 29 #include "precompiled_sfx2.hxx" 30 31 #include <tools/urlobj.hxx> 32 33 #include <sfx2/objsh.hxx> 34 #include <sfx2/docfile.hxx> 35 #include "openflag.hxx" 36 37 #include <svtools/htmlkywd.hxx> 38 #include <svtools/htmltokn.h> 39 #include <svtools/imap.hxx> 40 #include <svtools/imapcirc.hxx> 41 #include <svtools/imapobj.hxx> 42 #include <svtools/imappoly.hxx> 43 #include <svtools/imaprect.hxx> 44 #ifndef _SVSTDARR_ULONGS_DECL 45 #define _SVSTDARR_ULONGS 46 #include <svl/svstdarr.hxx> 47 #endif 48 #include <svl/zforlist.hxx> 49 #include <rtl/tencinfo.h> 50 #include <tools/tenccvt.hxx> 51 52 #include <sfx2/sfxhtml.hxx> 53 54 #include <com/sun/star/beans/XPropertyContainer.hpp> 55 56 57 using namespace ::com::sun::star; 58 59 60 sal_Char __FAR_DATA sHTML_MIME_text[] = "text/"; 61 sal_Char __FAR_DATA sHTML_MIME_application[] = "application/"; 62 sal_Char __FAR_DATA sHTML_MIME_experimental[] = "x-"; 63 64 // <INPUT TYPE=xxx> 65 static HTMLOptionEnum __READONLY_DATA aAreaShapeOptEnums[] = 66 { 67 { OOO_STRING_SVTOOLS_HTML_SH_rect, IMAP_OBJ_RECTANGLE }, 68 { OOO_STRING_SVTOOLS_HTML_SH_rectangle, IMAP_OBJ_RECTANGLE }, 69 { OOO_STRING_SVTOOLS_HTML_SH_circ, IMAP_OBJ_CIRCLE }, 70 { OOO_STRING_SVTOOLS_HTML_SH_circle, IMAP_OBJ_CIRCLE }, 71 { OOO_STRING_SVTOOLS_HTML_SH_poly, IMAP_OBJ_POLYGON }, 72 { OOO_STRING_SVTOOLS_HTML_SH_polygon, IMAP_OBJ_POLYGON }, 73 { 0, 0 } 74 }; 75 76 SfxHTMLParser::SfxHTMLParser( SvStream& rStream, sal_Bool bIsNewDoc, 77 SfxMedium *pMed ) : 78 HTMLParser( rStream, bIsNewDoc ), 79 pMedium( pMed ), pDLMedium( 0 ), 80 nMetaTags( 0 ) 81 { 82 DBG_ASSERT( RTL_TEXTENCODING_DONTKNOW == GetSrcEncoding( ), 83 "SfxHTMLParser::SfxHTMLParser: Wo kommt der ZS her?" ); 84 DBG_ASSERT( !IsSwitchToUCS2(), 85 "SfxHTMLParser::SfxHTMLParser: Switch to UCS2?" ); 86 87 // Altough the real default encoding is ISO8859-1, we use MS-1252 88 // als default encoding. 89 SetSrcEncoding( GetExtendedCompatibilityTextEncoding( RTL_TEXTENCODING_ISO_8859_1 ) ); 90 91 // If the file starts with a BOM, switch to UCS2. 92 SetSwitchToUCS2( sal_True ); 93 } 94 95 __EXPORT SfxHTMLParser::~SfxHTMLParser() 96 { 97 DBG_ASSERT( !pDLMedium, "Da ist ein File-Download stehengeblieben" ); 98 delete pDLMedium; 99 } 100 101 sal_Bool SfxHTMLParser::ParseMapOptions(ImageMap * pImageMap, 102 const HTMLOptions * pOptions) 103 { 104 DBG_ASSERT( pImageMap, "ParseMapOptions: keine Image-Map" ); 105 DBG_ASSERT( pOptions, "ParseMapOptions: keine Optionen" ); 106 107 String aName; 108 109 for( sal_uInt16 i=pOptions->Count(); i; ) 110 { 111 const HTMLOption *pOption = (*pOptions)[--i]; 112 switch( pOption->GetToken() ) 113 { 114 case HTML_O_NAME: 115 aName = pOption->GetString(); 116 break; 117 } 118 } 119 120 if( aName.Len() ) 121 pImageMap->SetName( aName ); 122 123 return aName.Len() > 0; 124 } 125 126 sal_Bool SfxHTMLParser::ParseAreaOptions(ImageMap * pImageMap, const String& rBaseURL, 127 const HTMLOptions * pOptions, 128 sal_uInt16 nEventMouseOver, 129 sal_uInt16 nEventMouseOut ) 130 { 131 DBG_ASSERT( pImageMap, "ParseAreaOptions: keine Image-Map" ); 132 DBG_ASSERT( pOptions, "ParseAreaOptions: keine Optionen" ); 133 134 sal_uInt16 nShape = IMAP_OBJ_RECTANGLE; 135 SvULongs aCoords; 136 String aName, aHRef, aAlt, aTarget, sEmpty; 137 sal_Bool bNoHRef = sal_False; 138 SvxMacroTableDtor aMacroTbl; 139 140 for( sal_uInt16 i=pOptions->Count(); i; ) 141 { 142 sal_uInt16 nEvent = 0; 143 ScriptType eScrpType = STARBASIC; 144 const HTMLOption *pOption = (*pOptions)[--i]; 145 switch( pOption->GetToken() ) 146 { 147 case HTML_O_NAME: 148 aName = pOption->GetString(); 149 break; 150 case HTML_O_SHAPE: 151 pOption->GetEnum( nShape, aAreaShapeOptEnums ); 152 break; 153 case HTML_O_COORDS: 154 pOption->GetNumbers( aCoords, sal_True ); 155 break; 156 case HTML_O_HREF: 157 aHRef = INetURLObject::GetAbsURL( rBaseURL, pOption->GetString() ); 158 break; 159 case HTML_O_NOHREF: 160 bNoHRef = sal_True; 161 break; 162 case HTML_O_ALT: 163 aAlt = pOption->GetString(); 164 break; 165 case HTML_O_TARGET: 166 aTarget = pOption->GetString(); 167 break; 168 169 case HTML_O_ONMOUSEOVER: 170 eScrpType = JAVASCRIPT; 171 case HTML_O_SDONMOUSEOVER: 172 nEvent = nEventMouseOver; 173 goto IMAPOBJ_SETEVENT; 174 175 case HTML_O_ONMOUSEOUT: 176 eScrpType = JAVASCRIPT; 177 case HTML_O_SDONMOUSEOUT: 178 nEvent = nEventMouseOut; 179 goto IMAPOBJ_SETEVENT; 180 IMAPOBJ_SETEVENT: 181 if( nEvent ) 182 { 183 String sTmp( pOption->GetString() ); 184 if( sTmp.Len() ) 185 { 186 sTmp.ConvertLineEnd(); 187 aMacroTbl.Insert( nEvent, 188 new SvxMacro( sTmp, sEmpty, eScrpType )); 189 } 190 } 191 break; 192 } 193 } 194 195 if( bNoHRef ) 196 aHRef.Erase(); 197 198 sal_Bool bNewArea = sal_True; 199 switch( nShape ) 200 { 201 case IMAP_OBJ_RECTANGLE: 202 if( aCoords.Count() >=4 ) 203 { 204 Rectangle aRec( aCoords[0], aCoords[1], 205 aCoords[2], aCoords[3] ); 206 IMapRectangleObject aMapRObj( aRec, aHRef, aAlt, String(), aTarget, aName, 207 !bNoHRef ); 208 if( aMacroTbl.Count() ) 209 aMapRObj.SetMacroTable( aMacroTbl ); 210 pImageMap->InsertIMapObject( aMapRObj ); 211 } 212 break; 213 case IMAP_OBJ_CIRCLE: 214 if( aCoords.Count() >=3 ) 215 { 216 Point aPoint( aCoords[0], aCoords[1] ); 217 IMapCircleObject aMapCObj( aPoint, aCoords[2],aHRef, aAlt, String(), 218 aTarget, aName, !bNoHRef ); 219 if( aMacroTbl.Count() ) 220 aMapCObj.SetMacroTable( aMacroTbl ); 221 pImageMap->InsertIMapObject( aMapCObj ); 222 } 223 break; 224 case IMAP_OBJ_POLYGON: 225 if( aCoords.Count() >=6 ) 226 { 227 sal_uInt16 nCount = aCoords.Count() / 2; 228 Polygon aPoly( nCount ); 229 for( sal_uInt16 i=0; i<nCount; i++ ) 230 aPoly[i] = Point( aCoords[2*i], aCoords[2*i+1] ); 231 IMapPolygonObject aMapPObj( aPoly, aHRef, aAlt, String(), aTarget, aName, 232 !bNoHRef ); 233 if( aMacroTbl.Count() ) 234 aMapPObj.SetMacroTable( aMacroTbl ); 235 pImageMap->InsertIMapObject( aMapPObj ); 236 } 237 break; 238 default: 239 bNewArea = sal_False; 240 } 241 242 return bNewArea; 243 } 244 245 246 void SfxHTMLParser::StartFileDownload( const String& rURL, int nToken, 247 SfxObjectShell *pSh ) 248 { 249 DBG_ASSERT( !pDLMedium, "StartFileDwonload bei aktivem Download" ); 250 if( pDLMedium ) 251 return; 252 253 pDLMedium = new SfxMedium( rURL, SFX_STREAM_READONLY, sal_False ); 254 if( pSh ) 255 { 256 // Medium registrieren, damit abgebrochen werden kann 257 pSh->RegisterTransfer( *pDLMedium ); 258 259 // Target-Frame uebertragen, damit auch javascript:-URLs 260 // "geladen" werden koennen. 261 //const SfxMedium *pShMedium = pSh->GetMedium(); 262 //if( pShMedium ) 263 // pDLMedium->SetLoadTargetFrame( pShMedium->GetLoadTargetFrame() ); 264 } 265 266 // Download anstossen (Achtung: Kann auch synchron sein). 267 if ( sal_True /*pMedium->GetDoneLink() == Link()*/ ) 268 pDLMedium->DownLoad(); 269 else 270 { 271 // Downloading-Flag auf sal_True setzen. Es werden dann auch 272 // Data-Available-Links, wenn wir in den Pending-Staus gelangen. 273 SetDownloadingFile( sal_True ); 274 pDLMedium->DownLoad( STATIC_LINK( this, SfxHTMLParser, FileDownloadDone ) ); 275 276 // Wenn das Dowsnloading-Flag noch gesetzt ist erfolgt der Download 277 // asynchron. Wir gehen dann in den Pedning-Staus und warten dort. 278 // Solange sind alle Aufrufe des Data-Avaialble-Link gesperrt. 279 if( IsDownloadingFile() ) 280 { 281 // Den aktuellen Zustand einfrieren und in den Pending-Status gehen. 282 // Wenn der Download beendet oder abgebrochen wurde, wird ueber 283 // NewDataRead ein Continue mit dem uebergeben Token angesteossen. 284 SaveState( nToken ); 285 eState = SVPAR_PENDING; 286 } 287 } 288 } 289 290 sal_Bool SfxHTMLParser::GetFileDownloadMIME( String& rMIME ) 291 { 292 return pDLMedium && pDLMedium->GetErrorCode()==0 && 293 pDLMedium->GetMIMEAndRedirect(rMIME)==0; 294 } 295 296 sal_Bool SfxHTMLParser::FinishFileDownload( String& rStr ) 297 { 298 String aStr; 299 300 sal_Bool bOK = pDLMedium && pDLMedium->GetErrorCode()==0; 301 if( bOK ) 302 { 303 SvStream* pStream = pDLMedium->GetInStream(); 304 DBG_ASSERT( pStream, "Kein In-Stream vom Medium erhalten" ); 305 306 SvMemoryStream aStream; 307 if( pStream ) // HACK wegen #65563# 308 aStream << *pStream; 309 310 aStream.Seek( STREAM_SEEK_TO_END ); 311 DBG_ASSERT( aStream.Tell() < STRING_MAXLEN, 312 "File zu lang fuer einen String, Ende abgeschnitten" ); 313 xub_StrLen nLen = aStream.Tell() < STRING_MAXLEN 314 ? (xub_StrLen)aStream.Tell() 315 : STRING_MAXLEN; 316 317 // TODO: untested!!! 318 rtl_TextEncoding eEnc = 319 GetExtendedCompatibilityTextEncoding( RTL_TEXTENCODING_ISO_8859_1 ); 320 String sMime; 321 if( pDLMedium->GetMIMEAndRedirect( sMime ) == 0 ) 322 { 323 rtl_TextEncoding eMimeEnc = GetEncodingByMIME( sMime ); 324 if( RTL_TEXTENCODING_DONTKNOW != eMimeEnc ) 325 eEnc = eMimeEnc; 326 } 327 328 ByteString sBuffer; 329 sal_Char* pBuffer = sBuffer.AllocBuffer(nLen); 330 aStream.Seek( 0 ); 331 aStream.Read((void*)pBuffer, nLen); 332 rStr = String( S2U(pBuffer) ); 333 } 334 335 delete pDLMedium; 336 pDLMedium = 0; 337 338 return bOK; 339 } 340 341 IMPL_STATIC_LINK( SfxHTMLParser, FileDownloadDone, void*, EMPTYARG ) 342 { 343 // Der Download ist jetzt abgeschlossen. Ausserdem muss/darf der 344 // Data-Available-Link wieder durchgelassen werden. 345 pThis->SetDownloadingFile( sal_False ); 346 347 // ... und einmal aufrufen, damit weitergelesen wird. 348 pThis->CallAsyncCallLink(); 349 350 return 0; 351 } 352 353 void SfxHTMLParser::GetScriptType_Impl( SvKeyValueIterator *pHTTPHeader ) 354 { 355 aScriptType = DEFINE_CONST_UNICODE(SVX_MACRO_LANGUAGE_JAVASCRIPT); 356 eScriptType = JAVASCRIPT; 357 if( pHTTPHeader ) 358 { 359 SvKeyValue aKV; 360 for( sal_Bool bCont = pHTTPHeader->GetFirst( aKV ); bCont; 361 bCont = pHTTPHeader->GetNext( aKV ) ) 362 { 363 if( aKV.GetKey().EqualsIgnoreCaseAscii( 364 OOO_STRING_SVTOOLS_HTML_META_content_script_type ) ) 365 { 366 if( aKV.GetValue().Len() ) 367 { 368 String aTmp( aKV.GetValue() ); 369 if( aTmp.EqualsIgnoreCaseAscii( sHTML_MIME_text, 0, 5 ) ) 370 aTmp.Erase( 0, 5 ); 371 else if( aTmp.EqualsIgnoreCaseAscii( sHTML_MIME_application, 372 0, 12 ) ) 373 aTmp.Erase( 0, 12 ); 374 else 375 break; 376 377 if( aTmp.EqualsIgnoreCaseAscii( sHTML_MIME_experimental, 0, 378 2 ) ) 379 { 380 aTmp.Erase( 0, 2 ); 381 } 382 383 if( aTmp.EqualsIgnoreCaseAscii( OOO_STRING_SVTOOLS_HTML_LG_starbasic ) ) 384 { 385 eScriptType = STARBASIC; 386 aScriptType = DEFINE_CONST_UNICODE(SVX_MACRO_LANGUAGE_STARBASIC); 387 } 388 if( !aTmp.EqualsIgnoreCaseAscii( OOO_STRING_SVTOOLS_HTML_LG_javascript ) ) 389 { 390 eScriptType = EXTENDED_STYPE; 391 aScriptType = aTmp; 392 } 393 } 394 break; 395 } 396 } 397 } 398 } 399 400 ScriptType SfxHTMLParser::GetScriptType( SvKeyValueIterator *pHTTPHeader ) const 401 { 402 if( !aScriptType.Len() ) 403 ((SfxHTMLParser *)this)->GetScriptType_Impl( pHTTPHeader ); 404 405 return eScriptType; 406 } 407 408 const String& SfxHTMLParser::GetScriptTypeString( 409 SvKeyValueIterator *pHTTPHeader ) const 410 { 411 if( !aScriptType.Len() ) 412 ((SfxHTMLParser *)this)->GetScriptType_Impl( pHTTPHeader ); 413 414 return aScriptType; 415 } 416 417 double SfxHTMLParser::GetTableDataOptionsValNum( sal_uInt32& nNumForm, 418 LanguageType& eNumLang, const String& aValStr, const String& aNumStr, 419 SvNumberFormatter& rFormatter ) 420 { 421 LanguageType eParseLang = (LanguageType )aNumStr.ToInt32(); 422 sal_uInt32 nParseForm = 423 rFormatter.GetFormatForLanguageIfBuiltIn( 0, eParseLang ); 424 double fVal; 425 rFormatter.IsNumberFormat( aValStr, nParseForm, fVal ); 426 if ( aNumStr.GetTokenCount( ';' ) > 2 ) 427 { 428 eNumLang = (LanguageType)aNumStr.GetToken( 1, ';' ).ToInt32(); 429 xub_StrLen nPos = aNumStr.Search( ';' ); 430 nPos = aNumStr.Search( ';', nPos + 1 ); 431 String aFormat( aNumStr.Copy( nPos + 1 ) ); 432 xub_StrLen nCheckPos; 433 short nType; 434 if ( eNumLang != LANGUAGE_SYSTEM ) 435 rFormatter.PutEntry( aFormat, nCheckPos, nType, nNumForm, eNumLang ); 436 else 437 rFormatter.PutandConvertEntry( aFormat, nCheckPos, nType, nNumForm, 438 eParseLang, eNumLang ); 439 } 440 else 441 { 442 eNumLang = LANGUAGE_SYSTEM; 443 nNumForm = rFormatter.GetFormatForLanguageIfBuiltIn( 0, eNumLang ); 444 } 445 return fVal; 446 } 447 448