transex3/inc/tagtest.hxx

   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: tagtest.hxx,v $
  10  * $Revision: 1.12 $
  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 #ifndef _TAGTEST_HXX_
  32 #define _TAGTEST_HXX_
  33
  34 #include <tools/string.hxx>
  35 #include <tools/list.hxx>
  36 #include <hash_map> /* std::hashmap*/
  37
  38 class GSILine;
  39
  40 typedef USHORT TokenId;
  41
  42 #define TOK_INVALIDPOS  USHORT( 0xFFFF )
  43
  44 class ParserMessage;
  45
  46 DECLARE_LIST( Impl_ParserMessageList, ParserMessage* )
  47 class ParserMessageList;
  48
  49
  50 struct equalByteString{
  51         bool operator()( const ByteString& rKey1, const ByteString& rKey2 ) const {
  52             return rKey1.CompareTo( rKey2 )==COMPARE_EQUAL;
  53     }
  54 };
  55 struct lessByteString{
  56         bool operator()( const ByteString& rKey1, const ByteString& rKey2 ) const {
  57             return rKey1.CompareTo( rKey2 )==COMPARE_LESS;
  58     }
  59 };
  60
  61 struct hashByteString{
  62     size_t operator()( const ByteString& rName ) const{
  63                 std::hash< const char* > myHash;
  64                 return myHash( rName.GetBuffer() );
  65     }
  66 };
  67
  68
  69
  70 typedef std::hash_map<ByteString , String , hashByteString,equalByteString>
  71                                 StringHashMap;
  72
  73 class TokenInfo
  74 {
  75 private:
  76     void SplitTag( ParserMessageList &rErrorList );
  77
  78     String aTagName;
  79     StringHashMap aProperties;
  80     BOOL bClosed;    // tag is closed  <sdnf/>
  81     BOOL bCloseTag;  // tag is close Tag  </sdnf>
  82
  83
  84     BOOL bIsBroken;
  85     BOOL bHasBeenFixed;
  86     BOOL bDone;
  87
  88 public:
  89
  90     String aTokenString;
  91     TokenId nId;
  92     USHORT nPos;            // Position in String
  93
  94     TokenInfo():bClosed(FALSE),bCloseTag(FALSE),bIsBroken(FALSE),bHasBeenFixed(FALSE),bDone(FALSE),nId( 0 ){;}
  95 explicit    TokenInfo( TokenId pnId, USHORT nP ):bClosed(FALSE),bCloseTag(FALSE),bIsBroken(FALSE),bHasBeenFixed(FALSE),bDone(FALSE),nId( pnId ),nPos(nP){;}
  96 explicit    TokenInfo( TokenId pnId, USHORT nP, String paStr ):bClosed(FALSE),bCloseTag(FALSE),bIsBroken(FALSE),bHasBeenFixed(FALSE),bDone(FALSE),aTokenString( paStr ),nId( pnId ),nPos(nP) {;}
  97 explicit    TokenInfo( TokenId pnId, USHORT nP, String paStr, ParserMessageList &rErrorList );
  98
  99     String GetTagName() const;
 100
 101     String MakeTag() const;
 102
 103     /**
 104         Is the property to be ignored or does it have the default value anyways
 105     **/
 106     BOOL IsPropertyRelevant( const ByteString &aName, const String &aValue ) const;
 107     BOOL IsPropertyValueValid( const ByteString &aName, const String &aValue ) const;
 108     /**
 109         Does the property contain the same value for all languages
 110         e.g.: the href in a link tag
 111     **/
 112     BOOL IsPropertyInvariant( const ByteString &aName, const String &aValue ) const;
 113     /**
 114         a subset of IsPropertyInvariant but containing only those that are fixable
 115         we dont wat to fix e.g.: ahelp :: visibility
 116     **/
 117     BOOL IsPropertyFixable( const ByteString &aName ) const;
 118     BOOL MatchesTranslation( TokenInfo& rInfo, BOOL bGenErrors, ParserMessageList &rErrorList, BOOL bFixTags = FALSE ) const;
 119
 120     BOOL IsDone() const { return bDone; }
 121     void SetDone( BOOL bNew = TRUE ) { bDone = bNew; }
 122
 123     BOOL HasBeenFixed() const { return bHasBeenFixed; }
 124     void SetHasBeenFixed( BOOL bNew = TRUE ) { bHasBeenFixed = bNew; }
 125 };
 126
 127
 128 class ParserMessageList : public Impl_ParserMessageList
 129 {
 130 public:
 131     void AddError( USHORT nErrorNr, ByteString aErrorText, const TokenInfo &rTag );
 132     void AddWarning( USHORT nErrorNr, ByteString aErrorText, const TokenInfo &rTag );
 133
 134     BOOL HasErrors();
 135 };
 136
 137
 138 #define TAG_GROUPMASK                           0xF000
 139 #define TAG_GROUPSHIFT                          12
 140
 141 #define TAG_GROUP( nTag )                       (( nTag & TAG_GROUPMASK ) >> TAG_GROUPSHIFT )
 142 #define TAG_NOGROUP( nTag )                     ( nTag & ~TAG_GROUPMASK )       // ~ = Bitweises NOT
 143
 144 #define TAG_NOMORETAGS                          0x0
 145
 146 #define TAG_GROUP_FORMAT                        0x1
 147 #define TAG_ON                                          0x100
 148 #define TAG_BOLDON                                      ( TAG_GROUP_FORMAT << TAG_GROUPSHIFT | TAG_ON | 0x001 )
 149 #define TAG_BOLDOFF                                     ( TAG_GROUP_FORMAT << TAG_GROUPSHIFT |          0x001 )
 150 #define TAG_ITALICON                            ( TAG_GROUP_FORMAT << TAG_GROUPSHIFT | TAG_ON | 0x002 )
 151 #define TAG_ITALICOFF                           ( TAG_GROUP_FORMAT << TAG_GROUPSHIFT |          0x002 )
 152 #define TAG_UNDERLINEON                         ( TAG_GROUP_FORMAT << TAG_GROUPSHIFT | TAG_ON | 0x004 )
 153 #define TAG_UNDERLINEOFF                        ( TAG_GROUP_FORMAT << TAG_GROUPSHIFT |          0x004 )
 154
 155 #define TAG_GROUP_NOTALLOWED            0x2
 156 #define TAG_HELPID                                      ( TAG_GROUP_NOTALLOWED << TAG_GROUPSHIFT | 0x001 )
 157 #define TAG_MODIFY                                      ( TAG_GROUP_NOTALLOWED << TAG_GROUPSHIFT | 0x002 )
 158 #define TAG_REFNR                                       ( TAG_GROUP_NOTALLOWED << TAG_GROUPSHIFT | 0x004 )
 159
 160 #define TAG_GROUP_STRUCTURE                     0x3
 161 #define TAG_NAME                                        ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x001 )
 162 #define TAG_HREF                                        ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x002 )
 163 #define TAG_AVIS                                        ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x004 )
 164 #define TAG_AHID                                        ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x008 )
 165
 166 #define TAG_TITEL                                       ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x020 )
 167 #define TAG_KEY                                         ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x040 )
 168 #define TAG_INDEX                                       ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x080 )
 169
 170 #define TAG_REFSTART                            ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x100 )
 171
 172 #define TAG_GRAPHIC                                     ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x200 )
 173 #define TAG_NEXTVERSION                         ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x400 )
 174
 175 #define TAG_GROUP_SYSSWITCH                     0x4
 176 #define TAG_WIN                                         ( TAG_GROUP_SYSSWITCH << TAG_GROUPSHIFT | 0x001 )
 177 #define TAG_UNIX                                        ( TAG_GROUP_SYSSWITCH << TAG_GROUPSHIFT | 0x002 )
 178 #define TAG_MAC                                         ( TAG_GROUP_SYSSWITCH << TAG_GROUPSHIFT | 0x004 )
 179 #define TAG_OS2                                         ( TAG_GROUP_SYSSWITCH << TAG_GROUPSHIFT | 0x008 )
 180
 181 #define TAG_GROUP_PROGSWITCH            0x5
 182 #define TAG_WRITER                                      ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x001 )
 183 #define TAG_CALC                                        ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x002 )
 184 #define TAG_DRAW                                        ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x004 )
 185 #define TAG_IMPRESS                                     ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x008 )
 186 #define TAG_SCHEDULE                            ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x010 )
 187 #define TAG_IMAGE                                       ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x020 )
 188 #define TAG_MATH                                        ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x040 )
 189 #define TAG_CHART                                       ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x080 )
 190 #define TAG_OFFICE                                      ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x100 )
 191
 192
 193 #define TAG_GROUP_META                          0x6
 194 #define TAG_OFFICEFULLNAME                      ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x001 )
 195 #define TAG_OFFICENAME                          ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x002 )
 196 #define TAG_OFFICEPATH                          ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x004 )
 197 #define TAG_OFFICEVERSION                       ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x008 )
 198 #define TAG_PORTALNAME                          ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x010 )
 199 #define TAG_PORTALFULLNAME                      ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x020 )
 200 #define TAG_PORTALPATH                          ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x040 )
 201 #define TAG_PORTALVERSION                       ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x080 )
 202 #define TAG_PORTALSHORTNAME                     ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x100 )
 203
 204
 205 #define TAG_GROUP_SINGLE            0x7
 206 #define TAG_REFINSERT               ( TAG_GROUP_SINGLE << TAG_GROUPSHIFT | 0x001 )
 207
 208
 209 #define TAG_GROUP_MULTI                         0x8
 210 #define TAG_END                                         ( TAG_GROUP_MULTI << TAG_GROUPSHIFT | 0x010 )
 211 #define TAG_ELSE                                        ( TAG_GROUP_MULTI << TAG_GROUPSHIFT | 0x020 )
 212 #define TAG_AEND                                        ( TAG_GROUP_MULTI << TAG_GROUPSHIFT | 0x040 )
 213 #define TAG_VERSIONEND                          ( TAG_GROUP_MULTI << TAG_GROUPSHIFT | 0x080 )
 214 #define TAG_ENDGRAPHIC                          ( TAG_GROUP_MULTI << TAG_GROUPSHIFT | 0x100 )
 215
 216 #define TAG_GROUP_MISC                          0x9
 217 #define TAG_COMMONSTART                         ( TAG_GROUP_MISC << TAG_GROUPSHIFT | 0x001 )
 218 #define TAG_COMMONEND                           ( TAG_GROUP_MISC << TAG_GROUPSHIFT | 0x002 )
 219
 220 #define TAG_UNKNOWN_TAG                         ( TAG_GROUP_MULTI << TAG_GROUPSHIFT | 0x800 )
 221
 222 DECLARE_LIST( TokenListImpl, TokenInfo* )
 223
 224 class TokenList : private TokenListImpl
 225 {
 226 private:
 227
 228     TokenList&   operator =( const TokenList& rList );
 229 //                { TokenListImpl::operator =( rList ); return *this; }
 230
 231
 232 public:
 233     using TokenListImpl::Count;
 234
 235
 236     TokenList() : TokenListImpl(){};
 237     ~TokenList(){ Clear(); };
 238
 239     void                Clear()
 240         {
 241             for ( ULONG i = 0 ; i < Count() ; i++ )
 242                 delete TokenListImpl::GetObject( i );
 243             TokenListImpl::Clear();
 244         }
 245     void                Insert( TokenInfo p, ULONG nIndex = LIST_APPEND )
 246         { TokenListImpl::Insert( new TokenInfo(p), nIndex ); }
 247 /*    TokenInfo         Remove( ULONG nIndex )
 248         {
 249             TokenInfo aT = GetObject( nIndex );
 250             delete TokenListImpl::GetObject( nIndex );
 251             TokenListImpl::Remove( nIndex );
 252             return aT;
 253         }*/
 254 //    TokenInfo         Remove( TokenInfo p ){ return Remove( GetPos( p ) ); }
 255 //    TokenInfo         GetCurObject() const { return *TokenListImpl::GetCurObject(); }
 256     TokenInfo&          GetObject( ULONG nIndex ) const
 257         {
 258 //                      if ( TokenListImpl::GetObject(nIndex) )
 259                 return *TokenListImpl::GetObject(nIndex);
 260 //                      else
 261 //                              return TokenInfo();
 262         }
 263 /*    ULONG             GetPos( const TokenInfo p ) const
 264         {
 265             for ( ULONG i = 0 ; i < Count() ; i++ )
 266                 if ( p == GetObject( i ) )
 267                     return i;
 268             return LIST_ENTRY_NOTFOUND;
 269         }*/
 270
 271     TokenList( const TokenList& rList );
 272 /*              {
 273             for ( ULONG i = 0 ; i < rList.Count() ; i++ )
 274             {
 275                 Insert( rList.GetObject( i ), LIST_APPEND );
 276             }
 277         }*/
 278 };
 279
 280 class ParserMessage
 281 {
 282     USHORT nErrorNr;
 283     ByteString aErrorText;
 284     USHORT nTagBegin,nTagLength;
 285
 286 protected:
 287     ParserMessage( USHORT PnErrorNr, ByteString PaErrorText, const TokenInfo &rTag );
 288 public:
 289
 290     USHORT GetErrorNr() { return nErrorNr; }
 291     ByteString GetErrorText() { return aErrorText; }
 292
 293     USHORT GetTagBegin() { return nTagBegin; }
 294     USHORT GetTagLength() { return nTagLength; }
 295
 296     virtual ~ParserMessage() {}
 297     virtual BOOL IsError() =0;
 298     virtual ByteString Prefix() =0;
 299 };
 300
 301 class ParserError : public ParserMessage
 302 {
 303 public:
 304     ParserError( USHORT PnErrorNr, ByteString PaErrorText, const TokenInfo &rTag );
 305
 306     virtual BOOL IsError() {return TRUE;};
 307     virtual ByteString Prefix() {return "Error:"; };
 308 };
 309
 310 class ParserWarning : public ParserMessage
 311 {
 312 public:
 313     ParserWarning( USHORT PnErrorNr, ByteString PaErrorText, const TokenInfo &rTag );
 314
 315     virtual BOOL IsError() {return FALSE;};
 316     virtual ByteString Prefix() {return "Warning:"; };
 317 };
 318
 319 class SimpleParser
 320 {
 321 private:
 322     USHORT nPos;
 323     String aSource;
 324     String aLastToken;
 325     TokenList aTokenList;
 326
 327     TokenInfo aNextTag;     // to store closetag in case of combined tags like <br/>
 328
 329     String GetNextTokenString( ParserMessageList &rErrorList, USHORT &rTokeStartPos );
 330
 331 public:
 332     SimpleParser();
 333     void Parse( String PaSource );
 334     TokenInfo GetNextToken( ParserMessageList &rErrorList );
 335     static String GetLexem( TokenInfo const &aToken );
 336     TokenList& GetTokenList(){ return aTokenList; }
 337 };
 338
 339 class TokenParser
 340 {
 341     BOOL match( const TokenInfo &aCurrentToken, const TokenId &aExpectedToken );
 342     BOOL match( const TokenInfo &aCurrentToken, const TokenInfo &aExpectedToken );
 343     void ParseError( USHORT nErrNr, ByteString aErrMsg, const TokenInfo &rTag );
 344     void Paragraph();
 345     void PfCase();
 346     void PfCaseBegin();
 347     void AppCase();
 348     void AppCaseBegin();
 349     void CaseEnd();
 350     void SimpleTag();
 351     void TagPair();
 352     void TagRef();
 353
 354     SimpleParser aParser;
 355     TokenInfo aTag;
 356
 357     TokenId nPfCaseOptions;
 358     TokenId nAppCaseOptions;
 359     BOOL bPfCaseActive ,bAppCaseActive;
 360
 361     TokenId nActiveRefTypes;
 362
 363     ParserMessageList *pErrorList;
 364
 365 public:
 366     TokenParser();
 367     void Parse( const String &aCode, ParserMessageList* pList );
 368 //      ParserMessageList& GetErrors(){ return aErrorList; }
 369 //      BOOL HasErrors(){ return ( aErrorList.Count() > 0 ); }
 370     TokenList& GetTokenList(){ return aParser.GetTokenList(); }
 371 };
 372
 373 class LingTest
 374 {
 375 private:
 376     TokenParser aReferenceParser;
 377     TokenParser aTesteeParser;
 378     ParserMessageList aCompareWarningList;
 379     void CheckTags( TokenList &aReference, TokenList &aTestee, BOOL bFixTags );
 380     BOOL IsTagMandatory( TokenInfo const &aToken, TokenId &aMetaTokens );
 381     String aFixedTestee;
 382 public:
 383     void CheckReference( GSILine *aReference );
 384     void CheckTestee( GSILine *aTestee, BOOL bHasSourceLine, BOOL bFixTags );
 385
 386 //      ParserMessageList& GetReferenceErrors(){ return aReferenceParser.GetErrors(); }
 387 //      BOOL HasReferenceErrors(){ return aReferenceParser.HasErrors(); }
 388
 389 //      ParserMessageList& GetTesteeErrors(){ return aTesteeParser.GetErrors(); }
 390 //      BOOL HasTesteeErrors(){ return aTesteeParser.HasErrors(); }
 391
 392     ParserMessageList& GetCompareWarnings(){ return aCompareWarningList; }
 393     BOOL HasCompareWarnings(){ return ( aCompareWarningList.Count() > 0 ); }
 394
 395     String GetFixedTestee(){ return aFixedTestee; }
 396 };
 397
 398 #endif
 399