l10ntools/inc/tagtest.hxx

   1 /* -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*- */
   2 /*************************************************************************
   3  *
   4  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
   5  *
   6  * Copyright 2000, 2010 Oracle and/or its affiliates.
   7  *
   8  * OpenOffice.org - a multi-platform office productivity suite
   9  *
  10  * This file is part of OpenOffice.org.
  11  *
  12  * OpenOffice.org is free software: you can redistribute it and/or modify
  13  * it under the terms of the GNU Lesser General Public License version 3
  14  * only, as published by the Free Software Foundation.
  15  *
  16  * OpenOffice.org is distributed in the hope that it will be useful,
  17  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  18  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19  * GNU Lesser General Public License version 3 for more details
  20  * (a copy is included in the LICENSE file that accompanied this code).
  21  *
  22  * You should have received a copy of the GNU Lesser General Public License
  23  * version 3 along with OpenOffice.org.  If not, see
  24  * <http://www.openoffice.org/license.html>
  25  * for a copy of the LGPLv3 License.
  26  *
  27  ************************************************************************/
  28
  29 #ifndef _TAGTEST_HXX_
  30 #define _TAGTEST_HXX_
  31
  32 #include <boost/unordered_map.hpp>
  33 #include <vector>
  34
  35 class GSILine;
  36
  37 typedef sal_Int32 TokenId;
  38
  39 #define TOK_INVALIDPOS (-1)
  40
  41 class ParserMessage;
  42 typedef ::std::vector< ParserMessage* > Impl_ParserMessageList;
  43
  44 class ParserMessageList;
  45
  46 typedef boost::unordered_map<rtl::OString, rtl::OUString, rtl::OStringHash> StringHashMap;
  47
  48 class TokenInfo
  49 {
  50 private:
  51     void SplitTag( ParserMessageList &rErrorList );
  52
  53     rtl::OUString aTagName;
  54     StringHashMap aProperties;
  55     sal_Bool bClosed;    // tag is closed  <sdnf/>
  56     sal_Bool bCloseTag;  // tag is close Tag  </sdnf>
  57
  58
  59     sal_Bool bIsBroken;
  60     sal_Bool bHasBeenFixed;
  61     sal_Bool bDone;
  62
  63 public:
  64
  65     rtl::OUString aTokenString;
  66     TokenId nId;
  67     sal_Int32 nPos;            // Position in String
  68
  69     TokenInfo():bClosed(sal_False),bCloseTag(sal_False),bIsBroken(sal_False),bHasBeenFixed(sal_False),bDone(sal_False),nId( 0 ){;}
  70 explicit    TokenInfo( TokenId pnId, sal_Int32 nP ):bClosed(sal_False),bCloseTag(sal_False),bIsBroken(sal_False),bHasBeenFixed(sal_False),bDone(sal_False),nId( pnId ),nPos(nP){;}
  71     explicit    TokenInfo( TokenId pnId, sal_Int32 nP, rtl::OUString const & paStr ):bClosed(sal_False),bCloseTag(sal_False),bIsBroken(sal_False),bHasBeenFixed(sal_False),bDone(sal_False),aTokenString( paStr ),nId( pnId ),nPos(nP) {;}
  72     explicit    TokenInfo( TokenId pnId, sal_Int32 nP, rtl::OUString const & paStr, ParserMessageList &rErrorList );
  73
  74     rtl::OUString GetTagName() const;
  75
  76     rtl::OUString MakeTag() const;
  77
  78     /**
  79         Is the property to be ignored or does it have the default value anyways
  80     **/
  81     sal_Bool IsPropertyRelevant( const rtl::OString &rName, const rtl::OUString &rValue ) const;
  82     sal_Bool IsPropertyValueValid( const rtl::OString &rName, const rtl::OUString &rValue ) const;
  83     /**
  84         Does the property contain the same value for all languages
  85         e.g.: the href in a link tag
  86     **/
  87     sal_Bool IsPropertyInvariant( const rtl::OString &rName, const rtl::OUString &rValue ) const;
  88     /**
  89         a subset of IsPropertyInvariant but containing only those that are fixable
  90         we dont wat to fix e.g.: ahelp :: visibility
  91     **/
  92     sal_Bool IsPropertyFixable( const rtl::OString &rName ) const;
  93     sal_Bool MatchesTranslation( TokenInfo& rInfo, sal_Bool bGenErrors, ParserMessageList &rErrorList, sal_Bool bFixTags = sal_False ) const;
  94
  95     sal_Bool IsDone() const { return bDone; }
  96     void SetDone( sal_Bool bNew = sal_True ) { bDone = bNew; }
  97
  98     sal_Bool HasBeenFixed() const { return bHasBeenFixed; }
  99     void SetHasBeenFixed( sal_Bool bNew = sal_True ) { bHasBeenFixed = bNew; }
 100 };
 101
 102
 103 class ParserMessageList
 104 {
 105 private:
 106     Impl_ParserMessageList maList;
 107
 108 public:
 109     ~ParserMessageList() { clear(); }
 110     void AddError( sal_Int32 nErrorNr, const rtl::OString& rErrorText, const TokenInfo &rTag );
 111     void AddWarning( sal_Int32 nErrorNr, const rtl::OString& rErrorText, const TokenInfo &rTag );
 112
 113     sal_Bool HasErrors();
 114     bool empty() const { return maList.empty(); }
 115     size_t size() const { return maList.size(); }
 116     ParserMessage* operator [] ( size_t i ) { return ( i < maList.size() ) ? maList[ i ] : NULL; }
 117     void clear();
 118 };
 119
 120
 121 #define TAG_GROUPMASK               0xF000
 122 #define TAG_GROUPSHIFT              12
 123
 124 #define TAG_GROUP( nTag )           (( nTag & TAG_GROUPMASK ) >> TAG_GROUPSHIFT )
 125 #define TAG_NOGROUP( nTag )         ( nTag & ~TAG_GROUPMASK )   // ~ = Bitweises NOT
 126
 127 #define TAG_NOMORETAGS              0x0
 128
 129 #define TAG_GROUP_FORMAT            0x1
 130 #define TAG_ON                      0x100
 131 #define TAG_BOLDON                  ( TAG_GROUP_FORMAT << TAG_GROUPSHIFT | TAG_ON | 0x001 )
 132 #define TAG_BOLDOFF                 ( TAG_GROUP_FORMAT << TAG_GROUPSHIFT |          0x001 )
 133 #define TAG_ITALICON                ( TAG_GROUP_FORMAT << TAG_GROUPSHIFT | TAG_ON | 0x002 )
 134 #define TAG_ITALICOFF               ( TAG_GROUP_FORMAT << TAG_GROUPSHIFT |          0x002 )
 135 #define TAG_UNDERLINEON             ( TAG_GROUP_FORMAT << TAG_GROUPSHIFT | TAG_ON | 0x004 )
 136 #define TAG_UNDERLINEOFF            ( TAG_GROUP_FORMAT << TAG_GROUPSHIFT |          0x004 )
 137
 138 #define TAG_GROUP_NOTALLOWED        0x2
 139 #define TAG_HELPID                  ( TAG_GROUP_NOTALLOWED << TAG_GROUPSHIFT | 0x001 )
 140 #define TAG_MODIFY                  ( TAG_GROUP_NOTALLOWED << TAG_GROUPSHIFT | 0x002 )
 141 #define TAG_REFNR                   ( TAG_GROUP_NOTALLOWED << TAG_GROUPSHIFT | 0x004 )
 142
 143 #define TAG_GROUP_STRUCTURE         0x3
 144 #define TAG_NAME                    ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x001 )
 145 #define TAG_HREF                    ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x002 )
 146 #define TAG_AVIS                    ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x004 )
 147 #define TAG_AHID                    ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x008 )
 148
 149 #define TAG_TITEL                   ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x020 )
 150 #define TAG_KEY                     ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x040 )
 151 #define TAG_INDEX                   ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x080 )
 152
 153 #define TAG_REFSTART                ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x100 )
 154
 155 #define TAG_GRAPHIC                 ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x200 )
 156 #define TAG_NEXTVERSION             ( TAG_GROUP_STRUCTURE << TAG_GROUPSHIFT | 0x400 )
 157
 158 #define TAG_GROUP_SYSSWITCH         0x4
 159 #define TAG_WIN                     ( TAG_GROUP_SYSSWITCH << TAG_GROUPSHIFT | 0x001 )
 160 #define TAG_UNIX                    ( TAG_GROUP_SYSSWITCH << TAG_GROUPSHIFT | 0x002 )
 161 #define TAG_MAC                     ( TAG_GROUP_SYSSWITCH << TAG_GROUPSHIFT | 0x004 )
 162 #define TAG_OS2                     ( TAG_GROUP_SYSSWITCH << TAG_GROUPSHIFT | 0x008 )
 163
 164 #define TAG_GROUP_PROGSWITCH        0x5
 165 #define TAG_WRITER                  ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x001 )
 166 #define TAG_CALC                    ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x002 )
 167 #define TAG_DRAW                    ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x004 )
 168 #define TAG_IMPRESS                 ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x008 )
 169 #define TAG_SCHEDULE                ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x010 )
 170 #define TAG_IMAGE                   ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x020 )
 171 #define TAG_MATH                    ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x040 )
 172 #define TAG_CHART                   ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x080 )
 173 #define TAG_OFFICE                  ( TAG_GROUP_PROGSWITCH << TAG_GROUPSHIFT | 0x100 )
 174
 175
 176 #define TAG_GROUP_META              0x6
 177 #define TAG_OFFICEFULLNAME          ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x001 )
 178 #define TAG_OFFICENAME              ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x002 )
 179 #define TAG_OFFICEPATH              ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x004 )
 180 #define TAG_OFFICEVERSION           ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x008 )
 181 #define TAG_PORTALNAME              ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x010 )
 182 #define TAG_PORTALFULLNAME          ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x020 )
 183 #define TAG_PORTALPATH              ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x040 )
 184 #define TAG_PORTALVERSION           ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x080 )
 185 #define TAG_PORTALSHORTNAME         ( TAG_GROUP_META << TAG_GROUPSHIFT | 0x100 )
 186
 187
 188 #define TAG_GROUP_SINGLE            0x7
 189 #define TAG_REFINSERT               ( TAG_GROUP_SINGLE << TAG_GROUPSHIFT | 0x001 )
 190
 191
 192 #define TAG_GROUP_MULTI             0x8
 193 #define TAG_END                     ( TAG_GROUP_MULTI << TAG_GROUPSHIFT | 0x010 )
 194 #define TAG_ELSE                    ( TAG_GROUP_MULTI << TAG_GROUPSHIFT | 0x020 )
 195 #define TAG_AEND                    ( TAG_GROUP_MULTI << TAG_GROUPSHIFT | 0x040 )
 196 #define TAG_VERSIONEND              ( TAG_GROUP_MULTI << TAG_GROUPSHIFT | 0x080 )
 197 #define TAG_ENDGRAPHIC              ( TAG_GROUP_MULTI << TAG_GROUPSHIFT | 0x100 )
 198
 199 #define TAG_GROUP_MISC              0x9
 200 #define TAG_COMMONSTART             ( TAG_GROUP_MISC << TAG_GROUPSHIFT | 0x001 )
 201 #define TAG_COMMONEND               ( TAG_GROUP_MISC << TAG_GROUPSHIFT | 0x002 )
 202
 203 #define TAG_UNKNOWN_TAG             ( TAG_GROUP_MULTI << TAG_GROUPSHIFT | 0x800 )
 204
 205 typedef ::std::vector< TokenInfo* > TokenListImpl;
 206
 207 class TokenList
 208 {
 209 private:
 210     TokenListImpl maList;
 211     TokenList&   operator =( const TokenList& rList );
 212
 213 public:
 214     TokenList() {}
 215     ~TokenList(){ clear(); }
 216
 217     size_t size() const { return maList.size(); }
 218     void clear()
 219         {
 220             for ( size_t i = 0 ; i < maList.size() ; i++ )
 221                 delete maList[ i ];
 222             maList.clear();
 223         }
 224
 225     void insert( TokenInfo p, size_t nIndex = size_t(-1) )
 226         {
 227             if ( nIndex < maList.size() ) {
 228                 TokenListImpl::iterator it = maList.begin();
 229                 ::std::advance( it, nIndex );
 230                 maList.insert( it, new TokenInfo(p) );
 231             } else {
 232                 maList.push_back( new TokenInfo(p) );
 233             }
 234         }
 235     TokenInfo& operator [] ( size_t nIndex ) const
 236         {
 237             return *maList[ nIndex ];
 238         }
 239
 240     TokenList( const TokenList& rList );
 241 };
 242
 243 class ParserMessage
 244 {
 245     sal_Int32 nErrorNr;
 246     rtl::OString aErrorText;
 247     sal_Int32 nTagBegin,nTagLength;
 248
 249 protected:
 250     ParserMessage( sal_Int32 PnErrorNr, const rtl::OString &rPaErrorText, const TokenInfo &rTag );
 251 public:
 252
 253     sal_Int32 GetErrorNr() { return nErrorNr; }
 254     rtl::OString GetErrorText() { return aErrorText; }
 255
 256     sal_Int32 GetTagBegin() { return nTagBegin; }
 257     sal_Int32 GetTagLength() { return nTagLength; }
 258
 259     virtual ~ParserMessage() {}
 260     virtual sal_Bool IsError() =0;
 261     virtual rtl::OString Prefix() =0;
 262 };
 263
 264 class ParserError : public ParserMessage
 265 {
 266 public:
 267     ParserError( sal_Int32 PnErrorNr, const rtl::OString &rPaErrorText, const TokenInfo &rTag );
 268
 269     virtual sal_Bool IsError() {return sal_True;}
 270     virtual rtl::OString Prefix() {return rtl::OString(RTL_CONSTASCII_STRINGPARAM("Error:")); }
 271 };
 272
 273 class ParserWarning : public ParserMessage
 274 {
 275 public:
 276     ParserWarning( sal_Int32 PnErrorNr, const rtl::OString &rPaErrorText, const TokenInfo &rTag );
 277
 278     virtual sal_Bool IsError() {return sal_False;}
 279     virtual rtl::OString Prefix() {return rtl::OString(RTL_CONSTASCII_STRINGPARAM("Warning:")); }
 280 };
 281
 282 class SimpleParser
 283 {
 284 private:
 285     sal_Int32 nPos;
 286     rtl::OUString aSource;
 287     rtl::OUString aLastToken;
 288     TokenList aTokenList;
 289
 290     TokenInfo aNextTag;     // to store closetag in case of combined tags like <br/>
 291
 292     rtl::OUString GetNextTokenString( ParserMessageList &rErrorList, sal_Int32 &rTokeStartPos );
 293
 294 public:
 295     SimpleParser();
 296     void Parse( rtl::OUString const & PaSource );
 297     TokenInfo GetNextToken( ParserMessageList &rErrorList );
 298     static rtl::OUString GetLexem( TokenInfo const &aToken );
 299     TokenList& GetTokenList(){ return aTokenList; }
 300 };
 301
 302 class TokenParser
 303 {
 304     sal_Bool match( const TokenInfo &aCurrentToken, const TokenId &aExpectedToken );
 305     sal_Bool match( const TokenInfo &aCurrentToken, const TokenInfo &aExpectedToken );
 306     void ParseError( sal_Int32 nErrNr, const rtl::OString &rErrMsg, const TokenInfo &rTag );
 307     void Paragraph();
 308     void PfCase();
 309     void PfCaseBegin();
 310     void AppCase();
 311     void AppCaseBegin();
 312     void CaseEnd();
 313     void SimpleTag();
 314     void TagPair();
 315     void TagRef();
 316
 317     SimpleParser aParser;
 318     TokenInfo aTag;
 319
 320     TokenId nPfCaseOptions;
 321     TokenId nAppCaseOptions;
 322     sal_Bool bPfCaseActive ,bAppCaseActive;
 323
 324     TokenId nActiveRefTypes;
 325
 326     ParserMessageList *pErrorList;
 327
 328 public:
 329     TokenParser();
 330     void Parse( const rtl::OUString &aCode, ParserMessageList* pList );
 331     TokenList& GetTokenList(){ return aParser.GetTokenList(); }
 332 };
 333
 334 class LingTest
 335 {
 336 private:
 337     TokenParser aReferenceParser;
 338     TokenParser aTesteeParser;
 339     ParserMessageList aCompareWarningList;
 340     void CheckTags( TokenList &aReference, TokenList &aTestee, sal_Bool bFixTags );
 341     sal_Bool IsTagMandatory( TokenInfo const &aToken, TokenId &aMetaTokens );
 342     rtl::OUString aFixedTestee;
 343 public:
 344     void CheckReference( GSILine *aReference );
 345     void CheckTestee( GSILine *aTestee, sal_Bool bHasSourceLine, sal_Bool bFixTags );
 346
 347     ParserMessageList& GetCompareWarnings(){ return aCompareWarningList; }
 348     sal_Bool HasCompareWarnings(){ return ( !aCompareWarningList.empty() ); }
 349
 350     rtl::OUString GetFixedTestee(){ return aFixedTestee; }
 351 };
 352
 353 #endif
 354
 355 /* vim:set shiftwidth=4 softtabstop=4 expandtab: */