htmlparsertokenmanager.java

Lucene是Java社区中著名的全文检索工具.它是一个开源工具包,使用Java实现.

         {
            switch(jjstateSet[--i])
            {
               case 1:
               case 0:
                  kind = 27;
                  jjCheckNAdd(0);
                  break;
               default : break;
            }
         } while(i != startsAt);
      }
      else
      {
         int i2 = (curChar & 0xff) >> 6;
         long l2 = 1L << (curChar & 077);
         MatchLoop: do
         {
            switch(jjstateSet[--i])
            {
               case 1:
               case 0:
                  if ((jjbitVec0[i2] & l2) == 0L)
                     break;
                  if (kind > 27)
                     kind = 27;
                  jjCheckNAdd(0);
                  break;
               default : break;
            }
         } while(i != startsAt);
      }
      if (kind != 0x7fffffff)
      {
         jjmatchedKind = kind;
         jjmatchedPos = curPos;
         kind = 0x7fffffff;
      }
      ++curPos;
      if ((i = jjnewStateCnt) == (startsAt = 2 - (jjnewStateCnt = startsAt)))
         return curPos;
      try { curChar = input_stream.readChar(); }
      catch(java.io.IOException e) { return curPos; }
   }
}
private final int jjMoveStringLiteralDfa0_1()
{
   return jjMoveNfa_1(1, 0);
}
private final int jjMoveNfa_1(int startState, int curPos)
{
   int[] nextStates;
   int startsAt = 0;
   jjnewStateCnt = 12;
   int i = 1;
   jjstateSet[0] = startState;
   int j, kind = 0x7fffffff;
   for (;;)
   {
      if (++jjround == 0x7fffffff)
         ReInitRounds();
      if (curChar < 64)
      {
         long l = 1L << curChar;
         MatchLoop: do
         {
            switch(jjstateSet[--i])
            {
               case 1:
                  if ((0xafffffffffffffffL & l) != 0L)
                  {
                     if (kind > 14)
                        kind = 14;
                     jjCheckNAdd(0);
                  }
                  else if ((0x5000000000000000L & l) != 0L)
                  {
                     if (kind > 14)
                        kind = 14;
                  }
                  if (curChar == 60)
                     jjstateSet[jjnewStateCnt++] = 10;
                  break;
               case 0:
                  if ((0xafffffffffffffffL & l) == 0L)
                     break;
                  if (kind > 14)
                     kind = 14;
                  jjCheckNAdd(0);
                  break;
               case 3:
                  if ((0xafffffffffffffffL & l) != 0L)
                     jjAddStates(18, 19);
                  break;
               case 4:
                  if (curChar == 62 && kind > 15)
                     kind = 15;
                  break;
               case 10:
                  if (curChar == 47)
                     jjstateSet[jjnewStateCnt++] = 9;
                  break;
               case 11:
                  if (curChar == 60)
                     jjstateSet[jjnewStateCnt++] = 10;
                  break;
               default : break;
            }
         } while(i != startsAt);
      }
      else if (curChar < 128)
      {
         long l = 1L << (curChar & 077);
         MatchLoop: do
         {
            switch(jjstateSet[--i])
            {
               case 1:
               case 0:
                  if (kind > 14)
                     kind = 14;
                  jjCheckNAdd(0);
                  break;
               case 2:
                  if (curChar == 116)
                     jjCheckNAddTwoStates(3, 4);
                  break;
               case 3:
                  jjCheckNAddTwoStates(3, 4);
                  break;
               case 5:
                  if (curChar == 112)
                     jjstateSet[jjnewStateCnt++] = 2;
                  break;
               case 6:
                  if (curChar == 105)
                     jjstateSet[jjnewStateCnt++] = 5;
                  break;
               case 7:
                  if (curChar == 114)
                     jjstateSet[jjnewStateCnt++] = 6;
                  break;
               case 8:
                  if (curChar == 99)
                     jjstateSet[jjnewStateCnt++] = 7;
                  break;
               case 9:
                  if (curChar == 115)
                     jjstateSet[jjnewStateCnt++] = 8;
                  break;
               default : break;
            }
         } while(i != startsAt);
      }
      else
      {
         int i2 = (curChar & 0xff) >> 6;
         long l2 = 1L << (curChar & 077);
         MatchLoop: do
         {
            switch(jjstateSet[--i])
            {
               case 1:
               case 0:
                  if ((jjbitVec0[i2] & l2) == 0L)
                     break;
                  if (kind > 14)
                     kind = 14;
                  jjCheckNAdd(0);
                  break;
               case 3:
                  if ((jjbitVec0[i2] & l2) != 0L)
                     jjAddStates(18, 19);
                  break;
               default : break;
            }
         } while(i != startsAt);
      }
      if (kind != 0x7fffffff)
      {
         jjmatchedKind = kind;
         jjmatchedPos = curPos;
         kind = 0x7fffffff;
      }
      ++curPos;
      if ((i = jjnewStateCnt) == (startsAt = 12 - (jjnewStateCnt = startsAt)))
         return curPos;
      try { curChar = input_stream.readChar(); }
      catch(java.io.IOException e) { return curPos; }
   }
}
private final int jjStopStringLiteralDfa_2(int pos, long active0)
{
   switch (pos)
   {
      default :
         return -1;
   }
}
private final int jjStartNfa_2(int pos, long active0)
{
   return jjMoveNfa_2(jjStopStringLiteralDfa_2(pos, active0), pos + 1);
}
private final int jjStartNfaWithStates_2(int pos, int kind, int state)
{
   jjmatchedKind = kind;
   jjmatchedPos = pos;
   try { curChar = input_stream.readChar(); }
   catch(java.io.IOException e) { return pos + 1; }
   return jjMoveNfa_2(state, pos + 1);
}
private final int jjMoveStringLiteralDfa0_2()
{
   switch(curChar)
   {
      case 34:
         return jjStopAtPos(0, 21);
      case 39:
         return jjStopAtPos(0, 20);
      case 61:
         return jjStartNfaWithStates_2(0, 17, 3);
      default :
         return jjMoveNfa_2(0, 0);
   }
}
private final int jjMoveNfa_2(int startState, int curPos)
{
   int[] nextStates;
   int startsAt = 0;
   jjnewStateCnt = 6;
   int i = 1;
   jjstateSet[0] = startState;
   int j, kind = 0x7fffffff;
   for (;;)
   {
      if (++jjround == 0x7fffffff)
         ReInitRounds();
      if (curChar < 64)
      {
         long l = 1L << curChar;
         MatchLoop: do
         {
            switch(jjstateSet[--i])
            {
               case 0:
                  if ((0x9fffff7affffd9ffL & l) != 0L)
                  {
                     if (kind > 16)
                        kind = 16;
                     jjCheckNAdd(1);
                  }
                  else if ((0x100002600L & l) != 0L)
                  {
                     if (kind > 22)
                        kind = 22;
                     jjCheckNAdd(5);
                  }
                  else if (curChar == 61)
                     jjstateSet[jjnewStateCnt++] = 3;
                  else if (curChar == 62)
                  {
                     if (kind > 18)
                        kind = 18;
                  }
                  break;
               case 1:
                  if ((0x9ffffffeffffd9ffL & l) == 0L)
                     break;
                  if (kind > 16)
                     kind = 16;
                  jjCheckNAdd(1);
                  break;
               case 2:
               case 3:
                  if (curChar == 62 && kind > 18)
                     kind = 18;
                  break;
               case 4:
                  if (curChar == 61)
                     jjstateSet[jjnewStateCnt++] = 3;
                  break;
               case 5:
                  if ((0x100002600L & l) == 0L)
                     break;
                  kind = 22;
                  jjCheckNAdd(5);
                  break;
               default : break;
            }
         } while(i != startsAt);
      }
      else if (curChar < 128)
      {
         long l = 1L << (curChar & 077);
         MatchLoop: do
         {
            switch(jjstateSet[--i])
            {
               case 0:
               case 1:
                  if (kind > 16)
                     kind = 16;
                  jjCheckNAdd(1);
                  break;
               default : break;
            }
         } while(i != startsAt);
      }
      else
      {
         int i2 = (curChar & 0xff) >> 6;
         long l2 = 1L << (curChar & 077);
         MatchLoop: do
         {
            switch(jjstateSet[--i])
            {
               case 0:
               case 1:
                  if ((jjbitVec0[i2] & l2) == 0L)
                     break;
                  if (kind > 16)
                     kind = 16;
                  jjCheckNAdd(1);
                  break;
               default : break;
            }
         } while(i != startsAt);
      }
      if (kind != 0x7fffffff)
      {
         jjmatchedKind = kind;
         jjmatchedPos = curPos;
         kind = 0x7fffffff;
      }
      ++curPos;
      if ((i = jjnewStateCnt) == (startsAt = 6 - (jjnewStateCnt = startsAt)))
         return curPos;
      try { curChar = input_stream.readChar(); }
      catch(java.io.IOException e) { return curPos; }
   }
}
static final int[] jjnextStates = {
   20, 21, 24, 12, 14, 16, 5, 8, 0, 4, 6, 0, 4, 6, 5, 0, 
   4, 6, 3, 4, 
};
public static final String[] jjstrLiteralImages = {
"", "\74\163\143\162\151\160\164", null, null, "\74\41\55\55", "\74\41", null, 
null, null, null, null, null, null, null, null, null, null, "\75", null, null, 
"\47", "\42", null, null, null, null, null, null, "\55\55\76", null, "\76", };
public static final String[] lexStateNames = {
   "DEFAULT", 
   "WithinScript", 
   "WithinTag", 
   "AfterEquals", 
   "WithinQuote1", 
   "WithinQuote2", 
   "WithinComment1", 
   "WithinComment2", 
};
public static final int[] jjnewLexState = {
   -1, 1, 2, 2, 6, 7, -1, -1, -1, -1, -1, -1, -1, -1, -1, 0, -1, 3, 0, 2, 4, 5, -1, -1, 2, 
   -1, 2, -1, 0, -1, 0, 
};
static final long[] jjtoToken = {
   0x7fbfec7fL, 
};
static final long[] jjtoSkip = {
   0x400000L, 
};
protected SimpleCharStream input_stream;
private final int[] jjrounds = new int[28];
private final int[] jjstateSet = new int[56];
protected char curChar;
public HTMLParserTokenManager(SimpleCharStream stream)
{
   if (SimpleCharStream.staticFlag)
      throw new Error("ERROR: Cannot use a static CharStream class with a non-static lexical analyzer.");
   input_stream = stream;
}
public HTMLParserTokenManager(SimpleCharStream stream, int lexState)
{
   this(stream);
   SwitchTo(lexState);
}
public void ReInit(SimpleCharStream stream)
{
   jjmatchedPos = jjnewStateCnt = 0;
   curLexState = defaultLexState;
   input_stream = stream;
   ReInitRounds();
}
private final void ReInitRounds()
{
   int i;
   jjround = 0x80000001;
   for (i = 28; i-- > 0;)
      jjrounds[i] = 0x80000000;
}
public void ReInit(SimpleCharStream stream, int lexState)
{
   ReInit(stream);
   SwitchTo(lexState);
}
public void SwitchTo(int lexState)
{
   if (lexState >= 8 || lexState < 0)
      throw new TokenMgrError("Error: Ignoring invalid lexical state : " + lexState + ". State unchanged.", TokenMgrError.INVALID_LEXICAL_STATE);
   else
      curLexState = lexState;
}

protected Token jjFillToken()
{
   Token t = Token.newToken(jjmatchedKind);
   t.kind = jjmatchedKind;
   String im = jjstrLiteralImages[jjmatchedKind];
   t.image = (im == null) ? input_stream.GetImage() : im;
   t.beginLine = input_stream.getBeginLine();
   t.beginColumn = input_stream.getBeginColumn();
   t.endLine = input_stream.getEndLine();
   t.endColumn = input_stream.getEndColumn();
   return t;
}

int curLexState = 0;
int defaultLexState = 0;
int jjnewStateCnt;
int jjround;
int jjmatchedPos;
int jjmatchedKind;

public Token getNextToken() 
{
  int kind;
  Token specialToken = null;
  Token matchedToken;
  int curPos = 0;

  EOFLoop :
  for (;;)
  {   
   try   
   {     
      curChar = input_stream.BeginToken();
   }     
   catch(java.io.IOException e)
   {        
      jjmatchedKind = 0;
      matchedToken = jjFillToken();
      return matchedToken;
   }

   switch(curLexState)
   {
     case 0:
       jjmatchedKind = 0x7fffffff;
       jjmatchedPos = 0;
       curPos = jjMoveStringLiteralDfa0_0();
       if (jjmatchedPos == 0 && jjmatchedKind > 13)
       {
          jjmatchedKind = 13;
       }
       break;
     case 1:
       jjmatchedKind = 0x7fffffff;
       jjmatchedPos = 0;
       curPos = jjMoveStringLiteralDfa0_1();
       break;
     case 2:
       jjmatchedKind = 0x7fffffff;
       jjmatchedPos = 0;
       curPos = jjMoveStringLiteralDfa0_2();
       break;
     case 3:
       jjmatchedKind = 0x7fffffff;
       jjmatchedPos = 0;
       curPos = jjMoveStringLiteralDfa0_3();
       break;
     case 4:
       jjmatchedKind = 0x7fffffff;
       jjmatchedPos = 0;
       curPos = jjMoveStringLiteralDfa0_4();
       break;
     case 5:
       jjmatchedKind = 0x7fffffff;
       jjmatchedPos = 0;
       curPos = jjMoveStringLiteralDfa0_5();
       break;
     case 6:
       jjmatchedKind = 0x7fffffff;
       jjmatchedPos = 0;
       curPos = jjMoveStringLiteralDfa0_6();
       break;
     case 7:
       jjmatchedKind = 0x7fffffff;
       jjmatchedPos = 0;
       curPos = jjMoveStringLiteralDfa0_7();
       break;
   }
     if (jjmatchedKind != 0x7fffffff)
     {
        if (jjmatchedPos + 1 < curPos)
           input_stream.backup(curPos - jjmatchedPos - 1);
        if ((jjtoToken[jjmatchedKind >> 6] & (1L << (jjmatchedKind & 077))) != 0L)
        {
           matchedToken = jjFillToken();
       if (jjnewLexState[jjmatchedKind] != -1)
         curLexState = jjnewLexState[jjmatchedKind];
           return matchedToken;
        }
        else
        {
         if (jjnewLexState[jjmatchedKind] != -1)
           curLexState = jjnewLexState[jjmatchedKind];
           continue EOFLoop;
        }
     }
     int error_line = input_stream.getEndLine();
     int error_column = input_stream.getEndColumn();
     String error_after = null;
     boolean EOFSeen = false;
     try { input_stream.readChar(); input_stream.backup(1); }
     catch (java.io.IOException e1) {
        EOFSeen = true;
        error_after = curPos <= 1 ? "" : input_stream.GetImage();
        if (curChar == '\n' || curChar == '\r') {
           error_line++;
           error_column = 0;
        }
        else
           error_column++;
     }
     if (!EOFSeen) {
        input_stream.backup(1);
        error_after = curPos <= 1 ? "" : input_stream.GetImage();
     }
     throw new TokenMgrError(EOFSeen, curLexState, error_line, error_column, error_after, curChar, TokenMgrError.LEXICAL_ERROR);
  }
}

}