perl5compiler.java

java实现正则表达式的代码

    if(__input._getOffset() == 0) {
      __input._setOffset(-1);
      __getNextChar();
    } else {
      __input._decrement();
      __getNextChar();
    }

    value = __input._getValue();

    while(value != CharStringPointer._END_OF_STRING &&
	  value != '|' && value != ')') {
      flags &= ~__TRYAGAIN;
      latest = __parseBranch(retFlags);

      if(latest == OpCode._NULL_OFFSET) {
	if((flags & __TRYAGAIN) != 0){
	  value = __input._getValue();
	  continue;
	}
	return OpCode._NULL_OFFSET;
      }

      retFlags[0] |= (flags & __NONNULL);

      if(chain == OpCode._NULL_OFFSET)
	retFlags[0] |= (flags & __SPSTART);
      else {
	++__cost;
	__programAddTail(chain, latest);
      }
      chain = latest;
      value = __input._getValue();
    }

    // If loop was never entered.
    if(chain == OpCode._NULL_OFFSET)
      __emitNode(OpCode._NOTHING);

    return offset;
  }


  private int __parseAtom(int[] retFlags) throws MalformedPatternException {
    boolean doDefault;
    char value;
    int offset, flags[] = { 0 };
    
    
    retFlags[0] = __WORSTCASE;
    doDefault = false;
    offset = OpCode._NULL_OFFSET;

  tryAgain:
    while(true) {

      value = __input._getValue();

      switch(value) {
      case '^' :
	__getNextChar();
	// The order here is important in order to support /ms.
	// /m takes precedence over /s for ^ and $, but not for .
	if((__modifierFlags[0] & __MULTILINE) != 0)
	  offset = __emitNode(OpCode._MBOL);
	else if((__modifierFlags[0] & __SINGLELINE) != 0)
	  offset = __emitNode(OpCode._SBOL);
	else
	  offset = __emitNode(OpCode._BOL);
	break tryAgain;

      case '$':
	__getNextChar();
	// The order here is important in order to support /ms.
	// /m takes precedence over /s for ^ and $, but not for .
	if((__modifierFlags[0] & __MULTILINE) != 0)
	  offset = __emitNode(OpCode._MEOL);
	else if((__modifierFlags[0] & __SINGLELINE) != 0)
	  offset = __emitNode(OpCode._SEOL);
	else
	  offset = __emitNode(OpCode._EOL);
	break tryAgain;

      case '.':
	__getNextChar();
	// The order here is important in order to support /ms.
	// /m takes precedence over /s for ^ and $, but not for .
	if((__modifierFlags[0] & __SINGLELINE) != 0)
	  offset = __emitNode(OpCode._SANY);
	else
	  offset = __emitNode(OpCode._ANY);
	++__cost;
	retFlags[0] |= (__NONNULL | __SIMPLE);
	break tryAgain;

      case '[':
	__input._increment();
	offset = __parseUnicodeClass();
	retFlags[0] |= (__NONNULL | __SIMPLE);
	break tryAgain;

      case '(':
	__getNextChar();
	offset = __parseExpression(true, flags);
	if(offset == OpCode._NULL_OFFSET) {
	  if((flags[0] & __TRYAGAIN) != 0)
	    continue tryAgain;
	  return OpCode._NULL_OFFSET;
	}
	retFlags[0] |= (flags[0] & (__NONNULL | __SPSTART));
	break tryAgain;

      case '|':
      case ')':
	if((flags[0] & __TRYAGAIN) != 0) {
	  retFlags[0] |= __TRYAGAIN;
	  return OpCode._NULL_OFFSET;
	}

	throw new MalformedPatternException("Error in expression at " +
				   __input._toString(__input._getOffset()));
	//break tryAgain;

      case '?':
      case '+':
      case '*':
	throw new MalformedPatternException(
                 "?+* follows nothing in expression");
	//break tryAgain;

      case '\\':
	value = __input._increment();

	switch(value) {
	case 'A' :
	  offset = __emitNode(OpCode._SBOL);
	  retFlags[0] |= __SIMPLE;
	  __getNextChar();
	  break;
	case 'G':
	  offset = __emitNode(OpCode._GBOL);
	  retFlags[0] |= __SIMPLE;
	  __getNextChar();
	  break;
	case 'Z':
	  offset = __emitNode(OpCode._SEOL);
	  retFlags[0] |= __SIMPLE;
	  __getNextChar();
	  break;
	case 'w':
	  offset = __emitNode(OpCode._ALNUM);
	  retFlags[0] |= (__NONNULL | __SIMPLE);
	  __getNextChar();
	  break;
	case 'W':
	  offset = __emitNode(OpCode._NALNUM);
	  retFlags[0] |= (__NONNULL | __SIMPLE);
	  __getNextChar();
	  break;
	case 'b':
	  offset = __emitNode(OpCode._BOUND);
	  retFlags[0] |= __SIMPLE;
	  __getNextChar();
	  break;
	case 'B':
	  offset = __emitNode(OpCode._NBOUND);
	  retFlags[0] |= __SIMPLE;
	  __getNextChar();
	  break;
	case 's':
	  offset = __emitNode(OpCode._SPACE);
	  retFlags[0] |= (__NONNULL | __SIMPLE);
	  __getNextChar();
	  break;
	case 'S':
	  offset = __emitNode(OpCode._NSPACE);
	  retFlags[0] |= (__NONNULL | __SIMPLE);
	  __getNextChar();
	  break;
	case 'd':
	  offset = __emitNode(OpCode._DIGIT);
	  retFlags[0] |= (__NONNULL | __SIMPLE);
	  __getNextChar();
	  break;
	case 'D':
	  offset = __emitNode(OpCode._NDIGIT);
	  retFlags[0] |= (__NONNULL | __SIMPLE);
	  __getNextChar();
	  break;
	case 'n': case 'r': case 't': case 'f': case 'e': case 'a': case 'x':
	case 'c': case '0':
	  doDefault = true;
	  break tryAgain;
	case '1': case '2': case '3': case '4': case '5': case '6': case '7':
	case '8': case '9':
	  int num;
	  StringBuffer buffer = new StringBuffer(10);

	  num = 0;
	  value = __input._getValueRelative(num);

	  while(Character.isDigit(value)) {
	    buffer.append(value);
	    ++num;
	    value = __input._getValueRelative(num);
	  }

	  try {
	    num = Integer.parseInt(buffer.toString());
	  } catch(NumberFormatException e) {
	    throw new MalformedPatternException(
	   "Unexpected number format exception.  Please report this bug." +
	   "NumberFormatException message: " + e.getMessage());
	  }

	  if(num > 9 && num >= __numParentheses) {
	    doDefault = true;
	    break tryAgain;
	  } else {
	    // A backreference may only occur AFTER its group
	    if(num >= __numParentheses)
	      throw new MalformedPatternException("Invalid backreference: \\" +
						  num);
	    __sawBackreference = true;
	    offset = __emitArgNode(OpCode._REF, (char)num);
	    retFlags[0] |= __NONNULL;

	    value = __input._getValue();
	    while(Character.isDigit(value))
	      value = __input._increment();

	    __input._decrement();
	    __getNextChar();
	  }
	  break;
	case '\0':
	case CharStringPointer._END_OF_STRING:
	  if(__input._isAtEnd())
	    throw new
	      MalformedPatternException("Trailing \\ in expression.");

	// fall through to default
	default:
	  doDefault = true;
	  break tryAgain;
	}
	break tryAgain;

      case '#':
	// skip over comments
	if((__modifierFlags[0] & __EXTENDED) != 0) {
	  while(!__input._isAtEnd() && __input._getValue() != '\n')
	    __input._increment();
	  if(!__input._isAtEnd())
	    continue tryAgain;
	}
	// fall through to default
      default:
	__input._increment();
	doDefault = true;
	break tryAgain;
      }// end master switch
    } // end tryAgain


    if(doDefault) {
      char ender;
      int length, pOffset, maxOffset, lastOffset, numLength[];

      offset = __emitNode(OpCode._EXACTLY);
      // Not sure that it's ok to use 0 to mark end.
      //__emitCode((char)0);
      __emitCode((char)CharStringPointer._END_OF_STRING);

    forLoop:
      for(length = 0, pOffset = __input._getOffset() - 1,
	    maxOffset = __input._getLength();
	  length < 127 && pOffset < maxOffset; ++length) {

	lastOffset = pOffset;
	value = __input._getValue(pOffset);

	switch(value) {
	case '^': case '$': case '.': case '[': case '(': case ')':
	case '|':
	  break forLoop;
	case '\\':
	  value = __input._getValue(++pOffset);

	  switch(value) {
	  case 'A': case 'G': case 'Z': case 'w': case 'W': case 'b':
	  case 'B': case 's': case 'S': case 'd': case 'D':
	    --pOffset;
	    break forLoop;
	  case 'n':
	    ender = '\n';
	    ++pOffset;
	    break;
	  case 'r':
	    ender = '\r';
	    ++pOffset;
	    break;
	  case 't':
	    ender = '\t';
	    ++pOffset;
	    break;
	  case 'f':
	    ender = '\f';
	    ++pOffset;
	    break;
	  case 'e':
	    ender = '\033';
	    ++pOffset;
	    break;
	  case 'a':
	    ender = '\007';
	    ++pOffset;
	    break;
	  case 'x':
	    numLength = new int[1];
	    ender = (char)__parseHex(__input._array, ++pOffset, 2, numLength);
	    pOffset+=numLength[0];
	    break;
	  case 'c':
	    ++pOffset;
	    ender = __input._getValue(pOffset++);
	    if(Character.isLowerCase(ender))
	      ender = Character.toUpperCase(ender);
	    ender ^= 64;
	    break;
	  case '0': case '1': case '2': case'3': case '4': case '5':
	  case '6': case '7': case '8': case '9':
	    boolean doOctal = false;
	    value = __input._getValue(pOffset);

	    if(value == '0')
	      doOctal = true;
	    value = __input._getValue(pOffset + 1);

	    if(Character.isDigit(value)) {
	      int num;
	      StringBuffer buffer = new StringBuffer(10);

	      num = pOffset;
	      value = __input._getValue(num);

	      while(Character.isDigit(value)){
		buffer.append(value);
		++num;
		value = __input._getValue(num);
	      }

	      try {
		num = Integer.parseInt(buffer.toString());
	      } catch(NumberFormatException e) {
		throw new MalformedPatternException(
	     "Unexpected number format exception.  Please report this bug." +
	     "NumberFormatException message: " + e.getMessage());
	      }

	      if(!doOctal)
		doOctal = (num >= __numParentheses);
	    }

	    if(doOctal) {
	      numLength = new int[1];
	      ender = (char)__parseOctal(__input._array, pOffset, 3, numLength);
	      pOffset+=numLength[0];
	    } else {
	      --pOffset;
	      break forLoop;
	    }
	    break;
	  case CharStringPointer._END_OF_STRING:
	  case '\0':
	    if(pOffset >= maxOffset)
	      throw new
		MalformedPatternException("Trailing \\ in expression.");
	    // fall through to default
	  default:
	    ender = __input._getValue(pOffset++);
	    break;
	  } // end backslash switch
	  break;
	case '#':
	  if((__modifierFlags[0] & __EXTENDED) != 0) {
	    while(pOffset < maxOffset && __input._getValue(pOffset) != '\n')
	      ++pOffset;
	  }
	  // fall through to whitespace handling
	case ' ': case '\t': case '\n': case '\r': case '\f': case '\013':
	  if((__modifierFlags[0] & __EXTENDED) != 0) {
	    ++pOffset;
	    --length;
	    continue;
	  }
	  // fall through to default
	default:
	  ender = __input._getValue(pOffset++);
	  break;

	}   // end master switch

	if((__modifierFlags[0] & __CASE_INSENSITIVE) != 0 &&
	   Character.isUpperCase(ender))
	  ender = Character.toLowerCase(ender);

	if(pOffset < maxOffset && __isComplexRepetitionOp(__input._array, pOffset)) {
	  if(length > 0)
	    pOffset = lastOffset;
	  else {
	    ++length;
	    __emitCode(ender);
	  }
	  break;
	}

	__emitCode(ender);


      } // end for loop


      __input._setOffset(pOffset - 1);
      __getNextChar();

      if(length < 0)
	throw new MalformedPatternException(
         "Unexpected compilation failure.  Please report this bug!");
      if(length > 0)
	retFlags[0] |= __NONNULL;
      if(length == 1)
	retFlags[0] |= __SIMPLE;
      if(__program!= null)
	__program[OpCode._getOperand(offset)] = (char)length;
      //__emitCode('\0'); // debug
      __emitCode(CharStringPointer._END_OF_STRING);
    }

    return offset;
  }

  // These are the original 8-bit character class handling methods.
  // We don't want to delete them just yet only to have to dig it out
  // of revision control later.
  /*
  // Set the bits in a character class.  Only recognizes ascii.
  private void __setCharacterClassBits(char[] bits, int offset, char deflt,
				       char ch)
  {
    if(__program== null || ch >= 256)
      return;
    ch &= 0xffff;

    if(deflt == 0) {
      bits[offset + (ch >> 4)] |= (1 << (ch & 0xf));
    } else {
      bits[offset + (ch >> 4)] &= ~(1 << (ch & 0xf));
    }
  }

  private int __parseCharacterClass() throws MalformedPatternException {
    boolean range = false, skipTest;
    char clss, deflt, lastclss = Character.MAX_VALUE;
    int offset, bits, numLength[] = { 0 };

    offset = __emitNode(OpCode._ANYOF);

    if(__input._getValue() == '^') {
      ++__cost;
      __input._increment();
      deflt = 0;
    } else {
      deflt = 0xffff;
    }

    bits = __programSize;
    for(clss = 0; clss < 16; clss++)
      __emitCode(deflt);

    clss = __input._getValue();

    if(clss == ']' || clss == '-')
      skipTest = true;
    else
      skipTest = false;

    while((!__input._isAtEnd() && (clss = __input._getValue()) != ']')
	  || skipTest) {
      // It sucks, but we have to make this assignment every time
      skipTest = false;
      __input._increment();
      if(clss == '\\') {
	clss = __input._postIncrement();

	switch(clss){
	case 'w':
	  for(clss = 0; clss < 256; clss++)
	    if(OpCode._isWordCharacter(clss))
	      __setCharacterClassBits(__program, bits, deflt, clss);
	  lastclss = Character.MAX_VALUE;
	  continue;