/**
* Checks if this is a (or the start of a) block-comment-ending delimiter character.
*
* @param c the character input.
* @return true if so, false if not.
*/
protected boolean isCommentEndDelimiterStart(char c) {
return kernel.getEndCommentDelimStartTable().contains(c);
}
/**
* Gets the next token. If there are no tokens left to read, this will return null.
*
* @return the next token, or null if no more tokens to read.
* @throws IOException if a token cannot be read by the underlying Reader.
*/
public Token nextToken() throws IOException {
boolean breakloop = false;
while (!breakloop) {
char c = 0;
if (isOnDelimBreak()) {
c = delimBreakChar;
clearDelimBreak();
} else c = readChar();
switch (getState()) {
case TYPE_END_OF_LEXER:
{
breakloop = true;
}
break;
case TYPE_UNKNOWN:
{
if (isLexerEnd(c)) {
setState(TYPE_END_OF_LEXER);
breakloop = true;
} else if (isStreamEnd(c)) {
if (kernel.willIncludeStreamBreak()) {
setState(TYPE_END_OF_STREAM);
breakloop = true;
}
streamStack.pop();
} else if (isNewline(c)) {
if (kernel.willIncludeNewlines()) {
setState(TYPE_DELIM_NEWLINE);
breakloop = true;
}
} else if (isSpace(c)) {
if (kernel.willIncludeSpaces()) {
setState(TYPE_DELIM_SPACE);
breakloop = true;
}
} else if (isTab(c)) {
if (kernel.willIncludeTabs()) {
setState(TYPE_DELIM_TAB);
breakloop = true;
}
} else if (isWhitespace(c)) {
} else if (isPoint(c) && isDelimiterStart(c)) {
setState(TYPE_POINT);
saveChar(c);
} else if (isPoint(c) && !isDelimiterStart(c)) {
setState(TYPE_FLOAT);
saveChar(c);
} else if (isSpecialStart(c)) {
setState(TYPE_SPECIAL);
setSpecialType(c);
saveChar(c);
} else if (isStringStart(c)) {
setState(TYPE_STRING);
setStringStartAndEnd(c);
} else if (isDelimiterStart(c)) {
setState(TYPE_DELIMITER);
saveChar(c);
} else if (isUnderscore(c)) {
setState(TYPE_IDENTIFIER);
saveChar(c);
} else if (isLetter(c)) {
setState(TYPE_IDENTIFIER);
saveChar(c);
} else if (c == '0') {
setState(TYPE_HEX_INTEGER0);
saveChar(c);
} else if (isDigit(c)) {
setState(TYPE_NUMBER);
saveChar(c);
} else {
setState(TYPE_ILLEGAL);
saveChar(c);
}
}
break; // end TYPE_START_OF_LEXER
case TYPE_ILLEGAL:
{
if (isStreamEnd(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isNewline(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isSpace(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isTab(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isWhitespace(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isSpecialStart(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isStringStart(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isDelimiterStart(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isLetter(c)) {
saveChar(c);
} else if (isDigit(c)) {
saveChar(c);
} else {
saveChar(c);
}
}
break; // end TYPE_ILLEGAL
case TYPE_POINT: // decimal point is seen, but it is a delimiter.
{
if (isStreamEnd(c)) {
setState(TYPE_DELIMITER);
setDelimBreak(c);
breakloop = true;
} else if (isNewline(c)) {
setState(TYPE_DELIMITER);
setDelimBreak(c);
breakloop = true;
} else if (isSpace(c)) {
setState(TYPE_DELIMITER);
setDelimBreak(c);
breakloop = true;
} else if (isTab(c)) {
setState(TYPE_DELIMITER);
setDelimBreak(c);
breakloop = true;
} else if (isWhitespace(c)) {
setState(TYPE_DELIMITER);
setDelimBreak(c);
breakloop = true;
} else if (isSpecialStart(c)) {
setState(TYPE_DELIMITER);
setDelimBreak(c);
breakloop = true;
} else if (isStringStart(c)) {
setState(TYPE_DELIMITER);
setDelimBreak(c);
breakloop = true;
} else if (isDigit(c)) {
setState(TYPE_FLOAT);
saveChar(c);
} else {
setState(TYPE_DELIMITER);
if (kernel.getDelimTable().containsKey(getCurrentLexeme() + c)) saveChar(c);
else {
setDelimBreak(c);
breakloop = true;
}
}
}
break; // end TYPE_POINT
case TYPE_FLOAT:
{
if (isStreamEnd(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isNewline(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isSpace(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isTab(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isWhitespace(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isExponent(c)) {
setState(TYPE_EXPONENT);
saveChar(c);
} else if (isSpecialStart(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isStringStart(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isDigit(c)) {
saveChar(c);
} else if (isDelimiterStart(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else {
setState(TYPE_ILLEGAL);
saveChar(c);
}
}
break; // end TYPE_FLOAT
case TYPE_IDENTIFIER:
{
if (isStreamEnd(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isNewline(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isSpace(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isTab(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isWhitespace(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isSpecialStart(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isStringStart(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isDelimiterStart(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isLetter(c)) {
saveChar(c);
} else if (isDigit(c)) {
saveChar(c);
} else if (isUnderscore(c)) {
saveChar(c);
} else {
setState(TYPE_ILLEGAL);
saveChar(c);
}
}
break; // end TYPE_IDENTIFIER
case TYPE_SPECIAL:
{
if (isStreamEnd(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isNewline(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isSpace(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isTab(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isWhitespace(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isSpecialStart(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isStringStart(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isDelimiterStart(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isLetter(c)) {
saveChar(c);
} else if (isDigit(c)) {
saveChar(c);
} else {
setState(TYPE_ILLEGAL);
saveChar(c);
}
}
break; // end TYPE_IDENTIFIER
case TYPE_HEX_INTEGER0:
{
if (isStreamEnd(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isNewline(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isSpace(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isTab(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isWhitespace(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isPoint(c)) {
setState(TYPE_FLOAT);
saveChar(c);
} else if (isSpecialStart(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isStringStart(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isDelimiterStart(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (c == 'x' || c == 'X') {
setState(TYPE_HEX_INTEGER1);
saveChar(c);
} else if (isLetter(c)) {
setState(TYPE_ILLEGAL);
saveChar(c);
} else if (isDigit(c)) {
setState(TYPE_NUMBER);
saveChar(c);
} else {
setState(TYPE_ILLEGAL);
saveChar(c);
}
}
break; // end TYPE_HEX_INTEGER0
case TYPE_HEX_INTEGER1:
{
if (isStreamEnd(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isNewline(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isSpace(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isTab(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isWhitespace(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isPoint(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isSpecialStart(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isStringStart(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isDelimiterStart(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isHexDigit(c)) {
setState(TYPE_HEX_INTEGER);
saveChar(c);
} else if (isLetter(c)) {
setState(TYPE_ILLEGAL);
saveChar(c);
} else {
setState(TYPE_ILLEGAL);
saveChar(c);
}
}
break; // end TYPE_HEX_INTEGER1
case TYPE_HEX_INTEGER:
{
if (isStreamEnd(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isNewline(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isSpace(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isTab(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isWhitespace(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isSpecialStart(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isStringStart(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isDelimiterStart(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isHexDigit(c)) {
saveChar(c);
} else if (isLetter(c)) {
setState(TYPE_ILLEGAL);
saveChar(c);
} else {
setState(TYPE_ILLEGAL);
saveChar(c);
}
}
break; // end TYPE_HEX_INTEGER
case TYPE_NUMBER:
{
if (isStreamEnd(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isNewline(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isSpace(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isTab(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isWhitespace(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isPoint(c)) {
setState(TYPE_FLOAT);
saveChar(c);
} else if (isExponent(c)) {
setState(TYPE_EXPONENT);
saveChar(c);
} else if (isSpecialStart(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isStringStart(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isDelimiterStart(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isLetter(c)) {
setState(TYPE_ILLEGAL);
saveChar(c);
} else if (isDigit(c)) {
saveChar(c);
} else {
setState(TYPE_ILLEGAL);
saveChar(c);
}
}
break; // end TYPE_NUMBER
case TYPE_EXPONENT:
{
if (isStreamEnd(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isNewline(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isSpace(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isTab(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isWhitespace(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isExponentSign(c)) {
setState(TYPE_EXPONENT_POWER);
saveChar(c);
} else if (isSpecialStart(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isStringStart(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isDelimiterStart(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isLetter(c)) {
setState(TYPE_ILLEGAL);
saveChar(c);
} else if (isDigit(c)) {
setState(TYPE_EXPONENT_POWER);
saveChar(c);
} else {
setState(TYPE_ILLEGAL);
saveChar(c);
}
}
break; // end TYPE_EXPONENT
case TYPE_EXPONENT_POWER:
{
if (isStreamEnd(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isNewline(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isSpace(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isTab(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isWhitespace(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isSpecialStart(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isStringStart(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else if (isDigit(c)) {
saveChar(c);
} else if (isDelimiterStart(c)) {
setState(TYPE_NUMBER);
setDelimBreak(c);
breakloop = true;
} else {
setState(TYPE_ILLEGAL);
saveChar(c);
}
}
break; // end TYPE_EXPONENT_POWER
case TYPE_STRING:
{
if (isStreamEnd(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isNewline(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else if (isStringEnd(c)) {
breakloop = true;
} else if (isStringEscape(c)) {
c = readChar();
if (isStringEnd(c)) saveChar(c);
else if (isStringEscape(c)) saveChar(c);
else
switch (c) {
case '0':
saveChar('\0');
break;
case 'b':
saveChar('\b');
break;
case 't':
saveChar('\t');
break;
case 'n':
saveChar('\n');
break;
case 'f':
saveChar('\f');
break;
case 'r':
saveChar('\r');
break;
case '/':
saveChar('/');
break;
case 'u':
{
StringBuilder sb = new StringBuilder();
for (int i = 0; i < 4; i++) {
c = readChar();
if (!isHexDigit(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else sb.append(c);
}
if (!breakloop) {
saveChar((char) (Integer.parseInt(sb.toString(), 16) & 0x0ffff));
}
}
break;
case 'x':
{
StringBuilder sb = new StringBuilder();
for (int i = 0; i < 2; i++) {
c = readChar();
if (!isHexDigit(c)) {
setState(TYPE_ILLEGAL);
setDelimBreak(c);
breakloop = true;
} else sb.append(c);
}
if (!breakloop) {
saveChar((char) (Integer.parseInt(sb.toString(), 16) & 0x0ff));
}
}
break;
}
} else {
saveChar(c);
}
}
break; // end TYPE_STRING
case TYPE_DELIMITER:
{
if (isStreamEnd(c)) {
setDelimBreak(c);
breakloop = true;
} else if (kernel.getCommentStartTable().containsKey(getCurrentLexeme() + c)) {
clearCurrentLexeme();
setState(TYPE_COMMENT);
} else if (kernel.getCommentLineTable().containsKey(getCurrentLexeme() + c)) {
clearCurrentLexeme();
setState(TYPE_LINE_COMMENT);
} else if (kernel.getDelimTable().containsKey(getCurrentLexeme() + c)) {
saveChar(c);
} else if (isNewline(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isSpace(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isTab(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isWhitespace(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isSpecialStart(c)) {
setDelimBreak(c);
breakloop = true;
} else if (isStringStart(c)) {
setDelimBreak(c);
breakloop = true;
} else {
setDelimBreak(c);
breakloop = true;
}
}
break; // end TYPE_DELIMITER
case TYPE_COMMENT:
{
if (isStreamEnd(c)) {
clearCurrentLexeme();
setState(TYPE_UNKNOWN);
} else if (kernel.getCommentEndTable().containsKey(getCurrentLexeme())) {
clearCurrentLexeme();
setState(TYPE_UNKNOWN);
} else if (isCommentEndDelimiterStart(c)) {
setState(TYPE_DELIM_COMMENT);
saveChar(c);
}
}
break; // end TYPE_COMMENT
case TYPE_DELIM_COMMENT:
{
if (isStreamEnd(c)) {
clearCurrentLexeme();
setState(TYPE_COMMENT);
} else if (kernel.getCommentEndTable().containsKey(getCurrentLexeme() + c)) {
clearCurrentLexeme();
setState(TYPE_UNKNOWN);
} else if (isWhitespace(c)) {
clearCurrentLexeme();
setState(TYPE_COMMENT);
} else {
clearCurrentLexeme();
saveChar(c);
}
}
break; // end TYPE_DELIM_COMMENT
case TYPE_LINE_COMMENT:
{
if (isStreamEnd(c)) {
clearCurrentLexeme();
setState(TYPE_UNKNOWN);
} else if (isNewline(c)) {
clearCurrentLexeme();
setState(TYPE_UNKNOWN);
}
}
break; // end TYPE_DELIM_COMMENT
}
}
// send token.
int type = getState();
String lexeme = getCurrentLexeme();
clearCurrentLexeme();
switch (getState()) {
case TYPE_DELIM_SPACE:
{
type = TYPE_DELIM_SPACE;
lexeme = " ";
}
break;
case TYPE_DELIM_TAB:
{
type = TYPE_DELIM_TAB;
lexeme = "\t";
}
break;
case TYPE_DELIM_NEWLINE:
{
type = TYPE_DELIM_NEWLINE;
lexeme = "";
}
break;
case TYPE_DELIMITER:
{
type = TYPE_DELIMITER;
if (kernel.getCommentStartTable().containsKey(lexeme))
type = kernel.getCommentStartTable().get(lexeme);
else if (kernel.getCommentEndTable().containsKey(lexeme))
type = kernel.getCommentEndTable().get(lexeme);
else if (kernel.getCommentLineTable().containsKey(lexeme))
type = kernel.getCommentLineTable().get(lexeme);
else if (kernel.getDelimTable().containsKey(lexeme))
type = kernel.getDelimTable().get(lexeme);
}
break;
case TYPE_IDENTIFIER:
{
type = TYPE_IDENTIFIER;
if (kernel.getKeywordTable().containsKey(lexeme))
type = kernel.getKeywordTable().get(lexeme);
else if (kernel.getCaseInsensitiveKeywordTable().containsKey(lexeme))
type = kernel.getCaseInsensitiveKeywordTable().get(lexeme);
}
break;
case TYPE_SPECIAL:
type = specialType;
break;
}
Token out = null;
if (getState() != TYPE_END_OF_LEXER) {
out = makeToken(type, lexeme);
setState(TYPE_UNKNOWN);
}
if (DEBUG) System.out.println(out);
return out;
}
/**
* Gets the special type for a special char.
*
* @param c the character input.
* @return the corresponding type, or {@link LexerKernel#TYPE_UNKNOWN} if no type.
*/
protected int getSpecialType(char c) {
if (!kernel.getSpecialDelimTable().containsKey(c)) return TYPE_UNKNOWN;
return kernel.getSpecialDelimTable().get(c);
}
/**
* Gets the character that ends a String, using the starting character.
*
* @param c the starting character.
* @return the corresponding end character, or the null character ('\0') if this does not end a
* string.
*/
protected char getStringEnd(char c) {
if (!isStringStart(c)) return '\0';
return kernel.getStringDelimTable().get(c);
}
/**
* Checks if this is a character that starts a special token.
*
* @param c the character to test.
* @return true if so, false if not.
*/
protected boolean isSpecialStart(char c) {
return kernel.getSpecialDelimTable().containsKey(c);
}
/**
* Checks if this is a character that starts a String.
*
* @param c the character to test.
* @return true if so, false if not.
*/
protected boolean isStringStart(char c) {
return kernel.getStringDelimTable().containsKey(c);
}
/**
* Checks if a character is a decimal point (depends on locale/kernel).
*
* @param c the character to test.
* @return true if so, false if not.
*/
protected boolean isPoint(char c) {
return kernel.getDecimalSeparator() == c;
}