public char[][] taskTags = null;
public char[][] taskPriorities = null;
public static final boolean DEBUG = false;
+ public static final boolean TRACE = false;
public Scanner() {
this(false, false);
}
return false;
}
}
- public int getNextToken() throws InvalidInputException {
- int htmlPosition = currentPosition;
+ public int getCastOrParen() {
+ int tempPosition = currentPosition;
+ char tempCharacter = currentCharacter;
+ int tempToken = TokenNameLPAREN;
+ boolean found = false;
+ StringBuffer buf = new StringBuffer();
try {
- while (!phpMode) {
+ do {
currentCharacter = source[currentPosition++];
- if (currentCharacter == '<') {
- if (getNextChar('?')) {
- currentCharacter = source[currentPosition++];
- if ((currentCharacter == ' ')
- || Character.isWhitespace(currentCharacter)) {
- // <?
- startPosition = currentPosition;
- phpMode = true;
- if (tokenizeWhiteSpace) {
- // && (whiteStart != currentPosition - 1)) {
- // reposition scanner in case we are interested by spaces as
- // tokens
- startPosition = htmlPosition;
- return TokenNameHTML;
+ } while (currentCharacter == ' ' || currentCharacter == '\t');
+ while ((currentCharacter >= 'a' && currentCharacter <= 'z')
+ || (currentCharacter >= 'A' && currentCharacter <= 'Z')) {
+ buf.append(currentCharacter);
+ currentCharacter = source[currentPosition++];
+ }
+ if (buf.length() >= 3 && buf.length() <= 7) {
+ char[] data = buf.toString().toCharArray();
+ int index = 0;
+ switch (data.length) {
+ case 3 :
+ // int
+ if ((data[index] == 'i') && (data[++index] == 'n')
+ && (data[++index] == 't')) {
+ found = true;
+ tempToken = TokenNameintCAST;
+ }
+ break;
+ case 4 :
+ // bool real
+ if ((data[index] == 'b') && (data[++index] == 'o')
+ && (data[++index] == 'o') && (data[++index] == 'l')) {
+ found = true;
+ tempToken = TokenNameboolCAST;
+ } else {
+ index = 0;
+ if ((data[index] == 'r') && (data[++index] == 'e')
+ && (data[++index] == 'a') && (data[++index] == 'l')) {
+ found = true;
+ tempToken = TokenNamedoubleCAST;
}
+ }
+ break;
+ case 5 :
+ // array unset float
+ if ((data[index] == 'a') && (data[++index] == 'r')
+ && (data[++index] == 'r') && (data[++index] == 'a')
+ && (data[++index] == 'y')) {
+ found = true;
+ tempToken = TokenNamearrayCAST;
} else {
- boolean phpStart = (currentCharacter == 'P')
- || (currentCharacter == 'p');
- if (phpStart) {
- int test = getNextChar('H', 'h');
- if (test >= 0) {
- test = getNextChar('P', 'p');
- if (test >= 0) {
- // <?PHP <?php
- startPosition = currentPosition;
- phpMode = true;
- if (tokenizeWhiteSpace) {
- // && (whiteStart != currentPosition - 1)) {
- // reposition scanner in case we are interested by spaces
- // as tokens
- startPosition = htmlPosition;
- return TokenNameHTML;
- }
- }
+ index = 0;
+ if ((data[index] == 'u') && (data[++index] == 'n')
+ && (data[++index] == 's') && (data[++index] == 'e')
+ && (data[++index] == 't')) {
+ found = true;
+ tempToken = TokenNameunsetCAST;
+ } else {
+ index = 0;
+ if ((data[index] == 'f') && (data[++index] == 'l')
+ && (data[++index] == 'o') && (data[++index] == 'a')
+ && (data[++index] == 't')) {
+ found = true;
+ tempToken = TokenNamedoubleCAST;
}
}
}
- }
+ break;
+ case 6 :
+ // object string double
+ if ((data[index] == 'o') && (data[++index] == 'b')
+ && (data[++index] == 'j') && (data[++index] == 'e')
+ && (data[++index] == 'c') && (data[++index] == 't')) {
+ found = true;
+ tempToken = TokenNameobjectCAST;
+ } else {
+ index = 0;
+ if ((data[index] == 's') && (data[++index] == 't')
+ && (data[++index] == 'r') && (data[++index] == 'i')
+ && (data[++index] == 'n') && (data[++index] == 'g')) {
+ found = true;
+ tempToken = TokenNamestringCAST;
+ } else {
+ index = 0;
+ if ((data[index] == 'd') && (data[++index] == 'o')
+ && (data[++index] == 'u') && (data[++index] == 'b')
+ && (data[++index] == 'l') && (data[++index] == 'e')) {
+ found = true;
+ tempToken = TokenNamedoubleCAST;
+ }
+ }
+ }
+ break;
+ case 7 :
+ // boolean integer
+ if ((data[index] == 'b') && (data[++index] == 'o')
+ && (data[++index] == 'o') && (data[++index] == 'l')
+ && (data[++index] == 'e') && (data[++index] == 'a')
+ && (data[++index] == 'n')) {
+ found = true;
+ tempToken = TokenNameboolCAST;
+ } else {
+ index = 0;
+ if ((data[index] == 'i') && (data[++index] == 'n')
+ && (data[++index] == 't') && (data[++index] == 'e')
+ && (data[++index] == 'g') && (data[++index] == 'e')
+ && (data[++index] == 'r')) {
+ found = true;
+ tempToken = TokenNameintCAST;
+ }
+ }
+ break;
}
- if ((currentCharacter == '\r') || (currentCharacter == '\n')) {
- if (recordLineSeparator) {
- pushLineSeparator();
- } else {
- currentLine = null;
+ if (found) {
+ while (currentCharacter == ' ' || currentCharacter == '\t') {
+ currentCharacter = source[currentPosition++];
+ }
+ if (currentCharacter == ')') {
+ return tempToken;
}
}
}
- } //-----------------end switch while try--------------------
- catch (IndexOutOfBoundsException e) {
- if (tokenizeWhiteSpace) {
- // && (whiteStart != currentPosition - 1)) {
- // reposition scanner in case we are interested by spaces as tokens
- startPosition = htmlPosition;
- }
- return TokenNameEOF;
+ } catch (IndexOutOfBoundsException e) {
+ }
+ currentCharacter = tempCharacter;
+ currentPosition = tempPosition;
+ return TokenNameLPAREN;
+ }
+ public int getNextToken() throws InvalidInputException {
+ if (!phpMode) {
+ return getInlinedHTML(currentPosition);
}
if (phpMode) {
this.wasAcr = false;
// ---------Identify the next token-------------
switch (currentCharacter) {
case '(' :
- return TokenNameLPAREN;
+ return getCastOrParen();
case ')' :
return TokenNameRPAREN;
case '{' :
case ',' :
return TokenNameCOMMA;
case '.' :
+ if (getNextChar('='))
+ return TokenNameDOT_EQUAL;
if (getNextCharAsDigit())
return scanNumber(true);
return TokenNameDOT;
return TokenNameREMAINDER;
case '<' :
{
- int test;
- if ((test = getNextChar('=', '<')) == 0)
- return TokenNameLESS_EQUAL;
- if (test > 0) {
- if (getNextChar('='))
- return TokenNameLEFT_SHIFT_EQUAL;
- if (getNextChar('<')) {
- int heredocStart = currentPosition;
- int heredocLength = 0;
- currentCharacter = source[currentPosition++];
- if (isPHPIdentifierStart(currentCharacter)) {
- currentCharacter = source[currentPosition++];
- } else {
- return TokenNameERROR;
- }
- while (isPHPIdentifierPart(currentCharacter)) {
+ int oldPosition = currentPosition;
+ try {
+ currentCharacter = source[currentPosition++];
+ } catch (IndexOutOfBoundsException e) {
+ currentPosition = oldPosition;
+ return TokenNameLESS;
+ }
+ switch (currentCharacter) {
+ case '=' :
+ return TokenNameLESS_EQUAL;
+ case '>' :
+ return TokenNameNOT_EQUAL;
+ case '<' :
+ if (getNextChar('='))
+ return TokenNameLEFT_SHIFT_EQUAL;
+ if (getNextChar('<')) {
+ int heredocStart = currentPosition;
+ int heredocLength = 0;
currentCharacter = source[currentPosition++];
- }
- heredocLength = currentPosition - heredocStart - 1;
- // heredoc end-tag determination
- boolean endTag = true;
- char ch;
- do {
- ch = source[currentPosition++];
- if (ch == '\r' || ch == '\n') {
- if (recordLineSeparator) {
- pushLineSeparator();
- } else {
- currentLine = null;
- }
- for (int i = 0; i < heredocLength; i++) {
- if (source[currentPosition + i] != source[heredocStart
- + i]) {
- endTag = false;
- break;
+ if (isPHPIdentifierStart(currentCharacter)) {
+ currentCharacter = source[currentPosition++];
+ } else {
+ return TokenNameERROR;
+ }
+ while (isPHPIdentifierPart(currentCharacter)) {
+ currentCharacter = source[currentPosition++];
+ }
+ heredocLength = currentPosition - heredocStart - 1;
+ // heredoc end-tag determination
+ boolean endTag = true;
+ char ch;
+ do {
+ ch = source[currentPosition++];
+ if (ch == '\r' || ch == '\n') {
+ if (recordLineSeparator) {
+ pushLineSeparator();
+ } else {
+ currentLine = null;
+ }
+ for (int i = 0; i < heredocLength; i++) {
+ if (source[currentPosition + i] != source[heredocStart
+ + i]) {
+ endTag = false;
+ break;
+ }
+ }
+ if (endTag) {
+ currentPosition += heredocLength - 1;
+ currentCharacter = source[currentPosition++];
+ break; // do...while loop
+ } else {
+ endTag = true;
}
}
- if (endTag) {
- currentPosition += heredocLength - 1;
- currentCharacter = source[currentPosition++];
- break; // do...while loop
- } else {
- endTag = true;
- }
- }
- } while (true);
- return TokenNameHEREDOC;
- }
- return TokenNameLEFT_SHIFT;
+ } while (true);
+ return TokenNameHEREDOC;
+ }
+ return TokenNameLEFT_SHIFT;
}
+ currentPosition = oldPosition;
return TokenNameLESS;
}
case '>' :
case '?' :
if (getNextChar('>')) {
phpMode = false;
- return TokenNameStopPHP;
+ if (currentPosition==source.length) {
+ phpMode = true;
+ return TokenNameINLINE_HTML;
+ }
+ return getInlinedHTML(currentPosition - 2);
}
return TokenNameQUESTION;
case ':' :
case '#' :
case '/' :
{
+ char startChar = currentCharacter;
+ if (getNextChar('=')) {
+ return TokenNameDIVIDE_EQUAL;
+ }
int test;
- if ((currentCharacter == '#')
- || (test = getNextChar('/', '*')) == 0) {
+ if ((startChar == '#') || (test = getNextChar('/', '*')) == 0) {
//line comment
int endPositionForLineComment = 0;
try { //get the next char
if (getNextChar('>')) {
startPosition = currentPosition - 2;
phpMode = false;
- return TokenNameStopPHP;
+ return TokenNameINLINE_HTML;
}
}
//get the next char
}
break;
}
- if (getNextChar('='))
- return TokenNameDIVIDE_EQUAL;
return TokenNameDIVIDE;
}
case '\u001a' :
throw new InvalidInputException("Ctrl-Z"); //$NON-NLS-1$
default :
if (currentCharacter == '$') {
- while ((currentCharacter = source[currentPosition++]) == '$') {
+ int oldPosition = currentPosition;
+ try {
+ currentCharacter = source[currentPosition++];
+ if (isPHPIdentifierStart(currentCharacter)) {
+ return scanIdentifierOrKeyword(true);
+ } else {
+ currentPosition = oldPosition;
+ return TokenNameDOLLAR;
+ }
+ } catch (IndexOutOfBoundsException e) {
+ currentPosition = oldPosition;
+ return TokenNameDOLLAR;
}
- if (currentCharacter == '{')
- return TokenNameDOLLAR_LBRACE;
- if (isPHPIdentifierStart(currentCharacter))
- return scanIdentifierOrKeyword(true);
- return TokenNameERROR;
}
if (isPHPIdentifierStart(currentCharacter))
return scanIdentifierOrKeyword(false);
}
return TokenNameEOF;
}
+ /**
+ * @return @throws
+ * InvalidInputException
+ */
+ private int getInlinedHTML(int start) throws InvalidInputException {
+ // int htmlPosition = start;
+ if (currentPosition>source.length) {
+ currentPosition = source.length;
+ return TokenNameEOF;
+ }
+ startPosition = start;
+ try {
+ while (!phpMode) {
+ currentCharacter = source[currentPosition++];
+ if (currentCharacter == '<') {
+ if (getNextChar('?')) {
+ currentCharacter = source[currentPosition++];
+ if ((currentCharacter == ' ')
+ || Character.isWhitespace(currentCharacter)) {
+ // <?
+ phpMode = true;
+ return TokenNameINLINE_HTML;
+ } else {
+ boolean phpStart = (currentCharacter == 'P')
+ || (currentCharacter == 'p');
+ if (phpStart) {
+ int test = getNextChar('H', 'h');
+ if (test >= 0) {
+ test = getNextChar('P', 'p');
+ if (test >= 0) {
+ // <?PHP <?php
+ phpMode = true;
+ return TokenNameINLINE_HTML;
+ }
+ }
+ }
+ }
+ }
+ }
+ if ((currentCharacter == '\r') || (currentCharacter == '\n')) {
+ if (recordLineSeparator) {
+ pushLineSeparator();
+ } else {
+ currentLine = null;
+ }
+ }
+ } //-----------------while--------------------
+ phpMode = true;
+ return TokenNameINLINE_HTML;
+ } //-----------------try--------------------
+ catch (IndexOutOfBoundsException e) {
+ startPosition = start;
+ currentPosition--;
+ }
+ phpMode = true;
+ return TokenNameINLINE_HTML;
+ }
// public final void getNextUnicodeChar()
// throws IndexOutOfBoundsException, InvalidInputException {
// //VOID
while (getNextCharAsJavaIdentifierPart()) {
};
if (isVariable) {
- if (new String(getCurrentTokenSource()).equals("$this")) {
- return TokenNamethis;
- }
+ // if (new String(getCurrentTokenSource()).equals("$this")) {
+ // return TokenNamethis;
+ // }
return TokenNameVariable;
}
int index, length;
case 3 :
//and
if ((data[++index] == 'n') && (data[++index] == 'd')) {
- return TokenNameAND;
+ return TokenNameand;
} else {
return TokenNameIdentifier;
}
return TokenNameIdentifier;
}
case 'c' :
- //case catch class const continue
+ //case catch class clone const continue
switch (length) {
case 4 :
if ((data[++index] == 'a') && (data[++index] == 's')
if ((data[++index] == 'a') && (data[++index] == 't')
&& (data[++index] == 'c') && (data[++index] == 'h'))
return TokenNamecatch;
- if ((data[index] == 'l') && (data[++index] == 'a')
+ index = 0;
+ if ((data[++index] == 'l') && (data[++index] == 'a')
&& (data[++index] == 's') && (data[++index] == 's'))
return TokenNameclass;
- if ((data[index] == 'o') && (data[++index] == 'n')
+ index = 0;
+ if ((data[++index] == 'l') && (data[++index] == 'o')
+ && (data[++index] == 'n') && (data[++index] == 'e'))
+ return TokenNameclone;
+ index = 0;
+ if ((data[++index] == 'o') && (data[++index] == 'n')
&& (data[++index] == 's') && (data[++index] == 't'))
return TokenNameconst;
else
else
return TokenNameIdentifier;
case 5 :
-// if ((data[++index] == 'a') && (data[++index] == 'l')
-// && (data[++index] == 's') && (data[++index] == 'e'))
-// return TokenNamefalse;
+ // if ((data[++index] == 'a') && (data[++index] == 'l')
+ // && (data[++index] == 's') && (data[++index] == 'e'))
+ // return TokenNamefalse;
if ((data[++index] == 'i') && (data[++index] == 'n')
&& (data[++index] == 'a') && (data[++index] == 'l'))
return TokenNamefinal;
return TokenNamenew;
else
return TokenNameIdentifier;
-// case 4 :
-// if ((data[++index] == 'u') && (data[++index] == 'l')
-// && (data[++index] == 'l'))
-// return TokenNamenull;
-// else
-// return TokenNameIdentifier;
+ // case 4 :
+ // if ((data[++index] == 'u') && (data[++index] == 'l')
+ // && (data[++index] == 'l'))
+ // return TokenNamenull;
+ // else
+ // return TokenNameIdentifier;
default :
return TokenNameIdentifier;
}
// or old_function
if (length == 2) {
if (data[++index] == 'r') {
- return TokenNameOR;
+ return TokenNameor;
}
}
// if (length == 12) {
return TokenNametry;
else
return TokenNameIdentifier;
-// case 4 :
-// if ((data[++index] == 'r') && (data[++index] == 'u')
-// && (data[++index] == 'e'))
-// return TokenNametrue;
-// else
-// return TokenNameIdentifier;
+ // case 4 :
+ // if ((data[++index] == 'r') && (data[++index] == 'u')
+ // && (data[++index] == 'e'))
+ // return TokenNametrue;
+ // else
+ // return TokenNameIdentifier;
case 5 :
if ((data[++index] == 'h') && (data[++index] == 'r')
&& (data[++index] == 'o') && (data[++index] == 'w'))
switch (length) {
case 3 :
if ((data[++index] == 'o') && (data[++index] == 'r'))
- return TokenNameXOR;
+ return TokenNamexor;
else
return TokenNameIdentifier;
default :
case TokenNameERROR :
return "ScannerError"; // + new String(getCurrentTokenSource()) + ")";
// //$NON-NLS-1$
- case TokenNameStopPHP :
- return "StopPHP(" + new String(getCurrentTokenSource()) + ")"; //$NON-NLS-1$ //$NON-NLS-2$
+ case TokenNameINLINE_HTML :
+ return "Inline-HTML(" + new String(getCurrentTokenSource()) + ")"; //$NON-NLS-1$ //$NON-NLS-2$
case TokenNameIdentifier :
return "Identifier(" + new String(getCurrentTokenSource()) + ")"; //$NON-NLS-1$ //$NON-NLS-2$
case TokenNameVariable :
return "Variable(" + new String(getCurrentTokenSource()) + ")"; //$NON-NLS-1$ //$NON-NLS-2$
case TokenNameabstract :
return "abstract"; //$NON-NLS-1$
+ case TokenNameand :
+ return "AND"; //$NON-NLS-1$
case TokenNamearray :
return "array"; //$NON-NLS-1$
case TokenNameas :
return "case"; //$NON-NLS-1$
case TokenNameclass :
return "class"; //$NON-NLS-1$
+ case TokenNameclone :
+ //$NON-NLS-1$
+ return "clone";
+ case TokenNameconst :
+ //$NON-NLS-1$
+ return "const";
case TokenNamecontinue :
return "continue"; //$NON-NLS-1$
case TokenNamedefault :
return "endwhile"; //$NON-NLS-1$
case TokenNameextends :
return "extends"; //$NON-NLS-1$
-// case TokenNamefalse :
-// return "false"; //$NON-NLS-1$
+ // case TokenNamefalse :
+ // return "false"; //$NON-NLS-1$
case TokenNamefinal :
return "final"; //$NON-NLS-1$
case TokenNamefor :
return "include_once"; //$NON-NLS-1$
case TokenNameinterface :
return "interface"; //$NON-NLS-1$
+ case TokenNameisset :
+ return "isset"; //$NON-NLS-1$
case TokenNamelist :
return "list"; //$NON-NLS-1$
case TokenNamenew :
return "new"; //$NON-NLS-1$
-// case TokenNamenull :
-// return "null"; //$NON-NLS-1$
+ // case TokenNamenull :
+ // return "null"; //$NON-NLS-1$
+ case TokenNameor :
+ return "OR"; //$NON-NLS-1$
case TokenNameprint :
return "print"; //$NON-NLS-1$
case TokenNameprivate :
return "static"; //$NON-NLS-1$
case TokenNameswitch :
return "switch"; //$NON-NLS-1$
-// case TokenNametrue :
-// return "true"; //$NON-NLS-1$
+ // case TokenNametrue :
+ // return "true"; //$NON-NLS-1$
case TokenNameunset :
return "unset"; //$NON-NLS-1$
case TokenNamevar :
return "var"; //$NON-NLS-1$
case TokenNamewhile :
return "while"; //$NON-NLS-1$
- case TokenNamethis :
- return "$this"; //$NON-NLS-1$
+ case TokenNamexor :
+ return "XOR"; //$NON-NLS-1$
+ // case TokenNamethis :
+ // return "$this"; //$NON-NLS-1$
case TokenNameIntegerLiteral :
return "Integer(" + new String(getCurrentTokenSource()) + ")"; //$NON-NLS-1$ //$NON-NLS-2$
case TokenNameDoubleLiteral :
return "^="; //$NON-NLS-1$
case TokenNameREMAINDER_EQUAL :
return "%="; //$NON-NLS-1$
+ case TokenNameDOT_EQUAL :
+ return ".="; //$NON-NLS-1$
case TokenNameLEFT_SHIFT_EQUAL :
return "<<="; //$NON-NLS-1$
case TokenNameRIGHT_SHIFT_EQUAL :
return "="; //$NON-NLS-1$
case TokenNameAT :
return "@";
- case TokenNameDOLLAR_LBRACE :
- return "${";
+ case TokenNameDOLLAR :
+ return "$";
+ // case TokenNameDOLLAR_LBRACE :
+ // return "${";
case TokenNameEOF :
return "EOF"; //$NON-NLS-1$
case TokenNameWHITESPACE :
return "COMMENT_BLOCK(" + new String(getCurrentTokenSource()) + ")"; //$NON-NLS-1$
case TokenNameCOMMENT_PHPDOC :
return "COMMENT_PHPDOC(" + new String(getCurrentTokenSource()) + ")"; //$NON-NLS-1$
- case TokenNameHTML :
- return "HTML(" + new String(getCurrentTokenSource()) + ")"; //$NON-NLS-1$
+ // case TokenNameHTML :
+ // return "HTML(" + new String(getCurrentTokenSource()) + ")";
+ // //$NON-NLS-1$
case TokenNameFILE :
return "__FILE__"; //$NON-NLS-1$
case TokenNameLINE :
case TokenNameMETHOD_C :
return "__METHOD__"; //$NON-NLS-1$
case TokenNameFUNC_C :
- return "__FUNCTION__"; //$NON-NLS-1$
+ return "__FUNCTION__"; //$NON-NLS-1
+ case TokenNameboolCAST :
+ return "( bool )"; //$NON-NLS-1$
+ case TokenNameintCAST :
+ return "( int )"; //$NON-NLS-1$
+ case TokenNamedoubleCAST :
+ return "( double )"; //$NON-NLS-1$
+ case TokenNameobjectCAST :
+ return "( object )"; //$NON-NLS-1$
+ case TokenNamestringCAST :
+ return "( string )"; //$NON-NLS-1$
default :
return "not-a-token(" + (new Integer(act)) + ") "
+ new String(getCurrentTokenSource()); //$NON-NLS-1$
git://git.phpeclipse.com / phpeclipse.git / blobdiff