1 // Derived from org.mozilla.javascript.TokenStream [NPL]
4 * The contents of this file are subject to the Netscape Public
5 * License Version 1.1 (the "License"); you may not use this file
6 * except in compliance with the License. You may obtain a copy of
7 * the License at http://www.mozilla.org/NPL/
9 * Software distributed under the License is distributed on an "AS
10 * IS" basis, WITHOUT WARRANTY OF ANY KIND, either express or
11 * implied. See the License for the specific language governing
12 * rights and limitations under the License.
14 * The Initial Developer of the Original Code is Netscape
15 * Communications Corporation.
17 * Contributor(s): Roger Lawrence, Mike McCabe
23 /** Lexes a stream of characters into a stream of Tokens */
24 class Lexer implements Tokens {
27 public static void main(String[] s) throws Exception {
28 Lexer l = new Lexer(new InputStreamReader(System.in), "stdin", 0);
30 while((tok = l.getToken()) != -1) System.out.println(codeToString[tok]);
33 /** the token that was just parsed */
36 /** the most recently parsed token, <i>regardless of pushbacks</i> */
37 protected int mostRecentlyReadToken;
39 /** if the token just parsed was a NUMBER, this is the numeric value */
40 protected Number number = null;
42 /** if the token just parsed was a NAME or STRING, this is the string value */
43 protected String string = null;
45 /** the line number of the most recently <i>lexed</i> token */
48 /** the line number of the most recently <i>parsed</i> token */
49 protected int parserLine = 0;
51 /** the column number of the current token */
52 protected int col = 0;
54 /** the name of the source code file being lexed */
55 protected String sourceName;
57 private SmartReader in;
58 public Lexer(Reader r, String sourceName, int line) throws IOException {
59 this.sourceName = sourceName;
61 this.parserLine = line;
62 in = new SmartReader(r);
66 // Predicates ///////////////////////////////////////////////////////////////////////
68 private static boolean isAlpha(int c) { return ((c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z')); }
69 private static boolean isDigit(int c) { return (c >= '0' && c <= '9'); }
70 private static int xDigitToInt(int c) {
71 if ('0' <= c && c <= '9') return c - '0';
72 else if ('a' <= c && c <= 'f') return c - ('a' - 10);
73 else if ('A' <= c && c <= 'F') return c - ('A' - 10);
78 // Token Subtype Handlers /////////////////////////////////////////////////////////
80 private int getKeyword(String s) throws IOException {
83 case 2: c=s.charAt(1);
84 if (c=='f') { if (s.charAt(0)=='i') return IF; }
85 else if (c=='n') { if (s.charAt(0)=='i') return IN; }
86 else if (c=='o') { if (s.charAt(0)=='d') return DO; }
88 case 3: switch (s.charAt(0)) {
89 case 'a': if (s.charAt(2)=='d' && s.charAt(1)=='n') return AND; break;
90 case 'f': if (s.charAt(2)=='r' && s.charAt(1)=='o') return FOR; break;
91 case 'i': if (s.charAt(2)=='t' && s.charAt(1)=='n') return RESERVED;
92 case 'n': if (s.charAt(2)=='w' && s.charAt(1)=='e') return RESERVED;
93 case 't': if (s.charAt(2)=='y' && s.charAt(1)=='r') return TRY; break;
94 case 'v': if (s.charAt(2)=='r' && s.charAt(1)=='a') return VAR; break;
96 case 4: switch (s.charAt(0)) {
97 case 'b': return s.equals("byte") ? RESERVED : -1;
98 case 'c': c=s.charAt(3);
99 if (c=='e') { if (s.charAt(2)=='s' && s.charAt(1)=='a') return CASE; }
100 else if (c=='r') { if (s.charAt(2)=='a' && s.charAt(1)=='h') return RESERVED; }
102 case 'e': c=s.charAt(3);
103 if (c=='e') { if (s.charAt(2)=='s' && s.charAt(1)=='l') return ELSE; }
104 else if (c=='m') { if (s.charAt(2)=='u' && s.charAt(1)=='n') return RESERVED; }
106 case 'g': return s.equals("goto") ? RESERVED : -1;
107 case 'l': return s.equals("long") ? RESERVED : -1;
108 case 'n': return s.equals("null") ? NULL : -1;
109 case 't': c=s.charAt(3);
110 if (c=='e') { if (s.charAt(2)=='u' && s.charAt(1)=='r') return TRUE; }
112 case 'w': if (s.equals("with")) return RESERVED; else return -1;
113 case 'v': if (s.equals("void")) return RESERVED; else return -1;
115 case 5: switch (s.charAt(2)) {
116 case 'a': return s.equals("class") ? RESERVED : -1;
117 case 'e': return s.equals("break") ? BREAK : -1;
118 case 'i': return s.equals("while") ? WHILE : -1;
119 case 'l': return s.equals("false") ? FALSE : -1;
120 case 'n': c=s.charAt(0);
121 if (s.equals("const")) return RESERVED;
122 else if (s.equals("final")) return RESERVED;
124 case 'o': c=s.charAt(0);
125 if (c == 'c') return s.equals("float") ? RESERVED : -1;
126 else if (c=='s') return s.equals("final") ? RESERVED : -1;
128 case 'p': return s.equals("super") ? RESERVED : -1;
129 case 'r': return s.equals("throw") ? THROW : -1;
130 case 't': return s.equals("catch") ? CATCH : -1;
132 case 6: switch (s.charAt(1)) {
133 case 'a': return s.equals("class") ? RESERVED : -1;
134 case 'e': c=s.charAt(0);
135 if (s.equals("delete")) return RESERVED;
136 else if (c=='r') return s.equals("return") ? RETURN : -1;
138 case 'h': return s.equals("throws") ? RESERVED : -1;
139 case 'o': return s.equals("double") ? RESERVED : -1;
140 case 's': return s.equals("assert") ? ASSERT : -1;
141 case 'u': return s.equals("public") ? RESERVED : -1;
142 case 'w': return s.equals("switch") ? SWITCH : -1;
143 case 'y': return s.equals("typeof") ? TYPEOF : -1;
145 case 7: switch (s.charAt(1)) {
146 case 'a': return s.equals("package") ? RESERVED : -1;
147 case 'e': return s.equals("default") ? DEFAULT : -1;
148 case 'i': return s.equals("finally") ? FINALLY : -1;
149 case 'o': return s.equals("boolean") ? RESERVED : -1;
150 case 'r': return s.equals("private") ? RESERVED : -1;
151 case 'x': return s.equals("extends") ? RESERVED : -1;
153 case 8: switch (s.charAt(0)) {
154 case 'a': return s.equals("abstract") ? RESERVED : -1;
155 case 'c': return s.equals("continue") ? CONTINUE : -1;
156 case 'd': return s.equals("debugger") ? RESERVED : -1;
157 case 'f': return s.equals("function") ? FUNCTION : -1;
158 case 'v': return s.equals("volatile") ? RESERVED : -1;
160 case 9: c=s.charAt(0);
161 if (c=='i') return s.equals("interface") ? RESERVED : -1;
162 else if (c=='p') return s.equals("protected") ? RESERVED : -1;
163 else if (c=='t') return s.equals("transient") ? RESERVED : -1;
165 case 10: c=s.charAt(1);
166 if (c=='m') return s.equals("implements") ? RESERVED : -1;
167 else if (c=='n' && s.equals("instanceof")) return RESERVED;
169 case 12: return s.equals("synchronized") ? RESERVED : -1;
174 private int getIdentifier(int c) throws IOException {
176 while (Character.isJavaIdentifierPart((char)(c = in.read())));
178 String str = in.getString();
179 int result = getKeyword(str);
180 if (result == RESERVED) throw new LexerException("The reserved word \"" + str + "\" is not permitted in XWT scripts");
181 if (result != -1) return result;
182 this.string = str.intern();
186 private int getNumber(int c) throws IOException {
189 double dval = Double.NaN;
191 boolean isInteger = true;
193 // figure out what base we're using
195 if (Character.toLowerCase((char)(c = in.read())) == 'x') { base = 16; in.startString(); }
196 else if (isDigit(c)) base = 8;
199 while (0 <= xDigitToInt(c) && !(base < 16 && isAlpha(c))) c = in.read();
200 if (base == 10 && (c == '.' || c == 'e' || c == 'E')) {
202 if (c == '.') do { c = in.read(); } while (isDigit(c));
203 if (c == 'e' || c == 'E') {
205 if (c == '+' || c == '-') c = in.read();
206 if (!isDigit(c)) throw new LexerException("float listeral did not have an exponent value");
207 do { c = in.read(); } while (isDigit(c));
212 String numString = in.getString();
213 if (base == 10 && !isInteger) {
214 try { dval = (Double.valueOf(numString)).doubleValue(); }
215 catch (NumberFormatException ex) { throw new LexerException("invalid numeric literal: \"" + numString + "\""); }
218 longval = Long.parseLong(numString, base);
219 dval = (double)longval;
221 dval = Double.parseDouble(numString);
222 longval = (long) dval;
223 if (longval == dval) isInteger = true;
227 if (!isInteger) this.number = new Double(dval);
228 else if (Byte.MIN_VALUE <= longval && longval <= Byte.MAX_VALUE) this.number = new Byte((byte)longval);
229 else if (Short.MIN_VALUE <= longval && longval <= Short.MAX_VALUE) this.number = new Short((short)longval);
230 else if (Integer.MIN_VALUE <= longval && longval <= Integer.MAX_VALUE) this.number = new Integer((int)longval);
231 else this.number = new Double(longval);
235 private int getString(int c) throws IOException {
236 StringBuffer stringBuf = null;
240 in.startString(); // start after the first "
241 while(c != quoteChar) {
242 if (c == '\n' || c == -1) throw new LexerException("unterminated string literal");
244 if (stringBuf == null) {
245 in.unread(); // Don't include the backslash
246 stringBuf = new StringBuffer(in.getString());
249 switch (c = in.read()) {
250 case 'b': c = '\b'; break;
251 case 'f': c = '\f'; break;
252 case 'n': c = '\n'; break;
253 case 'r': c = '\r'; break;
254 case 't': c = '\t'; break;
255 case 'v': c = '\u000B'; break;
256 case '\\': c = '\\'; break;
259 for(int i=0; i<4; i++) {
261 if (!((ci >= '0' && ci <= '9') || (ci >= 'a' && ci <= 'f') || (ci >= 'A' && ci <= 'F')))
262 throw new LexerException("illegal character '" + ((char)c) + "' in \\u unicode escape sequence");
263 v = (v << 8) | Integer.parseInt(ci + "", 16);
269 // just use the character that was escaped
273 if (stringBuf != null) stringBuf.append((char) c);
276 if (stringBuf != null) this.string = stringBuf.toString().intern();
278 in.unread(); // miss the trailing "
279 this.string = in.getString().intern();
285 private int _getToken() throws IOException {
287 do { c = in.read(); } while (c == '\u0020' || c == '\u0009' || c == '\u000C' || c == '\u000B' || c == '\n' );
288 if (c == -1) return -1;
289 if (c == '\\' || Character.isJavaIdentifierStart((char)c)) return getIdentifier(c);
290 if (isDigit(c) || (c == '.' && isDigit(in.peek()))) return getNumber(c);
291 if (c == '"' || c == '\'') return getString(c);
293 case ';': return SEMI;
300 case ',': return COMMA;
301 case '?': return HOOK;
302 case ':': return COLON;
303 case '.': return DOT;
304 case '|': return in.match('|') ? OR : (in.match('=') ? ASSIGN_BITOR : BITOR);
305 case '^': return in.match('=') ? ASSIGN_BITXOR : BITXOR;
306 case '&': return in.match('&') ? AND : in.match('=') ? ASSIGN_BITAND : BITAND;
307 case '=': return !in.match('=') ? ASSIGN : in.match('=') ? SHEQ : EQ;
308 case '!': return !in.match('=') ? BANG : in.match('=') ? SHNE : NE;
309 case '%': return in.match('=') ? ASSIGN_MOD : MOD;
310 case '~': return BITNOT;
311 case '+': return in.match('=') ? ASSIGN_ADD : in.match('+') ? INC : ADD;
312 case '-': return in.match('=') ? ASSIGN_SUB: in.match('-') ? DEC : SUB;
313 case '*': return in.match('=') ? ASSIGN_MUL : MUL;
314 case '<': return !in.match('<') ? (in.match('=') ? LE : LT) : in.match('=') ? ASSIGN_LSH : LSH;
315 case '>': return !in.match('>') ? (in.match('=') ? GE : GT) :
316 in.match('>') ? (in.match('=') ? ASSIGN_URSH : URSH) : (in.match('=') ? ASSIGN_RSH : RSH);
318 if (in.match('=')) return ASSIGN_DIV;
319 if (in.match('/')) { while ((c = in.read()) != -1 && c != '\n'); in.unread(); return getToken(); }
320 if (!in.match('*')) return DIV;
321 while ((c = in.read()) != -1 && !(c == '*' && in.match('/'))) {
322 if (c == '\n' || c != '/' || !in.match('*')) continue;
323 if (in.match('/')) return getToken();
324 throw new LexerException("nested comments are not permitted");
326 if (c == -1) throw new LexerException("unterminated comment");
327 return getToken(); // `goto retry'
328 default: throw new LexerException("illegal character: \'" + ((char)c) + "\'");
333 // SmartReader ////////////////////////////////////////////////////////////////
335 /** a Reader that tracks line numbers and can push back tokens */
336 private class SmartReader {
337 PushbackReader reader = null;
340 public SmartReader(Reader r) { reader = new PushbackReader(r); }
341 public void unread() throws IOException { unread((char)lastread); }
342 public void unread(char c) throws IOException {
344 if(c == '\n') col = -1;
346 if (accumulator != null) accumulator.setLength(accumulator.length() - 1);
348 public boolean match(char c) throws IOException { if (peek() == c) { reader.read(); return true; } else return false; }
349 public int peek() throws IOException {
350 int peeked = reader.read();
351 if (peeked != -1) reader.unread((char)peeked);
354 public int read() throws IOException {
355 lastread = reader.read();
356 if (accumulator != null) accumulator.append((char)lastread);
357 if (lastread != '\n' && lastread != '\r') col++;
358 if (lastread == '\n') {
359 // col is -1 if we just unread a newline, this is sort of ugly
360 if (col != -1) parserLine = ++line;
366 // FEATURE: could be much more efficient
367 StringBuffer accumulator = null;
368 public void startString() {
369 accumulator = new StringBuffer();
370 accumulator.append((char)lastread);
372 public String getString() throws IOException {
373 String ret = accumulator.toString();
380 // Token PushBack code ////////////////////////////////////////////////////////////
382 private int pushBackDepth = 0;
383 private int[] pushBackInts = new int[10];
384 private Object[] pushBackObjects = new Object[10];
386 /** push back a token */
387 public final void pushBackToken(int op, Object obj) {
388 if (pushBackDepth >= pushBackInts.length - 1) {
389 int[] newInts = new int[pushBackInts.length * 2];
390 System.arraycopy(pushBackInts, 0, newInts, 0, pushBackInts.length);
391 pushBackInts = newInts;
392 Object[] newObjects = new Object[pushBackObjects.length * 2];
393 System.arraycopy(pushBackObjects, 0, newObjects, 0, pushBackObjects.length);
394 pushBackObjects = newObjects;
396 pushBackInts[pushBackDepth] = op;
397 pushBackObjects[pushBackDepth] = obj;
401 /** push back the most recently read token */
402 public final void pushBackToken() { pushBackToken(op, number != null ? (Object)number : (Object)string); }
404 /** read a token but leave it in the stream */
405 public final int peekToken() throws IOException {
406 int ret = getToken();
412 public final int getToken() throws IOException {
415 if (pushBackDepth == 0) {
416 mostRecentlyReadToken = op;
417 return op = _getToken();
420 op = pushBackInts[pushBackDepth];
421 if (pushBackObjects[pushBackDepth] != null) {
422 number = pushBackObjects[pushBackDepth] instanceof Number ? (Number)pushBackObjects[pushBackDepth] : null;
423 string = pushBackObjects[pushBackDepth] instanceof String ? (String)pushBackObjects[pushBackDepth] : null;
428 class LexerException extends IOException {
429 public LexerException(String s) { super(sourceName + ":" + line + "," + col + ": " + s); }