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 name) throws IOException {
85 case "and": return AND;
86 case "for": return FOR;
87 case "int": return RESERVED;
88 case "new": return RESERVED;
89 case "try": return TRY;
90 case "var": return VAR;
91 case "byte": return RESERVED;
92 case "case": return CASE;
93 case "char": return RESERVED;
94 case "else": return ELSE;
95 case "enum": return RESERVED;
96 case "goto": return RESERVED;
97 case "long": return RESERVED;
98 case "null": return NULL;
99 case "true": return TRUE;
100 case "with": return RESERVED;
101 case "void": return RESERVED;
102 case "class": return RESERVED;
103 case "break": return BREAK;
104 case "while": return WHILE;
105 case "false": return FALSE;
106 case "const": return RESERVED;
107 case "final": return RESERVED;
108 case "super": return RESERVED;
109 case "throw": return THROW;
110 case "catch": return CATCH;
111 case "class": return RESERVED;
112 case "delete": return RESERVED;
113 case "return": return RETURN;
114 case "throws": return RESERVED;
115 case "double": return RESERVED;
116 case "assert": return ASSERT;
117 case "public": return RESERVED;
118 case "switch": return SWITCH;
119 case "typeof": return TYPEOF;
120 case "package": return RESERVED;
121 case "default": return DEFAULT;
122 case "finally": return FINALLY;
123 case "boolean": return RESERVED;
124 case "private": return RESERVED;
125 case "extends": return RESERVED;
126 case "abstract": return RESERVED;
127 case "continue": return CONTINUE;
128 case "debugger": return RESERVED;
129 case "function": return FUNCTION;
130 case "volatile": return RESERVED;
131 case "interface": return RESERVED;
132 case "protected": return RESERVED;
133 case "transient": return RESERVED;
134 case "implements": return RESERVED;
135 case "instanceof": return RESERVED;
136 case "synchronized": return RESERVED;
141 private int getIdentifier(int c) throws IOException {
143 while (Character.isJavaIdentifierPart((char)(c = in.read())));
145 String str = in.getString();
146 int result = getKeyword(str);
147 if (result == RESERVED) throw new LexerException("The reserved word \"" + str + "\" is not permitted in XWT scripts");
148 if (result != -1) return result;
149 this.string = str.intern();
153 private int getNumber(int c) throws IOException {
156 double dval = Double.NaN;
158 boolean isInteger = true;
160 // figure out what base we're using
162 if (Character.toLowerCase((char)(c = in.read())) == 'x') { base = 16; in.startString(); }
163 else if (isDigit(c)) base = 8;
166 while (0 <= xDigitToInt(c) && !(base < 16 && isAlpha(c))) c = in.read();
167 if (base == 10 && (c == '.' || c == 'e' || c == 'E')) {
169 if (c == '.') do { c = in.read(); } while (isDigit(c));
170 if (c == 'e' || c == 'E') {
172 if (c == '+' || c == '-') c = in.read();
173 if (!isDigit(c)) throw new LexerException("float listeral did not have an exponent value");
174 do { c = in.read(); } while (isDigit(c));
179 String numString = in.getString();
180 if (base == 10 && !isInteger) {
181 try { dval = (Double.valueOf(numString)).doubleValue(); }
182 catch (NumberFormatException ex) { throw new LexerException("invalid numeric literal: \"" + numString + "\""); }
185 longval = Long.parseLong(numString, base);
186 dval = (double)longval;
188 dval = Double.parseDouble(numString);
189 longval = (long) dval;
190 if (longval == dval) isInteger = true;
194 if (!isInteger) this.number = JS.N(dval);
195 else this.number = JS.N(longval);
199 private int getString(int c) throws IOException {
200 StringBuffer stringBuf = null;
204 in.startString(); // start after the first "
205 while(c != quoteChar) {
206 if (c == '\n' || c == -1) throw new LexerException("unterminated string literal");
208 if (stringBuf == null) {
209 in.unread(); // Don't include the backslash
210 stringBuf = new StringBuffer(in.getString());
213 switch (c = in.read()) {
214 case 'b': c = '\b'; break;
215 case 'f': c = '\f'; break;
216 case 'n': c = '\n'; break;
217 case 'r': c = '\r'; break;
218 case 't': c = '\t'; break;
219 case 'v': c = '\u000B'; break;
220 case '\\': c = '\\'; break;
223 for(int i=0; i<4; i++) {
225 if (!((ci >= '0' && ci <= '9') || (ci >= 'a' && ci <= 'f') || (ci >= 'A' && ci <= 'F')))
226 throw new LexerException("illegal character '" + ((char)c) + "' in \\u unicode escape sequence");
227 v = (v << 8) | Integer.parseInt(ci + "", 16);
233 // just use the character that was escaped
237 if (stringBuf != null) stringBuf.append((char) c);
240 if (stringBuf != null) this.string = stringBuf.toString().intern();
242 in.unread(); // miss the trailing "
243 this.string = in.getString().intern();
249 private int _getToken() throws IOException {
251 do { c = in.read(); } while (c == '\u0020' || c == '\u0009' || c == '\u000C' || c == '\u000B' || c == '\n' );
252 if (c == -1) return -1;
253 if (c == '\\' || Character.isJavaIdentifierStart((char)c)) return getIdentifier(c);
254 if (isDigit(c) || (c == '.' && isDigit(in.peek()))) return getNumber(c);
255 if (c == '"' || c == '\'') return getString(c);
257 case ';': return SEMI;
264 case ',': return COMMA;
265 case '?': return HOOK;
266 case ':': return COLON;
267 case '.': return DOT;
268 case '|': return in.match('|') ? OR : (in.match('=') ? ASSIGN_BITOR : BITOR);
269 case '^': return in.match('=') ? ASSIGN_BITXOR : BITXOR;
270 case '&': return in.match('&') ? AND : in.match('=') ? ASSIGN_BITAND : BITAND;
271 case '=': return !in.match('=') ? ASSIGN : in.match('=') ? SHEQ : EQ;
272 case '!': return !in.match('=') ? BANG : in.match('=') ? SHNE : NE;
273 case '%': return in.match('=') ? ASSIGN_MOD : MOD;
274 case '~': return BITNOT;
275 case '+': return in.match('=') ? ASSIGN_ADD : in.match('+') ? INC : ADD;
276 case '-': return in.match('=') ? ASSIGN_SUB: in.match('-') ? DEC : SUB;
277 case '*': return in.match('=') ? ASSIGN_MUL : MUL;
278 case '<': return !in.match('<') ? (in.match('=') ? LE : LT) : in.match('=') ? ASSIGN_LSH : LSH;
279 case '>': return !in.match('>') ? (in.match('=') ? GE : GT) :
280 in.match('>') ? (in.match('=') ? ASSIGN_URSH : URSH) : (in.match('=') ? ASSIGN_RSH : RSH);
282 if (in.match('=')) return ASSIGN_DIV;
283 if (in.match('/')) { while ((c = in.read()) != -1 && c != '\n'); in.unread(); return getToken(); }
284 if (!in.match('*')) return DIV;
285 while ((c = in.read()) != -1 && !(c == '*' && in.match('/'))) {
286 if (c == '\n' || c != '/' || !in.match('*')) continue;
287 if (in.match('/')) return getToken();
288 throw new LexerException("nested comments are not permitted");
290 if (c == -1) throw new LexerException("unterminated comment");
291 return getToken(); // `goto retry'
292 default: throw new LexerException("illegal character: \'" + ((char)c) + "\'");
297 // SmartReader ////////////////////////////////////////////////////////////////
299 /** a Reader that tracks line numbers and can push back tokens */
300 private class SmartReader {
301 PushbackReader reader = null;
304 public SmartReader(Reader r) { reader = new PushbackReader(r); }
305 public void unread() throws IOException { unread((char)lastread); }
306 public void unread(char c) throws IOException {
308 if(c == '\n') col = -1;
310 if (accumulator != null) accumulator.setLength(accumulator.length() - 1);
312 public boolean match(char c) throws IOException { if (peek() == c) { reader.read(); return true; } else return false; }
313 public int peek() throws IOException {
314 int peeked = reader.read();
315 if (peeked != -1) reader.unread((char)peeked);
318 public int read() throws IOException {
319 lastread = reader.read();
320 if (accumulator != null) accumulator.append((char)lastread);
321 if (lastread != '\n' && lastread != '\r') col++;
322 if (lastread == '\n') {
323 // col is -1 if we just unread a newline, this is sort of ugly
324 if (col != -1) parserLine = ++line;
330 // FEATURE: could be much more efficient
331 StringBuffer accumulator = null;
332 public void startString() {
333 accumulator = new StringBuffer();
334 accumulator.append((char)lastread);
336 public String getString() throws IOException {
337 String ret = accumulator.toString().intern();
344 // Token PushBack code ////////////////////////////////////////////////////////////
346 private int pushBackDepth = 0;
347 private int[] pushBackInts = new int[10];
348 private Object[] pushBackObjects = new Object[10];
350 /** push back a token */
351 public final void pushBackToken(int op, Object obj) {
352 if (pushBackDepth >= pushBackInts.length - 1) {
353 int[] newInts = new int[pushBackInts.length * 2];
354 System.arraycopy(pushBackInts, 0, newInts, 0, pushBackInts.length);
355 pushBackInts = newInts;
356 Object[] newObjects = new Object[pushBackObjects.length * 2];
357 System.arraycopy(pushBackObjects, 0, newObjects, 0, pushBackObjects.length);
358 pushBackObjects = newObjects;
360 pushBackInts[pushBackDepth] = op;
361 pushBackObjects[pushBackDepth] = obj;
365 /** push back the most recently read token */
366 public final void pushBackToken() { pushBackToken(op, number != null ? (Object)number : (Object)string); }
368 /** read a token but leave it in the stream */
369 public final int peekToken() throws IOException {
370 int ret = getToken();
376 public final int getToken() throws IOException {
379 if (pushBackDepth == 0) {
380 mostRecentlyReadToken = op;
381 return op = _getToken();
384 op = pushBackInts[pushBackDepth];
385 if (pushBackObjects[pushBackDepth] != null) {
386 number = pushBackObjects[pushBackDepth] instanceof Number ? (Number)pushBackObjects[pushBackDepth] : null;
387 string = pushBackObjects[pushBackDepth] instanceof String ? (String)pushBackObjects[pushBackDepth] : null;
392 class LexerException extends IOException {
393 public LexerException(String s) { super(sourceName + ":" + line + "," + col + ": " + s); }