-// This file was derived from org.mozilla.javascript.TokenStream; it
-// is covered by the NPL 1.1.
+// Derived from org.mozilla.javascript.TokenStream [NPL]
/**
* The contents of this file are subject to the Netscape Public
* Contributor(s): Roger Lawrence, Mike McCabe
*/
+// FIXME: mark lots of these methods 'final' so they get inlined
+
package org.xwt.js;
import java.io.*;
-class Lexer {
+/** Lexes a stream of characters into a stream of Tokens */
+class Lexer implements Tokens {
- private SmartReader in;
- private boolean pushedBack = false;
-
- private int op;
- public Number number;
- public String string;
-
- public Lexer(Reader r) throws IOException { in = new SmartReader(r); }
- public int peekToken() throws IOException { int ret = peekToken(); pushBackToken(); return ret; }
- public void pushBackToken() { if (pushedBack) throw new Error("can't push back twice"); pushedBack = true; }
-
- // Token Constants //////////////////////////////////////////////////////////
-
- public final static int
- EOL = 1, // end of line
- RETURN = 5, // return
- GOTO = 6, // goto
- BITOR = 11, // |
- ASSIGN_BITOR = 211, // |=
- BITXOR = 12, // ^
- ASSIGN_BITXOR= 212, // ^=
- BITAND = 13, // &
- ASSIGN_BITAND= 213, // &=
- EQ = 14, // ==
- NE = 15, // !=
- LT = 16, // <
- LE = 17, // <=
- GT = 18, // >
- GE = 19, // >=
- LSH = 20, // <<
- ASSIGN_LSH = 220, // <<=
- RSH = 21, // >>
- ASSIGN_RSH = 221, // >>=
- URSH = 22, // >>>
- ASSIGN_URSH = 222, // >>>=
- ADD = 23, // +
- ASSIGN_ADD = 223, // +=
- SUB = 24, // -
- ASSIGN_SUB = 224, // -=
- MUL = 25, // *
- ASSIGN_MUL = 225, // *=
- DIV = 26, // /
- ASSIGN_DIV = 226, // /=
- MOD = 27, // %
- ASSIGN_MOD = 227, // %=
- BITNOT = 28, // ~
- ASSIGN_BITNOT= 228, // ~=
- DELPROP = 31, // delete
- TYPEOF = 32, // typeof
- NAME = 44, // *** identifiers ***
- NUMBER = 45, // *** numeric literals ***
- STRING = 46, // *** string literals ***
- NULL = 49, // null
- THIS = 50, // this
- FALSE = 51, // false
- TRUE = 52, // true
- SHEQ = 53, // ===
- SHNE = 54, // !==
- THROW = 62, // throw
- IN = 63, // in
- INSTANCEOF = 64, // instanceof
- TRY = 75, // try
- SEMI = 89, // ;
- LB = 90, // [
- RB = 91, // ]
- LC = 92, // {
- RC = 93, // }
- LP = 94, // (
- RP = 95, // )
- COMMA = 96, // ,
- ASSIGN = 97, // =
- HOOK = 98, // ?
- COLON = 99, // :
- OR = 100, // ||
- AND = 101, // &&
- INC = 106, // ++
- DEC = 107, // --
- DOT = 108, // .
- FUNCTION = 110, // function
- IF = 113, // if keyword
- ELSE = 114, // else keyword
- SWITCH = 115, // switch keyword
- CASE = 116, // case keyword
- DEFAULT = 117, // default keyword
- WHILE = 118, // while keyword
- DO = 119, // do keyword
- FOR = 120, // for keyword
- BREAK = 121, // break keyword
- CONTINUE = 122, // continue keyword
- VAR = 123, // var keyword
- WITH = 124, // with keyword
- CATCH = 125, // catch keyword
- FINALLY = 126, // finally keyword
- RESERVED = 127, // reserved keywords
- NOP = 128, // NOP
- VOID = 132, // void keyword
- MOD_ASSIGN = 133, // %=
- BANG = 134, // %=
- ASSERT = 150; // assert keyword
+ /** for debugging */
+ public static void main(String[] s) throws Exception {
+ Lexer l = new Lexer(new InputStreamReader(System.in), "stdin", 0);
+ int tok = 0;
+ while((tok = l.getToken()) != -1) System.out.println(codeToString[tok]);
+ }
+ /** the token that was just parsed */
+ protected int op;
+
+ /** the most recently parsed token, <i>regardless of pushbacks</i> */
+ protected int mostRecentlyReadToken;
- // Predicates ///////////////////////////////////////////////////////////////////////
+ /** if the token just parsed was a NUMBER, this is the numeric value */
+ protected Number number = null;
- protected static boolean isJSIdentifier(String s) {
- int length = s.length();
- if (length == 0 || !Character.isJavaIdentifierStart(s.charAt(0))) return false;
- for (int i=1; i<length; i++) {
- char c = s.charAt(i);
- if (!Character.isJavaIdentifierPart(c) && c == '\\' && !((i + 5) < length) &&
- (s.charAt(i + 1) == 'u') && 0 <= xDigitToInt(s.charAt(i + 2)) && 0 <= xDigitToInt(s.charAt(i + 3)) &&
- 0 <= xDigitToInt(s.charAt(i + 4)) && 0 <= xDigitToInt(s.charAt(i + 5)))
- return false;
- }
- return true;
+ /** if the token just parsed was a NAME or STRING, this is the string value */
+ protected String string = null;
+
+ /** the line number of the most recently <i>lexed</i> token */
+ private int line = 0;
+
+ /** the line number of the most recently <i>parsed</i> token */
+ protected int parserLine = 0;
+
+ /** the column number of the current token */
+ protected int col = 0;
+
+ /** the name of the source code file being lexed */
+ protected String sourceName;
+
+ private SmartReader in;
+ public Lexer(Reader r, String sourceName, int line) throws IOException {
+ this.sourceName = sourceName;
+ this.line = line;
+ this.parserLine = line;
+ in = new SmartReader(r);
}
+
+ // Predicates ///////////////////////////////////////////////////////////////////////
+
private static boolean isAlpha(int c) { return ((c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z')); }
private static boolean isDigit(int c) { return (c >= '0' && c <= '9'); }
- private static boolean isLineBreak(int c) { return (c == '\n' || c == '\r' || c == 0x2028 || c == 0x2029); }
private static int xDigitToInt(int c) {
if ('0' <= c && c <= '9') return c - '0';
- if ('a' <= c && c <= 'f') return c - ('a' - 10);
- if ('A' <= c && c <= 'F') return c - ('A' - 10);
- return -1;
- }
- public static boolean isWhiteSpace(int c) {
- if (c == '\u0020' || c == '\u0009' || c == '\u000C' || c == '\u000B' || c == '\u00A0') return true;
- return Character.getType((char)c) == Character.SPACE_SEPARATOR;
+ else if ('a' <= c && c <= 'f') return c - ('a' - 10);
+ else if ('A' <= c && c <= 'F') return c - ('A' - 10);
+ else return -1;
}
// Token Subtype Handlers /////////////////////////////////////////////////////////
- private int getKeyword(String name) throws IOException {
- final int
- Id_break = BREAK,
- Id_case = CASE,
- Id_continue = CONTINUE,
- Id_default = DEFAULT,
- Id_delete = DELPROP,
- Id_do = DO,
- Id_else = ELSE,
- Id_false = FALSE,
- Id_for = FOR,
- Id_function = FUNCTION,
- Id_if = IF,
- Id_in = IN,
- Id_null = NULL,
- Id_return = RETURN,
- Id_switch = SWITCH,
- Id_this = THIS,
- Id_true = TRUE,
- Id_typeof = TYPEOF,
- Id_var = VAR,
- Id_void = VOID,
- Id_while = WHILE,
- Id_with = WITH,
-
- // the following are #ifdef RESERVE_JAVA_KEYWORDS in jsscan.c
- Id_abstract = RESERVED,
- Id_boolean = RESERVED,
- Id_byte = RESERVED,
- Id_catch = CATCH,
- Id_char = RESERVED,
- Id_class = RESERVED,
- Id_const = RESERVED,
- Id_debugger = RESERVED,
- Id_double = RESERVED,
- Id_enum = RESERVED,
- Id_extends = RESERVED,
- Id_final = RESERVED,
- Id_finally = FINALLY,
- Id_float = RESERVED,
- Id_goto = RESERVED,
- Id_implements = RESERVED,
- Id_instanceof = INSTANCEOF,
- Id_int = RESERVED,
- Id_interface = RESERVED,
- Id_long = RESERVED,
- Id_native = RESERVED,
- Id_package = RESERVED,
- Id_private = RESERVED,
- Id_protected = RESERVED,
- Id_public = RESERVED,
- Id_assert = ASSERT,
- Id_short = RESERVED,
- Id_static = RESERVED,
- Id_super = RESERVED,
- Id_synchronized = RESERVED,
- Id_throw = THROW,
- Id_throws = RESERVED,
- Id_transient = RESERVED,
- Id_try = TRY,
- Id_volatile = RESERVED;
-
- int id;
- String s = name;
- L0: { id = -1; String X = null; int c;
- L: switch (s.length()) {
+ private int getKeyword(String s) throws IOException {
+ char c;
+ switch (s.length()) {
case 2: c=s.charAt(1);
- if (c=='f') { if (s.charAt(0)=='i') {id=Id_if; break L0;} }
- else if (c=='n') { if (s.charAt(0)=='i') {id=Id_in; break L0;} }
- else if (c=='o') { if (s.charAt(0)=='d') {id=Id_do; break L0;} }
- break L;
+ if (c=='f') { if (s.charAt(0)=='i') return IF; }
+ else if (c=='n') { if (s.charAt(0)=='i') return IN; }
+ else if (c=='o') { if (s.charAt(0)=='d') return DO; }
+ break;
case 3: switch (s.charAt(0)) {
- case 'f': if (s.charAt(2)=='r' && s.charAt(1)=='o') {id=Id_for; break L0;} break L;
- case 'i': if (s.charAt(2)=='t' && s.charAt(1)=='n') {id=Id_int; break L0;} break L;
- case 'n': if (s.charAt(2)=='w' && s.charAt(1)=='e')
- throw new IOException("the new keyword is not permitted in XWT scripts");
- break L;
- case 't': if (s.charAt(2)=='y' && s.charAt(1)=='r') {id=Id_try; break L0;} break L;
- case 'v': if (s.charAt(2)=='r' && s.charAt(1)=='a') {id=Id_var; break L0;} break L;
- } break L;
+ case 'a': if (s.charAt(2)=='d' && s.charAt(1)=='n') return AND; break;
+ case 'f': if (s.charAt(2)=='r' && s.charAt(1)=='o') return FOR; break;
+ case 'i': if (s.charAt(2)=='t' && s.charAt(1)=='n') return RESERVED;
+ case 'n': if (s.charAt(2)=='w' && s.charAt(1)=='e') return RESERVED;
+ case 't': if (s.charAt(2)=='y' && s.charAt(1)=='r') return TRY; break;
+ case 'v': if (s.charAt(2)=='r' && s.charAt(1)=='a') return VAR; break;
+ } break;
case 4: switch (s.charAt(0)) {
- case 'b': X="byte";id=Id_byte; break L;
+ case 'b': return s.equals("byte") ? RESERVED : -1;
case 'c': c=s.charAt(3);
- if (c=='e') { if (s.charAt(2)=='s' && s.charAt(1)=='a') {id=Id_case; break L0;} }
- else if (c=='r') { if (s.charAt(2)=='a' && s.charAt(1)=='h') {id=Id_char; break L0;} }
- break L;
+ if (c=='e') { if (s.charAt(2)=='s' && s.charAt(1)=='a') return CASE; }
+ else if (c=='r') { if (s.charAt(2)=='a' && s.charAt(1)=='h') return RESERVED; }
+ return -1;
case 'e': c=s.charAt(3);
- if (c=='e') { if (s.charAt(2)=='s' && s.charAt(1)=='l') {id=Id_else; break L0;} }
- else if (c=='m') { if (s.charAt(2)=='u' && s.charAt(1)=='n') {id=Id_enum; break L0;} }
- break L;
- case 'g': X="goto";id=Id_goto; break L;
- case 'l': X="long";id=Id_long; break L;
- case 'n': X="null";id=Id_null; break L;
+ if (c=='e') { if (s.charAt(2)=='s' && s.charAt(1)=='l') return ELSE; }
+ else if (c=='m') { if (s.charAt(2)=='u' && s.charAt(1)=='n') return RESERVED; }
+ return -1;
+ case 'g': return s.equals("goto") ? RESERVED : -1;
+ case 'l': return s.equals("long") ? RESERVED : -1;
+ case 'n': return s.equals("null") ? NULL : -1;
case 't': c=s.charAt(3);
- if (c=='e') { if (s.charAt(2)=='u' && s.charAt(1)=='r') {id=Id_true; break L0;} }
- else if (c=='s') { if (s.charAt(2)=='i' && s.charAt(1)=='h') {id=Id_this; break L0;} }
- break L;
- case 'v': X="void";id=Id_void; break L;
- case 'w': X="with";id=Id_with; break L;
- } break L;
+ if (c=='e') { if (s.charAt(2)=='u' && s.charAt(1)=='r') return TRUE; }
+ else if (c=='s') { if (s.charAt(2)=='i' && s.charAt(1)=='h') return THIS; }
+ return -1;
+ case 'w': if (s.equals("with")) return RESERVED; else return -1;
+ case 'v': if (s.equals("void")) return RESERVED; else return -1;
+ } break;
case 5: switch (s.charAt(2)) {
- case 'a': X="class";id=Id_class; break L;
- case 'e': X="break";id=Id_break; break L;
- case 'i': X="while";id=Id_while; break L;
- case 'l': X="false";id=Id_false; break L;
+ case 'a': return s.equals("class") ? RESERVED : -1;
+ case 'e': return s.equals("break") ? BREAK : -1;
+ case 'i': return s.equals("while") ? WHILE : -1;
+ case 'l': return s.equals("false") ? FALSE : -1;
case 'n': c=s.charAt(0);
- if (c=='c') { X="const"; throw new IOException("the const keyword is not permitted in XWT"); }
- else if (c=='f') { X="final";id=Id_final; }
- break L;
+ if (s.equals("const")) return RESERVED;
+ else if (s.equals("final")) return RESERVED;
+ return -1;
case 'o': c=s.charAt(0);
- if (c=='f') { X="float";id=Id_float; }
- else if (c=='s') { X="short";id=Id_short; }
- break L;
- case 'p': X="super";id=Id_super; break L;
- case 'r': X="throw";id=Id_throw; break L;
- case 't': X="catch";id=Id_catch; break L;
- } break L;
+ if (c == 'c') return s.equals("float") ? RESERVED : -1;
+ else if (c=='s') return s.equals("final") ? RESERVED : -1;
+ break;
+ case 'p': return s.equals("super") ? RESERVED : -1;
+ case 'r': return s.equals("throw") ? THROW : -1;
+ case 't': return s.equals("catch") ? CATCH : -1;
+ } break;
case 6: switch (s.charAt(1)) {
- case 'a': X="native";id=Id_native; break L;
+ case 'a': return s.equals("class") ? RESERVED : -1;
case 'e': c=s.charAt(0);
- if (c=='d') { X="delete"; throw new IOException("the delete keyword is not permitted in XWT scripts"); }
- else if (c=='r') { X="return";id=Id_return; }
- break L;
- case 'h': X="throws";id=Id_throws; break L;
- case 'o': X="double";id=Id_double; break L;
- case 's': X="assert";id=Id_assert; break L;
- case 'u': X="public";id=Id_public; break L;
- case 'w': X="switch";id=Id_switch; break L;
- case 'y': X="typeof";id=Id_typeof; break L;
- } break L;
+ if (s.equals("delete")) return RESERVED;
+ else if (c=='r') return s.equals("return") ? RETURN : -1;
+ break;
+ case 'h': return s.equals("throws") ? RESERVED : -1;
+ case 'o': return s.equals("double") ? RESERVED : -1;
+ case 's': return s.equals("assert") ? ASSERT : -1;
+ case 'u': return s.equals("public") ? RESERVED : -1;
+ case 'w': return s.equals("switch") ? SWITCH : -1;
+ case 'y': return s.equals("typeof") ? TYPEOF : -1;
+ } break;
case 7: switch (s.charAt(1)) {
- case 'a': X="package";id=Id_package; break L;
- case 'e': X="default";id=Id_default; break L;
- case 'i': X="finally";id=Id_finally; break L;
- case 'o': X="boolean";id=Id_boolean; break L;
- case 'r': X="private";id=Id_private; break L;
- case 'x': X="extends";id=Id_extends; break L;
- } break L;
+ case 'a': return s.equals("package") ? RESERVED : -1;
+ case 'e': return s.equals("default") ? DEFAULT : -1;
+ case 'i': return s.equals("finally") ? FINALLY : -1;
+ case 'o': return s.equals("boolean") ? RESERVED : -1;
+ case 'r': return s.equals("private") ? RESERVED : -1;
+ case 'x': return s.equals("extends") ? RESERVED : -1;
+ } break;
case 8: switch (s.charAt(0)) {
- case 'a': X="abstract";id=Id_abstract; break L;
- case 'c': X="continue";id=Id_continue; break L;
- case 'd': X="debugger";id=Id_debugger; break L;
- case 'f': X="function";id=Id_function; break L;
- case 'v': X="volatile";id=Id_volatile; break L;
- } break L;
+ case 'a': return s.equals("abstract") ? RESERVED : -1;
+ case 'c': return s.equals("continue") ? CONTINUE : -1;
+ case 'd': return s.equals("debugger") ? RESERVED : -1;
+ case 'f': return s.equals("function") ? FUNCTION : -1;
+ case 'v': return s.equals("volatile") ? RESERVED : -1;
+ } break;
case 9: c=s.charAt(0);
- if (c=='i') { X="interface";id=Id_interface; }
- else if (c=='p') { X="protected";id=Id_protected; }
- else if (c=='t') { X="transient";id=Id_transient; }
- break L;
+ if (c=='i') return s.equals("interface") ? RESERVED : -1;
+ else if (c=='p') return s.equals("protected") ? RESERVED : -1;
+ else if (c=='t') return s.equals("transient") ? RESERVED : -1;
+ break;
case 10: c=s.charAt(1);
- if (c=='m') { X="implements";id=Id_implements; }
- else if (c=='n') { X="instanceof"; throw new IOException("the instanceof keyword is not permitted in XWT scripts"); }
- break L;
- case 12: X="synchronized";id=Id_synchronized; break L;
+ if (c=='m') return s.equals("implements") ? RESERVED : -1;
+ else if (c=='n' && s.equals("instanceof")) return RESERVED;
+ break;
+ case 12: return s.equals("synchronized") ? RESERVED : -1;
}
- if (X!=null && X!=s && !X.equals(s)) id = -1;
- }
- if (id == -1) { return -1; }
- this.op = id >> 8;
- return id & 0xff;
+ return -1;
}
private int getIdentifier(int c) throws IOException {
in.unread();
String str = in.getString();
int result = getKeyword(str);
+ if (result == RESERVED) throw new LexerException("The reserved word \"" + str + "\" is not permitted in XWT scripts");
if (result != -1) return result;
- this.string = str;
+ this.string = str.intern();
return NAME;
}
if (c == 'e' || c == 'E') {
c = in.read();
if (c == '+' || c == '-') c = in.read();
- if (!isDigit(c)) throw new IOException("msg.missing.exponent");
+ if (!isDigit(c)) throw new LexerException("float listeral did not have an exponent value");
do { c = in.read(); } while (isDigit(c));
}
}
String numString = in.getString();
if (base == 10 && !isInteger) {
try { dval = (Double.valueOf(numString)).doubleValue(); }
- catch (NumberFormatException ex) { throw new IOException("msg.caught.nfe"); }
+ catch (NumberFormatException ex) { throw new LexerException("invalid numeric literal: \"" + numString + "\""); }
} else {
if (isInteger) {
longval = Long.parseLong(numString, base);
dval = (double)longval;
- } else {
- // FIXME: we're not handling hex/octal fractions... does that matter?
+ } else {
dval = Double.parseDouble(numString);
longval = (long) dval;
if (longval == dval) isInteger = true;
c = in.read();
in.startString(); // start after the first "
while(c != quoteChar) {
- if (c == '\n' || c == -1) throw new IOException("msg.unterminated.string.lit");
+ if (c == '\n' || c == -1) throw new LexerException("unterminated string literal");
if (c == '\\') {
if (stringBuf == null) {
in.unread(); // Don't include the backslash
case 'r': c = '\r'; break;
case 't': c = '\t'; break;
case 'v': c = '\u000B'; break;
- default: throw new IOException("\\u and \\0 escapes not currently supported -- use XML entities");
+ case '\\': c = '\\'; break;
+ case 'u': {
+ int v = 0;
+ for(int i=0; i<4; i++) {
+ int ci = in.read();
+ if (!((ci >= '0' && ci <= '9') || (ci >= 'a' && ci <= 'f') || (ci >= 'A' && ci <= 'F')))
+ throw new LexerException("illegal character '" + ((char)c) + "' in \\u unicode escape sequence");
+ v = (v << 8) | Integer.parseInt(ci + "", 16);
+ }
+ c = (char)v;
+ break;
+ }
+ default:
+ // just use the character that was escaped
+ break;
}
}
if (stringBuf != null) stringBuf.append((char) c);
c = in.read();
}
- if (stringBuf != null) this.string = stringBuf.toString();
+ if (stringBuf != null) this.string = stringBuf.toString().intern();
else {
in.unread(); // miss the trailing "
- this.string = in.getString();
+ this.string = in.getString().intern();
in.read();
}
return STRING;
}
- public int getToken() throws IOException {
- if (pushedBack) { pushedBack = false; return op; }
- return (op = getToken());
- }
-
public int _getToken() throws IOException {
int c;
- do { if ((c = in.read()) == '\n') break; } while (isWhiteSpace(c) || c == '\n');
+ do { c = in.read(); } while (c == '\u0020' || c == '\u0009' || c == '\u000C' || c == '\u000B' || c == '\n' );
if (c == -1) return -1;
- if (c == '\\' && in.peek() == 'u') throw new IOException("\\u and \\0 escapes not currently supported -- use XML entities");
- if (Character.isJavaIdentifierStart((char)c)) return getIdentifier(c);
+ if (c == '\\' || Character.isJavaIdentifierStart((char)c)) return getIdentifier(c);
if (isDigit(c) || (c == '.' && isDigit(in.peek()))) return getNumber(c);
if (c == '"' || c == '\'') return getString(c);
switch (c) {
- case '\n': return EOL;
case ';': return SEMI;
case '[': return LB;
case ']': return RB;
case '&': return in.match('&') ? AND : in.match('=') ? ASSIGN_BITAND : BITAND;
case '=': return !in.match('=') ? ASSIGN : in.match('=') ? SHEQ : EQ;
case '!': return !in.match('=') ? BANG : in.match('=') ? SHNE : NE;
- case '%': return in.match('=') ? MOD_ASSIGN : MOD;
+ case '%': return in.match('=') ? ASSIGN_MOD : MOD;
case '~': return BITNOT;
case '+': return in.match('=') ? ASSIGN_ADD : in.match('+') ? INC : ADD;
case '-': return in.match('=') ? ASSIGN_SUB: in.match('-') ? DEC : SUB;
case '*': return in.match('=') ? ASSIGN_MUL : MUL;
- case '<': return !in.match('<') ?
- (in.match('=') ? LE : LT) :
- in.match('=') ? ASSIGN_LSH : LSH;
- case '>': return !in.match('>') ?
- (in.match('=') ? GE : GT) :
- in.match('>') ?
- (in.match('=') ? ASSIGN_URSH : URSH) :
- (in.match('=') ? ASSIGN_RSH : RSH);
+ case '<': return !in.match('<') ? (in.match('=') ? LE : LT) : in.match('=') ? ASSIGN_LSH : LSH;
+ case '>': return !in.match('>') ? (in.match('=') ? GE : GT) :
+ in.match('>') ? (in.match('=') ? ASSIGN_URSH : URSH) : (in.match('=') ? ASSIGN_RSH : RSH);
case '/':
if (in.match('=')) return ASSIGN_DIV;
if (in.match('/')) { while ((c = in.read()) != -1 && c != '\n'); in.unread(); return getToken(); }
while ((c = in.read()) != -1 && !(c == '*' && in.match('/'))) {
if (c == '\n' || c != '/' || !in.match('*')) continue;
if (in.match('/')) return getToken();
- throw new IOException("msg.nested.comment");
+ throw new LexerException("nested comments are not permitted");
}
- if (c == -1) throw new IOException("msg.unterminated.comment");
+ if (c == -1) throw new LexerException("unterminated comment");
return getToken(); // `goto retry'
- default: throw new IOException("illegal character: " + c);
+ default: throw new LexerException("illegal character: \'" + ((char)c) + "\'");
}
}
- private static class SmartReader {
+
+ // SmartReader ////////////////////////////////////////////////////////////////
+
+ /** a Reader that tracks line numbers and can push back tokens */
+ private class SmartReader {
PushbackReader reader = null;
int lastread = -1;
public SmartReader(Reader r) { reader = new PushbackReader(r); }
- public void unread() throws IOException { reader.unread(lastread); }
+ public void unread() throws IOException { unread((char)lastread); }
+ public void unread(char c) throws IOException {
+ reader.unread(c);
+ if (accumulator != null) accumulator.setLength(accumulator.length() - 1);
+ }
public boolean match(char c) throws IOException { if (peek() == c) { reader.read(); return true; } else return false; }
public int peek() throws IOException {
int peeked = reader.read();
}
public int read() throws IOException {
lastread = reader.read();
- if (accumulator != null) accumulator.append(lastread);
+ if (accumulator != null) accumulator.append((char)lastread);
+ if (lastread != '\n' && lastread != '\r') col++;
+ if (lastread == '\n') { parserLine = ++line; col = 0; }
return lastread;
}
- // FIXME: could be much more efficient
+ // FEATURE: could be much more efficient
StringBuffer accumulator = null;
- public void startString() { accumulator = new StringBuffer(); }
+ public void startString() {
+ accumulator = new StringBuffer();
+ accumulator.append((char)lastread);
+ }
public String getString() throws IOException {
String ret = accumulator.toString();
accumulator = null;
}
}
+
+ // Token PushBack code ////////////////////////////////////////////////////////////
+
+ private int pushBackDepth = 0;
+ private int[] pushBackInts = new int[10];
+ private Object[] pushBackObjects = new Object[10];
+
+ /** push back a token */
+ public void pushBackToken(int op, Object obj) {
+ if (pushBackDepth >= pushBackInts.length - 1) {
+ int[] newInts = new int[pushBackInts.length * 2];
+ System.arraycopy(pushBackInts, 0, newInts, 0, pushBackInts.length);
+ pushBackInts = newInts;
+ Object[] newObjects = new Object[pushBackObjects.length * 2];
+ System.arraycopy(pushBackObjects, 0, newObjects, 0, pushBackObjects.length);
+ pushBackObjects = newObjects;
+ }
+ pushBackInts[pushBackDepth] = op;
+ pushBackObjects[pushBackDepth] = obj;
+ pushBackDepth++;
+ }
+
+ /** push back the most recently read token */
+ public void pushBackToken() { pushBackToken(op, number != null ? (Object)number : (Object)string); }
+
+ /** read a token but leave it in the stream */
+ public int peekToken() throws IOException {
+ int ret = getToken();
+ pushBackToken();
+ return ret;
+ }
+
+ /** read a token */
+ public int getToken() throws IOException {
+ number = null;
+ string = null;
+ if (pushBackDepth == 0) {
+ mostRecentlyReadToken = op;
+ return op = _getToken();
+ }
+ pushBackDepth--;
+ op = pushBackInts[pushBackDepth];
+ if (pushBackObjects[pushBackDepth] != null) {
+ number = pushBackObjects[pushBackDepth] instanceof Number ? (Number)pushBackObjects[pushBackDepth] : null;
+ string = pushBackObjects[pushBackDepth] instanceof String ? (String)pushBackObjects[pushBackDepth] : null;
+ }
+ return op;
+ }
+
+ class LexerException extends IOException {
+ public LexerException(String s) { super(sourceName + ":" + line + "," + col + ": " + s); }
+ }
}