class Lexer {
+ public static void main(String[] s) throws Exception {
+ Lexer l = new Lexer(new InputStreamReader(System.in));
+ int tok = 0;
+ while((tok = l.getToken()) != -1)
+ System.out.println(codeToString[tok]);
+ }
+
private SmartReader in;
private boolean pushedBack = false;
- private int op;
+ public int op;
+ public int twoBack;
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; }
+ public int peekToken() throws IOException { int ret = getToken(); pushBackToken(); return ret; }
+ public void pushBackToken() { pushBackToken(op); }
+ public void pushBackToken(int i) { if (pushedBack) throw new Error("can't push back twice"); pushedBack = true; op = i; }
// 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, // +=
+ RETURN = 2, // return
+ GOTO = 3, // goto
+ BITOR = 4, // |
+ ASSIGN_BITOR = 5, // |=
+ BITXOR = 6, // ^
+ ASSIGN_BITXOR= 7, // ^=
+ BITAND = 8, // &
+ ASSIGN_BITAND= 9, // &=
+ EQ = 10, // ==
+ NE = 11, // !=
+ LT = 12, // <
+ LE = 13, // <=
+ GT = 14, // >
+ GE = 15, // >=
+ LSH = 16, // <<
+ ASSIGN_LSH = 17, // <<=
+ RSH = 18, // >>
+ ASSIGN_RSH = 19, // >>=
+ URSH = 20, // >>>
+ ASSIGN_URSH = 21, // >>>=
+ ADD = 22, // +
+ ASSIGN_ADD = 23, // +=
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
+ ASSIGN_SUB = 25, // -=
+ MUL = 26, // *
+ ASSIGN_MUL = 27, // *=
+ DIV = 28, // /
+ ASSIGN_DIV = 29, // /=
+ MOD = 30, // %
+ ASSIGN_MOD = 31, // %=
+ BITNOT = 32, // ~
+ ASSIGN_BITNOT= 33, // ~=
+ DELPROP = 34, // delete
+ TYPEOF = 35, // typeof
+ NAME = 36, // *** identifiers ***
+ NUMBER = 37, // *** numeric literals ***
+ STRING = 38, // *** string literals ***
+ NULL = 39, // null
+ THIS = 40, // this
+ FALSE = 41, // false
+ TRUE = 42, // true
+ SHEQ = 43, // ===
+ SHNE = 44, // !==
+ THROW = 45, // throw
+ IN = 46, // in
+ INSTANCEOF = 47, // instanceof
+ TRY = 48, // try
+ SEMI = 49, // ;
+ LB = 50, // [
+ RB = 51, // ]
+ LC = 52, // {
+ RC = 53, // }
+ LP = 54, // (
+ RP = 55, // )
+ COMMA = 56, // ,
+ ASSIGN = 57, // =
+ HOOK = 58, // ?
+ COLON = 59, // :
+ OR = 60, // ||
+ AND = 61, // &&
+ INC = 62, // ++
+ DEC = 63, // --
+ DOT = 64, // .
+ FUNCTION = 65, // function
+ IF = 66, // if keyword
+ ELSE = 67, // else keyword
+ SWITCH = 68, // switch keyword
+ CASE = 69, // case keyword
+ DEFAULT = 70, // default keyword
+ WHILE = 71, // while keyword
+ DO = 72, // do keyword
+ FOR = 73, // for keyword
+ BREAK = 74, // break keyword
+ CONTINUE = 75, // continue keyword
+ VAR = 76, // var keyword
+ WITH = 77, // with keyword
+ CATCH = 78, // catch keyword
+ FINALLY = 79, // finally keyword
+ RESERVED = 80, // reserved keywords
+ NOP = 81, // NOP
+ VOID = 82, // void keyword
+ MOD_ASSIGN = 83, // %=
+ BANG = 84, // %=
+ ASSERT = 85; // assert keyword
+
+ public static final int MAX_TOKEN = ASSERT;
+
+ public final static String[] codeToString = new String[] {
+ "0", "EOL", "RETURN", "GOTO", "BITOR", "ASSIGN_BITOR",
+ "BITXOR", "ASSIGN_BITXOR", "BITAND", "ASSIGN_BITAND", "EQ",
+ "NE", "LT", "LE", "GT", "GE", "LSH", "ASSIGN_LSH", "RSH",
+ "ASSIGN_RSH", "URSH", "ASSIGN_URSH", "ADD", "ASSIGN_ADD",
+ "SUB", "ASSIGN_SUB", "MUL", "ASSIGN_MUL", "DIV", "ASSIGN_DIV",
+ "MOD", "ASSIGN_MOD", "BITNOT", "ASSIGN_BITNOT=", "DELPROP",
+ "TYPEOF", "NAME", "NUMBER", "STRING", "NULL", "THIS", "FALSE",
+ "TRUE", "SHEQ", "SHNE", "THROW", "IN", "INSTANCEOF", "TRY",
+ "SEMI", "LB", "RB", "LC", "RC", "LP", "RP", "COMMA", "ASSIGN",
+ "HOOK", "COLON", "OR", "AND", "INC", "DEC", "DOT", "FUNCTION",
+ "IF", "ELSE", "SWITCH", "CASE", "DEFAULT", "WHILE", "DO",
+ "FOR", "BREAK", "CONTINUE", "VAR", "WITH", "CATCH", "FINALLY",
+ "RESERVED", "NOP", "VOID", "MOD_ASSIGN", "BANG", "ASSERT" };
// Predicates ///////////////////////////////////////////////////////////////////////
// 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 '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; break;
+ case 'n': if (s.charAt(2)=='w' && s.charAt(1)=='e') throw new IOException("the new keyword is not permitted in XWT scripts");
+ 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") ? GOTO : -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 'v': return s.equals("void") ? RESERVED : -1;
+ case 'w': return s.equals("with") ? WITH : -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;
- 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 (s.equals("const")) throw new IOException("the const keyword is not permitted in XWT");
+ else if (s.equals("final")) return RESERVED;
+ return -1;
+ case 'o': c=s.charAt(0);
+ 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")) throw new IOException("the delete keyword is not permitted in XWT scripts");
+ 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")) throw new IOException("the instanceof keyword is not permitted in XWT scripts");
+ 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 {
public int getToken() throws IOException {
if (pushedBack) { pushedBack = false; return op; }
- return (op = getToken());
+ do {
+ if (op != EOL) twoBack = op;
+ op = _getToken();
+ } while (op == EOL); // FIXME
+ //if (op == SEMI) throw new Error();
+ return op;
}
public int _getToken() throws IOException {
int lastread = -1;
public SmartReader(Reader r) { reader = new PushbackReader(r); }
- public void unread() throws IOException { reader.unread(lastread); }
+ public void unread() throws IOException {
+ reader.unread(lastread);
+ 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);
return lastread;
}
// FIXME: 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;