+// Copyright 2002 Adam Megacz, see the COPYING file for licensing [GPL]
package org.xwt.js;
+
import org.xwt.util.*;
import java.io.*;
-// FIXME: for..in
-// FIXME: delete keyword
+/** parses a stream of lexed tokens into a tree of Expr's */
public class Parser extends Lexer {
+ // Constructors //////////////////////////////////////////////////////
+
public Parser(Reader r) throws IOException { super(r); }
- private Parser skipToken() throws IOException { getToken(); return this; }
-
+
+ public static void main(String[] s) throws Exception {
+ Parser p = new Parser(new InputStreamReader(System.in));
+ while(true) {
+ Expr block = p.parseBlock(false);
+ if (block == null) return;
+ System.out.println(block);
+ if (p.peekToken() == -1) return;
+ }
+ }
+
+
+ // Statics ////////////////////////////////////////////////////////////
+
+ static byte[] precedence = new byte[MAX_TOKEN + 1];
+ static {
+ precedence[COMMA] = 1;
+ precedence[ASSIGN] = 2;
+ precedence[GT] = precedence[GE] = 3;
+ precedence[OR] = precedence[AND] = 4;
+ precedence[BITOR] = 5;
+ precedence[BITXOR] = 6;
+ precedence[BITAND] = 7;
+ precedence[EQ] = precedence[NE] = 8;
+ precedence[LT] = precedence[LE] = 9;
+ precedence[SHEQ] = precedence[SHNE] = 10;
+ precedence[LSH] = precedence[RSH] = precedence[URSH] = 11;
+ precedence[ADD] = precedence[SUB] = 12;
+ precedence[MUL] = precedence[DIV] = precedence[MOD] = 13;
+ precedence[BITNOT] = precedence[INSTANCEOF] = 14;
+ precedence[INC] = precedence[DEC] = 15;
+ precedence[LP] = 16;
+ precedence[DOT] = 17;
+ }
+
+
+ // Useful Types /////////////////////////////////////////////////////////
+
/** sorta like gcc trees */
public static class Expr {
+
int code = -1;
Expr left = null;
Expr right = null;
- Expr extra = null;
-
Expr next = null; // if this expr is part of a list
String string = null;
+ Number number = null;
- public Expr(String s) { this.string = s; } // an identifier or label
- public Expr(int code) { this(code, null, null, null); }
- public Expr(int code, Expr left) { this(code, left, null, null); }
- public Expr(int code, Expr left, Expr right) { this(code, left, right, null); }
- public Expr(int code, Expr left, Expr right, Expr extra) { this.left = left; this.right = right; this.extra = extra; this.code = code; }
- }
-
- /** parses a single statement */
- public Expr parseStatement() throws IOException {
- int tok;
- Expr ret;
- switch(tok = peekToken()) {
-
- case LC:
- ret = parseBlock(true);
-
- case THROW: case RETURN: case ASSERT:
- ret = new Expr(ASSERT, skipToken().parseExpr());
-
- case GOTO: case BREAK: case CONTINUE:
- skipToken();
- if (getToken() == NAME)
- ret = new Expr(tok, new Expr(string));
- else if (tok == GOTO)
- throw new Error("goto must be followed by a label");
- else
- ret = new Expr(tok);
-
- default:
- ret = parseExpr();
+ public String toString() { return toString(0); }
+ public String toString(int indent) {
+ String ret = "";
+ for(int i=0; i<indent; i++) ret += " ";
+ ret += codeToString[code];
+ if (code == NUMBER) ret += " " + number;
+ else if (string != null) ret += " \"" + string + "\"";
+ ret += "\n";
+ if (left != null) ret += left.toString(indent + 2);
+ if (right != null) ret += right.toString(indent + 2);
+ if (next != null) ret += next.toString(indent);
+ return ret;
}
- if (getToken() != SEMI) throw new Error("expected ;");
- return ret;
+ public Expr(String s) { code = STRING; this.string = s; } // an identifier or label
+ public Expr(Number n) { code = NUMBER; this.number = n; } // an identifier or label
+ public Expr(int code) { this(code, null, null); }
+ public Expr(int code, String s) { this.code = code; string = s; }
+ public Expr(int code, Expr left) { this(code, left, null); }
+ public Expr(int code, Expr left, Expr right) { this.code = code; this.left = left; this.right = right; }
}
-
+
/** a block is either a single statement or a list of statements surrounded by curly braces; all expressions are also statements */
public Expr parseBlock(boolean requireBraces) throws IOException {
+ Expr ret = null;
int tok = peekToken();
- if (requireBraces && tok != LC) throw new Error("expected {");
- if (tok != LC) return parseStatement();
- skipToken();
+ boolean braced = tok == LC;
+ if (requireBraces && !braced) throw new Error("expected {");
+ if (braced) getToken();
Expr head = null;
Expr tail = null;
- while(peekToken() != RC)
- if (head == null) head = tail = parseStatement(); else tail = tail.next = parseStatement();
- skipToken();
+ OUTER: while(true) {
+ Expr smt;
+ switch(tok = peekToken()) {
+ case -1: break OUTER;
+ case LC: smt = parseBlock(true); break;
+ case THROW: case RETURN: case ASSERT:
+ getToken();
+ smt = new Expr(tok, parseMaximalExpr());
+ if (getToken() != SEMI) throw new Error("expected ;");
+ break;
+ case GOTO: case BREAK: case CONTINUE:
+ getToken();
+ if (getToken() == NAME)
+ smt = new Expr(tok, new Expr(string));
+ else if (tok == GOTO)
+ throw new Error("goto must be followed by a label");
+ else
+ smt = new Expr(tok);
+ if (getToken() != SEMI) throw new Error("expected ;");
+ break;
+
+ case RC:
+ if (braced) getToken();
+ break OUTER;
+
+ case SEMI:
+ getToken();
+ if (!braced) break OUTER;
+ continue;
+
+ default:
+ smt = parseMaximalExpr();
+ if (smt == null) {
+ if (head == null) throw new Error("empty statement list");
+ break OUTER;
+ }
+ break;
+ }
+ if (head == null) head = tail = smt; else tail = (tail.next = smt);
+ }
return new Expr(LC, head);
}
- /** Subexpressions come in two flavors: starters and continuers.
- * Starters can appear at the start of an expression or after a
- * continuer, and continuers, which can appear after a starter.
- */
- public Expr parseExpr() throws IOException {
- Expr e = parseStarter();
- while(true) {
- Expr e2 = parseContinuer(e);
- if (e2 == null) return e;
- e = e2;
- }
+ /** throws an error if the next token is not <tt>code</tt> */
+ public void expect(int code) throws IOException {
+ int got = peekToken();
+ if (got != code)
+ throw new Error("expected " + codeToString[got] + ", got " + (got == -1 ? "EOL" : codeToString[got]));
}
- public Expr parseStarter() throws IOException {
- Expr e1 = null;
- Expr e2 = null;
- Expr e3 = null;
- Expr head = null;
- Expr tail = null;
- int tok = getToken();
- switch(tok) {
+ /** parses the largest possible expression */
+ public Expr parseMaximalExpr() throws IOException { return parseMaximalExpr(null, -1); }
+ public Expr parseMaximalExpr(Expr prefix, int minPrecedence) throws IOException {
+ Expr save = null;
+ do {
+ save = prefix;
+ if (peekToken() == -1) break;
+ prefix = parseSingleExpr(prefix, minPrecedence);
+ if (prefix == null) throw new Error("parseSingleExpr_() returned null");
+ } while (save != prefix);
+ return prefix;
+ }
+
+ /** parses the smallest possible complete expression */
+ public Expr parseSingleExpr() throws IOException { return parseSingleExpr(null, 0); }
+
+ /** parses the smallest possible complete expression beginning with prefix and only using operators with at least minPrecedence */
+ public Expr parseSingleExpr(Expr prefix, int minPrecedence) throws IOException {
+ Expr e1 = null, e2 = null, e3 = null, head = null, tail = null, ret = null;
+
+ int tok = peekToken();
+ if (minPrecedence > 0 && tok < precedence.length && precedence[tok] != 0 && precedence[tok] <= minPrecedence) return prefix;
+ getToken();
+
+ // these case arms match the precedence of operators; each arm is a precedence level.
+ switch (tok) {
+
+ case WITH: throw new Error("XWT does not allow the WITH keyword");
+ case VOID: case RESERVED: throw new Error("reserved word that you shouldn't be using");
+ case NAME: if (prefix != null) { pushBackToken(); return prefix; } else return parseMaximalExpr(new Expr(NAME, string), minPrecedence);
+ case STRING: if (prefix != null) { pushBackToken(); return prefix; } else return new Expr(string);
+ case NUMBER: if (prefix != null) { pushBackToken(); return prefix; } else return new Expr(number);
+ case NULL: case TRUE: case FALSE: case NOP: if (prefix != null) { pushBackToken(); return prefix; } else return new Expr(tok);
+
+ case COMMA: case ASSIGN: case GT: case GE: case OR: case AND:
+ case BITOR: case BITXOR: case BITAND: case EQ: case NE: case LT:
+ case LE: case SHEQ: case SHNE: case LSH: case RSH: case URSH:
+ case ADD: case SUB: case MUL: case DIV: case MOD: case DOT:
+ return new Expr(tok, prefix, parseMaximalExpr(null, precedence[tok]));
+
+ case BITNOT: case INSTANCEOF:
+ if (prefix != null) throw new Error("didn't expect non-null prefix!");
+ return new Expr(tok, parseMaximalExpr(null, precedence[tok]));
+
+ case INC: case DEC:
+ if (prefix == null) {
+ // prefix
+ return new Expr(tok, parseMaximalExpr(null, precedence[tok]));
+ } else {
+ // postfix
+ return new Expr(tok, null, prefix);
+ }
+
+ case LP:
+ if (prefix == null) { // grouping
+ Expr r = parseMaximalExpr();
+ expect(RP);
+ return r;
+
+ } else { // invocation
+ while(peekToken() != RP) {
+ Expr e = parseMaximalExpr(null, precedence[COMMA]);
+ if (head == null) head = tail = e; else tail = tail.next = e;
+ tok = getToken();
+ if (tok == RP) { pushBackToken(); break; }
+ if (tok != COMMA) throw new Error("expected comma or right paren, got " + codeToString[tok]);
+ }
+ getToken();
+ return new Expr(LP, prefix, head);
+ }
+
+ case LB:
+ if (prefix != null) {
+ // subscripting
+ e1 = parseSingleExpr();
+ if (getToken() != RB) throw new Error("expected a right brace");
+ return new Expr(LB, prefix, e1);
+ } else {
+ // array ctor
+ tok = getToken();
+ while(true) {
+ if (tok == RB) return new Expr(LB, prefix, head);
+ if (head == null) head = tail = parseSingleExpr(); else tail = tail.next = parseSingleExpr();
+ tok = getToken();
+ if (tok != COMMA && tok != RP) throw new Error("expected right bracket or comma");
+ }
+ }
+
+ case LC:
+ if (prefix != null) throw new Error("didn't expect non-null prefix");
+ tok = getToken();
+ while(true) {
+ if (tok == RC) return new Expr(LC, head);
+ if (tok != NAME) throw new Error("expecting name");
+ Expr name = parseSingleExpr();
+ if (tok != COLON) throw new Error("expecting colon");
+ e1 = new Expr(COLON, name, parseSingleExpr());
+ if (head == null) head = tail = e1; else tail = tail.next = e1;
+ tok = getToken();
+ if (tok != COMMA && tok != RP) throw new Error("expected right curly or comma");
+ }
+
+ case HOOK:
+ e2 = parseSingleExpr();
+ if (getToken() != COLON) throw new Error("expected colon to close ?: expression");
+ e3 = parseSingleExpr();
+ e2.next = e3;
+ return new Expr(HOOK, prefix, e2);
case SWITCH: {
+ if (prefix != null) throw new Error("didn't expect non-null prefix");
if (getToken() != LP) throw new Error("expected left paren");
- Expr switchExpr = parseExpr();
+ Expr switchExpr = parseSingleExpr();
if (getToken() != RP) throw new Error("expected left paren");
if (getToken() != LC) throw new Error("expected left brace");
Expr firstExpr = null;
Expr lastExpr = null;
while(true) {
if (getToken() != CASE) throw new Error("expected CASE");
- Expr caseExpr = parseExpr();
+ Expr caseExpr = parseSingleExpr();
if (getToken() != COLON) throw new Error("expected COLON");
Expr e = new Expr(CASE, caseExpr, parseBlock(false));
if (lastExpr == null) firstExpr = e;
}
case FUNCTION: {
+ if (prefix != null) throw new Error("didn't expect non-null prefix");
if (getToken() != LP) throw new Error("function keyword must be followed by a left paren");
Expr formalArgs = null, cur = null;
tok = getToken();
while(tok != RP) {
if (tok != NAME) throw new Error("expected a variable name");
if (cur == null) { formalArgs = cur = new Expr(string); }
- else { cur.next = new Expr(string); cur = cur.next; }
+ else { cur.next = new Expr(NAME, string); cur = cur.next; }
tok = getToken();
if (tok == RP) break;
if (tok != COMMA) throw new Error("function argument list must consist of alternating NAMEs and COMMAs");
tok = getToken();
}
- return new Expr(tok, formalArgs, parseBlock(true));
+ return new Expr(FUNCTION, formalArgs, parseBlock(true));
}
case VAR:
+ if (prefix != null) throw new Error("didn't expect non-null prefix");
while(true) {
if (getToken() != NAME) throw new Error("variable declarations must start with a variable name");
- Expr name = new Expr(string);
+ Expr name = new Expr(NAME, string);
Expr initVal = null;
tok = peekToken();
+ Expr e = null;
if (tok == ASSIGN) {
- skipToken();
- initVal = parseExpr();
+ getToken();
+ initVal = parseSingleExpr();
tok = peekToken();
+ e = new Expr(ASSIGN, name, initVal);
+ } else {
+ e = new Expr(NAME, name);
}
- Expr e = new Expr(VAR, name, initVal);
if (head == null) head = tail = e; else tail = tail.next = e;
if (tok != COMMA) break;
- skipToken();
+ getToken();
}
return new Expr(VAR, head);
-
- case LC:
- tok = getToken();
- while(true) {
- if (tok == RP) return new Expr(LC, head);
- if (tok != NAME) throw new Error("expecting name");
- Expr name = parseExpr();
- if (tok != COLON) throw new Error("expecting colon");
- e1 = new Expr(COLON, name, parseExpr());
- if (head == null) head = tail = e1; else tail = tail.next = e1;
- tok = getToken();
- if (tok != COMMA && tok != RP) throw new Error("expected right curly or comma");
- }
-
- case LB:
- tok = getToken();
- while(true) {
- if (tok == RB) return new Expr(LB, head);
- if (head == null) head = tail = parseExpr(); else tail = tail.next = parseExpr();
- tok = getToken();
- if (tok != COMMA && tok != RP) throw new Error("expected right bracket or comma");
- }
-
- case NAME:
- return new Expr(string);
-
- case INC: case DEC: case TYPEOF:
- return new Expr(tok, parseExpr());
-
- case TRUE: case FALSE: case NOP:
- return new Expr(tok);
-
+
case TRY: {
- // FIXME: we deliberately allow you to omit braces in catch{}/finally{} if they are single statements...
+ // We deliberately allow you to omit braces in catch{}/finally{} if they are single statements...
+ if (prefix != null) throw new Error("didn't expect non-null prefix");
Expr tryBlock = parseBlock(true);
- while ((tok = peekToken()) == CATCH)
- if (head == null) head = tail = parseBlock(false); else tail = tail.next = parseBlock(false);
- if (head == null) throw new Error("try without catch");
- return new Expr(TRY, tryBlock, head, tok == FINALLY ? skipToken().parseBlock(false) : null);
+ while ((tok = peekToken()) == CATCH || tok == FINALLY) {
+ getToken();
+ if (getToken() != LP) throw new Error("expected (");
+ if (getToken() != NAME) throw new Error("expected name");
+ Expr name = new Expr(NAME, string);
+ if (getToken() != RP) throw new Error("expected )");
+ e1 = new Expr(tok, name, parseBlock(false));
+ if (head == null) head = tail = e1; else tail = tail.next = e1;
+ }
+ if (head == null) throw new Error("try without catch or finally");
+ return new Expr(TRY, tryBlock, head);
}
-
+
case IF: case WHILE: {
+ if (prefix != null) throw new Error("didn't expect non-null prefix");
if (getToken() != LP) throw new Error("expected left paren");
- Expr parenExpr = parseExpr();
- if (getToken() != RP) throw new Error("expected right paren");
+ Expr parenExpr = parseMaximalExpr(null, -1);
+ int t;
+ if ((t = getToken()) != RP) throw new Error("expected right paren, but got " + codeToString[t]);
Expr firstBlock = parseBlock(false);
- if (tok == IF && peekToken() == ELSE) return new Expr(tok, parenExpr, firstBlock, skipToken().parseBlock(false));
+ if (tok == IF && peekToken() == ELSE) {
+ getToken();
+ firstBlock.next = parseBlock(false);
+ return new Expr(tok, parenExpr, firstBlock);
+ }
return new Expr(tok, parenExpr, firstBlock);
}
+ case IN: return prefix;
case FOR:
- // FIXME: for..in
+ if (prefix != null) throw new Error("didn't expect non-null prefix");
if (getToken() != LP) throw new Error("expected left paren");
- e1 = parseStatement();
- e2 = parseStatement();
- e3 = parseStatement(); // FIXME: this guy has to be okay with ending via a )
- if (getToken() != RP) throw new Error("expected right paren");
- throw new Error("not yet implemented");
- //return new Expr(FOR, e1, e2, e3, parseBlock(false));
+ e1 = parseMaximalExpr(null, -1);
+ if (e1.code == NAME && peekToken() == IN) {
+ getToken();
+ e2 = parseMaximalExpr(null, -1);
+ if (getToken() != RP) throw new Error("expected right paren");
+ return new Expr(FOR, new Expr(IN, e1, e2), parseBlock(false));
+
+ } else {
+ if (getToken() != SEMI) throw new Error("expected ;");
+ e2 = parseMaximalExpr(null, -1);
+ if (getToken() != SEMI) throw new Error("expected ;");
+ e3 = parseMaximalExpr(null, -1);
+ if (getToken() != RP) throw new Error("expected right paren");
+ return new Expr(LC, e1, new Expr(WHILE, e2, new Expr(LC, parseBlock(false), e3)));
+ }
case DO: {
+ if (prefix != null) throw new Error("didn't expect non-null prefix");
Expr firstBlock = parseBlock(false);
if (getToken() != WHILE) throw new Error("expecting WHILE");
if (getToken() != LP) throw new Error("expected left paren");
- Expr whileExpr = parseExpr();
+ Expr whileExpr = parseSingleExpr();
if (getToken() != RP) throw new Error("expected right paren");
if (getToken() != SEMI) throw new Error("semicolon");
return new Expr(DO, firstBlock, whileExpr);
}
- case VOID: case RESERVED:
- throw new Error("reserved word that you shouldn't be using");
-
- case WITH:
- throw new Error("WITH not yet implemented"); // FIXME
-
- default: throw new Error("I wasn't expecting a " + tok);
- }
- }
-
- // called after each parseExpr(); returns null if we can't make the expression any bigger
- public Expr parseContinuer(Expr prefix) throws IOException {
- Expr head = null;
- Expr tail = null;
- Expr e1, e2, e3;
- Expr ret = null;
- int tok;
-
- // FIXME: postfix and infix operators -- need to handle precedence
- switch (tok = getToken()) {
-
- case BITOR: case BITXOR: case BITAND: case EQ: case NE: case LT: case LE:
- case GT: case GE: case LSH: case RSH: case URSH: case ADD: case SUB: case MUL:
- case DIV: case MOD: case BITNOT: case SHEQ: case SHNE: case INSTANCEOF:
- case OR: case AND: case COMMA: case INC: case DEC:
- throw new Error("haven't figured out how to handle postfix/infix operators yet");
- //return new Expr(tok, prefix, (tok == INC || tok == DEC) ? null : parseExpr());
-
- case ASSIGN:
- throw new Error("haven't figured out how to handle postfix/infix operators yet");
-
- case LP:
- while(peekToken() != RP) {
- if (head == null) head = tail = parseExpr(); else tail = tail.next = parseExpr();
- tok = getToken();
- if (tok == RP) break;
- if (tok != COMMA) throw new Error("expected comma or right paren");
- }
- return new Expr(LP, prefix, head);
-
- case LB:
- e1 = parseExpr();
- if (getToken() != RB) throw new Error("expected a right brace");
- return new Expr(LB, prefix, e1);
-
- case HOOK:
- e2 = parseExpr();
- if (getToken() != COLON) throw new Error("expected colon to close ?: expression");
- e3 = parseExpr();
- return new Expr(HOOK, prefix, e2, e3);
-
default:
pushBackToken();
- return null;
+ return prefix;
}
}
projects / org.ibex.core.git / blobdiff