initial commit

This commit is contained in:
2025-11-18 17:31:44 +01:00
commit 669709e6aa
70 changed files with 11141 additions and 0 deletions

View File

@@ -0,0 +1,561 @@
package ssw.mj.impl;
import javassist.bytecode.analysis.ControlFlow;
import javassist.compiler.ast.MethodDecl;
import ssw.mj.Errors;
import ssw.mj.Errors.Message;
import ssw.mj.scanner.Token;
import javax.xml.stream.FactoryConfigurationError;
import java.util.EnumSet;
import static ssw.mj.Errors.Message.*;
import static ssw.mj.scanner.Token.Kind.*;
public final class Parser {
/**
* Maximum number of global variables per program
*/
private static final int MAX_GLOBALS = 32767;
/**
* Maximum number of fields per class
*/
private static final int MAX_FIELDS = 32767;
/**
* Maximum number of local variables per method
*/
private static final int MAX_LOCALS = 127;
/**
* Last recognized token;
*/
private Token t;
/**
* Lookahead token (not recognized).)
*/
private Token la;
/**
* Shortcut to kind attribute of lookahead token (la).
*/
private Token.Kind sym;
/**
* According scanner
*/
public final Scanner scanner;
/**
* According code buffer
*/
public final Code code;
/**
* According symbol table
*/
public final Tab tab;
private static int MIN_ERROR_DIST = 3;
private int errorDist = MIN_ERROR_DIST;
public Parser(Scanner scanner) {
this.scanner = scanner;
tab = new Tab(this);
code = new Code(this);
// Pseudo token to avoid crash when 1st symbol has scanner error.
la = new Token(none, 1, 1);
}
/**
* Reads ahead one symbol.
*/
private void scan() {
t = la;
la = scanner.next();
sym = la.kind;
errorDist++;
}
/**
* Verifies symbol and reads ahead.
*/
private void check(Token.Kind expected) {
if (sym == expected) {
scan();
} else {
error(TOKEN_EXPECTED, expected);
}
}
/**
* Adds error message to the list of errors.
*/
public void error(Message msg, Object... msgParams) {
// TODO Exercise UE-P-3: Replace panic mode with error recovery (i.e., keep track of error distance)
// TODO Exercise UE-P-3: Hint: Replacing panic mode also affects scan() method
if (errorDist >= MIN_ERROR_DIST) {
scanner.errors.error(la.line, la.col, msg, msgParams);
}
errorDist = 0;
}
/**
* Starts the analysis.
*/
public void parse() {
scan(); // scan first symbol, initializes look-ahead
Program(); // start analysis
check(eof);
}
// ===============================================
// TODO Exercise UE-P-2: Implementation of parser
// TODO Exercise UE-P-3: Error recovery methods
// TODO Exercise UE-P-4: Symbol table handling
// TODO Exercise UE-P-5-6: Code generation
// ===============================================
// TODO Exercise UE-P-3: Error distance
// TODO Exercise UE-P-2 + Exercise 3: Sets to handle certain first, follow, and recover sets
// Initialize first and follow sets.
static final EnumSet<Token.Kind> firstConstDecl = EnumSet.of(Token.Kind.final_);
static final EnumSet<Token.Kind> firstVarDecl = EnumSet.of(Token.Kind.ident);
static final EnumSet<Token.Kind> firstClassDecl = EnumSet.of(Token.Kind.class_);
static final EnumSet<Token.Kind> breakDecl = EnumSet.of(lbrace, eof);
static final EnumSet<Token.Kind> firstStatement = EnumSet.of(ident, if_, while_, break_, return_, read, print, lbrace, semicolon);
static final EnumSet<Token.Kind> breakStatement = EnumSet.of(rbrace, else_, eof);
static final EnumSet<Token.Kind> firstAsignop = EnumSet.of(assign, plusas, minusas, timesas, slashas, remas);
static final EnumSet<Token.Kind> firstFactor = EnumSet.of(ident, number, charConst, new_, lpar);
static final EnumSet<Token.Kind> firstMulop = EnumSet.of(times, slash, rem);
static final EnumSet<Token.Kind> firstMethodDecl = EnumSet.of(ident, void_);
static final EnumSet<Token.Kind> breakMethodDecl = EnumSet.of(rbrace, eof);
static <T extends Enum<T>> EnumSet<T> enumUnion(EnumSet<T> first, EnumSet<T> ...sets) {
final var copy = EnumSet.copyOf(first);
for (final var set : sets) {
copy.addAll(set);
}
return copy;
}
// ---------------------------------
// TODO Exercise UE-P-2: One top-down parsing method per production
/**
* Program = <br>
* "program" ident <br>
* { ConstDecl | VarDecl | ClassDecl } <br>
* "{" { MethodDecl } "}" .
*/
private void Program() {
// TODO Exercise UE-P-2
check(Token.Kind.program);
check(Token.Kind.ident);
while (!breakDecl.contains(sym)) {
if (firstVarDecl.contains(sym)) {
VarDecl();
} else if (firstClassDecl.contains(sym)) {
ClassDecl();
} else if (firstConstDecl.contains(sym)){
ConstDecl();
} else {
recoverDecl();
}
}
check(lbrace);
while (!breakMethodDecl.contains(sym)) {
if (firstMethodDecl.contains(sym)) {
MethodDecl();
} else {
recoverMethodDecl();
}
}
check(rbrace);
}
private void ConstDecl() {
check(Token.Kind.final_);
Type();
check(Token.Kind.ident);
check(Token.Kind.assign);
if (sym == Token.Kind.number) {
scan();
} else if (sym == Token.Kind.charConst) {
scan();
} else {
error(TOKEN_EXPECTED, "number or character constant");
}
check(Token.Kind.semicolon);
}
private void VarDecl() {
Type();
check(ident);
while (sym == Token.Kind.comma) {
scan();
check(ident);
}
check(semicolon);
}
private void ClassDecl() {
check(class_);
check(ident);
check(lbrace);
while (firstVarDecl.contains(sym)) {
VarDecl();
}
check(rbrace);
}
private void Type() {
check(ident);
if (sym == lbrack) {
scan();
check(rbrack);
}
}
private void MethodDecl() {
if (sym == ident) {
Type();
} else if (sym == void_) {
scan();
}
check(ident);
check(lpar);
if (sym == ident) {
FormPars();
}
check(rpar);
while (firstVarDecl.contains(sym)) {
VarDecl();
}
Block();
}
private void FormPars() {
Type();
check(ident);
while (sym == comma) {
scan();
Type();
check(ident);
}
}
private void Block() {
check(lbrace);
Statement();
check(rbrace);
}
private void Statement() {
while (!breakStatement.contains(sym)) {
switch (sym) {
case ident -> {
Designator();
if (firstAsignop.contains(sym)) {
AssignOp();
Expr();
} else if (sym == lpar) {
ActPars();
} else if (sym == pplus) {
scan();
} else if (sym == mminus) {
scan();
} else {
error(TOKEN_EXPECTED, "unexpected token. assignment token (=, +=, -=, *=, /=, %=), method call (\"(\"), increment (++) or decrement (--)");
}
check(semicolon);
}
case if_ -> {
scan();
check(lpar);
Condition();
check(rpar);
Statement();
if (sym == else_) {
scan();
Statement();
}
}
case while_ -> {
scan();
check(lpar);
Condition();
check(rpar);
Statement();
}
case break_-> {
scan();
check(semicolon);
}
case return_ -> {
scan();
if (sym == minus || firstFactor.contains(sym)) {
Expr();
}
check(semicolon);
}
case read -> {
scan();
check(lpar);
Designator();
check(rpar);
check(semicolon);
}
case print -> {
scan();
check(lpar);
Expr();
if (sym == comma) {
scan();
check(number);
}
check(rpar);
check(semicolon);
}
case lbrace -> Block();
case semicolon -> scan();
default -> recoverStatement();
};
}
}
private void Designator() {
check(ident);
while (sym == period || sym == lbrack) {
if (sym == period) {
scan();
check(ident);
} else {
scan();
if (sym == tilde) {
scan();
}
Expr();
check(rbrack);
}
}
}
private void AssignOp() {
switch (sym) {
case assign -> scan();
case plusas -> scan();
case minusas -> scan();
case timesas -> scan();
case slashas -> scan();
case remas -> scan();
default -> error(TOKEN_EXPECTED, "unexpected token. assignment token (=, +=, -=, *=, /=, %=), method call (\"(\"), increment (++) or decrement (--)");
}
}
private void Expr() {
if (sym == minus) {
scan();
}
Term();
while (sym == plus || sym == minus) {
Addop();
Term();
}
}
private void ActPars() {
check(lpar);
if (sym == minus || firstFactor.contains(sym)) {
Expr();
while (sym == comma) {
scan();
Expr();
}
}
check(rpar);
}
private void Term() {
Factor();
while (firstMulop.contains(sym)) {
Mulop();
Factor();
}
}
private void Addop() {
if (sym == plus) {
scan();
} else if (sym == minus) {
scan();
} else {
error(TOKEN_EXPECTED, "minus or plus expected");
}
}
private void Factor() {
switch(sym) {
case ident -> Designator();
case number -> scan();
case charConst -> scan();
case new_ -> {
scan();
check(ident);
if (sym == lbrack) {
scan();
Expr();
check(rbrack);
}
}
case lpar -> {
scan();
Expr();
check(rpar);
}
default -> error(TOKEN_EXPECTED, "unexpected token. identifier, number, character constant, new or \"(\"");
}
}
private void Mulop() {
if (sym == times) {
scan();
} else if (sym == slash) {
scan();
} else if (sym == rem) {
scan();
} else {
error(TOKEN_EXPECTED, "expected *, /, %");
}
}
private void Condition() {
CondTerm();
while (sym == or) {
scan();
CondTerm();
}
}
private void CondTerm() {
CondFact();
while (sym == and) {
scan();
CondFact();
}
}
private void CondFact() {
Expr();
Relop();
Expr();
}
private void Relop() {
switch (sym) {
case eql -> scan();
case neq -> scan();
case gtr -> scan();
case lss -> scan();
case leq -> scan();
case geq -> scan();
default -> error(TOKEN_EXPECTED, "unexpected token. ==, !=, >, >=, <, <=");
}
}
// ...
// ------------------------------------
// TODO Exercise UE-P-3: Error recovery methods: recoverDecl, recoverMethodDecl and recoverStat (+ TODO Exercise UE-P-5: Check idents for Type kind)
private void recoverDecl() {
final var firstDeclarationUnion = enumUnion(firstConstDecl, firstVarDecl, firstClassDecl);
final var recoverDeclSet = enumUnion(breakDecl, firstDeclarationUnion);
error(DECLARATION_RECOVERY);
do {
scan();
} while (!recoverDeclSet.contains(sym));
errorDist = 0;
}
private void recoverMethodDecl() {
error(METHOD_DECL_RECOVERY);
final var recoveryMethoDeclSet = enumUnion(breakMethodDecl, firstMethodDecl);
do {
scan();
} while (!recoveryMethoDeclSet.contains(sym));
errorDist = 0;
}
private void recoverStatement() {
error(STATEMENT_RECOVERY);
var recoveryStatementSet = enumUnion(breakStatement, firstStatement);
recoveryStatementSet.remove(ident);
recoveryStatementSet.remove(lbrace);
do {
scan();
} while (!recoveryStatementSet.contains(sym));
errorDist = 0;
}
// ====================================
// ====================================
}