2020-12-26 03:32:21 +03:00
|
|
|
#include <stdio.h>
|
|
|
|
#include <stdlib.h>
|
2020-12-26 12:39:29 +03:00
|
|
|
#include <string.h>
|
2020-12-28 10:07:52 +03:00
|
|
|
#include <sys/types.h>
|
2020-12-26 03:32:21 +03:00
|
|
|
|
|
|
|
#include "kuroko.h"
|
|
|
|
#include "compiler.h"
|
2020-12-27 09:58:32 +03:00
|
|
|
#include "memory.h"
|
2020-12-26 03:32:21 +03:00
|
|
|
#include "scanner.h"
|
|
|
|
#include "object.h"
|
2020-12-27 03:33:28 +03:00
|
|
|
#include "debug.h"
|
2020-12-28 03:08:35 +03:00
|
|
|
#include "vm.h"
|
2020-12-26 03:32:21 +03:00
|
|
|
|
|
|
|
typedef struct {
|
|
|
|
KrkToken current;
|
|
|
|
KrkToken previous;
|
|
|
|
int hadError;
|
|
|
|
int panicMode;
|
|
|
|
} Parser;
|
|
|
|
|
|
|
|
typedef enum {
|
|
|
|
PREC_NONE,
|
|
|
|
PREC_ASSIGNMENT, /* = */
|
|
|
|
PREC_OR, /* or */
|
|
|
|
PREC_AND, /* and */
|
|
|
|
PREC_EQUALITY, /* == != in */
|
|
|
|
PREC_COMPARISON, /* < > <= >= */
|
|
|
|
PREC_TERM, /* + - */
|
|
|
|
PREC_FACTOR, /* * */
|
|
|
|
PREC_UNARY, /* ! - not */
|
|
|
|
PREC_CALL, /* . () */
|
|
|
|
PREC_PRIMARY
|
|
|
|
} Precedence;
|
|
|
|
|
2020-12-26 10:53:15 +03:00
|
|
|
typedef void (*ParseFn)(int);
|
2020-12-26 03:32:21 +03:00
|
|
|
|
|
|
|
typedef struct {
|
|
|
|
ParseFn prefix;
|
|
|
|
ParseFn infix;
|
|
|
|
Precedence precedence;
|
|
|
|
} ParseRule;
|
|
|
|
|
2020-12-26 12:39:29 +03:00
|
|
|
typedef struct {
|
|
|
|
KrkToken name;
|
|
|
|
ssize_t depth;
|
2020-12-27 07:02:26 +03:00
|
|
|
int isCaptured;
|
2020-12-26 12:39:29 +03:00
|
|
|
} Local;
|
|
|
|
|
2020-12-27 07:02:26 +03:00
|
|
|
typedef struct {
|
|
|
|
size_t index;
|
|
|
|
int isLocal;
|
|
|
|
} Upvalue;
|
|
|
|
|
2020-12-27 03:33:28 +03:00
|
|
|
typedef enum {
|
|
|
|
TYPE_FUNCTION,
|
|
|
|
TYPE_MODULE,
|
2020-12-27 11:53:46 +03:00
|
|
|
TYPE_METHOD,
|
|
|
|
TYPE_INIT,
|
2020-12-27 03:33:28 +03:00
|
|
|
} FunctionType;
|
|
|
|
|
|
|
|
#define MAX_LOCALS 256
|
|
|
|
typedef struct Compiler {
|
|
|
|
struct Compiler * enclosing;
|
|
|
|
KrkFunction * function;
|
|
|
|
FunctionType type;
|
|
|
|
Local locals[MAX_LOCALS];
|
2020-12-26 12:39:29 +03:00
|
|
|
size_t localCount;
|
|
|
|
size_t scopeDepth;
|
2020-12-27 07:02:26 +03:00
|
|
|
Upvalue upvalues[MAX_LOCALS];
|
2020-12-26 12:39:29 +03:00
|
|
|
} Compiler;
|
|
|
|
|
2020-12-27 11:53:46 +03:00
|
|
|
typedef struct ClassCompiler {
|
|
|
|
struct ClassCompiler * enclosing;
|
|
|
|
KrkToken name;
|
2020-12-27 16:40:35 +03:00
|
|
|
int hasSuperClass;
|
2020-12-27 11:53:46 +03:00
|
|
|
} ClassCompiler;
|
|
|
|
|
2020-12-26 03:32:21 +03:00
|
|
|
Parser parser;
|
2020-12-26 12:39:29 +03:00
|
|
|
Compiler * current = NULL;
|
2020-12-27 11:53:46 +03:00
|
|
|
ClassCompiler * currentClass = NULL;
|
2020-12-26 12:39:29 +03:00
|
|
|
|
2020-12-26 03:32:21 +03:00
|
|
|
static KrkChunk * currentChunk() {
|
2020-12-27 03:33:28 +03:00
|
|
|
return ¤t->function->chunk;
|
2020-12-26 03:32:21 +03:00
|
|
|
}
|
|
|
|
|
2020-12-27 07:02:26 +03:00
|
|
|
#define EMIT_CONSTANT_OP(opc, arg) do { if (arg < 256) { emitBytes(opc, arg); } \
|
|
|
|
else { emitBytes(opc ## _LONG, arg >> 16); emitBytes(arg >> 8, arg); } } while (0)
|
|
|
|
|
2020-12-27 03:33:28 +03:00
|
|
|
static void initCompiler(Compiler * compiler, FunctionType type) {
|
|
|
|
compiler->enclosing = current;
|
|
|
|
compiler->function = NULL;
|
|
|
|
compiler->type = type;
|
2020-12-26 12:39:29 +03:00
|
|
|
compiler->localCount = 0;
|
|
|
|
compiler->scopeDepth = 0;
|
2020-12-28 05:11:50 +03:00
|
|
|
compiler->function = krk_newFunction();
|
2020-12-26 12:39:29 +03:00
|
|
|
current = compiler;
|
2020-12-27 03:33:28 +03:00
|
|
|
|
|
|
|
if (type != TYPE_MODULE) {
|
2020-12-28 05:11:50 +03:00
|
|
|
current->function->name = krk_copyString(parser.previous.start, parser.previous.length);
|
2020-12-27 03:33:28 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
Local * local = ¤t->locals[current->localCount++];
|
|
|
|
local->depth = 0;
|
2020-12-27 07:02:26 +03:00
|
|
|
local->isCaptured = 0;
|
2020-12-27 11:53:46 +03:00
|
|
|
|
|
|
|
if (type != TYPE_FUNCTION) {
|
|
|
|
local->name.start = "self";
|
|
|
|
local->name.length = 4;
|
|
|
|
} else {
|
|
|
|
local->name.start = "";
|
|
|
|
local->name.length = 0;
|
|
|
|
}
|
2020-12-26 12:39:29 +03:00
|
|
|
}
|
|
|
|
|
2020-12-26 03:32:21 +03:00
|
|
|
static void parsePrecedence(Precedence precedence);
|
2020-12-26 12:39:29 +03:00
|
|
|
static ssize_t parseVariable(const char * errorMessage);
|
2020-12-27 16:40:35 +03:00
|
|
|
static void variable(int canAssign);
|
2020-12-26 10:53:15 +03:00
|
|
|
static void defineVariable(size_t global);
|
2020-12-27 03:33:28 +03:00
|
|
|
static uint8_t argumentList();
|
2020-12-26 12:39:29 +03:00
|
|
|
static ssize_t identifierConstant(KrkToken * name);
|
|
|
|
static ssize_t resolveLocal(Compiler * compiler, KrkToken * name);
|
2020-12-26 03:32:21 +03:00
|
|
|
static ParseRule * getRule(KrkTokenType type);
|
2020-12-27 03:33:28 +03:00
|
|
|
static void defDeclaration();
|
2020-12-26 10:53:15 +03:00
|
|
|
static void expression();
|
|
|
|
static void statement();
|
|
|
|
static void declaration();
|
2020-12-26 14:39:47 +03:00
|
|
|
static void or_(int canAssign);
|
|
|
|
static void and_(int canAssign);
|
2020-12-27 10:45:34 +03:00
|
|
|
static void classDeclaration();
|
|
|
|
static void declareVariable();
|
2020-12-27 11:53:46 +03:00
|
|
|
static void namedVariable(KrkToken name, int canAssign);
|
2020-12-27 16:40:35 +03:00
|
|
|
static void addLocal(KrkToken name);
|
2020-12-26 03:32:21 +03:00
|
|
|
|
|
|
|
static void errorAt(KrkToken * token, const char * message) {
|
|
|
|
if (parser.panicMode) return;
|
|
|
|
parser.panicMode = 1;
|
|
|
|
|
|
|
|
fprintf(stderr, "[line %d] Error", (int)token->line);
|
|
|
|
if (token->type == TOKEN_EOF) {
|
|
|
|
fprintf(stderr, " at end");
|
|
|
|
} else if (token->type != TOKEN_ERROR) {
|
|
|
|
fprintf(stderr, " at '%.*s'", (int)token->length, token->start);
|
|
|
|
}
|
|
|
|
|
|
|
|
fprintf(stderr, ": %s\n", message);
|
|
|
|
parser.hadError = 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void error(const char * message) {
|
|
|
|
errorAt(&parser.previous, message);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void errorAtCurrent(const char * message) {
|
2020-12-26 14:39:47 +03:00
|
|
|
errorAt(&parser.previous, "(token before actual error)");
|
|
|
|
parser.panicMode = 0;
|
2020-12-26 03:32:21 +03:00
|
|
|
errorAt(&parser.current, message);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void advance() {
|
|
|
|
parser.previous = parser.current;
|
|
|
|
|
|
|
|
for (;;) {
|
|
|
|
parser.current = krk_scanToken();
|
2020-12-28 04:54:25 +03:00
|
|
|
|
|
|
|
#ifdef ENABLE_SCAN_TRACING
|
2020-12-28 14:38:26 +03:00
|
|
|
if (vm.flags & KRK_ENABLE_SCAN_TRACING) {
|
2020-12-28 04:54:25 +03:00
|
|
|
fprintf(stderr, "Token %d '%.*s' on line %d\n", parser.current.type,
|
|
|
|
(int)parser.current.length,
|
|
|
|
parser.current.start,
|
|
|
|
(int)parser.current.line);
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2020-12-26 03:32:21 +03:00
|
|
|
if (parser.current.type == TOKEN_RETRY) continue;
|
|
|
|
if (parser.current.type != TOKEN_ERROR) break;
|
|
|
|
|
|
|
|
errorAtCurrent(parser.current.start);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void consume(KrkTokenType type, const char * message) {
|
|
|
|
if (parser.current.type == type) {
|
|
|
|
advance();
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
errorAtCurrent(message);
|
|
|
|
}
|
|
|
|
|
2020-12-26 10:53:15 +03:00
|
|
|
static int check(KrkTokenType type) {
|
|
|
|
return parser.current.type == type;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int match(KrkTokenType type) {
|
|
|
|
if (!check(type)) return 0;
|
|
|
|
advance();
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2020-12-27 16:40:35 +03:00
|
|
|
static int identifiersEqual(KrkToken * a, KrkToken * b) {
|
|
|
|
return (a->length == b->length && memcmp(a->start, b->start, a->length) == 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
static KrkToken syntheticToken(const char * text) {
|
|
|
|
KrkToken token;
|
|
|
|
token.start = text;
|
|
|
|
token.length = (int)strlen(text);
|
|
|
|
return token;
|
|
|
|
}
|
|
|
|
|
2020-12-26 03:32:21 +03:00
|
|
|
static void emitByte(uint8_t byte) {
|
|
|
|
krk_writeChunk(currentChunk(), byte, parser.previous.line);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void emitBytes(uint8_t byte1, uint8_t byte2) {
|
|
|
|
emitByte(byte1);
|
|
|
|
emitByte(byte2);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void emitReturn() {
|
2020-12-27 11:53:46 +03:00
|
|
|
if (current->type == TYPE_INIT) {
|
|
|
|
emitBytes(OP_GET_LOCAL, 0);
|
2020-12-28 04:54:25 +03:00
|
|
|
} else if (current->type == TYPE_MODULE) {
|
|
|
|
/* Un-pop the last stack value */
|
|
|
|
emitBytes(OP_GET_LOCAL, 1);
|
2020-12-27 11:53:46 +03:00
|
|
|
} else {
|
|
|
|
emitByte(OP_NONE);
|
|
|
|
}
|
|
|
|
emitByte(OP_RETURN);
|
2020-12-26 03:32:21 +03:00
|
|
|
}
|
|
|
|
|
2020-12-27 03:33:28 +03:00
|
|
|
static KrkFunction * endCompiler() {
|
2020-12-26 03:32:21 +03:00
|
|
|
emitReturn();
|
2020-12-27 03:33:28 +03:00
|
|
|
KrkFunction * function = current->function;
|
2020-12-28 03:08:35 +03:00
|
|
|
#ifdef ENABLE_DEBUGGING
|
2020-12-28 14:38:26 +03:00
|
|
|
if ((vm.flags & KRK_ENABLE_DEBUGGING) && !parser.hadError) {
|
2020-12-27 03:33:28 +03:00
|
|
|
krk_disassembleChunk(currentChunk(), function->name != NULL ? function->name->chars : "<module>");
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
current = current->enclosing;
|
|
|
|
return function;
|
2020-12-26 03:32:21 +03:00
|
|
|
}
|
|
|
|
|
2020-12-26 10:53:15 +03:00
|
|
|
static void endOfLine() {
|
|
|
|
if (!(match(TOKEN_EOL) || match(TOKEN_EOF))) {
|
|
|
|
errorAtCurrent("Expected end of line.");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-12-27 09:58:32 +03:00
|
|
|
static size_t emitConstant(KrkValue value) {
|
|
|
|
return krk_writeConstant(currentChunk(), value, parser.previous.line);
|
2020-12-26 03:32:21 +03:00
|
|
|
}
|
|
|
|
|
2020-12-26 10:53:15 +03:00
|
|
|
static void number(int canAssign) {
|
2020-12-27 04:14:52 +03:00
|
|
|
const char * start = parser.previous.start;
|
2020-12-26 03:32:21 +03:00
|
|
|
int base = 10;
|
2020-12-27 04:14:52 +03:00
|
|
|
|
|
|
|
/* These special cases for hexadecimal, binary, octal values. */
|
2020-12-26 03:32:21 +03:00
|
|
|
if (start[0] == '0' && (start[1] == 'x' || start[1] == 'X')) {
|
2020-12-27 04:14:52 +03:00
|
|
|
base = 16;
|
2020-12-26 03:32:21 +03:00
|
|
|
start += 2;
|
|
|
|
} else if (start[0] == '0' && (start[1] == 'b' || start[1] == 'B')) {
|
|
|
|
base = 2;
|
|
|
|
start += 2;
|
2020-12-27 04:14:52 +03:00
|
|
|
} else if (start[0] == '0' && (start[1] == 'o' || start[1] == 'O')) {
|
2020-12-26 03:32:21 +03:00
|
|
|
base = 8;
|
2020-12-27 04:14:52 +03:00
|
|
|
start += 2;
|
2020-12-26 03:32:21 +03:00
|
|
|
}
|
2020-12-27 04:14:52 +03:00
|
|
|
|
|
|
|
/* If it wasn't a special base, it may be a floating point value. */
|
2020-12-26 03:32:21 +03:00
|
|
|
if (base == 10) {
|
|
|
|
for (size_t j = 0; j < parser.previous.length; ++j) {
|
|
|
|
if (parser.previous.start[j] == '.') {
|
|
|
|
double value = strtod(start, NULL);
|
|
|
|
emitConstant(FLOATING_VAL(value));
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
2020-12-27 04:14:52 +03:00
|
|
|
|
|
|
|
/* If we got here, it's an integer of some sort. */
|
2020-12-26 03:32:21 +03:00
|
|
|
int value = strtol(start, NULL, base);
|
|
|
|
emitConstant(INTEGER_VAL(value));
|
|
|
|
}
|
|
|
|
|
2020-12-26 10:53:15 +03:00
|
|
|
static void binary(int canAssign) {
|
2020-12-26 03:32:21 +03:00
|
|
|
KrkTokenType operatorType = parser.previous.type;
|
|
|
|
ParseRule * rule = getRule(operatorType);
|
|
|
|
parsePrecedence((Precedence)(rule->precedence + 1));
|
|
|
|
|
|
|
|
switch (operatorType) {
|
|
|
|
case TOKEN_BANG_EQUAL: emitBytes(OP_EQUAL, OP_NOT); break;
|
|
|
|
case TOKEN_EQUAL_EQUAL: emitByte(OP_EQUAL); break;
|
|
|
|
case TOKEN_GREATER: emitByte(OP_GREATER); break;
|
|
|
|
case TOKEN_GREATER_EQUAL: emitBytes(OP_LESS, OP_NOT); break;
|
|
|
|
case TOKEN_LESS: emitByte(OP_LESS); break;
|
|
|
|
case TOKEN_LESS_EQUAL: emitBytes(OP_GREATER, OP_NOT); break;
|
|
|
|
|
|
|
|
case TOKEN_PLUS: emitByte(OP_ADD); break;
|
|
|
|
case TOKEN_MINUS: emitByte(OP_SUBTRACT); break;
|
|
|
|
case TOKEN_ASTERISK: emitByte(OP_MULTIPLY); break;
|
|
|
|
case TOKEN_SOLIDUS: emitByte(OP_DIVIDE); break;
|
|
|
|
default: return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-12-27 03:33:28 +03:00
|
|
|
static void call(int canAssign) {
|
|
|
|
uint8_t argCount = argumentList();
|
|
|
|
emitBytes(OP_CALL, argCount);
|
|
|
|
}
|
|
|
|
|
2020-12-27 16:40:35 +03:00
|
|
|
static void get_(int canAssign) {
|
|
|
|
/* Synthesize get */
|
|
|
|
KrkToken _get = syntheticToken("__get__");
|
|
|
|
KrkToken _set = syntheticToken("__set__");
|
|
|
|
size_t indGet = identifierConstant(&_get);
|
|
|
|
size_t indSet = identifierConstant(&_set);
|
|
|
|
size_t offset = currentChunk()->count + 1;
|
|
|
|
emitBytes(OP_GET_PROPERTY, indGet); /* TODO what if it's > 256 */
|
|
|
|
expression();
|
|
|
|
consume(TOKEN_RIGHT_SQUARE, "Expected ending square bracket...");
|
|
|
|
if (canAssign && match(TOKEN_EQUAL)) {
|
|
|
|
expression();
|
|
|
|
currentChunk()->code[offset] = indSet;
|
|
|
|
emitBytes(OP_CALL, 2);
|
|
|
|
} else {
|
|
|
|
emitBytes(OP_CALL, 1);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-12-27 10:45:34 +03:00
|
|
|
static void dot(int canAssign) {
|
|
|
|
consume(TOKEN_IDENTIFIER, "Expected propert name");
|
|
|
|
size_t ind = identifierConstant(&parser.previous);
|
|
|
|
if (canAssign && match(TOKEN_EQUAL)) {
|
|
|
|
expression();
|
|
|
|
EMIT_CONSTANT_OP(OP_SET_PROPERTY, ind);
|
|
|
|
} else {
|
|
|
|
EMIT_CONSTANT_OP(OP_GET_PROPERTY, ind);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-12-26 10:53:15 +03:00
|
|
|
static void literal(int canAssign) {
|
2020-12-26 03:32:21 +03:00
|
|
|
switch (parser.previous.type) {
|
|
|
|
case TOKEN_FALSE: emitByte(OP_FALSE); break;
|
|
|
|
case TOKEN_NONE: emitByte(OP_NONE); break;
|
|
|
|
case TOKEN_TRUE: emitByte(OP_TRUE); break;
|
|
|
|
default: return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void expression() {
|
|
|
|
parsePrecedence(PREC_ASSIGNMENT);
|
|
|
|
}
|
|
|
|
|
2020-12-26 10:53:15 +03:00
|
|
|
static void varDeclaration() {
|
2020-12-27 16:40:35 +03:00
|
|
|
ssize_t ind = parseVariable("Expected variable name.");
|
2020-12-26 10:53:15 +03:00
|
|
|
|
|
|
|
if (match(TOKEN_EQUAL)) {
|
|
|
|
expression();
|
|
|
|
} else {
|
|
|
|
emitByte(OP_NONE);
|
|
|
|
}
|
|
|
|
|
2020-12-27 16:40:35 +03:00
|
|
|
defineVariable(ind);
|
2020-12-26 10:53:15 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static void printStatement() {
|
|
|
|
expression();
|
|
|
|
emitByte(OP_PRINT);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void synchronize() {
|
|
|
|
parser.panicMode = 0;
|
|
|
|
while (parser.current.type != TOKEN_EOF) {
|
|
|
|
if (parser.previous.type == TOKEN_EOL) return;
|
|
|
|
|
|
|
|
switch (parser.current.type) {
|
|
|
|
case TOKEN_CLASS:
|
|
|
|
case TOKEN_DEF:
|
|
|
|
case TOKEN_LET:
|
|
|
|
case TOKEN_FOR:
|
|
|
|
case TOKEN_IF:
|
|
|
|
case TOKEN_WHILE:
|
|
|
|
case TOKEN_PRINT:
|
|
|
|
case TOKEN_RETURN:
|
|
|
|
return;
|
|
|
|
default: break;
|
|
|
|
}
|
|
|
|
|
|
|
|
advance();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void declaration() {
|
2020-12-27 03:33:28 +03:00
|
|
|
if (check(TOKEN_DEF)) {
|
|
|
|
defDeclaration();
|
|
|
|
} else if (match(TOKEN_LET)) {
|
2020-12-26 10:53:15 +03:00
|
|
|
varDeclaration();
|
2020-12-27 10:45:34 +03:00
|
|
|
} else if (check(TOKEN_CLASS)) {
|
|
|
|
classDeclaration();
|
2020-12-26 14:39:47 +03:00
|
|
|
} else if (check(TOKEN_EOL)) {
|
|
|
|
return;
|
2020-12-26 10:53:15 +03:00
|
|
|
} else {
|
|
|
|
statement();
|
|
|
|
}
|
|
|
|
|
|
|
|
if (parser.panicMode) synchronize();
|
|
|
|
}
|
|
|
|
|
|
|
|
static void expressionStatement() {
|
|
|
|
expression();
|
|
|
|
emitByte(OP_POP);
|
|
|
|
}
|
|
|
|
|
2020-12-27 03:33:28 +03:00
|
|
|
static void beginScope() {
|
|
|
|
current->scopeDepth++;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void endScope() {
|
|
|
|
current->scopeDepth--;
|
|
|
|
while (current->localCount > 0 &&
|
2020-12-28 05:11:50 +03:00
|
|
|
current->locals[current->localCount - 1].depth > (ssize_t)current->scopeDepth) {
|
2020-12-27 07:02:26 +03:00
|
|
|
if (current->locals[current->localCount - 1].isCaptured) {
|
|
|
|
emitByte(OP_CLOSE_UPVALUE);
|
|
|
|
} else {
|
|
|
|
emitByte(OP_POP);
|
|
|
|
}
|
2020-12-27 03:33:28 +03:00
|
|
|
current->localCount--;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-12-28 05:11:50 +03:00
|
|
|
static void block(size_t indentation) {
|
2020-12-26 12:39:29 +03:00
|
|
|
if (match(TOKEN_EOL)) {
|
|
|
|
/* Begin actual blocks */
|
|
|
|
if (check(TOKEN_INDENTATION)) {
|
|
|
|
size_t currentIndentation = parser.current.length;
|
2020-12-26 14:39:47 +03:00
|
|
|
if (currentIndentation <= indentation) {
|
|
|
|
errorAtCurrent("Unexpected indentation level for new block");
|
|
|
|
}
|
2020-12-26 12:39:29 +03:00
|
|
|
do {
|
2020-12-28 06:16:44 +03:00
|
|
|
if (parser.current.length < currentIndentation) break;
|
|
|
|
advance(); /* Pass indentation */
|
2020-12-26 12:39:29 +03:00
|
|
|
declaration();
|
2020-12-26 14:39:47 +03:00
|
|
|
if (check(TOKEN_EOL)) endOfLine();
|
2020-12-26 12:39:29 +03:00
|
|
|
} while (check(TOKEN_INDENTATION));
|
2020-12-28 07:35:02 +03:00
|
|
|
#ifdef ENABLE_DEBUGGING
|
2020-12-28 14:38:26 +03:00
|
|
|
if (vm.flags & KRK_ENABLE_DEBUGGING) {
|
2020-12-28 07:35:02 +03:00
|
|
|
fprintf(stderr, "On line %d, ", (int)parser.current.line);
|
|
|
|
if (check(TOKEN_INDENTATION)) {
|
|
|
|
fprintf(stderr, "Exiting block from %d to %d\n",
|
|
|
|
(int)currentIndentation, (int)parser.current.length);
|
|
|
|
} else {
|
|
|
|
fprintf(stderr, "Exiting block from %d to something that isn't indentation.\n",
|
|
|
|
(int)currentIndentation);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
#endif
|
2020-12-26 12:39:29 +03:00
|
|
|
} else {
|
|
|
|
errorAtCurrent("Expected indentation for block");
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
errorAtCurrent("Unsupported single-line block");
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-12-28 05:11:50 +03:00
|
|
|
static void function(FunctionType type, size_t blockWidth) {
|
2020-12-27 03:33:28 +03:00
|
|
|
Compiler compiler;
|
|
|
|
initCompiler(&compiler, type);
|
2020-12-28 04:54:25 +03:00
|
|
|
compiler.function->chunk.filename = compiler.enclosing->function->chunk.filename;
|
2020-12-27 03:33:28 +03:00
|
|
|
|
|
|
|
beginScope();
|
|
|
|
|
|
|
|
consume(TOKEN_LEFT_PAREN, "Expected start of parameter list after function name.");
|
|
|
|
if (!check(TOKEN_RIGHT_PAREN)) {
|
|
|
|
do {
|
2020-12-27 11:53:46 +03:00
|
|
|
if (match(TOKEN_SELF)) {
|
|
|
|
if (type != TYPE_INIT && type != TYPE_METHOD) {
|
|
|
|
error("Invalid use of `self` as a function paramenter.");
|
|
|
|
}
|
|
|
|
continue;
|
|
|
|
}
|
2020-12-27 03:33:28 +03:00
|
|
|
current->function->arity++;
|
|
|
|
if (current->function->arity > 255) errorAtCurrent("too many function parameters");
|
|
|
|
ssize_t paramConstant = parseVariable("Expect parameter name.");
|
|
|
|
defineVariable(paramConstant);
|
|
|
|
} while (match(TOKEN_COMMA));
|
|
|
|
}
|
|
|
|
consume(TOKEN_RIGHT_PAREN, "Expected end of parameter list.");
|
|
|
|
|
|
|
|
consume(TOKEN_COLON, "Expected colon after function signature.");
|
|
|
|
block(blockWidth);
|
|
|
|
|
|
|
|
KrkFunction * function = endCompiler();
|
2020-12-27 07:02:26 +03:00
|
|
|
size_t ind = krk_addConstant(currentChunk(), OBJECT_VAL(function));
|
|
|
|
EMIT_CONSTANT_OP(OP_CLOSURE, ind);
|
|
|
|
|
|
|
|
for (size_t i = 0; i < function->upvalueCount; ++i) {
|
|
|
|
/* TODO: if the maximum count changes, fix the sizes for this */
|
|
|
|
emitByte(compiler.upvalues[i].isLocal ? 1 : 0);
|
|
|
|
emitByte(compiler.upvalues[i].index);
|
|
|
|
}
|
2020-12-27 03:33:28 +03:00
|
|
|
}
|
|
|
|
|
2020-12-28 05:11:50 +03:00
|
|
|
static void method(size_t blockWidth) {
|
2020-12-27 11:53:46 +03:00
|
|
|
/* This is actually "inside of a class definition", and that might mean
|
|
|
|
* arbitrary blank lines we need to accept... Sorry. */
|
|
|
|
if (check(TOKEN_EOL)) return;
|
|
|
|
|
|
|
|
/* def method(...): - just like functions; unlike Python, I'm just always
|
|
|
|
* going to assign `self` because Lox always assigns `this`; it should not
|
|
|
|
* show up in the initializer list; I may add support for it being there
|
|
|
|
* as a redundant thing, just to make more Python stuff work with changes. */
|
|
|
|
consume(TOKEN_DEF, "expected a definition, got nothing");
|
|
|
|
consume(TOKEN_IDENTIFIER, "expected method name");
|
|
|
|
size_t ind = identifierConstant(&parser.previous);
|
|
|
|
FunctionType type = TYPE_METHOD;
|
|
|
|
|
|
|
|
if (parser.previous.length == 8 && memcmp(parser.previous.start, "__init__", 8) == 0) {
|
|
|
|
type = TYPE_INIT;
|
|
|
|
}
|
|
|
|
|
|
|
|
function(type, blockWidth);
|
|
|
|
EMIT_CONSTANT_OP(OP_METHOD, ind);
|
|
|
|
}
|
|
|
|
|
2020-12-27 10:45:34 +03:00
|
|
|
static void classDeclaration() {
|
2020-12-28 05:11:50 +03:00
|
|
|
size_t blockWidth = (parser.previous.type == TOKEN_INDENTATION) ? parser.previous.length : 0;
|
2020-12-27 10:45:34 +03:00
|
|
|
advance(); /* Collect the `class` */
|
|
|
|
|
|
|
|
consume(TOKEN_IDENTIFIER, "Expected class name.");
|
2020-12-27 11:53:46 +03:00
|
|
|
KrkToken className = parser.previous;
|
2020-12-27 10:45:34 +03:00
|
|
|
size_t constInd = identifierConstant(&parser.previous);
|
|
|
|
declareVariable();
|
|
|
|
|
|
|
|
EMIT_CONSTANT_OP(OP_CLASS, constInd);
|
|
|
|
defineVariable(constInd);
|
|
|
|
|
2020-12-27 11:53:46 +03:00
|
|
|
ClassCompiler classCompiler;
|
|
|
|
classCompiler.name = parser.previous;
|
2020-12-27 16:40:35 +03:00
|
|
|
classCompiler.hasSuperClass = 0;
|
2020-12-27 11:53:46 +03:00
|
|
|
classCompiler.enclosing = currentClass;
|
|
|
|
currentClass = &classCompiler;
|
|
|
|
|
2020-12-27 16:40:35 +03:00
|
|
|
if (match(TOKEN_LEFT_PAREN)) {
|
|
|
|
if (match(TOKEN_IDENTIFIER)) {
|
|
|
|
variable(0);
|
|
|
|
if (identifiersEqual(&className, &parser.previous)) {
|
|
|
|
error("A class can not inherit from itself.");
|
|
|
|
}
|
|
|
|
|
|
|
|
beginScope();
|
|
|
|
addLocal(syntheticToken("super"));
|
|
|
|
defineVariable(0);
|
|
|
|
|
|
|
|
namedVariable(className, 0);
|
|
|
|
emitByte(OP_INHERIT);
|
|
|
|
classCompiler.hasSuperClass = 1;
|
|
|
|
}
|
|
|
|
consume(TOKEN_RIGHT_PAREN, "Expected closing brace after superclass.");
|
|
|
|
}
|
|
|
|
|
2020-12-27 11:53:46 +03:00
|
|
|
namedVariable(className, 0);
|
|
|
|
|
2020-12-27 10:45:34 +03:00
|
|
|
consume(TOKEN_COLON, "Expected colon after class");
|
2020-12-27 11:53:46 +03:00
|
|
|
if (match(TOKEN_EOL)) {
|
|
|
|
if (check(TOKEN_INDENTATION)) {
|
|
|
|
size_t currentIndentation = parser.current.length;
|
|
|
|
if (currentIndentation <= blockWidth) {
|
|
|
|
errorAtCurrent("Unexpected indentation level for class");
|
|
|
|
}
|
|
|
|
do {
|
2020-12-28 06:16:44 +03:00
|
|
|
if (parser.current.length < currentIndentation) break;
|
2020-12-27 11:53:46 +03:00
|
|
|
advance(); /* Pass the indentation */
|
|
|
|
method(currentIndentation);
|
|
|
|
if (check(TOKEN_EOL)) endOfLine();
|
|
|
|
} while (check(TOKEN_INDENTATION));
|
|
|
|
/* Exit from block */
|
|
|
|
}
|
|
|
|
} /* else empty class (and at end of file?) we'll allow it for now... */
|
|
|
|
emitByte(OP_POP);
|
2020-12-27 16:40:35 +03:00
|
|
|
if (classCompiler.hasSuperClass) {
|
|
|
|
endScope();
|
|
|
|
}
|
2020-12-27 11:53:46 +03:00
|
|
|
currentClass = currentClass->enclosing;
|
2020-12-27 10:45:34 +03:00
|
|
|
}
|
|
|
|
|
2020-12-27 03:33:28 +03:00
|
|
|
static void markInitialized() {
|
|
|
|
if (current->scopeDepth == 0) return;
|
|
|
|
current->locals[current->localCount - 1].depth = current->scopeDepth;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void defDeclaration() {
|
2020-12-28 05:11:50 +03:00
|
|
|
size_t blockWidth = (parser.previous.type == TOKEN_INDENTATION) ? parser.previous.length : 0;
|
2020-12-27 03:33:28 +03:00
|
|
|
advance(); /* Collect the `def` */
|
|
|
|
|
|
|
|
ssize_t global = parseVariable("Expected function name.");
|
|
|
|
markInitialized();
|
|
|
|
function(TYPE_FUNCTION, blockWidth);
|
|
|
|
defineVariable(global);
|
|
|
|
}
|
|
|
|
|
2020-12-26 14:39:47 +03:00
|
|
|
static int emitJump(uint8_t opcode) {
|
|
|
|
emitByte(opcode);
|
|
|
|
emitBytes(0xFF, 0xFF);
|
|
|
|
return currentChunk()->count - 2;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void patchJump(int offset) {
|
|
|
|
int jump = currentChunk()->count - offset - 2;
|
|
|
|
if (jump > 0xFFFF) {
|
|
|
|
error("Unsupported far jump (we'll get there)");
|
|
|
|
}
|
|
|
|
|
|
|
|
currentChunk()->code[offset] = (jump >> 8) & 0xFF;
|
|
|
|
currentChunk()->code[offset + 1] = (jump) & 0xFF;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void emitLoop(int loopStart) {
|
|
|
|
emitByte(OP_LOOP);
|
|
|
|
int offset = currentChunk()->count - loopStart + 2;
|
|
|
|
if (offset > 0xFFFF) error("offset too big");
|
|
|
|
emitBytes(offset >> 8, offset);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void ifStatement() {
|
|
|
|
/* Figure out what block level contains us so we can match our partner else */
|
2020-12-28 05:11:50 +03:00
|
|
|
size_t blockWidth = (parser.previous.type == TOKEN_INDENTATION) ? parser.previous.length : 0;
|
2020-12-26 14:39:47 +03:00
|
|
|
|
|
|
|
/* Collect the if token that started this statement */
|
|
|
|
advance();
|
|
|
|
|
|
|
|
/* Collect condition expression */
|
|
|
|
expression();
|
|
|
|
|
|
|
|
/* if EXPR: */
|
|
|
|
consume(TOKEN_COLON, "Expect ':' after condition.");
|
|
|
|
|
|
|
|
int thenJump = emitJump(OP_JUMP_IF_FALSE);
|
|
|
|
emitByte(OP_POP);
|
|
|
|
|
|
|
|
/* Start a new scope and enter a block */
|
|
|
|
beginScope();
|
|
|
|
block(blockWidth);
|
|
|
|
endScope();
|
|
|
|
|
|
|
|
int elseJump = emitJump(OP_JUMP);
|
|
|
|
patchJump(thenJump);
|
|
|
|
emitByte(OP_POP);
|
|
|
|
|
|
|
|
/* See if we have a matching else block */
|
|
|
|
if (blockWidth == 0 || (check(TOKEN_INDENTATION) && (parser.current.length == blockWidth))) {
|
2020-12-28 06:16:44 +03:00
|
|
|
/* This is complicated */
|
|
|
|
KrkToken previous;
|
|
|
|
if (blockWidth) {
|
|
|
|
previous = parser.previous;
|
|
|
|
advance();
|
|
|
|
}
|
2020-12-26 14:39:47 +03:00
|
|
|
if (match(TOKEN_ELSE)) {
|
|
|
|
/* TODO ELIF or ELSE IF */
|
|
|
|
consume(TOKEN_COLON, "Expect ':' after else.");
|
|
|
|
beginScope();
|
|
|
|
block(blockWidth);
|
|
|
|
endScope();
|
2020-12-28 06:16:44 +03:00
|
|
|
} else {
|
|
|
|
krk_ungetToken(parser.current);
|
|
|
|
parser.current = parser.previous;
|
|
|
|
if (blockWidth) {
|
|
|
|
parser.previous = previous;
|
|
|
|
}
|
2020-12-26 14:39:47 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
patchJump(elseJump);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void whileStatement() {
|
2020-12-28 05:11:50 +03:00
|
|
|
size_t blockWidth = (parser.previous.type == TOKEN_INDENTATION) ? parser.previous.length : 0;
|
2020-12-26 14:39:47 +03:00
|
|
|
advance();
|
|
|
|
|
|
|
|
int loopStart = currentChunk()->count;
|
|
|
|
|
|
|
|
expression();
|
|
|
|
consume(TOKEN_COLON, "Expect ':' after condition.");
|
|
|
|
|
|
|
|
int exitJump = emitJump(OP_JUMP_IF_FALSE);
|
|
|
|
emitByte(OP_POP);
|
|
|
|
|
|
|
|
beginScope();
|
|
|
|
block(blockWidth);
|
|
|
|
endScope();
|
|
|
|
|
|
|
|
emitLoop(loopStart);
|
|
|
|
|
|
|
|
patchJump(exitJump);
|
|
|
|
emitByte(OP_POP);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void forStatement() {
|
|
|
|
/* I'm not sure if I want this to be more like Python or C/Lox/etc. */
|
2020-12-28 05:11:50 +03:00
|
|
|
size_t blockWidth = (parser.previous.type == TOKEN_INDENTATION) ? parser.previous.length : 0;
|
2020-12-26 14:39:47 +03:00
|
|
|
advance();
|
|
|
|
|
|
|
|
/* For now this is going to be kinda broken */
|
|
|
|
beginScope();
|
|
|
|
|
2020-12-28 07:35:02 +03:00
|
|
|
ssize_t loopInd = current->localCount;
|
2020-12-26 14:39:47 +03:00
|
|
|
varDeclaration();
|
2020-12-27 03:33:28 +03:00
|
|
|
|
2020-12-28 07:35:02 +03:00
|
|
|
int loopStart;
|
|
|
|
int exitJump;
|
2020-12-26 14:39:47 +03:00
|
|
|
|
2020-12-28 07:35:02 +03:00
|
|
|
if (match(TOKEN_IN)) {
|
|
|
|
defineVariable(loopInd);
|
2020-12-26 14:39:47 +03:00
|
|
|
|
2020-12-28 07:35:02 +03:00
|
|
|
KrkToken _it = syntheticToken("__loop_iter");
|
|
|
|
KrkToken _iter = syntheticToken("__iter__");
|
2020-12-28 05:11:50 +03:00
|
|
|
size_t indLoopIter = current->localCount;
|
2020-12-28 07:35:02 +03:00
|
|
|
|
|
|
|
/* __loop_iter = */
|
|
|
|
addLocal(_it);
|
|
|
|
defineVariable(indLoopIter);
|
|
|
|
|
|
|
|
/* ITERABLE.__iter__() */
|
2020-12-26 14:39:47 +03:00
|
|
|
expression();
|
2020-12-28 07:35:02 +03:00
|
|
|
ssize_t ind = identifierConstant(&_iter);
|
|
|
|
EMIT_CONSTANT_OP(OP_GET_PROPERTY, ind);
|
|
|
|
emitBytes(OP_CALL, 0);
|
|
|
|
|
|
|
|
/* assign */
|
|
|
|
EMIT_CONSTANT_OP(OP_SET_LOCAL, indLoopIter);
|
|
|
|
|
|
|
|
/* LOOP STARTS HERE */
|
|
|
|
loopStart = currentChunk()->count;
|
|
|
|
emitByte(0xFF);
|
|
|
|
|
|
|
|
/* Call the iterator */
|
|
|
|
EMIT_CONSTANT_OP(OP_GET_LOCAL, indLoopIter);
|
|
|
|
emitBytes(OP_CALL, 0);
|
|
|
|
|
|
|
|
/* Assign the result to our loop index */
|
|
|
|
EMIT_CONSTANT_OP(OP_SET_LOCAL, loopInd);
|
|
|
|
|
|
|
|
/* Get the loop iterator again */
|
|
|
|
EMIT_CONSTANT_OP(OP_GET_LOCAL, indLoopIter);
|
|
|
|
emitByte(OP_EQUAL);
|
|
|
|
emitByte(OP_NOT);
|
|
|
|
exitJump = emitJump(OP_JUMP_IF_FALSE);
|
|
|
|
emitByte(OP_POP);
|
|
|
|
|
|
|
|
} else {
|
|
|
|
consume(TOKEN_COMMA,"expect ,");
|
|
|
|
loopStart = currentChunk()->count;
|
|
|
|
|
|
|
|
|
|
|
|
expression(); /* condition */
|
|
|
|
exitJump = emitJump(OP_JUMP_IF_FALSE);
|
2020-12-26 14:39:47 +03:00
|
|
|
emitByte(OP_POP);
|
|
|
|
|
2020-12-28 07:35:02 +03:00
|
|
|
if (check(TOKEN_COMMA)) {
|
|
|
|
advance();
|
|
|
|
int bodyJump = emitJump(OP_JUMP);
|
|
|
|
int incrementStart = currentChunk()->count;
|
|
|
|
expression();
|
|
|
|
emitByte(OP_POP);
|
|
|
|
|
|
|
|
emitLoop(loopStart);
|
|
|
|
loopStart = incrementStart;
|
|
|
|
patchJump(bodyJump);
|
|
|
|
}
|
2020-12-26 14:39:47 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
consume(TOKEN_COLON,"expect :");
|
|
|
|
|
|
|
|
block(blockWidth);
|
|
|
|
|
|
|
|
emitLoop(loopStart);
|
|
|
|
patchJump(exitJump);
|
|
|
|
emitByte(OP_POP);
|
|
|
|
endScope();
|
|
|
|
}
|
|
|
|
|
2020-12-27 03:33:28 +03:00
|
|
|
static void returnStatement() {
|
|
|
|
if (check(TOKEN_EOL) || check(TOKEN_EOF)) {
|
|
|
|
emitReturn();
|
|
|
|
} else {
|
2020-12-27 11:53:46 +03:00
|
|
|
if (current->type == TYPE_INIT) {
|
|
|
|
error("Can not return values from __init__");
|
|
|
|
}
|
2020-12-27 03:33:28 +03:00
|
|
|
expression();
|
|
|
|
emitByte(OP_RETURN);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-12-27 12:55:52 +03:00
|
|
|
static void importStatement() {
|
|
|
|
consume(TOKEN_IDENTIFIER, "Expected module name");
|
|
|
|
declareVariable();
|
|
|
|
size_t ind = identifierConstant(&parser.previous);
|
|
|
|
EMIT_CONSTANT_OP(OP_IMPORT, ind);
|
|
|
|
defineVariable(ind);
|
|
|
|
}
|
|
|
|
|
2020-12-28 13:25:33 +03:00
|
|
|
static void exportStatement() {
|
|
|
|
consume(TOKEN_IDENTIFIER, "Expected variable name");
|
|
|
|
namedVariable(parser.previous, 0);
|
|
|
|
size_t ind = identifierConstant(&parser.previous);
|
|
|
|
EMIT_CONSTANT_OP(OP_DEFINE_GLOBAL, ind);
|
|
|
|
EMIT_CONSTANT_OP(OP_SET_GLOBAL, ind);
|
|
|
|
}
|
|
|
|
|
2020-12-26 10:53:15 +03:00
|
|
|
static void statement() {
|
2020-12-26 14:39:47 +03:00
|
|
|
if (check(TOKEN_EOL)) {
|
|
|
|
return; /* Meaningless blank line */
|
|
|
|
}
|
2020-12-26 12:39:29 +03:00
|
|
|
|
2020-12-26 10:53:15 +03:00
|
|
|
if (match(TOKEN_PRINT)) {
|
|
|
|
printStatement();
|
2020-12-28 13:25:33 +03:00
|
|
|
} else if (match(TOKEN_EXPORT)) {
|
|
|
|
exportStatement();
|
2020-12-26 14:39:47 +03:00
|
|
|
} else if (check(TOKEN_IF)) {
|
|
|
|
/*
|
|
|
|
* We check rather than match because we need to look at the indentation
|
|
|
|
* token that came before this (if it was one) to figure out what block
|
|
|
|
* indentation level we're at, so that we can match our companion else
|
|
|
|
* and make sure it's not the else for a higher if block.
|
|
|
|
*
|
|
|
|
* TODO: Are there other things where we want to do this?
|
|
|
|
*/
|
|
|
|
ifStatement();
|
|
|
|
} else if (check(TOKEN_WHILE)) {
|
|
|
|
whileStatement();
|
|
|
|
} else if (check(TOKEN_FOR)) {
|
|
|
|
forStatement();
|
2020-12-27 03:33:28 +03:00
|
|
|
} else if (match(TOKEN_RETURN)) {
|
|
|
|
returnStatement();
|
2020-12-27 12:55:52 +03:00
|
|
|
} else if (match(TOKEN_IMPORT)) {
|
|
|
|
importStatement();
|
2020-12-26 10:53:15 +03:00
|
|
|
} else {
|
|
|
|
expressionStatement();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void grouping(int canAssign) {
|
2020-12-26 03:32:21 +03:00
|
|
|
expression();
|
|
|
|
consume(TOKEN_RIGHT_PAREN, "Expect ')' after expression.");
|
|
|
|
}
|
|
|
|
|
2020-12-26 10:53:15 +03:00
|
|
|
static void unary(int canAssign) {
|
2020-12-26 03:32:21 +03:00
|
|
|
KrkTokenType operatorType = parser.previous.type;
|
|
|
|
|
|
|
|
parsePrecedence(PREC_UNARY);
|
|
|
|
|
|
|
|
switch (operatorType) {
|
|
|
|
case TOKEN_MINUS: emitByte(OP_NEGATE); break;
|
|
|
|
|
|
|
|
/* These are equivalent */
|
|
|
|
case TOKEN_BANG:
|
|
|
|
case TOKEN_NOT:
|
|
|
|
emitByte(OP_NOT);
|
|
|
|
break;
|
|
|
|
|
|
|
|
default: return;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-12-26 10:53:15 +03:00
|
|
|
static void string(int canAssign) {
|
2020-12-28 05:11:50 +03:00
|
|
|
emitConstant(OBJECT_VAL(krk_copyString(parser.previous.start + 1, parser.previous.length - 2)));
|
2020-12-26 03:32:21 +03:00
|
|
|
}
|
|
|
|
|
2020-12-27 04:58:36 +03:00
|
|
|
/* TODO
|
|
|
|
static void codepoint(int canAssign) {
|
|
|
|
// Convert utf8 bytes to single codepoint; error on multiple codepoints.
|
|
|
|
// Emit as constant Integer value? Or as separate Codepoint value?
|
|
|
|
// The latter could add to strings as utf8 bytes, but compare to
|
|
|
|
// Integers as the numerical value...
|
|
|
|
}
|
|
|
|
*/
|
|
|
|
|
2020-12-27 07:02:26 +03:00
|
|
|
static size_t addUpvalue(Compiler * compiler, ssize_t index, int isLocal) {
|
|
|
|
size_t upvalueCount = compiler->function->upvalueCount;
|
|
|
|
for (size_t i = 0; i < upvalueCount; ++i) {
|
|
|
|
Upvalue * upvalue = &compiler->upvalues[i];
|
2020-12-28 05:11:50 +03:00
|
|
|
if ((ssize_t)upvalue->index == index && upvalue->isLocal == isLocal) {
|
2020-12-27 07:02:26 +03:00
|
|
|
return i;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
if (upvalueCount == MAX_LOCALS) {
|
|
|
|
error("Too many closure variables in function.");
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
compiler->upvalues[upvalueCount].isLocal = isLocal;
|
|
|
|
compiler->upvalues[upvalueCount].index = index;
|
|
|
|
return compiler->function->upvalueCount++;
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t resolveUpvalue(Compiler * compiler, KrkToken * name) {
|
|
|
|
if (compiler->enclosing == NULL) return -1;
|
|
|
|
ssize_t local = resolveLocal(compiler->enclosing, name);
|
|
|
|
if (local != -1) {
|
|
|
|
compiler->enclosing->locals[local].isCaptured = 1;
|
|
|
|
return addUpvalue(compiler, local, 1);
|
|
|
|
}
|
|
|
|
ssize_t upvalue = resolveUpvalue(compiler->enclosing, name);
|
|
|
|
if (upvalue != -1) {
|
|
|
|
return addUpvalue(compiler, upvalue, 0);
|
|
|
|
}
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
#define DO_VARIABLE(opset,opget) do { \
|
|
|
|
if (canAssign && match(TOKEN_EQUAL)) { \
|
|
|
|
expression(); \
|
|
|
|
EMIT_CONSTANT_OP(opset, arg); \
|
|
|
|
} else { \
|
|
|
|
EMIT_CONSTANT_OP(opget, arg); \
|
|
|
|
} } while (0)
|
2020-12-26 10:53:15 +03:00
|
|
|
|
|
|
|
static void namedVariable(KrkToken name, int canAssign) {
|
2020-12-26 12:39:29 +03:00
|
|
|
ssize_t arg = resolveLocal(current, &name);
|
|
|
|
if (arg != -1) {
|
2020-12-27 07:02:26 +03:00
|
|
|
DO_VARIABLE(OP_SET_LOCAL, OP_GET_LOCAL);
|
|
|
|
} else if ((arg = resolveUpvalue(current, &name)) != -1) {
|
|
|
|
DO_VARIABLE(OP_SET_UPVALUE, OP_GET_UPVALUE);
|
2020-12-26 10:53:15 +03:00
|
|
|
} else {
|
2020-12-26 12:39:29 +03:00
|
|
|
arg = identifierConstant(&name);
|
2020-12-27 07:02:26 +03:00
|
|
|
DO_VARIABLE(OP_SET_GLOBAL, OP_GET_GLOBAL);
|
2020-12-26 10:53:15 +03:00
|
|
|
}
|
|
|
|
}
|
2020-12-27 07:02:26 +03:00
|
|
|
#undef DO_VARIABLE
|
2020-12-26 10:53:15 +03:00
|
|
|
|
|
|
|
static void variable(int canAssign) {
|
|
|
|
namedVariable(parser.previous, canAssign);
|
|
|
|
}
|
|
|
|
|
2020-12-27 11:53:46 +03:00
|
|
|
static void self(int canAssign) {
|
|
|
|
if (currentClass == NULL) {
|
|
|
|
error("Invalid reference to `self` outside of a class method.");
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
variable(0);
|
|
|
|
}
|
|
|
|
|
2020-12-27 16:40:35 +03:00
|
|
|
static void super_(int canAssign) {
|
|
|
|
if (currentClass == NULL) {
|
|
|
|
error("Invalid reference to `super` outside of a class.");
|
|
|
|
} else if (!currentClass->hasSuperClass) {
|
|
|
|
error("Invalid reference to `super` from a base class.");
|
|
|
|
}
|
|
|
|
consume(TOKEN_LEFT_PAREN, "Expected `super` to be called.");
|
|
|
|
consume(TOKEN_RIGHT_PAREN, "`super` can not take arguments.");
|
|
|
|
consume(TOKEN_DOT, "Expected a field of `super()` to be referenced.");
|
|
|
|
consume(TOKEN_IDENTIFIER, "Expected a field name.");
|
|
|
|
size_t ind = identifierConstant(&parser.previous);
|
|
|
|
namedVariable(syntheticToken("self"), 0);
|
|
|
|
namedVariable(syntheticToken("super"), 0);
|
|
|
|
EMIT_CONSTANT_OP(OP_GET_SUPER, ind);
|
|
|
|
}
|
|
|
|
|
2020-12-26 03:32:21 +03:00
|
|
|
ParseRule rules[] = {
|
2020-12-27 03:33:28 +03:00
|
|
|
[TOKEN_LEFT_PAREN] = {grouping, call, PREC_CALL},
|
2020-12-26 03:32:21 +03:00
|
|
|
[TOKEN_RIGHT_PAREN] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_LEFT_BRACE] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_RIGHT_BRACE] = {NULL, NULL, PREC_NONE},
|
2020-12-27 16:40:35 +03:00
|
|
|
[TOKEN_LEFT_SQUARE] = {NULL, get_, PREC_CALL},
|
2020-12-26 03:32:21 +03:00
|
|
|
[TOKEN_RIGHT_SQUARE] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_COLON] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_COMMA] = {NULL, NULL, PREC_NONE},
|
2020-12-27 10:45:34 +03:00
|
|
|
[TOKEN_DOT] = {NULL, dot, PREC_CALL},
|
2020-12-26 03:32:21 +03:00
|
|
|
[TOKEN_MINUS] = {unary, binary, PREC_TERM},
|
|
|
|
[TOKEN_PLUS] = {NULL, binary, PREC_TERM},
|
|
|
|
[TOKEN_SEMICOLON] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_SOLIDUS] = {NULL, binary, PREC_FACTOR},
|
|
|
|
[TOKEN_ASTERISK] = {NULL, binary, PREC_FACTOR},
|
|
|
|
[TOKEN_BANG] = {unary, NULL, PREC_NONE},
|
|
|
|
[TOKEN_BANG_EQUAL] = {NULL, binary, PREC_EQUALITY},
|
|
|
|
[TOKEN_EQUAL] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_EQUAL_EQUAL] = {NULL, binary, PREC_EQUALITY},
|
|
|
|
[TOKEN_GREATER] = {NULL, binary, PREC_COMPARISON},
|
|
|
|
[TOKEN_GREATER_EQUAL] = {NULL, binary, PREC_COMPARISON},
|
|
|
|
[TOKEN_LESS] = {NULL, binary, PREC_COMPARISON},
|
|
|
|
[TOKEN_LESS_EQUAL] = {NULL, binary, PREC_COMPARISON},
|
2020-12-26 10:53:15 +03:00
|
|
|
[TOKEN_IDENTIFIER] = {variable, NULL, PREC_NONE},
|
2020-12-26 03:32:21 +03:00
|
|
|
[TOKEN_STRING] = {string, NULL, PREC_NONE},
|
|
|
|
[TOKEN_NUMBER] = {number, NULL, PREC_NONE},
|
2020-12-27 04:58:36 +03:00
|
|
|
[TOKEN_CODEPOINT] = {NULL, NULL, PREC_NONE}, /* TODO */
|
2020-12-26 14:39:47 +03:00
|
|
|
[TOKEN_AND] = {NULL, and_, PREC_AND},
|
2020-12-26 03:32:21 +03:00
|
|
|
[TOKEN_CLASS] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_ELSE] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_FALSE] = {literal, NULL, PREC_NONE},
|
|
|
|
[TOKEN_FOR] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_DEF] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_IF] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_IN] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_LET] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_NONE] = {literal, NULL, PREC_NONE},
|
|
|
|
[TOKEN_NOT] = {unary, NULL, PREC_NONE},
|
2020-12-26 14:39:47 +03:00
|
|
|
[TOKEN_OR] = {NULL, or_, PREC_OR},
|
2020-12-26 03:32:21 +03:00
|
|
|
[TOKEN_PRINT] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_RETURN] = {NULL, NULL, PREC_NONE},
|
2020-12-27 11:53:46 +03:00
|
|
|
[TOKEN_SELF] = {self, NULL, PREC_NONE},
|
2020-12-27 16:40:35 +03:00
|
|
|
[TOKEN_SUPER] = {super_, NULL, PREC_NONE},
|
2020-12-26 03:32:21 +03:00
|
|
|
[TOKEN_TRUE] = {literal, NULL, PREC_NONE},
|
|
|
|
[TOKEN_WHILE] = {NULL, NULL, PREC_NONE},
|
|
|
|
|
|
|
|
/* This is going to get interesting */
|
|
|
|
[TOKEN_INDENTATION] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_ERROR] = {NULL, NULL, PREC_NONE},
|
|
|
|
[TOKEN_EOF] = {NULL, NULL, PREC_NONE},
|
|
|
|
};
|
|
|
|
|
|
|
|
static void parsePrecedence(Precedence precedence) {
|
|
|
|
advance();
|
|
|
|
ParseFn prefixRule = getRule(parser.previous.type)->prefix;
|
|
|
|
if (prefixRule == NULL) {
|
|
|
|
error("expect expression");
|
|
|
|
return;
|
|
|
|
}
|
2020-12-26 10:53:15 +03:00
|
|
|
int canAssign = precedence <= PREC_ASSIGNMENT;
|
|
|
|
prefixRule(canAssign);
|
2020-12-26 03:32:21 +03:00
|
|
|
while (precedence <= getRule(parser.current.type)->precedence) {
|
|
|
|
advance();
|
|
|
|
ParseFn infixRule = getRule(parser.previous.type)->infix;
|
2020-12-26 10:53:15 +03:00
|
|
|
infixRule(canAssign);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (canAssign && match(TOKEN_EQUAL)) {
|
|
|
|
error("invalid assignment target");
|
2020-12-26 03:32:21 +03:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2020-12-26 12:39:29 +03:00
|
|
|
static ssize_t identifierConstant(KrkToken * name) {
|
2020-12-28 05:11:50 +03:00
|
|
|
return krk_addConstant(currentChunk(), OBJECT_VAL(krk_copyString(name->start, name->length)));
|
2020-12-26 10:53:15 +03:00
|
|
|
}
|
|
|
|
|
2020-12-26 12:39:29 +03:00
|
|
|
static ssize_t resolveLocal(Compiler * compiler, KrkToken * name) {
|
|
|
|
for (ssize_t i = compiler->localCount - 1; i >= 0; i--) {
|
|
|
|
Local * local = &compiler->locals[i];
|
|
|
|
if (identifiersEqual(name, &local->name)) {
|
|
|
|
if (local->depth == -1) {
|
|
|
|
error("can not initialize value recursively (are you shadowing something?)");
|
|
|
|
}
|
|
|
|
return i;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void addLocal(KrkToken name) {
|
2020-12-27 03:33:28 +03:00
|
|
|
if (current->localCount == MAX_LOCALS) {
|
2020-12-26 12:39:29 +03:00
|
|
|
error("too many locals");
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
Local * local = ¤t->locals[current->localCount++];
|
|
|
|
local->name = name;
|
|
|
|
local->depth = -1;
|
2020-12-27 07:02:26 +03:00
|
|
|
local->isCaptured = 0;
|
2020-12-26 12:39:29 +03:00
|
|
|
}
|
|
|
|
|
|
|
|
static void declareVariable() {
|
|
|
|
if (current->scopeDepth == 0) return;
|
|
|
|
KrkToken * name = &parser.previous;
|
|
|
|
/* Detect duplicate definition */
|
|
|
|
for (ssize_t i = current->localCount - 1; i >= 0; i--) {
|
|
|
|
Local * local = ¤t->locals[i];
|
2020-12-28 05:11:50 +03:00
|
|
|
if (local->depth != -1 && local->depth < (ssize_t)current->scopeDepth) break;
|
2020-12-26 12:39:29 +03:00
|
|
|
if (identifiersEqual(name, &local->name)) error("Duplicate definition");
|
|
|
|
}
|
|
|
|
addLocal(*name);
|
|
|
|
}
|
|
|
|
|
|
|
|
static ssize_t parseVariable(const char * errorMessage) {
|
2020-12-26 10:53:15 +03:00
|
|
|
consume(TOKEN_IDENTIFIER, errorMessage);
|
2020-12-26 12:39:29 +03:00
|
|
|
|
|
|
|
declareVariable();
|
|
|
|
if (current->scopeDepth > 0) return 0;
|
|
|
|
|
2020-12-26 10:53:15 +03:00
|
|
|
return identifierConstant(&parser.previous);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void defineVariable(size_t global) {
|
2020-12-26 12:39:29 +03:00
|
|
|
if (current->scopeDepth > 0) {
|
|
|
|
markInitialized();
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2020-12-26 10:53:15 +03:00
|
|
|
EMIT_CONSTANT_OP(OP_DEFINE_GLOBAL, global);
|
|
|
|
}
|
|
|
|
|
2020-12-27 03:33:28 +03:00
|
|
|
static uint8_t argumentList() {
|
|
|
|
uint8_t argCount = 0;
|
|
|
|
if (!check(TOKEN_RIGHT_PAREN)) {
|
|
|
|
do {
|
|
|
|
expression();
|
|
|
|
if (argCount == 255) error("Too many arguments to function."); // Need long call...
|
|
|
|
argCount++;
|
|
|
|
} while (match(TOKEN_COMMA));
|
|
|
|
}
|
|
|
|
consume(TOKEN_RIGHT_PAREN, "Expected ')' after arguments.");
|
|
|
|
return argCount;
|
|
|
|
}
|
|
|
|
|
2020-12-26 14:39:47 +03:00
|
|
|
static void and_(int canAssign) {
|
|
|
|
int endJump = emitJump(OP_JUMP_IF_FALSE);
|
|
|
|
emitByte(OP_POP);
|
|
|
|
parsePrecedence(PREC_AND);
|
|
|
|
patchJump(endJump);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void or_(int canAssign) {
|
|
|
|
int endJump = emitJump(OP_JUMP_IF_TRUE);
|
|
|
|
emitByte(OP_POP);
|
|
|
|
parsePrecedence(PREC_OR);
|
|
|
|
patchJump(endJump);
|
|
|
|
}
|
|
|
|
|
2020-12-26 03:32:21 +03:00
|
|
|
static ParseRule * getRule(KrkTokenType type) {
|
|
|
|
return &rules[type];
|
|
|
|
}
|
|
|
|
|
2020-12-28 04:54:25 +03:00
|
|
|
KrkFunction * krk_compile(const char * src, int newScope, char * fileName) {
|
2020-12-26 03:32:21 +03:00
|
|
|
krk_initScanner(src);
|
2020-12-26 12:39:29 +03:00
|
|
|
Compiler compiler;
|
2020-12-27 03:33:28 +03:00
|
|
|
initCompiler(&compiler, TYPE_MODULE);
|
2020-12-28 05:11:50 +03:00
|
|
|
compiler.function->chunk.filename = krk_copyString(fileName, strlen(fileName));
|
2020-12-26 03:32:21 +03:00
|
|
|
|
2020-12-27 12:55:52 +03:00
|
|
|
if (newScope) beginScope();
|
|
|
|
|
2020-12-26 03:32:21 +03:00
|
|
|
parser.hadError = 0;
|
|
|
|
parser.panicMode = 0;
|
|
|
|
|
|
|
|
advance();
|
2020-12-26 10:53:15 +03:00
|
|
|
|
|
|
|
while (!match(TOKEN_EOF)) {
|
|
|
|
declaration();
|
2020-12-26 14:39:47 +03:00
|
|
|
if (check(TOKEN_EOL)) advance();
|
2020-12-26 10:53:15 +03:00
|
|
|
}
|
2020-12-26 03:32:21 +03:00
|
|
|
|
2020-12-27 03:33:28 +03:00
|
|
|
KrkFunction * function = endCompiler();
|
|
|
|
return parser.hadError ? NULL : function;
|
2020-12-26 03:32:21 +03:00
|
|
|
}
|
2020-12-27 09:58:32 +03:00
|
|
|
|
|
|
|
void krk_markCompilerRoots() {
|
|
|
|
Compiler * compiler = current;
|
|
|
|
while (compiler != NULL) {
|
|
|
|
krk_markObject((KrkObj*)compiler->function);
|
|
|
|
compiler = compiler->enclosing;
|
|
|
|
}
|
|
|
|
}
|