zongor
/
undar-lang
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

wip symbol table, remove malloc_imm, fix docs

This commit is contained in:
zongor 2025-11-29 08:26:12 -08:00
parent 4f73339efb
commit 9d2053aef0
17 changed files with 471 additions and 443 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
README.org
View File

@ -55,33 +55,37 @@ You can view some examples in the =.ul.ir= files in =/test=
**Sample Program: =hello.ul.ir=** **Sample Program: =hello.ul.ir=**
#+BEGIN_SRC sh #+BEGIN_SRC sh
function main () global str terminal_namespace = "/dev/term/0"
str hello is $0 global str new_line = "\n"
global str message = "nuqneH 'u'?"
malloc_immediate "nuqneH 'u'?" -> hello function main ()
call pln hello str hello $0
load_immediate message -> hello
call pln hello -> void
exit 0 exit 0
function pln (str message is $0) function pln (str message $0)
str ts is $1 str ts $1
int msg_length is $2 int mode $5
str nl is $3 int msg_length $2
int nl_length is $4 str nl $3
int mode is $5 int nl_length $4
malloc_immediate "/dev/term/0" -> ts # get terminal device load_immediate terminal_namespace -> ts
load_immediate 0 -> mode load_immediate 0 -> mode
syscall OPEN ts mode -> ts syscall OPEN ts mode -> ts
strlen message -> msg_length strlen message -> msg_length
syscall WRITE ts message msg_length syscall WRITE ts message msg_length
malloc_immediate "\n" -> nl load_immediate new_line -> nl
strlen nl -> nl_length strlen nl -> nl_length
syscall WRITE ts nl nl_length syscall WRITE ts nl nl_length
return return
#+END_SRC #+END_SRC
#+BEGIN_SRC sh #+BEGIN_SRC sh
./build/linux/undar-linux-debug ./test/hello.asm.lisp ./build/linux/undar-linux-debug ./test/hello.ul.ir
#+END_SRC #+END_SRC
Running the compiler without arguments will put it in "REPL" mode. It will function similar to a LISP repl. Running the compiler without arguments will put it in "REPL" mode. It will function similar to a LISP repl.

317
src/tools/assembler/assembler.c
View File

@ -15,8 +15,6 @@ void emit_u32(VM *vm, u32 value) {
vm->cp += 4; vm->cp += 4;
} }
void emit_opcode(VM *vm, Opcode op) { emit_byte(vm, op); }
SymbolTable *symbol_table_init() { SymbolTable *symbol_table_init() {
SymbolTable *table = malloc(sizeof(SymbolTable)); SymbolTable *table = malloc(sizeof(SymbolTable));
table->symbols = malloc(16 * sizeof(Symbol)); table->symbols = malloc(16 * sizeof(Symbol));
@ -56,8 +54,8 @@ u32 get_ref(VM *vm, SymbolTable *st, const char *name, ScopeType scope) {
return sym->ref; return sym->ref;
} }
Token nextTokenIs(TokenType type) { Token next_token_is(TokenType type) {
Token token = nextToken(); Token token = next_token();
if (token.type != type) { if (token.type != type) {
printf("ERROR at line %d: %.*s\n", token.line, token.length, token.start); printf("ERROR at line %d: %.*s\n", token.line, token.length, token.start);
exit(1); exit(1);
@ -71,7 +69,7 @@ Token nextTokenIs(TokenType type) {
bool define_global(VM *vm, SymbolTable *st) { bool define_global(VM *vm, SymbolTable *st) {
Symbol s; Symbol s;
Token token_type = nextToken(); Token token_type = next_token();
switch (token_type.type) { switch (token_type.type) {
case TOKEN_TYPE_BOOL: case TOKEN_TYPE_BOOL:
s.type = BOOL; s.type = BOOL;
@ -114,8 +112,8 @@ bool define_global(VM *vm, SymbolTable *st) {
return false; return false;
} }
Token eq = nextTokenIs(TOKEN_EQ); Token eq = next_token_is(TOKEN_EQ);
Token name = nextTokenIs(TOKEN_IDENTIFIER); Token name = next_token_is(TOKEN_IDENTIFIER);
if (name.length > MAX_SYMBOL_NAME_LENGTH) { if (name.length > MAX_SYMBOL_NAME_LENGTH) {
return false; return false;
@ -127,21 +125,21 @@ bool define_global(VM *vm, SymbolTable *st) {
s.ref = addr; s.ref = addr;
s.scope = GLOBAL; s.scope = GLOBAL;
Token value = nextToken(); Token value = next_token();
switch (value.type) { switch (value.type) {
case TOKEN_KEYWORD_TRUE: { case TOKEN_KEYWORD_TRUE: {
u32 addr = vm->mp; u32 addr = vm->mp;
write_u8(vm, memory, addr, 1); write_u8(vm, memory, addr, 1);
vm->mp += 1; vm->mp += s.size;
vm->frames[vm->fp].end += 1; vm->frames[vm->fp].end += s.size;
} }
case TOKEN_KEYWORD_FALSE: { case TOKEN_KEYWORD_FALSE: {
u32 addr = vm->mp; u32 addr = vm->mp;
write_u8(vm, memory, addr, 0); write_u8(vm, memory, addr, 0);
vm->mp += 1; vm->mp += s.size;
vm->frames[vm->fp].end += 1; vm->frames[vm->fp].end += s.size;
} }
case TOKEN_LITERAL_INT: { case TOKEN_LITERAL_INT: {
i32 out = atoi(value.start); i32 out = atoi(value.start);
@ -266,7 +264,7 @@ void define_var(VM *vm, SymbolTable *st, Token regType) {
break; break;
} }
case TOKEN_TYPE_REAL: { case TOKEN_TYPE_REAL: {
s.type = REAL; s.type = F32;
s.size = 4; s.size = 4;
break; break;
} }
@ -286,7 +284,7 @@ void define_var(VM *vm, SymbolTable *st, Token regType) {
exit(1); exit(1);
} }
Token name = nextTokenIs(TOKEN_IDENTIFIER); Token name = next_token_is(TOKEN_IDENTIFIER);
if (name.length > MAX_SYMBOL_NAME_LENGTH) { if (name.length > MAX_SYMBOL_NAME_LENGTH) {
printf("VARIABLE NAME TOO LONG at line %d: %.*s\n", regType.line, printf("VARIABLE NAME TOO LONG at line %d: %.*s\n", regType.line,
regType.length, regType.start); regType.length, regType.start);
@ -295,9 +293,9 @@ void define_var(VM *vm, SymbolTable *st, Token regType) {
memcpy(s.name, name.start, name.length); memcpy(s.name, name.start, name.length);
nextTokenIs(TOKEN_BIG_MONEY); next_token_is(TOKEN_BIG_MONEY);
Token reg_num = nextTokenIs(TOKEN_LITERAL_INT); Token reg_num = next_token_is(TOKEN_LITERAL_INT);
s.ref = atoi(reg_num.start); s.ref = atoi(reg_num.start);
symbol_table_add(st, s); symbol_table_add(st, s);
} }
@ -305,33 +303,33 @@ void define_var(VM *vm, SymbolTable *st, Token regType) {
/** /**
* function . * function .
*/ */
void define_function(vm *vm, SymbolTable *st) { void define_function(VM *vm, SymbolTable *st) {
Symbol s; Symbol s;
s.scope = LOCAL; s.scope = LOCAL;
s.type = FUNCTION; s.type = FUNCTION;
Token name = nextTokenIs(TOKEN_IDENTIFIER); Token name = next_token_is(TOKEN_IDENTIFIER);
if (name.length > MAX_SYMBOL_NAME_LENGTH) { if (name.length > MAX_SYMBOL_NAME_LENGTH) {
printf("FUNCITON NAME TOO LONG at line %d: %.*s\n", regType.line, printf("FUNCITON NAME TOO LONG at line %d: %.*s\n", name.line,
regType.length, regType.start); name.length, name.start);
exit(1); exit(1);
} }
memcpy(s.name, name.start, name.length); memcpy(s.name, name.start, name.length);
nextTokenIs(TOKEN_LPAREN); next_token_is(TOKEN_LPAREN);
Token next = nextToken(); Token next = next_token();
while (next.type != TOKEN_RPAREN) { while (next.type != TOKEN_RPAREN) {
Token regType = nextToken(); Token regType = next_token();
define_var(vm, st, regType); define_var(vm, st, regType);
Token comma = nextToken(); Token next = next_token();
if (comma.type == TOKEN_COMMA) { if (next.type == TOKEN_COMMA) {
continue; continue;
} else if (comma.type == TOKEN_RPAREN) { } else if (next.type == TOKEN_RPAREN) {
break; break;
} else { } else {
printf("ERROR at line %d: %.*s\n", comma.line, comma.length, comma.start); printf("ERROR at line %d: %.*s\n", next.line, next.length, next.start);
exit(1); exit(1);
} }
} }
@ -347,10 +345,10 @@ void define_branch(VM *vm, SymbolTable *st) {
s.scope = LOCAL; s.scope = LOCAL;
s.type = VOID; s.type = VOID;
token name = nextTokenIs(TOKEN_IDENTIFIER); Token name = next_token_is(TOKEN_IDENTIFIER);
if (name.length > MAX_SYMBOL_NAME_LENGTH) { if (name.length > MAX_SYMBOL_NAME_LENGTH) {
printf("BRANCH NAME TOO LONG at line %d: %.*s\n", regType.line, printf("BRANCH NAME TOO LONG at line %d: %.*s\n", name.line,
regType.length, regType.start); name.length, name.start);
exit(1); exit(1);
} }
memcpy(s.name, name.start, name.length); memcpy(s.name, name.start, name.length);
@ -364,136 +362,150 @@ void define_branch(VM *vm, SymbolTable *st) {
*/ */
void build_symbol_table(VM *vm, char *source, SymbolTable *st) { void build_symbol_table(VM *vm, char *source, SymbolTable *st) {
Token token; Token token;
initLexer(source); init_lexer(source);
do { do {
token = nextToken(); token = next_token();
if (token.type == TOKEN_ERROR) { if (token.type == TOKEN_ERROR) {
printf("ERROR at line %d: %.*s\n", token.line, token.length, token.start); printf("ERROR at line %d: %.*s\n", token.line, token.length, token.start);
exit(1); exit(1);
} }
if (token.type != TOKEN_EOF) {
printf("Line %d [%s]: %.*s\n", token.line, tokenTypeToString(token.type),
token.length, token.start);
if (token.type == TOKEN_KEYWORD_GLOBAL) { if (token.type == TOKEN_KEYWORD_GLOBAL) {
define_global(vm, st); define_global(vm, st);
continue; continue;
} }
if (token.type == TOKEN_KEYWORD_FN) { if (token.type == TOKEN_KEYWORD_FN) {
define_function(vm, st); define_function(vm, st);
continue; continue;
} }
if (token.type == TOKEN_KEYWORD_PLEX || token.type == TOKEN_TYPE_I8 || if (token.type == TOKEN_KEYWORD_PLEX || token.type == TOKEN_TYPE_I8 ||
token.type == TOKEN_TYPE_I16 || token.type == TOKEN_TYPE_INT || token.type == TOKEN_TYPE_I16 || token.type == TOKEN_TYPE_INT ||
token.type == TOKEN_TYPE_U8 || token.type == TOKEN_TYPE_U16 || token.type == TOKEN_TYPE_U8 || token.type == TOKEN_TYPE_U16 ||
token.type == TOKEN_TYPE_NAT || token.type == TOKEN_TYPE_REAL || token.type == TOKEN_TYPE_NAT || token.type == TOKEN_TYPE_REAL ||
token.type == TOKEN_TYPE_STR || token.type == TOKEN_TYPE_BOOL) { token.type == TOKEN_TYPE_STR || token.type == TOKEN_TYPE_BOOL) {
define_var(vm, st, token); define_var(vm, st, token);
continue; continue;
} }
if (token.type == TOKEN_KEYWORD_LOOP || token.type == TOKEN_KEYWORD_IF || if (token.type == TOKEN_KEYWORD_LOOP || token.type == TOKEN_KEYWORD_IF ||
token.type == TOKEN_KEYWORD_ELSE || token.type == TOKEN_KEYWORD_DO || token.type == TOKEN_KEYWORD_ELSE || token.type == TOKEN_KEYWORD_DO ||
token.type == TOKEN_KEYWORD_FOR) { token.type == TOKEN_KEYWORD_FOR) {
define_branch(vm, st); define_branch(vm, st);
continue; continue;
} }
if (token.type == TOKEN_IDENTIFIER) { if (token.type == TOKEN_IDENTIFIER) {
// check to see if it is an opcode first // check to see if it is an opcode first
if (streq(token.start, "exit")) { if (streq(token.start, "exit")) {
} else if (streq(token.start, "call")) { vm->pc++;
} else if (streq(token.start, "syscall")) {
} else if (streq(token.start, "load_immediate")) { next_token_is(TOKEN_LITERAL_NAT);
} else if (streq(token.start, "load_indirect_8")) { vm->pc+=4;
} else if (streq(token.start, "load_indirect_16")) { } else if (streq(token.start, "call")) {
} else if (streq(token.start, "load_indirect_32")) { vm->pc++;
} else if (streq(token.start, "load_absolute_8")) {
} else if (streq(token.start, "load_absolute_16")) { next_token_is(TOKEN_IDENTIFIER);
} else if (streq(token.start, "load_absolute_32")) { vm->pc+=4;
} else if (streq(token.start, "load_offset_8")) {
} else if (streq(token.start, "load_offset_16")) { vm->pc++; /* number of args (implied) */
} else if (streq(token.start, "load_offset_32")) {
} else if (streq(token.start, "store_absolute_8")) { Token next = next_token();
} else if (streq(token.start, "store_absolute_16")) { while (next.type != TOKEN_ARROW_LEFT) {
} else if (streq(token.start, "store_absolute_32")) { vm->pc++;
} else if (streq(token.start, "store_indirect_8")) { Token next = next_token();
} else if (streq(token.start, "store_indirect_16")) {
} else if (streq(token.start, "store_indirect_32")) {
} else if (streq(token.start, "store_offset_8")) {
} else if (streq(token.start, "store_offset_16")) {
} else if (streq(token.start, "store_offset_32")) {
} else if (streq(token.start, "malloc")) {
} else if (streq(token.start, "malloc_immediate")) {
} else if (streq(token.start, "memset_8")) {
} else if (streq(token.start, "memset_16")) {
} else if (streq(token.start, "memset_32")) {
} else if (streq(token.start, "register_move")) {
} else if (streq(token.start, "add_int")) {
} else if (streq(token.start, "sub_int")) {
} else if (streq(token.start, "mul_int")) {
} else if (streq(token.start, "div_int")) {
} else if (streq(token.start, "abs_int")) {
} else if (streq(token.start, "neg_int")) {
} else if (streq(token.start, "add_nat")) {
} else if (streq(token.start, "sub_nat")) {
} else if (streq(token.start, "mul_nat")) {
} else if (streq(token.start, "div_nat")) {
} else if (streq(token.start, "abs_nat")) {
} else if (streq(token.start, "neg_nat")) {
} else if (streq(token.start, "add_real")) {
} else if (streq(token.start, "sub_real")) {
} else if (streq(token.start, "mul_real")) {
} else if (streq(token.start, "div_real")) {
} else if (streq(token.start, "abs_real")) {
} else if (streq(token.start, "neg_real")) {
} else if (streq(token.start, "int_to_real")) {
} else if (streq(token.start, "nat_to_real")) {
} else if (streq(token.start, "real_to_int")) {
} else if (streq(token.start, "real_to_nat")) {
} else if (streq(token.start, "bit_shift_left")) {
} else if (streq(token.start, "bit_shift_right")) {
} else if (streq(token.start, "bit_shift_r_ext")) {
} else if (streq(token.start, "bit_and")) {
} else if (streq(token.start, "bit_or")) {
} else if (streq(token.start, "bit_xor")) {
} else if (streq(token.start, "jump")) {
} else if (streq(token.start, "jump_if_flag")) {
} else if (streq(token.start, "jump_eq_int")) {
} else if (streq(token.start, "jump_neq_int")) {
} else if (streq(token.start, "jump_gt_int")) {
} else if (streq(token.start, "jump_lt_int")) {
} else if (streq(token.start, "jump_le_int")) {
} else if (streq(token.start, "jump_ge_int")) {
} else if (streq(token.start, "jump_eq_nat")) {
} else if (streq(token.start, "jump_neq_nat")) {
} else if (streq(token.start, "jump_gt_nat")) {
} else if (streq(token.start, "jump_lt_nat")) {
} else if (streq(token.start, "jump_le_nat")) {
} else if (streq(token.start, "jump_ge_nat")) {
} else if (streq(token.start, "jump_eq_real")) {
} else if (streq(token.start, "jump_neq_real")) {
} else if (streq(token.start, "jump_ge_real")) {
} else if (streq(token.start, "jump_gt_real")) {
} else if (streq(token.start, "jump_lt_real")) {
} else if (streq(token.start, "jump_le_real")) {
} else if (streq(token.start, "string_length")) {
} else if (streq(token.start, "string_eq")) {
} else if (streq(token.start, "string_concat")) {
} else if (streq(token.start, "string_get_char")) {
} else if (streq(token.start, "string_find_char")) {
} else if (streq(token.start, "string_slice")) {
} else if (streq(token.start, "int_to_string")) {
} else if (streq(token.start, "nat_to_string")) {
} else if (streq(token.start, "real_to_string")) {
} else if (streq(token.start, "string_to_int")) {
} else if (streq(token.start, "string_to_nat")) {
} else if (streq(token.start, "string_to_real")) {
} else {
// some other identifier
} }
/* return type */
next = next_token();
vm->pc++;
} else if (streq(token.start, "syscall")) {
} else if (streq(token.start, "load_immediate")) {
} else if (streq(token.start, "load_indirect_8")) {
} else if (streq(token.start, "load_indirect_16")) {
} else if (streq(token.start, "load_indirect_32")) {
} else if (streq(token.start, "load_absolute_8")) {
} else if (streq(token.start, "load_absolute_16")) {
} else if (streq(token.start, "load_absolute_32")) {
} else if (streq(token.start, "load_offset_8")) {
} else if (streq(token.start, "load_offset_16")) {
} else if (streq(token.start, "load_offset_32")) {
} else if (streq(token.start, "store_absolute_8")) {
} else if (streq(token.start, "store_absolute_16")) {
} else if (streq(token.start, "store_absolute_32")) {
} else if (streq(token.start, "store_indirect_8")) {
} else if (streq(token.start, "store_indirect_16")) {
} else if (streq(token.start, "store_indirect_32")) {
} else if (streq(token.start, "store_offset_8")) {
} else if (streq(token.start, "store_offset_16")) {
} else if (streq(token.start, "store_offset_32")) {
} else if (streq(token.start, "malloc")) {
} else if (streq(token.start, "memset_8")) {
} else if (streq(token.start, "memset_16")) {
} else if (streq(token.start, "memset_32")) {
} else if (streq(token.start, "register_move")) {
} else if (streq(token.start, "add_int")) {
} else if (streq(token.start, "sub_int")) {
} else if (streq(token.start, "mul_int")) {
} else if (streq(token.start, "div_int")) {
} else if (streq(token.start, "abs_int")) {
} else if (streq(token.start, "neg_int")) {
} else if (streq(token.start, "add_nat")) {
} else if (streq(token.start, "sub_nat")) {
} else if (streq(token.start, "mul_nat")) {
} else if (streq(token.start, "div_nat")) {
} else if (streq(token.start, "abs_nat")) {
} else if (streq(token.start, "neg_nat")) {
} else if (streq(token.start, "add_real")) {
} else if (streq(token.start, "sub_real")) {
} else if (streq(token.start, "mul_real")) {
} else if (streq(token.start, "div_real")) {
} else if (streq(token.start, "abs_real")) {
} else if (streq(token.start, "neg_real")) {
} else if (streq(token.start, "int_to_real")) {
} else if (streq(token.start, "nat_to_real")) {
} else if (streq(token.start, "real_to_int")) {
} else if (streq(token.start, "real_to_nat")) {
} else if (streq(token.start, "bit_shift_left")) {
} else if (streq(token.start, "bit_shift_right")) {
} else if (streq(token.start, "bit_shift_r_ext")) {
} else if (streq(token.start, "bit_and")) {
} else if (streq(token.start, "bit_or")) {
} else if (streq(token.start, "bit_xor")) {
} else if (streq(token.start, "jump")) {
} else if (streq(token.start, "jump_if_flag")) {
} else if (streq(token.start, "jump_eq_int")) {
} else if (streq(token.start, "jump_neq_int")) {
} else if (streq(token.start, "jump_gt_int")) {
} else if (streq(token.start, "jump_lt_int")) {
} else if (streq(token.start, "jump_le_int")) {
} else if (streq(token.start, "jump_ge_int")) {
} else if (streq(token.start, "jump_eq_nat")) {
} else if (streq(token.start, "jump_neq_nat")) {
} else if (streq(token.start, "jump_gt_nat")) {
} else if (streq(token.start, "jump_lt_nat")) {
} else if (streq(token.start, "jump_le_nat")) {
} else if (streq(token.start, "jump_ge_nat")) {
} else if (streq(token.start, "jump_eq_real")) {
} else if (streq(token.start, "jump_neq_real")) {
} else if (streq(token.start, "jump_ge_real")) {
} else if (streq(token.start, "jump_gt_real")) {
} else if (streq(token.start, "jump_lt_real")) {
} else if (streq(token.start, "jump_le_real")) {
} else if (streq(token.start, "string_length")) {
} else if (streq(token.start, "string_eq")) {
} else if (streq(token.start, "string_concat")) {
} else if (streq(token.start, "string_get_char")) {
} else if (streq(token.start, "string_find_char")) {
} else if (streq(token.start, "string_slice")) {
} else if (streq(token.start, "int_to_string")) {
} else if (streq(token.start, "nat_to_string")) {
} else if (streq(token.start, "real_to_string")) {
} else if (streq(token.start, "string_to_int")) {
} else if (streq(token.start, "string_to_nat")) {
} else if (streq(token.start, "string_to_real")) {
} else {
// some other identifier
} }
} }
} while (token.type != TOKEN_EOF); } while (token.type != TOKEN_EOF);
@ -504,15 +516,15 @@ void build_symbol_table(VM *vm, char *source, SymbolTable *st) {
*/ */
void emit_bytecode(VM *vm, char *source, SymbolTable *st) { void emit_bytecode(VM *vm, char *source, SymbolTable *st) {
Token token; Token token;
initLexer(source); init_lexer(source);
do { do {
token = nextToken(); token = next_token();
if (token.type == TOKEN_ERROR) { if (token.type == TOKEN_ERROR) {
printf("ERROR at line %d: %.*s\n", token.line, token.length, token.start); printf("ERROR at line %d: %.*s\n", token.line, token.length, token.start);
break; break;
} }
if (token.type != TOKEN_EOF) { if (token.type != TOKEN_EOF) {
printf("Line %d [%s]: %.*s\n", token.line, tokenTypeToString(token.type), printf("Line %d [%s]: %.*s\n", token.line, token_type_to_string(token.type),
token.length, token.start); token.length, token.start);
if (token.type == TOKEN_KEYWORD_GLOBAL) { if (token.type == TOKEN_KEYWORD_GLOBAL) {
@ -561,7 +573,6 @@ void emit_bytecode(VM *vm, char *source, SymbolTable *st) {
} else if (streq(token.start, "store_offset_16")) { } else if (streq(token.start, "store_offset_16")) {
} else if (streq(token.start, "store_offset_32")) { } else if (streq(token.start, "store_offset_32")) {
} else if (streq(token.start, "malloc")) { } else if (streq(token.start, "malloc")) {
} else if (streq(token.start, "malloc_immediate")) {
} else if (streq(token.start, "memset_8")) { } else if (streq(token.start, "memset_8")) {
} else if (streq(token.start, "memset_16")) { } else if (streq(token.start, "memset_16")) {
} else if (streq(token.start, "memset_32")) { } else if (streq(token.start, "memset_32")) {

194
src/tools/assembler/lexer.c
View File

@ -6,24 +6,24 @@
typedef struct { typedef struct {
const char *start; const char *start;
const char *current; const char *current;
int line; i32 line;
} Lexer; } Lexer;
Lexer lexer; Lexer lexer;
void initLexer(const char *source) { void init_lexer(const char *source) {
lexer.start = source; lexer.start = source;
lexer.current = source; lexer.current = source;
lexer.line = 1; lexer.line = 1;
} }
static bool isAlpha(char c) { static bool is_alpha(char c) {
return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || c == '_'; return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || c == '_';
} }
static bool isDigit(char c) { return c >= '0' && c <= '9'; } static bool is_digit(char c) { return c >= '0' && c <= '9'; }
static bool isAtEnd() { return *lexer.current == '\0'; } static bool is_at_end() { return *lexer.current == '\0'; }
static char advance() { static char advance() {
lexer.current++; lexer.current++;
@ -32,14 +32,14 @@ static char advance() {
static char peek() { return *lexer.current; } static char peek() { return *lexer.current; }
static char peekNext() { static char peek_next() {
if (isAtEnd()) if (is_at_end())
return '\0'; return '\0';
return lexer.current[1]; return lexer.current[1];
} }
static bool match(char expected) { static bool match(char expected) {
if (isAtEnd()) if (is_at_end())
return false; return false;
if (*lexer.current != expected) if (*lexer.current != expected)
return false; return false;
@ -47,25 +47,25 @@ static bool match(char expected) {
return true; return true;
} }
static Token makeToken(TokenType type) { static Token make_token(TokenType type) {
Token token; Token token;
token.type = type; token.type = type;
token.start = lexer.start; token.start = lexer.start;
token.length = (int)(lexer.current - lexer.start); token.length = (i32)(lexer.current - lexer.start);
token.line = lexer.line; token.line = lexer.line;
return token; return token;
} }
static Token errorToken(const char *message) { static Token error_token(const char *message) {
Token token; Token token;
token.type = TOKEN_ERROR; token.type = TOKEN_ERROR;
token.start = message; token.start = message;
token.length = (int)strlen(message); token.length = (i32)strlen(message);
token.line = lexer.line; token.line = lexer.line;
return token; return token;
} }
static void skipWhitespace() { static void skip_whitespace() {
for (;;) { for (;;) {
char c = peek(); char c = peek();
switch (c) { switch (c) {
@ -79,19 +79,19 @@ static void skipWhitespace() {
advance(); advance();
break; break;
case '/': case '/':
if (peekNext() == '/') { if (peek_next() == '/') {
// Single-line comment: skip until newline or end of file // Single-line comment: skip until newline or end of file
advance(); advance();
while (peek() != '\n' && !isAtEnd()) while (peek() != '\n' && !is_at_end())
advance(); advance();
} else if (peekNext() == '*') { } else if (peek_next() == '*') {
// Multi-line comment: skip until '*/' or end of file // Multi-line comment: skip until '*/' or end of file
advance(); advance();
advance(); advance();
while (!isAtEnd()) { while (!is_at_end()) {
if (peek() == '\n') if (peek() == '\n')
lexer.line++; lexer.line++;
if (peek() == '*' && peekNext() == '/') { if (peek() == '*' && peek_next() == '/') {
advance(); advance();
advance(); advance();
break; // Exit loop, comment ended break; // Exit loop, comment ended
@ -108,7 +108,7 @@ static void skipWhitespace() {
} }
} }
static TokenType checkKeyword(int start, int length, const char *rest, static TokenType check_keyword(i32 start, i32 length, const char *rest,
TokenType type) { TokenType type) {
if (lexer.current - lexer.start == start + length && if (lexer.current - lexer.start == start + length &&
memcmp(lexer.start + start, rest, length) == 0) { memcmp(lexer.start + start, rest, length) == 0) {
@ -124,9 +124,9 @@ static TokenType identifierType() {
if (lexer.current - lexer.start > 1) { if (lexer.current - lexer.start > 1) {
switch (lexer.start[1]) { switch (lexer.start[1]) {
case 'n': case 'n':
return checkKeyword(2, 1, "d", TOKEN_OPERATOR_AND); return check_keyword(2, 1, "d", TOKEN_OPERATOR_AND);
case 's': case 's':
return checkKeyword(2, 0, "", TOKEN_KEYWORD_AS); return check_keyword(2, 0, "", TOKEN_KEYWORD_AS);
} }
} }
break; break;
@ -134,47 +134,47 @@ static TokenType identifierType() {
if (lexer.current - lexer.start > 1) { if (lexer.current - lexer.start > 1) {
switch (lexer.start[1]) { switch (lexer.start[1]) {
case 'l': case 'l':
return checkKeyword(2, 3, "ose", TOKEN_KEYWORD_CLOSE); return check_keyword(2, 3, "ose", TOKEN_KEYWORD_CLOSE);
case 'o': case 'o':
return checkKeyword(2, 3, "nst", TOKEN_KEYWORD_CONST); return check_keyword(2, 3, "nst", TOKEN_KEYWORD_CONST);
} }
} }
break; break;
case 'e': case 'e':
return checkKeyword(1, 3, "lse", TOKEN_KEYWORD_ELSE); return check_keyword(1, 3, "lse", TOKEN_KEYWORD_ELSE);
case 'f': case 'f':
if (lexer.current - lexer.start > 1) { if (lexer.current - lexer.start > 1) {
switch (lexer.start[1]) { switch (lexer.start[1]) {
case 'a': case 'a':
return checkKeyword(2, 3, "lse", TOKEN_KEYWORD_FALSE); return check_keyword(2, 3, "lse", TOKEN_KEYWORD_FALSE);
case 'o': case 'o':
return checkKeyword(2, 1, "r", TOKEN_KEYWORD_FOR); return check_keyword(2, 1, "r", TOKEN_KEYWORD_FOR);
case '3': case '3':
return checkKeyword(1, 1, "2", TOKEN_TYPE_REAL); return check_keyword(1, 1, "2", TOKEN_TYPE_REAL);
} }
return checkKeyword(1, 7, "unction", TOKEN_KEYWORD_FN); return check_keyword(1, 7, "unction", TOKEN_KEYWORD_FN);
} }
break; break;
case 'i': case 'i':
if (lexer.current - lexer.start > 1) { if (lexer.current - lexer.start > 1) {
switch (lexer.start[1]) { switch (lexer.start[1]) {
case 'f': case 'f':
return checkKeyword(2, 0, "", TOKEN_KEYWORD_IF); return check_keyword(2, 0, "", TOKEN_KEYWORD_IF);
case 's': case 's':
return checkKeyword(2, 0, "", TOKEN_KEYWORD_IS); return check_keyword(2, 0, "", TOKEN_KEYWORD_IS);
case '8': case '8':
return checkKeyword(2, 0, "", TOKEN_TYPE_I8); return check_keyword(2, 0, "", TOKEN_TYPE_I8);
case '1': case '1':
return checkKeyword(2, 1, "6", TOKEN_TYPE_I16); return check_keyword(2, 1, "6", TOKEN_TYPE_I16);
case '3': case '3':
return checkKeyword(2, 1, "2", TOKEN_TYPE_INT); return check_keyword(2, 1, "2", TOKEN_TYPE_INT);
case 'n': case 'n':
if (lexer.current - lexer.start > 2) { if (lexer.current - lexer.start > 2) {
switch (lexer.start[2]) { switch (lexer.start[2]) {
case 'i': case 'i':
return checkKeyword(3, 2, "t", TOKEN_KEYWORD_INIT); return check_keyword(3, 2, "t", TOKEN_KEYWORD_INIT);
case 't': case 't':
return checkKeyword(3, 0, "", TOKEN_TYPE_INT); return check_keyword(3, 0, "", TOKEN_TYPE_INT);
} }
} }
break; break;
@ -185,9 +185,9 @@ static TokenType identifierType() {
if (lexer.current - lexer.start > 1) { if (lexer.current - lexer.start > 1) {
switch (lexer.start[1]) { switch (lexer.start[1]) {
case 'a': case 'a':
return checkKeyword(2, 1, "t", TOKEN_TYPE_NAT); return check_keyword(2, 1, "t", TOKEN_TYPE_NAT);
case 'i': case 'i':
return checkKeyword(2, 1, "l", TOKEN_KEYWORD_NIL); return check_keyword(2, 1, "l", TOKEN_KEYWORD_NIL);
} }
} }
break; break;
@ -195,9 +195,9 @@ static TokenType identifierType() {
if (lexer.current - lexer.start > 1) { if (lexer.current - lexer.start > 1) {
switch (lexer.start[1]) { switch (lexer.start[1]) {
case 'p': case 'p':
return checkKeyword(2, 2, "en", TOKEN_KEYWORD_OPEN); return check_keyword(2, 2, "en", TOKEN_KEYWORD_OPEN);
case 'r': case 'r':
return checkKeyword(2, 0, "", TOKEN_OPERATOR_OR); return check_keyword(2, 0, "", TOKEN_OPERATOR_OR);
} }
} }
break; break;
@ -205,7 +205,7 @@ static TokenType identifierType() {
if (lexer.current - lexer.start > 1) { if (lexer.current - lexer.start > 1) {
switch (lexer.start[1]) { switch (lexer.start[1]) {
case 'l': case 'l':
return checkKeyword(2, 2, "ex", TOKEN_KEYWORD_PLEX); return check_keyword(2, 2, "ex", TOKEN_KEYWORD_PLEX);
} }
} }
break; break;
@ -216,11 +216,11 @@ static TokenType identifierType() {
if (lexer.current - lexer.start > 2) { if (lexer.current - lexer.start > 2) {
switch (lexer.start[2]) { switch (lexer.start[2]) {
case 'a': case 'a':
return checkKeyword(3, 1, "d", TOKEN_KEYWORD_READ); return check_keyword(3, 1, "d", TOKEN_KEYWORD_READ);
case 'f': case 'f':
return checkKeyword(3, 4, "resh", TOKEN_KEYWORD_REFRESH); return check_keyword(3, 4, "resh", TOKEN_KEYWORD_REFRESH);
case 't': case 't':
return checkKeyword(3, 3, "urn", TOKEN_KEYWORD_RETURN); return check_keyword(3, 3, "urn", TOKEN_KEYWORD_RETURN);
} }
} }
break; break;
@ -231,7 +231,7 @@ static TokenType identifierType() {
if (lexer.current - lexer.start > 1) { if (lexer.current - lexer.start > 1) {
switch (lexer.start[1]) { switch (lexer.start[1]) {
case 't': case 't':
return checkKeyword(2, 1, "r", TOKEN_TYPE_STR); return check_keyword(2, 1, "r", TOKEN_TYPE_STR);
} }
} }
break; break;
@ -239,9 +239,9 @@ static TokenType identifierType() {
if (lexer.current - lexer.start > 1) { if (lexer.current - lexer.start > 1) {
switch (lexer.start[1]) { switch (lexer.start[1]) {
case 'h': case 'h':
return checkKeyword(2, 2, "is", TOKEN_KEYWORD_THIS); return check_keyword(2, 2, "is", TOKEN_KEYWORD_THIS);
case 'r': case 'r':
return checkKeyword(2, 2, "ue", TOKEN_KEYWORD_TRUE); return check_keyword(2, 2, "ue", TOKEN_KEYWORD_TRUE);
} }
} }
break; break;
@ -249,13 +249,13 @@ static TokenType identifierType() {
if (lexer.current - lexer.start > 1) { if (lexer.current - lexer.start > 1) {
switch (lexer.start[1]) { switch (lexer.start[1]) {
case 's': case 's':
return checkKeyword(2, 1, "e", TOKEN_KEYWORD_USE); return check_keyword(2, 1, "e", TOKEN_KEYWORD_USE);
case '8': case '8':
return checkKeyword(2, 0, "", TOKEN_TYPE_U8); return check_keyword(2, 0, "", TOKEN_TYPE_U8);
case '1': case '1':
return checkKeyword(2, 1, "6", TOKEN_TYPE_U16); return check_keyword(2, 1, "6", TOKEN_TYPE_U16);
case '3': case '3':
return checkKeyword(2, 1, "2", TOKEN_TYPE_NAT); return check_keyword(2, 1, "2", TOKEN_TYPE_NAT);
} }
} }
break; break;
@ -263,129 +263,133 @@ static TokenType identifierType() {
if (lexer.current - lexer.start > 1) { if (lexer.current - lexer.start > 1) {
switch (lexer.start[1]) { switch (lexer.start[1]) {
case 'h': case 'h':
return checkKeyword(2, 3, "ile", TOKEN_KEYWORD_WHILE); return check_keyword(2, 3, "ile", TOKEN_KEYWORD_WHILE);
case 'r': case 'r':
return checkKeyword(2, 3, "ite", TOKEN_KEYWORD_WRITE); return check_keyword(2, 3, "ite", TOKEN_KEYWORD_WRITE);
} }
} }
break; break;
case 'g': case 'g':
return checkKeyword(1, 5, "lobal", TOKEN_KEYWORD_GLOBAL); return check_keyword(1, 5, "lobal", TOKEN_KEYWORD_GLOBAL);
case 'l': case 'l':
return checkKeyword(1, 4, "oop", TOKEN_KEYWORD_LOOP); return check_keyword(1, 4, "oop", TOKEN_KEYWORD_LOOP);
case 'd':
return check_keyword(1, 1, "o", TOKEN_KEYWORD_DO);
case 'v':
return check_keyword(1, 3, "oid", TOKEN_TYPE_VOID);
} }
return TOKEN_IDENTIFIER; return TOKEN_IDENTIFIER;
} }
static Token identifier() { static Token identifier() {
while (isAlpha(peek()) || isDigit(peek())) while (is_alpha(peek()) || is_digit(peek()))
advance(); advance();
return makeToken(identifierType()); return make_token(identifierType());
} }
static Token number() { static Token number() {
while (isDigit(peek())) while (is_digit(peek()))
advance(); advance();
/* Look for a fractional part. */ /* Look for a fractional part. */
if (peek() == '.' && isDigit(peekNext())) { if (peek() == '.' && is_digit(peek_next())) {
/* Consume the ".". */ /* Consume the ".". */
advance(); advance();
while (isDigit(peek())) while (is_digit(peek()))
advance(); advance();
return makeToken(TOKEN_LITERAL_REAL); return make_token(TOKEN_LITERAL_REAL);
} }
return makeToken(TOKEN_LITERAL_INT); return make_token(TOKEN_LITERAL_INT);
} }
static Token string() { static Token string() {
while (peek() != '"' && !isAtEnd()) { while (peek() != '"' && !is_at_end()) {
if (peek() == '\n') if (peek() == '\n')
lexer.line++; lexer.line++;
advance(); advance();
} }
if (isAtEnd()) if (is_at_end())
return errorToken("Unterminated string."); return error_token("Unterminated string.");
/* The closing quote. */ /* The closing quote. */
advance(); advance();
return makeToken(TOKEN_LITERAL_STR); return make_token(TOKEN_LITERAL_STR);
} }
Token nextToken() { Token next_token() {
skipWhitespace(); skip_whitespace();
lexer.start = lexer.current; lexer.start = lexer.current;
if (isAtEnd()) if (is_at_end())
return makeToken(TOKEN_EOF); return make_token(TOKEN_EOF);
char c = advance(); char c = advance();
if (isAlpha(c)) if (is_alpha(c))
return identifier(); return identifier();
if (isDigit(c)) if (is_digit(c))
return number(); return number();
switch (c) { switch (c) {
case '(': case '(':
return makeToken(TOKEN_LPAREN); return make_token(TOKEN_LPAREN);
case ')': case ')':
return makeToken(TOKEN_RPAREN); return make_token(TOKEN_RPAREN);
case '{': case '{':
return makeToken(TOKEN_LBRACE); return make_token(TOKEN_LBRACE);
case '}': case '}':
return makeToken(TOKEN_RBRACE); return make_token(TOKEN_RBRACE);
case '[': case '[':
return makeToken(TOKEN_LBRACKET); return make_token(TOKEN_LBRACKET);
case ']': case ']':
return makeToken(TOKEN_RBRACKET); return make_token(TOKEN_RBRACKET);
case ';': case ';':
return makeToken(TOKEN_SEMICOLON); return make_token(TOKEN_SEMICOLON);
case ',': case ',':
return makeToken(TOKEN_COMMA); return make_token(TOKEN_COMMA);
case '.': case '.':
return makeToken(TOKEN_DOT); return make_token(TOKEN_DOT);
case '-': case '-':
return makeToken(match('>') ? TOKEN_ARROW_LEFT : TOKEN_MINUS); return make_token(match('>') ? TOKEN_ARROW_LEFT : TOKEN_MINUS);
case '+': case '+':
return makeToken(TOKEN_PLUS); return make_token(TOKEN_PLUS);
case '/': case '/':
return makeToken(TOKEN_SLASH); return make_token(TOKEN_SLASH);
case '&': case '&':
return makeToken(match('&') ? TOKEN_AND_AND : TOKEN_AND); return make_token(match('&') ? TOKEN_AND_AND : TOKEN_AND);
case '#': case '#':
return makeToken(TOKEN_MESH); return make_token(TOKEN_MESH);
case '$': case '$':
return makeToken(TOKEN_BIG_MONEY); return make_token(TOKEN_BIG_MONEY);
case '*': case '*':
return makeToken(TOKEN_STAR); return make_token(TOKEN_STAR);
case '!': case '!':
return makeToken(match('=') ? TOKEN_BANG_EQ : TOKEN_BANG); return make_token(match('=') ? TOKEN_BANG_EQ : TOKEN_BANG);
case '=': case '=':
return makeToken(match('=') ? TOKEN_EQ_EQ : TOKEN_EQ); return make_token(match('=') ? TOKEN_EQ_EQ : TOKEN_EQ);
case '<': case '<':
return makeToken(match('=') ? TOKEN_LTE : TOKEN_LT); return make_token(match('=') ? TOKEN_LTE : TOKEN_LT);
case '>': case '>':
return makeToken(match('=') ? TOKEN_GTE : TOKEN_GT); return make_token(match('=') ? TOKEN_GTE : TOKEN_GT);
case '"': case '"':
return string(); return string();
} }
return errorToken("Unexpected character."); return error_token("Unexpected character.");
} }
const char *tokenTypeToString(TokenType type) { const char *token_type_to_string(TokenType type) {
switch (type) { switch (type) {
case TOKEN_EOF: case TOKEN_EOF:
return "EOF"; return "EOF";
case TOKEN_IDENTIFIER: case TOKEN_IDENTIFIER:
return "IDENTIFIER"; return "IDENTIFIER";
case TOKEN_LITERAL_INT: case TOKEN_LITERAL_INT:
return "LITERAL_INT"; return "LITERAL_i32";
case TOKEN_LITERAL_NAT: case TOKEN_LITERAL_NAT:
return "LITERAL_NAT"; return "LITERAL_NAT";
case TOKEN_LITERAL_REAL: case TOKEN_LITERAL_REAL:
@ -393,7 +397,7 @@ const char *tokenTypeToString(TokenType type) {
case TOKEN_LITERAL_STR: case TOKEN_LITERAL_STR:
return "LITERAL_STR"; return "LITERAL_STR";
case TOKEN_TYPE_INT: case TOKEN_TYPE_INT:
return "TYPE_INT"; return "TYPE_i32";
case TOKEN_TYPE_NAT: case TOKEN_TYPE_NAT:
return "TYPE_NAT"; return "TYPE_NAT";
case TOKEN_TYPE_REAL: case TOKEN_TYPE_REAL:

9
src/tools/assembler/lexer.h
View File

@ -16,6 +16,8 @@ typedef enum {
TOKEN_TYPE_NAT, TOKEN_TYPE_NAT,
TOKEN_TYPE_REAL, TOKEN_TYPE_REAL,
TOKEN_TYPE_STR, TOKEN_TYPE_STR,
TOKEN_TYPE_BOOL,
TOKEN_TYPE_VOID,
TOKEN_KEYWORD_PLEX, TOKEN_KEYWORD_PLEX,
TOKEN_KEYWORD_FN, TOKEN_KEYWORD_FN,
TOKEN_KEYWORD_CONST, TOKEN_KEYWORD_CONST,
@ -36,6 +38,7 @@ typedef enum {
TOKEN_KEYWORD_REFRESH, TOKEN_KEYWORD_REFRESH,
TOKEN_KEYWORD_CLOSE, TOKEN_KEYWORD_CLOSE,
TOKEN_KEYWORD_LOOP, TOKEN_KEYWORD_LOOP,
TOKEN_KEYWORD_DO,
TOKEN_KEYWORD_NIL, TOKEN_KEYWORD_NIL,
TOKEN_KEYWORD_TRUE, TOKEN_KEYWORD_TRUE,
TOKEN_KEYWORD_FALSE, TOKEN_KEYWORD_FALSE,
@ -79,8 +82,8 @@ typedef struct {
int line; int line;
} Token; } Token;
void initLexer(const char *source); void init_lexer(const char *source);
Token nextToken(); Token next_token();
const char* tokenTypeToString(TokenType type); const char* token_type_to_string(TokenType type);
#endif #endif

12
src/tools/compiler/compiler.c
View File

@ -184,14 +184,14 @@ Symbol *global(VM *vm) {
s.ref.global = vm->mp; s.ref.global = vm->mp;
Token token_type = nextToken(); Token token_type = next_token();
Token array_or_eq = nextToken(); Token array_or_eq = next_token();
if (array_or_eq.type == TOKEN_LBRACKET) { if (array_or_eq.type == TOKEN_LBRACKET) {
Token rb = nextToken(); Token rb = next_token();
if (rb.type != TOKEN_RBRACKET) if (rb.type != TOKEN_RBRACKET)
return nil; return nil;
Token eq = nextToken(); Token eq = next_token();
if (eq.type != TOKEN_EQ) if (eq.type != TOKEN_EQ)
return nil; return nil;
@ -268,7 +268,7 @@ Symbol *global(VM *vm) {
s.type = t; s.type = t;
Token value = nextToken(); Token value = next_token();
return nil; return nil;
} }
@ -345,7 +345,7 @@ void advance() {
parser.previous = parser.current; parser.previous = parser.current;
for (;;) { for (;;) {
parser.current = nextToken(); parser.current = next_token();
if (parser.current.type != TOKEN_ERROR) if (parser.current.type != TOKEN_ERROR)
break; break;

4
src/tools/compiler/lexer.c
View File

@ -315,7 +315,7 @@ static Token string() {
return makeToken(TOKEN_LITERAL_STR); return makeToken(TOKEN_LITERAL_STR);
} }
Token nextToken() { Token next_token() {
skipWhitespace(); skipWhitespace();
lexer.start = lexer.current; lexer.start = lexer.current;
@ -376,7 +376,7 @@ Token nextToken() {
return errorToken("Unexpected character."); return errorToken("Unexpected character.");
} }
const char *tokenTypeToString(TokenType type) { const char *token_type_to_string(TokenType type) {
switch (type) { switch (type) {
case TOKEN_EOF: case TOKEN_EOF:
return "EOF"; return "EOF";

4
src/tools/compiler/lexer.h
View File

@ -79,7 +79,7 @@ typedef struct {
} Token; } Token;
void initLexer(const char *source); void initLexer(const char *source);
Token nextToken(); Token next_token();
const char* tokenTypeToString(TokenType type); const char* token_type_to_string(TokenType type);
#endif #endif

1
src/vm/opcodes.h
View File

@ -28,7 +28,6 @@ typedef enum {
OP_STORE_OFF_16, /* store_offset_16 : memory[locals[dest] + offset] = locals[src1] && 0xFFFF */ OP_STORE_OFF_16, /* store_offset_16 : memory[locals[dest] + offset] = locals[src1] && 0xFFFF */
OP_STORE_OFF_32, /* store_offset_32 : memory[locals[dest] + offset] = locals[src1] */ OP_STORE_OFF_32, /* store_offset_32 : memory[locals[dest] + offset] = locals[src1] */
OP_MALLOC, /* malloc : dest = fat ptr to memory of ((src1 as size) + 4) */ OP_MALLOC, /* malloc : dest = fat ptr to memory of ((src1 as size) + 4) */
OP_MALLOC_IMM, /* malloc_immediate : dest = fat ptr to memory of raw <str or array or plex>*/
OP_MEMSET_8, /* memset_8 : dest <-> dest+count = src1 as u8 */ OP_MEMSET_8, /* memset_8 : dest <-> dest+count = src1 as u8 */
OP_MEMSET_16, /* memset_16 : dest <-> dest+count = src1 as u8 */ OP_MEMSET_16, /* memset_16 : dest <-> dest+count = src1 as u8 */
OP_MEMSET_32, /* memset_32 : dest <-> dest+count = src1 as u32 */ OP_MEMSET_32, /* memset_32 : dest <-> dest+count = src1 as u32 */

2
test/add.ul.ir
View File

@ -11,7 +11,7 @@ function main ()
load_absolute_32 y -> $1 load_absolute_32 y -> $1
call add $0 $1 -> ans call add $0 $1 -> ans
int_to_string ans -> ans_string int_to_string ans -> ans_string
call pln ans_string call pln ans_string -> void
exit 0 exit 0
function add (int a $0, int b $1) function add (int a $0, int b $1)

14
test/fib.ul.ir
View File

@ -6,23 +6,23 @@ function main ()
int str_n $1 int str_n $1
load_immediate 35 -> n load_immediate 35 -> n
call &fib n -> n call fib n -> n
int_to_string n -> str_n int_to_string n -> str_n
call &pln str_n call pln str_n -> void
exit 0 exit 0
function fib (int n $0) function fib (int n $0)
load_immediate 2 -> $1 load_immediate 2 -> $1
jump_lt_int &base_case n $1 jump_lt_int base_case n $1
load_immediate 2 -> $3 load_immediate 2 -> $3
sub_int n $3 -> $4 sub_int n $3 -> $4
call &fib $4 -> $5 call fib $4 -> $5
load_immediate 1 -> $3 load_immediate 1 -> $3
sub_int n $3 -> $4 sub_int n $3 -> $4
call &fib $4 -> $6 call fib $4 -> $6
add_int $6 $5 -> $7 add_int $6 $5 -> $7
return $7 return $7
@ -37,12 +37,12 @@ function pln (str message $0)
str nl $3 str nl $3
int nl_length $4 int nl_length $4
load_immediate &terminal_namespace -> ts load_immediate terminal_namespace -> ts
load_immediate 0 -> mode load_immediate 0 -> mode
syscall OPEN ts mode -> ts syscall OPEN ts mode -> ts
strlen message -> msg_length strlen message -> msg_length
syscall WRITE ts message msg_length syscall WRITE ts message msg_length
load_immediate &new_line -> nl load_immediate new_line -> nl
strlen nl -> nl_length strlen nl -> nl_length
syscall WRITE ts nl nl_length syscall WRITE ts nl nl_length
return return

9
test/hello.ul.ir
View File

@ -5,8 +5,8 @@ global str message = "nuqneH 'u'?"
function main () function main ()
str hello $0 str hello $0
load_immediate &message -> hello load_immediate message -> hello
call pln hello call pln hello -> void
exit 0 exit 0
function pln (str message $0) function pln (str message $0)
@ -16,13 +16,12 @@ function pln (str message $0)
str nl $3 str nl $3
int nl_length $4 int nl_length $4
load_immediate terminal_namespace -> ts
load_immediate &terminal_namespace -> ts
load_immediate 0 -> mode load_immediate 0 -> mode
syscall OPEN ts mode -> ts syscall OPEN ts mode -> ts
strlen message -> msg_length strlen message -> msg_length
syscall WRITE ts message msg_length syscall WRITE ts message msg_length
load_immediate &new_line -> nl load_immediate new_line -> nl
strlen nl -> nl_length strlen nl -> nl_length
syscall WRITE ts nl nl_length syscall WRITE ts nl nl_length
return return

51
test/loop.ul.ir
View File

@ -1,9 +1,12 @@
global str terminal_namespace = "/dev/term/0"
global str prompt = "Enter a string:"
global str new_line = "\n"
function main () function main ()
real a is $0 real a $0
int i is $1 int i $1
int mode is $11 int mode $11
str term is $10 str term $10
load_immediate 5.0 -> a load_immediate 5.0 -> a
load_immediate 5000 -> i load_immediate 5000 -> i
@ -15,41 +18,41 @@ function main ()
add_int i $3 -> i add_int i $3 -> i
jump_ge_int loop_body i $2 jump_ge_int loop_body i $2
malloc_immediate "/dev/term/0" -> term load_immediate terminal_namespace -> term
load_immediate 0 -> mode load_immediate 0 -> mode
syscall OPEN term mode -> term # Terminal term = open("/dev/term/0", 0); syscall OPEN term mode -> term // Terminal term = open("/dev/term/0", 0);
nat b is $1 nat b $1
real_to_nat a -> b real_to_nat a -> b
malloc_immediate "Enter a string:" -> $7 load_immediate prompt -> $7
string_length $7 -> $8 string_length $7 -> $8
syscall WRITE term $7 $8 # print prompt syscall WRITE term $7 $8 // print prompt
str user_string is $9 str user_string $9
load_immediate 32 -> $8 load_immediate 32 -> $8
malloc $8 -> user_string malloc $8 -> user_string
syscall READ term user_string $8 # read in max 32 byte string syscall READ term user_string $8 // read in max 32 byte string
call pln user_string call pln user_string -> void
nat_to_string b -> $4 nat_to_string b -> $4
call pln $4 call pln $4 -> void
real_to_string a -> $3 real_to_string a -> $3
call pln $3 call pln $3 -> void
exit 0 exit 0
function pln (str message is $0) function pln (str message $0)
str ts is $1 str ts $1
int mode is $5 int mode $5
int msg_length is $2 int msg_length $2
str nl is $3 str nl $3
int nl_length is $4 int nl_length $4
malloc_immediate "/dev/term/0" -> ts load_immediate terminal_namespace -> ts
load_immediate 0 -> mode load_immediate 0 -> mode
syscall OPEN ts mode -> ts # get terminal device syscall OPEN ts mode -> ts
strlen message -> msg_length strlen message -> msg_length
syscall WRITE ts message msg_length syscall WRITE ts message msg_length
malloc_immediate "\n" -> nl load_immediate new_line -> nl
strlen nl -> nl_length strlen nl -> nl_length
syscall WRITE ts nl nl_length syscall WRITE ts nl nl_length
return return

25
test/malloc.ul.ir
View File

@ -1,22 +1,25 @@
global str terminal_namespace = "/dev/term/0"
global str prompt = "Enter a string:"
global str new_line = "\n"
function main () function main ()
int mode $11; int mode $11;
str term $10; str term $10;
malloc_immediate "/dev/term/0" -> term load_immediate terminal_namespace -> term
load_immediate 0 -> mode load_immediate 0 -> mode
syscall OPEN term mode -> term # Terminal term = open("/dev/term/0", 0); syscall OPEN term mode -> term // Terminal term = open("/dev/term/0", 0);
malloc_immediate "Enter a string:" -> $7 load_immediate prompt -> $7
string_length $7 -> $8 string_length $7 -> $8
syscall WRITE term $7 $8 # print prompt syscall WRITE term $7 $8 // print prompt
str user_string $9 str user_string $9
load_immediate 32 -> $8 load_immediate 32 -> $8
malloc $8 -> user_string malloc $8 -> user_string
syscall READ term user_string $8 # read in max 32 byte string syscall READ term user_string $8 // read in max 32 byte string
call pln user_string; call pln user_string -> void
exit 0 exit 0
function pln (str message $0) function pln (str message $0)
@ -24,14 +27,14 @@ function pln (str message $0)
int mode $5 int mode $5
int msg_length $2 int msg_length $2
str nl $3 str nl $3
int nl_length $4 int nl_length $4
malloc_immediate "/dev/term/0" -> ts load_immediate terminal_namespace -> ts
load_immediate 0 -> mode load_immediate 0 -> mode
syscall OPEN ts mode -> ts # get terminal device syscall OPEN ts mode -> ts
strlen message -> msg_length strlen message -> msg_length
syscall WRITE ts message msg_length syscall WRITE ts message msg_length
malloc_immediate "\n" -> nl load_immediate new_line -> nl
strlen nl -> nl_length strlen nl -> nl_length
syscall WRITE ts nl nl_length syscall WRITE ts nl nl_length
return return

195
test/paint-bw.ul.ir
View File

@ -8,177 +8,178 @@ global const byte LIGHT_GRAY = 182
global byte SELECTED_COLOR = 255 global byte SELECTED_COLOR = 255
function main () function main ()
# Open screen // Open screen
plex screen is $0 plex screen $0
str screen_name is $18 str screen_name $18
int mode is $11 int mode $11
nat screen_buffer is $21 nat screen_buffer $21
# use load immediate because it is a pointer to a string, not a value // use load immediate because it a pointer to a string, not a value
load_address &screen_namespace -> screen_name load_address screen_namespace -> screen_name
load_immediate 0 -> mode load_immediate 0 -> mode
syscall OPEN screen_name mode -> screen # Screen screen = open("/dev/screen/0", 0); syscall OPEN screen_name mode -> screen // Screen screen = open("/dev/screen/0", 0);
nat width is $20 nat width $20
nat size is $22 nat size $22
load_offset_32 screen 8 -> width # load width load_offset_32 screen 8 -> width // load width
load_offset_32 screen 12 -> size # load size load_offset_32 screen 12 -> size // load size
load_immediate 16 -> $1 # offset for screen buffer load_immediate 16 -> $1 // offset for screen buffer
add_nat screen $1 -> screen_buffer add_nat screen $1 -> screen_buffer
# open mouse // open mouse
plex mouse is $15 plex mouse $15
str mouse_name is $16 str mouse_name $16
load_address &mouse_namespace -> mouse_name load_address mouse_namespace -> mouse_name
syscall OPEN mouse_name mode -> mouse # Mouse mouse = open("/dev/mouse/0", 0); syscall OPEN mouse_name mode -> mouse // Mouse mouse = open("/dev/mouse/0", 0);
byte color is $1 byte color $1
nat x_pos is $12 nat x_pos $12
nat y_pos is $13 nat y_pos $13
load_absolute_32 &BLACK -> color load_absolute_32 BLACK -> color
load_immediate 1 -> x_pos load_immediate 1 -> x_pos
load_immediate 1 -> y_pos load_immediate 1 -> y_pos
call &draw_outlined_swatch screen_buffer color x_pos y_pos width call draw_outlined_swatch screen_buffer color x_pos y_pos width -> void
load_absolute_32 &WHITE -> color load_absolute_32 WHITE -> color
load_immediate 21 -> x_pos load_immediate 21 -> x_pos
load_immediate 1 -> y_pos load_immediate 1 -> y_pos
call &draw_outlined_swatch screen_buffer color x_pos y_pos width call draw_outlined_swatch screen_buffer color x_pos y_pos width -> void
# screen.draw# // screen.draw
syscall WRITE screen screen_buffer size syscall WRITE screen screen_buffer size
nat zero is $11 nat zero $11
draw_loop: loop draw_loop
# load mouse click data // load mouse click data
syscall REFRESH mouse syscall REFRESH mouse
byte left_down is $9 byte left_down $9
load_offset_8 mouse 16 -> left_down # load btn1 pressed load_offset_8 mouse 16 -> left_down // load btn1 pressed
jump_eq_nat &draw_loop left_down zero jump_eq_nat draw_loop left_down zero
nat mouse_x is $7 nat mouse_x $7
nat mouse_y is $8 nat mouse_y $8
load_offset_32 mouse 8 -> mouse_x # load x load_offset_32 mouse 8 -> mouse_x // load x
load_offset_32 mouse 12 -> mouse_y # load y load_offset_32 mouse 12 -> mouse_y // load y
nat box_size is $14 nat box_size $14
load_immediate 20 -> box_size load_immediate 20 -> box_size
# first row // first row
load_absolute_32 &BLACK -> color load_absolute_32 BLACK -> color
load_immediate 1 -> x_pos load_immediate 1 -> x_pos
load_immediate 1 -> y_pos load_immediate 1 -> y_pos
call &draw_outlined_swatch screen_buffer color x_pos y_pos width call draw_outlined_swatch screen_buffer color x_pos y_pos width -> void
call &set_color_if_clicked mouse_x mouse_y x_pos y_pos color box_size call set_color_if_clicked mouse_x mouse_y x_pos y_pos color box_size -> void
load_absolute_32 &WHITE -> color load_absolute_32 WHITE -> color
load_immediate 21 -> x_pos load_immediate 21 -> x_pos
load_immediate 1 -> y_pos load_immediate 1 -> y_pos
call &draw_outlined_swatch screen_buffer color x_pos y_pos width call draw_outlined_swatch screen_buffer color x_pos y_pos width -> void
call &set_color_if_clicked mouse_x mouse_y x_pos y_pos color box_size call set_color_if_clicked mouse_x mouse_y x_pos y_pos color box_size -> void
syscall WRITE screen screen_buffer size syscall WRITE screen screen_buffer size
byte selected_color is $25 byte selected_color $25
load_absolute_32 &SELECTED_COLOR -> selected_color load_absolute_32 SELECTED_COLOR -> selected_color
nat brush_size is $19 nat brush_size $19
load_immediate 5 -> brush_size load_immediate 5 -> brush_size
call &draw_box screen_buffer width selected_color mouse_x mouse_y brush_size brush_size call draw_box screen_buffer width selected_color mouse_x mouse_y brush_size brush_size -> void
jump &draw_loop jump draw_loop
# Flush and exit // Flush and exit
exit 0 exit 0
function set_color_if_clicked (int click_x is $0, int click_y is $1, function set_color_if_clicked (int click_x $0, int click_y $1,
int box_x is $2, int box_y is $3, byte color is $4, int box_size is $5) int box_x $2, int box_y $3, byte color $4, int box_size $5)
# Compute right // Compute right
int right_edge is $6 int right_edge $6
add_int box_x box_size -> right_edge add_int box_x box_size -> right_edge
# Compute bottom = box_y + box_size // Compute bottom = box_y + box_size
int bottom_edge is $7 int bottom_edge $7
add_int box_y box_size -> bottom_edge add_int box_y box_size -> bottom_edge
# Bounds check: x in [box_x, right] and y in [box_y, bottom] // Bounds check: x in [box_x, right] and y in [box_y, bottom]
jump_lt_int &fail click_x box_x jump_lt_int fail click_x box_x
jump_ge_int &fail click_x right_edge jump_ge_int fail click_x right_edge
jump_lt_int &fail click_y box_y jump_lt_int fail click_y box_y
jump_ge_int &fail click_y bottom_edge jump_ge_int fail click_y bottom_edge
store_absolute_8 &SELECTED_COLOR color store_absolute_8 SELECTED_COLOR color
fail: else fail
return return
function draw_outlined_swatch(nat base is $0, function draw_outlined_swatch(nat base $0,
byte color is $1, int x is $2, int y is $3, int width is $4) byte color $1, int x $2, int y $3, int width $4)
# Constants // Constants
nat background_color is $5 nat background_color $5
load_absolute_32 &GRAY -> background_color load_absolute_32 GRAY -> background_color
byte selected_color is $10 byte selected_color $10
load_absolute_32 &SELECTED_COLOR -> selected_color load_absolute_32 SELECTED_COLOR -> selected_color
jump_eq_int &set_selected selected_color color jump_eq_int set_selected selected_color color
jump &end_set_selected jump end_set_selected
set_selected: set_selected:
load_absolute_32 &DARK_GRAY -> background_color load_absolute_32 DARK_GRAY -> background_color
end_set_selected: end_set_selected:
nat outline_size is $6 nat outline_size $6
load_immediate 20 -> outline_size load_immediate 20 -> outline_size
nat fill_size is $7 nat fill_size $7
load_immediate 17 -> fill_size load_immediate 17 -> fill_size
nat offset is $8 nat offset $8
load_immediate 2 -> offset load_immediate 2 -> offset
call &draw_box base width background_color x y outline_size outline_size call draw_box base width background_color x y outline_size outline_size -> void
add_int x offset -> $9 # x + 2 add_int x offset -> $9 // x + 2
add_int y offset -> $10 # y + 2 add_int y offset -> $10 // y + 2
call &draw_box base width color $9 $10 fill_size fill_size call draw_box base width color $9 $10 fill_size fill_size -> void
return return
function draw_box (nat base is $0, nat screen_width is $1, function draw_box (nat base $0, nat screen_width $1,
byte color is $2, nat x_start is $3, nat y_start is $4, nat width is $5, nat height is $6) byte color $2, nat x_start $3, nat y_start $4,
nat width $5, nat height $6)
# Compute start address: base + y*640 + x // Compute start address: base + y*640 + x
nat offset is $15 nat offset $15
mul_int y_start screen_width -> offset mul_int y_start screen_width -> offset
add_int offset x_start -> offset add_int offset x_start -> offset
add_nat offset base -> offset add_nat offset base -> offset
nat fat_ptr_size is $25 nat fat_ptr_size $25
load_immediate 4 -> fat_ptr_size load_immediate 4 -> fat_ptr_size
add_nat offset fat_ptr_size -> offset # need to add offset for fat pointer size add_nat offset fat_ptr_size -> offset // need to add offset for fat pointer size
int i is $30 int i $30
load_immediate 1 -> i load_immediate 1 -> i
int zero is $26 int zero $26
load_immediate 0 -> zero load_immediate 0 -> zero
int row_end is $27 int row_end $27
nat pixel_ptr is $29 nat pixel_ptr $29
draw_box_outer: loop draw_box_outer
add_int offset width -> row_end # current + width add_int offset width -> row_end // current + width
register_move offset -> pixel_ptr # set pixel point register_move offset -> pixel_ptr // set pixel point
memset_8 pixel_ptr color width # draw row memset_8 pixel_ptr color width // draw row
add_int offset screen_width -> offset # next row += 640 add_int offset screen_width -> offset // next row += 640
sub_int height i -> height # decrement row count sub_int height i -> height // decrement row count
jump_gt_int &draw_box_outer height zero jump_gt_int draw_box_outer height zero
return return

20
test/paint.ul.ir
View File

@ -39,12 +39,12 @@ function main ()
load_absolute_32 BLACK -> color load_absolute_32 BLACK -> color
load_immediate 1 -> x_pos load_immediate 1 -> x_pos
load_immediate 1 -> y_pos load_immediate 1 -> y_pos
call draw_outlined_swatch screen_buffer color x_pos y_pos width call draw_outlined_swatch [screen_buffer color x_pos y_pos width] -> void
load_absolute_32 WHITE -> color load_absolute_32 WHITE -> color
load_immediate 21 -> x_pos load_immediate 21 -> x_pos
load_immediate 1 -> y_pos load_immediate 1 -> y_pos
call draw_outlined_swatch screen_buffer color x_pos y_pos width call draw_outlined_swatch screen_buffer color x_pos y_pos width -> void
// screen.draw// // screen.draw//
syscall WRITE screen screen_buffer size syscall WRITE screen screen_buffer size
@ -72,15 +72,15 @@ function main ()
load_absolute_32 BLACK -> color load_absolute_32 BLACK -> color
load_immediate 1 -> x_pos load_immediate 1 -> x_pos
load_immediate 1 -> y_pos load_immediate 1 -> y_pos
call draw_outlined_swatch screen_buffer color x_pos y_pos width call draw_outlined_swatch screen_buffer color x_pos y_pos width -> void
call set_color_if_clicked mouse_x mouse_y x_pos y_pos color box_size call set_color_if_clicked mouse_x mouse_y x_pos y_pos color box_size -> void
load_absolute_32 WHITE -> color load_absolute_32 WHITE -> color
load_immediate 21 -> x_pos load_immediate 21 -> x_pos
load_immediate 1 -> y_pos load_immediate 1 -> y_pos
call draw_outlined_swatch screen_buffer color x_pos y_pos width call draw_outlined_swatch screen_buffer color x_pos y_pos width -> void
call set_color_if_clicked mouse_x mouse_y x_pos y_pos color box_size call set_color_if_clicked mouse_x mouse_y x_pos y_pos color box_size -> void
syscall WRITE screen screen_buffer size syscall WRITE screen screen_buffer size
@ -90,7 +90,7 @@ function main ()
nat brush_size $19 nat brush_size $19
load_immediate 5 -> brush_size load_immediate 5 -> brush_size
call draw_box screen_buffer width selected_color mouse_x mouse_y brush_size brush_size call draw_box screen_buffer width selected_color mouse_x mouse_y brush_size brush_size -> void
jump draw_loop jump draw_loop
@ -144,12 +144,12 @@ function draw_outlined_swatch(nat base $0,
nat offset $8 nat offset $8
load_immediate 2 -> offset load_immediate 2 -> offset
call draw_box base width background_color x y outline_size outline_size call draw_box base width background_color x y outline_size outline_size -> void
add_int x offset -> $9 // x + 2 add_int x offset -> $9 // x + 2
add_int y offset -> $10 // y + 2 add_int y offset -> $10 // y + 2
call draw_box base width color $9 $10 fill_size fill_size call draw_box base width color $9 $10 fill_size fill_size -> void
return return
@ -180,5 +180,5 @@ function draw_box (nat base $0, nat screen_width $1,
memset_8 pixel_ptr color width // draw row memset_8 pixel_ptr color width // draw row
add_int offset screen_width -> offset // next row += 640 add_int offset screen_width -> offset // next row += 640
sub_int height i -> height // decrement row count sub_int height i -> height // decrement row count
jump_gt_int draw_box_outer height zero jump_gt_int draw_box_outer height zero
return return

2
test/simple.ul.ir
View File

@ -11,7 +11,7 @@ function main ()
add_real x y -> result add_real x y -> result
str result_str $3 str result_str $3
real_to_string result -> result_str real_to_string result -> result_str
call &pln result_str call &pln result_str -> void
exit 0 exit 0
function pln (str message $0) function pln (str message $0)

25
test/window.ul.ir
View File

@ -24,21 +24,21 @@ function main ()
syscall OPEN screen mode -> screen syscall OPEN screen mode -> screen
nat_to_string screen -> tmp_str nat_to_string screen -> tmp_str
call pln tmp_str call pln tmp_str -> void
load_offset_32 screen 8 -> width load_offset_32 screen 8 -> width
nat_to_string width -> tmp_str nat_to_string width -> tmp_str
call pln tmp_str call pln tmp_str -> void
load_offset_32 screen 12 -> buffer_size load_offset_32 screen 12 -> buffer_size
nat_to_string buffer_size -> tmp_str nat_to_string buffer_size -> tmp_str
call pln tmp_str call pln tmp_str -> void
load_immediate 16 -> offset_temp load_immediate 16 -> offset_temp
add_nat screen offset_temp -> screen_buffer add_nat screen offset_temp -> screen_buffer
nat_to_string screen_buffer -> tmp_str nat_to_string screen_buffer -> tmp_str
call pln tmp_str call pln tmp_str -> void
// open mouse // open mouse
load_immediate mouse_namespace -> mouse load_immediate mouse_namespace -> mouse
@ -72,17 +72,18 @@ function main ()
exit 0 exit 0
function pln (str message $0) function pln (str message $0)
str term $1 str ts $1
int msg_length $2
str nl $3
int nl_length $4
int mode $5 int mode $5
int msg_length $2
str nl $3
int nl_length $4
load_immediate terminal_namespace -> ts
load_immediate 0 -> mode load_immediate 0 -> mode
syscall OPEN terminal_namespace mode -> term syscall OPEN ts mode -> ts
strlen message -> msg_length strlen message -> msg_length
syscall WRITE term message msg_length syscall WRITE ts message msg_length
load_address new_line -> nl load_immediate new_line -> nl
strlen nl -> nl_length strlen nl -> nl_length
syscall WRITE term nl nl_length syscall WRITE ts nl nl_length
return return