add more parses and fix minor memory leak

This commit is contained in:
2025-05-31 20:03:06 +01:00
parent d18ff96f8f
commit a614eab8fc
16 changed files with 188 additions and 50 deletions

View File

@@ -19,6 +19,13 @@ debug: $(CFILES) $(LEXER_C) $(LEXER_H)
mkdir -p bin
gcc -g -O0 -o $(BINARY) $(CFILES) $(CFLAGS)
optimised: $(CFILES) $(LEXER_C) $(LEXER_H)
mkdir -p bin
gcc -O3 -fprofile-generate -o $(BINARY) $(CFILES) $(CFLAGS)
${BINARY}
gcc -O3 -fprofile-use -o $(BINARY) $(CFILES) $(CFLAGS)
clean:
rm -rf bin
rm -f $(LEXER_C) $(LEXER_H)

View File

@@ -15,8 +15,7 @@ void darray_init(DArray *arr, size_t element_size) {
}
void darray_resize(DArray *arr, size_t new_size) {
size_t new_capacity = ((new_size + CHUNK_SIZE - 1) / CHUNK_SIZE) * CHUNK_SIZE;
size_t new_capacity = ((new_size + CHUNK_SIZE) / CHUNK_SIZE) * CHUNK_SIZE;
if (new_capacity != arr->capacity) {
void *new_data = realloc(arr->data, new_capacity * arr->element_size);
if (!new_data) {
@@ -37,7 +36,7 @@ void darray_push(DArray *arr, void *element) {
arr->size++;
}
void *target = (char *)arr->data + (arr->size - 1) * arr->element_size;
void *target = (void *)arr->data + (arr->size - 1) * arr->element_size;
memcpy(target, element, arr->element_size);
}
@@ -48,7 +47,7 @@ void darray_pop(DArray *arr, void (*free_data)(void *)) {
arr->size--;
if (free_data) {
void *target = (char *)arr->data + arr->size * arr->element_size;
void *target = (void *)arr->data + arr->size * arr->element_size;
free_data(target);
}
@@ -60,13 +59,13 @@ void *darray_get(DArray *arr, size_t index) {
fprintf(stderr, "darray_get: index out of bounds\n");
exit(EXIT_FAILURE);
}
return (char *)arr->data + index * arr->element_size;
return (void *)arr->data + index * arr->element_size;
}
void darray_free(DArray *arr, void (*free_data)(void *)) {
if (free_data) {
for (size_t i = 0; i < arr->size; ++i) {
void *element = (char *)arr->data + i * arr->element_size;
void *element = (void *)arr->data + i * arr->element_size;
free_data(element);
}
}

View File

@@ -18,6 +18,16 @@ int yywrap(void *) {
"," {return TOKEN_COMMA; }
":" {return TOKEN_COLON; }
"=" { return TOKEN_ASSIGN; }
"+=" { return TOKEN_ASSIGN_PLUS; }
"-=" { return TOKEN_ASSIGN_MINUS; }
"//=" { return TOKEN_ASSIGN_FLOORDIV; }
"/=" { return TOKEN_ASSIGN_SLASH; }
"%=" { return TOKEN_ASSIGN_MODULO; }
"*=" { return TOKEN_ASSIGN_STAR; }
"^=" { return TOKEN_ASSIGN_CARET; }
"not"[ \t]+"in" { return TOKEN_NOT_IN; }
"&&" { return TOKEN_AND; }
"||" { return TOKEN_OR; }
@@ -25,7 +35,6 @@ int yywrap(void *) {
">=" { return TOKEN_GE; }
"!=" { return TOKEN_NE; }
"==" { return TOKEN_EQ; }
"=" { return TOKEN_ASSIGN; }
"//" { return TOKEN_FLOORDIV; }
"<" { return TOKEN_LT; }
">" { return TOKEN_GT; }

View File

@@ -19,6 +19,7 @@ void lexer(LexerState state) {
yyget_text(scanner)
);
darray_push(state.tokens, token_struct);
free(token_struct);
if (token == TOKEN_NEW_LINE) {
state.current_column = 0;
} else {

View File

@@ -1,3 +1,6 @@
#ifndef LEXER_H
#define LEXER_H
#include "token.h"
#include "../dynamic_array/darray.h"
#include <stdio.h>
@@ -11,3 +14,6 @@ typedef struct {
} LexerState;
void lexer(LexerState state);
#endif // LEXER_H

View File

@@ -10,6 +10,15 @@ typedef enum {
TOKEN_NEW_LINE,
TOKEN_INDENT,
TOKEN_ASSIGN,
TOKEN_ASSIGN_PLUS,
TOKEN_ASSIGN_MINUS,
TOKEN_ASSIGN_FLOORDIV,
TOKEN_ASSIGN_SLASH,
TOKEN_ASSIGN_MODULO,
TOKEN_ASSIGN_STAR,
TOKEN_ASSIGN_CARET,
// Operators
TOKEN_AND, // &&
TOKEN_OR, // ||
@@ -20,7 +29,6 @@ typedef enum {
TOKEN_GT, // >
TOKEN_NE, // !=
TOKEN_EQ, // ==
TOKEN_ASSIGN,
TOKEN_PLUS, // +
TOKEN_MINUS, // -
TOKEN_MODULO, // %
@@ -72,5 +80,5 @@ typedef struct {
} Token;
Token *create_token(TokenType type, int line, int column, char *value);
void free_token(void * ptr);
void free_token(void *ptr);
#endif

View File

@@ -10,7 +10,7 @@
int main() {
ar_memory_init();
const char * path = "test.ar";
char * path = "test.ar";
DArray tokens;
darray_init(&tokens, sizeof(Token));
@@ -22,16 +22,17 @@ int main() {
&tokens
};
lexer(state);
fclose(state.file);
DArray parsed;
DArray ast;
darray_init(&parsed, sizeof(ParsedValue));
darray_init(&ast, sizeof(ParsedValue));
parser(&parsed, &tokens, false);
parser(path,&ast, &tokens, false);
darray_free(&tokens, free_token);
darray_free(&parsed,free_parsed_value);
darray_free(&ast,free_parsed);
return 0;
}

View File

@@ -3,26 +3,10 @@
#include <string.h>
#include <stdlib.h> // for malloc/free (temp arena fallback)
static char* temp_arena = NULL;
static size_t temp_arena_capacity = 0;
static size_t temp_arena_offset = 0;
#define TEMP_ARENA_INITIAL_CAPACITY 4096
void ar_memory_init() {
GC_INIT();
}
void ar_memory_shutdown() {
// No-op for Boehm, but could clean up temp arena here
if (temp_arena) {
free(temp_arena);
temp_arena = NULL;
temp_arena_capacity = 0;
temp_arena_offset = 0;
}
}
void* ar_alloc(size_t size) {
return GC_MALLOC(size);
}

View File

@@ -8,12 +8,7 @@ void* ar_alloc(size_t size);
void* ar_alloc_atomic(size_t size);
char* ar_strdup(const char* str);
// Optional: temporary/arena allocations (e.g., for parsing)
void* ar_temp_alloc(size_t size);
void ar_temp_free_all();
// Memory init/shutdown
void ar_memory_init();
void ar_memory_shutdown();
#endif // ARGON_MEMORY_H

View File

@@ -0,0 +1,44 @@
#include "assign.h"
#include "../../lexer/token.h"
#include "../parser.h"
#include <stdbool.h>
#include <stdio.h>
#include <stdlib.h>
ParsedValue *parse_assign(char*file,DArray *parsed, DArray *tokens,
ParsedValue *assign_to, size_t *index) {
Token *token = darray_get(tokens, *index);
ParsedAssign *assign = malloc(sizeof(ParsedAssign));
assign->to = assign_to;
assign->type = token->type;
(*index)++;
token = darray_get(tokens, *index);
switch (token->type) {
case TOKEN_ASSIGN:
case TOKEN_ASSIGN_CARET:
case TOKEN_ASSIGN_FLOORDIV:
case TOKEN_ASSIGN_MINUS:
case TOKEN_ASSIGN_MODULO:
case TOKEN_ASSIGN_PLUS:
case TOKEN_ASSIGN_SLASH:
case TOKEN_ASSIGN_STAR:
fprintf(stderr, "%s:%u:%u error: invalid syntax\n", file, token->line,
token->column);
exit(EXIT_FAILURE);
default:
break;
}
assign->from = parse_token(file,parsed, tokens, index, true);
ParsedValue *parsedValue = malloc(sizeof(ParsedValue));
parsedValue->type = AST_ASSIGN;
parsedValue->data = assign;
return parsedValue;
}
void free_parse_assign(void*ptr) {
ParsedValue * parsedValue = ptr;
ParsedAssign* parsedAssign = parsedValue->data;
free_parsed(parsedAssign->to);
free_parsed(parsedAssign->from);
free(parsedAssign);
}

View File

@@ -0,0 +1,17 @@
#ifndef ASSIGN_H
#define ASSIGN_H
#include "../parser.h"
#include "../../lexer/token.h"
typedef struct {
ParsedValue * to;
TokenType type;
ParsedValue * from;
} ParsedAssign;
ParsedValue *parse_assign(char*file,DArray *parsed, DArray *tokens,
ParsedValue *assign_to, size_t *index);
void free_parse_assign(void*ptr);
#endif // ASSIGN_H

View File

@@ -1,35 +1,67 @@
#include "parser.h"
#include "../dynamic_array/darray.h"
#include "../lexer/token.h"
#include "assign/assign.h"
#include "string/string.h"
#include <stdbool.h>
#include <stddef.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
ParsedValue *parse_token(DArray *tokens, size_t *index) {
ParsedValue *parse_token(char *file, DArray *parsed, DArray *tokens,
size_t *index, bool inline_flag) {
Token *token = darray_get(tokens, *index);
if (!inline_flag) {
switch (token->type) {
default:
break;
};
}
switch (token->type) {
case TOKEN_STRING:
(*index)++;
return parse_string(*token);
case TOKEN_NEW_LINE:
(*index)++;
return NULL;
while (token->type == TOKEN_NEW_LINE && *index+1 <= tokens->size) {
(*index)++;
token = darray_get(tokens, *index);
}
if (token->type == TOKEN_NEW_LINE) return NULL;
return parse_token(file, parsed, tokens, index, inline_flag);
case TOKEN_INDENT:
fprintf(stderr, "error: \n");
fprintf(stderr, "%s:%u:%u error: invalid indentation\n", file, token->line,
token->column);
exit(EXIT_FAILURE);
case TOKEN_ASSIGN:
case TOKEN_ASSIGN_CARET:
case TOKEN_ASSIGN_FLOORDIV:
case TOKEN_ASSIGN_MINUS:
case TOKEN_ASSIGN_MODULO:
case TOKEN_ASSIGN_PLUS:
case TOKEN_ASSIGN_SLASH:
case TOKEN_ASSIGN_STAR:
if (parsed->size == 0) {
fprintf(stderr, "%s:%u:%u error: assigning to nothing\n", file, token->line,
token->column);
exit(EXIT_FAILURE);
}
ParsedValue *assign_to = malloc(sizeof(ParsedValue));
memcpy(assign_to, darray_get(parsed, parsed->size-1), sizeof(ParsedValue));
darray_resize(parsed, parsed->size-1);
return parse_assign(file, parsed, tokens, assign_to, index);
default:
fprintf(stderr, "Panic: unreachable\n");
exit(EXIT_FAILURE);
}
}
void parser(DArray *parsed, DArray *tokens, bool inline_flag) {
void parser(char *file, DArray *parsed, DArray *tokens, bool inline_flag) {
size_t index = 0;
size_t length = tokens->size;
while (index < length) {
ParsedValue *parsed_code = parse_token(tokens, &index);
ParsedValue *parsed_code =
parse_token(file, parsed, tokens, &index, inline_flag);
if (parsed_code) {
darray_push(parsed, parsed_code);
free(parsed_code);
@@ -37,11 +69,14 @@ void parser(DArray *parsed, DArray *tokens, bool inline_flag) {
}
}
void free_parsed_value(void *ptr) {
ParsedValue *tagged = ptr;
switch (tagged->type) {
void free_parsed(void *ptr) {
ParsedValue *parsed = ptr;
switch (parsed->type) {
case AST_STRING:
free(tagged->data);
free(parsed->data);
break;
case AST_ASSIGN:
free_parse_assign(parsed);
break;
}
}

View File

@@ -10,6 +10,7 @@ typedef struct LinkedList LinkedList;
typedef enum {
AST_STRING,
AST_ASSIGN,
} ValueType;
typedef struct {
@@ -18,11 +19,11 @@ typedef struct {
} ParsedValue;
void parser(DArray * parsed, DArray * tokens, bool inline_flag);
void parser(char*file,DArray *parsed, DArray *tokens, bool inline_flag);
ParsedValue * parse_token(DArray * tokens, size_t *index);
ParsedValue *parse_token(char*file,DArray *parsed, DArray *tokens, size_t *index, bool inline_flag);
void free_parsed_value(void *ptr);
void free_parsed(void *ptr);
#endif // PARSER_H

View File

@@ -0,0 +1,12 @@
#include "translator.h"
#include "../dynamic_array/darray.h"
#include <stdlib.h>
#include <stdint.h>
Translated * init_translator() {
Translated *translated = malloc(sizeof(Translated));
if (!translated) return NULL;
darray_init(&translated->bytecode, sizeof(uint8_t));
return translated;
}

View File

@@ -0,0 +1,18 @@
#ifndef TRANSLATOR_H
#define TRANSLATOR_H
#include <stddef.h>
#include "../dynamic_array/darray.h"
typedef enum {
OP_INIT_STRING
} OperationType;
typedef struct {
size_t registerCount;
DArray bytecode;
} Translated;
#endif

View File

@@ -1 +1,2 @@
'hello world'
"hello world"
"hello world"="hello world"