add identifier to parser

This commit is contained in:
2025-06-01 02:28:27 +01:00
parent b7e9493171
commit d2518afb8e
20 changed files with 733 additions and 251 deletions

View File

@@ -1,30 +1,27 @@
#include "lex.yy.h"
#include "lexer.h"
#include "lex.yy.h"
void lexer(LexerState state) {
yyscan_t scanner;
yyscan_t scanner;
yylex_init(&scanner);
yylex_init(&scanner);
yyset_extra(&state, scanner);
yyset_extra(&state, scanner);
yyset_in(state.file, scanner);
yyset_in(state.file, scanner);
int token;
while ((token = yylex(scanner)) != 0) {
Token * token_struct = create_token(
token,
yyget_lineno(scanner),
state.current_column+1,
yyget_text(scanner)
);
darray_push(state.tokens, token_struct);
free(token_struct);
if (token == TOKEN_NEW_LINE) {
state.current_column = 0;
} else {
state.current_column += yyget_leng(scanner);
}
int token;
while ((token = yylex(scanner)) != 0) {
Token *token_struct =
create_token(token, yyget_lineno(scanner), state.current_column + 1,
yyget_text(scanner));
darray_push(state.tokens, token_struct);
free(token_struct);
if (token == TOKEN_NEW_LINE) {
state.current_column = 0;
} else {
state.current_column += yyget_leng(scanner);
}
yylex_destroy(scanner);
}
yylex_destroy(scanner);
}

View File

@@ -1,19 +1,18 @@
#ifndef LEXER_H
#define LEXER_H
#include "token.h"
#include "../dynamic_array/darray.h"
#include "token.h"
#include <stdio.h>
typedef struct {
const char *path;
FILE *file;
int current_column;
DArray* tokens;
// add more fields as needed
const char *path;
FILE *file;
int current_column;
DArray *tokens;
// add more fields as needed
} LexerState;
void lexer(LexerState state);
#endif // LEXER_H

View File

@@ -1,17 +1,17 @@
#include "token.h"
#include <stdlib.h>
#include "../string/string.h"
#include <stdlib.h>
Token *create_token(TokenType type, int line, int column, char *value) {
Token * token = malloc(sizeof(Token));
token->type = type;
token->line=line;
token->column=column;
token->value=cloneString(value);
return token;
Token *token = malloc(sizeof(Token));
token->type = type;
token->line = line;
token->column = column;
token->value = cloneString(value);
return token;
}
void free_token(void * ptr) {
Token* token = ptr;
free(token->value);
void free_token(void *ptr) {
Token *token = ptr;
free(token->value);
}

View File

@@ -20,15 +20,15 @@ typedef enum {
TOKEN_ASSIGN_CARET,
// Operators
TOKEN_AND, // &&
TOKEN_OR, // ||
TOKEN_NOT_IN, // not in
TOKEN_LE, // <=
TOKEN_GE, // >=
TOKEN_LT, // <
TOKEN_GT, // >
TOKEN_NE, // !=
TOKEN_EQ, // ==
TOKEN_AND, // &&
TOKEN_OR, // ||
TOKEN_NOT_IN, // not in
TOKEN_LE, // <=
TOKEN_GE, // >=
TOKEN_LT, // <
TOKEN_GT, // >
TOKEN_NE, // !=
TOKEN_EQ, // ==
TOKEN_PLUS, // +
TOKEN_MINUS, // -
TOKEN_MODULO, // %

View File

@@ -1,97 +1,99 @@
#include "list.h"
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "list.h"
LinkedList *create_list(size_t data_size) {
LinkedList *list = malloc(sizeof(LinkedList));
list->head = NULL;
list->data_size = data_size;
list->length = 0;
return list;
LinkedList *list = malloc(sizeof(LinkedList));
list->head = NULL;
list->data_size = data_size;
list->length = 0;
return list;
}
void append(LinkedList *list, void *element) {
Node *new_node = malloc(sizeof(Node));
new_node->data = malloc(list->data_size);
memcpy(new_node->data, element, list->data_size);
new_node->next = NULL;
Node *new_node = malloc(sizeof(Node));
new_node->data = malloc(list->data_size);
memcpy(new_node->data, element, list->data_size);
new_node->next = NULL;
if (!list->head) {
list->head = new_node;
} else {
Node *temp = list->head;
while (temp->next) temp = temp->next;
temp->next = new_node;
}
list->length++;
if (!list->head) {
list->head = new_node;
} else {
Node *temp = list->head;
while (temp->next)
temp = temp->next;
temp->next = new_node;
}
list->length++;
}
void *get_element_at(LinkedList *list, size_t index) {
if (index >= list->length) return NULL;
if (index >= list->length)
return NULL;
Node *current = list->head;
for (size_t i = 0; i < index; ++i)
current = current->next;
Node *current = list->head;
for (size_t i = 0; i < index; ++i)
current = current->next;
return current->data;
return current->data;
}
int set_element_at(LinkedList *list, size_t index, void *element) {
if (index >= list->length) return 0;
if (index >= list->length)
return 0;
Node *current = list->head;
for (size_t i = 0; i < index; ++i)
current = current->next;
Node *current = list->head;
for (size_t i = 0; i < index; ++i)
current = current->next;
memcpy(current->data, element, list->data_size);
return 1;
memcpy(current->data, element, list->data_size);
return 1;
}
int remove_at(LinkedList *list, size_t index) {
if (index >= list->length) return 0;
if (index >= list->length)
return 0;
Node *temp = list->head;
Node *prev = NULL;
Node *temp = list->head;
Node *prev = NULL;
for (size_t i = 0; i < index; ++i) {
prev = temp;
temp = temp->next;
}
for (size_t i = 0; i < index; ++i) {
prev = temp;
temp = temp->next;
}
if (prev)
prev->next = temp->next;
else
list->head = temp->next;
if (prev)
prev->next = temp->next;
else
list->head = temp->next;
free(temp->data);
free(temp);
list->length--;
return 1;
free(temp->data);
free(temp);
list->length--;
return 1;
}
size_t list_length(LinkedList *list) {
return list->length;
}
size_t list_length(LinkedList *list) { return list->length; }
void print_list(LinkedList *list, void (*print_func)(void *)) {
Node *current = list->head;
while (current) {
print_func(current->data);
current = current->next;
}
Node *current = list->head;
while (current) {
print_func(current->data);
current = current->next;
}
}
void free_list(LinkedList *list, void (*free_data)(void *)) {
Node *current = list->head;
while (current) {
Node *next = current->next;
Node *current = list->head;
while (current) {
Node *next = current->next;
if (free_data) // Safe to pass NULL if you don't need it
free_data(current->data);
if (free_data) // Safe to pass NULL if you don't need it
free_data(current->data);
free(current);
current = next;
}
free(list);
free(current);
current = next;
}
free(list);
}

View File

@@ -5,14 +5,14 @@
// Node structure (opaque to user)
typedef struct Node {
void *data;
struct Node *next;
void *data;
struct Node *next;
} Node;
typedef struct LinkedList {
Node *head;
size_t data_size;
size_t length;
Node *head;
size_t data_size;
size_t length;
} LinkedList;
// Create a new list for the given data type size

View File

@@ -1,38 +1,34 @@
#include "dynamic_array/darray.h"
#include "lexer/lexer.h"
#include "lexer/token.h"
#include "parser/parser.h"
#include "memory.h"
#include "dynamic_array/darray.h"
#include "parser/parser.h"
#include <stdbool.h>
#include <stddef.h>
#include <stdio.h>
int main() {
ar_memory_init();
char * path = "test.ar";
DArray tokens;
int main(int argc, char *argv[]) {
if (argc <= 1)
return -1;
ar_memory_init();
char *path = argv[1];
DArray tokens;
darray_init(&tokens, sizeof(Token));
darray_init(&tokens, sizeof(Token));
LexerState state = {
path,
fopen(path, "r"),
0,
&tokens
};
lexer(state);
fclose(state.file);
LexerState state = {path, fopen(path, "r"), 0, &tokens};
lexer(state);
fclose(state.file);
DArray ast;
DArray ast;
darray_init(&ast, sizeof(ParsedValue));
darray_init(&ast, sizeof(ParsedValue));
parser(path, &ast, &tokens, false);
darray_free(&tokens, free_token);
parser(path,&ast, &tokens, false);
darray_free(&tokens, free_token);
darray_free(&ast,free_parsed);
darray_free(&ast, free_parsed);
return 0;
return 0;
}

View File

@@ -1,23 +1,17 @@
#include "memory.h"
#include <gc.h>
#include <string.h>
#include <stdlib.h> // for malloc/free (temp arena fallback)
#include <string.h>
void ar_memory_init() {
GC_INIT();
}
void ar_memory_init() { GC_INIT(); }
void* ar_alloc(size_t size) {
return GC_MALLOC(size);
}
void *ar_alloc(size_t size) { return GC_MALLOC(size); }
void* ar_alloc_atomic(size_t size) {
return GC_MALLOC_ATOMIC(size);
}
void *ar_alloc_atomic(size_t size) { return GC_MALLOC_ATOMIC(size); }
char* ar_strdup(const char* str) {
size_t len = strlen(str) + 1;
char* copy = (char*)GC_MALLOC(len);
memcpy(copy, str, len);
return copy;
char *ar_strdup(const char *str) {
size_t len = strlen(str) + 1;
char *copy = (char *)GC_MALLOC(len);
memcpy(copy, str, len);
return copy;
}

View File

@@ -4,9 +4,9 @@
#include <stddef.h> // for size_t
// GC-managed allocations
void* ar_alloc(size_t size);
void* ar_alloc_atomic(size_t size);
char* ar_strdup(const char* str);
void *ar_alloc(size_t size);
void *ar_alloc_atomic(size_t size);
char *ar_strdup(const char *str);
// Memory init/shutdown
void ar_memory_init();

View File

@@ -5,7 +5,7 @@
#include <stdio.h>
#include <stdlib.h>
ParsedValue *parse_assign(char*file,DArray *parsed, DArray *tokens,
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));
@@ -24,20 +24,20 @@ ParsedValue *parse_assign(char*file,DArray *parsed, DArray *tokens,
case TOKEN_ASSIGN_STAR:
fprintf(stderr, "%s:%u:%u error: invalid syntax\n", file, token->line,
token->column);
exit(EXIT_FAILURE);
default:
break;
exit(EXIT_FAILURE);
default:
break;
}
assign->from = parse_token(file,parsed, tokens, index, true);
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;
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 +1,12 @@
#include "identifier.h"
#include "../../lexer/token.h"
#include "../parser.h"
#include <stdlib.h>
#include <string.h>
ParsedValue *parse_identifier(Token *token) {
ParsedValue *parsedValue = malloc(sizeof(ParsedValue));
parsedValue->type = AST_IDENTIFIER;
parsedValue->data = strcpy(malloc(sizeof(token->value)), token->value);
return parsedValue;
}

View File

@@ -0,0 +1,11 @@
// parser.h
#ifndef IDENTIFIER_H
#define IDENTIFIER_H
#include "../parser.h"
#include "../../lexer/token.h" // for Token
// Function declaration for parsing an identifier
ParsedValue * parse_identifier(Token * token);
#endif // IDENTIFIER_H

View File

@@ -2,6 +2,7 @@
#include "../dynamic_array/darray.h"
#include "../lexer/token.h"
#include "assign/assign.h"
#include "identifier/identifier.h"
#include "string/string.h"
#include <stdbool.h>
#include <stddef.h>
@@ -14,8 +15,8 @@ ParsedValue *parse_token(char *file, DArray *parsed, DArray *tokens,
Token *token = darray_get(tokens, *index);
if (!inline_flag) {
switch (token->type) {
default:
break;
default:
break;
};
}
switch (token->type) {
@@ -23,11 +24,11 @@ ParsedValue *parse_token(char *file, DArray *parsed, DArray *tokens,
(*index)++;
return parse_string(*token);
case TOKEN_NEW_LINE:
while (token->type == TOKEN_NEW_LINE && *index+1 <= tokens->size) {
(*index)++;
while (token->type == TOKEN_NEW_LINE && ++(*index) < tokens->size) {
token = darray_get(tokens, *index);
}
if (token->type == TOKEN_NEW_LINE) return NULL;
if (token->type == TOKEN_NEW_LINE)
return NULL;
return parse_token(file, parsed, tokens, index, inline_flag);
case TOKEN_INDENT:
fprintf(stderr, "%s:%u:%u error: invalid indentation\n", file, token->line,
@@ -43,13 +44,17 @@ ParsedValue *parse_token(char *file, DArray *parsed, DArray *tokens,
case TOKEN_ASSIGN_STAR:
if (parsed->size == 0) {
fprintf(stderr, "%s:%u:%u error: syntax error\n", file, token->line,
token->column);
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);
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);
case TOKEN_IDENTIFIER:
(*index)++;
return parse_identifier(token);
default:
fprintf(stderr, "Panic: unreachable\n");
exit(EXIT_FAILURE);
@@ -72,6 +77,7 @@ void parser(char *file, DArray *parsed, DArray *tokens, bool inline_flag) {
void free_parsed(void *ptr) {
ParsedValue *parsed = ptr;
switch (parsed->type) {
case AST_IDENTIFIER:
case AST_STRING:
free(parsed->data);
break;

View File

@@ -1,29 +1,28 @@
#ifndef PARSER_H
#define PARSER_H
#include "../dynamic_array/darray.h"
#include <stdbool.h>
#include <stddef.h>
#include "../dynamic_array/darray.h"
typedef struct LinkedList LinkedList;
typedef enum {
AST_STRING,
AST_ASSIGN,
AST_IDENTIFIER,
} ValueType;
typedef struct {
ValueType type;
void *data;
} ParsedValue;
void parser(char*file,DArray *parsed, DArray *tokens, bool inline_flag);
void parser(char *file, DArray *parsed, DArray *tokens, bool inline_flag);
ParsedValue *parse_token(char*file,DArray *parsed, DArray *tokens, size_t *index, bool inline_flag);
ParsedValue *parse_token(char *file, DArray *parsed, DArray *tokens,
size_t *index, bool inline_flag);
void free_parsed(void *ptr);
#endif // PARSER_H

View File

@@ -42,6 +42,7 @@ char *unquote(char *str) {
cJSON *json = cJSON_Parse(str);
if (!json || !cJSON_IsString(json)) {
cJSON_Delete(json);
if (swapped)
free(swapped);
return NULL;
@@ -71,9 +72,8 @@ char *unquote(char *str) {
return unescaped;
}
ParsedValue * parse_string(Token token) {
ParsedValue * parsedValue = malloc(sizeof(ParsedValue));
ParsedValue *parse_string(Token token) {
ParsedValue *parsedValue = malloc(sizeof(ParsedValue));
parsedValue->type = AST_STRING;
parsedValue->data = unquote(token.value);
return parsedValue;

View File

@@ -6,69 +6,57 @@
const char *WHITE_SPACE = " \t\n\r\f\v";
char *cloneString(char *str)
{
if (str == NULL)
{
return NULL;
}
char *cloneString(char *str) {
if (str == NULL) {
return NULL;
}
size_t len = strlen(str);
char *clone = malloc((len + 1) * sizeof(char));
size_t len = strlen(str);
char *clone = malloc((len + 1) * sizeof(char));
if (clone == NULL)
{
return NULL;
}
if (clone == NULL) {
return NULL;
}
strcpy(clone, str);
return clone;
strcpy(clone, str);
return clone;
}
void stripString(char *str, const char *chars)
{
if (str == NULL || chars == NULL)
{
return;
}
size_t len = strlen(str);
size_t charsLen = strlen(chars);
if (len == 0 || charsLen == 0)
{
return;
}
size_t i = 0;
while (i < len)
{
if (strchr(chars, str[i]) == NULL)
{
break;
}
i++;
}
if (i > 0)
{
memmove(str, str + i, len - i + 1);
}
size_t j = len-i - 1;
while (j > 0)
{
if (strchr(chars, str[j]) == NULL)
{
break;
}
j--;
}
if (j < len)
{
str[j + 1] = '\0';
}
str = realloc(str, (j + 1) * sizeof(char));
void stripString(char *str, const char *chars) {
if (str == NULL || chars == NULL) {
return;
}
size_t len = strlen(str);
size_t charsLen = strlen(chars);
if (len == 0 || charsLen == 0) {
return;
}
size_t i = 0;
while (i < len) {
if (strchr(chars, str[i]) == NULL) {
break;
}
i++;
}
if (i > 0) {
memmove(str, str + i, len - i + 1);
}
size_t j = len - i - 1;
while (j > 0) {
if (strchr(chars, str[j]) == NULL) {
break;
}
j--;
}
if (j < len) {
str[j + 1] = '\0';
}
str = realloc(str, (j + 1) * sizeof(char));
return;
}

View File

@@ -1,11 +1,10 @@
#ifndef CLONESTRING_H
#define CLONESTRING_H
extern const char * WHITE_SPACE;
extern const char *WHITE_SPACE;
char* cloneString(char* str);
void stripString(char* str, const char* chars);
char *cloneString(char *str);
void stripString(char *str, const char *chars);
#endif // CLONESTRING_H

View File

@@ -1,11 +1,12 @@
#include "translator.h"
#include "../dynamic_array/darray.h"
#include <stdlib.h>
#include <stdint.h>
#include <stdlib.h>
Translated * init_translator() {
Translated *init_translator() {
Translated *translated = malloc(sizeof(Translated));
if (!translated) return NULL;
if (!translated)
return NULL;
darray_init(&translated->bytecode, sizeof(uint8_t));
return translated;

View File

@@ -1,14 +1,10 @@
#ifndef TRANSLATOR_H
#define TRANSLATOR_H
#include <stddef.h>
#include "../dynamic_array/darray.h"
#include <stddef.h>
typedef enum {
OP_INIT_STRING
} OperationType;
typedef enum { OP_INIT_STRING } OperationType;
typedef struct {
size_t registerCount;