Compare commits

..

7 Commits

Author SHA1 Message Date
seeseemelk b6d0a78d06 Add integration test 2026-05-01 09:44:46 +02:00
seeseemelk 3bdccf2000 Add integration test framework 2026-04-30 22:21:08 +02:00
seeseemelk 177fb971e4 Rename AST structures to Tree and relocate freeing logic 2026-04-30 21:46:15 +02:00
seeseemelk ea55dedd07 Refactor AST and Parser into modular subdirectories
- Split ast.h into granular headers in v0/ast/
- Split parser.c into modular implementation files in v0/parser/
- Move and rename parser tests to v0/parser/test_*.c
- Update build system (include.mk) with modular sub-makefiles
- Maintain v0/ast.h and v0/parser.h as umbrella headers
2026-04-30 21:23:07 +02:00
seeseemelk 4bd66ea216 More variable stuff 2026-04-30 20:25:53 +02:00
seeseemelk 0704284726 Can parse variables 2026-04-29 21:39:48 +02:00
seeseemelk 94ae665a0a Add initial variable work 2026-04-29 21:20:52 +02:00
40 changed files with 902 additions and 534 deletions
+2 -3
View File
@@ -26,9 +26,8 @@ There will be no `test_buffer.h`. Instead, `test.c` will directly
Every syntax error path identified in the parser MUST have a corresponding test.
## Language Syntax
Since this is a compiler for a new language, do not assume anything
of its syntax.
Always check the `specs` directory.
Since this is a compiler for a new language, do not assume anything of its syntax.
Always check the `specs` directory to see examples and documentation about the language.
If there is anything unclear, ask the user for clarification.
It is certainly possible that there are contradictions in the
+7 -1
View File
@@ -1,6 +1,6 @@
.PHONY: all test clean
all: c2 test
all: c2 test integration-test
c2: v0/bin/c2
cp $< $@
@@ -13,3 +13,9 @@ clean::
rm -f c2
include v0/include.mk
integration-test: v0/bin/c2 v0/bin/test_integration
./v0/bin/test_integration
v0/bin/test_integration: v0/test_integration.c
$(CC) $(CFLAGS) -o $@ $<
+16
View File
@@ -13,5 +13,21 @@ In order to run the tests, run `make test`.
## Versioning
The current version is v0. Its source code lives in the `v0` directory.
## Testing
### Unit Tests
Run unit tests with:
```bash
make test
```
### Integration Tests
Integration tests compare the compiler output with expected C files.
To add a new integration test, create a new directory under `v0/integration_tests/` with `input.c2` and `expected.c` files.
Run integration tests with:
```bash
make integration-test
```
## Languages Specifications
See the specs directory for information on the actual language syntax.
+1
View File
@@ -0,0 +1 @@
Hello, world
+2 -2
View File
@@ -6,10 +6,10 @@ Global variables can be defined as such:
```c2
// Defines a global variable called my_var.
int32 my_var;
i32 my_var;
// Defines a const variable.
const int32 my_var;
const i32 my_var;
// Defines a global variable whose type is determined automatically.
// The value will be determined at runtime.
+7 -72
View File
@@ -4,83 +4,18 @@
#ifndef AST_H
#define AST_H
#include "bool.h"
#include "token.h"
#include <stddef.h>
typedef struct {
/** @brief The name of the module being imported. */
char* module_name;
/** @brief Whether the import is public or not. */
bool is_public;
} ImportDeclaration;
typedef enum {
TYPE_EXPRESSION_BUILTIN,
TYPE_EXPRESSION_ARRAY
} TypeExpressionTag;
#include "ast/expression.h"
#include "ast/declaration.h"
#include "ast/module.h"
/**
* An expression that evaluates to a type.
* Frees a module and all its children.
*/
typedef struct TypeExpression TypeExpression;
struct TypeExpression{
/** @brief defines which entry in the union is valid */
TypeExpressionTag tag;
union {
/** @brief Evaluates to an array of the given type. */
struct {
/** @brief A pointer to the type of the elements stored in the array. */
TypeExpression* array;
} array;
/** @brief Evaluates to a builtin integer type.*/
struct {
/**
* @brief The number of bits in the integer.
* Typical values are 8, 16, 32, and 64.
*/
int bitSize;
/** @brief `true` if the type is signed, `false` if it's unsigned. */
bool isSigned;
} builtin;
};
};
void ast_free_module(ModuleTree* module);
/**
* A declaration that aliases one type to another.
* Frees a type expression.
*/
typedef struct {
/** @brief The name of the alias. */
char* name;
/** @brief The value of the alias. */
TypeExpression value;
/** @brief Whether the import is public or not. */
bool is_public;
} AliasDeclaration;
/**
* The top-level model.
* Every file matches an entire Module.
*/
typedef struct {
/** @brief The name of the module. */
char* name;
/** @brief The list of imports in the module. */
ImportDeclaration* imports;
/** @brief The number of imports in the module. */
size_t import_count;
/** @brief The list of aliases in the module. */
AliasDeclaration* aliases;
/** @brief The number of aliases in the module. */
size_t alias_count;
} Module;
void ast_free_type(TypeTree* type);
#endif
+49
View File
@@ -0,0 +1,49 @@
#ifndef AST_DECLARATION_H
#define AST_DECLARATION_H
#include "expression.h"
#include "../bool.h"
typedef struct {
/** @brief The name of the module being imported. */
char* module_name;
/** @brief Whether the import is public or not. */
bool is_public;
} ImportTree;
/**
* A declaration that aliases one type to another.
*/
typedef struct {
/** @brief The name of the alias. */
const char* name;
/** @brief The value of the alias. */
TypeTree value;
} AliasTree;
/**
* A declaration of a variable, which may be a constant or not, and may be static or not.
*/
typedef struct {
/** @brief The name of the variable. */
char* name;
/** @brief The type of the variable. */
TypeTree type;
/** @brief The optional initializer expression. */
ExpressionTree* initializer;
/** @brief Whether the variable is public or not. */
bool is_public;
/** @brief Whether the variable is static or not. */
bool is_static;
/** @brief Whether the variable is a constant or not. */
bool is_const;
} VariableTree;
#endif
+9
View File
@@ -0,0 +1,9 @@
#include "expression.h"
#include <stdlib.h>
void ast_free_type(TypeTree* expr) {
if (expr->tag == TYPE_TREE_ARRAY) {
ast_free_type(expr->array.array);
free(expr->array.array);
}
}
+52
View File
@@ -0,0 +1,52 @@
#ifndef AST_EXPRESSION_H
#define AST_EXPRESSION_H
#include "../bool.h"
typedef enum {
EXPRESSION_TREE_INTEGER,
EXPRESSION_TREE_STRING,
EXPRESSION_TREE_BOOLEAN
} ExpressionTreeTag;
typedef struct {
ExpressionTreeTag tag;
union {
int integer;
const char* string;
bool boolean;
};
} ExpressionTree;
typedef enum {
TYPE_TREE_BUILTIN,
TYPE_TREE_ARRAY
} TypeTreeTag;
/**
* An expression that evaluates to a type.
*/
typedef struct TypeTree TypeTree;
struct TypeTree {
/** @brief defines which entry in the union is valid */
TypeTreeTag tag;
union {
/** @brief Evaluates to an array of the given type. */
struct {
/** @brief A pointer to the type of the elements stored in the array. */
TypeTree* array;
} array;
/** @brief Evaluates to a builtin integer type.*/
struct {
/**
* @brief The number of bits in the integer.
* Typical values are 8, 16, 32, and 64.
*/
int bitSize;
/** @brief `true` if the type is signed, `false` if it's unsigned. */
bool isSigned;
} builtin;
};
};
#endif
+3
View File
@@ -0,0 +1,3 @@
# There are currently no .c files in the ast directory.
# This file is provided for future consistency.
AST_SRC := v0/ast/module.c v0/ast/expression.c
+43
View File
@@ -0,0 +1,43 @@
#include "module.h"
#include "expression.h"
#include <stdlib.h>
void ast_free_type(TypeTree* type);
void ast_free_module(ModuleTree* module) {
if (module == NULL) {
return;
}
if (module->imports != NULL) {
for(size_t i = 0; i < module->import_count; i++) {
free(module->imports[i].module_name);
}
free(module->imports);
}
if (module->aliases != NULL) {
for(size_t i = 0; i < module->alias_count; i++) {
free((void*)module->aliases[i].name);
ast_free_type(&module->aliases[i].value);
}
free(module->aliases);
}
if (module->variables != NULL) {
for(size_t i = 0; i < module->variable_count; i++) {
free(module->variables[i].name);
ast_free_type(&module->variables[i].type);
if (module->variables[i].initializer) {
if (module->variables[i].initializer->tag == EXPRESSION_TREE_STRING) {
free((void*)module->variables[i].initializer->string);
}
free(module->variables[i].initializer);
}
}
free(module->variables);
}
free(module->name);
free(module);
}
+34
View File
@@ -0,0 +1,34 @@
#ifndef AST_MODULE_H
#define AST_MODULE_H
#include "declaration.h"
#include <stddef.h>
/**
* The top-level model.
* Every file matches an entire Module.
*/
typedef struct {
/** @brief The name of the module. */
char* name;
/** @brief The list of imports in the module. */
ImportTree* imports;
/** @brief The number of imports in the module. */
size_t import_count;
/** @brief The list of aliases in the module. */
AliasTree* aliases;
/** @brief The number of aliases in the module. */
size_t alias_count;
/** @brief The list of variables in the module. */
VariableTree* variables;
/** @brief The number of variables in the module. */
size_t variable_count;
} ModuleTree;
#endif
+4 -1
View File
@@ -1,4 +1,7 @@
V0_SRC := v0/main.c v0/util.c v0/token.c v0/parser.c v0/log.c v0/str.c
include v0/ast/include.mk
include v0/parser/include.mk
V0_SRC := v0/main.c v0/util.c v0/token.c $(AST_SRC) $(PARSER_SRC) v0/log.c v0/str.c
# V0_TEST must only include `v0/test.c` itself, as all other test Csource files are
# included directly into `v0/test.c` using `#include "test_xyz.c"`.
+4
View File
@@ -0,0 +1,4 @@
#include <stdint.h>
// u32 simple:x
static uint32_t v_6simple_1x = 123;
+2
View File
@@ -0,0 +1,2 @@
module simple;
u32 x = 123;
-338
View File
@@ -1,338 +0,0 @@
#include "parser.h"
#include "log.h"
#include <stdlib.h>
#include <string.h>
#include <stdio.h>
typedef struct {
TokenStream* ts;
Token token;
} Parser;
/**
* Reads a new token into p->token.
*/
static void parser_next_token(Parser* p) {
p->token = tokenstream_next(p->ts);
}
/**
* Reads a new token if the current token is equal to the expected token.
*
* If they are equal, it continues to the next token.
*
* @param p
* @param token The expected token.
* @returns `true` if the current token matches the expected, `false` if it does not.
*/
static bool parser_accept(Parser* p, TokenType token) {
if (p->token.token == token) {
parser_next_token(p);
return true;
}
return false;
}
/**
* Consumes the expected token if present, otherwise logs an error.
*
* @param p The parser state.
* @param token The token type to expect.
* @param msg The error message if the token is not found.
* @return true if the token was consumed, false otherwise.
*/
static bool parser_expect(Parser* p, TokenType token, const char* msg) {
if (parser_accept(p, token)) {
return true;
}
log_on_line(&p->token.location, msg);
return false;
}
/**
* Checks if the current token matches the expected token type without consuming it.
*
* @param p The parser state.
* @param token The token type to check.
* @return true if the current token matches, false otherwise.
*/
static bool parser_peek(Parser* p, TokenType token) {
if (p->token.token == token) {
return true;
}
return false;
}
/**
* Checks if the current token matches the expected token type without consuming it.
*
* @param p The parser state.
* @param token The token type to check.
* @return true if the current token matches, false otherwise.
*/
static bool parser_require(Parser* p, TokenType token, const char* msg) {
if (parser_peek(p, token)) {
return true;
}
log_on_line(&p->token.location, msg);
return false;
}
/**
* Returns the text of the current token and advances the parser to the next token.
*
* @param p The parser state.
* @return A newly allocated string containing the current token's text.
*/
static char* parser_to_text(Parser* p) {
char* str = string_copy(p->token.text);
parser_next_token(p);
return str;
}
/**
* Parses the "module" keyword
*/
static bool parse_module_declaration(Parser* p, Module* module) {
if (!parser_expect(p, TOKEN_MODULE, "expected keyword 'module'")) {
return false;
}
if (!parser_require(p, TOKEN_IDENTIFIER, "expected module identifier")) {
return false;
}
module->name = parser_to_text(p);
return parser_expect(p, TOKEN_SEMICOLON, "expected ';' after module name");
}
/**
* Parses an import declaration.
*
* @param p The parser state.
* @param module The module to add the import to.
* @param is_public Whether the import is public.
* @return true if successful, false otherwise.
*/
static bool parse_import_declaration(Parser* p, Module* module, bool is_public) {
module->import_count++;
module->imports = realloc(module->imports, sizeof(ImportDeclaration) * module->import_count);
ImportDeclaration* import = &module->imports[module->import_count - 1];
memset(import, 0, sizeof(ImportDeclaration));
import->is_public = is_public;
if (!parser_require(p, TOKEN_IDENTIFIER, "expected module identifier")) {
return false;
}
import->module_name = parser_to_text(p);
if (!parser_expect(p, TOKEN_SEMICOLON, "expected ';' after import")) {
return false;
}
return true;
}
/**
* Parses a primitive type expression.
*
* @param p The parser state.
* @param expr The type expression to populate.
* @return true if successful, false otherwise.
*/
static bool parse_primitive_type_expression(Parser* p, TypeExpression* expr) {
if (parser_accept(p, TOKEN_U8)) {
expr->tag = TYPE_EXPRESSION_BUILTIN;
expr->builtin.bitSize = 8;
expr->builtin.isSigned = false;
return true;
} else if (parser_accept(p, TOKEN_U16)) {
expr->tag = TYPE_EXPRESSION_BUILTIN;
expr->builtin.bitSize = 16;
expr->builtin.isSigned = false;
return true;
} else if (parser_accept(p, TOKEN_U32)) {
expr->tag = TYPE_EXPRESSION_BUILTIN;
expr->builtin.bitSize = 32;
expr->builtin.isSigned = false;
return true;
} else if (parser_accept(p, TOKEN_U64)) {
expr->tag = TYPE_EXPRESSION_BUILTIN;
expr->builtin.bitSize = 64;
expr->builtin.isSigned = false;
return true;
} else if (parser_accept(p, TOKEN_I8)) {
expr->tag = TYPE_EXPRESSION_BUILTIN;
expr->builtin.bitSize = 8;
expr->builtin.isSigned = true;
return true;
} else if (parser_accept(p, TOKEN_I16)) {
expr->tag = TYPE_EXPRESSION_BUILTIN;
expr->builtin.bitSize = 16;
expr->builtin.isSigned = true;
return true;
} else if (parser_accept(p, TOKEN_I32)) {
expr->tag = TYPE_EXPRESSION_BUILTIN;
expr->builtin.bitSize = 32;
expr->builtin.isSigned = true;
return true;
} else if (parser_accept(p, TOKEN_I64)) {
expr->tag = TYPE_EXPRESSION_BUILTIN;
expr->builtin.bitSize = 64;
expr->builtin.isSigned = true;
return true;
} else {
log_on_line(&p->token.location, "expected type expression");
return false;
}
}
/**
* Parses an array type expression.
*
* @param p The parser state.
* @param expr The type expression to populate.
* @return true if successful, false otherwise.
*/
static bool parse_array_type_expression(Parser* p, TypeExpression* expr) {
TypeExpression elementType;
if (!parse_primitive_type_expression(p, &elementType)) {
return false;
}
if (parser_accept(p, TOKEN_BRACKET_OPEN)) {
expr->tag = TYPE_EXPRESSION_ARRAY;
expr->array.array = malloc(sizeof(TypeExpression));
*expr->array.array = elementType;
if (!parser_expect(p, TOKEN_BRACKET_CLOSE, "expected ']' to end array type")) {
return false;
}
} else {
*expr = elementType;
return true;
}
return true;
}
/**
* Parses a type expression.
*
* @param p The parser state.
* @param expr The type expression to populate.
* @return true if successful, false otherwise.
*/
static bool parse_type_expression(Parser* p, TypeExpression* expr) {
return parse_array_type_expression(p, expr);
}
/**
* Parses an alias declaration.
*
* @param p The parser state.
* @param module The module to add the alias to.
* @param is_public Whether the alias is public.
* @return true if successful, false otherwise.
*/
static bool parse_alias_declaration(Parser* p, Module* module, bool is_public) {
module->alias_count++;
module->aliases = realloc(module->aliases, sizeof(AliasDeclaration) * module->alias_count);
AliasDeclaration* alias = &module->aliases[module->alias_count - 1];
memset(alias, 0, sizeof(AliasDeclaration));
alias->is_public = is_public;
if (!parser_require(p, TOKEN_IDENTIFIER, "expected alias identifier")) {
return false;
}
alias->name = parser_to_text(p);
if (!parser_expect(p, TOKEN_ASSIGN, "expected '=' after alias name")) {
return false;
}
if (!parse_type_expression(p, &alias->value)) {
return false;
}
if (!parser_expect(p, TOKEN_SEMICOLON, "expected ';' after alias declaration")) {
return false;
}
return true;
}
Module* parser_parse(TokenStream* ts) {
Parser* p = malloc(sizeof(Parser));
p->ts = ts;
parser_next_token(p);
Module* module = malloc(sizeof(Module));
memset(module, 0, sizeof(Module));
if (!parse_module_declaration(p, module)) {
goto fail;
}
while (!parser_peek(p, TOKEN_EOF)) {
bool is_public = false;
bool terminal = false;
do {
if (parser_accept(p, TOKEN_IMPORT)) {
if (!parse_import_declaration(p, module, is_public)) {
goto fail;
}
terminal = true;
} else if (parser_accept(p, TOKEN_ALIAS)) {
if (!parse_alias_declaration(p, module, is_public)) {
goto fail;
}
terminal = true;
} else if (parser_accept(p, TOKEN_PUBLIC)) {
is_public = true;
} else {
log_on_line(&p->token.location, "unexpected token");
goto fail;
}
} while (!terminal);
}
free(p);
return module;
fail:
free(p);
parser_free(module);
return NULL;
}
void free_type_expression(TypeExpression* expr) {
if (expr->tag == TYPE_EXPRESSION_ARRAY) {
free_type_expression(expr->array.array);
free(expr->array.array);
}
}
void parser_free(Module* module) {
if (module == NULL) {
return;
}
if (module->imports != NULL) {
for(size_t i = 0; i < module->import_count; i++) {
free(module->imports[i].module_name);
}
free(module->imports);
}
if (module->aliases != NULL) {
for(size_t i = 0; i < module->alias_count; i++) {
free(module->aliases[i].name);
free_type_expression(&module->aliases[i].value);
}
free(module->aliases);
}
free(module->name);
free(module);
}
+1 -8
View File
@@ -10,13 +10,6 @@
* @param ts The TokenStream to read.
* @returns The parsed module.
*/
Module* parser_parse(TokenStream* ts);
/**
* Frees the parsed AST.
*
* @param module The AST return by parser_parse.
*/
void parser_free(Module* module);
ModuleTree* parser_parse(TokenStream* ts);
#endif
+52
View File
@@ -0,0 +1,52 @@
#include "internal.h"
#include "../str.h"
#include "../log.h"
#include <stdlib.h>
void parser_next_token(Parser* p) {
p->token = tokenstream_next(p->ts);
}
bool parser_accept(Parser* p, TokenType token) {
if (p->token.token == token) {
parser_next_token(p);
return true;
}
return false;
}
bool parser_expect(Parser* p, TokenType token, const char* msg) {
if (parser_accept(p, token)) {
return true;
}
log_on_line(&p->token.location, msg);
return false;
}
bool parser_peek(Parser* p, TokenType token) {
if (p->token.token == token) {
return true;
}
return false;
}
bool parser_require(Parser* p, TokenType token, const char* msg) {
if (parser_peek(p, token)) {
return true;
}
log_on_line(&p->token.location, msg);
return false;
}
char* parser_to_text(Parser* p) {
char* str = string_copy(p->token.text);
parser_next_token(p);
return str;
}
bool parser_accept_primitive(Parser* p) {
return parser_peek(p, TOKEN_I8) || parser_peek(p, TOKEN_I16) ||
parser_peek(p, TOKEN_I32) || parser_peek(p, TOKEN_I64) ||
parser_peek(p, TOKEN_U8) || parser_peek(p, TOKEN_U16) ||
parser_peek(p, TOKEN_U32) || parser_peek(p, TOKEN_U64);
}
+87
View File
@@ -0,0 +1,87 @@
#include "internal.h"
#include <stdlib.h>
#include <string.h>
bool parse_import_declaration(Parser* p, ModuleTree* module, bool is_public) {
module->import_count++;
module->imports = realloc(module->imports, sizeof(ImportTree) * module->import_count);
ImportTree* import = &module->imports[module->import_count - 1];
memset(import, 0, sizeof(ImportTree));
import->is_public = is_public;
if (!parser_require(p, TOKEN_IDENTIFIER, "expected module identifier")) {
return false;
}
import->module_name = parser_to_text(p);
if (!parser_expect(p, TOKEN_SEMICOLON, "expected ';' after import")) {
return false;
}
return true;
}
bool parse_alias_declaration(Parser* p, ModuleTree* module, bool is_public) {
(void)is_public;
module->alias_count++;
module->aliases = realloc(module->aliases, sizeof(AliasTree) * module->alias_count);
AliasTree* alias = &module->aliases[module->alias_count - 1];
memset(alias, 0, sizeof(AliasTree));
if (!parser_require(p, TOKEN_IDENTIFIER, "expected alias identifier")) {
return false;
}
alias->name = parser_to_text(p);
if (!parser_expect(p, TOKEN_ASSIGN, "expected '=' after alias name")) {
return false;
}
if (!parse_type_expression(p, &alias->value)) {
return false;
}
if (!parser_expect(p, TOKEN_SEMICOLON, "expected ';' after alias declaration")) {
return false;
}
return true;
}
bool parse_variable_declaration(Parser* p, ModuleTree* module, bool is_public, bool is_static, bool is_const) {
module->variable_count++;
module->variables = realloc(module->variables, sizeof(VariableTree) * module->variable_count);
VariableTree* var = &module->variables[module->variable_count - 1];
memset(var, 0, sizeof(VariableTree));
var->is_public = is_public;
var->is_static = is_static;
var->is_const = is_const;
if (parser_accept_primitive(p)) {
if (!parse_type_expression(p, &var->type)) {
return false;
}
}
if (!parser_require(p, TOKEN_IDENTIFIER, "expected variable identifier")) {
return false;
}
var->name = parser_to_text(p);
if (parser_accept(p, TOKEN_ASSIGN)) {
var->initializer = malloc(sizeof(ExpressionTree));
if (!parse_expression(p, var->initializer)) {
return false;
}
}
if (!parser_expect(p, TOKEN_SEMICOLON, "expected ';' after variable declaration")) {
return false;
}
return true;
}
+98
View File
@@ -0,0 +1,98 @@
#include "internal.h"
#include "../log.h"
#include <stdlib.h>
bool parse_primitive_type_expression(Parser* p, TypeTree* expr) {
if (parser_accept(p, TOKEN_U8)) {
expr->tag = TYPE_TREE_BUILTIN;
expr->builtin.bitSize = 8;
expr->builtin.isSigned = false;
return true;
} else if (parser_accept(p, TOKEN_U16)) {
expr->tag = TYPE_TREE_BUILTIN;
expr->builtin.bitSize = 16;
expr->builtin.isSigned = false;
return true;
} else if (parser_accept(p, TOKEN_U32)) {
expr->tag = TYPE_TREE_BUILTIN;
expr->builtin.bitSize = 32;
expr->builtin.isSigned = false;
return true;
} else if (parser_accept(p, TOKEN_U64)) {
expr->tag = TYPE_TREE_BUILTIN;
expr->builtin.bitSize = 64;
expr->builtin.isSigned = false;
return true;
} else if (parser_accept(p, TOKEN_I8)) {
expr->tag = TYPE_TREE_BUILTIN;
expr->builtin.bitSize = 8;
expr->builtin.isSigned = true;
return true;
} else if (parser_accept(p, TOKEN_I16)) {
expr->tag = TYPE_TREE_BUILTIN;
expr->builtin.bitSize = 16;
expr->builtin.isSigned = true;
return true;
} else if (parser_accept(p, TOKEN_I32)) {
expr->tag = TYPE_TREE_BUILTIN;
expr->builtin.bitSize = 32;
expr->builtin.isSigned = true;
return true;
} else if (parser_accept(p, TOKEN_I64)) {
expr->tag = TYPE_TREE_BUILTIN;
expr->builtin.bitSize = 64;
expr->builtin.isSigned = true;
return true;
} else {
log_on_line(&p->token.location, "expected type expression");
return false;
}
}
bool parse_array_type_expression(Parser* p, TypeTree* expr) {
TypeTree elementType;
if (!parse_primitive_type_expression(p, &elementType)) {
return false;
}
if (parser_accept(p, TOKEN_BRACKET_OPEN)) {
expr->tag = TYPE_TREE_ARRAY;
expr->array.array = malloc(sizeof(TypeTree));
*expr->array.array = elementType;
if (!parser_expect(p, TOKEN_BRACKET_CLOSE, "expected ']' to end array type")) {
return false;
}
} else {
*expr = elementType;
return true;
}
return true;
}
bool parse_type_expression(Parser* p, TypeTree* expr) {
return parse_array_type_expression(p, expr);
}
bool parse_expression(Parser* p, ExpressionTree* expr) {
if (parser_peek(p, TOKEN_INTEGER)) {
expr->tag = EXPRESSION_TREE_INTEGER;
expr->integer = atoi(p->token.text.data);
parser_next_token(p);
return true;
} else if (parser_peek(p, TOKEN_STRING)) {
expr->tag = EXPRESSION_TREE_STRING;
expr->string = parser_to_text(p);
return true;
} else if (parser_accept(p, TOKEN_TRUE)) {
expr->tag = EXPRESSION_TREE_BOOLEAN;
expr->boolean = true;
return true;
} else if (parser_accept(p, TOKEN_FALSE)) {
expr->tag = EXPRESSION_TREE_BOOLEAN;
expr->boolean = false;
return true;
}
log_on_line(&p->token.location, "expected expression");
return false;
}
+1
View File
@@ -0,0 +1 @@
PARSER_SRC := v0/parser/core.c v0/parser/expression.c v0/parser/declaration.c v0/parser/module.c
+36
View File
@@ -0,0 +1,36 @@
#ifndef PARSER_INTERNAL_H
#define PARSER_INTERNAL_H
#include "../parser.h"
#include "../token.h"
#include "../ast.h"
typedef struct {
TokenStream* ts;
Token token;
} Parser;
// Core functions
void parser_next_token(Parser* p);
bool parser_accept(Parser* p, TokenType token);
bool parser_expect(Parser* p, TokenType token, const char* msg);
bool parser_peek(Parser* p, TokenType token);
bool parser_require(Parser* p, TokenType token, const char* msg);
char* parser_to_text(Parser* p);
bool parser_accept_primitive(Parser* p);
// Base parsing (expressions, types)
bool parse_primitive_type_expression(Parser* p, TypeTree* expr);
bool parse_array_type_expression(Parser* p, TypeTree* expr);
bool parse_type_expression(Parser* p, TypeTree* expr);
bool parse_expression(Parser* p, ExpressionTree* expr);
// Declaration parsing
bool parse_import_declaration(Parser* p, ModuleTree* module, bool is_public);
bool parse_alias_declaration(Parser* p, ModuleTree* module, bool is_public);
bool parse_variable_declaration(Parser* p, ModuleTree* module, bool is_public, bool is_static, bool is_const);
// Module parsing
bool parse_module_declaration(Parser* p, ModuleTree* module);
#endif
+87
View File
@@ -0,0 +1,87 @@
#include "internal.h"
#include "../log.h"
#include <stdlib.h>
#include <string.h>
bool parse_module_declaration(Parser* p, ModuleTree* module) {
if (!parser_expect(p, TOKEN_MODULE, "expected keyword 'module'")) {
return false;
}
if (!parser_require(p, TOKEN_IDENTIFIER, "expected module identifier")) {
return false;
}
module->name = parser_to_text(p);
return parser_expect(p, TOKEN_SEMICOLON, "expected ';' after module name");
}
ModuleTree* parser_parse(TokenStream* ts) {
Parser* p = malloc(sizeof(Parser));
p->ts = ts;
parser_next_token(p);
ModuleTree* module = malloc(sizeof(ModuleTree));
memset(module, 0, sizeof(ModuleTree));
if (!parse_module_declaration(p, module)) {
goto fail;
}
while (!parser_peek(p, TOKEN_EOF)) {
bool is_public = false;
bool is_static = false;
bool is_const = false;
bool terminal = false;
while (!terminal) {
if (parser_accept(p, TOKEN_IMPORT)) {
if (is_static) {
log_on_line(&p->token.location, "import declarations cannot be static or const");
goto fail;
}
if (is_const) {
log_on_line(&p->token.location, "import declarations cannot be static or const");
goto fail;
}
if (!parse_import_declaration(p, module, is_public)) {
goto fail;
}
terminal = true;
} else if (parser_accept(p, TOKEN_ALIAS)) {
if (is_static) {
log_on_line(&p->token.location, "alias declarations cannot be static or const");
goto fail;
}
if (is_const) {
log_on_line(&p->token.location, "alias declarations cannot be static or const");
goto fail;
}
if (!parse_alias_declaration(p, module, is_public)) {
goto fail;
}
terminal = true;
} else if (parser_accept(p, TOKEN_PUBLIC)) {
is_public = true;
} else if (parser_accept(p, TOKEN_STATIC)) {
is_static = true;
} else if (parser_accept(p, TOKEN_CONST)) {
is_const = true;
} else if (parser_accept(p, TOKEN_VAR) || parser_accept_primitive(p)) {
if (!parse_variable_declaration(p, module, is_public, is_static, is_const)) {
goto fail;
}
terminal = true;
} else {
log_on_line(&p->token.location, "unexpected token");
goto fail;
}
}
}
free(p);
return module;
fail:
free(p);
ast_free_module(module);
return NULL;
}
+8
View File
@@ -0,0 +1,8 @@
#include "../test.h"
#include "../parser.h"
// Currently core utilities are tested indirectly through other parser tests.
// Placeholder for future explicit core utility tests.
static void test_parser_core_placeholder(void) {
// No-op
}
+89
View File
@@ -0,0 +1,89 @@
#include "../test.h"
#include "../parser.h"
#include <string.h>
#include <stdlib.h>
static void test_parser_missing_semicolon_import(void) {
test_get_ast();
assert_log_file("expected error for missing semicolon");
}
static void test_parser_bad_import_name(void) {
test_get_ast();
assert_log_file("expected error for bad import name");
}
static void test_parser_imports(void) {
ModuleTree* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_str("my_module", m->name, "expected name 'my_module'");
assert_not_null(m->imports, "expected imports to be parsed");
assert_int(1, (int)m->import_count, "expected one import");
assert_str("other_module", m->imports[0].module_name, "expected import name 'other_module'");
assert_false(m->imports[0].is_public, "expected import to not be public");
}
static void test_parser_public_imports(void) {
ModuleTree* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_str("my_module", m->name, "expected name 'my_module'");
assert_not_null(m->imports, "expected imports to be parsed");
assert_int(1, (int)m->import_count, "expected one import");
assert_str("other_module", m->imports[0].module_name, "expected import name 'other_module'");
assert_true(m->imports[0].is_public, "expected import to be public");
}
static void test_parser_alias_simple(void) {
ModuleTree* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_int(1, (int)m->alias_count, "expected correct number of aliases");
AliasTree alias = m->aliases[0];
assert_str("myalias", alias.name, "expected correct alias name");
}
static void test_parser_variable_simple(void) {
ModuleTree* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_int(1, (int)m->variable_count, "expected correct number of variables");
VariableTree var = m->variables[0];
assert_str("my_var", var.name, "expected correct variable name");
assert_false(var.is_const, "expected not const");
assert_false(var.is_static, "expected not static");
}
static void test_parser_variable_const(void) {
ModuleTree* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_int(1, (int)m->variable_count, "expected correct number of variables");
VariableTree var = m->variables[0];
assert_str("my_const", var.name, "expected correct variable name");
assert_true(var.is_const, "expected const");
assert_false(var.is_static, "expected not static");
}
static void test_parser_variable_static(void) {
ModuleTree* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_int(1, (int)m->variable_count, "expected correct number of variables");
VariableTree var = m->variables[0];
assert_str("my_static", var.name, "expected correct variable name");
assert_false(var.is_const, "expected not const");
assert_true(var.is_static, "expected static");
}
static void test_parser_multiple_vars(void) {
ModuleTree* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_int(2, (int)m->variable_count, "expected correct number of variables");
assert_str("var1", m->variables[0].name, "expected first variable name 'var1'");
assert_str("var2", m->variables[1].name, "expected second variable name 'var2'");
}
+52
View File
@@ -0,0 +1,52 @@
#include "../test.h"
#include "../parser.h"
#include <string.h>
#include <stdlib.h>
static void test_parser_alias_simple_type(void) {
ModuleTree* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_int(1, (int)m->alias_count, "expected correct number of aliases");
AliasTree alias = m->aliases[0];
assert_int(TYPE_TREE_BUILTIN, alias.value.tag, "expected correct alias tag");
assert_int(32, alias.value.builtin.bitSize, "expected bitSize 32");
assert_true(alias.value.builtin.isSigned, "expected signed");
}
static void test_parser_alias_array(void) {
ModuleTree* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_int(1, (int)m->alias_count, "expected correct number of aliases");
AliasTree alias = m->aliases[0];
assert_int(TYPE_TREE_ARRAY, alias.value.tag, "expected correct alias tag");
TypeTree* valueType = alias.value.array.array;
assert_not_null(valueType, "expected pointer to array type");
assert_int(TYPE_TREE_BUILTIN, valueType->tag, "expected correct type tag");
assert_int(32, valueType->builtin.bitSize, "expected bitSize 32");
assert_true(valueType->builtin.isSigned, "expected signed");
}
static void test_parser_variable_init(void) {
ModuleTree* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_int(1, (int)m->variable_count, "expected 1 variable");
VariableTree* var = &m->variables[0];
assert_str("x", var->name, "expected variable name 'x'");
assert_not_null(var->initializer, "expected variable to have an initializer");
assert_int(EXPRESSION_TREE_INTEGER, var->initializer->tag, "expected integer initializer");
assert_int(123, var->initializer->integer, "expected value 123");
}
static void test_parser_variable_simple_type(void) {
ModuleTree* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_int(1, (int)m->variable_count, "expected correct number of variables");
VariableTree var = m->variables[0];
assert_int(TYPE_TREE_BUILTIN, var.type.tag, "expected correct type tag");
assert_int(32, var.type.builtin.bitSize, "expected bitSize 32");
assert_true(var.type.builtin.isSigned, "expected signed");
}
+21
View File
@@ -0,0 +1,21 @@
#include "../test.h"
#include "../parser.h"
#include <string.h>
#include <stdlib.h>
static void test_parser_module_name(void) {
ModuleTree* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_str("my_module", m->name, "expected name 'my_module'");
}
static void test_parser_bad_module_name(void) {
test_get_ast();
assert_log_file("expected error to be logged for bad module name");
}
static void test_parser_missing_semicolon_module(void) {
test_get_ast();
assert_log_file("expected error for missing semicolon");
}
+22 -12
View File
@@ -14,7 +14,7 @@ static char* s_logOutput = NULL;
static const char* s_currentTestName = NULL;
static char* s_testSource = NULL;
static Module* s_currentModule = NULL;
static ModuleTree* s_currentModule = NULL;
static TokenStream* s_currentTokenStream = NULL;
void fail(const char* msg) {
@@ -90,7 +90,7 @@ TokenStream* test_get_tokenstream(void) {
return s_currentTokenStream;
}
Module* test_get_ast(void) {
ModuleTree* test_get_ast(void) {
if (s_currentModule == NULL) {
s_currentModule = parser_parse(test_get_tokenstream());
}
@@ -189,7 +189,10 @@ typedef struct {
} TestCase;
#include "test_token.c"
#include "test_parser.c"
#include "parser/test_module.c"
#include "parser/test_declaration.c"
#include "parser/test_expression.c"
#include "parser/test_core.c"
#include "test_log.c"
static int s_totalTests;
@@ -201,16 +204,23 @@ static TestCase s_tests[] = {
TEST(test_log_error)
TEST(test_log_on_line_variadic)
TEST(test_log_on_line)
TEST(test_parser_alias_and_import_mix)
TEST(test_parser_alias_array)
TEST(test_parser_alias_simple)
TEST(test_parser_bad_import_name)
TEST(test_parser_bad_module_name)
TEST(test_parser_imports)
TEST(test_parser_missing_semicolon_import)
TEST(test_parser_missing_semicolon_module)
TEST(test_parser_module_name)
TEST(test_parser_bad_module_name)
TEST(test_parser_missing_semicolon_module)
TEST(test_parser_missing_semicolon_import)
TEST(test_parser_bad_import_name)
TEST(test_parser_imports)
TEST(test_parser_public_imports)
TEST(test_parser_alias_simple)
TEST(test_parser_alias_simple_type)
TEST(test_parser_alias_array)
TEST(test_parser_variable_simple)
TEST(test_parser_variable_simple_type)
TEST(test_parser_variable_const)
TEST(test_parser_variable_init)
TEST(test_parser_variable_static)
TEST(test_parser_multiple_vars)
TEST(test_parser_core_placeholder)
TEST(test_tokenstream_comma)
TEST(test_tokenstream_info)
TEST(test_tokenstream_keywords_and_symbols)
@@ -267,7 +277,7 @@ int main(int argc, char** argv) {
// Free AST and TokenStream after each test
if (s_currentModule) {
parser_free(s_currentModule);
ast_free_module(s_currentModule);
s_currentModule = NULL;
}
if (s_currentTokenStream) {
+1 -1
View File
@@ -89,6 +89,6 @@ TokenStream* test_get_tokenstream(void);
*
* At the end of the test, the AST will be freed automatically by the test harness.
*/
Module* test_get_ast(void);
ModuleTree* test_get_ast(void);
#endif
+63
View File
@@ -0,0 +1,63 @@
#define _DEFAULT_SOURCE
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <dirent.h>
int run_test(const char* dir_name) {
char cmd[2048];
char input_path[1024];
char expected_path[1024];
snprintf(input_path, sizeof(input_path), "v0/integration_tests/%s/input.c2", dir_name);
snprintf(expected_path, sizeof(expected_path), "v0/integration_tests/%s/expected.c", dir_name);
if (snprintf(cmd, sizeof(cmd), "./v0/bin/c2 %s > actual.c", input_path) >= sizeof(cmd)) {
printf("Command buffer too small for %s\n", dir_name);
return 1;
}
if (system(cmd) != 0) {
printf("Failed to run compiler for %s\n", dir_name);
return 1;
}
if (snprintf(cmd, sizeof(cmd), "diff -u %s actual.c", expected_path) >= sizeof(cmd)) {
printf("Command buffer too small for %s\n", dir_name);
return 1;
}
if (system(cmd) != 0) {
printf("Test %s failed: Output mismatch\n", dir_name);
return 1;
}
printf("Test %s passed\n", dir_name);
return 0;
}
int main() {
DIR* d = opendir("v0/integration_tests");
if (!d) {
perror("opendir");
return 1;
}
struct dirent* dir;
int passed = 0;
int failed = 0;
while ((dir = readdir(d)) != NULL) {
if (dir->d_type == DT_DIR && strcmp(dir->d_name, ".") != 0 && strcmp(dir->d_name, "..") != 0) {
if (run_test(dir->d_name) == 0) {
passed++;
} else {
failed++;
}
}
}
closedir(d);
printf("\nTotal tests: %d, Passed: %d, Failed: %d\n", passed + failed, passed, failed);
return failed > 0 ? 1 : 0;
}
-95
View File
@@ -1,95 +0,0 @@
#include "test.h"
#include "parser.h"
#include <string.h>
#include <stdlib.h>
static void test_parser_module_name(void) {
Module* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_str("my_module", m->name, "expected name 'my_module'");
}
static void test_parser_bad_module_name(void) {
test_get_ast();
assert_log_file("expected error to be logged for bad module name");
}
static void test_parser_missing_semicolon_module(void) {
test_get_ast();
assert_log_file("expected error for missing semicolon");
}
static void test_parser_missing_semicolon_import(void) {
test_get_ast();
assert_log_file("expected error for missing semicolon");
}
static void test_parser_bad_import_name(void) {
test_get_ast();
assert_log_file("expected error for bad import name");
}
static void test_parser_imports(void) {
Module* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_str("my_module", m->name, "expected name 'my_module'");
assert_not_null(m->imports, "expected imports to be parsed");
assert_int(1, (int)m->import_count, "expected one import");
assert_str("other_module", m->imports[0].module_name, "expected import name 'other_module'");
assert_false(m->imports[0].is_public, "expected import to not be public");
}
static void test_parser_public_imports(void) {
Module* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_str("my_module", m->name, "expected name 'my_module'");
assert_not_null(m->imports, "expected imports to be parsed");
assert_int(1, (int)m->import_count, "expected one import");
assert_str("other_module", m->imports[0].module_name, "expected import name 'other_module'");
assert_true(m->imports[0].is_public, "expected import to be public");
}
static void test_parser_alias_simple(void) {
Module* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_int(1, (int)m->alias_count, "expected correct number of aliases");
AliasDeclaration alias = m->aliases[0];
assert_str("myalias", alias.name, "expected correct alias name");
assert_int(TYPE_EXPRESSION_BUILTIN, alias.value.tag, "expected correct alias tag");
assert_int(32, alias.value.builtin.bitSize, "expected bitSize 32");
assert_true(alias.value.builtin.isSigned, "expected signed");
}
static void test_parser_alias_array(void) {
Module* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_int(1, (int)m->alias_count, "expected correct number of aliases");
AliasDeclaration alias = m->aliases[0];
assert_str("myalias", alias.name, "expected correct alias name");
assert_int(TYPE_EXPRESSION_ARRAY, alias.value.tag, "expected correct alias tag");
TypeExpression* valueType = alias.value.array.array;
assert_not_null(valueType, "expected pointer to array type");
assert_int(TYPE_EXPRESSION_BUILTIN, valueType->tag, "expected correct type tag");
assert_int(32, valueType->builtin.bitSize, "expected bitSize 32");
assert_true(valueType->builtin.isSigned, "expected signed");
}
static void test_parser_alias_and_import_mix(void) {
Module* m = test_get_ast();
assert_not_null(m, "expected module to be parsed");
assert_int(2, (int)m->import_count, "expected 2 imports");
assert_int(2, (int)m->alias_count, "expected 2 aliases");
assert_str("foo", m->imports[0].module_name, "expected import 1 name 'foo'");
assert_str("bar", m->imports[1].module_name, "expected import 2 name 'bar'");
assert_str("myalias", m->aliases[0].name, "expected alias 1 name 'myalias'");
assert_str("otheralias", m->aliases[1].name, "expected alias 2 name 'otheralias'");
}
+3
View File
@@ -0,0 +1,3 @@
module mymodule;
alias myalias = i32;
+4
View File
@@ -0,0 +1,4 @@
module test_multiple_vars;
i32 var1;
i32 var2;
+3
View File
@@ -0,0 +1,3 @@
module test_const_var;
const i32 my_const;
+2
View File
@@ -0,0 +1,2 @@
module mymodule;
var x = 123;
+4
View File
@@ -0,0 +1,4 @@
module my_module;
// Defines a global variable called my_var.
i32 my_var;
+4
View File
@@ -0,0 +1,4 @@
module my_module;
// Defines a global variable called my_var.
i32 my_var;
+3
View File
@@ -0,0 +1,3 @@
module test_static_var;
static i32 my_static;
+21
View File
@@ -44,6 +44,8 @@ static const KeywordMap keywords[] = {
{"u16", TOKEN_U16},
{"u32", TOKEN_U32},
{"u64", TOKEN_U64},
{"true", TOKEN_TRUE},
{"false", TOKEN_FALSE},
};
/**
@@ -234,8 +236,27 @@ Token tokenstream_next(TokenStream* ts) {
case ',': return create_token(ts, TOKEN_COMMA, start_text, 1, start_line, start_column, line_start);
case ';': return create_token(ts, TOKEN_SEMICOLON, start_text, 1, start_line, start_column, line_start);
case '=': return create_token(ts, TOKEN_ASSIGN, start_text, 1, start_line, start_column, line_start);
case '"': {
size_t len = 0;
const char* start = &ts->code[ts->pos];
while (peek_char(ts) != '"' && peek_char(ts) != '\0') {
read_char(ts);
len++;
}
if (peek_char(ts) == '"') read_char(ts);
return create_token(ts, TOKEN_STRING, start, len, start_line, start_column + 1, line_start);
}
}
if (isdigit(c)) {
size_t len = 1;
while (isdigit(peek_char(ts))) {
read_char(ts);
len++;
}
return create_token(ts, TOKEN_INTEGER, start_text, len, start_line, start_column, line_start);
}
/* Keywords and identifiers */
if (is_identifier_start(c)) {
/* Declarations first for C89 */
+4
View File
@@ -38,6 +38,10 @@ typedef enum {
TOKEN_U16,
TOKEN_U32,
TOKEN_U64,
TOKEN_STRING,
TOKEN_INTEGER,
TOKEN_TRUE,
TOKEN_FALSE,
/* Variable */
TOKEN_IDENTIFIER,