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
parent 4bd66ea216
commit ea55dedd07
20 changed files with 656 additions and 637 deletions
@@ -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);
+}
@@ -0,0 +1,87 @@
+#include "internal.h"
+#include <stdlib.h>
+#include <string.h>
+
+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;
+}
+
+bool parse_alias_declaration(Parser* p, Module* module, bool is_public) {
+    (void)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));
+
+    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, Module* module, bool is_public, bool is_static, bool is_const) {
+    module->variable_count++;
+    module->variables = realloc(module->variables, sizeof(VariableDeclaration) * module->variable_count);
+
+    VariableDeclaration* var = &module->variables[module->variable_count - 1];
+    memset(var, 0, sizeof(VariableDeclaration));
+    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(Expression));
+        if (!parse_expression(p, var->initializer)) {
+            return false;
+        }
+    }
+
+    if (!parser_expect(p, TOKEN_SEMICOLON, "expected ';' after variable declaration")) {
+        return false;
+    }
+
+    return true;
+}
@@ -0,0 +1,105 @@
+#include "internal.h"
+#include "../log.h"
+#include <stdlib.h>
+
+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;
+    }
+}
+
+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;
+}
+
+bool parse_type_expression(Parser* p, TypeExpression* expr) {
+    return parse_array_type_expression(p, expr);
+}
+
+bool parse_expression(Parser* p, Expression* expr) {
+    if (parser_peek(p, TOKEN_INTEGER)) {
+        expr->tag = EXPRESSION_INTEGER;
+        expr->integer = atoi(p->token.text.data);
+        parser_next_token(p);
+        return true;
+    } else if (parser_peek(p, TOKEN_STRING)) {
+        expr->tag = EXPRESSION_STRING;
+        expr->string = parser_to_text(p);
+        return true;
+    } else if (parser_accept(p, TOKEN_TRUE)) {
+        expr->tag = EXPRESSION_BOOLEAN;
+        expr->boolean = true;
+        return true;
+    } else if (parser_accept(p, TOKEN_FALSE)) {
+        expr->tag = EXPRESSION_BOOLEAN;
+        expr->boolean = false;
+        return true;
+    }
+    log_on_line(&p->token.location, "expected expression");
+    return false;
+}
+
+void free_type_expression(TypeExpression* expr) {
+    if (expr->tag == TYPE_EXPRESSION_ARRAY) {
+        free_type_expression(expr->array.array);
+        free(expr->array.array);
+    }
+}
@@ -0,0 +1 @@
+PARSER_SRC := v0/parser/core.c v0/parser/expression.c v0/parser/declaration.c v0/parser/module.c
@@ -0,0 +1,37 @@
+#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, TypeExpression* expr);
+bool parse_array_type_expression(Parser* p, TypeExpression* expr);
+bool parse_type_expression(Parser* p, TypeExpression* expr);
+bool parse_expression(Parser* p, Expression* expr);
+void free_type_expression(TypeExpression* expr);
+
+// Declaration parsing
+bool parse_import_declaration(Parser* p, Module* module, bool is_public);
+bool parse_alias_declaration(Parser* p, Module* module, bool is_public);
+bool parse_variable_declaration(Parser* p, Module* module, bool is_public, bool is_static, bool is_const);
+
+// Module parsing
+bool parse_module_declaration(Parser* p, Module* module);
+
+#endif
@@ -0,0 +1,125 @@
+#include "internal.h"
+#include "../log.h"
+#include <stdlib.h>
+#include <string.h>
+
+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");
+}
+
+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 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);
+    parser_free(module);
+    return NULL;
+}
+
+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((void*)module->aliases[i].name);
+            free_type_expression(&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);
+            free_type_expression(&module->variables[i].type);
+            if (module->variables[i].initializer) {
+                if (module->variables[i].initializer->tag == EXPRESSION_STRING) {
+                    free((void*)module->variables[i].initializer->string);
+                }
+                free(module->variables[i].initializer);
+            }
+        }
+        free(module->variables);
+    }
+
+    free(module->name);
+    free(module);
+}
@@ -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
+}
@@ -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) {
+	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");
+}
+
+static void test_parser_variable_simple(void) {
+	Module* 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");
+	VariableDeclaration 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) {
+	Module* 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");
+	VariableDeclaration 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) {
+	Module* 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");
+	VariableDeclaration 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) {
+	Module* 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'");
+}
@@ -0,0 +1,52 @@
+#include "../test.h"
+#include "../parser.h"
+#include <string.h>
+#include <stdlib.h>
+
+static void test_parser_alias_simple_type(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_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_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_variable_init(void) {
+	Module* m = test_get_ast();
+    
+	assert_not_null(m, "expected module to be parsed");
+    assert_int(1, (int)m->variable_count, "expected 1 variable");
+    VariableDeclaration* 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_INTEGER, var->initializer->tag, "expected integer initializer");
+    assert_int(123, var->initializer->integer, "expected value 123");
+}
+
+static void test_parser_variable_simple_type(void) {
+	Module* 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");
+	VariableDeclaration var = m->variables[0];
+	assert_int(TYPE_EXPRESSION_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");
+}
@@ -0,0 +1,21 @@
+#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");
+}
				`@@ -0,0 +1 @@`
				`PARSER_SRC := v0/parser/core.c v0/parser/expression.c v0/parser/declaration.c v0/parser/module.c`