From da12345b6e3c297e530764c4424b7d9e6a1cb65e Mon Sep 17 00:00:00 2001 From: Zuhaitz Méndez Fernández de Aránguiz Date: Wed, 28 Jan 2026 01:19:20 +0000 Subject: Fix for #141 --- src/parser/parser_core.c | 2 +- src/parser/parser_expr.c | 37 +++++++++++++++++++++++++++++++++++-- src/parser/parser_stmt.c | 6 +++--- 3 files changed, 39 insertions(+), 6 deletions(-) (limited to 'src/parser') diff --git a/src/parser/parser_core.c b/src/parser/parser_core.c index e9a418e..ba6f2fe 100644 --- a/src/parser/parser_core.c +++ b/src/parser/parser_core.c @@ -700,7 +700,7 @@ static ASTNode *generate_derive_impls(ParserContext *ctx, ASTNode *strct, char * { // Simplistic Debug for now, I know. code = xmalloc(1024); - sprintf(code, "impl %s { fn to_string(self) -> char* { return \"%s { ... }\"; } }", + sprintf(code, "impl %s { fn to_string(self) -> char* { return \"%s {{ ... }}\"; } }", name, name); } else if (0 == strcmp(trait, "Copy")) diff --git a/src/parser/parser_expr.c b/src/parser/parser_expr.c index df2dcf6..332d9c3 100644 --- a/src/parser/parser_expr.c +++ b/src/parser/parser_expr.c @@ -935,6 +935,17 @@ static ASTNode *create_fstring_block(ParserContext *ctx, const char *content) free(txt); } + // Handle escape {{ + if (brace[1] == '{') + { + ASTNode *cat = ast_create(NODE_RAW_STMT); + cat->raw_stmt.content = xstrdup("strcat(_b, \"{\");"); + tail->next = cat; + tail = cat; + cur = brace + 2; + continue; + } + char *end_brace = strchr(brace, '}'); if (!end_brace) { @@ -1100,8 +1111,30 @@ static ASTNode *parse_float_literal(Token t) } // Parse string literal -static ASTNode *parse_string_literal(Token t) +static ASTNode *parse_string_literal(ParserContext *ctx, Token t) { + // Check for implicit interpolation + int has_interpolation = 0; + for (int i = 1; i < t.len - 1; i++) + { + if (t.start[i] == '{') + { + has_interpolation = 1; + break; + } + } + + if (has_interpolation) + { + + char *inner = xmalloc(t.len); + strncpy(inner, t.start + 1, t.len - 2); + inner[t.len - 2] = 0; + ASTNode *node = create_fstring_block(ctx, inner); + free(inner); + return node; + } + ASTNode *node = ast_create(NODE_EXPR_LITERAL); node->literal.type_kind = LITERAL_STRING; node->literal.string_val = xmalloc(t.len); @@ -1276,7 +1309,7 @@ ASTNode *parse_primary(ParserContext *ctx, Lexer *l) } else if (t.type == TOK_STRING) { - node = parse_string_literal(t); + node = parse_string_literal(ctx, t); } else if (t.type == TOK_FSTRING) { diff --git a/src/parser/parser_stmt.c b/src/parser/parser_stmt.c index ab5b89c..85e9825 100644 --- a/src/parser/parser_stmt.c +++ b/src/parser/parser_stmt.c @@ -1877,7 +1877,7 @@ ASTNode *parse_statement(ParserContext *ctx, Lexer *l) lexer_next(&lookahead); TokenType next_type = lexer_peek(&lookahead).type; - if (next_type == TOK_SEMICOLON || next_type == TOK_DOTDOT) + if (next_type == TOK_SEMICOLON || next_type == TOK_DOTDOT || next_type == TOK_RBRACE) { Token t = lexer_next(l); // consume string @@ -1894,8 +1894,7 @@ ASTNode *parse_statement(ParserContext *ctx, Lexer *l) inner[t.len - 2] = 0; } - // ; means println (end of line), .. means print (continuation) - int is_ln = (next_type == TOK_SEMICOLON); + int is_ln = (next_type == TOK_SEMICOLON || next_type == TOK_RBRACE); char **used_syms = NULL; int used_count = 0; char *code = @@ -1913,6 +1912,7 @@ ASTNode *parse_statement(ParserContext *ctx, Lexer *l) lexer_next(l); // consume optional ; } } + // If TOK_RBRACE, do not consume it, so parse_block can see it and terminate loop. ASTNode *n = ast_create(NODE_RAW_STMT); // Append semicolon to Statement Expression to make it a valid statement -- cgit v1.2.3 From 279bc13582cb681867bc4ebadb4287449a2c7383 Mon Sep 17 00:00:00 2001 From: Zuhaitz Méndez Fernández de Aránguiz Date: Wed, 28 Jan 2026 10:31:05 +0000 Subject: Fix for #145 --- src/parser/parser_expr.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'src/parser') diff --git a/src/parser/parser_expr.c b/src/parser/parser_expr.c index 332d9c3..ee1f96f 100644 --- a/src/parser/parser_expr.c +++ b/src/parser/parser_expr.c @@ -3625,7 +3625,7 @@ ASTNode *parse_expr_prec(ParserContext *ctx, Lexer *l, Precedence min_prec) } else if (t->kind == TYPE_STRING) { - strcat(fmt, "%s"); + strcat(fmt, "%ms"); } else if (t->kind == TYPE_CHAR || t->kind == TYPE_I8 || t->kind == TYPE_U8 || t->kind == TYPE_BYTE) -- cgit v1.2.3 From f8d9b233952357d327e856100835adf3cef47f23 Mon Sep 17 00:00:00 2001 From: Zuhaitz Méndez Fernández de Aránguiz Date: Wed, 28 Jan 2026 15:58:55 +0000 Subject: Fix for #148 --- src/parser/parser_stmt.c | 8 +------- 1 file changed, 1 insertion(+), 7 deletions(-) (limited to 'src/parser') diff --git a/src/parser/parser_stmt.c b/src/parser/parser_stmt.c index 85e9825..a471fe6 100644 --- a/src/parser/parser_stmt.c +++ b/src/parser/parser_stmt.c @@ -462,13 +462,7 @@ ASTNode *parse_defer(ParserContext *ctx, Lexer *l) } else { - s = ast_create(NODE_RAW_STMT); - char *raw_content = consume_and_rewrite(ctx, l); - // consume_and_rewrite strips the semicolon, so we must add it back for proper C generation - char *safe_content = xmalloc(strlen(raw_content) + 2); - sprintf(safe_content, "%s;", raw_content); - free(raw_content); - s->raw_stmt.content = safe_content; + s = parse_statement(ctx, l); } ctx->in_defer_block = prev_in_defer; -- cgit v1.2.3 From fd692ab7bb9f7b1e8f5d878a16154a4a03d0f6f9 Mon Sep 17 00:00:00 2001 From: Zuhaitz Méndez Fernández de Aránguiz Date: Wed, 28 Jan 2026 19:52:37 +0000 Subject: Opaque structs and aliases + some improvements --- README.md | 72 ++++++++++++++++++- src/ast/ast.c | 17 +++++ src/ast/ast.h | 10 +++ src/lexer/token.c | 4 ++ src/lsp/lsp_analysis.c | 71 ++++++++++++++++++- src/lsp/lsp_index.c | 52 +++++++++++++- src/parser/parser.h | 26 +++---- src/parser/parser_core.c | 30 ++++++-- src/parser/parser_decl.c | 64 ++++++++++++++++- src/parser/parser_expr.c | 133 +++++++++++++++++++++++++++++++++++- src/parser/parser_struct.c | 12 ++-- src/parser/parser_type.c | 19 +++++- src/parser/parser_utils.c | 14 +++- src/zprep.h | 1 + tests/features/_opaque_alias_lib.zc | 13 ++++ tests/features/_opaque_lib.zc | 15 ++++ tests/features/test_opaque.zc | 18 +++++ tests/features/test_opaque_alias.zc | 13 ++++ 18 files changed, 549 insertions(+), 35 deletions(-) create mode 100644 tests/features/_opaque_alias_lib.zc create mode 100644 tests/features/_opaque_lib.zc create mode 100644 tests/features/test_opaque.zc create mode 100644 tests/features/test_opaque_alias.zc (limited to 'src/parser') diff --git a/README.md b/README.md index 9550208..a35e0de 100644 --- a/README.md +++ b/README.md @@ -43,9 +43,11 @@ Join the discussion, share demos, ask questions, or report bugs in the official - [Arrays](#arrays) - [Tuples](#tuples) - [Structs](#structs) + - [Opaque Structs](#opaque-structs) - [Enums](#enums) - [Unions](#unions) - [Type Aliases](#type-aliases) + - [Opaque Type Aliases](#opaque-type-aliases) - [4. Functions & Lambdas](#4-functions--lambdas) - [Functions](#functions) - [Const Arguments](#const-arguments) @@ -91,6 +93,7 @@ Join the discussion, share demos, ask questions, or report bugs in the official - [Volatile](#volatile) - [Named Constraints](#named-constraints) - [15. Build Directives](#15-build-directives) + - [16. Keywords](#16-keywords) - [Standard Library](#standard-library) - [Tooling](#tooling) - [Language Server (LSP)](#language-server-lsp) @@ -261,6 +264,29 @@ struct Flags { > **Note**: Structs use [Move Semantics](#move-semantics--copy-safety) by default. Fields can be accessed via `.` even on pointers (Auto-Dereference). +#### Opaque Structs +You can define a struct as `opaque` to restrict access to its fields to the defining module only, while still allowing the struct to be allocated on the stack (size is known). + +```zc +// In user.zc +opaque struct User { + id: int; + name: string; +} + +fn new_user(name: string) -> User { + return User{id: 1, name: name}; // OK: Inside module +} + +// In main.zc +import "user.zc"; + +fn main() { + let u = new_user("Alice"); + // let id = u.id; // Error: Cannot access private field 'id' +} +``` + #### Enums Tagged unions (Sum types) capable of holding data. ```zc @@ -287,6 +313,27 @@ alias ID = int; alias PointMap = Map; ``` +#### Opaque Type Aliases +You can define a type alias as `opaque` to create a new type that is distinct from its underlying type outside of the defining module. This provides strong encapsulation and type safety without the runtime overhead of a wrapper struct. + +```zc +// In library.zc +opaque alias Handle = int; + +fn make_handle(v: int) -> Handle { + return v; // Implicit conversion allowed inside module +} + +// In main.zc +import "library.zc"; + +fn main() { + let h: Handle = make_handle(42); + // let i: int = h; // Error: Type validation failed + // let h2: Handle = 10; // Error: Type validation failed +} +``` + ### 4. Functions & Lambdas #### Functions @@ -980,6 +1027,29 @@ import "raylib.h" fn main() { ... } ``` +### 16. Keywords + +The following keywords are reserved in Zen C. + +#### Declarations +`alias`, `def`, `enum`, `fn`, `impl`, `import`, `let`, `module`, `opaque`, `struct`, `trait`, `union`, `use` + +#### Control Flow +`async`, `await`, `break`, `catch`, `continue`, `defer`, `else`, `for`, `goto`, `guard`, `if`, `loop`, `match`, `return`, `try`, `unless`, `while` + +#### Special +`asm`, `assert`, `autofree`, `comptime`, `const`, `embed`, `launch`, `ref`, `sizeof`, `static`, `test`, `volatile` + +#### Constants +`true`, `false`, `null` + +#### C Reserved +The following identifiers are reserved because they are keywords in C11: +`auto`, `case`, `char`, `default`, `do`, `double`, `extern`, `float`, `inline`, `int`, `long`, `register`, `restrict`, `short`, `signed`, `switch`, `typedef`, `unsigned`, `void`, `_Atomic`, `_Bool`, `_Complex`, `_Generic`, `_Imaginary`, `_Noreturn`, `_Static_assert`, `_Thread_local` + +#### Operators +`and`, `or` + --- ## Standard Library @@ -1190,7 +1260,7 @@ fn add_kernel(a: float*, b: float*, c: float*, n: int) { } fn main() { - const N = 1024; + def N = 1024; let d_a = cuda_alloc(N); let d_b = cuda_alloc(N); let d_c = cuda_alloc(N); diff --git a/src/ast/ast.c b/src/ast/ast.c index 0799845..f4922a6 100644 --- a/src/ast/ast.c +++ b/src/ast/ast.c @@ -168,6 +168,18 @@ int type_eq(Type *a, Type *b) { return 0 == strcmp(a->name, b->name); } + if (a->kind == TYPE_ALIAS) + { + if (a->alias.is_opaque_alias) + { + if (b->kind != TYPE_ALIAS || !b->alias.is_opaque_alias) + { + return 0; + } + return 0 == strcmp(a->name, b->name); + } + return type_eq(a->inner, b); + } if (a->kind == TYPE_POINTER || a->kind == TYPE_ARRAY) { return type_eq(a->inner, b->inner); @@ -340,6 +352,8 @@ static char *type_to_string_impl(Type *t) } return xstrdup(t->name); } + case TYPE_ALIAS: + return xstrdup(t->name); default: return xstrdup("unknown"); @@ -524,6 +538,9 @@ static char *type_to_c_string_impl(Type *t) case TYPE_GENERIC: return xstrdup(t->name); + case TYPE_ALIAS: + return type_to_c_string(t->inner); + case TYPE_ENUM: return xstrdup(t->name); diff --git a/src/ast/ast.h b/src/ast/ast.h index b272cae..a868bf0 100644 --- a/src/ast/ast.h +++ b/src/ast/ast.h @@ -58,6 +58,7 @@ typedef enum TYPE_ARRAY, ///< Fixed size array [N]. TYPE_FUNCTION, ///< Function pointer or reference. TYPE_GENERIC, ///< Generic type parameter (T). + TYPE_ALIAS, ///< Opaque type alias. TYPE_UNKNOWN ///< Unknown/unresolved type. } TypeKind; @@ -84,6 +85,11 @@ typedef struct Type int has_drop; ///< 1 if type implements Drop trait (RAII). int has_iterable; ///< 1 if type implements Iterable trait. } traits; + struct + { + int is_opaque_alias; + char *alias_defined_in_file; + } alias; }; } Type; @@ -263,6 +269,8 @@ struct ASTNode { char *alias; char *original_type; + int is_opaque; + char *defined_in_file; } type_alias; struct @@ -436,6 +444,8 @@ struct ASTNode Attribute *attributes; // Custom attributes char **used_structs; // Names of structs used/mixed-in int used_struct_count; + int is_opaque; + char *defined_in_file; // File where the struct is defined (for privacy check) } strct; struct diff --git a/src/lexer/token.c b/src/lexer/token.c index 6696a5c..095815d 100644 --- a/src/lexer/token.c +++ b/src/lexer/token.c @@ -201,6 +201,10 @@ Token lexer_next(Lexer *l) { return (Token){TOK_OR, s, 2, start_line, start_col}; } + if (len == 6 && strncmp(s, "opaque", 6) == 0) + { + return (Token){TOK_OPAQUE, s, 6, start_line, start_col}; + } // F-Strings if (len == 1 && s[0] == 'f' && s[1] == '"') diff --git a/src/lsp/lsp_analysis.c b/src/lsp/lsp_analysis.c index 088bede..0367d93 100644 --- a/src/lsp/lsp_analysis.c +++ b/src/lsp/lsp_analysis.c @@ -454,11 +454,80 @@ void lsp_completion(const char *uri, int line, int col, int id) cJSON_AddStringToObject(item, "label", s->name); cJSON_AddNumberToObject(item, "kind", 22); // Struct char detail[256]; - sprintf(detail, "struct %s", s->name); + sprintf(detail, "%sstruct %s", + (s->node && s->node->type == NODE_STRUCT && s->node->strct.is_opaque) + ? "opaque " + : "", + s->name); cJSON_AddStringToObject(item, "detail", detail); cJSON_AddItemToArray(items, item); s = s->next; } + + // Globals and Constants + StructRef *g = g_project->ctx->parsed_globals_list; + while (g) + { + if (g->node) + { + cJSON *item = cJSON_CreateObject(); + char *name = + (g->node->type == NODE_CONST) ? g->node->var_decl.name : g->node->var_decl.name; + cJSON_AddStringToObject(item, "label", name); + cJSON_AddNumberToObject(item, "kind", 21); // Constant/Variable + char detail[256]; + sprintf(detail, "%s %s", (g->node->type == NODE_CONST) ? "const" : "var", name); + cJSON_AddStringToObject(item, "detail", detail); + cJSON_AddItemToArray(items, item); + } + g = g->next; + } + + // Enums + StructRef *e = g_project->ctx->parsed_enums_list; + while (e) + { + if (e->node) + { + cJSON *item = cJSON_CreateObject(); + cJSON_AddStringToObject(item, "label", e->node->enm.name); + cJSON_AddNumberToObject(item, "kind", 13); // Enum + char detail[256]; + sprintf(detail, "enum %s", e->node->enm.name); + cJSON_AddStringToObject(item, "detail", detail); + cJSON_AddItemToArray(items, item); + } + e = e->next; + } + + // Type Aliases + TypeAlias *ta = g_project->ctx->type_aliases; + while (ta) + { + cJSON *item = cJSON_CreateObject(); + cJSON_AddStringToObject(item, "label", ta->alias); + cJSON_AddNumberToObject(item, "kind", 8); // Interface/Reference + char detail[256]; + sprintf(detail, "alias %s = %s", ta->alias, ta->original_type); + cJSON_AddStringToObject(item, "detail", detail); + cJSON_AddItemToArray(items, item); + ta = ta->next; + } + + // Keywords + const char *keywords[] = { + "fn", "struct", "enum", "alias", "return", "if", "else", "for", "while", + "break", "continue", "true", "false", "int", "char", "bool", "string", "void", + "import", "module", "test", "assert", "defer", "sizeof", "opaque", "unsafe", "asm", + "trait", "impl", "u8", "u16", "u32", "u64", "i8", "i16", "i32", + "i64", "f32", "f64", "usize", "isize", "const", "var", NULL}; + for (int i = 0; keywords[i]; i++) + { + cJSON *item = cJSON_CreateObject(); + cJSON_AddStringToObject(item, "label", keywords[i]); + cJSON_AddNumberToObject(item, "kind", 14); // Keyword + cJSON_AddItemToArray(items, item); + } } cJSON_AddItemToObject(root, "result", items); diff --git a/src/lsp/lsp_index.c b/src/lsp/lsp_index.c index 285dec3..975e153 100644 --- a/src/lsp/lsp_index.c +++ b/src/lsp/lsp_index.c @@ -151,7 +151,32 @@ void lsp_walk_node(LSPIndex *idx, ASTNode *node) else if (node->type == NODE_STRUCT) { char hover[256]; - sprintf(hover, "struct %s", node->strct.name); + if (node->strct.is_opaque) + { + sprintf(hover, "opaque struct %s", node->strct.name); + } + else + { + sprintf(hover, "struct %s", node->strct.name); + } + lsp_index_add_def(idx, node->token, hover, node); + } + else if (node->type == NODE_ENUM) + { + char hover[256]; + sprintf(hover, "enum %s", node->enm.name); + lsp_index_add_def(idx, node->token, hover, node); + } + else if (node->type == NODE_TYPE_ALIAS) + { + char hover[256]; + sprintf(hover, "alias %s = %s", node->type_alias.alias, node->type_alias.original_type); + lsp_index_add_def(idx, node->token, hover, node); + } + else if (node->type == NODE_TRAIT) + { + char hover[256]; + sprintf(hover, "trait %s", node->trait.name); lsp_index_add_def(idx, node->token, hover, node); } @@ -196,6 +221,31 @@ void lsp_walk_node(LSPIndex *idx, ASTNode *node) lsp_walk_node(idx, node->call.callee); lsp_walk_node(idx, node->call.args); break; + case NODE_MATCH: + lsp_walk_node(idx, node->match_stmt.expr); + lsp_walk_node(idx, node->match_stmt.cases); + break; + case NODE_MATCH_CASE: + lsp_walk_node(idx, node->match_case.guard); + lsp_walk_node(idx, node->match_case.body); + break; + case NODE_FOR: + lsp_walk_node(idx, node->for_stmt.init); + lsp_walk_node(idx, node->for_stmt.condition); + lsp_walk_node(idx, node->for_stmt.step); + lsp_walk_node(idx, node->for_stmt.body); + break; + case NODE_FOR_RANGE: + lsp_walk_node(idx, node->for_range.start); + lsp_walk_node(idx, node->for_range.end); + lsp_walk_node(idx, node->for_range.body); + break; + case NODE_LOOP: + lsp_walk_node(idx, node->loop_stmt.body); + break; + case NODE_DEFER: + lsp_walk_node(idx, node->defer_stmt.stmt); + break; default: break; } diff --git a/src/parser/parser.h b/src/parser/parser.h index cf57971..262c359 100644 --- a/src/parser/parser.h +++ b/src/parser/parser.h @@ -261,6 +261,8 @@ typedef struct TypeAlias char *alias; ///< New type name. char *original_type; ///< Original type. struct TypeAlias *next; + int is_opaque; + char *defined_in_file; } TypeAlias; /** @@ -514,12 +516,9 @@ char *sanitize_mangled_name(const char *name); /** * @brief Registers a type alias. */ -void register_type_alias(ParserContext *ctx, const char *alias, const char *original); - -/** - * @brief Finds a type alias. - */ -const char *find_type_alias(ParserContext *ctx, const char *alias); +TypeAlias *find_type_alias_node(ParserContext *ctx, const char *name); +void register_type_alias(ParserContext *ctx, const char *alias, const char *original, int is_opaque, + const char *defined_in_file); /** * @brief Registers an implementation. @@ -681,10 +680,6 @@ void register_selective_import(ParserContext *ctx, const char *symbol, const cha SelectiveImport *find_selective_import(ParserContext *ctx, const char *name); // Type Aliases -/** - * @brief Registers a type alias. - */ -void register_type_alias(ParserContext *ctx, const char *alias, const char *original); /** * @brief Finds a type alias. @@ -740,6 +735,11 @@ FuncSig *find_func(ParserContext *ctx, const char *name); */ Type *parse_type_formal(ParserContext *ctx, Lexer *l); +/** + * @brief Checks compatibility of opaque aliases (allows access within defining file). + */ +int check_opaque_alias_compat(ParserContext *ctx, Type *a, Type *b); + /** * @brief Parses a type. */ @@ -889,7 +889,7 @@ ASTNode *parse_def(ParserContext *ctx, Lexer *l); /** * @brief Parses a type alias. */ -ASTNode *parse_type_alias(ParserContext *ctx, Lexer *l); +ASTNode *parse_type_alias(ParserContext *ctx, Lexer *l, int is_opaque); /** * @brief Parses a function definition. @@ -899,7 +899,7 @@ ASTNode *parse_function(ParserContext *ctx, Lexer *l, int is_async); /** * @brief Parses a struct definition. */ -ASTNode *parse_struct(ParserContext *ctx, Lexer *l, int is_union); +ASTNode *parse_struct(ParserContext *ctx, Lexer *l, int is_union, int is_opaque); /** * @brief Parses an enum definition. @@ -963,4 +963,4 @@ char *patch_self_args(const char *args, const char *struct_name); */ ASTNode *parse_program_nodes(ParserContext *ctx, Lexer *l); -#endif // PARSER_H +#endif // PARSER_H \ No newline at end of file diff --git a/src/parser/parser_core.c b/src/parser/parser_core.c index ba6f2fe..43137b1 100644 --- a/src/parser/parser_core.c +++ b/src/parser/parser_core.c @@ -338,7 +338,7 @@ ASTNode *parse_program_nodes(ParserContext *ctx, Lexer *l) } else if (0 == strncmp(t.start, "struct", 6) && 6 == t.len) { - s = parse_struct(ctx, l, 0); + s = parse_struct(ctx, l, 0, 0); if (s && s->type == NODE_STRUCT) { s->strct.is_packed = attr_packed; @@ -436,7 +436,7 @@ ASTNode *parse_program_nodes(ParserContext *ctx, Lexer *l) } else if (0 == strncmp(t.start, "type", 4) && 4 == t.len) { - s = parse_type_alias(ctx, l); + s = parse_type_alias(ctx, l, 0); } else if (0 == strncmp(t.start, "raw", 3) && 3 == t.len) { @@ -482,9 +482,31 @@ ASTNode *parse_program_nodes(ParserContext *ctx, Lexer *l) lexer_next(l); } } + else if (t.type == TOK_OPAQUE) + { + lexer_next(l); // eat opaque + Token next = lexer_peek(l); + if (0 == strncmp(next.start, "struct", 6) && 6 == next.len) + { + s = parse_struct(ctx, l, 0, 1); + if (s && s->type == NODE_STRUCT) + { + s->strct.is_packed = attr_packed; + s->strct.align = attr_align; + } + } + else if (next.type == TOK_ALIAS) + { + s = parse_type_alias(ctx, l, 1); + } + else + { + zpanic_at(next, "Expected 'struct' or 'alias' after 'opaque'"); + } + } else if (t.type == TOK_ALIAS) { - s = parse_type_alias(ctx, l); + s = parse_type_alias(ctx, l, 0); } else if (t.type == TOK_ASYNC) { @@ -506,7 +528,7 @@ ASTNode *parse_program_nodes(ParserContext *ctx, Lexer *l) else if (t.type == TOK_UNION) { - s = parse_struct(ctx, l, 1); + s = parse_struct(ctx, l, 1, 0); } else if (t.type == TOK_TRAIT) { diff --git a/src/parser/parser_decl.c b/src/parser/parser_decl.c index 98f46e1..c96ca36 100644 --- a/src/parser/parser_decl.c +++ b/src/parser/parser_decl.c @@ -579,6 +579,10 @@ ASTNode *parse_var_decl(ParserContext *ctx, Lexer *l) { type_obj->inner = init->type_info->inner; // Shallow copy for inner } + if (init->type_info->kind == TYPE_ALIAS) + { + type_obj->alias = init->type_info->alias; + } // Copy function type args for lambda/closure support if (init->type_info->args && init->type_info->arg_count > 0) { @@ -631,6 +635,59 @@ ASTNode *parse_var_decl(ParserContext *ctx, Lexer *l) // Register in symbol table with actual token add_symbol_with_token(ctx, name, type, type_obj, name_tok); + if (init && type_obj) + { + Type *t = init->type_info; + if (!t && init->type == NODE_EXPR_VAR) + { + t = find_symbol_type_info(ctx, init->var_ref.name); + } + + // Literal type construction for validation + Type *temp_literal_type = NULL; + if (!t && init->type == NODE_EXPR_LITERAL) + { + if (init->literal.type_kind == LITERAL_INT) + { + temp_literal_type = type_new(TYPE_INT); + } + else if (init->literal.type_kind == LITERAL_FLOAT) + { + temp_literal_type = type_new(TYPE_FLOAT); + } + else if (init->literal.type_kind == LITERAL_STRING) + { + temp_literal_type = type_new(TYPE_STRING); + } + else if (init->literal.type_kind == LITERAL_CHAR) + { + temp_literal_type = type_new(TYPE_CHAR); + } + t = temp_literal_type; + } + + // Special case for literals: if implicit conversion works + if (t && !type_eq(type_obj, t)) + { + // Allow integer compatibility if types are roughly ints (lax check in type_eq handles + // most, but let's be safe) + if (!check_opaque_alias_compat(ctx, type_obj, t)) + { + char *expected = type_to_string(type_obj); + char *got = type_to_string(t); + zpanic_at(init->token, "Type validation failed. Expected '%s', but got '%s'", + expected, got); + free(expected); + free(got); + } + } + + if (temp_literal_type) + { + free(temp_literal_type); // Simple free, shallow + } + } + // NEW: Capture Const Integer Values if (init && init->type == NODE_EXPR_LITERAL && init->literal.type_kind == LITERAL_INT) { @@ -839,7 +896,7 @@ ASTNode *parse_def(ParserContext *ctx, Lexer *l) return o; } -ASTNode *parse_type_alias(ParserContext *ctx, Lexer *l) +ASTNode *parse_type_alias(ParserContext *ctx, Lexer *l, int is_opaque) { lexer_next(l); // consume 'type' or 'alias' Token n = lexer_next(l); @@ -859,8 +916,11 @@ ASTNode *parse_type_alias(ParserContext *ctx, Lexer *l) strncpy(node->type_alias.alias, n.start, n.len); node->type_alias.alias[n.len] = 0; node->type_alias.original_type = o; + node->type_alias.is_opaque = is_opaque; + node->type_alias.defined_in_file = g_current_filename ? xstrdup(g_current_filename) : NULL; - register_type_alias(ctx, node->type_alias.alias, o); + register_type_alias(ctx, node->type_alias.alias, o, is_opaque, + node->type_alias.defined_in_file); return node; } diff --git a/src/parser/parser_expr.c b/src/parser/parser_expr.c index ee1f96f..8f3579a 100644 --- a/src/parser/parser_expr.c +++ b/src/parser/parser_expr.c @@ -1,3 +1,43 @@ +#include "../codegen/codegen.h" + +int check_opaque_alias_compat(ParserContext *ctx, Type *a, Type *b) +{ + if (!a || !b) + { + return 0; + } + + int a_is_opaque = (a->kind == TYPE_ALIAS && a->alias.is_opaque_alias); + int b_is_opaque = (b->kind == TYPE_ALIAS && b->alias.is_opaque_alias); + + if (!a_is_opaque && !b_is_opaque) + { + return 1; + } + + if (a_is_opaque) + { + if (a->alias.alias_defined_in_file && g_current_filename && + strcmp(a->alias.alias_defined_in_file, g_current_filename) == 0) + { + return check_opaque_alias_compat(ctx, a->inner, b); + } + return 0; + } + + if (b_is_opaque) + { + if (b->alias.alias_defined_in_file && g_current_filename && + strcmp(b->alias.alias_defined_in_file, g_current_filename) == 0) + { + return check_opaque_alias_compat(ctx, a, b->inner); + } + return 0; + } + + return 0; +} + #include "../zen/zen_facts.h" #include "parser.h" #include @@ -1083,6 +1123,7 @@ static ASTNode *create_fstring_block(ParserContext *ctx, const char *content) static ASTNode *parse_int_literal(Token t) { ASTNode *node = ast_create(NODE_EXPR_LITERAL); + node->token = t; node->literal.type_kind = LITERAL_INT; node->type_info = type_new(TYPE_INT); char *s = token_strdup(t); @@ -1104,6 +1145,7 @@ static ASTNode *parse_int_literal(Token t) static ASTNode *parse_float_literal(Token t) { ASTNode *node = ast_create(NODE_EXPR_LITERAL); + node->token = t; node->literal.type_kind = LITERAL_FLOAT; node->literal.float_val = atof(t.start); node->type_info = type_new(TYPE_F64); @@ -1136,6 +1178,7 @@ static ASTNode *parse_string_literal(ParserContext *ctx, Token t) } ASTNode *node = ast_create(NODE_EXPR_LITERAL); + node->token = t; node->literal.type_kind = LITERAL_STRING; node->literal.string_val = xmalloc(t.len); strncpy(node->literal.string_val, t.start + 1, t.len - 2); @@ -1159,6 +1202,7 @@ static ASTNode *parse_fstring_literal(ParserContext *ctx, Token t) static ASTNode *parse_char_literal(Token t) { ASTNode *node = ast_create(NODE_EXPR_LITERAL); + node->token = t; node->literal.type_kind = LITERAL_CHAR; node->literal.string_val = token_strdup(t); node->type_info = type_new(TYPE_I8); @@ -2051,6 +2095,20 @@ ASTNode *parse_primary(ParserContext *ctx, Lexer *l) sprintf(prefixed, "%s_%s", ctx->current_module_prefix, acc); struct_name = prefixed; } + + // Opaque Struct Check + ASTNode *def = find_struct_def(ctx, struct_name); + if (def && def->type == NODE_STRUCT && def->strct.is_opaque) + { + if (!def->strct.defined_in_file || + (g_current_filename && + strcmp(def->strct.defined_in_file, g_current_filename) != 0)) + { + zpanic_at(lexer_peek(l), + "Cannot initialize opaque struct '%s' outside its module", + struct_name); + } + } lexer_next(l); node = ast_create(NODE_EXPR_STRUCT_INIT); node->struct_init.struct_name = struct_name; @@ -4045,6 +4103,30 @@ ASTNode *parse_expr_prec(ParserContext *ctx, Lexer *l, Precedence min_prec) node->member.field = token_strdup(field); node->member.is_pointer_access = 1; + // Opaque Check + int is_ptr_dummy = 0; + char *alloc_name = NULL; + char *sname = + resolve_struct_name_from_type(ctx, lhs->type_info, &is_ptr_dummy, &alloc_name); + if (sname) + { + ASTNode *def = find_struct_def(ctx, sname); + if (def && def->type == NODE_STRUCT && def->strct.is_opaque) + { + if (!def->strct.defined_in_file || + (g_current_filename && + strcmp(def->strct.defined_in_file, g_current_filename) != 0)) + { + zpanic_at(field, "Cannot access private field '%s' of opaque struct '%s'", + node->member.field, sname); + } + } + if (alloc_name) + { + free(alloc_name); + } + } + node->type_info = get_field_type(ctx, lhs->type_info, node->member.field); if (node->type_info) { @@ -4074,6 +4156,30 @@ ASTNode *parse_expr_prec(ParserContext *ctx, Lexer *l, Precedence min_prec) node->member.field = token_strdup(field); node->member.is_pointer_access = 2; + // Opaque Check + int is_ptr_dummy = 0; + char *alloc_name = NULL; + char *sname = + resolve_struct_name_from_type(ctx, lhs->type_info, &is_ptr_dummy, &alloc_name); + if (sname) + { + ASTNode *def = find_struct_def(ctx, sname); + if (def && def->type == NODE_STRUCT && def->strct.is_opaque) + { + if (!def->strct.defined_in_file || + (g_current_filename && + strcmp(def->strct.defined_in_file, g_current_filename) != 0)) + { + zpanic_at(field, "Cannot access private field '%s' of opaque struct '%s'", + node->member.field, sname); + } + } + if (alloc_name) + { + free(alloc_name); + } + } + node->type_info = get_field_type(ctx, lhs->type_info, node->member.field); if (node->type_info) { @@ -4653,6 +4759,30 @@ ASTNode *parse_expr_prec(ParserContext *ctx, Lexer *l, Precedence min_prec) node->member.field = token_strdup(field); node->member.is_pointer_access = 0; + // Opaque Check + int is_ptr_dummy = 0; + char *alloc_name = NULL; + char *sname = + resolve_struct_name_from_type(ctx, lhs->type_info, &is_ptr_dummy, &alloc_name); + if (sname) + { + ASTNode *def = find_struct_def(ctx, sname); + if (def && def->type == NODE_STRUCT && def->strct.is_opaque) + { + if (!def->strct.defined_in_file || + (g_current_filename && + strcmp(def->strct.defined_in_file, g_current_filename) != 0)) + { + zpanic_at(field, "Cannot access private field '%s' of opaque struct '%s'", + node->member.field, sname); + } + } + if (alloc_name) + { + free(alloc_name); + } + } + node->member.field = token_strdup(field); node->member.is_pointer_access = 0; @@ -5540,7 +5670,8 @@ ASTNode *parse_expr_prec(ParserContext *ctx, Lexer *l, Precedence min_prec) } else { - if (type_eq(lhs->type_info, rhs->type_info)) + if (type_eq(lhs->type_info, rhs->type_info) || + check_opaque_alias_compat(ctx, lhs->type_info, rhs->type_info)) { bin->type_info = lhs->type_info; } diff --git a/src/parser/parser_struct.c b/src/parser/parser_struct.c index 84450ba..82dd346 100644 --- a/src/parser/parser_struct.c +++ b/src/parser/parser_struct.c @@ -652,7 +652,7 @@ ASTNode *parse_impl(ParserContext *ctx, Lexer *l) } } -ASTNode *parse_struct(ParserContext *ctx, Lexer *l, int is_union) +ASTNode *parse_struct(ParserContext *ctx, Lexer *l, int is_union, int is_opaque) { lexer_next(l); // eat struct or union @@ -705,6 +705,7 @@ ASTNode *parse_struct(ParserContext *ctx, Lexer *l, int is_union) n->strct.is_union = is_union; n->strct.fields = NULL; n->strct.is_incomplete = 1; + n->strct.is_opaque = is_opaque; return n; } @@ -800,10 +801,6 @@ ASTNode *parse_struct(ParserContext *ctx, Lexer *l, int is_union) ASTNode *nf = ast_create(NODE_FIELD); nf->field.name = xstrdup(f->field.name); nf->field.type = xstrdup(f->field.type); - // Copy type info? Ideally deep copy or ref - // For now, we leave it NULL or shallow copy if needed, but mixins usually - // aren't generic params themselves in the same way. - // Let's shallow copy for safety if it exists. nf->type_info = f->type_info; if (!h) @@ -829,7 +826,6 @@ ASTNode *parse_struct(ParserContext *ctx, Lexer *l, int is_union) free(use_name); continue; } - // --------------------------------------- if (t.type == TOK_IDENT) { @@ -904,8 +900,10 @@ ASTNode *parse_struct(ParserContext *ctx, Lexer *l, int is_union) node->strct.generic_params = gps; node->strct.generic_param_count = gp_count; node->strct.is_union = is_union; + node->strct.is_opaque = is_opaque; node->strct.used_structs = temp_used_structs; node->strct.used_struct_count = temp_used_count; + node->strct.defined_in_file = g_current_filename ? xstrdup(g_current_filename) : NULL; if (gp_count > 0) { @@ -1102,7 +1100,7 @@ ASTNode *parse_enum(ParserContext *ctx, Lexer *l) node->enm.name = ename; node->enm.variants = h; - node->enm.generic_param = gp; // 3. Store generic param + node->enm.generic_param = gp; // Store generic param if (gp) { diff --git a/src/parser/parser_type.c b/src/parser/parser_type.c index c01f061..65f2848 100644 --- a/src/parser/parser_type.c +++ b/src/parser/parser_type.c @@ -33,12 +33,25 @@ Type *parse_type_base(ParserContext *ctx, Lexer *l) char *name = token_strdup(t); // Check for alias - const char *aliased = find_type_alias(ctx, name); - if (aliased) + TypeAlias *alias_node = find_type_alias_node(ctx, name); + if (alias_node) { free(name); Lexer tmp; - lexer_init(&tmp, aliased); + lexer_init(&tmp, alias_node->original_type); + + if (alias_node->is_opaque) + { + Type *underlying = parse_type_formal(ctx, &tmp); + Type *wrapper = type_new(TYPE_ALIAS); + wrapper->name = xstrdup(alias_node->alias); + wrapper->inner = underlying; + wrapper->alias.is_opaque_alias = 1; + wrapper->alias.alias_defined_in_file = + alias_node->defined_in_file ? xstrdup(alias_node->defined_in_file) : NULL; + return wrapper; + } + return parse_type_formal(ctx, &tmp); } diff --git a/src/parser/parser_utils.c b/src/parser/parser_utils.c index 4e85500..48418b6 100644 --- a/src/parser/parser_utils.c +++ b/src/parser/parser_utils.c @@ -402,23 +402,33 @@ void add_to_struct_list(ParserContext *ctx, ASTNode *node) ctx->parsed_structs_list = r; } -void register_type_alias(ParserContext *ctx, const char *alias, const char *original) +void register_type_alias(ParserContext *ctx, const char *alias, const char *original, int is_opaque, + const char *defined_in_file) { TypeAlias *ta = xmalloc(sizeof(TypeAlias)); ta->alias = xstrdup(alias); ta->original_type = xstrdup(original); + ta->is_opaque = is_opaque; + ta->defined_in_file = defined_in_file ? xstrdup(defined_in_file) : NULL; ta->next = ctx->type_aliases; ctx->type_aliases = ta; } const char *find_type_alias(ParserContext *ctx, const char *alias) +{ + TypeAlias *ta = find_type_alias_node(ctx, alias); + return ta ? ta->original_type : NULL; +} + +TypeAlias *find_type_alias_node(ParserContext *ctx, const char *alias) { TypeAlias *ta = ctx->type_aliases; while (ta) { if (strcmp(ta->alias, alias) == 0) { - return ta->original_type; + // printf("DEBUG: Found Alias '%s' (Opaque: %d)\n", alias, ta->is_opaque); + return ta; } ta = ta->next; } diff --git a/src/zprep.h b/src/zprep.h index e248871..a943f3f 100644 --- a/src/zprep.h +++ b/src/zprep.h @@ -108,6 +108,7 @@ typedef enum TOK_PREPROC, ///< Preprocessor directive (#...). TOK_ALIAS, ///< 'alias' keyword. TOK_COMMENT, ///< Comment (usually skipped). + TOK_OPAQUE, ///< 'opaque' keyword. TOK_UNKNOWN ///< Unknown token. } TokenType; diff --git a/tests/features/_opaque_alias_lib.zc b/tests/features/_opaque_alias_lib.zc new file mode 100644 index 0000000..7ca4abc --- /dev/null +++ b/tests/features/_opaque_alias_lib.zc @@ -0,0 +1,13 @@ +opaque alias Handle = int; + +fn new_handle(v: int) -> Handle { + return v; // Implicit cast int -> Handle (OK in module) +} + +fn get_val(h: Handle) -> int { + return h; // Implicit cast Handle -> int (OK in module) +} + +fn compare_handles(a: Handle, b: Handle) -> bool { + return a == b; // Strict equality (OK) +} diff --git a/tests/features/_opaque_lib.zc b/tests/features/_opaque_lib.zc new file mode 100644 index 0000000..de4d4c4 --- /dev/null +++ b/tests/features/_opaque_lib.zc @@ -0,0 +1,15 @@ +opaque struct SecretBox { + value: int; +} + +fn new_box(v: int) -> SecretBox { + return SecretBox { value: v }; +} + +fn get_value(b: SecretBox*) -> int { + return b.value; +} + +fn set_value(b: SecretBox*, v: int) { + b.value = v; +} diff --git a/tests/features/test_opaque.zc b/tests/features/test_opaque.zc new file mode 100644 index 0000000..5c84b2a --- /dev/null +++ b/tests/features/test_opaque.zc @@ -0,0 +1,18 @@ +import "_opaque_lib.zc"; + +fn main() { + let b = new_box(42); + + // Stack allocation should work (size known) + let b2: SecretBox; + b2 = b; + + // Public methods should work + let v = get_value(&b2); + assert(v == 42, "Value should be 42"); + + set_value(&b2, 100); + assert(get_value(&b2) == 100, "Value should be 100"); + + println "Opaque struct test passed"; +} diff --git a/tests/features/test_opaque_alias.zc b/tests/features/test_opaque_alias.zc new file mode 100644 index 0000000..062fa1f --- /dev/null +++ b/tests/features/test_opaque_alias.zc @@ -0,0 +1,13 @@ +import "_opaque_alias_lib.zc"; + +fn main() { + let h = new_handle(42); + let v = get_val(h); + + assert(v == 42, "Opaque Alias FAIL"); + + let h2 = new_handle(42); + assert(compare_handles(h, h2), "Equality FAIL"); + + println "Opaque Alias OK"; +} -- cgit v1.2.3 From ee090168fd7f678e40150b3699e335625b90d947 Mon Sep 17 00:00:00 2001 From: Zuhaitz Méndez Fernández de Aránguiz Date: Thu, 29 Jan 2026 22:45:43 +0000 Subject: Fix for #150 --- src/parser/parser_expr.c | 15 +++++++++++++++ 1 file changed, 15 insertions(+) (limited to 'src/parser') diff --git a/src/parser/parser_expr.c b/src/parser/parser_expr.c index 8f3579a..51d2baa 100644 --- a/src/parser/parser_expr.c +++ b/src/parser/parser_expr.c @@ -156,6 +156,14 @@ int is_type_copy(ParserContext *ctx, Type *t) { return 1; } + + // If the struct is NOT defined (opaque/C type) and does NOT implement Drop, + // treat it as Copy (C behavior). + if (!find_struct_def(ctx, t->name) && !check_impl(ctx, "Drop", t->name)) + { + return 1; + } + return 0; case TYPE_ARRAY: @@ -164,6 +172,13 @@ int is_type_copy(ParserContext *ctx, Type *t) // but if it's a value assignment, C doesn't support it anyway unless wrapped in struct. return 0; + case TYPE_ALIAS: + if (t->alias.is_opaque_alias) + { + return 1; + } + return is_type_copy(ctx, t->inner); + default: return 1; } -- cgit v1.2.3 From 0427d254207a69e394499d1abaea768f484f1cb5 Mon Sep 17 00:00:00 2001 From: Zuhaitz Méndez Fernández de Aránguiz Date: Fri, 30 Jan 2026 02:24:10 +0000 Subject: Improvements related to C23 (#112) --- README.md | 12 +++++++ README_ES.md | 11 ++++++ README_ZH_CN.md | 11 ++++++ README_ZH_TW.md | 11 ++++++ src/ast/ast.c | 25 ++++++++++++++ src/ast/ast.h | 4 ++- src/codegen/codegen_decl.c | 4 +++ src/codegen/compat.h | 6 +++- src/parser/parser_expr.c | 2 ++ src/parser/parser_type.c | 84 ++++++++++++++++++++++++++++++++++++++++++++++ 10 files changed, 168 insertions(+), 2 deletions(-) (limited to 'src/parser') diff --git a/README.md b/README.md index 75c4624..f03f605 100644 --- a/README.md +++ b/README.md @@ -110,6 +110,7 @@ Join the discussion, share demos, ask questions, or report bugs in the official - [C++ Interop](#c-interop) - [CUDA Interop](#cuda-interop) - [Objective-C Interop](#objective-c-interop) + - [C23 Support](#c23-support) - [Contributing](#contributing) - [Attributions](#attributions) @@ -197,6 +198,8 @@ let y: const int = 10; // Read-only (Type qualified) // y = 20; // Error: cannot assign to const ``` +> **Type Inference**: Zen C automatically infers types for initialized variables. It compiles to C23 `auto` on supported compilers, or GCC's `__auto_type` extension otherwise. + ### 2. Primitive Types | Type | C Equivalent | Description | @@ -211,6 +214,8 @@ let y: const int = 10; // Read-only (Type qualified) | `char` | `char` | Single character | | `string` | `char*` | C-string (null-terminated) | | `U0`, `u0`, `void` | `void` | Empty type | +| `iN` (for example, `i256`) | `_BitInt(N)` | Arbitrary bit-width signed integer (C23) | +| `uN` (for example, `u42`) | `unsigned _BitInt(N)` | Arbitrary bit-width unsigned integer (C23) | ### 3. Aggregate Types @@ -1337,6 +1342,13 @@ let tid = local_id(); > **Note:** The `--cuda` flag sets `nvcc` as the compiler and implies `--cpp` mode. Requires the NVIDIA CUDA Toolkit. +### C23 Support + +Zen C supports modern C23 features when using a compatible backend compiler (GCC 14+, Clang 14+, TCC (partial)). + +- **`auto`**: Zen C automatically maps type inference to standard C23 `auto` if `__STDC_VERSION__ >= 202300L`. +- **`_BitInt(N)`**: Use `iN` and `uN` types (e.g., `i256`, `u12`, `i24`) to access C23 arbitrary-width integers. + ### Objective-C Interop Zen C can compile to Objective-C (`.m`) using the `--objc` flag, allowing you to use Objective-C frameworks (like Cocoa/Foundation) and syntax. diff --git a/README_ES.md b/README_ES.md index c5655e3..9040ea8 100644 --- a/README_ES.md +++ b/README_ES.md @@ -197,6 +197,8 @@ let y: const int = 10; // Solo lectura (Calificado por tipo) // y = 20; // Error: no se puede asignar a una constante ``` +> **Inferencia de tipos**: Zen C infiere automáticamente los tipos para variables inicializadas. Se compila a `auto` de C23 en compiladores compatibles, o a la extensión `__auto_type` de GCC en otros casos. + ### 2. Tipos Primitivos | Tipo | Equivalente en C | Descripción | @@ -211,6 +213,8 @@ let y: const int = 10; // Solo lectura (Calificado por tipo) | `char` | `char` | Carácter único | | `string` | `char*` | Cadena de C (terminada en null) | | `U0`, `u0`, `void` | `void` | Tipo vacío | +| `iN` (ej. `i256`) | `_BitInt(N)` | Entero con signo de ancho arbitrario (C23) | +| `uN` (ej. `u42`) | `unsigned _BitInt(N)` | Entero sin signo de ancho arbitrario (C23) | ### 3. Tipos Agregados @@ -1337,6 +1341,13 @@ let tid = local_id(); > **Nota:** La flag `--cuda` establece `nvcc` como el compilador e implica el modo `--cpp`. Requiere el NVIDIA CUDA Toolkit. +### Soporte C23 + +Zen C soporta características modernas de C23 cuando se utiliza un compilador backend compatible (GCC 14+, Clang 14+). + +- **`auto`**: Zen C mapea automáticamente la inferencia de tipos a `auto` estándar de C23 si `__STDC_VERSION__ >= 202300L`. +- **`_BitInt(N)`**: Use tipos `iN` y `uN` (ej. `i256`, `u12`, `i24`) para acceder a enteros de ancho arbitrario de C23. + ### Interop con Objective-C Zen C puede compilarse a Objective-C (`.m`) usando la flag `--objc`, permitiéndote usar frameworks de Objective-C (como Cocoa/Foundation) y su sintaxis. diff --git a/README_ZH_CN.md b/README_ZH_CN.md index 6fce2d2..217e9ec 100644 --- a/README_ZH_CN.md +++ b/README_ZH_CN.md @@ -197,6 +197,8 @@ let y: const int = 10; // 只读 (类型修饰) // y = 20; // 错误：无法赋值给 const 变量 ``` +> **类型推导**：Zen C 自动推导初始化变量的类型。在支持的编译器上编译为 C23 的 `auto`，否则使用 GCC 的 `__auto_type` 扩展。 + ### 2. 原始类型 | 类型 | C 等效类型 | 描述 | @@ -211,6 +213,8 @@ let y: const int = 10; // 只读 (类型修饰) | `char` | `char` | 单个字符 | | `string` | `char*` | C-string (以 null 结尾) | | `U0`, `u0`, `void` | `void` | 空类型 | +| `iN` (例 `i256`) | `_BitInt(N)` | 任意位宽有符号整数 (C23) | +| `uN` (例 `u42`) | `unsigned _BitInt(N)` | 任意位宽无符号整数 (C23) | ### 3. 复合类型 @@ -1337,6 +1341,13 @@ let tid = local_id(); > **注意：** `--cuda` 标志设置 `nvcc` 为编译器并隐含 `--cpp` 模式。需要安装 NVIDIA CUDA Toolkit。 +### C23 支持 + +当使用兼容的后端编译器（GCC 14+, Clang 14+）时，Zen C 支持现代 C23特性。 + +- **`auto`**: 如果 `__STDC_VERSION__ >= 202300L`，Zen C 会自动将类型推导映射到标准 C23 `auto`。 +- **`_BitInt(N)`**: 使用 `iN` 和 `uN` 类型（例如 `i256`, `u12`, `i24`）访问 C23 任意位宽整数。 + ### Objective-C 互操作 Zen C 可以通过 `--objc` 标志编译为 Objective-C (`.m`)，允许你使用 Objective-C 框架（如 Cocoa/Foundation）和语法。 diff --git a/README_ZH_TW.md b/README_ZH_TW.md index fc9cef5..8618540 100644 --- a/README_ZH_TW.md +++ b/README_ZH_TW.md @@ -197,6 +197,8 @@ let y: const int = 10; // 只讀 (類型修飾) // y = 20; // 錯誤：無法賦值給 const 變量 ``` +> **型別推導**：Zen C 自動推導初始化變數的型別。在支援的編譯器上編譯為 C23 的 `auto`，否則使用 GCC 的 `__auto_type` 擴充功能。 + ### 2. 原始類型 | 類型 | C 等效類型 | 描述 | @@ -211,6 +213,8 @@ let y: const int = 10; // 只讀 (類型修飾) | `char` | `char` | 單個字符 | | `string` | `char*` | C-string (以 null 結尾) | | `U0`, `u0`, `void` | `void` | 空類型 | +| `iN` (例 `i256`) | `_BitInt(N)` | 任意位元寬度有號整數 (C23) | +| `uN` (例 `u42`) | `unsigned _BitInt(N)` | 任意位元寬度無號整數 (C23) | ### 3. 複合類型 @@ -1337,6 +1341,13 @@ let tid = local_id(); > **注意：** `--cuda` 標誌設置 `nvcc` 為編譯器並隱含 `--cpp` 模式。需要安裝 NVIDIA CUDA Toolkit。 +### C23 支援 + +當使用相容的後端編譯器（GCC 14+, Clang 14+）時，Zen C 支援現代 C23 特性。 + +- **`auto`**: 如果 `__STDC_VERSION__ >= 202300L`，Zen C 會自動將型別推導映射到標準 C23 `auto`。 +- **`_BitInt(N)`**: 使用 `iN` 和 `uN` 型別（例如 `i256`, `u12`, `i24`）存取 C23 任意位元寬度整數。 + ### Objective-C 互操作 Zen C 可以通過 `--objc` 標誌編譯為 Objective-C (`.m`)，允許你使用 Objective-C 框架（如 Cocoa/Foundation）和語法。 diff --git a/src/ast/ast.c b/src/ast/ast.c index f4922a6..439a9f5 100644 --- a/src/ast/ast.c +++ b/src/ast/ast.c @@ -100,6 +100,7 @@ int is_integer_type(Type *t) t->kind == TYPE_I64 || t->kind == TYPE_U64 || t->kind == TYPE_USIZE || t->kind == TYPE_ISIZE || t->kind == TYPE_BYTE || t->kind == TYPE_RUNE || t->kind == TYPE_UINT || t->kind == TYPE_I128 || t->kind == TYPE_U128 || + t->kind == TYPE_BITINT || t->kind == TYPE_UBITINT || (t->kind == TYPE_STRUCT && t->name && (0 == strcmp(t->name, "int8_t") || 0 == strcmp(t->name, "uint8_t") || 0 == strcmp(t->name, "int16_t") || 0 == strcmp(t->name, "uint16_t") || @@ -262,6 +263,18 @@ static char *type_to_string_impl(Type *t) return xstrdup("int"); case TYPE_FLOAT: return xstrdup("float"); + case TYPE_BITINT: + { + char *res = xmalloc(32); + sprintf(res, "i%d", t->array_size); + return res; + } + case TYPE_UBITINT: + { + char *res = xmalloc(32); + sprintf(res, "u%d", t->array_size); + return res; + } case TYPE_POINTER: { @@ -452,6 +465,18 @@ static char *type_to_c_string_impl(Type *t) return xstrdup("int"); case TYPE_FLOAT: return xstrdup("float"); + case TYPE_BITINT: + { + char *res = xmalloc(32); + sprintf(res, "_BitInt(%d)", t->array_size); + return res; + } + case TYPE_UBITINT: + { + char *res = xmalloc(40); + sprintf(res, "unsigned _BitInt(%d)", t->array_size); + return res; + } case TYPE_POINTER: { diff --git a/src/ast/ast.h b/src/ast/ast.h index a868bf0..71d9943 100644 --- a/src/ast/ast.h +++ b/src/ast/ast.h @@ -59,6 +59,8 @@ typedef enum TYPE_FUNCTION, ///< Function pointer or reference. TYPE_GENERIC, ///< Generic type parameter (T). TYPE_ALIAS, ///< Opaque type alias. + TYPE_BITINT, ///< C23 _BitInt(N). + TYPE_UBITINT, ///< C23 unsigned _BitInt(N). TYPE_UNKNOWN ///< Unknown/unresolved type. } TypeKind; @@ -75,7 +77,7 @@ typedef struct Type int is_const; ///< 1 if const-qualified. int is_explicit_struct; ///< 1 if defined with "struct" keyword explicitly. int is_raw; // Raw function pointer (fn*) - int array_size; ///< Size for fixed-size arrays. + int array_size; ///< Size for fixed-size arrays. For TYPE_BITINT, this is the bit width. union { int is_varargs; ///< 1 if function type is variadic. diff --git a/src/codegen/codegen_decl.c b/src/codegen/codegen_decl.c index 31513ef..0b78676 100644 --- a/src/codegen/codegen_decl.c +++ b/src/codegen/codegen_decl.c @@ -85,7 +85,11 @@ void emit_preamble(ParserContext *ctx, FILE *out) else { // C mode + fputs("#if defined(__STDC_VERSION__) && __STDC_VERSION__ >= 202300L\n", out); + fputs("#define ZC_AUTO auto\n", out); + fputs("#else\n", out); fputs("#define ZC_AUTO __auto_type\n", out); + fputs("#endif\n", out); fputs("#define ZC_CAST(T, x) ((T)(x))\n", out); fputs(ZC_TCC_COMPAT_STR, out); fputs("static inline const char* _z_bool_str(_Bool b) { return b ? \"true\" : " diff --git a/src/codegen/compat.h b/src/codegen/compat.h index 63a5af5..f8d9a4e 100644 --- a/src/codegen/compat.h +++ b/src/codegen/compat.h @@ -14,7 +14,11 @@ #define ZC_EXTERN_C_END } #else /* C mode */ -#define ZC_AUTO __auto_type ///< Auto type inference. +#if defined(__STDC_VERSION__) && __STDC_VERSION__ >= 202300L +#define ZC_AUTO auto ///< C23 standard auto. +#else +#define ZC_AUTO __auto_type ///< GCC/Clang extension. +#endif #define ZC_CAST(T, x) ((T)(x)) ///< Explicit cast. #define ZC_REINTERPRET(T, x) ((T)(x)) ///< Reinterpret cast. #define ZC_EXTERN_C ///< Extern "C" (no-op in C). diff --git a/src/parser/parser_expr.c b/src/parser/parser_expr.c index 51d2baa..6156cc0 100644 --- a/src/parser/parser_expr.c +++ b/src/parser/parser_expr.c @@ -148,6 +148,8 @@ int is_type_copy(ParserContext *ctx, Type *t) case TYPE_POINTER: // Pointers are Copy case TYPE_FUNCTION: case TYPE_ENUM: // Enums are integers + case TYPE_BITINT: + case TYPE_UBITINT: return 1; case TYPE_STRUCT: diff --git a/src/parser/parser_type.c b/src/parser/parser_type.c index 65f2848..49e961c 100644 --- a/src/parser/parser_type.c +++ b/src/parser/parser_type.c @@ -300,6 +300,90 @@ Type *parse_type_base(ParserContext *ctx, Lexer *l) free(name); return type_new(TYPE_I16); } + + // C23 BitInt Support (i42, u256, etc.) + if ((name[0] == 'i' || name[0] == 'u') && isdigit(name[1])) + { + // Verify it is a purely numeric suffix + int valid = 1; + for (size_t k = 1; k < strlen(name); k++) + { + if (!isdigit(name[k])) + { + valid = 0; + break; + } + } + if (valid) + { + int width = atoi(name + 1); + if (width > 0) + { + // Map standard widths to standard types for standard ABI/C compabitility + if (name[0] == 'i') + { + if (width == 8) + { + free(name); + return type_new(TYPE_I8); + } + if (width == 16) + { + free(name); + return type_new(TYPE_I16); + } + if (width == 32) + { + free(name); + return type_new(TYPE_I32); + } + if (width == 64) + { + free(name); + return type_new(TYPE_I64); + } + if (width == 128) + { + free(name); + return type_new(TYPE_I128); + } + } + else + { + if (width == 8) + { + free(name); + return type_new(TYPE_U8); + } + if (width == 16) + { + free(name); + return type_new(TYPE_U16); + } + if (width == 32) + { + free(name); + return type_new(TYPE_U32); + } + if (width == 64) + { + free(name); + return type_new(TYPE_U64); + } + if (width == 128) + { + free(name); + return type_new(TYPE_U128); + } + } + + Type *t = type_new(name[0] == 'u' ? TYPE_UBITINT : TYPE_BITINT); + t->array_size = width; + free(name); + return t; + } + } + } if (strcmp(name, "u16") == 0) { free(name); -- cgit v1.2.3 From 472434301940015365f7ed303f52d71c505ac487 Mon Sep 17 00:00:00 2001 From: Zuhaitz Méndez Fernández de Aránguiz Date: Fri, 30 Jan 2026 19:44:32 +0000 Subject: Improvements for the standard library + '@ctype'. --- Makefile | 2 +- README.md | 1 + README_ES.md | 1 + README_ZH_CN.md | 1 + README_ZH_TW.md | 1 + src/ast/ast.h | 2 ++ src/codegen/codegen_utils.c | 8 +++++- src/parser/parser.h | 3 ++- src/parser/parser_decl.c | 6 +++-- src/parser/parser_struct.c | 2 +- src/parser/parser_utils.c | 53 ++++++++++++++++++++++++++++++++++++-- std/core.zc | 6 ++--- std/cuda.zc | 2 ++ std/env.zc | 29 +++++++-------------- std/fs.zc | 62 ++++++++++++++++++--------------------------- std/io.zc | 43 ++++++++++++++++--------------- std/map.zc | 22 +++++++++------- std/net.zc | 25 ++++++++++-------- std/process.zc | 21 ++++++++------- std/set.zc | 20 +++++++-------- std/thread.zc | 5 ++++ std/time.zc | 3 +++ 22 files changed, 192 insertions(+), 126 deletions(-) (limited to 'src/parser') diff --git a/Makefile b/Makefile index d15f556..b2d8e29 100644 --- a/Makefile +++ b/Makefile @@ -188,7 +188,7 @@ clean: @echo "=> Clean complete!" # Test -test: $(TARGET) +test: $(TARGET) $(PLUGINS) ./tests/run_tests.sh ./tests/run_codegen_tests.sh ./tests/run_example_transpile.sh diff --git a/README.md b/README.md index f03f605..001e590 100644 --- a/README.md +++ b/README.md @@ -939,6 +939,7 @@ Decorate functions and structs to modify compiler behavior. | `@host` | Fn | CUDA: Host function (`__host__`). | | `@comptime` | Fn | Helper function available for compile-time execution. | | `@derive(...)` | Struct | Auto-implement traits. Supports `Debug`, `Eq` (Smart Derive), `Copy`, `Clone`. | +| `@ctype("type")` | Fn Param | Overrides generated C type for a parameter. | | `@` | Any | Passes generic attributes to C (e.g. `@flatten`, `@alias("name")`). | #### Custom Attributes diff --git a/README_ES.md b/README_ES.md index 9040ea8..999a7b4 100644 --- a/README_ES.md +++ b/README_ES.md @@ -938,6 +938,7 @@ Decora funciones y structs para modificar el comportamiento del compilador. | `@host` | Fn | CUDA: Función de host (`__host__`). | | `@comptime` | Fn | Función auxiliar disponible para ejecución en tiempo de compilación. | | `@derive(...)` | Struct | Implementa traits automáticamente. Soporta `Debug`, `Eq` (Derivación Inteligente), `Copy`, `Clone`. | +| `@ctype("tipo")` | Parámetro Fn | Sobrescribe el tipo C generado para un parámetro. | | `@` | Cualquier | Pasa atributos genéricos a C (ej. `@flatten`, `@alias("nombre")`). | #### Atributos Personalizados diff --git a/README_ZH_CN.md b/README_ZH_CN.md index 217e9ec..daa8a3d 100644 --- a/README_ZH_CN.md +++ b/README_ZH_CN.md @@ -938,6 +938,7 @@ let re = regex! { ^[a-z]+$ }; | `@host` | 函数 | CUDA: 主机函数 (`__host__`)。 | | `@comptime` | 函数 | 用于编译时执行的辅助函数。 | | `@derive(...)` | 结构体 | 自动实现 Trait。支持 `Debug`, `Eq` (智能派生), `Copy`, `Clone`。 | +| `@ctype("type")` | 函数参数 | 覆盖参数生成的 C 类型。 | | `@` | 任意 | 将泛型属性传递给 C (例如 `@flatten`, `@alias("name")`)。 | #### 自定义属性 diff --git a/README_ZH_TW.md b/README_ZH_TW.md index 8618540..2707caa 100644 --- a/README_ZH_TW.md +++ b/README_ZH_TW.md @@ -938,6 +938,7 @@ let re = regex! { ^[a-z]+$ }; | `@host` | 函數 | CUDA: 主機函數 (`__host__`)。 | | `@comptime` | 函數 | 用於編譯時執行的輔助函數。 | | `@derive(...)` | 結構體 | 自動實現 Trait。支持 `Debug`, `Eq` (智能派生), `Copy`, `Clone`。 | +| `@ctype("type")` | 函數參數 | 覆蓋參數生成的 C 類型。 | | `@` | 任意 | 將泛型屬性傳遞給 C (例如 `@flatten`, `@alias("name")`)。 | #### 自定義屬性 diff --git a/src/ast/ast.h b/src/ast/ast.h index 71d9943..4498d7c 100644 --- a/src/ast/ast.h +++ b/src/ast/ast.h @@ -229,6 +229,8 @@ struct ASTNode int cuda_device; // @device -> __device__ int cuda_host; // @host -> __host__ + char **c_type_overrides; // @ctype("...") per parameter + Attribute *attributes; // Custom attributes } func; diff --git a/src/codegen/codegen_utils.c b/src/codegen/codegen_utils.c index 391ebd3..0d03661 100644 --- a/src/codegen/codegen_utils.c +++ b/src/codegen/codegen_utils.c @@ -714,7 +714,12 @@ void emit_func_signature(FILE *out, ASTNode *func, const char *name_override) } char *type_str = NULL; - if (func->func.arg_types && func->func.arg_types[i]) + // Check for @ctype override first + if (func->func.c_type_overrides && func->func.c_type_overrides[i]) + { + type_str = xstrdup(func->func.c_type_overrides[i]); + } + else if (func->func.arg_types && func->func.arg_types[i]) { type_str = codegen_type_to_string(func->func.arg_types[i]); } @@ -724,6 +729,7 @@ void emit_func_signature(FILE *out, ASTNode *func, const char *name_override) } const char *name = ""; + if (func->func.param_names && func->func.param_names[i]) { name = func->func.param_names[i]; diff --git a/src/parser/parser.h b/src/parser/parser.h index 262c359..23c2920 100644 --- a/src/parser/parser.h +++ b/src/parser/parser.h @@ -561,7 +561,8 @@ ASTNode *parse_arrow_lambda_multi(ParserContext *ctx, Lexer *l, char **param_nam * @brief Parses and converts arguments. */ char *parse_and_convert_args(ParserContext *ctx, Lexer *l, char ***defaults_out, int *count_out, - Type ***types_out, char ***names_out, int *is_varargs_out); + Type ***types_out, char ***names_out, int *is_varargs_out, + char ***ctype_overrides_out); /** * @brief Checks if a file has been imported. diff --git a/src/parser/parser_decl.c b/src/parser/parser_decl.c index c96ca36..93a124d 100644 --- a/src/parser/parser_decl.c +++ b/src/parser/parser_decl.c @@ -99,10 +99,11 @@ ASTNode *parse_function(ParserContext *ctx, Lexer *l, int is_async) int count; Type **arg_types; char **param_names; + char **ctype_overrides; int is_varargs = 0; - char *args = - parse_and_convert_args(ctx, l, &defaults, &count, &arg_types, ¶m_names, &is_varargs); + char *args = parse_and_convert_args(ctx, l, &defaults, &count, &arg_types, ¶m_names, + &is_varargs, &ctype_overrides); char *ret = "void"; Type *ret_type_obj = type_new(TYPE_VOID); @@ -191,6 +192,7 @@ ASTNode *parse_function(ParserContext *ctx, Lexer *l, int is_async) node->func.defaults = defaults; node->func.ret_type_info = ret_type_obj; node->func.is_varargs = is_varargs; + node->func.c_type_overrides = ctype_overrides; if (gen_param) { diff --git a/src/parser/parser_struct.c b/src/parser/parser_struct.c index 82dd346..109eeee 100644 --- a/src/parser/parser_struct.c +++ b/src/parser/parser_struct.c @@ -100,7 +100,7 @@ ASTNode *parse_trait(ParserContext *ctx, Lexer *l) char **param_names = NULL; int is_varargs = 0; char *args = parse_and_convert_args(ctx, l, &defaults, &arg_count, &arg_types, ¶m_names, - &is_varargs); + &is_varargs, NULL); char *ret = xstrdup("void"); if (lexer_peek(l).type == TOK_ARROW) diff --git a/src/parser/parser_utils.c b/src/parser/parser_utils.c index 48418b6..28d2c11 100644 --- a/src/parser/parser_utils.c +++ b/src/parser/parser_utils.c @@ -3392,7 +3392,8 @@ char *consume_and_rewrite(ParserContext *ctx, Lexer *l) } char *parse_and_convert_args(ParserContext *ctx, Lexer *l, char ***defaults_out, int *count_out, - Type ***types_out, char ***names_out, int *is_varargs_out) + Type ***types_out, char ***names_out, int *is_varargs_out, + char ***ctype_overrides_out) { Token t = lexer_next(l); if (t.type != TOK_LPAREN) @@ -3406,18 +3407,52 @@ char *parse_and_convert_args(ParserContext *ctx, Lexer *l, char ***defaults_out, char **defaults = xmalloc(sizeof(char *) * 16); Type **types = xmalloc(sizeof(Type *) * 16); char **names = xmalloc(sizeof(char *) * 16); + char **ctype_overrides = xmalloc(sizeof(char *) * 16); for (int i = 0; i < 16; i++) { defaults[i] = NULL; types[i] = NULL; names[i] = NULL; + ctype_overrides[i] = NULL; } if (lexer_peek(l).type != TOK_RPAREN) { while (1) { + // Check for @ctype("...") before parameter + char *ctype_override = NULL; + if (lexer_peek(l).type == TOK_AT) + { + lexer_next(l); // eat @ + Token attr = lexer_next(l); + if (attr.type == TOK_IDENT && attr.len == 5 && strncmp(attr.start, "ctype", 5) == 0) + { + if (lexer_next(l).type != TOK_LPAREN) + { + zpanic_at(lexer_peek(l), "Expected ( after @ctype"); + } + Token ctype_tok = lexer_next(l); + if (ctype_tok.type != TOK_STRING) + { + zpanic_at(ctype_tok, "@ctype requires a string argument"); + } + // Extract string content (strip quotes) + ctype_override = xmalloc(ctype_tok.len - 1); + strncpy(ctype_override, ctype_tok.start + 1, ctype_tok.len - 2); + ctype_override[ctype_tok.len - 2] = 0; + if (lexer_next(l).type != TOK_RPAREN) + { + zpanic_at(lexer_peek(l), "Expected ) after @ctype string"); + } + } + else + { + zpanic_at(attr, "Unknown parameter attribute @%.*s", attr.len, attr.start); + } + } + Token t = lexer_next(l); // Handle 'self' if (t.type == TOK_IDENT && strncmp(t.start, "self", 4) == 0 && t.len == 4) @@ -3470,6 +3505,7 @@ char *parse_and_convert_args(ParserContext *ctx, Lexer *l, char ***defaults_out, types[count] = type_new_ptr(type_new(TYPE_VOID)); add_symbol(ctx, "self", "void*", types[count]); } + ctype_overrides[count] = ctype_override; count++; } else @@ -3514,11 +3550,20 @@ char *parse_and_convert_args(ParserContext *ctx, Lexer *l, char ***defaults_out, } else { - strcat(buf, type_str); + // Use @ctype override if present + if (ctype_override) + { + strcat(buf, ctype_override); + } + else + { + strcat(buf, type_str); + } strcat(buf, " "); strcat(buf, name); } + ctype_overrides[count] = ctype_override; count++; if (lexer_peek(l).type == TOK_OP && is_token(lexer_peek(l), "=")) @@ -3593,6 +3638,10 @@ char *parse_and_convert_args(ParserContext *ctx, Lexer *l, char ***defaults_out, *count_out = count; *types_out = types; *names_out = names; + if (ctype_overrides_out) + { + *ctype_overrides_out = ctype_overrides; + } return buf; } diff --git a/std/core.zc b/std/core.zc index f450517..7379db4 100644 --- a/std/core.zc +++ b/std/core.zc @@ -7,11 +7,11 @@ include let __zen_hash_seed: usize = 14695981039346656037; -raw { -void _zen_panic(const char* file, int line, const char* func, const char* msg) { +extern fn exit(code: int); + +fn _zen_panic(file: const char*, line: int, func: const char*, msg: const char*) { fprintf(stderr, "%s:%d (%s): Panic: %s\n", file, line, func, msg); exit(1); } -} #define panic(msg) _zen_panic(__FILE__, __LINE__, __func__, msg) \ No newline at end of file diff --git a/std/cuda.zc b/std/cuda.zc index c6a9403..8fc6545 100644 --- a/std/cuda.zc +++ b/std/cuda.zc @@ -101,6 +101,8 @@ fn cuda_ok() -> bool { } +// Minimal raw block: required for cudaDeviceProp struct field access +// The cudaDeviceProp struct cannot be declared in Zen-C without type conflicts raw { void _z_cuda_get_props(int dev, char* name, size_t* total_mem, int* sm_count, int* major, int* minor, int* max_threads, int* warp_size) { struct cudaDeviceProp prop; diff --git a/std/env.zc b/std/env.zc index 959784f..c63fd3d 100644 --- a/std/env.zc +++ b/std/env.zc @@ -2,23 +2,12 @@ import "./core.zc" import "./option.zc" import "./string.zc" -raw { - char *_z_env_get(char *name) { - return getenv(name); - } - - int _z_env_set(char *name, char *value, int overwrite) { - return setenv(name, value, overwrite); - } - - int _z_env_unset(char *name) { - return unsetenv(name); - } -} +include -extern fn _z_env_get(name: char*) -> char*; -extern fn _z_env_set(name: char*, value: char*, overwrite: int) -> int; -extern fn _z_env_unset(name: char*) -> int; +// Direct externs with const char* to match C stdlib declarations +extern fn getenv(name: const char*) -> char*; +extern fn setenv(name: const char*, value: const char*, overwrite: int) -> int; +extern fn unsetenv(name: const char*) -> int; @derive(Eq) enum EnvRes { @@ -30,7 +19,7 @@ struct Env {} impl Env { fn get(name: string) -> Option { - let value: string = _z_env_get(name); + let value: string = getenv(name); if (value == NULL) { return Option::None(); } @@ -39,7 +28,7 @@ impl Env { } fn get_dup(name: string) -> Option { - let value: string = _z_env_get(name); + let value: string = getenv(name); if (value == NULL) { return Option::None(); } @@ -52,13 +41,13 @@ impl Env { } fn set(name: string, value: string) -> EnvRes { - let ret: int = _z_env_set(name, value, 1); + let ret: int = setenv(name, value, 1); return (ret == 0) ? EnvRes::OK() : EnvRes::ERR(); } fn unset(name: string) -> EnvRes { - let ret: int = _z_env_unset(name); + let ret: int = unsetenv(name); return (ret == 0) ? EnvRes::OK() : EnvRes::ERR(); } diff --git a/std/fs.zc b/std/fs.zc index 4547b30..a00993b 100644 --- a/std/fs.zc +++ b/std/fs.zc @@ -14,12 +14,20 @@ include include include -// TODO: restructure this tomorrow (lol). +// Direct externs for simple functions with const char* parameters +extern fn access(pathname: const char*, mode: int) -> int; +extern fn unlink(pathname: const char*) -> int; +extern fn rmdir(pathname: const char*) -> int; +extern fn malloc(size: usize) -> void*; +extern fn free(ptr: void*); + +// Minimal raw block: required for opaque FILE*/DIR* types and C struct access +// These cannot be expressed in Zen-C extern declarations without type conflicts raw { typedef struct DirEntry* DirEntryPtr; - // Wrappers for FILE* handling due to opaque pointer casting - void* _z_fs_fopen(char* path, char* mode) { + // FILE* wrappers - fopen/fclose/etc use FILE* which conflicts with void* + void* _z_fs_fopen(const char* path, const char* mode) { return fopen(path, mode); } @@ -43,22 +51,8 @@ raw { return (int64_t)ftell((FILE*)stream); } - // Wrappers needed because C headers declare these with 'const char*' - // but Zen C externs generate 'char*', leading to conflicting types. - int _z_fs_access(char* pathname, int mode) { - return access(pathname, mode); - } - - int _z_fs_unlink(char* pathname) { - return unlink(pathname); - } - - int _z_fs_rmdir(char* pathname) { - return rmdir(pathname); - } - - // Wrappers for DIR* handling - void* _z_fs_opendir(char* name) { + // DIR* wrappers - opendir/closedir/readdir use DIR* which conflicts with void* + void* _z_fs_opendir(const char* name) { return opendir(name); } @@ -66,8 +60,8 @@ raw { return closedir((DIR*)dir); } - // struct stat / struct dirent helpers - int _z_fs_get_metadata(char* path, uint64_t* size, int* is_dir, int* is_file) { + // struct stat access - cannot define matching Zen-C struct for stat + int _z_fs_get_metadata(const char* path, uint64_t* size, int* is_dir, int* is_file) { struct stat st; if (stat(path, &st) != 0) return -1; *size = st.st_size; @@ -76,6 +70,7 @@ raw { return 0; } + // struct dirent access - readdir returns struct dirent* int _z_fs_read_entry(void* dir, char* out_name, int buf_size, int* is_dir) { struct dirent* ent = readdir((DIR*)dir); if (!ent) return 0; @@ -85,7 +80,8 @@ raw { return 1; } - int _z_fs_mkdir(char* path) { + // mkdir has different signatures on Windows vs POSIX + int _z_fs_mkdir(const char* path) { #ifdef _WIN32 return mkdir(path); #else @@ -94,26 +90,18 @@ raw { } } -// Direct externs -extern fn malloc(size: usize) -> void*; -extern fn free(ptr: void*); - -extern fn _z_fs_mkdir(path: char*) -> int; -extern fn _z_fs_get_metadata(path: char*, size: U64*, is_dir: int*, is_file: int*) -> int; +extern fn _z_fs_mkdir(path: const char*) -> int; +extern fn _z_fs_get_metadata(path: const char*, size: U64*, is_dir: int*, is_file: int*) -> int; extern fn _z_fs_read_entry(dir: void*, out_name: char*, buf_size: int, is_dir: int*) -> int; -extern fn _z_fs_fopen(path: char*, mode: char*) -> void*; +extern fn _z_fs_fopen(path: const char*, mode: const char*) -> void*; extern fn _z_fs_fclose(stream: void*) -> int; extern fn _z_fs_fread(ptr: void*, size: usize, nmemb: usize, stream: void*) -> usize; extern fn _z_fs_fwrite(ptr: void*, size: usize, nmemb: usize, stream: void*) -> usize; extern fn _z_fs_fseek(stream: void*, offset: I64, whence: int) -> int; extern fn _z_fs_ftell(stream: void*) -> I64; -extern fn _z_fs_opendir(name: char*) -> void*; +extern fn _z_fs_opendir(name: const char*) -> void*; extern fn _z_fs_closedir(dir: void*) -> int; -extern fn _z_fs_access(pathname: char*, mode: int) -> int; -extern fn _z_fs_unlink(pathname: char*) -> int; -extern fn _z_fs_rmdir(pathname: char*) -> int; - struct File { handle: void*; @@ -203,7 +191,7 @@ impl File { } fn exists(path: char*) -> bool { - return _z_fs_access(path, Z_F_OK) == 0; + return access(path, Z_F_OK) == 0; } fn metadata(path: char*) -> Result { @@ -230,14 +218,14 @@ impl File { } fn remove_file(path: char*) -> Result { - if (_z_fs_unlink(path) != 0) { + if (unlink(path) != 0) { return Result::Err("Failed to remove file"); } return Result::Ok(true); } fn remove_dir(path: char*) -> Result { - if (_z_fs_rmdir(path) != 0) { + if (rmdir(path) != 0) { return Result::Err("Failed to remove directory"); } return Result::Ok(true); diff --git a/std/io.zc b/std/io.zc index 2793ecf..cfb9179 100644 --- a/std/io.zc +++ b/std/io.zc @@ -2,28 +2,32 @@ import "./core.zc" import "./string.zc" -raw { - int _z_vprintf(char* fmt, va_list ap) { - return vprintf((const char*)fmt, ap); - } - - int _z_vsnprintf(char* str, size_t size, char* fmt, va_list ap) { - return vsnprintf(str, size, (const char*)fmt, ap); - } +include +include - void* _z_get_stdin() { return stdin; } - int _z_get_eof() { return EOF; } +// These work directly with const char* in extern declarations +extern fn vprintf(fmt: const char*, ap: va_list) -> int; +extern fn vsnprintf(str: char*, size: usize, fmt: const char*, ap: va_list) -> int; + +// EOF is typically -1, but we define it for portability +def Z_EOF = -1; + +// Minimal raw block: only for truly opaque FILE* types that can't be +// represented in Zen-C extern declarations without type conflicts. +// These wrappers use void* to avoid FILE* declaration conflicts. +raw { + void* _z_get_stdin(void) { return stdin; } int _z_fgetc(void* stream) { return fgetc((FILE*)stream); } } -extern fn _z_vprintf(fmt: char*, ap: va_list) -> int; -extern fn _z_vsnprintf(str: char*, size: usize, fmt: char*, ap: va_list) -> int; +extern fn _z_get_stdin() -> void*; +extern fn _z_fgetc(stream: void*) -> int; fn format(fmt: char*, ...) -> char* { static let buffer: char[1024]; let ap: va_list; va_start(ap, fmt); - _z_vsnprintf(buffer, 1024, fmt, ap); + vsnprintf(buffer, 1024, fmt, ap); va_end(ap); return (char*)buffer; } @@ -31,7 +35,7 @@ fn format(fmt: char*, ...) -> char* { fn format_into(buffer: char*, size: usize, fmt: char*, ...) -> int { let ap: va_list; va_start(ap, fmt); - let ret = _z_vsnprintf(buffer, size, fmt, ap); + let ret = vsnprintf(buffer, size, fmt, ap); va_end(ap); return ret; } @@ -42,7 +46,7 @@ fn format_new(fmt: char*, ...) -> char* { let ap: va_list; va_start(ap, fmt); - _z_vsnprintf(buffer, 1024, fmt, ap); + vsnprintf(buffer, 1024, fmt, ap); va_end(ap); return buffer; } @@ -50,7 +54,7 @@ fn format_new(fmt: char*, ...) -> char* { fn print(fmt: char*, ...) -> int { let ap: va_list; va_start(ap, fmt); - let ret = _z_vprintf(fmt, ap); + let ret = vprintf(fmt, ap); va_end(ap); return ret; } @@ -58,7 +62,7 @@ fn print(fmt: char*, ...) -> int { fn println(fmt: char*, ...) -> int { let ap: va_list; va_start(ap, fmt); - let ret = _z_vprintf(fmt, ap); + let ret = vprintf(fmt, ap); va_end(ap); puts(""); return ret + 1; @@ -72,11 +76,10 @@ fn readln() -> char* { let c: int; let std_in = _z_get_stdin(); - let eof_val = _z_get_eof(); while (true) { c = _z_fgetc(std_in); - if (c == eof_val) break; + if (c == Z_EOF) break; if (c == 10) break; // '\n' if (len + 1 >= cap) { @@ -93,7 +96,7 @@ fn readln() -> char* { len = len + 1; } - if (len == 0 && c == eof_val) { + if (len == 0 && c == Z_EOF) { free(line); return NULL; } diff --git a/std/map.zc b/std/map.zc index 70d6ad2..8376da2 100644 --- a/std/map.zc +++ b/std/map.zc @@ -3,16 +3,20 @@ import "./core.zc" import "./option.zc" import "./mem.zc" -raw { - extern size_t __zen_hash_seed; - size_t _map_hash_str(const char* str) { - size_t hash = __zen_hash_seed; - while (*str) { - hash ^= (unsigned char)*str++; - hash *= 1099511628211UL; - } - return hash; +// Pure Zen-C string hash using FNV-1a algorithm +fn _map_hash_str(str: const char*) -> usize { + let hash = __zen_hash_seed; + let i: usize = 0; + + while (str[i] != 0) { + // Cast char to U8 for unsigned byte value + let b: U8 = (U8)str[i]; + hash = hash ^ (usize)b; + hash = hash * (usize)1099511628211; + i = i + 1; } + + return hash; } struct Map { diff --git a/std/net.zc b/std/net.zc index dd10642..d410a4d 100644 --- a/std/net.zc +++ b/std/net.zc @@ -13,8 +13,17 @@ import "./mem.zc" def Z_AF_INET = 2; def Z_SOCK_STREAM = 1; +// Direct externs for simple socket functions +extern fn socket(domain: int, type: int, proto: int) -> int; +extern fn close(fd: int) -> int; +extern fn read(fd: int, buf: void*, count: usize) -> isize; + +// Minimal raw block: required for struct sockaddr_in usage +// These functions encapsulate sockaddr_in setup because the struct layout +// cannot be declared in Zen-C without type conflicts with C headers. +// Also includes inet_pton, htons, bind, connect, listen, accept wrappers. raw { - static int _z_net_bind(int fd, char *host, int port) { + static int _z_net_bind(int fd, const char *host, int port) { struct sockaddr_in addr; addr.sin_family = AF_INET; addr.sin_port = htons(port); @@ -28,7 +37,7 @@ raw { return 0; } - static int _z_net_connect(int fd, char *host, int port) { + static int _z_net_connect(int fd, const char *host, int port) { struct sockaddr_in addr; addr.sin_family = AF_INET; addr.sin_port = htons(port); @@ -42,19 +51,15 @@ raw { return accept(fd, NULL, NULL); } - static ssize_t _z_net_write(int fd, char* buf, size_t n) { + static ssize_t _z_net_write(int fd, const char* buf, size_t n) { return write(fd, (const void*)buf, n); } } -extern fn socket(domain: int, type: int, proto: int) -> int; -extern fn close(fd: int) -> int; -extern fn read(fd: int, buf: void*, count: usize) -> isize; - -extern fn _z_net_bind(fd: int, host: char*, port: int) -> int; -extern fn _z_net_connect(fd: int, host: char*, port: int) -> int; +extern fn _z_net_bind(fd: int, host: const char*, port: int) -> int; +extern fn _z_net_connect(fd: int, host: const char*, port: int) -> int; extern fn _z_net_accept(fd: int) -> int; -extern fn _z_net_write(fd: int, buf: char*, n: usize) -> isize; +extern fn _z_net_write(fd: int, buf: const char*, n: usize) -> isize; struct TcpStream { diff --git a/std/process.zc b/std/process.zc index d0b09a9..3ce43b6 100644 --- a/std/process.zc +++ b/std/process.zc @@ -5,8 +5,16 @@ import "./mem.zc"; import "./string.zc"; import "./option.zc"; -raw { - void *_z_popen(char *command, char *type) { +include +include + +// system() can be externed directly with const char* +extern fn system(command: const char*) -> int; + +// Minimal raw block: only for opaque FILE* types +// popen/pclose/fgets use FILE* which conflicts with void* +raw { + void *_z_popen(const char *command, const char *type) { return (void *)popen(command, type); } @@ -17,16 +25,11 @@ raw { char *_z_fgets(char *s, int size, void *stream) { return fgets(s, size, (FILE *)stream); } - - int _z_system(char *command) { - return system(command); - } } -extern fn _z_popen(command: char*, type: char*) -> void*; +extern fn _z_popen(command: const char*, type: const char*) -> void*; extern fn _z_pclose(stream: void*) -> int; extern fn _z_fgets(s: char*, size: int, stream: void*) -> char*; -extern fn _z_system(command: char*) -> int; struct Output { stdout: String; @@ -105,7 +108,7 @@ impl Command { fn status(self) -> int { let cmd_str = self._build_cmd(); - let code = _z_system(cmd_str.c_str()); + let code = system(cmd_str.c_str()); cmd_str.free(); return code; } diff --git a/std/set.zc b/std/set.zc index ba6c93f..e1faab3 100644 --- a/std/set.zc +++ b/std/set.zc @@ -2,17 +2,17 @@ import "./core.zc" import "./option.zc" -raw { - extern size_t __zen_hash_seed; - size_t _set_hash(const void* data, size_t len) { - size_t hash = __zen_hash_seed; - const unsigned char* bytes = (const unsigned char*)data; - for (size_t i = 0; i < len; i++) { - hash ^= bytes[i]; - hash *= 1099511628211UL; - } - return hash; +// Pure Zen-C generic hash using FNV-1a algorithm +fn _set_hash(data: const void*, len: usize) -> usize { + let hash = __zen_hash_seed; + let bytes: U8* = (U8*)data; + + for (let i: usize = 0; i < len; i = i + 1) { + hash = hash ^ (usize)bytes[i]; + hash = hash * (usize)1099511628211; } + + return hash; } struct Set { diff --git a/std/thread.zc b/std/thread.zc index 98f080e..16f3ca1 100644 --- a/std/thread.zc +++ b/std/thread.zc @@ -7,6 +7,11 @@ import "./core.zc" import "./result.zc" import "./mem.zc" +// Essential raw block: required for pthread operations and closure trampolining +// This block cannot be eliminated because: +// 1. z_closure_T is an internal compiler type for Zen-C closures +// 2. pthread_t, pthread_mutex_t are opaque types that can't be extern'd with void* +// 3. The trampoline function needs to cast and execute Zen-C closures raw { typedef void (*ZenThreadFunc)(void*); diff --git a/std/time.zc b/std/time.zc index 865dd6d..fa764ed 100644 --- a/std/time.zc +++ b/std/time.zc @@ -5,6 +5,9 @@ include include include +// Minimal raw block: required because gettimeofday() uses struct timeval +// which can't be declared in Zen-C without type conflicts, and time() +// has conflicting type signature (time_t* vs void*) raw { static uint64_t _time_now_impl(void) { struct timeval tv; -- cgit v1.2.3 From 856c9fe56b412779e045ef86a767b93d5c7f563b Mon Sep 17 00:00:00 2001 From: Zuhaitz Méndez Fernández de Aránguiz Date: Sat, 31 Jan 2026 01:15:25 +0000 Subject: Improvements for slice + better iteration for arrays --- README.md | 11 ++- README_ES.md | 12 ++- README_ZH_CN.md | 27 ++++-- README_ZH_TW.md | 11 ++- docs/std/slice.md | 90 ++++++++++++++++++ src/codegen/codegen.c | 88 +++++++++++++++++- src/codegen/codegen_utils.c | 25 ++++- src/parser/parser_stmt.c | 152 ++++++++++++++++++++++++++++--- std/slice.zc | 35 +++++++ tests/std/test_direct_array_iteration.zc | 37 ++++++++ tests/std/test_slice_iteration.zc | 29 ++++++ 11 files changed, 479 insertions(+), 38 deletions(-) create mode 100644 docs/std/slice.md create mode 100644 tests/std/test_direct_array_iteration.zc create mode 100644 tests/std/test_slice_iteration.zc (limited to 'src/parser') diff --git a/README.md b/README.md index ab51b83..ecdf8fc 100644 --- a/README.md +++ b/README.md @@ -494,8 +494,15 @@ for i in 0..<10 { ... } // Exclusive (Explicit) for i in 0..=10 { ... } // Inclusive (0 to 10) for i in 0..10 step 2 { ... } -// Iterator (Vec, Array, or custom Iterable) -for item in collection { ... } +// Iterator (Vec or custom Iterable) +for item in vec { ... } + +// Iterate over fixed-size arrays directly +let arr: int[5] = [1, 2, 3, 4, 5]; +for val in arr { + // val is int + println "{val}"; +} // While while x < 10 { ... } diff --git a/README_ES.md b/README_ES.md index d2cfbbb..d73e9ca 100644 --- a/README_ES.md +++ b/README_ES.md @@ -493,8 +493,15 @@ for i in 0..<10 { ... } // Exclusivo (Explícito) for i in 0..=10 { ... } // Inclusivo (0 al 10) for i in 0..10 step 2 { ... } -// Iterador (Vec, Array, o Iterable personalizado) -for item in coleccion { ... } +// Iterador (Vec o Iterable personalizado) +for item in vec { ... } + +// Iterar sobre arrays de tamaño fijo directamente +let arr: int[5] = [1, 2, 3, 4, 5]; +for val in arr { + // val es int + println "{val}"; +} // While while x < 10 { ... } @@ -508,6 +515,7 @@ externo: loop { for _ in 0..5 { ... } ``` + #### Control Avanzado ```zc // Guard: Ejecuta else y retorna si la condición es falsa diff --git a/README_ZH_CN.md b/README_ZH_CN.md index 2ac38a2..51689f6 100644 --- a/README_ZH_CN.md +++ b/README_ZH_CN.md @@ -485,26 +485,33 @@ match opt { } ``` -#### 循环 +#### 循環 ```zc -// 区间迭代 -for i in 0..10 { ... } // 左闭右开 (0 到 9) -for i in 0..<10 { ... } // 左闭右开 (显式) -for i in 0..=10 { ... } // 全闭 (0 到 10) +// 區間迭代 +for i in 0..10 { ... } // 左閉右開 (0 到 9) +for i in 0..<10 { ... } // 左閉右開 (顯式) +for i in 0..=10 { ... } // 全閉 (0 到 10) for i in 0..10 step 2 { ... } -// 迭代器 (Vec, Array, 或自定义 Iterable) -for item in collection { ... } +// 迭代器 (Vec 或自定義 Iterable) +for item in vec { ... } -// While 循环 +// 直接迭代固定大小数组 +let arr: int[5] = [1, 2, 3, 4, 5]; +for val in arr { + // val 是 int + println "{val}"; +} + +// While 循環 while x < 10 { ... } -// 带标签的无限循环 +// 帶標籤的無限循環 outer: loop { if done { break outer; } } -// 重复 N 次 +// 重複 N 次 for _ in 0..5 { ... } ``` diff --git a/README_ZH_TW.md b/README_ZH_TW.md index 13591cf..6fa0dbd 100644 --- a/README_ZH_TW.md +++ b/README_ZH_TW.md @@ -493,8 +493,15 @@ for i in 0..<10 { ... } // 左閉右開 (顯式) for i in 0..=10 { ... } // 全閉 (0 到 10) for i in 0..10 step 2 { ... } -// 迭代器 (Vec, Array, 或自定義 Iterable) -for item in collection { ... } +// 迭代器 (Vec 或自定義 Iterable) +for item in vec { ... } + +// 直接迭代固定大小數組 +let arr: int[5] = [1, 2, 3, 4, 5]; +for val in arr { + // val 是 int + println "{val}"; +} // While 循環 while x < 10 { ... } diff --git a/docs/std/slice.md b/docs/std/slice.md new file mode 100644 index 0000000..b70c5fe --- /dev/null +++ b/docs/std/slice.md @@ -0,0 +1,90 @@ +# Standard Library: Slice (`std/slice.zc`) + +`Slice` is a lightweight, non-owning view into a contiguous sequence of elements. It's particularly useful for working with fixed-size arrays and enabling iteration. + +## Usage + +```zc +import "std/slice.zc" + +fn main() { + let arr: int[5] = [1, 2, 3, 4, 5]; + + // Direct iteration (Recommended) + for val in arr { + println "{val}"; + } + + // Manual slice creation (for partial views or specific needs) + let slice = Slice::from_array((int*)(&arr), 5); + for val in slice { + println "{val}"; + } +} +``` + +## Structure + +```zc +struct Slice { + data: T*; + len: usize; +} +``` + +## Methods + +### Construction + +| Method | Signature | Description | +| :--- | :--- | :--- | +| **from_array** | `Slice::from_array(arr: T*, len: usize) -> Slice` | Creates a slice view over an array. | + +### Iteration + +| Method | Signature | Description | +| :--- | :--- | :--- | +| **iterator** | `iterator(self) -> SliceIter` | Returns an iterator for `for-in` loops. | + +`SliceIter` implements the iterator protocol with a `next() -> Option` method. + +### Access & Query + +| Method | Signature | Description | +| :--- | :--- | :--- | +| **length** | `length(self) -> usize` | Returns the number of elements. | +| **is_empty** | `is_empty(self) -> bool` | Returns true if length is 0. | +| **get** | `get(self, idx: usize) -> Option` | Returns the element at index, or None if out of bounds. | +| **at** | `at(self, idx: usize) -> Option` | Alias for `get`. | + +## Examples + +### Iterating over fixed-size arrays + +```zc +let numbers: int[3] = [10, 20, 30]; +let slice = Slice::from_array((int*)(&numbers), 3); + +for n in slice { + println "Number: {n}"; +} +``` + +### Safe indexed access + +```zc +let arr: int[3] = [1, 2, 3]; +let slice = Slice::from_array((int*)(&arr), 3); + +let opt = slice.get(1); +if (!opt.is_none()) { + println "Value: {opt.unwrap()}"; // Prints: Value: 2 +} +``` + +## Notes + +- `Slice` does not own its data - it's just a view +- No memory management needed (no `free()` method) +- Must specify the generic type explicitly: `Slice`, `Slice`, etc. +- The array pointer cast `(T*)(&arr)` is required for fixed-size arrays diff --git a/src/codegen/codegen.c b/src/codegen/codegen.c index a66f179..7a67428 100644 --- a/src/codegen/codegen.c +++ b/src/codegen/codegen.c @@ -65,6 +65,52 @@ static void codegen_var_expr(ParserContext *ctx, ASTNode *node, FILE *out) zwarn_at(node->token, "%s\n = help: %s", msg, help); } } + + // Check for static method call pattern: Type::method or Slice::method + char *double_colon = strstr(node->var_ref.name, "::"); + if (double_colon) + { + // Extract type name and method name + int type_len = double_colon - node->var_ref.name; + char *type_name = xmalloc(type_len + 1); + strncpy(type_name, node->var_ref.name, type_len); + type_name[type_len] = 0; + + char *method_name = double_colon + 2; // Skip :: + + // Handle generic types: Slice -> Slice_int + char mangled_type[256]; + if (strchr(type_name, '<')) + { + // Generic type - need to mangle it + char *lt = strchr(type_name, '<'); + char *gt = strchr(type_name, '>'); + + if (lt && gt) + { + // Extract base type and type argument + *lt = 0; + char *type_arg = lt + 1; + *gt = 0; + + sprintf(mangled_type, "%s_%s", type_name, type_arg); + } + else + { + strcpy(mangled_type, type_name); + } + } + else + { + strcpy(mangled_type, type_name); + } + + // Output as Type__method + fprintf(out, "%s__%s", mangled_type, method_name); + free(type_name); + return; + } + fprintf(out, "%s", node->var_ref.name); } @@ -348,6 +394,7 @@ void codegen_expression(ParserContext *ctx, ASTNode *node, FILE *out) } // Check for Static Enum Variant Call: Enum.Variant(...) + if (target->type == NODE_EXPR_VAR) { ASTNode *def = find_struct_def(ctx, target->var_ref.name); @@ -418,11 +465,43 @@ void codegen_expression(ParserContext *ctx, ASTNode *node, FILE *out) base += 7; } + char *mangled_base = base; + char base_buf[256]; + + // Mangle generic types: Slice -> Slice_int, Vec -> Vec_Point + char *lt = strchr(base, '<'); + if (lt) + { + char *gt = strchr(lt, '>'); + if (gt) + { + int prefix_len = lt - base; + int arg_len = gt - lt - 1; + snprintf(base_buf, 255, "%.*s_%.*s", prefix_len, base, arg_len, lt + 1); + mangled_base = base_buf; + } + } + if (!strchr(type, '*') && target->type == NODE_EXPR_CALL) { - fprintf(out, "({ %s _t = ", type); + char *type_mangled = type; + char type_buf[256]; + char *t_lt = strchr(type, '<'); + if (t_lt) + { + char *t_gt = strchr(t_lt, '>'); + if (t_gt) + { + int p_len = t_lt - type; + int a_len = t_gt - t_lt - 1; + snprintf(type_buf, 255, "%.*s_%.*s", p_len, type, a_len, t_lt + 1); + type_mangled = type_buf; + } + } + + fprintf(out, "({ %s _t = ", type_mangled); codegen_expression(ctx, target, out); - fprintf(out, "; %s__%s(&_t", base, method); + fprintf(out, "; %s__%s(&_t", mangled_base, method); ASTNode *arg = node->call.args; while (arg) { @@ -435,10 +514,11 @@ void codegen_expression(ParserContext *ctx, ASTNode *node, FILE *out) else { // Mixin Lookup Logic - char *call_base = base; + char *call_base = mangled_base; + int need_cast = 0; char mixin_func_name[128]; - sprintf(mixin_func_name, "%s__%s", base, method); + sprintf(mixin_func_name, "%s__%s", call_base, method); char *resolved_method_suffix = NULL; diff --git a/src/codegen/codegen_utils.c b/src/codegen/codegen_utils.c index 0d03661..8d9cb28 100644 --- a/src/codegen/codegen_utils.c +++ b/src/codegen/codegen_utils.c @@ -64,13 +64,28 @@ void emit_c_decl(FILE *out, const char *type_str, const char *name) } else if (generic && (!bracket || generic < bracket)) { - // Strip generic part for C output - int base_len = generic - type_str; - fprintf(out, "%.*s %s", base_len, type_str, name); + char *gt = strchr(generic, '>'); + if (gt) + { + int base_len = generic - type_str; + int arg_len = gt - generic - 1; + + fprintf(out, "%.*s_%.*s %s", base_len, type_str, arg_len, generic + 1, name); - if (bracket) + if (bracket) + { + fprintf(out, "%s", bracket); + } + } + else { - fprintf(out, "%s", bracket); + int base_len = generic - type_str; + fprintf(out, "%.*s %s", base_len, type_str, name); + + if (bracket) + { + fprintf(out, "%s", bracket); + } } } else if (bracket) diff --git a/src/parser/parser_stmt.c b/src/parser/parser_stmt.c index a471fe6..4c24de3 100644 --- a/src/parser/parser_stmt.c +++ b/src/parser/parser_stmt.c @@ -1133,6 +1133,7 @@ ASTNode *parse_for(ParserContext *ctx, Lexer *l) ASTNode *obj_expr = start_expr; char *iter_method = "iterator"; + ASTNode *slice_decl = NULL; // Track if we need to add a slice declaration // Check for reference iteration: for x in &vec if (obj_expr->type == NODE_EXPR_UNARY && obj_expr->unary.op && @@ -1142,6 +1143,78 @@ ASTNode *parse_for(ParserContext *ctx, Lexer *l) iter_method = "iter_ref"; } + // Check for array iteration: wrap with Slice::from_array + if (obj_expr->type_info && obj_expr->type_info->kind == TYPE_ARRAY && + obj_expr->type_info->array_size > 0) + { + // Create a var decl for the slice + slice_decl = ast_create(NODE_VAR_DECL); + slice_decl->var_decl.name = xstrdup("__zc_arr_slice"); + + // Build type string: Slice + char *elem_type_str = type_to_string(obj_expr->type_info->inner); + char slice_type[256]; + sprintf(slice_type, "Slice<%s>", elem_type_str); + slice_decl->var_decl.type_str = xstrdup(slice_type); + + ASTNode *from_array_call = ast_create(NODE_EXPR_CALL); + ASTNode *static_method = ast_create(NODE_EXPR_VAR); + + // The function name for static methods is Type::method format + char func_name[512]; + snprintf(func_name, 511, "%s::from_array", slice_type); + static_method->var_ref.name = xstrdup(func_name); + + from_array_call->call.callee = static_method; + + // Create arguments + ASTNode *arr_addr = ast_create(NODE_EXPR_UNARY); + arr_addr->unary.op = xstrdup("&"); + arr_addr->unary.operand = obj_expr; + + ASTNode *arr_cast = ast_create(NODE_EXPR_CAST); + char cast_type[256]; + sprintf(cast_type, "%s*", elem_type_str); + arr_cast->cast.target_type = xstrdup(cast_type); + arr_cast->cast.expr = arr_addr; + + ASTNode *size_arg = ast_create(NODE_EXPR_LITERAL); + size_arg->literal.type_kind = LITERAL_INT; + size_arg->literal.int_val = obj_expr->type_info->array_size; + char size_buf[32]; + sprintf(size_buf, "%d", obj_expr->type_info->array_size); + size_arg->literal.string_val = xstrdup(size_buf); + + arr_cast->next = size_arg; + from_array_call->call.args = arr_cast; + from_array_call->call.arg_count = 2; + + slice_decl->var_decl.init_expr = from_array_call; + + // Manually trigger generic instantiation for Slice + // This ensures that Slice_int, Slice_float, etc. structures are generated + Token dummy_tok = {0}; + instantiate_generic(ctx, "Slice", elem_type_str, elem_type_str, dummy_tok); + + // Instantiate SliceIter and Option too for the loop logic + char iter_type[256]; + sprintf(iter_type, "SliceIter<%s>", elem_type_str); + instantiate_generic(ctx, "SliceIter", elem_type_str, elem_type_str, dummy_tok); + + char option_type[256]; + sprintf(option_type, "Option<%s>", elem_type_str); + instantiate_generic(ctx, "Option", elem_type_str, elem_type_str, dummy_tok); + + // Replace obj_expr with a reference to the slice variable + ASTNode *slice_ref = ast_create(NODE_EXPR_VAR); + slice_ref->var_ref.name = xstrdup("__zc_arr_slice"); + slice_ref->resolved_type = + xstrdup(slice_type); // Explicitly set type for codegen + obj_expr = slice_ref; + + free(elem_type_str); + } + // var __it = obj.iterator(); ASTNode *it_decl = ast_create(NODE_VAR_DECL); it_decl->var_decl.name = xstrdup("__it"); @@ -1182,6 +1255,34 @@ ASTNode *parse_for(ParserContext *ctx, Lexer *l) stmts_tail = node; \ } + char *iter_type_ptr = NULL; + char *option_type_ptr = NULL; + + if (slice_decl) + { + char *slice_t = slice_decl->var_decl.type_str; + char *start = strchr(slice_t, '<'); + if (start) + { + char *end = strrchr(slice_t, '>'); + if (end) + { + int len = end - start - 1; + char *elem = xmalloc(len + 1); + strncpy(elem, start + 1, len); + elem[len] = 0; + + iter_type_ptr = xmalloc(256); + sprintf(iter_type_ptr, "SliceIter<%s>", elem); + + option_type_ptr = xmalloc(256); + sprintf(option_type_ptr, "Option<%s>", elem); + + free(elem); + } + } + } + // var __opt = __it.next(); ASTNode *opt_decl = ast_create(NODE_VAR_DECL); opt_decl->var_decl.name = xstrdup("__opt"); @@ -1192,6 +1293,10 @@ ASTNode *parse_for(ParserContext *ctx, Lexer *l) ASTNode *memb_next = ast_create(NODE_EXPR_MEMBER); ASTNode *it_ref = ast_create(NODE_EXPR_VAR); it_ref->var_ref.name = xstrdup("__it"); + if (iter_type_ptr) + { + it_ref->resolved_type = xstrdup(iter_type_ptr); + } memb_next->member.target = it_ref; memb_next->member.field = xstrdup("next"); call_next->call.callee = memb_next; @@ -1204,15 +1309,22 @@ ASTNode *parse_for(ParserContext *ctx, Lexer *l) ASTNode *memb_is_none = ast_create(NODE_EXPR_MEMBER); ASTNode *opt_ref1 = ast_create(NODE_EXPR_VAR); opt_ref1->var_ref.name = xstrdup("__opt"); + if (option_type_ptr) + { + opt_ref1->resolved_type = xstrdup(option_type_ptr); + } memb_is_none->member.target = opt_ref1; memb_is_none->member.field = xstrdup("is_none"); call_is_none->call.callee = memb_is_none; + call_is_none->call.args = NULL; + call_is_none->call.arg_count = 0; - ASTNode *break_stmt = ast_create(NODE_BREAK); - + // if (__opt.is_none()) break; ASTNode *if_break = ast_create(NODE_IF); if_break->if_stmt.condition = call_is_none; + ASTNode *break_stmt = ast_create(NODE_BREAK); if_break->if_stmt.then_body = break_stmt; + if_break->if_stmt.else_body = NULL; APPEND_STMT(if_break); // var = __opt.unwrap(); @@ -1225,25 +1337,28 @@ ASTNode *parse_for(ParserContext *ctx, Lexer *l) ASTNode *memb_unwrap = ast_create(NODE_EXPR_MEMBER); ASTNode *opt_ref2 = ast_create(NODE_EXPR_VAR); opt_ref2->var_ref.name = xstrdup("__opt"); + if (option_type_ptr) + { + opt_ref2->resolved_type = xstrdup(option_type_ptr); + } memb_unwrap->member.target = opt_ref2; memb_unwrap->member.field = xstrdup("unwrap"); call_unwrap->call.callee = memb_unwrap; + call_unwrap->call.args = NULL; + call_unwrap->call.arg_count = 0; user_var_decl->var_decl.init_expr = call_unwrap; APPEND_STMT(user_var_decl); - // User Body + // User body statements enter_scope(ctx); add_symbol(ctx, var_name, NULL, NULL); - ASTNode *user_body_node; - if (lexer_peek(l).type == TOK_LBRACE) - { - user_body_node = parse_block(ctx, l); - } - else + // Body block + ASTNode *stmt = parse_statement(ctx, l); + ASTNode *user_body_node = stmt; + if (stmt && stmt->type != NODE_BLOCK) { - ASTNode *stmt = parse_statement(ctx, l); ASTNode *blk = ast_create(NODE_BLOCK); blk->block.statements = stmt; user_body_node = blk; @@ -1256,10 +1371,21 @@ ASTNode *parse_for(ParserContext *ctx, Lexer *l) loop_body->block.statements = stmts_head; while_loop->while_stmt.body = loop_body; - // Wrap entire thing in a block to scope _it + // Wrap entire thing in a block to scope __it (and __zc_arr_slice if present) ASTNode *outer_block = ast_create(NODE_BLOCK); - it_decl->next = while_loop; - outer_block->block.statements = it_decl; + if (slice_decl) + { + // Chain: slice_decl -> it_decl -> while_loop + slice_decl->next = it_decl; + it_decl->next = while_loop; + outer_block->block.statements = slice_decl; + } + else + { + // Chain: it_decl -> while_loop + it_decl->next = while_loop; + outer_block->block.statements = it_decl; + } return outer_block; } diff --git a/std/slice.zc b/std/slice.zc index 778c6ed..7ace396 100644 --- a/std/slice.zc +++ b/std/slice.zc @@ -1,10 +1,45 @@ +import "./option.zc" + struct Slice { data: T*; len: usize; } +struct SliceIter { + data: T*; + count: usize; + idx: usize; +} + +impl SliceIter { + fn next(self) -> Option { + if (self.idx < self.count) { + let item = self.data[self.idx]; + self.idx = self.idx + 1; + return Option::Some(item); + } + return Option::None(); + } + + fn iterator(self) -> SliceIter { + return *self; + } +} + impl Slice { + fn from_array(arr: T*, len: usize) -> Slice { + return Slice { data: arr, len: len }; + } + + fn iterator(self) -> SliceIter { + return SliceIter { + data: self.data, + count: self.len, + idx: 0 + }; + } + fn length(self) -> usize { return self.len; } diff --git a/tests/std/test_direct_array_iteration.zc b/tests/std/test_direct_array_iteration.zc new file mode 100644 index 0000000..359951f --- /dev/null +++ b/tests/std/test_direct_array_iteration.zc @@ -0,0 +1,37 @@ +import "std/slice.zc" + +test "direct array iteration" { + let arr: int[5] = [1, 2, 3, 4, 5]; + + let sum = 0; + for val in arr { + sum = sum + val; + } + + assert(sum == 15, "Sum should be 1+2+3+4+5 = 15"); +} + +test "direct array iteration with different types" { + let floats: float[3] = [1.5, 2.5, 3.0]; + let count = 0; + + for f in floats { + count = count + 1; + } + + assert(count == 3, "Should iterate over all 3 elements"); +} + +// TODO: Nested array iteration needs special handling +// test "nested array iteration" { +// let matrix: int[2][3] = [[1, 2, 3], [4, 5, 6]]; +// let total = 0; +// +// for row in matrix { +// for val in row { +// total = total + val; +// } +// } +// +// assert(total == 21, "Sum should be 1+2+3+4+5+6 = 21"); +// } diff --git a/tests/std/test_slice_iteration.zc b/tests/std/test_slice_iteration.zc new file mode 100644 index 0000000..b7eddf4 --- /dev/null +++ b/tests/std/test_slice_iteration.zc @@ -0,0 +1,29 @@ +import "std/slice.zc" +import "std/io.zc" + +test "slice from array iteration" { + let ints: int[5] = [1, 2, 3, 4, 5]; + let slice = Slice::from_array((int*)(&ints), 5); + + // Test iteration + let sum = 0; + for val in slice { + sum = sum + val; + } + + if (sum != 15) { + panic("Slice iteration failed: expected sum 15"); + } +} + +test "slice methods" { + let arr: int[3] = [10, 20, 30]; + let slice = Slice::from_array((int*)(&arr), 3); + + if (slice.length() != 3) panic("Slice length wrong"); + if (slice.is_empty()) panic("Slice should not be empty"); + + let opt = slice.get(1); + if (opt.is_none()) panic("Slice get failed"); + if (opt.unwrap() != 20) panic("Slice get returned wrong value"); +} -- cgit v1.2.3 From 71297486445992de9d97affabc77908da79e8c89 Mon Sep 17 00:00:00 2001 From: Zuhaitz Méndez Fernández de Aránguiz Date: Sat, 31 Jan 2026 01:33:46 +0000 Subject: Fix compilation error --- src/parser/parser_stmt.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'src/parser') diff --git a/src/parser/parser_stmt.c b/src/parser/parser_stmt.c index 4c24de3..7758ae3 100644 --- a/src/parser/parser_stmt.c +++ b/src/parser/parser_stmt.c @@ -3305,7 +3305,7 @@ char *run_comptime_block(ParserContext *ctx, Lexer *l) ASTNode *fn = ref->node; if (fn && fn->type == NODE_FUNCTION && fn->func.is_comptime) { - emit_func_signature(f, fn, NULL); + emit_func_signature(ctx, f, fn, NULL); fprintf(f, ";\n"); codegen_node_single(ctx, fn, f); } -- cgit v1.2.3 From 051400c70a4d5384923113cfbcbc69e8e58d27a0 Mon Sep 17 00:00:00 2001 From: Zuhaitz Méndez Fernández de Aránguiz Date: Sat, 31 Jan 2026 11:58:40 +0000 Subject: Github copilot instructions --- .github/copilot-instructions.md | 87 +++++++++++++++++++++++++++++++++++++++++ README.md | 2 +- src/parser/parser_expr.c | 1 + 3 files changed, 89 insertions(+), 1 deletion(-) create mode 100644 .github/copilot-instructions.md (limited to 'src/parser') diff --git a/.github/copilot-instructions.md b/.github/copilot-instructions.md new file mode 100644 index 0000000..fac102a --- /dev/null +++ b/.github/copilot-instructions.md @@ -0,0 +1,87 @@ +# Zen C Copilot Instructions + +These instructions are **MANDATORY** for all code generation and review tasks in the Zen C project. + +## 1. Memory Management (Critical) +* **Arena Allocation**: The compiler uses a bump-pointer arena allocator. + * **MUST USE**: `xmalloc`, `xcalloc`, `xrealloc`, `xstrdup` (defined in `src/utils/utils.c`). + * **NEVER USE**: Standard `malloc`, `calloc`, `free` (unless interfacing with an external library that strictly requires owned heap memory, like CJSON). +* **Destructors/Freeing**: + * `free(ptr)` is `#defined` to `((void)0)` in `src/zprep.h`. It is a no-op. + * Do **NOT** attempt to free AST nodes or types. `ast_free` is a no-op. + * Memory is reclaimed only when the process exits. Design your data structures accordingly (append-only is fine). + +## 2. AST and Type System +* **Creation**: Use `ast_create(NODE_TYPE)` to allocate new nodes. +* **Type Representation**: + * Use `Type` struct (defined in `src/ast/ast.h`). + * Use `type_new(TYPE_KIND)` helper. + * `type_to_string(t)` and `type_to_c_string(t)` return arena-allocated strings. Do not worry about freeing them. +* **Traversal**: + * The compiler uses **Recursive Descent** with **Switch Statements** on `node->type`. + * Do NOT introduce Visitor patterns or callback tables unless consistent with existing code. + +## 3. Parser Patterns +* **Context**: `ParserContext *ctx` is the god-object. It MUST be passed to almost every function in parsing, analysis, and codegen. + * **Signature Rule**: `ReturnType func_name(ParserContext *ctx, ...)` +* **Token Consumption**: + * Use `expect(lexer, TOKEN_TYPE, "error message")` for mandatory tokens. + * For optional tokens, check `l->token.type` and assume `lexer_next(l)` is used to advance (verify specific helper availability). +* **Error Handling**: + * **Fatal**: `zpanic("msg")` or `zpanic_at(token, "msg")`. Exits immediately (or delegates to LSP handler). + * **Warning**: `zwarn("msg")` or `zwarn_at(token, "msg")`. + * **Semantic Errors**: Prefer `zpanic_at` for type errors to give line/col info. + +## 4. Code Generation (C Backend) +* **Generic Mangling**: + * Generic structs (e.g., `Slice`) are mangled to `Slice_int` **IF AND ONLY IF** the instantiated struct exists (checked via `find_struct_def_codegen`). + * **Fallback**: If the mangled struct (e.g. `Async_int`) does not exist, use the base name (`Async`). This handles opaque types like `Async` correctly. + * See `emit_c_decl` in `src/codegen/codegen_utils.c` for the canonical implementation. +* **Function Signatures**: + * Use `emit_func_signature(ctx, out, node, override_name)` to handle modifiers, return types, and arguments correctly. +* **Output**: Use `fprintf(out, ...)` where `out` is the `FILE*`. + +## 5. Coding Style & Conventions +* **Formatting**: + * Indentation: 4 spaces. + * Braces: **ALWAYS** Use braces `{}` for control flow (`if`, `while`, `for`), even for single lines. + * Opening brace on the **NEXT** line (Allman style). +* **Naming**: + * Structs/Enums: `PascalCase`. + * Functions/Variables: `snake_case`. + * Macros/Constants: `SCREAMING_SNAKE_CASE`. + * Private/Static: No strict prefix, but `static` keyword is mandatory for internal file-scope functions. +* **Iterators**: + * When implementing iteration in compiler (C code): Use `while (node) { ... node = node->next; }` for linked lists (`ASTNode`, `StructRef`). + +## 6. Standard Library (Zen C Definitions) +* **Arrays**: Use `for val in arr` syntax (direct iteration). +* **Vectors**: `Vec` is a dynamic array. +* **Strings**: `string` in Zen C maps to `char*` in C. `kstring` or `zstr` are higher-level wrappers. + +## 7. Common Pitfalls +* **Unused Variables**: The compiler builds with `-Wall -Werror` (or similar strictness). Cast unused vars to void: `(void)var_name;`. +## 8. Zen C Language Rules (For writing .zc files) +* **Syntax**: + * Variables: `let x = 10;`, `let y: const int = 20;`. + * Constants: `def MAX = 100;` (compile-time). + * Functions: `fn name(arg: type) -> ret { ... }`. + * Structs: `struct Point { x: int; y: int; }`. + * Enums: `enum Shape { Circle(float), Rect(float, float) }`. +* **Memory & Resources**: + * **Move Semantics**: Structs/Enums are moved by default on assignment/pass-by-value. + * **Defer**: Use `defer stmt;` to run cleanup at scope exit. + * **Drop**: Implement `impl Drop for T` for RAII. +* **Arrays & Slices**: + * **Iteration**: Use `for val in arr` (direct iteration supported). + * **Slices**: `Slice` is a view. `int[N]` auto-converts to slice in loops. +* **Generics**: + * Syntax: `struct Box { val: T; }`. +* **Concurrency**: + * Use `async fn` and `await` keyword. + * `Async` is the opaque future type. +* **Standard Library**: + * Import with `import "std/io.zc"`, `import "std/vec.zc"`. + * Use `println "msg"` (shorthand) or `printf`. +* **Testing**: + * Use `test "name" { ... }` blocks for unit tests. diff --git a/README.md b/README.md index ecdf8fc..bf4962e 100644 --- a/README.md +++ b/README.md @@ -165,7 +165,7 @@ zc repl ### Environment Variables -You can set `ZC_ROOT` to specify the location of the Standard Library (standard imports like `import "std/vector.zc"`). This allows you to run `zc` from any directory. +You can set `ZC_ROOT` to specify the location of the Standard Library (standard imports like `import "std/vec.zc"`). This allows you to run `zc` from any directory. ```bash export ZC_ROOT=/path/to/Zen-C diff --git a/src/parser/parser_expr.c b/src/parser/parser_expr.c index 6156cc0..28dc465 100644 --- a/src/parser/parser_expr.c +++ b/src/parser/parser_expr.c @@ -5421,6 +5421,7 @@ ASTNode *parse_expr_prec(ParserContext *ctx, Lexer *l, Precedence min_prec) // This gives a warning as "unused" but it's needed for the rewrite. char *r_name = resolve_struct_name_from_type(ctx, rhs->type_info, &is_rhs_ptr, &r_alloc); + (void)r_name; if (r_alloc) { free(r_alloc); -- cgit v1.2.3 From ccc53b11a0e273f46cb40e5f0eb32a74ab6750bf Mon Sep 17 00:00:00 2001 From: Zuhaitz Méndez Fernández de Aránguiz Date: Sat, 31 Jan 2026 15:31:41 +0000 Subject: Fix for #159 --- docs/std/slice.md | 15 +++-- src/codegen/codegen_main.c | 33 +++++++++++ src/parser/parser_stmt.c | 114 ++++++++++++++++++++++++++++++++++++ src/parser/parser_struct.c | 115 +++++++++++++++++++++++++++++++++++++ std/mem.zc | 23 +------- std/slice.zc | 5 ++ tests/memory/test_memory_safety.zc | 9 ++- 7 files changed, 283 insertions(+), 31 deletions(-) (limited to 'src/parser') diff --git a/docs/std/slice.md b/docs/std/slice.md index b70c5fe..f029995 100644 --- a/docs/std/slice.md +++ b/docs/std/slice.md @@ -10,12 +10,12 @@ import "std/slice.zc" fn main() { let arr: int[5] = [1, 2, 3, 4, 5]; - // Direct iteration (Recommended) + // Direct iteration (auto-imports std/slice.zc) for val in arr { println "{val}"; } - // Manual slice creation (for partial views or specific needs) + // Manual slice creation let slice = Slice::from_array((int*)(&arr), 5); for val in slice { println "{val}"; @@ -39,6 +39,7 @@ struct Slice { | Method | Signature | Description | | :--- | :--- | :--- | | **from_array** | `Slice::from_array(arr: T*, len: usize) -> Slice` | Creates a slice view over an array. | +| **new** | `Slice::new(data: T*, len: usize) -> Slice` | Alias for `from_array` (backwards compat). | ### Iteration @@ -62,10 +63,10 @@ struct Slice { ### Iterating over fixed-size arrays ```zc +// std/slice.zc is auto-imported when using for-in on arrays let numbers: int[3] = [10, 20, 30]; -let slice = Slice::from_array((int*)(&numbers), 3); -for n in slice { +for n in numbers { println "Number: {n}"; } ``` @@ -73,6 +74,8 @@ for n in slice { ### Safe indexed access ```zc +import "std/slice.zc" + let arr: int[3] = [1, 2, 3]; let slice = Slice::from_array((int*)(&arr), 3); @@ -84,7 +87,7 @@ if (!opt.is_none()) { ## Notes -- `Slice` does not own its data - it's just a view +- `Slice` does not own its data — it's just a view - No memory management needed (no `free()` method) -- Must specify the generic type explicitly: `Slice`, `Slice`, etc. +- **Auto-import**: `std/slice.zc` is automatically imported when using `for val in arr` on a fixed-size array - The array pointer cast `(T*)(&arr)` is required for fixed-size arrays diff --git a/src/codegen/codegen_main.c b/src/codegen/codegen_main.c index b298700..82fc3ce 100644 --- a/src/codegen/codegen_main.c +++ b/src/codegen/codegen_main.c @@ -448,6 +448,39 @@ void codegen_node(ParserContext *ctx, ASTNode *node, FILE *out) emit_type_aliases(kids, out); // Emit local aliases (redundant but safe) emit_trait_defs(kids, out); + // Also emit traits from parsed_globals_list (from auto-imported files like std/mem.zc) + // but only if they weren't already emitted from kids + StructRef *trait_ref = ctx->parsed_globals_list; + while (trait_ref) + { + if (trait_ref->node && trait_ref->node->type == NODE_TRAIT) + { + // Check if this trait was already in kids (explicitly imported) + int already_in_kids = 0; + ASTNode *k = kids; + while (k) + { + if (k->type == NODE_TRAIT && k->trait.name && trait_ref->node->trait.name && + strcmp(k->trait.name, trait_ref->node->trait.name) == 0) + { + already_in_kids = 1; + break; + } + k = k->next; + } + + if (!already_in_kids) + { + // Create a temporary single-node list for emit_trait_defs + ASTNode *saved_next = trait_ref->node->next; + trait_ref->node->next = NULL; + emit_trait_defs(trait_ref->node, out); + trait_ref->node->next = saved_next; + } + } + trait_ref = trait_ref->next; + } + // Track emitted raw statements to prevent duplicates EmittedContent *emitted_raw = NULL; diff --git a/src/parser/parser_stmt.c b/src/parser/parser_stmt.c index 7758ae3..0677cf5 100644 --- a/src/parser/parser_stmt.c +++ b/src/parser/parser_stmt.c @@ -14,6 +14,118 @@ char *curr_func_ret = NULL; char *run_comptime_block(ParserContext *ctx, Lexer *l); +extern char *g_current_filename; + +/** + * @brief Auto-imports std/slice.zc if not already imported. + * + * This is called when array iteration is detected in for-in loops, + * to ensure the Slice, SliceIter, and Option templates are available. + */ +static void auto_import_std_slice(ParserContext *ctx) +{ + // Check if already imported via templates + GenericTemplate *t = ctx->templates; + while (t) + { + if (strcmp(t->name, "Slice") == 0) + { + return; // Already have the Slice template + } + t = t->next; + } + + // Try to find and import std/slice.zc + static const char *std_paths[] = {"std/slice.zc", "./std/slice.zc", NULL}; + static const char *system_paths[] = {"/usr/local/share/zenc", "/usr/share/zenc", NULL}; + + char resolved_path[1024]; + int found = 0; + + // First, try relative to current file + if (g_current_filename) + { + char *current_dir = xstrdup(g_current_filename); + char *last_slash = strrchr(current_dir, '/'); + if (last_slash) + { + *last_slash = 0; + snprintf(resolved_path, sizeof(resolved_path), "%s/std/slice.zc", current_dir); + if (access(resolved_path, R_OK) == 0) + { + found = 1; + } + } + free(current_dir); + } + + // Try relative paths + if (!found) + { + for (int i = 0; std_paths[i] && !found; i++) + { + if (access(std_paths[i], R_OK) == 0) + { + strncpy(resolved_path, std_paths[i], sizeof(resolved_path) - 1); + resolved_path[sizeof(resolved_path) - 1] = '\0'; + found = 1; + } + } + } + + // Try system paths + if (!found) + { + for (int i = 0; system_paths[i] && !found; i++) + { + snprintf(resolved_path, sizeof(resolved_path), "%s/std/slice.zc", system_paths[i]); + if (access(resolved_path, R_OK) == 0) + { + found = 1; + } + } + } + + if (!found) + { + return; // Could not find std/slice.zc, instantiate_generic will error + } + + // Canonicalize path + char *real_fn = realpath(resolved_path, NULL); + if (real_fn) + { + strncpy(resolved_path, real_fn, sizeof(resolved_path) - 1); + resolved_path[sizeof(resolved_path) - 1] = '\0'; + free(real_fn); + } + + // Check if already imported + if (is_file_imported(ctx, resolved_path)) + { + return; + } + mark_file_imported(ctx, resolved_path); + + // Load and parse the file + char *src = load_file(resolved_path); + if (!src) + { + return; // Could not load file + } + + Lexer i; + lexer_init(&i, src); + + // Save and restore filename context + char *saved_fn = g_current_filename; + g_current_filename = resolved_path; + + // Parse the slice module contents + parse_program_nodes(ctx, &i); + + g_current_filename = saved_fn; +} static void check_assignment_condition(ASTNode *cond) { @@ -1193,6 +1305,8 @@ ASTNode *parse_for(ParserContext *ctx, Lexer *l) // Manually trigger generic instantiation for Slice // This ensures that Slice_int, Slice_float, etc. structures are generated + // First, ensure std/slice.zc is imported (auto-import if needed) + auto_import_std_slice(ctx); Token dummy_tok = {0}; instantiate_generic(ctx, "Slice", elem_type_str, elem_type_str, dummy_tok); diff --git a/src/parser/parser_struct.c b/src/parser/parser_struct.c index 109eeee..e53b56c 100644 --- a/src/parser/parser_struct.c +++ b/src/parser/parser_struct.c @@ -12,6 +12,114 @@ #include "zprep_plugin.h" #include "../codegen/codegen.h" +extern char *g_current_filename; + +/** + * @brief Auto-imports std/mem.zc if not already imported. + * + * This is called when the Drop trait is used (impl Drop for X). + */ +static void auto_import_std_mem(ParserContext *ctx) +{ + // Check if Drop trait is already registered (means mem.zc was imported) + if (check_impl(ctx, "Drop", "__trait_marker__")) + { + // Check_impl returns 0 if not found, but we need a different check + // Let's check if we can find any indicator that mem.zc was loaded + } + + // Try to find and import std/mem.zc + static const char *std_paths[] = {"std/mem.zc", "./std/mem.zc", NULL}; + static const char *system_paths[] = {"/usr/local/share/zenc", "/usr/share/zenc", NULL}; + + char resolved_path[1024]; + int found = 0; + + // First, try relative to current file + if (g_current_filename) + { + char *current_dir = xstrdup(g_current_filename); + char *last_slash = strrchr(current_dir, '/'); + if (last_slash) + { + *last_slash = 0; + snprintf(resolved_path, sizeof(resolved_path), "%s/std/mem.zc", current_dir); + if (access(resolved_path, R_OK) == 0) + { + found = 1; + } + } + free(current_dir); + } + + // Try relative paths + if (!found) + { + for (int i = 0; std_paths[i] && !found; i++) + { + if (access(std_paths[i], R_OK) == 0) + { + strncpy(resolved_path, std_paths[i], sizeof(resolved_path) - 1); + resolved_path[sizeof(resolved_path) - 1] = '\0'; + found = 1; + } + } + } + + // Try system paths + if (!found) + { + for (int i = 0; system_paths[i] && !found; i++) + { + snprintf(resolved_path, sizeof(resolved_path), "%s/std/mem.zc", system_paths[i]); + if (access(resolved_path, R_OK) == 0) + { + found = 1; + } + } + } + + if (!found) + { + return; // Could not find std/mem.zc + } + + // Canonicalize path + char *real_fn = realpath(resolved_path, NULL); + if (real_fn) + { + strncpy(resolved_path, real_fn, sizeof(resolved_path) - 1); + resolved_path[sizeof(resolved_path) - 1] = '\0'; + free(real_fn); + } + + // Check if already imported + if (is_file_imported(ctx, resolved_path)) + { + return; + } + mark_file_imported(ctx, resolved_path); + + // Load and parse the file + char *src = load_file(resolved_path); + if (!src) + { + return; // Could not load file + } + + Lexer i; + lexer_init(&i, src); + + // Save and restore filename context + char *saved_fn = g_current_filename; + g_current_filename = resolved_path; + + // Parse the mem module contents + parse_program_nodes(ctx, &i); + + g_current_filename = saved_fn; +} + // Trait Parsing ASTNode *parse_trait(ParserContext *ctx, Lexer *l) { @@ -149,6 +257,7 @@ ASTNode *parse_trait(ParserContext *ctx, Lexer *l) } register_trait(name); + add_to_global_list(ctx, n_node); // Track for codegen (VTable emission) return n_node; } @@ -206,6 +315,12 @@ ASTNode *parse_impl(ParserContext *ctx, Lexer *l) register_generic(ctx, target_gen_param); } + // Auto-import std/mem.zc if implementing Drop, Copy, or Clone traits + if (strcmp(name1, "Drop") == 0 || strcmp(name1, "Copy") == 0 || strcmp(name1, "Clone") == 0) + { + auto_import_std_mem(ctx); + } + register_impl(ctx, name1, name2); // RAII: Check for "Drop" trait implementation diff --git a/std/mem.zc b/std/mem.zc index 6ee96e8..f1a5f5a 100644 --- a/std/mem.zc +++ b/std/mem.zc @@ -49,28 +49,7 @@ impl Box { } } -struct Slice { - data: T*; - len: usize; -} - -impl Slice { - fn new(data: T*, len: usize) -> Self { - return Self { data: data, len: len }; - } - - fn get(self, i: usize) -> T { - return self.data[i]; - } - - fn set(self, i: usize, val: T) { - self.data[i] = val; - } - - fn is_empty(self) -> bool { - return self.len == 0; - } -} +// Note: Slice is defined in std/slice.zc with iteration support fn mem_zero(ptr: T*, count: usize) { memset(ptr, 0, sizeof(T) * count); diff --git a/std/slice.zc b/std/slice.zc index 7ace396..3c317ca 100644 --- a/std/slice.zc +++ b/std/slice.zc @@ -32,6 +32,11 @@ impl Slice { return Slice { data: arr, len: len }; } + // Alias for backwards compatibility with std/mem.zc + fn new(data: T*, len: usize) -> Slice { + return Slice { data: data, len: len }; + } + fn iterator(self) -> SliceIter { return SliceIter { data: self.data, diff --git a/tests/memory/test_memory_safety.zc b/tests/memory/test_memory_safety.zc index a5cc960..b672cc9 100644 --- a/tests/memory/test_memory_safety.zc +++ b/tests/memory/test_memory_safety.zc @@ -1,5 +1,6 @@ import "std/mem.zc" +import "std/slice.zc" // ** Globals ** let DROP_COUNT = 0; @@ -127,11 +128,13 @@ test "test_slice" { let data: int[5] = [1, 2, 3, 4, 5]; let s = Slice::new(&data[0], 5); f" Slice len: {(int)s.len}"; - let v2 = s.get(2); + let opt_v2 = s.get(2); + let v2 = opt_v2.unwrap(); f" Slice[2]: {v2}"; assert(v2 == 3, "Slice get failed"); - s.set(0, 99); - let v0 = s.get(0); + s.data[0] = 99; + let opt_v0 = s.get(0); + let v0 = opt_v0.unwrap(); f" After set: Slice[0] = {v0}"; assert(v0 == 99, "Slice set failed"); " ✓ Slice works!"; -- cgit v1.2.3 From aced94a89dd732d8ae8fddd27de9e1f1094c449a Mon Sep 17 00:00:00 2001 From: Zuhaitz Méndez Fernández de Aránguiz Date: Sat, 31 Jan 2026 15:48:07 +0000 Subject: Fix for #158 --- src/parser/parser_expr.c | 10 ++++++++++ std/slice.zc | 4 ++-- 2 files changed, 12 insertions(+), 2 deletions(-) (limited to 'src/parser') diff --git a/src/parser/parser_expr.c b/src/parser/parser_expr.c index 28dc465..7c53d96 100644 --- a/src/parser/parser_expr.c +++ b/src/parser/parser_expr.c @@ -5644,7 +5644,17 @@ ASTNode *parse_expr_prec(ParserContext *ctx, Lexer *l, Precedence min_prec) char *t1 = type_to_string(lhs->type_info); char *t2 = type_to_string(rhs->type_info); // Skip type check if either operand is void* (escape hatch type) + // or if either operand is a generic type parameter (T, K, V, etc.) int skip_check = (strcmp(t1, "void*") == 0 || strcmp(t2, "void*") == 0); + if (lhs->type_info->kind == TYPE_GENERIC || rhs->type_info->kind == TYPE_GENERIC) + { + skip_check = 1; + } + // Also check if type name is a single uppercase letter (common generic param) + if ((strlen(t1) == 1 && isupper(t1[0])) || (strlen(t2) == 1 && isupper(t2[0]))) + { + skip_check = 1; + } // Allow comparing pointers/strings with integer literal 0 (NULL) if (!skip_check) diff --git a/std/slice.zc b/std/slice.zc index 3c317ca..c757fbd 100644 --- a/std/slice.zc +++ b/std/slice.zc @@ -28,8 +28,8 @@ impl SliceIter { } impl Slice { - fn from_array(arr: T*, len: usize) -> Slice { - return Slice { data: arr, len: len }; + fn from_array(ptr: T*, len: usize) -> Slice { + return Slice { data: ptr, len: len }; } // Alias for backwards compatibility with std/mem.zc -- cgit v1.2.3