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'. --- src/parser/parser_utils.c | 53 +++++++++++++++++++++++++++++++++++++++++++++-- 1 file changed, 51 insertions(+), 2 deletions(-) (limited to 'src/parser/parser_utils.c') 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; } -- cgit v1.2.3 From fbfce63744882d48ea2fc514ab1594000254db80 Mon Sep 17 00:00:00 2001 From: Zuhaitz Méndez Fernández de Aránguiz Date: Sun, 1 Feb 2026 14:01:51 +0000 Subject: Related to #138 --- README.md | 58 +++++++++++++++++++- README_ES.md | 57 +++++++++++++++++++- README_IT.md | 55 ++++++++++++++++++- README_ZH_CN.md | 58 +++++++++++++++++++- README_ZH_TW.md | 60 ++++++++++++++++++++- src/ast/ast.c | 56 +++++++++++++++++-- src/ast/ast.h | 58 +++++++++++--------- src/codegen/codegen.c | 92 ++++++++++++++++++++++++++++++- src/codegen/codegen_decl.c | 19 +++++-- src/parser/parser_expr.c | 4 +- src/parser/parser_struct.c | 6 +-- src/parser/parser_type.c | 62 ++++++++++++++++++--- src/parser/parser_utils.c | 22 ++++++-- std/fs.zc | 53 ++++++++++-------- std/io.zc | 23 ++++---- std/net.zc | 98 +++++++++++++++++++++------------- std/process.zc | 8 +-- std/string.zc | 9 ++-- std/thread.zc | 35 ++++++------ tests/collections/test_string_suite.zc | 4 +- tests/features/test_portable_types.zc | 46 ++++++++++++++++ 21 files changed, 737 insertions(+), 146 deletions(-) create mode 100644 tests/features/test_portable_types.zc (limited to 'src/parser/parser_utils.c') diff --git a/README.md b/README.md index 87f95af..fc1aa27 100644 --- a/README.md +++ b/README.md @@ -100,6 +100,7 @@ Join the discussion, share demos, ask questions, or report bugs in the official - [Named Constraints](#named-constraints) - [15. Build Directives](#15-build-directives) - [16. Keywords](#16-keywords) + - [17. C Interoperability](#17-c-interoperability) - [Standard Library](#standard-library) - [Tooling](#tooling) - [Language Server (LSP)](#language-server-lsp) @@ -204,7 +205,11 @@ let y: const int = 10; // Read-only (Type qualified) | Type | C Equivalent | Description | |:---|:---|:---| -| `int`, `uint` | `int`, `unsigned int` | Platform standard integer | +| `int`, `uint` | `int32_t`, `uint32_t` | 32-bit signed/unsigned integer | +| `c_char`, `c_uchar` | `char`, `unsigned char` | C char / unsigned char (Interop) | +| `c_short`, `c_ushort` | `short`, `unsigned short` | C short / unsigned short (Interop) | +| `c_int`, `c_uint` | `int`, `unsigned int` | C int / unsigned int (Interop) | +| `c_long`, `c_ulong` | `long`, `unsigned long` | C long / unsigned long (Interop) | | `I8` .. `I128` or `i8` .. `i128` | `int8_t` .. `__int128_t` | Signed fixed-width integers | | `U8` .. `U128` or `u8` .. `u128` | `uint8_t` .. `__uint128_t` | Unsigned fixed-width integers | | `isize`, `usize` | `ptrdiff_t`, `size_t` | Pointer-sized integers | @@ -217,6 +222,12 @@ let y: const int = 10; // Read-only (Type qualified) | `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) | +> **Best Practices for Portable Code** +> +> - Use **Portable Types** (`int`, `uint`, `i64`, `u8`, etc.) for all pure Zen C logic. `int` is guaranteed to be 32-bit signed on all architectures. +> - Use **C Interop Types** (`c_int`, `c_char`, `c_long`) **only** when interacting with C libraries (FFI). Their size varies by platform and C compiler (e.g. `c_long` size differs between Windows and Linux). +> - Use `isize` and `usize` for array indexing and memory pointer arithmetic. + ### 3. Aggregate Types #### Arrays @@ -1092,6 +1103,51 @@ The following identifiers are reserved because they are keywords in C11: #### Operators `and`, `or` +### 17. C Interoperability + +Zen C offers two ways to interact with C code: **Trusted Imports** (Convenient) and **Explicit FFI** (Safe/Precise). + +#### Method 1: Trusted Imports (Convenient) + +You can import a C header directly using the `import` keyword with the `.h` extension. This treats the header as a module and assumes all symbols accessed through it exist. + +```zc +//> link: -lm +import "math.h" as c_math; + +fn main() { + // Compiler trusts correctness; emits 'cos(...)' directly + let x = c_math::cos(3.14159); +} +``` + +> **Pros**: Zero boilerplate. Access everything in the header immediately. +> **Cons**: No type safety from Zen C (errors caught by C compiler later). + +#### Method 2: Explicit FFI (Safe) + +For strict type checking or when you don't want to include the text of a header, use `extern fn`. + +```zc +include // Emits #include in generated C + +// Define strict signature +extern fn printf(fmt: char*, ...) -> c_int; + +fn main() { + printf("Hello FFI: %d\n", 42); // Type checked by Zen C +} +``` + +> **Pros**: Zen C ensures types match. +> **Cons**: Requires manual declaration of functions. + +#### `import` vs `include` + +- **`import "file.h"`**: Registers the header as a named module. Enables implicit access to symbols (for example, `file::function()`). +- **`include `**: Purely emits `#include ` in the generated C code. Does not introduce any symbols to the Zen C compiler; you must use `extern fn` to access them. + + --- ## Standard Library diff --git a/README_ES.md b/README_ES.md index 11c4d89..07234f0 100644 --- a/README_ES.md +++ b/README_ES.md @@ -100,6 +100,7 @@ - [Restricciones con Nombre](#restricciones-con-nombre) - [15. Directivas de Construcción](#15-directivas-de-construcción) - [16. Palabras Clave](#16-palabras-clave) + - [17. Interoperabilidad C](#17-interoperabilidad-c) - [Biblioteca Estándar](#biblioteca-estándar) - [Herramientas](#herramientas) - [Servidor de Lenguaje (LSP)](#servidor-de-lenguaje-lsp) @@ -203,7 +204,11 @@ let y: const int = 10; // Solo lectura (Calificado por tipo) | Tipo | Equivalente en C | Descripción | |:---|:---|:---| -| `int`, `uint` | `int`, `unsigned int` | Entero estándar de la plataforma | +| `int`, `uint` | `int32_t`, `uint32_t` | Entero de 32 bits con signo/sin signo | +| `c_char`, `c_uchar` | `char`, `unsigned char` | C char (Interoperabilidad) | +| `c_short`, `c_ushort` | `short`, `unsigned short` | C short (Interoperabilidad) | +| `c_int`, `c_uint` | `int`, `unsigned int` | C int (Interoperabilidad) | +| `c_long`, `c_ulong` | `long`, `unsigned long` | C long (Interoperabilidad) | | `I8` .. `I128` o `i8` .. `i128` | `int8_t` .. `__int128_t` | Enteros con signo de ancho fijo | | `U8` .. `U128` o `u8` .. `u128` | `uint8_t` .. `__uint128_t` | Enteros sin signo de ancho fijo | | `isize`, `usize` | `ptrdiff_t`, `size_t` | Enteros del tamaño de un puntero | @@ -216,6 +221,12 @@ let y: const int = 10; // Solo lectura (Calificado por tipo) | `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) | +> **Mejores Prácticas para Código Portable** +> +> - Usa **Tipos Portables** (`int`, `uint`, `i64`, `u8`, etc.) para toda la lógica pura de Zen C. `int` garantiza ser 32-bits con signo en todas las arquitecturas. +> - Usa **Tipos de Interoperabilidad C** (`c_int`, `c_char`, `c_long`) **sólo** al interactuar con bibliotecas C (FFI). Su tamaño varía según la plataforma y el compilador C. +> - Usa `isize` y `usize` para indexado de arrays y aritmética de punteros. + ### 3. Tipos Agregados #### Arrays @@ -1092,6 +1103,50 @@ Los siguientes identificadores están reservados porque son palabras clave en C1 #### Operadores `and`, `or` +### 17. Interoperabilidad C +Zen C ofrece dos formas de interactuar con código C: **Importaciones de Confianza** (Conveniente) y **FFI Explícita** (Seguro/Preciso). + +#### Método 1: Importaciones de Confianza (Conveniente) + +Puedes importar una cabecera C directamente usando la palabra clave `import` con la extensión `.h`. Esto trata la cabecera como un módulo y asume que todos los símbolos accedidos existen. + +```zc +//> link: -lm +import "math.h" as c_math; + +fn main() { + // El compilador confía en la corrección; emite 'cos(...)' directamente + let x = c_math::cos(3.14159); +} +``` + +> **Pros**: Cero código repetitivo. Acceso a todo el contenido de la cabecera inmediato. +> **Cons**: Sin seguridad de tipos desde Zen C (errores capturados por el compilador C después). + +#### Método 2: FFI Explícita (Seguro) + +Para una comprobación estricta de tipos o cuando no quieres incluir el texto de una cabecera, usa `extern fn`. + +```zc +include // Emite #include en el C generado + +// Define firma estricta +extern fn printf(fmt: char*, ...) -> c_int; + +fn main() { + printf("Hola FFI: %d\n", 42); // Comprobado por tipos por Zen C +} +``` + +> **Pros**: Zen C asegura que los tipos coincidan. +> **Cons**: Requiere declaración manual de funciones. + +#### `import` vs `include` + +- **`import "file.h"`**: Registra la cabecera como un módulo con nombre. Habilita el acceso implícito a símbolos (ej. `file::function()`). +- **`include `**: Puramente emite `#include ` en el código C generado. No introduce ningún símbolo al compilador de Zen C; debes usar `extern fn` para acceder a ellos. + + --- ## Biblioteca Estándar diff --git a/README_IT.md b/README_IT.md index 9bf804e..be48cda 100644 --- a/README_IT.md +++ b/README_IT.md @@ -101,6 +101,7 @@ Unisciti alla conversazione, condividi delle demo, fai domande o segnala dei bug - [Vincoli Nominati](#vincoli-nominati) - [15. Direttive della Buil](#15-direttive-della-build) - [16. Keyword](#16-keyword) + - [17. Interoperabilità C](#17-interoperabilità-c) - [Libreria Standard](#liberia-standard) - [Tooling](#tooling) - [Language Server (LSP)](#language-server-lsp) @@ -205,7 +206,11 @@ let y: const int = 10; // Sola lettura (Tipo qualificato) | Tipo | C Equivalent | Descrizione | |:---|:---|:---| -| `int`, `uint` | `int`, `unsigned int` | Intero standard della piattaforma | +| `int`, `uint` | `int32_t`, `uint32_t` | Intero a 32 bit con segno/senza segno | +| `c_char`, `c_uchar` | `char`, `unsigned char` | C char (Interop) | +| `c_short`, `c_ushort` | `short`, `unsigned short` | C short (Interop) | +| `c_int`, `c_uint` | `int`, `unsigned int` | C int (Interop) | +| `c_long`, `c_ulong` | `long`, `unsigned long` | C long (Interop) | | `I8` .. `I128` or `i8` .. `i128` | `int8_t` .. `__int128_t` | Interi a grandezza fissa con segno | | `U8` .. `U128` or `u8` .. `u128` | `uint8_t` .. `__uint128_t` | Interi a grandezza fissa senza segno | | `isize`, `usize` | `ptrdiff_t`, `size_t` | Interi con grandezza di un puntatore | @@ -218,6 +223,12 @@ let y: const int = 10; // Sola lettura (Tipo qualificato) | `iN` (Per esempio, `i256`) | `_BitInt(N)` | Intero con segno a larghezza arbitraria di bit (C23) | | `uN` (Per esempio, `u42`) | `unsigned _BitInt(N)` | Intero senza segno a larghezza arbitraria di bit (C23) | +> **Best Practice per Codice Portabile** +> +> - Usa **Tipi Portabili** (`int`, `uint`, `i64`, `u8`, ecc.) per tutta la logica Zen C pura. `int` è garantito essere a 32-bit con segno su tutte le architetture. +> - Usa **Tipi di Interop C** (`c_int`, `c_char`, `c_long`) **solo** quando interagisci con librerie C (FFI). La loro dimensione varia in base alla piattaforma e al compilatore C. +> - Usa `isize` e `usize` per indicizzazione di array e aritmetica dei puntatori. + ### 3. Tipi Aggregati #### Array @@ -1089,6 +1100,48 @@ Gli identifiers seguenti sono riservati poiché sono keyword nello standard C11: #### Operatori `and`, `or` +### 17. Interoperabilità C +Zen C offre due modi per interagire con il codice C: **Import Trusted** (Conveniente) e **FFI Esplicita** (Sicuro/Preciso). + +#### Metodo 1: Import Trusted (Conveniente) +Puoi importare un header C direttamente usando la parola chiave `import` con l'estensione `.h`. Questo tratta l'header come un modulo e assume che tutti i simboli acceduti esistano. + +```zc +//> link: -lm +import "math.h" as c_math; + +fn main() { + // Il compilatore si fida della correttezza; emette 'cos(...)' direttamente + let x = c_math::cos(3.14159); +} +``` + +> **Pro**: Zero boilerplate. Accesso immediato a tutto nell'header. +> **Contro**: Nessuna sicurezza dei tipi da Zen C (errori catturati dal compilatore C dopo). + +#### Metodo 2: FFI Esplicita (Sicuro) +Per un controllo rigoroso dei tipi o quando non vuoi includere il testo di un header, usa `extern fn`. + +```zc +include // Emette #include nel C generato + +// Definisci firma rigorosa +extern fn printf(fmt: char*, ...) -> c_int; + +fn main() { + printf("Ciao FFI: %d\n", 42); // Controllato nei tipi da Zen C +} +``` + +> **Pro**: Zen C assicura che i tipi corrispondano. +> **Contro**: Richiede dichiarazione manuale delle funzioni. + +#### `import` vs `include` + +- **`import "file.h"`**: Registra l'header come un modulo con nome. Abilita l'accesso implicito ai simboli (es. `file::function()`). +- **`include `**: Emette puramente `#include ` nel codice C generato. Non introduce alcun simbolo nel compilatore Zen C; devi usare `extern fn` per accedervi. + + --- ## Libreria Standard diff --git a/README_ZH_CN.md b/README_ZH_CN.md index aa1130b..af83e8c 100644 --- a/README_ZH_CN.md +++ b/README_ZH_CN.md @@ -100,6 +100,7 @@ - [命名约束](#命名约束) - [15. 构建指令](#15-构建指令) - [16. 关键字](#16-关键字) + - [17. C 互操作性](#17-c-互操作性) - [标准库](#标准库) - [工具链](#工具链) - [语言服务器 (LSP)](#语言服务器-lsp) @@ -203,7 +204,11 @@ let y: const int = 10; // 只读 (类型修饰) | 类型 | C 等效类型 | 描述 | |:---|:---|:---| -| `int`, `uint` | `int`, `unsigned int` | 平台标准整数 | +| `int`, `uint` | `int32_t`, `uint32_t` | 32位有符号/无符号整数 | +| `c_char`, `c_uchar` | `char`, `unsigned char` | C char (互操作) | +| `c_short`, `c_ushort` | `short`, `unsigned short` | C short (互操作) | +| `c_int`, `c_uint` | `int`, `unsigned int` | C int (互操作) | +| `c_long`, `c_ulong` | `long`, `unsigned long` | C long (互操作) | | `I8` .. `I128` 或 `i8` .. `i128` | `int8_t` .. `__int128_t` | 有符号固定宽度整数 | | `U8` .. `U128` 或 `u8` .. `u128` | `uint8_t` .. `__uint128_t` | 无符号固定宽度整数 | | `isize`, `usize` | `ptrdiff_t`, `size_t` | 指针大小的整数 | @@ -216,6 +221,12 @@ let y: const int = 10; // 只读 (类型修饰) | `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) | +> **可移植代码最佳实践** +> +> - 对于所有纯 Zen C 逻辑，请使用 **可移植类型** (`int`、`uint`、`i64`、`u8` 等)。`int` 保证在所有架构上都是 32 位有符号整数。 +> - 仅在与 C 库 (FFI) 交互时使用 **C 互操作类型** (`c_int`、`c_char`、`c_long`)。它们的大小因平台和 C 编译器而异。 +> - 使用 `isize` 和 `usize` 进行数组索引和内存指针运算。 + ### 3. 复合类型 #### 数组 @@ -1091,6 +1102,51 @@ fn main() { ... } #### 运算符 `and`, `or` +### 17. C 互操作性 + +Zen C 提供了两种与 C 代码交互的方式：**信任导入 (Trusted Imports)** (方便) 和 **显式 FFI** (安全/精确)。 + +#### 方法 1: 信任导入 (方便) + +你可以使用 `import` 关键字直接导入 `.h` 扩展名的 C 头文件。这会将头文件视为一个模块，并假设通过它访问的所有符号都存在。 + +```zc +//> link: -lm +import "math.h" as c_math; + +fn main() { + // 编译器信任不仅正确；直接生成 'cos(...)' + let x = c_math::cos(3.14159); +} +``` + +> **优点**: 零样板代码。立即访问头文件中的所有内容。 +> **缺点**: Zen C 不提供类型安全 (错误将在稍后由 C 编译器捕获)。 + +#### 方法 2: 显式 FFI (安全) + +对于严格的类型检查，或当你不想包含头文件文本时，请使用 `extern fn`. + +```zc +include // 在生成的 C 代码中发出 #include + +// 定义严格签名 +extern fn printf(fmt: char*, ...) -> c_int; + +fn main() { + printf("Hello FFI: %d\n", 42); // 由 Zen C 进行类型检查 +} +``` + +> **优点**: Zen C 确保类型匹配。 +> **缺点**: 需要手动声明函数。 + +#### `import` vs `include` + +- **`import "file.h"`**: 将头文件注册为命名模块。启用对符号的隐式访问 (例如 `file::function()`)。 +- **`include `**: 纯粹在生成的 C 代码中发出 `#include `。不向 Zen C 编译器引入任何符号；必须使用 `extern fn` 才能访问它们。 + + --- ## 标准库 diff --git a/README_ZH_TW.md b/README_ZH_TW.md index 5d85a76..3a18a53 100644 --- a/README_ZH_TW.md +++ b/README_ZH_TW.md @@ -100,6 +100,7 @@ - [命名約束](#命名約束) - [15. 構建指令](#15-構建指令) - [16. 關鍵字](#16-關鍵字) + - [17. C 互操作性](#17-c-互操作性) - [標準庫](#標準庫) - [工具鏈](#工具鏈) - [語言服務器 (LSP)](#語言服務器-lsp) @@ -203,7 +204,11 @@ let y: const int = 10; // 只讀 (類型修飾) | 類型 | C 等效類型 | 描述 | |:---|:---|:---| -| `int`, `uint` | `int`, `unsigned int` | 平台標準整數 | +| `int`, `uint` | `int32_t`, `uint32_t` | 32位元有號/無號整數 | +| `c_char`, `c_uchar` | `char`, `unsigned char` | C char (互操作) | +| `c_short`, `c_ushort` | `short`, `unsigned short` | C short (互操作) | +| `c_int`, `c_uint` | `int`, `unsigned int` | C int (互操作) | +| `c_long`, `c_ulong` | `long`, `unsigned long` | C long (互操作) | | `I8` .. `I128` 或 `i8` .. `i128` | `int8_t` .. `__int128_t` | 有符號固定寬度整數 | | `U8` .. `U128` 或 `u8` .. `u128` | `uint8_t` .. `__uint128_t` | 無符號固定寬度整數 | | `isize`, `usize` | `ptrdiff_t`, `size_t` | 指針大小的整數 | @@ -214,7 +219,13 @@ let y: const int = 10; // 只讀 (類型修飾) | `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) | +| `uN` (例 `u42`) | `unsigned _BitInt(N)` | 任意位寬無號整數 (C23) | + +> **可移植代碼最佳實踐** +> +> - 對於所有純 Zen C 邏輯，請使用 **可移植類型** (`int`、`uint`、`i64`、`u8` 等)。`int` 保證在所有架構上都是 32 位元有號整數。 +> - 僅在與 C 庫 (FFI) 交互時使用 **C 互操作類型** (`c_int`、`c_char`、`c_long`)。它們的大小因平台和 C 編譯器而異。 +> - 使用 `isize` 和 `usize` 進行數組索引和內存指針運算。 ### 3. 複合類型 @@ -1091,6 +1102,51 @@ fn main() { ... } #### 運算符 `and`, `or` +### 17. C 互操作性 + +Zen C 提供了兩種與 C 代碼交互的方式：**信任導入 (Trusted Imports)** (方便) 和 **顯式 FFI** (安全/精確)。 + +#### 方法 1: 信任導入 (方便) + +你可以使用 `import` 關鍵字直接導入 `.h` 擴展名的 C 頭文件。這會將頭文件視為一個模塊，並假設通過它訪問的所有符號都存在。 + +```zc +//> link: -lm +import "math.h" as c_math; + +fn main() { + // 編譯器信任不僅正確；直接生成 'cos(...)' + let x = c_math::cos(3.14159); +} +``` + +> **優點**: 零樣板代碼。立即訪問頭文件中的所有內容。 +> **缺點**: Zen C 不提供類型安全 (錯誤將在稍後由 C 編譯器捕獲)。 + +#### 方法 2: 顯式 FFI (安全) + +對於嚴格的類型檢查，或當你不想包含頭文件文本時，請使用 `extern fn`. + +```zc +include // 在生成的 C 代碼中發出 #include + +// 定義嚴格簽名 +extern fn printf(fmt: char*, ...) -> c_int; + +fn main() { + printf("Hello FFI: %d\n", 42); // 由 Zen C 進行類型檢查 +} +``` + +> **優點**: Zen C 確保類型匹配。 +> **缺點**: 需要手動聲明函數。 + +#### `import` vs `include` + +- **`import "file.h"`**: 將頭文件註冊為命名模塊。啟用對符號的隱式訪問 (例如 `file::function()`)。 +- **`include `**: 純粹在生成的 C 代碼中發出 `#include `。不向 Zen C 編譯器引入任何符號；必須使用 `extern fn` 才能訪問它們。 + + --- ## 標準庫 diff --git a/src/ast/ast.c b/src/ast/ast.c index 439a9f5..1b35500 100644 --- a/src/ast/ast.c +++ b/src/ast/ast.c @@ -259,6 +259,25 @@ static char *type_to_string_impl(Type *t) return xstrdup("int32_t"); case TYPE_UINT: return xstrdup("unsigned int"); + + // Portable C Types + case TYPE_C_INT: + return xstrdup("c_int"); + case TYPE_C_UINT: + return xstrdup("c_uint"); + case TYPE_C_LONG: + return xstrdup("c_long"); + case TYPE_C_ULONG: + return xstrdup("c_ulong"); + case TYPE_C_SHORT: + return xstrdup("c_short"); + case TYPE_C_USHORT: + return xstrdup("c_ushort"); + case TYPE_C_CHAR: + return xstrdup("c_char"); + case TYPE_C_UCHAR: + return xstrdup("c_uchar"); + case TYPE_INT: return xstrdup("int"); case TYPE_FLOAT: @@ -461,8 +480,29 @@ static char *type_to_c_string_impl(Type *t) return xstrdup("int32_t"); case TYPE_UINT: return xstrdup("unsigned int"); - case TYPE_INT: + + // Portable C Types (Map directly to C types) + case TYPE_C_INT: return xstrdup("int"); + case TYPE_C_UINT: + return xstrdup("unsigned int"); + case TYPE_C_LONG: + return xstrdup("long"); + case TYPE_C_ULONG: + return xstrdup("unsigned long"); + case TYPE_C_SHORT: + return xstrdup("short"); + case TYPE_C_USHORT: + return xstrdup("unsigned short"); + case TYPE_C_CHAR: + return xstrdup("char"); + case TYPE_C_UCHAR: + return xstrdup("unsigned char"); + + case TYPE_INT: + // 'int' in Zen C maps to 'i32' now for portability. + // FFI should use c_int. + return xstrdup("int32_t"); case TYPE_FLOAT: return xstrdup("float"); case TYPE_BITINT: @@ -519,8 +559,11 @@ static char *type_to_c_string_impl(Type *t) return res; } - char *res = xmalloc(strlen(inner) + 7); - sprintf(res, "Slice_%s", inner); + char *inner_zens = type_to_string(t->inner); + char *res = xmalloc(strlen(inner_zens) + 7); + sprintf(res, "Slice_%s", inner_zens); + free(inner_zens); + free(inner); return res; } @@ -561,7 +604,12 @@ static char *type_to_c_string_impl(Type *t) return xstrdup("z_closure_T"); case TYPE_GENERIC: - return xstrdup(t->name); + // Use type_to_string to get the mangled name (e.g. Option_int) instead of raw C string + // composition This ensures consistency with struct definitions. + { + char *s = type_to_string(t); + return s; + } case TYPE_ALIAS: return type_to_c_string(t->inner); diff --git a/src/ast/ast.h b/src/ast/ast.h index 4498d7c..fa67043 100644 --- a/src/ast/ast.h +++ b/src/ast/ast.h @@ -28,30 +28,40 @@ typedef enum */ typedef enum { - TYPE_VOID, ///< `void` type. - TYPE_BOOL, ///< `bool` type. - TYPE_CHAR, ///< `char` type. - TYPE_STRING, ///< `string` type. - TYPE_U0, ///< `u0` type. - TYPE_I8, ///< `i8` type. - TYPE_U8, ///< `u8` type. - TYPE_I16, ///< `i16` type. - TYPE_U16, ///< `u16` type. - TYPE_I32, ///< `i32` type. - TYPE_U32, ///< `u32` type. - TYPE_I64, ///< `i64` type. - TYPE_U64, ///< `u64` type. - TYPE_I128, ///< `i128` type. - TYPE_U128, ///< `u128` type. - TYPE_F32, ///< `f32` type. - TYPE_F64, ///< `f64` type. - TYPE_INT, ///< `int` (alias, usually i32). - TYPE_FLOAT, ///< `float` (alias). - TYPE_USIZE, ///< `usize` (pointer size unsigned). - TYPE_ISIZE, ///< `isize` (pointer size signed). - TYPE_BYTE, ///< `byte`. - TYPE_RUNE, ///< `rune`. - TYPE_UINT, ///< `uint` (alias). + TYPE_VOID, ///< `void` type. + TYPE_BOOL, ///< `bool` type. + TYPE_CHAR, ///< `char` type. + TYPE_STRING, ///< `string` type. + TYPE_U0, ///< `u0` type. + TYPE_I8, ///< `i8` type. + TYPE_U8, ///< `u8` type. + TYPE_I16, ///< `i16` type. + TYPE_U16, ///< `u16` type. + TYPE_I32, ///< `i32` type. + TYPE_U32, ///< `u32` type. + TYPE_I64, ///< `i64` type. + TYPE_U64, ///< `u64` type. + TYPE_I128, ///< `i128` type. + TYPE_U128, ///< `u128` type. + TYPE_F32, ///< `f32` type. + TYPE_F64, ///< `f64` type. + TYPE_INT, ///< `int` (alias, usually i32). + TYPE_FLOAT, ///< `float` (alias). + TYPE_USIZE, ///< `usize` (pointer size unsigned). + TYPE_ISIZE, ///< `isize` (pointer size signed). + TYPE_BYTE, ///< `byte`. + TYPE_RUNE, ///< `rune`. + TYPE_UINT, ///< `uint` (alias). + // Portable C Types (FFI) + TYPE_C_INT, ///< `c_int` (int). + TYPE_C_UINT, ///< `c_uint` (unsigned int). + TYPE_C_LONG, ///< `c_long` (long). + TYPE_C_ULONG, ///< `c_ulong` (unsigned long). + TYPE_C_SHORT, ///< `c_short` (short). + TYPE_C_USHORT, ///< `c_ushort` (unsigned short). + TYPE_C_CHAR, ///< `c_char` (char). + TYPE_C_UCHAR, ///< `c_uchar` (unsigned char). + TYPE_STRUCT, ///< Struct type. TYPE_ENUM, ///< Enum type. TYPE_POINTER, ///< Pointer type (*). diff --git a/src/codegen/codegen.c b/src/codegen/codegen.c index 0496a46..384820b 100644 --- a/src/codegen/codegen.c +++ b/src/codegen/codegen.c @@ -1150,14 +1150,102 @@ void codegen_expression(ParserContext *ctx, ASTNode *node, FILE *out) break; } case NODE_EXPR_CAST: - fprintf(out, "((%s)(", node->cast.target_type); + { + const char *t = node->cast.target_type; + const char *mapped = t; + if (strcmp(t, "c_int") == 0) + { + mapped = "int"; + } + else if (strcmp(t, "c_uint") == 0) + { + mapped = "unsigned int"; + } + else if (strcmp(t, "c_long") == 0) + { + mapped = "long"; + } + else if (strcmp(t, "c_ulong") == 0) + { + mapped = "unsigned long"; + } + else if (strcmp(t, "c_short") == 0) + { + mapped = "short"; + } + else if (strcmp(t, "c_ushort") == 0) + { + mapped = "unsigned short"; + } + else if (strcmp(t, "c_char") == 0) + { + mapped = "char"; + } + else if (strcmp(t, "c_uchar") == 0) + { + mapped = "unsigned char"; + } + else if (strcmp(t, "int") == 0) + { + mapped = "int32_t"; + } + else if (strcmp(t, "uint") == 0) + { + mapped = "unsigned int"; + } + + fprintf(out, "((%s)(", mapped); codegen_expression(ctx, node->cast.expr, out); fprintf(out, "))"); break; + } case NODE_EXPR_SIZEOF: if (node->size_of.target_type) { - fprintf(out, "sizeof(%s)", node->size_of.target_type); + const char *t = node->size_of.target_type; + const char *mapped = t; + if (strcmp(t, "c_int") == 0) + { + mapped = "int"; + } + else if (strcmp(t, "c_uint") == 0) + { + mapped = "unsigned int"; + } + else if (strcmp(t, "c_long") == 0) + { + mapped = "long"; + } + else if (strcmp(t, "c_ulong") == 0) + { + mapped = "unsigned long"; + } + else if (strcmp(t, "c_short") == 0) + { + mapped = "short"; + } + else if (strcmp(t, "c_ushort") == 0) + { + mapped = "unsigned short"; + } + else if (strcmp(t, "c_char") == 0) + { + mapped = "char"; + } + else if (strcmp(t, "c_uchar") == 0) + { + mapped = "unsigned char"; + } + else if (strcmp(t, "int") == 0) + { + mapped = "int32_t"; // Strict mapping + } + else if (strcmp(t, "uint") == 0) + { + mapped = "unsigned int"; // uint alias + } + + fprintf(out, "sizeof(%s)", mapped); } else { diff --git a/src/codegen/codegen_decl.c b/src/codegen/codegen_decl.c index 31bd2ee..1623ffc 100644 --- a/src/codegen/codegen_decl.c +++ b/src/codegen/codegen_decl.c @@ -1130,12 +1130,23 @@ void print_type_defs(ParserContext *ctx, FILE *out, ASTNode *nodes) fprintf(out, "typedef struct Tuple_%s Tuple_%s;\nstruct Tuple_%s { ", t->sig, t->sig, t->sig); char *s = xstrdup(t->sig); - char *p = strtok(s, "_"); + char *current = s; + char *next_sep = strstr(current, "__"); int i = 0; - while (p) + while (current) { - fprintf(out, "%s v%d; ", p, i++); - p = strtok(NULL, "_"); + if (next_sep) + { + *next_sep = 0; + fprintf(out, "%s v%d; ", current, i++); + current = next_sep + 2; + next_sep = strstr(current, "__"); + } + else + { + fprintf(out, "%s v%d; ", current, i++); + break; + } } free(s); fprintf(out, "};\n"); diff --git a/src/parser/parser_expr.c b/src/parser/parser_expr.c index 7c53d96..5bf0089 100644 --- a/src/parser/parser_expr.c +++ b/src/parser/parser_expr.c @@ -1908,7 +1908,7 @@ ASTNode *parse_primary(ParserContext *ctx, Lexer *l) { Type *formal_type = parse_type_formal(ctx, l); concrete_types[arg_count] = type_to_string(formal_type); - unmangled_types[arg_count] = type_to_c_string(formal_type); + unmangled_types[arg_count] = type_to_string(formal_type); arg_count++; if (lexer_peek(l).type == TOK_COMMA) @@ -2944,7 +2944,7 @@ ASTNode *parse_primary(ParserContext *ctx, Lexer *l) { if (i > 0) { - strcat(sig, "_"); + strcat(sig, "__"); } strcat(sig, type_strs[i]); } diff --git a/src/parser/parser_struct.c b/src/parser/parser_struct.c index e53b56c..c89ad34 100644 --- a/src/parser/parser_struct.c +++ b/src/parser/parser_struct.c @@ -863,7 +863,7 @@ ASTNode *parse_struct(ParserContext *ctx, Lexer *l, int is_union, int is_opaque) Token field_name = lexer_next(l); lexer_next(l); // eat : Type *ft = parse_type_formal(ctx, l); - char *field_type_str = type_to_string(ft); + char *field_type_str = type_to_c_string(ft); expect(l, TOK_SEMICOLON, "Expected ;"); ASTNode *nf = ast_create(NODE_FIELD); @@ -947,7 +947,7 @@ ASTNode *parse_struct(ParserContext *ctx, Lexer *l, int is_union, int is_opaque) Token f_name = lexer_next(l); expect(l, TOK_COLON, "Expected :"); Type *ft = parse_type_formal(ctx, l); - char *f_type = type_to_string(ft); + char *f_type = type_to_c_string(ft); ASTNode *f = ast_create(NODE_FIELD); f->field.name = token_strdup(f_name); @@ -1120,7 +1120,7 @@ ASTNode *parse_enum(ParserContext *ctx, Lexer *l) while (lexer_peek(l).type == TOK_COMMA) { lexer_next(l); // eat , - strcat(sig, "_"); + strcat(sig, "__"); Type *next_t = parse_type_obj(ctx, l); char *ns = type_to_string(next_t); if (strlen(sig) + strlen(ns) + 2 > 510) diff --git a/src/parser/parser_type.c b/src/parser/parser_type.c index 49e961c..fcbe12d 100644 --- a/src/parser/parser_type.c +++ b/src/parser/parser_type.c @@ -427,13 +427,13 @@ Type *parse_type_base(ParserContext *ctx, Lexer *l) if (strcmp(name, "uint") == 0) { free(name); - return type_new(TYPE_UINT); + return type_new(TYPE_U32); // Strict uint32_t } if (strcmp(name, "int") == 0) { free(name); - return type_new(TYPE_INT); + return type_new(TYPE_I32); // Strict int32_t } if (strcmp(name, "float") == 0) { @@ -467,23 +467,31 @@ Type *parse_type_base(ParserContext *ctx, Lexer *l) } if (strcmp(name, "long") == 0) { + zwarn_at(t, "'long' is treated as portable 'int64_t' in Zen C. Use 'c_long' for " + "platform-dependent C long."); free(name); return type_new(TYPE_I64); } if (strcmp(name, "short") == 0) { + zwarn_at(t, "'short' is treated as portable 'int16_t' in Zen C. Use 'c_short' for " + "platform-dependent C short."); free(name); return type_new(TYPE_I16); } if (strcmp(name, "unsigned") == 0) { + zwarn_at(t, "'unsigned' is treated as portable 'uint32_t' in Zen C. Use 'c_uint' for " + "platform-dependent C unsigned int."); free(name); - return type_new(TYPE_UINT); + return type_new(TYPE_U32); } if (strcmp(name, "signed") == 0) { + zwarn_at(t, "'signed' is treated as portable 'int32_t' in Zen C. Use 'c_int' for " + "platform-dependent C int."); free(name); - return type_new(TYPE_INT); + return type_new(TYPE_I32); } if (strcmp(name, "int8_t") == 0) { @@ -536,6 +544,48 @@ Type *parse_type_base(ParserContext *ctx, Lexer *l) return type_new(TYPE_ISIZE); } + // Portable C Types + if (strcmp(name, "c_int") == 0) + { + free(name); + return type_new(TYPE_C_INT); + } + if (strcmp(name, "c_uint") == 0) + { + free(name); + return type_new(TYPE_C_UINT); + } + if (strcmp(name, "c_long") == 0) + { + free(name); + return type_new(TYPE_C_LONG); + } + if (strcmp(name, "c_ulong") == 0) + { + free(name); + return type_new(TYPE_C_ULONG); + } + if (strcmp(name, "c_short") == 0) + { + free(name); + return type_new(TYPE_C_SHORT); + } + if (strcmp(name, "c_ushort") == 0) + { + free(name); + return type_new(TYPE_C_USHORT); + } + if (strcmp(name, "c_char") == 0) + { + free(name); + return type_new(TYPE_C_CHAR); + } + if (strcmp(name, "c_uchar") == 0) + { + free(name); + return type_new(TYPE_C_UCHAR); + } + // Relaxed Type Check: If explicit 'struct Name', trust the user. if (explicit_struct) { @@ -677,7 +727,7 @@ Type *parse_type_base(ParserContext *ctx, Lexer *l) zpanic_at(t, "Expected > after generic"); } - char *unmangled_arg = type_to_c_string(first_arg); + char *unmangled_arg = type_to_string(first_arg); int is_single_dep = 0; for (int k = 0; k < ctx->known_generics_count; ++k) @@ -791,7 +841,7 @@ Type *parse_type_base(ParserContext *ctx, Lexer *l) if (lexer_peek(l).type == TOK_COMMA) { lexer_next(l); - strcat(sig, "_"); + strcat(sig, "__"); } else { diff --git a/src/parser/parser_utils.c b/src/parser/parser_utils.c index 28d2c11..8ea2934 100644 --- a/src/parser/parser_utils.c +++ b/src/parser/parser_utils.c @@ -691,16 +691,22 @@ void register_tuple(ParserContext *ctx, const char *sig) s_def->strct.name = xstrdup(struct_name); char *s_sig = xstrdup(sig); - char *tok = strtok(s_sig, "_"); + char *current = s_sig; + char *next_sep = strstr(current, "__"); ASTNode *head = NULL, *tail = NULL; int i = 0; - while (tok) + while (current) { + if (next_sep) + { + *next_sep = 0; + } + ASTNode *f = ast_create(NODE_FIELD); char fname[32]; sprintf(fname, "v%d", i++); f->field.name = xstrdup(fname); - f->field.type = xstrdup(tok); + f->field.type = xstrdup(current); if (!head) { @@ -712,7 +718,15 @@ void register_tuple(ParserContext *ctx, const char *sig) } tail = f; - tok = strtok(NULL, "_"); + if (next_sep) + { + current = next_sep + 2; + next_sep = strstr(current, "__"); + } + else + { + break; + } } free(s_sig); s_def->strct.fields = head; diff --git a/std/fs.zc b/std/fs.zc index a00993b..5b2cb21 100644 --- a/std/fs.zc +++ b/std/fs.zc @@ -15,9 +15,9 @@ include include // 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 access(pathname: const char*, mode: c_int) -> c_int; +extern fn unlink(pathname: const char*) -> c_int; +extern fn rmdir(pathname: const char*) -> c_int; extern fn malloc(size: usize) -> void*; extern fn free(ptr: void*); @@ -90,17 +90,17 @@ raw { } } -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_mkdir(path: const char*) -> c_int; +extern fn _z_fs_get_metadata(path: const char*, size: U64*, is_dir: c_int*, is_file: c_int*) -> c_int; +extern fn _z_fs_read_entry(dir: void*, out_name: char*, buf_size: c_int, is_dir: c_int*) -> c_int; extern fn _z_fs_fopen(path: const char*, mode: const char*) -> void*; -extern fn _z_fs_fclose(stream: void*) -> int; +extern fn _z_fs_fclose(stream: void*) -> c_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_fseek(stream: void*, offset: I64, whence: c_int) -> c_int; extern fn _z_fs_ftell(stream: void*) -> I64; extern fn _z_fs_opendir(name: const char*) -> void*; -extern fn _z_fs_closedir(dir: void*) -> int; +extern fn _z_fs_closedir(dir: void*) -> c_int; struct File { @@ -191,41 +191,50 @@ impl File { } fn exists(path: char*) -> bool { - return access(path, Z_F_OK) == 0; + let zero: c_int = 0; + return access(path, Z_F_OK) == zero; } fn metadata(path: char*) -> Result { let size: uint64_t; - let is_d: int; - let is_f: int; + let is_d: c_int; + let is_f: c_int; - if (_z_fs_get_metadata(path, &size, &is_d, &is_f) != 0) { + let res = _z_fs_get_metadata(path, &size, &is_d, &is_f); + let non_zero: c_int = 0; + if (res != non_zero) { return Result::Err("Failed to get metadata"); } return Result::Ok(Metadata { size: (U64)size, - is_dir: is_d != 0, - is_file: is_f != 0 + is_dir: is_d != non_zero, + is_file: is_f != non_zero }); } fn create_dir(path: char*) -> Result { - if (_z_fs_mkdir(path) != 0) { + let res = _z_fs_mkdir(path); + let zero: c_int = 0; + if (res != zero) { return Result::Err("Failed to create directory"); } return Result::Ok(true); } fn remove_file(path: char*) -> Result { - if (unlink(path) != 0) { + let res = unlink(path); + let zero: c_int = 0; + if (res != zero) { return Result::Err("Failed to remove file"); } return Result::Ok(true); } fn remove_dir(path: char*) -> Result { - if (rmdir(path) != 0) { + let res = rmdir(path); + let zero: c_int = 0; + if (res != zero) { return Result::Err("Failed to remove directory"); } return Result::Ok(true); @@ -245,17 +254,19 @@ impl File { return Result< Vec >::Err("Out of memory"); } - let is_d: int = 0; + let is_d: c_int = 0; + let is_d_zero: c_int = 0; while (_z_fs_read_entry(dir, name_buf, 256, &is_d)) { - if (strcmp(name_buf, ".") == 0 || strcmp(name_buf, "..") == 0) { + let zero_cmp: c_int = 0; + if (strcmp(name_buf, ".") == zero_cmp || strcmp(name_buf, "..") == zero_cmp) { continue; } let s = String::new(name_buf); let ent = DirEntry { name: s, - is_dir: is_d != 0 + is_dir: is_d != is_d_zero }; // Transfer ownership: String -> DirEntry diff --git a/std/io.zc b/std/io.zc index a5a7359..d9829dd 100644 --- a/std/io.zc +++ b/std/io.zc @@ -6,11 +6,11 @@ include include // These work directly with const char* in extern declarations -extern fn vprintf(fmt: const char*, ap: va_list) -> int; +extern fn vprintf(fmt: const char*, ap: va_list) -> c_int; // vsnprintf is problematic on macOS because it's a macro that expands to a builtin with a different signature // so we wrap it in a C function to avoid the conflict -extern fn _z_vsnprintf(str: char*, size: usize, fmt: const char*, ap: va_list) -> int; +extern fn _z_vsnprintf(str: char*, size: usize, fmt: const char*, ap: va_list) -> c_int; // EOF is typically -1, but we define it for portability def Z_EOF = -1; @@ -27,7 +27,7 @@ raw { } extern fn _z_get_stdin() -> void*; -extern fn _z_fgetc(stream: void*) -> int; +extern fn _z_fgetc(stream: void*) -> c_int; fn format(fmt: char*, ...) -> char* { static let buffer: char[1024]; @@ -40,7 +40,7 @@ fn format(fmt: char*, ...) -> char* { return (char*)buffer; } -fn format_into(buffer: char*, size: usize, fmt: char*, ...) -> int { +fn format_into(buffer: char*, size: usize, fmt: char*, ...) -> c_int { let ap: va_list; va_start(ap, fmt); @@ -63,7 +63,7 @@ fn format_new(fmt: char*, ...) -> char* { return buffer; } -fn print(fmt: char*, ...) -> int { +fn print(fmt: char*, ...) -> c_int { let ap: va_list; va_start(ap, fmt); let ret = vprintf(fmt, ap); @@ -71,7 +71,7 @@ fn print(fmt: char*, ...) -> int { return ret; } -fn println(fmt: char*, ...) -> int { +fn println(fmt: char*, ...) -> c_int { let ap: va_list; va_start(ap, fmt); let ret = vprintf(fmt, ap); @@ -86,13 +86,15 @@ fn readln() -> char* { let line: char* = malloc(cap); if (line == NULL) return NULL; - let c: int; + let c: c_int; let std_in = _z_get_stdin(); while (true) { c = _z_fgetc(std_in); - if (c == Z_EOF) break; - if (c == 10) break; // '\n' + let eof_c: c_int = Z_EOF; + if (c == eof_c) break; + let nl_c: c_int = 10; + if (c == nl_c) break; // '\n' if (len + 1 >= cap) { cap = cap * 2; @@ -108,7 +110,8 @@ fn readln() -> char* { len = len + 1; } - if (len == 0 && c == Z_EOF) { + let eof_final: c_int = Z_EOF; + if (len == 0 && c == eof_final) { free(line); return NULL; } diff --git a/std/net.zc b/std/net.zc index 826b795..dce1a01 100644 --- a/std/net.zc +++ b/std/net.zc @@ -14,9 +14,9 @@ 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; +extern fn socket(domain: c_int, type: c_int, proto: c_int) -> c_int; +extern fn close(fd: c_int) -> c_int; +extern fn read(fd: c_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 @@ -56,47 +56,63 @@ raw { } } -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: const char*, n: usize) -> isize; +extern fn _z_net_bind(fd: c_int, host: const char*, port: c_int) -> c_int; +extern fn _z_net_connect(fd: c_int, host: const char*, port: c_int) -> c_int; +extern fn _z_net_accept(fd: c_int) -> c_int; +extern fn _z_net_write(fd: c_int, buf: const char*, n: usize) -> isize; struct TcpStream { - handle: int; + handle: c_int; } -extern fn strerror(errnum: int) -> char*; +extern fn strerror(errnum: c_int) -> char*; impl TcpStream { fn read(self, buf: char*, len: usize) -> Result { let n = read(self.handle - 1, (void*)buf, len); - if (n < 0) return Result::Err(strerror(errno)); + let zero: c_int = 0; + if (n < (isize)zero) return Result::Err(strerror(errno)); return Result::Ok((usize)n); } - fn write(self, buf: char*, len: usize) -> Result { - let n = _z_net_write(self.handle - 1, buf, len); - if (n < 0) return Result::Err("Write failed"); + fn write(self, buf: u8*, len: usize) -> Result { + let one: c_int = 1; + let n: isize = _z_net_write(self.handle - one, buf, len); + let zero: isize = 0; + if (n < zero) return Result::Err("Write failed"); return Result::Ok((usize)n); } fn close(self) { - if (self.handle > 0) { - close(self.handle - 1); - self.handle = 0; + let zero: c_int = 0; + if (self.handle > zero) { + let one: c_int = 1; + close(self.handle - one); + self.handle = zero; } } - fn connect(host: char*, port: int) -> Result { - let fd = socket(Z_AF_INET, Z_SOCK_STREAM, 0); - if (fd < 0) return Result::Err("Failed to create socket"); + fn connect(host: char*, port: c_int) -> Result { + let zero: c_int = 0; + let fd = socket(Z_AF_INET, Z_SOCK_STREAM, zero); + if (fd < zero) return Result::Err("Failed to create socket"); + + // C constants like -1 + let neg_one: c_int = -1; let res = _z_net_connect(fd, host, port); - if (res == -1) { close(fd); return Result::Err("Invalid address"); } - if (res == -2) { close(fd); return Result::Err("Connection failed"); } + if (res == neg_one) { close(fd); return Result::Err("Invalid address"); } + // _z_net_connect might return -2? The original code had it. + // Assuming -2 is also possible... check implementation or just assume logic was correct. + // But wait, the original code had: + // if (res == -1) ... if (res == -2) ... + // I will keep it but cast strict. + let neg_two: c_int = -2; + if (res == neg_two) { close(fd); return Result::Err("Connection failed"); } - return Result::Ok(TcpStream { handle: fd + 1 }); + let one: c_int = 1; + return Result::Ok(TcpStream { handle: fd + one }); } } @@ -107,32 +123,42 @@ impl Drop for TcpStream { } struct TcpListener { - handle: int; + handle: c_int; } impl TcpListener { - fn bind(host: char*, port: int) -> Result { - let fd = socket(Z_AF_INET, Z_SOCK_STREAM, 0); - if (fd < 0) return Result::Err("Failed to create socket"); + fn bind(host: char*, port: c_int) -> Result { + let zero: c_int = 0; + let fd = socket(Z_AF_INET, Z_SOCK_STREAM, zero); + if (fd < zero) return Result::Err("Failed to create socket"); let res = _z_net_bind(fd, host, port); - if (res == -1) { close(fd); return Result::Err("Invalid address"); } - if (res == -2) { close(fd); return Result::Err("Bind failed"); } - if (res == -3) { close(fd); return Result::Err("Listen failed"); } + let neg_one: c_int = -1; + let neg_two: c_int = -2; + let neg_three: c_int = -3; + + if (res == neg_one) { close(fd); return Result::Err("Invalid address"); } + if (res == neg_two) { close(fd); return Result::Err("Bind failed"); } + if (res == neg_three) { close(fd); return Result::Err("Listen failed"); } - return Result::Ok(TcpListener { handle: fd + 1 }); + let one: c_int = 1; + return Result::Ok(TcpListener { handle: fd + one }); } fn accept(self) -> Result { - let client_fd = _z_net_accept(self.handle - 1); - if (client_fd < 0) return Result::Err("Accept failed"); - return Result::Ok(TcpStream { handle: client_fd + 1 }); + let one: c_int = 1; + let client_fd = _z_net_accept(self.handle - one); + let zero: c_int = 0; + if (client_fd < zero) return Result::Err("Accept failed"); + return Result::Ok(TcpStream { handle: client_fd + one }); } fn close(self) { - if (self.handle > 0) { - close(self.handle - 1); - self.handle = 0; + let zero: c_int = 0; + if (self.handle > zero) { + let one: c_int = 1; + close(self.handle - one); + self.handle = zero; } } } diff --git a/std/process.zc b/std/process.zc index 3ce43b6..9f432c0 100644 --- a/std/process.zc +++ b/std/process.zc @@ -9,7 +9,7 @@ include include // system() can be externed directly with const char* -extern fn system(command: const char*) -> int; +extern fn system(command: const char*) -> c_int; // Minimal raw block: only for opaque FILE* types // popen/pclose/fgets use FILE* which conflicts with void* @@ -28,8 +28,8 @@ raw { } 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_pclose(stream: void*) -> c_int; +extern fn _z_fgets(s: char*, size: c_int, stream: void*) -> char*; struct Output { stdout: String; @@ -88,7 +88,7 @@ impl Command { let buf = (char*)malloc(buf_size); while (true) { - let res = _z_fgets(buf, (int)buf_size, fp); + let res = _z_fgets(buf, (c_int)buf_size, fp); if (res == 0) break; let chunk = String::from(buf); diff --git a/std/string.zc b/std/string.zc index 0bc9539..54f11b2 100644 --- a/std/string.zc +++ b/std/string.zc @@ -90,7 +90,8 @@ impl String { } fn eq(self, other: String*) -> bool { - return strcmp(self.c_str(), (*other).c_str()) == 0; + let zero: c_int = 0; + return strcmp(self.c_str(), (*other).c_str()) == zero; } fn length(self) -> usize { @@ -146,7 +147,8 @@ impl String { fn starts_with(self, prefix: char*) -> bool { let plen = strlen(prefix); if plen > self.length() { return false; } - return strncmp(self.c_str(), prefix, plen) == 0; + let zero: c_int = 0; + return strncmp(self.c_str(), prefix, plen) == zero; } fn ends_with(self, suffix: char*) -> bool { @@ -154,7 +156,8 @@ impl String { let len = self.length(); if slen > len { return false; } let offset = (int)(len - slen); - return strcmp(self.c_str() + offset, suffix) == 0; + let zero: c_int = 0; + return strcmp(self.c_str() + offset, suffix) == zero; } fn free(self) { diff --git a/std/thread.zc b/std/thread.zc index 16f3ca1..78d2547 100644 --- a/std/thread.zc +++ b/std/thread.zc @@ -35,11 +35,11 @@ raw { if (ret == 0) { *out_handle = (size_t)pt; } - return ret; + return (int)ret; } static int _z_thread_join(void *handle) { - return pthread_join((pthread_t)handle, NULL); + return (int)pthread_join((pthread_t)handle, NULL); } static void _z_mutex_init(void *ptr) { @@ -63,13 +63,13 @@ raw { } } -extern fn _z_thread_spawn(ctx: void*, out: usize*) -> int; -extern fn _z_thread_join(handle: void*) -> int; +extern fn _z_thread_spawn(ctx: void*, out: usize*) -> c_int; +extern fn _z_thread_join(handle: void*) -> c_int; extern fn _z_mutex_init(ptr: void*); extern fn _z_mutex_lock(ptr: void*); extern fn _z_mutex_unlock(ptr: void*); extern fn _z_mutex_destroy(ptr: void*); -extern fn _z_usleep(micros: int); +extern fn _z_usleep(micros: c_int); @@ -79,26 +79,28 @@ struct Thread { impl Thread { fn spawn(func: fn()) -> Result { - let t: usize = 0; + let out_handle: usize = 0; - let ctx_copy = malloc(16); // z_closure_T is 16 bytes - if (ctx_copy == NULL) return Result::Err("OOM"); + let ctx = malloc(16); // z_closure_T is 16 bytes + if (ctx == NULL) return Result::Err("OOM"); - memcpy(ctx_copy, &func, 16); + memcpy(ctx, &func, 16); - let ret = _z_thread_spawn(ctx_copy, &t); - - if (ret != 0) { - free(ctx_copy); + let ret = _z_thread_spawn(ctx, &out_handle); + let zero: c_int = 0; + if (ret != zero) { + // Failed to spawn + free(ctx); return Result::Err("Failed to create thread"); } - return Result::Ok(Thread { handle: (void*)t }); + return Result::Ok(Thread { handle: (void*)out_handle }); } fn join(self) -> Result { let ret = _z_thread_join(self.handle); - if (ret != 0) return Result::Err("Join failed"); + let zero: c_int = 0; + if (ret != zero) return Result::Err("Join failed"); return Result::Ok(true); } } @@ -138,5 +140,6 @@ impl Drop for Mutex { } fn sleep_ms(ms: int) { - _z_usleep(ms * 1000); + let micros: c_int = (c_int)(ms * 1000); + _z_usleep(micros); } diff --git a/tests/collections/test_string_suite.zc b/tests/collections/test_string_suite.zc index afe08af..64ed9d8 100644 --- a/tests/collections/test_string_suite.zc +++ b/tests/collections/test_string_suite.zc @@ -91,7 +91,9 @@ test "test_fstrings_return" { let inner = f"Inner({x})"; let outer = f"Outer({inner})"; println "Composed: {outer}"; - assert(strcmp(outer, "Outer(Inner(100))") == 0, "Composed f-string failed"); + let outer_res = strcmp(outer, "Outer(Inner(100))"); + let zero: c_int = 0; + assert(outer_res == zero, "Composed f-string failed"); } test "test_string_std_ops" { diff --git a/tests/features/test_portable_types.zc b/tests/features/test_portable_types.zc new file mode 100644 index 0000000..8f54fcb --- /dev/null +++ b/tests/features/test_portable_types.zc @@ -0,0 +1,46 @@ + +import "std/io.zc"; + +// This test verifies the new portable integer types and C interop types. + +extern fn abs(x: c_int) -> c_int; +extern fn labs(x: c_long) -> c_long; + +fn main() -> int { + // Portable types + let a: i32 = -42; + let b: u32 = 42; + let c: i64 = -1000000; + let d: u64 = 1000000; + + if (a != -42) return 1; + if (b != 42) return 2; + if (c != -1000000) return 3; + if (d != 1000000) return 4; + + // C Types + let ca: c_int = -10; + let cb: c_long = -20; + let cc: c_short = -5; + let cd: c_char = 65; // 'A' + + // Test C interaction + let abs_val = abs(ca); + let expected_abs: c_int = 10; + if (abs_val != expected_abs) return 5; + + let labs_val = labs(cb); + let expected_labs: c_long = 20; + if (labs_val != expected_labs) return 6; + + // Size checks (these are platform dependent but we can check relations) + // sizeof(c_char) is always 1 + if (sizeof(c_char) != 1) return 7; + + // sizeof(c_short) <= sizeof(c_int) <= sizeof(c_long) + if (sizeof(c_short) > sizeof(c_int)) return 8; + if (sizeof(c_int) > sizeof(c_long)) return 9; + + printf("Portable types test passed.\n"); + return 0; +} -- cgit v1.2.3