miniextendr_macros/lib.rs
1//! # miniextendr-macros - Procedural macros for Rust <-> R interop
2//!
3//! This crate provides the procedural macros that power miniextendr's code
4//! generation. Most users should depend on `miniextendr-api` and use its
5//! re-exports, but this crate can be used directly when you only need macros.
6//!
7//! Primary macros and derives:
8//! - `#[miniextendr]` on functions, impl blocks, trait defs, and trait impls.
9//! - `#[r_ffi_checked]` for main-thread routing of C-ABI wrappers.
10//! - Derives: `ExternalPtr`, `RNativeType`, ALTREP derives, `RFactor`.
11//! - Helpers: `typed_list` for typed list builders.
12//!
13//! R wrapper generation is driven by Rust doc comments (roxygen tags are
14//! extracted). During package build, the wrapper-gen pass loads the installed
15//! shared object into R and calls `miniextendr_write_wrappers`, which walks the
16//! linkme `#[distributed_slice]` tables and writes `R/<pkg>-wrappers.R`.
17//!
18//! ## Quick start
19//!
20//! ```ignore
21//! use miniextendr_api::miniextendr;
22//!
23//! #[miniextendr]
24//! fn add(a: i32, b: i32) -> i32 {
25//! a + b
26//! }
27//! ```
28//!
29//! ## Macro expansion pipeline
30//!
31//! ### Overview
32//!
33//! ```text
34//! ┌──────────────────────────────────────────────────────────────────────────┐
35//! │ #[miniextendr] on fn │
36//! │ │
37//! │ 1. Parse: syn::ItemFn → MiniextendrFunctionParsed │
38//! │ 2. Analyze return type (Result<T>, Option<T>, raw SEXP, etc.) │
39//! │ 3. Generate: │
40//! │ ├── C wrapper: extern "C-unwind" fn C_<crate>_<name>(...) → SEXP │
41//! │ ├── R wrapper: const R_WRAPPER_<NAME>: &str = "..." │
42//! │ └── Registration: const call_method_def_<name>: R_CallMethodDef │
43//! │ 4. Original function preserved (with added attributes) │
44//! └──────────────────────────────────────────────────────────────────────────┘
45//!
46//! ┌──────────────────────────────────────────────────────────────────────────┐
47//! │ #[miniextendr(env|r6|s3|s4|s7)] on impl │
48//! │ │
49//! │ 1. Parse: syn::ItemImpl → extract methods │
50//! │ 2. For each method: │
51//! │ ├── Generate C wrapper (handles self parameter) │
52//! │ ├── Generate R method wrapper string │
53//! │ └── Generate registration entry │
54//! │ 3. Generate class definition code per class system: │
55//! │ ├── env: new.env() + method assignment │
56//! │ ├── r6: R6Class() definition │
57//! │ ├── s3: S3 generics + methods │
58//! │ ├── s4: setClass() + setMethod() │
59//! │ └── s7: new_class() definition │
60//! │ 4. Emit const with combined R code │
61//! └──────────────────────────────────────────────────────────────────────────┘
62//!
63//! ┌──────────────────────────────────────────────────────────────────────────┐
64//! │ #[miniextendr] on trait │
65//! │ │
66//! │ 1. Parse: syn::ItemTrait → extract method signatures │
67//! │ 2. Generate: │
68//! │ ├── Trait tag constant: const TAG_<TRAIT>: mx_tag = ... │
69//! │ ├── Vtable struct: struct __vtable_<Trait> { ... } │
70//! │ └── CCalls table: static MX_CCALL_<TRAIT>: [...] = ... │
71//! │ 3. Original trait preserved │
72//! └──────────────────────────────────────────────────────────────────────────┘
73//!
74//! ┌──────────────────────────────────────────────────────────────────────────┐
75//! │ #[miniextendr] impl Trait for Type │
76//! │ │
77//! │ 1. Parse: syn::ItemImpl (trait impl) │
78//! │ 2. Generate: │
79//! │ ├── Vtable instance: static __VTABLE_<CRATE>_<TRAIT>_FOR_<TYPE> │
80//! │ ├── Wrapper struct: struct __MxWrapper<Type> { erased, data } │
81//! │ ├── Query function: fn __mx_query_<type>(tag) → vtable ptr │
82//! │ └── Base vtable: static __MX_BASE_VTABLE_<TYPE>: ... │
83//! │ 3. Original impl preserved │
84//! └──────────────────────────────────────────────────────────────────────────┘
85//!
86//! ```
87//!
88//! ### Key Modules
89//!
90//! | Module | Purpose |
91//! |--------|---------|
92//! | `miniextendr_fn` | Function parsing and attribute handling |
93//! | `c_wrapper_builder` | C wrapper generation (`extern "C-unwind"`) |
94//! | `r_wrapper_builder` | R wrapper code generation |
95//! | `rust_conversion_builder` | Rust→SEXP return value conversion |
96//! | `miniextendr_impl` | `impl Type` block processing |
97//! | `r_class_formatter` | Class system code generation (env/r6/s3/s4/s7) |
98//! | `miniextendr_trait` | Trait ABI metadata generation |
99//! | `miniextendr_impl_trait` | `impl Trait for Type` vtable generation |
100//! | `altrep` / `altrep_derive` | ALTREP struct derivation |
101//! | `externalptr_derive` | `#[derive(ExternalPtr)]` |
102//! | `roxygen` | Roxygen doc comment handling |
103//!
104//! ### Generated Symbol Naming
105//!
106//! Every `#[no_mangle]` symbol is prefixed with the consuming crate's name
107//! (from `CARGO_CRATE_NAME`) so two packages loaded into one webR session
108//! can't collide on a C symbol (#1273; helpers in `naming.rs`). For a
109//! function `my_func` in a crate `mypkg`:
110//! - C wrapper: `C_mypkg_my_func`
111//! - R wrapper const: `R_WRAPPER_MY_FUNC`
112//! - Registration: `call_method_def_my_func`
113//!
114//! For a type `MyType` with trait `Counter` in a crate `mypkg`:
115//! - Vtable: `__VTABLE_MYPKG_COUNTER_FOR_MYTYPE`
116//! - Wrapper: `__MxWrapperMyType`
117//! - Query: `__mx_query_mytype`
118//!
119//! ## Return Type Handling
120//!
121//! The `return_type_analysis` module determines how to convert Rust returns to SEXP:
122//!
123//! | Rust Type | Strategy | R Result |
124//! |-----------|----------|----------|
125//! | `T: IntoR` | `.into_sexp()` | Converted value |
126//! | `Result<T, E>` | Unwrap or R error | Value or error |
127//! | `Option<T>` | `Some` → value, `None` → `NULL` | Value or NULL |
128//! | `SEXP` | Pass through | Raw SEXP |
129//! | `()` | Invisible NULL | `invisible(NULL)` |
130//!
131//! Use `#[miniextendr(unwrap_in_r)]` to return `Result<T, E>` to R without unwrapping.
132//!
133//! ## Thread Strategy
134//!
135//! By default, `#[miniextendr]` functions run on R's main thread. Opt into
136//! worker-thread execution with `#[miniextendr(worker)]` (requires the
137//! `worker-thread` feature on `miniextendr-api`). A worker opt-in is ignored
138//! when the signature requires main-thread execution (returns/takes `SEXP`,
139//! uses variadic dots, or sets `check_interrupt`).
140//!
141//! **Note**: `ExternalPtr<T>` is `Send` — it can be returned from worker
142//! thread functions. All R API operations on `ExternalPtr` are serialized
143//! through `with_r_thread`.
144//!
145//! ## Class Systems
146//!
147//! The `r_class_formatter` module generates R code for different class systems:
148//!
149//! | System | Generated R Code | Self Parameter |
150//! |--------|------------------|----------------|
151//! | `env` | `new.env()` with methods | `self` environment |
152//! | `r6` | `R6Class()` | `self` environment |
153//! | `s3` | `structure()` + generics | First argument |
154//! | `s4` | `setClass()` + `setMethod()` | First argument |
155//! | `s7` | `new_class()` | `self` property |
156
157// miniextendr-macros procedural macros
158
159mod altrep;
160mod c_wrapper_builder;
161mod list_macro;
162mod match_arg_keys;
163mod miniextendr_fn;
164mod type_inspect;
165mod typed_dataframe;
166mod typed_list;
167mod util;
168use crate::miniextendr_fn::{MiniextendrFnAttrs, MiniextendrFunctionParsed};
169mod miniextendr_impl;
170mod r_wrapper_builder;
171/// Builder utilities for formatting R wrapper arguments and calls.
172pub(crate) use r_wrapper_builder::RArgumentBuilder;
173mod rust_conversion_builder;
174/// Helper for generating Rust→R conversion code for return values.
175pub(crate) use rust_conversion_builder::RustConversionBuilder;
176mod method_return_builder;
177/// Helpers for shaping method return handling (R vs Rust wrapper code).
178pub(crate) use method_return_builder::{MethodReturnBuilder, ReturnStrategy};
179mod altrep_derive;
180mod dataframe_derive;
181mod lifecycle;
182mod list_derive;
183mod r_class_formatter;
184mod r_preconditions;
185mod return_type_analysis;
186mod roxygen;
187
188// Trait ABI support modules
189mod externalptr_derive;
190mod miniextendr_impl_trait;
191mod miniextendr_trait;
192mod typed_external_macro;
193
194// Factor support
195mod factor_derive;
196mod match_arg_derive;
197mod newtype_derive;
198
199// Struct/enum dispatch for #[miniextendr] on structs and enums
200mod struct_enum_dispatch;
201
202// r! proc-macro implementation
203mod r_macro;
204
205// vctrs support
206#[cfg(feature = "vctrs")]
207mod vctrs_derive;
208mod vctrs_generics;
209
210mod naming;
211pub(crate) use naming::r_wrapper_const_ident_for;
212
213// Feature default mutual exclusivity guards
214#[cfg(all(feature = "r6-default", feature = "s7-default"))]
215compile_error!(
216 "features \"r6-default\" and \"s7-default\" are mutually exclusive — \
217 enable exactly one, or omit both to fall back to the unspecified default"
218);
219// Note: default-main-thread was removed — main thread is now the hardcoded default.
220// worker-default still opts into worker thread execution.
221
222pub(crate) use type_inspect::{
223 SeveralOkContainer, classify_several_ok_container, first_type_argument,
224 is_main_thread_bound_input, is_main_thread_bound_return, is_sexp_type, second_type_argument,
225};
226pub(crate) use util::{extract_cfg_attrs, r_wrapper_raw_literal, source_location_doc};
227
228/// Validate the signature of an `extern "C-unwind"` fn exported via `#[miniextendr]`.
229///
230/// R's `.Call` interface passes all arguments as `SEXP` and expects a `SEXP`
231/// return value. For `extern "C-unwind"` functions the user writes the C symbol
232/// directly, so the signature must satisfy those invariants statically —
233/// otherwise the generated registration produces UB at runtime.
234///
235/// Called before any codegen so we fail fast on an invalid extern signature
236/// rather than emitting a wrapper that would only matter after the error.
237fn validate_extern_signature(
238 abi: &syn::Abi,
239 attrs: &[syn::Attribute],
240 inputs: &syn::punctuated::Punctuated<syn::FnArg, syn::Token![,]>,
241 output: &syn::ReturnType,
242) -> syn::Result<()> {
243 use syn::spanned::Spanned;
244
245 // Reject `self` receivers up front — they are never valid for `.Call`
246 // exports, and a missing-return-type error would only hide the real
247 // problem (users write `fn foo(self)` to ask "can I export a method?").
248 for input in inputs.iter() {
249 if let syn::FnArg::Receiver(recv) = input {
250 return Err(syn::Error::new_spanned(
251 recv,
252 "self parameter not allowed in standalone functions; \
253 use #[miniextendr(env|r6|s3|s4|s7)] on impl blocks instead",
254 ));
255 }
256 }
257
258 // Require one of #[no_mangle] / #[unsafe(no_mangle)] / #[export_name].
259 let has_no_mangle = attrs.iter().any(|attr| {
260 attr.path().is_ident("no_mangle")
261 || attr
262 .parse_nested_meta(|meta| {
263 if meta.path.is_ident("no_mangle") {
264 Err(meta.error("found #[no_mangle]"))
265 } else {
266 Ok(())
267 }
268 })
269 .is_err()
270 });
271 let has_export_name = attrs.iter().any(|attr| attr.path().is_ident("export_name"));
272 if !has_no_mangle && !has_export_name {
273 return Err(syn::Error::new(
274 attrs
275 .first()
276 .map(|attr| attr.span())
277 .unwrap_or_else(|| abi.span()),
278 "extern \"C-unwind\" functions need a visible C symbol for R's .Call interface. \
279 Add one of:\n \
280 - `#[unsafe(no_mangle)]` (Rust 2024 edition)\n \
281 - `#[no_mangle]` (Rust 2021 edition)\n \
282 - `#[export_name = \"my_symbol\"]` (custom symbol name)",
283 ));
284 }
285
286 // Return type must be SEXP.
287 match output {
288 non_return_type @ syn::ReturnType::Default => {
289 return Err(syn::Error::new(
290 non_return_type.span(),
291 "extern \"C-unwind\" functions used with #[miniextendr] must return SEXP. \
292 Add `-> miniextendr_api::SEXP` as the return type. \
293 If you want automatic type conversion, remove `extern \"C-unwind\"` and let \
294 the macro generate the C wrapper.",
295 ));
296 }
297 syn::ReturnType::Type(_rarrow, output_type) => match output_type.as_ref() {
298 syn::Type::Path(type_path) => {
299 if let Some(path_to_sexp) = type_path.path.segments.last().map(|x| &x.ident)
300 && path_to_sexp != "SEXP"
301 {
302 return Err(syn::Error::new(
303 path_to_sexp.span(),
304 format!(
305 "extern \"C-unwind\" functions must return SEXP, found `{path_to_sexp}`. \
306 R's .Call interface expects SEXP return values. \
307 Change the return type to `miniextendr_api::SEXP`, or remove \
308 `extern \"C-unwind\"` to let the macro handle type conversion.",
309 ),
310 ));
311 }
312 }
313 _ => {
314 return Err(syn::Error::new(
315 output_type.span(),
316 "extern \"C-unwind\" functions must return SEXP. \
317 R's .Call interface expects SEXP return values. \
318 Change the return type to `miniextendr_api::SEXP`, or remove \
319 `extern \"C-unwind\"` to let the macro handle type conversion.",
320 ));
321 }
322 },
323 }
324
325 // Every input must be SEXP. Reject variadics and receivers.
326 for input in inputs.iter() {
327 match input {
328 syn::FnArg::Receiver(recv) => {
329 return Err(syn::Error::new_spanned(
330 recv,
331 "extern \"C-unwind\" functions cannot have a `self` parameter. \
332 R's .Call interface only accepts SEXP arguments. \
333 Use `#[miniextendr(env|r6|s3|s4|s7)]` on an impl block for methods.",
334 ));
335 }
336 syn::FnArg::Typed(pat_type) => {
337 if let syn::Pat::Rest(_) = pat_type.pat.as_ref() {
338 return Err(syn::Error::new_spanned(
339 pat_type,
340 "extern functions cannot use variadic (...) - .Call passes fixed arguments",
341 ));
342 }
343 let is_sexp = match pat_type.ty.as_ref() {
344 syn::Type::Path(type_path) => type_path
345 .path
346 .segments
347 .last()
348 .is_some_and(|seg| seg.ident == "SEXP"),
349 _ => false,
350 };
351 if !is_sexp {
352 let is_dots_type = if let syn::Type::Reference(type_ref) = pat_type.ty.as_ref()
353 {
354 if let syn::Type::Path(inner) = type_ref.elem.as_ref() {
355 inner
356 .path
357 .segments
358 .last()
359 .is_some_and(|seg| seg.ident == "Dots")
360 } else {
361 false
362 }
363 } else if let syn::Type::Path(type_path) = pat_type.ty.as_ref() {
364 type_path
365 .path
366 .segments
367 .last()
368 .is_some_and(|seg| seg.ident == "Dots")
369 } else {
370 false
371 };
372 let msg = if is_dots_type {
373 "extern functions cannot use Dots; use `...` syntax in non-extern #[miniextendr] functions instead"
374 } else {
375 "extern function parameters must be SEXP - .Call passes all arguments as SEXP"
376 };
377 return Err(syn::Error::new_spanned(&pat_type.ty, msg));
378 }
379 }
380 }
381 }
382
383 Ok(())
384}
385
386/// Builds the `let dots_typed = <dots>.typed(<spec>)...` statement injected
387/// at the start of a function body when `#[miniextendr(dots =
388/// typed_list!(...))]` is used.
389///
390/// Uses `unwrap_or_else(|e| panic!("...: {e}"))` rather than
391/// `Result::expect`, which formats the error with `Debug` instead of
392/// `Display` — for `TypedListError` that leaks PascalCase enum-variant names
393/// (`Missing { name: "x" }`) into the R-visible message instead of the
394/// human-phrased text (`missing required field: "x"`) that the direct
395/// `typed_list!` path already produces via `Display`. See audit A8.
396pub(crate) fn build_dots_validation_stmt(
397 dots_param: &syn::Ident,
398 spec_tokens: &proc_macro2::TokenStream,
399) -> syn::Stmt {
400 syn::parse_quote! {
401 let dots_typed = #dots_param.typed(#spec_tokens)
402 .unwrap_or_else(|e| panic!("dots validation failed: {e}"));
403 }
404}
405
406/// Emit the `extern "C-unwind"` helper + `R_CallMethodDef` registration for
407/// each standalone-fn `match_arg` param.
408///
409/// Each helper returns the enum's `CHOICES` wrapped in a STRSXP so the R
410/// wrapper's prelude can call `match.arg(x, .Call(helper, ...))`. Factored
411/// out of the `miniextendr` fn body so the entry point doesn't own the
412/// `quote!` scaffolding.
413fn build_match_arg_helpers(
414 match_arg_param_info: &[(String, String, &syn::Type)],
415 parsed: &miniextendr_fn::MiniextendrFunctionParsed,
416 c_ident_str: &str,
417 cfg_attrs: &[syn::Attribute],
418) -> Vec<proc_macro2::TokenStream> {
419 match_arg_param_info
420 .iter()
421 .map(|(r_param, rust_name, param_ty)| {
422 // For several_ok, the param type is e.g. Vec<Mode>/Box<[Mode]>/[Mode; N]/&[Mode]
423 // — extract inner Mode for choices_sexp.
424 let choices_ty: &syn::Type = if parsed.has_several_ok(rust_name) {
425 classify_several_ok_container(param_ty)
426 .map(|(_, t)| t)
427 .unwrap_or(param_ty)
428 } else {
429 param_ty
430 };
431 let helper_fn_name = crate::match_arg_keys::choices_helper_c_name(c_ident_str, r_param);
432 let helper_fn_ident = syn::Ident::new(&helper_fn_name, proc_macro2::Span::call_site());
433 let helper_def_ident =
434 crate::match_arg_keys::choices_helper_def_ident(c_ident_str, r_param);
435 let helper_c_name = syn::LitCStr::new(
436 std::ffi::CString::new(helper_fn_name.clone())
437 .expect("valid C string")
438 .as_c_str(),
439 proc_macro2::Span::call_site(),
440 );
441 quote::quote! {
442 #(#cfg_attrs)*
443 #[allow(non_snake_case)]
444 #[unsafe(no_mangle)]
445 pub extern "C-unwind" fn #helper_fn_ident(
446 __miniextendr_call: ::miniextendr_api::SEXP,
447 ) -> ::miniextendr_api::SEXP {
448 ::miniextendr_api::choices_sexp::<#choices_ty>()
449 }
450
451 #(#cfg_attrs)*
452 #[cfg_attr(not(target_arch = "wasm32"), ::miniextendr_api::linkme::distributed_slice(::miniextendr_api::registry::MX_CALL_DEFS), linkme(crate = ::miniextendr_api::linkme))]
453 #[allow(non_upper_case_globals)]
454 #[allow(non_snake_case)]
455 static #helper_def_ident: ::miniextendr_api::sys::R_CallMethodDef = unsafe {
456 ::miniextendr_api::sys::R_CallMethodDef {
457 name: #helper_c_name.as_ptr(),
458 fun: Some(std::mem::transmute::<
459 unsafe extern "C-unwind" fn(
460 ::miniextendr_api::SEXP,
461 ) -> ::miniextendr_api::SEXP,
462 unsafe extern "C-unwind" fn() -> *mut ::std::os::raw::c_void,
463 >(#helper_fn_ident)),
464 numArgs: 1i32,
465 }
466 };
467 }
468 })
469 .collect()
470}
471
472/// Export Rust items to R.
473///
474/// `#[miniextendr]` can be applied to:
475/// - `fn` items (generate C + R wrappers)
476/// - `impl` blocks (generate R class methods)
477/// - `trait` items (generate trait ABI metadata)
478/// - ALTREP wrapper structs (generate `RegisterAltrep` impls)
479///
480/// # Functions
481///
482/// ```ignore
483/// use miniextendr_api::miniextendr;
484///
485/// #[miniextendr]
486/// fn add(a: i32, b: i32) -> i32 { a + b }
487/// ```
488///
489/// This produces a C wrapper `C_<crate>_add` and an R wrapper `add()`.
490/// Registration is automatic via linkme distributed slices.
491///
492/// ## `extern "C-unwind"`
493///
494/// If the function is declared `extern "C-unwind"` and exported with
495/// `#[no_mangle]` (2021), `#[unsafe(no_mangle)]` (2024), or `#[export_name = "..."]`,
496/// the function itself is the C symbol and the R wrapper is prefixed with
497/// `unsafe_` to signal bypassed safety (no worker isolation or conversion).
498///
499/// ## Variadics (`...`)
500///
501/// Use `...` as the last argument. The Rust parameter becomes `_dots: &Dots`.
502/// Use `name: ...` to give it a custom name (e.g., `args: ...` → `args: &Dots`).
503///
504/// ### Typed Dots Validation
505///
506/// Use `#[miniextendr(dots = typed_list!(...))]` to automatically validate dots
507/// and create a `dots_typed` variable with typed accessors:
508///
509/// ```ignore
510/// #[miniextendr(dots = typed_list!(x => numeric(), y => integer(), z? => character()))]
511/// pub fn my_func(...) -> String {
512/// let x: f64 = dots_typed.get("x").expect("x");
513/// let y: i32 = dots_typed.get("y").expect("y");
514/// let z: Option<String> = dots_typed.get_opt("z").expect("z");
515/// format!("x={}, y={}", x, y)
516/// }
517/// ```
518///
519/// Type specs: `numeric()`, `integer()`, `logical()`, `character()`, `list()`,
520/// `raw()`, `complex()`, or `"class_name"` for class inheritance checks.
521/// Add `(n)` for exact length: `numeric(4)`. Use `?` suffix for optional fields.
522/// Use `@exact;` prefix for strict mode (reject extra fields).
523///
524/// ## Attributes
525///
526/// - `#[miniextendr(worker)]` — opt into worker-thread execution
527/// - `#[miniextendr(invisible)]` / `#[miniextendr(visible)]` — control return visibility
528/// - `#[miniextendr(check_interrupt)]` — check for user interrupt after call
529/// - `#[miniextendr(coerce)]` — coerce R type before conversion (also usable per-parameter)
530/// - `#[miniextendr(strict)]` — reject lossy conversions for i64/u64/isize/usize
531/// - `#[miniextendr(unwrap_in_r)]` — return `Result<T, E>` to R without unwrapping
532/// - `#[miniextendr(dots = typed_list!(...))]` — validate dots, create `dots_typed`
533/// - `#[miniextendr(internal)]` — adds `@keywords internal` to R wrapper
534/// - `#[miniextendr(noexport)]` — suppresses `@export` from R wrapper
535///
536/// # Impl blocks (class systems)
537///
538/// Apply `#[miniextendr(env|r6|s7|s3|s4)]` to an `impl Type` block.
539/// Use `#[miniextendr(label = "...")]` to disambiguate multiple impl blocks
540/// on the same type.
541/// Registration is automatic.
542///
543/// ## R6 Active Bindings
544///
545/// For R6 classes, use `#[miniextendr(r6(active))]` on methods to create
546/// active bindings (computed properties accessed without parentheses):
547///
548/// ```ignore
549/// use miniextendr_api::miniextendr;
550///
551/// pub struct Rectangle {
552/// width: f64,
553/// height: f64,
554/// }
555///
556/// #[miniextendr(r6)]
557/// impl Rectangle {
558/// pub fn new(width: f64, height: f64) -> Self {
559/// Self { width, height }
560/// }
561///
562/// /// Returns the area (width * height).
563/// #[miniextendr(r6(active))]
564/// pub fn area(&self) -> f64 {
565/// self.width * self.height
566/// }
567///
568/// /// Regular method (requires parentheses).
569/// pub fn scale(&mut self, factor: f64) {
570/// self.width *= factor;
571/// self.height *= factor;
572/// }
573/// }
574/// ```
575///
576/// In R:
577/// ```r
578/// r <- Rectangle$new(3, 4)
579/// r$area # 12 (active binding - no parentheses!)
580/// r$scale(2) # Regular method call
581/// r$area # 24
582/// ```
583///
584/// Active bindings must be getter-only methods taking only `&self`.
585///
586/// ## S7 Properties
587///
588/// For S7 classes, use `#[miniextendr(s7(getter))]` and `#[miniextendr(s7(setter))]`
589/// to create computed properties accessed via `@`:
590///
591/// ```ignore
592/// use miniextendr_api::{miniextendr, ExternalPtr};
593///
594/// #[derive(ExternalPtr)]
595/// pub struct Range {
596/// start: f64,
597/// end: f64,
598/// }
599///
600/// #[miniextendr(s7)]
601/// impl Range {
602/// pub fn new(start: f64, end: f64) -> Self {
603/// Self { start, end }
604/// }
605///
606/// /// Computed property (read-only): length of the range.
607/// #[miniextendr(s7(getter))]
608/// pub fn length(&self) -> f64 {
609/// self.end - self.start
610/// }
611///
612/// /// Dynamic property getter.
613/// #[miniextendr(s7(getter, prop = "midpoint"))]
614/// pub fn get_midpoint(&self) -> f64 {
615/// (self.start + self.end) / 2.0
616/// }
617///
618/// /// Dynamic property setter.
619/// #[miniextendr(s7(setter, prop = "midpoint"))]
620/// pub fn set_midpoint(&mut self, value: f64) {
621/// let half = self.length() / 2.0;
622/// self.start = value - half;
623/// self.end = value + half;
624/// }
625/// }
626/// ```
627///
628/// In R:
629/// ```r
630/// r <- Range(0, 10)
631/// r@length # 10 (computed, read-only)
632/// r@midpoint # 5 (dynamic property)
633/// r@midpoint <- 20 # Adjusts start/end to center at 20
634/// ```
635///
636/// ### Property Attributes
637///
638/// - `#[miniextendr(s7(getter))]` - Read-only computed property
639/// - `#[miniextendr(s7(getter, prop = "name"))]` - Named property getter
640/// - `#[miniextendr(s7(setter, prop = "name"))]` - Named property setter
641/// - `#[miniextendr(s7(getter, default = "0.0"))]` - Property with default value
642/// - `#[miniextendr(s7(getter, required))]` - Required property (error if not provided)
643/// - `#[miniextendr(s7(getter, frozen))]` - Property that can only be set once
644/// - `#[miniextendr(s7(getter, deprecated = "Use X instead"))]` - Deprecated property
645/// - `#[miniextendr(s7(validate))]` - Validator function for property
646///
647/// ## S7 Generic Dispatch Control
648///
649/// Control how S7 generics are created:
650///
651/// - `#[miniextendr(s7(no_dots))]` - Create strict generic without `...`
652/// - `#[miniextendr(s7(dispatch = "x,y"))]` - Multi-dispatch on multiple arguments
653/// - `#[miniextendr(s7(fallback))]` - Register method for `class_any` (catch-all).
654/// The generated R wrapper uses `tryCatch(x@.ptr, error = function(e) x)` to
655/// safely extract the self argument, so non-miniextendr objects won't crash with
656/// a slot-access error. Instead, incompatible objects produce a Rust type-conversion
657/// error when the method tries to interpret the argument as `&Self`.
658///
659/// ```ignore
660/// #[miniextendr(s7)]
661/// impl MyClass {
662/// /// Strict generic: function(x) instead of function(x, ...)
663/// #[miniextendr(s7(no_dots))]
664/// pub fn strict_method(&self) -> i32 { 42 }
665///
666/// /// Fallback method dispatched on class_any.
667/// /// Calling this on a non-MyClass object produces a type-conversion error,
668/// /// not a slot-access crash.
669/// #[miniextendr(s7(fallback))]
670/// pub fn describe(&self) -> String { "generic description".into() }
671/// }
672/// ```
673///
674/// ## S7 Type Conversion (`convert`)
675///
676/// Use `convert_from` and `convert_to` to enable S7's `convert()` for type coercion:
677///
678/// ```ignore
679/// use miniextendr_api::{miniextendr, ExternalPtr};
680///
681/// #[derive(ExternalPtr)]
682/// pub struct Celsius { value: f64 }
683///
684/// #[derive(ExternalPtr)]
685/// pub struct Fahrenheit { value: f64 }
686///
687/// #[miniextendr(s7)]
688/// impl Fahrenheit {
689/// pub fn new(value: f64) -> Self { Self { value } }
690///
691/// /// Convert FROM Celsius TO Fahrenheit.
692/// /// Usage: S7::convert(celsius_obj, Fahrenheit)
693/// #[miniextendr(s7(convert_from = "Celsius"))]
694/// pub fn from_celsius(c: ExternalPtr<Celsius>) -> Self {
695/// Fahrenheit { value: c.value * 9.0 / 5.0 + 32.0 }
696/// }
697///
698/// /// Convert FROM Fahrenheit TO Celsius.
699/// /// Usage: S7::convert(fahrenheit_obj, Celsius)
700/// #[miniextendr(s7(convert_to = "Celsius"))]
701/// pub fn to_celsius(&self) -> Celsius {
702/// Celsius { value: (self.value - 32.0) * 5.0 / 9.0 }
703/// }
704/// }
705/// ```
706///
707/// In R:
708/// ```r
709/// c <- Celsius(100)
710/// f <- S7::convert(c, Fahrenheit) # Uses convert_from
711/// c2 <- S7::convert(f, Celsius) # Uses convert_to
712/// ```
713///
714/// **Note:** Classes must be defined before they can be referenced in convert methods.
715/// Define the "from" class before the "to" class to avoid forward reference issues.
716///
717/// # Traits (ABI)
718///
719/// Apply `#[miniextendr]` to a trait to generate ABI metadata, then use
720/// `#[miniextendr] impl Trait for Type`. Registration is automatic.
721///
722/// # ALTREP
723///
724/// `#[miniextendr]` no longer generates ALTREP classes — `class`/`base`
725/// attributes on a one-field struct are a compile error. Use the per-family
726/// derives (`#[derive(AltrepInteger)]`, …) with `#[altrep(class = "...")]`
727/// instead; registration is automatic there.
728#[proc_macro_attribute]
729pub fn miniextendr(
730 attr: proc_macro::TokenStream,
731 item: proc_macro::TokenStream,
732) -> proc_macro::TokenStream {
733 // Try to parse as function first
734 if syn::parse::<syn::ItemFn>(item.clone()).is_ok() {
735 // Continue with function handling below
736 } else if syn::parse::<syn::ItemImpl>(item.clone()).is_ok() {
737 // Delegate to impl block parser
738 return miniextendr_impl::expand_impl(attr, item);
739 } else if syn::parse::<syn::ItemTrait>(item.clone()).is_ok() {
740 // Delegate to trait ABI generator
741 return miniextendr_trait::expand_trait(attr, item);
742 } else {
743 // Delegate to struct/enum dispatch (handles ALTREP, ExternalPtr, list, dataframe, factor, match_arg)
744 return struct_enum_dispatch::expand_struct_or_enum(attr, item);
745 }
746
747 let MiniextendrFnAttrs {
748 force_worker,
749 force_invisible,
750 check_interrupt,
751 coerce_all,
752 rng,
753 unwrap_in_r,
754 no_preconditions,
755 no_call_attribution,
756 return_pref,
757 return_pref_span,
758 s3_generic,
759 s3_class,
760 dots_spec,
761 dots_span,
762 lifecycle,
763 strict,
764 internal,
765 noexport,
766 export,
767 doc,
768 c_symbol,
769 r_name: fn_r_name,
770 r_entry,
771 r_post_checks,
772 r_on_exit,
773 } = syn::parse_macro_input!(attr as MiniextendrFnAttrs);
774
775 let mut parsed = syn::parse_macro_input!(item as MiniextendrFunctionParsed);
776
777 // Reject async functions
778 if let Some(asyncness) = &parsed.item().sig.asyncness {
779 return syn::Error::new_spanned(
780 asyncness,
781 "async functions are not supported by #[miniextendr]; \
782 R's C API is synchronous and incompatible with async executors",
783 )
784 .into_compile_error()
785 .into();
786 }
787
788 // Validate: reject type/const generic functions.
789 // Lifetime params ARE allowed — they are erased at codegen and produce a single monomorphic
790 // symbol, so `#[no_mangle] extern "C-unwind" fn f<'a>(...)` is valid.
791 // Type and const params require monomorphization → multiple symbols → cannot be #[no_mangle].
792 {
793 let params = &parsed.item().sig.generics.params;
794 let has_type_or_const = params
795 .iter()
796 .any(|p| matches!(p, syn::GenericParam::Type(_) | syn::GenericParam::Const(_)));
797
798 if has_type_or_const {
799 let err = syn::Error::new_spanned(
800 &parsed.item().sig.generics,
801 "#[miniextendr] functions cannot have generic type or const parameters. \
802 Generic functions are incompatible with `extern \"C-unwind\"` and `#[no_mangle]` \
803 required for R FFI. Consider using trait objects or monomorphization instead. \
804 Explicit lifetime parameters are allowed (lifetimes are erased at codegen).",
805 );
806 return err.into_compile_error().into();
807 }
808 }
809
810 // NB: no automatic `#[track_caller]` (dropped in #1121). It made a *direct*
811 // panic's hook-location resolve to the generated wrapper glue instead of the
812 // user's `panic!` line, which defeats the `(at file:line)` suffix now folded
813 // into the R error message from the panic hook. See docs/TRACK_CALLER.md.
814 parsed.add_inline_never_if_needed();
815
816 // Extract commonly used values
817 let uses_internal_c_wrapper = parsed.uses_internal_c_wrapper();
818 let c_ident = if let Some(ref sym) = c_symbol {
819 syn::Ident::new(sym, parsed.c_wrapper_ident().span())
820 } else {
821 parsed.c_wrapper_ident()
822 };
823 let r_wrapper_generator = parsed.r_wrapper_const_ident();
824
825 // Extract references to parsed components
826 let rust_ident = parsed.ident();
827 let inputs = parsed.inputs();
828 let output = parsed.output();
829 let abi = parsed.abi();
830 let attrs = parsed.attrs();
831 let vis = parsed.vis();
832 let generics = parsed.generics();
833 let has_dots = parsed.has_dots();
834 let named_dots = parsed.named_dots().cloned();
835
836 // Fail fast on invalid extern "C-unwind" signatures *before* any codegen,
837 // so we never emit a wrapper that would be discarded by the surfaced error.
838 if let Some(user_abi) = abi
839 && let Err(e) = validate_extern_signature(user_abi, attrs, inputs, output)
840 {
841 return e.into_compile_error().into();
842 }
843
844 // Check for @title/@description conflicts with implicit values (doc-lint feature)
845 // Skip when `doc` attribute overrides the roxygen — implicit docs are irrelevant then.
846 let doc_lint_warnings = if doc.is_some() {
847 proc_macro2::TokenStream::new()
848 } else {
849 crate::roxygen::doc_conflict_warnings(attrs, rust_ident.span())
850 };
851
852 // calling the rust function with
853 let rust_inputs: Vec<syn::Ident> = inputs
854 .iter()
855 .filter_map(|arg| {
856 if let syn::FnArg::Typed(pt) = arg
857 && let syn::Pat::Ident(p) = pt.pat.as_ref()
858 {
859 return Some(p.ident.clone());
860 }
861 None
862 })
863 .collect();
864 // dbg!(&rust_inputs);
865
866 // Validate dots_spec usage (actual injection happens later in the function body)
867 if dots_spec.is_some() && !has_dots {
868 let err = syn::Error::new(
869 dots_span.unwrap_or_else(proc_macro2::Span::call_site),
870 "#[miniextendr(dots = typed_list!(...))] requires a `...` parameter in the function signature",
871 );
872 return err.into_compile_error().into();
873 }
874
875 // Analyze return type to determine:
876 // - Whether it returns SEXP (affects thread strategy)
877 // - Whether result should be invisible
878 let rust_result_ident =
879 syn::Ident::new("__miniextendr_rust_result", proc_macro2::Span::mixed_site());
880 let return_analysis = return_type_analysis::analyze_return_type(
881 output,
882 &rust_result_ident,
883 rust_ident,
884 unwrap_in_r,
885 strict,
886 );
887
888 let returns_sexp = return_analysis.returns_sexp;
889 let is_invisible_return_type = return_analysis.is_invisible;
890
891 // Apply explicit visibility override from #[miniextendr(invisible)] or #[miniextendr(visible)]
892 let is_invisible_return_type = force_invisible.unwrap_or(is_invisible_return_type);
893
894 // Check if any input parameter is main-thread-bound (SEXP or a !Send
895 // framework wrapper like AltrepSexp — neither can move into the worker
896 // closure, so the function must stay on the main thread)
897 let has_sexp_inputs = inputs.iter().any(|arg| {
898 if let syn::FnArg::Typed(pat_type) = arg {
899 is_main_thread_bound_input(pat_type.ty.as_ref())
900 } else {
901 false
902 }
903 });
904
905 // Check if the return type is main-thread-bound (BuiltDataFrame /
906 // DataFrameShape / an R-backed view — all hold R memory and are !Send, so
907 // they cannot cross back from the worker thread; `run_on_worker` requires
908 // `T: Send`). Walks nested positions: Result<BuiltDataFrame, E>,
909 // Option<DataFrameShape>, Vec<...>, tuples.
910 let has_main_thread_bound_return = match output {
911 syn::ReturnType::Default => false,
912 syn::ReturnType::Type(_, ty) => is_main_thread_bound_return(ty),
913 };
914
915 // ═══════════════════════════════════════════════════════════════════════════
916 // Thread Strategy Selection
917 // ═══════════════════════════════════════════════════════════════════════════
918 //
919 // miniextendr supports two execution strategies:
920 //
921 // 1. **Main Thread Strategy** (with_r_unwind_protect) — DEFAULT
922 // - All code runs on R's main thread
923 // - Required when SEXP types are involved (not Send)
924 // - Required for R API calls (Rf_*, R_*)
925 // - Panic handling via R_UnwindProtect (Rust destructors run correctly)
926 // - Errors are always returned as tagged SEXP values; the R wrapper
927 // inspects the tag and raises a structured condition (`rust_*` class
928 // layering) past the Rust boundary.
929 // - Simpler execution model, better R integration
930 //
931 // 2. **Worker Thread Strategy** (run_on_worker + catch_unwind) — OPT-IN
932 // - Argument conversion on main thread (SEXP → Rust types)
933 // - Function execution on dedicated worker thread (clean panic isolation)
934 // - Result conversion on main thread (Rust types → SEXP)
935 // - Panic handling via catch_unwind (prevents unwinding across FFI boundary)
936 // - Opt in with #[miniextendr(worker)]
937 // - ExternalPtr<T> is Send: can be returned from worker thread functions
938 // - R API calls from worker use with_r_thread (serialized to main thread)
939 //
940 // Default: Main thread (simpler execution model, compatible with the
941 // tagged-SEXP error transport)
942 // Override: Use worker thread with #[miniextendr(worker)]
943 //
944 // Thread strategy:
945 // - Main thread is always used unless force_worker is set
946 // - force_worker cannot override hard requirements for main thread
947 // - Hard requirements: returns_sexp, has_sexp_inputs,
948 // has_main_thread_bound_return, has_dots, check_interrupt
949 let requires_main_thread = returns_sexp
950 || has_sexp_inputs
951 || has_main_thread_bound_return
952 || has_dots
953 || check_interrupt;
954 let use_main_thread = !force_worker || requires_main_thread;
955
956 // Extract cfg attributes to apply to generated items (needed by CWrapperContext)
957 let cfg_attrs = extract_cfg_attrs(parsed.attrs());
958
959 let source_loc_doc = source_location_doc(rust_ident.span());
960
961 // Build individual per-parameter coerce and match_arg_several_ok lists
962 let mut coerce_params_list: Vec<String> = Vec::new();
963 let mut match_arg_several_ok_params_list: Vec<String> = Vec::new();
964 for input in inputs.iter() {
965 if let syn::FnArg::Typed(pt) = input
966 && let syn::Pat::Ident(pat_ident) = pt.pat.as_ref()
967 {
968 let param_name = pat_ident.ident.to_string();
969 if parsed.has_coerce_attr(¶m_name) {
970 coerce_params_list.push(param_name.clone());
971 }
972 if parsed.has_match_arg_attr(¶m_name) && parsed.has_several_ok(¶m_name) {
973 match_arg_several_ok_params_list.push(param_name);
974 }
975 }
976 }
977
978 // Build the call expression: rust_ident(rust_input_1, rust_input_2, ...)
979 let fn_call_expr = quote::quote! { #rust_ident(#(#rust_inputs),*) };
980
981 // Determine return handling: use standalone-fn semantics (OptionIntoR for Option<T>)
982 // and handle unwrap_in_r (Result<T, E> → IntoR to pass result list to R).
983 let return_pref_is_set = !matches!(return_pref, crate::miniextendr_fn::ReturnPref::Auto);
984 let fn_return_handling = if unwrap_in_r && crate::return_type_analysis::output_is_result(output)
985 {
986 // In unwrap_in_r mode the IntoR here operates on the whole Result<T,E> (the
987 // framework's IntoR for Result encodes it as a tagged list for R to decode), NOT on
988 // the inner T. prefer= always targets that inner T, so there is nothing for it to
989 // wrap here — reject rather than silently drop it (see apply_return_pref docstring
990 // and the BUG4 audit finding).
991 if return_pref_is_set {
992 let span = return_pref_span.unwrap_or_else(proc_macro2::Span::call_site);
993 return syn::Error::new(
994 span,
995 "`prefer = ...` cannot be combined with `unwrap_in_r` on a function returning \
996 `Result<T, E>`: `unwrap_in_r` converts the whole `Result<T, E>` via a single \
997 `IntoR` impl (a tagged list for R to decode), not the inner `T` alone, so \
998 there is no plain `T` for `prefer=` to wrap. Remove `prefer=`.",
999 )
1000 .into_compile_error()
1001 .into();
1002 }
1003 c_wrapper_builder::ReturnHandling::IntoR
1004 } else {
1005 let auto = c_wrapper_builder::detect_return_handling_standalone_fn(output);
1006 // Apply return_pref override: wraps the result in AsList/AsExternalPtr/AsRNative.
1007 // Only applies to the plain IntoR variant — Option*/Result*/Unit/RawSexp/ExternalPtr
1008 // variants have no bare T to wrap and now hard-error instead of silently ignoring
1009 // prefer= (see apply_return_pref docstring).
1010 match apply_return_pref(auto, return_pref, return_pref_span) {
1011 Ok(handling) => handling,
1012 Err(err) => return err.into_compile_error().into(),
1013 }
1014 };
1015
1016 let thread_strategy = if use_main_thread {
1017 c_wrapper_builder::ThreadStrategy::MainThread
1018 } else {
1019 c_wrapper_builder::ThreadStrategy::WorkerThread
1020 };
1021
1022 // Build CWrapperContext for the standalone fn
1023 let mut c_wrapper_builder =
1024 c_wrapper_builder::CWrapperContext::builder(rust_ident.clone(), c_ident.clone())
1025 .r_wrapper_const(r_wrapper_generator.clone())
1026 .inputs(inputs.iter().cloned().collect())
1027 .output(output.clone())
1028 .call_expr(fn_call_expr)
1029 .thread_strategy(thread_strategy)
1030 .return_handling(fn_return_handling)
1031 .cfg_attrs(cfg_attrs.clone())
1032 .vis(vis.clone())
1033 .generics(generics.clone())
1034 .preserve_param_names();
1035
1036 if uses_internal_c_wrapper {
1037 // Normal Rust fn: generate full C wrapper
1038 } else {
1039 // extern "C-unwind" fn: user wrote the C symbol; only emit R_CallMethodDef
1040 c_wrapper_builder = c_wrapper_builder.skip_wrapper();
1041 }
1042
1043 if coerce_all {
1044 c_wrapper_builder = c_wrapper_builder.coerce_all();
1045 }
1046 for param in &coerce_params_list {
1047 c_wrapper_builder = c_wrapper_builder.with_coerce_param(param.clone());
1048 }
1049 for param in match_arg_several_ok_params_list {
1050 c_wrapper_builder = c_wrapper_builder.match_arg_several_ok(param);
1051 }
1052 if check_interrupt {
1053 c_wrapper_builder = c_wrapper_builder.check_interrupt();
1054 }
1055 if rng {
1056 c_wrapper_builder = c_wrapper_builder.rng();
1057 }
1058 if strict {
1059 c_wrapper_builder = c_wrapper_builder.strict();
1060 }
1061
1062 let c_wrapper = c_wrapper_builder.build().generate();
1063
1064 // region: R wrappers generation in `fn`
1065 // Build R formal parameters and call arguments using shared builder
1066 let mut arg_builder = RArgumentBuilder::new(inputs);
1067 if has_dots {
1068 arg_builder = arg_builder.with_dots(named_dots.clone().map(|id| id.to_string()));
1069 }
1070 // Add user-specified parameter defaults (Missing<T> defaults handled via body prelude)
1071 let mut merged_defaults = parsed.param_defaults();
1072 // For match_arg params, always use the choices placeholder as the formal
1073 // default — the write-time pass replaces it with `c("a", "b", ...)`.
1074 // If the user supplied `default = "\"X\""`, capture X as `preferred_default`
1075 // so the write-time pass rotates X to position 1; the user's literal does
1076 // NOT become the formal (otherwise R's `match.arg` would only see one
1077 // choice).
1078 //
1079 // Tuple: (placeholder, rust_param, preferred_default_unquoted_or_empty)
1080 let mut match_arg_placeholders: Vec<(String, String, String)> = Vec::new();
1081 // (r_name, rust_param) pairs — used later to build @param doc placeholders
1082 let mut match_arg_r_names: Vec<(String, String)> = Vec::new();
1083 for match_arg_param in parsed.match_arg_params() {
1084 let r_name = r_wrapper_builder::normalize_r_arg_string(match_arg_param);
1085 match_arg_r_names.push((r_name.clone(), match_arg_param.clone()));
1086 let preferred = match merged_defaults.get(&r_name) {
1087 Some(raw) => crate::match_arg_keys::extract_match_arg_default(raw),
1088 None => String::new(),
1089 };
1090 let placeholder = crate::match_arg_keys::choices_placeholder(&c_ident.to_string(), &r_name);
1091 merged_defaults.insert(r_name.clone(), placeholder.clone());
1092 match_arg_placeholders.push((placeholder, match_arg_param.clone(), preferred));
1093 }
1094 // Add c("a", "b", "c") default for choices params (idiomatic R match.arg pattern)
1095 for (param_name, choices) in parsed.choices_params() {
1096 let r_name = r_wrapper_builder::normalize_r_arg_string(param_name);
1097 let quoted: Vec<String> = choices.iter().map(|c| format!("\"{}\"", c)).collect();
1098 merged_defaults
1099 .entry(r_name)
1100 .or_insert_with(|| format!("c({})", quoted.join(", ")));
1101 }
1102 arg_builder = arg_builder.with_defaults(merged_defaults);
1103
1104 let r_formals = arg_builder.build_formals();
1105 let mut r_call_args_strs = arg_builder.build_call_args_vec();
1106
1107 // Prepend .call parameter if using internal C wrapper.
1108 // `#[miniextendr(no_call_attribution)]` / `fast` emits `.call = NULL`
1109 // instead of `match.call()` — saves ~1200 ns/call. The R-side
1110 // .miniextendr_raise_condition helper falls back to sys.call() so the
1111 // error UX is preserved (positional args instead of named).
1112 if uses_internal_c_wrapper {
1113 let call_arg = if no_call_attribution {
1114 ".call = NULL".to_string()
1115 } else {
1116 ".call = match.call()".to_string()
1117 };
1118 r_call_args_strs.insert(0, call_arg);
1119 }
1120
1121 // Build the R body string consistently
1122 let c_ident_str = c_ident.to_string();
1123 let call_args_joined = r_call_args_strs.join(", ");
1124 let call_expr = if r_call_args_strs.is_empty() {
1125 format!(".Call({})", c_ident_str)
1126 } else {
1127 format!(".Call({}, {})", c_ident_str, call_args_joined)
1128 };
1129 let r_wrapper_return_str = {
1130 // Capture result, check for tagged condition value, raise R condition if present.
1131 let final_return = if is_invisible_return_type {
1132 "invisible(.val)"
1133 } else {
1134 ".val"
1135 };
1136 crate::method_return_builder::standalone_body(&call_expr, final_return, " ")
1137 };
1138 // Determine R function name and S3-specific comments
1139 let is_s3_method = s3_generic.is_some() || s3_class.is_some();
1140 let r_wrapper_ident_str: String;
1141 let s3_method_comment: String;
1142
1143 if is_s3_method {
1144 // For S3 methods, function name is generic.class
1145 // generic defaults to Rust function name if not specified
1146 let generic = s3_generic.clone().unwrap_or_else(|| rust_ident.to_string());
1147 // s3_class is guaranteed to be Some here because MiniextendrFnAttrs::parse
1148 // validates that s3(...) always has class specified
1149 let class = s3_class.as_ref().expect("s3_class validated at parse time");
1150 r_wrapper_ident_str = format!("{}.{}", generic, class);
1151 // Add @importFrom for vctrs generics so roxygen registers the dependency
1152 let import_comment = if crate::vctrs_generics::is_vctrs_generic(&generic) {
1153 format!("#' @importFrom vctrs {}\n", generic)
1154 } else {
1155 String::new()
1156 };
1157 s3_method_comment = format!("{}#' @method {} {}\n", import_comment, generic, class);
1158 } else if let Some(ref custom_name) = fn_r_name {
1159 r_wrapper_ident_str = custom_name.clone();
1160 s3_method_comment = String::new();
1161 } else if abi.is_some() {
1162 r_wrapper_ident_str = format!("unsafe_{}", rust_ident);
1163 s3_method_comment = String::new();
1164 } else {
1165 r_wrapper_ident_str = rust_ident.to_string();
1166 s3_method_comment = String::new();
1167 };
1168
1169 // Stable, consistent R formatting style: brace on same line, body indented, closing brace on its own line
1170 // r_formals is already a joined string from build_formals()
1171 let formals_joined = r_formals;
1172 let mut roxygen_tags = if let Some(ref doc_text) = doc {
1173 // Custom doc override: each line becomes a separate roxygen tag entry
1174 doc_text.lines().map(|l| l.to_string()).collect()
1175 } else {
1176 crate::roxygen::roxygen_tags_from_attrs(attrs)
1177 };
1178
1179 // Determine lifecycle: explicit attr > #[deprecated] extraction
1180 let lifecycle_spec = lifecycle.or_else(|| {
1181 attrs
1182 .iter()
1183 .find_map(crate::lifecycle::parse_rust_deprecated)
1184 });
1185
1186 // Inject lifecycle badge into roxygen tags if present
1187 if let Some(ref spec) = lifecycle_spec {
1188 crate::lifecycle::inject_lifecycle_badge(&mut roxygen_tags, spec);
1189 }
1190
1191 // Auto-generate @param tags for every non-dots parameter the user didn't
1192 // already document. Priority, per param:
1193 // 1. choices(...) — quoted list, "One of ..." / "One or more of ..."
1194 // 2. match_arg — placeholder resolved at write time (#210)
1195 // 3. everything else — "(no documentation available)" fallback
1196 //
1197 // Collect (doc_placeholder, rust_param) as we go so the write-time resolver
1198 // registry gets an MX_MATCH_ARG_PARAM_DOCS entry for every match_arg param.
1199 let mut match_arg_param_doc_placeholders: Vec<(String, String)> = Vec::new();
1200 for arg in inputs.iter() {
1201 let syn::FnArg::Typed(pt) = arg else {
1202 continue;
1203 };
1204 let syn::Pat::Ident(pat_ident) = pt.pat.as_ref() else {
1205 continue;
1206 };
1207 if parsed.is_dots_param(&pat_ident.ident) {
1208 continue;
1209 }
1210 let rust_name = pat_ident.ident.to_string();
1211 let r_name = r_wrapper_builder::normalize_r_arg_ident(&pat_ident.ident).to_string();
1212 let already_documented = crate::roxygen::param_documented(&roxygen_tags, &r_name);
1213 if already_documented {
1214 continue;
1215 }
1216
1217 if let Some(choices) = parsed.choices_for_param(&rust_name) {
1218 let quoted: Vec<String> = choices.iter().map(|c| format!("\"{}\"", c)).collect();
1219 let prefix = if parsed.has_several_ok(&rust_name) {
1220 "One or more of"
1221 } else {
1222 "One of"
1223 };
1224 roxygen_tags.push(format!("@param {r_name} {prefix} {}.", quoted.join(", ")));
1225 } else if parsed.has_match_arg_attr(&rust_name) {
1226 let doc_placeholder =
1227 crate::match_arg_keys::param_doc_placeholder(&c_ident.to_string(), &r_name);
1228 roxygen_tags.push(format!("@param {r_name} {doc_placeholder}"));
1229 match_arg_param_doc_placeholders.push((doc_placeholder, rust_name));
1230 } else {
1231 roxygen_tags.push(format!("@param {r_name} (no documentation available)"));
1232 }
1233 }
1234
1235 // A standalone function's reference page is titled by its R wrapper name — never
1236 // the doc-comment prose. rustdoc summaries are markdown (intra-doc links, code
1237 // spans) that roxygen2 can't resolve as a `\title`; the prose is promoted to
1238 // `@description` by `roxygen_tags_from_attrs` instead. Without a `@title`, roxygen2
1239 // skips the `.Rd` entirely (#1054), so inject the wrapper name when none exists.
1240 if !roxygen_tags.is_empty() && !crate::roxygen::has_roxygen_tag(&roxygen_tags, "title") {
1241 roxygen_tags.insert(0, format!("@title {}", r_wrapper_ident_str));
1242 }
1243
1244 let roxygen_tags_str = crate::roxygen::format_roxygen_tags(&roxygen_tags);
1245 let has_export_tag = crate::roxygen::has_roxygen_tag(&roxygen_tags, "export");
1246 let has_no_rd_tag = crate::roxygen::has_roxygen_tag(&roxygen_tags, "noRd");
1247 let has_internal_tag = crate::roxygen::has_roxygen_tag(&roxygen_tags, "keywords internal");
1248 // Add roxygen comments: @source for traceability, @export if public
1249 let source_comment = format!(
1250 "#' @source Generated by miniextendr from Rust fn `{}`\n",
1251 rust_ident
1252 );
1253 // Inject @keywords internal if #[miniextendr(internal)] and not already present
1254 let internal_comment = if internal && !has_internal_tag {
1255 "#' @keywords internal\n"
1256 } else {
1257 ""
1258 };
1259 // S3 methods need both @method (for registration) AND @export (for NAMESPACE)
1260 // Don't auto-export functions marked with @noRd, @keywords internal, or attr flags
1261 // #[miniextendr(export)] forces @export even on non-pub functions
1262 let export_comment = if (matches!(vis, syn::Visibility::Public(_)) || export)
1263 && !has_export_tag
1264 && !has_no_rd_tag
1265 && !has_internal_tag
1266 && !internal
1267 && !noexport
1268 {
1269 "#' @export\n".to_string()
1270 } else {
1271 String::new()
1272 };
1273 // `noexport` means no man page at all (docs/CLASS_SYSTEMS.md export-control
1274 // table): inject @noRd so roxygen2 skips the Rd and the write-time
1275 // @rdname-by-file-stem grouping (registry.rs) leaves the fn out of shared
1276 // pages. Without this, an unexported fn keeps a \usage entry in man/,
1277 // which R CMD check flags as a code/documentation mismatch.
1278 let no_rd_comment = if noexport && !has_no_rd_tag {
1279 "#' @noRd\n"
1280 } else {
1281 ""
1282 };
1283 // Generate match.arg prelude for parameters with #[miniextendr(match_arg)]
1284 // Collect (r_param_name, rust_name, rust_type) for each match_arg param
1285 let match_arg_param_info: Vec<(String, String, &syn::Type)> = inputs
1286 .iter()
1287 .filter_map(|arg| {
1288 if let syn::FnArg::Typed(pt) = arg
1289 && let syn::Pat::Ident(pat_ident) = pt.pat.as_ref()
1290 {
1291 let rust_name = pat_ident.ident.to_string();
1292 if parsed.has_match_arg_attr(&rust_name) {
1293 let r_name =
1294 r_wrapper_builder::normalize_r_arg_ident(&pat_ident.ident).to_string();
1295 return Some((r_name, rust_name, pt.ty.as_ref()));
1296 }
1297 }
1298 None
1299 })
1300 .collect();
1301
1302 let match_arg_prelude = if match_arg_param_info.is_empty() {
1303 String::new()
1304 } else {
1305 let mut lines = Vec::new();
1306 for (r_param, rust_name, _) in &match_arg_param_info {
1307 // factor → character normalization
1308 lines.push(format!(
1309 "{param} <- if (is.factor({param})) as.character({param}) else {param}",
1310 param = r_param,
1311 ));
1312 // match.arg pulls the choice list off the formal default (which the
1313 // write-time pass has populated as `c("a", "b", ...)`), so no
1314 // explicit second arg is needed.
1315 if parsed.has_several_ok(rust_name) {
1316 lines.push(format!(
1317 "{param} <- base::match.arg({param}, several.ok = TRUE)",
1318 param = r_param,
1319 ));
1320 } else {
1321 lines.push(format!(
1322 "{param} <- base::match.arg({param})",
1323 param = r_param,
1324 ));
1325 }
1326 }
1327 lines.join("\n ")
1328 };
1329
1330 // Generate idiomatic match.arg prelude for choices params
1331 // These use the simpler pattern: `param <- match.arg(param)` (no C helper call needed)
1332 // With `several_ok`, emit `match.arg(param, several.ok = TRUE)` for multi-value selection
1333 let choices_prelude = {
1334 let mut lines = Vec::new();
1335 for arg in inputs.iter() {
1336 if let syn::FnArg::Typed(pt) = arg
1337 && let syn::Pat::Ident(pat_ident) = pt.pat.as_ref()
1338 {
1339 let rust_name = pat_ident.ident.to_string();
1340 if parsed.choices_for_param(&rust_name).is_some() {
1341 let r_name =
1342 r_wrapper_builder::normalize_r_arg_ident(&pat_ident.ident).to_string();
1343 if parsed.has_several_ok(&rust_name) {
1344 lines.push(format!(
1345 "{r_name} <- match.arg({r_name}, several.ok = TRUE)"
1346 ));
1347 } else {
1348 lines.push(format!("{r_name} <- match.arg({r_name})"));
1349 }
1350 }
1351 }
1352 }
1353 if lines.is_empty() {
1354 String::new()
1355 } else {
1356 lines.join("\n ")
1357 }
1358 };
1359
1360 // Generate lifecycle prelude if needed
1361 let lifecycle_prelude = lifecycle_spec
1362 .as_ref()
1363 .and_then(|spec| spec.r_prelude(&r_wrapper_ident_str));
1364
1365 // Generate R-side precondition checks (stopifnot + fallback precheck calls)
1366 // Skip both match_arg and choices params (already validated by match.arg)
1367 let mut skip_params: std::collections::HashSet<String> =
1368 parsed.match_arg_params().cloned().collect();
1369 for (param_name, _) in parsed.choices_params() {
1370 skip_params.insert(r_wrapper_builder::normalize_r_arg_string(param_name));
1371 }
1372 // `#[miniextendr(no_preconditions)]` / `fast` opts out of the stopifnot
1373 // prelude entirely. TryFromSexp still raises a typed Rust error on
1374 // mismatched input — see analysis/scaffolding-deep-findings-2026-05-20.md
1375 // for why this is ~1230 ns / 1-arg or ~3900 ns / 5-arg of savings.
1376 let precondition_prelude = if no_preconditions {
1377 String::new()
1378 } else {
1379 // A coerced integer-element vector reads via `&[i32]` (INTSXP-only), so its
1380 // precondition tightens to `is.integer` (issue #616). `coerce_params_list`
1381 // holds Rust names; normalize to R names.
1382 let precondition_opts = r_preconditions::PreconditionOptions {
1383 coerce_all,
1384 coerce_params: coerce_params_list
1385 .iter()
1386 .map(|p| r_wrapper_builder::normalize_r_arg_string(p))
1387 .collect(),
1388 };
1389 let precondition_output =
1390 r_preconditions::build_precondition_checks(inputs, &skip_params, &precondition_opts);
1391 if precondition_output.static_checks.is_empty() {
1392 String::new()
1393 } else {
1394 precondition_output.static_checks.join("\n ")
1395 }
1396 };
1397
1398 // Combine all preludes: r_entry, on.exit, lifecycle, static preconditions, match.arg, choices, r_post_checks
1399 // (Missing<T> forwarding lives inline in the `.Call()` args — see
1400 // `build_call_args_vec` — because a prelude binding of the missing
1401 // sentinel errors on lookup.)
1402 let on_exit_str = r_on_exit.as_ref().map(|oe| oe.to_r_code());
1403 let combined_prelude = {
1404 let mut parts = Vec::new();
1405 if let Some(ref entry) = r_entry {
1406 parts.push(entry.as_str());
1407 }
1408 if let Some(ref s) = on_exit_str {
1409 parts.push(s.as_str());
1410 }
1411 if let Some(ref lc) = lifecycle_prelude {
1412 parts.push(lc.as_str());
1413 }
1414 if !precondition_prelude.is_empty() {
1415 parts.push(&precondition_prelude);
1416 }
1417 if !match_arg_prelude.is_empty() {
1418 parts.push(&match_arg_prelude);
1419 }
1420 if !choices_prelude.is_empty() {
1421 parts.push(&choices_prelude);
1422 }
1423 if let Some(ref post) = r_post_checks {
1424 parts.push(post.as_str());
1425 }
1426 if parts.is_empty() {
1427 None
1428 } else {
1429 Some(parts.join("\n "))
1430 }
1431 };
1432
1433 let r_wrapper_string = if let Some(prelude) = combined_prelude {
1434 format!(
1435 "{}{}{}{}{}{}{} <- function({}) {{\n {}\n {}\n}}",
1436 roxygen_tags_str,
1437 source_comment,
1438 s3_method_comment,
1439 internal_comment,
1440 no_rd_comment,
1441 export_comment,
1442 r_wrapper_ident_str,
1443 formals_joined,
1444 prelude,
1445 r_wrapper_return_str
1446 )
1447 } else {
1448 format!(
1449 "{}{}{}{}{}{}{} <- function({}) {{\n {}\n}}",
1450 roxygen_tags_str,
1451 source_comment,
1452 s3_method_comment,
1453 internal_comment,
1454 no_rd_comment,
1455 export_comment,
1456 r_wrapper_ident_str,
1457 formals_joined,
1458 r_wrapper_return_str
1459 )
1460 };
1461 // Use a raw string literal for better readability in macro expansion
1462 let r_wrapper_str = r_wrapper_raw_literal(&r_wrapper_string);
1463
1464 // endregion
1465
1466 // Generate doc strings with links
1467 let r_wrapper_doc = format!(
1468 "R wrapper code for [`{}`], calls [`{}`].",
1469 rust_ident, c_ident
1470 );
1471 let source_start = rust_ident.span().start();
1472 let source_line_lit = syn::LitInt::new(&source_start.line.to_string(), rust_ident.span());
1473 let source_col_lit =
1474 syn::LitInt::new(&(source_start.column + 1).to_string(), rust_ident.span());
1475
1476 // Get the normalized item for output, with roxygen tags stripped from docs.
1477 // Roxygen tags are for R documentation and shouldn't appear in rustdoc.
1478 let mut original_item = parsed.item_without_roxygen();
1479 // Strip only the miniextendr attributes; keep everything else.
1480 original_item
1481 .attrs
1482 .retain(|attr| !attr.path().is_ident("miniextendr"));
1483
1484 // Inject dots_typed binding into function body if dots = typed_list!(...) was specified
1485 if let Some(ref spec_tokens) = dots_spec {
1486 let dots_param = named_dots.clone().unwrap_or_else(|| {
1487 syn::Ident::new("__miniextendr_dots", proc_macro2::Span::call_site())
1488 });
1489 let validation_stmt = build_dots_validation_stmt(&dots_param, spec_tokens);
1490 original_item.block.stmts.insert(0, validation_stmt);
1491 }
1492
1493 let original_item = original_item;
1494
1495 // Generate match_arg choices helper C wrappers and R_CallMethodDef entries
1496 let match_arg_helpers = build_match_arg_helpers(
1497 &match_arg_param_info,
1498 &parsed,
1499 &c_ident.to_string(),
1500 &cfg_attrs,
1501 );
1502
1503 // Generate MX_MATCH_ARG_CHOICES entries for placeholder → choices replacement
1504 // Resolve the `MatchArg`-bound type used in the choices_str closure: for
1505 // `several_ok` params that's the inner element of the container, otherwise
1506 // it's the param type directly.
1507 let choices_ty_for = |rust_param: &str| -> Option<&syn::Type> {
1508 let (_, _, param_ty) = match_arg_param_info
1509 .iter()
1510 .find(|(_, rn, _)| rn == rust_param)?;
1511 let ty: &syn::Type = if parsed.has_several_ok(rust_param) {
1512 classify_several_ok_container(param_ty)
1513 .map(|(_, t)| t)
1514 .unwrap_or(param_ty)
1515 } else {
1516 param_ty
1517 };
1518 Some(ty)
1519 };
1520
1521 let match_arg_choices_entries: Vec<proc_macro2::TokenStream> = match_arg_placeholders
1522 .iter()
1523 .filter_map(|(placeholder, rust_param, preferred_default)| {
1524 let choices_ty = choices_ty_for(rust_param)?;
1525 let entry_ident = syn::Ident::new(
1526 &format!(
1527 "match_arg_choices_entry_{}",
1528 crate::match_arg_keys::placeholder_ident_suffix(placeholder)
1529 ),
1530 proc_macro2::Span::call_site(),
1531 );
1532 Some(crate::match_arg_keys::choices_entry_tokens(
1533 &cfg_attrs,
1534 &entry_ident,
1535 placeholder,
1536 choices_ty,
1537 preferred_default,
1538 ))
1539 })
1540 .collect();
1541
1542 // Generate MX_MATCH_ARG_PARAM_DOCS entries for @param doc placeholder → choice description
1543 let match_arg_param_doc_entries: Vec<proc_macro2::TokenStream> =
1544 match_arg_param_doc_placeholders
1545 .iter()
1546 .filter_map(|(doc_placeholder, rust_param)| {
1547 let choices_ty = choices_ty_for(rust_param)?;
1548 let several_ok_lit = parsed.has_several_ok(rust_param);
1549 let entry_ident = syn::Ident::new(
1550 &format!(
1551 "match_arg_param_doc_entry_{}",
1552 crate::match_arg_keys::placeholder_ident_suffix(doc_placeholder)
1553 ),
1554 proc_macro2::Span::call_site(),
1555 );
1556 Some(crate::match_arg_keys::param_doc_entry_tokens(
1557 &cfg_attrs,
1558 &entry_ident,
1559 doc_placeholder,
1560 several_ok_lit,
1561 choices_ty,
1562 ))
1563 })
1564 .collect();
1565
1566 // Generate doc comment linking to C wrapper and R wrapper constant
1567 let fn_r_wrapper_doc = format!(
1568 "See [`{}`] for C wrapper, [`{}`] for R wrapper.",
1569 c_ident, r_wrapper_generator
1570 );
1571
1572 let expanded: proc_macro::TokenStream = quote::quote! {
1573 // rust function with doc link to R wrapper
1574 #[doc = #fn_r_wrapper_doc]
1575 #original_item
1576
1577 // C wrapper
1578 #(#cfg_attrs)*
1579 #c_wrapper
1580
1581 // R wrapper (self-registers via distributed_slice)
1582 #(#cfg_attrs)*
1583 #[doc = #r_wrapper_doc]
1584 #[doc = concat!("Wraps Rust function `", stringify!(#rust_ident), "`.")]
1585 #[doc = #source_loc_doc]
1586 #[doc = concat!("Generated from source file `", file!(), "`.")]
1587 #[cfg_attr(not(target_arch = "wasm32"), ::miniextendr_api::linkme::distributed_slice(::miniextendr_api::registry::MX_R_WRAPPERS), linkme(crate = ::miniextendr_api::linkme))]
1588 #[allow(non_upper_case_globals)]
1589 #[allow(non_snake_case)]
1590 static #r_wrapper_generator: ::miniextendr_api::registry::RWrapperEntry =
1591 ::miniextendr_api::registry::RWrapperEntry {
1592 priority: ::miniextendr_api::registry::RWrapperPriority::Function,
1593 source_file: file!(),
1594 content: concat!(
1595 "# Generated from Rust fn `",
1596 stringify!(#rust_ident),
1597 "` (",
1598 file!(),
1599 ":",
1600 #source_line_lit,
1601 ":",
1602 #source_col_lit,
1603 ")",
1604 #r_wrapper_str
1605 ),
1606 };
1607
1608 // match_arg choices helpers (C wrappers + R_CallMethodDef entries)
1609 // Each helper's call_method_def self-registers via distributed_slice
1610 #(#match_arg_helpers)*
1611
1612 // match_arg choices entries for R wrapper default replacement
1613 #(#match_arg_choices_entries)*
1614
1615 // match_arg @param doc entries for R wrapper roxygen doc replacement
1616 #(#match_arg_param_doc_entries)*
1617
1618 // doc-lint warnings (if any)
1619 #doc_lint_warnings
1620 }
1621 .into();
1622
1623 expanded
1624}
1625
1626/// Maps a `ReturnPref` attribute value onto an auto-detected `ReturnHandling`.
1627///
1628/// Only the plain `IntoR` variant has a bare `T` for `prefer=` to wrap, so it is the
1629/// only variant substituted with its pref-specific counterpart
1630/// (`AsListOf`/`AsExternalPtrOf`/`AsNativeOf`). Every other variant (`Unit`, `RawSexp`,
1631/// `ExternalPtr`, `Option*`, `Result*`) has its own fixed SEXP-shape rule that
1632/// `prefer=` cannot compose with — returning a compile error for those is better than
1633/// silently dropping the attribute (see the BUG4 audit finding: `prefer = "list"` on an
1634/// `Option<T>` return used to be accepted and silently ignored).
1635fn apply_return_pref(
1636 auto: c_wrapper_builder::ReturnHandling,
1637 pref: crate::miniextendr_fn::ReturnPref,
1638 pref_span: Option<proc_macro2::Span>,
1639) -> syn::Result<c_wrapper_builder::ReturnHandling> {
1640 use crate::miniextendr_fn::ReturnPref;
1641 use c_wrapper_builder::ReturnHandling;
1642
1643 let wrapped = match pref {
1644 ReturnPref::Auto => return Ok(auto),
1645 ReturnPref::List => match auto {
1646 ReturnHandling::IntoR => Some(ReturnHandling::AsListOf),
1647 _ => None,
1648 },
1649 ReturnPref::ExternalPtr => match auto {
1650 ReturnHandling::IntoR => Some(ReturnHandling::AsExternalPtrOf),
1651 _ => None,
1652 },
1653 ReturnPref::Native => match auto {
1654 ReturnHandling::IntoR => Some(ReturnHandling::AsNativeOf),
1655 _ => None,
1656 },
1657 };
1658
1659 wrapped.ok_or_else(|| {
1660 let (pref_name, wrapper_name) = return_pref_names(pref);
1661 let span = pref_span.unwrap_or_else(proc_macro2::Span::call_site);
1662 syn::Error::new(
1663 span,
1664 format!(
1665 "`prefer = \"{pref_name}\"` cannot be honored on this return type. \
1666 `prefer=` only applies to a function returning a plain `T: IntoR` value, \
1667 which it wraps in `{wrapper_name}` before conversion. This function's return \
1668 type falls into a different codegen category ({}) with its own fixed \
1669 SEXP-shape rule, so there is no plain `T` for `prefer=` to wrap. Remove \
1670 `prefer=`, or change the return type to a plain `T`.",
1671 return_handling_category_description(&auto),
1672 ),
1673 )
1674 })
1675}
1676
1677/// Human-readable `(attribute value, wrapper type)` pair for a [`ReturnPref`](crate::miniextendr_fn::ReturnPref),
1678/// used to phrase the `apply_return_pref` compile error.
1679fn return_pref_names(pref: crate::miniextendr_fn::ReturnPref) -> (&'static str, &'static str) {
1680 use crate::miniextendr_fn::ReturnPref;
1681 match pref {
1682 ReturnPref::Auto => ("auto", ""),
1683 ReturnPref::List => ("list", "AsList"),
1684 ReturnPref::ExternalPtr => ("externalptr", "AsExternalPtr"),
1685 ReturnPref::Native => ("native", "AsRNative"),
1686 }
1687}
1688
1689/// Human-readable description of a [`ReturnHandling`](c_wrapper_builder::ReturnHandling)
1690/// category, for the `apply_return_pref` compile error. Only describes the categories
1691/// [`c_wrapper_builder::detect_return_handling_standalone_fn`] can actually produce;
1692/// the wildcard arm covers variants that never reach `apply_return_pref` as `auto`
1693/// (`IntoR` itself, method-only `SelfHandle`, and the `As*Of` variants `apply_return_pref`
1694/// produces as *output*, never takes as input).
1695fn return_handling_category_description(rh: &c_wrapper_builder::ReturnHandling) -> &'static str {
1696 use c_wrapper_builder::ReturnHandling;
1697 match rh {
1698 ReturnHandling::Unit => "the unit return type `()`",
1699 ReturnHandling::RawSexp => "a raw `SEXP` return type",
1700 ReturnHandling::ExternalPtr => {
1701 "a `Self`-returning constructor, already converted via `ExternalPtr::new`"
1702 }
1703 ReturnHandling::OptionUnit => "`Option<()>`",
1704 ReturnHandling::OptionSexp => "`Option<SEXP>`",
1705 ReturnHandling::OptionIntoR | ReturnHandling::OptionIntoRUnwrap => "`Option<T>`",
1706 ReturnHandling::ResultUnit => "`Result<(), E>`",
1707 ReturnHandling::ResultSexp => "`Result<SEXP, E>`",
1708 ReturnHandling::ResultIntoR => "`Result<T, E>`",
1709 ReturnHandling::ResultNullOnErr => "`Result<T, ()>`",
1710 _ => "this return type",
1711 }
1712}
1713
1714/// Generate thread-safe wrappers for R FFI functions.
1715///
1716/// Apply this to an `extern "C-unwind"` block to generate, **for each
1717/// non-variadic function**, a pair of entry points:
1718///
1719/// - The original name (e.g. `Rf_allocVector`) — a safe Rust wrapper that
1720/// runs directly on R's main thread, routes through
1721/// `miniextendr_api::worker::with_r_thread` from an active miniextendr
1722/// worker context, and panics for arbitrary off-main callers.
1723/// - A `*_unchecked` sibling (`Rf_allocVector_unchecked`) — the raw
1724/// `extern "C-unwind"` declaration with no main-thread assertion and no
1725/// worker round-trip.
1726///
1727/// User code should reach for the checked variant by default; the unchecked
1728/// sibling exists for three known-safe contexts:
1729///
1730/// 1. **Inside ALTREP callbacks** — R is already calling us on the main
1731/// thread, so the assertion would always pass and the route would
1732/// deadlock the call back to R.
1733/// 2. **Inside a `with_r_unwind_protect` body** — the guard has established
1734/// main-thread context, and re-entering `with_r_thread` would nest two
1735/// `R_UnwindProtect` frames (paying the longjmp-leak cost twice).
1736/// 3. **Inside a `with_r_thread` body** — the assertion is redundant; you
1737/// are already where you needed to be.
1738///
1739/// The build-time lint **MXL301** enforces this: calling `*_unchecked`
1740/// outside one of those three contexts is a compile-time error. Without the
1741/// `worker-thread` feature, the checked variant still enforces the recorded
1742/// main-thread contract; it simply has no worker route available.
1743///
1744/// # Tradeoffs at a glance
1745///
1746/// | Variant | Asserts main thread | Routes to main | When to use |
1747/// |---|---|---|---|
1748/// | `Rf_foo` (checked) | yes (debug) | yes (from worker) | default |
1749/// | `Rf_foo_unchecked` | no | no | ALTREP callbacks, `with_r_unwind_protect`, `with_r_thread` |
1750///
1751/// # Behavior
1752///
1753/// All non-variadic functions are routed to the main thread via `with_r_thread`
1754/// when called from a worker thread. The return value is wrapped in `Sendable`
1755/// and sent back to the caller. This applies to both value-returning functions
1756/// (SEXP, i32, etc.) and pointer-returning functions (`*const T`, `*mut T`).
1757///
1758/// Pointer-returning functions (like `INTEGER`, `REAL`) are safe to route because
1759/// the underlying SEXP must be GC-protected by the caller, and R's GC only runs
1760/// during R API calls which are serialized through `with_r_thread`.
1761///
1762/// # Initialization Requirement
1763///
1764/// `miniextendr_runtime_init()` must be called before using any wrapped function.
1765/// Calling before initialization will panic with a descriptive error message.
1766///
1767/// # Limitations
1768///
1769/// - Variadic functions are passed through unchanged (no wrapper)
1770/// - Statics are passed through unchanged
1771/// - Functions with `#[link_name]` are passed through unchanged
1772///
1773/// # Example
1774///
1775/// ```ignore
1776/// #[r_ffi_checked]
1777/// unsafe extern "C-unwind" {
1778/// // Routed to main thread via with_r_thread when called from worker
1779/// pub fn Rf_ScalarInteger(arg1: i32) -> SEXP;
1780/// pub fn INTEGER(x: SEXP) -> *mut i32;
1781/// }
1782/// ```
1783#[proc_macro_attribute]
1784pub fn r_ffi_checked(
1785 _attr: proc_macro::TokenStream,
1786 item: proc_macro::TokenStream,
1787) -> proc_macro::TokenStream {
1788 let foreign_mod = syn::parse_macro_input!(item as syn::ItemForeignMod);
1789
1790 let foreign_mod_attrs = &foreign_mod.attrs;
1791 let abi = &foreign_mod.abi;
1792 let mut unchecked_items = Vec::new();
1793 let mut checked_wrappers = Vec::new();
1794
1795 for item in &foreign_mod.items {
1796 match item {
1797 syn::ForeignItem::Fn(fn_item) => {
1798 let is_variadic = fn_item.sig.variadic.is_some();
1799
1800 // Check if function already has #[link_name] - if so, pass through unchanged
1801 let has_link_name = fn_item
1802 .attrs
1803 .iter()
1804 .any(|attr| attr.path().is_ident("link_name"));
1805
1806 if is_variadic || has_link_name {
1807 // Pass through variadic functions and functions with explicit link_name unchanged
1808 unchecked_items.push(item.clone());
1809 } else {
1810 // Generate checked wrapper for non-variadic functions
1811 let vis = &fn_item.vis;
1812 let fn_name = &fn_item.sig.ident;
1813 let fn_name_str = fn_name.to_string();
1814 let unchecked_name = quote::format_ident!("{}_unchecked", fn_name);
1815 let unchecked_name_str = unchecked_name.to_string();
1816 let inputs = &fn_item.sig.inputs;
1817 let output = &fn_item.sig.output;
1818 // Filter out link_name attributes (already checked above, but be safe)
1819 let attrs: Vec<_> = fn_item
1820 .attrs
1821 .iter()
1822 .filter(|attr| !attr.path().is_ident("link_name"))
1823 .collect();
1824 let checked_doc = format!(
1825 "Checked wrapper for `{}`. Calls `{}` and routes through `with_r_thread`.",
1826 fn_name_str, unchecked_name_str
1827 );
1828 let checked_doc_lit = syn::LitStr::new(&checked_doc, fn_name.span());
1829 let source_loc_doc = crate::source_location_doc(fn_name.span());
1830 let source_loc_doc_lit = syn::LitStr::new(&source_loc_doc, fn_name.span());
1831
1832 // Generate the unchecked FFI binding with #[link_name]
1833 // Same visibility as the checked variant
1834 let link_name = syn::LitStr::new(&fn_name_str, fn_name.span());
1835 let unchecked_fn: syn::ForeignItem = syn::parse_quote! {
1836 #(#attrs)*
1837 #[doc = concat!("Unchecked FFI binding for `", stringify!(#fn_name), "`.")]
1838 #[doc = #source_loc_doc_lit]
1839 #[doc = concat!("Generated from source file `", file!(), "`.")]
1840 #[link_name = #link_name]
1841 #vis fn #unchecked_name(#inputs) #output;
1842 };
1843 unchecked_items.push(unchecked_fn);
1844
1845 // Generate a checked wrapper function
1846 let arg_names: Vec<_> = inputs
1847 .iter()
1848 .filter_map(|arg| {
1849 if let syn::FnArg::Typed(pat_type) = arg
1850 && let syn::Pat::Ident(pat_ident) = pat_type.pat.as_ref()
1851 {
1852 Some(pat_ident.ident.clone())
1853 } else {
1854 None
1855 }
1856 })
1857 .collect();
1858
1859 let is_never = matches!(output, syn::ReturnType::Type(_, ty) if matches!(**ty, syn::Type::Never(_)));
1860
1861 let wrapper = if is_never {
1862 // Never-returning functions (like Rf_error)
1863 quote::quote! {
1864 #(#attrs)*
1865 #[doc = #checked_doc_lit]
1866 #[doc = #source_loc_doc_lit]
1867 #[doc = concat!("Generated from source file `", file!(), "`.")]
1868 #[inline(always)]
1869 #[allow(non_snake_case)]
1870 #vis unsafe fn #fn_name(#inputs) #output {
1871 ::miniextendr_api::worker::with_r_thread(move || unsafe {
1872 #unchecked_name(#(#arg_names),*)
1873 })
1874 }
1875 }
1876 } else {
1877 // Normal functions - route via with_r_thread
1878 quote::quote! {
1879 #(#attrs)*
1880 #[doc = #checked_doc_lit]
1881 #[doc = #source_loc_doc_lit]
1882 #[doc = concat!("Generated from source file `", file!(), "`.")]
1883 #[inline(always)]
1884 #[allow(non_snake_case)]
1885 #vis unsafe fn #fn_name(#inputs) #output {
1886 let result = ::miniextendr_api::worker::with_r_thread(move || {
1887 ::miniextendr_api::worker::Sendable(unsafe {
1888 #unchecked_name(#(#arg_names),*)
1889 })
1890 });
1891 result.0
1892 }
1893 }
1894 };
1895 checked_wrappers.push(wrapper);
1896 }
1897 }
1898 _ => {
1899 // Pass through statics and other items unchanged
1900 unchecked_items.push(item.clone());
1901 }
1902 }
1903 }
1904
1905 let expanded = quote::quote! {
1906 #(#foreign_mod_attrs)*
1907 unsafe #abi {
1908 #(#unchecked_items)*
1909 }
1910
1911 #(#checked_wrappers)*
1912 };
1913
1914 expanded.into()
1915}
1916
1917/// Derive macro for implementing `RNativeType` on a newtype wrapper.
1918///
1919/// This allows newtype wrappers around R native types to work with `Vec<T>`,
1920/// `&[T]` conversions and the `Coerce<R>` traits.
1921/// The inner type must implement `RNativeType`.
1922///
1923/// # Supported Struct Forms
1924///
1925/// Both tuple structs and single-field named structs are supported:
1926///
1927/// ```ignore
1928/// use miniextendr_api::RNativeType;
1929///
1930/// // Tuple struct (most common)
1931/// #[derive(Clone, Copy, RNativeType)]
1932/// struct UserId(i32);
1933///
1934/// // Named single-field struct
1935/// #[derive(Clone, Copy, RNativeType)]
1936/// struct Temperature { celsius: f64 }
1937/// ```
1938///
1939/// # Generated Code
1940///
1941/// For `struct UserId(i32)`, this generates:
1942///
1943/// ```ignore
1944/// impl RNativeType for UserId {
1945/// const SEXP_TYPE: SEXPTYPE = <i32 as RNativeType>::SEXP_TYPE;
1946/// const R_NA: Self = UserId(<i32 as RNativeType>::R_NA);
1947///
1948/// unsafe fn dataptr_mut(sexp: SEXP) -> *mut Self {
1949/// <i32 as RNativeType>::dataptr_mut(sexp).cast()
1950/// }
1951/// }
1952/// ```
1953///
1954/// # Using the Newtype with Coerce
1955///
1956/// Once `RNativeType` is derived, you can implement `Coerce` to/from the newtype:
1957///
1958/// ```ignore
1959/// impl Coerce<UserId> for i32 {
1960/// fn coerce(self) -> UserId { UserId(self) }
1961/// }
1962///
1963/// let id: UserId = 42.coerce();
1964/// ```
1965///
1966/// # Requirements
1967///
1968/// - Must be a newtype struct (exactly one field, tuple or named)
1969/// - The inner type must implement `RNativeType` (`i32`, `f64`, `RLogical`, `u8`, `Rcomplex`)
1970/// - Should also derive `Copy` (required by `RNativeType: Copy`)
1971#[proc_macro_derive(RNativeType)]
1972pub fn derive_rnative_type(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
1973 let input = syn::parse_macro_input!(input as syn::DeriveInput);
1974 let name = &input.ident;
1975 let (impl_generics, ty_generics, where_clause) = input.generics.split_for_impl();
1976
1977 // Extract inner type and constructor — must be a newtype (single field)
1978 let (inner_ty, elt_ctor): (syn::Type, proc_macro2::TokenStream) = match &input.data {
1979 syn::Data::Struct(data) => match &data.fields {
1980 syn::Fields::Unnamed(fields) if fields.unnamed.len() == 1 => {
1981 let ty = fields.unnamed.first().unwrap().ty.clone();
1982 let ctor = quote::quote! { Self(val) };
1983 (ty, ctor)
1984 }
1985 syn::Fields::Named(fields) if fields.named.len() == 1 => {
1986 let field = fields.named.first().unwrap();
1987 let ty = field.ty.clone();
1988 let field_name = field.ident.as_ref().unwrap();
1989 let ctor = quote::quote! { Self { #field_name: val } };
1990 (ty, ctor)
1991 }
1992 _ => {
1993 return syn::Error::new_spanned(
1994 name,
1995 "#[derive(RNativeType)] requires a newtype struct with exactly one field",
1996 )
1997 .into_compile_error()
1998 .into();
1999 }
2000 },
2001 _ => {
2002 return syn::Error::new_spanned(name, "#[derive(RNativeType)] only works on structs")
2003 .into_compile_error()
2004 .into();
2005 }
2006 };
2007
2008 let expanded = quote::quote! {
2009 impl #impl_generics ::miniextendr_api::RNativeType for #name #ty_generics #where_clause {
2010 const SEXP_TYPE: ::miniextendr_api::SEXPTYPE =
2011 <#inner_ty as ::miniextendr_api::RNativeType>::SEXP_TYPE;
2012
2013 const R_NA: Self = {
2014 let val = <#inner_ty as ::miniextendr_api::RNativeType>::R_NA;
2015 #elt_ctor
2016 };
2017
2018 #[inline]
2019 unsafe fn dataptr_mut(sexp: ::miniextendr_api::SEXP) -> *mut Self {
2020 // Newtype is repr(transparent), so we can cast the pointer
2021 unsafe {
2022 <#inner_ty as ::miniextendr_api::RNativeType>::dataptr_mut(sexp).cast()
2023 }
2024 }
2025
2026 #[inline]
2027 fn elt(sexp: ::miniextendr_api::SEXP, i: isize) -> Self {
2028 let val = <#inner_ty as ::miniextendr_api::RNativeType>::elt(sexp, i);
2029 #elt_ctor
2030 }
2031 }
2032
2033 };
2034
2035 expanded.into()
2036}
2037
2038/// Derive macro for implementing `TypedExternal` on a type.
2039///
2040/// This makes the type compatible with `ExternalPtr<T>` for storing in R's external pointers.
2041///
2042/// # Basic Usage
2043///
2044/// ```ignore
2045/// use miniextendr_api::TypedExternal;
2046///
2047/// #[derive(ExternalPtr)]
2048/// struct MyData {
2049/// value: i32,
2050/// }
2051///
2052/// // Now you can use ExternalPtr<MyData>
2053/// let ptr = ExternalPtr::new(MyData { value: 42 });
2054/// ```
2055///
2056/// # Trait ABI
2057///
2058/// Trait dispatch wrappers are automatically generated:
2059///
2060/// ```ignore
2061/// use miniextendr_api::miniextendr;
2062///
2063/// #[derive(ExternalPtr)]
2064/// struct MyCounter {
2065/// value: i32,
2066/// }
2067///
2068/// #[miniextendr]
2069/// impl Counter for MyCounter {
2070/// fn value(&self) -> i32 { self.value }
2071/// fn increment(&mut self) { self.value += 1; }
2072/// }
2073/// ```
2074///
2075/// This generates additional infrastructure for type-erased trait dispatch:
2076/// - `__MxWrapperMyCounter` - Type-erased wrapper struct
2077/// - `__MX_BASE_VTABLE_MYCOUNTER` - Base vtable with drop/query
2078/// - `__mx_wrap_mycounter()` - Constructor returning `*mut mx_erased`
2079///
2080/// # Generated Code (Basic)
2081///
2082/// For a type `MyData` without traits:
2083///
2084/// ```ignore
2085/// impl TypedExternal for MyData {
2086/// const TYPE_NAME: &'static str = "MyData";
2087/// const TYPE_NAME_CSTR: &'static [u8] = b"MyData\0";
2088/// }
2089/// ```
2090#[proc_macro_derive(ExternalPtr, attributes(externalptr, r_data))]
2091pub fn derive_external_ptr(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2092 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2093
2094 // Standalone `#[derive(ExternalPtr)]` always emits the `IntoExternalPtr`
2095 // marker (enabling the blanket `IntoR`); only the struct-level
2096 // `prefer = "native"` dispatch path suppresses it (#1283).
2097 externalptr_derive::derive_external_ptr(input, true)
2098 .unwrap_or_else(|e| e.into_compile_error())
2099 .into()
2100}
2101
2102/// Derive macro for ALTREP integer vector data types.
2103///
2104/// Auto-implements `AltrepLen`, `AltIntegerData`, and the low-level ALTREP
2105/// trait impls (`Altrep`, `AltVec`, `AltInteger`, `InferBase`).
2106///
2107/// # Attributes
2108///
2109/// - `#[altrep(len = "field_name")]` - Specify length field (auto-detects "len" or "length")
2110/// - `#[altrep(elt = "field_name")]` - For constant vectors, specify which field provides elements
2111/// - `#[altrep(dataptr)]` - Enable direct data-pointer access
2112/// - `#[altrep(serialize)]` - Enable ALTREP serialization support
2113/// - `#[altrep(subset)]` - Enable `Extract_subset` optimization
2114/// - `#[altrep(no_lowlevel)]` - Skip the automatic low-level trait impls
2115///
2116/// # Example (Constant Vector - Zero Boilerplate!)
2117///
2118/// ```ignore
2119/// #[derive(ExternalPtr, AltrepInteger)]
2120/// #[altrep(elt = "value")] // All elements return this field
2121/// pub struct ConstantIntData {
2122/// value: i32,
2123/// len: usize,
2124/// }
2125///
2126/// // That's it! 3 lines instead of 30!
2127/// // AltrepLen, AltIntegerData, and low-level impls are auto-generated
2128///
2129/// #[miniextendr(class = "ConstantInt")]
2130/// pub struct ConstantIntClass(pub ConstantIntData);
2131/// ```
2132///
2133/// # Example (Custom elt() - Override One Method)
2134///
2135/// ```ignore
2136/// #[derive(ExternalPtr, AltrepInteger)]
2137/// pub struct ArithSeqData {
2138/// start: i32,
2139/// step: i32,
2140/// len: usize,
2141/// }
2142///
2143/// // Auto-generates AltrepLen and stub AltIntegerData
2144/// // Just override elt() for custom logic:
2145/// impl AltIntegerData for ArithSeqData {
2146/// fn elt(&self, i: usize) -> i32 {
2147/// self.start + (i as i32) * self.step
2148/// }
2149/// }
2150/// ```
2151#[proc_macro_derive(AltrepInteger, attributes(altrep))]
2152pub fn derive_altrep_integer(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2153 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2154 altrep_derive::derive_altrep_integer(input)
2155 .unwrap_or_else(|e| e.into_compile_error())
2156 .into()
2157}
2158
2159/// Derive macro for ALTREP real vector data types.
2160///
2161/// Auto-implements `AltrepLen` and `AltRealData` traits.
2162/// Supports the same `#[altrep(...)]` attributes as [`AltrepInteger`](derive@AltrepInteger).
2163#[proc_macro_derive(AltrepReal, attributes(altrep))]
2164pub fn derive_altrep_real(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2165 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2166 altrep_derive::derive_altrep_real(input)
2167 .unwrap_or_else(|e| e.into_compile_error())
2168 .into()
2169}
2170
2171/// Derive macro for ALTREP logical vector data types.
2172///
2173/// Auto-implements `AltrepLen` and `AltLogicalData` traits.
2174/// Supports the same `#[altrep(...)]` attributes as [`AltrepInteger`](derive@AltrepInteger).
2175#[proc_macro_derive(AltrepLogical, attributes(altrep))]
2176pub fn derive_altrep_logical(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2177 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2178 altrep_derive::derive_altrep_logical(input)
2179 .unwrap_or_else(|e| e.into_compile_error())
2180 .into()
2181}
2182
2183/// Derive macro for ALTREP raw vector data types.
2184///
2185/// Auto-implements `AltrepLen` and `AltRawData` traits.
2186/// Supports the same `#[altrep(...)]` attributes as [`AltrepInteger`](derive@AltrepInteger).
2187#[proc_macro_derive(AltrepRaw, attributes(altrep))]
2188pub fn derive_altrep_raw(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2189 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2190 altrep_derive::derive_altrep_raw(input)
2191 .unwrap_or_else(|e| e.into_compile_error())
2192 .into()
2193}
2194
2195/// Derive macro for ALTREP string vector data types.
2196///
2197/// Auto-implements `AltrepLen` and `AltStringData` traits.
2198/// Supports the same `#[altrep(...)]` attributes as [`AltrepInteger`](derive@AltrepInteger).
2199#[proc_macro_derive(AltrepString, attributes(altrep))]
2200pub fn derive_altrep_string(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2201 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2202 altrep_derive::derive_altrep_string(input)
2203 .unwrap_or_else(|e| e.into_compile_error())
2204 .into()
2205}
2206
2207/// Derive macro for ALTREP complex vector data types.
2208///
2209/// Auto-implements `AltrepLen` and `AltComplexData` traits.
2210/// Supports the same `#[altrep(...)]` attributes as [`AltrepInteger`](derive@AltrepInteger).
2211#[proc_macro_derive(AltrepComplex, attributes(altrep))]
2212pub fn derive_altrep_complex(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2213 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2214 altrep_derive::derive_altrep_complex(input)
2215 .unwrap_or_else(|e| e.into_compile_error())
2216 .into()
2217}
2218
2219/// Derive macro for ALTREP list vector data types.
2220///
2221/// Auto-implements `AltrepLen` and `AltListData` traits.
2222/// Supports the same `#[altrep(...)]` attributes as [`AltrepInteger`](derive@AltrepInteger),
2223/// except `dataptr` and `subset` which are not supported for list ALTREP.
2224#[proc_macro_derive(AltrepList, attributes(altrep))]
2225pub fn derive_altrep_list(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2226 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2227 altrep_derive::derive_altrep_list(input)
2228 .unwrap_or_else(|e| e.into_compile_error())
2229 .into()
2230}
2231
2232/// Derive ALTREP registration for a data struct.
2233///
2234/// Generates `TypedExternal`, `AltrepClass`, `RegisterAltrep`, `IntoR`,
2235/// linkme registration entry, and `Ref`/`Mut` accessor types.
2236///
2237/// The struct must already have low-level ALTREP traits implemented.
2238/// For most use cases, prefer a family-specific derive:
2239/// `#[derive(AltrepInteger)]`, `#[derive(AltrepReal)]`, etc.
2240/// Use `#[altrep(manual)]` on a family derive to skip data trait generation
2241/// when you provide your own `AltrepLen` + `Alt*Data` impls.
2242///
2243/// # Attributes
2244///
2245/// - `#[altrep(class = "Name")]` — custom ALTREP class name (defaults to struct name)
2246///
2247/// # Example
2248///
2249/// ```ignore
2250/// // Prefer family derives with manual:
2251/// #[derive(AltrepInteger)]
2252/// #[altrep(manual, class = "MyCustom", serialize)]
2253/// struct MyData { ... }
2254///
2255/// impl AltrepLen for MyData { ... }
2256/// impl AltIntegerData for MyData { ... }
2257/// ```
2258#[proc_macro_derive(Altrep, attributes(altrep))]
2259pub fn derive_altrep(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2260 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2261 altrep::derive_altrep(input)
2262 .unwrap_or_else(|e| e.into_compile_error())
2263 .into()
2264}
2265
2266/// Derive `IntoList` for a struct (Rust → R list).
2267///
2268/// - Named structs → named R list: `list(x = 1L, y = 2L)`
2269/// - Tuple structs → unnamed R list: `list(1L, 2L)`
2270/// - Fields annotated `#[into_list(ignore)]` are skipped
2271#[proc_macro_derive(IntoList, attributes(into_list))]
2272pub fn derive_into_list(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2273 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2274 list_derive::derive_into_list(input)
2275 .unwrap_or_else(|e| e.into_compile_error())
2276 .into()
2277}
2278
2279/// Derive `TryFromList` for a struct (R list → Rust).
2280///
2281/// - Named structs: extract by field name
2282/// - Tuple structs: extract by position (0, 1, 2, ...)
2283/// - Fields annotated `#[into_list(ignore)]` are not read and are initialized with `Default::default()`
2284#[proc_macro_derive(TryFromList, attributes(into_list))]
2285pub fn derive_try_from_list(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2286 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2287 list_derive::derive_try_from_list(input)
2288 .unwrap_or_else(|e| e.into_compile_error())
2289 .into()
2290}
2291
2292/// Derive `PreferList`: emits an `IntoR` impl selecting list as the type's default
2293/// Rust→R conversion (via `IntoList::into_list`).
2294///
2295/// A type carries exactly one representation default: stacking two `Prefer*`
2296/// derives is a compile error. Each `Prefer*` derive emits a fixed-name marker
2297/// const, so a second one triggers a guided `duplicate definitions with name
2298/// __miniextendr_conflicting_Prefer_derives__keep_ONE_or_use_call_site_As_wrappers`
2299/// error (alongside the raw conflicting-`IntoR`-impl error) — keep one `Prefer*`,
2300/// or drop them all and choose a representation per return value at the call site
2301/// with an `As*` wrapper (`AsList`, `AsExternalPtr`, `AsDataFrame`, ...).
2302///
2303/// # Example
2304///
2305/// ```ignore
2306/// #[derive(IntoList, PreferList)]
2307/// struct Config { verbose: bool, threads: i32 }
2308/// // IntoR produces list(verbose = TRUE, threads = 4L)
2309/// ```
2310#[proc_macro_derive(PreferList)]
2311pub fn derive_prefer_list(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2312 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2313 list_derive::derive_prefer_list(input)
2314 .unwrap_or_else(|e| e.into_compile_error())
2315 .into()
2316}
2317
2318/// Derive `PreferDataFrame`: when a type implements both `IntoDataFrame` (via `DataFrameRow`)
2319/// and other conversion paths, this selects data.frame as the default `IntoR` conversion.
2320///
2321/// # Example
2322///
2323/// ```ignore
2324/// #[derive(DataFrameRow, PreferDataFrame)]
2325/// struct Obs { time: f64, value: f64 }
2326/// // IntoR produces data.frame(time = ..., value = ...)
2327/// ```
2328#[proc_macro_derive(PreferDataFrame)]
2329pub fn derive_prefer_data_frame(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2330 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2331 list_derive::derive_prefer_data_frame(input)
2332 .unwrap_or_else(|e| e.into_compile_error())
2333 .into()
2334}
2335
2336/// Derive `PreferExternalPtr`: when a type implements both `ExternalPtr` and
2337/// other conversion paths (e.g., `IntoList`), this selects `ExternalPtr` wrapping
2338/// as the default `IntoR` conversion.
2339///
2340/// # Example
2341///
2342/// ```ignore
2343/// #[derive(ExternalPtr, IntoList, PreferExternalPtr)]
2344/// struct Model { weights: Vec<f64> }
2345/// // IntoR wraps as ExternalPtr (opaque R object), not list
2346/// ```
2347#[proc_macro_derive(PreferExternalPtr)]
2348pub fn derive_prefer_externalptr(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2349 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2350 list_derive::derive_prefer_externalptr(input)
2351 .unwrap_or_else(|e| e.into_compile_error())
2352 .into()
2353}
2354
2355/// Derive `PreferRNativeType`: when a newtype wraps an `RNativeType` and also
2356/// implements other conversions, this selects the native R vector conversion
2357/// as the default `IntoR` path.
2358///
2359/// # Example
2360///
2361/// ```ignore
2362/// #[derive(Copy, Clone, RNativeType, PreferRNativeType)]
2363/// struct Meters(f64);
2364/// // IntoR produces a numeric scalar, not an ExternalPtr
2365/// ```
2366#[proc_macro_derive(PreferRNativeType)]
2367pub fn derive_prefer_rnative(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2368 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2369 list_derive::derive_prefer_rnative(input)
2370 .unwrap_or_else(|e| e.into_compile_error())
2371 .into()
2372}
2373
2374/// Derive `PreferVctrs`: emits an `IntoR` impl converting the type to its R vctrs object via
2375/// `IntoVctrs::into_vctrs`.
2376///
2377/// Pair with `#[derive(Vctrs)]` (which supplies `IntoVctrs`) so the type can be returned
2378/// directly from `#[miniextendr]` functions instead of `value.into_vctrs().map_err(...)`.
2379///
2380/// # Example
2381///
2382/// ```ignore
2383/// #[derive(Vctrs, PreferVctrs)]
2384/// #[vctrs(class = "percent", base = "double")]
2385/// struct Percent { #[vctrs(data)] values: Vec<f64> }
2386/// // IntoR builds the `percent` vctrs vector; a build failure becomes an R error.
2387/// ```
2388#[cfg(feature = "vctrs")]
2389#[proc_macro_derive(PreferVctrs)]
2390pub fn derive_prefer_vctrs(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2391 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2392 list_derive::derive_prefer_vctrs(input)
2393 .unwrap_or_else(|e| e.into_compile_error())
2394 .into()
2395}
2396
2397/// Derive `DataFrameRow`: generates a companion `*DataFrame` type with collection fields,
2398/// plus `IntoR` / `TryFromSexp` / `IntoDataFrame` impls for seamless R data.frame conversion.
2399///
2400/// # Example
2401///
2402/// ```ignore
2403/// #[derive(DataFrameRow)]
2404/// struct Measurement {
2405/// time: f64,
2406/// value: f64,
2407/// }
2408///
2409/// // Generates MeasurementDataFrame { time: Vec<f64>, value: Vec<f64> }
2410/// // plus conversion impls
2411/// ```
2412///
2413/// # Struct-level attributes
2414///
2415/// - `#[dataframe(name = "CustomDf")]` — custom name for the generated DataFrame type
2416/// - `#[dataframe(align)]` — pad shorter columns with NA to match longest
2417/// - `#[dataframe(tag = "my_tag")]` — attach a tag attribute to the data.frame
2418/// - `#[dataframe(conflicts = "string")]` — resolve conflicting column types as strings
2419///
2420/// # Field-level attributes
2421///
2422/// - `#[dataframe(skip)]` — omit this field from the DataFrame
2423/// - `#[dataframe(rename = "col")]` — custom column name
2424/// - `#[dataframe(as_list)]` — keep collection as single list column (no expansion)
2425/// - `#[dataframe(expand)]` / `#[dataframe(unnest)]` — expand collection into suffixed columns
2426/// - `#[dataframe(width = N)]` — pin expansion width (shorter rows get NA)
2427///
2428/// # Public surface (which verbs to call)
2429///
2430/// Every capability the derive provides has a documented, trait-based (or `std`)
2431/// verb — reach for these, not any incidental inherent plumbing:
2432///
2433/// - **Rows → R `data.frame`**: `rows.into_dataframe()?` (owned, GC-rooted
2434/// `BuiltDataFrame`) or `rows.wrap_data_frame()` (deferred `IntoR` wrapper);
2435/// parallel variant `rows.into_dataframe_par()?`. From the `IntoDataFrame` /
2436/// `AsDataFrameExt` traits (both re-exported from `miniextendr_api::prelude`).
2437/// - **R `data.frame` → rows**: `Vec::<Row>::from_dataframe(&df)?` (parallel:
2438/// `Vec::<Row>::from_dataframe_par(&df)?`), from the `FromDataFrame` trait — or
2439/// the one-call `Row::try_from_dataframe(sexp)` reader on the row type.
2440/// - **Rows ↔ the pure-Rust columnar companion** (`<Row>DataFrame`, `Vec`-columns,
2441/// no R involved): the `ColumnarFrame` trait (in the prelude) —
2442/// `<Row>DataFrame::from_rows(rows)` / `from_rows_par(rows)` (parallel build of
2443/// the *companion*, which `into_dataframe_par` does not give you) and, for
2444/// row-iterable companions, `companion.into_rows()`. `Vec<Row>: Into<companion>`
2445/// and the companion's `IntoIterator` are the equivalent `std` verbs.
2446/// - **Enum split representation**: `rows.into_dataframe_split()` returns one
2447/// `data.frame` per variant as an R list (only that variant's columns — no NA
2448/// fill), from the `IntoDataFrameSplit` trait (in the prelude). Enum rows
2449/// only; struct derives don't partition.
2450///
2451/// The generated `<Row>DataFrame` / `<Row>DataFrameIter` types are intermediate
2452/// column-oriented companions; you rarely name them directly.
2453#[proc_macro_derive(DataFrameRow, attributes(dataframe))]
2454pub fn derive_dataframe_row(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2455 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2456 dataframe_derive::derive_dataframe_row(input)
2457 .unwrap_or_else(|e| e.into_compile_error())
2458 .into()
2459}
2460
2461/// Derive `RFactor`: enables conversion between Rust enums and R factors.
2462///
2463/// # Usage
2464///
2465/// ```ignore
2466/// #[derive(Copy, Clone, RFactor)]
2467/// enum Color {
2468/// Red,
2469/// Green,
2470/// Blue,
2471/// }
2472/// ```
2473///
2474/// # Attributes
2475///
2476/// - `#[r_factor(rename = "name")]` - Rename a variant's level string
2477/// - `#[r_factor(rename_all = "snake_case")]` - Rename all variants (snake_case, kebab-case, lower, upper)
2478#[proc_macro_derive(RFactor, attributes(r_factor))]
2479pub fn derive_r_factor(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2480 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2481 factor_derive::derive_r_factor(input)
2482 .unwrap_or_else(|e| e.into_compile_error())
2483 .into()
2484}
2485
2486/// Derive `MatchArg`: enables conversion between Rust enums and R character strings
2487/// with `match.arg` semantics (partial matching, informative errors).
2488///
2489/// # Usage
2490///
2491/// ```ignore
2492/// #[derive(Copy, Clone, MatchArg)]
2493/// enum Mode {
2494/// Fast,
2495/// Safe,
2496/// Debug,
2497/// }
2498/// ```
2499///
2500/// # Attributes
2501///
2502/// - `#[match_arg(rename = "name")]` - Rename a variant's choice string
2503/// - `#[match_arg(rename_all = "snake_case")]` - Rename all variants (snake_case, kebab-case, lower, upper)
2504///
2505/// # Generated Implementations
2506///
2507/// - `MatchArg` - Choice metadata and bidirectional conversion
2508/// - `TryFromSexp` - Convert R STRSXP/factor to enum (with partial matching)
2509/// - `IntoR` - Convert enum to R character scalar
2510#[proc_macro_derive(MatchArg, attributes(match_arg))]
2511pub fn derive_match_arg(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2512 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2513 match_arg_derive::derive_match_arg(input)
2514 .unwrap_or_else(|e| e.into_compile_error())
2515 .into()
2516}
2517
2518/// Derive `TryFromSexp` for a single-field newtype: forward the R → Rust
2519/// conversion to the inner type.
2520///
2521/// Generates a scalar `TryFromSexp` impl that delegates to the inner type (so the
2522/// newtype inherits its exact SEXPTYPE checks, NA policy, and error text), plus a
2523/// `FromRNewtype` marker impl. The marker lets `miniextendr-api`'s container
2524/// blankets light up `Vec<T>` / `Option<T>` / `Vec<Option<T>>` automatically.
2525///
2526/// # Usage
2527///
2528/// ```ignore
2529/// use uuid::Uuid;
2530///
2531/// #[derive(TryFromSexp)] // R -> Rust only
2532/// struct Pattern(regex::Regex);
2533///
2534/// #[derive(TryFromSexp, IntoR)] // round-trip; Vec/Option containers work too
2535/// struct UserId(Uuid);
2536/// ```
2537///
2538/// Direction is chosen by which derive you list — derive only `TryFromSexp` for
2539/// inner types that read from R but cannot be written back (e.g. `regex::Regex`).
2540#[proc_macro_derive(TryFromSexp)]
2541pub fn derive_try_from_sexp(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2542 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2543 newtype_derive::derive_try_from_sexp(input)
2544 .unwrap_or_else(|e| e.into_compile_error())
2545 .into()
2546}
2547
2548/// Derive `IntoR` for a single-field newtype: forward the Rust → R conversion to
2549/// the inner type.
2550///
2551/// Generates a scalar `IntoR` impl that delegates to the inner type, plus an
2552/// `IntoRNewtype` marker (powering the `Option<T>` / `Vec<Option<T>>` container
2553/// blankets) and a concrete `IntoRVecElement` impl (powering `Vec<T>`). See
2554/// `#[derive(TryFromSexp)]` for usage.
2555///
2556/// Do not derive both `IntoR` and `MatchArg` on the same type: both feed the
2557/// single `IntoR for Vec<T>` blanket slot and would collide (E0119).
2558#[proc_macro_derive(IntoR)]
2559pub fn derive_into_r(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2560 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2561 newtype_derive::derive_into_r(input)
2562 .unwrap_or_else(|e| e.into_compile_error())
2563 .into()
2564}
2565
2566/// Derive `Vctrs`: enables creating vctrs-compatible S3 vector classes from Rust structs.
2567///
2568/// # Usage
2569///
2570/// ```ignore
2571/// #[derive(Vctrs)]
2572/// #[vctrs(class = "percent", base = "double")]
2573/// pub struct Percent {
2574/// data: Vec<f64>,
2575/// }
2576/// ```
2577///
2578/// # Attributes
2579///
2580/// - `#[vctrs(class = "name")]` - R class name (required)
2581/// - `#[vctrs(base = "type")]` - Base type: double, integer, logical, character, raw, list, record
2582/// - `#[vctrs(abbr = "abbr")]` - Abbreviation for `vec_ptype_abbr`
2583/// - `#[vctrs(inherit_base = true|false)]` - Whether to include base type in class vector
2584///
2585/// # Generated Implementations
2586///
2587/// - `VctrsClass` - Metadata trait for vctrs class information
2588/// - `VctrsRecord` (for `base = "record"`) - Field names for record types
2589#[cfg(feature = "vctrs")]
2590#[proc_macro_derive(Vctrs, attributes(vctrs))]
2591pub fn derive_vctrs(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2592 let input = syn::parse_macro_input!(input as syn::DeriveInput);
2593 vctrs_derive::derive_vctrs(input)
2594 .unwrap_or_else(|e| e.into_compile_error())
2595 .into()
2596}
2597
2598/// Create a `TypedListSpec` for validating `...` arguments or lists.
2599///
2600/// This macro provides ergonomic syntax for defining typed list specifications
2601/// that can be used with `Dots::typed()` to validate the structure of
2602/// `...` arguments passed from R.
2603///
2604/// # Syntax
2605///
2606/// ```text
2607/// typed_list!(
2608/// name => type_spec, // required field with type
2609/// name? => type_spec, // optional field with type
2610/// name, // required field, any type
2611/// name?, // optional field, any type
2612/// )
2613/// ```
2614///
2615/// For strict mode (no extra fields allowed):
2616/// ```text
2617/// typed_list!(@exact; name => type_spec, ...)
2618/// ```
2619///
2620/// # Type Specifications
2621///
2622/// ## Base types (with optional length)
2623/// - `numeric()` / `numeric(4)` - Real/double vector
2624/// - `integer()` / `integer(4)` - Integer vector
2625/// - `logical()` / `logical(4)` - Logical vector
2626/// - `character()` / `character(4)` - Character vector
2627/// - `raw()` / `raw(4)` - Raw vector
2628/// - `complex()` / `complex(4)` - Complex vector
2629/// - `list()` / `list(4)` - List (VECSXP)
2630///
2631/// ## Special types
2632/// - `data_frame()` - Data frame
2633/// - `factor()` - Factor
2634/// - `matrix()` - Matrix
2635/// - `array()` - Array
2636/// - `function()` - Function
2637/// - `environment()` - Environment
2638/// - `null()` - NULL only
2639/// - `any()` - Any type
2640///
2641/// ## String literals
2642/// - `"numeric"`, `"integer"`, etc. - Same as call syntax
2643/// - `"data.frame"` - Data frame (alias)
2644/// - `"MyClass"` - Any other string is treated as a class name (uses `Rf_inherits`)
2645///
2646/// # Examples
2647///
2648/// ## Basic usage
2649///
2650/// ```ignore
2651/// use miniextendr_api::{miniextendr, typed_list, Dots};
2652///
2653/// #[miniextendr]
2654/// pub fn process_args(dots: ...) -> Result<i32, String> {
2655/// let args = dots.typed(typed_list!(
2656/// alpha => numeric(4),
2657/// beta => list(),
2658/// gamma? => "character",
2659/// )).map_err(|e| e.to_string())?;
2660///
2661/// let alpha: Vec<f64> = args.get("alpha").map_err(|e| e.to_string())?;
2662/// Ok(alpha.len() as i32)
2663/// }
2664/// ```
2665///
2666/// ## Strict mode
2667///
2668/// ```ignore
2669/// // Reject any extra named fields
2670/// let args = dots.typed(typed_list!(@exact;
2671/// x => numeric(),
2672/// y => numeric(),
2673/// ))?;
2674/// ```
2675///
2676/// ## Class checking
2677///
2678/// ```ignore
2679/// // Check for specific R class (uses Rf_inherits semantics)
2680/// let args = dots.typed(typed_list!(
2681/// data => "data.frame",
2682/// model => "lm",
2683/// ))?;
2684/// ```
2685///
2686/// ## Attribute sugar
2687///
2688/// Instead of calling `.typed()` manually, you can use `typed_list!` directly in the
2689/// `#[miniextendr]` attribute for automatic validation:
2690///
2691/// ```ignore
2692/// #[miniextendr(dots = typed_list!(x => numeric(), y => numeric()))]
2693/// pub fn my_func(...) -> String {
2694/// // `dots_typed` is automatically created and validated
2695/// let x: f64 = dots_typed.get("x").expect("x");
2696/// let y: f64 = dots_typed.get("y").expect("y");
2697/// format!("x={}, y={}", x, y)
2698/// }
2699/// ```
2700///
2701/// This injects validation at the start of the function body:
2702/// ```ignore
2703/// let dots_typed = _dots.typed(typed_list!(...))
2704/// .unwrap_or_else(|e| panic!("dots validation failed: {e}"));
2705/// ```
2706///
2707/// See the [`#[miniextendr]`](macro@miniextendr) attribute documentation for more details.
2708///
2709#[proc_macro]
2710pub fn typed_list(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2711 let parsed = syn::parse_macro_input!(input as typed_list::TypedListInput);
2712 typed_list::expand_typed_list(parsed).into()
2713}
2714
2715/// Define a compile-time-validated wrapper for an R `data.frame` input.
2716///
2717/// `typed_dataframe!` mirrors [`typed_list!`] for the data.frame shape:
2718/// declare the columns once, get a struct that implements `TryFromSexp`
2719/// (validating both the `data.frame` class and per-column SEXPTYPE) plus
2720/// per-column borrowed accessors that return `&[T]`.
2721///
2722/// # Syntax
2723///
2724/// ```ignore
2725/// typed_dataframe! {
2726/// /// The shape we accept for the Theoph PK dataset.
2727/// pub TheophDf {
2728/// subject: i32,
2729/// weight: f64,
2730/// dose: f64,
2731/// flag: Option<i32>, // optional column
2732/// }
2733/// }
2734/// ```
2735///
2736/// For strict mode (reject any column not declared):
2737/// ```ignore
2738/// typed_dataframe! {
2739/// @exact;
2740/// pub Strict { x: i32 }
2741/// }
2742/// ```
2743///
2744/// # Supported element types
2745///
2746/// v1 supports column element types that implement
2747/// `miniextendr_api::RNativeType`:
2748///
2749/// - `i32` — `INTSXP`
2750/// - `f64` — `REALSXP`
2751/// - `u8` — `RAWSXP`
2752/// - `miniextendr_api::RLogical` — `LGLSXP`
2753/// - `miniextendr_api::Rcomplex` — `CPLXSXP`
2754///
2755/// `String`/`&str` column types are not yet supported (character vectors
2756/// don't expose a contiguous slice). `bool` is also not yet supported as
2757/// a direct field type — use `RLogical` and convert per-element, or
2758/// follow the open follow-up issues from PR #698.
2759///
2760/// # Generated API
2761///
2762/// For each `name: T` column the macro emits:
2763/// - `pub fn name(&self) -> &[T]` (required)
2764/// - `pub fn name(&self) -> Option<&[T]>` (optional, `Option<T>`)
2765///
2766/// Plus housekeeping:
2767/// - `pub fn nrow(&self) -> usize`
2768/// - `pub fn ncol(&self) -> usize` (count of *declared* columns)
2769/// - `pub fn as_sexp(&self) -> SEXP`
2770///
2771/// All borrowed accessors are bound to `&self`; the SEXP is protected
2772/// by the surrounding `#[miniextendr]` call wrapper while the struct is
2773/// alive.
2774///
2775/// # Error reporting
2776///
2777/// `TryFromSexp::try_from_sexp` batches every per-column error into a
2778/// single `SexpError::InvalidValue`, so the R user sees one diagnostic
2779/// covering all missing or wrong-typed columns rather than a sequence of
2780/// stop-on-first-failure messages.
2781///
2782/// # Example
2783///
2784/// ```ignore
2785/// use miniextendr_api::{miniextendr, typed_dataframe};
2786///
2787/// typed_dataframe! {
2788/// pub TheophDf {
2789/// subject: i32,
2790/// weight: f64,
2791/// dose: f64,
2792/// }
2793/// }
2794///
2795/// #[miniextendr]
2796/// pub fn theoph_nrow(df: TheophDf) -> i32 {
2797/// // df.subject() -> &[i32], df.weight() -> &[f64]
2798/// // Lengths are guaranteed equal across columns (data.frame invariant).
2799/// df.nrow() as i32
2800/// }
2801/// ```
2802///
2803/// [`typed_list!`]: macro@typed_list
2804#[proc_macro]
2805pub fn typed_dataframe(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2806 let parsed = syn::parse_macro_input!(input as typed_dataframe::TypedDataframeInput);
2807 typed_dataframe::expand_typed_dataframe(parsed).into()
2808}
2809
2810/// Construct an R list from Rust values.
2811///
2812/// This macro provides a convenient way to create R lists in Rust code,
2813/// using R-like syntax. Values are converted to R objects via the [`IntoR`] trait.
2814///
2815/// # Syntax
2816///
2817/// ```ignore
2818/// // Named entries (like R's list())
2819/// list!(
2820/// alpha = 1,
2821/// beta = "hello",
2822/// "my-name" = vec![1, 2, 3],
2823/// )
2824///
2825/// // Unnamed entries
2826/// list!(1, "hello", vec![1, 2, 3])
2827///
2828/// // Mixed (unnamed entries get empty string names)
2829/// list!(alpha = 1, 2, beta = "hello")
2830///
2831/// // Empty list
2832/// list!()
2833/// ```
2834///
2835/// # Examples
2836///
2837/// ```ignore
2838/// use miniextendr_api::{list, IntoR};
2839///
2840/// // Create a named list
2841/// let my_list = list!(
2842/// x = 42,
2843/// y = "hello world",
2844/// z = vec![1.0, 2.0, 3.0],
2845/// );
2846///
2847/// // In R this is equivalent to:
2848/// // list(x = 42L, y = "hello world", z = c(1, 2, 3))
2849/// ```
2850///
2851/// [`IntoR`]: https://docs.rs/miniextendr-api/latest/miniextendr_api/into_r/trait.IntoR.html
2852#[proc_macro]
2853pub fn list(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2854 let parsed = syn::parse_macro_input!(input as list_macro::ListInput);
2855 list_macro::expand_list(parsed).into()
2856}
2857
2858/// Evaluate R code written as **Rust tokens**, validated at compile time.
2859///
2860/// `r!` takes a single R expression as a token stream, `stringify!`s it into a
2861/// static R source string at build time, and evaluates it via
2862/// `miniextendr_api::expression::r_eval_str` (the same protect-safe parse + eval
2863/// path as `r_str!`).
2864///
2865/// # What you get today
2866///
2867/// Because the argument is a Rust token tree, the Rust front-end already
2868/// rejects **unbalanced delimiters** (`r!(f(1, 2)` won't compile) and
2869/// lexically invalid tokens before R ever sees the string — a cheap
2870/// compile-time guard over the pure-runtime `r_str!`. The source is lowered to
2871/// a `&'static str` (`stringify!`), so there is no `format!` allocation at the
2872/// call site.
2873///
2874/// This proc-macro additionally validates a conservative subset of known-bad
2875/// R syntax constructs (trailing binary operators, consecutive non-unary
2876/// binary operators, bare `if`/`while`/`for` without a body, etc.) and emits
2877/// a precise compile error pointing at the offending token. Empty (missing)
2878/// call arguments — `f(, x)`, `matrix(, 2, 2)` — are valid R and pass.
2879///
2880/// # What is deferred
2881///
2882/// Direct `Rf_lang*` call-tree lowering (skipping the runtime parser entirely)
2883/// is tracked as a follow-up in issue #938 (item 2). Until then `r!` parses
2884/// its static string at first evaluation, exactly like `r_str!`.
2885///
2886/// # Non-goals
2887///
2888/// A complete R grammar validator is not achievable over Rust tokens:
2889/// - Single-quoted strings (`'hello'`) and backtick-quoted names (`` `foo` ``)
2890/// already die at the Rust lexer — nothing to validate.
2891/// - `%op%` tokenises as `%`, ident, `%` and is accepted without analysis.
2892/// - Anything the validator cannot confidently classify as wrong passes through
2893/// unvalidated (conservative reject-only-known-bad design).
2894///
2895/// # Forms
2896///
2897/// - `r!(R tokens…)` — evaluate in `R_GlobalEnv`.
2898/// - `r!(env: e; R tokens…)` — evaluate in the environment SEXP `e`. The
2899/// leading `env: <expr> ;` is consumed as Rust, the rest is R source.
2900///
2901/// Both evaluate to `Result<SEXP, String>`; the `SEXP` is **unprotected**.
2902///
2903/// # Safety
2904///
2905/// Expands to an `unsafe` block; the underlying FFI is `#[r_ffi_checked]`, so
2906/// calls from a worker thread are serialized onto the R thread.
2907///
2908/// # Example
2909///
2910/// ```ignore
2911/// let three = r!(1L + 2L)?;
2912/// let rows = r!(getFromNamespace(".theoph_rows", "dataframeflows")())?;
2913/// let in_env = r!(env: my_env; x + 1)?;
2914/// ```
2915#[proc_macro]
2916pub fn r(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2917 r_macro::expand(input)
2918}
2919
2920/// Internal proc macro used by TPIE (Trait-Provided Impl Expansion).
2921///
2922/// Called by `__mx_impl_<Trait>!` macro_rules macros generated by `#[miniextendr]` on traits.
2923/// Do not call directly.
2924#[proc_macro]
2925#[doc(hidden)]
2926pub fn __mx_trait_impl_expand(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2927 miniextendr_impl_trait::expand_tpie(input)
2928}
2929
2930/// Generate `TypedExternal` and `IntoExternalPtr` impls for a concrete monomorphization
2931/// of a generic type.
2932///
2933/// Since `#[derive(ExternalPtr)]` rejects generic types, use this macro to generate
2934/// the necessary impls for a specific type instantiation.
2935///
2936/// # Example
2937///
2938/// ```ignore
2939/// struct Wrapper<T> { inner: T }
2940///
2941/// impl_typed_external!(Wrapper<i32>);
2942/// impl_typed_external!(Wrapper<String>);
2943/// ```
2944#[proc_macro]
2945pub fn impl_typed_external(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2946 match typed_external_macro::impl_typed_external(input.into()) {
2947 Ok(tokens) => tokens.into(),
2948 Err(err) => err.into_compile_error().into(),
2949 }
2950}
2951
2952/// Generate the `R_init_*` entry point for a miniextendr R package.
2953///
2954/// This macro consolidates all package initialization into a single line.
2955/// It generates an `extern "C-unwind"` function that R calls when loading
2956/// the shared library.
2957///
2958/// # Usage
2959///
2960/// ```ignore
2961/// // Auto-detects package name from CARGO_CRATE_NAME (recommended):
2962/// miniextendr_api::miniextendr_init!();
2963///
2964/// // Or specify explicitly (for edge cases):
2965/// miniextendr_api::miniextendr_init!(mypkg);
2966/// ```
2967///
2968/// The generated function calls `miniextendr_api::init::package_init` which
2969/// handles panic hooks, runtime init, locale assertion, ALTREP setup, trait ABI
2970/// registration, routine registration, and symbol locking.
2971#[proc_macro]
2972pub fn miniextendr_init(input: proc_macro::TokenStream) -> proc_macro::TokenStream {
2973 let pkg_name: syn::Ident = if input.is_empty() {
2974 // Auto-detect from CARGO_CRATE_NAME (set by cargo during compilation).
2975 // Cargo normalizes hyphens → underscores, so this is almost always a
2976 // valid Rust/C identifier. Still parse through syn so malformed values
2977 // surface as a compile error rather than an ICE.
2978 let name = match std::env::var("CARGO_CRATE_NAME") {
2979 Ok(n) => n,
2980 Err(_) => {
2981 return syn::Error::new(
2982 proc_macro2::Span::call_site(),
2983 "CARGO_CRATE_NAME not set. Either pass the package name explicitly: \
2984 miniextendr_init!(mypkg), or ensure you're building with cargo.",
2985 )
2986 .into_compile_error()
2987 .into();
2988 }
2989 };
2990 match syn::parse_str::<syn::Ident>(&name) {
2991 Ok(id) => id,
2992 Err(_) => {
2993 return syn::Error::new(
2994 proc_macro2::Span::call_site(),
2995 format!(
2996 "CARGO_CRATE_NAME `{name}` is not a valid C identifier; \
2997 R_init_<pkg> must match `[A-Za-z_][A-Za-z0-9_]*`. \
2998 Pass the name explicitly: miniextendr_init!(my_pkg)."
2999 ),
3000 )
3001 .into_compile_error()
3002 .into();
3003 }
3004 }
3005 } else {
3006 syn::parse_macro_input!(input as syn::Ident)
3007 };
3008 let fn_name = syn::Ident::new(&format!("R_init_{}", pkg_name), pkg_name.span());
3009 let unload_name = syn::Ident::new(&format!("R_unload_{}", pkg_name), pkg_name.span());
3010
3011 // Build a byte string literal with NUL terminator for the package name.
3012 let mut name_bytes = pkg_name.to_string().into_bytes();
3013 name_bytes.push(0);
3014 let byte_lit = syn::LitByteStr::new(&name_bytes, pkg_name.span());
3015
3016 let expanded = quote::quote! {
3017 // wasm32: pull in the host-generated `wasm_registry.rs` snapshot
3018 // (committed by the user crate, regenerated by `just wasm-prepare`).
3019 // The path is relative to the file invoking `miniextendr_init!()` —
3020 // by convention `<crate>/src/rust/lib.rs`, so the snapshot sits at
3021 // `<crate>/src/rust/wasm_registry.rs`. Module is `#[doc(hidden)]`
3022 // because it's purely an internal bridge between the user crate's
3023 // wrapper / vtable / register-fn `#[no_mangle]` exports and
3024 // `miniextendr_api::registry::install_wasm_runtime_slices`.
3025 #[cfg(target_arch = "wasm32")]
3026 #[path = "wasm_registry.rs"]
3027 #[doc(hidden)]
3028 mod __miniextendr_wasm_registry;
3029
3030 #[unsafe(no_mangle)]
3031 pub unsafe extern "C-unwind" fn #fn_name(
3032 dll: *mut ::miniextendr_api::sys::DllInfo,
3033 ) {
3034 // wasm32: install the pre-generated runtime tables before
3035 // package_init runs. linkme didn't gather anything (the slices
3036 // are OnceLock-backed on wasm32), so register_routines /
3037 // universal_query would otherwise see empty slices.
3038 #[cfg(target_arch = "wasm32")]
3039 ::miniextendr_api::registry::install_wasm_runtime_slices(
3040 __miniextendr_wasm_registry::MX_CALL_DEFS_WASM,
3041 __miniextendr_wasm_registry::MX_ALTREP_REGISTRATIONS_WASM,
3042 __miniextendr_wasm_registry::MX_TRAIT_DISPATCH_WASM,
3043 );
3044
3045 unsafe {
3046 // SAFETY: byte literal is a valid NUL-terminated string produced by the macro.
3047 let pkg_name = ::std::ffi::CStr::from_bytes_with_nul_unchecked(#byte_lit);
3048 ::miniextendr_api::init::package_init(dll, pkg_name);
3049 }
3050 }
3051
3052 /// R_unload_<pkg> entry point — R calls this on `detach(unload=TRUE)` /
3053 /// `dyn.unload()`. Signals the miniextendr worker thread (if enabled)
3054 /// to exit cleanly. See `#103`.
3055 #[unsafe(no_mangle)]
3056 pub unsafe extern "C-unwind" fn #unload_name(
3057 _dll: *mut ::miniextendr_api::sys::DllInfo,
3058 ) {
3059 ::miniextendr_api::worker::miniextendr_runtime_shutdown();
3060 }
3061
3062 /// Linker anchor: stub.c takes the address of this symbol to force the
3063 /// linker to pull in the user crate's archive member from the staticlib.
3064 /// With codegen-units = 1, this single member contains all linkme
3065 /// distributed_slice entries. The name is package-independent so stub.c
3066 /// doesn't need configure substitution.
3067 ///
3068 /// Defined as a function rather than a static so it stays exported under
3069 /// the webR wasm RUSTFLAG -Zdefault-visibility=hidden, which keeps
3070 /// no_mangle functions exported (like the R_init entry point) but hides
3071 /// no_mangle statics. A hidden anchor breaks wasm side-module dlopen
3072 /// (bad export type, undefined). See miniextendr webR notes (#494).
3073 #[unsafe(no_mangle)]
3074 pub extern "C" fn miniextendr_force_link() {}
3075 };
3076
3077 expanded.into()
3078}
3079
3080#[cfg(test)]
3081mod tests;