From 6cb0746f8223cc674f1a4496f80b21d7cf231799 Mon Sep 17 00:00:00 2001 From: Jonas Schievink Date: Mon, 18 Jul 2022 14:31:21 +0200 Subject: [PATCH] Support the 1.64 nightly proc macro ABI --- .../proc-macro-srv/src/abis/abi_1_64/mod.rs | 104 ++ .../abis/abi_1_64/proc_macro/bridge/buffer.rs | 156 +++ .../abis/abi_1_64/proc_macro/bridge/client.rs | 529 ++++++++ .../abi_1_64/proc_macro/bridge/closure.rs | 32 + .../abis/abi_1_64/proc_macro/bridge/handle.rs | 89 ++ .../abis/abi_1_64/proc_macro/bridge/mod.rs | 493 ++++++++ .../abis/abi_1_64/proc_macro/bridge/rpc.rs | 304 +++++ .../abi_1_64/proc_macro/bridge/scoped_cell.rs | 81 ++ .../proc_macro/bridge/selfless_reify.rs | 84 ++ .../abis/abi_1_64/proc_macro/bridge/server.rs | 339 +++++ .../abis/abi_1_64/proc_macro/diagnostic.rs | 166 +++ .../src/abis/abi_1_64/proc_macro/mod.rs | 1125 +++++++++++++++++ .../src/abis/abi_1_64/proc_macro/quote.rs | 139 ++ .../src/abis/abi_1_64/rustc_server.rs | 792 ++++++++++++ crates/proc-macro-srv/src/abis/mod.rs | 11 +- 15 files changed, 4443 insertions(+), 1 deletion(-) create mode 100644 crates/proc-macro-srv/src/abis/abi_1_64/mod.rs create mode 100644 crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/buffer.rs create mode 100644 crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/client.rs create mode 100644 crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/closure.rs create mode 100644 crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/handle.rs create mode 100644 crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/mod.rs create mode 100644 crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/rpc.rs create mode 100644 crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/scoped_cell.rs create mode 100644 crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/selfless_reify.rs create mode 100644 crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/server.rs create mode 100644 crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/diagnostic.rs create mode 100644 crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/mod.rs create mode 100644 crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/quote.rs create mode 100644 crates/proc-macro-srv/src/abis/abi_1_64/rustc_server.rs diff --git a/crates/proc-macro-srv/src/abis/abi_1_64/mod.rs b/crates/proc-macro-srv/src/abis/abi_1_64/mod.rs new file mode 100644 index 00000000000..8a365223a30 --- /dev/null +++ b/crates/proc-macro-srv/src/abis/abi_1_64/mod.rs @@ -0,0 +1,104 @@ +//! Macro ABI for version 1.63 of rustc + +#[allow(dead_code)] +#[doc(hidden)] +mod proc_macro; + +#[allow(dead_code)] +#[doc(hidden)] +mod rustc_server; + +use libloading::Library; +use proc_macro_api::ProcMacroKind; + +use super::PanicMessage; + +pub(crate) struct Abi { + exported_macros: Vec, +} + +impl From for PanicMessage { + fn from(p: proc_macro::bridge::PanicMessage) -> Self { + Self { message: p.as_str().map(|s| s.to_string()) } + } +} + +impl Abi { + pub unsafe fn from_lib(lib: &Library, symbol_name: String) -> Result { + let macros: libloading::Symbol<&&[proc_macro::bridge::client::ProcMacro]> = + lib.get(symbol_name.as_bytes())?; + Ok(Self { exported_macros: macros.to_vec() }) + } + + pub fn expand( + &self, + macro_name: &str, + macro_body: &tt::Subtree, + attributes: Option<&tt::Subtree>, + ) -> Result { + let parsed_body = rustc_server::TokenStream::with_subtree(macro_body.clone()); + + let parsed_attributes = attributes.map_or(rustc_server::TokenStream::new(), |attr| { + rustc_server::TokenStream::with_subtree(attr.clone()) + }); + + for proc_macro in &self.exported_macros { + match proc_macro { + proc_macro::bridge::client::ProcMacro::CustomDerive { + trait_name, client, .. + } if *trait_name == macro_name => { + let res = client.run( + &proc_macro::bridge::server::SameThread, + rustc_server::Rustc::default(), + parsed_body, + true, + ); + return res.map(|it| it.into_subtree()).map_err(PanicMessage::from); + } + proc_macro::bridge::client::ProcMacro::Bang { name, client } + if *name == macro_name => + { + let res = client.run( + &proc_macro::bridge::server::SameThread, + rustc_server::Rustc::default(), + parsed_body, + true, + ); + return res.map(|it| it.into_subtree()).map_err(PanicMessage::from); + } + proc_macro::bridge::client::ProcMacro::Attr { name, client } + if *name == macro_name => + { + let res = client.run( + &proc_macro::bridge::server::SameThread, + rustc_server::Rustc::default(), + parsed_attributes, + parsed_body, + true, + ); + return res.map(|it| it.into_subtree()).map_err(PanicMessage::from); + } + _ => continue, + } + } + + Err(proc_macro::bridge::PanicMessage::String("Nothing to expand".to_string()).into()) + } + + pub fn list_macros(&self) -> Vec<(String, ProcMacroKind)> { + self.exported_macros + .iter() + .map(|proc_macro| match proc_macro { + proc_macro::bridge::client::ProcMacro::CustomDerive { trait_name, .. } => { + (trait_name.to_string(), ProcMacroKind::CustomDerive) + } + proc_macro::bridge::client::ProcMacro::Bang { name, .. } => { + (name.to_string(), ProcMacroKind::FuncLike) + } + proc_macro::bridge::client::ProcMacro::Attr { name, .. } => { + (name.to_string(), ProcMacroKind::Attr) + } + }) + .collect() + } +} diff --git a/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/buffer.rs b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/buffer.rs new file mode 100644 index 00000000000..48030f8d82d --- /dev/null +++ b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/buffer.rs @@ -0,0 +1,156 @@ +//! Buffer management for same-process client<->server communication. + +use std::io::{self, Write}; +use std::mem; +use std::ops::{Deref, DerefMut}; +use std::slice; + +#[repr(C)] +pub struct Buffer { + data: *mut u8, + len: usize, + capacity: usize, + reserve: extern "C" fn(Buffer, usize) -> Buffer, + drop: extern "C" fn(Buffer), +} + +unsafe impl Sync for Buffer {} +unsafe impl Send for Buffer {} + +impl Default for Buffer { + #[inline] + fn default() -> Self { + Self::from(vec![]) + } +} + +impl Deref for Buffer { + type Target = [u8]; + #[inline] + fn deref(&self) -> &[u8] { + unsafe { slice::from_raw_parts(self.data as *const u8, self.len) } + } +} + +impl DerefMut for Buffer { + #[inline] + fn deref_mut(&mut self) -> &mut [u8] { + unsafe { slice::from_raw_parts_mut(self.data, self.len) } + } +} + +impl Buffer { + #[inline] + pub(super) fn new() -> Self { + Self::default() + } + + #[inline] + pub(super) fn clear(&mut self) { + self.len = 0; + } + + #[inline] + pub(super) fn take(&mut self) -> Self { + mem::take(self) + } + + // We have the array method separate from extending from a slice. This is + // because in the case of small arrays, codegen can be more efficient + // (avoiding a memmove call). With extend_from_slice, LLVM at least + // currently is not able to make that optimization. + #[inline] + pub(super) fn extend_from_array(&mut self, xs: &[u8; N]) { + if xs.len() > (self.capacity - self.len) { + let b = self.take(); + *self = (b.reserve)(b, xs.len()); + } + unsafe { + xs.as_ptr().copy_to_nonoverlapping(self.data.add(self.len), xs.len()); + self.len += xs.len(); + } + } + + #[inline] + pub(super) fn extend_from_slice(&mut self, xs: &[u8]) { + if xs.len() > (self.capacity - self.len) { + let b = self.take(); + *self = (b.reserve)(b, xs.len()); + } + unsafe { + xs.as_ptr().copy_to_nonoverlapping(self.data.add(self.len), xs.len()); + self.len += xs.len(); + } + } + + #[inline] + pub(super) fn push(&mut self, v: u8) { + // The code here is taken from Vec::push, and we know that reserve() + // will panic if we're exceeding isize::MAX bytes and so there's no need + // to check for overflow. + if self.len == self.capacity { + let b = self.take(); + *self = (b.reserve)(b, 1); + } + unsafe { + *self.data.add(self.len) = v; + self.len += 1; + } + } +} + +impl Write for Buffer { + #[inline] + fn write(&mut self, xs: &[u8]) -> io::Result { + self.extend_from_slice(xs); + Ok(xs.len()) + } + + #[inline] + fn write_all(&mut self, xs: &[u8]) -> io::Result<()> { + self.extend_from_slice(xs); + Ok(()) + } + + #[inline] + fn flush(&mut self) -> io::Result<()> { + Ok(()) + } +} + +impl Drop for Buffer { + #[inline] + fn drop(&mut self) { + let b = self.take(); + (b.drop)(b); + } +} + +impl From> for Buffer { + fn from(mut v: Vec) -> Self { + let (data, len, capacity) = (v.as_mut_ptr(), v.len(), v.capacity()); + mem::forget(v); + + // This utility function is nested in here because it can *only* + // be safely called on `Buffer`s created by *this* `proc_macro`. + fn to_vec(b: Buffer) -> Vec { + unsafe { + let Buffer { data, len, capacity, .. } = b; + mem::forget(b); + Vec::from_raw_parts(data, len, capacity) + } + } + + extern "C" fn reserve(b: Buffer, additional: usize) -> Buffer { + let mut v = to_vec(b); + v.reserve(additional); + Buffer::from(v) + } + + extern "C" fn drop(b: Buffer) { + mem::drop(to_vec(b)); + } + + Buffer { data, len, capacity, reserve, drop } + } +} diff --git a/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/client.rs b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/client.rs new file mode 100644 index 00000000000..22bda8ba5a7 --- /dev/null +++ b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/client.rs @@ -0,0 +1,529 @@ +//! Client-side types. + +use super::*; + +use std::marker::PhantomData; + +macro_rules! define_handles { + ( + 'owned: $($oty:ident,)* + 'interned: $($ity:ident,)* + ) => { + #[repr(C)] + #[allow(non_snake_case)] + pub struct HandleCounters { + $($oty: AtomicUsize,)* + $($ity: AtomicUsize,)* + } + + impl HandleCounters { + // FIXME(eddyb) use a reference to the `static COUNTERS`, instead of + // a wrapper `fn` pointer, once `const fn` can reference `static`s. + extern "C" fn get() -> &'static Self { + static COUNTERS: HandleCounters = HandleCounters { + $($oty: AtomicUsize::new(1),)* + $($ity: AtomicUsize::new(1),)* + }; + &COUNTERS + } + } + + // FIXME(eddyb) generate the definition of `HandleStore` in `server.rs`. + #[repr(C)] + #[allow(non_snake_case)] + pub(super) struct HandleStore { + $($oty: handle::OwnedStore,)* + $($ity: handle::InternedStore,)* + } + + impl HandleStore { + pub(super) fn new(handle_counters: &'static HandleCounters) -> Self { + HandleStore { + $($oty: handle::OwnedStore::new(&handle_counters.$oty),)* + $($ity: handle::InternedStore::new(&handle_counters.$ity),)* + } + } + } + + $( + #[repr(C)] + pub(crate) struct $oty { + handle: handle::Handle, + // Prevent Send and Sync impls. `!Send`/`!Sync` is the usual + // way of doing this, but that requires unstable features. + // rust-analyzer uses this code and avoids unstable features. + _marker: PhantomData<*mut ()>, + } + + // Forward `Drop::drop` to the inherent `drop` method. + impl Drop for $oty { + fn drop(&mut self) { + $oty { + handle: self.handle, + _marker: PhantomData, + }.drop(); + } + } + + impl Encode for $oty { + fn encode(self, w: &mut Writer, s: &mut S) { + let handle = self.handle; + mem::forget(self); + handle.encode(w, s); + } + } + + impl DecodeMut<'_, '_, HandleStore>> + for Marked + { + fn decode(r: &mut Reader<'_>, s: &mut HandleStore>) -> Self { + s.$oty.take(handle::Handle::decode(r, &mut ())) + } + } + + impl Encode for &$oty { + fn encode(self, w: &mut Writer, s: &mut S) { + self.handle.encode(w, s); + } + } + + impl<'s, S: server::Types> Decode<'_, 's, HandleStore>> + for &'s Marked + { + fn decode(r: &mut Reader<'_>, s: &'s HandleStore>) -> Self { + &s.$oty[handle::Handle::decode(r, &mut ())] + } + } + + impl Encode for &mut $oty { + fn encode(self, w: &mut Writer, s: &mut S) { + self.handle.encode(w, s); + } + } + + impl<'s, S: server::Types> DecodeMut<'_, 's, HandleStore>> + for &'s mut Marked + { + fn decode( + r: &mut Reader<'_>, + s: &'s mut HandleStore> + ) -> Self { + &mut s.$oty[handle::Handle::decode(r, &mut ())] + } + } + + impl Encode>> + for Marked + { + fn encode(self, w: &mut Writer, s: &mut HandleStore>) { + s.$oty.alloc(self).encode(w, s); + } + } + + impl DecodeMut<'_, '_, S> for $oty { + fn decode(r: &mut Reader<'_>, s: &mut S) -> Self { + $oty { + handle: handle::Handle::decode(r, s), + _marker: PhantomData, + } + } + } + )* + + $( + #[repr(C)] + #[derive(Copy, Clone, PartialEq, Eq, Hash)] + pub(crate) struct $ity { + handle: handle::Handle, + // Prevent Send and Sync impls. `!Send`/`!Sync` is the usual + // way of doing this, but that requires unstable features. + // rust-analyzer uses this code and avoids unstable features. + _marker: PhantomData<*mut ()>, + } + + impl Encode for $ity { + fn encode(self, w: &mut Writer, s: &mut S) { + self.handle.encode(w, s); + } + } + + impl DecodeMut<'_, '_, HandleStore>> + for Marked + { + fn decode(r: &mut Reader<'_>, s: &mut HandleStore>) -> Self { + s.$ity.copy(handle::Handle::decode(r, &mut ())) + } + } + + impl Encode>> + for Marked + { + fn encode(self, w: &mut Writer, s: &mut HandleStore>) { + s.$ity.alloc(self).encode(w, s); + } + } + + impl DecodeMut<'_, '_, S> for $ity { + fn decode(r: &mut Reader<'_>, s: &mut S) -> Self { + $ity { + handle: handle::Handle::decode(r, s), + _marker: PhantomData, + } + } + } + )* + } +} +define_handles! { + 'owned: + FreeFunctions, + TokenStream, + Literal, + SourceFile, + MultiSpan, + Diagnostic, + + 'interned: + Ident, + Span, +} + +// FIXME(eddyb) generate these impls by pattern-matching on the +// names of methods - also could use the presence of `fn drop` +// to distinguish between 'owned and 'interned, above. +// Alternatively, special "modes" could be listed of types in with_api +// instead of pattern matching on methods, here and in server decl. + +impl Clone for TokenStream { + fn clone(&self) -> Self { + self.clone() + } +} + +impl Clone for Literal { + fn clone(&self) -> Self { + self.clone() + } +} + +impl fmt::Debug for Literal { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.debug_struct("Literal") + // format the kind without quotes, as in `kind: Float` + .field("kind", &format_args!("{}", &self.debug_kind())) + .field("symbol", &self.symbol()) + // format `Some("...")` on one line even in {:#?} mode + .field("suffix", &format_args!("{:?}", &self.suffix())) + .field("span", &self.span()) + .finish() + } +} + +impl Clone for SourceFile { + fn clone(&self) -> Self { + self.clone() + } +} + +impl Span { + pub(crate) fn def_site() -> Span { + Bridge::with(|bridge| bridge.globals.def_site) + } + + pub(crate) fn call_site() -> Span { + Bridge::with(|bridge| bridge.globals.call_site) + } + + pub(crate) fn mixed_site() -> Span { + Bridge::with(|bridge| bridge.globals.mixed_site) + } +} + +impl fmt::Debug for Span { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.write_str(&self.debug()) + } +} + +macro_rules! define_client_side { + ($($name:ident { + $(fn $method:ident($($arg:ident: $arg_ty:ty),* $(,)?) $(-> $ret_ty:ty)*;)* + }),* $(,)?) => { + $(impl $name { + $(pub(crate) fn $method($($arg: $arg_ty),*) $(-> $ret_ty)* { + Bridge::with(|bridge| { + let mut buf = bridge.cached_buffer.take(); + + buf.clear(); + api_tags::Method::$name(api_tags::$name::$method).encode(&mut buf, &mut ()); + reverse_encode!(buf; $($arg),*); + + buf = bridge.dispatch.call(buf); + + let r = Result::<_, PanicMessage>::decode(&mut &buf[..], &mut ()); + + bridge.cached_buffer = buf; + + r.unwrap_or_else(|e| panic::resume_unwind(e.into())) + }) + })* + })* + } +} +with_api!(self, self, define_client_side); + +struct Bridge<'a> { + /// Reusable buffer (only `clear`-ed, never shrunk), primarily + /// used for making requests. + cached_buffer: Buffer, + + /// Server-side function that the client uses to make requests. + dispatch: closure::Closure<'a, Buffer, Buffer>, + + /// Provided globals for this macro expansion. + globals: ExpnGlobals, +} + +enum BridgeState<'a> { + /// No server is currently connected to this client. + NotConnected, + + /// A server is connected and available for requests. + Connected(Bridge<'a>), + + /// Access to the bridge is being exclusively acquired + /// (e.g., during `BridgeState::with`). + InUse, +} + +enum BridgeStateL {} + +impl<'a> scoped_cell::ApplyL<'a> for BridgeStateL { + type Out = BridgeState<'a>; +} + +thread_local! { + static BRIDGE_STATE: scoped_cell::ScopedCell = + scoped_cell::ScopedCell::new(BridgeState::NotConnected); +} + +impl BridgeState<'_> { + /// Take exclusive control of the thread-local + /// `BridgeState`, and pass it to `f`, mutably. + /// The state will be restored after `f` exits, even + /// by panic, including modifications made to it by `f`. + /// + /// N.B., while `f` is running, the thread-local state + /// is `BridgeState::InUse`. + fn with(f: impl FnOnce(&mut BridgeState<'_>) -> R) -> R { + BRIDGE_STATE.with(|state| { + state.replace(BridgeState::InUse, |mut state| { + // FIXME(#52812) pass `f` directly to `replace` when `RefMutL` is gone + f(&mut *state) + }) + }) + } +} + +impl Bridge<'_> { + fn with(f: impl FnOnce(&mut Bridge<'_>) -> R) -> R { + BridgeState::with(|state| match state { + BridgeState::NotConnected => { + panic!("procedural macro API is used outside of a procedural macro"); + } + BridgeState::InUse => { + panic!("procedural macro API is used while it's already in use"); + } + BridgeState::Connected(bridge) => f(bridge), + }) + } +} + +pub(crate) fn is_available() -> bool { + BridgeState::with(|state| match state { + BridgeState::Connected(_) | BridgeState::InUse => true, + BridgeState::NotConnected => false, + }) +} + +/// A client-side RPC entry-point, which may be using a different `proc_macro` +/// from the one used by the server, but can be invoked compatibly. +/// +/// Note that the (phantom) `I` ("input") and `O` ("output") type parameters +/// decorate the `Client` with the RPC "interface" of the entry-point, but +/// do not themselves participate in ABI, at all, only facilitate type-checking. +/// +/// E.g. `Client` is the common proc macro interface, +/// used for `#[proc_macro] fn foo(input: TokenStream) -> TokenStream`, +/// indicating that the RPC input and output will be serialized token streams, +/// and forcing the use of APIs that take/return `S::TokenStream`, server-side. +#[repr(C)] +pub struct Client { + // FIXME(eddyb) use a reference to the `static COUNTERS`, instead of + // a wrapper `fn` pointer, once `const fn` can reference `static`s. + pub(super) get_handle_counters: extern "C" fn() -> &'static HandleCounters, + + pub(super) run: extern "C" fn(BridgeConfig<'_>) -> Buffer, + + pub(super) _marker: PhantomData O>, +} + +impl Copy for Client {} +impl Clone for Client { + fn clone(&self) -> Self { + *self + } +} + +fn maybe_install_panic_hook(force_show_panics: bool) { + // Hide the default panic output within `proc_macro` expansions. + // NB. the server can't do this because it may use a different libstd. + static HIDE_PANICS_DURING_EXPANSION: Once = Once::new(); + HIDE_PANICS_DURING_EXPANSION.call_once(|| { + let prev = panic::take_hook(); + panic::set_hook(Box::new(move |info| { + let show = BridgeState::with(|state| match state { + BridgeState::NotConnected => true, + BridgeState::Connected(_) | BridgeState::InUse => force_show_panics, + }); + if show { + prev(info) + } + })); + }); +} + +/// Client-side helper for handling client panics, entering the bridge, +/// deserializing input and serializing output. +// FIXME(eddyb) maybe replace `Bridge::enter` with this? +fn run_client DecodeMut<'a, 's, ()>, R: Encode<()>>( + config: BridgeConfig<'_>, + f: impl FnOnce(A) -> R, +) -> Buffer { + let BridgeConfig { input: mut buf, dispatch, force_show_panics, .. } = config; + + panic::catch_unwind(panic::AssertUnwindSafe(|| { + maybe_install_panic_hook(force_show_panics); + + let reader = &mut &buf[..]; + let (globals, input) = <(ExpnGlobals, A)>::decode(reader, &mut ()); + + // Put the buffer we used for input back in the `Bridge` for requests. + let new_state = + BridgeState::Connected(Bridge { cached_buffer: buf.take(), dispatch, globals }); + + BRIDGE_STATE.with(|state| { + state.set(new_state, || { + let output = f(input); + + // Take the `cached_buffer` back out, for the output value. + buf = Bridge::with(|bridge| bridge.cached_buffer.take()); + + // HACK(eddyb) Separate encoding a success value (`Ok(output)`) + // from encoding a panic (`Err(e: PanicMessage)`) to avoid + // having handles outside the `bridge.enter(|| ...)` scope, and + // to catch panics that could happen while encoding the success. + // + // Note that panics should be impossible beyond this point, but + // this is defensively trying to avoid any accidental panicking + // reaching the `extern "C"` (which should `abort` but might not + // at the moment, so this is also potentially preventing UB). + buf.clear(); + Ok::<_, ()>(output).encode(&mut buf, &mut ()); + }) + }) + })) + .map_err(PanicMessage::from) + .unwrap_or_else(|e| { + buf.clear(); + Err::<(), _>(e).encode(&mut buf, &mut ()); + }); + buf +} + +impl Client { + pub const fn expand1( + f: impl Fn(super::super::TokenStream) -> super::super::TokenStream + Copy, + ) -> Self { + Client { + get_handle_counters: HandleCounters::get, + run: super::selfless_reify::reify_to_extern_c_fn_hrt_bridge(move |bridge| { + run_client(bridge, |input| f(super::super::TokenStream(input)).0) + }), + _marker: PhantomData, + } + } +} + +impl Client<(super::super::TokenStream, super::super::TokenStream), super::super::TokenStream> { + pub const fn expand2( + f: impl Fn(super::super::TokenStream, super::super::TokenStream) -> super::super::TokenStream + + Copy, + ) -> Self { + Client { + get_handle_counters: HandleCounters::get, + run: super::selfless_reify::reify_to_extern_c_fn_hrt_bridge(move |bridge| { + run_client(bridge, |(input, input2)| { + f(super::super::TokenStream(input), super::super::TokenStream(input2)).0 + }) + }), + _marker: PhantomData, + } + } +} + +#[repr(C)] +#[derive(Copy, Clone)] +pub enum ProcMacro { + CustomDerive { + trait_name: &'static str, + attributes: &'static [&'static str], + client: Client, + }, + + Attr { + name: &'static str, + client: Client< + (super::super::TokenStream, super::super::TokenStream), + super::super::TokenStream, + >, + }, + + Bang { + name: &'static str, + client: Client, + }, +} + +impl ProcMacro { + pub fn name(&self) -> &'static str { + match self { + ProcMacro::CustomDerive { trait_name, .. } => trait_name, + ProcMacro::Attr { name, .. } => name, + ProcMacro::Bang { name, .. } => name, + } + } + + pub const fn custom_derive( + trait_name: &'static str, + attributes: &'static [&'static str], + expand: impl Fn(super::super::TokenStream) -> super::super::TokenStream + Copy, + ) -> Self { + ProcMacro::CustomDerive { trait_name, attributes, client: Client::expand1(expand) } + } + + pub const fn attr( + name: &'static str, + expand: impl Fn(super::super::TokenStream, super::super::TokenStream) -> super::super::TokenStream + + Copy, + ) -> Self { + ProcMacro::Attr { name, client: Client::expand2(expand) } + } + + pub const fn bang( + name: &'static str, + expand: impl Fn(super::super::TokenStream) -> super::super::TokenStream + Copy, + ) -> Self { + ProcMacro::Bang { name, client: Client::expand1(expand) } + } +} diff --git a/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/closure.rs b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/closure.rs new file mode 100644 index 00000000000..d371ae3cea0 --- /dev/null +++ b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/closure.rs @@ -0,0 +1,32 @@ +//! Closure type (equivalent to `&mut dyn FnMut(A) -> R`) that's `repr(C)`. + +use std::marker::PhantomData; + +#[repr(C)] +pub struct Closure<'a, A, R> { + call: unsafe extern "C" fn(*mut Env, A) -> R, + env: *mut Env, + // Prevent Send and Sync impls. `!Send`/`!Sync` is the usual way of doing + // this, but that requires unstable features. rust-analyzer uses this code + // and avoids unstable features. + // + // The `'a` lifetime parameter represents the lifetime of `Env`. + _marker: PhantomData<*mut &'a mut ()>, +} + +struct Env; + +impl<'a, A, R, F: FnMut(A) -> R> From<&'a mut F> for Closure<'a, A, R> { + fn from(f: &'a mut F) -> Self { + unsafe extern "C" fn call R>(env: *mut Env, arg: A) -> R { + (*(env as *mut _ as *mut F))(arg) + } + Closure { call: call::, env: f as *mut _ as *mut Env, _marker: PhantomData } + } +} + +impl<'a, A, R> Closure<'a, A, R> { + pub fn call(&mut self, arg: A) -> R { + unsafe { (self.call)(self.env, arg) } + } +} diff --git a/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/handle.rs b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/handle.rs new file mode 100644 index 00000000000..c219a9465d3 --- /dev/null +++ b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/handle.rs @@ -0,0 +1,89 @@ +//! Server-side handles and storage for per-handle data. + +use std::collections::{BTreeMap, HashMap}; +use std::hash::{BuildHasher, Hash}; +use std::num::NonZeroU32; +use std::ops::{Index, IndexMut}; +use std::sync::atomic::{AtomicUsize, Ordering}; + +pub(super) type Handle = NonZeroU32; + +/// A store that associates values of type `T` with numeric handles. A value can +/// be looked up using its handle. +pub(super) struct OwnedStore { + counter: &'static AtomicUsize, + data: BTreeMap, +} + +impl OwnedStore { + pub(super) fn new(counter: &'static AtomicUsize) -> Self { + // Ensure the handle counter isn't 0, which would panic later, + // when `NonZeroU32::new` (aka `Handle::new`) is called in `alloc`. + assert_ne!(counter.load(Ordering::SeqCst), 0); + + OwnedStore { counter, data: BTreeMap::new() } + } +} + +impl OwnedStore { + pub(super) fn alloc(&mut self, x: T) -> Handle { + let counter = self.counter.fetch_add(1, Ordering::SeqCst); + let handle = Handle::new(counter as u32).expect("`proc_macro` handle counter overflowed"); + assert!(self.data.insert(handle, x).is_none()); + handle + } + + pub(super) fn take(&mut self, h: Handle) -> T { + self.data.remove(&h).expect("use-after-free in `proc_macro` handle") + } +} + +impl Index for OwnedStore { + type Output = T; + fn index(&self, h: Handle) -> &T { + self.data.get(&h).expect("use-after-free in `proc_macro` handle") + } +} + +impl IndexMut for OwnedStore { + fn index_mut(&mut self, h: Handle) -> &mut T { + self.data.get_mut(&h).expect("use-after-free in `proc_macro` handle") + } +} + +// HACK(eddyb) deterministic `std::collections::hash_map::RandomState` replacement +// that doesn't require adding any dependencies to `proc_macro` (like `rustc-hash`). +#[derive(Clone)] +struct NonRandomState; + +impl BuildHasher for NonRandomState { + type Hasher = std::collections::hash_map::DefaultHasher; + #[inline] + fn build_hasher(&self) -> Self::Hasher { + Self::Hasher::new() + } +} + +/// Like `OwnedStore`, but avoids storing any value more than once. +pub(super) struct InternedStore { + owned: OwnedStore, + interner: HashMap, +} + +impl InternedStore { + pub(super) fn new(counter: &'static AtomicUsize) -> Self { + InternedStore { + owned: OwnedStore::new(counter), + interner: HashMap::with_hasher(NonRandomState), + } + } + + pub(super) fn alloc(&mut self, x: T) -> Handle { + let owned = &mut self.owned; + *self.interner.entry(x).or_insert_with(|| owned.alloc(x)) + } + + pub(super) fn copy(&mut self, h: Handle) -> T { + self.owned[h] + } +} diff --git a/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/mod.rs b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/mod.rs new file mode 100644 index 00000000000..ffd44079323 --- /dev/null +++ b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/mod.rs @@ -0,0 +1,493 @@ +//! Internal interface for communicating between a `proc_macro` client +//! (a proc macro crate) and a `proc_macro` server (a compiler front-end). +//! +//! Serialization (with C ABI buffers) and unique integer handles are employed +//! to allow safely interfacing between two copies of `proc_macro` built +//! (from the same source) by different compilers with potentially mismatching +//! Rust ABIs (e.g., stage0/bin/rustc vs stage1/bin/rustc during bootstrap). + +#![deny(unsafe_code)] + +pub use super::{Delimiter, Level, LineColumn, Spacing}; +use std::fmt; +use std::hash::Hash; +use std::marker; +use std::mem; +use std::ops::Bound; +use std::panic; +use std::sync::atomic::AtomicUsize; +use std::sync::Once; +use std::thread; + +/// Higher-order macro describing the server RPC API, allowing automatic +/// generation of type-safe Rust APIs, both client-side and server-side. +/// +/// `with_api!(MySelf, my_self, my_macro)` expands to: +/// ```rust,ignore (pseudo-code) +/// my_macro! { +/// // ... +/// Literal { +/// // ... +/// fn character(ch: char) -> MySelf::Literal; +/// // ... +/// fn span(my_self: &MySelf::Literal) -> MySelf::Span; +/// fn set_span(my_self: &mut MySelf::Literal, span: MySelf::Span); +/// }, +/// // ... +/// } +/// ``` +/// +/// The first two arguments serve to customize the arguments names +/// and argument/return types, to enable several different usecases: +/// +/// If `my_self` is just `self`, then each `fn` signature can be used +/// as-is for a method. If it's anything else (`self_` in practice), +/// then the signatures don't have a special `self` argument, and +/// can, therefore, have a different one introduced. +/// +/// If `MySelf` is just `Self`, then the types are only valid inside +/// a trait or a trait impl, where the trait has associated types +/// for each of the API types. If non-associated types are desired, +/// a module name (`self` in practice) can be used instead of `Self`. +macro_rules! with_api { + ($S:ident, $self:ident, $m:ident) => { + $m! { + FreeFunctions { + fn drop($self: $S::FreeFunctions); + fn track_env_var(var: &str, value: Option<&str>); + fn track_path(path: &str); + }, + TokenStream { + fn drop($self: $S::TokenStream); + fn clone($self: &$S::TokenStream) -> $S::TokenStream; + fn is_empty($self: &$S::TokenStream) -> bool; + fn expand_expr($self: &$S::TokenStream) -> Result<$S::TokenStream, ()>; + fn from_str(src: &str) -> $S::TokenStream; + fn to_string($self: &$S::TokenStream) -> String; + fn from_token_tree( + tree: TokenTree<$S::TokenStream, $S::Span, $S::Ident, $S::Literal>, + ) -> $S::TokenStream; + fn concat_trees( + base: Option<$S::TokenStream>, + trees: Vec>, + ) -> $S::TokenStream; + fn concat_streams( + base: Option<$S::TokenStream>, + streams: Vec<$S::TokenStream>, + ) -> $S::TokenStream; + fn into_trees( + $self: $S::TokenStream + ) -> Vec>; + }, + Ident { + fn new(string: &str, span: $S::Span, is_raw: bool) -> $S::Ident; + fn span($self: $S::Ident) -> $S::Span; + fn with_span($self: $S::Ident, span: $S::Span) -> $S::Ident; + }, + Literal { + fn drop($self: $S::Literal); + fn clone($self: &$S::Literal) -> $S::Literal; + fn from_str(s: &str) -> Result<$S::Literal, ()>; + fn to_string($self: &$S::Literal) -> String; + fn debug_kind($self: &$S::Literal) -> String; + fn symbol($self: &$S::Literal) -> String; + fn suffix($self: &$S::Literal) -> Option; + fn integer(n: &str) -> $S::Literal; + fn typed_integer(n: &str, kind: &str) -> $S::Literal; + fn float(n: &str) -> $S::Literal; + fn f32(n: &str) -> $S::Literal; + fn f64(n: &str) -> $S::Literal; + fn string(string: &str) -> $S::Literal; + fn character(ch: char) -> $S::Literal; + fn byte_string(bytes: &[u8]) -> $S::Literal; + fn span($self: &$S::Literal) -> $S::Span; + fn set_span($self: &mut $S::Literal, span: $S::Span); + fn subspan( + $self: &$S::Literal, + start: Bound, + end: Bound, + ) -> Option<$S::Span>; + }, + SourceFile { + fn drop($self: $S::SourceFile); + fn clone($self: &$S::SourceFile) -> $S::SourceFile; + fn eq($self: &$S::SourceFile, other: &$S::SourceFile) -> bool; + fn path($self: &$S::SourceFile) -> String; + fn is_real($self: &$S::SourceFile) -> bool; + }, + MultiSpan { + fn drop($self: $S::MultiSpan); + fn new() -> $S::MultiSpan; + fn push($self: &mut $S::MultiSpan, span: $S::Span); + }, + Diagnostic { + fn drop($self: $S::Diagnostic); + fn new(level: Level, msg: &str, span: $S::MultiSpan) -> $S::Diagnostic; + fn sub( + $self: &mut $S::Diagnostic, + level: Level, + msg: &str, + span: $S::MultiSpan, + ); + fn emit($self: $S::Diagnostic); + }, + Span { + fn debug($self: $S::Span) -> String; + fn source_file($self: $S::Span) -> $S::SourceFile; + fn parent($self: $S::Span) -> Option<$S::Span>; + fn source($self: $S::Span) -> $S::Span; + fn start($self: $S::Span) -> LineColumn; + fn end($self: $S::Span) -> LineColumn; + fn before($self: $S::Span) -> $S::Span; + fn after($self: $S::Span) -> $S::Span; + fn join($self: $S::Span, other: $S::Span) -> Option<$S::Span>; + fn resolved_at($self: $S::Span, at: $S::Span) -> $S::Span; + fn source_text($self: $S::Span) -> Option; + fn save_span($self: $S::Span) -> usize; + fn recover_proc_macro_span(id: usize) -> $S::Span; + }, + } + }; +} + +// FIXME(eddyb) this calls `encode` for each argument, but in reverse, +// to match the ordering in `reverse_decode`. +macro_rules! reverse_encode { + ($writer:ident;) => {}; + ($writer:ident; $first:ident $(, $rest:ident)*) => { + reverse_encode!($writer; $($rest),*); + $first.encode(&mut $writer, &mut ()); + } +} + +// FIXME(eddyb) this calls `decode` for each argument, but in reverse, +// to avoid borrow conflicts from borrows started by `&mut` arguments. +macro_rules! reverse_decode { + ($reader:ident, $s:ident;) => {}; + ($reader:ident, $s:ident; $first:ident: $first_ty:ty $(, $rest:ident: $rest_ty:ty)*) => { + reverse_decode!($reader, $s; $($rest: $rest_ty),*); + let $first = <$first_ty>::decode(&mut $reader, $s); + } +} + +#[allow(unsafe_code)] +mod buffer; +#[forbid(unsafe_code)] +pub mod client; +#[allow(unsafe_code)] +mod closure; +#[forbid(unsafe_code)] +mod handle; +#[macro_use] +#[forbid(unsafe_code)] +mod rpc; +#[allow(unsafe_code)] +mod scoped_cell; +#[allow(unsafe_code)] +mod selfless_reify; +#[forbid(unsafe_code)] +pub mod server; + +use buffer::Buffer; +pub use rpc::PanicMessage; +use rpc::{Decode, DecodeMut, Encode, Reader, Writer}; + +/// Configuration for establishing an active connection between a server and a +/// client. The server creates the bridge config (`run_server` in `server.rs`), +/// then passes it to the client through the function pointer in the `run` field +/// of `client::Client`. The client constructs a local `Bridge` from the config +/// in TLS during its execution (`Bridge::{enter, with}` in `client.rs`). +#[repr(C)] +pub struct BridgeConfig<'a> { + /// Buffer used to pass initial input to the client. + input: Buffer, + + /// Server-side function that the client uses to make requests. + dispatch: closure::Closure<'a, Buffer, Buffer>, + + /// If 'true', always invoke the default panic hook + force_show_panics: bool, + + // Prevent Send and Sync impls. `!Send`/`!Sync` is the usual way of doing + // this, but that requires unstable features. rust-analyzer uses this code + // and avoids unstable features. + _marker: marker::PhantomData<*mut ()>, +} + +#[forbid(unsafe_code)] +#[allow(non_camel_case_types)] +mod api_tags { + use super::rpc::{DecodeMut, Encode, Reader, Writer}; + + macro_rules! declare_tags { + ($($name:ident { + $(fn $method:ident($($arg:ident: $arg_ty:ty),* $(,)?) $(-> $ret_ty:ty)*;)* + }),* $(,)?) => { + $( + pub(super) enum $name { + $($method),* + } + rpc_encode_decode!(enum $name { $($method),* }); + )* + + pub(super) enum Method { + $($name($name)),* + } + rpc_encode_decode!(enum Method { $($name(m)),* }); + } + } + with_api!(self, self, declare_tags); +} + +/// Helper to wrap associated types to allow trait impl dispatch. +/// That is, normally a pair of impls for `T::Foo` and `T::Bar` +/// can overlap, but if the impls are, instead, on types like +/// `Marked` and `Marked`, they can't. +trait Mark { + type Unmarked; + fn mark(unmarked: Self::Unmarked) -> Self; +} + +/// Unwrap types wrapped by `Mark::mark` (see `Mark` for details). +trait Unmark { + type Unmarked; + fn unmark(self) -> Self::Unmarked; +} + +#[derive(Copy, Clone, PartialEq, Eq, Hash)] +struct Marked { + value: T, + _marker: marker::PhantomData, +} + +impl Mark for Marked { + type Unmarked = T; + fn mark(unmarked: Self::Unmarked) -> Self { + Marked { value: unmarked, _marker: marker::PhantomData } + } +} +impl Unmark for Marked { + type Unmarked = T; + fn unmark(self) -> Self::Unmarked { + self.value + } +} +impl<'a, T, M> Unmark for &'a Marked { + type Unmarked = &'a T; + fn unmark(self) -> Self::Unmarked { + &self.value + } +} +impl<'a, T, M> Unmark for &'a mut Marked { + type Unmarked = &'a mut T; + fn unmark(self) -> Self::Unmarked { + &mut self.value + } +} + +impl Mark for Vec { + type Unmarked = Vec; + fn mark(unmarked: Self::Unmarked) -> Self { + // Should be a no-op due to std's in-place collect optimizations. + unmarked.into_iter().map(T::mark).collect() + } +} +impl Unmark for Vec { + type Unmarked = Vec; + fn unmark(self) -> Self::Unmarked { + // Should be a no-op due to std's in-place collect optimizations. + self.into_iter().map(T::unmark).collect() + } +} + +macro_rules! mark_noop { + ($($ty:ty),* $(,)?) => { + $( + impl Mark for $ty { + type Unmarked = Self; + fn mark(unmarked: Self::Unmarked) -> Self { + unmarked + } + } + impl Unmark for $ty { + type Unmarked = Self; + fn unmark(self) -> Self::Unmarked { + self + } + } + )* + } +} +mark_noop! { + (), + bool, + char, + &'_ [u8], + &'_ str, + String, + u8, + usize, + Delimiter, + Level, + LineColumn, + Spacing, +} + +rpc_encode_decode!( + enum Delimiter { + Parenthesis, + Brace, + Bracket, + None, + } +); +rpc_encode_decode!( + enum Level { + Error, + Warning, + Note, + Help, + } +); +rpc_encode_decode!(struct LineColumn { line, column }); +rpc_encode_decode!( + enum Spacing { + Alone, + Joint, + } +); + +macro_rules! mark_compound { + (struct $name:ident <$($T:ident),+> { $($field:ident),* $(,)? }) => { + impl<$($T: Mark),+> Mark for $name <$($T),+> { + type Unmarked = $name <$($T::Unmarked),+>; + fn mark(unmarked: Self::Unmarked) -> Self { + $name { + $($field: Mark::mark(unmarked.$field)),* + } + } + } + impl<$($T: Unmark),+> Unmark for $name <$($T),+> { + type Unmarked = $name <$($T::Unmarked),+>; + fn unmark(self) -> Self::Unmarked { + $name { + $($field: Unmark::unmark(self.$field)),* + } + } + } + }; + (enum $name:ident <$($T:ident),+> { $($variant:ident $(($field:ident))?),* $(,)? }) => { + impl<$($T: Mark),+> Mark for $name <$($T),+> { + type Unmarked = $name <$($T::Unmarked),+>; + fn mark(unmarked: Self::Unmarked) -> Self { + match unmarked { + $($name::$variant $(($field))? => { + $name::$variant $((Mark::mark($field)))? + })* + } + } + } + impl<$($T: Unmark),+> Unmark for $name <$($T),+> { + type Unmarked = $name <$($T::Unmarked),+>; + fn unmark(self) -> Self::Unmarked { + match self { + $($name::$variant $(($field))? => { + $name::$variant $((Unmark::unmark($field)))? + })* + } + } + } + } +} + +macro_rules! compound_traits { + ($($t:tt)*) => { + rpc_encode_decode!($($t)*); + mark_compound!($($t)*); + }; +} + +compound_traits!( + enum Bound { + Included(x), + Excluded(x), + Unbounded, + } +); + +compound_traits!( + enum Option { + Some(t), + None, + } +); + +compound_traits!( + enum Result { + Ok(t), + Err(e), + } +); + +#[derive(Copy, Clone)] +pub struct DelimSpan { + pub open: Span, + pub close: Span, + pub entire: Span, +} + +impl DelimSpan { + pub fn from_single(span: Span) -> Self { + DelimSpan { open: span, close: span, entire: span } + } +} + +compound_traits!(struct DelimSpan { open, close, entire }); + +#[derive(Clone)] +pub struct Group { + pub delimiter: Delimiter, + pub stream: Option, + pub span: DelimSpan, +} + +compound_traits!(struct Group { delimiter, stream, span }); + +#[derive(Clone)] +pub struct Punct { + pub ch: u8, + pub joint: bool, + pub span: Span, +} + +compound_traits!(struct Punct { ch, joint, span }); + +#[derive(Clone)] +pub enum TokenTree { + Group(Group), + Punct(Punct), + Ident(Ident), + Literal(Literal), +} + +compound_traits!( + enum TokenTree { + Group(tt), + Punct(tt), + Ident(tt), + Literal(tt), + } +); + +/// Globals provided alongside the initial inputs for a macro expansion. +/// Provides values such as spans which are used frequently to avoid RPC. +#[derive(Clone)] +pub struct ExpnGlobals { + pub def_site: Span, + pub call_site: Span, + pub mixed_site: Span, +} + +compound_traits!( + struct ExpnGlobals { def_site, call_site, mixed_site } +); diff --git a/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/rpc.rs b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/rpc.rs new file mode 100644 index 00000000000..e9d7a46c06f --- /dev/null +++ b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/rpc.rs @@ -0,0 +1,304 @@ +//! Serialization for client-server communication. + +use std::any::Any; +use std::char; +use std::io::Write; +use std::num::NonZeroU32; +use std::str; + +pub(super) type Writer = super::buffer::Buffer; + +pub(super) trait Encode: Sized { + fn encode(self, w: &mut Writer, s: &mut S); +} + +pub(super) type Reader<'a> = &'a [u8]; + +pub(super) trait Decode<'a, 's, S>: Sized { + fn decode(r: &mut Reader<'a>, s: &'s S) -> Self; +} + +pub(super) trait DecodeMut<'a, 's, S>: Sized { + fn decode(r: &mut Reader<'a>, s: &'s mut S) -> Self; +} + +macro_rules! rpc_encode_decode { + (le $ty:ty) => { + impl Encode for $ty { + fn encode(self, w: &mut Writer, _: &mut S) { + w.extend_from_array(&self.to_le_bytes()); + } + } + + impl DecodeMut<'_, '_, S> for $ty { + fn decode(r: &mut Reader<'_>, _: &mut S) -> Self { + const N: usize = ::std::mem::size_of::<$ty>(); + + let mut bytes = [0; N]; + bytes.copy_from_slice(&r[..N]); + *r = &r[N..]; + + Self::from_le_bytes(bytes) + } + } + }; + (struct $name:ident $(<$($T:ident),+>)? { $($field:ident),* $(,)? }) => { + impl),+)?> Encode for $name $(<$($T),+>)? { + fn encode(self, w: &mut Writer, s: &mut S) { + $(self.$field.encode(w, s);)* + } + } + + impl<'a, S, $($($T: for<'s> DecodeMut<'a, 's, S>),+)?> DecodeMut<'a, '_, S> + for $name $(<$($T),+>)? + { + fn decode(r: &mut Reader<'a>, s: &mut S) -> Self { + $name { + $($field: DecodeMut::decode(r, s)),* + } + } + } + }; + (enum $name:ident $(<$($T:ident),+>)? { $($variant:ident $(($field:ident))*),* $(,)? }) => { + impl),+)?> Encode for $name $(<$($T),+>)? { + fn encode(self, w: &mut Writer, s: &mut S) { + // HACK(eddyb): `Tag` enum duplicated between the + // two impls as there's no other place to stash it. + #[allow(non_upper_case_globals)] + mod tag { + #[repr(u8)] enum Tag { $($variant),* } + + $(pub const $variant: u8 = Tag::$variant as u8;)* + } + + match self { + $($name::$variant $(($field))* => { + tag::$variant.encode(w, s); + $($field.encode(w, s);)* + })* + } + } + } + + impl<'a, S, $($($T: for<'s> DecodeMut<'a, 's, S>),+)?> DecodeMut<'a, '_, S> + for $name $(<$($T),+>)? + { + fn decode(r: &mut Reader<'a>, s: &mut S) -> Self { + // HACK(eddyb): `Tag` enum duplicated between the + // two impls as there's no other place to stash it. + #[allow(non_upper_case_globals)] + mod tag { + #[repr(u8)] enum Tag { $($variant),* } + + $(pub const $variant: u8 = Tag::$variant as u8;)* + } + + match u8::decode(r, s) { + $(tag::$variant => { + $(let $field = DecodeMut::decode(r, s);)* + $name::$variant $(($field))* + })* + _ => unreachable!(), + } + } + } + } +} + +impl Encode for () { + fn encode(self, _: &mut Writer, _: &mut S) {} +} + +impl DecodeMut<'_, '_, S> for () { + fn decode(_: &mut Reader<'_>, _: &mut S) -> Self {} +} + +impl Encode for u8 { + fn encode(self, w: &mut Writer, _: &mut S) { + w.push(self); + } +} + +impl DecodeMut<'_, '_, S> for u8 { + fn decode(r: &mut Reader<'_>, _: &mut S) -> Self { + let x = r[0]; + *r = &r[1..]; + x + } +} + +rpc_encode_decode!(le u32); +rpc_encode_decode!(le usize); + +impl Encode for bool { + fn encode(self, w: &mut Writer, s: &mut S) { + (self as u8).encode(w, s); + } +} + +impl DecodeMut<'_, '_, S> for bool { + fn decode(r: &mut Reader<'_>, s: &mut S) -> Self { + match u8::decode(r, s) { + 0 => false, + 1 => true, + _ => unreachable!(), + } + } +} + +impl Encode for char { + fn encode(self, w: &mut Writer, s: &mut S) { + (self as u32).encode(w, s); + } +} + +impl DecodeMut<'_, '_, S> for char { + fn decode(r: &mut Reader<'_>, s: &mut S) -> Self { + char::from_u32(u32::decode(r, s)).unwrap() + } +} + +impl Encode for NonZeroU32 { + fn encode(self, w: &mut Writer, s: &mut S) { + self.get().encode(w, s); + } +} + +impl DecodeMut<'_, '_, S> for NonZeroU32 { + fn decode(r: &mut Reader<'_>, s: &mut S) -> Self { + Self::new(u32::decode(r, s)).unwrap() + } +} + +impl, B: Encode> Encode for (A, B) { + fn encode(self, w: &mut Writer, s: &mut S) { + self.0.encode(w, s); + self.1.encode(w, s); + } +} + +impl<'a, S, A: for<'s> DecodeMut<'a, 's, S>, B: for<'s> DecodeMut<'a, 's, S>> DecodeMut<'a, '_, S> + for (A, B) +{ + fn decode(r: &mut Reader<'a>, s: &mut S) -> Self { + (DecodeMut::decode(r, s), DecodeMut::decode(r, s)) + } +} + +impl Encode for &[u8] { + fn encode(self, w: &mut Writer, s: &mut S) { + self.len().encode(w, s); + w.write_all(self).unwrap(); + } +} + +impl<'a, S> DecodeMut<'a, '_, S> for &'a [u8] { + fn decode(r: &mut Reader<'a>, s: &mut S) -> Self { + let len = usize::decode(r, s); + let xs = &r[..len]; + *r = &r[len..]; + xs + } +} + +impl Encode for &str { + fn encode(self, w: &mut Writer, s: &mut S) { + self.as_bytes().encode(w, s); + } +} + +impl<'a, S> DecodeMut<'a, '_, S> for &'a str { + fn decode(r: &mut Reader<'a>, s: &mut S) -> Self { + str::from_utf8(<&[u8]>::decode(r, s)).unwrap() + } +} + +impl Encode for String { + fn encode(self, w: &mut Writer, s: &mut S) { + self[..].encode(w, s); + } +} + +impl DecodeMut<'_, '_, S> for String { + fn decode(r: &mut Reader<'_>, s: &mut S) -> Self { + <&str>::decode(r, s).to_string() + } +} + +impl> Encode for Vec { + fn encode(self, w: &mut Writer, s: &mut S) { + self.len().encode(w, s); + for x in self { + x.encode(w, s); + } + } +} + +impl<'a, S, T: for<'s> DecodeMut<'a, 's, S>> DecodeMut<'a, '_, S> for Vec { + fn decode(r: &mut Reader<'a>, s: &mut S) -> Self { + let len = usize::decode(r, s); + let mut vec = Vec::with_capacity(len); + for _ in 0..len { + vec.push(T::decode(r, s)); + } + vec + } +} + +/// Simplified version of panic payloads, ignoring +/// types other than `&'static str` and `String`. +pub enum PanicMessage { + StaticStr(&'static str), + String(String), + Unknown, +} + +impl From> for PanicMessage { + fn from(payload: Box) -> Self { + if let Some(s) = payload.downcast_ref::<&'static str>() { + return PanicMessage::StaticStr(s); + } + if let Ok(s) = payload.downcast::() { + return PanicMessage::String(*s); + } + PanicMessage::Unknown + } +} + +impl Into> for PanicMessage { + fn into(self) -> Box { + match self { + PanicMessage::StaticStr(s) => Box::new(s), + PanicMessage::String(s) => Box::new(s), + PanicMessage::Unknown => { + struct UnknownPanicMessage; + Box::new(UnknownPanicMessage) + } + } + } +} + +impl PanicMessage { + pub fn as_str(&self) -> Option<&str> { + match self { + PanicMessage::StaticStr(s) => Some(s), + PanicMessage::String(s) => Some(s), + PanicMessage::Unknown => None, + } + } +} + +impl Encode for PanicMessage { + fn encode(self, w: &mut Writer, s: &mut S) { + self.as_str().encode(w, s); + } +} + +impl DecodeMut<'_, '_, S> for PanicMessage { + fn decode(r: &mut Reader<'_>, s: &mut S) -> Self { + match Option::::decode(r, s) { + Some(s) => PanicMessage::String(s), + None => PanicMessage::Unknown, + } + } +} diff --git a/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/scoped_cell.rs b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/scoped_cell.rs new file mode 100644 index 00000000000..2cde1f65adf --- /dev/null +++ b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/scoped_cell.rs @@ -0,0 +1,81 @@ +//! `Cell` variant for (scoped) existential lifetimes. + +use std::cell::Cell; +use std::mem; +use std::ops::{Deref, DerefMut}; + +/// Type lambda application, with a lifetime. +#[allow(unused_lifetimes)] +pub trait ApplyL<'a> { + type Out; +} + +/// Type lambda taking a lifetime, i.e., `Lifetime -> Type`. +pub trait LambdaL: for<'a> ApplyL<'a> {} + +impl ApplyL<'a>> LambdaL for T {} + +// HACK(eddyb) work around projection limitations with a newtype +// FIXME(#52812) replace with `&'a mut >::Out` +pub struct RefMutL<'a, 'b, T: LambdaL>(&'a mut >::Out); + +impl<'a, 'b, T: LambdaL> Deref for RefMutL<'a, 'b, T> { + type Target = >::Out; + fn deref(&self) -> &Self::Target { + self.0 + } +} + +impl<'a, 'b, T: LambdaL> DerefMut for RefMutL<'a, 'b, T> { + fn deref_mut(&mut self) -> &mut Self::Target { + self.0 + } +} + +pub struct ScopedCell(Cell<>::Out>); + +impl ScopedCell { + pub const fn new(value: >::Out) -> Self { + ScopedCell(Cell::new(value)) + } + + /// Sets the value in `self` to `replacement` while + /// running `f`, which gets the old value, mutably. + /// The old value will be restored after `f` exits, even + /// by panic, including modifications made to it by `f`. + pub fn replace<'a, R>( + &self, + replacement: >::Out, + f: impl for<'b, 'c> FnOnce(RefMutL<'b, 'c, T>) -> R, + ) -> R { + /// Wrapper that ensures that the cell always gets filled + /// (with the original state, optionally changed by `f`), + /// even if `f` had panicked. + struct PutBackOnDrop<'a, T: LambdaL> { + cell: &'a ScopedCell, + value: Option<>::Out>, + } + + impl<'a, T: LambdaL> Drop for PutBackOnDrop<'a, T> { + fn drop(&mut self) { + self.cell.0.set(self.value.take().unwrap()); + } + } + + let mut put_back_on_drop = PutBackOnDrop { + cell: self, + value: Some(self.0.replace(unsafe { + let erased = mem::transmute_copy(&replacement); + mem::forget(replacement); + erased + })), + }; + + f(RefMutL(put_back_on_drop.value.as_mut().unwrap())) + } + + /// Sets the value in `self` to `value` while running `f`. + pub fn set(&self, value: >::Out, f: impl FnOnce() -> R) -> R { + self.replace(value, |_| f()) + } +} diff --git a/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/selfless_reify.rs b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/selfless_reify.rs new file mode 100644 index 00000000000..907ad256e4b --- /dev/null +++ b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/selfless_reify.rs @@ -0,0 +1,84 @@ +//! Abstraction for creating `fn` pointers from any callable that *effectively* +//! has the equivalent of implementing `Default`, even if the compiler neither +//! provides `Default` nor allows reifying closures (i.e. creating `fn` pointers) +//! other than those with absolutely no captures. +//! +//! More specifically, for a closure-like type to be "effectively `Default`": +//! * it must be a ZST (zero-sized type): no information contained within, so +//! that `Default`'s return value (if it were implemented) is unambiguous +//! * it must be `Copy`: no captured "unique ZST tokens" or any other similar +//! types that would make duplicating values at will unsound +//! * combined with the ZST requirement, this confers a kind of "telecopy" +//! ability: similar to `Copy`, but without keeping the value around, and +//! instead "reconstructing" it (a noop given it's a ZST) when needed +//! * it must be *provably* inhabited: no captured uninhabited types or any +//! other types that cannot be constructed by the user of this abstraction +//! * the proof is a value of the closure-like type itself, in a sense the +//! "seed" for the "telecopy" process made possible by ZST + `Copy` +//! * this requirement is the only reason an abstraction limited to a specific +//! usecase is required: ZST + `Copy` can be checked with *at worst* a panic +//! at the "attempted `::default()` call" time, but that doesn't guarantee +//! that the value can be soundly created, and attempting to use the typical +//! "proof ZST token" approach leads yet again to having a ZST + `Copy` type +//! that is not proof of anything without a value (i.e. isomorphic to a +//! newtype of the type it's trying to prove the inhabitation of) +//! +//! A more flexible (and safer) solution to the general problem could exist once +//! `const`-generic parameters can have type parameters in their types: +//! +//! ```rust,ignore (needs future const-generics) +//! extern "C" fn ffi_wrapper< +//! A, R, +//! F: Fn(A) -> R, +//! const f: F, // <-- this `const`-generic is not yet allowed +//! >(arg: A) -> R { +//! f(arg) +//! } +//! ``` + +use std::mem; + +// FIXME(eddyb) this could be `trait` impls except for the `const fn` requirement. +macro_rules! define_reify_functions { + ($( + fn $name:ident $(<$($param:ident),*>)? + for $(extern $abi:tt)? fn($($arg:ident: $arg_ty:ty),*) -> $ret_ty:ty; + )+) => { + $(pub const fn $name< + $($($param,)*)? + F: Fn($($arg_ty),*) -> $ret_ty + Copy + >(f: F) -> $(extern $abi)? fn($($arg_ty),*) -> $ret_ty { + // FIXME(eddyb) describe the `F` type (e.g. via `type_name::`) once panic + // formatting becomes possible in `const fn`. + assert!(mem::size_of::() == 0, "selfless_reify: closure must be zero-sized"); + + $(extern $abi)? fn wrapper< + $($($param,)*)? + F: Fn($($arg_ty),*) -> $ret_ty + Copy + >($($arg: $arg_ty),*) -> $ret_ty { + let f = unsafe { + // SAFETY: `F` satisfies all criteria for "out of thin air" + // reconstructability (see module-level doc comment). + mem::MaybeUninit::::uninit().assume_init() + }; + f($($arg),*) + } + let _f_proof = f; + wrapper::< + $($($param,)*)? + F + > + })+ + } +} + +define_reify_functions! { + fn _reify_to_extern_c_fn_unary for extern "C" fn(arg: A) -> R; + + // HACK(eddyb) this abstraction is used with `for<'a> fn(BridgeConfig<'a>) + // -> T` but that doesn't work with just `reify_to_extern_c_fn_unary` + // because of the `fn` pointer type being "higher-ranked" (i.e. the + // `for<'a>` binder). + // FIXME(eddyb) try to remove the lifetime from `BridgeConfig`, that'd help. + fn reify_to_extern_c_fn_hrt_bridge for extern "C" fn(bridge: super::BridgeConfig<'_>) -> R; +} diff --git a/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/server.rs b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/server.rs new file mode 100644 index 00000000000..6e7a8d8c10d --- /dev/null +++ b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/bridge/server.rs @@ -0,0 +1,339 @@ +//! Server-side traits. + +use super::*; + +// FIXME(eddyb) generate the definition of `HandleStore` in `server.rs`. +use super::client::HandleStore; + +pub trait Types { + type FreeFunctions: 'static; + type TokenStream: 'static + Clone; + type Ident: 'static + Copy + Eq + Hash; + type Literal: 'static + Clone; + type SourceFile: 'static + Clone; + type MultiSpan: 'static; + type Diagnostic: 'static; + type Span: 'static + Copy + Eq + Hash; +} + +/// Declare an associated fn of one of the traits below, adding necessary +/// default bodies. +macro_rules! associated_fn { + (fn drop(&mut self, $arg:ident: $arg_ty:ty)) => + (fn drop(&mut self, $arg: $arg_ty) { mem::drop($arg) }); + + (fn clone(&mut self, $arg:ident: $arg_ty:ty) -> $ret_ty:ty) => + (fn clone(&mut self, $arg: $arg_ty) -> $ret_ty { $arg.clone() }); + + ($($item:tt)*) => ($($item)*;) +} + +macro_rules! declare_server_traits { + ($($name:ident { + $(fn $method:ident($($arg:ident: $arg_ty:ty),* $(,)?) $(-> $ret_ty:ty)?;)* + }),* $(,)?) => { + $(pub trait $name: Types { + $(associated_fn!(fn $method(&mut self, $($arg: $arg_ty),*) $(-> $ret_ty)?);)* + })* + + pub trait Server: Types $(+ $name)* { + fn globals(&mut self) -> ExpnGlobals; + } + } +} +with_api!(Self, self_, declare_server_traits); + +pub(super) struct MarkedTypes(S); + +impl Server for MarkedTypes { + fn globals(&mut self) -> ExpnGlobals { + <_>::mark(Server::globals(&mut self.0)) + } +} + +macro_rules! define_mark_types_impls { + ($($name:ident { + $(fn $method:ident($($arg:ident: $arg_ty:ty),* $(,)?) $(-> $ret_ty:ty)?;)* + }),* $(,)?) => { + impl Types for MarkedTypes { + $(type $name = Marked;)* + } + + $(impl $name for MarkedTypes { + $(fn $method(&mut self, $($arg: $arg_ty),*) $(-> $ret_ty)? { + <_>::mark($name::$method(&mut self.0, $($arg.unmark()),*)) + })* + })* + } +} +with_api!(Self, self_, define_mark_types_impls); + +struct Dispatcher { + handle_store: HandleStore, + server: S, +} + +macro_rules! define_dispatcher_impl { + ($($name:ident { + $(fn $method:ident($($arg:ident: $arg_ty:ty),* $(,)?) $(-> $ret_ty:ty)?;)* + }),* $(,)?) => { + // FIXME(eddyb) `pub` only for `ExecutionStrategy` below. + pub trait DispatcherTrait { + // HACK(eddyb) these are here to allow `Self::$name` to work below. + $(type $name;)* + fn dispatch(&mut self, buf: Buffer) -> Buffer; + } + + impl DispatcherTrait for Dispatcher> { + $(type $name = as Types>::$name;)* + fn dispatch(&mut self, mut buf: Buffer) -> Buffer { + let Dispatcher { handle_store, server } = self; + + let mut reader = &buf[..]; + match api_tags::Method::decode(&mut reader, &mut ()) { + $(api_tags::Method::$name(m) => match m { + $(api_tags::$name::$method => { + let mut call_method = || { + reverse_decode!(reader, handle_store; $($arg: $arg_ty),*); + $name::$method(server, $($arg),*) + }; + // HACK(eddyb) don't use `panic::catch_unwind` in a panic. + // If client and server happen to use the same `libstd`, + // `catch_unwind` asserts that the panic counter was 0, + // even when the closure passed to it didn't panic. + let r = if thread::panicking() { + Ok(call_method()) + } else { + panic::catch_unwind(panic::AssertUnwindSafe(call_method)) + .map_err(PanicMessage::from) + }; + + buf.clear(); + r.encode(&mut buf, handle_store); + })* + }),* + } + buf + } + } + } +} +with_api!(Self, self_, define_dispatcher_impl); + +pub trait ExecutionStrategy { + fn run_bridge_and_client( + &self, + dispatcher: &mut impl DispatcherTrait, + input: Buffer, + run_client: extern "C" fn(BridgeConfig<'_>) -> Buffer, + force_show_panics: bool, + ) -> Buffer; +} + +pub struct SameThread; + +impl ExecutionStrategy for SameThread { + fn run_bridge_and_client( + &self, + dispatcher: &mut impl DispatcherTrait, + input: Buffer, + run_client: extern "C" fn(BridgeConfig<'_>) -> Buffer, + force_show_panics: bool, + ) -> Buffer { + let mut dispatch = |buf| dispatcher.dispatch(buf); + + run_client(BridgeConfig { + input, + dispatch: (&mut dispatch).into(), + force_show_panics, + _marker: marker::PhantomData, + }) + } +} + +// NOTE(eddyb) Two implementations are provided, the second one is a bit +// faster but neither is anywhere near as fast as same-thread execution. + +pub struct CrossThread1; + +impl ExecutionStrategy for CrossThread1 { + fn run_bridge_and_client( + &self, + dispatcher: &mut impl DispatcherTrait, + input: Buffer, + run_client: extern "C" fn(BridgeConfig<'_>) -> Buffer, + force_show_panics: bool, + ) -> Buffer { + use std::sync::mpsc::channel; + + let (req_tx, req_rx) = channel(); + let (res_tx, res_rx) = channel(); + + let join_handle = thread::spawn(move || { + let mut dispatch = |buf| { + req_tx.send(buf).unwrap(); + res_rx.recv().unwrap() + }; + + run_client(BridgeConfig { + input, + dispatch: (&mut dispatch).into(), + force_show_panics, + _marker: marker::PhantomData, + }) + }); + + for b in req_rx { + res_tx.send(dispatcher.dispatch(b)).unwrap(); + } + + join_handle.join().unwrap() + } +} + +pub struct CrossThread2; + +impl ExecutionStrategy for CrossThread2 { + fn run_bridge_and_client( + &self, + dispatcher: &mut impl DispatcherTrait, + input: Buffer, + run_client: extern "C" fn(BridgeConfig<'_>) -> Buffer, + force_show_panics: bool, + ) -> Buffer { + use std::sync::{Arc, Mutex}; + + enum State { + Req(T), + Res(T), + } + + let mut state = Arc::new(Mutex::new(State::Res(Buffer::new()))); + + let server_thread = thread::current(); + let state2 = state.clone(); + let join_handle = thread::spawn(move || { + let mut dispatch = |b| { + *state2.lock().unwrap() = State::Req(b); + server_thread.unpark(); + loop { + thread::park(); + if let State::Res(b) = &mut *state2.lock().unwrap() { + break b.take(); + } + } + }; + + let r = run_client(BridgeConfig { + input, + dispatch: (&mut dispatch).into(), + force_show_panics, + _marker: marker::PhantomData, + }); + + // Wake up the server so it can exit the dispatch loop. + drop(state2); + server_thread.unpark(); + + r + }); + + // Check whether `state2` was dropped, to know when to stop. + while Arc::get_mut(&mut state).is_none() { + thread::park(); + let mut b = match &mut *state.lock().unwrap() { + State::Req(b) => b.take(), + _ => continue, + }; + b = dispatcher.dispatch(b.take()); + *state.lock().unwrap() = State::Res(b); + join_handle.thread().unpark(); + } + + join_handle.join().unwrap() + } +} + +fn run_server< + S: Server, + I: Encode>>, + O: for<'a, 's> DecodeMut<'a, 's, HandleStore>>, +>( + strategy: &impl ExecutionStrategy, + handle_counters: &'static client::HandleCounters, + server: S, + input: I, + run_client: extern "C" fn(BridgeConfig<'_>) -> Buffer, + force_show_panics: bool, +) -> Result { + let mut dispatcher = + Dispatcher { handle_store: HandleStore::new(handle_counters), server: MarkedTypes(server) }; + + let globals = dispatcher.server.globals(); + + let mut buf = Buffer::new(); + (globals, input).encode(&mut buf, &mut dispatcher.handle_store); + + buf = strategy.run_bridge_and_client(&mut dispatcher, buf, run_client, force_show_panics); + + Result::decode(&mut &buf[..], &mut dispatcher.handle_store) +} + +impl client::Client { + pub fn run( + &self, + strategy: &impl ExecutionStrategy, + server: S, + input: S::TokenStream, + force_show_panics: bool, + ) -> Result + where + S: Server, + S::TokenStream: Default, + { + let client::Client { get_handle_counters, run, _marker } = *self; + run_server( + strategy, + get_handle_counters(), + server, + as Types>::TokenStream::mark(input), + run, + force_show_panics, + ) + .map(|s| as Types>::TokenStream>>::unmark(s).unwrap_or_default()) + } +} + +impl + client::Client< + (super::super::TokenStream, super::super::TokenStream), + super::super::TokenStream, + > +{ + pub fn run( + &self, + strategy: &impl ExecutionStrategy, + server: S, + input: S::TokenStream, + input2: S::TokenStream, + force_show_panics: bool, + ) -> Result + where + S: Server, + S::TokenStream: Default, + { + let client::Client { get_handle_counters, run, _marker } = *self; + run_server( + strategy, + get_handle_counters(), + server, + ( + as Types>::TokenStream::mark(input), + as Types>::TokenStream::mark(input2), + ), + run, + force_show_panics, + ) + .map(|s| as Types>::TokenStream>>::unmark(s).unwrap_or_default()) + } +} diff --git a/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/diagnostic.rs b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/diagnostic.rs new file mode 100644 index 00000000000..3fade2dc4f9 --- /dev/null +++ b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/diagnostic.rs @@ -0,0 +1,166 @@ +//! lib-proc-macro diagnostic +//! +//! Copy from +//! augmented with removing unstable features + +use super::Span; + +/// An enum representing a diagnostic level. +#[derive(Copy, Clone, Debug)] +#[non_exhaustive] +pub enum Level { + /// An error. + Error, + /// A warning. + Warning, + /// A note. + Note, + /// A help message. + Help, +} + +/// Trait implemented by types that can be converted into a set of `Span`s. +pub trait MultiSpan { + /// Converts `self` into a `Vec`. + fn into_spans(self) -> Vec; +} + +impl MultiSpan for Span { + fn into_spans(self) -> Vec { + vec![self] + } +} + +impl MultiSpan for Vec { + fn into_spans(self) -> Vec { + self + } +} + +impl<'a> MultiSpan for &'a [Span] { + fn into_spans(self) -> Vec { + self.to_vec() + } +} + +/// A structure representing a diagnostic message and associated children +/// messages. +#[derive(Clone, Debug)] +pub struct Diagnostic { + level: Level, + message: String, + spans: Vec, + children: Vec, +} + +macro_rules! diagnostic_child_methods { + ($spanned:ident, $regular:ident, $level:expr) => { + #[doc = concat!("Adds a new child diagnostics message to `self` with the [`", + stringify!($level), "`] level, and the given `spans` and `message`.")] + pub fn $spanned(mut self, spans: S, message: T) -> Diagnostic + where + S: MultiSpan, + T: Into, + { + self.children.push(Diagnostic::spanned(spans, $level, message)); + self + } + + #[doc = concat!("Adds a new child diagnostic message to `self` with the [`", + stringify!($level), "`] level, and the given `message`.")] + pub fn $regular>(mut self, message: T) -> Diagnostic { + self.children.push(Diagnostic::new($level, message)); + self + } + }; +} + +/// Iterator over the children diagnostics of a `Diagnostic`. +#[derive(Debug, Clone)] +pub struct Children<'a>(std::slice::Iter<'a, Diagnostic>); + +impl<'a> Iterator for Children<'a> { + type Item = &'a Diagnostic; + + fn next(&mut self) -> Option { + self.0.next() + } +} + +impl Diagnostic { + /// Creates a new diagnostic with the given `level` and `message`. + pub fn new>(level: Level, message: T) -> Diagnostic { + Diagnostic { level, message: message.into(), spans: vec![], children: vec![] } + } + + /// Creates a new diagnostic with the given `level` and `message` pointing to + /// the given set of `spans`. + pub fn spanned(spans: S, level: Level, message: T) -> Diagnostic + where + S: MultiSpan, + T: Into, + { + Diagnostic { level, message: message.into(), spans: spans.into_spans(), children: vec![] } + } + + diagnostic_child_methods!(span_error, error, Level::Error); + diagnostic_child_methods!(span_warning, warning, Level::Warning); + diagnostic_child_methods!(span_note, note, Level::Note); + diagnostic_child_methods!(span_help, help, Level::Help); + + /// Returns the diagnostic `level` for `self`. + pub fn level(&self) -> Level { + self.level + } + + /// Sets the level in `self` to `level`. + pub fn set_level(&mut self, level: Level) { + self.level = level; + } + + /// Returns the message in `self`. + pub fn message(&self) -> &str { + &self.message + } + + /// Sets the message in `self` to `message`. + pub fn set_message>(&mut self, message: T) { + self.message = message.into(); + } + + /// Returns the `Span`s in `self`. + pub fn spans(&self) -> &[Span] { + &self.spans + } + + /// Sets the `Span`s in `self` to `spans`. + pub fn set_spans(&mut self, spans: S) { + self.spans = spans.into_spans(); + } + + /// Returns an iterator over the children diagnostics of `self`. + pub fn children(&self) -> Children<'_> { + Children(self.children.iter()) + } + + /// Emit the diagnostic. + pub fn emit(self) { + fn to_internal(spans: Vec) -> super::bridge::client::MultiSpan { + let mut multi_span = super::bridge::client::MultiSpan::new(); + for span in spans { + multi_span.push(span.0); + } + multi_span + } + + let mut diag = super::bridge::client::Diagnostic::new( + self.level, + &self.message[..], + to_internal(self.spans), + ); + for c in self.children { + diag.sub(c.level, &c.message[..], to_internal(c.spans)); + } + diag.emit(); + } +} diff --git a/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/mod.rs b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/mod.rs new file mode 100644 index 00000000000..be62c73ef32 --- /dev/null +++ b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/mod.rs @@ -0,0 +1,1125 @@ +//! A support library for macro authors when defining new macros. +//! +//! This library, provided by the standard distribution, provides the types +//! consumed in the interfaces of procedurally defined macro definitions such as +//! function-like macros `#[proc_macro]`, macro attributes `#[proc_macro_attribute]` and +//! custom derive attributes`#[proc_macro_derive]`. +//! +//! See [the book] for more. +//! +//! [the book]: ../book/ch19-06-macros.html#procedural-macros-for-generating-code-from-attributes + +#[doc(hidden)] +pub mod bridge; + +mod diagnostic; + +pub use diagnostic::{Diagnostic, Level, MultiSpan}; + +use std::cmp::Ordering; +use std::ops::RangeBounds; +use std::path::PathBuf; +use std::str::FromStr; +use std::{error, fmt, iter, mem}; + +/// Determines whether proc_macro has been made accessible to the currently +/// running program. +/// +/// The proc_macro crate is only intended for use inside the implementation of +/// procedural macros. All the functions in this crate panic if invoked from +/// outside of a procedural macro, such as from a build script or unit test or +/// ordinary Rust binary. +/// +/// With consideration for Rust libraries that are designed to support both +/// macro and non-macro use cases, `proc_macro::is_available()` provides a +/// non-panicking way to detect whether the infrastructure required to use the +/// API of proc_macro is presently available. Returns true if invoked from +/// inside of a procedural macro, false if invoked from any other binary. +pub fn is_available() -> bool { + bridge::client::is_available() +} + +/// The main type provided by this crate, representing an abstract stream of +/// tokens, or, more specifically, a sequence of token trees. +/// The type provide interfaces for iterating over those token trees and, conversely, +/// collecting a number of token trees into one stream. +/// +/// This is both the input and output of `#[proc_macro]`, `#[proc_macro_attribute]` +/// and `#[proc_macro_derive]` definitions. +#[derive(Clone)] +pub struct TokenStream(Option); + +/// Error returned from `TokenStream::from_str`. +#[non_exhaustive] +#[derive(Debug)] +pub struct LexError; + +impl fmt::Display for LexError { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.write_str("cannot parse string into token stream") + } +} + +impl error::Error for LexError {} + +/// Error returned from `TokenStream::expand_expr`. +#[non_exhaustive] +#[derive(Debug)] +pub struct ExpandError; + +impl fmt::Display for ExpandError { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.write_str("macro expansion failed") + } +} + +impl error::Error for ExpandError {} + +impl TokenStream { + /// Returns an empty `TokenStream` containing no token trees. + pub fn new() -> TokenStream { + TokenStream(None) + } + + /// Checks if this `TokenStream` is empty. + pub fn is_empty(&self) -> bool { + self.0.as_ref().map(|h| h.is_empty()).unwrap_or(true) + } + + /// Parses this `TokenStream` as an expression and attempts to expand any + /// macros within it. Returns the expanded `TokenStream`. + /// + /// Currently only expressions expanding to literals will succeed, although + /// this may be relaxed in the future. + /// + /// NOTE: In error conditions, `expand_expr` may leave macros unexpanded, + /// report an error, failing compilation, and/or return an `Err(..)`. The + /// specific behavior for any error condition, and what conditions are + /// considered errors, is unspecified and may change in the future. + pub fn expand_expr(&self) -> Result { + let stream = self.0.as_ref().ok_or(ExpandError)?; + match bridge::client::TokenStream::expand_expr(stream) { + Ok(stream) => Ok(TokenStream(Some(stream))), + Err(_) => Err(ExpandError), + } + } +} + +/// Attempts to break the string into tokens and parse those tokens into a token stream. +/// May fail for a number of reasons, for example, if the string contains unbalanced delimiters +/// or characters not existing in the language. +/// All tokens in the parsed stream get `Span::call_site()` spans. +/// +/// NOTE: some errors may cause panics instead of returning `LexError`. We reserve the right to +/// change these errors into `LexError`s later. +impl FromStr for TokenStream { + type Err = LexError; + + fn from_str(src: &str) -> Result { + Ok(TokenStream(Some(bridge::client::TokenStream::from_str(src)))) + } +} + +/// Prints the token stream as a string that is supposed to be losslessly convertible back +/// into the same token stream (modulo spans), except for possibly `TokenTree::Group`s +/// with `Delimiter::None` delimiters and negative numeric literals. +impl fmt::Display for TokenStream { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.write_str(&self.to_string()) + } +} + +/// Prints token in a form convenient for debugging. +impl fmt::Debug for TokenStream { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.write_str("TokenStream ")?; + f.debug_list().entries(self.clone()).finish() + } +} + +impl Default for TokenStream { + fn default() -> Self { + TokenStream::new() + } +} + +pub use quote::{quote, quote_span}; + +fn tree_to_bridge_tree( + tree: TokenTree, +) -> bridge::TokenTree< + bridge::client::TokenStream, + bridge::client::Span, + bridge::client::Ident, + bridge::client::Literal, +> { + match tree { + TokenTree::Group(tt) => bridge::TokenTree::Group(tt.0), + TokenTree::Punct(tt) => bridge::TokenTree::Punct(tt.0), + TokenTree::Ident(tt) => bridge::TokenTree::Ident(tt.0), + TokenTree::Literal(tt) => bridge::TokenTree::Literal(tt.0), + } +} + +/// Creates a token stream containing a single token tree. +impl From for TokenStream { + fn from(tree: TokenTree) -> TokenStream { + TokenStream(Some(bridge::client::TokenStream::from_token_tree(tree_to_bridge_tree(tree)))) + } +} + +/// Non-generic helper for implementing `FromIterator` and +/// `Extend` with less monomorphization in calling crates. +struct ConcatStreamsHelper { + streams: Vec, +} + +impl ConcatStreamsHelper { + fn new(capacity: usize) -> Self { + ConcatStreamsHelper { streams: Vec::with_capacity(capacity) } + } + + fn push(&mut self, stream: TokenStream) { + if let Some(stream) = stream.0 { + self.streams.push(stream); + } + } + + fn build(mut self) -> TokenStream { + if self.streams.len() <= 1 { + TokenStream(self.streams.pop()) + } else { + TokenStream(Some(bridge::client::TokenStream::concat_streams(None, self.streams))) + } + } + + fn append_to(mut self, stream: &mut TokenStream) { + if self.streams.is_empty() { + return; + } + let base = stream.0.take(); + if base.is_none() && self.streams.len() == 1 { + stream.0 = self.streams.pop(); + } else { + stream.0 = Some(bridge::client::TokenStream::concat_streams(base, self.streams)); + } + } +} + +/// Collects a number of token trees into a single stream. +impl iter::FromIterator for TokenStream { + fn from_iter>(trees: I) -> Self { + trees.into_iter().map(TokenStream::from).collect() + } +} + +/// A "flattening" operation on token streams, collects token trees +/// from multiple token streams into a single stream. +impl iter::FromIterator for TokenStream { + fn from_iter>(streams: I) -> Self { + let iter = streams.into_iter(); + let mut builder = ConcatStreamsHelper::new(iter.size_hint().0); + iter.for_each(|stream| builder.push(stream)); + builder.build() + } +} + +impl Extend for TokenStream { + fn extend>(&mut self, trees: I) { + self.extend(trees.into_iter().map(TokenStream::from)); + } +} + +impl Extend for TokenStream { + fn extend>(&mut self, streams: I) { + // FIXME(eddyb) Use an optimized implementation if/when possible. + *self = iter::once(mem::replace(self, Self::new())).chain(streams).collect(); + } +} + +/// Public implementation details for the `TokenStream` type, such as iterators. +pub mod token_stream { + use super::{bridge, Group, Ident, Literal, Punct, TokenStream, TokenTree}; + + /// An iterator over `TokenStream`'s `TokenTree`s. + /// The iteration is "shallow", e.g., the iterator doesn't recurse into delimited groups, + /// and returns whole groups as token trees. + #[derive(Clone)] + pub struct IntoIter( + std::vec::IntoIter< + bridge::TokenTree< + bridge::client::TokenStream, + bridge::client::Span, + bridge::client::Ident, + bridge::client::Literal, + >, + >, + ); + + impl Iterator for IntoIter { + type Item = TokenTree; + + fn next(&mut self) -> Option { + self.0.next().map(|tree| match tree { + bridge::TokenTree::Group(tt) => TokenTree::Group(Group(tt)), + bridge::TokenTree::Punct(tt) => TokenTree::Punct(Punct(tt)), + bridge::TokenTree::Ident(tt) => TokenTree::Ident(Ident(tt)), + bridge::TokenTree::Literal(tt) => TokenTree::Literal(Literal(tt)), + }) + } + } + + impl IntoIterator for TokenStream { + type Item = TokenTree; + type IntoIter = IntoIter; + + fn into_iter(self) -> IntoIter { + IntoIter(self.0.map(|v| v.into_trees()).unwrap_or_default().into_iter()) + } + } +} + +#[doc(hidden)] +mod quote; + +/// A region of source code, along with macro expansion information. +#[derive(Copy, Clone)] +pub struct Span(bridge::client::Span); + +macro_rules! diagnostic_method { + ($name:ident, $level:expr) => { + /// Creates a new `Diagnostic` with the given `message` at the span + /// `self`. + pub fn $name>(self, message: T) -> Diagnostic { + Diagnostic::spanned(self, $level, message) + } + }; +} + +impl Span { + /// A span that resolves at the macro definition site. + pub fn def_site() -> Span { + Span(bridge::client::Span::def_site()) + } + + /// The span of the invocation of the current procedural macro. + /// Identifiers created with this span will be resolved as if they were written + /// directly at the macro call location (call-site hygiene) and other code + /// at the macro call site will be able to refer to them as well. + pub fn call_site() -> Span { + Span(bridge::client::Span::call_site()) + } + + /// A span that represents `macro_rules` hygiene, and sometimes resolves at the macro + /// definition site (local variables, labels, `$crate`) and sometimes at the macro + /// call site (everything else). + /// The span location is taken from the call-site. + pub fn mixed_site() -> Span { + Span(bridge::client::Span::mixed_site()) + } + + /// The original source file into which this span points. + pub fn source_file(&self) -> SourceFile { + SourceFile(self.0.source_file()) + } + + /// The `Span` for the tokens in the previous macro expansion from which + /// `self` was generated from, if any. + pub fn parent(&self) -> Option { + self.0.parent().map(Span) + } + + /// The span for the origin source code that `self` was generated from. If + /// this `Span` wasn't generated from other macro expansions then the return + /// value is the same as `*self`. + pub fn source(&self) -> Span { + Span(self.0.source()) + } + + /// Gets the starting line/column in the source file for this span. + pub fn start(&self) -> LineColumn { + self.0.start().add_1_to_column() + } + + /// Gets the ending line/column in the source file for this span. + pub fn end(&self) -> LineColumn { + self.0.end().add_1_to_column() + } + + /// Creates an empty span pointing to directly before this span. + pub fn before(&self) -> Span { + Span(self.0.before()) + } + + /// Creates an empty span pointing to directly after this span. + pub fn after(&self) -> Span { + Span(self.0.after()) + } + + /// Creates a new span encompassing `self` and `other`. + /// + /// Returns `None` if `self` and `other` are from different files. + pub fn join(&self, other: Span) -> Option { + self.0.join(other.0).map(Span) + } + + /// Creates a new span with the same line/column information as `self` but + /// that resolves symbols as though it were at `other`. + pub fn resolved_at(&self, other: Span) -> Span { + Span(self.0.resolved_at(other.0)) + } + + /// Creates a new span with the same name resolution behavior as `self` but + /// with the line/column information of `other`. + pub fn located_at(&self, other: Span) -> Span { + other.resolved_at(*self) + } + + /// Compares to spans to see if they're equal. + pub fn eq(&self, other: &Span) -> bool { + self.0 == other.0 + } + + /// Returns the source text behind a span. This preserves the original source + /// code, including spaces and comments. It only returns a result if the span + /// corresponds to real source code. + /// + /// Note: The observable result of a macro should only rely on the tokens and + /// not on this source text. The result of this function is a best effort to + /// be used for diagnostics only. + pub fn source_text(&self) -> Option { + self.0.source_text() + } + + // Used by the implementation of `Span::quote` + #[doc(hidden)] + pub fn save_span(&self) -> usize { + self.0.save_span() + } + + // Used by the implementation of `Span::quote` + #[doc(hidden)] + pub fn recover_proc_macro_span(id: usize) -> Span { + Span(bridge::client::Span::recover_proc_macro_span(id)) + } + + diagnostic_method!(error, Level::Error); + diagnostic_method!(warning, Level::Warning); + diagnostic_method!(note, Level::Note); + diagnostic_method!(help, Level::Help); +} + +/// Prints a span in a form convenient for debugging. +impl fmt::Debug for Span { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + self.0.fmt(f) + } +} + +/// A line-column pair representing the start or end of a `Span`. +#[derive(Copy, Clone, Debug, PartialEq, Eq)] +pub struct LineColumn { + /// The 1-indexed line in the source file on which the span starts or ends (inclusive). + pub line: usize, + /// The 1-indexed column (number of bytes in UTF-8 encoding) in the source + /// file on which the span starts or ends (inclusive). + pub column: usize, +} + +impl LineColumn { + fn add_1_to_column(self) -> Self { + LineColumn { line: self.line, column: self.column + 1 } + } +} + +impl Ord for LineColumn { + fn cmp(&self, other: &Self) -> Ordering { + self.line.cmp(&other.line).then(self.column.cmp(&other.column)) + } +} + +impl PartialOrd for LineColumn { + fn partial_cmp(&self, other: &Self) -> Option { + Some(self.cmp(other)) + } +} + +/// The source file of a given `Span`. +#[derive(Clone)] +pub struct SourceFile(bridge::client::SourceFile); + +impl SourceFile { + /// Gets the path to this source file. + /// + /// ### Note + /// If the code span associated with this `SourceFile` was generated by an external macro, this + /// macro, this might not be an actual path on the filesystem. Use [`is_real`] to check. + /// + /// Also note that even if `is_real` returns `true`, if `--remap-path-prefix` was passed on + /// the command line, the path as given might not actually be valid. + /// + /// [`is_real`]: Self::is_real + pub fn path(&self) -> PathBuf { + PathBuf::from(self.0.path()) + } + + /// Returns `true` if this source file is a real source file, and not generated by an external + /// macro's expansion. + pub fn is_real(&self) -> bool { + // This is a hack until intercrate spans are implemented and we can have real source files + // for spans generated in external macros. + // https://github.com/rust-lang/rust/pull/43604#issuecomment-333334368 + self.0.is_real() + } +} + +impl fmt::Debug for SourceFile { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.debug_struct("SourceFile") + .field("path", &self.path()) + .field("is_real", &self.is_real()) + .finish() + } +} + +impl PartialEq for SourceFile { + fn eq(&self, other: &Self) -> bool { + self.0.eq(&other.0) + } +} + +impl Eq for SourceFile {} + +/// A single token or a delimited sequence of token trees (e.g., `[1, (), ..]`). +#[derive(Clone)] +pub enum TokenTree { + /// A token stream surrounded by bracket delimiters. + Group(Group), + /// An identifier. + Ident(Ident), + /// A single punctuation character (`+`, `,`, `$`, etc.). + Punct(Punct), + /// A literal character (`'a'`), string (`"hello"`), number (`2.3`), etc. + Literal(Literal), +} + +impl TokenTree { + /// Returns the span of this tree, delegating to the `span` method of + /// the contained token or a delimited stream. + pub fn span(&self) -> Span { + match *self { + TokenTree::Group(ref t) => t.span(), + TokenTree::Ident(ref t) => t.span(), + TokenTree::Punct(ref t) => t.span(), + TokenTree::Literal(ref t) => t.span(), + } + } + + /// Configures the span for *only this token*. + /// + /// Note that if this token is a `Group` then this method will not configure + /// the span of each of the internal tokens, this will simply delegate to + /// the `set_span` method of each variant. + pub fn set_span(&mut self, span: Span) { + match *self { + TokenTree::Group(ref mut t) => t.set_span(span), + TokenTree::Ident(ref mut t) => t.set_span(span), + TokenTree::Punct(ref mut t) => t.set_span(span), + TokenTree::Literal(ref mut t) => t.set_span(span), + } + } +} + +/// Prints token tree in a form convenient for debugging. +impl fmt::Debug for TokenTree { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + // Each of these has the name in the struct type in the derived debug, + // so don't bother with an extra layer of indirection + match *self { + TokenTree::Group(ref tt) => tt.fmt(f), + TokenTree::Ident(ref tt) => tt.fmt(f), + TokenTree::Punct(ref tt) => tt.fmt(f), + TokenTree::Literal(ref tt) => tt.fmt(f), + } + } +} + +impl From for TokenTree { + fn from(g: Group) -> TokenTree { + TokenTree::Group(g) + } +} + +impl From for TokenTree { + fn from(g: Ident) -> TokenTree { + TokenTree::Ident(g) + } +} + +impl From for TokenTree { + fn from(g: Punct) -> TokenTree { + TokenTree::Punct(g) + } +} + +impl From for TokenTree { + fn from(g: Literal) -> TokenTree { + TokenTree::Literal(g) + } +} + +/// Prints the token tree as a string that is supposed to be losslessly convertible back +/// into the same token tree (modulo spans), except for possibly `TokenTree::Group`s +/// with `Delimiter::None` delimiters and negative numeric literals. +impl fmt::Display for TokenTree { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.write_str(&self.to_string()) + } +} + +/// A delimited token stream. +/// +/// A `Group` internally contains a `TokenStream` which is surrounded by `Delimiter`s. +#[derive(Clone)] +pub struct Group(bridge::Group); + +/// Describes how a sequence of token trees is delimited. +#[derive(Copy, Clone, Debug, PartialEq, Eq)] +pub enum Delimiter { + /// `( ... )` + Parenthesis, + /// `{ ... }` + Brace, + /// `[ ... ]` + Bracket, + /// `Ø ... Ø` + /// An invisible delimiter, that may, for example, appear around tokens coming from a + /// "macro variable" `$var`. It is important to preserve operator priorities in cases like + /// `$var * 3` where `$var` is `1 + 2`. + /// Invisible delimiters might not survive roundtrip of a token stream through a string. + None, +} + +impl Group { + /// Creates a new `Group` with the given delimiter and token stream. + /// + /// This constructor will set the span for this group to + /// `Span::call_site()`. To change the span you can use the `set_span` + /// method below. + pub fn new(delimiter: Delimiter, stream: TokenStream) -> Group { + Group(bridge::Group { + delimiter, + stream: stream.0, + span: bridge::DelimSpan::from_single(Span::call_site().0), + }) + } + + /// Returns the delimiter of this `Group` + pub fn delimiter(&self) -> Delimiter { + self.0.delimiter + } + + /// Returns the `TokenStream` of tokens that are delimited in this `Group`. + /// + /// Note that the returned token stream does not include the delimiter + /// returned above. + pub fn stream(&self) -> TokenStream { + TokenStream(self.0.stream.clone()) + } + + /// Returns the span for the delimiters of this token stream, spanning the + /// entire `Group`. + /// + /// ```text + /// pub fn span(&self) -> Span { + /// ^^^^^^^ + /// ``` + pub fn span(&self) -> Span { + Span(self.0.span.entire) + } + + /// Returns the span pointing to the opening delimiter of this group. + /// + /// ```text + /// pub fn span_open(&self) -> Span { + /// ^ + /// ``` + pub fn span_open(&self) -> Span { + Span(self.0.span.open) + } + + /// Returns the span pointing to the closing delimiter of this group. + /// + /// ```text + /// pub fn span_close(&self) -> Span { + /// ^ + /// ``` + pub fn span_close(&self) -> Span { + Span(self.0.span.close) + } + + /// Configures the span for this `Group`'s delimiters, but not its internal + /// tokens. + /// + /// This method will **not** set the span of all the internal tokens spanned + /// by this group, but rather it will only set the span of the delimiter + /// tokens at the level of the `Group`. + pub fn set_span(&mut self, span: Span) { + self.0.span = bridge::DelimSpan::from_single(span.0); + } +} + +/// Prints the group as a string that should be losslessly convertible back +/// into the same group (modulo spans), except for possibly `TokenTree::Group`s +/// with `Delimiter::None` delimiters. +impl fmt::Display for Group { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.write_str(&self.to_string()) + } +} + +impl fmt::Debug for Group { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.debug_struct("Group") + .field("delimiter", &self.delimiter()) + .field("stream", &self.stream()) + .field("span", &self.span()) + .finish() + } +} + +/// A `Punct` is a single punctuation character such as `+`, `-` or `#`. +/// +/// Multi-character operators like `+=` are represented as two instances of `Punct` with different +/// forms of `Spacing` returned. +#[derive(Clone)] +pub struct Punct(bridge::Punct); + +/// Describes whether a `Punct` is followed immediately by another `Punct` ([`Spacing::Joint`]) or +/// by a different token or whitespace ([`Spacing::Alone`]). +#[derive(Copy, Clone, Debug, PartialEq, Eq)] +pub enum Spacing { + /// A `Punct` is not immediately followed by another `Punct`. + /// E.g. `+` is `Alone` in `+ =`, `+ident` and `+()`. + Alone, + /// A `Punct` is immediately followed by another `Punct`. + /// E.g. `+` is `Joint` in `+=` and `++`. + /// + /// Additionally, single quote `'` can join with identifiers to form lifetimes: `'ident`. + Joint, +} + +impl Punct { + /// Creates a new `Punct` from the given character and spacing. + /// The `ch` argument must be a valid punctuation character permitted by the language, + /// otherwise the function will panic. + /// + /// The returned `Punct` will have the default span of `Span::call_site()` + /// which can be further configured with the `set_span` method below. + pub fn new(ch: char, spacing: Spacing) -> Punct { + const LEGAL_CHARS: &[char] = &[ + '=', '<', '>', '!', '~', '+', '-', '*', '/', '%', '^', '&', '|', '@', '.', ',', ';', + ':', '#', '$', '?', '\'', + ]; + if !LEGAL_CHARS.contains(&ch) { + panic!("unsupported character `{:?}`", ch); + } + Punct(bridge::Punct { + ch: ch as u8, + joint: spacing == Spacing::Joint, + span: Span::call_site().0, + }) + } + + /// Returns the value of this punctuation character as `char`. + pub fn as_char(&self) -> char { + self.0.ch as char + } + + /// Returns the spacing of this punctuation character, indicating whether it's immediately + /// followed by another `Punct` in the token stream, so they can potentially be combined into + /// a multi-character operator (`Joint`), or it's followed by some other token or whitespace + /// (`Alone`) so the operator has certainly ended. + pub fn spacing(&self) -> Spacing { + if self.0.joint { + Spacing::Joint + } else { + Spacing::Alone + } + } + + /// Returns the span for this punctuation character. + pub fn span(&self) -> Span { + Span(self.0.span) + } + + /// Configure the span for this punctuation character. + pub fn set_span(&mut self, span: Span) { + self.0.span = span.0; + } +} + +/// Prints the punctuation character as a string that should be losslessly convertible +/// back into the same character. +impl fmt::Display for Punct { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.write_str(&self.to_string()) + } +} + +impl fmt::Debug for Punct { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.debug_struct("Punct") + .field("ch", &self.as_char()) + .field("spacing", &self.spacing()) + .field("span", &self.span()) + .finish() + } +} + +impl PartialEq for Punct { + fn eq(&self, rhs: &char) -> bool { + self.as_char() == *rhs + } +} + +impl PartialEq for char { + fn eq(&self, rhs: &Punct) -> bool { + *self == rhs.as_char() + } +} + +/// An identifier (`ident`). +#[derive(Clone)] +pub struct Ident(bridge::client::Ident); + +impl Ident { + /// Creates a new `Ident` with the given `string` as well as the specified + /// `span`. + /// The `string` argument must be a valid identifier permitted by the + /// language (including keywords, e.g. `self` or `fn`). Otherwise, the function will panic. + /// + /// Note that `span`, currently in rustc, configures the hygiene information + /// for this identifier. + /// + /// As of this time `Span::call_site()` explicitly opts-in to "call-site" hygiene + /// meaning that identifiers created with this span will be resolved as if they were written + /// directly at the location of the macro call, and other code at the macro call site will be + /// able to refer to them as well. + /// + /// Later spans like `Span::def_site()` will allow to opt-in to "definition-site" hygiene + /// meaning that identifiers created with this span will be resolved at the location of the + /// macro definition and other code at the macro call site will not be able to refer to them. + /// + /// Due to the current importance of hygiene this constructor, unlike other + /// tokens, requires a `Span` to be specified at construction. + pub fn new(string: &str, span: Span) -> Ident { + Ident(bridge::client::Ident::new(string, span.0, false)) + } + + /// Same as `Ident::new`, but creates a raw identifier (`r#ident`). + /// The `string` argument be a valid identifier permitted by the language + /// (including keywords, e.g. `fn`). Keywords which are usable in path segments + /// (e.g. `self`, `super`) are not supported, and will cause a panic. + pub fn new_raw(string: &str, span: Span) -> Ident { + Ident(bridge::client::Ident::new(string, span.0, true)) + } + + /// Returns the span of this `Ident`, encompassing the entire string returned + /// by [`to_string`](Self::to_string). + pub fn span(&self) -> Span { + Span(self.0.span()) + } + + /// Configures the span of this `Ident`, possibly changing its hygiene context. + pub fn set_span(&mut self, span: Span) { + self.0 = self.0.with_span(span.0); + } +} + +/// Prints the identifier as a string that should be losslessly convertible +/// back into the same identifier. +impl fmt::Display for Ident { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.write_str(&self.to_string()) + } +} + +impl fmt::Debug for Ident { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.debug_struct("Ident") + .field("ident", &self.to_string()) + .field("span", &self.span()) + .finish() + } +} + +/// A literal string (`"hello"`), byte string (`b"hello"`), +/// character (`'a'`), byte character (`b'a'`), an integer or floating point number +/// with or without a suffix (`1`, `1u8`, `2.3`, `2.3f32`). +/// Boolean literals like `true` and `false` do not belong here, they are `Ident`s. +#[derive(Clone)] +pub struct Literal(bridge::client::Literal); + +macro_rules! suffixed_int_literals { + ($($name:ident => $kind:ident,)*) => ($( + /// Creates a new suffixed integer literal with the specified value. + /// + /// This function will create an integer like `1u32` where the integer + /// value specified is the first part of the token and the integral is + /// also suffixed at the end. + /// Literals created from negative numbers might not survive round-trips through + /// `TokenStream` or strings and may be broken into two tokens (`-` and positive literal). + /// + /// Literals created through this method have the `Span::call_site()` + /// span by default, which can be configured with the `set_span` method + /// below. + pub fn $name(n: $kind) -> Literal { + Literal(bridge::client::Literal::typed_integer(&n.to_string(), stringify!($kind))) + } + )*) +} + +macro_rules! unsuffixed_int_literals { + ($($name:ident => $kind:ident,)*) => ($( + /// Creates a new unsuffixed integer literal with the specified value. + /// + /// This function will create an integer like `1` where the integer + /// value specified is the first part of the token. No suffix is + /// specified on this token, meaning that invocations like + /// `Literal::i8_unsuffixed(1)` are equivalent to + /// `Literal::u32_unsuffixed(1)`. + /// Literals created from negative numbers might not survive rountrips through + /// `TokenStream` or strings and may be broken into two tokens (`-` and positive literal). + /// + /// Literals created through this method have the `Span::call_site()` + /// span by default, which can be configured with the `set_span` method + /// below. + pub fn $name(n: $kind) -> Literal { + Literal(bridge::client::Literal::integer(&n.to_string())) + } + )*) +} + +impl Literal { + suffixed_int_literals! { + u8_suffixed => u8, + u16_suffixed => u16, + u32_suffixed => u32, + u64_suffixed => u64, + u128_suffixed => u128, + usize_suffixed => usize, + i8_suffixed => i8, + i16_suffixed => i16, + i32_suffixed => i32, + i64_suffixed => i64, + i128_suffixed => i128, + isize_suffixed => isize, + } + + unsuffixed_int_literals! { + u8_unsuffixed => u8, + u16_unsuffixed => u16, + u32_unsuffixed => u32, + u64_unsuffixed => u64, + u128_unsuffixed => u128, + usize_unsuffixed => usize, + i8_unsuffixed => i8, + i16_unsuffixed => i16, + i32_unsuffixed => i32, + i64_unsuffixed => i64, + i128_unsuffixed => i128, + isize_unsuffixed => isize, + } + + /// Creates a new unsuffixed floating-point literal. + /// + /// This constructor is similar to those like `Literal::i8_unsuffixed` where + /// the float's value is emitted directly into the token but no suffix is + /// used, so it may be inferred to be a `f64` later in the compiler. + /// Literals created from negative numbers might not survive rountrips through + /// `TokenStream` or strings and may be broken into two tokens (`-` and positive literal). + /// + /// # Panics + /// + /// This function requires that the specified float is finite, for + /// example if it is infinity or NaN this function will panic. + pub fn f32_unsuffixed(n: f32) -> Literal { + if !n.is_finite() { + panic!("Invalid float literal {n}"); + } + let mut repr = n.to_string(); + if !repr.contains('.') { + repr.push_str(".0"); + } + Literal(bridge::client::Literal::float(&repr)) + } + + /// Creates a new suffixed floating-point literal. + /// + /// This constructor will create a literal like `1.0f32` where the value + /// specified is the preceding part of the token and `f32` is the suffix of + /// the token. This token will always be inferred to be an `f32` in the + /// compiler. + /// Literals created from negative numbers might not survive rountrips through + /// `TokenStream` or strings and may be broken into two tokens (`-` and positive literal). + /// + /// # Panics + /// + /// This function requires that the specified float is finite, for + /// example if it is infinity or NaN this function will panic. + pub fn f32_suffixed(n: f32) -> Literal { + if !n.is_finite() { + panic!("Invalid float literal {n}"); + } + Literal(bridge::client::Literal::f32(&n.to_string())) + } + + /// Creates a new unsuffixed floating-point literal. + /// + /// This constructor is similar to those like `Literal::i8_unsuffixed` where + /// the float's value is emitted directly into the token but no suffix is + /// used, so it may be inferred to be a `f64` later in the compiler. + /// Literals created from negative numbers might not survive rountrips through + /// `TokenStream` or strings and may be broken into two tokens (`-` and positive literal). + /// + /// # Panics + /// + /// This function requires that the specified float is finite, for + /// example if it is infinity or NaN this function will panic. + pub fn f64_unsuffixed(n: f64) -> Literal { + if !n.is_finite() { + panic!("Invalid float literal {n}"); + } + let mut repr = n.to_string(); + if !repr.contains('.') { + repr.push_str(".0"); + } + Literal(bridge::client::Literal::float(&repr)) + } + + /// Creates a new suffixed floating-point literal. + /// + /// This constructor will create a literal like `1.0f64` where the value + /// specified is the preceding part of the token and `f64` is the suffix of + /// the token. This token will always be inferred to be an `f64` in the + /// compiler. + /// Literals created from negative numbers might not survive rountrips through + /// `TokenStream` or strings and may be broken into two tokens (`-` and positive literal). + /// + /// # Panics + /// + /// This function requires that the specified float is finite, for + /// example if it is infinity or NaN this function will panic. + pub fn f64_suffixed(n: f64) -> Literal { + if !n.is_finite() { + panic!("Invalid float literal {n}"); + } + Literal(bridge::client::Literal::f64(&n.to_string())) + } + + /// String literal. + pub fn string(string: &str) -> Literal { + Literal(bridge::client::Literal::string(string)) + } + + /// Character literal. + pub fn character(ch: char) -> Literal { + Literal(bridge::client::Literal::character(ch)) + } + + /// Byte string literal. + pub fn byte_string(bytes: &[u8]) -> Literal { + Literal(bridge::client::Literal::byte_string(bytes)) + } + + /// Returns the span encompassing this literal. + pub fn span(&self) -> Span { + Span(self.0.span()) + } + + /// Configures the span associated for this literal. + pub fn set_span(&mut self, span: Span) { + self.0.set_span(span.0); + } + + /// Returns a `Span` that is a subset of `self.span()` containing only the + /// source bytes in range `range`. Returns `None` if the would-be trimmed + /// span is outside the bounds of `self`. + // FIXME(SergioBenitez): check that the byte range starts and ends at a + // UTF-8 boundary of the source. otherwise, it's likely that a panic will + // occur elsewhere when the source text is printed. + // FIXME(SergioBenitez): there is no way for the user to know what + // `self.span()` actually maps to, so this method can currently only be + // called blindly. For example, `to_string()` for the character 'c' returns + // "'\u{63}'"; there is no way for the user to know whether the source text + // was 'c' or whether it was '\u{63}'. + pub fn subspan>(&self, range: R) -> Option { + self.0.subspan(range.start_bound().cloned(), range.end_bound().cloned()).map(Span) + } +} + +/// Parse a single literal from its stringified representation. +/// +/// In order to parse successfully, the input string must not contain anything +/// but the literal token. Specifically, it must not contain whitespace or +/// comments in addition to the literal. +/// +/// The resulting literal token will have a `Span::call_site()` span. +/// +/// NOTE: some errors may cause panics instead of returning `LexError`. We +/// reserve the right to change these errors into `LexError`s later. +impl FromStr for Literal { + type Err = LexError; + + fn from_str(src: &str) -> Result { + match bridge::client::Literal::from_str(src) { + Ok(literal) => Ok(Literal(literal)), + Err(()) => Err(LexError), + } + } +} + +/// Prints the literal as a string that should be losslessly convertible +/// back into the same literal (except for possible rounding for floating point literals). +impl fmt::Display for Literal { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + f.write_str(&self.to_string()) + } +} + +impl fmt::Debug for Literal { + fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { + self.0.fmt(f) + } +} + +/// Tracked access to environment variables. +pub mod tracked_env { + use std::env::{self, VarError}; + use std::ffi::OsStr; + + /// Retrieve an environment variable and add it to build dependency info. + /// Build system executing the compiler will know that the variable was accessed during + /// compilation, and will be able to rerun the build when the value of that variable changes. + /// Besides the dependency tracking this function should be equivalent to `env::var` from the + /// standard library, except that the argument must be UTF-8. + pub fn var + AsRef>(key: K) -> Result { + let key: &str = key.as_ref(); + let value = env::var(key); + super::bridge::client::FreeFunctions::track_env_var(key, value.as_deref().ok()); + value + } +} + +/// Tracked access to additional files. +pub mod tracked_path { + + /// Track a file explicitly. + /// + /// Commonly used for tracking asset preprocessing. + pub fn path>(path: P) { + let path: &str = path.as_ref(); + super::bridge::client::FreeFunctions::track_path(path); + } +} diff --git a/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/quote.rs b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/quote.rs new file mode 100644 index 00000000000..39309faa412 --- /dev/null +++ b/crates/proc-macro-srv/src/abis/abi_1_64/proc_macro/quote.rs @@ -0,0 +1,139 @@ +//! # Quasiquoter +//! This file contains the implementation internals of the quasiquoter provided by `quote!`. + +//! This quasiquoter uses macros 2.0 hygiene to reliably access +//! items from `proc_macro`, to build a `proc_macro::TokenStream`. + +use super::{Delimiter, Group, Ident, Literal, Punct, Spacing, Span, TokenStream, TokenTree}; + +macro_rules! quote_tt { + (($($t:tt)*)) => { Group::new(Delimiter::Parenthesis, quote!($($t)*)) }; + ([$($t:tt)*]) => { Group::new(Delimiter::Bracket, quote!($($t)*)) }; + ({$($t:tt)*}) => { Group::new(Delimiter::Brace, quote!($($t)*)) }; + (,) => { Punct::new(',', Spacing::Alone) }; + (.) => { Punct::new('.', Spacing::Alone) }; + (;) => { Punct::new(';', Spacing::Alone) }; + (!) => { Punct::new('!', Spacing::Alone) }; + (<) => { Punct::new('<', Spacing::Alone) }; + (>) => { Punct::new('>', Spacing::Alone) }; + (&) => { Punct::new('&', Spacing::Alone) }; + (=) => { Punct::new('=', Spacing::Alone) }; + ($i:ident) => { Ident::new(stringify!($i), Span::def_site()) }; +} + +macro_rules! quote_ts { + ((@ $($t:tt)*)) => { $($t)* }; + (::) => { + [ + TokenTree::from(Punct::new(':', Spacing::Joint)), + TokenTree::from(Punct::new(':', Spacing::Alone)), + ].iter() + .cloned() + .map(|mut x| { + x.set_span(Span::def_site()); + x + }) + .collect::() + }; + ($t:tt) => { TokenTree::from(quote_tt!($t)) }; +} + +/// Simpler version of the real `quote!` macro, implemented solely +/// through `macro_rules`, for bootstrapping the real implementation +/// (see the `quote` function), which does not have access to the +/// real `quote!` macro due to the `proc_macro` crate not being +/// able to depend on itself. +/// +/// Note: supported tokens are a subset of the real `quote!`, but +/// unquoting is different: instead of `$x`, this uses `(@ expr)`. +macro_rules! quote { + () => { TokenStream::new() }; + ($($t:tt)*) => { + [ + $(TokenStream::from(quote_ts!($t)),)* + ].iter().cloned().collect::() + }; +} + +/// Quote a `TokenStream` into a `TokenStream`. +/// This is the actual implementation of the `quote!()` proc macro. +/// +/// It is loaded by the compiler in `register_builtin_macros`. +pub fn quote(stream: TokenStream) -> TokenStream { + if stream.is_empty() { + return quote!(super::TokenStream::new()); + } + let proc_macro_crate = quote!(crate); + let mut after_dollar = false; + let tokens = stream + .into_iter() + .filter_map(|tree| { + if after_dollar { + after_dollar = false; + match tree { + TokenTree::Ident(_) => { + return Some(quote!(Into::::into( + Clone::clone(&(@ tree))),)); + } + TokenTree::Punct(ref tt) if tt.as_char() == '$' => {} + _ => panic!("`$` must be followed by an ident or `$` in `quote!`"), + } + } else if let TokenTree::Punct(ref tt) = tree { + if tt.as_char() == '$' { + after_dollar = true; + return None; + } + } + + Some(quote!(super::TokenStream::from((@ match tree { + TokenTree::Punct(tt) => quote!(super::TokenTree::Punct(super::Punct::new( + (@ TokenTree::from(Literal::character(tt.as_char()))), + (@ match tt.spacing() { + Spacing::Alone => quote!(super::Spacing::Alone), + Spacing::Joint => quote!(super::Spacing::Joint), + }), + ))), + TokenTree::Group(tt) => quote!(super::TokenTree::Group(super::Group::new( + (@ match tt.delimiter() { + Delimiter::Parenthesis => quote!(super::Delimiter::Parenthesis), + Delimiter::Brace => quote!(super::Delimiter::Brace), + Delimiter::Bracket => quote!(super::Delimiter::Bracket), + Delimiter::None => quote!(super::Delimiter::None), + }), + (@ quote(tt.stream())), + ))), + TokenTree::Ident(tt) => quote!(super::TokenTree::Ident(super::Ident::new( + (@ TokenTree::from(Literal::string(&tt.to_string()))), + (@ quote_span(proc_macro_crate.clone(), tt.span())), + ))), + TokenTree::Literal(tt) => quote!(super::TokenTree::Literal({ + let mut iter = (@ TokenTree::from(Literal::string(&tt.to_string()))) + .parse::() + .unwrap() + .into_iter(); + if let (Some(super::TokenTree::Literal(mut lit)), None) = + (iter.next(), iter.next()) + { + lit.set_span((@ quote_span(proc_macro_crate.clone(), tt.span()))); + lit + } else { + unreachable!() + } + })) + })),)) + }) + .collect::(); + + if after_dollar { + panic!("unexpected trailing `$` in `quote!`"); + } + + quote!([(@ tokens)].iter().cloned().collect::()) +} + +/// Quote a `Span` into a `TokenStream`. +/// This is needed to implement a custom quoter. +pub fn quote_span(proc_macro_crate: TokenStream, span: Span) -> TokenStream { + let id = span.save_span(); + quote!((@ proc_macro_crate ) ::Span::recover_proc_macro_span((@ TokenTree::from(Literal::usize_unsuffixed(id))))) +} diff --git a/crates/proc-macro-srv/src/abis/abi_1_64/rustc_server.rs b/crates/proc-macro-srv/src/abis/abi_1_64/rustc_server.rs new file mode 100644 index 00000000000..757454a9262 --- /dev/null +++ b/crates/proc-macro-srv/src/abis/abi_1_64/rustc_server.rs @@ -0,0 +1,792 @@ +//! Rustc proc-macro server implementation with tt +//! +//! Based on idea from +//! The lib-proc-macro server backend is `TokenStream`-agnostic, such that +//! we could provide any TokenStream implementation. +//! The original idea from fedochet is using proc-macro2 as backend, +//! we use tt instead for better integration with RA. +//! +//! FIXME: No span and source file information is implemented yet + +use super::proc_macro::bridge::{self, server}; + +use std::collections::HashMap; +use std::hash::Hash; +use std::iter::FromIterator; +use std::ops::Bound; +use std::{ascii, vec::IntoIter}; + +type Group = tt::Subtree; +type TokenTree = tt::TokenTree; +type Punct = tt::Punct; +type Spacing = tt::Spacing; +type Literal = tt::Literal; +type Span = tt::TokenId; + +#[derive(Debug, Default, Clone)] +pub struct TokenStream { + pub token_trees: Vec, +} + +impl TokenStream { + pub fn new() -> Self { + TokenStream::default() + } + + pub fn with_subtree(subtree: tt::Subtree) -> Self { + if subtree.delimiter.is_some() { + TokenStream { token_trees: vec![TokenTree::Subtree(subtree)] } + } else { + TokenStream { token_trees: subtree.token_trees } + } + } + + pub fn into_subtree(self) -> tt::Subtree { + tt::Subtree { delimiter: None, token_trees: self.token_trees } + } + + pub fn is_empty(&self) -> bool { + self.token_trees.is_empty() + } +} + +/// Creates a token stream containing a single token tree. +impl From for TokenStream { + fn from(tree: TokenTree) -> TokenStream { + TokenStream { token_trees: vec![tree] } + } +} + +/// Collects a number of token trees into a single stream. +impl FromIterator for TokenStream { + fn from_iter>(trees: I) -> Self { + trees.into_iter().map(TokenStream::from).collect() + } +} + +/// A "flattening" operation on token streams, collects token trees +/// from multiple token streams into a single stream. +impl FromIterator for TokenStream { + fn from_iter>(streams: I) -> Self { + let mut builder = TokenStreamBuilder::new(); + streams.into_iter().for_each(|stream| builder.push(stream)); + builder.build() + } +} + +impl Extend for TokenStream { + fn extend>(&mut self, trees: I) { + self.extend(trees.into_iter().map(TokenStream::from)); + } +} + +impl Extend for TokenStream { + fn extend>(&mut self, streams: I) { + for item in streams { + for tkn in item { + match tkn { + tt::TokenTree::Subtree(subtree) if subtree.delimiter.is_none() => { + self.token_trees.extend(subtree.token_trees); + } + _ => { + self.token_trees.push(tkn); + } + } + } + } + } +} + +#[derive(Clone)] +pub struct SourceFile { + // FIXME stub +} + +type Level = super::proc_macro::Level; +type LineColumn = super::proc_macro::LineColumn; + +/// A structure representing a diagnostic message and associated children +/// messages. +#[derive(Clone, Debug)] +pub struct Diagnostic { + level: Level, + message: String, + spans: Vec, + children: Vec, +} + +impl Diagnostic { + /// Creates a new diagnostic with the given `level` and `message`. + pub fn new>(level: Level, message: T) -> Diagnostic { + Diagnostic { level, message: message.into(), spans: vec![], children: vec![] } + } +} + +// Rustc Server Ident has to be `Copyable` +// We use a stub here for bypassing +#[derive(Hash, Eq, PartialEq, Copy, Clone)] +pub struct IdentId(u32); + +#[derive(Clone, Hash, Eq, PartialEq)] +struct IdentData(tt::Ident); + +#[derive(Default)] +struct IdentInterner { + idents: HashMap, + ident_data: Vec, +} + +impl IdentInterner { + fn intern(&mut self, data: &IdentData) -> u32 { + if let Some(index) = self.idents.get(data) { + return *index; + } + + let index = self.idents.len() as u32; + self.ident_data.push(data.clone()); + self.idents.insert(data.clone(), index); + index + } + + fn get(&self, index: u32) -> &IdentData { + &self.ident_data[index as usize] + } + + #[allow(unused)] + fn get_mut(&mut self, index: u32) -> &mut IdentData { + self.ident_data.get_mut(index as usize).expect("Should be consistent") + } +} + +pub struct TokenStreamBuilder { + acc: TokenStream, +} + +/// Public implementation details for the `TokenStream` type, such as iterators. +pub mod token_stream { + use std::str::FromStr; + + use super::{TokenStream, TokenTree}; + + /// An iterator over `TokenStream`'s `TokenTree`s. + /// The iteration is "shallow", e.g., the iterator doesn't recurse into delimited groups, + /// and returns whole groups as token trees. + impl IntoIterator for TokenStream { + type Item = TokenTree; + type IntoIter = super::IntoIter; + + fn into_iter(self) -> Self::IntoIter { + self.token_trees.into_iter() + } + } + + type LexError = String; + + /// Attempts to break the string into tokens and parse those tokens into a token stream. + /// May fail for a number of reasons, for example, if the string contains unbalanced delimiters + /// or characters not existing in the language. + /// All tokens in the parsed stream get `Span::call_site()` spans. + /// + /// NOTE: some errors may cause panics instead of returning `LexError`. We reserve the right to + /// change these errors into `LexError`s later. + impl FromStr for TokenStream { + type Err = LexError; + + fn from_str(src: &str) -> Result { + let (subtree, _token_map) = + mbe::parse_to_token_tree(src).ok_or("Failed to parse from mbe")?; + + let subtree = subtree_replace_token_ids_with_unspecified(subtree); + Ok(TokenStream::with_subtree(subtree)) + } + } + + impl ToString for TokenStream { + fn to_string(&self) -> String { + tt::pretty(&self.token_trees) + } + } + + fn subtree_replace_token_ids_with_unspecified(subtree: tt::Subtree) -> tt::Subtree { + tt::Subtree { + delimiter: subtree + .delimiter + .map(|d| tt::Delimiter { id: tt::TokenId::unspecified(), ..d }), + token_trees: subtree + .token_trees + .into_iter() + .map(token_tree_replace_token_ids_with_unspecified) + .collect(), + } + } + + fn token_tree_replace_token_ids_with_unspecified(tt: tt::TokenTree) -> tt::TokenTree { + match tt { + tt::TokenTree::Leaf(leaf) => { + tt::TokenTree::Leaf(leaf_replace_token_ids_with_unspecified(leaf)) + } + tt::TokenTree::Subtree(subtree) => { + tt::TokenTree::Subtree(subtree_replace_token_ids_with_unspecified(subtree)) + } + } + } + + fn leaf_replace_token_ids_with_unspecified(leaf: tt::Leaf) -> tt::Leaf { + match leaf { + tt::Leaf::Literal(lit) => { + tt::Leaf::Literal(tt::Literal { id: tt::TokenId::unspecified(), ..lit }) + } + tt::Leaf::Punct(punct) => { + tt::Leaf::Punct(tt::Punct { id: tt::TokenId::unspecified(), ..punct }) + } + tt::Leaf::Ident(ident) => { + tt::Leaf::Ident(tt::Ident { id: tt::TokenId::unspecified(), ..ident }) + } + } + } +} + +impl TokenStreamBuilder { + fn new() -> TokenStreamBuilder { + TokenStreamBuilder { acc: TokenStream::new() } + } + + fn push(&mut self, stream: TokenStream) { + self.acc.extend(stream.into_iter()) + } + + fn build(self) -> TokenStream { + self.acc + } +} + +pub struct FreeFunctions; + +#[derive(Clone)] +pub struct TokenStreamIter { + trees: IntoIter, +} + +#[derive(Default)] +pub struct Rustc { + ident_interner: IdentInterner, + // FIXME: store span information here. +} + +impl server::Types for Rustc { + type FreeFunctions = FreeFunctions; + type TokenStream = TokenStream; + type Ident = IdentId; + type Literal = Literal; + type SourceFile = SourceFile; + type Diagnostic = Diagnostic; + type Span = Span; + type MultiSpan = Vec; +} + +impl server::FreeFunctions for Rustc { + fn track_env_var(&mut self, _var: &str, _value: Option<&str>) { + // FIXME: track env var accesses + // https://github.com/rust-lang/rust/pull/71858 + } + fn track_path(&mut self, _path: &str) {} +} + +impl server::TokenStream for Rustc { + fn is_empty(&mut self, stream: &Self::TokenStream) -> bool { + stream.is_empty() + } + fn from_str(&mut self, src: &str) -> Self::TokenStream { + use std::str::FromStr; + + Self::TokenStream::from_str(src).expect("cannot parse string") + } + fn to_string(&mut self, stream: &Self::TokenStream) -> String { + stream.to_string() + } + fn from_token_tree( + &mut self, + tree: bridge::TokenTree, + ) -> Self::TokenStream { + match tree { + bridge::TokenTree::Group(group) => { + let group = Group { + delimiter: delim_to_internal(group.delimiter), + token_trees: match group.stream { + Some(stream) => stream.into_iter().collect(), + None => Vec::new(), + }, + }; + let tree = TokenTree::from(group); + Self::TokenStream::from_iter(vec![tree]) + } + + bridge::TokenTree::Ident(IdentId(index)) => { + let IdentData(ident) = self.ident_interner.get(index).clone(); + let ident: tt::Ident = ident; + let leaf = tt::Leaf::from(ident); + let tree = TokenTree::from(leaf); + Self::TokenStream::from_iter(vec![tree]) + } + + bridge::TokenTree::Literal(literal) => { + let leaf = tt::Leaf::from(literal); + let tree = TokenTree::from(leaf); + Self::TokenStream::from_iter(vec![tree]) + } + + bridge::TokenTree::Punct(p) => { + let punct = tt::Punct { + char: p.ch as char, + spacing: if p.joint { Spacing::Joint } else { Spacing::Alone }, + id: p.span, + }; + let leaf = tt::Leaf::from(punct); + let tree = TokenTree::from(leaf); + Self::TokenStream::from_iter(vec![tree]) + } + } + } + + fn expand_expr(&mut self, self_: &Self::TokenStream) -> Result { + Ok(self_.clone()) + } + + fn concat_trees( + &mut self, + base: Option, + trees: Vec>, + ) -> Self::TokenStream { + let mut builder = TokenStreamBuilder::new(); + if let Some(base) = base { + builder.push(base); + } + for tree in trees { + builder.push(self.from_token_tree(tree)); + } + builder.build() + } + + fn concat_streams( + &mut self, + base: Option, + streams: Vec, + ) -> Self::TokenStream { + let mut builder = TokenStreamBuilder::new(); + if let Some(base) = base { + builder.push(base); + } + for stream in streams { + builder.push(stream); + } + builder.build() + } + + fn into_trees( + &mut self, + stream: Self::TokenStream, + ) -> Vec> { + stream + .into_iter() + .map(|tree| match tree { + tt::TokenTree::Leaf(tt::Leaf::Ident(ident)) => { + bridge::TokenTree::Ident(IdentId(self.ident_interner.intern(&IdentData(ident)))) + } + tt::TokenTree::Leaf(tt::Leaf::Literal(lit)) => bridge::TokenTree::Literal(lit), + tt::TokenTree::Leaf(tt::Leaf::Punct(punct)) => { + bridge::TokenTree::Punct(bridge::Punct { + ch: punct.char as u8, + joint: punct.spacing == Spacing::Joint, + span: punct.id, + }) + } + tt::TokenTree::Subtree(subtree) => bridge::TokenTree::Group(bridge::Group { + delimiter: delim_to_external(subtree.delimiter), + stream: if subtree.token_trees.is_empty() { + None + } else { + Some(subtree.token_trees.into_iter().collect()) + }, + span: bridge::DelimSpan::from_single( + subtree.delimiter.map_or(Span::unspecified(), |del| del.id), + ), + }), + }) + .collect() + } +} + +fn delim_to_internal(d: bridge::Delimiter) -> Option { + let kind = match d { + bridge::Delimiter::Parenthesis => tt::DelimiterKind::Parenthesis, + bridge::Delimiter::Brace => tt::DelimiterKind::Brace, + bridge::Delimiter::Bracket => tt::DelimiterKind::Bracket, + bridge::Delimiter::None => return None, + }; + Some(tt::Delimiter { id: tt::TokenId::unspecified(), kind }) +} + +fn delim_to_external(d: Option) -> bridge::Delimiter { + match d.map(|it| it.kind) { + Some(tt::DelimiterKind::Parenthesis) => bridge::Delimiter::Parenthesis, + Some(tt::DelimiterKind::Brace) => bridge::Delimiter::Brace, + Some(tt::DelimiterKind::Bracket) => bridge::Delimiter::Bracket, + None => bridge::Delimiter::None, + } +} + +fn spacing_to_internal(spacing: bridge::Spacing) -> Spacing { + match spacing { + bridge::Spacing::Alone => Spacing::Alone, + bridge::Spacing::Joint => Spacing::Joint, + } +} + +fn spacing_to_external(spacing: Spacing) -> bridge::Spacing { + match spacing { + Spacing::Alone => bridge::Spacing::Alone, + Spacing::Joint => bridge::Spacing::Joint, + } +} + +impl server::Ident for Rustc { + fn new(&mut self, string: &str, span: Self::Span, _is_raw: bool) -> Self::Ident { + IdentId(self.ident_interner.intern(&IdentData(tt::Ident { text: string.into(), id: span }))) + } + + fn span(&mut self, ident: Self::Ident) -> Self::Span { + self.ident_interner.get(ident.0).0.id + } + fn with_span(&mut self, ident: Self::Ident, span: Self::Span) -> Self::Ident { + let data = self.ident_interner.get(ident.0); + let new = IdentData(tt::Ident { id: span, ..data.0.clone() }); + IdentId(self.ident_interner.intern(&new)) + } +} + +impl server::Literal for Rustc { + fn debug_kind(&mut self, _literal: &Self::Literal) -> String { + // r-a: debug_kind and suffix are unsupported; corresponding client code has been changed to not call these. + // They must still be present to be ABI-compatible and work with upstream proc_macro. + "".to_owned() + } + fn from_str(&mut self, s: &str) -> Result { + Ok(Literal { text: s.into(), id: tt::TokenId::unspecified() }) + } + fn symbol(&mut self, literal: &Self::Literal) -> String { + literal.text.to_string() + } + fn suffix(&mut self, _literal: &Self::Literal) -> Option { + None + } + + fn to_string(&mut self, literal: &Self::Literal) -> String { + literal.to_string() + } + + fn integer(&mut self, n: &str) -> Self::Literal { + let n = match n.parse::() { + Ok(n) => n.to_string(), + Err(_) => n.parse::().unwrap().to_string(), + }; + Literal { text: n.into(), id: tt::TokenId::unspecified() } + } + + fn typed_integer(&mut self, n: &str, kind: &str) -> Self::Literal { + macro_rules! def_suffixed_integer { + ($kind:ident, $($ty:ty),*) => { + match $kind { + $( + stringify!($ty) => { + let n: $ty = n.parse().unwrap(); + format!(concat!("{}", stringify!($ty)), n) + } + )* + _ => unimplemented!("unknown args for typed_integer: n {}, kind {}", n, $kind), + } + } + } + + let text = def_suffixed_integer! {kind, u8, u16, u32, u64, u128, usize, i8, i16, i32, i64, i128, isize}; + + Literal { text: text.into(), id: tt::TokenId::unspecified() } + } + + fn float(&mut self, n: &str) -> Self::Literal { + let n: f64 = n.parse().unwrap(); + let mut text = f64::to_string(&n); + if !text.contains('.') { + text += ".0" + } + Literal { text: text.into(), id: tt::TokenId::unspecified() } + } + + fn f32(&mut self, n: &str) -> Self::Literal { + let n: f32 = n.parse().unwrap(); + let text = format!("{}f32", n); + Literal { text: text.into(), id: tt::TokenId::unspecified() } + } + + fn f64(&mut self, n: &str) -> Self::Literal { + let n: f64 = n.parse().unwrap(); + let text = format!("{}f64", n); + Literal { text: text.into(), id: tt::TokenId::unspecified() } + } + + fn string(&mut self, string: &str) -> Self::Literal { + let mut escaped = String::new(); + for ch in string.chars() { + escaped.extend(ch.escape_debug()); + } + Literal { text: format!("\"{}\"", escaped).into(), id: tt::TokenId::unspecified() } + } + + fn character(&mut self, ch: char) -> Self::Literal { + Literal { text: format!("'{}'", ch).into(), id: tt::TokenId::unspecified() } + } + + fn byte_string(&mut self, bytes: &[u8]) -> Self::Literal { + let string = bytes + .iter() + .cloned() + .flat_map(ascii::escape_default) + .map(Into::::into) + .collect::(); + + Literal { text: format!("b\"{}\"", string).into(), id: tt::TokenId::unspecified() } + } + + fn span(&mut self, literal: &Self::Literal) -> Self::Span { + literal.id + } + + fn set_span(&mut self, literal: &mut Self::Literal, span: Self::Span) { + literal.id = span; + } + + fn subspan( + &mut self, + _literal: &Self::Literal, + _start: Bound, + _end: Bound, + ) -> Option { + // FIXME handle span + None + } +} + +impl server::SourceFile for Rustc { + // FIXME these are all stubs + fn eq(&mut self, _file1: &Self::SourceFile, _file2: &Self::SourceFile) -> bool { + true + } + fn path(&mut self, _file: &Self::SourceFile) -> String { + String::new() + } + fn is_real(&mut self, _file: &Self::SourceFile) -> bool { + true + } +} + +impl server::Diagnostic for Rustc { + fn new(&mut self, level: Level, msg: &str, spans: Self::MultiSpan) -> Self::Diagnostic { + let mut diag = Diagnostic::new(level, msg); + diag.spans = spans; + diag + } + + fn sub( + &mut self, + _diag: &mut Self::Diagnostic, + _level: Level, + _msg: &str, + _spans: Self::MultiSpan, + ) { + // FIXME handle diagnostic + // + } + + fn emit(&mut self, _diag: Self::Diagnostic) { + // FIXME handle diagnostic + // diag.emit() + } +} + +impl server::Span for Rustc { + fn debug(&mut self, span: Self::Span) -> String { + format!("{:?}", span.0) + } + fn source_file(&mut self, _span: Self::Span) -> Self::SourceFile { + SourceFile {} + } + fn save_span(&mut self, _span: Self::Span) -> usize { + // FIXME stub + 0 + } + fn recover_proc_macro_span(&mut self, _id: usize) -> Self::Span { + // FIXME stub + tt::TokenId::unspecified() + } + /// Recent feature, not yet in the proc_macro + /// + /// See PR: + /// https://github.com/rust-lang/rust/pull/55780 + fn source_text(&mut self, _span: Self::Span) -> Option { + None + } + + fn parent(&mut self, _span: Self::Span) -> Option { + // FIXME handle span + None + } + fn source(&mut self, span: Self::Span) -> Self::Span { + // FIXME handle span + span + } + fn start(&mut self, _span: Self::Span) -> LineColumn { + // FIXME handle span + LineColumn { line: 0, column: 0 } + } + fn end(&mut self, _span: Self::Span) -> LineColumn { + // FIXME handle span + LineColumn { line: 0, column: 0 } + } + fn join(&mut self, first: Self::Span, _second: Self::Span) -> Option { + // Just return the first span again, because some macros will unwrap the result. + Some(first) + } + fn resolved_at(&mut self, _span: Self::Span, _at: Self::Span) -> Self::Span { + // FIXME handle span + tt::TokenId::unspecified() + } + + fn after(&mut self, _self_: Self::Span) -> Self::Span { + tt::TokenId::unspecified() + } + + fn before(&mut self, _self_: Self::Span) -> Self::Span { + tt::TokenId::unspecified() + } +} + +impl server::MultiSpan for Rustc { + fn new(&mut self) -> Self::MultiSpan { + // FIXME handle span + vec![] + } + + fn push(&mut self, other: &mut Self::MultiSpan, span: Self::Span) { + //TODP + other.push(span) + } +} + +impl server::Server for Rustc { + fn globals(&mut self) -> bridge::ExpnGlobals { + bridge::ExpnGlobals { + def_site: Span::unspecified(), + call_site: Span::unspecified(), + mixed_site: Span::unspecified(), + } + } +} + +#[cfg(test)] +mod tests { + use super::super::proc_macro::bridge::server::Literal; + use super::*; + + #[test] + fn test_rustc_server_literals() { + let mut srv = Rustc { ident_interner: IdentInterner::default() }; + assert_eq!(srv.integer("1234").text, "1234"); + + assert_eq!(srv.typed_integer("12", "u8").text, "12u8"); + assert_eq!(srv.typed_integer("255", "u16").text, "255u16"); + assert_eq!(srv.typed_integer("1234", "u32").text, "1234u32"); + assert_eq!(srv.typed_integer("15846685", "u64").text, "15846685u64"); + assert_eq!(srv.typed_integer("15846685258", "u128").text, "15846685258u128"); + assert_eq!(srv.typed_integer("156788984", "usize").text, "156788984usize"); + assert_eq!(srv.typed_integer("127", "i8").text, "127i8"); + assert_eq!(srv.typed_integer("255", "i16").text, "255i16"); + assert_eq!(srv.typed_integer("1234", "i32").text, "1234i32"); + assert_eq!(srv.typed_integer("15846685", "i64").text, "15846685i64"); + assert_eq!(srv.typed_integer("15846685258", "i128").text, "15846685258i128"); + assert_eq!(srv.float("0").text, "0.0"); + assert_eq!(srv.float("15684.5867").text, "15684.5867"); + assert_eq!(srv.f32("15684.58").text, "15684.58f32"); + assert_eq!(srv.f64("15684.58").text, "15684.58f64"); + + assert_eq!(srv.string("hello_world").text, "\"hello_world\""); + assert_eq!(srv.character('c').text, "'c'"); + assert_eq!(srv.byte_string(b"1234586\x88").text, "b\"1234586\\x88\""); + + // u128::max + assert_eq!( + srv.integer("340282366920938463463374607431768211455").text, + "340282366920938463463374607431768211455" + ); + // i128::min + assert_eq!( + srv.integer("-170141183460469231731687303715884105728").text, + "-170141183460469231731687303715884105728" + ); + } + + #[test] + fn test_rustc_server_to_string() { + let s = TokenStream { + token_trees: vec![ + tt::TokenTree::Leaf(tt::Leaf::Ident(tt::Ident { + text: "struct".into(), + id: tt::TokenId::unspecified(), + })), + tt::TokenTree::Leaf(tt::Leaf::Ident(tt::Ident { + text: "T".into(), + id: tt::TokenId::unspecified(), + })), + tt::TokenTree::Subtree(tt::Subtree { + delimiter: Some(tt::Delimiter { + id: tt::TokenId::unspecified(), + kind: tt::DelimiterKind::Brace, + }), + token_trees: vec![], + }), + ], + }; + + assert_eq!(s.to_string(), "struct T {}"); + } + + #[test] + fn test_rustc_server_from_str() { + use std::str::FromStr; + let subtree_paren_a = tt::TokenTree::Subtree(tt::Subtree { + delimiter: Some(tt::Delimiter { + id: tt::TokenId::unspecified(), + kind: tt::DelimiterKind::Parenthesis, + }), + token_trees: vec![tt::TokenTree::Leaf(tt::Leaf::Ident(tt::Ident { + text: "a".into(), + id: tt::TokenId::unspecified(), + }))], + }); + + let t1 = TokenStream::from_str("(a)").unwrap(); + assert_eq!(t1.token_trees.len(), 1); + assert_eq!(t1.token_trees[0], subtree_paren_a); + + let t2 = TokenStream::from_str("(a);").unwrap(); + assert_eq!(t2.token_trees.len(), 2); + assert_eq!(t2.token_trees[0], subtree_paren_a); + + let underscore = TokenStream::from_str("_").unwrap(); + assert_eq!( + underscore.token_trees[0], + tt::TokenTree::Leaf(tt::Leaf::Ident(tt::Ident { + text: "_".into(), + id: tt::TokenId::unspecified(), + })) + ); + } +} diff --git a/crates/proc-macro-srv/src/abis/mod.rs b/crates/proc-macro-srv/src/abis/mod.rs index 815f2e96380..0f3ebe44635 100644 --- a/crates/proc-macro-srv/src/abis/mod.rs +++ b/crates/proc-macro-srv/src/abis/mod.rs @@ -30,6 +30,7 @@ mod abi_1_57; mod abi_1_58; mod abi_1_63; +mod abi_1_64; use super::dylib::LoadProcMacroDylibError; pub(crate) use abi_1_48::Abi as Abi_1_48; @@ -38,6 +39,7 @@ pub(crate) use abi_1_57::Abi as Abi_1_57; pub(crate) use abi_1_58::Abi as Abi_1_58; pub(crate) use abi_1_63::Abi as Abi_1_63; +pub(crate) use abi_1_64::Abi as Abi_1_64; use libloading::Library; use proc_macro_api::{ProcMacroKind, RustCInfo}; @@ -58,6 +60,7 @@ pub(crate) enum Abi { Abi1_57(Abi_1_57), Abi1_58(Abi_1_58), Abi1_63(Abi_1_63), + Abi1_64(Abi_1_64), } impl Abi { @@ -98,10 +101,14 @@ pub fn from_lib( let inner = unsafe { Abi_1_58::from_lib(lib, symbol_name) }?; Ok(Abi::Abi1_58(inner)) } - (1, 63..) => { + (1, 63) => { let inner = unsafe { Abi_1_63::from_lib(lib, symbol_name) }?; Ok(Abi::Abi1_63(inner)) } + (1, 64..) => { + let inner = unsafe { Abi_1_64::from_lib(lib, symbol_name) }?; + Ok(Abi::Abi1_64(inner)) + } _ => Err(LoadProcMacroDylibError::UnsupportedABI), } } @@ -119,6 +126,7 @@ pub fn expand( Self::Abi1_57(abi) => abi.expand(macro_name, macro_body, attributes), Self::Abi1_58(abi) => abi.expand(macro_name, macro_body, attributes), Self::Abi1_63(abi) => abi.expand(macro_name, macro_body, attributes), + Self::Abi1_64(abi) => abi.expand(macro_name, macro_body, attributes), } } @@ -130,6 +138,7 @@ pub fn list_macros(&self) -> Vec<(String, ProcMacroKind)> { Self::Abi1_57(abi) => abi.list_macros(), Self::Abi1_58(abi) => abi.list_macros(), Self::Abi1_63(abi) => abi.list_macros(), + Self::Abi1_64(abi) => abi.list_macros(), } } } -- 2.44.0