1384 lines
46 KiB
Rust
Raw Normal View History

use crate::expand::{self, AstFragment, Invocation};
use crate::module::DirOwnership;
use rustc_ast::attr::MarkedAttrs;
use rustc_ast::ptr::P;
use rustc_ast::token::{self, Nonterminal};
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 18:33:17 -05:00
use rustc_ast::tokenstream::{CanSynthesizeMissingTokens, TokenStream};
use rustc_ast::visit::{AssocCtxt, Visitor};
use rustc_ast::{self as ast, AstLike, Attribute, Item, NodeId, PatKind};
use rustc_attr::{self as attr, Deprecation, Stability};
use rustc_data_structures::fx::FxHashMap;
use rustc_data_structures::sync::{self, Lrc};
use rustc_errors::{Applicability, DiagnosticBuilder, ErrorReported};
Extend `proc_macro_back_compat` lint to `procedural-masquerade` We now lint on *any* use of `procedural-masquerade` crate. While this crate still exists, its main reverse dependency (`cssparser`) no longer depends on it. Any crates still depending off should stop doing so, as it only exists to support very old Rust versions. If a crate actually needs to support old versions of rustc via `procedural-masquerade`, then they'll just need to accept the warning until we remove it entirely (at the same time as the back-compat hack). The latest version of `procedural-masquerade` does not work with the latest rustc, but trying to check for the version seems like more trouble than it's worth. While working on this, I realized that the `proc-macro-hack` check was never actually doing anything. The corresponding enum variant in `proc-macro-hack` is named `Value` or `Nested` - it has never been called `Input`. Due to a strange Crater issue, the Crater run that tested adding this did *not* end up testing it - some of the crates that would have failed did not actually have their tests checked, making it seem as though the `proc-macro-hack` check was working. The Crater issue is being discussed at https://rust-lang.zulipchat.com/#narrow/stream/242791-t-infra/topic/Nearly.20identical.20Crater.20runs.20processed.20a.20crate.20differently/near/230406661 Despite the `proc-macro-hack` check not actually doing anything, we haven't gotten any reports from users about their build being broken. I went ahead and removed it entirely, since it's clear that no one is being affected by the `proc-macro-hack` regression in practice.
2021-03-15 15:54:25 -04:00
use rustc_lint_defs::builtin::PROC_MACRO_BACK_COMPAT;
use rustc_lint_defs::BuiltinLintDiagnostics;
use rustc_parse::{self, nt_to_tokenstream, parser, MACRO_ARGUMENTS};
use rustc_session::{parse::ParseSess, Limit, Session};
2021-05-02 21:19:28 +02:00
use rustc_span::def_id::{CrateNum, DefId, LocalDefId};
2020-01-01 19:40:49 +01:00
use rustc_span::edition::Edition;
2021-06-25 20:43:04 +02:00
use rustc_span::hygiene::{AstPass, ExpnData, ExpnKind, LocalExpnId};
use rustc_span::source_map::SourceMap;
2020-01-01 19:30:57 +01:00
use rustc_span::symbol::{kw, sym, Ident, Symbol};
use rustc_span::{FileName, MultiSpan, Span, DUMMY_SP};
use smallvec::{smallvec, SmallVec};
2019-02-07 02:33:01 +09:00
2019-12-22 17:42:04 -05:00
use std::default::Default;
use std::iter;
use std::path::PathBuf;
use std::rc::Rc;
crate use rustc_span::hygiene::MacroKind;
// When adding new variants, make sure to
// adjust the `visit_*` / `flat_map_*` calls in `InvocationCollector`
// to use `assign_id!`
2019-12-22 17:42:04 -05:00
#[derive(Debug, Clone)]
pub enum Annotatable {
Item(P<ast::Item>),
TraitItem(P<ast::AssocItem>),
ImplItem(P<ast::AssocItem>),
ForeignItem(P<ast::ForeignItem>),
Stmt(P<ast::Stmt>),
Expr(P<ast::Expr>),
Arm(ast::Arm),
ExprField(ast::ExprField),
PatField(ast::PatField),
GenericParam(ast::GenericParam),
Param(ast::Param),
FieldDef(ast::FieldDef),
Variant(ast::Variant),
}
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 18:33:17 -05:00
impl Annotatable {
pub fn span(&self) -> Span {
match *self {
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 18:33:17 -05:00
Annotatable::Item(ref item) => item.span,
Annotatable::TraitItem(ref trait_item) => trait_item.span,
Annotatable::ImplItem(ref impl_item) => impl_item.span,
Annotatable::ForeignItem(ref foreign_item) => foreign_item.span,
Annotatable::Stmt(ref stmt) => stmt.span,
Annotatable::Expr(ref expr) => expr.span,
Annotatable::Arm(ref arm) => arm.span,
Annotatable::ExprField(ref field) => field.span,
Annotatable::PatField(ref fp) => fp.pat.span,
Annotatable::GenericParam(ref gp) => gp.ident.span,
Annotatable::Param(ref p) => p.span,
Annotatable::FieldDef(ref sf) => sf.span,
Annotatable::Variant(ref v) => v.span,
}
}
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 18:33:17 -05:00
pub fn visit_attrs(&mut self, f: impl FnOnce(&mut Vec<Attribute>)) {
match self {
Overhaul `syntax::fold::Folder`. This commit changes `syntax::fold::Folder` from a functional style (where most methods take a `T` and produce a new `T`) to a more imperative style (where most methods take and modify a `&mut T`), and renames it `syntax::mut_visit::MutVisitor`. The first benefit is speed. The functional style does not require any reallocations, due to the use of `P::map` and `MoveMap::move_{,flat_}map`. However, every field in the AST must be overwritten; even those fields that are unchanged are overwritten with the same value. This causes a lot of unnecessary memory writes. The imperative style reduces instruction counts by 1--3% across a wide range of workloads, particularly incremental workloads. The second benefit is conciseness; the imperative style is usually more concise. E.g. compare the old functional style: ``` fn fold_abc(&mut self, abc: ABC) { ABC { a: fold_a(abc.a), b: fold_b(abc.b), c: abc.c, } } ``` with the imperative style: ``` fn visit_abc(&mut self, ABC { a, b, c: _ }: &mut ABC) { visit_a(a); visit_b(b); } ``` (The reductions get larger in more complex examples.) Overall, the patch removes over 200 lines of code -- even though the new code has more comments -- and a lot of the remaining lines have fewer characters. Some notes: - The old style used methods called `fold_*`. The new style mostly uses methods called `visit_*`, but there are a few methods that map a `T` to something other than a `T`, which are called `flat_map_*` (`T` maps to multiple `T`s) or `filter_map_*` (`T` maps to 0 or 1 `T`s). - `move_map.rs`/`MoveMap`/`move_map`/`move_flat_map` are renamed `map_in_place.rs`/`MapInPlace`/`map_in_place`/`flat_map_in_place` to reflect their slightly changed signatures. - Although this commit renames the `fold` module as `mut_visit`, it keeps it in the `fold.rs` file, so as not to confuse git. The next commit will rename the file.
2019-02-05 15:20:55 +11:00
Annotatable::Item(item) => item.visit_attrs(f),
Annotatable::TraitItem(trait_item) => trait_item.visit_attrs(f),
Annotatable::ImplItem(impl_item) => impl_item.visit_attrs(f),
Annotatable::ForeignItem(foreign_item) => foreign_item.visit_attrs(f),
Annotatable::Stmt(stmt) => stmt.visit_attrs(f),
Annotatable::Expr(expr) => expr.visit_attrs(f),
Annotatable::Arm(arm) => arm.visit_attrs(f),
Annotatable::ExprField(field) => field.visit_attrs(f),
Annotatable::PatField(fp) => fp.visit_attrs(f),
Annotatable::GenericParam(gp) => gp.visit_attrs(f),
Annotatable::Param(p) => p.visit_attrs(f),
Annotatable::FieldDef(sf) => sf.visit_attrs(f),
Annotatable::Variant(v) => v.visit_attrs(f),
}
}
pub fn visit_with<'a, V: Visitor<'a>>(&'a self, visitor: &mut V) {
match self {
Annotatable::Item(item) => visitor.visit_item(item),
Annotatable::TraitItem(item) => visitor.visit_assoc_item(item, AssocCtxt::Trait),
Annotatable::ImplItem(item) => visitor.visit_assoc_item(item, AssocCtxt::Impl),
Annotatable::ForeignItem(foreign_item) => visitor.visit_foreign_item(foreign_item),
Annotatable::Stmt(stmt) => visitor.visit_stmt(stmt),
Annotatable::Expr(expr) => visitor.visit_expr(expr),
Annotatable::Arm(arm) => visitor.visit_arm(arm),
Annotatable::ExprField(field) => visitor.visit_expr_field(field),
Annotatable::PatField(fp) => visitor.visit_pat_field(fp),
Annotatable::GenericParam(gp) => visitor.visit_generic_param(gp),
Annotatable::Param(p) => visitor.visit_param(p),
Annotatable::FieldDef(sf) => visitor.visit_field_def(sf),
Annotatable::Variant(v) => visitor.visit_variant(v),
}
}
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 18:33:17 -05:00
pub fn into_nonterminal(self) -> Nonterminal {
match self {
Annotatable::Item(item) => token::NtItem(item),
Annotatable::TraitItem(item) | Annotatable::ImplItem(item) => {
token::NtItem(P(item.and_then(ast::AssocItem::into_item)))
}
Annotatable::ForeignItem(item) => {
token::NtItem(P(item.and_then(ast::ForeignItem::into_item)))
}
Annotatable::Stmt(stmt) => token::NtStmt(stmt.into_inner()),
Annotatable::Expr(expr) => token::NtExpr(expr),
Annotatable::Arm(..)
| Annotatable::ExprField(..)
| Annotatable::PatField(..)
| Annotatable::GenericParam(..)
| Annotatable::Param(..)
| Annotatable::FieldDef(..)
| Annotatable::Variant(..) => panic!("unexpected annotatable"),
}
}
crate fn into_tokens(self, sess: &ParseSess) -> TokenStream {
Implement token-based handling of attributes during expansion This PR modifies the macro expansion infrastructure to handle attributes in a fully token-based manner. As a result: * Derives macros no longer lose spans when their input is modified by eager cfg-expansion. This is accomplished by performing eager cfg-expansion on the token stream that we pass to the derive proc-macro * Inner attributes now preserve spans in all cases, including when we have multiple inner attributes in a row. This is accomplished through the following changes: * New structs `AttrAnnotatedTokenStream` and `AttrAnnotatedTokenTree` are introduced. These are very similar to a normal `TokenTree`, but they also track the position of attributes and attribute targets within the stream. They are built when we collect tokens during parsing. An `AttrAnnotatedTokenStream` is converted to a regular `TokenStream` when we invoke a macro. * Token capturing and `LazyTokenStream` are modified to work with `AttrAnnotatedTokenStream`. A new `ReplaceRange` type is introduced, which is created during the parsing of a nested AST node to make the 'outer' AST node aware of the attributes and attribute target stored deeper in the token stream. * When we need to perform eager cfg-expansion (either due to `#[derive]` or `#[cfg_eval]`), we tokenize and reparse our target, capturing additional information about the locations of `#[cfg]` and `#[cfg_attr]` attributes at any depth within the target. This is a performance optimization, allowing us to perform less work in the typical case where captured tokens never have eager cfg-expansion run.
2020-11-28 18:33:17 -05:00
nt_to_tokenstream(&self.into_nonterminal(), sess, CanSynthesizeMissingTokens::No)
}
pub fn expect_item(self) -> P<ast::Item> {
match self {
Annotatable::Item(i) => i,
2019-12-22 17:42:04 -05:00
_ => panic!("expected Item"),
}
}
pub fn expect_trait_item(self) -> P<ast::AssocItem> {
match self {
Annotatable::TraitItem(i) => i,
2019-12-22 17:42:04 -05:00
_ => panic!("expected Item"),
}
}
pub fn expect_impl_item(self) -> P<ast::AssocItem> {
match self {
Annotatable::ImplItem(i) => i,
2019-12-22 17:42:04 -05:00
_ => panic!("expected Item"),
}
}
pub fn expect_foreign_item(self) -> P<ast::ForeignItem> {
match self {
Annotatable::ForeignItem(i) => i,
2019-12-22 17:42:04 -05:00
_ => panic!("expected foreign item"),
}
}
pub fn expect_stmt(self) -> ast::Stmt {
match self {
Annotatable::Stmt(stmt) => stmt.into_inner(),
_ => panic!("expected statement"),
}
}
pub fn expect_expr(self) -> P<ast::Expr> {
match self {
Annotatable::Expr(expr) => expr,
_ => panic!("expected expression"),
}
}
pub fn expect_arm(self) -> ast::Arm {
match self {
Annotatable::Arm(arm) => arm,
2019-12-22 17:42:04 -05:00
_ => panic!("expected match arm"),
}
}
pub fn expect_expr_field(self) -> ast::ExprField {
match self {
Annotatable::ExprField(field) => field,
2019-12-22 17:42:04 -05:00
_ => panic!("expected field"),
}
}
pub fn expect_pat_field(self) -> ast::PatField {
match self {
Annotatable::PatField(fp) => fp,
2019-12-22 17:42:04 -05:00
_ => panic!("expected field pattern"),
}
}
pub fn expect_generic_param(self) -> ast::GenericParam {
match self {
Annotatable::GenericParam(gp) => gp,
2019-12-22 17:42:04 -05:00
_ => panic!("expected generic parameter"),
}
}
pub fn expect_param(self) -> ast::Param {
match self {
Annotatable::Param(param) => param,
2019-12-22 17:42:04 -05:00
_ => panic!("expected parameter"),
}
}
pub fn expect_field_def(self) -> ast::FieldDef {
match self {
Annotatable::FieldDef(sf) => sf,
2019-12-22 17:42:04 -05:00
_ => panic!("expected struct field"),
}
}
pub fn expect_variant(self) -> ast::Variant {
match self {
Annotatable::Variant(v) => v,
2019-12-22 17:42:04 -05:00
_ => panic!("expected variant"),
}
}
}
/// Result of an expansion that may need to be retried.
/// Consider using this for non-`MultiItemModifier` expanders as well.
pub enum ExpandResult<T, U> {
/// Expansion produced a result (possibly dummy).
Ready(T),
/// Expansion could not produce a result and needs to be retried.
Retry(U),
}
// `meta_item` is the attribute, and `item` is the item being modified.
pub trait MultiItemModifier {
2019-12-22 17:42:04 -05:00
fn expand(
&self,
ecx: &mut ExtCtxt<'_>,
span: Span,
meta_item: &ast::MetaItem,
item: Annotatable,
) -> ExpandResult<Vec<Annotatable>, Annotatable>;
}
2019-12-30 21:38:43 +03:00
impl<F> MultiItemModifier for F
2019-12-22 17:42:04 -05:00
where
2019-12-30 21:38:43 +03:00
F: Fn(&mut ExtCtxt<'_>, Span, &ast::MetaItem, Annotatable) -> Vec<Annotatable>,
{
2019-12-22 17:42:04 -05:00
fn expand(
&self,
ecx: &mut ExtCtxt<'_>,
span: Span,
meta_item: &ast::MetaItem,
item: Annotatable,
) -> ExpandResult<Vec<Annotatable>, Annotatable> {
ExpandResult::Ready(self(ecx, span, meta_item, item))
}
}
pub trait ProcMacro {
2020-03-17 10:09:18 +01:00
fn expand<'cx>(
&self,
ecx: &'cx mut ExtCtxt<'_>,
span: Span,
ts: TokenStream,
) -> Result<TokenStream, ErrorReported>;
}
impl<F> ProcMacro for F
2019-12-22 17:42:04 -05:00
where
F: Fn(TokenStream) -> TokenStream,
{
2020-03-17 10:09:18 +01:00
fn expand<'cx>(
&self,
_ecx: &'cx mut ExtCtxt<'_>,
_span: Span,
ts: TokenStream,
) -> Result<TokenStream, ErrorReported> {
// FIXME setup implicit context in TLS before calling self.
2020-09-28 23:10:48 +03:00
Ok(self(ts))
}
}
pub trait AttrProcMacro {
2019-12-22 17:42:04 -05:00
fn expand<'cx>(
&self,
ecx: &'cx mut ExtCtxt<'_>,
span: Span,
annotation: TokenStream,
annotated: TokenStream,
2020-03-17 10:56:00 +01:00
) -> Result<TokenStream, ErrorReported>;
}
impl<F> AttrProcMacro for F
2019-12-22 17:42:04 -05:00
where
F: Fn(TokenStream, TokenStream) -> TokenStream,
{
2019-12-22 17:42:04 -05:00
fn expand<'cx>(
&self,
_ecx: &'cx mut ExtCtxt<'_>,
_span: Span,
annotation: TokenStream,
annotated: TokenStream,
2020-03-17 10:56:00 +01:00
) -> Result<TokenStream, ErrorReported> {
// FIXME setup implicit context in TLS before calling self.
2020-09-28 23:10:48 +03:00
Ok(self(annotation, annotated))
}
}
/// Represents a thing that maps token trees to Macro Results
pub trait TTMacroExpander {
fn expand<'cx>(
&self,
2019-02-07 02:33:01 +09:00
ecx: &'cx mut ExtCtxt<'_>,
span: Span,
input: TokenStream,
2019-12-22 17:42:04 -05:00
) -> Box<dyn MacResult + 'cx>;
}
pub type MacroExpanderFn =
2019-12-22 17:42:04 -05:00
for<'cx> fn(&'cx mut ExtCtxt<'_>, Span, TokenStream) -> Box<dyn MacResult + 'cx>;
impl<F> TTMacroExpander for F
2019-12-22 17:42:04 -05:00
where
F: for<'cx> Fn(&'cx mut ExtCtxt<'_>, Span, TokenStream) -> Box<dyn MacResult + 'cx>,
{
fn expand<'cx>(
&self,
2019-02-07 02:33:01 +09:00
ecx: &'cx mut ExtCtxt<'_>,
span: Span,
input: TokenStream,
2019-12-22 17:42:04 -05:00
) -> Box<dyn MacResult + 'cx> {
2020-09-28 23:10:48 +03:00
self(ecx, span, input)
}
}
2013-07-08 15:55:14 -07:00
// Use a macro because forwarding to a simple function has type system issues
macro_rules! make_stmts_default {
($me:expr) => {
2019-12-22 17:42:04 -05:00
$me.make_expr().map(|e| {
smallvec![ast::Stmt {
id: ast::DUMMY_NODE_ID,
span: e.span,
kind: ast::StmtKind::Expr(e),
}]
})
};
}
/// The result of a macro expansion. The return values of the various
/// methods are spliced into the AST at the callsite of the macro.
pub trait MacResult {
2019-02-08 14:53:55 +01:00
/// Creates an expression.
2014-09-13 19:06:01 +03:00
fn make_expr(self: Box<Self>) -> Option<P<ast::Expr>> {
None
}
2021-08-08 11:49:13 -03:00
2019-02-08 14:53:55 +01:00
/// Creates zero or more items.
2018-08-30 11:42:16 +02:00
fn make_items(self: Box<Self>) -> Option<SmallVec<[P<ast::Item>; 1]>> {
None
}
2019-02-08 14:53:55 +01:00
/// Creates zero or more impl items.
fn make_impl_items(self: Box<Self>) -> Option<SmallVec<[P<ast::AssocItem>; 1]>> {
None
}
2019-02-08 14:53:55 +01:00
/// Creates zero or more trait items.
fn make_trait_items(self: Box<Self>) -> Option<SmallVec<[P<ast::AssocItem>; 1]>> {
None
}
2019-02-08 14:53:55 +01:00
/// Creates zero or more items in an `extern {}` block
fn make_foreign_items(self: Box<Self>) -> Option<SmallVec<[P<ast::ForeignItem>; 1]>> {
2019-12-22 17:42:04 -05:00
None
}
2019-02-08 14:53:55 +01:00
/// Creates a pattern.
2014-09-13 19:06:01 +03:00
fn make_pat(self: Box<Self>) -> Option<P<ast::Pat>> {
2014-05-19 13:32:51 -07:00
None
}
2019-02-08 14:53:55 +01:00
/// Creates zero or more statements.
///
/// By default this attempts to create an expression statement,
/// returning None if that fails.
2018-08-30 11:42:16 +02:00
fn make_stmts(self: Box<Self>) -> Option<SmallVec<[ast::Stmt; 1]>> {
make_stmts_default!(self)
}
fn make_ty(self: Box<Self>) -> Option<P<ast::Ty>> {
None
}
fn make_arms(self: Box<Self>) -> Option<SmallVec<[ast::Arm; 1]>> {
None
}
fn make_expr_fields(self: Box<Self>) -> Option<SmallVec<[ast::ExprField; 1]>> {
None
}
fn make_pat_fields(self: Box<Self>) -> Option<SmallVec<[ast::PatField; 1]>> {
None
}
fn make_generic_params(self: Box<Self>) -> Option<SmallVec<[ast::GenericParam; 1]>> {
None
}
fn make_params(self: Box<Self>) -> Option<SmallVec<[ast::Param; 1]>> {
None
}
fn make_field_defs(self: Box<Self>) -> Option<SmallVec<[ast::FieldDef; 1]>> {
None
}
fn make_variants(self: Box<Self>) -> Option<SmallVec<[ast::Variant; 1]>> {
None
}
}
2013-07-08 15:55:14 -07:00
macro_rules! make_MacEager {
( $( $fld:ident: $t:ty, )* ) => {
/// `MacResult` implementation for the common case where you've already
/// built each form of AST that you might return.
#[derive(Default)]
pub struct MacEager {
$(
pub $fld: Option<$t>,
)*
}
impl MacEager {
$(
pub fn $fld(v: $t) -> Box<dyn MacResult> {
Box::new(MacEager {
$fld: Some(v),
..Default::default()
})
}
)*
}
}
}
make_MacEager! {
expr: P<ast::Expr>,
pat: P<ast::Pat>,
2018-08-30 11:42:16 +02:00
items: SmallVec<[P<ast::Item>; 1]>,
impl_items: SmallVec<[P<ast::AssocItem>; 1]>,
trait_items: SmallVec<[P<ast::AssocItem>; 1]>,
foreign_items: SmallVec<[P<ast::ForeignItem>; 1]>,
2018-08-30 11:42:16 +02:00
stmts: SmallVec<[ast::Stmt; 1]>,
ty: P<ast::Ty>,
2014-05-19 13:32:51 -07:00
}
impl MacResult for MacEager {
fn make_expr(self: Box<Self>) -> Option<P<ast::Expr>> {
self.expr
2014-05-19 13:32:51 -07:00
}
2018-08-30 11:42:16 +02:00
fn make_items(self: Box<Self>) -> Option<SmallVec<[P<ast::Item>; 1]>> {
self.items
2014-05-19 13:32:51 -07:00
}
fn make_impl_items(self: Box<Self>) -> Option<SmallVec<[P<ast::AssocItem>; 1]>> {
self.impl_items
}
fn make_trait_items(self: Box<Self>) -> Option<SmallVec<[P<ast::AssocItem>; 1]>> {
self.trait_items
}
fn make_foreign_items(self: Box<Self>) -> Option<SmallVec<[P<ast::ForeignItem>; 1]>> {
self.foreign_items
}
2018-08-30 11:42:16 +02:00
fn make_stmts(self: Box<Self>) -> Option<SmallVec<[ast::Stmt; 1]>> {
match self.stmts.as_ref().map_or(0, |s| s.len()) {
0 => make_stmts_default!(self),
_ => self.stmts,
}
}
fn make_pat(self: Box<Self>) -> Option<P<ast::Pat>> {
if let Some(p) = self.pat {
return Some(p);
}
if let Some(e) = self.expr {
if let ast::ExprKind::Lit(_) = e.kind {
return Some(P(ast::Pat {
id: ast::DUMMY_NODE_ID,
span: e.span,
2019-09-26 16:18:31 +01:00
kind: PatKind::Lit(e),
tokens: None,
}));
}
}
None
}
fn make_ty(self: Box<Self>) -> Option<P<ast::Ty>> {
self.ty
}
}
/// Fill-in macro expansion result, to allow compilation to continue
/// after hitting errors.
2015-03-30 09:38:59 -04:00
#[derive(Copy, Clone)]
pub struct DummyResult {
2018-12-20 03:57:48 +03:00
is_error: bool,
span: Span,
2012-07-06 14:29:50 -07:00
}
impl DummyResult {
2019-02-08 14:53:55 +01:00
/// Creates a default MacResult that can be anything.
///
/// Use this as a return value after hitting any errors and
/// calling `span_err`.
2019-12-22 17:42:04 -05:00
pub fn any(span: Span) -> Box<dyn MacResult + 'static> {
Box::new(DummyResult { is_error: true, span })
2018-12-20 03:57:48 +03:00
}
/// Same as `any`, but must be a valid fragment, not error.
2019-12-22 17:42:04 -05:00
pub fn any_valid(span: Span) -> Box<dyn MacResult + 'static> {
Box::new(DummyResult { is_error: false, span })
}
/// A plain dummy expression.
2018-12-20 03:57:48 +03:00
pub fn raw_expr(sp: Span, is_error: bool) -> P<ast::Expr> {
2014-09-13 19:06:01 +03:00
P(ast::Expr {
id: ast::DUMMY_NODE_ID,
kind: if is_error { ast::ExprKind::Err } else { ast::ExprKind::Tup(Vec::new()) },
span: sp,
2019-12-03 16:38:34 +01:00
attrs: ast::AttrVec::new(),
2020-05-19 16:56:20 -04:00
tokens: None,
2014-09-13 19:06:01 +03:00
})
}
2014-05-19 13:32:51 -07:00
/// A plain dummy pattern.
2014-09-13 19:06:01 +03:00
pub fn raw_pat(sp: Span) -> ast::Pat {
ast::Pat { id: ast::DUMMY_NODE_ID, kind: PatKind::Wild, span: sp, tokens: None }
2014-05-19 13:32:51 -07:00
}
/// A plain dummy type.
2018-12-20 03:57:48 +03:00
pub fn raw_ty(sp: Span, is_error: bool) -> P<ast::Ty> {
P(ast::Ty {
id: ast::DUMMY_NODE_ID,
2019-09-26 17:25:31 +01:00
kind: if is_error { ast::TyKind::Err } else { ast::TyKind::Tup(Vec::new()) },
2019-12-22 17:42:04 -05:00
span: sp,
tokens: None,
})
}
}
impl MacResult for DummyResult {
2014-09-13 19:06:01 +03:00
fn make_expr(self: Box<DummyResult>) -> Option<P<ast::Expr>> {
2018-12-20 03:57:48 +03:00
Some(DummyResult::raw_expr(self.span, self.is_error))
}
2014-09-13 19:06:01 +03:00
fn make_pat(self: Box<DummyResult>) -> Option<P<ast::Pat>> {
Some(P(DummyResult::raw_pat(self.span)))
2014-05-19 13:32:51 -07:00
}
2018-08-30 11:42:16 +02:00
fn make_items(self: Box<DummyResult>) -> Option<SmallVec<[P<ast::Item>; 1]>> {
Some(SmallVec::new())
}
fn make_impl_items(self: Box<DummyResult>) -> Option<SmallVec<[P<ast::AssocItem>; 1]>> {
Some(SmallVec::new())
}
fn make_trait_items(self: Box<DummyResult>) -> Option<SmallVec<[P<ast::AssocItem>; 1]>> {
Some(SmallVec::new())
}
fn make_foreign_items(self: Box<Self>) -> Option<SmallVec<[P<ast::ForeignItem>; 1]>> {
Some(SmallVec::new())
}
2018-08-30 11:42:16 +02:00
fn make_stmts(self: Box<DummyResult>) -> Option<SmallVec<[ast::Stmt; 1]>> {
2018-08-13 22:15:16 +03:00
Some(smallvec![ast::Stmt {
2016-06-17 02:30:01 +00:00
id: ast::DUMMY_NODE_ID,
2019-09-26 17:34:50 +01:00
kind: ast::StmtKind::Expr(DummyResult::raw_expr(self.span, self.is_error)),
2016-06-17 02:30:01 +00:00
span: self.span,
2018-08-13 22:15:16 +03:00
}])
}
fn make_ty(self: Box<DummyResult>) -> Option<P<ast::Ty>> {
2018-12-20 03:57:48 +03:00
Some(DummyResult::raw_ty(self.span, self.is_error))
}
fn make_arms(self: Box<DummyResult>) -> Option<SmallVec<[ast::Arm; 1]>> {
2019-12-22 17:42:04 -05:00
Some(SmallVec::new())
}
fn make_expr_fields(self: Box<DummyResult>) -> Option<SmallVec<[ast::ExprField; 1]>> {
Some(SmallVec::new())
}
fn make_pat_fields(self: Box<DummyResult>) -> Option<SmallVec<[ast::PatField; 1]>> {
Some(SmallVec::new())
}
fn make_generic_params(self: Box<DummyResult>) -> Option<SmallVec<[ast::GenericParam; 1]>> {
Some(SmallVec::new())
}
fn make_params(self: Box<DummyResult>) -> Option<SmallVec<[ast::Param; 1]>> {
Some(SmallVec::new())
}
fn make_field_defs(self: Box<DummyResult>) -> Option<SmallVec<[ast::FieldDef; 1]>> {
Some(SmallVec::new())
}
fn make_variants(self: Box<DummyResult>) -> Option<SmallVec<[ast::Variant; 1]>> {
Some(SmallVec::new())
}
}
/// A syntax extension kind.
pub enum SyntaxExtensionKind {
/// A token-based function-like macro.
Bang(
/// An expander with signature TokenStream -> TokenStream.
Box<dyn ProcMacro + sync::Sync + sync::Send>,
),
/// An AST-based function-like macro.
LegacyBang(
/// An expander with signature TokenStream -> AST.
Box<dyn TTMacroExpander + sync::Sync + sync::Send>,
),
/// A token-based attribute macro.
Attr(
/// An expander with signature (TokenStream, TokenStream) -> TokenStream.
/// The first TokenSteam is the attribute itself, the second is the annotated item.
/// The produced TokenSteam replaces the input TokenSteam.
Box<dyn AttrProcMacro + sync::Sync + sync::Send>,
),
/// An AST-based attribute macro.
LegacyAttr(
/// An expander with signature (AST, AST) -> AST.
/// The first AST fragment is the attribute itself, the second is the annotated item.
/// The produced AST fragment replaces the input AST fragment.
Box<dyn MultiItemModifier + sync::Sync + sync::Send>,
),
/// A trivial attribute "macro" that does nothing,
/// only keeps the attribute and marks it as inert,
/// thus making it ineligible for further expansion.
2021-08-05 17:58:59 -05:00
NonMacroAttr,
/// A token-based derive macro.
Derive(
/// An expander with signature TokenStream -> TokenStream (not yet).
/// The produced TokenSteam is appended to the input TokenSteam.
Box<dyn MultiItemModifier + sync::Sync + sync::Send>,
),
/// An AST-based derive macro.
LegacyDerive(
/// An expander with signature AST -> AST.
/// The produced AST fragment is appended to the input AST fragment.
Box<dyn MultiItemModifier + sync::Sync + sync::Send>,
),
}
/// A struct representing a macro definition in "lowered" form ready for expansion.
pub struct SyntaxExtension {
/// A syntax extension kind.
pub kind: SyntaxExtensionKind,
/// Span of the macro definition.
pub span: Span,
/// List of unstable features that are treated as stable inside this macro.
pub allow_internal_unstable: Option<Lrc<[Symbol]>>,
/// Suppresses the `unsafe_code` lint for code produced by this macro.
pub allow_internal_unsafe: bool,
/// Enables the macro helper hack (`ident!(...)` -> `$crate::ident!(...)`) for this macro.
pub local_inner_macros: bool,
/// The macro's stability info.
pub stability: Option<Stability>,
/// The macro's deprecation info.
pub deprecation: Option<Deprecation>,
/// Names of helper attributes registered by this macro.
pub helper_attrs: Vec<Symbol>,
/// Edition of the crate in which this macro is defined.
pub edition: Edition,
/// Built-in macros have a couple of special properties like availability
/// in `#[no_implicit_prelude]` modules, so we have to keep this flag.
pub builtin_name: Option<Symbol>,
}
impl SyntaxExtension {
2019-02-08 14:53:55 +01:00
/// Returns which kind of macro calls this syntax extension.
pub fn macro_kind(&self) -> MacroKind {
match self.kind {
2019-12-22 17:42:04 -05:00
SyntaxExtensionKind::Bang(..) | SyntaxExtensionKind::LegacyBang(..) => MacroKind::Bang,
SyntaxExtensionKind::Attr(..)
| SyntaxExtensionKind::LegacyAttr(..)
2021-08-05 17:58:59 -05:00
| SyntaxExtensionKind::NonMacroAttr => MacroKind::Attr,
2019-12-22 17:42:04 -05:00
SyntaxExtensionKind::Derive(..) | SyntaxExtensionKind::LegacyDerive(..) => {
MacroKind::Derive
}
}
}
2017-03-22 08:39:51 +00:00
/// Constructs a syntax extension with default properties.
pub fn default(kind: SyntaxExtensionKind, edition: Edition) -> SyntaxExtension {
SyntaxExtension {
span: DUMMY_SP,
allow_internal_unstable: None,
allow_internal_unsafe: false,
local_inner_macros: false,
stability: None,
deprecation: None,
helper_attrs: Vec::new(),
edition,
builtin_name: None,
kind,
2018-06-24 19:54:23 +03:00
}
}
/// Constructs a syntax extension with the given properties
/// and other properties converted from attributes.
pub fn new(
sess: &Session,
kind: SyntaxExtensionKind,
span: Span,
helper_attrs: Vec<Symbol>,
edition: Edition,
2020-04-19 13:00:18 +02:00
name: Symbol,
attrs: &[ast::Attribute],
) -> SyntaxExtension {
let allow_internal_unstable =
attr::allow_internal_unstable(sess, &attrs).collect::<Vec<Symbol>>();
let mut local_inner_macros = false;
if let Some(macro_export) = sess.find_by_name(attrs, sym::macro_export) {
if let Some(l) = macro_export.meta_item_list() {
local_inner_macros = attr::list_contains_name(&l, sym::local_inner_macros);
}
}
let (builtin_name, helper_attrs) = sess
2021-01-09 16:54:12 +01:00
.find_by_name(attrs, sym::rustc_builtin_macro)
.map(|attr| {
// Override `helper_attrs` passed above if it's a built-in macro,
// marking `proc_macro_derive` macros as built-in is not a realistic use case.
parse_macro_name_and_helper_attrs(sess.diagnostic(), attr, "built-in").map_or_else(
|| (Some(name), Vec::new()),
|(name, helper_attrs)| (Some(name), helper_attrs),
)
})
.unwrap_or_else(|| (None, helper_attrs));
let (stability, const_stability) = attr::find_stability(&sess, attrs, span);
if let Some((_, sp)) = const_stability {
sess.parse_sess
.span_diagnostic
.struct_span_err(sp, "macros cannot have const stability attributes")
.span_label(sp, "invalid const stability attribute")
.span_label(
sess.source_map().guess_head_span(span),
"const stability attribute affects this macro",
)
.emit();
}
SyntaxExtension {
kind,
span,
allow_internal_unstable: (!allow_internal_unstable.is_empty())
.then(|| allow_internal_unstable.into()),
allow_internal_unsafe: sess.contains_name(attrs, sym::allow_internal_unsafe),
local_inner_macros,
stability: stability.map(|(s, _)| s),
deprecation: attr::find_deprecation(&sess, attrs).map(|(d, _)| d),
helper_attrs,
edition,
builtin_name,
}
}
pub fn dummy_bang(edition: Edition) -> SyntaxExtension {
2019-12-22 17:42:04 -05:00
fn expander<'cx>(
_: &'cx mut ExtCtxt<'_>,
span: Span,
_: TokenStream,
) -> Box<dyn MacResult + 'cx> {
DummyResult::any(span)
}
SyntaxExtension::default(SyntaxExtensionKind::LegacyBang(Box::new(expander)), edition)
}
pub fn dummy_derive(edition: Edition) -> SyntaxExtension {
2019-12-22 17:42:04 -05:00
fn expander(
_: &mut ExtCtxt<'_>,
_: Span,
_: &ast::MetaItem,
_: Annotatable,
) -> Vec<Annotatable> {
Vec::new()
}
SyntaxExtension::default(SyntaxExtensionKind::Derive(Box::new(expander)), edition)
}
2021-08-05 17:58:59 -05:00
pub fn non_macro_attr(edition: Edition) -> SyntaxExtension {
SyntaxExtension::default(SyntaxExtensionKind::NonMacroAttr, edition)
}
pub fn expn_data(
&self,
2021-06-25 20:43:04 +02:00
parent: LocalExpnId,
call_site: Span,
descr: Symbol,
macro_def_id: Option<DefId>,
2021-06-28 19:29:55 +02:00
parent_module: Option<DefId>,
) -> ExpnData {
ExpnData::new(
ExpnKind::Macro(self.macro_kind(), descr),
2021-06-25 20:43:04 +02:00
parent.to_expn_id(),
call_site,
self.span,
self.allow_internal_unstable.clone(),
self.allow_internal_unsafe,
self.local_inner_macros,
self.edition,
macro_def_id,
2021-06-28 19:29:55 +02:00
parent_module,
)
}
}
/// Error type that denotes indeterminacy.
pub struct Indeterminate;
pub type DeriveResolutions = Vec<(ast::Path, Annotatable, Option<Lrc<SyntaxExtension>>)>;
2020-06-27 23:51:28 +03:00
pub trait ResolverExpand {
fn next_node_id(&mut self) -> NodeId;
2021-05-02 21:19:28 +02:00
fn invocation_parent(&self, id: LocalExpnId) -> LocalDefId;
fn resolve_dollar_crates(&mut self);
2021-06-25 20:43:04 +02:00
fn visit_ast_fragment_with_placeholders(
&mut self,
expn_id: LocalExpnId,
fragment: &AstFragment,
);
2021-01-10 14:36:30 +03:00
fn register_builtin_macro(&mut self, name: Symbol, ext: SyntaxExtensionKind);
fn expansion_for_ast_pass(
&mut self,
call_site: Span,
pass: AstPass,
features: &[Symbol],
parent_module_id: Option<NodeId>,
2021-06-25 20:43:04 +02:00
) -> LocalExpnId;
2016-11-10 10:11:25 +00:00
fn resolve_imports(&mut self);
fn resolve_macro_invocation(
2019-12-22 17:42:04 -05:00
&mut self,
invoc: &Invocation,
2021-06-25 20:43:04 +02:00
eager_expansion_root: LocalExpnId,
2019-12-22 17:42:04 -05:00
force: bool,
) -> Result<Lrc<SyntaxExtension>, Indeterminate>;
fn check_unused_macros(&mut self);
2020-11-19 01:50:16 +03:00
// Resolver interfaces for specific built-in macros.
/// Does `#[derive(...)]` attribute with the given `ExpnId` have built-in `Copy` inside it?
2021-06-25 20:43:04 +02:00
fn has_derive_copy(&self, expn_id: LocalExpnId) -> bool;
/// Resolve paths inside the `#[derive(...)]` attribute with the given `ExpnId`.
fn resolve_derives(
&mut self,
2021-06-25 20:43:04 +02:00
expn_id: LocalExpnId,
force: bool,
derive_paths: &dyn Fn() -> DeriveResolutions,
) -> Result<(), Indeterminate>;
/// Take resolutions for paths inside the `#[derive(...)]` attribute with the given `ExpnId`
/// back from resolver.
2021-06-25 20:43:04 +02:00
fn take_derive_resolutions(&mut self, expn_id: LocalExpnId) -> Option<DeriveResolutions>;
2020-11-19 01:50:16 +03:00
/// Path resolution logic for `#[cfg_accessible(path)]`.
2021-06-25 20:43:04 +02:00
fn cfg_accessible(
&mut self,
expn_id: LocalExpnId,
path: &ast::Path,
) -> Result<bool, Indeterminate>;
Implement span quoting for proc-macros This PR implements span quoting, allowing proc-macros to produce spans pointing *into their own crate*. This is used by the unstable `proc_macro::quote!` macro, allowing us to get error messages like this: ``` error[E0412]: cannot find type `MissingType` in this scope --> $DIR/auxiliary/span-from-proc-macro.rs:37:20 | LL | pub fn error_from_attribute(_args: TokenStream, _input: TokenStream) -> TokenStream { | ----------------------------------------------------------------------------------- in this expansion of procedural macro `#[error_from_attribute]` ... LL | field: MissingType | ^^^^^^^^^^^ not found in this scope | ::: $DIR/span-from-proc-macro.rs:8:1 | LL | #[error_from_attribute] | ----------------------- in this macro invocation ``` Here, `MissingType` occurs inside the implementation of the proc-macro `#[error_from_attribute]`. Previosuly, this would always result in a span pointing at `#[error_from_attribute]` This will make many proc-macro-related error message much more useful - when a proc-macro generates code containing an error, users will get an error message pointing directly at that code (within the macro definition), instead of always getting a span pointing at the macro invocation site. This is implemented as follows: * When a proc-macro crate is being *compiled*, it causes the `quote!` macro to get run. This saves all of the sapns in the input to `quote!` into the metadata of *the proc-macro-crate* (which we are currently compiling). The `quote!` macro then expands to a call to `proc_macro::Span::recover_proc_macro_span(id)`, where `id` is an opaque identifier for the span in the crate metadata. * When the same proc-macro crate is *run* (e.g. it is loaded from disk and invoked by some consumer crate), the call to `proc_macro::Span::recover_proc_macro_span` causes us to load the span from the proc-macro crate's metadata. The proc-macro then produces a `TokenStream` containing a `Span` pointing into the proc-macro crate itself. The recursive nature of 'quote!' can be difficult to understand at first. The file `src/test/ui/proc-macro/quote-debug.stdout` shows the output of the `quote!` macro, which should make this eaier to understand. This PR also supports custom quoting spans in custom quote macros (e.g. the `quote` crate). All span quoting goes through the `proc_macro::quote_span` method, which can be called by a custom quote macro to perform span quoting. An example of this usage is provided in `src/test/ui/proc-macro/auxiliary/custom-quote.rs` Custom quoting currently has a few limitations: In order to quote a span, we need to generate a call to `proc_macro::Span::recover_proc_macro_span`. However, proc-macros support renaming the `proc_macro` crate, so we can't simply hardcode this path. Previously, the `quote_span` method used the path `crate::Span` - however, this only works when it is called by the builtin `quote!` macro in the same crate. To support being called from arbitrary crates, we need access to the name of the `proc_macro` crate to generate a path. This PR adds an additional argument to `quote_span` to specify the name of the `proc_macro` crate. Howver, this feels kind of hacky, and we may want to change this before stabilizing anything quote-related. Additionally, using `quote_span` currently requires enabling the `proc_macro_internals` feature. The builtin `quote!` macro has an `#[allow_internal_unstable]` attribute, but this won't work for custom quote implementations. This will likely require some additional tricks to apply `allow_internal_unstable` to the span of `proc_macro::Span::recover_proc_macro_span`.
2020-08-02 19:52:16 -04:00
/// Decodes the proc-macro quoted span in the specified crate, with the specified id.
/// No caching is performed.
fn get_proc_macro_quoted_span(&self, krate: CrateNum, id: usize) -> Span;
2021-07-16 22:22:08 +02:00
/// The order of items in the HIR is unrelated to the order of
/// items in the AST. However, we generate proc macro harnesses
/// based on the AST order, and later refer to these harnesses
/// from the HIR. This field keeps track of the order in which
/// we generated proc macros harnesses, so that we can map
/// HIR proc macros items back to their harness items.
fn declare_proc_macro(&mut self, id: NodeId);
}
#[derive(Clone, Default)]
pub struct ModuleData {
/// Path to the module starting from the crate name, like `my_crate::foo::bar`.
2020-04-19 13:00:18 +02:00
pub mod_path: Vec<Ident>,
/// Stack of paths to files loaded by out-of-line module items,
/// used to detect and report recursive module inclusions.
pub file_path_stack: Vec<PathBuf>,
/// Directory to search child module files in,
/// often (but not necessarily) the parent of the top file path on the `file_path_stack`.
pub dir_path: PathBuf,
}
impl ModuleData {
pub fn with_dir_path(&self, dir_path: PathBuf) -> ModuleData {
ModuleData {
mod_path: self.mod_path.clone(),
file_path_stack: self.file_path_stack.clone(),
dir_path,
}
}
}
#[derive(Clone)]
pub struct ExpansionData {
2021-06-25 20:43:04 +02:00
pub id: LocalExpnId,
pub depth: usize,
pub module: Rc<ModuleData>,
pub dir_ownership: DirOwnership,
pub prior_type_ascription: Option<(Span, bool)>,
/// Some parent node that is close to this macro call
pub lint_node_id: NodeId,
Display an extra note for trailing semicolon lint with trailing macro Currently, we parse macros at the end of a block (e.g. `fn foo() { my_macro!() }`) as expressions, rather than statements. This means that a macro invoked in this position cannot expand to items or semicolon-terminated expressions. In the future, we might want to start parsing these kinds of macros as statements. This would make expansion more 'token-based' (i.e. macro expansion behaves (almost) as if you just textually replaced the macro invocation with its output). However, this is a breaking change (see PR #78991), so it will require further discussion. Since the current behavior will not be changing any time soon, we need to address the interaction with the `SEMICOLON_IN_EXPRESSIONS_FROM_MACROS` lint. Since we are parsing the result of macro expansion as an expression, we will emit a lint if there's a trailing semicolon in the macro output. However, this results in a somewhat confusing message for users, since it visually looks like there should be no problem with having a semicolon at the end of a block (e.g. `fn foo() { my_macro!() }` => `fn foo() { produced_expr; }`) To help reduce confusion, this commit adds a note explaining that the macro is being interpreted as an expression. Additionally, we suggest adding a semicolon after the macro *invocation* - this will cause us to parse the macro call as a statement. We do *not* use a structured suggestion for this, since the user may actually want to remove the semicolon from the macro definition (allowing the block to evaluate to the expression produced by the macro).
2021-07-22 11:24:42 -05:00
pub is_trailing_mac: bool,
}
type OnExternModLoaded<'a> =
Option<&'a dyn Fn(Ident, Vec<Attribute>, Vec<P<Item>>, Span) -> (Vec<Attribute>, Vec<P<Item>>)>;
2014-06-09 13:12:30 -07:00
/// One of these is made during expansion and incrementally updated as we go;
/// when a macro expansion occurs, the resulting nodes have the `backtrace()
/// -> expn_data` of their expansion context stored into their span.
2013-12-25 11:10:33 -07:00
pub struct ExtCtxt<'a> {
pub sess: &'a Session,
pub ecfg: expand::ExpansionConfig<'a>,
pub reduced_recursion_limit: Option<Limit>,
pub root_path: PathBuf,
2020-06-27 23:51:28 +03:00
pub resolver: &'a mut dyn ResolverExpand,
pub current_expansion: ExpansionData,
/// Error recovery mode entered when expansion is stuck
/// (or during eager expansion, but that's a hack).
pub force_mode: bool,
pub expansions: FxHashMap<Span, Vec<String>>,
2020-03-16 00:43:37 +01:00
/// Called directly after having parsed an external `mod foo;` in expansion.
///
/// `Ident` is the module name.
pub(super) extern_mod_loaded: OnExternModLoaded<'a>,
/// When we 'expand' an inert attribute, we leave it
/// in the AST, but insert it here so that we know
/// not to expand it again.
pub(super) expanded_inert_attrs: MarkedAttrs,
}
2013-03-15 15:24:24 -04:00
2013-12-25 11:10:33 -07:00
impl<'a> ExtCtxt<'a> {
2019-12-22 17:42:04 -05:00
pub fn new(
sess: &'a Session,
2019-12-22 17:42:04 -05:00
ecfg: expand::ExpansionConfig<'a>,
2020-06-27 23:51:28 +03:00
resolver: &'a mut dyn ResolverExpand,
extern_mod_loaded: OnExternModLoaded<'a>,
2019-12-22 17:42:04 -05:00
) -> ExtCtxt<'a> {
2013-12-27 17:17:36 -07:00
ExtCtxt {
sess,
ecfg,
reduced_recursion_limit: None,
resolver,
2020-03-16 00:43:37 +01:00
extern_mod_loaded,
root_path: PathBuf::new(),
current_expansion: ExpansionData {
2021-06-25 20:43:04 +02:00
id: LocalExpnId::ROOT,
depth: 0,
module: Default::default(),
dir_ownership: DirOwnership::Owned { relative: None },
prior_type_ascription: None,
lint_node_id: ast::CRATE_NODE_ID,
Display an extra note for trailing semicolon lint with trailing macro Currently, we parse macros at the end of a block (e.g. `fn foo() { my_macro!() }`) as expressions, rather than statements. This means that a macro invoked in this position cannot expand to items or semicolon-terminated expressions. In the future, we might want to start parsing these kinds of macros as statements. This would make expansion more 'token-based' (i.e. macro expansion behaves (almost) as if you just textually replaced the macro invocation with its output). However, this is a breaking change (see PR #78991), so it will require further discussion. Since the current behavior will not be changing any time soon, we need to address the interaction with the `SEMICOLON_IN_EXPRESSIONS_FROM_MACROS` lint. Since we are parsing the result of macro expansion as an expression, we will emit a lint if there's a trailing semicolon in the macro output. However, this results in a somewhat confusing message for users, since it visually looks like there should be no problem with having a semicolon at the end of a block (e.g. `fn foo() { my_macro!() }` => `fn foo() { produced_expr; }`) To help reduce confusion, this commit adds a note explaining that the macro is being interpreted as an expression. Additionally, we suggest adding a semicolon after the macro *invocation* - this will cause us to parse the macro call as a statement. We do *not* use a structured suggestion for this, since the user may actually want to remove the semicolon from the macro definition (allowing the block to evaluate to the expression produced by the macro).
2021-07-22 11:24:42 -05:00
is_trailing_mac: false,
},
force_mode: false,
expansions: FxHashMap::default(),
expanded_inert_attrs: MarkedAttrs::new(),
}
}
/// Returns a `Folder` for deeply expanding all macros in an AST node.
pub fn expander<'b>(&'b mut self) -> expand::MacroExpander<'b, 'a> {
expand::MacroExpander::new(self, false)
}
2019-02-08 14:53:55 +01:00
/// Returns a `Folder` that deeply expands all macros and assigns all `NodeId`s in an AST node.
/// Once `NodeId`s are assigned, the node may not be expanded, removed, or otherwise modified.
pub fn monotonic_expander<'b>(&'b mut self) -> expand::MacroExpander<'b, 'a> {
expand::MacroExpander::new(self, true)
}
pub fn new_parser_from_tts(&self, stream: TokenStream) -> parser::Parser<'a> {
rustc_parse::stream_to_parser(&self.sess.parse_sess, stream, MACRO_ARGUMENTS)
}
2019-12-22 17:42:04 -05:00
pub fn source_map(&self) -> &'a SourceMap {
self.sess.parse_sess.source_map()
2019-12-22 17:42:04 -05:00
}
pub fn parse_sess(&self) -> &'a ParseSess {
&self.sess.parse_sess
2019-12-22 17:42:04 -05:00
}
pub fn call_site(&self) -> Span {
self.current_expansion.id.expn_data().call_site
2017-03-17 04:04:41 +00:00
}
/// Equivalent of `Span::def_site` from the proc macro API,
/// except that the location is taken from the span passed as an argument.
pub fn with_def_site_ctxt(&self, span: Span) -> Span {
2021-06-25 20:43:04 +02:00
span.with_def_site_ctxt(self.current_expansion.id.to_expn_id())
}
/// Equivalent of `Span::call_site` from the proc macro API,
/// except that the location is taken from the span passed as an argument.
pub fn with_call_site_ctxt(&self, span: Span) -> Span {
2021-06-25 20:43:04 +02:00
span.with_call_site_ctxt(self.current_expansion.id.to_expn_id())
}
/// Equivalent of `Span::mixed_site` from the proc macro API,
/// except that the location is taken from the span passed as an argument.
pub fn with_mixed_site_ctxt(&self, span: Span) -> Span {
2021-06-25 20:43:04 +02:00
span.with_mixed_site_ctxt(self.current_expansion.id.to_expn_id())
}
/// Returns span for the macro which originally caused the current expansion to happen.
///
/// Stops backtracing at include! boundary.
2017-05-15 09:41:05 +00:00
pub fn expansion_cause(&self) -> Option<Span> {
self.current_expansion.id.expansion_cause()
}
2019-12-22 17:42:04 -05:00
pub fn struct_span_err<S: Into<MultiSpan>>(&self, sp: S, msg: &str) -> DiagnosticBuilder<'a> {
self.sess.parse_sess.span_diagnostic.struct_span_err(sp, msg)
2015-12-21 10:00:43 +13:00
}
/// Emit `msg` attached to `sp`, without immediately stopping
/// compilation.
///
/// Compilation will be stopped in the near future (at the end of
/// the macro expansion phase).
pub fn span_err<S: Into<MultiSpan>>(&self, sp: S, msg: &str) {
self.sess.parse_sess.span_diagnostic.span_err(sp, msg);
}
pub fn span_warn<S: Into<MultiSpan>>(&self, sp: S, msg: &str) {
self.sess.parse_sess.span_diagnostic.span_warn(sp, msg);
}
pub fn span_bug<S: Into<MultiSpan>>(&self, sp: S, msg: &str) -> ! {
self.sess.parse_sess.span_diagnostic.span_bug(sp, msg);
}
pub fn trace_macros_diag(&mut self) {
for (sp, notes) in self.expansions.iter() {
let mut db = self.sess.parse_sess.span_diagnostic.span_note_diag(*sp, "trace_macro");
for note in notes {
db.note(note);
}
db.emit();
}
// Fixme: does this result in errors?
self.expansions.clear();
}
pub fn bug(&self, msg: &str) -> ! {
self.sess.parse_sess.span_diagnostic.bug(msg);
}
pub fn trace_macros(&self) -> bool {
self.ecfg.trace_mac
}
2013-12-28 22:35:38 -07:00
pub fn set_trace_macros(&mut self, x: bool) {
self.ecfg.trace_mac = x
}
2020-04-19 13:00:18 +02:00
pub fn std_path(&self, components: &[Symbol]) -> Vec<Ident> {
let def_site = self.with_def_site_ctxt(DUMMY_SP);
2019-05-11 17:41:37 +03:00
iter::once(Ident::new(kw::DollarCrate, def_site))
.chain(components.iter().map(|&s| Ident::with_dummy_span(s)))
.collect()
}
pub fn def_site_path(&self, components: &[Symbol]) -> Vec<Ident> {
let def_site = self.with_def_site_ctxt(DUMMY_SP);
components.iter().map(|&s| Ident::new(s, def_site)).collect()
}
2017-05-11 10:26:07 +02:00
pub fn check_unused_macros(&mut self) {
2017-05-11 10:26:07 +02:00
self.resolver.check_unused_macros();
}
2019-06-22 21:51:51 +02:00
2021-05-18 21:46:41 -04:00
/// Resolves a `path` mentioned inside Rust code, returning an absolute path.
2019-06-22 21:51:51 +02:00
///
2021-05-18 21:46:41 -04:00
/// This unifies the logic used for resolving `include_X!`.
2019-06-22 21:51:51 +02:00
///
2021-05-18 21:46:41 -04:00
/// FIXME: move this to `rustc_builtin_macros` and make it private.
pub fn resolve_path(
&self,
path: impl Into<PathBuf>,
span: Span,
) -> Result<PathBuf, DiagnosticBuilder<'a>> {
2019-06-22 21:51:51 +02:00
let path = path.into();
// Relative paths are resolved relative to the file in which they are found
// after macro expansion (that is, they are unhygienic).
if !path.is_absolute() {
let callsite = span.source_callsite();
let mut result = match self.source_map().span_to_filename(callsite) {
FileName::Real(name) => name
.into_local_path()
.expect("attempting to resolve a file path in an external file"),
2019-06-22 21:51:51 +02:00
FileName::DocTest(path, _) => path,
2019-12-22 17:42:04 -05:00
other => {
return Err(self.struct_span_err(
span,
&format!(
"cannot resolve relative path in non-file source `{}`",
self.source_map().filename_for_diagnostics(&other)
),
2019-12-22 17:42:04 -05:00
));
}
2019-06-22 21:51:51 +02:00
};
result.pop();
result.push(path);
Ok(result)
2019-06-22 21:51:51 +02:00
} else {
Ok(path)
2019-06-22 21:51:51 +02:00
}
}
}
2019-02-08 14:53:55 +01:00
/// Extracts a string literal from the macro expanded version of `expr`,
/// returning a diagnostic error of `err_msg` if `expr` is not a string literal.
/// The returned bool indicates whether an applicable suggestion has already been
/// added to the diagnostic to avoid emitting multiple suggestions. `Err(None)`
/// indicates that an ast error was encountered.
pub fn expr_to_spanned_string<'a>(
2019-02-07 02:33:01 +09:00
cx: &'a mut ExtCtxt<'_>,
expr: P<ast::Expr>,
err_msg: &str,
) -> Result<(Symbol, ast::StrStyle, Span), Option<(DiagnosticBuilder<'a>, bool)>> {
// Perform eager expansion on the expression.
// We want to be able to handle e.g., `concat!("foo", "bar")`.
let expr = cx.expander().fully_expand_fragment(AstFragment::Expr(expr)).make_expr();
Err(match expr.kind {
2019-09-26 16:56:53 +01:00
ast::ExprKind::Lit(ref l) => match l.kind {
ast::LitKind::Str(s, style) => return Ok((s, style, expr.span)),
ast::LitKind::ByteStr(_) => {
let mut err = cx.struct_span_err(l.span, err_msg);
err.span_suggestion(
expr.span.shrink_to_lo(),
"consider removing the leading `b`",
String::new(),
Applicability::MaybeIncorrect,
);
Some((err, true))
}
ast::LitKind::Err(_) => None,
_ => Some((cx.struct_span_err(l.span, err_msg), false)),
},
ast::ExprKind::Err => None,
_ => Some((cx.struct_span_err(expr.span, err_msg), false)),
})
}
/// Extracts a string literal from the macro expanded version of `expr`,
/// emitting `err_msg` if `expr` is not a string literal. This does not stop
/// compilation on error, merely emits a non-fatal error and returns `None`.
2019-12-22 17:42:04 -05:00
pub fn expr_to_string(
cx: &mut ExtCtxt<'_>,
expr: P<ast::Expr>,
err_msg: &str,
) -> Option<(Symbol, ast::StrStyle)> {
expr_to_spanned_string(cx, expr, err_msg)
2020-02-02 09:47:58 +10:00
.map_err(|err| {
err.map(|(mut err, _)| {
2020-02-02 09:47:58 +10:00
err.emit();
})
})
.ok()
.map(|(symbol, style, _)| (symbol, style))
2016-09-02 22:01:35 +00:00
}
/// Non-fatally assert that `tts` is empty. Note that this function
/// returns even when `tts` is non-empty, macros that *need* to stop
/// compilation should call
/// `cx.parse_sess.span_diagnostic.abort_if_errors()` (this should be
/// done as rarely as possible).
2019-12-22 17:42:04 -05:00
pub fn check_zero_tts(cx: &ExtCtxt<'_>, sp: Span, tts: TokenStream, name: &str) {
if !tts.is_empty() {
cx.span_err(sp, &format!("{} takes no arguments", name));
}
}
2020-03-17 08:29:34 +01:00
/// Parse an expression. On error, emit it, advancing to `Eof`, and return `None`.
pub fn parse_expr(p: &mut parser::Parser<'_>) -> Option<P<ast::Expr>> {
2020-03-17 08:29:34 +01:00
match p.parse_expr() {
Ok(e) => return Some(e),
Err(mut err) => err.emit(),
}
while p.token != token::Eof {
p.bump();
}
None
}
/// Interpreting `tts` as a comma-separated sequence of expressions,
2019-02-08 14:53:55 +01:00
/// expect exactly one string literal, or emit an error and return `None`.
2019-12-22 17:42:04 -05:00
pub fn get_single_str_from_tts(
cx: &mut ExtCtxt<'_>,
sp: Span,
tts: TokenStream,
name: &str,
) -> Option<String> {
let mut p = cx.new_parser_from_tts(tts);
2014-11-02 23:10:09 -08:00
if p.token == token::Eof {
cx.span_err(sp, &format!("{} takes 1 argument", name));
2019-12-22 17:42:04 -05:00
return None;
2014-11-02 23:10:09 -08:00
}
2020-03-17 08:29:34 +01:00
let ret = parse_expr(&mut p)?;
let _ = p.eat(&token::Comma);
if p.token != token::Eof {
cx.span_err(sp, &format!("{} takes 1 argument", name));
}
2019-12-22 17:42:04 -05:00
expr_to_string(cx, ret, "argument must be a string literal").map(|(s, _)| s.to_string())
}
2020-03-17 08:29:34 +01:00
/// Extracts comma-separated expressions from `tts`.
/// On error, emit it, and return `None`.
2019-12-22 17:42:04 -05:00
pub fn get_exprs_from_tts(
cx: &mut ExtCtxt<'_>,
sp: Span,
tts: TokenStream,
) -> Option<Vec<P<ast::Expr>>> {
let mut p = cx.new_parser_from_tts(tts);
let mut es = Vec::new();
2014-10-27 19:22:52 +11:00
while p.token != token::Eof {
2020-03-17 08:29:34 +01:00
let expr = parse_expr(&mut p)?;
// Perform eager expansion on the expression.
// We want to be able to handle e.g., `concat!("foo", "bar")`.
let expr = cx.expander().fully_expand_fragment(AstFragment::Expr(expr)).make_expr();
Overhaul `syntax::fold::Folder`. This commit changes `syntax::fold::Folder` from a functional style (where most methods take a `T` and produce a new `T`) to a more imperative style (where most methods take and modify a `&mut T`), and renames it `syntax::mut_visit::MutVisitor`. The first benefit is speed. The functional style does not require any reallocations, due to the use of `P::map` and `MoveMap::move_{,flat_}map`. However, every field in the AST must be overwritten; even those fields that are unchanged are overwritten with the same value. This causes a lot of unnecessary memory writes. The imperative style reduces instruction counts by 1--3% across a wide range of workloads, particularly incremental workloads. The second benefit is conciseness; the imperative style is usually more concise. E.g. compare the old functional style: ``` fn fold_abc(&mut self, abc: ABC) { ABC { a: fold_a(abc.a), b: fold_b(abc.b), c: abc.c, } } ``` with the imperative style: ``` fn visit_abc(&mut self, ABC { a, b, c: _ }: &mut ABC) { visit_a(a); visit_b(b); } ``` (The reductions get larger in more complex examples.) Overall, the patch removes over 200 lines of code -- even though the new code has more comments -- and a lot of the remaining lines have fewer characters. Some notes: - The old style used methods called `fold_*`. The new style mostly uses methods called `visit_*`, but there are a few methods that map a `T` to something other than a `T`, which are called `flat_map_*` (`T` maps to multiple `T`s) or `filter_map_*` (`T` maps to 0 or 1 `T`s). - `move_map.rs`/`MoveMap`/`move_map`/`move_flat_map` are renamed `map_in_place.rs`/`MapInPlace`/`map_in_place`/`flat_map_in_place` to reflect their slightly changed signatures. - Although this commit renames the `fold` module as `mut_visit`, it keeps it in the `fold.rs` file, so as not to confuse git. The next commit will rename the file.
2019-02-05 15:20:55 +11:00
es.push(expr);
if p.eat(&token::Comma) {
continue;
}
2014-10-27 19:22:52 +11:00
if p.token != token::Eof {
cx.span_err(sp, "expected token: `,`");
return None;
}
}
Some(es)
}
Extend `proc_macro_back_compat` lint to `procedural-masquerade` We now lint on *any* use of `procedural-masquerade` crate. While this crate still exists, its main reverse dependency (`cssparser`) no longer depends on it. Any crates still depending off should stop doing so, as it only exists to support very old Rust versions. If a crate actually needs to support old versions of rustc via `procedural-masquerade`, then they'll just need to accept the warning until we remove it entirely (at the same time as the back-compat hack). The latest version of `procedural-masquerade` does not work with the latest rustc, but trying to check for the version seems like more trouble than it's worth. While working on this, I realized that the `proc-macro-hack` check was never actually doing anything. The corresponding enum variant in `proc-macro-hack` is named `Value` or `Nested` - it has never been called `Input`. Due to a strange Crater issue, the Crater run that tested adding this did *not* end up testing it - some of the crates that would have failed did not actually have their tests checked, making it seem as though the `proc-macro-hack` check was working. The Crater issue is being discussed at https://rust-lang.zulipchat.com/#narrow/stream/242791-t-infra/topic/Nearly.20identical.20Crater.20runs.20processed.20a.20crate.20differently/near/230406661 Despite the `proc-macro-hack` check not actually doing anything, we haven't gotten any reports from users about their build being broken. I went ahead and removed it entirely, since it's clear that no one is being affected by the `proc-macro-hack` regression in practice.
2021-03-15 15:54:25 -04:00
pub fn parse_macro_name_and_helper_attrs(
diag: &rustc_errors::Handler,
attr: &Attribute,
descr: &str,
) -> Option<(Symbol, Vec<Symbol>)> {
// Once we've located the `#[proc_macro_derive]` attribute, verify
// that it's of the form `#[proc_macro_derive(Foo)]` or
// `#[proc_macro_derive(Foo, attributes(A, ..))]`
let list = match attr.meta_item_list() {
Some(list) => list,
None => return None,
};
if list.len() != 1 && list.len() != 2 {
diag.span_err(attr.span, "attribute must have either one or two arguments");
return None;
}
let trait_attr = match list[0].meta_item() {
Some(meta_item) => meta_item,
_ => {
diag.span_err(list[0].span(), "not a meta item");
return None;
}
};
let trait_ident = match trait_attr.ident() {
Some(trait_ident) if trait_attr.is_word() => trait_ident,
_ => {
diag.span_err(trait_attr.span, "must only be one word");
return None;
}
};
if !trait_ident.name.can_be_raw() {
diag.span_err(
trait_attr.span,
&format!("`{}` cannot be a name of {} macro", trait_ident, descr),
);
}
let attributes_attr = list.get(1);
let proc_attrs: Vec<_> = if let Some(attr) = attributes_attr {
if !attr.has_name(sym::attributes) {
diag.span_err(attr.span(), "second argument must be `attributes`")
}
attr.meta_item_list()
.unwrap_or_else(|| {
diag.span_err(attr.span(), "attribute must be of form: `attributes(foo, bar)`");
&[]
})
.iter()
.filter_map(|attr| {
let attr = match attr.meta_item() {
Some(meta_item) => meta_item,
_ => {
diag.span_err(attr.span(), "not a meta item");
return None;
}
};
let ident = match attr.ident() {
Some(ident) if attr.is_word() => ident,
_ => {
diag.span_err(attr.span, "must only be one word");
return None;
}
};
if !ident.name.can_be_raw() {
diag.span_err(
attr.span,
&format!("`{}` cannot be a name of derive helper attribute", ident),
);
}
Some(ident.name)
})
.collect()
} else {
Vec::new()
};
Some((trait_ident.name, proc_attrs))
}
Extend `proc_macro_back_compat` lint to `procedural-masquerade` We now lint on *any* use of `procedural-masquerade` crate. While this crate still exists, its main reverse dependency (`cssparser`) no longer depends on it. Any crates still depending off should stop doing so, as it only exists to support very old Rust versions. If a crate actually needs to support old versions of rustc via `procedural-masquerade`, then they'll just need to accept the warning until we remove it entirely (at the same time as the back-compat hack). The latest version of `procedural-masquerade` does not work with the latest rustc, but trying to check for the version seems like more trouble than it's worth. While working on this, I realized that the `proc-macro-hack` check was never actually doing anything. The corresponding enum variant in `proc-macro-hack` is named `Value` or `Nested` - it has never been called `Input`. Due to a strange Crater issue, the Crater run that tested adding this did *not* end up testing it - some of the crates that would have failed did not actually have their tests checked, making it seem as though the `proc-macro-hack` check was working. The Crater issue is being discussed at https://rust-lang.zulipchat.com/#narrow/stream/242791-t-infra/topic/Nearly.20identical.20Crater.20runs.20processed.20a.20crate.20differently/near/230406661 Despite the `proc-macro-hack` check not actually doing anything, we haven't gotten any reports from users about their build being broken. I went ahead and removed it entirely, since it's clear that no one is being affected by the `proc-macro-hack` regression in practice.
2021-03-15 15:54:25 -04:00
/// This nonterminal looks like some specific enums from
/// `proc-macro-hack` and `procedural-masquerade` crates.
/// We need to maintain some special pretty-printing behavior for them due to incorrect
/// asserts in old versions of those crates and their wide use in the ecosystem.
/// See issue #73345 for more details.
/// FIXME(#73933): Remove this eventually.
pub(crate) fn pretty_printing_compatibility_hack(nt: &Nonterminal, sess: &ParseSess) -> bool {
let item = match nt {
Nonterminal::NtItem(item) => item,
Nonterminal::NtStmt(stmt) => match &stmt.kind {
ast::StmtKind::Item(item) => item,
_ => return false,
},
_ => return false,
};
let name = item.ident.name;
if name == sym::ProceduralMasqueradeDummyType {
if let ast::ItemKind::Enum(enum_def, _) = &item.kind {
if let [variant] = &*enum_def.variants {
if variant.ident.name == sym::Input {
sess.buffer_lint_with_diagnostic(
&PROC_MACRO_BACK_COMPAT,
item.ident.span,
ast::CRATE_NODE_ID,
"using `procedural-masquerade` crate",
BuiltinLintDiagnostics::ProcMacroBackCompat(
"The `procedural-masquerade` crate has been unnecessary since Rust 1.30.0. \
Versions of this crate below 0.1.7 will eventually stop compiling.".to_string())
);
return true;
}
}
}
}
false
}