diff options
Diffstat (limited to 'compiler/rustc_parse/src')
| -rw-r--r-- | compiler/rustc_parse/src/errors.rs | 27 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/lib.rs | 5 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/attr.rs | 6 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/attr_wrapper.rs | 22 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/diagnostics.rs | 58 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/expr.rs | 19 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/item.rs | 45 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/mod.rs | 173 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/nonterminal.rs | 2 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/pat.rs | 76 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/stmt.rs | 3 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/validate_attr.rs | 11 |
12 files changed, 226 insertions, 221 deletions
diff --git a/compiler/rustc_parse/src/errors.rs b/compiler/rustc_parse/src/errors.rs index 06c09960727..26f38c9156a 100644 --- a/compiler/rustc_parse/src/errors.rs +++ b/compiler/rustc_parse/src/errors.rs @@ -1435,6 +1435,13 @@ pub(crate) struct AsyncBlockIn2015 { } #[derive(Diagnostic)] +#[diag(parse_async_move_block_in_2015)] +pub(crate) struct AsyncMoveBlockIn2015 { + #[primary_span] + pub span: Span, +} + +#[derive(Diagnostic)] #[diag(parse_self_argument_pointer)] pub(crate) struct SelfArgumentPointer { #[primary_span] @@ -1809,6 +1816,12 @@ pub struct UnknownPrefix<'a> { } #[derive(Subdiagnostic)] +#[note(parse_macro_expands_to_adt_field)] +pub struct MacroExpandsToAdtField<'a> { + pub adt_ty: &'a str, +} + +#[derive(Subdiagnostic)] pub enum UnknownPrefixSugg { #[suggestion( parse_suggestion_br, @@ -2731,3 +2744,17 @@ pub(crate) struct WhereClauseBeforeConstBodySugg { #[suggestion_part(code = "")] pub right: Span, } + +#[derive(Diagnostic)] +#[diag(parse_generic_args_in_pat_require_turbofish_syntax)] +pub(crate) struct GenericArgsInPatRequireTurbofishSyntax { + #[primary_span] + pub span: Span, + #[suggestion( + parse_sugg_turbofish_syntax, + style = "verbose", + code = "::", + applicability = "maybe-incorrect" + )] + pub suggest_turbofish: Span, +} diff --git a/compiler/rustc_parse/src/lib.rs b/compiler/rustc_parse/src/lib.rs index 47233d0615c..892be36aae7 100644 --- a/compiler/rustc_parse/src/lib.rs +++ b/compiler/rustc_parse/src/lib.rs @@ -8,6 +8,7 @@ #![feature(never_type)] #![feature(rustc_attrs)] #![recursion_limit = "256"] +#![cfg_attr(not(bootstrap), allow(internal_features))] #[macro_use] extern crate tracing; @@ -205,7 +206,7 @@ pub fn stream_to_parser<'a>( stream: TokenStream, subparser_name: Option<&'static str>, ) -> Parser<'a> { - Parser::new(sess, stream, false, subparser_name) + Parser::new(sess, stream, subparser_name) } /// Runs the given subparser `f` on the tokens of the given `attr`'s item. @@ -215,7 +216,7 @@ pub fn parse_in<'a, T>( name: &'static str, mut f: impl FnMut(&mut Parser<'a>) -> PResult<'a, T>, ) -> PResult<'a, T> { - let mut parser = Parser::new(sess, tts, false, Some(name)); + let mut parser = Parser::new(sess, tts, Some(name)); let result = f(&mut parser)?; if parser.token != token::Eof { parser.unexpected()?; diff --git a/compiler/rustc_parse/src/parser/attr.rs b/compiler/rustc_parse/src/parser/attr.rs index ee0abba1c17..104de47b97d 100644 --- a/compiler/rustc_parse/src/parser/attr.rs +++ b/compiler/rustc_parse/src/parser/attr.rs @@ -36,7 +36,7 @@ impl<'a> Parser<'a> { pub(super) fn parse_outer_attributes(&mut self) -> PResult<'a, AttrWrapper> { let mut outer_attrs = ast::AttrVec::new(); let mut just_parsed_doc_comment = false; - let start_pos = self.token_cursor.num_next_calls; + let start_pos = self.num_bump_calls; loop { let attr = if self.check(&token::Pound) { let prev_outer_attr_sp = outer_attrs.last().map(|attr| attr.span); @@ -277,7 +277,7 @@ impl<'a> Parser<'a> { pub(crate) fn parse_inner_attributes(&mut self) -> PResult<'a, ast::AttrVec> { let mut attrs = ast::AttrVec::new(); loop { - let start_pos: u32 = self.token_cursor.num_next_calls.try_into().unwrap(); + let start_pos: u32 = self.num_bump_calls.try_into().unwrap(); // Only try to parse if it is an inner attribute (has `!`). let attr = if self.check(&token::Pound) && self.look_ahead(1, |t| t == &token::Not) { Some(self.parse_attribute(InnerAttrPolicy::Permitted)?) @@ -298,7 +298,7 @@ impl<'a> Parser<'a> { None }; if let Some(attr) = attr { - let end_pos: u32 = self.token_cursor.num_next_calls.try_into().unwrap(); + let end_pos: u32 = self.num_bump_calls.try_into().unwrap(); // If we are currently capturing tokens, mark the location of this inner attribute. // If capturing ends up creating a `LazyAttrTokenStream`, we will include // this replace range with it, removing the inner attribute from the final diff --git a/compiler/rustc_parse/src/parser/attr_wrapper.rs b/compiler/rustc_parse/src/parser/attr_wrapper.rs index 158ab2a2956..5d6c574baa6 100644 --- a/compiler/rustc_parse/src/parser/attr_wrapper.rs +++ b/compiler/rustc_parse/src/parser/attr_wrapper.rs @@ -213,6 +213,7 @@ impl<'a> Parser<'a> { let start_token = (self.token.clone(), self.token_spacing); let cursor_snapshot = self.token_cursor.clone(); + let start_pos = self.num_bump_calls; let has_outer_attrs = !attrs.attrs.is_empty(); let prev_capturing = std::mem::replace(&mut self.capture_state.capturing, Capturing::Yes); @@ -273,8 +274,7 @@ impl<'a> Parser<'a> { let replace_ranges_end = self.capture_state.replace_ranges.len(); - let cursor_snapshot_next_calls = cursor_snapshot.num_next_calls; - let mut end_pos = self.token_cursor.num_next_calls; + let mut end_pos = self.num_bump_calls; let mut captured_trailing = false; @@ -301,12 +301,12 @@ impl<'a> Parser<'a> { // then extend the range of captured tokens to include it, since the parser // was not actually bumped past it. When the `LazyAttrTokenStream` gets converted // into an `AttrTokenStream`, we will create the proper token. - if self.token_cursor.break_last_token { + if self.break_last_token { assert!(!captured_trailing, "Cannot set break_last_token and have trailing token"); end_pos += 1; } - let num_calls = end_pos - cursor_snapshot_next_calls; + let num_calls = end_pos - start_pos; // If we have no attributes, then we will never need to // use any replace ranges. @@ -316,7 +316,7 @@ impl<'a> Parser<'a> { // Grab any replace ranges that occur *inside* the current AST node. // We will perform the actual replacement when we convert the `LazyAttrTokenStream` // to an `AttrTokenStream`. - let start_calls: u32 = cursor_snapshot_next_calls.try_into().unwrap(); + let start_calls: u32 = start_pos.try_into().unwrap(); self.capture_state.replace_ranges[replace_ranges_start..replace_ranges_end] .iter() .cloned() @@ -331,7 +331,7 @@ impl<'a> Parser<'a> { start_token, num_calls, cursor_snapshot, - break_last_token: self.token_cursor.break_last_token, + break_last_token: self.break_last_token, replace_ranges, }); @@ -359,14 +359,10 @@ impl<'a> Parser<'a> { // with a `FlatToken::AttrTarget`. If this AST node is inside an item // that has `#[derive]`, then this will allow us to cfg-expand this // AST node. - let start_pos = - if has_outer_attrs { attrs.start_pos } else { cursor_snapshot_next_calls }; + let start_pos = if has_outer_attrs { attrs.start_pos } else { start_pos }; let new_tokens = vec![(FlatToken::AttrTarget(attr_data), Spacing::Alone)]; - assert!( - !self.token_cursor.break_last_token, - "Should not have unglued last token with cfg attr" - ); + assert!(!self.break_last_token, "Should not have unglued last token with cfg attr"); let range: Range<u32> = (start_pos.try_into().unwrap())..(end_pos.try_into().unwrap()); self.capture_state.replace_ranges.push((range, new_tokens)); self.capture_state.replace_ranges.extend(inner_attr_replace_ranges); @@ -464,6 +460,6 @@ mod size_asserts { use rustc_data_structures::static_assert_size; // tidy-alphabetical-start static_assert_size!(AttrWrapper, 16); - static_assert_size!(LazyAttrTokenStreamImpl, 120); + static_assert_size!(LazyAttrTokenStreamImpl, 104); // tidy-alphabetical-end } diff --git a/compiler/rustc_parse/src/parser/diagnostics.rs b/compiler/rustc_parse/src/parser/diagnostics.rs index e6de51a673c..4e639a54cf7 100644 --- a/compiler/rustc_parse/src/parser/diagnostics.rs +++ b/compiler/rustc_parse/src/parser/diagnostics.rs @@ -4,10 +4,11 @@ use super::{ TokenExpectType, TokenType, }; use crate::errors::{ - AmbiguousPlus, AttributeOnParamType, BadQPathStage2, BadTypePlus, BadTypePlusSub, ColonAsSemi, - ComparisonOperatorsCannotBeChained, ComparisonOperatorsCannotBeChainedSugg, - ConstGenericWithoutBraces, ConstGenericWithoutBracesSugg, DocCommentDoesNotDocumentAnything, - DocCommentOnParamType, DoubleColonInBound, ExpectedIdentifier, ExpectedSemi, ExpectedSemiSugg, + AmbiguousPlus, AsyncMoveBlockIn2015, AttributeOnParamType, BadQPathStage2, BadTypePlus, + BadTypePlusSub, ColonAsSemi, ComparisonOperatorsCannotBeChained, + ComparisonOperatorsCannotBeChainedSugg, ConstGenericWithoutBraces, + ConstGenericWithoutBracesSugg, DocCommentDoesNotDocumentAnything, DocCommentOnParamType, + DoubleColonInBound, ExpectedIdentifier, ExpectedSemi, ExpectedSemiSugg, GenericParamsWithoutAngleBrackets, GenericParamsWithoutAngleBracketsSugg, HelpIdentifierStartsWithNumber, InInTypo, IncorrectAwait, IncorrectSemicolon, IncorrectUseOfAwait, ParenthesesInForHead, ParenthesesInForHeadSugg, @@ -573,6 +574,12 @@ impl<'a> Parser<'a> { return Err(self.sess.create_err(UseEqInstead { span: self.token.span })); } + if self.token.is_keyword(kw::Move) && self.prev_token.is_keyword(kw::Async) { + // The 2015 edition is in use because parsing of `async move` has failed. + let span = self.prev_token.span.to(self.token.span); + return Err(self.sess.create_err(AsyncMoveBlockIn2015 { span })); + } + let expect = tokens_to_string(&expected); let actual = super::token_descr(&self.token); let (msg_exp, (label_sp, label_exp)) = if expected.len() > 1 { @@ -2100,7 +2107,7 @@ impl<'a> Parser<'a> { } pub(super) fn recover_arg_parse(&mut self) -> PResult<'a, (P<ast::Pat>, P<ast::Ty>)> { - let pat = self.parse_pat_no_top_alt(Some(Expected::ArgumentName))?; + let pat = self.parse_pat_no_top_alt(Some(Expected::ArgumentName), None)?; self.expect(&token::Colon)?; let ty = self.parse_ty()?; @@ -2508,7 +2515,7 @@ impl<'a> Parser<'a> { // Skip the `:`. snapshot_pat.bump(); snapshot_type.bump(); - match snapshot_pat.parse_pat_no_top_alt(expected) { + match snapshot_pat.parse_pat_no_top_alt(expected, None) { Err(inner_err) => { inner_err.cancel(); } @@ -2634,6 +2641,7 @@ impl<'a> Parser<'a> { pub(crate) fn maybe_recover_unexpected_comma( &mut self, lo: Span, + is_mac_invoc: bool, rt: CommaRecoveryMode, ) -> PResult<'a, ()> { if self.token != token::Comma { @@ -2654,24 +2662,28 @@ impl<'a> Parser<'a> { let seq_span = lo.to(self.prev_token.span); let mut err = self.struct_span_err(comma_span, "unexpected `,` in pattern"); if let Ok(seq_snippet) = self.span_to_snippet(seq_span) { - err.multipart_suggestion( - format!( - "try adding parentheses to match on a tuple{}", - if let CommaRecoveryMode::LikelyTuple = rt { "" } else { "..." }, - ), - vec![ - (seq_span.shrink_to_lo(), "(".to_string()), - (seq_span.shrink_to_hi(), ")".to_string()), - ], - Applicability::MachineApplicable, - ); - if let CommaRecoveryMode::EitherTupleOrPipe = rt { - err.span_suggestion( - seq_span, - "...or a vertical bar to match on multiple alternatives", - seq_snippet.replace(',', " |"), + if is_mac_invoc { + err.note(fluent::parse_macro_expands_to_match_arm); + } else { + err.multipart_suggestion( + format!( + "try adding parentheses to match on a tuple{}", + if let CommaRecoveryMode::LikelyTuple = rt { "" } else { "..." }, + ), + vec![ + (seq_span.shrink_to_lo(), "(".to_string()), + (seq_span.shrink_to_hi(), ")".to_string()), + ], Applicability::MachineApplicable, ); + if let CommaRecoveryMode::EitherTupleOrPipe = rt { + err.span_suggestion( + seq_span, + "...or a vertical bar to match on multiple alternatives", + seq_snippet.replace(',', " |"), + Applicability::MachineApplicable, + ); + } } } Err(err) @@ -2772,7 +2784,7 @@ impl<'a> Parser<'a> { /// sequence of patterns until `)` is reached. fn skip_pat_list(&mut self) -> PResult<'a, ()> { while !self.check(&token::CloseDelim(Delimiter::Parenthesis)) { - self.parse_pat_no_top_alt(None)?; + self.parse_pat_no_top_alt(None, None)?; if !self.eat(&token::Comma) { return Ok(()); } diff --git a/compiler/rustc_parse/src/parser/expr.rs b/compiler/rustc_parse/src/parser/expr.rs index 55f857aa31c..e409c7c6781 100644 --- a/compiler/rustc_parse/src/parser/expr.rs +++ b/compiler/rustc_parse/src/parser/expr.rs @@ -857,7 +857,7 @@ impl<'a> Parser<'a> { let msg = format!( "cast cannot be followed by {}", match with_postfix.kind { - ExprKind::Index(_, _) => "indexing", + ExprKind::Index(..) => "indexing", ExprKind::Try(_) => "`?`", ExprKind::Field(_, _) => "a field access", ExprKind::MethodCall(_) => "a method call", @@ -1167,7 +1167,7 @@ impl<'a> Parser<'a> { DestructuredFloat::TrailingDot(sym, sym_span, dot_span) => { assert!(suffix.is_none()); // Analogous to `Self::break_and_eat` - self.token_cursor.break_last_token = true; + self.break_last_token = true; // This might work, in cases like `1. 2`, and might not, // in cases like `offset_of!(Ty, 1.)`. It depends on what comes // after the float-like token, and therefore we have to make @@ -1304,7 +1304,10 @@ impl<'a> Parser<'a> { let index = self.parse_expr()?; self.suggest_missing_semicolon_before_array(prev_span, open_delim_span)?; self.expect(&token::CloseDelim(Delimiter::Bracket))?; - Ok(self.mk_expr(lo.to(self.prev_token.span), self.mk_index(base, index))) + Ok(self.mk_expr( + lo.to(self.prev_token.span), + self.mk_index(base, index, open_delim_span.to(self.prev_token.span)), + )) } /// Assuming we have just parsed `.`, continue parsing into an expression. @@ -2338,7 +2341,7 @@ impl<'a> Parser<'a> { let lo = self.token.span; let attrs = self.parse_outer_attributes()?; self.collect_tokens_trailing_token(attrs, ForceCollect::No, |this, attrs| { - let pat = this.parse_pat_no_top_alt(Some(Expected::ParameterName))?; + let pat = this.parse_pat_no_top_alt(Some(Expected::ParameterName), None)?; let ty = if this.eat(&token::Colon) { this.parse_ty()? } else { @@ -2599,7 +2602,7 @@ impl<'a> Parser<'a> { // Recover from missing expression in `for` loop if matches!(expr.kind, ExprKind::Block(..)) - && !matches!(self.token.kind, token::OpenDelim(token::Delimiter::Brace)) + && !matches!(self.token.kind, token::OpenDelim(Delimiter::Brace)) && self.may_recover() { self.sess @@ -2781,7 +2784,7 @@ impl<'a> Parser<'a> { return None; } let pre_pat_snapshot = self.create_snapshot_for_diagnostic(); - match self.parse_pat_no_top_alt(None) { + match self.parse_pat_no_top_alt(None, None) { Ok(_pat) => { if self.token.kind == token::FatArrow { // Reached arm end. @@ -3366,8 +3369,8 @@ impl<'a> Parser<'a> { ExprKind::Binary(binop, lhs, rhs) } - fn mk_index(&self, expr: P<Expr>, idx: P<Expr>) -> ExprKind { - ExprKind::Index(expr, idx) + fn mk_index(&self, expr: P<Expr>, idx: P<Expr>, brackets_span: Span) -> ExprKind { + ExprKind::Index(expr, idx, brackets_span) } fn mk_call(&self, f: P<Expr>, args: ThinVec<P<Expr>>) -> ExprKind { diff --git a/compiler/rustc_parse/src/parser/item.rs b/compiler/rustc_parse/src/parser/item.rs index 1301ed3e388..24c65d061f9 100644 --- a/compiler/rustc_parse/src/parser/item.rs +++ b/compiler/rustc_parse/src/parser/item.rs @@ -1,20 +1,20 @@ -use crate::errors; - use super::diagnostics::{dummy_arg, ConsumeClosingDelim}; use super::ty::{AllowPlus, RecoverQPath, RecoverReturnSign}; use super::{AttrWrapper, FollowedByType, ForceCollect, Parser, PathStyle, TrailingToken}; +use crate::errors::{self, MacroExpandsToAdtField}; +use crate::fluent_generated as fluent; use ast::StaticItem; use rustc_ast::ast::*; use rustc_ast::ptr::P; use rustc_ast::token::{self, Delimiter, TokenKind}; use rustc_ast::tokenstream::{DelimSpan, TokenStream, TokenTree}; use rustc_ast::util::case::Case; +use rustc_ast::MacCall; use rustc_ast::{self as ast, AttrVec, Attribute, DUMMY_NODE_ID}; use rustc_ast::{Async, Const, Defaultness, IsAuto, Mutability, Unsafe, UseTree, UseTreeKind}; use rustc_ast::{BindingAnnotation, Block, FnDecl, FnSig, Param, SelfKind}; use rustc_ast::{EnumDef, FieldDef, Generics, TraitRef, Ty, TyKind, Variant, VariantData}; use rustc_ast::{FnHeader, ForeignItem, Path, PathSegment, Visibility, VisibilityKind}; -use rustc_ast::{MacCall, MacDelimiter}; use rustc_ast_pretty::pprust; use rustc_errors::{ struct_span_err, Applicability, DiagnosticBuilder, ErrorGuaranteed, IntoDiagnostic, PResult, @@ -1450,6 +1450,17 @@ impl<'a> Parser<'a> { } let ident = this.parse_field_ident("enum", vlo)?; + if this.token == token::Not { + if let Err(mut err) = this.unexpected::<()>() { + err.note(fluent::parse_macro_expands_to_enum_variant).emit(); + } + + this.bump(); + this.parse_delim_args()?; + + return Ok((None, TrailingToken::MaybeComma)); + } + let struct_def = if this.check(&token::OpenDelim(Delimiter::Brace)) { // Parse a struct variant. let (fields, recovered) = @@ -1477,7 +1488,7 @@ impl<'a> Parser<'a> { Ok((Some(vr), TrailingToken::MaybeComma)) }, - ).map_err(|mut err|{ + ).map_err(|mut err| { err.help("enum variants can be `Variant`, `Variant = <integer>`, `Variant(Type, ..., TypeN)` or `Variant { fields: Types }`"); err }) @@ -1687,7 +1698,8 @@ impl<'a> Parser<'a> { self.collect_tokens_trailing_token(attrs, ForceCollect::No, |this, attrs| { let lo = this.token.span; let vis = this.parse_visibility(FollowedByType::No)?; - Ok((this.parse_single_struct_field(adt_ty, lo, vis, attrs)?, TrailingToken::None)) + this.parse_single_struct_field(adt_ty, lo, vis, attrs) + .map(|field| (field, TrailingToken::None)) }) } @@ -1821,8 +1833,8 @@ impl<'a> Parser<'a> { "field names and their types are separated with `:`", ":", Applicability::MachineApplicable, - ); - err.emit(); + ) + .emit(); } else { return Err(err); } @@ -1839,6 +1851,23 @@ impl<'a> Parser<'a> { attrs: AttrVec, ) -> PResult<'a, FieldDef> { let name = self.parse_field_ident(adt_ty, lo)?; + // Parse the macro invocation and recover + if self.token.kind == token::Not { + if let Err(mut err) = self.unexpected::<FieldDef>() { + err.subdiagnostic(MacroExpandsToAdtField { adt_ty }).emit(); + self.bump(); + self.parse_delim_args()?; + return Ok(FieldDef { + span: DUMMY_SP, + ident: None, + vis, + id: DUMMY_NODE_ID, + ty: self.mk_ty(DUMMY_SP, TyKind::Err), + attrs, + is_placeholder: false, + }); + } + } self.expect_field_ty_separator()?; let ty = self.parse_ty()?; if self.token.kind == token::Colon && self.look_ahead(1, |tok| tok.kind != token::Colon) { @@ -1968,7 +1997,7 @@ impl<'a> Parser<'a> { let arrow = TokenTree::token_alone(token::FatArrow, pspan.between(bspan)); // `=>` let tokens = TokenStream::new(vec![params, arrow, body]); let dspan = DelimSpan::from_pair(pspan.shrink_to_lo(), bspan.shrink_to_hi()); - P(DelimArgs { dspan, delim: MacDelimiter::Brace, tokens }) + P(DelimArgs { dspan, delim: Delimiter::Brace, tokens }) } else { return self.unexpected(); }; diff --git a/compiler/rustc_parse/src/parser/mod.rs b/compiler/rustc_parse/src/parser/mod.rs index bd547ed790a..c5b46b809b1 100644 --- a/compiler/rustc_parse/src/parser/mod.rs +++ b/compiler/rustc_parse/src/parser/mod.rs @@ -24,8 +24,8 @@ use rustc_ast::tokenstream::{TokenStream, TokenTree, TokenTreeCursor}; use rustc_ast::util::case::Case; use rustc_ast::AttrId; use rustc_ast::DUMMY_NODE_ID; -use rustc_ast::{self as ast, AnonConst, AttrStyle, Const, DelimArgs, Extern}; -use rustc_ast::{Async, AttrArgs, AttrArgsEq, Expr, ExprKind, MacDelimiter, Mutability, StrLit}; +use rustc_ast::{self as ast, AnonConst, Const, DelimArgs, Extern}; +use rustc_ast::{Async, AttrArgs, AttrArgsEq, Expr, ExprKind, Mutability, StrLit}; use rustc_ast::{HasAttrs, HasTokens, Unsafe, Visibility, VisibilityKind}; use rustc_ast_pretty::pprust; use rustc_data_structures::fx::FxHashMap; @@ -38,7 +38,7 @@ use rustc_session::parse::ParseSess; use rustc_span::source_map::{Span, DUMMY_SP}; use rustc_span::symbol::{kw, sym, Ident, Symbol}; use std::ops::Range; -use std::{cmp, mem, slice}; +use std::{mem, slice}; use thin_vec::ThinVec; use tracing::debug; @@ -135,9 +135,24 @@ pub struct Parser<'a> { pub capture_cfg: bool, restrictions: Restrictions, expected_tokens: Vec<TokenType>, - // Important: This must only be advanced from `bump` to ensure that - // `token_cursor.num_next_calls` is updated properly. token_cursor: TokenCursor, + // The number of calls to `bump`, i.e. the position in the token stream. + num_bump_calls: usize, + // During parsing we may sometimes need to 'unglue' a glued token into two + // component tokens (e.g. '>>' into '>' and '>), so the parser can consume + // them one at a time. This process bypasses the normal capturing mechanism + // (e.g. `num_bump_calls` will not be incremented), since the 'unglued' + // tokens due not exist in the original `TokenStream`. + // + // If we end up consuming both unglued tokens, this is not an issue. We'll + // end up capturing the single 'glued' token. + // + // However, sometimes we may want to capture just the first 'unglued' + // token. For example, capturing the `Vec<u8>` in `Option<Vec<u8>>` + // requires us to unglue the trailing `>>` token. The `break_last_token` + // field is used to track this token. It gets appended to the captured + // stream when we evaluate a `LazyAttrTokenStream`. + break_last_token: bool, /// This field is used to keep track of how many left angle brackets we have seen. This is /// required in order to detect extra leading left angle brackets (`<` characters) and error /// appropriately. @@ -161,7 +176,7 @@ pub struct Parser<'a> { // This type is used a lot, e.g. it's cloned when matching many declarative macro rules with nonterminals. Make sure // it doesn't unintentionally get bigger. #[cfg(all(target_arch = "x86_64", target_pointer_width = "64"))] -rustc_data_structures::static_assert_size!(Parser<'_>, 272); +rustc_data_structures::static_assert_size!(Parser<'_>, 264); /// Stores span information about a closure. #[derive(Clone)] @@ -223,37 +238,6 @@ struct TokenCursor { // tokens are in `stack[n-1]`. `stack[0]` (when present) has no delimiters // because it's the outermost token stream which never has delimiters. stack: Vec<(TokenTreeCursor, Delimiter, DelimSpan)>, - - // We need to desugar doc comments from `/// foo` form into `#[doc = - // r"foo"]` form when parsing declarative macro inputs in `parse_tt`, - // because some declarative macros look for `doc` attributes. - desugar_doc_comments: bool, - - // Counts the number of calls to `{,inlined_}next`. - num_next_calls: usize, - - // During parsing, we may sometimes need to 'unglue' a - // glued token into two component tokens - // (e.g. '>>' into '>' and '>), so that the parser - // can consume them one at a time. This process - // bypasses the normal capturing mechanism - // (e.g. `num_next_calls` will not be incremented), - // since the 'unglued' tokens due not exist in - // the original `TokenStream`. - // - // If we end up consuming both unglued tokens, - // then this is not an issue - we'll end up - // capturing the single 'glued' token. - // - // However, in certain circumstances, we may - // want to capture just the first 'unglued' token. - // For example, capturing the `Vec<u8>` - // in `Option<Vec<u8>>` requires us to unglue - // the trailing `>>` token. The `break_last_token` - // field is used to track this token - it gets - // appended to the captured stream when - // we evaluate a `LazyAttrTokenStream`. - break_last_token: bool, } impl TokenCursor { @@ -265,29 +249,17 @@ impl TokenCursor { #[inline(always)] fn inlined_next(&mut self) -> (Token, Spacing) { loop { - // FIXME: we currently don't return `Delimiter` open/close delims. To fix #67062 we will - // need to, whereupon the `delim != Delimiter::Invisible` conditions below can be - // removed. + // FIXME: we currently don't return `Delimiter::Invisible` open/close delims. To fix + // #67062 we will need to, whereupon the `delim != Delimiter::Invisible` conditions + // below can be removed. if let Some(tree) = self.tree_cursor.next_ref() { match tree { &TokenTree::Token(ref token, spacing) => { - match (self.desugar_doc_comments, token) { - ( - true, - &Token { kind: token::DocComment(_, attr_style, data), span }, - ) => { - let desugared = self.desugar(attr_style, data, span); - self.tree_cursor.replace_prev_and_rewind(desugared); - // Continue to get the first token of the desugared doc comment. - } - _ => { - debug_assert!(!matches!( - token.kind, - token::OpenDelim(_) | token::CloseDelim(_) - )); - return (token.clone(), spacing); - } - } + debug_assert!(!matches!( + token.kind, + token::OpenDelim(_) | token::CloseDelim(_) + )); + return (token.clone(), spacing); } &TokenTree::Delimited(sp, delim, ref tts) => { let trees = tts.clone().into_trees(); @@ -311,52 +283,6 @@ impl TokenCursor { } } } - - // Desugar a doc comment into something like `#[doc = r"foo"]`. - fn desugar(&mut self, attr_style: AttrStyle, data: Symbol, span: Span) -> Vec<TokenTree> { - // Searches for the occurrences of `"#*` and returns the minimum number of `#`s - // required to wrap the text. E.g. - // - `abc d` is wrapped as `r"abc d"` (num_of_hashes = 0) - // - `abc "d"` is wrapped as `r#"abc "d""#` (num_of_hashes = 1) - // - `abc "##d##"` is wrapped as `r###"abc ##"d"##"###` (num_of_hashes = 3) - let mut num_of_hashes = 0; - let mut count = 0; - for ch in data.as_str().chars() { - count = match ch { - '"' => 1, - '#' if count > 0 => count + 1, - _ => 0, - }; - num_of_hashes = cmp::max(num_of_hashes, count); - } - - // `/// foo` becomes `doc = r"foo"`. - let delim_span = DelimSpan::from_single(span); - let body = TokenTree::Delimited( - delim_span, - Delimiter::Bracket, - [ - TokenTree::token_alone(token::Ident(sym::doc, false), span), - TokenTree::token_alone(token::Eq, span), - TokenTree::token_alone( - TokenKind::lit(token::StrRaw(num_of_hashes), data, None), - span, - ), - ] - .into_iter() - .collect::<TokenStream>(), - ); - - if attr_style == AttrStyle::Inner { - vec![ - TokenTree::token_alone(token::Pound, span), - TokenTree::token_alone(token::Not, span), - body, - ] - } else { - vec![TokenTree::token_alone(token::Pound, span), body] - } - } } #[derive(Debug, Clone, PartialEq)] @@ -451,8 +377,7 @@ pub(super) fn token_descr(token: &Token) -> String { impl<'a> Parser<'a> { pub fn new( sess: &'a ParseSess, - tokens: TokenStream, - desugar_doc_comments: bool, + stream: TokenStream, subparser_name: Option<&'static str>, ) -> Self { let mut parser = Parser { @@ -463,13 +388,9 @@ impl<'a> Parser<'a> { capture_cfg: false, restrictions: Restrictions::empty(), expected_tokens: Vec::new(), - token_cursor: TokenCursor { - tree_cursor: tokens.into_trees(), - stack: Vec::new(), - num_next_calls: 0, - desugar_doc_comments, - break_last_token: false, - }, + token_cursor: TokenCursor { tree_cursor: stream.into_trees(), stack: Vec::new() }, + num_bump_calls: 0, + break_last_token: false, unmatched_angle_bracket_count: 0, max_angle_bracket_count: 0, last_unexpected_token_span: None, @@ -772,7 +693,7 @@ impl<'a> Parser<'a> { // If we consume any additional tokens, then this token // is not needed (we'll capture the entire 'glued' token), // and `bump` will set this field to `None` - self.token_cursor.break_last_token = true; + self.break_last_token = true; // Use the spacing of the glued token as the spacing // of the unglued second token. self.bump_with((Token::new(second, second_span), self.token_spacing)); @@ -1114,11 +1035,11 @@ impl<'a> Parser<'a> { // Note: destructuring here would give nicer code, but it was found in #96210 to be slower // than `.0`/`.1` access. let mut next = self.token_cursor.inlined_next(); - self.token_cursor.num_next_calls += 1; + self.num_bump_calls += 1; // We've retrieved an token from the underlying // cursor, so we no longer need to worry about // an unglued token. See `break_and_eat` for more details - self.token_cursor.break_last_token = false; + self.break_last_token = false; if next.0.span.is_dummy() { // Tweak the location for better diagnostics, but keep syntactic context intact. let fallback_span = self.token.span; @@ -1172,7 +1093,7 @@ impl<'a> Parser<'a> { } i += 1; } - return looker(&token); + looker(&token) } /// Returns whether any of the given keywords are `dist` tokens ahead of the current one. @@ -1295,12 +1216,10 @@ impl<'a> Parser<'a> { || self.check(&token::OpenDelim(Delimiter::Brace)); delimited.then(|| { - // We've confirmed above that there is a delimiter so unwrapping is OK. let TokenTree::Delimited(dspan, delim, tokens) = self.parse_token_tree() else { unreachable!() }; - - DelimArgs { dspan, delim: MacDelimiter::from_token(delim).unwrap(), tokens } + DelimArgs { dspan, delim, tokens } }) } @@ -1316,12 +1235,11 @@ impl<'a> Parser<'a> { } /// Parses a single token tree from the input. - pub(crate) fn parse_token_tree(&mut self) -> TokenTree { + pub fn parse_token_tree(&mut self) -> TokenTree { match self.token.kind { token::OpenDelim(..) => { // Grab the tokens within the delimiters. - let tree_cursor = &self.token_cursor.tree_cursor; - let stream = tree_cursor.stream.clone(); + let stream = self.token_cursor.tree_cursor.stream.clone(); let (_, delim, span) = *self.token_cursor.stack.last().unwrap(); // Advance the token cursor through the entire delimited @@ -1352,15 +1270,6 @@ impl<'a> Parser<'a> { } } - /// Parses a stream of tokens into a list of `TokenTree`s, up to EOF. - pub fn parse_all_token_trees(&mut self) -> PResult<'a, Vec<TokenTree>> { - let mut tts = Vec::new(); - while self.token != token::Eof { - tts.push(self.parse_token_tree()); - } - Ok(tts) - } - pub fn parse_tokens(&mut self) -> TokenStream { let mut result = Vec::new(); loop { @@ -1520,7 +1429,7 @@ impl<'a> Parser<'a> { } pub fn approx_token_stream_pos(&self) -> usize { - self.token_cursor.num_next_calls + self.num_bump_calls } } diff --git a/compiler/rustc_parse/src/parser/nonterminal.rs b/compiler/rustc_parse/src/parser/nonterminal.rs index adb0d372a40..f5681532b3a 100644 --- a/compiler/rustc_parse/src/parser/nonterminal.rs +++ b/compiler/rustc_parse/src/parser/nonterminal.rs @@ -131,7 +131,7 @@ impl<'a> Parser<'a> { }, NonterminalKind::PatParam { .. } | NonterminalKind::PatWithOr { .. } => { token::NtPat(self.collect_tokens_no_attrs(|this| match kind { - NonterminalKind::PatParam { .. } => this.parse_pat_no_top_alt(None), + NonterminalKind::PatParam { .. } => this.parse_pat_no_top_alt(None, None), NonterminalKind::PatWithOr { .. } => this.parse_pat_allow_top_alt( None, RecoverComma::No, diff --git a/compiler/rustc_parse/src/parser/pat.rs b/compiler/rustc_parse/src/parser/pat.rs index fed16278db5..4aadb7d7ca5 100644 --- a/compiler/rustc_parse/src/parser/pat.rs +++ b/compiler/rustc_parse/src/parser/pat.rs @@ -2,10 +2,11 @@ use super::{ForceCollect, Parser, PathStyle, TrailingToken}; use crate::errors::{ self, AmbiguousRangePattern, DotDotDotForRemainingFields, DotDotDotRangeToPatternNotAllowed, DotDotDotRestPattern, EnumPatternInsteadOfIdentifier, ExpectedBindingLeftOfAt, - ExpectedCommaAfterPatternField, InclusiveRangeExtraEquals, InclusiveRangeMatchArrow, - InclusiveRangeNoEnd, InvalidMutInPattern, PatternOnWrongSideOfAt, RefMutOrderIncorrect, - RemoveLet, RepeatedMutInPattern, TopLevelOrPatternNotAllowed, TopLevelOrPatternNotAllowedSugg, - TrailingVertNotAllowed, UnexpectedLifetimeInPattern, UnexpectedVertVertBeforeFunctionParam, + ExpectedCommaAfterPatternField, GenericArgsInPatRequireTurbofishSyntax, + InclusiveRangeExtraEquals, InclusiveRangeMatchArrow, InclusiveRangeNoEnd, InvalidMutInPattern, + PatternOnWrongSideOfAt, RefMutOrderIncorrect, RemoveLet, RepeatedMutInPattern, + TopLevelOrPatternNotAllowed, TopLevelOrPatternNotAllowedSugg, TrailingVertNotAllowed, + UnexpectedLifetimeInPattern, UnexpectedVertVertBeforeFunctionParam, UnexpectedVertVertInPattern, }; use crate::{maybe_recover_from_interpolated_ty_qpath, maybe_whole}; @@ -80,7 +81,8 @@ enum EatOrResult { } /// The syntax location of a given pattern. Used for diagnostics. -pub(super) enum PatternLocation { +#[derive(Clone, Copy)] +pub enum PatternLocation { LetBinding, FunctionParameter, } @@ -91,8 +93,12 @@ impl<'a> Parser<'a> { /// Corresponds to `pat<no_top_alt>` in RFC 2535 and does not admit or-patterns /// at the top level. Used when parsing the parameters of lambda expressions, /// functions, function pointers, and `pat` macro fragments. - pub fn parse_pat_no_top_alt(&mut self, expected: Option<Expected>) -> PResult<'a, P<Pat>> { - self.parse_pat_with_range_pat(true, expected) + pub fn parse_pat_no_top_alt( + &mut self, + expected: Option<Expected>, + syntax_loc: Option<PatternLocation>, + ) -> PResult<'a, P<Pat>> { + self.parse_pat_with_range_pat(true, expected, syntax_loc) } /// Parses a pattern. @@ -110,7 +116,7 @@ impl<'a> Parser<'a> { ra: RecoverColon, rt: CommaRecoveryMode, ) -> PResult<'a, P<Pat>> { - self.parse_pat_allow_top_alt_inner(expected, rc, ra, rt).map(|(pat, _)| pat) + self.parse_pat_allow_top_alt_inner(expected, rc, ra, rt, None).map(|(pat, _)| pat) } /// Returns the pattern and a bool indicating whether we recovered from a trailing vert (true = @@ -121,6 +127,7 @@ impl<'a> Parser<'a> { rc: RecoverComma, ra: RecoverColon, rt: CommaRecoveryMode, + syntax_loc: Option<PatternLocation>, ) -> PResult<'a, (P<Pat>, bool)> { // Keep track of whether we recovered from a trailing vert so that we can avoid duplicated // suggestions (which bothers rustfix). @@ -133,9 +140,13 @@ impl<'a> Parser<'a> { }; // Parse the first pattern (`p_0`). - let mut first_pat = self.parse_pat_no_top_alt(expected)?; + let mut first_pat = self.parse_pat_no_top_alt(expected, syntax_loc)?; if rc == RecoverComma::Yes { - self.maybe_recover_unexpected_comma(first_pat.span, rt)?; + self.maybe_recover_unexpected_comma( + first_pat.span, + matches!(first_pat.kind, PatKind::MacCall(_)), + rt, + )?; } // If the next token is not a `|`, @@ -172,12 +183,12 @@ impl<'a> Parser<'a> { break; } } - let pat = self.parse_pat_no_top_alt(expected).map_err(|mut err| { + let pat = self.parse_pat_no_top_alt(expected, syntax_loc).map_err(|mut err| { err.span_label(lo, WHILE_PARSING_OR_MSG); err })?; if rc == RecoverComma::Yes { - self.maybe_recover_unexpected_comma(pat.span, rt)?; + self.maybe_recover_unexpected_comma(pat.span, false, rt)?; } pats.push(pat); } @@ -208,6 +219,7 @@ impl<'a> Parser<'a> { rc, RecoverColon::No, CommaRecoveryMode::LikelyTuple, + Some(syntax_loc), )?; let colon = self.eat(&token::Colon); @@ -319,6 +331,7 @@ impl<'a> Parser<'a> { &mut self, allow_range_pat: bool, expected: Option<Expected>, + syntax_loc: Option<PatternLocation>, ) -> PResult<'a, P<Pat>> { maybe_recover_from_interpolated_ty_qpath!(self, true); maybe_whole!(self, NtPat, |x| x); @@ -358,11 +371,11 @@ impl<'a> Parser<'a> { // Parse _ PatKind::Wild } else if self.eat_keyword(kw::Mut) { - self.parse_pat_ident_mut()? + self.parse_pat_ident_mut(syntax_loc)? } else if self.eat_keyword(kw::Ref) { // Parse ref ident @ pat / ref mut ident @ pat let mutbl = self.parse_mutability(); - self.parse_pat_ident(BindingAnnotation(ByRef::Yes, mutbl))? + self.parse_pat_ident(BindingAnnotation(ByRef::Yes, mutbl), syntax_loc)? } else if self.eat_keyword(kw::Box) { self.parse_pat_box()? } else if self.check_inline_const(0) { @@ -384,7 +397,7 @@ impl<'a> Parser<'a> { // Parse `ident @ pat` // This can give false positives and parse nullary enums, // they are dealt with later in resolve. - self.parse_pat_ident(BindingAnnotation::NONE)? + self.parse_pat_ident(BindingAnnotation::NONE, syntax_loc)? } else if self.is_start_of_pat_with_path() { // Parse pattern starting with a path let (qself, path) = if self.eat_lt() { @@ -485,7 +498,7 @@ impl<'a> Parser<'a> { // At this point we attempt to parse `@ $pat_rhs` and emit an error. self.bump(); // `@` - let mut rhs = self.parse_pat_no_top_alt(None)?; + let mut rhs = self.parse_pat_no_top_alt(None, None)?; let whole_span = lhs.span.to(rhs.span); if let PatKind::Ident(_, _, sub @ None) = &mut rhs.kind { @@ -541,7 +554,7 @@ impl<'a> Parser<'a> { } let mutbl = self.parse_mutability(); - let subpat = self.parse_pat_with_range_pat(false, expected)?; + let subpat = self.parse_pat_with_range_pat(false, expected, None)?; Ok(PatKind::Ref(subpat, mutbl)) } @@ -566,12 +579,12 @@ impl<'a> Parser<'a> { } /// Parse a mutable binding with the `mut` token already eaten. - fn parse_pat_ident_mut(&mut self) -> PResult<'a, PatKind> { + fn parse_pat_ident_mut(&mut self, syntax_loc: Option<PatternLocation>) -> PResult<'a, PatKind> { let mut_span = self.prev_token.span; if self.eat_keyword(kw::Ref) { self.sess.emit_err(RefMutOrderIncorrect { span: mut_span.to(self.prev_token.span) }); - return self.parse_pat_ident(BindingAnnotation::REF_MUT); + return self.parse_pat_ident(BindingAnnotation::REF_MUT, syntax_loc); } self.recover_additional_muts(); @@ -584,7 +597,7 @@ impl<'a> Parser<'a> { } // Parse the pattern we hope to be an identifier. - let mut pat = self.parse_pat_no_top_alt(Some(Expected::Identifier))?; + let mut pat = self.parse_pat_no_top_alt(Some(Expected::Identifier), None)?; // If we don't have `mut $ident (@ pat)?`, error. if let PatKind::Ident(BindingAnnotation(ByRef::No, m @ Mutability::Not), ..) = &mut pat.kind @@ -810,10 +823,25 @@ impl<'a> Parser<'a> { /// Parses `ident` or `ident @ pat`. /// Used by the copy foo and ref foo patterns to give a good /// error message when parsing mistakes like `ref foo(a, b)`. - fn parse_pat_ident(&mut self, binding_annotation: BindingAnnotation) -> PResult<'a, PatKind> { + fn parse_pat_ident( + &mut self, + binding_annotation: BindingAnnotation, + syntax_loc: Option<PatternLocation>, + ) -> PResult<'a, PatKind> { let ident = self.parse_ident()?; + + if !matches!(syntax_loc, Some(PatternLocation::FunctionParameter)) + && self.check_noexpect(&token::Lt) + && self.look_ahead(1, |t| t.can_begin_type()) + { + return Err(self.sess.create_err(GenericArgsInPatRequireTurbofishSyntax { + span: self.token.span, + suggest_turbofish: self.token.span.shrink_to_lo(), + })); + } + let sub = if self.eat(&token::At) { - Some(self.parse_pat_no_top_alt(Some(Expected::BindingPattern))?) + Some(self.parse_pat_no_top_alt(Some(Expected::BindingPattern), None)?) } else { None }; @@ -902,14 +930,14 @@ impl<'a> Parser<'a> { // We cannot use `parse_pat_ident()` since it will complain `box` // is not an identifier. let sub = if self.eat(&token::At) { - Some(self.parse_pat_no_top_alt(Some(Expected::BindingPattern))?) + Some(self.parse_pat_no_top_alt(Some(Expected::BindingPattern), None)?) } else { None }; Ok(PatKind::Ident(BindingAnnotation::NONE, Ident::new(kw::Box, box_span), sub)) } else { - let pat = self.parse_pat_with_range_pat(false, None)?; + let pat = self.parse_pat_with_range_pat(false, None, None)?; self.sess.gated_spans.gate(sym::box_patterns, box_span.to(self.prev_token.span)); Ok(PatKind::Box(pat)) } diff --git a/compiler/rustc_parse/src/parser/stmt.rs b/compiler/rustc_parse/src/parser/stmt.rs index 1cdf2efa764..12c267351b9 100644 --- a/compiler/rustc_parse/src/parser/stmt.rs +++ b/compiler/rustc_parse/src/parser/stmt.rs @@ -193,10 +193,9 @@ impl<'a> Parser<'a> { /// At this point, the `!` token after the path has already been eaten. fn parse_stmt_mac(&mut self, lo: Span, attrs: AttrVec, path: ast::Path) -> PResult<'a, Stmt> { let args = self.parse_delim_args()?; - let delim = args.delim.to_token(); let hi = self.prev_token.span; - let style = match delim { + let style = match args.delim { Delimiter::Brace => MacStmtStyle::Braces, _ => MacStmtStyle::NoBraces, }; diff --git a/compiler/rustc_parse/src/validate_attr.rs b/compiler/rustc_parse/src/validate_attr.rs index 2011083019c..f7396598220 100644 --- a/compiler/rustc_parse/src/validate_attr.rs +++ b/compiler/rustc_parse/src/validate_attr.rs @@ -2,9 +2,10 @@ use crate::{errors, parse_in}; +use rustc_ast::token::Delimiter; use rustc_ast::tokenstream::DelimSpan; use rustc_ast::MetaItemKind; -use rustc_ast::{self as ast, AttrArgs, AttrArgsEq, Attribute, DelimArgs, MacDelimiter, MetaItem}; +use rustc_ast::{self as ast, AttrArgs, AttrArgsEq, Attribute, DelimArgs, MetaItem}; use rustc_ast_pretty::pprust; use rustc_errors::{Applicability, FatalError, PResult}; use rustc_feature::{AttributeTemplate, BuiltinAttribute, BUILTIN_ATTRIBUTE_MAP}; @@ -84,8 +85,8 @@ pub fn parse_meta<'a>(sess: &'a ParseSess, attr: &Attribute) -> PResult<'a, Meta }) } -pub fn check_meta_bad_delim(sess: &ParseSess, span: DelimSpan, delim: MacDelimiter) { - if let ast::MacDelimiter::Parenthesis = delim { +pub fn check_meta_bad_delim(sess: &ParseSess, span: DelimSpan, delim: Delimiter) { + if let Delimiter::Parenthesis = delim { return; } sess.emit_err(errors::MetaBadDelim { @@ -94,8 +95,8 @@ pub fn check_meta_bad_delim(sess: &ParseSess, span: DelimSpan, delim: MacDelimit }); } -pub fn check_cfg_attr_bad_delim(sess: &ParseSess, span: DelimSpan, delim: MacDelimiter) { - if let ast::MacDelimiter::Parenthesis = delim { +pub fn check_cfg_attr_bad_delim(sess: &ParseSess, span: DelimSpan, delim: Delimiter) { + if let Delimiter::Parenthesis = delim { return; } sess.emit_err(errors::CfgAttrBadDelim { |