diff options
| author | bors <bors@rust-lang.org> | 2024-08-13 15:01:50 +0000 |
|---|---|---|
| committer | bors <bors@rust-lang.org> | 2024-08-13 15:01:50 +0000 |
| commit | f96e296927cc0c6e9dd611edb943f6349001eca5 (patch) | |
| tree | 76c5e72c0570998ece04f11ac90e09b5d2613abc /compiler/rustc_parse/src | |
| parent | 00423bb1d85f3513e534abdb26f54a6966e88d47 (diff) | |
| parent | 28af7e09581c3b39cdbf2850df2f157690ab7e56 (diff) | |
| download | rust-f96e296927cc0c6e9dd611edb943f6349001eca5.tar.gz rust-f96e296927cc0c6e9dd611edb943f6349001eca5.zip | |
Auto merge of #17880 - lnicola:sync-from-rust, r=lnicola
minor: sync from downstream
Diffstat (limited to 'compiler/rustc_parse/src')
23 files changed, 562 insertions, 436 deletions
diff --git a/compiler/rustc_parse/src/errors.rs b/compiler/rustc_parse/src/errors.rs index 2e81d2a876b..0d4512be480 100644 --- a/compiler/rustc_parse/src/errors.rs +++ b/compiler/rustc_parse/src/errors.rs @@ -2,9 +2,10 @@ use std::borrow::Cow; use rustc_ast::token::Token; use rustc_ast::{Path, Visibility}; +use rustc_errors::codes::*; use rustc_errors::{ - codes::*, Applicability, Diag, DiagCtxtHandle, Diagnostic, EmissionGuarantee, Level, - SubdiagMessageOp, Subdiagnostic, + Applicability, Diag, DiagCtxtHandle, Diagnostic, EmissionGuarantee, Level, SubdiagMessageOp, + Subdiagnostic, }; use rustc_macros::{Diagnostic, Subdiagnostic}; use rustc_session::errors::ExprParenthesesNeeded; @@ -554,12 +555,7 @@ pub(crate) enum MissingInInForLoopSub { code = "in" )] InNotOf(#[primary_span] Span), - #[suggestion( - parse_add_in, - style = "verbose", - applicability = "maybe-incorrect", - code = " in " - )] + #[suggestion(parse_add_in, style = "verbose", applicability = "maybe-incorrect", code = " in ")] AddIn(#[primary_span] Span), } @@ -2729,15 +2725,24 @@ impl HelpUseLatestEdition { #[derive(Diagnostic)] #[diag(parse_box_syntax_removed)] -pub struct BoxSyntaxRemoved<'a> { +pub struct BoxSyntaxRemoved { #[primary_span] - #[suggestion( - code = "Box::new({code})", - applicability = "machine-applicable", - style = "verbose" - )] pub span: Span, - pub code: &'a str, + #[subdiagnostic] + pub sugg: AddBoxNew, +} + +#[derive(Subdiagnostic)] +#[multipart_suggestion( + parse_box_syntax_removed_suggestion, + applicability = "machine-applicable", + style = "verbose" +)] +pub struct AddBoxNew { + #[suggestion_part(code = "Box::new(")] + pub box_kw_and_lo: Span, + #[suggestion_part(code = ")")] + pub hi: Span, } #[derive(Diagnostic)] @@ -3187,6 +3192,7 @@ pub(crate) struct DotDotRangeAttribute { #[note] pub struct InvalidAttrUnsafe { #[primary_span] + #[label] pub span: Span, pub name: Path, } diff --git a/compiler/rustc_parse/src/lexer/diagnostics.rs b/compiler/rustc_parse/src/lexer/diagnostics.rs index 1247e2e44fb..4d5d1ce099e 100644 --- a/compiler/rustc_parse/src/lexer/diagnostics.rs +++ b/compiler/rustc_parse/src/lexer/diagnostics.rs @@ -1,9 +1,10 @@ -use super::UnmatchedDelim; use rustc_ast::token::Delimiter; use rustc_errors::Diag; use rustc_span::source_map::SourceMap; use rustc_span::Span; +use super::UnmatchedDelim; + #[derive(Default)] pub(super) struct TokenTreeDiagInfo { /// Stack of open delimiters and their spans. Used for error message. diff --git a/compiler/rustc_parse/src/lexer/mod.rs b/compiler/rustc_parse/src/lexer/mod.rs index 511805cf8d6..f30939093c2 100644 --- a/compiler/rustc_parse/src/lexer/mod.rs +++ b/compiler/rustc_parse/src/lexer/mod.rs @@ -1,25 +1,26 @@ use std::ops::Range; -use crate::errors; -use crate::lexer::unicode_chars::UNICODE_ARRAY; -use crate::make_unclosed_delims_error; use rustc_ast::ast::{self, AttrStyle}; use rustc_ast::token::{self, CommentKind, Delimiter, IdentIsRaw, Token, TokenKind}; use rustc_ast::tokenstream::TokenStream; use rustc_ast::util::unicode::contains_text_flow_control_chars; -use rustc_errors::{codes::*, Applicability, Diag, DiagCtxtHandle, StashKey}; +use rustc_errors::codes::*; +use rustc_errors::{Applicability, Diag, DiagCtxtHandle, StashKey}; use rustc_lexer::unescape::{self, EscapeError, Mode}; -use rustc_lexer::{Base, DocStyle, RawStrError}; -use rustc_lexer::{Cursor, LiteralKind}; +use rustc_lexer::{Base, Cursor, DocStyle, LiteralKind, RawStrError}; use rustc_session::lint::builtin::{ RUST_2021_PREFIXES_INCOMPATIBLE_SYNTAX, TEXT_DIRECTION_CODEPOINT_IN_COMMENT, }; use rustc_session::lint::BuiltinLintDiag; use rustc_session::parse::ParseSess; +use rustc_span::edition::Edition; use rustc_span::symbol::Symbol; -use rustc_span::{edition::Edition, BytePos, Pos, Span}; +use rustc_span::{BytePos, Pos, Span}; use tracing::debug; +use crate::lexer::unicode_chars::UNICODE_ARRAY; +use crate::{errors, make_unclosed_delims_error}; + mod diagnostics; mod tokentrees; mod unescape_error_reporting; diff --git a/compiler/rustc_parse/src/lexer/tokentrees.rs b/compiler/rustc_parse/src/lexer/tokentrees.rs index 8e543454691..fb4ed5b93b2 100644 --- a/compiler/rustc_parse/src/lexer/tokentrees.rs +++ b/compiler/rustc_parse/src/lexer/tokentrees.rs @@ -1,14 +1,15 @@ -use super::diagnostics::report_suspicious_mismatch_block; -use super::diagnostics::same_indentation_level; -use super::diagnostics::TokenTreeDiagInfo; -use super::{StringReader, UnmatchedDelim}; -use crate::Parser; use rustc_ast::token::{self, Delimiter, Token}; use rustc_ast::tokenstream::{DelimSpacing, DelimSpan, Spacing, TokenStream, TokenTree}; use rustc_ast_pretty::pprust::token_to_string; use rustc_errors::{Applicability, PErr}; use rustc_span::symbol::kw; +use super::diagnostics::{ + report_suspicious_mismatch_block, same_indentation_level, TokenTreeDiagInfo, +}; +use super::{StringReader, UnmatchedDelim}; +use crate::Parser; + pub(super) struct TokenTreesReader<'psess, 'src> { string_reader: StringReader<'psess, 'src>, /// The "next" token, which has been obtained from the `StringReader` but @@ -72,16 +73,31 @@ impl<'psess, 'src> TokenTreesReader<'psess, 'src> { fn eof_err(&mut self) -> PErr<'psess> { let msg = "this file contains an unclosed delimiter"; let mut err = self.string_reader.dcx().struct_span_err(self.token.span, msg); - for &(_, sp) in &self.diag_info.open_braces { - err.span_label(sp, "unclosed delimiter"); + + let unclosed_delimiter_show_limit = 5; + let len = usize::min(unclosed_delimiter_show_limit, self.diag_info.open_braces.len()); + for &(_, span) in &self.diag_info.open_braces[..len] { + err.span_label(span, "unclosed delimiter"); self.diag_info.unmatched_delims.push(UnmatchedDelim { found_delim: None, found_span: self.token.span, - unclosed_span: Some(sp), + unclosed_span: Some(span), candidate_span: None, }); } + if let Some((_, span)) = self.diag_info.open_braces.get(unclosed_delimiter_show_limit) + && self.diag_info.open_braces.len() >= unclosed_delimiter_show_limit + 2 + { + err.span_label( + *span, + format!( + "another {} unclosed delimiters begin from here", + self.diag_info.open_braces.len() - unclosed_delimiter_show_limit + ), + ); + } + if let Some((delim, _)) = self.diag_info.open_braces.last() { report_suspicious_mismatch_block( &mut err, diff --git a/compiler/rustc_parse/src/lexer/unescape_error_reporting.rs b/compiler/rustc_parse/src/lexer/unescape_error_reporting.rs index b7a790fcf83..efa53f0962b 100644 --- a/compiler/rustc_parse/src/lexer/unescape_error_reporting.rs +++ b/compiler/rustc_parse/src/lexer/unescape_error_reporting.rs @@ -40,7 +40,8 @@ pub(crate) fn emit_unescape_error( dcx.emit_err(UnescapeError::InvalidUnicodeEscape { span: err_span, surrogate: false }) } EscapeError::MoreThanOneChar => { - use unicode_normalization::{char::is_combining_mark, UnicodeNormalization}; + use unicode_normalization::char::is_combining_mark; + use unicode_normalization::UnicodeNormalization; let mut sugg = None; let mut note = None; diff --git a/compiler/rustc_parse/src/lexer/unicode_chars.rs b/compiler/rustc_parse/src/lexer/unicode_chars.rs index 0a82ede3b75..d78b3664b1e 100644 --- a/compiler/rustc_parse/src/lexer/unicode_chars.rs +++ b/compiler/rustc_parse/src/lexer/unicode_chars.rs @@ -1,12 +1,12 @@ //! Characters and their corresponding confusables were collected from //! <https://www.unicode.org/Public/security/10.0.0/confusables.txt> +use rustc_span::symbol::kw; +use rustc_span::{BytePos, Pos, Span}; + use super::StringReader; -use crate::{ - errors::TokenSubstitution, - token::{self, Delimiter}, -}; -use rustc_span::{symbol::kw, BytePos, Pos, Span}; +use crate::errors::TokenSubstitution; +use crate::token::{self, Delimiter}; #[rustfmt::skip] // for line breaks pub(super) const UNICODE_ARRAY: &[(char, &str, &str)] = &[ diff --git a/compiler/rustc_parse/src/lib.rs b/compiler/rustc_parse/src/lib.rs index 4454747ea02..37079271493 100644 --- a/compiler/rustc_parse/src/lib.rs +++ b/compiler/rustc_parse/src/lib.rs @@ -5,6 +5,7 @@ #![allow(rustc::diagnostic_outside_of_impl)] #![allow(rustc::untranslatable_diagnostic)] #![feature(array_windows)] +#![feature(assert_matches)] #![feature(box_patterns)] #![feature(debug_closure_helpers)] #![feature(if_let_guard)] @@ -12,18 +13,17 @@ #![feature(let_chains)] // tidy-alphabetical-end +use std::path::Path; + use rustc_ast as ast; -use rustc_ast::token; use rustc_ast::tokenstream::TokenStream; -use rustc_ast::{AttrItem, Attribute, MetaItem}; +use rustc_ast::{token, AttrItem, Attribute, MetaItem}; use rustc_ast_pretty::pprust; use rustc_data_structures::sync::Lrc; use rustc_errors::{Diag, FatalError, PResult}; use rustc_session::parse::ParseSess; use rustc_span::{FileName, SourceFile, Span}; -use std::path::Path; - pub const MACRO_ARGUMENTS: Option<&str> = Some("macro arguments"); #[macro_use] diff --git a/compiler/rustc_parse/src/parser/attr.rs b/compiler/rustc_parse/src/parser/attr.rs index 535b53a836e..8fdfbcee385 100644 --- a/compiler/rustc_parse/src/parser/attr.rs +++ b/compiler/rustc_parse/src/parser/attr.rs @@ -1,16 +1,16 @@ -use crate::errors; -use crate::fluent_generated as fluent; -use crate::maybe_whole; - -use super::{AttrWrapper, Capturing, FnParseMode, ForceCollect, Parser, PathStyle}; use rustc_ast as ast; use rustc_ast::attr; use rustc_ast::token::{self, Delimiter}; -use rustc_errors::{codes::*, Diag, PResult}; -use rustc_span::{sym, symbol::kw, BytePos, Span}; +use rustc_errors::codes::*; +use rustc_errors::{Diag, PResult}; +use rustc_span::symbol::kw; +use rustc_span::{sym, BytePos, Span}; use thin_vec::ThinVec; use tracing::debug; +use super::{AttrWrapper, Capturing, FnParseMode, ForceCollect, Parser, ParserRange, PathStyle}; +use crate::{errors, fluent_generated as fluent, maybe_whole}; + // Public for rustfmt usage #[derive(Debug)] pub enum InnerAttrPolicy { @@ -31,6 +31,12 @@ enum OuterAttributeType { Attribute, } +#[derive(Clone, Copy, PartialEq, Eq)] +pub enum AllowLeadingUnsafe { + Yes, + No, +} + impl<'a> Parser<'a> { /// Parses attributes that appear before an item. pub(super) fn parse_outer_attributes(&mut self) -> PResult<'a, AttrWrapper> { @@ -307,8 +313,8 @@ impl<'a> Parser<'a> { // inner attribute, for possible later processing in a `LazyAttrTokenStream`. if let Capturing::Yes = self.capture_state.capturing { let end_pos = self.num_bump_calls; - let range = start_pos..end_pos; - self.capture_state.inner_attr_ranges.insert(attr.id, range); + let parser_range = ParserRange(start_pos..end_pos); + self.capture_state.inner_attr_parser_ranges.insert(attr.id, parser_range); } attrs.push(attr); } else { @@ -332,7 +338,7 @@ impl<'a> Parser<'a> { /// Parses `cfg_attr(pred, attr_item_list)` where `attr_item_list` is comma-delimited. pub fn parse_cfg_attr(&mut self) -> PResult<'a, (ast::MetaItem, Vec<(ast::AttrItem, Span)>)> { - let cfg_predicate = self.parse_meta_item()?; + let cfg_predicate = self.parse_meta_item(AllowLeadingUnsafe::No)?; self.expect(&token::Comma)?; // Presumably, the majority of the time there will only be one attr. @@ -368,7 +374,10 @@ impl<'a> Parser<'a> { /// MetaItem = SimplePath ( '=' UNSUFFIXED_LIT | '(' MetaSeq? ')' )? ; /// MetaSeq = MetaItemInner (',' MetaItemInner)* ','? ; /// ``` - pub fn parse_meta_item(&mut self) -> PResult<'a, ast::MetaItem> { + pub fn parse_meta_item( + &mut self, + unsafe_allowed: AllowLeadingUnsafe, + ) -> PResult<'a, ast::MetaItem> { // We can't use `maybe_whole` here because it would bump in the `None` // case, which we don't want. if let token::Interpolated(nt) = &self.token.kind @@ -384,7 +393,11 @@ impl<'a> Parser<'a> { } let lo = self.token.span; - let is_unsafe = self.eat_keyword(kw::Unsafe); + let is_unsafe = if unsafe_allowed == AllowLeadingUnsafe::Yes { + self.eat_keyword(kw::Unsafe) + } else { + false + }; let unsafety = if is_unsafe { let unsafe_span = self.prev_token.span; self.psess.gated_spans.gate(sym::unsafe_attributes, unsafe_span); @@ -427,7 +440,7 @@ impl<'a> Parser<'a> { Err(err) => err.cancel(), // we provide a better error below } - match self.parse_meta_item() { + match self.parse_meta_item(AllowLeadingUnsafe::No) { Ok(mi) => return Ok(ast::NestedMetaItem::MetaItem(mi)), Err(err) => err.cancel(), // we provide a better error below } diff --git a/compiler/rustc_parse/src/parser/attr_wrapper.rs b/compiler/rustc_parse/src/parser/attr_wrapper.rs index 5dc49ea51d1..abf61036c2d 100644 --- a/compiler/rustc_parse/src/parser/attr_wrapper.rs +++ b/compiler/rustc_parse/src/parser/attr_wrapper.rs @@ -1,14 +1,19 @@ -use super::{Capturing, FlatToken, ForceCollect, Parser, ReplaceRange, TokenCursor}; +use std::{iter, mem}; + use rustc_ast::token::{Delimiter, Token, TokenKind}; -use rustc_ast::tokenstream::{AttrTokenStream, AttrTokenTree, AttrsTarget, DelimSpacing}; -use rustc_ast::tokenstream::{DelimSpan, LazyAttrTokenStream, Spacing, ToAttrTokenStream}; -use rustc_ast::{self as ast}; -use rustc_ast::{AttrVec, Attribute, HasAttrs, HasTokens}; +use rustc_ast::tokenstream::{ + AttrTokenStream, AttrTokenTree, AttrsTarget, DelimSpacing, DelimSpan, LazyAttrTokenStream, + Spacing, ToAttrTokenStream, +}; +use rustc_ast::{self as ast, AttrVec, Attribute, HasAttrs, HasTokens}; use rustc_errors::PResult; use rustc_session::parse::ParseSess; use rustc_span::{sym, Span, DUMMY_SP}; -use std::{iter, mem}; +use super::{ + Capturing, FlatToken, ForceCollect, NodeRange, NodeReplacement, Parser, ParserRange, + TokenCursor, +}; /// A wrapper type to ensure that the parser handles outer attributes correctly. /// When we parse outer attributes, we need to ensure that we capture tokens @@ -26,8 +31,8 @@ use std::{iter, mem}; #[derive(Debug, Clone)] pub struct AttrWrapper { attrs: AttrVec, - // The start of the outer attributes in the token cursor. - // This allows us to create a `ReplaceRange` for the entire attribute + // The start of the outer attributes in the parser's token stream. + // This lets us create a `NodeReplacement` for the entire attribute // target, including outer attributes. start_pos: u32, } @@ -51,10 +56,9 @@ impl AttrWrapper { /// Prepend `self.attrs` to `attrs`. // FIXME: require passing an NT to prevent misuse of this method - pub(crate) fn prepend_to_nt_inner(self, attrs: &mut AttrVec) { - let mut self_attrs = self.attrs; - mem::swap(attrs, &mut self_attrs); - attrs.extend(self_attrs); + pub(crate) fn prepend_to_nt_inner(mut self, attrs: &mut AttrVec) { + mem::swap(attrs, &mut self.attrs); + attrs.extend(self.attrs); } pub fn is_empty(&self) -> bool { @@ -87,7 +91,7 @@ struct LazyAttrTokenStreamImpl { cursor_snapshot: TokenCursor, num_calls: u32, break_last_token: bool, - replace_ranges: Box<[ReplaceRange]>, + node_replacements: Box<[NodeReplacement]>, } impl ToAttrTokenStream for LazyAttrTokenStreamImpl { @@ -102,21 +106,24 @@ impl ToAttrTokenStream for LazyAttrTokenStreamImpl { .chain(iter::repeat_with(|| FlatToken::Token(cursor_snapshot.next()))) .take(self.num_calls as usize); - if self.replace_ranges.is_empty() { + if self.node_replacements.is_empty() { make_attr_token_stream(tokens, self.break_last_token) } else { let mut tokens: Vec<_> = tokens.collect(); - let mut replace_ranges = self.replace_ranges.to_vec(); - replace_ranges.sort_by_key(|(range, _)| range.start); + let mut node_replacements = self.node_replacements.to_vec(); + node_replacements.sort_by_key(|(range, _)| range.0.start); #[cfg(debug_assertions)] - for [(range, tokens), (next_range, next_tokens)] in replace_ranges.array_windows() { + for [(node_range, tokens), (next_node_range, next_tokens)] in + node_replacements.array_windows() + { assert!( - range.end <= next_range.start || range.end >= next_range.end, - "Replace ranges should either be disjoint or nested: ({:?}, {:?}) ({:?}, {:?})", - range, + node_range.0.end <= next_node_range.0.start + || node_range.0.end >= next_node_range.0.end, + "Node ranges should be disjoint or nested: ({:?}, {:?}) ({:?}, {:?})", + node_range, tokens, - next_range, + next_node_range, next_tokens, ); } @@ -134,20 +141,23 @@ impl ToAttrTokenStream for LazyAttrTokenStreamImpl { // start position, we ensure that any (outer) replace range which // encloses another (inner) replace range will fully overwrite the // inner range's replacement. - for (range, target) in replace_ranges.into_iter().rev() { - assert!(!range.is_empty(), "Cannot replace an empty range: {range:?}"); + for (node_range, target) in node_replacements.into_iter().rev() { + assert!( + !node_range.0.is_empty(), + "Cannot replace an empty node range: {:?}", + node_range.0 + ); // Replace the tokens in range with zero or one `FlatToken::AttrsTarget`s, plus // enough `FlatToken::Empty`s to fill up the rest of the range. This keeps the // total length of `tokens` constant throughout the replacement process, allowing - // us to use all of the `ReplaceRanges` entries without adjusting indices. + // us to do all replacements without adjusting indices. let target_len = target.is_some() as usize; tokens.splice( - (range.start as usize)..(range.end as usize), - target - .into_iter() - .map(|target| FlatToken::AttrsTarget(target)) - .chain(iter::repeat(FlatToken::Empty).take(range.len() - target_len)), + (node_range.0.start as usize)..(node_range.0.end as usize), + target.into_iter().map(|target| FlatToken::AttrsTarget(target)).chain( + iter::repeat(FlatToken::Empty).take(node_range.0.len() - target_len), + ), ); } make_attr_token_stream(tokens.into_iter(), self.break_last_token) @@ -214,7 +224,7 @@ impl<'a> Parser<'a> { let cursor_snapshot = self.token_cursor.clone(); let start_pos = self.num_bump_calls; let has_outer_attrs = !attrs.attrs.is_empty(); - let replace_ranges_start = self.capture_state.replace_ranges.len(); + let parser_replacements_start = self.capture_state.parser_replacements.len(); // We set and restore `Capturing::Yes` on either side of the call to // `f`, so we can distinguish the outermost call to @@ -269,7 +279,7 @@ impl<'a> Parser<'a> { return Ok(ret); } - let replace_ranges_end = self.capture_state.replace_ranges.len(); + let parser_replacements_end = self.capture_state.parser_replacements.len(); assert!( !(self.break_last_token && capture_trailing), @@ -286,15 +296,16 @@ impl<'a> Parser<'a> { let num_calls = end_pos - start_pos; - // Take the captured ranges for any inner attributes that we parsed in - // `Parser::parse_inner_attributes`, and pair them in a `ReplaceRange` - // with `None`, which means the relevant tokens will be removed. (More - // details below.) - let mut inner_attr_replace_ranges = Vec::new(); + // Take the captured `ParserRange`s for any inner attributes that we parsed in + // `Parser::parse_inner_attributes`, and pair them in a `ParserReplacement` with `None`, + // which means the relevant tokens will be removed. (More details below.) + let mut inner_attr_parser_replacements = Vec::new(); for attr in ret.attrs() { if attr.style == ast::AttrStyle::Inner { - if let Some(attr_range) = self.capture_state.inner_attr_ranges.remove(&attr.id) { - inner_attr_replace_ranges.push((attr_range, None)); + if let Some(inner_attr_parser_range) = + self.capture_state.inner_attr_parser_ranges.remove(&attr.id) + { + inner_attr_parser_replacements.push((inner_attr_parser_range, None)); } else { self.dcx().span_delayed_bug(attr.span, "Missing token range for attribute"); } @@ -303,37 +314,41 @@ impl<'a> Parser<'a> { // This is hot enough for `deep-vector` that checking the conditions for an empty iterator // is measurably faster than actually executing the iterator. - let replace_ranges: Box<[ReplaceRange]> = - if replace_ranges_start == replace_ranges_end && inner_attr_replace_ranges.is_empty() { - Box::new([]) - } else { - // Grab any replace ranges that occur *inside* the current AST node. We will - // perform the actual replacement only when we convert the `LazyAttrTokenStream` to - // an `AttrTokenStream`. - self.capture_state.replace_ranges[replace_ranges_start..replace_ranges_end] - .iter() - .cloned() - .chain(inner_attr_replace_ranges.iter().cloned()) - .map(|(range, data)| ((range.start - start_pos)..(range.end - start_pos), data)) - .collect() - }; + let node_replacements: Box<[_]> = if parser_replacements_start == parser_replacements_end + && inner_attr_parser_replacements.is_empty() + { + Box::new([]) + } else { + // Grab any replace ranges that occur *inside* the current AST node. Convert them + // from `ParserRange` form to `NodeRange` form. We will perform the actual + // replacement only when we convert the `LazyAttrTokenStream` to an + // `AttrTokenStream`. + self.capture_state.parser_replacements + [parser_replacements_start..parser_replacements_end] + .iter() + .cloned() + .chain(inner_attr_parser_replacements.iter().cloned()) + .map(|(parser_range, data)| (NodeRange::new(parser_range, start_pos), data)) + .collect() + }; // What is the status here when parsing the example code at the top of this method? // // When parsing `g`: // - `start_pos..end_pos` is `12..33` (`fn g { ... }`, excluding the outer attr). - // - `inner_attr_replace_ranges` has one entry (`5..15`, when counting from `fn`), to + // - `inner_attr_parser_replacements` has one entry (`ParserRange(17..27)`), to // delete the inner attr's tokens. - // - This entry is put into the lazy tokens for `g`, i.e. deleting the inner attr from - // those tokens (if they get evaluated). + // - This entry is converted to `NodeRange(5..15)` (relative to the `fn`) and put into + // the lazy tokens for `g`, i.e. deleting the inner attr from those tokens (if they get + // evaluated). // - Those lazy tokens are also put into an `AttrsTarget` that is appended to `self`'s // replace ranges at the bottom of this function, for processing when parsing `m`. - // - `replace_ranges_start..replace_ranges_end` is empty. + // - `parser_replacements_start..parser_replacements_end` is empty. // // When parsing `m`: // - `start_pos..end_pos` is `0..34` (`mod m`, excluding the `#[cfg_eval]` attribute). - // - `inner_attr_replace_ranges` is empty. - // - `replace_range_start..replace_ranges_end` has one entry. + // - `inner_attr_parser_replacements` is empty. + // - `parser_replacements_start..parser_replacements_end` has one entry. // - One `AttrsTarget` (added below when parsing `g`) to replace all of `g` (`3..33`, // including its outer attribute), with: // - `attrs`: includes the outer and the inner attr. @@ -344,7 +359,7 @@ impl<'a> Parser<'a> { num_calls, cursor_snapshot, break_last_token: self.break_last_token, - replace_ranges, + node_replacements, }); // If we support tokens and don't already have them, store the newly captured tokens. @@ -365,7 +380,7 @@ impl<'a> Parser<'a> { // What is the status here when parsing the example code at the top of this method? // // When parsing `g`, we add one entry: - // - The `start_pos..end_pos` (`3..33`) entry has a new `AttrsTarget` with: + // - The pushed entry (`ParserRange(3..33)`) has a new `AttrsTarget` with: // - `attrs`: includes the outer and the inner attr. // - `tokens`: lazy tokens for `g` (with its inner attr deleted). // @@ -376,12 +391,14 @@ impl<'a> Parser<'a> { // cfg-expand this AST node. let start_pos = if has_outer_attrs { attrs.start_pos } else { start_pos }; let target = AttrsTarget { attrs: ret.attrs().iter().cloned().collect(), tokens }; - self.capture_state.replace_ranges.push((start_pos..end_pos, Some(target))); + self.capture_state + .parser_replacements + .push((ParserRange(start_pos..end_pos), Some(target))); } else if matches!(self.capture_state.capturing, Capturing::No) { // Only clear the ranges once we've finished capturing entirely, i.e. we've finished // the outermost call to this method. - self.capture_state.replace_ranges.clear(); - self.capture_state.inner_attr_ranges.clear(); + self.capture_state.parser_replacements.clear(); + self.capture_state.inner_attr_parser_ranges.clear(); } Ok(ret) } @@ -469,8 +486,9 @@ fn needs_tokens(attrs: &[ast::Attribute]) -> bool { // Some types are used a lot. Make sure they don't unintentionally get bigger. #[cfg(target_pointer_width = "64")] mod size_asserts { - use super::*; use rustc_data_structures::static_assert_size; + + use super::*; // tidy-alphabetical-start static_assert_size!(AttrWrapper, 16); static_assert_size!(LazyAttrTokenStreamImpl, 96); diff --git a/compiler/rustc_parse/src/parser/diagnostics.rs b/compiler/rustc_parse/src/parser/diagnostics.rs index 1a0d9aa6378..47ca85ba060 100644 --- a/compiler/rustc_parse/src/parser/diagnostics.rs +++ b/compiler/rustc_parse/src/parser/diagnostics.rs @@ -1,25 +1,6 @@ -use super::pat::Expected; -use super::{ - BlockMode, CommaRecoveryMode, Parser, PathStyle, Restrictions, SemiColonMode, SeqSep, TokenType, -}; -use crate::errors::{ - AddParen, AmbiguousPlus, AsyncMoveBlockIn2015, AttributeOnParamType, AwaitSuggestion, - BadQPathStage2, BadTypePlus, BadTypePlusSub, ColonAsSemi, ComparisonOperatorsCannotBeChained, - ComparisonOperatorsCannotBeChainedSugg, ConstGenericWithoutBraces, - ConstGenericWithoutBracesSugg, DocCommentDoesNotDocumentAnything, DocCommentOnParamType, - DoubleColonInBound, ExpectedIdentifier, ExpectedSemi, ExpectedSemiSugg, - GenericParamsWithoutAngleBrackets, GenericParamsWithoutAngleBracketsSugg, - HelpIdentifierStartsWithNumber, HelpUseLatestEdition, InInTypo, IncorrectAwait, - IncorrectSemicolon, IncorrectUseOfAwait, PatternMethodParamWithoutBody, QuestionMarkInType, - QuestionMarkInTypeSugg, SelfParamNotFirst, StructLiteralBodyWithoutPath, - StructLiteralBodyWithoutPathSugg, StructLiteralNeedingParens, StructLiteralNeedingParensSugg, - SuggAddMissingLetStmt, SuggEscapeIdentifier, SuggRemoveComma, TernaryOperator, - UnexpectedConstInGenericParam, UnexpectedConstParamDeclaration, - UnexpectedConstParamDeclarationSugg, UnmatchedAngleBrackets, UseEqInstead, WrapType, -}; -use crate::fluent_generated as fluent; -use crate::parser; -use crate::parser::attr::InnerAttrPolicy; +use std::mem::take; +use std::ops::{Deref, DerefMut}; + use ast::token::IdentIsRaw; use rustc_ast as ast; use rustc_ast::ptr::P; @@ -41,11 +22,31 @@ use rustc_session::errors::ExprParenthesesNeeded; use rustc_span::source_map::Spanned; use rustc_span::symbol::{kw, sym, Ident}; use rustc_span::{BytePos, Span, SpanSnippetError, Symbol, DUMMY_SP}; -use std::mem::take; -use std::ops::{Deref, DerefMut}; use thin_vec::{thin_vec, ThinVec}; use tracing::{debug, trace}; +use super::pat::Expected; +use super::{ + BlockMode, CommaRecoveryMode, Parser, PathStyle, Restrictions, SemiColonMode, SeqSep, TokenType, +}; +use crate::errors::{ + AddParen, AmbiguousPlus, AsyncMoveBlockIn2015, AttributeOnParamType, AwaitSuggestion, + BadQPathStage2, BadTypePlus, BadTypePlusSub, ColonAsSemi, ComparisonOperatorsCannotBeChained, + ComparisonOperatorsCannotBeChainedSugg, ConstGenericWithoutBraces, + ConstGenericWithoutBracesSugg, DocCommentDoesNotDocumentAnything, DocCommentOnParamType, + DoubleColonInBound, ExpectedIdentifier, ExpectedSemi, ExpectedSemiSugg, + GenericParamsWithoutAngleBrackets, GenericParamsWithoutAngleBracketsSugg, + HelpIdentifierStartsWithNumber, HelpUseLatestEdition, InInTypo, IncorrectAwait, + IncorrectSemicolon, IncorrectUseOfAwait, PatternMethodParamWithoutBody, QuestionMarkInType, + QuestionMarkInTypeSugg, SelfParamNotFirst, StructLiteralBodyWithoutPath, + StructLiteralBodyWithoutPathSugg, StructLiteralNeedingParens, StructLiteralNeedingParensSugg, + SuggAddMissingLetStmt, SuggEscapeIdentifier, SuggRemoveComma, TernaryOperator, + UnexpectedConstInGenericParam, UnexpectedConstParamDeclaration, + UnexpectedConstParamDeclarationSugg, UnmatchedAngleBrackets, UseEqInstead, WrapType, +}; +use crate::parser::attr::InnerAttrPolicy; +use crate::{fluent_generated as fluent, parser}; + /// Creates a placeholder argument. pub(super) fn dummy_arg(ident: Ident, guar: ErrorGuaranteed) -> Param { let pat = P(Pat { diff --git a/compiler/rustc_parse/src/parser/expr.rs b/compiler/rustc_parse/src/parser/expr.rs index 389a6d11e19..cf5d65708ab 100644 --- a/compiler/rustc_parse/src/parser/expr.rs +++ b/compiler/rustc_parse/src/parser/expr.rs @@ -1,30 +1,22 @@ // ignore-tidy-filelength -use super::diagnostics::SnapshotParser; -use super::pat::{CommaRecoveryMode, Expected, RecoverColon, RecoverComma}; -use super::ty::{AllowPlus, RecoverQPath, RecoverReturnSign}; -use super::{ - AttrWrapper, BlockMode, ClosureSpans, ForceCollect, Parser, PathStyle, Restrictions, - SemiColonMode, SeqSep, TokenType, Trailing, -}; +use core::mem; +use core::ops::ControlFlow; -use crate::errors; -use crate::maybe_recover_from_interpolated_ty_qpath; use ast::mut_visit::{self, MutVisitor}; use ast::token::IdentIsRaw; use ast::{CoroutineKind, ForLoopKind, GenBlockKind, MatchKind, Pat, Path, PathSegment, Recovered}; -use core::mem; -use core::ops::ControlFlow; use rustc_ast::ptr::P; use rustc_ast::token::{self, Delimiter, Token, TokenKind}; use rustc_ast::util::case::Case; use rustc_ast::util::classify; use rustc_ast::util::parser::{prec_let_scrutinee_needs_par, AssocOp, Fixity}; use rustc_ast::visit::{walk_expr, Visitor}; -use rustc_ast::{self as ast, AttrStyle, AttrVec, CaptureBy, ExprField, UnOp, DUMMY_NODE_ID}; -use rustc_ast::{AnonConst, BinOp, BinOpKind, FnDecl, FnRetTy, MacCall, Param, Ty, TyKind}; -use rustc_ast::{Arm, BlockCheckMode, Expr, ExprKind, Label, Movability, RangeLimits}; -use rustc_ast::{ClosureBinder, MetaItemLit, StmtKind}; +use rustc_ast::{ + self as ast, AnonConst, Arm, AttrStyle, AttrVec, BinOp, BinOpKind, BlockCheckMode, CaptureBy, + ClosureBinder, Expr, ExprField, ExprKind, FnDecl, FnRetTy, Label, MacCall, MetaItemLit, + Movability, Param, RangeLimits, StmtKind, Ty, TyKind, UnOp, DUMMY_NODE_ID, +}; use rustc_ast_pretty::pprust; use rustc_data_structures::stack::ensure_sufficient_stack; use rustc_errors::{Applicability, Diag, PResult, StashKey, Subdiagnostic}; @@ -39,13 +31,14 @@ use rustc_span::{BytePos, ErrorGuaranteed, Pos, Span}; use thin_vec::{thin_vec, ThinVec}; use tracing::instrument; -#[derive(Debug)] -pub(super) enum LhsExpr { - // Already parsed just the outer attributes. - Unparsed { attrs: AttrWrapper }, - // Already parsed the expression. - Parsed { expr: P<Expr>, starts_statement: bool }, -} +use super::diagnostics::SnapshotParser; +use super::pat::{CommaRecoveryMode, Expected, RecoverColon, RecoverComma}; +use super::ty::{AllowPlus, RecoverQPath, RecoverReturnSign}; +use super::{ + AttrWrapper, BlockMode, ClosureSpans, ForceCollect, Parser, PathStyle, Restrictions, + SemiColonMode, SeqSep, TokenType, Trailing, +}; +use crate::{errors, maybe_recover_from_interpolated_ty_qpath}; #[derive(Debug)] enum DestructuredFloat { @@ -112,30 +105,31 @@ impl<'a> Parser<'a> { r: Restrictions, attrs: AttrWrapper, ) -> PResult<'a, P<Expr>> { - self.with_res(r, |this| this.parse_expr_assoc_with(0, LhsExpr::Unparsed { attrs })) + self.with_res(r, |this| this.parse_expr_assoc_with(0, attrs)) } /// Parses an associative expression with operators of at least `min_prec` precedence. pub(super) fn parse_expr_assoc_with( &mut self, min_prec: usize, - lhs: LhsExpr, + attrs: AttrWrapper, ) -> PResult<'a, P<Expr>> { - let mut starts_stmt = false; - let mut lhs = match lhs { - LhsExpr::Parsed { expr, starts_statement } => { - starts_stmt = starts_statement; - expr - } - LhsExpr::Unparsed { attrs } => { - if self.token.is_range_separator() { - return self.parse_expr_prefix_range(attrs); - } else { - self.parse_expr_prefix(attrs)? - } - } + let lhs = if self.token.is_range_separator() { + return self.parse_expr_prefix_range(attrs); + } else { + self.parse_expr_prefix(attrs)? }; + self.parse_expr_assoc_rest_with(min_prec, false, lhs) + } + /// Parses the rest of an associative expression (i.e. the part after the lhs) with operators + /// of at least `min_prec` precedence. + pub(super) fn parse_expr_assoc_rest_with( + &mut self, + min_prec: usize, + starts_stmt: bool, + mut lhs: P<Expr>, + ) -> PResult<'a, P<Expr>> { if !self.should_continue_as_assoc_expr(&lhs) { return Ok(lhs); } @@ -271,7 +265,7 @@ impl<'a> Parser<'a> { }; let rhs = self.with_res(restrictions - Restrictions::STMT_EXPR, |this| { let attrs = this.parse_outer_attributes()?; - this.parse_expr_assoc_with(prec + prec_adjustment, LhsExpr::Unparsed { attrs }) + this.parse_expr_assoc_with(prec + prec_adjustment, attrs) })?; let span = self.mk_expr_sp(&lhs, lhs_span, rhs.span); @@ -446,7 +440,7 @@ impl<'a> Parser<'a> { let maybe_lt = self.token.clone(); let attrs = self.parse_outer_attributes()?; Some( - self.parse_expr_assoc_with(prec + 1, LhsExpr::Unparsed { attrs }) + self.parse_expr_assoc_with(prec + 1, attrs) .map_err(|err| self.maybe_err_dotdotlt_syntax(maybe_lt, err))?, ) } else { @@ -503,12 +497,9 @@ impl<'a> Parser<'a> { let (span, opt_end) = if this.is_at_start_of_range_notation_rhs() { // RHS must be parsed with more associativity than the dots. let attrs = this.parse_outer_attributes()?; - this.parse_expr_assoc_with( - op.unwrap().precedence() + 1, - LhsExpr::Unparsed { attrs }, - ) - .map(|x| (lo.to(x.span), Some(x))) - .map_err(|err| this.maybe_err_dotdotlt_syntax(maybe_lt, err))? + this.parse_expr_assoc_with(op.unwrap().precedence() + 1, attrs) + .map(|x| (lo.to(x.span), Some(x))) + .map_err(|err| this.maybe_err_dotdotlt_syntax(maybe_lt, err))? } else { (lo, None) }; @@ -617,10 +608,12 @@ impl<'a> Parser<'a> { /// Parse `box expr` - this syntax has been removed, but we still parse this /// for now to provide a more useful error fn parse_expr_box(&mut self, box_kw: Span) -> PResult<'a, (Span, ExprKind)> { - let (span, _) = self.parse_expr_prefix_common(box_kw)?; - let inner_span = span.with_lo(box_kw.hi()); - let code = self.psess.source_map().span_to_snippet(inner_span).unwrap(); - let guar = self.dcx().emit_err(errors::BoxSyntaxRemoved { span: span, code: code.trim() }); + let (span, expr) = self.parse_expr_prefix_common(box_kw)?; + // Make a multipart suggestion instead of `span_to_snippet` in case source isn't available + let box_kw_and_lo = box_kw.until(self.interpolated_or_expr_span(&expr)); + let hi = span.shrink_to_hi(); + let sugg = errors::AddBoxNew { box_kw_and_lo, hi }; + let guar = self.dcx().emit_err(errors::BoxSyntaxRemoved { span, sugg }); Ok((span, ExprKind::Err(guar))) } @@ -701,12 +694,12 @@ impl<'a> Parser<'a> { // `foo: ` ExprKind::Path(None, ast::Path { segments, .. }), token::Ident(kw::For | kw::Loop | kw::While, IdentIsRaw::No), - ) if segments.len() == 1 => { + ) if let [segment] = segments.as_slice() => { let snapshot = self.create_snapshot_for_diagnostic(); let label = Label { ident: Ident::from_str_and_span( - &format!("'{}", segments[0].ident), - segments[0].ident.span, + &format!("'{}", segment.ident), + segment.ident.span, ), }; match self.parse_expr_labeled(label, false) { @@ -886,7 +879,7 @@ impl<'a> Parser<'a> { mut e: P<Expr>, lo: Span, ) -> PResult<'a, P<Expr>> { - let res = ensure_sufficient_stack(|| { + let mut res = ensure_sufficient_stack(|| { loop { let has_question = if self.prev_token.kind == TokenKind::Ident(kw::Return, IdentIsRaw::No) { @@ -933,17 +926,13 @@ impl<'a> Parser<'a> { // Stitch the list of outer attributes onto the return value. A little // bit ugly, but the best way given the current code structure. - if attrs.is_empty() { - res - } else { - res.map(|expr| { - expr.map(|mut expr| { - attrs.extend(expr.attrs); - expr.attrs = attrs; - expr - }) - }) + if !attrs.is_empty() + && let Ok(expr) = &mut res + { + mem::swap(&mut expr.attrs, &mut attrs); + expr.attrs.extend(attrs) } + res } pub(super) fn parse_dot_suffix_expr( @@ -2644,10 +2633,7 @@ impl<'a> Parser<'a> { self.expect(&token::Eq)?; } let attrs = self.parse_outer_attributes()?; - let expr = self.parse_expr_assoc_with( - 1 + prec_let_scrutinee_needs_par(), - LhsExpr::Unparsed { attrs }, - )?; + let expr = self.parse_expr_assoc_with(1 + prec_let_scrutinee_needs_par(), attrs)?; let span = lo.to(expr.span); Ok(self.mk_expr(span, ExprKind::Let(pat, expr, span, recovered))) } @@ -3152,7 +3138,8 @@ impl<'a> Parser<'a> { if !require_comma { arm_body = Some(expr); - this.eat(&token::Comma); + // Eat a comma if it exists, though. + let _ = this.eat(&token::Comma); Ok(Recovered::No) } else if let Some((span, guar)) = this.parse_arm_body_missing_braces(&expr, arrow_span) @@ -3653,7 +3640,7 @@ impl<'a> Parser<'a> { fields.push(f); } self.recover_stmt_(SemiColonMode::Comma, BlockMode::Ignore); - self.eat(&token::Comma); + let _ = self.eat(&token::Comma); } } } diff --git a/compiler/rustc_parse/src/parser/generics.rs b/compiler/rustc_parse/src/parser/generics.rs index 523538e9643..9124c15707d 100644 --- a/compiler/rustc_parse/src/parser/generics.rs +++ b/compiler/rustc_parse/src/parser/generics.rs @@ -1,21 +1,19 @@ -use crate::errors::{ - self, MultipleWhereClauses, UnexpectedDefaultValueForLifetimeInGenericParameters, - UnexpectedSelfInGenericParameters, WhereClauseBeforeTupleStructBody, - WhereClauseBeforeTupleStructBodySugg, -}; - -use super::{ForceCollect, Parser}; - use ast::token::Delimiter; -use rustc_ast::token; use rustc_ast::{ - self as ast, AttrVec, GenericBounds, GenericParam, GenericParamKind, TyKind, WhereClause, + self as ast, token, AttrVec, GenericBounds, GenericParam, GenericParamKind, TyKind, WhereClause, }; use rustc_errors::{Applicability, PResult}; use rustc_span::symbol::{kw, Ident}; use rustc_span::Span; use thin_vec::ThinVec; +use super::{ForceCollect, Parser}; +use crate::errors::{ + self, MultipleWhereClauses, UnexpectedDefaultValueForLifetimeInGenericParameters, + UnexpectedSelfInGenericParameters, WhereClauseBeforeTupleStructBody, + WhereClauseBeforeTupleStructBodySugg, +}; + enum PredicateOrStructBody { Predicate(ast::WherePredicate), StructBody(ThinVec<ast::FieldDef>), @@ -180,7 +178,8 @@ impl<'a> Parser<'a> { span: this.prev_token.span, }); - this.eat(&token::Comma); + // Eat a trailing comma, if it exists. + let _ = this.eat(&token::Comma); } let param = if this.check_lifetime() { diff --git a/compiler/rustc_parse/src/parser/item.rs b/compiler/rustc_parse/src/parser/item.rs index 112855e6d1f..8775d792c3d 100644 --- a/compiler/rustc_parse/src/parser/item.rs +++ b/compiler/rustc_parse/src/parser/item.rs @@ -1,9 +1,6 @@ -use super::diagnostics::{dummy_arg, ConsumeClosingDelim}; -use super::ty::{AllowPlus, RecoverQPath, RecoverReturnSign}; -use super::{AttrWrapper, FollowedByType, ForceCollect, Parser, PathStyle, Trailing}; -use crate::errors::{self, MacroExpandsToAdtField}; -use crate::fluent_generated as fluent; -use crate::maybe_whole; +use std::fmt::Write; +use std::mem; + use ast::token::IdentIsRaw; use rustc_ast::ast::*; use rustc_ast::ptr::P; @@ -12,18 +9,21 @@ use rustc_ast::tokenstream::{DelimSpan, TokenStream, TokenTree}; use rustc_ast::util::case::Case; use rustc_ast::{self as ast}; use rustc_ast_pretty::pprust; -use rustc_errors::{codes::*, struct_span_code_err, Applicability, PResult, StashKey}; +use rustc_errors::codes::*; +use rustc_errors::{struct_span_code_err, Applicability, PResult, StashKey}; use rustc_span::edit_distance::edit_distance; use rustc_span::edition::Edition; -use rustc_span::source_map; use rustc_span::symbol::{kw, sym, Ident, Symbol}; -use rustc_span::ErrorGuaranteed; -use rustc_span::{Span, DUMMY_SP}; -use std::fmt::Write; -use std::mem; +use rustc_span::{source_map, ErrorGuaranteed, Span, DUMMY_SP}; use thin_vec::{thin_vec, ThinVec}; use tracing::debug; +use super::diagnostics::{dummy_arg, ConsumeClosingDelim}; +use super::ty::{AllowPlus, RecoverQPath, RecoverReturnSign}; +use super::{AttrWrapper, FollowedByType, ForceCollect, Parser, PathStyle, Trailing}; +use crate::errors::{self, MacroExpandsToAdtField}; +use crate::{fluent_generated as fluent, maybe_whole}; + impl<'a> Parser<'a> { /// Parses a source module as a crate. This is the main entry point for the parser. pub fn parse_crate_mod(&mut self) -> PResult<'a, ast::Crate> { @@ -471,9 +471,8 @@ impl<'a> Parser<'a> { Err(mut err) => { // Maybe the user misspelled `macro_rules` (issue #91227) if self.token.is_ident() - && path.segments.len() == 1 - && edit_distance("macro_rules", &path.segments[0].ident.to_string(), 2) - .is_some() + && let [segment] = path.segments.as_slice() + && edit_distance("macro_rules", &segment.ident.to_string(), 2).is_some() { err.span_suggestion( path.span, @@ -1192,13 +1191,14 @@ impl<'a> Parser<'a> { mut safety: Safety, ) -> PResult<'a, ItemInfo> { let abi = self.parse_abi(); // ABI? + // FIXME: This recovery should be tested better. if safety == Safety::Default && self.token.is_keyword(kw::Unsafe) && self.look_ahead(1, |t| t.kind == token::OpenDelim(Delimiter::Brace)) { self.expect(&token::OpenDelim(Delimiter::Brace)).unwrap_err().emit(); safety = Safety::Unsafe(self.token.span); - self.eat_keyword(kw::Unsafe); + let _ = self.eat_keyword(kw::Unsafe); } let module = ast::ForeignMod { safety, @@ -1759,7 +1759,7 @@ impl<'a> Parser<'a> { } } } - self.eat(&token::CloseDelim(Delimiter::Brace)); + self.expect(&token::CloseDelim(Delimiter::Brace))?; } else { let token_str = super::token_descr(&self.token); let where_str = if parsed_where { "" } else { "`where`, or " }; @@ -1902,7 +1902,7 @@ impl<'a> Parser<'a> { if let Some(_guar) = guar { // Handle a case like `Vec<u8>>,` where we can continue parsing fields // after the comma - self.eat(&token::Comma); + let _ = self.eat(&token::Comma); // `check_trailing_angle_brackets` already emitted a nicer error, as // proven by the presence of `_guar`. We can continue parsing. diff --git a/compiler/rustc_parse/src/parser/mod.rs b/compiler/rustc_parse/src/parser/mod.rs index e7240869a39..9b3b6d5f9ad 100644 --- a/compiler/rustc_parse/src/parser/mod.rs +++ b/compiler/rustc_parse/src/parser/mod.rs @@ -10,18 +10,21 @@ mod path; mod stmt; mod ty; -use crate::lexer::UnmatchedDelim; +use std::assert_matches::debug_assert_matches; +use std::ops::Range; +use std::{fmt, mem, slice}; + use attr_wrapper::AttrWrapper; pub use diagnostics::AttemptLocalParseRecovery; pub(crate) use expr::ForbiddenLetReason; pub(crate) use item::FnParseMode; pub use pat::{CommaRecoveryMode, RecoverColon, RecoverComma}; use path::PathStyle; - use rustc_ast::ptr::P; use rustc_ast::token::{self, Delimiter, IdentIsRaw, Nonterminal, Token, TokenKind}; -use rustc_ast::tokenstream::{AttrsTarget, DelimSpacing, DelimSpan, Spacing}; -use rustc_ast::tokenstream::{TokenStream, TokenTree, TokenTreeCursor}; +use rustc_ast::tokenstream::{ + AttrsTarget, DelimSpacing, DelimSpan, Spacing, TokenStream, TokenTree, TokenTreeCursor, +}; use rustc_ast::util::case::Case; use rustc_ast::{ self as ast, AnonConst, AttrArgs, AttrArgsEq, AttrId, ByRef, Const, CoroutineKind, DelimArgs, @@ -35,14 +38,13 @@ use rustc_errors::{Applicability, Diag, FatalError, MultiSpan, PResult}; use rustc_session::parse::ParseSess; use rustc_span::symbol::{kw, sym, Ident, Symbol}; use rustc_span::{Span, DUMMY_SP}; -use std::ops::Range; -use std::{fmt, mem, slice}; use thin_vec::ThinVec; use tracing::debug; use crate::errors::{ self, IncorrectVisibilityRestriction, MismatchedClosingDelimiter, NonStringAbiLiteral, }; +use crate::lexer::UnmatchedDelim; #[cfg(test)] mod tests; @@ -191,24 +193,54 @@ struct ClosureSpans { body: Span, } -/// Indicates a range of tokens that should be replaced by -/// the tokens in the provided `AttrsTarget`. This is used in two -/// places during token collection: +/// A token range within a `Parser`'s full token stream. +#[derive(Clone, Debug)] +struct ParserRange(Range<u32>); + +/// A token range within an individual AST node's (lazy) token stream, i.e. +/// relative to that node's first token. Distinct from `ParserRange` so the two +/// kinds of range can't be mixed up. +#[derive(Clone, Debug)] +struct NodeRange(Range<u32>); + +/// Indicates a range of tokens that should be replaced by an `AttrsTarget` +/// (replacement) or be replaced by nothing (deletion). This is used in two +/// places during token collection. /// -/// 1. During the parsing of an AST node that may have a `#[derive]` -/// attribute, we parse a nested AST node that has `#[cfg]` or `#[cfg_attr]` -/// In this case, we use a `ReplaceRange` to replace the entire inner AST node -/// with `FlatToken::AttrsTarget`, allowing us to perform eager cfg-expansion -/// on an `AttrTokenStream`. +/// 1. Replacement. During the parsing of an AST node that may have a +/// `#[derive]` attribute, when we parse a nested AST node that has `#[cfg]` +/// or `#[cfg_attr]`, we replace the entire inner AST node with +/// `FlatToken::AttrsTarget`. This lets us perform eager cfg-expansion on an +/// `AttrTokenStream`. /// -/// 2. When we parse an inner attribute while collecting tokens. We -/// remove inner attributes from the token stream entirely, and -/// instead track them through the `attrs` field on the AST node. -/// This allows us to easily manipulate them (for example, removing -/// the first macro inner attribute to invoke a proc-macro). -/// When create a `TokenStream`, the inner attributes get inserted -/// into the proper place in the token stream. -type ReplaceRange = (Range<u32>, Option<AttrsTarget>); +/// 2. Deletion. We delete inner attributes from all collected token streams, +/// and instead track them through the `attrs` field on the AST node. This +/// lets us manipulate them similarly to outer attributes. When we create a +/// `TokenStream`, the inner attributes are inserted into the proper place +/// in the token stream. +/// +/// Each replacement starts off in `ParserReplacement` form but is converted to +/// `NodeReplacement` form when it is attached to a single AST node, via +/// `LazyAttrTokenStreamImpl`. +type ParserReplacement = (ParserRange, Option<AttrsTarget>); + +/// See the comment on `ParserReplacement`. +type NodeReplacement = (NodeRange, Option<AttrsTarget>); + +impl NodeRange { + // Converts a range within a parser's tokens to a range within a + // node's tokens beginning at `start_pos`. + // + // For example, imagine a parser with 50 tokens in its token stream, a + // function that spans `ParserRange(20..40)` and an inner attribute within + // that function that spans `ParserRange(30..35)`. We would find the inner + // attribute's range within the function's tokens by subtracting 20, which + // is the position of the function's start token. This gives + // `NodeRange(10..15)`. + fn new(ParserRange(parser_range): ParserRange, start_pos: u32) -> NodeRange { + NodeRange((parser_range.start - start_pos)..(parser_range.end - start_pos)) + } +} /// Controls how we capture tokens. Capturing can be expensive, /// so we try to avoid performing capturing in cases where @@ -225,8 +257,8 @@ enum Capturing { #[derive(Clone, Debug)] struct CaptureState { capturing: Capturing, - replace_ranges: Vec<ReplaceRange>, - inner_attr_ranges: FxHashMap<AttrId, Range<u32>>, + parser_replacements: Vec<ParserReplacement>, + inner_attr_parser_ranges: FxHashMap<AttrId, ParserRange>, } /// Iterator over a `TokenStream` that produces `Token`s. It's a bit odd that @@ -416,8 +448,8 @@ impl<'a> Parser<'a> { subparser_name, capture_state: CaptureState { capturing: Capturing::No, - replace_ranges: Vec::new(), - inner_attr_ranges: Default::default(), + parser_replacements: Vec::new(), + inner_attr_parser_ranges: Default::default(), }, current_closure: None, recovery: Recovery::Allowed, @@ -546,6 +578,7 @@ impl<'a> Parser<'a> { } #[inline] + #[must_use] fn check_noexpect(&self, tok: &TokenKind) -> bool { self.token == *tok } @@ -555,6 +588,7 @@ impl<'a> Parser<'a> { /// the main purpose of this function is to reduce the cluttering of the suggestions list /// which using the normal eat method could introduce in some cases. #[inline] + #[must_use] fn eat_noexpect(&mut self, tok: &TokenKind) -> bool { let is_present = self.check_noexpect(tok); if is_present { @@ -565,6 +599,7 @@ impl<'a> Parser<'a> { /// Consumes a token 'tok' if it exists. Returns whether the given token was present. #[inline] + #[must_use] pub fn eat(&mut self, tok: &TokenKind) -> bool { let is_present = self.check(tok); if is_present { @@ -576,12 +611,14 @@ impl<'a> Parser<'a> { /// If the next token is the given keyword, returns `true` without eating it. /// An expectation is also added for diagnostics purposes. #[inline] + #[must_use] fn check_keyword(&mut self, kw: Symbol) -> bool { self.expected_tokens.push(TokenType::Keyword(kw)); self.token.is_keyword(kw) } #[inline] + #[must_use] fn check_keyword_case(&mut self, kw: Symbol, case: Case) -> bool { if self.check_keyword(kw) { return true; @@ -601,6 +638,7 @@ impl<'a> Parser<'a> { /// Otherwise, returns `false`. An expectation is also added for diagnostics purposes. // Public for rustc_builtin_macros and rustfmt usage. #[inline] + #[must_use] pub fn eat_keyword(&mut self, kw: Symbol) -> bool { if self.check_keyword(kw) { self.bump(); @@ -614,6 +652,7 @@ impl<'a> Parser<'a> { /// If the case differs (and is ignored) an error is issued. /// This is useful for recovery. #[inline] + #[must_use] fn eat_keyword_case(&mut self, kw: Symbol, case: Case) -> bool { if self.eat_keyword(kw) { return true; @@ -635,6 +674,7 @@ impl<'a> Parser<'a> { /// Otherwise, returns `false`. No expectation is added. // Public for rustc_builtin_macros usage. #[inline] + #[must_use] pub fn eat_keyword_noexpect(&mut self, kw: Symbol) -> bool { if self.token.is_keyword(kw) { self.bump(); @@ -647,7 +687,7 @@ impl<'a> Parser<'a> { /// If the given word is not a keyword, signals an error. /// If the next token is not the given word, signals an error. /// Otherwise, eats it. - fn expect_keyword(&mut self, kw: Symbol) -> PResult<'a, ()> { + pub fn expect_keyword(&mut self, kw: Symbol) -> PResult<'a, ()> { if !self.eat_keyword(kw) { self.unexpected() } else { Ok(()) } } @@ -1024,8 +1064,11 @@ impl<'a> Parser<'a> { f: impl FnMut(&mut Parser<'a>) -> PResult<'a, T>, ) -> PResult<'a, (ThinVec<T>, Trailing)> { let (val, trailing, recovered) = self.parse_seq_to_before_end(ket, sep, f)?; - if matches!(recovered, Recovered::No) { - self.eat(ket); + if matches!(recovered, Recovered::No) && !self.eat(ket) { + self.dcx().span_delayed_bug( + self.token.span, + "recovered but `parse_seq_to_before_end` did not give us the ket token", + ); } Ok((val, trailing)) } @@ -1124,10 +1167,12 @@ impl<'a> Parser<'a> { match self.token_cursor.tree_cursor.look_ahead(0) { Some(tree) => { // Indexing stayed within the current token tree. - return match tree { - TokenTree::Token(token, _) => looker(token), - TokenTree::Delimited(dspan, _, delim, _) => { - looker(&Token::new(token::OpenDelim(*delim), dspan.open)) + match tree { + TokenTree::Token(token, _) => return looker(token), + &TokenTree::Delimited(dspan, _, delim, _) => { + if delim != Delimiter::Invisible { + return looker(&Token::new(token::OpenDelim(delim), dspan.open)); + } } }; } @@ -1208,9 +1253,6 @@ impl<'a> Parser<'a> { if self.eat_keyword_case(kw::Unsafe, case) { Safety::Unsafe(self.prev_token.uninterpolated_span()) } else if self.eat_keyword_case(kw::Safe, case) { - self.psess - .gated_spans - .gate(sym::unsafe_extern_blocks, self.prev_token.uninterpolated_span()); Safety::Safe(self.prev_token.uninterpolated_span()) } else { Safety::Default @@ -1249,7 +1291,7 @@ impl<'a> Parser<'a> { if pat { self.psess.gated_spans.gate(sym::inline_const_pat, span); } - self.eat_keyword(kw::Const); + self.expect_keyword(kw::Const)?; let (attrs, blk) = self.parse_inner_attrs_and_block()?; let anon_const = AnonConst { id: DUMMY_NODE_ID, @@ -1346,7 +1388,7 @@ impl<'a> Parser<'a> { // can capture these tokens if necessary. self.bump(); if self.token_cursor.stack.len() == target_depth { - debug_assert!(matches!(self.token.kind, token::CloseDelim(_))); + debug_assert_matches!(self.token.kind, token::CloseDelim(_)); break; } } diff --git a/compiler/rustc_parse/src/parser/mut_visit/tests.rs b/compiler/rustc_parse/src/parser/mut_visit/tests.rs index 677bcdf7fcd..b82c295732d 100644 --- a/compiler/rustc_parse/src/parser/mut_visit/tests.rs +++ b/compiler/rustc_parse/src/parser/mut_visit/tests.rs @@ -1,10 +1,11 @@ -use crate::parser::tests::{matches_codepattern, string_to_crate}; use rustc_ast as ast; use rustc_ast::mut_visit::MutVisitor; use rustc_ast_pretty::pprust; use rustc_span::create_default_session_globals_then; use rustc_span::symbol::Ident; +use crate::parser::tests::{matches_codepattern, string_to_crate}; + // This version doesn't care about getting comments or doc-strings in. fn print_crate_items(krate: &ast::Crate) -> String { krate.items.iter().map(|i| pprust::item_to_string(i)).collect::<Vec<_>>().join(" ") diff --git a/compiler/rustc_parse/src/parser/nonterminal.rs b/compiler/rustc_parse/src/parser/nonterminal.rs index 886d6af1735..999f6f0eeb0 100644 --- a/compiler/rustc_parse/src/parser/nonterminal.rs +++ b/compiler/rustc_parse/src/parser/nonterminal.rs @@ -1,7 +1,8 @@ use rustc_ast::ptr::P; -use rustc_ast::token::{ - self, Delimiter, Nonterminal::*, NonterminalKind, NtExprKind::*, NtPatKind::*, Token, -}; +use rustc_ast::token::Nonterminal::*; +use rustc_ast::token::NtExprKind::*; +use rustc_ast::token::NtPatKind::*; +use rustc_ast::token::{self, Delimiter, NonterminalKind, Token}; use rustc_ast::HasTokens; use rustc_ast_pretty::pprust; use rustc_data_structures::sync::Lrc; @@ -38,6 +39,7 @@ impl<'a> Parser<'a> { } match kind { + // `expr_2021` and earlier NonterminalKind::Expr(Expr2021 { .. }) => { token.can_begin_expr() // This exception is here for backwards compatibility. @@ -45,8 +47,16 @@ impl<'a> Parser<'a> { // This exception is here for backwards compatibility. && !token.is_keyword(kw::Const) } + // Current edition expressions NonterminalKind::Expr(Expr) => { - token.can_begin_expr() + // In Edition 2024, `_` is considered an expression, so we + // need to allow it here because `token.can_begin_expr()` does + // not consider `_` to be an expression. + // + // Because `can_begin_expr` is used elsewhere, we need to reduce + // the scope of where the `_` is considered an expression to + // just macro parsing code. + (token.can_begin_expr() || token.is_keyword(kw::Underscore)) // This exception is here for backwards compatibility. && !token.is_keyword(kw::Let) } diff --git a/compiler/rustc_parse/src/parser/pat.rs b/compiler/rustc_parse/src/parser/pat.rs index aa818878cd8..5bfb8bdf776 100644 --- a/compiler/rustc_parse/src/parser/pat.rs +++ b/compiler/rustc_parse/src/parser/pat.rs @@ -1,17 +1,3 @@ -use super::{ForceCollect, Parser, PathStyle, Restrictions, Trailing}; -use crate::errors::{ - self, AmbiguousRangePattern, DotDotDotForRemainingFields, DotDotDotRangeToPatternNotAllowed, - DotDotDotRestPattern, EnumPatternInsteadOfIdentifier, ExpectedBindingLeftOfAt, - ExpectedCommaAfterPatternField, GenericArgsInPatRequireTurbofishSyntax, - InclusiveRangeExtraEquals, InclusiveRangeMatchArrow, InclusiveRangeNoEnd, InvalidMutInPattern, - ParenRangeSuggestion, PatternOnWrongSideOfAt, RemoveLet, RepeatedMutInPattern, - SwitchRefBoxOrder, TopLevelOrPatternNotAllowed, TopLevelOrPatternNotAllowedSugg, - TrailingVertNotAllowed, UnexpectedExpressionInPattern, UnexpectedLifetimeInPattern, - UnexpectedParenInRangePat, UnexpectedParenInRangePatSugg, - UnexpectedVertVertBeforeFunctionParam, UnexpectedVertVertInPattern, WrapInParens, -}; -use crate::parser::expr::{could_be_unclosed_char_literal, LhsExpr}; -use crate::{maybe_recover_from_interpolated_ty_qpath, maybe_whole}; use rustc_ast::mut_visit::{walk_pat, MutVisitor}; use rustc_ast::ptr::P; use rustc_ast::token::{self, BinOpToken, Delimiter, Token}; @@ -27,6 +13,21 @@ use rustc_span::symbol::{kw, sym, Ident}; use rustc_span::{BytePos, ErrorGuaranteed, Span}; use thin_vec::{thin_vec, ThinVec}; +use super::{ForceCollect, Parser, PathStyle, Restrictions, Trailing}; +use crate::errors::{ + self, AmbiguousRangePattern, DotDotDotForRemainingFields, DotDotDotRangeToPatternNotAllowed, + DotDotDotRestPattern, EnumPatternInsteadOfIdentifier, ExpectedBindingLeftOfAt, + ExpectedCommaAfterPatternField, GenericArgsInPatRequireTurbofishSyntax, + InclusiveRangeExtraEquals, InclusiveRangeMatchArrow, InclusiveRangeNoEnd, InvalidMutInPattern, + ParenRangeSuggestion, PatternOnWrongSideOfAt, RemoveLet, RepeatedMutInPattern, + SwitchRefBoxOrder, TopLevelOrPatternNotAllowed, TopLevelOrPatternNotAllowedSugg, + TrailingVertNotAllowed, UnexpectedExpressionInPattern, UnexpectedLifetimeInPattern, + UnexpectedParenInRangePat, UnexpectedParenInRangePatSugg, + UnexpectedVertVertBeforeFunctionParam, UnexpectedVertVertInPattern, WrapInParens, +}; +use crate::parser::expr::could_be_unclosed_char_literal; +use crate::{maybe_recover_from_interpolated_ty_qpath, maybe_whole}; + #[derive(PartialEq, Copy, Clone)] pub enum Expected { ParameterName, @@ -402,8 +403,9 @@ impl<'a> Parser<'a> { // Parse an associative expression such as `+ expr`, `% expr`, ... // Assignements, ranges and `|` are disabled by [`Restrictions::IS_PAT`]. - let lhs = LhsExpr::Parsed { expr, starts_statement: false }; - if let Ok(expr) = snapshot.parse_expr_assoc_with(0, lhs).map_err(|err| err.cancel()) { + if let Ok(expr) = + snapshot.parse_expr_assoc_rest_with(0, false, expr).map_err(|err| err.cancel()) + { // We got a valid expression. self.restore_snapshot(snapshot); self.restrictions.remove(Restrictions::IS_PAT); diff --git a/compiler/rustc_parse/src/parser/path.rs b/compiler/rustc_parse/src/parser/path.rs index b9014dea726..6f82d6b9826 100644 --- a/compiler/rustc_parse/src/parser/path.rs +++ b/compiler/rustc_parse/src/parser/path.rs @@ -1,8 +1,5 @@ -use super::ty::{AllowPlus, RecoverQPath, RecoverReturnSign}; -use super::{Parser, Restrictions, TokenType}; -use crate::errors::PathSingleColon; -use crate::parser::{CommaRecoveryMode, RecoverColon, RecoverComma}; -use crate::{errors, maybe_whole}; +use std::mem; + use ast::token::IdentIsRaw; use rustc_ast::ptr::P; use rustc_ast::token::{self, Delimiter, Token, TokenKind}; @@ -14,10 +11,15 @@ use rustc_ast::{ use rustc_errors::{Applicability, Diag, PResult}; use rustc_span::symbol::{kw, sym, Ident}; use rustc_span::{BytePos, Span}; -use std::mem; use thin_vec::ThinVec; use tracing::debug; +use super::ty::{AllowPlus, RecoverQPath, RecoverReturnSign}; +use super::{Parser, Restrictions, TokenType}; +use crate::errors::PathSingleColon; +use crate::parser::{CommaRecoveryMode, RecoverColon, RecoverComma}; +use crate::{errors, maybe_whole}; + /// Specifies how to parse a path. #[derive(Copy, Clone, PartialEq)] pub(super) enum PathStyle { @@ -311,7 +313,8 @@ impl<'a> Parser<'a> { } // Generic arguments are found - `<`, `(`, `::<` or `::(`. - self.eat(&token::PathSep); + // First, eat `::` if it exists. + let _ = self.eat(&token::PathSep); let lo = self.token.span; let args = if self.eat_lt() { // `<'a, T, A = U>` @@ -823,7 +826,8 @@ impl<'a> Parser<'a> { // We can only resolve single-segment paths at the moment, because multi-segment paths // require type-checking: see `visit_generic_arg` in `src/librustc_resolve/late.rs`. ast::ExprKind::Path(None, path) - if path.segments.len() == 1 && path.segments[0].args.is_none() => + if let [segment] = path.segments.as_slice() + && segment.args.is_none() => { true } diff --git a/compiler/rustc_parse/src/parser/stmt.rs b/compiler/rustc_parse/src/parser/stmt.rs index d8de7c1bfa1..b206f134f0e 100644 --- a/compiler/rustc_parse/src/parser/stmt.rs +++ b/compiler/rustc_parse/src/parser/stmt.rs @@ -1,31 +1,30 @@ -use super::attr::InnerAttrForbiddenReason; -use super::diagnostics::AttemptLocalParseRecovery; -use super::expr::LhsExpr; -use super::pat::{PatternLocation, RecoverComma}; -use super::path::PathStyle; -use super::{ - AttrWrapper, BlockMode, FnParseMode, ForceCollect, Parser, Restrictions, SemiColonMode, -}; -use crate::errors; -use crate::maybe_whole; +use std::borrow::Cow; +use std::mem; -use crate::errors::MalformedLoopLabel; use ast::Label; use rustc_ast as ast; use rustc_ast::ptr::P; use rustc_ast::token::{self, Delimiter, TokenKind}; use rustc_ast::util::classify::{self, TrailingBrace}; -use rustc_ast::{AttrStyle, AttrVec, LocalKind, MacCall, MacCallStmt, MacStmtStyle}; -use rustc_ast::{Block, BlockCheckMode, Expr, ExprKind, HasAttrs, Local, Recovered, Stmt}; -use rustc_ast::{StmtKind, DUMMY_NODE_ID}; +use rustc_ast::{ + AttrStyle, AttrVec, Block, BlockCheckMode, Expr, ExprKind, HasAttrs, Local, LocalKind, MacCall, + MacCallStmt, MacStmtStyle, Recovered, Stmt, StmtKind, DUMMY_NODE_ID, +}; use rustc_errors::{Applicability, Diag, PResult}; use rustc_span::symbol::{kw, sym, Ident}; use rustc_span::{BytePos, ErrorGuaranteed, Span}; - -use std::borrow::Cow; -use std::mem; use thin_vec::{thin_vec, ThinVec}; +use super::attr::InnerAttrForbiddenReason; +use super::diagnostics::AttemptLocalParseRecovery; +use super::pat::{PatternLocation, RecoverComma}; +use super::path::PathStyle; +use super::{ + AttrWrapper, BlockMode, FnParseMode, ForceCollect, Parser, Restrictions, SemiColonMode, +}; +use crate::errors::MalformedLoopLabel; +use crate::{errors, maybe_whole}; + impl<'a> Parser<'a> { /// Parses a statement. This stops just before trailing semicolons on everything but items. /// e.g., a `StmtKind::Semi` parses to a `StmtKind::Expr`, leaving the trailing `;` unconsumed. @@ -66,7 +65,12 @@ impl<'a> Parser<'a> { } Ok(Some(if self.token.is_keyword(kw::Let) { - self.parse_local_mk(lo, attrs, capture_semi, force_collect)? + self.collect_tokens_trailing_token(attrs, force_collect, |this, attrs| { + this.expect_keyword(kw::Let)?; + let local = this.parse_local(attrs)?; + let trailing = capture_semi && this.token.kind == token::Semi; + Ok((this.mk_stmt(lo.to(this.prev_token.span), StmtKind::Let(local)), trailing)) + })? } else if self.is_kw_followed_by_ident(kw::Mut) && self.may_recover() { self.recover_stmt_local_after_let( lo, @@ -112,13 +116,12 @@ impl<'a> Parser<'a> { } } } else if let Some(item) = self.parse_item_common( - attrs.clone(), + attrs.clone(), // FIXME: unwanted clone of attrs false, true, FnParseMode { req_name: |_| true, req_body: true }, force_collect, )? { - // FIXME: Bad copy of attrs self.mk_stmt(lo.to(item.span), StmtKind::Item(P(item))) } else if self.eat(&token::Semi) { // Do not attempt to parse an expression if we're done here. @@ -173,7 +176,7 @@ impl<'a> Parser<'a> { // Perform this outside of the `collect_tokens_trailing_token` closure, // since our outer attributes do not apply to this part of the expression let expr = self.with_res(Restrictions::STMT_EXPR, |this| { - this.parse_expr_assoc_with(0, LhsExpr::Parsed { expr, starts_statement: true }) + this.parse_expr_assoc_rest_with(0, true, expr) })?; Ok(self.mk_stmt(lo.to(self.prev_token.span), StmtKind::Expr(expr))) } else { @@ -206,8 +209,7 @@ impl<'a> Parser<'a> { let e = self.mk_expr(lo.to(hi), ExprKind::MacCall(mac)); let e = self.maybe_recover_from_bad_qpath(e)?; let e = self.parse_expr_dot_or_call_with(attrs, e, lo)?; - let e = self - .parse_expr_assoc_with(0, LhsExpr::Parsed { expr: e, starts_statement: false })?; + let e = self.parse_expr_assoc_rest_with(0, false, e)?; StmtKind::Expr(e) }; Ok(self.mk_stmt(lo.to(hi), kind)) @@ -247,21 +249,6 @@ impl<'a> Parser<'a> { Ok(stmt) } - fn parse_local_mk( - &mut self, - lo: Span, - attrs: AttrWrapper, - capture_semi: bool, - force_collect: ForceCollect, - ) -> PResult<'a, Stmt> { - self.collect_tokens_trailing_token(attrs, force_collect, |this, attrs| { - this.expect_keyword(kw::Let)?; - let local = this.parse_local(attrs)?; - let trailing = capture_semi && this.token.kind == token::Semi; - Ok((this.mk_stmt(lo.to(this.prev_token.span), StmtKind::Let(local)), trailing)) - }) - } - /// Parses a local variable declaration. fn parse_local(&mut self, attrs: AttrVec) -> PResult<'a, P<Local>> { let lo = self.prev_token.span; @@ -421,10 +408,14 @@ impl<'a> Parser<'a> { fn parse_initializer(&mut self, eq_optional: bool) -> PResult<'a, Option<P<Expr>>> { let eq_consumed = match self.token.kind { token::BinOpEq(..) => { - // Recover `let x <op>= 1` as `let x = 1` + // Recover `let x <op>= 1` as `let x = 1` We must not use `+ BytePos(1)` here + // because `<op>` can be a multi-byte lookalike that was recovered, e.g. `➖=` (the + // `➖` is a U+2796 Heavy Minus Sign Unicode Character) that was recovered as a + // `-=`. + let extra_op_span = self.psess.source_map().start_point(self.token.span); self.dcx().emit_err(errors::CompoundAssignmentExpressionInLet { span: self.token.span, - suggestion: self.token.span.with_hi(self.token.span.lo() + BytePos(1)), + suggestion: extra_op_span, }); self.bump(); true @@ -685,7 +676,7 @@ impl<'a> Parser<'a> { match &expr.kind { ExprKind::Path(None, ast::Path { segments, .. }) - if segments.len() == 1 => + if let [segment] = segments.as_slice() => { if self.token == token::Colon && self.look_ahead(1, |token| { @@ -702,8 +693,8 @@ impl<'a> Parser<'a> { let snapshot = self.create_snapshot_for_diagnostic(); let label = Label { ident: Ident::from_str_and_span( - &format!("'{}", segments[0].ident), - segments[0].ident.span, + &format!("'{}", segment.ident), + segment.ident.span, ), }; match self.parse_expr_labeled(label, false) { diff --git a/compiler/rustc_parse/src/parser/tests.rs b/compiler/rustc_parse/src/parser/tests.rs index 491aa71155a..cb8e8d30988 100644 --- a/compiler/rustc_parse/src/parser/tests.rs +++ b/compiler/rustc_parse/src/parser/tests.rs @@ -1,30 +1,28 @@ -use crate::parser::ForceCollect; -use crate::{ - new_parser_from_source_str, parser::Parser, source_str_to_stream, unwrap_or_emit_fatal, -}; +use std::assert_matches::assert_matches; +use std::io::prelude::*; +use std::iter::Peekable; +use std::path::{Path, PathBuf}; +use std::sync::{Arc, Mutex}; +use std::{io, str}; + use ast::token::IdentIsRaw; use rustc_ast::ptr::P; use rustc_ast::token::{self, Delimiter, Token}; use rustc_ast::tokenstream::{DelimSpacing, DelimSpan, Spacing, TokenStream, TokenTree}; -use rustc_ast::visit; -use rustc_ast::{self as ast, PatKind}; +use rustc_ast::{self as ast, visit, PatKind}; use rustc_ast_pretty::pprust::item_to_string; use rustc_data_structures::sync::Lrc; use rustc_errors::emitter::HumanEmitter; use rustc_errors::{DiagCtxt, MultiSpan, PResult}; use rustc_session::parse::ParseSess; -use rustc_span::create_default_session_globals_then; use rustc_span::source_map::{FilePathMapping, SourceMap}; use rustc_span::symbol::{kw, sym, Symbol}; -use rustc_span::{BytePos, FileName, Pos, Span}; -use std::io; -use std::io::prelude::*; -use std::iter::Peekable; -use std::path::{Path, PathBuf}; -use std::str; -use std::sync::{Arc, Mutex}; +use rustc_span::{create_default_session_globals_then, BytePos, FileName, Pos, Span}; use termcolor::WriteColor; +use crate::parser::{ForceCollect, Parser}; +use crate::{new_parser_from_source_str, source_str_to_stream, unwrap_or_emit_fatal}; + fn psess() -> ParseSess { ParseSess::new(vec![crate::DEFAULT_LOCALE_RESOURCE, crate::DEFAULT_LOCALE_RESOURCE]) } @@ -1750,7 +1748,7 @@ fn out_of_line_mod() { .unwrap(); let ast::ItemKind::Mod(_, mod_kind) = &item.kind else { panic!() }; - assert!(matches!(mod_kind, ast::ModKind::Loaded(items, ..) if items.len() == 2)); + assert_matches!(mod_kind, ast::ModKind::Loaded(items, ..) if items.len() == 2); }); } diff --git a/compiler/rustc_parse/src/parser/tokenstream/tests.rs b/compiler/rustc_parse/src/parser/tokenstream/tests.rs index 9be00a14791..d518dfee2b2 100644 --- a/compiler/rustc_parse/src/parser/tokenstream/tests.rs +++ b/compiler/rustc_parse/src/parser/tokenstream/tests.rs @@ -1,8 +1,8 @@ -use crate::parser::tests::string_to_stream; use rustc_ast::token::{self, IdentIsRaw}; use rustc_ast::tokenstream::{TokenStream, TokenTree}; -use rustc_span::create_default_session_globals_then; -use rustc_span::{BytePos, Span, Symbol}; +use rustc_span::{create_default_session_globals_then, BytePos, Span, Symbol}; + +use crate::parser::tests::string_to_stream; fn string_to_ts(string: &str) -> TokenStream { string_to_stream(string.to_owned()) diff --git a/compiler/rustc_parse/src/parser/ty.rs b/compiler/rustc_parse/src/parser/ty.rs index f95ecd254ce..352ddd9eac4 100644 --- a/compiler/rustc_parse/src/parser/ty.rs +++ b/compiler/rustc_parse/src/parser/ty.rs @@ -1,13 +1,3 @@ -use super::{Parser, PathStyle, SeqSep, TokenType, Trailing}; - -use crate::errors::{ - self, DynAfterMut, ExpectedFnPathFoundFnKeyword, ExpectedMutOrConstInRawPointerType, - FnPointerCannotBeAsync, FnPointerCannotBeConst, FnPtrWithGenerics, FnPtrWithGenericsSugg, - HelpUseLatestEdition, InvalidDynKeyword, LifetimeAfterMut, NeedPlusAfterTraitObjectLifetime, - NestedCVariadicType, ReturnTypesUseThinArrow, -}; -use crate::{maybe_recover_from_interpolated_ty_qpath, maybe_whole}; - use rustc_ast::ptr::P; use rustc_ast::token::{self, BinOpToken, Delimiter, Token, TokenKind}; use rustc_ast::util::case::Case; @@ -21,6 +11,15 @@ use rustc_span::symbol::{kw, sym, Ident}; use rustc_span::{ErrorGuaranteed, Span, Symbol}; use thin_vec::{thin_vec, ThinVec}; +use super::{Parser, PathStyle, SeqSep, TokenType, Trailing}; +use crate::errors::{ + self, DynAfterMut, ExpectedFnPathFoundFnKeyword, ExpectedMutOrConstInRawPointerType, + FnPointerCannotBeAsync, FnPointerCannotBeConst, FnPtrWithGenerics, FnPtrWithGenericsSugg, + HelpUseLatestEdition, InvalidDynKeyword, LifetimeAfterMut, NeedPlusAfterTraitObjectLifetime, + NestedCVariadicType, ReturnTypesUseThinArrow, +}; +use crate::{maybe_recover_from_interpolated_ty_qpath, maybe_whole}; + /// Signals whether parsing a type should allow `+`. /// /// For example, let T be the type `impl Default + 'static` diff --git a/compiler/rustc_parse/src/validate_attr.rs b/compiler/rustc_parse/src/validate_attr.rs index 3d5e6371f4c..a64c00f3b6c 100644 --- a/compiler/rustc_parse/src/validate_attr.rs +++ b/compiler/rustc_parse/src/validate_attr.rs @@ -1,7 +1,5 @@ //! Meta-syntax validation logic of attributes for post-expansion. -use crate::{errors, parse_in}; - use rustc_ast::token::Delimiter; use rustc_ast::tokenstream::DelimSpan; use rustc_ast::{ @@ -18,6 +16,8 @@ use rustc_session::lint::BuiltinLintDiag; use rustc_session::parse::ParseSess; use rustc_span::{sym, BytePos, Span, Symbol}; +use crate::{errors, parse_in}; + pub fn check_attr(features: &Features, psess: &ParseSess, attr: &Attribute) { if attr.is_doc_comment() { return; @@ -26,76 +26,35 @@ pub fn check_attr(features: &Features, psess: &ParseSess, attr: &Attribute) { let attr_info = attr.ident().and_then(|ident| BUILTIN_ATTRIBUTE_MAP.get(&ident.name)); let attr_item = attr.get_normal_item(); - let is_unsafe_attr = attr_info.is_some_and(|attr| attr.safety == AttributeSafety::Unsafe); - - if features.unsafe_attributes { - if is_unsafe_attr { - if let ast::Safety::Default = attr_item.unsafety { - let path_span = attr_item.path.span; - - // If the `attr_item`'s span is not from a macro, then just suggest - // wrapping it in `unsafe(...)`. Otherwise, we suggest putting the - // `unsafe(`, `)` right after and right before the opening and closing - // square bracket respectively. - let diag_span = if attr_item.span().can_be_used_for_suggestions() { - attr_item.span() - } else { - attr.span - .with_lo(attr.span.lo() + BytePos(2)) - .with_hi(attr.span.hi() - BytePos(1)) - }; - - if attr.span.at_least_rust_2024() { - psess.dcx().emit_err(errors::UnsafeAttrOutsideUnsafe { - span: path_span, - suggestion: errors::UnsafeAttrOutsideUnsafeSuggestion { - left: diag_span.shrink_to_lo(), - right: diag_span.shrink_to_hi(), - }, - }); - } else { - psess.buffer_lint( - UNSAFE_ATTR_OUTSIDE_UNSAFE, - path_span, - ast::CRATE_NODE_ID, - BuiltinLintDiag::UnsafeAttrOutsideUnsafe { - attribute_name_span: path_span, - sugg_spans: (diag_span.shrink_to_lo(), diag_span.shrink_to_hi()), - }, - ); - } - } - } else { - if let Safety::Unsafe(unsafe_span) = attr_item.unsafety { - psess.dcx().emit_err(errors::InvalidAttrUnsafe { - span: unsafe_span, - name: attr_item.path.clone(), - }); - } - } - } + // All non-builtin attributes are considered safe + let safety = attr_info.map(|x| x.safety).unwrap_or(AttributeSafety::Normal); + check_attribute_safety(features, psess, safety, attr); // Check input tokens for built-in and key-value attributes. match attr_info { // `rustc_dummy` doesn't have any restrictions specific to built-in attributes. Some(BuiltinAttribute { name, template, .. }) if *name != sym::rustc_dummy => { match parse_meta(psess, attr) { - Ok(meta) => check_builtin_meta_item(psess, &meta, attr.style, *name, *template), + // Don't check safety again, we just did that + Ok(meta) => check_builtin_meta_item( + features, psess, &meta, attr.style, *name, *template, false, + ), Err(err) => { err.emit(); } } } - _ if let AttrArgs::Eq(..) = attr_item.args => { - // All key-value attributes are restricted to meta-item syntax. - match parse_meta(psess, attr) { - Ok(_) => {} - Err(err) => { - err.emit(); + _ => { + if let AttrArgs::Eq(..) = attr_item.args { + // All key-value attributes are restricted to meta-item syntax. + match parse_meta(psess, attr) { + Ok(_) => {} + Err(err) => { + err.emit(); + } } } } - _ => {} } } @@ -198,12 +157,85 @@ fn is_attr_template_compatible(template: &AttributeTemplate, meta: &ast::MetaIte } } +pub fn check_attribute_safety( + features: &Features, + psess: &ParseSess, + safety: AttributeSafety, + attr: &Attribute, +) { + if !features.unsafe_attributes { + return; + } + + let attr_item = attr.get_normal_item(); + + if safety == AttributeSafety::Unsafe { + if let ast::Safety::Default = attr_item.unsafety { + let path_span = attr_item.path.span; + + // If the `attr_item`'s span is not from a macro, then just suggest + // wrapping it in `unsafe(...)`. Otherwise, we suggest putting the + // `unsafe(`, `)` right after and right before the opening and closing + // square bracket respectively. + let diag_span = if attr_item.span().can_be_used_for_suggestions() { + attr_item.span() + } else { + attr.span.with_lo(attr.span.lo() + BytePos(2)).with_hi(attr.span.hi() - BytePos(1)) + }; + + if attr.span.at_least_rust_2024() { + psess.dcx().emit_err(errors::UnsafeAttrOutsideUnsafe { + span: path_span, + suggestion: errors::UnsafeAttrOutsideUnsafeSuggestion { + left: diag_span.shrink_to_lo(), + right: diag_span.shrink_to_hi(), + }, + }); + } else { + psess.buffer_lint( + UNSAFE_ATTR_OUTSIDE_UNSAFE, + path_span, + ast::CRATE_NODE_ID, + BuiltinLintDiag::UnsafeAttrOutsideUnsafe { + attribute_name_span: path_span, + sugg_spans: (diag_span.shrink_to_lo(), diag_span.shrink_to_hi()), + }, + ); + } + } + } else { + if let Safety::Unsafe(unsafe_span) = attr_item.unsafety { + psess.dcx().emit_err(errors::InvalidAttrUnsafe { + span: unsafe_span, + name: attr_item.path.clone(), + }); + } + } +} + +// Called by `check_builtin_meta_item` and code that manually denies +// `unsafe(...)` in `cfg` +pub fn deny_builtin_meta_unsafety(features: &Features, psess: &ParseSess, meta: &MetaItem) { + // This only supports denying unsafety right now - making builtin attributes + // support unsafety will requite us to thread the actual `Attribute` through + // for the nice diagnostics. + if features.unsafe_attributes { + if let Safety::Unsafe(unsafe_span) = meta.unsafety { + psess + .dcx() + .emit_err(errors::InvalidAttrUnsafe { span: unsafe_span, name: meta.path.clone() }); + } + } +} + pub fn check_builtin_meta_item( + features: &Features, psess: &ParseSess, meta: &MetaItem, style: ast::AttrStyle, name: Symbol, template: AttributeTemplate, + deny_unsafety: bool, ) { // Some special attributes like `cfg` must be checked // before the generic check, so we skip them here. @@ -212,6 +244,10 @@ pub fn check_builtin_meta_item( if !should_skip(name) && !is_attr_template_compatible(&template, &meta.kind) { emit_malformed_attribute(psess, style, meta.span, name, template); } + + if deny_unsafety { + deny_builtin_meta_unsafety(features, psess, meta); + } } fn emit_malformed_attribute( |
