diff options
Diffstat (limited to 'compiler/rustc_parse/src')
| -rw-r--r-- | compiler/rustc_parse/src/errors.rs | 15 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/lexer/mod.rs | 107 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/lexer/unescape_error_reporting.rs | 42 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/attr_wrapper.rs | 12 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/diagnostics.rs | 36 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/expr.rs | 43 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/item.rs | 108 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/mod.rs | 58 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/path.rs | 4 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/stmt.rs | 12 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/ty.rs | 7 |
11 files changed, 282 insertions, 162 deletions
diff --git a/compiler/rustc_parse/src/errors.rs b/compiler/rustc_parse/src/errors.rs index 0924c853715..d59982f7063 100644 --- a/compiler/rustc_parse/src/errors.rs +++ b/compiler/rustc_parse/src/errors.rs @@ -421,6 +421,15 @@ pub(crate) struct ExpectedExpressionFoundLet { } #[derive(Diagnostic)] +#[diag(parser_expect_eq_instead_of_eqeq)] +pub(crate) struct ExpectedEqForLetExpr { + #[primary_span] + pub span: Span, + #[suggestion(applicability = "maybe-incorrect", code = "=", style = "verbose")] + pub sugg_span: Span, +} + +#[derive(Diagnostic)] #[diag(parser_expected_else_block)] pub(crate) struct ExpectedElseBlock { #[primary_span] @@ -1156,10 +1165,12 @@ pub(crate) struct ParenthesesInForHead { #[derive(Subdiagnostic)] #[multipart_suggestion(suggestion, applicability = "machine-applicable")] pub(crate) struct ParenthesesInForHeadSugg { - #[suggestion_part(code = "")] + #[suggestion_part(code = "{left_snippet}")] pub left: Span, - #[suggestion_part(code = "")] + pub left_snippet: String, + #[suggestion_part(code = "{right_snippet}")] pub right: Span, + pub right_snippet: String, } #[derive(Diagnostic)] diff --git a/compiler/rustc_parse/src/lexer/mod.rs b/compiler/rustc_parse/src/lexer/mod.rs index de8f1c00c12..645262bd2f1 100644 --- a/compiler/rustc_parse/src/lexer/mod.rs +++ b/compiler/rustc_parse/src/lexer/mod.rs @@ -353,55 +353,55 @@ impl<'a> StringReader<'a> { fn cook_lexer_literal( &self, start: BytePos, - suffix_start: BytePos, + end: BytePos, kind: rustc_lexer::LiteralKind, ) -> (token::LitKind, Symbol) { - // prefix means `"` or `br"` or `r###"`, ... - let (lit_kind, mode, prefix_len, postfix_len) = match kind { + match kind { rustc_lexer::LiteralKind::Char { terminated } => { if !terminated { self.sess.span_diagnostic.span_fatal_with_code( - self.mk_sp(start, suffix_start), + self.mk_sp(start, end), "unterminated character literal", error_code!(E0762), ) } - (token::Char, Mode::Char, 1, 1) // ' ' + self.cook_quoted(token::Char, Mode::Char, start, end, 1, 1) // ' ' } rustc_lexer::LiteralKind::Byte { terminated } => { if !terminated { self.sess.span_diagnostic.span_fatal_with_code( - self.mk_sp(start + BytePos(1), suffix_start), + self.mk_sp(start + BytePos(1), end), "unterminated byte constant", error_code!(E0763), ) } - (token::Byte, Mode::Byte, 2, 1) // b' ' + self.cook_quoted(token::Byte, Mode::Byte, start, end, 2, 1) // b' ' } rustc_lexer::LiteralKind::Str { terminated } => { if !terminated { self.sess.span_diagnostic.span_fatal_with_code( - self.mk_sp(start, suffix_start), + self.mk_sp(start, end), "unterminated double quote string", error_code!(E0765), ) } - (token::Str, Mode::Str, 1, 1) // " " + self.cook_quoted(token::Str, Mode::Str, start, end, 1, 1) // " " } rustc_lexer::LiteralKind::ByteStr { terminated } => { if !terminated { self.sess.span_diagnostic.span_fatal_with_code( - self.mk_sp(start + BytePos(1), suffix_start), + self.mk_sp(start + BytePos(1), end), "unterminated double quote byte string", error_code!(E0766), ) } - (token::ByteStr, Mode::ByteStr, 2, 1) // b" " + self.cook_quoted(token::ByteStr, Mode::ByteStr, start, end, 2, 1) // b" " } rustc_lexer::LiteralKind::RawStr { n_hashes } => { if let Some(n_hashes) = n_hashes { let n = u32::from(n_hashes); - (token::StrRaw(n_hashes), Mode::RawStr, 2 + n, 1 + n) // r##" "## + let kind = token::StrRaw(n_hashes); + self.cook_quoted(kind, Mode::RawStr, start, end, 2 + n, 1 + n) // r##" "## } else { self.report_raw_str_error(start, 1); } @@ -409,56 +409,59 @@ impl<'a> StringReader<'a> { rustc_lexer::LiteralKind::RawByteStr { n_hashes } => { if let Some(n_hashes) = n_hashes { let n = u32::from(n_hashes); - (token::ByteStrRaw(n_hashes), Mode::RawByteStr, 3 + n, 1 + n) // br##" "## + let kind = token::ByteStrRaw(n_hashes); + self.cook_quoted(kind, Mode::RawByteStr, start, end, 3 + n, 1 + n) // br##" "## } else { self.report_raw_str_error(start, 2); } } rustc_lexer::LiteralKind::Int { base, empty_int } => { - return if empty_int { + if empty_int { self.sess .span_diagnostic .struct_span_err_with_code( - self.mk_sp(start, suffix_start), + self.mk_sp(start, end), "no valid digits found for number", error_code!(E0768), ) .emit(); (token::Integer, sym::integer(0)) } else { - self.validate_int_literal(base, start, suffix_start); - (token::Integer, self.symbol_from_to(start, suffix_start)) - }; + if matches!(base, Base::Binary | Base::Octal) { + let base = base as u32; + let s = self.str_from_to(start + BytePos(2), end); + for (idx, c) in s.char_indices() { + if c != '_' && c.to_digit(base).is_none() { + self.err_span_( + start + BytePos::from_usize(2 + idx), + start + BytePos::from_usize(2 + idx + c.len_utf8()), + &format!("invalid digit for a base {} literal", base), + ); + } + } + } + (token::Integer, self.symbol_from_to(start, end)) + } } rustc_lexer::LiteralKind::Float { base, empty_exponent } => { if empty_exponent { self.err_span_(start, self.pos, "expected at least one digit in exponent"); } - match base { - Base::Hexadecimal => self.err_span_( - start, - suffix_start, - "hexadecimal float literal is not supported", - ), + Base::Hexadecimal => { + self.err_span_(start, end, "hexadecimal float literal is not supported") + } Base::Octal => { - self.err_span_(start, suffix_start, "octal float literal is not supported") + self.err_span_(start, end, "octal float literal is not supported") } Base::Binary => { - self.err_span_(start, suffix_start, "binary float literal is not supported") + self.err_span_(start, end, "binary float literal is not supported") } - _ => (), + _ => {} } - - let id = self.symbol_from_to(start, suffix_start); - return (token::Float, id); + (token::Float, self.symbol_from_to(start, end)) } - }; - let content_start = start + BytePos(prefix_len); - let content_end = suffix_start - BytePos(postfix_len); - let id = self.symbol_from_to(content_start, content_end); - self.validate_literal_escape(mode, content_start, content_end, prefix_len, postfix_len); - (lit_kind, id) + } } #[inline] @@ -649,20 +652,22 @@ impl<'a> StringReader<'a> { ) } - fn validate_literal_escape( + fn cook_quoted( &self, + kind: token::LitKind, mode: Mode, - content_start: BytePos, - content_end: BytePos, + start: BytePos, + end: BytePos, prefix_len: u32, postfix_len: u32, - ) { + ) -> (token::LitKind, Symbol) { + let content_start = start + BytePos(prefix_len); + let content_end = end - BytePos(postfix_len); let lit_content = self.str_from_to(content_start, content_end); unescape::unescape_literal(lit_content, mode, &mut |range, result| { // Here we only check for errors. The actual unescaping is done later. if let Err(err) = result { - let span_with_quotes = self - .mk_sp(content_start - BytePos(prefix_len), content_end + BytePos(postfix_len)); + let span_with_quotes = self.mk_sp(start, end); let (start, end) = (range.start as u32, range.end as u32); let lo = content_start + BytePos(start); let hi = lo + BytePos(end - start); @@ -678,23 +683,7 @@ impl<'a> StringReader<'a> { ); } }); - } - - fn validate_int_literal(&self, base: Base, content_start: BytePos, content_end: BytePos) { - let base = match base { - Base::Binary => 2, - Base::Octal => 8, - _ => return, - }; - let s = self.str_from_to(content_start + BytePos(2), content_end); - for (idx, c) in s.char_indices() { - let idx = idx as u32; - if c != '_' && c.to_digit(base).is_none() { - let lo = content_start + BytePos(2 + idx); - let hi = content_start + BytePos(2 + idx + c.len_utf8() as u32); - self.err_span_(lo, hi, &format!("invalid digit for a base {} literal", base)); - } - } + (kind, Symbol::intern(lit_content)) } } diff --git a/compiler/rustc_parse/src/lexer/unescape_error_reporting.rs b/compiler/rustc_parse/src/lexer/unescape_error_reporting.rs index f075de71426..6373f5b4fd6 100644 --- a/compiler/rustc_parse/src/lexer/unescape_error_reporting.rs +++ b/compiler/rustc_parse/src/lexer/unescape_error_reporting.rs @@ -108,7 +108,7 @@ pub(crate) fn emit_unescape_error( } if !has_help { - let (prefix, msg) = if mode.is_bytes() { + let (prefix, msg) = if mode.is_byte() { ("b", "if you meant to write a byte string literal, use double quotes") } else { ("", "if you meant to write a `str` literal, use double quotes") @@ -142,7 +142,7 @@ pub(crate) fn emit_unescape_error( EscapeError::EscapeOnlyChar => { let (c, char_span) = last_char(); - let msg = if mode.is_bytes() { + let msg = if mode.is_byte() { "byte constant must be escaped" } else { "character constant must be escaped" @@ -182,11 +182,11 @@ pub(crate) fn emit_unescape_error( let (c, span) = last_char(); let label = - if mode.is_bytes() { "unknown byte escape" } else { "unknown character escape" }; + if mode.is_byte() { "unknown byte escape" } else { "unknown character escape" }; let ec = escaped_char(c); let mut diag = handler.struct_span_err(span, &format!("{}: `{}`", label, ec)); diag.span_label(span, label); - if c == '{' || c == '}' && !mode.is_bytes() { + if c == '{' || c == '}' && !mode.is_byte() { diag.help( "if used in a formatting string, curly braces are escaped with `{{` and `}}`", ); @@ -196,7 +196,7 @@ pub(crate) fn emit_unescape_error( version control settings", ); } else { - if !mode.is_bytes() { + if !mode.is_byte() { diag.span_suggestion( span_with_quotes, "if you meant to write a literal backslash (perhaps escaping in a regular expression), consider a raw string literal", @@ -231,16 +231,23 @@ pub(crate) fn emit_unescape_error( .emit(); } EscapeError::NonAsciiCharInByte => { - assert!(mode.is_bytes()); let (c, span) = last_char(); - let mut err = handler.struct_span_err(span, "non-ASCII character in byte constant"); + let desc = match mode { + Mode::Byte => "byte literal", + Mode::ByteStr => "byte string literal", + Mode::RawByteStr => "raw byte string literal", + _ => panic!("non-is_byte literal paired with NonAsciiCharInByte"), + }; + let mut err = handler.struct_span_err(span, format!("non-ASCII character in {}", desc)); let postfix = if unicode_width::UnicodeWidthChar::width(c).unwrap_or(1) == 0 { format!(" but is {:?}", c) } else { String::new() }; - err.span_label(span, &format!("byte constant must be ASCII{}", postfix)); - if (c as u32) <= 0xFF { + err.span_label(span, &format!("must be ASCII{}", postfix)); + // Note: the \\xHH suggestions are not given for raw byte string + // literals, because they are araw and so cannot use any escapes. + if (c as u32) <= 0xFF && mode != Mode::RawByteStr { err.span_suggestion( span, &format!( @@ -250,9 +257,9 @@ pub(crate) fn emit_unescape_error( format!("\\x{:X}", c as u32), Applicability::MaybeIncorrect, ); - } else if matches!(mode, Mode::Byte) { + } else if mode == Mode::Byte { err.span_label(span, "this multibyte character does not fit into a single byte"); - } else if matches!(mode, Mode::ByteStr) { + } else if mode != Mode::RawByteStr { let mut utf8 = String::new(); utf8.push(c); err.span_suggestion( @@ -270,19 +277,6 @@ pub(crate) fn emit_unescape_error( } err.emit(); } - EscapeError::NonAsciiCharInByteString => { - assert!(mode.is_bytes()); - let (c, span) = last_char(); - let postfix = if unicode_width::UnicodeWidthChar::width(c).unwrap_or(1) == 0 { - format!(" but is {:?}", c) - } else { - String::new() - }; - handler - .struct_span_err(span, "raw byte string must be ASCII") - .span_label(span, &format!("must be ASCII{}", postfix)) - .emit(); - } EscapeError::OutOfRangeHexEscape => { handler .struct_span_err(span, "out of range hex escape") diff --git a/compiler/rustc_parse/src/parser/attr_wrapper.rs b/compiler/rustc_parse/src/parser/attr_wrapper.rs index 1b16ecb5ec2..c8160548763 100644 --- a/compiler/rustc_parse/src/parser/attr_wrapper.rs +++ b/compiler/rustc_parse/src/parser/attr_wrapper.rs @@ -5,7 +5,8 @@ use rustc_ast::tokenstream::{AttrTokenTree, DelimSpan, LazyAttrTokenStream, Spac use rustc_ast::{self as ast}; use rustc_ast::{AttrVec, Attribute, HasAttrs, HasTokens}; use rustc_errors::PResult; -use rustc_span::{sym, Span}; +use rustc_session::parse::ParseSess; +use rustc_span::{sym, Span, DUMMY_SP}; use std::convert::TryInto; use std::ops::Range; @@ -39,8 +40,13 @@ impl AttrWrapper { pub fn empty() -> AttrWrapper { AttrWrapper { attrs: AttrVec::new(), start_pos: usize::MAX } } - // FIXME: Delay span bug here? - pub(crate) fn take_for_recovery(self) -> AttrVec { + + pub(crate) fn take_for_recovery(self, sess: &ParseSess) -> AttrVec { + sess.span_diagnostic.delay_span_bug( + self.attrs.get(0).map(|attr| attr.span).unwrap_or(DUMMY_SP), + "AttrVec is taken for recovery but no error is produced", + ); + self.attrs } diff --git a/compiler/rustc_parse/src/parser/diagnostics.rs b/compiler/rustc_parse/src/parser/diagnostics.rs index 7dc4fd0044f..0bbe073fe2a 100644 --- a/compiler/rustc_parse/src/parser/diagnostics.rs +++ b/compiler/rustc_parse/src/parser/diagnostics.rs @@ -1653,15 +1653,29 @@ impl<'a> Parser<'a> { (token::CloseDelim(Delimiter::Parenthesis), Some(begin_par_sp)) => { self.bump(); + let sm = self.sess.source_map(); + let left = begin_par_sp; + let right = self.prev_token.span; + let left_snippet = if let Ok(snip) = sm.span_to_prev_source(left) && + !snip.ends_with(" ") { + " ".to_string() + } else { + "".to_string() + }; + + let right_snippet = if let Ok(snip) = sm.span_to_next_source(right) && + !snip.starts_with(" ") { + " ".to_string() + } else { + "".to_string() + }; + self.sess.emit_err(ParenthesesInForHead { - span: vec![begin_par_sp, self.prev_token.span], + span: vec![left, right], // With e.g. `for (x) in y)` this would replace `(x) in y)` // with `x) in y)` which is syntactically invalid. // However, this is prevented before we get here. - sugg: ParenthesesInForHeadSugg { - left: begin_par_sp, - right: self.prev_token.span, - }, + sugg: ParenthesesInForHeadSugg { left, right, left_snippet, right_snippet }, }); // Unwrap `(pat)` into `pat` to avoid the `unused_parens` lint. @@ -2468,11 +2482,15 @@ impl<'a> Parser<'a> { } pub(crate) fn maybe_recover_unexpected_block_label(&mut self) -> bool { - let Some(label) = self.eat_label().filter(|_| { - self.eat(&token::Colon) && self.token.kind == token::OpenDelim(Delimiter::Brace) - }) else { + // Check for `'a : {` + if !(self.check_lifetime() + && self.look_ahead(1, |tok| tok.kind == token::Colon) + && self.look_ahead(2, |tok| tok.kind == token::OpenDelim(Delimiter::Brace))) + { return false; - }; + } + let label = self.eat_label().expect("just checked if a label exists"); + self.bump(); // eat `:` let span = label.ident.span.to(self.prev_token.span); let mut err = self.struct_span_err(span, "block label not supported here"); err.span_label(span, "not supported here"); diff --git a/compiler/rustc_parse/src/parser/expr.rs b/compiler/rustc_parse/src/parser/expr.rs index 4a1162b9599..b072573af23 100644 --- a/compiler/rustc_parse/src/parser/expr.rs +++ b/compiler/rustc_parse/src/parser/expr.rs @@ -9,9 +9,9 @@ use crate::errors::{ ArrayBracketsInsteadOfSpaces, ArrayBracketsInsteadOfSpacesSugg, AsyncMoveOrderIncorrect, BinaryFloatLiteralNotSupported, BracesForStructLiteral, CatchAfterTry, CommaAfterBaseStruct, ComparisonInterpretedAsGeneric, ComparisonOrShiftInterpretedAsGenericSugg, - DoCatchSyntaxRemoved, DotDotDot, EqFieldInit, ExpectedElseBlock, ExpectedExpressionFoundLet, - FieldExpressionWithGeneric, FloatLiteralRequiresIntegerPart, FoundExprWouldBeStmt, - HexadecimalFloatLiteralNotSupported, IfExpressionMissingCondition, + DoCatchSyntaxRemoved, DotDotDot, EqFieldInit, ExpectedElseBlock, ExpectedEqForLetExpr, + ExpectedExpressionFoundLet, FieldExpressionWithGeneric, FloatLiteralRequiresIntegerPart, + FoundExprWouldBeStmt, HexadecimalFloatLiteralNotSupported, IfExpressionMissingCondition, IfExpressionMissingThenBlock, IfExpressionMissingThenBlockSub, IntLiteralTooLarge, InvalidBlockMacroSegment, InvalidComparisonOperator, InvalidComparisonOperatorSub, InvalidFloatLiteralSuffix, InvalidFloatLiteralWidth, InvalidIntLiteralWidth, @@ -33,6 +33,7 @@ use core::mem; use rustc_ast::ptr::P; use rustc_ast::token::{self, Delimiter, Token, TokenKind}; use rustc_ast::tokenstream::Spacing; +use rustc_ast::util::case::Case; use rustc_ast::util::classify; use rustc_ast::util::literal::LitError; use rustc_ast::util::parser::{prec_let_scrutinee_needs_par, AssocOp, Fixity}; @@ -833,16 +834,11 @@ impl<'a> Parser<'a> { ("cast", None) }; - // Save the memory location of expr before parsing any following postfix operators. - // This will be compared with the memory location of the output expression. - // If they different we can assume we parsed another expression because the existing expression is not reallocated. - let addr_before = &*cast_expr as *const _ as usize; let with_postfix = self.parse_dot_or_call_expr_with_(cast_expr, span)?; - let changed = addr_before != &*with_postfix as *const _ as usize; // Check if an illegal postfix operator has been added after the cast. - // If the resulting expression is not a cast, or has a different memory location, it is an illegal postfix operator. - if !matches!(with_postfix.kind, ExprKind::Cast(_, _) | ExprKind::Type(_, _)) || changed { + // If the resulting expression is not a cast, it is an illegal postfix operator. + if !matches!(with_postfix.kind, ExprKind::Cast(_, _) | ExprKind::Type(_, _)) { let msg = format!( "{cast_kind} cannot be followed by {}", match with_postfix.kind { @@ -2095,7 +2091,7 @@ impl<'a> Parser<'a> { if self.eat_keyword(kw::Static) { Movability::Static } else { Movability::Movable }; let asyncness = if self.token.uninterpolated_span().rust_2018() { - self.parse_asyncness() + self.parse_asyncness(Case::Sensitive) } else { Async::No }; @@ -2276,7 +2272,7 @@ impl<'a> Parser<'a> { self.mk_block_err(cond_span.shrink_to_hi()) } } else { - let attrs = self.parse_outer_attributes()?.take_for_recovery(); // For recovery. + let attrs = self.parse_outer_attributes()?; // For recovery. let block = if self.check(&token::OpenDelim(Delimiter::Brace)) { self.parse_block()? } else { @@ -2293,7 +2289,7 @@ impl<'a> Parser<'a> { })? } }; - self.error_on_if_block_attrs(lo, false, block.span, &attrs); + self.error_on_if_block_attrs(lo, false, block.span, attrs); block }; let els = if self.eat_keyword(kw::Else) { Some(self.parse_else_expr()?) } else { None }; @@ -2334,7 +2330,15 @@ impl<'a> Parser<'a> { RecoverColon::Yes, CommaRecoveryMode::LikelyTuple, )?; - self.expect(&token::Eq)?; + if self.token == token::EqEq { + self.sess.emit_err(ExpectedEqForLetExpr { + span: self.token.span, + sugg_span: self.token.span, + }); + self.bump(); + } else { + self.expect(&token::Eq)?; + } let expr = self.with_res(self.restrictions | Restrictions::NO_STRUCT_LITERAL, |this| { this.parse_assoc_expr_with(1 + prec_let_scrutinee_needs_par(), None.into()) })?; @@ -2346,7 +2350,7 @@ impl<'a> Parser<'a> { /// Parses an `else { ... }` expression (`else` token already eaten). fn parse_else_expr(&mut self) -> PResult<'a, P<Expr>> { let else_span = self.prev_token.span; // `else` - let attrs = self.parse_outer_attributes()?.take_for_recovery(); // For recovery. + let attrs = self.parse_outer_attributes()?; // For recovery. let expr = if self.eat_keyword(kw::If) { self.parse_if_expr()? } else if self.check(&TokenKind::OpenDelim(Delimiter::Brace)) { @@ -2381,7 +2385,7 @@ impl<'a> Parser<'a> { }, } }; - self.error_on_if_block_attrs(else_span, true, expr.span, &attrs); + self.error_on_if_block_attrs(else_span, true, expr.span, attrs); Ok(expr) } @@ -2390,8 +2394,13 @@ impl<'a> Parser<'a> { ctx_span: Span, is_ctx_else: bool, branch_span: Span, - attrs: &[ast::Attribute], + attrs: AttrWrapper, ) { + if attrs.is_empty() { + return; + } + + let attrs: &[ast::Attribute] = &attrs.take_for_recovery(self.sess); let (attributes, last) = match attrs { [] => return, [x0 @ xn] | [x0, .., xn] => (x0.span.to(xn.span), xn.span), diff --git a/compiler/rustc_parse/src/parser/item.rs b/compiler/rustc_parse/src/parser/item.rs index bda301c52e9..494f0cf56a8 100644 --- a/compiler/rustc_parse/src/parser/item.rs +++ b/compiler/rustc_parse/src/parser/item.rs @@ -8,6 +8,7 @@ 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::{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}; @@ -34,7 +35,7 @@ impl<'a> Parser<'a> { /// Parses a `mod <foo> { ... }` or `mod <foo>;` item. fn parse_item_mod(&mut self, attrs: &mut AttrVec) -> PResult<'a, ItemInfo> { - let unsafety = self.parse_unsafety(); + let unsafety = self.parse_unsafety(Case::Sensitive); self.expect_keyword(kw::Mod)?; let id = self.parse_ident()?; let mod_kind = if self.eat(&token::Semi) { @@ -143,8 +144,15 @@ impl<'a> Parser<'a> { let lo = self.token.span; let vis = self.parse_visibility(FollowedByType::No)?; let mut def = self.parse_defaultness(); - let kind = - self.parse_item_kind(&mut attrs, mac_allowed, lo, &vis, &mut def, fn_parse_mode)?; + let kind = self.parse_item_kind( + &mut attrs, + mac_allowed, + lo, + &vis, + &mut def, + fn_parse_mode, + Case::Sensitive, + )?; if let Some((ident, kind)) = kind { self.error_on_unconsumed_default(def, &kind); let span = lo.to(self.prev_token.span); @@ -205,16 +213,17 @@ impl<'a> Parser<'a> { vis: &Visibility, def: &mut Defaultness, fn_parse_mode: FnParseMode, + case: Case, ) -> PResult<'a, Option<ItemInfo>> { let def_final = def == &Defaultness::Final; - let mut def = || mem::replace(def, Defaultness::Final); + let mut def_ = || mem::replace(def, Defaultness::Final); - let info = if self.eat_keyword(kw::Use) { + let info = if self.eat_keyword_case(kw::Use, case) { self.parse_use_item()? - } else if self.check_fn_front_matter(def_final) { + } else if self.check_fn_front_matter(def_final, case) { // FUNCTION ITEM let (ident, sig, generics, body) = self.parse_fn(attrs, fn_parse_mode, lo, vis)?; - (ident, ItemKind::Fn(Box::new(Fn { defaultness: def(), sig, generics, body }))) + (ident, ItemKind::Fn(Box::new(Fn { defaultness: def_(), sig, generics, body }))) } else if self.eat_keyword(kw::Extern) { if self.eat_keyword(kw::Crate) { // EXTERN CRATE @@ -225,7 +234,7 @@ impl<'a> Parser<'a> { } } else if self.is_unsafe_foreign_mod() { // EXTERN BLOCK - let unsafety = self.parse_unsafety(); + let unsafety = self.parse_unsafety(Case::Sensitive); self.expect_keyword(kw::Extern)?; self.parse_item_foreign_mod(attrs, unsafety)? } else if self.is_static_global() { @@ -234,15 +243,15 @@ impl<'a> Parser<'a> { let m = self.parse_mutability(); let (ident, ty, expr) = self.parse_item_global(Some(m))?; (ident, ItemKind::Static(ty, m, expr)) - } else if let Const::Yes(const_span) = self.parse_constness() { + } else if let Const::Yes(const_span) = self.parse_constness(Case::Sensitive) { // CONST ITEM if self.token.is_keyword(kw::Impl) { // recover from `const impl`, suggest `impl const` - self.recover_const_impl(const_span, attrs, def())? + self.recover_const_impl(const_span, attrs, def_())? } else { self.recover_const_mut(const_span); let (ident, ty, expr) = self.parse_item_global(None)?; - (ident, ItemKind::Const(def(), ty, expr)) + (ident, ItemKind::Const(def_(), ty, expr)) } } else if self.check_keyword(kw::Trait) || self.check_auto_or_unsafe_trait_item() { // TRAIT ITEM @@ -251,7 +260,7 @@ impl<'a> Parser<'a> { || self.check_keyword(kw::Unsafe) && self.is_keyword_ahead(1, &[kw::Impl]) { // IMPL ITEM - self.parse_item_impl(attrs, def())? + self.parse_item_impl(attrs, def_())? } else if self.check_keyword(kw::Mod) || self.check_keyword(kw::Unsafe) && self.is_keyword_ahead(1, &[kw::Mod]) { @@ -259,7 +268,7 @@ impl<'a> Parser<'a> { self.parse_item_mod(attrs)? } else if self.eat_keyword(kw::Type) { // TYPE ITEM - self.parse_type_alias(def())? + self.parse_type_alias(def_())? } else if self.eat_keyword(kw::Enum) { // ENUM ITEM self.parse_item_enum()? @@ -286,6 +295,19 @@ impl<'a> Parser<'a> { } else if self.isnt_macro_invocation() && vis.kind.is_pub() { self.recover_missing_kw_before_item()?; return Ok(None); + } else if self.isnt_macro_invocation() && case == Case::Sensitive { + _ = def_; + + // Recover wrong cased keywords + return self.parse_item_kind( + attrs, + macros_allowed, + lo, + vis, + def, + fn_parse_mode, + Case::Insensitive, + ); } else if macros_allowed && self.check_path() { // MACRO INVOCATION ITEM (Ident::empty(), ItemKind::MacCall(P(self.parse_item_macro(vis)?))) @@ -538,7 +560,7 @@ impl<'a> Parser<'a> { attrs: &mut AttrVec, defaultness: Defaultness, ) -> PResult<'a, ItemInfo> { - let unsafety = self.parse_unsafety(); + let unsafety = self.parse_unsafety(Case::Sensitive); self.expect_keyword(kw::Impl)?; // First, parse generic parameters if necessary. @@ -552,7 +574,7 @@ impl<'a> Parser<'a> { generics }; - let constness = self.parse_constness(); + let constness = self.parse_constness(Case::Sensitive); if let Const::Yes(span) = constness { self.sess.gated_spans.gate(sym::const_trait_impl, span); } @@ -796,7 +818,7 @@ impl<'a> Parser<'a> { /// Parses `unsafe? auto? trait Foo { ... }` or `trait Foo = Bar;`. fn parse_item_trait(&mut self, attrs: &mut AttrVec, lo: Span) -> PResult<'a, ItemInfo> { - let unsafety = self.parse_unsafety(); + let unsafety = self.parse_unsafety(Case::Sensitive); // Parse optional `auto` prefix. let is_auto = if self.eat_keyword(kw::Auto) { IsAuto::Yes } else { IsAuto::No }; @@ -971,6 +993,23 @@ impl<'a> Parser<'a> { if self.eat(&token::ModSep) { self.parse_use_tree_glob_or_nested()? } else { + // Recover from using a colon as path separator. + while self.eat_noexpect(&token::Colon) { + self.struct_span_err(self.prev_token.span, "expected `::`, found `:`") + .span_suggestion_short( + self.prev_token.span, + "use double colon", + "::", + Applicability::MachineApplicable, + ) + .note_once("import paths are delimited using `::`") + .emit(); + + // We parse the rest of the path and append it to the original prefix. + self.parse_path_segments(&mut prefix.segments, PathStyle::Mod, None)?; + prefix.span = lo.to(self.prev_token.span); + } + UseTreeKind::Simple(self.parse_rename()?, DUMMY_NODE_ID, DUMMY_NODE_ID) } }; @@ -1745,7 +1784,7 @@ impl<'a> Parser<'a> { let (ident, is_raw) = self.ident_or_err()?; if !is_raw && ident.is_reserved() { let snapshot = self.create_snapshot_for_diagnostic(); - let err = if self.check_fn_front_matter(false) { + let err = if self.check_fn_front_matter(false, Case::Sensitive) { let inherited_vis = Visibility { span: rustc_span::DUMMY_SP, kind: VisibilityKind::Inherited, @@ -2155,7 +2194,7 @@ impl<'a> Parser<'a> { /// /// `check_pub` adds additional `pub` to the checks in case users place it /// wrongly, can be used to ensure `pub` never comes after `default`. - pub(super) fn check_fn_front_matter(&mut self, check_pub: bool) -> bool { + pub(super) fn check_fn_front_matter(&mut self, check_pub: bool, case: Case) -> bool { // We use an over-approximation here. // `const const`, `fn const` won't parse, but we're not stepping over other syntax either. // `pub` is added in case users got confused with the ordering like `async pub fn`, @@ -2165,23 +2204,30 @@ impl<'a> Parser<'a> { } else { &[kw::Const, kw::Async, kw::Unsafe, kw::Extern] }; - self.check_keyword(kw::Fn) // Definitely an `fn`. + self.check_keyword_case(kw::Fn, case) // Definitely an `fn`. // `$qual fn` or `$qual $qual`: - || quals.iter().any(|&kw| self.check_keyword(kw)) + || quals.iter().any(|&kw| self.check_keyword_case(kw, case)) && self.look_ahead(1, |t| { // `$qual fn`, e.g. `const fn` or `async fn`. - t.is_keyword(kw::Fn) + t.is_keyword_case(kw::Fn, case) // Two qualifiers `$qual $qual` is enough, e.g. `async unsafe`. - || t.is_non_raw_ident_where(|i| quals.contains(&i.name) - // Rule out 2015 `const async: T = val`. - && i.is_reserved() + || ( + ( + t.is_non_raw_ident_where(|i| + quals.contains(&i.name) + // Rule out 2015 `const async: T = val`. + && i.is_reserved() + ) + || case == Case::Insensitive + && t.is_non_raw_ident_where(|i| quals.iter().any(|qual| qual.as_str() == i.name.as_str().to_lowercase())) + ) // Rule out unsafe extern block. && !self.is_unsafe_foreign_mod()) }) // `extern ABI fn` - || self.check_keyword(kw::Extern) + || self.check_keyword_case(kw::Extern, case) && self.look_ahead(1, |t| t.can_begin_literal_maybe_minus()) - && self.look_ahead(2, |t| t.is_keyword(kw::Fn)) + && self.look_ahead(2, |t| t.is_keyword_case(kw::Fn, case)) } /// Parses all the "front matter" (or "qualifiers") for a `fn` declaration, @@ -2197,22 +2243,22 @@ impl<'a> Parser<'a> { /// `Visibility::Inherited` when no visibility is known. pub(super) fn parse_fn_front_matter(&mut self, orig_vis: &Visibility) -> PResult<'a, FnHeader> { let sp_start = self.token.span; - let constness = self.parse_constness(); + let constness = self.parse_constness(Case::Insensitive); let async_start_sp = self.token.span; - let asyncness = self.parse_asyncness(); + let asyncness = self.parse_asyncness(Case::Insensitive); let unsafe_start_sp = self.token.span; - let unsafety = self.parse_unsafety(); + let unsafety = self.parse_unsafety(Case::Insensitive); let ext_start_sp = self.token.span; - let ext = self.parse_extern(); + let ext = self.parse_extern(Case::Insensitive); if let Async::Yes { span, .. } = asyncness { self.ban_async_in_2015(span); } - if !self.eat_keyword(kw::Fn) { + if !self.eat_keyword_case(kw::Fn, Case::Insensitive) { // It is possible for `expect_one_of` to recover given the contents of // `self.expected_tokens`, therefore, do not use `self.unexpected()` which doesn't // account for this. diff --git a/compiler/rustc_parse/src/parser/mod.rs b/compiler/rustc_parse/src/parser/mod.rs index 2e59c005e31..14dc490fb02 100644 --- a/compiler/rustc_parse/src/parser/mod.rs +++ b/compiler/rustc_parse/src/parser/mod.rs @@ -22,6 +22,7 @@ use rustc_ast::token::{self, Delimiter, Nonterminal, Token, TokenKind}; use rustc_ast::tokenstream::AttributesData; use rustc_ast::tokenstream::{self, DelimSpan, Spacing}; use rustc_ast::tokenstream::{TokenStream, TokenTree}; +use rustc_ast::util::case::Case; use rustc_ast::AttrId; use rustc_ast::DUMMY_NODE_ID; use rustc_ast::{self as ast, AnonConst, AttrStyle, AttrVec, Const, Extern}; @@ -636,6 +637,20 @@ impl<'a> Parser<'a> { self.token.is_keyword(kw) } + fn check_keyword_case(&mut self, kw: Symbol, case: Case) -> bool { + if self.check_keyword(kw) { + return true; + } + + if case == Case::Insensitive + && let Some((ident, /* is_raw */ false)) = self.token.ident() + && ident.as_str().to_lowercase() == kw.as_str().to_lowercase() { + true + } else { + false + } + } + /// If the next token is the given keyword, eats it and returns `true`. /// Otherwise, returns `false`. An expectation is also added for diagnostics purposes. // Public for rustfmt usage. @@ -648,6 +663,33 @@ impl<'a> Parser<'a> { } } + /// Eats a keyword, optionally ignoring the case. + /// If the case differs (and is ignored) an error is issued. + /// This is useful for recovery. + fn eat_keyword_case(&mut self, kw: Symbol, case: Case) -> bool { + if self.eat_keyword(kw) { + return true; + } + + if case == Case::Insensitive + && let Some((ident, /* is_raw */ false)) = self.token.ident() + && ident.as_str().to_lowercase() == kw.as_str().to_lowercase() { + self + .struct_span_err(ident.span, format!("keyword `{kw}` is written in a wrong case")) + .span_suggestion( + ident.span, + "write it in the correct case", + kw, + Applicability::MachineApplicable + ).emit(); + + self.bump(); + return true; + } + + false + } + fn eat_keyword_noexpect(&mut self, kw: Symbol) -> bool { if self.token.is_keyword(kw) { self.bump(); @@ -1127,8 +1169,8 @@ impl<'a> Parser<'a> { } /// Parses asyncness: `async` or nothing. - fn parse_asyncness(&mut self) -> Async { - if self.eat_keyword(kw::Async) { + fn parse_asyncness(&mut self, case: Case) -> Async { + if self.eat_keyword_case(kw::Async, case) { let span = self.prev_token.uninterpolated_span(); Async::Yes { span, closure_id: DUMMY_NODE_ID, return_impl_trait_id: DUMMY_NODE_ID } } else { @@ -1137,8 +1179,8 @@ impl<'a> Parser<'a> { } /// Parses unsafety: `unsafe` or nothing. - fn parse_unsafety(&mut self) -> Unsafe { - if self.eat_keyword(kw::Unsafe) { + fn parse_unsafety(&mut self, case: Case) -> Unsafe { + if self.eat_keyword_case(kw::Unsafe, case) { Unsafe::Yes(self.prev_token.uninterpolated_span()) } else { Unsafe::No @@ -1146,10 +1188,10 @@ impl<'a> Parser<'a> { } /// Parses constness: `const` or nothing. - fn parse_constness(&mut self) -> Const { + fn parse_constness(&mut self, case: Case) -> Const { // Avoid const blocks to be parsed as const items if self.look_ahead(1, |t| t != &token::OpenDelim(Delimiter::Brace)) - && self.eat_keyword(kw::Const) + && self.eat_keyword_case(kw::Const, case) { Const::Yes(self.prev_token.uninterpolated_span()) } else { @@ -1404,8 +1446,8 @@ impl<'a> Parser<'a> { } /// Parses `extern string_literal?`. - fn parse_extern(&mut self) -> Extern { - if self.eat_keyword(kw::Extern) { + fn parse_extern(&mut self, case: Case) -> Extern { + if self.eat_keyword_case(kw::Extern, case) { let mut extern_span = self.prev_token.span; let abi = self.parse_abi(); if let Some(abi) = abi { diff --git a/compiler/rustc_parse/src/parser/path.rs b/compiler/rustc_parse/src/parser/path.rs index fdc1af27f82..d46565dea89 100644 --- a/compiler/rustc_parse/src/parser/path.rs +++ b/compiler/rustc_parse/src/parser/path.rs @@ -631,7 +631,9 @@ impl<'a> Parser<'a> { /// - A single-segment path. pub(super) fn expr_is_valid_const_arg(&self, expr: &P<rustc_ast::Expr>) -> bool { match &expr.kind { - ast::ExprKind::Block(_, _) | ast::ExprKind::Lit(_) => true, + ast::ExprKind::Block(_, _) + | ast::ExprKind::Lit(_) + | ast::ExprKind::IncludedBytes(..) => true, ast::ExprKind::Unary(ast::UnOp::Neg, expr) => { matches!(expr.kind, ast::ExprKind::Lit(_)) } diff --git a/compiler/rustc_parse/src/parser/stmt.rs b/compiler/rustc_parse/src/parser/stmt.rs index 12753c6785c..9684145ad99 100644 --- a/compiler/rustc_parse/src/parser/stmt.rs +++ b/compiler/rustc_parse/src/parser/stmt.rs @@ -19,7 +19,7 @@ use rustc_ast as ast; use rustc_ast::ptr::P; use rustc_ast::token::{self, Delimiter, TokenKind}; use rustc_ast::util::classify; -use rustc_ast::{AttrStyle, AttrVec, Attribute, LocalKind, MacCall, MacCallStmt, MacStmtStyle}; +use rustc_ast::{AttrStyle, AttrVec, LocalKind, MacCall, MacCallStmt, MacStmtStyle}; use rustc_ast::{Block, BlockCheckMode, Expr, ExprKind, HasAttrs, Local, Stmt}; use rustc_ast::{StmtKind, DUMMY_NODE_ID}; use rustc_errors::{Applicability, DiagnosticBuilder, ErrorGuaranteed, PResult}; @@ -101,7 +101,7 @@ impl<'a> Parser<'a> { 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. - self.error_outer_attrs(&attrs.take_for_recovery()); + self.error_outer_attrs(attrs); self.mk_stmt(lo, StmtKind::Empty) } else if self.token != token::CloseDelim(Delimiter::Brace) { // Remainder are line-expr stmts. @@ -120,7 +120,7 @@ impl<'a> Parser<'a> { } self.mk_stmt(lo.to(e.span), StmtKind::Expr(e)) } else { - self.error_outer_attrs(&attrs.take_for_recovery()); + self.error_outer_attrs(attrs); return Ok(None); })) } @@ -199,8 +199,10 @@ impl<'a> Parser<'a> { /// Error on outer attributes in this context. /// Also error if the previous token was a doc comment. - fn error_outer_attrs(&self, attrs: &[Attribute]) { - if let [.., last] = attrs { + fn error_outer_attrs(&self, attrs: AttrWrapper) { + if !attrs.is_empty() + && let attrs = attrs.take_for_recovery(self.sess) + && let attrs @ [.., last] = &*attrs { if last.is_doc_comment() { self.sess.emit_err(DocCommentDoesNotDocumentAnything { span: last.span, diff --git a/compiler/rustc_parse/src/parser/ty.rs b/compiler/rustc_parse/src/parser/ty.rs index 2a8512acf8c..4d78c5bd0e2 100644 --- a/compiler/rustc_parse/src/parser/ty.rs +++ b/compiler/rustc_parse/src/parser/ty.rs @@ -4,6 +4,7 @@ use crate::{maybe_recover_from_interpolated_ty_qpath, maybe_whole}; use rustc_ast::ptr::P; use rustc_ast::token::{self, Delimiter, Token, TokenKind}; +use rustc_ast::util::case::Case; use rustc_ast::{ self as ast, BareFnTy, FnRetTy, GenericBound, GenericBounds, GenericParam, Generics, Lifetime, MacCall, MutTy, Mutability, PolyTraitRef, TraitBoundModifier, TraitObjectSyntax, Ty, TyKind, @@ -267,7 +268,7 @@ impl<'a> Parser<'a> { } else if self.eat_keyword(kw::Underscore) { // A type to be inferred `_` TyKind::Infer - } else if self.check_fn_front_matter(false) { + } else if self.check_fn_front_matter(false, Case::Sensitive) { // Function pointer type self.parse_ty_bare_fn(lo, Vec::new(), recover_return_sign)? } else if self.check_keyword(kw::For) { @@ -275,7 +276,7 @@ impl<'a> Parser<'a> { // `for<'lt> [unsafe] [extern "ABI"] fn (&'lt S) -> T` // `for<'lt> Trait1<'lt> + Trait2 + 'a` let lifetime_defs = self.parse_late_bound_lifetime_defs()?; - if self.check_fn_front_matter(false) { + if self.check_fn_front_matter(false, Case::Sensitive) { self.parse_ty_bare_fn(lo, lifetime_defs, recover_return_sign)? } else { let path = self.parse_path(PathStyle::Type)?; @@ -401,7 +402,7 @@ impl<'a> Parser<'a> { .span_suggestions( span.shrink_to_hi(), "add `mut` or `const` here", - ["mut ".to_string(), "const ".to_string()].into_iter(), + ["mut ".to_string(), "const ".to_string()], Applicability::HasPlaceholders, ) .emit(); |