diff options
Diffstat (limited to 'compiler/rustc_parse/src')
| -rw-r--r-- | compiler/rustc_parse/src/errors.rs | 91 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/lexer/mod.rs | 119 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/lexer/unescape_error_reporting.rs | 42 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/attr_wrapper.rs | 13 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/diagnostics.rs | 24 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/expr.rs | 77 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/item.rs | 108 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/mod.rs | 90 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/stmt.rs | 65 | ||||
| -rw-r--r-- | compiler/rustc_parse/src/parser/ty.rs | 7 |
10 files changed, 414 insertions, 222 deletions
diff --git a/compiler/rustc_parse/src/errors.rs b/compiler/rustc_parse/src/errors.rs index a3fbc06293f..d59982f7063 100644 --- a/compiler/rustc_parse/src/errors.rs +++ b/compiler/rustc_parse/src/errors.rs @@ -64,7 +64,7 @@ pub(crate) struct BadQPathStage2 { #[diag(parser_incorrect_semicolon)] pub(crate) struct IncorrectSemicolon<'a> { #[primary_span] - #[suggestion_short(code = "", applicability = "machine-applicable")] + #[suggestion(style = "short", code = "", applicability = "machine-applicable")] pub span: Span, #[help] pub opt_help: Option<()>, @@ -136,7 +136,12 @@ pub(crate) struct InvalidComparisonOperator { #[derive(Subdiagnostic)] pub(crate) enum InvalidComparisonOperatorSub { - #[suggestion_short(use_instead, applicability = "machine-applicable", code = "{correct}")] + #[suggestion( + use_instead, + style = "short", + applicability = "machine-applicable", + code = "{correct}" + )] Correctable { #[primary_span] span: Span, @@ -160,14 +165,16 @@ pub(crate) struct InvalidLogicalOperator { #[derive(Subdiagnostic)] pub(crate) enum InvalidLogicalOperatorSub { - #[suggestion_short( + #[suggestion( use_amp_amp_for_conjunction, + style = "short", applicability = "machine-applicable", code = "&&" )] Conjunction(#[primary_span] Span), - #[suggestion_short( + #[suggestion( use_pipe_pipe_for_disjunction, + style = "short", applicability = "machine-applicable", code = "||" )] @@ -178,7 +185,7 @@ pub(crate) enum InvalidLogicalOperatorSub { #[diag(parser_tilde_is_not_unary_operator)] pub(crate) struct TildeAsUnaryOperator( #[primary_span] - #[suggestion_short(applicability = "machine-applicable", code = "!")] + #[suggestion(style = "short", applicability = "machine-applicable", code = "!")] pub Span, ); @@ -194,22 +201,25 @@ pub(crate) struct NotAsNegationOperator { #[derive(Subdiagnostic)] pub enum NotAsNegationOperatorSub { - #[suggestion_short( + #[suggestion( parser_unexpected_token_after_not_default, + style = "short", applicability = "machine-applicable", code = "!" )] SuggestNotDefault(#[primary_span] Span), - #[suggestion_short( + #[suggestion( parser_unexpected_token_after_not_bitwise, + style = "short", applicability = "machine-applicable", code = "!" )] SuggestNotBitwise(#[primary_span] Span), - #[suggestion_short( + #[suggestion( parser_unexpected_token_after_not_logical, + style = "short", applicability = "machine-applicable", code = "!" )] @@ -249,7 +259,7 @@ pub(crate) struct UnexpectedTokenAfterLabel { #[primary_span] #[label(parser_unexpected_token_after_label)] pub span: Span, - #[suggestion_verbose(suggestion_remove_label, code = "")] + #[suggestion(suggestion_remove_label, style = "verbose", code = "")] pub remove_label: Option<Span>, #[subdiagnostic] pub enclose_in_block: Option<UnexpectedTokenAfterLabelSugg>, @@ -272,7 +282,7 @@ pub(crate) struct RequireColonAfterLabeledExpression { pub span: Span, #[label] pub label: Span, - #[suggestion_short(applicability = "machine-applicable", code = ": ")] + #[suggestion(style = "short", applicability = "machine-applicable", code = ": ")] pub label_end: Span, } @@ -354,11 +364,20 @@ pub(crate) struct IntLiteralTooLarge { pub(crate) struct MissingSemicolonBeforeArray { #[primary_span] pub open_delim: Span, - #[suggestion_verbose(applicability = "maybe-incorrect", code = ";")] + #[suggestion(style = "verbose", applicability = "maybe-incorrect", code = ";")] pub semicolon: Span, } #[derive(Diagnostic)] +#[diag(parser_expect_dotdot_not_dotdotdot)] +pub(crate) struct MissingDotDot { + #[primary_span] + pub token_span: Span, + #[suggestion(applicability = "maybe-incorrect", code = "..", style = "verbose")] + pub sugg_span: Span, +} + +#[derive(Diagnostic)] #[diag(parser_invalid_block_macro_segment)] pub(crate) struct InvalidBlockMacroSegment { #[primary_span] @@ -402,6 +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] @@ -442,9 +470,9 @@ pub(crate) struct MissingInInForLoop { #[derive(Subdiagnostic)] pub(crate) enum MissingInInForLoopSub { // Has been misleading, at least in the past (closed Issue #48492), thus maybe-incorrect - #[suggestion_short(use_in_not_of, applicability = "maybe-incorrect", code = "in")] + #[suggestion(use_in_not_of, style = "short", applicability = "maybe-incorrect", code = "in")] InNotOf(#[primary_span] Span), - #[suggestion_short(add_in, applicability = "maybe-incorrect", code = " in ")] + #[suggestion(add_in, style = "short", applicability = "maybe-incorrect", code = " in ")] AddIn(#[primary_span] Span), } @@ -470,7 +498,7 @@ pub(crate) struct CatchAfterTry { pub(crate) struct CommaAfterBaseStruct { #[primary_span] pub span: Span, - #[suggestion_short(applicability = "machine-applicable", code = "")] + #[suggestion(style = "short", applicability = "machine-applicable", code = "")] pub comma: Span, } @@ -512,7 +540,7 @@ pub(crate) struct RemoveLet { #[diag(parser_use_eq_instead)] pub(crate) struct UseEqInstead { #[primary_span] - #[suggestion_short(applicability = "machine-applicable", code = "=")] + #[suggestion(style = "short", applicability = "machine-applicable", code = "=")] pub span: Span, } @@ -520,7 +548,7 @@ pub(crate) struct UseEqInstead { #[diag(parser_use_empty_block_not_semi)] pub(crate) struct UseEmptyBlockNotSemi { #[primary_span] - #[suggestion_hidden(applicability = "machine-applicable", code = "{{}}")] + #[suggestion(style = "hidden", applicability = "machine-applicable", code = "{{}}")] pub span: Span, } @@ -576,7 +604,12 @@ pub(crate) struct LeadingPlusNotSupported { #[primary_span] #[label] pub span: Span, - #[suggestion_verbose(suggestion_remove_plus, code = "", applicability = "machine-applicable")] + #[suggestion( + suggestion_remove_plus, + style = "verbose", + code = "", + applicability = "machine-applicable" + )] pub remove_plus: Option<Span>, #[subdiagnostic] pub add_parentheses: Option<ExprParenthesesNeeded>, @@ -843,7 +876,7 @@ pub(crate) struct InvalidCurlyInLetElse { #[help] pub(crate) struct CompoundAssignmentExpressionInLet { #[primary_span] - #[suggestion_short(code = "=", applicability = "maybe-incorrect")] + #[suggestion(style = "short", code = "=", applicability = "maybe-incorrect")] pub span: Span, } @@ -864,8 +897,9 @@ pub(crate) struct InvalidMetaItem { } #[derive(Subdiagnostic)] -#[suggestion_verbose( +#[suggestion( parser_sugg_escape_to_use_as_identifier, + style = "verbose", applicability = "maybe-incorrect", code = "r#" )] @@ -918,6 +952,7 @@ pub(crate) struct ExpectedIdentifier { } impl<'a, G: EmissionGuarantee> IntoDiagnostic<'a, G> for ExpectedIdentifier { + #[track_caller] fn into_diagnostic( self, handler: &'a rustc_errors::Handler, @@ -963,6 +998,7 @@ pub(crate) struct ExpectedSemi { } impl<'a, G: EmissionGuarantee> IntoDiagnostic<'a, G> for ExpectedSemi { + #[track_caller] fn into_diagnostic( self, handler: &'a rustc_errors::Handler, @@ -1003,7 +1039,12 @@ pub(crate) enum ExpectedSemiSugg { applicability = "machine-applicable" )] ChangeToSemi(#[primary_span] Span), - #[suggestion_short(parser_sugg_add_semi, code = ";", applicability = "machine-applicable")] + #[suggestion( + parser_sugg_add_semi, + style = "short", + code = ";", + applicability = "machine-applicable" + )] AddSemi(#[primary_span] Span), } @@ -1057,8 +1098,9 @@ pub(crate) struct GenericParamsWithoutAngleBracketsSugg { pub(crate) struct ComparisonOperatorsCannotBeChained { #[primary_span] pub span: Vec<Span>, - #[suggestion_verbose( + #[suggestion( parser_sugg_turbofish_syntax, + style = "verbose", code = "::", applicability = "maybe-incorrect" )] @@ -1072,8 +1114,9 @@ pub(crate) struct ComparisonOperatorsCannotBeChained { #[derive(Subdiagnostic)] pub(crate) enum ComparisonOperatorsCannotBeChainedSugg { - #[suggestion_verbose( + #[suggestion( sugg_split_comparison, + style = "verbose", code = " && {middle_term}", applicability = "maybe-incorrect" )] @@ -1217,7 +1260,7 @@ pub(crate) enum UnexpectedConstParamDeclarationSugg { pub(crate) struct UnexpectedConstInGenericParam { #[primary_span] pub span: Span, - #[suggestion_verbose(code = "", applicability = "maybe-incorrect")] + #[suggestion(style = "verbose", code = "", applicability = "maybe-incorrect")] pub to_remove: Option<Span>, } @@ -1225,7 +1268,7 @@ pub(crate) struct UnexpectedConstInGenericParam { #[diag(parser_async_move_order_incorrect)] pub(crate) struct AsyncMoveOrderIncorrect { #[primary_span] - #[suggestion_verbose(code = "async move", applicability = "maybe-incorrect")] + #[suggestion(style = "verbose", code = "async move", applicability = "maybe-incorrect")] pub span: Span, } diff --git a/compiler/rustc_parse/src/lexer/mod.rs b/compiler/rustc_parse/src/lexer/mod.rs index 462bce16ad7..645262bd2f1 100644 --- a/compiler/rustc_parse/src/lexer/mod.rs +++ b/compiler/rustc_parse/src/lexer/mod.rs @@ -175,20 +175,10 @@ impl<'a> StringReader<'a> { if string == "_" { self.sess .span_diagnostic - .struct_span_warn( + .struct_span_err( self.mk_sp(suffix_start, self.pos), "underscore literal suffix is not allowed", ) - .warn( - "this was previously accepted by the compiler but is \ - being phased out; it will become a hard error in \ - a future release!", - ) - .note( - "see issue #42326 \ - <https://github.com/rust-lang/rust/issues/42326> \ - for more information", - ) .emit(); None } else { @@ -363,55 +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); } @@ -419,56 +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] @@ -659,20 +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); @@ -688,23 +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 81c051b8f35..1b16ecb5ec2 100644 --- a/compiler/rustc_parse/src/parser/attr_wrapper.rs +++ b/compiler/rustc_parse/src/parser/attr_wrapper.rs @@ -273,16 +273,23 @@ impl<'a> Parser<'a> { let cursor_snapshot_next_calls = cursor_snapshot.num_next_calls; let mut end_pos = self.token_cursor.num_next_calls; + let mut captured_trailing = false; + // Capture a trailing token if requested by the callback 'f' match trailing { TrailingToken::None => {} + TrailingToken::Gt => { + assert_eq!(self.token.kind, token::Gt); + } TrailingToken::Semi => { assert_eq!(self.token.kind, token::Semi); end_pos += 1; + captured_trailing = true; } TrailingToken::MaybeComma => { if self.token.kind == token::Comma { end_pos += 1; + captured_trailing = true; } } } @@ -292,11 +299,7 @@ impl<'a> Parser<'a> { // was not actually bumped past it. When the `LazyAttrTokenStream` gets converted // into an `AttrTokenStream`, we will create the proper token. if self.token_cursor.break_last_token { - assert_eq!( - trailing, - TrailingToken::None, - "Cannot set `break_last_token` and have trailing token" - ); + assert!(!captured_trailing, "Cannot set break_last_token and have trailing token"); end_pos += 1; } diff --git a/compiler/rustc_parse/src/parser/diagnostics.rs b/compiler/rustc_parse/src/parser/diagnostics.rs index e9549e89998..0bbe073fe2a 100644 --- a/compiler/rustc_parse/src/parser/diagnostics.rs +++ b/compiler/rustc_parse/src/parser/diagnostics.rs @@ -769,6 +769,10 @@ impl<'a> Parser<'a> { segment: &PathSegment, end: &[&TokenKind], ) -> bool { + if !self.may_recover() { + return false; + } + // This function is intended to be invoked after parsing a path segment where there are two // cases: // @@ -863,6 +867,10 @@ impl<'a> Parser<'a> { /// Check if a method call with an intended turbofish has been written without surrounding /// angle brackets. pub(super) fn check_turbofish_missing_angle_brackets(&mut self, segment: &mut PathSegment) { + if !self.may_recover() { + return; + } + if token::ModSep == self.token.kind && segment.args.is_none() { let snapshot = self.create_snapshot_for_diagnostic(); self.bump(); @@ -1396,6 +1404,10 @@ impl<'a> Parser<'a> { &mut self, base: P<T>, ) -> PResult<'a, P<T>> { + if !self.may_recover() { + return Ok(base); + } + // Do not add `::` to expected tokens. if self.token == token::ModSep { if let Some(ty) = base.to_ty() { @@ -2470,11 +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 98520a446a6..7355730c9eb 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, @@ -20,9 +20,9 @@ use crate::errors::{ InvalidNumLiteralSuffix, LabeledLoopInBreak, LeadingPlusNotSupported, LeftArrowOperator, LifetimeInBorrowExpression, MacroInvocationWithQualifiedPath, MalformedLoopLabel, MatchArmBodyWithoutBraces, MatchArmBodyWithoutBracesSugg, MissingCommaAfterMatchArm, - MissingInInForLoop, MissingInInForLoopSub, MissingSemicolonBeforeArray, NoFieldsForFnCall, - NotAsNegationOperator, NotAsNegationOperatorSub, OctalFloatLiteralNotSupported, - OuterAttributeNotAllowedOnIfElse, ParenthesesWithStructFields, + MissingDotDot, MissingInInForLoop, MissingInInForLoopSub, MissingSemicolonBeforeArray, + NoFieldsForFnCall, NotAsNegationOperator, NotAsNegationOperatorSub, + OctalFloatLiteralNotSupported, OuterAttributeNotAllowedOnIfElse, ParenthesesWithStructFields, RequireColonAfterLabeledExpression, ShiftInterpretedAsGeneric, StructLiteralNotAllowedHere, StructLiteralNotAllowedHereSugg, TildeAsUnaryOperator, UnexpectedTokenAfterLabel, UnexpectedTokenAfterLabelSugg, WrapExpressionInParentheses, @@ -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}; @@ -132,7 +133,7 @@ impl<'a> Parser<'a> { Ok(expr) => Ok(expr), Err(mut err) => match self.token.ident() { Some((Ident { name: kw::Underscore, .. }, false)) - if self.look_ahead(1, |t| t == &token::Comma) => + if self.may_recover() && self.look_ahead(1, |t| t == &token::Comma) => { // Special-case handling of `foo(_, _, _)` err.emit(); @@ -456,7 +457,7 @@ impl<'a> Parser<'a> { return None; } (Some(op), _) => (op, self.token.span), - (None, Some((Ident { name: sym::and, span }, false))) => { + (None, Some((Ident { name: sym::and, span }, false))) if self.may_recover() => { self.sess.emit_err(InvalidLogicalOperator { span: self.token.span, incorrect: "and".into(), @@ -464,7 +465,7 @@ impl<'a> Parser<'a> { }); (AssocOp::LAnd, span) } - (None, Some((Ident { name: sym::or, span }, false))) => { + (None, Some((Ident { name: sym::or, span }, false))) if self.may_recover() => { self.sess.emit_err(InvalidLogicalOperator { span: self.token.span, incorrect: "or".into(), @@ -615,7 +616,7 @@ impl<'a> Parser<'a> { token::Ident(..) if this.token.is_keyword(kw::Box) => { make_it!(this, attrs, |this, _| this.parse_box_expr(lo)) } - token::Ident(..) if this.is_mistaken_not_ident_negation() => { + token::Ident(..) if this.may_recover() && this.is_mistaken_not_ident_negation() => { make_it!(this, attrs, |this, _| this.recover_not_expr(lo)) } _ => return this.parse_dot_or_call_expr(Some(attrs)), @@ -718,6 +719,10 @@ impl<'a> Parser<'a> { let cast_expr = match self.parse_as_cast_ty() { Ok(rhs) => mk_expr(self, lhs, rhs), Err(type_err) => { + if !self.may_recover() { + return Err(type_err); + } + // Rewind to before attempting to parse the type with generics, to recover // from situations like `x as usize < y` in which we first tried to parse // `usize < y` as a type with generic arguments. @@ -829,16 +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 { @@ -1197,6 +1197,10 @@ impl<'a> Parser<'a> { seq: &mut PResult<'a, P<Expr>>, snapshot: Option<(SnapshotParser<'a>, ExprKind)>, ) -> Option<P<Expr>> { + if !self.may_recover() { + return None; + } + match (seq.as_mut(), snapshot) { (Err(err), Some((mut snapshot, ExprKind::Path(None, path)))) => { snapshot.bump(); // `(` @@ -1360,7 +1364,7 @@ impl<'a> Parser<'a> { ) } else if self.check_inline_const(0) { self.parse_const_block(lo.to(self.token.span), false) - } else if self.is_do_catch_block() { + } else if self.may_recover() && self.is_do_catch_block() { self.recover_do_catch() } else if self.is_try_block() { self.expect_keyword(kw::Try)?; @@ -1532,6 +1536,7 @@ impl<'a> Parser<'a> { { self.parse_block_expr(label, lo, BlockCheckMode::Default) } else if !ate_colon + && self.may_recover() && (matches!(self.token.kind, token::CloseDelim(_) | token::Comma) || self.token.is_op()) { @@ -1999,6 +2004,10 @@ impl<'a> Parser<'a> { prev_span: Span, open_delim_span: Span, ) -> PResult<'a, ()> { + if !self.may_recover() { + return Ok(()); + } + if self.token.kind == token::Comma { if !self.sess.source_map().is_multiline(prev_span.until(self.token.span)) { return Ok(()); @@ -2039,7 +2048,7 @@ impl<'a> Parser<'a> { lo: Span, blk_mode: BlockCheckMode, ) -> PResult<'a, P<Expr>> { - if self.is_array_like_block() { + if self.may_recover() && self.is_array_like_block() { if let Some(arr) = self.maybe_suggest_brackets_instead_of_braces(lo) { return Ok(arr); } @@ -2082,7 +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 }; @@ -2112,6 +2121,8 @@ impl<'a> Parser<'a> { // HACK: This is needed so we can detect whether we're inside a macro, // where regular assumptions about what tokens can follow other tokens // don't necessarily apply. + && self.may_recover() + // FIXME(Nilstrieb): Remove this check once `may_recover` actually stops recovery && self.subparser_name.is_none() { // It is likely that the closure body is a block but where the @@ -2319,7 +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()) })?; @@ -2865,7 +2884,7 @@ impl<'a> Parser<'a> { }; while self.token != token::CloseDelim(close_delim) { - if self.eat(&token::DotDot) { + if self.eat(&token::DotDot) || self.recover_struct_field_dots(close_delim) { let exp_span = self.prev_token.span; // We permit `.. }` on the left-hand side of a destructuring assignment. if self.check(&token::CloseDelim(close_delim)) { @@ -3012,6 +3031,18 @@ impl<'a> Parser<'a> { self.recover_stmt(); } + fn recover_struct_field_dots(&mut self, close_delim: Delimiter) -> bool { + if !self.look_ahead(1, |t| *t == token::CloseDelim(close_delim)) + && self.eat(&token::DotDotDot) + { + // recover from typo of `...`, suggest `..` + let span = self.prev_token.span; + self.sess.emit_err(MissingDotDot { token_span: span, sugg_span: span }); + return true; + } + false + } + /// Parses `ident (COLON expr)?`. fn parse_expr_field(&mut self) -> PResult<'a, ExprField> { let attrs = self.parse_outer_attributes()?; @@ -3142,6 +3173,8 @@ impl<'a> Parser<'a> { && this.token.kind == token::Semi { TrailingToken::Semi + } else if this.token.kind == token::Gt { + TrailingToken::Gt } else { // FIXME - pass this through from the place where we know // we need a comma, rather than assuming that `#[attr] expr,` 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 b934e087608..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}; @@ -79,6 +80,7 @@ pub enum ForceCollect { pub enum TrailingToken { None, Semi, + Gt, /// If the trailing token is a comma, then capture it /// Otherwise, ignore the trailing token MaybeComma, @@ -103,6 +105,7 @@ macro_rules! maybe_whole { macro_rules! maybe_recover_from_interpolated_ty_qpath { ($self: expr, $allow_qpath_recovery: expr) => { if $allow_qpath_recovery + && $self.may_recover() && $self.look_ahead(1, |t| t == &token::ModSep) && let token::Interpolated(nt) = &$self.token.kind && let token::NtTy(ty) = &**nt @@ -114,6 +117,12 @@ macro_rules! maybe_recover_from_interpolated_ty_qpath { }; } +#[derive(Clone, Copy)] +pub enum Recovery { + Allowed, + Forbidden, +} + #[derive(Clone)] pub struct Parser<'a> { pub sess: &'a ParseSess, @@ -151,12 +160,15 @@ pub struct Parser<'a> { /// This allows us to recover when the user forget to add braces around /// multiple statements in the closure body. pub current_closure: Option<ClosureSpans>, + /// Whether the parser is allowed to do recovery. + /// This is disabled when parsing macro arguments, see #103534 + pub recovery: Recovery, } -// This type is used a lot, e.g. it's cloned when matching many declarative macro rules. Make sure +// This type is used a lot, e.g. it's cloned when matching many declarative macro rules with nonterminals. Make sure // it doesn't unintentionally get bigger. #[cfg(all(target_arch = "x86_64", target_pointer_width = "64"))] -rustc_data_structures::static_assert_size!(Parser<'_>, 328); +rustc_data_structures::static_assert_size!(Parser<'_>, 336); /// Stores span information about a closure. #[derive(Clone)] @@ -482,6 +494,7 @@ impl<'a> Parser<'a> { inner_attr_ranges: Default::default(), }, current_closure: None, + recovery: Recovery::Allowed, }; // Make parser point to the first token. @@ -490,6 +503,22 @@ impl<'a> Parser<'a> { parser } + pub fn forbid_recovery(mut self) -> Self { + self.recovery = Recovery::Forbidden; + self + } + + /// Whether the parser is allowed to recover from broken code. + /// + /// If this returns false, recovering broken code into valid code (especially if this recovery does lookahead) + /// is not allowed. All recovery done by the parser must be gated behind this check. + /// + /// Technically, this only needs to restrict eager recovery by doing lookahead at more tokens. + /// But making the distinction is very subtle, and simply forbidding all recovery is a lot simpler to uphold. + fn may_recover(&self) -> bool { + matches!(self.recovery, Recovery::Allowed) + } + pub fn unexpected<T>(&mut self) -> PResult<'a, T> { match self.expect_one_of(&[], &[]) { Err(e) => Err(e), @@ -608,6 +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. @@ -620,6 +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(); @@ -1099,8 +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 { @@ -1109,8 +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 @@ -1118,10 +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 { @@ -1376,8 +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/stmt.rs b/compiler/rustc_parse/src/parser/stmt.rs index a61e77b7c3b..12753c6785c 100644 --- a/compiler/rustc_parse/src/parser/stmt.rs +++ b/compiler/rustc_parse/src/parser/stmt.rs @@ -553,39 +553,46 @@ impl<'a> Parser<'a> { match stmt.kind { // Expression without semicolon. StmtKind::Expr(ref mut expr) - if self.token != token::Eof && classify::expr_requires_semi_to_be_stmt(expr) => - { + if self.token != token::Eof && classify::expr_requires_semi_to_be_stmt(expr) => { // Just check for errors and recover; do not eat semicolon yet. - if let Err(mut e) = - self.expect_one_of(&[], &[token::Semi, token::CloseDelim(Delimiter::Brace)]) - { - if let TokenKind::DocComment(..) = self.token.kind { - if let Ok(snippet) = self.span_to_snippet(self.token.span) { - let sp = self.token.span; - let marker = &snippet[..3]; - let (comment_marker, doc_comment_marker) = marker.split_at(2); - - e.span_suggestion( - sp.with_hi(sp.lo() + BytePos(marker.len() as u32)), - &format!( - "add a space before `{}` to use a regular comment", - doc_comment_marker, - ), - format!("{} {}", comment_marker, doc_comment_marker), - Applicability::MaybeIncorrect, - ); + // `expect_one_of` returns PResult<'a, bool /* recovered */> + let replace_with_err = + match self.expect_one_of(&[], &[token::Semi, token::CloseDelim(Delimiter::Brace)]) { + // Recover from parser, skip type error to avoid extra errors. + Ok(true) => true, + Err(mut e) => { + if let TokenKind::DocComment(..) = self.token.kind && + let Ok(snippet) = self.span_to_snippet(self.token.span) { + let sp = self.token.span; + let marker = &snippet[..3]; + let (comment_marker, doc_comment_marker) = marker.split_at(2); + + e.span_suggestion( + sp.with_hi(sp.lo() + BytePos(marker.len() as u32)), + &format!( + "add a space before `{}` to use a regular comment", + doc_comment_marker, + ), + format!("{} {}", comment_marker, doc_comment_marker), + Applicability::MaybeIncorrect, + ); } - } - if let Err(mut e) = - self.check_mistyped_turbofish_with_multiple_type_params(e, expr) - { - if recover.no() { - return Err(e); + + if let Err(mut e) = + self.check_mistyped_turbofish_with_multiple_type_params(e, expr) + { + if recover.no() { + return Err(e); + } + e.emit(); + self.recover_stmt(); } - e.emit(); - self.recover_stmt(); + true } - // Don't complain about type errors in body tail after parse error (#57383). + _ => false + }; + if replace_with_err { + // We already emitted an error, so don't emit another type error let sp = expr.span.to(self.prev_token.span); *expr = self.mk_expr_err(sp); } 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(); |