diff options
Diffstat (limited to 'compiler/rustc_parse/src/lexer/mod.rs')
-rw-r--r-- | compiler/rustc_parse/src/lexer/mod.rs | 77 |
1 files changed, 65 insertions, 12 deletions
diff --git a/compiler/rustc_parse/src/lexer/mod.rs b/compiler/rustc_parse/src/lexer/mod.rs index 9e856c9f2..c6e6b46e4 100644 --- a/compiler/rustc_parse/src/lexer/mod.rs +++ b/compiler/rustc_parse/src/lexer/mod.rs @@ -1,3 +1,5 @@ +use std::ops::Range; + use crate::errors; use crate::lexer::unicode_chars::UNICODE_ARRAY; use crate::make_unclosed_delims_error; @@ -6,7 +8,7 @@ use rustc_ast::token::{self, CommentKind, Delimiter, Token, TokenKind}; use rustc_ast::tokenstream::TokenStream; use rustc_ast::util::unicode::contains_text_flow_control_chars; use rustc_errors::{error_code, Applicability, Diagnostic, DiagnosticBuilder, StashKey}; -use rustc_lexer::unescape::{self, Mode}; +use rustc_lexer::unescape::{self, EscapeError, Mode}; use rustc_lexer::Cursor; use rustc_lexer::{Base, DocStyle, RawStrError}; use rustc_session::lint::builtin::{ @@ -67,7 +69,7 @@ pub(crate) fn parse_token_trees<'a>( match token_trees { Ok(stream) if unmatched_delims.is_empty() => Ok(stream), _ => { - // Return error if there are unmatched delimiters or unclosng delimiters. + // Return error if there are unmatched delimiters or unclosed delimiters. // We emit delimiter mismatch errors first, then emit the unclosing delimiter mismatch // because the delimiter mismatch is more likely to be the root cause of error @@ -204,16 +206,15 @@ impl<'a> StringReader<'a> { rustc_lexer::TokenKind::Literal { kind, suffix_start } => { let suffix_start = start + BytePos(suffix_start); let (kind, symbol) = self.cook_lexer_literal(start, suffix_start, kind); + if let token::LitKind::CStr | token::LitKind::CStrRaw(_) = kind { + self.sess.gated_spans.gate(sym::c_str_literals, self.mk_sp(start, self.pos)); + } let suffix = if suffix_start < self.pos { let string = self.str_from(suffix_start); if string == "_" { self.sess .span_diagnostic - .struct_span_err( - self.mk_sp(suffix_start, self.pos), - "underscore literal suffix is not allowed", - ) - .emit(); + .emit_err(errors::UnderscoreLiteralSuffix { span: self.mk_sp(suffix_start, self.pos) }); None } else { Some(Symbol::intern(string)) @@ -325,7 +326,7 @@ impl<'a> StringReader<'a> { ) -> DiagnosticBuilder<'a, !> { self.sess .span_diagnostic - .struct_span_fatal(self.mk_sp(from_pos, to_pos), &format!("{}: {}", m, escaped_char(c))) + .struct_span_fatal(self.mk_sp(from_pos, to_pos), format!("{}: {}", m, escaped_char(c))) } /// Detect usages of Unicode codepoints changing the direction of the text on screen and loudly @@ -419,6 +420,16 @@ impl<'a> StringReader<'a> { } self.cook_quoted(token::ByteStr, Mode::ByteStr, start, end, 2, 1) // b" " } + rustc_lexer::LiteralKind::CStr { terminated } => { + if !terminated { + self.sess.span_diagnostic.span_fatal_with_code( + self.mk_sp(start + BytePos(1), end), + "unterminated C string", + error_code!(E0767), + ) + } + self.cook_c_string(token::CStr, Mode::CStr, start, end, 2, 1) // c" " + } rustc_lexer::LiteralKind::RawStr { n_hashes } => { if let Some(n_hashes) = n_hashes { let n = u32::from(n_hashes); @@ -437,6 +448,15 @@ impl<'a> StringReader<'a> { self.report_raw_str_error(start, 2); } } + rustc_lexer::LiteralKind::RawCStr { n_hashes } => { + if let Some(n_hashes) = n_hashes { + let n = u32::from(n_hashes); + let kind = token::CStrRaw(n_hashes); + self.cook_c_string(kind, Mode::RawCStr, start, end, 3 + n, 1 + n) // cr##" "## + } else { + self.report_raw_str_error(start, 2); + } + } rustc_lexer::LiteralKind::Int { base, empty_int } => { if empty_int { let span = self.mk_sp(start, end); @@ -546,7 +566,7 @@ impl<'a> StringReader<'a> { err.span_label(self.mk_sp(start, start), "unterminated raw string"); if n_hashes > 0 { - err.note(&format!( + err.note(format!( "this raw string should be terminated with `\"{}`", "#".repeat(n_hashes as usize) )); @@ -642,7 +662,7 @@ impl<'a> StringReader<'a> { &RUST_2021_PREFIXES_INCOMPATIBLE_SYNTAX, prefix_span, ast::CRATE_NODE_ID, - &format!("prefix `{prefix}` is unknown"), + format!("prefix `{prefix}` is unknown"), BuiltinLintDiagnostics::ReservedPrefix(prefix_span), ); } @@ -652,7 +672,7 @@ impl<'a> StringReader<'a> { self.sess.emit_fatal(errors::TooManyHashes { span: self.mk_sp(start, self.pos), num }); } - fn cook_quoted( + fn cook_common( &self, kind: token::LitKind, mode: Mode, @@ -660,12 +680,13 @@ impl<'a> StringReader<'a> { end: BytePos, prefix_len: u32, postfix_len: u32, + unescape: fn(&str, Mode, &mut dyn FnMut(Range<usize>, Result<(), EscapeError>)), ) -> (token::LitKind, Symbol) { let mut has_fatal_err = false; 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| { + unescape(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(start, end); @@ -696,6 +717,38 @@ impl<'a> StringReader<'a> { (token::Err, self.symbol_from_to(start, end)) } } + + fn cook_quoted( + &self, + kind: token::LitKind, + mode: Mode, + start: BytePos, + end: BytePos, + prefix_len: u32, + postfix_len: u32, + ) -> (token::LitKind, Symbol) { + self.cook_common(kind, mode, start, end, prefix_len, postfix_len, |src, mode, callback| { + unescape::unescape_literal(src, mode, &mut |span, result| { + callback(span, result.map(drop)) + }) + }) + } + + fn cook_c_string( + &self, + kind: token::LitKind, + mode: Mode, + start: BytePos, + end: BytePos, + prefix_len: u32, + postfix_len: u32, + ) -> (token::LitKind, Symbol) { + self.cook_common(kind, mode, start, end, prefix_len, postfix_len, |src, mode, callback| { + unescape::unescape_c_string(src, mode, &mut |span, result| { + callback(span, result.map(drop)) + }) + }) + } } pub fn nfc_normalize(string: &str) -> Symbol { |