]>
Commit | Line | Data |
---|---|---|
136023e0 | 1 | use rustc_ast::ast::{self, AttrStyle}; |
3dfed10e | 2 | use rustc_ast::token::{self, CommentKind, Token, TokenKind}; |
1b1a35ee XL |
3 | use rustc_ast::tokenstream::{Spacing, TokenStream}; |
4 | use rustc_errors::{error_code, Applicability, DiagnosticBuilder, FatalError, PResult}; | |
5 | use rustc_lexer::unescape::{self, Mode}; | |
6 | use rustc_lexer::{Base, DocStyle, RawStrError}; | |
136023e0 XL |
7 | use rustc_session::lint::builtin::RUST_2021_PREFIXES_INCOMPATIBLE_SYNTAX; |
8 | use rustc_session::lint::BuiltinLintDiagnostics; | |
74b04a01 | 9 | use rustc_session::parse::ParseSess; |
dfeec247 | 10 | use rustc_span::symbol::{sym, Symbol}; |
136023e0 | 11 | use rustc_span::{edition::Edition, BytePos, Pos, Span}; |
1a4d82fc | 12 | |
3dfed10e | 13 | use tracing::debug; |
1a4d82fc | 14 | |
32a655c1 | 15 | mod tokentrees; |
60c5eb7d | 16 | mod unescape_error_reporting; |
dfeec247 | 17 | mod unicode_chars; |
ba9703b0 | 18 | |
5869c6ff | 19 | use unescape_error_reporting::{emit_unescape_error, escaped_char}; |
1a4d82fc | 20 | |
9fa01778 XL |
21 | #[derive(Clone, Debug)] |
22 | pub struct UnmatchedBrace { | |
23 | pub expected_delim: token::DelimToken, | |
e74abb32 | 24 | pub found_delim: Option<token::DelimToken>, |
9fa01778 XL |
25 | pub found_span: Span, |
26 | pub unclosed_span: Option<Span>, | |
27 | pub candidate_span: Option<Span>, | |
28 | } | |
29 | ||
1b1a35ee XL |
30 | crate fn parse_token_trees<'a>( |
31 | sess: &'a ParseSess, | |
32 | src: &'a str, | |
33 | start_pos: BytePos, | |
34 | override_span: Option<Span>, | |
35 | ) -> (PResult<'a, TokenStream>, Vec<UnmatchedBrace>) { | |
36 | StringReader { sess, start_pos, pos: start_pos, end_src_index: src.len(), src, override_span } | |
37 | .into_token_trees() | |
38 | } | |
39 | ||
40 | struct StringReader<'a> { | |
416331ca XL |
41 | sess: &'a ParseSess, |
42 | /// Initial position, read-only. | |
43 | start_pos: BytePos, | |
44 | /// The absolute offset within the source_map of the current character. | |
f9f354fc | 45 | pos: BytePos, |
94b46f34 | 46 | /// Stop reading src at this index. |
416331ca XL |
47 | end_src_index: usize, |
48 | /// Source text to tokenize. | |
1b1a35ee | 49 | src: &'a str, |
48663c56 | 50 | override_span: Option<Span>, |
cc61c64b XL |
51 | } |
52 | ||
32a655c1 | 53 | impl<'a> StringReader<'a> { |
416331ca | 54 | fn mk_sp(&self, lo: BytePos, hi: BytePos) -> Span { |
e1599b0c | 55 | self.override_span.unwrap_or_else(|| Span::with_root_ctxt(lo, hi)) |
8bb4bdeb XL |
56 | } |
57 | ||
1b1a35ee XL |
58 | /// Returns the next token, and info about preceding whitespace, if any. |
59 | fn next_token(&mut self) -> (Spacing, Token) { | |
60 | let mut spacing = Spacing::Joint; | |
61 | ||
62 | // Skip `#!` at the start of the file | |
416331ca XL |
63 | let start_src_index = self.src_index(self.pos); |
64 | let text: &str = &self.src[start_src_index..self.end_src_index]; | |
1b1a35ee XL |
65 | let is_beginning_of_file = self.pos == self.start_pos; |
66 | if is_beginning_of_file { | |
67 | if let Some(shebang_len) = rustc_lexer::strip_shebang(text) { | |
68 | self.pos = self.pos + BytePos::from_usize(shebang_len); | |
69 | spacing = Spacing::Alone; | |
70 | } | |
416331ca XL |
71 | } |
72 | ||
1b1a35ee XL |
73 | // Skip trivial (whitespace & comments) tokens |
74 | loop { | |
75 | let start_src_index = self.src_index(self.pos); | |
76 | let text: &str = &self.src[start_src_index..self.end_src_index]; | |
416331ca | 77 | |
1b1a35ee XL |
78 | if text.is_empty() { |
79 | let span = self.mk_sp(self.pos, self.pos); | |
80 | return (spacing, Token::new(token::Eof, span)); | |
416331ca | 81 | } |
416331ca | 82 | |
1b1a35ee | 83 | let token = rustc_lexer::first_token(text); |
416331ca | 84 | |
1b1a35ee XL |
85 | let start = self.pos; |
86 | self.pos = self.pos + BytePos::from_usize(token.len); | |
416331ca | 87 | |
1b1a35ee | 88 | debug!("next_token: {:?}({:?})", token.kind, self.str_from(start)); |
416331ca | 89 | |
1b1a35ee XL |
90 | match self.cook_lexer_token(token.kind, start) { |
91 | Some(kind) => { | |
92 | let span = self.mk_sp(start, self.pos); | |
93 | return (spacing, Token::new(kind, span)); | |
94 | } | |
95 | None => spacing = Spacing::Alone, | |
96 | } | |
97 | } | |
1a4d82fc JJ |
98 | } |
99 | ||
100 | /// Report a fatal lexical error with a given span. | |
94b46f34 | 101 | fn fatal_span(&self, sp: Span, m: &str) -> FatalError { |
32a655c1 | 102 | self.sess.span_diagnostic.span_fatal(sp, m) |
1a4d82fc JJ |
103 | } |
104 | ||
105 | /// Report a lexical error with a given span. | |
94b46f34 | 106 | fn err_span(&self, sp: Span, m: &str) { |
0731742a | 107 | self.sess.span_diagnostic.struct_span_err(sp, m).emit(); |
1a4d82fc JJ |
108 | } |
109 | ||
110 | /// Report a fatal error spanning [`from_pos`, `to_pos`). | |
92a42be0 | 111 | fn fatal_span_(&self, from_pos: BytePos, to_pos: BytePos, m: &str) -> FatalError { |
041b39d2 | 112 | self.fatal_span(self.mk_sp(from_pos, to_pos), m) |
1a4d82fc JJ |
113 | } |
114 | ||
115 | /// Report a lexical error spanning [`from_pos`, `to_pos`). | |
116 | fn err_span_(&self, from_pos: BytePos, to_pos: BytePos, m: &str) { | |
041b39d2 | 117 | self.err_span(self.mk_sp(from_pos, to_pos), m) |
1a4d82fc JJ |
118 | } |
119 | ||
dfeec247 XL |
120 | fn struct_fatal_span_char( |
121 | &self, | |
122 | from_pos: BytePos, | |
123 | to_pos: BytePos, | |
124 | m: &str, | |
125 | c: char, | |
126 | ) -> DiagnosticBuilder<'a> { | |
5869c6ff XL |
127 | self.sess |
128 | .span_diagnostic | |
129 | .struct_span_fatal(self.mk_sp(from_pos, to_pos), &format!("{}: {}", m, escaped_char(c))) | |
9cc50fc6 | 130 | } |
1a4d82fc | 131 | |
416331ca | 132 | /// Turns simple `rustc_lexer::TokenKind` enum into a rich |
cdc7bbd5 | 133 | /// `rustc_ast::TokenKind`. This turns strings into interned |
416331ca | 134 | /// symbols and runs additional validation. |
1b1a35ee XL |
135 | fn cook_lexer_token(&self, token: rustc_lexer::TokenKind, start: BytePos) -> Option<TokenKind> { |
136 | Some(match token { | |
3dfed10e | 137 | rustc_lexer::TokenKind::LineComment { doc_style } => { |
1b1a35ee XL |
138 | // Skip non-doc comments |
139 | let doc_style = doc_style?; | |
3dfed10e | 140 | |
1b1a35ee XL |
141 | // Opening delimiter of the length 3 is not included into the symbol. |
142 | let content_start = start + BytePos(3); | |
143 | let content = self.str_from(content_start); | |
144 | self.cook_doc_comment(content_start, content, CommentKind::Line, doc_style) | |
416331ca | 145 | } |
3dfed10e | 146 | rustc_lexer::TokenKind::BlockComment { doc_style, terminated } => { |
416331ca | 147 | if !terminated { |
3dfed10e XL |
148 | let msg = match doc_style { |
149 | Some(_) => "unterminated block doc-comment", | |
150 | None => "unterminated block comment", | |
416331ca XL |
151 | }; |
152 | let last_bpos = self.pos; | |
17df50a5 XL |
153 | self.sess.span_diagnostic.span_fatal_with_code( |
154 | self.mk_sp(start, last_bpos), | |
155 | msg, | |
156 | error_code!(E0758), | |
157 | ); | |
416331ca | 158 | } |
1b1a35ee XL |
159 | |
160 | // Skip non-doc comments | |
161 | let doc_style = doc_style?; | |
162 | ||
163 | // Opening delimiter of the length 3 and closing delimiter of the length 2 | |
164 | // are not included into the symbol. | |
165 | let content_start = start + BytePos(3); | |
166 | let content_end = self.pos - BytePos(if terminated { 2 } else { 0 }); | |
167 | let content = self.str_from_to(content_start, content_end); | |
168 | self.cook_doc_comment(content_start, content, CommentKind::Block, doc_style) | |
9cc50fc6 | 169 | } |
1b1a35ee | 170 | rustc_lexer::TokenKind::Whitespace => return None, |
136023e0 XL |
171 | rustc_lexer::TokenKind::Ident |
172 | | rustc_lexer::TokenKind::RawIdent | |
173 | | rustc_lexer::TokenKind::UnknownPrefix => { | |
416331ca | 174 | let is_raw_ident = token == rustc_lexer::TokenKind::RawIdent; |
136023e0 | 175 | let is_unknown_prefix = token == rustc_lexer::TokenKind::UnknownPrefix; |
416331ca XL |
176 | let mut ident_start = start; |
177 | if is_raw_ident { | |
178 | ident_start = ident_start + BytePos(2); | |
179 | } | |
136023e0 XL |
180 | if is_unknown_prefix { |
181 | self.report_unknown_prefix(start); | |
182 | } | |
dfeec247 | 183 | let sym = nfc_normalize(self.str_from(ident_start)); |
f9f354fc XL |
184 | let span = self.mk_sp(start, self.pos); |
185 | self.sess.symbol_gallery.insert(sym, span); | |
416331ca | 186 | if is_raw_ident { |
416331ca XL |
187 | if !sym.can_be_raw() { |
188 | self.err_span(span, &format!("`{}` cannot be a raw identifier", sym)); | |
189 | } | |
190 | self.sess.raw_identifier_spans.borrow_mut().push(span); | |
191 | } | |
192 | token::Ident(sym, is_raw_ident) | |
193 | } | |
194 | rustc_lexer::TokenKind::Literal { kind, suffix_start } => { | |
195 | let suffix_start = start + BytePos(suffix_start as u32); | |
196 | let (kind, symbol) = self.cook_lexer_literal(start, suffix_start, kind); | |
197 | let suffix = if suffix_start < self.pos { | |
198 | let string = self.str_from(suffix_start); | |
199 | if string == "_" { | |
dfeec247 XL |
200 | self.sess |
201 | .span_diagnostic | |
202 | .struct_span_warn( | |
203 | self.mk_sp(suffix_start, self.pos), | |
204 | "underscore literal suffix is not allowed", | |
205 | ) | |
206 | .warn( | |
207 | "this was previously accepted by the compiler but is \ | |
416331ca | 208 | being phased out; it will become a hard error in \ |
dfeec247 XL |
209 | a future release!", |
210 | ) | |
211 | .note( | |
74b04a01 XL |
212 | "see issue #42326 \ |
213 | <https://github.com/rust-lang/rust/issues/42326> \ | |
214 | for more information", | |
dfeec247 | 215 | ) |
416331ca XL |
216 | .emit(); |
217 | None | |
218 | } else { | |
219 | Some(Symbol::intern(string)) | |
220 | } | |
1a4d82fc | 221 | } else { |
416331ca | 222 | None |
1a4d82fc | 223 | }; |
416331ca XL |
224 | token::Literal(token::Lit { kind, symbol, suffix }) |
225 | } | |
226 | rustc_lexer::TokenKind::Lifetime { starts_with_number } => { | |
227 | // Include the leading `'` in the real identifier, for macro | |
228 | // expansion purposes. See #12512 for the gory details of why | |
229 | // this is necessary. | |
230 | let lifetime_name = self.str_from(start); | |
231 | if starts_with_number { | |
dfeec247 | 232 | self.err_span_(start, self.pos, "lifetimes cannot start with a number"); |
416331ca XL |
233 | } |
234 | let ident = Symbol::intern(lifetime_name); | |
235 | token::Lifetime(ident) | |
236 | } | |
237 | rustc_lexer::TokenKind::Semi => token::Semi, | |
238 | rustc_lexer::TokenKind::Comma => token::Comma, | |
416331ca XL |
239 | rustc_lexer::TokenKind::Dot => token::Dot, |
240 | rustc_lexer::TokenKind::OpenParen => token::OpenDelim(token::Paren), | |
241 | rustc_lexer::TokenKind::CloseParen => token::CloseDelim(token::Paren), | |
242 | rustc_lexer::TokenKind::OpenBrace => token::OpenDelim(token::Brace), | |
243 | rustc_lexer::TokenKind::CloseBrace => token::CloseDelim(token::Brace), | |
244 | rustc_lexer::TokenKind::OpenBracket => token::OpenDelim(token::Bracket), | |
245 | rustc_lexer::TokenKind::CloseBracket => token::CloseDelim(token::Bracket), | |
246 | rustc_lexer::TokenKind::At => token::At, | |
247 | rustc_lexer::TokenKind::Pound => token::Pound, | |
248 | rustc_lexer::TokenKind::Tilde => token::Tilde, | |
249 | rustc_lexer::TokenKind::Question => token::Question, | |
416331ca XL |
250 | rustc_lexer::TokenKind::Colon => token::Colon, |
251 | rustc_lexer::TokenKind::Dollar => token::Dollar, | |
416331ca | 252 | rustc_lexer::TokenKind::Eq => token::Eq, |
3dfed10e | 253 | rustc_lexer::TokenKind::Bang => token::Not, |
416331ca | 254 | rustc_lexer::TokenKind::Lt => token::Lt, |
416331ca | 255 | rustc_lexer::TokenKind::Gt => token::Gt, |
416331ca | 256 | rustc_lexer::TokenKind::Minus => token::BinOp(token::Minus), |
416331ca | 257 | rustc_lexer::TokenKind::And => token::BinOp(token::And), |
416331ca | 258 | rustc_lexer::TokenKind::Or => token::BinOp(token::Or), |
416331ca | 259 | rustc_lexer::TokenKind::Plus => token::BinOp(token::Plus), |
416331ca | 260 | rustc_lexer::TokenKind::Star => token::BinOp(token::Star), |
416331ca | 261 | rustc_lexer::TokenKind::Slash => token::BinOp(token::Slash), |
416331ca | 262 | rustc_lexer::TokenKind::Caret => token::BinOp(token::Caret), |
416331ca | 263 | rustc_lexer::TokenKind::Percent => token::BinOp(token::Percent), |
416331ca XL |
264 | |
265 | rustc_lexer::TokenKind::Unknown => { | |
266 | let c = self.str_from(start).chars().next().unwrap(); | |
dfeec247 XL |
267 | let mut err = |
268 | self.struct_fatal_span_char(start, self.pos, "unknown start of token", c); | |
416331ca XL |
269 | // FIXME: the lexer could be used to turn the ASCII version of unicode homoglyphs, |
270 | // instead of keeping a table in `check_for_substitution`into the token. Ideally, | |
271 | // this should be inside `rustc_lexer`. However, we should first remove compound | |
272 | // tokens like `<<` from `rustc_lexer`, and then add fancier error recovery to it, | |
273 | // as there will be less overall work to do this way. | |
1b1a35ee | 274 | let token = unicode_chars::check_for_substitution(self, start, c, &mut err); |
6a06907d XL |
275 | if c == '\x00' { |
276 | err.help("source files must contain UTF-8 encoded text, unexpected null bytes might occur when a different encoding is used"); | |
277 | } | |
416331ca | 278 | err.emit(); |
1b1a35ee | 279 | token? |
1a4d82fc | 280 | } |
1b1a35ee | 281 | }) |
416331ca | 282 | } |
b7449926 | 283 | |
3dfed10e XL |
284 | fn cook_doc_comment( |
285 | &self, | |
286 | content_start: BytePos, | |
287 | content: &str, | |
288 | comment_kind: CommentKind, | |
289 | doc_style: DocStyle, | |
290 | ) -> TokenKind { | |
291 | if content.contains('\r') { | |
292 | for (idx, _) in content.char_indices().filter(|&(_, c)| c == '\r') { | |
293 | self.err_span_( | |
294 | content_start + BytePos(idx as u32), | |
295 | content_start + BytePos(idx as u32 + 1), | |
296 | match comment_kind { | |
297 | CommentKind::Line => "bare CR not allowed in doc-comment", | |
298 | CommentKind::Block => "bare CR not allowed in block doc-comment", | |
299 | }, | |
300 | ); | |
301 | } | |
302 | } | |
303 | ||
304 | let attr_style = match doc_style { | |
305 | DocStyle::Outer => AttrStyle::Outer, | |
306 | DocStyle::Inner => AttrStyle::Inner, | |
307 | }; | |
308 | ||
309 | token::DocComment(comment_kind, attr_style, Symbol::intern(content)) | |
310 | } | |
311 | ||
416331ca XL |
312 | fn cook_lexer_literal( |
313 | &self, | |
314 | start: BytePos, | |
315 | suffix_start: BytePos, | |
dfeec247 | 316 | kind: rustc_lexer::LiteralKind, |
416331ca | 317 | ) -> (token::LitKind, Symbol) { |
f9f354fc XL |
318 | // prefix means `"` or `br"` or `r###"`, ... |
319 | let (lit_kind, mode, prefix_len, postfix_len) = match kind { | |
416331ca XL |
320 | rustc_lexer::LiteralKind::Char { terminated } => { |
321 | if !terminated { | |
17df50a5 XL |
322 | self.sess.span_diagnostic.span_fatal_with_code( |
323 | self.mk_sp(start, suffix_start), | |
324 | "unterminated character literal", | |
325 | error_code!(E0762), | |
326 | ) | |
416331ca | 327 | } |
f9f354fc | 328 | (token::Char, Mode::Char, 1, 1) // ' ' |
dfeec247 | 329 | } |
416331ca XL |
330 | rustc_lexer::LiteralKind::Byte { terminated } => { |
331 | if !terminated { | |
17df50a5 XL |
332 | self.sess.span_diagnostic.span_fatal_with_code( |
333 | self.mk_sp(start + BytePos(1), suffix_start), | |
334 | "unterminated byte constant", | |
335 | error_code!(E0763), | |
336 | ) | |
416331ca | 337 | } |
f9f354fc | 338 | (token::Byte, Mode::Byte, 2, 1) // b' ' |
dfeec247 | 339 | } |
416331ca XL |
340 | rustc_lexer::LiteralKind::Str { terminated } => { |
341 | if !terminated { | |
17df50a5 XL |
342 | self.sess.span_diagnostic.span_fatal_with_code( |
343 | self.mk_sp(start, suffix_start), | |
344 | "unterminated double quote string", | |
345 | error_code!(E0765), | |
346 | ) | |
416331ca | 347 | } |
f9f354fc | 348 | (token::Str, Mode::Str, 1, 1) // " " |
416331ca XL |
349 | } |
350 | rustc_lexer::LiteralKind::ByteStr { terminated } => { | |
351 | if !terminated { | |
17df50a5 XL |
352 | self.sess.span_diagnostic.span_fatal_with_code( |
353 | self.mk_sp(start + BytePos(1), suffix_start), | |
354 | "unterminated double quote byte string", | |
355 | error_code!(E0766), | |
356 | ) | |
416331ca | 357 | } |
f9f354fc | 358 | (token::ByteStr, Mode::ByteStr, 2, 1) // b" " |
416331ca | 359 | } |
f035d41b XL |
360 | rustc_lexer::LiteralKind::RawStr { n_hashes, err } => { |
361 | self.report_raw_str_error(start, err); | |
416331ca | 362 | let n = u32::from(n_hashes); |
f9f354fc | 363 | (token::StrRaw(n_hashes), Mode::RawStr, 2 + n, 1 + n) // r##" "## |
416331ca | 364 | } |
f035d41b XL |
365 | rustc_lexer::LiteralKind::RawByteStr { n_hashes, err } => { |
366 | self.report_raw_str_error(start, err); | |
416331ca | 367 | let n = u32::from(n_hashes); |
f9f354fc | 368 | (token::ByteStrRaw(n_hashes), Mode::RawByteStr, 3 + n, 1 + n) // br##" "## |
416331ca XL |
369 | } |
370 | rustc_lexer::LiteralKind::Int { base, empty_int } => { | |
f9f354fc | 371 | return if empty_int { |
f035d41b XL |
372 | self.sess |
373 | .span_diagnostic | |
374 | .struct_span_err_with_code( | |
375 | self.mk_sp(start, suffix_start), | |
376 | "no valid digits found for number", | |
377 | error_code!(E0768), | |
378 | ) | |
379 | .emit(); | |
416331ca XL |
380 | (token::Integer, sym::integer(0)) |
381 | } else { | |
382 | self.validate_int_literal(base, start, suffix_start); | |
383 | (token::Integer, self.symbol_from_to(start, suffix_start)) | |
f9f354fc | 384 | }; |
dfeec247 | 385 | } |
416331ca XL |
386 | rustc_lexer::LiteralKind::Float { base, empty_exponent } => { |
387 | if empty_exponent { | |
f035d41b | 388 | self.err_span_(start, self.pos, "expected at least one digit in exponent"); |
416331ca XL |
389 | } |
390 | ||
391 | match base { | |
dfeec247 XL |
392 | Base::Hexadecimal => self.err_span_( |
393 | start, | |
394 | suffix_start, | |
395 | "hexadecimal float literal is not supported", | |
396 | ), | |
416331ca | 397 | Base::Octal => { |
dfeec247 | 398 | self.err_span_(start, suffix_start, "octal float literal is not supported") |
416331ca XL |
399 | } |
400 | Base::Binary => { | |
dfeec247 | 401 | self.err_span_(start, suffix_start, "binary float literal is not supported") |
416331ca | 402 | } |
dfeec247 | 403 | _ => (), |
416331ca XL |
404 | } |
405 | ||
406 | let id = self.symbol_from_to(start, suffix_start); | |
f9f354fc | 407 | return (token::Float, id); |
dfeec247 | 408 | } |
f9f354fc XL |
409 | }; |
410 | let content_start = start + BytePos(prefix_len); | |
411 | let content_end = suffix_start - BytePos(postfix_len); | |
412 | let id = self.symbol_from_to(content_start, content_end); | |
5869c6ff | 413 | self.validate_literal_escape(mode, content_start, content_end, prefix_len, postfix_len); |
f035d41b | 414 | (lit_kind, id) |
f9f354fc XL |
415 | } |
416 | ||
94b46f34 XL |
417 | #[inline] |
418 | fn src_index(&self, pos: BytePos) -> usize { | |
416331ca | 419 | (pos - self.start_pos).to_usize() |
1a4d82fc JJ |
420 | } |
421 | ||
dc9dc135 XL |
422 | /// Slice of the source text from `start` up to but excluding `self.pos`, |
423 | /// meaning the slice does not include the character `self.ch`. | |
dfeec247 | 424 | fn str_from(&self, start: BytePos) -> &str { |
dc9dc135 | 425 | self.str_from_to(start, self.pos) |
1a4d82fc JJ |
426 | } |
427 | ||
dc9dc135 XL |
428 | /// As symbol_from, with an explicit endpoint. |
429 | fn symbol_from_to(&self, start: BytePos, end: BytePos) -> Symbol { | |
1a4d82fc | 430 | debug!("taking an ident from {:?} to {:?}", start, end); |
dc9dc135 | 431 | Symbol::intern(self.str_from_to(start, end)) |
1a4d82fc JJ |
432 | } |
433 | ||
dc9dc135 | 434 | /// Slice of the source text spanning from `start` up to but excluding `end`. |
dfeec247 | 435 | fn str_from_to(&self, start: BytePos, end: BytePos) -> &str { |
dc9dc135 | 436 | &self.src[self.src_index(start)..self.src_index(end)] |
1a4d82fc JJ |
437 | } |
438 | ||
f035d41b XL |
439 | fn report_raw_str_error(&self, start: BytePos, opt_err: Option<RawStrError>) { |
440 | match opt_err { | |
441 | Some(RawStrError::InvalidStarter { bad_char }) => { | |
442 | self.report_non_started_raw_string(start, bad_char) | |
443 | } | |
444 | Some(RawStrError::NoTerminator { expected, found, possible_terminator_offset }) => self | |
445 | .report_unterminated_raw_string(start, expected, possible_terminator_offset, found), | |
446 | Some(RawStrError::TooManyDelimiters { found }) => { | |
447 | self.report_too_many_hashes(start, found) | |
ba9703b0 | 448 | } |
f035d41b | 449 | None => (), |
ba9703b0 XL |
450 | } |
451 | } | |
452 | ||
f035d41b | 453 | fn report_non_started_raw_string(&self, start: BytePos, bad_char: char) -> ! { |
dfeec247 XL |
454 | self.struct_fatal_span_char( |
455 | start, | |
456 | self.pos, | |
f035d41b | 457 | "found invalid character; only `#` is allowed in raw string delimitation", |
dfeec247 XL |
458 | bad_char, |
459 | ) | |
460 | .emit(); | |
416331ca | 461 | FatalError.raise() |
1a4d82fc JJ |
462 | } |
463 | ||
ba9703b0 XL |
464 | fn report_unterminated_raw_string( |
465 | &self, | |
466 | start: BytePos, | |
467 | n_hashes: usize, | |
468 | possible_offset: Option<usize>, | |
469 | found_terminators: usize, | |
470 | ) -> ! { | |
74b04a01 XL |
471 | let mut err = self.sess.span_diagnostic.struct_span_fatal_with_code( |
472 | self.mk_sp(start, start), | |
473 | "unterminated raw string", | |
474 | error_code!(E0748), | |
475 | ); | |
ba9703b0 | 476 | |
dfeec247 | 477 | err.span_label(self.mk_sp(start, start), "unterminated raw string"); |
1a4d82fc | 478 | |
416331ca | 479 | if n_hashes > 0 { |
dfeec247 XL |
480 | err.note(&format!( |
481 | "this raw string should be terminated with `\"{}`", | |
ba9703b0 | 482 | "#".repeat(n_hashes) |
dfeec247 | 483 | )); |
48663c56 | 484 | } |
1a4d82fc | 485 | |
ba9703b0 XL |
486 | if let Some(possible_offset) = possible_offset { |
487 | let lo = start + BytePos(possible_offset as u32); | |
488 | let hi = lo + BytePos(found_terminators as u32); | |
489 | let span = self.mk_sp(lo, hi); | |
490 | err.span_suggestion( | |
491 | span, | |
492 | "consider terminating the string here", | |
493 | "#".repeat(n_hashes), | |
494 | Applicability::MaybeIncorrect, | |
495 | ); | |
496 | } | |
497 | ||
416331ca XL |
498 | err.emit(); |
499 | FatalError.raise() | |
1a4d82fc JJ |
500 | } |
501 | ||
136023e0 XL |
502 | // RFC 3101 introduced the idea of (reserved) prefixes. As of Rust 2021, |
503 | // using a (unknown) prefix is an error. In earlier editions, however, they | |
504 | // only result in a (allowed by default) lint, and are treated as regular | |
505 | // identifier tokens. | |
506 | fn report_unknown_prefix(&self, start: BytePos) { | |
507 | let prefix_span = self.mk_sp(start, self.pos); | |
508 | let msg = format!("prefix `{}` is unknown", self.str_from_to(start, self.pos)); | |
509 | ||
510 | let expn_data = prefix_span.ctxt().outer_expn_data(); | |
511 | ||
512 | if expn_data.edition >= Edition::Edition2021 { | |
513 | // In Rust 2021, this is a hard error. | |
514 | let mut err = self.sess.span_diagnostic.struct_span_err(prefix_span, &msg); | |
515 | err.span_label(prefix_span, "unknown prefix"); | |
516 | if expn_data.is_root() { | |
517 | err.span_suggestion_verbose( | |
518 | prefix_span.shrink_to_hi(), | |
519 | "consider inserting whitespace here", | |
520 | " ".into(), | |
521 | Applicability::MachineApplicable, | |
522 | ); | |
523 | } | |
524 | err.note("prefixed identifiers and literals are reserved since Rust 2021"); | |
525 | err.emit(); | |
526 | } else { | |
527 | // Before Rust 2021, only emit a lint for migration. | |
528 | self.sess.buffer_lint_with_diagnostic( | |
529 | &RUST_2021_PREFIXES_INCOMPATIBLE_SYNTAX, | |
530 | prefix_span, | |
531 | ast::CRATE_NODE_ID, | |
532 | &msg, | |
533 | BuiltinLintDiagnostics::ReservedPrefix(prefix_span), | |
534 | ); | |
535 | } | |
536 | } | |
537 | ||
fc512014 | 538 | /// Note: It was decided to not add a test case, because it would be too big. |
29967ef6 | 539 | /// <https://github.com/rust-lang/rust/pull/50296#issuecomment-392135180> |
f035d41b | 540 | fn report_too_many_hashes(&self, start: BytePos, found: usize) -> ! { |
ba9703b0 XL |
541 | self.fatal_span_( |
542 | start, | |
543 | self.pos, | |
f035d41b XL |
544 | &format!( |
545 | "too many `#` symbols: raw strings may be delimited \ | |
546 | by up to 65535 `#` symbols, but found {}", | |
547 | found | |
548 | ), | |
ba9703b0 XL |
549 | ) |
550 | .raise(); | |
1a4d82fc | 551 | } |
48663c56 | 552 | |
5869c6ff XL |
553 | fn validate_literal_escape( |
554 | &self, | |
555 | mode: Mode, | |
556 | content_start: BytePos, | |
557 | content_end: BytePos, | |
558 | prefix_len: u32, | |
559 | postfix_len: u32, | |
560 | ) { | |
f9f354fc XL |
561 | let lit_content = self.str_from_to(content_start, content_end); |
562 | unescape::unescape_literal(lit_content, mode, &mut |range, result| { | |
563 | // Here we only check for errors. The actual unescaping is done later. | |
564 | if let Err(err) = result { | |
5869c6ff XL |
565 | let span_with_quotes = self |
566 | .mk_sp(content_start - BytePos(prefix_len), content_end + BytePos(postfix_len)); | |
fc512014 XL |
567 | let (start, end) = (range.start as u32, range.end as u32); |
568 | let lo = content_start + BytePos(start); | |
569 | let hi = lo + BytePos(end - start); | |
570 | let span = self.mk_sp(lo, hi); | |
dc9dc135 XL |
571 | emit_unescape_error( |
572 | &self.sess.span_diagnostic, | |
f9f354fc XL |
573 | lit_content, |
574 | span_with_quotes, | |
fc512014 | 575 | span, |
f9f354fc | 576 | mode, |
dc9dc135 XL |
577 | range, |
578 | err, | |
f9f354fc | 579 | ); |
dc9dc135 | 580 | } |
f9f354fc | 581 | }); |
48663c56 | 582 | } |
1a4d82fc | 583 | |
416331ca XL |
584 | fn validate_int_literal(&self, base: Base, content_start: BytePos, content_end: BytePos) { |
585 | let base = match base { | |
586 | Base::Binary => 2, | |
587 | Base::Octal => 8, | |
588 | _ => return, | |
589 | }; | |
590 | let s = self.str_from_to(content_start + BytePos(2), content_end); | |
591 | for (idx, c) in s.char_indices() { | |
592 | let idx = idx as u32; | |
593 | if c != '_' && c.to_digit(base).is_none() { | |
594 | let lo = content_start + BytePos(2 + idx); | |
595 | let hi = content_start + BytePos(2 + idx + c.len_utf8() as u32); | |
dfeec247 | 596 | self.err_span_(lo, hi, &format!("invalid digit for a base {} literal", base)); |
416331ca XL |
597 | } |
598 | } | |
599 | } | |
9cc50fc6 | 600 | } |
dfeec247 XL |
601 | |
602 | pub fn nfc_normalize(string: &str) -> Symbol { | |
603 | use unicode_normalization::{is_nfc_quick, IsNormalized, UnicodeNormalization}; | |
604 | match is_nfc_quick(string.chars()) { | |
605 | IsNormalized::Yes => Symbol::intern(string), | |
606 | _ => { | |
607 | let normalized_str: String = string.chars().nfc().collect(); | |
608 | Symbol::intern(&normalized_str) | |
609 | } | |
610 | } | |
611 | } |