]>
Commit | Line | Data |
---|---|---|
3dfed10e XL |
1 | use rustc_ast::ast::AttrStyle; |
2 | use rustc_ast::token::{self, CommentKind, Token, TokenKind}; | |
60c5eb7d | 3 | use rustc_data_structures::sync::Lrc; |
ba9703b0 | 4 | use rustc_errors::{error_code, Applicability, DiagnosticBuilder, FatalError}; |
dfeec247 | 5 | use rustc_lexer::Base; |
f035d41b | 6 | use rustc_lexer::{unescape, RawStrError}; |
74b04a01 | 7 | use rustc_session::parse::ParseSess; |
dfeec247 XL |
8 | use rustc_span::symbol::{sym, Symbol}; |
9 | use rustc_span::{BytePos, Pos, Span}; | |
1a4d82fc | 10 | |
1a4d82fc | 11 | use std::char; |
3dfed10e | 12 | use tracing::debug; |
1a4d82fc | 13 | |
32a655c1 | 14 | mod tokentrees; |
60c5eb7d | 15 | mod unescape_error_reporting; |
dfeec247 | 16 | mod unicode_chars; |
ba9703b0 | 17 | |
3dfed10e | 18 | use rustc_lexer::{unescape::Mode, DocStyle}; |
60c5eb7d | 19 | use unescape_error_reporting::{emit_unescape_error, push_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 | ||
1a4d82fc | 30 | pub struct StringReader<'a> { |
416331ca XL |
31 | sess: &'a ParseSess, |
32 | /// Initial position, read-only. | |
33 | start_pos: BytePos, | |
34 | /// The absolute offset within the source_map of the current character. | |
f9f354fc | 35 | pos: BytePos, |
94b46f34 | 36 | /// Stop reading src at this index. |
416331ca XL |
37 | end_src_index: usize, |
38 | /// Source text to tokenize. | |
94b46f34 | 39 | src: Lrc<String>, |
48663c56 | 40 | override_span: Option<Span>, |
cc61c64b XL |
41 | } |
42 | ||
32a655c1 | 43 | impl<'a> StringReader<'a> { |
dfeec247 XL |
44 | pub fn new( |
45 | sess: &'a ParseSess, | |
46 | source_file: Lrc<rustc_span::SourceFile>, | |
47 | override_span: Option<Span>, | |
48 | ) -> Self { | |
ba9703b0 XL |
49 | // Make sure external source is loaded first, before accessing it. |
50 | // While this can't show up during normal parsing, `retokenize` may | |
51 | // be called with a source file from an external crate. | |
f035d41b | 52 | sess.source_map().ensure_source_file_source_present(Lrc::clone(&source_file)); |
ba9703b0 | 53 | |
ba9703b0 | 54 | let src = if let Some(src) = &source_file.src { |
f035d41b | 55 | Lrc::clone(&src) |
ba9703b0 | 56 | } else if let Some(src) = source_file.external_src.borrow().get_source() { |
f035d41b | 57 | Lrc::clone(&src) |
ba9703b0 | 58 | } else { |
dfeec247 XL |
59 | sess.span_diagnostic |
60 | .bug(&format!("cannot lex `source_file` without source: {}", source_file.name)); | |
ba9703b0 | 61 | }; |
c34b1796 | 62 | |
9e0c209e | 63 | StringReader { |
3b2f2976 | 64 | sess, |
416331ca | 65 | start_pos: source_file.start_pos, |
b7449926 | 66 | pos: source_file.start_pos, |
94b46f34 | 67 | end_src_index: src.len(), |
94b46f34 | 68 | src, |
94b46f34 | 69 | override_span, |
a7813a04 | 70 | } |
1a4d82fc JJ |
71 | } |
72 | ||
8bb4bdeb | 73 | pub fn retokenize(sess: &'a ParseSess, mut span: Span) -> Self { |
b7449926 XL |
74 | let begin = sess.source_map().lookup_byte_offset(span.lo()); |
75 | let end = sess.source_map().lookup_byte_offset(span.hi()); | |
8bb4bdeb XL |
76 | |
77 | // Make the range zero-length if the span is invalid. | |
60c5eb7d | 78 | if begin.sf.start_pos != end.sf.start_pos { |
0531ce1d | 79 | span = span.shrink_to_lo(); |
8bb4bdeb XL |
80 | } |
81 | ||
416331ca | 82 | let mut sr = StringReader::new(sess, begin.sf, None); |
8bb4bdeb XL |
83 | |
84 | // Seek the lexer to the right byte range. | |
94b46f34 | 85 | sr.end_src_index = sr.src_index(span.hi()); |
8bb4bdeb | 86 | |
416331ca XL |
87 | sr |
88 | } | |
8bb4bdeb | 89 | |
416331ca | 90 | fn mk_sp(&self, lo: BytePos, hi: BytePos) -> Span { |
e1599b0c | 91 | self.override_span.unwrap_or_else(|| Span::with_root_ctxt(lo, hi)) |
8bb4bdeb XL |
92 | } |
93 | ||
416331ca | 94 | /// Returns the next token, including trivia like whitespace or comments. |
416331ca XL |
95 | pub fn next_token(&mut self) -> Token { |
96 | let start_src_index = self.src_index(self.pos); | |
97 | let text: &str = &self.src[start_src_index..self.end_src_index]; | |
98 | ||
99 | if text.is_empty() { | |
100 | let span = self.mk_sp(self.pos, self.pos); | |
101 | return Token::new(token::Eof, span); | |
102 | } | |
103 | ||
104 | { | |
105 | let is_beginning_of_file = self.pos == self.start_pos; | |
106 | if is_beginning_of_file { | |
107 | if let Some(shebang_len) = rustc_lexer::strip_shebang(text) { | |
108 | let start = self.pos; | |
109 | self.pos = self.pos + BytePos::from_usize(shebang_len); | |
110 | ||
111 | let sym = self.symbol_from(start + BytePos::from_usize("#!".len())); | |
112 | let kind = token::Shebang(sym); | |
113 | ||
114 | let span = self.mk_sp(start, self.pos); | |
115 | return Token::new(kind, span); | |
116 | } | |
117 | } | |
118 | } | |
119 | ||
120 | let token = rustc_lexer::first_token(text); | |
121 | ||
122 | let start = self.pos; | |
123 | self.pos = self.pos + BytePos::from_usize(token.len); | |
124 | ||
125 | debug!("try_next_token: {:?}({:?})", token.kind, self.str_from(start)); | |
126 | ||
416331ca | 127 | let kind = self.cook_lexer_token(token.kind, start); |
416331ca XL |
128 | let span = self.mk_sp(start, self.pos); |
129 | Token::new(kind, span) | |
1a4d82fc JJ |
130 | } |
131 | ||
132 | /// Report a fatal lexical error with a given span. | |
94b46f34 | 133 | fn fatal_span(&self, sp: Span, m: &str) -> FatalError { |
32a655c1 | 134 | self.sess.span_diagnostic.span_fatal(sp, m) |
1a4d82fc JJ |
135 | } |
136 | ||
137 | /// Report a lexical error with a given span. | |
94b46f34 | 138 | fn err_span(&self, sp: Span, m: &str) { |
0731742a | 139 | self.sess.span_diagnostic.struct_span_err(sp, m).emit(); |
1a4d82fc JJ |
140 | } |
141 | ||
142 | /// Report a fatal error spanning [`from_pos`, `to_pos`). | |
92a42be0 | 143 | fn fatal_span_(&self, from_pos: BytePos, to_pos: BytePos, m: &str) -> FatalError { |
041b39d2 | 144 | self.fatal_span(self.mk_sp(from_pos, to_pos), m) |
1a4d82fc JJ |
145 | } |
146 | ||
147 | /// Report a lexical error spanning [`from_pos`, `to_pos`). | |
148 | fn err_span_(&self, from_pos: BytePos, to_pos: BytePos, m: &str) { | |
041b39d2 | 149 | self.err_span(self.mk_sp(from_pos, to_pos), m) |
1a4d82fc JJ |
150 | } |
151 | ||
dfeec247 XL |
152 | fn struct_fatal_span_char( |
153 | &self, | |
154 | from_pos: BytePos, | |
155 | to_pos: BytePos, | |
156 | m: &str, | |
157 | c: char, | |
158 | ) -> DiagnosticBuilder<'a> { | |
9cc50fc6 SL |
159 | let mut m = m.to_string(); |
160 | m.push_str(": "); | |
48663c56 | 161 | push_escaped_char(&mut m, c); |
b7449926 | 162 | |
041b39d2 | 163 | self.sess.span_diagnostic.struct_span_fatal(self.mk_sp(from_pos, to_pos), &m[..]) |
9cc50fc6 | 164 | } |
1a4d82fc | 165 | |
416331ca | 166 | /// Turns simple `rustc_lexer::TokenKind` enum into a rich |
74b04a01 | 167 | /// `librustc_ast::TokenKind`. This turns strings into interned |
416331ca | 168 | /// symbols and runs additional validation. |
dfeec247 | 169 | fn cook_lexer_token(&self, token: rustc_lexer::TokenKind, start: BytePos) -> TokenKind { |
416331ca | 170 | match token { |
3dfed10e XL |
171 | rustc_lexer::TokenKind::LineComment { doc_style } => { |
172 | match doc_style { | |
173 | Some(doc_style) => { | |
174 | // Opening delimiter of the length 3 is not included into the symbol. | |
175 | let content_start = start + BytePos(3); | |
176 | let content = self.str_from(content_start); | |
177 | ||
178 | self.cook_doc_comment(content_start, content, CommentKind::Line, doc_style) | |
179 | } | |
180 | None => token::Comment, | |
ba9703b0 | 181 | } |
416331ca | 182 | } |
3dfed10e | 183 | rustc_lexer::TokenKind::BlockComment { doc_style, terminated } => { |
416331ca | 184 | if !terminated { |
3dfed10e XL |
185 | let msg = match doc_style { |
186 | Some(_) => "unterminated block doc-comment", | |
187 | None => "unterminated block comment", | |
416331ca XL |
188 | }; |
189 | let last_bpos = self.pos; | |
f035d41b XL |
190 | self.sess |
191 | .span_diagnostic | |
192 | .struct_span_fatal_with_code( | |
193 | self.mk_sp(start, last_bpos), | |
194 | msg, | |
195 | error_code!(E0758), | |
196 | ) | |
197 | .emit(); | |
198 | FatalError.raise(); | |
416331ca | 199 | } |
3dfed10e XL |
200 | match doc_style { |
201 | Some(doc_style) => { | |
202 | // Opening delimiter of the length 3 and closing delimiter of the length 2 | |
203 | // are not included into the symbol. | |
204 | let content_start = start + BytePos(3); | |
205 | let content_end = self.pos - BytePos(if terminated { 2 } else { 0 }); | |
206 | let content = self.str_from_to(content_start, content_end); | |
207 | ||
208 | self.cook_doc_comment(content_start, content, CommentKind::Block, doc_style) | |
209 | } | |
210 | None => token::Comment, | |
ba9703b0 | 211 | } |
9cc50fc6 | 212 | } |
416331ca XL |
213 | rustc_lexer::TokenKind::Whitespace => token::Whitespace, |
214 | rustc_lexer::TokenKind::Ident | rustc_lexer::TokenKind::RawIdent => { | |
215 | let is_raw_ident = token == rustc_lexer::TokenKind::RawIdent; | |
216 | let mut ident_start = start; | |
217 | if is_raw_ident { | |
218 | ident_start = ident_start + BytePos(2); | |
219 | } | |
dfeec247 | 220 | let sym = nfc_normalize(self.str_from(ident_start)); |
f9f354fc XL |
221 | let span = self.mk_sp(start, self.pos); |
222 | self.sess.symbol_gallery.insert(sym, span); | |
416331ca | 223 | if is_raw_ident { |
416331ca XL |
224 | if !sym.can_be_raw() { |
225 | self.err_span(span, &format!("`{}` cannot be a raw identifier", sym)); | |
226 | } | |
227 | self.sess.raw_identifier_spans.borrow_mut().push(span); | |
228 | } | |
229 | token::Ident(sym, is_raw_ident) | |
230 | } | |
231 | rustc_lexer::TokenKind::Literal { kind, suffix_start } => { | |
232 | let suffix_start = start + BytePos(suffix_start as u32); | |
233 | let (kind, symbol) = self.cook_lexer_literal(start, suffix_start, kind); | |
234 | let suffix = if suffix_start < self.pos { | |
235 | let string = self.str_from(suffix_start); | |
236 | if string == "_" { | |
dfeec247 XL |
237 | self.sess |
238 | .span_diagnostic | |
239 | .struct_span_warn( | |
240 | self.mk_sp(suffix_start, self.pos), | |
241 | "underscore literal suffix is not allowed", | |
242 | ) | |
243 | .warn( | |
244 | "this was previously accepted by the compiler but is \ | |
416331ca | 245 | being phased out; it will become a hard error in \ |
dfeec247 XL |
246 | a future release!", |
247 | ) | |
248 | .note( | |
74b04a01 XL |
249 | "see issue #42326 \ |
250 | <https://github.com/rust-lang/rust/issues/42326> \ | |
251 | for more information", | |
dfeec247 | 252 | ) |
416331ca XL |
253 | .emit(); |
254 | None | |
255 | } else { | |
256 | Some(Symbol::intern(string)) | |
257 | } | |
1a4d82fc | 258 | } else { |
416331ca | 259 | None |
1a4d82fc | 260 | }; |
416331ca XL |
261 | token::Literal(token::Lit { kind, symbol, suffix }) |
262 | } | |
263 | rustc_lexer::TokenKind::Lifetime { starts_with_number } => { | |
264 | // Include the leading `'` in the real identifier, for macro | |
265 | // expansion purposes. See #12512 for the gory details of why | |
266 | // this is necessary. | |
267 | let lifetime_name = self.str_from(start); | |
268 | if starts_with_number { | |
dfeec247 | 269 | self.err_span_(start, self.pos, "lifetimes cannot start with a number"); |
416331ca XL |
270 | } |
271 | let ident = Symbol::intern(lifetime_name); | |
272 | token::Lifetime(ident) | |
273 | } | |
274 | rustc_lexer::TokenKind::Semi => token::Semi, | |
275 | rustc_lexer::TokenKind::Comma => token::Comma, | |
416331ca XL |
276 | rustc_lexer::TokenKind::Dot => token::Dot, |
277 | rustc_lexer::TokenKind::OpenParen => token::OpenDelim(token::Paren), | |
278 | rustc_lexer::TokenKind::CloseParen => token::CloseDelim(token::Paren), | |
279 | rustc_lexer::TokenKind::OpenBrace => token::OpenDelim(token::Brace), | |
280 | rustc_lexer::TokenKind::CloseBrace => token::CloseDelim(token::Brace), | |
281 | rustc_lexer::TokenKind::OpenBracket => token::OpenDelim(token::Bracket), | |
282 | rustc_lexer::TokenKind::CloseBracket => token::CloseDelim(token::Bracket), | |
283 | rustc_lexer::TokenKind::At => token::At, | |
284 | rustc_lexer::TokenKind::Pound => token::Pound, | |
285 | rustc_lexer::TokenKind::Tilde => token::Tilde, | |
286 | rustc_lexer::TokenKind::Question => token::Question, | |
416331ca XL |
287 | rustc_lexer::TokenKind::Colon => token::Colon, |
288 | rustc_lexer::TokenKind::Dollar => token::Dollar, | |
416331ca | 289 | rustc_lexer::TokenKind::Eq => token::Eq, |
3dfed10e | 290 | rustc_lexer::TokenKind::Bang => token::Not, |
416331ca | 291 | rustc_lexer::TokenKind::Lt => token::Lt, |
416331ca | 292 | rustc_lexer::TokenKind::Gt => token::Gt, |
416331ca | 293 | rustc_lexer::TokenKind::Minus => token::BinOp(token::Minus), |
416331ca | 294 | rustc_lexer::TokenKind::And => token::BinOp(token::And), |
416331ca | 295 | rustc_lexer::TokenKind::Or => token::BinOp(token::Or), |
416331ca | 296 | rustc_lexer::TokenKind::Plus => token::BinOp(token::Plus), |
416331ca | 297 | rustc_lexer::TokenKind::Star => token::BinOp(token::Star), |
416331ca | 298 | rustc_lexer::TokenKind::Slash => token::BinOp(token::Slash), |
416331ca | 299 | rustc_lexer::TokenKind::Caret => token::BinOp(token::Caret), |
416331ca | 300 | rustc_lexer::TokenKind::Percent => token::BinOp(token::Percent), |
416331ca XL |
301 | |
302 | rustc_lexer::TokenKind::Unknown => { | |
303 | let c = self.str_from(start).chars().next().unwrap(); | |
dfeec247 XL |
304 | let mut err = |
305 | self.struct_fatal_span_char(start, self.pos, "unknown start of token", c); | |
416331ca XL |
306 | // FIXME: the lexer could be used to turn the ASCII version of unicode homoglyphs, |
307 | // instead of keeping a table in `check_for_substitution`into the token. Ideally, | |
308 | // this should be inside `rustc_lexer`. However, we should first remove compound | |
309 | // tokens like `<<` from `rustc_lexer`, and then add fancier error recovery to it, | |
310 | // as there will be less overall work to do this way. | |
311 | let token = unicode_chars::check_for_substitution(self, start, c, &mut err) | |
312 | .unwrap_or_else(|| token::Unknown(self.symbol_from(start))); | |
313 | err.emit(); | |
314 | token | |
1a4d82fc JJ |
315 | } |
316 | } | |
416331ca | 317 | } |
b7449926 | 318 | |
3dfed10e XL |
319 | fn cook_doc_comment( |
320 | &self, | |
321 | content_start: BytePos, | |
322 | content: &str, | |
323 | comment_kind: CommentKind, | |
324 | doc_style: DocStyle, | |
325 | ) -> TokenKind { | |
326 | if content.contains('\r') { | |
327 | for (idx, _) in content.char_indices().filter(|&(_, c)| c == '\r') { | |
328 | self.err_span_( | |
329 | content_start + BytePos(idx as u32), | |
330 | content_start + BytePos(idx as u32 + 1), | |
331 | match comment_kind { | |
332 | CommentKind::Line => "bare CR not allowed in doc-comment", | |
333 | CommentKind::Block => "bare CR not allowed in block doc-comment", | |
334 | }, | |
335 | ); | |
336 | } | |
337 | } | |
338 | ||
339 | let attr_style = match doc_style { | |
340 | DocStyle::Outer => AttrStyle::Outer, | |
341 | DocStyle::Inner => AttrStyle::Inner, | |
342 | }; | |
343 | ||
344 | token::DocComment(comment_kind, attr_style, Symbol::intern(content)) | |
345 | } | |
346 | ||
416331ca XL |
347 | fn cook_lexer_literal( |
348 | &self, | |
349 | start: BytePos, | |
350 | suffix_start: BytePos, | |
dfeec247 | 351 | kind: rustc_lexer::LiteralKind, |
416331ca | 352 | ) -> (token::LitKind, Symbol) { |
f9f354fc XL |
353 | // prefix means `"` or `br"` or `r###"`, ... |
354 | let (lit_kind, mode, prefix_len, postfix_len) = match kind { | |
416331ca XL |
355 | rustc_lexer::LiteralKind::Char { terminated } => { |
356 | if !terminated { | |
f035d41b XL |
357 | self.sess |
358 | .span_diagnostic | |
359 | .struct_span_fatal_with_code( | |
360 | self.mk_sp(start, suffix_start), | |
361 | "unterminated character literal", | |
362 | error_code!(E0762), | |
363 | ) | |
364 | .emit(); | |
365 | FatalError.raise(); | |
416331ca | 366 | } |
f9f354fc | 367 | (token::Char, Mode::Char, 1, 1) // ' ' |
dfeec247 | 368 | } |
416331ca XL |
369 | rustc_lexer::LiteralKind::Byte { terminated } => { |
370 | if !terminated { | |
f035d41b XL |
371 | self.sess |
372 | .span_diagnostic | |
373 | .struct_span_fatal_with_code( | |
374 | self.mk_sp(start + BytePos(1), suffix_start), | |
375 | "unterminated byte constant", | |
376 | error_code!(E0763), | |
377 | ) | |
378 | .emit(); | |
379 | FatalError.raise(); | |
416331ca | 380 | } |
f9f354fc | 381 | (token::Byte, Mode::Byte, 2, 1) // b' ' |
dfeec247 | 382 | } |
416331ca XL |
383 | rustc_lexer::LiteralKind::Str { terminated } => { |
384 | if !terminated { | |
f035d41b XL |
385 | self.sess |
386 | .span_diagnostic | |
387 | .struct_span_fatal_with_code( | |
388 | self.mk_sp(start, suffix_start), | |
389 | "unterminated double quote string", | |
390 | error_code!(E0765), | |
391 | ) | |
392 | .emit(); | |
393 | FatalError.raise(); | |
416331ca | 394 | } |
f9f354fc | 395 | (token::Str, Mode::Str, 1, 1) // " " |
416331ca XL |
396 | } |
397 | rustc_lexer::LiteralKind::ByteStr { terminated } => { | |
398 | if !terminated { | |
f035d41b XL |
399 | self.sess |
400 | .span_diagnostic | |
401 | .struct_span_fatal_with_code( | |
402 | self.mk_sp(start + BytePos(1), suffix_start), | |
403 | "unterminated double quote byte string", | |
404 | error_code!(E0766), | |
405 | ) | |
406 | .emit(); | |
407 | FatalError.raise(); | |
416331ca | 408 | } |
f9f354fc | 409 | (token::ByteStr, Mode::ByteStr, 2, 1) // b" " |
416331ca | 410 | } |
f035d41b XL |
411 | rustc_lexer::LiteralKind::RawStr { n_hashes, err } => { |
412 | self.report_raw_str_error(start, err); | |
416331ca | 413 | let n = u32::from(n_hashes); |
f9f354fc | 414 | (token::StrRaw(n_hashes), Mode::RawStr, 2 + n, 1 + n) // r##" "## |
416331ca | 415 | } |
f035d41b XL |
416 | rustc_lexer::LiteralKind::RawByteStr { n_hashes, err } => { |
417 | self.report_raw_str_error(start, err); | |
416331ca | 418 | let n = u32::from(n_hashes); |
f9f354fc | 419 | (token::ByteStrRaw(n_hashes), Mode::RawByteStr, 3 + n, 1 + n) // br##" "## |
416331ca XL |
420 | } |
421 | rustc_lexer::LiteralKind::Int { base, empty_int } => { | |
f9f354fc | 422 | return if empty_int { |
f035d41b XL |
423 | self.sess |
424 | .span_diagnostic | |
425 | .struct_span_err_with_code( | |
426 | self.mk_sp(start, suffix_start), | |
427 | "no valid digits found for number", | |
428 | error_code!(E0768), | |
429 | ) | |
430 | .emit(); | |
416331ca XL |
431 | (token::Integer, sym::integer(0)) |
432 | } else { | |
433 | self.validate_int_literal(base, start, suffix_start); | |
434 | (token::Integer, self.symbol_from_to(start, suffix_start)) | |
f9f354fc | 435 | }; |
dfeec247 | 436 | } |
416331ca XL |
437 | rustc_lexer::LiteralKind::Float { base, empty_exponent } => { |
438 | if empty_exponent { | |
f035d41b | 439 | self.err_span_(start, self.pos, "expected at least one digit in exponent"); |
416331ca XL |
440 | } |
441 | ||
442 | match base { | |
dfeec247 XL |
443 | Base::Hexadecimal => self.err_span_( |
444 | start, | |
445 | suffix_start, | |
446 | "hexadecimal float literal is not supported", | |
447 | ), | |
416331ca | 448 | Base::Octal => { |
dfeec247 | 449 | self.err_span_(start, suffix_start, "octal float literal is not supported") |
416331ca XL |
450 | } |
451 | Base::Binary => { | |
dfeec247 | 452 | self.err_span_(start, suffix_start, "binary float literal is not supported") |
416331ca | 453 | } |
dfeec247 | 454 | _ => (), |
416331ca XL |
455 | } |
456 | ||
457 | let id = self.symbol_from_to(start, suffix_start); | |
f9f354fc | 458 | return (token::Float, id); |
dfeec247 | 459 | } |
f9f354fc XL |
460 | }; |
461 | let content_start = start + BytePos(prefix_len); | |
462 | let content_end = suffix_start - BytePos(postfix_len); | |
463 | let id = self.symbol_from_to(content_start, content_end); | |
464 | self.validate_literal_escape(mode, content_start, content_end); | |
f035d41b | 465 | (lit_kind, id) |
f9f354fc XL |
466 | } |
467 | ||
468 | pub fn pos(&self) -> BytePos { | |
469 | self.pos | |
1a4d82fc JJ |
470 | } |
471 | ||
94b46f34 XL |
472 | #[inline] |
473 | fn src_index(&self, pos: BytePos) -> usize { | |
416331ca | 474 | (pos - self.start_pos).to_usize() |
1a4d82fc JJ |
475 | } |
476 | ||
dc9dc135 XL |
477 | /// Slice of the source text from `start` up to but excluding `self.pos`, |
478 | /// meaning the slice does not include the character `self.ch`. | |
dfeec247 | 479 | fn str_from(&self, start: BytePos) -> &str { |
dc9dc135 | 480 | self.str_from_to(start, self.pos) |
1a4d82fc JJ |
481 | } |
482 | ||
dc9dc135 XL |
483 | /// Creates a Symbol from a given offset to the current offset. |
484 | fn symbol_from(&self, start: BytePos) -> Symbol { | |
c30ab7b3 | 485 | debug!("taking an ident from {:?} to {:?}", start, self.pos); |
dc9dc135 | 486 | Symbol::intern(self.str_from(start)) |
1a4d82fc JJ |
487 | } |
488 | ||
dc9dc135 XL |
489 | /// As symbol_from, with an explicit endpoint. |
490 | fn symbol_from_to(&self, start: BytePos, end: BytePos) -> Symbol { | |
1a4d82fc | 491 | debug!("taking an ident from {:?} to {:?}", start, end); |
dc9dc135 | 492 | Symbol::intern(self.str_from_to(start, end)) |
1a4d82fc JJ |
493 | } |
494 | ||
dc9dc135 | 495 | /// Slice of the source text spanning from `start` up to but excluding `end`. |
dfeec247 | 496 | fn str_from_to(&self, start: BytePos, end: BytePos) -> &str { |
dc9dc135 | 497 | &self.src[self.src_index(start)..self.src_index(end)] |
1a4d82fc JJ |
498 | } |
499 | ||
f035d41b XL |
500 | fn report_raw_str_error(&self, start: BytePos, opt_err: Option<RawStrError>) { |
501 | match opt_err { | |
502 | Some(RawStrError::InvalidStarter { bad_char }) => { | |
503 | self.report_non_started_raw_string(start, bad_char) | |
504 | } | |
505 | Some(RawStrError::NoTerminator { expected, found, possible_terminator_offset }) => self | |
506 | .report_unterminated_raw_string(start, expected, possible_terminator_offset, found), | |
507 | Some(RawStrError::TooManyDelimiters { found }) => { | |
508 | self.report_too_many_hashes(start, found) | |
ba9703b0 | 509 | } |
f035d41b | 510 | None => (), |
ba9703b0 XL |
511 | } |
512 | } | |
513 | ||
f035d41b | 514 | fn report_non_started_raw_string(&self, start: BytePos, bad_char: char) -> ! { |
dfeec247 XL |
515 | self.struct_fatal_span_char( |
516 | start, | |
517 | self.pos, | |
f035d41b | 518 | "found invalid character; only `#` is allowed in raw string delimitation", |
dfeec247 XL |
519 | bad_char, |
520 | ) | |
521 | .emit(); | |
416331ca | 522 | FatalError.raise() |
1a4d82fc JJ |
523 | } |
524 | ||
ba9703b0 XL |
525 | fn report_unterminated_raw_string( |
526 | &self, | |
527 | start: BytePos, | |
528 | n_hashes: usize, | |
529 | possible_offset: Option<usize>, | |
530 | found_terminators: usize, | |
531 | ) -> ! { | |
74b04a01 XL |
532 | let mut err = self.sess.span_diagnostic.struct_span_fatal_with_code( |
533 | self.mk_sp(start, start), | |
534 | "unterminated raw string", | |
535 | error_code!(E0748), | |
536 | ); | |
ba9703b0 | 537 | |
dfeec247 | 538 | err.span_label(self.mk_sp(start, start), "unterminated raw string"); |
1a4d82fc | 539 | |
416331ca | 540 | if n_hashes > 0 { |
dfeec247 XL |
541 | err.note(&format!( |
542 | "this raw string should be terminated with `\"{}`", | |
ba9703b0 | 543 | "#".repeat(n_hashes) |
dfeec247 | 544 | )); |
48663c56 | 545 | } |
1a4d82fc | 546 | |
ba9703b0 XL |
547 | if let Some(possible_offset) = possible_offset { |
548 | let lo = start + BytePos(possible_offset as u32); | |
549 | let hi = lo + BytePos(found_terminators as u32); | |
550 | let span = self.mk_sp(lo, hi); | |
551 | err.span_suggestion( | |
552 | span, | |
553 | "consider terminating the string here", | |
554 | "#".repeat(n_hashes), | |
555 | Applicability::MaybeIncorrect, | |
556 | ); | |
557 | } | |
558 | ||
416331ca XL |
559 | err.emit(); |
560 | FatalError.raise() | |
1a4d82fc JJ |
561 | } |
562 | ||
f035d41b XL |
563 | /// Note: It was decided to not add a test case, because it would be to big. |
564 | /// https://github.com/rust-lang/rust/pull/50296#issuecomment-392135180 | |
565 | fn report_too_many_hashes(&self, start: BytePos, found: usize) -> ! { | |
ba9703b0 XL |
566 | self.fatal_span_( |
567 | start, | |
568 | self.pos, | |
f035d41b XL |
569 | &format!( |
570 | "too many `#` symbols: raw strings may be delimited \ | |
571 | by up to 65535 `#` symbols, but found {}", | |
572 | found | |
573 | ), | |
ba9703b0 XL |
574 | ) |
575 | .raise(); | |
1a4d82fc | 576 | } |
48663c56 | 577 | |
f9f354fc XL |
578 | fn validate_literal_escape(&self, mode: Mode, content_start: BytePos, content_end: BytePos) { |
579 | let lit_content = self.str_from_to(content_start, content_end); | |
580 | unescape::unescape_literal(lit_content, mode, &mut |range, result| { | |
581 | // Here we only check for errors. The actual unescaping is done later. | |
582 | if let Err(err) = result { | |
583 | let span_with_quotes = | |
584 | self.mk_sp(content_start - BytePos(1), content_end + BytePos(1)); | |
dc9dc135 XL |
585 | emit_unescape_error( |
586 | &self.sess.span_diagnostic, | |
f9f354fc XL |
587 | lit_content, |
588 | span_with_quotes, | |
589 | mode, | |
dc9dc135 XL |
590 | range, |
591 | err, | |
f9f354fc | 592 | ); |
dc9dc135 | 593 | } |
f9f354fc | 594 | }); |
48663c56 | 595 | } |
1a4d82fc | 596 | |
416331ca XL |
597 | fn validate_int_literal(&self, base: Base, content_start: BytePos, content_end: BytePos) { |
598 | let base = match base { | |
599 | Base::Binary => 2, | |
600 | Base::Octal => 8, | |
601 | _ => return, | |
602 | }; | |
603 | let s = self.str_from_to(content_start + BytePos(2), content_end); | |
604 | for (idx, c) in s.char_indices() { | |
605 | let idx = idx as u32; | |
606 | if c != '_' && c.to_digit(base).is_none() { | |
607 | let lo = content_start + BytePos(2 + idx); | |
608 | let hi = content_start + BytePos(2 + idx + c.len_utf8() as u32); | |
dfeec247 | 609 | self.err_span_(lo, hi, &format!("invalid digit for a base {} literal", base)); |
416331ca XL |
610 | } |
611 | } | |
612 | } | |
9cc50fc6 | 613 | } |
dfeec247 XL |
614 | |
615 | pub fn nfc_normalize(string: &str) -> Symbol { | |
616 | use unicode_normalization::{is_nfc_quick, IsNormalized, UnicodeNormalization}; | |
617 | match is_nfc_quick(string.chars()) { | |
618 | IsNormalized::Yes => Symbol::intern(string), | |
619 | _ => { | |
620 | let normalized_str: String = string.chars().nfc().collect(); | |
621 | Symbol::intern(&normalized_str) | |
622 | } | |
623 | } | |
624 | } |