]>
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}; | |
c295e0f8 XL |
7 | use rustc_session::lint::builtin::{ |
8 | TEXT_DIRECTION_CODEPOINT_IN_COMMENT, RUST_2021_PREFIXES_INCOMPATIBLE_SYNTAX, | |
9 | }; | |
136023e0 | 10 | use rustc_session::lint::BuiltinLintDiagnostics; |
74b04a01 | 11 | use rustc_session::parse::ParseSess; |
dfeec247 | 12 | use rustc_span::symbol::{sym, Symbol}; |
136023e0 | 13 | use rustc_span::{edition::Edition, BytePos, Pos, Span}; |
1a4d82fc | 14 | |
3dfed10e | 15 | use tracing::debug; |
1a4d82fc | 16 | |
32a655c1 | 17 | mod tokentrees; |
60c5eb7d | 18 | mod unescape_error_reporting; |
dfeec247 | 19 | mod unicode_chars; |
ba9703b0 | 20 | |
5869c6ff | 21 | use unescape_error_reporting::{emit_unescape_error, escaped_char}; |
1a4d82fc | 22 | |
9fa01778 XL |
23 | #[derive(Clone, Debug)] |
24 | pub struct UnmatchedBrace { | |
25 | pub expected_delim: token::DelimToken, | |
e74abb32 | 26 | pub found_delim: Option<token::DelimToken>, |
9fa01778 XL |
27 | pub found_span: Span, |
28 | pub unclosed_span: Option<Span>, | |
29 | pub candidate_span: Option<Span>, | |
30 | } | |
31 | ||
1b1a35ee XL |
32 | crate fn parse_token_trees<'a>( |
33 | sess: &'a ParseSess, | |
34 | src: &'a str, | |
35 | start_pos: BytePos, | |
36 | override_span: Option<Span>, | |
37 | ) -> (PResult<'a, TokenStream>, Vec<UnmatchedBrace>) { | |
38 | StringReader { sess, start_pos, pos: start_pos, end_src_index: src.len(), src, override_span } | |
39 | .into_token_trees() | |
40 | } | |
41 | ||
42 | struct StringReader<'a> { | |
416331ca XL |
43 | sess: &'a ParseSess, |
44 | /// Initial position, read-only. | |
45 | start_pos: BytePos, | |
46 | /// The absolute offset within the source_map of the current character. | |
f9f354fc | 47 | pos: BytePos, |
94b46f34 | 48 | /// Stop reading src at this index. |
416331ca XL |
49 | end_src_index: usize, |
50 | /// Source text to tokenize. | |
1b1a35ee | 51 | src: &'a str, |
48663c56 | 52 | override_span: Option<Span>, |
cc61c64b XL |
53 | } |
54 | ||
32a655c1 | 55 | impl<'a> StringReader<'a> { |
416331ca | 56 | fn mk_sp(&self, lo: BytePos, hi: BytePos) -> Span { |
e1599b0c | 57 | self.override_span.unwrap_or_else(|| Span::with_root_ctxt(lo, hi)) |
8bb4bdeb XL |
58 | } |
59 | ||
1b1a35ee XL |
60 | /// Returns the next token, and info about preceding whitespace, if any. |
61 | fn next_token(&mut self) -> (Spacing, Token) { | |
62 | let mut spacing = Spacing::Joint; | |
63 | ||
64 | // Skip `#!` at the start of the file | |
416331ca XL |
65 | let start_src_index = self.src_index(self.pos); |
66 | let text: &str = &self.src[start_src_index..self.end_src_index]; | |
1b1a35ee XL |
67 | let is_beginning_of_file = self.pos == self.start_pos; |
68 | if is_beginning_of_file { | |
69 | if let Some(shebang_len) = rustc_lexer::strip_shebang(text) { | |
70 | self.pos = self.pos + BytePos::from_usize(shebang_len); | |
71 | spacing = Spacing::Alone; | |
72 | } | |
416331ca XL |
73 | } |
74 | ||
1b1a35ee XL |
75 | // Skip trivial (whitespace & comments) tokens |
76 | loop { | |
77 | let start_src_index = self.src_index(self.pos); | |
78 | let text: &str = &self.src[start_src_index..self.end_src_index]; | |
416331ca | 79 | |
1b1a35ee XL |
80 | if text.is_empty() { |
81 | let span = self.mk_sp(self.pos, self.pos); | |
82 | return (spacing, Token::new(token::Eof, span)); | |
416331ca | 83 | } |
416331ca | 84 | |
1b1a35ee | 85 | let token = rustc_lexer::first_token(text); |
416331ca | 86 | |
1b1a35ee XL |
87 | let start = self.pos; |
88 | self.pos = self.pos + BytePos::from_usize(token.len); | |
416331ca | 89 | |
1b1a35ee | 90 | debug!("next_token: {:?}({:?})", token.kind, self.str_from(start)); |
416331ca | 91 | |
1b1a35ee XL |
92 | match self.cook_lexer_token(token.kind, start) { |
93 | Some(kind) => { | |
94 | let span = self.mk_sp(start, self.pos); | |
95 | return (spacing, Token::new(kind, span)); | |
96 | } | |
97 | None => spacing = Spacing::Alone, | |
98 | } | |
99 | } | |
1a4d82fc JJ |
100 | } |
101 | ||
102 | /// Report a fatal lexical error with a given span. | |
94b46f34 | 103 | fn fatal_span(&self, sp: Span, m: &str) -> FatalError { |
32a655c1 | 104 | self.sess.span_diagnostic.span_fatal(sp, m) |
1a4d82fc JJ |
105 | } |
106 | ||
107 | /// Report a lexical error with a given span. | |
94b46f34 | 108 | fn err_span(&self, sp: Span, m: &str) { |
0731742a | 109 | self.sess.span_diagnostic.struct_span_err(sp, m).emit(); |
1a4d82fc JJ |
110 | } |
111 | ||
112 | /// Report a fatal error spanning [`from_pos`, `to_pos`). | |
92a42be0 | 113 | fn fatal_span_(&self, from_pos: BytePos, to_pos: BytePos, m: &str) -> FatalError { |
041b39d2 | 114 | self.fatal_span(self.mk_sp(from_pos, to_pos), m) |
1a4d82fc JJ |
115 | } |
116 | ||
117 | /// Report a lexical error spanning [`from_pos`, `to_pos`). | |
118 | fn err_span_(&self, from_pos: BytePos, to_pos: BytePos, m: &str) { | |
041b39d2 | 119 | self.err_span(self.mk_sp(from_pos, to_pos), m) |
1a4d82fc JJ |
120 | } |
121 | ||
dfeec247 XL |
122 | fn struct_fatal_span_char( |
123 | &self, | |
124 | from_pos: BytePos, | |
125 | to_pos: BytePos, | |
126 | m: &str, | |
127 | c: char, | |
128 | ) -> DiagnosticBuilder<'a> { | |
5869c6ff XL |
129 | self.sess |
130 | .span_diagnostic | |
131 | .struct_span_fatal(self.mk_sp(from_pos, to_pos), &format!("{}: {}", m, escaped_char(c))) | |
9cc50fc6 | 132 | } |
1a4d82fc | 133 | |
c295e0f8 XL |
134 | /// Detect usages of Unicode codepoints changing the direction of the text on screen and loudly |
135 | /// complain about it. | |
136 | fn lint_unicode_text_flow(&self, start: BytePos) { | |
137 | // Opening delimiter of the length 2 is not included into the comment text. | |
138 | let content_start = start + BytePos(2); | |
139 | let content = self.str_from(content_start); | |
140 | let span = self.mk_sp(start, self.pos); | |
141 | const UNICODE_TEXT_FLOW_CHARS: &[char] = &[ | |
142 | '\u{202A}', '\u{202B}', '\u{202D}', '\u{202E}', '\u{2066}', '\u{2067}', '\u{2068}', | |
143 | '\u{202C}', '\u{2069}', | |
144 | ]; | |
145 | if content.contains(UNICODE_TEXT_FLOW_CHARS) { | |
146 | self.sess.buffer_lint_with_diagnostic( | |
147 | &TEXT_DIRECTION_CODEPOINT_IN_COMMENT, | |
148 | span, | |
149 | ast::CRATE_NODE_ID, | |
150 | "unicode codepoint changing visible direction of text present in comment", | |
151 | BuiltinLintDiagnostics::UnicodeTextFlow(span, content.to_string()), | |
152 | ); | |
153 | } | |
154 | } | |
155 | ||
416331ca | 156 | /// Turns simple `rustc_lexer::TokenKind` enum into a rich |
cdc7bbd5 | 157 | /// `rustc_ast::TokenKind`. This turns strings into interned |
416331ca | 158 | /// symbols and runs additional validation. |
1b1a35ee XL |
159 | fn cook_lexer_token(&self, token: rustc_lexer::TokenKind, start: BytePos) -> Option<TokenKind> { |
160 | Some(match token { | |
3dfed10e | 161 | rustc_lexer::TokenKind::LineComment { doc_style } => { |
1b1a35ee | 162 | // Skip non-doc comments |
c295e0f8 XL |
163 | let doc_style = if let Some(doc_style) = doc_style { |
164 | doc_style | |
165 | } else { | |
166 | self.lint_unicode_text_flow(start); | |
167 | return None; | |
168 | }; | |
3dfed10e | 169 | |
1b1a35ee XL |
170 | // Opening delimiter of the length 3 is not included into the symbol. |
171 | let content_start = start + BytePos(3); | |
172 | let content = self.str_from(content_start); | |
173 | self.cook_doc_comment(content_start, content, CommentKind::Line, doc_style) | |
416331ca | 174 | } |
3dfed10e | 175 | rustc_lexer::TokenKind::BlockComment { doc_style, terminated } => { |
416331ca | 176 | if !terminated { |
3dfed10e XL |
177 | let msg = match doc_style { |
178 | Some(_) => "unterminated block doc-comment", | |
179 | None => "unterminated block comment", | |
416331ca XL |
180 | }; |
181 | let last_bpos = self.pos; | |
17df50a5 XL |
182 | self.sess.span_diagnostic.span_fatal_with_code( |
183 | self.mk_sp(start, last_bpos), | |
184 | msg, | |
185 | error_code!(E0758), | |
186 | ); | |
416331ca | 187 | } |
1b1a35ee XL |
188 | |
189 | // Skip non-doc comments | |
c295e0f8 XL |
190 | let doc_style = if let Some(doc_style) = doc_style { |
191 | doc_style | |
192 | } else { | |
193 | self.lint_unicode_text_flow(start); | |
194 | return None; | |
195 | }; | |
1b1a35ee XL |
196 | |
197 | // Opening delimiter of the length 3 and closing delimiter of the length 2 | |
198 | // are not included into the symbol. | |
199 | let content_start = start + BytePos(3); | |
200 | let content_end = self.pos - BytePos(if terminated { 2 } else { 0 }); | |
201 | let content = self.str_from_to(content_start, content_end); | |
202 | self.cook_doc_comment(content_start, content, CommentKind::Block, doc_style) | |
9cc50fc6 | 203 | } |
1b1a35ee | 204 | rustc_lexer::TokenKind::Whitespace => return None, |
136023e0 XL |
205 | rustc_lexer::TokenKind::Ident |
206 | | rustc_lexer::TokenKind::RawIdent | |
207 | | rustc_lexer::TokenKind::UnknownPrefix => { | |
416331ca | 208 | let is_raw_ident = token == rustc_lexer::TokenKind::RawIdent; |
136023e0 | 209 | let is_unknown_prefix = token == rustc_lexer::TokenKind::UnknownPrefix; |
416331ca XL |
210 | let mut ident_start = start; |
211 | if is_raw_ident { | |
212 | ident_start = ident_start + BytePos(2); | |
213 | } | |
136023e0 XL |
214 | if is_unknown_prefix { |
215 | self.report_unknown_prefix(start); | |
216 | } | |
dfeec247 | 217 | let sym = nfc_normalize(self.str_from(ident_start)); |
f9f354fc XL |
218 | let span = self.mk_sp(start, self.pos); |
219 | self.sess.symbol_gallery.insert(sym, span); | |
416331ca | 220 | if is_raw_ident { |
416331ca XL |
221 | if !sym.can_be_raw() { |
222 | self.err_span(span, &format!("`{}` cannot be a raw identifier", sym)); | |
223 | } | |
224 | self.sess.raw_identifier_spans.borrow_mut().push(span); | |
225 | } | |
226 | token::Ident(sym, is_raw_ident) | |
227 | } | |
228 | rustc_lexer::TokenKind::Literal { kind, suffix_start } => { | |
229 | let suffix_start = start + BytePos(suffix_start as u32); | |
230 | let (kind, symbol) = self.cook_lexer_literal(start, suffix_start, kind); | |
231 | let suffix = if suffix_start < self.pos { | |
232 | let string = self.str_from(suffix_start); | |
233 | if string == "_" { | |
dfeec247 XL |
234 | self.sess |
235 | .span_diagnostic | |
236 | .struct_span_warn( | |
237 | self.mk_sp(suffix_start, self.pos), | |
238 | "underscore literal suffix is not allowed", | |
239 | ) | |
240 | .warn( | |
241 | "this was previously accepted by the compiler but is \ | |
416331ca | 242 | being phased out; it will become a hard error in \ |
dfeec247 XL |
243 | a future release!", |
244 | ) | |
245 | .note( | |
74b04a01 XL |
246 | "see issue #42326 \ |
247 | <https://github.com/rust-lang/rust/issues/42326> \ | |
248 | for more information", | |
dfeec247 | 249 | ) |
416331ca XL |
250 | .emit(); |
251 | None | |
252 | } else { | |
253 | Some(Symbol::intern(string)) | |
254 | } | |
1a4d82fc | 255 | } else { |
416331ca | 256 | None |
1a4d82fc | 257 | }; |
416331ca XL |
258 | token::Literal(token::Lit { kind, symbol, suffix }) |
259 | } | |
260 | rustc_lexer::TokenKind::Lifetime { starts_with_number } => { | |
261 | // Include the leading `'` in the real identifier, for macro | |
262 | // expansion purposes. See #12512 for the gory details of why | |
263 | // this is necessary. | |
264 | let lifetime_name = self.str_from(start); | |
265 | if starts_with_number { | |
dfeec247 | 266 | self.err_span_(start, self.pos, "lifetimes cannot start with a number"); |
416331ca XL |
267 | } |
268 | let ident = Symbol::intern(lifetime_name); | |
269 | token::Lifetime(ident) | |
270 | } | |
271 | rustc_lexer::TokenKind::Semi => token::Semi, | |
272 | rustc_lexer::TokenKind::Comma => token::Comma, | |
416331ca XL |
273 | rustc_lexer::TokenKind::Dot => token::Dot, |
274 | rustc_lexer::TokenKind::OpenParen => token::OpenDelim(token::Paren), | |
275 | rustc_lexer::TokenKind::CloseParen => token::CloseDelim(token::Paren), | |
276 | rustc_lexer::TokenKind::OpenBrace => token::OpenDelim(token::Brace), | |
277 | rustc_lexer::TokenKind::CloseBrace => token::CloseDelim(token::Brace), | |
278 | rustc_lexer::TokenKind::OpenBracket => token::OpenDelim(token::Bracket), | |
279 | rustc_lexer::TokenKind::CloseBracket => token::CloseDelim(token::Bracket), | |
280 | rustc_lexer::TokenKind::At => token::At, | |
281 | rustc_lexer::TokenKind::Pound => token::Pound, | |
282 | rustc_lexer::TokenKind::Tilde => token::Tilde, | |
283 | rustc_lexer::TokenKind::Question => token::Question, | |
416331ca XL |
284 | rustc_lexer::TokenKind::Colon => token::Colon, |
285 | rustc_lexer::TokenKind::Dollar => token::Dollar, | |
416331ca | 286 | rustc_lexer::TokenKind::Eq => token::Eq, |
3dfed10e | 287 | rustc_lexer::TokenKind::Bang => token::Not, |
416331ca | 288 | rustc_lexer::TokenKind::Lt => token::Lt, |
416331ca | 289 | rustc_lexer::TokenKind::Gt => token::Gt, |
416331ca | 290 | rustc_lexer::TokenKind::Minus => token::BinOp(token::Minus), |
416331ca | 291 | rustc_lexer::TokenKind::And => token::BinOp(token::And), |
416331ca | 292 | rustc_lexer::TokenKind::Or => token::BinOp(token::Or), |
416331ca | 293 | rustc_lexer::TokenKind::Plus => token::BinOp(token::Plus), |
416331ca | 294 | rustc_lexer::TokenKind::Star => token::BinOp(token::Star), |
416331ca | 295 | rustc_lexer::TokenKind::Slash => token::BinOp(token::Slash), |
416331ca | 296 | rustc_lexer::TokenKind::Caret => token::BinOp(token::Caret), |
416331ca | 297 | rustc_lexer::TokenKind::Percent => token::BinOp(token::Percent), |
416331ca XL |
298 | |
299 | rustc_lexer::TokenKind::Unknown => { | |
300 | let c = self.str_from(start).chars().next().unwrap(); | |
dfeec247 XL |
301 | let mut err = |
302 | self.struct_fatal_span_char(start, self.pos, "unknown start of token", c); | |
416331ca XL |
303 | // FIXME: the lexer could be used to turn the ASCII version of unicode homoglyphs, |
304 | // instead of keeping a table in `check_for_substitution`into the token. Ideally, | |
305 | // this should be inside `rustc_lexer`. However, we should first remove compound | |
306 | // tokens like `<<` from `rustc_lexer`, and then add fancier error recovery to it, | |
307 | // as there will be less overall work to do this way. | |
1b1a35ee | 308 | let token = unicode_chars::check_for_substitution(self, start, c, &mut err); |
6a06907d XL |
309 | if c == '\x00' { |
310 | err.help("source files must contain UTF-8 encoded text, unexpected null bytes might occur when a different encoding is used"); | |
311 | } | |
416331ca | 312 | err.emit(); |
1b1a35ee | 313 | token? |
1a4d82fc | 314 | } |
1b1a35ee | 315 | }) |
416331ca | 316 | } |
b7449926 | 317 | |
3dfed10e XL |
318 | fn cook_doc_comment( |
319 | &self, | |
320 | content_start: BytePos, | |
321 | content: &str, | |
322 | comment_kind: CommentKind, | |
323 | doc_style: DocStyle, | |
324 | ) -> TokenKind { | |
325 | if content.contains('\r') { | |
326 | for (idx, _) in content.char_indices().filter(|&(_, c)| c == '\r') { | |
327 | self.err_span_( | |
328 | content_start + BytePos(idx as u32), | |
329 | content_start + BytePos(idx as u32 + 1), | |
330 | match comment_kind { | |
331 | CommentKind::Line => "bare CR not allowed in doc-comment", | |
332 | CommentKind::Block => "bare CR not allowed in block doc-comment", | |
333 | }, | |
334 | ); | |
335 | } | |
336 | } | |
337 | ||
338 | let attr_style = match doc_style { | |
339 | DocStyle::Outer => AttrStyle::Outer, | |
340 | DocStyle::Inner => AttrStyle::Inner, | |
341 | }; | |
342 | ||
343 | token::DocComment(comment_kind, attr_style, Symbol::intern(content)) | |
344 | } | |
345 | ||
416331ca XL |
346 | fn cook_lexer_literal( |
347 | &self, | |
348 | start: BytePos, | |
349 | suffix_start: BytePos, | |
dfeec247 | 350 | kind: rustc_lexer::LiteralKind, |
416331ca | 351 | ) -> (token::LitKind, Symbol) { |
f9f354fc XL |
352 | // prefix means `"` or `br"` or `r###"`, ... |
353 | let (lit_kind, mode, prefix_len, postfix_len) = match kind { | |
416331ca XL |
354 | rustc_lexer::LiteralKind::Char { terminated } => { |
355 | if !terminated { | |
17df50a5 XL |
356 | self.sess.span_diagnostic.span_fatal_with_code( |
357 | self.mk_sp(start, suffix_start), | |
358 | "unterminated character literal", | |
359 | error_code!(E0762), | |
360 | ) | |
416331ca | 361 | } |
f9f354fc | 362 | (token::Char, Mode::Char, 1, 1) // ' ' |
dfeec247 | 363 | } |
416331ca XL |
364 | rustc_lexer::LiteralKind::Byte { terminated } => { |
365 | if !terminated { | |
17df50a5 XL |
366 | self.sess.span_diagnostic.span_fatal_with_code( |
367 | self.mk_sp(start + BytePos(1), suffix_start), | |
368 | "unterminated byte constant", | |
369 | error_code!(E0763), | |
370 | ) | |
416331ca | 371 | } |
f9f354fc | 372 | (token::Byte, Mode::Byte, 2, 1) // b' ' |
dfeec247 | 373 | } |
416331ca XL |
374 | rustc_lexer::LiteralKind::Str { terminated } => { |
375 | if !terminated { | |
17df50a5 XL |
376 | self.sess.span_diagnostic.span_fatal_with_code( |
377 | self.mk_sp(start, suffix_start), | |
378 | "unterminated double quote string", | |
379 | error_code!(E0765), | |
380 | ) | |
416331ca | 381 | } |
f9f354fc | 382 | (token::Str, Mode::Str, 1, 1) // " " |
416331ca XL |
383 | } |
384 | rustc_lexer::LiteralKind::ByteStr { terminated } => { | |
385 | if !terminated { | |
17df50a5 XL |
386 | self.sess.span_diagnostic.span_fatal_with_code( |
387 | self.mk_sp(start + BytePos(1), suffix_start), | |
388 | "unterminated double quote byte string", | |
389 | error_code!(E0766), | |
390 | ) | |
416331ca | 391 | } |
f9f354fc | 392 | (token::ByteStr, Mode::ByteStr, 2, 1) // b" " |
416331ca | 393 | } |
f035d41b XL |
394 | rustc_lexer::LiteralKind::RawStr { n_hashes, err } => { |
395 | self.report_raw_str_error(start, err); | |
416331ca | 396 | let n = u32::from(n_hashes); |
f9f354fc | 397 | (token::StrRaw(n_hashes), Mode::RawStr, 2 + n, 1 + n) // r##" "## |
416331ca | 398 | } |
f035d41b XL |
399 | rustc_lexer::LiteralKind::RawByteStr { n_hashes, err } => { |
400 | self.report_raw_str_error(start, err); | |
416331ca | 401 | let n = u32::from(n_hashes); |
f9f354fc | 402 | (token::ByteStrRaw(n_hashes), Mode::RawByteStr, 3 + n, 1 + n) // br##" "## |
416331ca XL |
403 | } |
404 | rustc_lexer::LiteralKind::Int { base, empty_int } => { | |
f9f354fc | 405 | return if empty_int { |
f035d41b XL |
406 | self.sess |
407 | .span_diagnostic | |
408 | .struct_span_err_with_code( | |
409 | self.mk_sp(start, suffix_start), | |
410 | "no valid digits found for number", | |
411 | error_code!(E0768), | |
412 | ) | |
413 | .emit(); | |
416331ca XL |
414 | (token::Integer, sym::integer(0)) |
415 | } else { | |
416 | self.validate_int_literal(base, start, suffix_start); | |
417 | (token::Integer, self.symbol_from_to(start, suffix_start)) | |
f9f354fc | 418 | }; |
dfeec247 | 419 | } |
416331ca XL |
420 | rustc_lexer::LiteralKind::Float { base, empty_exponent } => { |
421 | if empty_exponent { | |
f035d41b | 422 | self.err_span_(start, self.pos, "expected at least one digit in exponent"); |
416331ca XL |
423 | } |
424 | ||
425 | match base { | |
dfeec247 XL |
426 | Base::Hexadecimal => self.err_span_( |
427 | start, | |
428 | suffix_start, | |
429 | "hexadecimal float literal is not supported", | |
430 | ), | |
416331ca | 431 | Base::Octal => { |
dfeec247 | 432 | self.err_span_(start, suffix_start, "octal float literal is not supported") |
416331ca XL |
433 | } |
434 | Base::Binary => { | |
dfeec247 | 435 | self.err_span_(start, suffix_start, "binary float literal is not supported") |
416331ca | 436 | } |
dfeec247 | 437 | _ => (), |
416331ca XL |
438 | } |
439 | ||
440 | let id = self.symbol_from_to(start, suffix_start); | |
f9f354fc | 441 | return (token::Float, id); |
dfeec247 | 442 | } |
f9f354fc XL |
443 | }; |
444 | let content_start = start + BytePos(prefix_len); | |
445 | let content_end = suffix_start - BytePos(postfix_len); | |
446 | let id = self.symbol_from_to(content_start, content_end); | |
5869c6ff | 447 | self.validate_literal_escape(mode, content_start, content_end, prefix_len, postfix_len); |
f035d41b | 448 | (lit_kind, id) |
f9f354fc XL |
449 | } |
450 | ||
94b46f34 XL |
451 | #[inline] |
452 | fn src_index(&self, pos: BytePos) -> usize { | |
416331ca | 453 | (pos - self.start_pos).to_usize() |
1a4d82fc JJ |
454 | } |
455 | ||
dc9dc135 XL |
456 | /// Slice of the source text from `start` up to but excluding `self.pos`, |
457 | /// meaning the slice does not include the character `self.ch`. | |
dfeec247 | 458 | fn str_from(&self, start: BytePos) -> &str { |
dc9dc135 | 459 | self.str_from_to(start, self.pos) |
1a4d82fc JJ |
460 | } |
461 | ||
dc9dc135 XL |
462 | /// As symbol_from, with an explicit endpoint. |
463 | fn symbol_from_to(&self, start: BytePos, end: BytePos) -> Symbol { | |
1a4d82fc | 464 | debug!("taking an ident from {:?} to {:?}", start, end); |
dc9dc135 | 465 | Symbol::intern(self.str_from_to(start, end)) |
1a4d82fc JJ |
466 | } |
467 | ||
dc9dc135 | 468 | /// Slice of the source text spanning from `start` up to but excluding `end`. |
dfeec247 | 469 | fn str_from_to(&self, start: BytePos, end: BytePos) -> &str { |
dc9dc135 | 470 | &self.src[self.src_index(start)..self.src_index(end)] |
1a4d82fc JJ |
471 | } |
472 | ||
f035d41b XL |
473 | fn report_raw_str_error(&self, start: BytePos, opt_err: Option<RawStrError>) { |
474 | match opt_err { | |
475 | Some(RawStrError::InvalidStarter { bad_char }) => { | |
476 | self.report_non_started_raw_string(start, bad_char) | |
477 | } | |
478 | Some(RawStrError::NoTerminator { expected, found, possible_terminator_offset }) => self | |
479 | .report_unterminated_raw_string(start, expected, possible_terminator_offset, found), | |
480 | Some(RawStrError::TooManyDelimiters { found }) => { | |
481 | self.report_too_many_hashes(start, found) | |
ba9703b0 | 482 | } |
f035d41b | 483 | None => (), |
ba9703b0 XL |
484 | } |
485 | } | |
486 | ||
f035d41b | 487 | fn report_non_started_raw_string(&self, start: BytePos, bad_char: char) -> ! { |
dfeec247 XL |
488 | self.struct_fatal_span_char( |
489 | start, | |
490 | self.pos, | |
f035d41b | 491 | "found invalid character; only `#` is allowed in raw string delimitation", |
dfeec247 XL |
492 | bad_char, |
493 | ) | |
494 | .emit(); | |
416331ca | 495 | FatalError.raise() |
1a4d82fc JJ |
496 | } |
497 | ||
ba9703b0 XL |
498 | fn report_unterminated_raw_string( |
499 | &self, | |
500 | start: BytePos, | |
501 | n_hashes: usize, | |
502 | possible_offset: Option<usize>, | |
503 | found_terminators: usize, | |
504 | ) -> ! { | |
74b04a01 XL |
505 | let mut err = self.sess.span_diagnostic.struct_span_fatal_with_code( |
506 | self.mk_sp(start, start), | |
507 | "unterminated raw string", | |
508 | error_code!(E0748), | |
509 | ); | |
ba9703b0 | 510 | |
dfeec247 | 511 | err.span_label(self.mk_sp(start, start), "unterminated raw string"); |
1a4d82fc | 512 | |
416331ca | 513 | if n_hashes > 0 { |
dfeec247 XL |
514 | err.note(&format!( |
515 | "this raw string should be terminated with `\"{}`", | |
ba9703b0 | 516 | "#".repeat(n_hashes) |
dfeec247 | 517 | )); |
48663c56 | 518 | } |
1a4d82fc | 519 | |
ba9703b0 XL |
520 | if let Some(possible_offset) = possible_offset { |
521 | let lo = start + BytePos(possible_offset as u32); | |
522 | let hi = lo + BytePos(found_terminators as u32); | |
523 | let span = self.mk_sp(lo, hi); | |
524 | err.span_suggestion( | |
525 | span, | |
526 | "consider terminating the string here", | |
527 | "#".repeat(n_hashes), | |
528 | Applicability::MaybeIncorrect, | |
529 | ); | |
530 | } | |
531 | ||
416331ca XL |
532 | err.emit(); |
533 | FatalError.raise() | |
1a4d82fc JJ |
534 | } |
535 | ||
136023e0 XL |
536 | // RFC 3101 introduced the idea of (reserved) prefixes. As of Rust 2021, |
537 | // using a (unknown) prefix is an error. In earlier editions, however, they | |
538 | // only result in a (allowed by default) lint, and are treated as regular | |
539 | // identifier tokens. | |
540 | fn report_unknown_prefix(&self, start: BytePos) { | |
541 | let prefix_span = self.mk_sp(start, self.pos); | |
94222f64 XL |
542 | let prefix_str = self.str_from_to(start, self.pos); |
543 | let msg = format!("prefix `{}` is unknown", prefix_str); | |
136023e0 XL |
544 | |
545 | let expn_data = prefix_span.ctxt().outer_expn_data(); | |
546 | ||
547 | if expn_data.edition >= Edition::Edition2021 { | |
548 | // In Rust 2021, this is a hard error. | |
549 | let mut err = self.sess.span_diagnostic.struct_span_err(prefix_span, &msg); | |
550 | err.span_label(prefix_span, "unknown prefix"); | |
94222f64 XL |
551 | if prefix_str == "rb" { |
552 | err.span_suggestion_verbose( | |
553 | prefix_span, | |
554 | "use `br` for a raw byte string", | |
555 | "br".to_string(), | |
556 | Applicability::MaybeIncorrect, | |
557 | ); | |
558 | } else if expn_data.is_root() { | |
136023e0 XL |
559 | err.span_suggestion_verbose( |
560 | prefix_span.shrink_to_hi(), | |
561 | "consider inserting whitespace here", | |
562 | " ".into(), | |
94222f64 | 563 | Applicability::MaybeIncorrect, |
136023e0 XL |
564 | ); |
565 | } | |
566 | err.note("prefixed identifiers and literals are reserved since Rust 2021"); | |
567 | err.emit(); | |
568 | } else { | |
569 | // Before Rust 2021, only emit a lint for migration. | |
570 | self.sess.buffer_lint_with_diagnostic( | |
571 | &RUST_2021_PREFIXES_INCOMPATIBLE_SYNTAX, | |
572 | prefix_span, | |
573 | ast::CRATE_NODE_ID, | |
574 | &msg, | |
575 | BuiltinLintDiagnostics::ReservedPrefix(prefix_span), | |
576 | ); | |
577 | } | |
578 | } | |
579 | ||
fc512014 | 580 | /// Note: It was decided to not add a test case, because it would be too big. |
29967ef6 | 581 | /// <https://github.com/rust-lang/rust/pull/50296#issuecomment-392135180> |
f035d41b | 582 | fn report_too_many_hashes(&self, start: BytePos, found: usize) -> ! { |
ba9703b0 XL |
583 | self.fatal_span_( |
584 | start, | |
585 | self.pos, | |
f035d41b XL |
586 | &format!( |
587 | "too many `#` symbols: raw strings may be delimited \ | |
588 | by up to 65535 `#` symbols, but found {}", | |
589 | found | |
590 | ), | |
ba9703b0 XL |
591 | ) |
592 | .raise(); | |
1a4d82fc | 593 | } |
48663c56 | 594 | |
5869c6ff XL |
595 | fn validate_literal_escape( |
596 | &self, | |
597 | mode: Mode, | |
598 | content_start: BytePos, | |
599 | content_end: BytePos, | |
600 | prefix_len: u32, | |
601 | postfix_len: u32, | |
602 | ) { | |
f9f354fc XL |
603 | let lit_content = self.str_from_to(content_start, content_end); |
604 | unescape::unescape_literal(lit_content, mode, &mut |range, result| { | |
605 | // Here we only check for errors. The actual unescaping is done later. | |
606 | if let Err(err) = result { | |
5869c6ff XL |
607 | let span_with_quotes = self |
608 | .mk_sp(content_start - BytePos(prefix_len), content_end + BytePos(postfix_len)); | |
fc512014 XL |
609 | let (start, end) = (range.start as u32, range.end as u32); |
610 | let lo = content_start + BytePos(start); | |
611 | let hi = lo + BytePos(end - start); | |
612 | let span = self.mk_sp(lo, hi); | |
dc9dc135 XL |
613 | emit_unescape_error( |
614 | &self.sess.span_diagnostic, | |
f9f354fc XL |
615 | lit_content, |
616 | span_with_quotes, | |
fc512014 | 617 | span, |
f9f354fc | 618 | mode, |
dc9dc135 XL |
619 | range, |
620 | err, | |
f9f354fc | 621 | ); |
dc9dc135 | 622 | } |
f9f354fc | 623 | }); |
48663c56 | 624 | } |
1a4d82fc | 625 | |
416331ca XL |
626 | fn validate_int_literal(&self, base: Base, content_start: BytePos, content_end: BytePos) { |
627 | let base = match base { | |
628 | Base::Binary => 2, | |
629 | Base::Octal => 8, | |
630 | _ => return, | |
631 | }; | |
632 | let s = self.str_from_to(content_start + BytePos(2), content_end); | |
633 | for (idx, c) in s.char_indices() { | |
634 | let idx = idx as u32; | |
635 | if c != '_' && c.to_digit(base).is_none() { | |
636 | let lo = content_start + BytePos(2 + idx); | |
637 | let hi = content_start + BytePos(2 + idx + c.len_utf8() as u32); | |
dfeec247 | 638 | self.err_span_(lo, hi, &format!("invalid digit for a base {} literal", base)); |
416331ca XL |
639 | } |
640 | } | |
641 | } | |
9cc50fc6 | 642 | } |
dfeec247 XL |
643 | |
644 | pub fn nfc_normalize(string: &str) -> Symbol { | |
645 | use unicode_normalization::{is_nfc_quick, IsNormalized, UnicodeNormalization}; | |
646 | match is_nfc_quick(string.chars()) { | |
647 | IsNormalized::Yes => Symbol::intern(string), | |
648 | _ => { | |
649 | let normalized_str: String = string.chars().nfc().collect(); | |
650 | Symbol::intern(&normalized_str) | |
651 | } | |
652 | } | |
653 | } |