]>
Commit | Line | Data |
---|---|---|
3c0e092e | 1 | use crate::lexer::unicode_chars::UNICODE_ARRAY; |
136023e0 | 2 | use rustc_ast::ast::{self, AttrStyle}; |
3dfed10e | 3 | use rustc_ast::token::{self, CommentKind, Token, TokenKind}; |
1b1a35ee | 4 | use rustc_ast::tokenstream::{Spacing, TokenStream}; |
3c0e092e | 5 | use rustc_ast::util::unicode::contains_text_flow_control_chars; |
5e7ed085 | 6 | use rustc_errors::{error_code, Applicability, DiagnosticBuilder, ErrorGuaranteed, PResult}; |
1b1a35ee XL |
7 | use rustc_lexer::unescape::{self, Mode}; |
8 | use rustc_lexer::{Base, DocStyle, RawStrError}; | |
c295e0f8 | 9 | use rustc_session::lint::builtin::{ |
3c0e092e | 10 | RUST_2021_PREFIXES_INCOMPATIBLE_SYNTAX, TEXT_DIRECTION_CODEPOINT_IN_COMMENT, |
c295e0f8 | 11 | }; |
136023e0 | 12 | use rustc_session::lint::BuiltinLintDiagnostics; |
74b04a01 | 13 | use rustc_session::parse::ParseSess; |
dfeec247 | 14 | use rustc_span::symbol::{sym, Symbol}; |
136023e0 | 15 | use rustc_span::{edition::Edition, BytePos, Pos, Span}; |
1a4d82fc | 16 | |
3dfed10e | 17 | use tracing::debug; |
1a4d82fc | 18 | |
32a655c1 | 19 | mod tokentrees; |
60c5eb7d | 20 | mod unescape_error_reporting; |
dfeec247 | 21 | mod unicode_chars; |
ba9703b0 | 22 | |
5869c6ff | 23 | use unescape_error_reporting::{emit_unescape_error, escaped_char}; |
1a4d82fc | 24 | |
9fa01778 XL |
25 | #[derive(Clone, Debug)] |
26 | pub struct UnmatchedBrace { | |
27 | pub expected_delim: token::DelimToken, | |
e74abb32 | 28 | pub found_delim: Option<token::DelimToken>, |
9fa01778 XL |
29 | pub found_span: Span, |
30 | pub unclosed_span: Option<Span>, | |
31 | pub candidate_span: Option<Span>, | |
32 | } | |
33 | ||
1b1a35ee XL |
34 | crate fn parse_token_trees<'a>( |
35 | sess: &'a ParseSess, | |
36 | src: &'a str, | |
37 | start_pos: BytePos, | |
38 | override_span: Option<Span>, | |
39 | ) -> (PResult<'a, TokenStream>, Vec<UnmatchedBrace>) { | |
40 | StringReader { sess, start_pos, pos: start_pos, end_src_index: src.len(), src, override_span } | |
41 | .into_token_trees() | |
42 | } | |
43 | ||
44 | struct StringReader<'a> { | |
416331ca XL |
45 | sess: &'a ParseSess, |
46 | /// Initial position, read-only. | |
47 | start_pos: BytePos, | |
48 | /// The absolute offset within the source_map of the current character. | |
f9f354fc | 49 | pos: BytePos, |
94b46f34 | 50 | /// Stop reading src at this index. |
416331ca XL |
51 | end_src_index: usize, |
52 | /// Source text to tokenize. | |
1b1a35ee | 53 | src: &'a str, |
48663c56 | 54 | override_span: Option<Span>, |
cc61c64b XL |
55 | } |
56 | ||
32a655c1 | 57 | impl<'a> StringReader<'a> { |
416331ca | 58 | fn mk_sp(&self, lo: BytePos, hi: BytePos) -> Span { |
e1599b0c | 59 | self.override_span.unwrap_or_else(|| Span::with_root_ctxt(lo, hi)) |
8bb4bdeb XL |
60 | } |
61 | ||
1b1a35ee XL |
62 | /// Returns the next token, and info about preceding whitespace, if any. |
63 | fn next_token(&mut self) -> (Spacing, Token) { | |
64 | let mut spacing = Spacing::Joint; | |
65 | ||
66 | // Skip `#!` at the start of the file | |
416331ca XL |
67 | let start_src_index = self.src_index(self.pos); |
68 | let text: &str = &self.src[start_src_index..self.end_src_index]; | |
1b1a35ee XL |
69 | let is_beginning_of_file = self.pos == self.start_pos; |
70 | if is_beginning_of_file { | |
71 | if let Some(shebang_len) = rustc_lexer::strip_shebang(text) { | |
72 | self.pos = self.pos + BytePos::from_usize(shebang_len); | |
73 | spacing = Spacing::Alone; | |
74 | } | |
416331ca XL |
75 | } |
76 | ||
1b1a35ee XL |
77 | // Skip trivial (whitespace & comments) tokens |
78 | loop { | |
79 | let start_src_index = self.src_index(self.pos); | |
80 | let text: &str = &self.src[start_src_index..self.end_src_index]; | |
416331ca | 81 | |
1b1a35ee XL |
82 | if text.is_empty() { |
83 | let span = self.mk_sp(self.pos, self.pos); | |
84 | return (spacing, Token::new(token::Eof, span)); | |
416331ca | 85 | } |
416331ca | 86 | |
1b1a35ee | 87 | let token = rustc_lexer::first_token(text); |
416331ca | 88 | |
1b1a35ee XL |
89 | let start = self.pos; |
90 | self.pos = self.pos + BytePos::from_usize(token.len); | |
416331ca | 91 | |
1b1a35ee | 92 | debug!("next_token: {:?}({:?})", token.kind, self.str_from(start)); |
416331ca | 93 | |
1b1a35ee XL |
94 | match self.cook_lexer_token(token.kind, start) { |
95 | Some(kind) => { | |
96 | let span = self.mk_sp(start, self.pos); | |
97 | return (spacing, Token::new(kind, span)); | |
98 | } | |
99 | None => spacing = Spacing::Alone, | |
100 | } | |
101 | } | |
1a4d82fc JJ |
102 | } |
103 | ||
104 | /// Report a fatal lexical error with a given span. | |
5e7ed085 | 105 | fn fatal_span(&self, sp: Span, m: &str) -> ! { |
32a655c1 | 106 | self.sess.span_diagnostic.span_fatal(sp, m) |
1a4d82fc JJ |
107 | } |
108 | ||
109 | /// Report a lexical error with a given span. | |
94b46f34 | 110 | fn err_span(&self, sp: Span, m: &str) { |
0731742a | 111 | self.sess.span_diagnostic.struct_span_err(sp, m).emit(); |
1a4d82fc JJ |
112 | } |
113 | ||
114 | /// Report a fatal error spanning [`from_pos`, `to_pos`). | |
5e7ed085 | 115 | fn fatal_span_(&self, from_pos: BytePos, to_pos: BytePos, m: &str) -> ! { |
041b39d2 | 116 | self.fatal_span(self.mk_sp(from_pos, to_pos), m) |
1a4d82fc JJ |
117 | } |
118 | ||
119 | /// Report a lexical error spanning [`from_pos`, `to_pos`). | |
120 | fn err_span_(&self, from_pos: BytePos, to_pos: BytePos, m: &str) { | |
041b39d2 | 121 | self.err_span(self.mk_sp(from_pos, to_pos), m) |
1a4d82fc JJ |
122 | } |
123 | ||
dfeec247 XL |
124 | fn struct_fatal_span_char( |
125 | &self, | |
126 | from_pos: BytePos, | |
127 | to_pos: BytePos, | |
128 | m: &str, | |
129 | c: char, | |
5e7ed085 | 130 | ) -> DiagnosticBuilder<'a, !> { |
5869c6ff XL |
131 | self.sess |
132 | .span_diagnostic | |
133 | .struct_span_fatal(self.mk_sp(from_pos, to_pos), &format!("{}: {}", m, escaped_char(c))) | |
9cc50fc6 | 134 | } |
1a4d82fc | 135 | |
5e7ed085 FG |
136 | fn struct_err_span_char( |
137 | &self, | |
138 | from_pos: BytePos, | |
139 | to_pos: BytePos, | |
140 | m: &str, | |
141 | c: char, | |
142 | ) -> DiagnosticBuilder<'a, ErrorGuaranteed> { | |
143 | self.sess | |
144 | .span_diagnostic | |
145 | .struct_span_err(self.mk_sp(from_pos, to_pos), &format!("{}: {}", m, escaped_char(c))) | |
146 | } | |
147 | ||
c295e0f8 XL |
148 | /// Detect usages of Unicode codepoints changing the direction of the text on screen and loudly |
149 | /// complain about it. | |
150 | fn lint_unicode_text_flow(&self, start: BytePos) { | |
151 | // Opening delimiter of the length 2 is not included into the comment text. | |
152 | let content_start = start + BytePos(2); | |
153 | let content = self.str_from(content_start); | |
3c0e092e XL |
154 | if contains_text_flow_control_chars(content) { |
155 | let span = self.mk_sp(start, self.pos); | |
c295e0f8 XL |
156 | self.sess.buffer_lint_with_diagnostic( |
157 | &TEXT_DIRECTION_CODEPOINT_IN_COMMENT, | |
158 | span, | |
159 | ast::CRATE_NODE_ID, | |
160 | "unicode codepoint changing visible direction of text present in comment", | |
161 | BuiltinLintDiagnostics::UnicodeTextFlow(span, content.to_string()), | |
162 | ); | |
163 | } | |
164 | } | |
165 | ||
416331ca | 166 | /// Turns simple `rustc_lexer::TokenKind` enum into a rich |
cdc7bbd5 | 167 | /// `rustc_ast::TokenKind`. This turns strings into interned |
416331ca | 168 | /// symbols and runs additional validation. |
1b1a35ee XL |
169 | fn cook_lexer_token(&self, token: rustc_lexer::TokenKind, start: BytePos) -> Option<TokenKind> { |
170 | Some(match token { | |
3dfed10e | 171 | rustc_lexer::TokenKind::LineComment { doc_style } => { |
1b1a35ee | 172 | // Skip non-doc comments |
5099ac24 | 173 | let Some(doc_style) = doc_style else { |
c295e0f8 XL |
174 | self.lint_unicode_text_flow(start); |
175 | return None; | |
176 | }; | |
3dfed10e | 177 | |
1b1a35ee XL |
178 | // Opening delimiter of the length 3 is not included into the symbol. |
179 | let content_start = start + BytePos(3); | |
180 | let content = self.str_from(content_start); | |
181 | self.cook_doc_comment(content_start, content, CommentKind::Line, doc_style) | |
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; | |
17df50a5 XL |
190 | self.sess.span_diagnostic.span_fatal_with_code( |
191 | self.mk_sp(start, last_bpos), | |
192 | msg, | |
193 | error_code!(E0758), | |
194 | ); | |
416331ca | 195 | } |
1b1a35ee XL |
196 | |
197 | // Skip non-doc comments | |
5099ac24 | 198 | let Some(doc_style) = doc_style else { |
c295e0f8 XL |
199 | self.lint_unicode_text_flow(start); |
200 | return None; | |
201 | }; | |
1b1a35ee XL |
202 | |
203 | // Opening delimiter of the length 3 and closing delimiter of the length 2 | |
204 | // are not included into the symbol. | |
205 | let content_start = start + BytePos(3); | |
206 | let content_end = self.pos - BytePos(if terminated { 2 } else { 0 }); | |
207 | let content = self.str_from_to(content_start, content_end); | |
208 | self.cook_doc_comment(content_start, content, CommentKind::Block, doc_style) | |
9cc50fc6 | 209 | } |
1b1a35ee | 210 | rustc_lexer::TokenKind::Whitespace => return None, |
136023e0 XL |
211 | rustc_lexer::TokenKind::Ident |
212 | | rustc_lexer::TokenKind::RawIdent | |
213 | | rustc_lexer::TokenKind::UnknownPrefix => { | |
416331ca | 214 | let is_raw_ident = token == rustc_lexer::TokenKind::RawIdent; |
136023e0 | 215 | let is_unknown_prefix = token == rustc_lexer::TokenKind::UnknownPrefix; |
416331ca XL |
216 | let mut ident_start = start; |
217 | if is_raw_ident { | |
218 | ident_start = ident_start + BytePos(2); | |
219 | } | |
136023e0 XL |
220 | if is_unknown_prefix { |
221 | self.report_unknown_prefix(start); | |
222 | } | |
dfeec247 | 223 | let sym = nfc_normalize(self.str_from(ident_start)); |
f9f354fc XL |
224 | let span = self.mk_sp(start, self.pos); |
225 | self.sess.symbol_gallery.insert(sym, span); | |
416331ca | 226 | if is_raw_ident { |
416331ca XL |
227 | if !sym.can_be_raw() { |
228 | self.err_span(span, &format!("`{}` cannot be a raw identifier", sym)); | |
229 | } | |
230 | self.sess.raw_identifier_spans.borrow_mut().push(span); | |
231 | } | |
232 | token::Ident(sym, is_raw_ident) | |
233 | } | |
3c0e092e XL |
234 | rustc_lexer::TokenKind::InvalidIdent |
235 | // Do not recover an identifier with emoji if the codepoint is a confusable | |
236 | // with a recoverable substitution token, like `➖`. | |
237 | if UNICODE_ARRAY | |
238 | .iter() | |
239 | .find(|&&(c, _, _)| { | |
240 | let sym = self.str_from(start); | |
241 | sym.chars().count() == 1 && c == sym.chars().next().unwrap() | |
242 | }) | |
243 | .is_none() => | |
244 | { | |
245 | let sym = nfc_normalize(self.str_from(start)); | |
246 | let span = self.mk_sp(start, self.pos); | |
247 | self.sess.bad_unicode_identifiers.borrow_mut().entry(sym).or_default().push(span); | |
248 | token::Ident(sym, false) | |
249 | } | |
416331ca XL |
250 | rustc_lexer::TokenKind::Literal { kind, suffix_start } => { |
251 | let suffix_start = start + BytePos(suffix_start as u32); | |
252 | let (kind, symbol) = self.cook_lexer_literal(start, suffix_start, kind); | |
253 | let suffix = if suffix_start < self.pos { | |
254 | let string = self.str_from(suffix_start); | |
255 | if string == "_" { | |
dfeec247 XL |
256 | self.sess |
257 | .span_diagnostic | |
258 | .struct_span_warn( | |
259 | self.mk_sp(suffix_start, self.pos), | |
260 | "underscore literal suffix is not allowed", | |
261 | ) | |
262 | .warn( | |
263 | "this was previously accepted by the compiler but is \ | |
416331ca | 264 | being phased out; it will become a hard error in \ |
dfeec247 XL |
265 | a future release!", |
266 | ) | |
267 | .note( | |
74b04a01 XL |
268 | "see issue #42326 \ |
269 | <https://github.com/rust-lang/rust/issues/42326> \ | |
270 | for more information", | |
dfeec247 | 271 | ) |
416331ca XL |
272 | .emit(); |
273 | None | |
274 | } else { | |
275 | Some(Symbol::intern(string)) | |
276 | } | |
1a4d82fc | 277 | } else { |
416331ca | 278 | None |
1a4d82fc | 279 | }; |
416331ca XL |
280 | token::Literal(token::Lit { kind, symbol, suffix }) |
281 | } | |
282 | rustc_lexer::TokenKind::Lifetime { starts_with_number } => { | |
283 | // Include the leading `'` in the real identifier, for macro | |
284 | // expansion purposes. See #12512 for the gory details of why | |
285 | // this is necessary. | |
286 | let lifetime_name = self.str_from(start); | |
287 | if starts_with_number { | |
dfeec247 | 288 | self.err_span_(start, self.pos, "lifetimes cannot start with a number"); |
416331ca XL |
289 | } |
290 | let ident = Symbol::intern(lifetime_name); | |
291 | token::Lifetime(ident) | |
292 | } | |
293 | rustc_lexer::TokenKind::Semi => token::Semi, | |
294 | rustc_lexer::TokenKind::Comma => token::Comma, | |
416331ca XL |
295 | rustc_lexer::TokenKind::Dot => token::Dot, |
296 | rustc_lexer::TokenKind::OpenParen => token::OpenDelim(token::Paren), | |
297 | rustc_lexer::TokenKind::CloseParen => token::CloseDelim(token::Paren), | |
298 | rustc_lexer::TokenKind::OpenBrace => token::OpenDelim(token::Brace), | |
299 | rustc_lexer::TokenKind::CloseBrace => token::CloseDelim(token::Brace), | |
300 | rustc_lexer::TokenKind::OpenBracket => token::OpenDelim(token::Bracket), | |
301 | rustc_lexer::TokenKind::CloseBracket => token::CloseDelim(token::Bracket), | |
302 | rustc_lexer::TokenKind::At => token::At, | |
303 | rustc_lexer::TokenKind::Pound => token::Pound, | |
304 | rustc_lexer::TokenKind::Tilde => token::Tilde, | |
305 | rustc_lexer::TokenKind::Question => token::Question, | |
416331ca XL |
306 | rustc_lexer::TokenKind::Colon => token::Colon, |
307 | rustc_lexer::TokenKind::Dollar => token::Dollar, | |
416331ca | 308 | rustc_lexer::TokenKind::Eq => token::Eq, |
3dfed10e | 309 | rustc_lexer::TokenKind::Bang => token::Not, |
416331ca | 310 | rustc_lexer::TokenKind::Lt => token::Lt, |
416331ca | 311 | rustc_lexer::TokenKind::Gt => token::Gt, |
416331ca | 312 | rustc_lexer::TokenKind::Minus => token::BinOp(token::Minus), |
416331ca | 313 | rustc_lexer::TokenKind::And => token::BinOp(token::And), |
416331ca | 314 | rustc_lexer::TokenKind::Or => token::BinOp(token::Or), |
416331ca | 315 | rustc_lexer::TokenKind::Plus => token::BinOp(token::Plus), |
416331ca | 316 | rustc_lexer::TokenKind::Star => token::BinOp(token::Star), |
416331ca | 317 | rustc_lexer::TokenKind::Slash => token::BinOp(token::Slash), |
416331ca | 318 | rustc_lexer::TokenKind::Caret => token::BinOp(token::Caret), |
416331ca | 319 | rustc_lexer::TokenKind::Percent => token::BinOp(token::Percent), |
416331ca | 320 | |
3c0e092e | 321 | rustc_lexer::TokenKind::Unknown | rustc_lexer::TokenKind::InvalidIdent => { |
416331ca | 322 | let c = self.str_from(start).chars().next().unwrap(); |
dfeec247 | 323 | let mut err = |
5e7ed085 | 324 | self.struct_err_span_char(start, self.pos, "unknown start of token", c); |
416331ca XL |
325 | // FIXME: the lexer could be used to turn the ASCII version of unicode homoglyphs, |
326 | // instead of keeping a table in `check_for_substitution`into the token. Ideally, | |
327 | // this should be inside `rustc_lexer`. However, we should first remove compound | |
328 | // tokens like `<<` from `rustc_lexer`, and then add fancier error recovery to it, | |
329 | // as there will be less overall work to do this way. | |
1b1a35ee | 330 | let token = unicode_chars::check_for_substitution(self, start, c, &mut err); |
6a06907d XL |
331 | if c == '\x00' { |
332 | err.help("source files must contain UTF-8 encoded text, unexpected null bytes might occur when a different encoding is used"); | |
333 | } | |
416331ca | 334 | err.emit(); |
1b1a35ee | 335 | token? |
1a4d82fc | 336 | } |
1b1a35ee | 337 | }) |
416331ca | 338 | } |
b7449926 | 339 | |
3dfed10e XL |
340 | fn cook_doc_comment( |
341 | &self, | |
342 | content_start: BytePos, | |
343 | content: &str, | |
344 | comment_kind: CommentKind, | |
345 | doc_style: DocStyle, | |
346 | ) -> TokenKind { | |
347 | if content.contains('\r') { | |
348 | for (idx, _) in content.char_indices().filter(|&(_, c)| c == '\r') { | |
349 | self.err_span_( | |
350 | content_start + BytePos(idx as u32), | |
351 | content_start + BytePos(idx as u32 + 1), | |
352 | match comment_kind { | |
353 | CommentKind::Line => "bare CR not allowed in doc-comment", | |
354 | CommentKind::Block => "bare CR not allowed in block doc-comment", | |
355 | }, | |
356 | ); | |
357 | } | |
358 | } | |
359 | ||
360 | let attr_style = match doc_style { | |
361 | DocStyle::Outer => AttrStyle::Outer, | |
362 | DocStyle::Inner => AttrStyle::Inner, | |
363 | }; | |
364 | ||
365 | token::DocComment(comment_kind, attr_style, Symbol::intern(content)) | |
366 | } | |
367 | ||
416331ca XL |
368 | fn cook_lexer_literal( |
369 | &self, | |
370 | start: BytePos, | |
371 | suffix_start: BytePos, | |
dfeec247 | 372 | kind: rustc_lexer::LiteralKind, |
416331ca | 373 | ) -> (token::LitKind, Symbol) { |
f9f354fc XL |
374 | // prefix means `"` or `br"` or `r###"`, ... |
375 | let (lit_kind, mode, prefix_len, postfix_len) = match kind { | |
416331ca XL |
376 | rustc_lexer::LiteralKind::Char { terminated } => { |
377 | if !terminated { | |
17df50a5 XL |
378 | self.sess.span_diagnostic.span_fatal_with_code( |
379 | self.mk_sp(start, suffix_start), | |
380 | "unterminated character literal", | |
381 | error_code!(E0762), | |
382 | ) | |
416331ca | 383 | } |
f9f354fc | 384 | (token::Char, Mode::Char, 1, 1) // ' ' |
dfeec247 | 385 | } |
416331ca XL |
386 | rustc_lexer::LiteralKind::Byte { terminated } => { |
387 | if !terminated { | |
17df50a5 XL |
388 | self.sess.span_diagnostic.span_fatal_with_code( |
389 | self.mk_sp(start + BytePos(1), suffix_start), | |
390 | "unterminated byte constant", | |
391 | error_code!(E0763), | |
392 | ) | |
416331ca | 393 | } |
f9f354fc | 394 | (token::Byte, Mode::Byte, 2, 1) // b' ' |
dfeec247 | 395 | } |
416331ca XL |
396 | rustc_lexer::LiteralKind::Str { terminated } => { |
397 | if !terminated { | |
17df50a5 XL |
398 | self.sess.span_diagnostic.span_fatal_with_code( |
399 | self.mk_sp(start, suffix_start), | |
400 | "unterminated double quote string", | |
401 | error_code!(E0765), | |
402 | ) | |
416331ca | 403 | } |
f9f354fc | 404 | (token::Str, Mode::Str, 1, 1) // " " |
416331ca XL |
405 | } |
406 | rustc_lexer::LiteralKind::ByteStr { terminated } => { | |
407 | if !terminated { | |
17df50a5 XL |
408 | self.sess.span_diagnostic.span_fatal_with_code( |
409 | self.mk_sp(start + BytePos(1), suffix_start), | |
410 | "unterminated double quote byte string", | |
411 | error_code!(E0766), | |
412 | ) | |
416331ca | 413 | } |
f9f354fc | 414 | (token::ByteStr, Mode::ByteStr, 2, 1) // b" " |
416331ca | 415 | } |
f035d41b XL |
416 | rustc_lexer::LiteralKind::RawStr { n_hashes, err } => { |
417 | self.report_raw_str_error(start, err); | |
416331ca | 418 | let n = u32::from(n_hashes); |
f9f354fc | 419 | (token::StrRaw(n_hashes), Mode::RawStr, 2 + n, 1 + n) // r##" "## |
416331ca | 420 | } |
f035d41b XL |
421 | rustc_lexer::LiteralKind::RawByteStr { n_hashes, err } => { |
422 | self.report_raw_str_error(start, err); | |
416331ca | 423 | let n = u32::from(n_hashes); |
f9f354fc | 424 | (token::ByteStrRaw(n_hashes), Mode::RawByteStr, 3 + n, 1 + n) // br##" "## |
416331ca XL |
425 | } |
426 | rustc_lexer::LiteralKind::Int { base, empty_int } => { | |
f9f354fc | 427 | return if empty_int { |
f035d41b XL |
428 | self.sess |
429 | .span_diagnostic | |
430 | .struct_span_err_with_code( | |
431 | self.mk_sp(start, suffix_start), | |
432 | "no valid digits found for number", | |
433 | error_code!(E0768), | |
434 | ) | |
435 | .emit(); | |
416331ca XL |
436 | (token::Integer, sym::integer(0)) |
437 | } else { | |
438 | self.validate_int_literal(base, start, suffix_start); | |
439 | (token::Integer, self.symbol_from_to(start, suffix_start)) | |
f9f354fc | 440 | }; |
dfeec247 | 441 | } |
416331ca XL |
442 | rustc_lexer::LiteralKind::Float { base, empty_exponent } => { |
443 | if empty_exponent { | |
f035d41b | 444 | self.err_span_(start, self.pos, "expected at least one digit in exponent"); |
416331ca XL |
445 | } |
446 | ||
447 | match base { | |
dfeec247 XL |
448 | Base::Hexadecimal => self.err_span_( |
449 | start, | |
450 | suffix_start, | |
451 | "hexadecimal float literal is not supported", | |
452 | ), | |
416331ca | 453 | Base::Octal => { |
dfeec247 | 454 | self.err_span_(start, suffix_start, "octal float literal is not supported") |
416331ca XL |
455 | } |
456 | Base::Binary => { | |
dfeec247 | 457 | self.err_span_(start, suffix_start, "binary float literal is not supported") |
416331ca | 458 | } |
dfeec247 | 459 | _ => (), |
416331ca XL |
460 | } |
461 | ||
462 | let id = self.symbol_from_to(start, suffix_start); | |
f9f354fc | 463 | return (token::Float, id); |
dfeec247 | 464 | } |
f9f354fc XL |
465 | }; |
466 | let content_start = start + BytePos(prefix_len); | |
467 | let content_end = suffix_start - BytePos(postfix_len); | |
468 | let id = self.symbol_from_to(content_start, content_end); | |
5869c6ff | 469 | self.validate_literal_escape(mode, content_start, content_end, prefix_len, postfix_len); |
f035d41b | 470 | (lit_kind, id) |
f9f354fc XL |
471 | } |
472 | ||
94b46f34 XL |
473 | #[inline] |
474 | fn src_index(&self, pos: BytePos) -> usize { | |
416331ca | 475 | (pos - self.start_pos).to_usize() |
1a4d82fc JJ |
476 | } |
477 | ||
dc9dc135 XL |
478 | /// Slice of the source text from `start` up to but excluding `self.pos`, |
479 | /// meaning the slice does not include the character `self.ch`. | |
dfeec247 | 480 | fn str_from(&self, start: BytePos) -> &str { |
dc9dc135 | 481 | self.str_from_to(start, self.pos) |
1a4d82fc JJ |
482 | } |
483 | ||
dc9dc135 XL |
484 | /// As symbol_from, with an explicit endpoint. |
485 | fn symbol_from_to(&self, start: BytePos, end: BytePos) -> Symbol { | |
1a4d82fc | 486 | debug!("taking an ident from {:?} to {:?}", start, end); |
dc9dc135 | 487 | Symbol::intern(self.str_from_to(start, end)) |
1a4d82fc JJ |
488 | } |
489 | ||
dc9dc135 | 490 | /// Slice of the source text spanning from `start` up to but excluding `end`. |
dfeec247 | 491 | fn str_from_to(&self, start: BytePos, end: BytePos) -> &str { |
dc9dc135 | 492 | &self.src[self.src_index(start)..self.src_index(end)] |
1a4d82fc JJ |
493 | } |
494 | ||
f035d41b XL |
495 | fn report_raw_str_error(&self, start: BytePos, opt_err: Option<RawStrError>) { |
496 | match opt_err { | |
497 | Some(RawStrError::InvalidStarter { bad_char }) => { | |
498 | self.report_non_started_raw_string(start, bad_char) | |
499 | } | |
500 | Some(RawStrError::NoTerminator { expected, found, possible_terminator_offset }) => self | |
501 | .report_unterminated_raw_string(start, expected, possible_terminator_offset, found), | |
502 | Some(RawStrError::TooManyDelimiters { found }) => { | |
503 | self.report_too_many_hashes(start, found) | |
ba9703b0 | 504 | } |
f035d41b | 505 | None => (), |
ba9703b0 XL |
506 | } |
507 | } | |
508 | ||
f035d41b | 509 | fn report_non_started_raw_string(&self, start: BytePos, bad_char: char) -> ! { |
dfeec247 XL |
510 | self.struct_fatal_span_char( |
511 | start, | |
512 | self.pos, | |
f035d41b | 513 | "found invalid character; only `#` is allowed in raw string delimitation", |
dfeec247 XL |
514 | bad_char, |
515 | ) | |
5e7ed085 | 516 | .emit() |
1a4d82fc JJ |
517 | } |
518 | ||
ba9703b0 XL |
519 | fn report_unterminated_raw_string( |
520 | &self, | |
521 | start: BytePos, | |
522 | n_hashes: usize, | |
523 | possible_offset: Option<usize>, | |
524 | found_terminators: usize, | |
525 | ) -> ! { | |
74b04a01 XL |
526 | let mut err = self.sess.span_diagnostic.struct_span_fatal_with_code( |
527 | self.mk_sp(start, start), | |
528 | "unterminated raw string", | |
529 | error_code!(E0748), | |
530 | ); | |
ba9703b0 | 531 | |
dfeec247 | 532 | err.span_label(self.mk_sp(start, start), "unterminated raw string"); |
1a4d82fc | 533 | |
416331ca | 534 | if n_hashes > 0 { |
dfeec247 XL |
535 | err.note(&format!( |
536 | "this raw string should be terminated with `\"{}`", | |
ba9703b0 | 537 | "#".repeat(n_hashes) |
dfeec247 | 538 | )); |
48663c56 | 539 | } |
1a4d82fc | 540 | |
ba9703b0 XL |
541 | if let Some(possible_offset) = possible_offset { |
542 | let lo = start + BytePos(possible_offset as u32); | |
543 | let hi = lo + BytePos(found_terminators as u32); | |
544 | let span = self.mk_sp(lo, hi); | |
545 | err.span_suggestion( | |
546 | span, | |
547 | "consider terminating the string here", | |
548 | "#".repeat(n_hashes), | |
549 | Applicability::MaybeIncorrect, | |
550 | ); | |
551 | } | |
552 | ||
5e7ed085 | 553 | err.emit() |
1a4d82fc JJ |
554 | } |
555 | ||
136023e0 XL |
556 | // RFC 3101 introduced the idea of (reserved) prefixes. As of Rust 2021, |
557 | // using a (unknown) prefix is an error. In earlier editions, however, they | |
558 | // only result in a (allowed by default) lint, and are treated as regular | |
559 | // identifier tokens. | |
560 | fn report_unknown_prefix(&self, start: BytePos) { | |
561 | let prefix_span = self.mk_sp(start, self.pos); | |
94222f64 XL |
562 | let prefix_str = self.str_from_to(start, self.pos); |
563 | let msg = format!("prefix `{}` is unknown", prefix_str); | |
136023e0 XL |
564 | |
565 | let expn_data = prefix_span.ctxt().outer_expn_data(); | |
566 | ||
567 | if expn_data.edition >= Edition::Edition2021 { | |
568 | // In Rust 2021, this is a hard error. | |
569 | let mut err = self.sess.span_diagnostic.struct_span_err(prefix_span, &msg); | |
570 | err.span_label(prefix_span, "unknown prefix"); | |
94222f64 XL |
571 | if prefix_str == "rb" { |
572 | err.span_suggestion_verbose( | |
573 | prefix_span, | |
574 | "use `br` for a raw byte string", | |
575 | "br".to_string(), | |
576 | Applicability::MaybeIncorrect, | |
577 | ); | |
578 | } else if expn_data.is_root() { | |
136023e0 XL |
579 | err.span_suggestion_verbose( |
580 | prefix_span.shrink_to_hi(), | |
581 | "consider inserting whitespace here", | |
582 | " ".into(), | |
94222f64 | 583 | Applicability::MaybeIncorrect, |
136023e0 XL |
584 | ); |
585 | } | |
586 | err.note("prefixed identifiers and literals are reserved since Rust 2021"); | |
587 | err.emit(); | |
588 | } else { | |
589 | // Before Rust 2021, only emit a lint for migration. | |
590 | self.sess.buffer_lint_with_diagnostic( | |
591 | &RUST_2021_PREFIXES_INCOMPATIBLE_SYNTAX, | |
592 | prefix_span, | |
593 | ast::CRATE_NODE_ID, | |
594 | &msg, | |
595 | BuiltinLintDiagnostics::ReservedPrefix(prefix_span), | |
596 | ); | |
597 | } | |
598 | } | |
599 | ||
f035d41b | 600 | fn report_too_many_hashes(&self, start: BytePos, found: usize) -> ! { |
ba9703b0 XL |
601 | self.fatal_span_( |
602 | start, | |
603 | self.pos, | |
f035d41b XL |
604 | &format!( |
605 | "too many `#` symbols: raw strings may be delimited \ | |
5e7ed085 | 606 | by up to 255 `#` symbols, but found {}", |
f035d41b XL |
607 | found |
608 | ), | |
ba9703b0 | 609 | ) |
1a4d82fc | 610 | } |
48663c56 | 611 | |
5869c6ff XL |
612 | fn validate_literal_escape( |
613 | &self, | |
614 | mode: Mode, | |
615 | content_start: BytePos, | |
616 | content_end: BytePos, | |
617 | prefix_len: u32, | |
618 | postfix_len: u32, | |
619 | ) { | |
f9f354fc XL |
620 | let lit_content = self.str_from_to(content_start, content_end); |
621 | unescape::unescape_literal(lit_content, mode, &mut |range, result| { | |
622 | // Here we only check for errors. The actual unescaping is done later. | |
623 | if let Err(err) = result { | |
5869c6ff XL |
624 | let span_with_quotes = self |
625 | .mk_sp(content_start - BytePos(prefix_len), content_end + BytePos(postfix_len)); | |
fc512014 XL |
626 | let (start, end) = (range.start as u32, range.end as u32); |
627 | let lo = content_start + BytePos(start); | |
628 | let hi = lo + BytePos(end - start); | |
629 | let span = self.mk_sp(lo, hi); | |
dc9dc135 XL |
630 | emit_unescape_error( |
631 | &self.sess.span_diagnostic, | |
f9f354fc XL |
632 | lit_content, |
633 | span_with_quotes, | |
fc512014 | 634 | span, |
f9f354fc | 635 | mode, |
dc9dc135 XL |
636 | range, |
637 | err, | |
f9f354fc | 638 | ); |
dc9dc135 | 639 | } |
f9f354fc | 640 | }); |
48663c56 | 641 | } |
1a4d82fc | 642 | |
416331ca XL |
643 | fn validate_int_literal(&self, base: Base, content_start: BytePos, content_end: BytePos) { |
644 | let base = match base { | |
645 | Base::Binary => 2, | |
646 | Base::Octal => 8, | |
647 | _ => return, | |
648 | }; | |
649 | let s = self.str_from_to(content_start + BytePos(2), content_end); | |
650 | for (idx, c) in s.char_indices() { | |
651 | let idx = idx as u32; | |
652 | if c != '_' && c.to_digit(base).is_none() { | |
653 | let lo = content_start + BytePos(2 + idx); | |
654 | let hi = content_start + BytePos(2 + idx + c.len_utf8() as u32); | |
dfeec247 | 655 | self.err_span_(lo, hi, &format!("invalid digit for a base {} literal", base)); |
416331ca XL |
656 | } |
657 | } | |
658 | } | |
9cc50fc6 | 659 | } |
dfeec247 XL |
660 | |
661 | pub fn nfc_normalize(string: &str) -> Symbol { | |
662 | use unicode_normalization::{is_nfc_quick, IsNormalized, UnicodeNormalization}; | |
663 | match is_nfc_quick(string.chars()) { | |
664 | IsNormalized::Yes => Symbol::intern(string), | |
665 | _ => { | |
666 | let normalized_str: String = string.chars().nfc().collect(); | |
667 | Symbol::intern(&normalized_str) | |
668 | } | |
669 | } | |
670 | } |