ra_ap_syntax/ast/
token_ext.rs

1//! There are many AstNodes, but only a few tokens, so we hand-write them here.
2
3use std::ops::Range;
4use std::{borrow::Cow, num::ParseIntError};
5
6use rustc_literal_escaper::{
7    EscapeError, MixedUnit, unescape_byte, unescape_byte_str, unescape_c_str, unescape_char,
8    unescape_str,
9};
10use stdx::always;
11
12use crate::{
13    TextRange, TextSize,
14    ast::{self, AstToken},
15};
16
17impl ast::Comment {
18    pub fn kind(&self) -> CommentKind {
19        CommentKind::from_text(self.text())
20    }
21
22    pub fn is_doc(&self) -> bool {
23        self.kind().doc.is_some()
24    }
25
26    pub fn is_inner(&self) -> bool {
27        self.kind().doc == Some(CommentPlacement::Inner)
28    }
29
30    pub fn is_outer(&self) -> bool {
31        self.kind().doc == Some(CommentPlacement::Outer)
32    }
33
34    pub fn prefix(&self) -> &'static str {
35        let &(prefix, _kind) = CommentKind::BY_PREFIX
36            .iter()
37            .find(|&(prefix, kind)| self.kind() == *kind && self.text().starts_with(prefix))
38            .unwrap();
39        prefix
40    }
41
42    /// Returns the textual content of a doc comment node as a single string with prefix and suffix
43    /// removed.
44    pub fn doc_comment(&self) -> Option<&str> {
45        let kind = self.kind();
46        match kind {
47            CommentKind { shape, doc: Some(_) } => {
48                let prefix = kind.prefix();
49                let text = &self.text()[prefix.len()..];
50                let text = if shape == CommentShape::Block {
51                    text.strip_suffix("*/").unwrap_or(text)
52                } else {
53                    text
54                };
55                Some(text)
56            }
57            _ => None,
58        }
59    }
60}
61
62#[derive(Debug, PartialEq, Eq, Clone, Copy)]
63pub struct CommentKind {
64    pub shape: CommentShape,
65    pub doc: Option<CommentPlacement>,
66}
67
68#[derive(Debug, PartialEq, Eq, Clone, Copy)]
69pub enum CommentShape {
70    Line,
71    Block,
72}
73
74impl CommentShape {
75    pub fn is_line(self) -> bool {
76        self == CommentShape::Line
77    }
78
79    pub fn is_block(self) -> bool {
80        self == CommentShape::Block
81    }
82}
83
84#[derive(Debug, PartialEq, Eq, Clone, Copy)]
85pub enum CommentPlacement {
86    Inner,
87    Outer,
88}
89
90impl CommentKind {
91    const BY_PREFIX: [(&'static str, CommentKind); 9] = [
92        ("/**/", CommentKind { shape: CommentShape::Block, doc: None }),
93        ("/***", CommentKind { shape: CommentShape::Block, doc: None }),
94        ("////", CommentKind { shape: CommentShape::Line, doc: None }),
95        ("///", CommentKind { shape: CommentShape::Line, doc: Some(CommentPlacement::Outer) }),
96        ("//!", CommentKind { shape: CommentShape::Line, doc: Some(CommentPlacement::Inner) }),
97        ("/**", CommentKind { shape: CommentShape::Block, doc: Some(CommentPlacement::Outer) }),
98        ("/*!", CommentKind { shape: CommentShape::Block, doc: Some(CommentPlacement::Inner) }),
99        ("//", CommentKind { shape: CommentShape::Line, doc: None }),
100        ("/*", CommentKind { shape: CommentShape::Block, doc: None }),
101    ];
102
103    pub(crate) fn from_text(text: &str) -> CommentKind {
104        let &(_prefix, kind) = CommentKind::BY_PREFIX
105            .iter()
106            .find(|&(prefix, _kind)| text.starts_with(prefix))
107            .unwrap();
108        kind
109    }
110
111    pub fn prefix(&self) -> &'static str {
112        let &(prefix, _) =
113            CommentKind::BY_PREFIX.iter().rev().find(|(_, kind)| kind == self).unwrap();
114        prefix
115    }
116}
117
118impl ast::Whitespace {
119    pub fn spans_multiple_lines(&self) -> bool {
120        let text = self.text();
121        text.find('\n').is_some_and(|idx| text[idx + 1..].contains('\n'))
122    }
123}
124
125#[derive(Debug)]
126pub struct QuoteOffsets {
127    pub quotes: (TextRange, TextRange),
128    pub contents: TextRange,
129}
130
131impl QuoteOffsets {
132    fn new(literal: &str) -> Option<QuoteOffsets> {
133        let left_quote = literal.find('"')?;
134        let right_quote = literal.rfind('"')?;
135        if left_quote == right_quote {
136            // `literal` only contains one quote
137            return None;
138        }
139
140        let start = TextSize::from(0);
141        let left_quote = TextSize::try_from(left_quote).unwrap() + TextSize::of('"');
142        let right_quote = TextSize::try_from(right_quote).unwrap();
143        let end = TextSize::of(literal);
144
145        let res = QuoteOffsets {
146            quotes: (TextRange::new(start, left_quote), TextRange::new(right_quote, end)),
147            contents: TextRange::new(left_quote, right_quote),
148        };
149        Some(res)
150    }
151}
152
153pub trait IsString: AstToken {
154    const RAW_PREFIX: &'static str;
155    fn unescape(s: &str, callback: impl FnMut(Range<usize>, Result<char, EscapeError>));
156    fn is_raw(&self) -> bool {
157        self.text().starts_with(Self::RAW_PREFIX)
158    }
159    fn quote_offsets(&self) -> Option<QuoteOffsets> {
160        let text = self.text();
161        let offsets = QuoteOffsets::new(text)?;
162        let o = self.syntax().text_range().start();
163        let offsets = QuoteOffsets {
164            quotes: (offsets.quotes.0 + o, offsets.quotes.1 + o),
165            contents: offsets.contents + o,
166        };
167        Some(offsets)
168    }
169    fn text_range_between_quotes(&self) -> Option<TextRange> {
170        self.quote_offsets().map(|it| it.contents)
171    }
172    fn text_without_quotes(&self) -> &str {
173        let text = self.text();
174        let Some(offsets) = self.text_range_between_quotes() else { return text };
175        &text[offsets - self.syntax().text_range().start()]
176    }
177    fn open_quote_text_range(&self) -> Option<TextRange> {
178        self.quote_offsets().map(|it| it.quotes.0)
179    }
180    fn close_quote_text_range(&self) -> Option<TextRange> {
181        self.quote_offsets().map(|it| it.quotes.1)
182    }
183    fn escaped_char_ranges(&self, cb: &mut dyn FnMut(TextRange, Result<char, EscapeError>)) {
184        let Some(text_range_no_quotes) = self.text_range_between_quotes() else { return };
185
186        let start = self.syntax().text_range().start();
187        let text = &self.text()[text_range_no_quotes - start];
188        let offset = text_range_no_quotes.start() - start;
189
190        Self::unescape(text, &mut |range: Range<usize>, unescaped_char| {
191            if let Some((s, e)) = range.start.try_into().ok().zip(range.end.try_into().ok()) {
192                cb(TextRange::new(s, e) + offset, unescaped_char);
193            }
194        });
195    }
196    fn map_range_up(&self, range: TextRange) -> Option<TextRange> {
197        let contents_range = self.text_range_between_quotes()?;
198        if always!(TextRange::up_to(contents_range.len()).contains_range(range)) {
199            Some(range + contents_range.start())
200        } else {
201            None
202        }
203    }
204}
205
206impl IsString for ast::String {
207    const RAW_PREFIX: &'static str = "r";
208    fn unescape(s: &str, cb: impl FnMut(Range<usize>, Result<char, EscapeError>)) {
209        unescape_str(s, cb)
210    }
211}
212
213impl ast::String {
214    pub fn value(&self) -> Result<Cow<'_, str>, EscapeError> {
215        let text = self.text();
216        let text_range = self.text_range_between_quotes().ok_or(EscapeError::LoneSlash)?;
217        let text = &text[text_range - self.syntax().text_range().start()];
218        if self.is_raw() {
219            return Ok(Cow::Borrowed(text));
220        }
221
222        let mut buf = String::new();
223        let mut prev_end = 0;
224        let mut has_error = None;
225        unescape_str(text, |char_range, unescaped_char| {
226            match (unescaped_char, buf.capacity() == 0) {
227                (Ok(c), false) => buf.push(c),
228                (Ok(_), true) if char_range.len() == 1 && char_range.start == prev_end => {
229                    prev_end = char_range.end
230                }
231                (Ok(c), true) => {
232                    buf.reserve_exact(text.len());
233                    buf.push_str(&text[..prev_end]);
234                    buf.push(c);
235                }
236                (Err(e), _) => has_error = Some(e),
237            }
238        });
239
240        match (has_error, buf.capacity() == 0) {
241            (Some(e), _) => Err(e),
242            (None, true) => Ok(Cow::Borrowed(text)),
243            (None, false) => Ok(Cow::Owned(buf)),
244        }
245    }
246}
247
248impl IsString for ast::ByteString {
249    const RAW_PREFIX: &'static str = "br";
250    fn unescape(s: &str, mut callback: impl FnMut(Range<usize>, Result<char, EscapeError>)) {
251        unescape_byte_str(s, |range, res| callback(range, res.map(char::from)))
252    }
253}
254
255impl ast::ByteString {
256    pub fn value(&self) -> Result<Cow<'_, [u8]>, EscapeError> {
257        let text = self.text();
258        let text_range = self.text_range_between_quotes().ok_or(EscapeError::LoneSlash)?;
259        let text = &text[text_range - self.syntax().text_range().start()];
260        if self.is_raw() {
261            return Ok(Cow::Borrowed(text.as_bytes()));
262        }
263
264        let mut buf: Vec<u8> = Vec::new();
265        let mut prev_end = 0;
266        let mut has_error = None;
267        unescape_byte_str(text, |char_range, unescaped_byte| {
268            match (unescaped_byte, buf.capacity() == 0) {
269                (Ok(b), false) => buf.push(b),
270                (Ok(_), true) if char_range.len() == 1 && char_range.start == prev_end => {
271                    prev_end = char_range.end
272                }
273                (Ok(b), true) => {
274                    buf.reserve_exact(text.len());
275                    buf.extend_from_slice(&text.as_bytes()[..prev_end]);
276                    buf.push(b);
277                }
278                (Err(e), _) => has_error = Some(e),
279            }
280        });
281
282        match (has_error, buf.capacity() == 0) {
283            (Some(e), _) => Err(e),
284            (None, true) => Ok(Cow::Borrowed(text.as_bytes())),
285            (None, false) => Ok(Cow::Owned(buf)),
286        }
287    }
288}
289
290impl IsString for ast::CString {
291    const RAW_PREFIX: &'static str = "cr";
292    // NOTE: This method should only be used for highlighting ranges. The unescaped
293    // char/byte is not used. For simplicity, we return an arbitrary placeholder char.
294    fn unescape(s: &str, mut callback: impl FnMut(Range<usize>, Result<char, EscapeError>)) {
295        unescape_c_str(s, |range, _res| callback(range, Ok('_')))
296    }
297}
298
299impl ast::CString {
300    pub fn value(&self) -> Result<Cow<'_, [u8]>, EscapeError> {
301        let text = self.text();
302        let text_range = self.text_range_between_quotes().ok_or(EscapeError::LoneSlash)?;
303        let text = &text[text_range - self.syntax().text_range().start()];
304        if self.is_raw() {
305            return Ok(Cow::Borrowed(text.as_bytes()));
306        }
307
308        let mut buf = Vec::new();
309        let mut prev_end = 0;
310        let mut has_error = None;
311        let extend_unit = |buf: &mut Vec<u8>, unit: MixedUnit| match unit {
312            MixedUnit::Char(c) => buf.extend(c.encode_utf8(&mut [0; 4]).as_bytes()),
313            MixedUnit::HighByte(b) => buf.push(b),
314        };
315        unescape_c_str(text, |char_range, unescaped| match (unescaped, buf.capacity() == 0) {
316            (Ok(u), false) => extend_unit(&mut buf, u),
317            (Ok(_), true) if char_range.len() == 1 && char_range.start == prev_end => {
318                prev_end = char_range.end
319            }
320            (Ok(u), true) => {
321                buf.reserve_exact(text.len());
322                buf.extend(&text.as_bytes()[..prev_end]);
323                extend_unit(&mut buf, u);
324            }
325            (Err(e), _) => has_error = Some(e),
326        });
327
328        match (has_error, buf.capacity() == 0) {
329            (Some(e), _) => Err(e),
330            (None, true) => Ok(Cow::Borrowed(text.as_bytes())),
331            (None, false) => Ok(Cow::Owned(buf)),
332        }
333    }
334}
335
336impl ast::IntNumber {
337    pub fn radix(&self) -> Radix {
338        match self.text().get(..2).unwrap_or_default() {
339            "0b" => Radix::Binary,
340            "0o" => Radix::Octal,
341            "0x" => Radix::Hexadecimal,
342            _ => Radix::Decimal,
343        }
344    }
345
346    pub fn split_into_parts(&self) -> (&str, &str, &str) {
347        let radix = self.radix();
348        let (prefix, mut text) = self.text().split_at(radix.prefix_len());
349
350        let is_suffix_start: fn(&(usize, char)) -> bool = match radix {
351            Radix::Hexadecimal => |(_, c)| matches!(c, 'g'..='z' | 'G'..='Z'),
352            _ => |(_, c)| c.is_ascii_alphabetic(),
353        };
354
355        let mut suffix = "";
356        if let Some((suffix_start, _)) = text.char_indices().find(is_suffix_start) {
357            let (text2, suffix2) = text.split_at(suffix_start);
358            text = text2;
359            suffix = suffix2;
360        };
361
362        (prefix, text, suffix)
363    }
364
365    pub fn value(&self) -> Result<u128, ParseIntError> {
366        let (_, text, _) = self.split_into_parts();
367        u128::from_str_radix(&text.replace('_', ""), self.radix() as u32)
368    }
369
370    pub fn suffix(&self) -> Option<&str> {
371        let (_, _, suffix) = self.split_into_parts();
372        if suffix.is_empty() { None } else { Some(suffix) }
373    }
374
375    pub fn value_string(&self) -> String {
376        let (_, text, _) = self.split_into_parts();
377        text.replace('_', "")
378    }
379}
380
381impl ast::FloatNumber {
382    pub fn split_into_parts(&self) -> (&str, &str) {
383        let text = self.text();
384        let mut float_text = self.text();
385        let mut suffix = "";
386        let mut indices = text.char_indices();
387        if let Some((mut suffix_start, c)) = indices.by_ref().find(|(_, c)| c.is_ascii_alphabetic())
388        {
389            if c == 'e' || c == 'E' {
390                if let Some(suffix_start_tuple) = indices.find(|(_, c)| c.is_ascii_alphabetic()) {
391                    suffix_start = suffix_start_tuple.0;
392
393                    float_text = &text[..suffix_start];
394                    suffix = &text[suffix_start..];
395                }
396            } else {
397                float_text = &text[..suffix_start];
398                suffix = &text[suffix_start..];
399            }
400        }
401
402        (float_text, suffix)
403    }
404
405    pub fn suffix(&self) -> Option<&str> {
406        let (_, suffix) = self.split_into_parts();
407        if suffix.is_empty() { None } else { Some(suffix) }
408    }
409
410    pub fn value_string(&self) -> String {
411        let (text, _) = self.split_into_parts();
412        text.replace('_', "")
413    }
414}
415
416#[derive(Debug, PartialEq, Eq, Copy, Clone)]
417pub enum Radix {
418    Binary = 2,
419    Octal = 8,
420    Decimal = 10,
421    Hexadecimal = 16,
422}
423
424impl Radix {
425    pub const ALL: &'static [Radix] =
426        &[Radix::Binary, Radix::Octal, Radix::Decimal, Radix::Hexadecimal];
427
428    const fn prefix_len(self) -> usize {
429        match self {
430            Self::Decimal => 0,
431            _ => 2,
432        }
433    }
434}
435
436impl ast::Char {
437    pub fn value(&self) -> Result<char, EscapeError> {
438        let mut text = self.text();
439        if text.starts_with('\'') {
440            text = &text[1..];
441        } else {
442            return Err(EscapeError::ZeroChars);
443        }
444        if text.ends_with('\'') {
445            text = &text[0..text.len() - 1];
446        }
447
448        unescape_char(text)
449    }
450}
451
452impl ast::Byte {
453    pub fn value(&self) -> Result<u8, EscapeError> {
454        let mut text = self.text();
455        if text.starts_with("b\'") {
456            text = &text[2..];
457        } else {
458            return Err(EscapeError::ZeroChars);
459        }
460        if text.ends_with('\'') {
461            text = &text[0..text.len() - 1];
462        }
463
464        unescape_byte(text)
465    }
466}
467
468#[cfg(test)]
469mod tests {
470    use rustc_apfloat::ieee::Quad as f128;
471
472    use crate::ast::{self, FloatNumber, IntNumber, make};
473
474    fn check_float_suffix<'a>(lit: &str, expected: impl Into<Option<&'a str>>) {
475        assert_eq!(FloatNumber { syntax: make::tokens::literal(lit) }.suffix(), expected.into());
476    }
477
478    fn check_int_suffix<'a>(lit: &str, expected: impl Into<Option<&'a str>>) {
479        assert_eq!(IntNumber { syntax: make::tokens::literal(lit) }.suffix(), expected.into());
480    }
481
482    // FIXME(#17451) Use `expected: f128` once `f128` is stabilised.
483    fn check_float_value(lit: &str, expected: &str) {
484        let expected = Some(expected.parse::<f128>().unwrap());
485        assert_eq!(
486            FloatNumber { syntax: make::tokens::literal(lit) }.value_string().parse::<f128>().ok(),
487            expected
488        );
489        assert_eq!(
490            IntNumber { syntax: make::tokens::literal(lit) }.value_string().parse::<f128>().ok(),
491            expected
492        );
493    }
494
495    fn check_int_value(lit: &str, expected: impl Into<Option<u128>>) {
496        assert_eq!(IntNumber { syntax: make::tokens::literal(lit) }.value().ok(), expected.into());
497    }
498
499    #[test]
500    fn test_float_number_suffix() {
501        check_float_suffix("123.0", None);
502        check_float_suffix("123f32", "f32");
503        check_float_suffix("123.0e", None);
504        check_float_suffix("123.0e4", None);
505        check_float_suffix("123.0ef16", "f16");
506        check_float_suffix("123.0E4f32", "f32");
507        check_float_suffix("1_2_3.0_f128", "f128");
508    }
509
510    #[test]
511    fn test_int_number_suffix() {
512        check_int_suffix("123", None);
513        check_int_suffix("123i32", "i32");
514        check_int_suffix("1_0_1_l_o_l", "l_o_l");
515        check_int_suffix("0b11", None);
516        check_int_suffix("0o11", None);
517        check_int_suffix("0xff", None);
518        check_int_suffix("0b11u32", "u32");
519        check_int_suffix("0o11u32", "u32");
520        check_int_suffix("0xffu32", "u32");
521    }
522
523    fn check_string_value<'a>(lit: &str, expected: impl Into<Option<&'a str>>) {
524        assert_eq!(
525            ast::String { syntax: make::tokens::literal(&format!("\"{lit}\"")) }
526                .value()
527                .as_deref()
528                .ok(),
529            expected.into()
530        );
531    }
532
533    #[test]
534    fn test_string_escape() {
535        check_string_value(r"foobar", "foobar");
536        check_string_value(r"\foobar", None);
537        check_string_value(r"\nfoobar", "\nfoobar");
538        check_string_value(r"C:\\Windows\\System32\\", "C:\\Windows\\System32\\");
539        check_string_value(r"\x61bcde", "abcde");
540        check_string_value(
541            r"a\
542bcde", "abcde",
543        );
544    }
545
546    fn check_byte_string_value<'a, const N: usize>(
547        lit: &str,
548        expected: impl Into<Option<&'a [u8; N]>>,
549    ) {
550        assert_eq!(
551            ast::ByteString { syntax: make::tokens::literal(&format!("b\"{lit}\"")) }
552                .value()
553                .as_deref()
554                .ok(),
555            expected.into().map(|value| &value[..])
556        );
557    }
558
559    #[test]
560    fn test_byte_string_escape() {
561        check_byte_string_value(r"foobar", b"foobar");
562        check_byte_string_value(r"\foobar", None::<&[u8; 0]>);
563        check_byte_string_value(r"\nfoobar", b"\nfoobar");
564        check_byte_string_value(r"C:\\Windows\\System32\\", b"C:\\Windows\\System32\\");
565        check_byte_string_value(r"\x61bcde", b"abcde");
566        check_byte_string_value(
567            r"a\
568bcde", b"abcde",
569        );
570    }
571
572    #[test]
573    fn test_value_underscores() {
574        check_float_value("1.3_4665449586950493453___6_f128", "1.346654495869504934536");
575        check_float_value("1.234567891011121_f64", "1.234567891011121");
576        check_float_value("1__0.__0__f32", "10.0");
577        check_float_value("3._0_f16", "3.0");
578        check_int_value("0b__1_0_", 2);
579        check_int_value("1_1_1_1_1_1", 111111);
580    }
581}