kcl_lib/parsing/token/
mod.rs

1// Clippy does not agree with rustc here for some reason.
2#![allow(clippy::needless_lifetimes)]
3
4use std::{fmt, iter::Enumerate, num::NonZeroUsize, str::FromStr};
5
6use anyhow::Result;
7use parse_display::Display;
8use schemars::JsonSchema;
9use serde::{Deserialize, Serialize};
10use tokeniser::Input;
11use tower_lsp::lsp_types::SemanticTokenType;
12use winnow::{
13    self,
14    error::ParseError,
15    stream::{ContainsToken, Stream},
16};
17
18use crate::{
19    errors::KclError,
20    parsing::ast::types::{ItemVisibility, VariableKind},
21    source_range::SourceRange,
22    CompilationError, ModuleId,
23};
24
25mod tokeniser;
26
27pub(crate) use tokeniser::RESERVED_WORDS;
28
29// Note the ordering, it's important that `m` comes after `mm` and `cm`.
30pub const NUM_SUFFIXES: [&str; 10] = ["mm", "cm", "m", "inch", "in", "ft", "yd", "deg", "rad", "?"];
31
32#[derive(Clone, Copy, Debug, Eq, PartialEq, Serialize, Deserialize, ts_rs::TS, JsonSchema)]
33#[repr(u32)]
34pub enum NumericSuffix {
35    None,
36    Count,
37    Length,
38    Angle,
39    Mm,
40    Cm,
41    M,
42    Inch,
43    Ft,
44    Yd,
45    Deg,
46    Rad,
47    Unknown,
48}
49
50impl NumericSuffix {
51    #[allow(dead_code)]
52    pub fn is_none(self) -> bool {
53        self == Self::None
54    }
55
56    pub fn is_some(self) -> bool {
57        self != Self::None
58    }
59
60    pub fn digestable_id(&self) -> &[u8] {
61        match self {
62            NumericSuffix::None => &[],
63            NumericSuffix::Count => b"_",
64            NumericSuffix::Unknown => b"?",
65            NumericSuffix::Length => b"Length",
66            NumericSuffix::Angle => b"Angle",
67            NumericSuffix::Mm => b"mm",
68            NumericSuffix::Cm => b"cm",
69            NumericSuffix::M => b"m",
70            NumericSuffix::Inch => b"in",
71            NumericSuffix::Ft => b"ft",
72            NumericSuffix::Yd => b"yd",
73            NumericSuffix::Deg => b"deg",
74            NumericSuffix::Rad => b"rad",
75        }
76    }
77}
78
79impl FromStr for NumericSuffix {
80    type Err = CompilationError;
81
82    fn from_str(s: &str) -> Result<Self, Self::Err> {
83        match s {
84            "_" | "Count" => Ok(NumericSuffix::Count),
85            "Length" => Ok(NumericSuffix::Length),
86            "Angle" => Ok(NumericSuffix::Angle),
87            "mm" | "millimeters" => Ok(NumericSuffix::Mm),
88            "cm" | "centimeters" => Ok(NumericSuffix::Cm),
89            "m" | "meters" => Ok(NumericSuffix::M),
90            "inch" | "in" => Ok(NumericSuffix::Inch),
91            "ft" | "feet" => Ok(NumericSuffix::Ft),
92            "yd" | "yards" => Ok(NumericSuffix::Yd),
93            "deg" | "degrees" => Ok(NumericSuffix::Deg),
94            "rad" | "radians" => Ok(NumericSuffix::Rad),
95            "?" => Ok(NumericSuffix::Unknown),
96            _ => Err(CompilationError::err(SourceRange::default(), "invalid unit of measure")),
97        }
98    }
99}
100
101impl fmt::Display for NumericSuffix {
102    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
103        match self {
104            NumericSuffix::None => Ok(()),
105            NumericSuffix::Count => write!(f, "_"),
106            NumericSuffix::Unknown => write!(f, "_?"),
107            NumericSuffix::Length => write!(f, "Length"),
108            NumericSuffix::Angle => write!(f, "Angle"),
109            NumericSuffix::Mm => write!(f, "mm"),
110            NumericSuffix::Cm => write!(f, "cm"),
111            NumericSuffix::M => write!(f, "m"),
112            NumericSuffix::Inch => write!(f, "in"),
113            NumericSuffix::Ft => write!(f, "ft"),
114            NumericSuffix::Yd => write!(f, "yd"),
115            NumericSuffix::Deg => write!(f, "deg"),
116            NumericSuffix::Rad => write!(f, "rad"),
117        }
118    }
119}
120
121#[derive(Clone, Debug, PartialEq)]
122pub(crate) struct TokenStream {
123    tokens: Vec<Token>,
124}
125
126impl TokenStream {
127    fn new(tokens: Vec<Token>) -> Self {
128        Self { tokens }
129    }
130
131    pub(super) fn remove_unknown(&mut self) -> Vec<Token> {
132        let tokens = std::mem::take(&mut self.tokens);
133        let (tokens, unknown_tokens): (Vec<Token>, Vec<Token>) = tokens
134            .into_iter()
135            .partition(|token| token.token_type != TokenType::Unknown);
136        self.tokens = tokens;
137        unknown_tokens
138    }
139
140    pub fn iter(&self) -> impl Iterator<Item = &Token> {
141        self.tokens.iter()
142    }
143
144    pub fn is_empty(&self) -> bool {
145        self.tokens.is_empty()
146    }
147
148    pub fn as_slice(&self) -> TokenSlice {
149        TokenSlice::from(self)
150    }
151}
152
153impl<'a> From<&'a TokenStream> for TokenSlice<'a> {
154    fn from(stream: &'a TokenStream) -> Self {
155        TokenSlice {
156            start: 0,
157            end: stream.tokens.len(),
158            stream,
159        }
160    }
161}
162
163impl IntoIterator for TokenStream {
164    type Item = Token;
165
166    type IntoIter = std::vec::IntoIter<Token>;
167
168    fn into_iter(self) -> Self::IntoIter {
169        self.tokens.into_iter()
170    }
171}
172
173#[derive(Debug, Clone)]
174pub(crate) struct TokenSlice<'a> {
175    stream: &'a TokenStream,
176    /// Current position of the leading Token in the stream
177    start: usize,
178    /// The number of total Tokens in the stream
179    end: usize,
180}
181
182impl<'a> std::ops::Deref for TokenSlice<'a> {
183    type Target = [Token];
184
185    fn deref(&self) -> &Self::Target {
186        &self.stream.tokens[self.start..self.end]
187    }
188}
189
190impl<'a> TokenSlice<'a> {
191    pub fn token(&self, i: usize) -> &Token {
192        &self.stream.tokens[i + self.start]
193    }
194
195    pub fn iter(&self) -> impl Iterator<Item = &Token> {
196        (**self).iter()
197    }
198
199    pub fn without_ends(&self) -> Self {
200        Self {
201            start: self.start + 1,
202            end: self.end - 1,
203            stream: self.stream,
204        }
205    }
206
207    pub fn as_source_range(&self) -> SourceRange {
208        let stream_len = self.stream.tokens.len();
209        let first_token = if stream_len == self.start {
210            &self.stream.tokens[self.start - 1]
211        } else {
212            self.token(0)
213        };
214        let last_token = if stream_len == self.end {
215            &self.stream.tokens[stream_len - 1]
216        } else {
217            self.token(self.end - self.start)
218        };
219        SourceRange::new(first_token.start, last_token.end, last_token.module_id)
220    }
221}
222
223impl<'a> IntoIterator for TokenSlice<'a> {
224    type Item = &'a Token;
225
226    type IntoIter = std::slice::Iter<'a, Token>;
227
228    fn into_iter(self) -> Self::IntoIter {
229        self.stream.tokens[self.start..self.end].iter()
230    }
231}
232
233impl<'a> Stream for TokenSlice<'a> {
234    type Token = Token;
235    type Slice = Self;
236    type IterOffsets = Enumerate<std::vec::IntoIter<Token>>;
237    type Checkpoint = Checkpoint;
238
239    fn iter_offsets(&self) -> Self::IterOffsets {
240        #[allow(clippy::unnecessary_to_owned)]
241        self.to_vec().into_iter().enumerate()
242    }
243
244    fn eof_offset(&self) -> usize {
245        self.len()
246    }
247
248    fn next_token(&mut self) -> Option<Self::Token> {
249        let token = self.first()?.clone();
250        self.start += 1;
251        Some(token)
252    }
253
254    fn offset_for<P>(&self, predicate: P) -> Option<usize>
255    where
256        P: Fn(Self::Token) -> bool,
257    {
258        self.iter().position(|b| predicate(b.clone()))
259    }
260
261    fn offset_at(&self, tokens: usize) -> Result<usize, winnow::error::Needed> {
262        if let Some(needed) = tokens.checked_sub(self.len()).and_then(NonZeroUsize::new) {
263            Err(winnow::error::Needed::Size(needed))
264        } else {
265            Ok(tokens)
266        }
267    }
268
269    fn next_slice(&mut self, offset: usize) -> Self::Slice {
270        assert!(self.start + offset <= self.end);
271
272        let next = TokenSlice {
273            stream: self.stream,
274            start: self.start,
275            end: self.start + offset,
276        };
277        self.start += offset;
278        next
279    }
280
281    fn checkpoint(&self) -> Self::Checkpoint {
282        Checkpoint(self.start, self.end)
283    }
284
285    fn reset(&mut self, checkpoint: &Self::Checkpoint) {
286        self.start = checkpoint.0;
287        self.end = checkpoint.1;
288    }
289
290    fn raw(&self) -> &dyn fmt::Debug {
291        self
292    }
293}
294
295impl<'a> winnow::stream::Offset for TokenSlice<'a> {
296    fn offset_from(&self, start: &Self) -> usize {
297        self.start - start.start
298    }
299}
300
301impl<'a> winnow::stream::Offset<Checkpoint> for TokenSlice<'a> {
302    fn offset_from(&self, start: &Checkpoint) -> usize {
303        self.start - start.0
304    }
305}
306
307impl winnow::stream::Offset for Checkpoint {
308    fn offset_from(&self, start: &Self) -> usize {
309        self.0 - start.0
310    }
311}
312
313impl<'a> winnow::stream::StreamIsPartial for TokenSlice<'a> {
314    type PartialState = ();
315
316    fn complete(&mut self) -> Self::PartialState {}
317
318    fn restore_partial(&mut self, _: Self::PartialState) {}
319
320    fn is_partial_supported() -> bool {
321        false
322    }
323}
324
325impl<'a> winnow::stream::FindSlice<&str> for TokenSlice<'a> {
326    fn find_slice(&self, substr: &str) -> Option<std::ops::Range<usize>> {
327        self.iter()
328            .enumerate()
329            .find_map(|(i, b)| if b.value == substr { Some(i..self.end) } else { None })
330    }
331}
332
333#[derive(Clone, Debug)]
334pub struct Checkpoint(usize, usize);
335
336/// The types of tokens.
337#[derive(Debug, PartialEq, Eq, Copy, Clone, Display)]
338#[display(style = "camelCase")]
339pub enum TokenType {
340    /// A number.
341    Number,
342    /// A word.
343    Word,
344    /// An operator.
345    Operator,
346    /// A string.
347    String,
348    /// A keyword.
349    Keyword,
350    /// A type.
351    Type,
352    /// A brace.
353    Brace,
354    /// A hash.
355    Hash,
356    /// A bang.
357    Bang,
358    /// A dollar sign.
359    Dollar,
360    /// Whitespace.
361    Whitespace,
362    /// A comma.
363    Comma,
364    /// A colon.
365    Colon,
366    /// A double colon: `::`
367    DoubleColon,
368    /// A period.
369    Period,
370    /// A double period: `..`.
371    DoublePeriod,
372    /// A double period and a less than: `..<`.
373    DoublePeriodLessThan,
374    /// A line comment.
375    LineComment,
376    /// A block comment.
377    BlockComment,
378    /// A function name.
379    Function,
380    /// Unknown lexemes.
381    Unknown,
382    /// The ? symbol, used for optional values.
383    QuestionMark,
384    /// The @ symbol.
385    At,
386    /// `;`
387    SemiColon,
388}
389
390/// Most KCL tokens correspond to LSP semantic tokens (but not all).
391impl TryFrom<TokenType> for SemanticTokenType {
392    type Error = anyhow::Error;
393    fn try_from(token_type: TokenType) -> Result<Self> {
394        // If you return a new kind of `SemanticTokenType`, make sure to update `SEMANTIC_TOKEN_TYPES`
395        // in the LSP implementation.
396        Ok(match token_type {
397            TokenType::Number => Self::NUMBER,
398            TokenType::Word => Self::VARIABLE,
399            TokenType::Keyword => Self::KEYWORD,
400            TokenType::Type => Self::TYPE,
401            TokenType::Operator => Self::OPERATOR,
402            TokenType::QuestionMark => Self::OPERATOR,
403            TokenType::String => Self::STRING,
404            TokenType::Bang => Self::OPERATOR,
405            TokenType::LineComment => Self::COMMENT,
406            TokenType::BlockComment => Self::COMMENT,
407            TokenType::Function => Self::FUNCTION,
408            TokenType::Whitespace
409            | TokenType::Brace
410            | TokenType::Comma
411            | TokenType::Colon
412            | TokenType::DoubleColon
413            | TokenType::Period
414            | TokenType::DoublePeriod
415            | TokenType::DoublePeriodLessThan
416            | TokenType::Hash
417            | TokenType::Dollar
418            | TokenType::At
419            | TokenType::SemiColon
420            | TokenType::Unknown => {
421                anyhow::bail!("unsupported token type: {:?}", token_type)
422            }
423        })
424    }
425}
426
427impl TokenType {
428    pub fn is_whitespace(&self) -> bool {
429        matches!(self, Self::Whitespace)
430    }
431
432    pub fn is_comment(&self) -> bool {
433        matches!(self, Self::LineComment | Self::BlockComment)
434    }
435}
436
437#[derive(Debug, PartialEq, Eq, Clone)]
438pub struct Token {
439    pub token_type: TokenType,
440    /// Offset in the source code where this token begins.
441    pub start: usize,
442    /// Offset in the source code where this token ends.
443    pub end: usize,
444    pub(super) module_id: ModuleId,
445    pub(super) value: String,
446}
447
448impl ContainsToken<Token> for (TokenType, &str) {
449    fn contains_token(&self, token: Token) -> bool {
450        self.0 == token.token_type && self.1 == token.value
451    }
452}
453
454impl ContainsToken<Token> for TokenType {
455    fn contains_token(&self, token: Token) -> bool {
456        *self == token.token_type
457    }
458}
459
460impl Token {
461    pub fn from_range(
462        range: std::ops::Range<usize>,
463        module_id: ModuleId,
464        token_type: TokenType,
465        value: String,
466    ) -> Self {
467        Self {
468            start: range.start,
469            end: range.end,
470            module_id,
471            value,
472            token_type,
473        }
474    }
475    pub fn is_code_token(&self) -> bool {
476        !matches!(
477            self.token_type,
478            TokenType::Whitespace | TokenType::LineComment | TokenType::BlockComment
479        )
480    }
481
482    pub fn as_source_range(&self) -> SourceRange {
483        SourceRange::new(self.start, self.end, self.module_id)
484    }
485
486    pub fn as_source_ranges(&self) -> Vec<SourceRange> {
487        vec![self.as_source_range()]
488    }
489
490    pub fn visibility_keyword(&self) -> Option<ItemVisibility> {
491        if !matches!(self.token_type, TokenType::Keyword) {
492            return None;
493        }
494        match self.value.as_str() {
495            "export" => Some(ItemVisibility::Export),
496            _ => None,
497        }
498    }
499
500    pub fn numeric_value(&self) -> Option<f64> {
501        if self.token_type != TokenType::Number {
502            return None;
503        }
504        let value = &self.value;
505        let value = value
506            .split_once(|c: char| c == '_' || c.is_ascii_alphabetic())
507            .map(|(s, _)| s)
508            .unwrap_or(value);
509        value.parse().ok()
510    }
511
512    pub fn uint_value(&self) -> Option<u32> {
513        if self.token_type != TokenType::Number {
514            return None;
515        }
516        let value = &self.value;
517        let value = value
518            .split_once(|c: char| c == '_' || c.is_ascii_alphabetic())
519            .map(|(s, _)| s)
520            .unwrap_or(value);
521        value.parse().ok()
522    }
523
524    pub fn numeric_suffix(&self) -> NumericSuffix {
525        if self.token_type != TokenType::Number {
526            return NumericSuffix::None;
527        }
528
529        if self.value.ends_with('_') {
530            return NumericSuffix::Count;
531        }
532
533        for suffix in NUM_SUFFIXES {
534            if self.value.ends_with(suffix) {
535                return suffix.parse().unwrap();
536            }
537        }
538
539        NumericSuffix::None
540    }
541
542    /// Is this token the beginning of a variable/function declaration?
543    /// If so, what kind?
544    /// If not, returns None.
545    pub fn declaration_keyword(&self) -> Option<VariableKind> {
546        if !matches!(self.token_type, TokenType::Keyword) {
547            return None;
548        }
549        Some(match self.value.as_str() {
550            "fn" => VariableKind::Fn,
551            "var" | "let" | "const" => VariableKind::Const,
552            _ => return None,
553        })
554    }
555}
556
557impl From<Token> for SourceRange {
558    fn from(token: Token) -> Self {
559        Self::new(token.start, token.end, token.module_id)
560    }
561}
562
563impl From<&Token> for SourceRange {
564    fn from(token: &Token) -> Self {
565        Self::new(token.start, token.end, token.module_id)
566    }
567}
568
569pub fn lex(s: &str, module_id: ModuleId) -> Result<TokenStream, KclError> {
570    tokeniser::lex(s, module_id).map_err(From::from)
571}
572
573impl From<ParseError<Input<'_>, winnow::error::ContextError>> for KclError {
574    fn from(err: ParseError<Input<'_>, winnow::error::ContextError>) -> Self {
575        let (input, offset): (Vec<char>, usize) = (err.input().chars().collect(), err.offset());
576        let module_id = err.input().state.module_id;
577
578        if offset >= input.len() {
579            // From the winnow docs:
580            //
581            // This is an offset, not an index, and may point to
582            // the end of input (input.len()) on eof errors.
583
584            return KclError::Lexical(crate::errors::KclErrorDetails::new(
585                "unexpected EOF while parsing".to_owned(),
586                vec![SourceRange::new(offset, offset, module_id)],
587            ));
588        }
589
590        // TODO: Add the Winnow tokenizer context to the error.
591        // See https://github.com/KittyCAD/modeling-app/issues/784
592        let bad_token = &input[offset];
593        // TODO: Add the Winnow parser context to the error.
594        // See https://github.com/KittyCAD/modeling-app/issues/784
595        KclError::Lexical(crate::errors::KclErrorDetails::new(
596            format!("found unknown token '{}'", bad_token),
597            vec![SourceRange::new(offset, offset + 1, module_id)],
598        ))
599    }
600}
kcl_lib/parsing/token/mod.rs

kcl_lib/parsing/token/
mod.rs
