harper_core/expr/
sequence_expr.rs

1use paste::paste;
2
3use crate::{
4    CharStringExt, Span, Token, TokenKind,
5    expr::{FirstMatchOf, FixedPhrase, LongestMatchOf},
6    patterns::{AnyPattern, IndefiniteArticle, WhitespacePattern, Word, WordSet},
7};
8
9use super::{Expr, Optional, Repeating, Step, UnlessStep};
10
11#[derive(Default)]
12pub struct SequenceExpr {
13    exprs: Vec<Box<dyn Expr>>,
14}
15
16/// Generate a `then_*` method from an available `is_*` function on [`TokenKind`].
17macro_rules! gen_then_from_is {
18    ($quality:ident) => {
19        paste! {
20            #[doc = concat!("Adds a step matching a token where [`TokenKind::is_", stringify!($quality), "()`] returns true.")]
21            pub fn [< then_$quality >] (self) -> Self{
22                self.then(|tok: &Token, _source: &[char]| {
23                    tok.kind.[< is_$quality >]()
24                })
25            }
26
27            #[doc = concat!("Adds an optional step matching a token where [`TokenKind::is_", stringify!($quality), "()`] returns true.")]
28            pub fn [< then_optional_$quality >] (self) -> Self{
29                self.then_optional(|tok: &Token, _source: &[char]| {
30                    tok.kind.[< is_$quality >]()
31                })
32            }
33
34            #[doc = concat!("Adds a step matching one or more consecutive tokens where [`TokenKind::is_", stringify!($quality), "()`] returns true.")]
35            pub fn [< then_one_or_more_$quality s >] (self) -> Self{
36                self.then_one_or_more(Box::new(|tok: &Token, _source: &[char]| {
37                    tok.kind.[< is_$quality >]()
38                }))
39            }
40
41            #[doc = concat!("Adds a step matching a token where [`TokenKind::is_", stringify!($quality), "()`] returns false.")]
42            pub fn [< then_anything_but_$quality >] (self) -> Self{
43                self.then(|tok: &Token, _source: &[char]| {
44                    if tok.kind.[< is_$quality >](){
45                        false
46                    }else{
47                        true
48                    }
49                })
50            }
51        }
52    };
53}
54
55impl Expr for SequenceExpr {
56    /// Run the expression starting at an index, returning the total matched window.
57    ///
58    /// If any step returns `None`, the entire expression does as well.
59    fn run(&self, mut cursor: usize, tokens: &[Token], source: &[char]) -> Option<Span<Token>> {
60        let mut window = Span::new_with_len(cursor, 0);
61
62        for cur_expr in &self.exprs {
63            let out = cur_expr.run(cursor, tokens, source)?;
64
65            // Only expand the window if the match actually covers some tokens
66            if out.end > out.start {
67                window.expand_to_include(out.start);
68                window.expand_to_include(out.end.checked_sub(1).unwrap_or(out.start));
69            }
70
71            // Only advance cursor if we actually matched something
72            if out.end > cursor {
73                cursor = out.end;
74            } else if out.start < cursor {
75                cursor = out.start;
76            }
77            // If both start and end are equal to cursor, don't move the cursor
78        }
79
80        Some(window)
81    }
82}
83
84impl SequenceExpr {
85    // Constructor methods
86
87    // Single word token methods
88
89    /// Construct a new sequence with a [`Word`] at the beginning of the operation list.
90    pub fn any_capitalization_of(word: &'static str) -> Self {
91        Self::default().then_any_capitalization_of(word)
92    }
93
94    /// Shorthand for [`Self::any_capitalization_of`].
95    pub fn aco(word: &'static str) -> Self {
96        Self::any_capitalization_of(word)
97    }
98
99    /// Match any word from the given set of words, case-insensitive.
100    pub fn word_set(words: &'static [&'static str]) -> Self {
101        Self::default().then_word_set(words)
102    }
103
104    /// Match any word.
105    pub fn any_word() -> Self {
106        Self::default().then_any_word()
107    }
108
109    // Expressions of more than one token
110
111    // Multiple expressions
112
113    /// Match the first of multiple expressions.
114    pub fn any_of(exprs: Vec<Box<dyn Expr>>) -> Self {
115        Self::default().then_any_of(exprs)
116    }
117
118    /// Will be accepted unless the condition matches.
119    pub fn unless(condition: impl Expr + 'static) -> Self {
120        Self::default().then_unless(condition)
121    }
122
123    // Builder methods
124
125    /// Push an [expression](Expr) to the operation list.
126    pub fn then(mut self, expr: impl Expr + 'static) -> Self {
127        self.exprs.push(Box::new(expr));
128        self
129    }
130
131    /// Pushes an expression that could move the cursor to the sequence, but does not require it.
132    pub fn then_optional(mut self, expr: impl Expr + 'static) -> Self {
133        self.exprs.push(Box::new(Optional::new(expr)));
134        self
135    }
136
137    /// Pushes an expression that will match any of the provided expressions.
138    ///
139    /// If more than one of the provided expressions match, this function provides no guarantee
140    /// as to which match will end up being used. If you need to get the longest of multiple
141    /// matches, use [`Self::then_longest_of()`] instead.
142    pub fn then_any_of(mut self, exprs: Vec<Box<dyn Expr>>) -> Self {
143        self.exprs.push(Box::new(FirstMatchOf::new(exprs)));
144        self
145    }
146
147    /// Pushes an expression that will match the longest of the provided expressions.
148    ///
149    /// If you don't need the longest match, prefer using the short-circuiting
150    /// [`Self::then_any_of()`] instead.
151    pub fn then_longest_of(mut self, exprs: Vec<Box<dyn Expr>>) -> Self {
152        self.exprs.push(Box::new(LongestMatchOf::new(exprs)));
153        self
154    }
155
156    /// Appends the steps in `other` onto the end of `self`.
157    /// This is more efficient than [`Self::then`] because it avoids pointer redirection.
158    pub fn then_seq(mut self, mut other: Self) -> Self {
159        self.exprs.append(&mut other.exprs);
160        self
161    }
162
163    /// Pushes an expression that will match any word from the given set of words, case-insensitive.
164    pub fn then_word_set(self, words: &'static [&'static str]) -> Self {
165        self.then(WordSet::new(words))
166    }
167
168    /// Matches any token whose `Kind` exactly matches.
169    pub fn then_strict(self, kind: TokenKind) -> Self {
170        self.then(move |tok: &Token, _source: &[char]| tok.kind == kind)
171    }
172
173    /// Match against one or more whitespace tokens.
174    pub fn then_whitespace(self) -> Self {
175        self.then(WhitespacePattern)
176    }
177
178    /// Shorthand for [`Self::then_whitespace`].
179    pub fn t_ws(self) -> Self {
180        self.then_whitespace()
181    }
182
183    pub fn then_one_or_more(self, expr: impl Expr + 'static) -> Self {
184        self.then(Repeating::new(Box::new(expr), 1))
185    }
186
187    /// Create a new condition that will step one token forward if met.
188    /// If the condition is _not_ met, the whole expression returns `None`.
189    ///
190    /// This can be used to build out exceptions to other rules.
191    ///
192    /// See [`UnlessStep`] for more info.
193    pub fn then_unless(self, condition: impl Expr + 'static) -> Self {
194        self.then(UnlessStep::new(condition, |_tok: &Token, _src: &[char]| {
195            true
196        }))
197    }
198
199    /// Match any single token.
200    ///
201    /// See [`AnyPattern`] for more info.
202    pub fn then_anything(self) -> Self {
203        self.then(AnyPattern)
204    }
205
206    /// Match any single token.
207    ///
208    /// Shorthand for [`Self::then_anything`].
209    pub fn t_any(self) -> Self {
210        self.then_anything()
211    }
212
213    // Word matching methods
214
215    /// Matches any word.
216    pub fn then_any_word(self) -> Self {
217        self.then(|tok: &Token, _source: &[char]| tok.kind.is_word())
218    }
219
220    /// Match examples of `word` that have any capitalization.
221    pub fn then_any_capitalization_of(self, word: &'static str) -> Self {
222        self.then(Word::new(word))
223    }
224
225    /// Shorthand for [`Self::then_any_capitalization_of`].
226    pub fn t_aco(self, word: &'static str) -> Self {
227        self.then_any_capitalization_of(word)
228    }
229
230    /// Match examples of `word` case-sensitively.
231    pub fn then_exact_word(self, word: &'static str) -> Self {
232        self.then(Word::new_exact(word))
233    }
234
235    /// Match a fixed phrase.
236    pub fn then_fixed_phrase(self, phrase: &'static str) -> Self {
237        self.then(FixedPhrase::from_phrase(phrase))
238    }
239
240    /// Match any word except the ones in `words`.
241    pub fn then_word_except(self, words: &'static [&'static str]) -> Self {
242        self.then(move |tok: &Token, src: &[char]| {
243            !tok.kind.is_word()
244                || !words
245                    .iter()
246                    .any(|&word| tok.span.get_content(src).eq_ignore_ascii_case_str(word))
247        })
248    }
249
250    // Token kind/predicate matching methods
251
252    // One kind
253
254    /// Match a token of a given kind which is not in the list of words.
255    pub fn then_kind_except<F>(self, pred_is: F, ex: &'static [&'static str]) -> Self
256    where
257        F: Fn(&TokenKind) -> bool + Send + Sync + 'static,
258    {
259        self.then(move |tok: &Token, src: &[char]| {
260            pred_is(&tok.kind)
261                && !ex
262                    .iter()
263                    .any(|&word| tok.span.get_content(src).eq_ignore_ascii_case_str(word))
264        })
265    }
266
267    // Two kinds
268
269    /// Match a token where both token kind predicates return true.
270    /// For instance, a word that can be both noun and verb.
271    pub fn then_kind_both<F1, F2>(self, pred_is_1: F1, pred_is_2: F2) -> Self
272    where
273        F1: Fn(&TokenKind) -> bool + Send + Sync + 'static,
274        F2: Fn(&TokenKind) -> bool + Send + Sync + 'static,
275    {
276        self.then(move |tok: &Token, _source: &[char]| pred_is_1(&tok.kind) && pred_is_2(&tok.kind))
277    }
278
279    /// Match a token where either of the two token kind predicates returns true.
280    /// For instance, an adjetive or an adverb.
281    pub fn then_kind_either<F1, F2>(self, pred_is_1: F1, pred_is_2: F2) -> Self
282    where
283        F1: Fn(&TokenKind) -> bool + Send + Sync + 'static,
284        F2: Fn(&TokenKind) -> bool + Send + Sync + 'static,
285    {
286        self.then(move |tok: &Token, _source: &[char]| pred_is_1(&tok.kind) || pred_is_2(&tok.kind))
287    }
288
289    /// Match a token where the first token kind predicate returns true and the second returns false.
290    /// For instance, a word that can be a noun but cannot be a verb.
291    pub fn then_kind_is_but_is_not<F1, F2>(self, pred_is: F1, pred_not: F2) -> Self
292    where
293        F1: Fn(&TokenKind) -> bool + Send + Sync + 'static,
294        F2: Fn(&TokenKind) -> bool + Send + Sync + 'static,
295    {
296        self.then(move |tok: &Token, _source: &[char]| pred_is(&tok.kind) && !pred_not(&tok.kind))
297    }
298
299    /// Match a token where the first token kind predicate returns true and the second returns false,
300    /// and the token is not in the list of exceptions.
301    pub fn then_kind_is_but_is_not_except<F1, F2>(
302        self,
303        pred_is: F1,
304        pred_not: F2,
305        ex: &'static [&'static str],
306    ) -> Self
307    where
308        F1: Fn(&TokenKind) -> bool + Send + Sync + 'static,
309        F2: Fn(&TokenKind) -> bool + Send + Sync + 'static,
310    {
311        self.then(move |tok: &Token, src: &[char]| {
312            pred_is(&tok.kind)
313                && !pred_not(&tok.kind)
314                && !ex
315                    .iter()
316                    .any(|&word| tok.span.get_content(src).eq_ignore_ascii_case_str(word))
317        })
318    }
319
320    gen_then_from_is!(sentence_terminator);
321    // More than two kinds
322
323    /// Match a token where any of the token kind predicates returns true.
324    /// Like `then_kind_either` but for more than two predicates.
325    pub fn then_kind_any<F>(self, preds_is: &'static [F]) -> Self
326    where
327        F: Fn(&TokenKind) -> bool + Send + Sync + 'static,
328    {
329        self.then(move |tok: &Token, _source: &[char]| preds_is.iter().any(|pred| pred(&tok.kind)))
330    }
331
332    /// Match a token where any of the token kind predicates returns true,
333    /// and the word is not in the list of exceptions.
334    pub fn then_kind_any_except<F>(
335        self,
336        preds_is: &'static [F],
337        ex: &'static [&'static str],
338    ) -> Self
339    where
340        F: Fn(&TokenKind) -> bool + Send + Sync + 'static,
341    {
342        self.then(move |tok: &Token, src: &[char]| {
343            preds_is.iter().any(|pred| pred(&tok.kind))
344                && !ex
345                    .iter()
346                    .any(|&word| tok.span.get_content(src).eq_ignore_ascii_case_str(word))
347        })
348    }
349
350    /// Match a token where any of the token kind predicates returns true,
351    /// or the token is in the list of words.
352    pub fn then_kind_any_or_words<F>(
353        self,
354        preds: &'static [F],
355        words: &'static [&'static str],
356    ) -> Self
357    where
358        F: Fn(&TokenKind) -> bool + Send + Sync + 'static,
359    {
360        self.then(move |tok: &Token, src: &[char]| {
361            preds.iter().any(|pred| pred(&tok.kind))
362                // && !words
363                || words
364                    .iter()
365                    .any(|&word| tok.span.get_content(src).eq_ignore_ascii_case_str(word))
366        })
367    }
368
369    /// Match a token where any of the first token kind predicates returns true,
370    /// the second returns false, and the token is not in the list of exceptions.    
371    pub fn then_kind_any_but_not_except<F1, F2>(
372        self,
373        preds_is: &'static [F1],
374        pred_not: F2,
375        ex: &'static [&'static str],
376    ) -> Self
377    where
378        F1: Fn(&TokenKind) -> bool + Send + Sync + 'static,
379        F2: Fn(&TokenKind) -> bool + Send + Sync + 'static,
380    {
381        self.then(move |tok: &Token, src: &[char]| {
382            preds_is.iter().any(|pred| pred(&tok.kind))
383                && !pred_not(&tok.kind)
384                && !ex
385                    .iter()
386                    .any(|&word| tok.span.get_content(src).eq_ignore_ascii_case_str(word))
387        })
388    }
389
390    // Word property matching methods
391
392    // Out-of-vocabulary word. (Words not in the dictionary)
393    gen_then_from_is!(oov);
394    gen_then_from_is!(swear);
395
396    // Part-of-speech matching methods
397
398    // Nominals (nouns and pronouns)
399
400    gen_then_from_is!(nominal);
401    gen_then_from_is!(plural_nominal);
402    gen_then_from_is!(non_plural_nominal);
403    gen_then_from_is!(possessive_nominal);
404
405    // Nouns
406
407    gen_then_from_is!(noun);
408    gen_then_from_is!(proper_noun);
409    gen_then_from_is!(mass_noun_only);
410
411    // Pronouns
412
413    gen_then_from_is!(pronoun);
414    gen_then_from_is!(personal_pronoun);
415    gen_then_from_is!(first_person_singular_pronoun);
416    gen_then_from_is!(first_person_plural_pronoun);
417    gen_then_from_is!(second_person_pronoun);
418    gen_then_from_is!(third_person_pronoun);
419    gen_then_from_is!(third_person_singular_pronoun);
420    gen_then_from_is!(third_person_plural_pronoun);
421    gen_then_from_is!(subject_pronoun);
422    gen_then_from_is!(object_pronoun);
423
424    // Verbs
425
426    gen_then_from_is!(verb);
427    gen_then_from_is!(auxiliary_verb);
428    gen_then_from_is!(linking_verb);
429    gen_then_from_is!(verb_lemma);
430    gen_then_from_is!(verb_simple_past_form);
431    gen_then_from_is!(verb_past_participle_form);
432
433    // Adjectives
434
435    gen_then_from_is!(adjective);
436    gen_then_from_is!(positive_adjective);
437    gen_then_from_is!(comparative_adjective);
438    gen_then_from_is!(superlative_adjective);
439
440    // Adverbs
441
442    gen_then_from_is!(adverb);
443
444    // Determiners
445
446    gen_then_from_is!(determiner);
447    gen_then_from_is!(demonstrative_determiner);
448    gen_then_from_is!(quantifier);
449    gen_then_from_is!(non_quantifier_determiner);
450
451    /// Push an [`IndefiniteArticle`] to the end of the operation list.
452    pub fn then_indefinite_article(self) -> Self {
453        self.then(IndefiniteArticle::default())
454    }
455
456    // Other parts of speech
457
458    gen_then_from_is!(conjunction);
459    gen_then_from_is!(preposition);
460
461    // Punctuation
462
463    gen_then_from_is!(punctuation);
464    gen_then_from_is!(apostrophe);
465    gen_then_from_is!(comma);
466    gen_then_from_is!(hyphen);
467    gen_then_from_is!(period);
468    gen_then_from_is!(semicolon);
469    gen_then_from_is!(quote);
470
471    // Other
472
473    gen_then_from_is!(number);
474    gen_then_from_is!(case_separator);
475    gen_then_from_is!(likely_homograph);
476}
477
478impl<S> From<S> for SequenceExpr
479where
480    S: Step + 'static,
481{
482    fn from(step: S) -> Self {
483        Self {
484            exprs: vec![Box::new(step)],
485        }
486    }
487}
488
489#[cfg(test)]
490mod tests {
491    use crate::{
492        Document, TokenKind,
493        expr::{ExprExt, SequenceExpr},
494        linting::tests::SpanVecExt,
495    };
496
497    #[test]
498    fn test_kind_both() {
499        let noun_and_verb =
500            SequenceExpr::default().then_kind_both(TokenKind::is_noun, TokenKind::is_verb);
501        let doc = Document::new_plain_english_curated("Use a good example.");
502        let matches = noun_and_verb.iter_matches_in_doc(&doc).collect::<Vec<_>>();
503        assert_eq!(matches.to_strings(&doc), vec!["Use", "good", "example"]);
504    }
505
506    #[test]
507    fn test_adjective_or_determiner() {
508        let expr = SequenceExpr::default()
509            .then_kind_either(TokenKind::is_adjective, TokenKind::is_determiner);
510        let doc = Document::new_plain_english_curated("Use a good example.");
511        let matches = expr.iter_matches_in_doc(&doc).collect::<Vec<_>>();
512        assert_eq!(matches.to_strings(&doc), vec!["a", "good"]);
513    }
514
515    #[test]
516    fn test_noun_but_not_adjective() {
517        let expr = SequenceExpr::default()
518            .then_kind_is_but_is_not(TokenKind::is_noun, TokenKind::is_adjective);
519        let doc = Document::new_plain_english_curated("Use a good example.");
520        let matches = expr.iter_matches_in_doc(&doc).collect::<Vec<_>>();
521        assert_eq!(matches.to_strings(&doc), vec!["Use", "example"]);
522    }
523}
harper_core/expr/sequence_expr.rs

harper_core/expr/
sequence_expr.rs
