harper_core/patterns/
mod.rs

1//! [`Pattern`]s are one of the more powerful ways to query text inside Harper, especially for beginners.
2//!
3//! Through the [`PatternLinter`](crate::linting::PatternLinter) trait, they make it much easier to
4//! build Harper [rules](crate::linting::Linter).
5//!
6//! See the page about [`SequencePattern`] for a concrete example of their use.
7
8use crate::{Document, Span, Token};
9
10mod all;
11mod any_pattern;
12mod either_pattern;
13mod exact_phrase;
14mod implies_quantity;
15mod indefinite_article;
16mod inflection_of_be;
17mod invert;
18mod naive_pattern_group;
19mod nominal_phrase;
20mod pattern_map;
21mod repeating_pattern;
22mod sequence_pattern;
23mod similar_to_phrase;
24mod split_compound_word;
25mod whitespace_pattern;
26mod within_edit_distance;
27mod word;
28mod word_pattern_group;
29mod word_set;
30
31pub use all::All;
32pub use any_pattern::AnyPattern;
33use blanket::blanket;
34pub use either_pattern::EitherPattern;
35pub use exact_phrase::ExactPhrase;
36pub use implies_quantity::ImpliesQuantity;
37pub use indefinite_article::IndefiniteArticle;
38pub use inflection_of_be::InflectionOfBe;
39pub use invert::Invert;
40pub use naive_pattern_group::NaivePatternGroup;
41pub use nominal_phrase::NominalPhrase;
42pub use pattern_map::PatternMap;
43pub use repeating_pattern::RepeatingPattern;
44pub use sequence_pattern::SequencePattern;
45pub use similar_to_phrase::SimilarToPhrase;
46pub use split_compound_word::SplitCompoundWord;
47pub use whitespace_pattern::WhitespacePattern;
48pub use word::Word;
49pub use word_pattern_group::WordPatternGroup;
50pub use word_set::WordSet;
51
52#[cfg(not(feature = "concurrent"))]
53#[blanket(derive(Rc, Arc))]
54pub trait Pattern {
55    /// Check if the pattern matches at the start of the given token slice.
56    ///
57    /// Returns the length of the match if successful, or `None` if not.
58    fn matches(&self, tokens: &[Token], source: &[char]) -> Option<usize>;
59}
60#[cfg(feature = "concurrent")]
61#[blanket(derive(Arc))]
62pub trait Pattern: Send + Sync {
63    /// Check if the pattern matches at the start of the given token slice.
64    ///
65    /// Returns the length of the match if successful, or `None` if not.
66    fn matches(&self, tokens: &[Token], source: &[char]) -> Option<usize>;
67}
68
69pub trait PatternExt {
70    fn iter_matches(&self, tokens: &[Token], source: &[char]) -> impl Iterator<Item = Span>;
71
72    /// Search through all tokens to locate all non-overlapping pattern matches.
73    fn find_all_matches(&self, tokens: &[Token], source: &[char]) -> Vec<Span> {
74        self.iter_matches(tokens, source).collect()
75    }
76}
77
78impl<P> PatternExt for P
79where
80    P: Pattern + ?Sized,
81{
82    fn find_all_matches(&self, tokens: &[Token], source: &[char]) -> Vec<Span> {
83        self.iter_matches(tokens, source).collect()
84    }
85    fn iter_matches(&self, tokens: &[Token], source: &[char]) -> impl Iterator<Item = Span> {
86        MatchIter::new(self, tokens, source)
87    }
88}
89
90struct MatchIter<'a, 'b, 'c, P: ?Sized> {
91    pattern: &'a P,
92    tokens: &'b [Token],
93    source: &'c [char],
94    index: usize,
95}
96impl<'a, 'b, 'c, P> MatchIter<'a, 'b, 'c, P>
97where
98    P: Pattern + ?Sized,
99{
100    fn new(pattern: &'a P, tokens: &'b [Token], source: &'c [char]) -> Self {
101        Self {
102            pattern,
103            tokens,
104            source,
105            index: 0,
106        }
107    }
108}
109impl<P> Iterator for MatchIter<'_, '_, '_, P>
110where
111    P: Pattern + ?Sized,
112{
113    type Item = Span;
114
115    fn next(&mut self) -> Option<Self::Item> {
116        while self.index < self.tokens.len() {
117            if let Some(len) = self
118                .pattern
119                .matches(&self.tokens[self.index..], self.source)
120            {
121                let span = Span::new_with_len(self.index, len);
122                self.index += len.max(1);
123                return Some(span);
124            } else {
125                self.index += 1;
126            }
127        }
128
129        None
130    }
131}
132
133pub trait OwnedPatternExt {
134    fn or(self, other: Box<dyn Pattern>) -> EitherPattern;
135}
136
137impl<P> OwnedPatternExt for P
138where
139    P: Pattern + 'static,
140{
141    fn or(self, other: Box<dyn Pattern>) -> EitherPattern {
142        EitherPattern::new(vec![Box::new(self), other])
143    }
144}
145
146#[cfg(feature = "concurrent")]
147impl<F> Pattern for F
148where
149    F: Fn(&Token, &[char]) -> bool,
150    F: Send + Sync,
151{
152    fn matches(&self, tokens: &[Token], source: &[char]) -> Option<usize> {
153        if self(tokens.first()?, source) {
154            Some(1)
155        } else {
156            None
157        }
158    }
159}
160
161#[cfg(not(feature = "concurrent"))]
162impl<F> Pattern for F
163where
164    F: Fn(&Token, &[char]) -> bool,
165{
166    fn matches(&self, tokens: &[Token], source: &[char]) -> Option<usize> {
167        if self(tokens.first()?, source) {
168            Some(1)
169        } else {
170            None
171        }
172    }
173}
174
175pub trait DocPattern {
176    fn find_all_matches_in_doc(&self, document: &Document) -> Vec<Span>;
177}
178
179impl<P: PatternExt> DocPattern for P {
180    fn find_all_matches_in_doc(&self, document: &Document) -> Vec<Span> {
181        self.find_all_matches(document.get_tokens(), document.get_source())
182    }
183}
harper_core/patterns/mod.rs

harper_core/patterns/
mod.rs
