harper_core/patterns/
mod.rs1use crate::{Document, LSend, Span, Token};
9
10mod any_pattern;
11mod derived_from;
12mod implies_quantity;
13mod indefinite_article;
14mod inflection_of_be;
15mod invert;
16mod modal_verb;
17mod nominal_phrase;
18mod prepositional_preceder;
19mod relative_pronoun;
20mod upos_set;
21mod whitespace_pattern;
22mod within_edit_distance;
23mod word;
24mod word_set;
25
26pub use any_pattern::AnyPattern;
27pub use derived_from::DerivedFrom;
28pub use implies_quantity::ImpliesQuantity;
29pub use indefinite_article::IndefiniteArticle;
30pub use inflection_of_be::InflectionOfBe;
31pub use invert::Invert;
32pub use modal_verb::ModalVerb;
33pub use nominal_phrase::NominalPhrase;
34pub use prepositional_preceder::{PrepositionalPrecederPattern, prepositional_preceder};
35pub use relative_pronoun::RelativePronoun;
36pub use upos_set::UPOSSet;
37pub use whitespace_pattern::WhitespacePattern;
38pub use within_edit_distance::WithinEditDistance;
39pub use word::Word;
40pub use word_set::WordSet;
41
42pub trait Pattern: LSend {
43 fn matches(&self, tokens: &[Token], source: &[char]) -> Option<usize>;
47}
48
49pub trait PatternExt {
50 fn iter_matches(&self, tokens: &[Token], source: &[char]) -> impl Iterator<Item = Span<Token>>;
51
52 fn find_all_matches(&self, tokens: &[Token], source: &[char]) -> Vec<Span<Token>> {
54 self.iter_matches(tokens, source).collect()
55 }
56}
57
58impl<P> PatternExt for P
59where
60 P: Pattern + ?Sized,
61{
62 fn iter_matches(&self, tokens: &[Token], source: &[char]) -> impl Iterator<Item = Span<Token>> {
63 MatchIter::new(self, tokens, source)
64 }
65}
66
67struct MatchIter<'a, 'b, 'c, P: ?Sized> {
68 pattern: &'a P,
69 tokens: &'b [Token],
70 source: &'c [char],
71 index: usize,
72}
73impl<'a, 'b, 'c, P> MatchIter<'a, 'b, 'c, P>
74where
75 P: Pattern + ?Sized,
76{
77 fn new(pattern: &'a P, tokens: &'b [Token], source: &'c [char]) -> Self {
78 Self {
79 pattern,
80 tokens,
81 source,
82 index: 0,
83 }
84 }
85}
86impl<P> Iterator for MatchIter<'_, '_, '_, P>
87where
88 P: Pattern + ?Sized,
89{
90 type Item = Span<Token>;
91
92 fn next(&mut self) -> Option<Self::Item> {
93 while self.index < self.tokens.len() {
94 if let Some(len) = self
95 .pattern
96 .matches(&self.tokens[self.index..], self.source)
97 {
98 let span = Span::new_with_len(self.index, len);
99 self.index += len.max(1);
100 return Some(span);
101 } else {
102 self.index += 1;
103 }
104 }
105
106 None
107 }
108}
109
110pub trait SingleTokenPattern: LSend {
113 fn matches_token(&self, token: &Token, source: &[char]) -> bool;
114}
115
116impl<S: SingleTokenPattern> Pattern for S {
117 fn matches(&self, tokens: &[Token], source: &[char]) -> Option<usize> {
118 if self.matches_token(tokens.first()?, source) {
119 Some(1)
120 } else {
121 None
122 }
123 }
124}
125
126impl<F: LSend + Fn(&Token, &[char]) -> bool> SingleTokenPattern for F {
127 fn matches_token(&self, token: &Token, source: &[char]) -> bool {
128 self(token, source)
129 }
130}
131
132pub trait DocPattern {
133 fn find_all_matches_in_doc(&self, document: &Document) -> Vec<Span<Token>>;
134}
135
136impl<P: PatternExt> DocPattern for P {
137 fn find_all_matches_in_doc(&self, document: &Document) -> Vec<Span<Token>> {
138 self.find_all_matches(document.get_tokens(), document.get_source())
139 }
140}