great_tokenizer/
matcher.rs

1use std::{any::Any, cell::LazyCell, fmt::{Debug, Display, Formatter}};
2
3use regex::Regex;
4
5pub trait MatcherTrait {
6    fn get(&self, src: &String) -> Option<String>;
7    fn get_at(&self, src: &String, index: usize) -> Option<String>;
8    fn _to_string(&self) -> String;
9}
10
11impl MatcherTrait for String {
12    fn get(&self, src: &String) -> Option<String> {
13        if src.starts_with(self) {
14            Some(self.clone())
15        } else {
16            None
17        }
18    }
19    fn get_at(&self, src: &String, index: usize) -> Option<String> {
20        if let Some(x) = src.match_indices(self).find(|x| x.0 == index) {
21            Some(x.1.to_string())
22        } else {
23            None
24        }
25    }
26    fn _to_string(&self) -> String {
27        self.clone()
28    }
29}
30
31impl MatcherTrait for Vec<String> {
32    fn get(&self, src: &String) -> Option<String> {
33        for v in self {
34            if let Some(x) = v.get(src) {
35                return Some(x);
36            }
37        }
38        None
39    }
40    fn get_at(&self, src: &String, index: usize) -> Option<String> {
41        for v in self {
42            if let Some(x) = v.get_at(src, index) {
43                return Some(x);
44            }
45        }
46        None
47    }
48    fn _to_string(&self) -> String {
49        serde_json::to_string(self).unwrap()
50    }
51}
52
53impl MatcherTrait for str {
54    fn get(&self, src: &String) -> Option<String> {
55        self.to_owned().get(src)
56    }
57    fn get_at(&self, src: &String, index: usize) -> Option<String> {
58        self.to_owned().get_at(src, index)
59    }
60    fn _to_string(&self) -> String {
61        self.to_owned()
62    }
63}
64
65impl MatcherTrait for &str {
66    fn get(&self, src: &String) -> Option<String> {
67        (*self).to_owned().get(src)
68    }
69    fn get_at(&self, src: &String, index: usize) -> Option<String> {
70        (*self).to_owned().get_at(src, index)
71    }
72    fn _to_string(&self) -> String {
73        (*self).to_owned()
74    }
75}
76
77impl<T> MatcherTrait for T
78where
79    T: Fn(&String) -> Option<String>,
80{
81    fn get(&self, src: &String) -> Option<String> {
82        self(src)
83    }
84    fn get_at(&self, src: &String, index: usize) -> Option<String> {
85        self(&src[index..].to_owned())
86    }
87    fn _to_string(&self) -> String {
88        "<Fn>".to_owned()
89    }
90}
91
92impl MatcherTrait for Regex {
93    fn get(&self, src: &String) -> Option<String> {
94        if let Some(s) = self.captures(src.as_str()) {
95            Some(s.get(0).unwrap().as_str().to_string())
96        } else {
97            None
98        }
99    }
100    fn get_at(&self, src: &String, index: usize) -> Option<String> {
101        if let Some(x) = self.captures_at(src.as_str(), index) {
102            Some(x.get(0).unwrap().as_str().to_string())
103        } else {
104            None
105        }
106    }
107    fn _to_string(&self) -> String {
108        self.to_string()
109    }
110}
111
112impl MatcherTrait for LazyCell<Regex> {
113    fn get(&self, src: &String) -> Option<String> {
114        (**self).get(src)
115    }
116    fn get_at(&self, src: &String, index: usize) -> Option<String> {
117        (**self).get_at(src, index)
118    }
119    fn _to_string(&self) -> String {
120        (**self)._to_string()
121    }
122}
123
124pub struct Matcher {
125    inner: Box<dyn MatcherTrait>,
126}
127
128impl Display for Matcher {
129    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
130        write!(f, "{}", self.inner._to_string())
131    }
132}
133
134impl Debug for Matcher {
135    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
136        <Self as Display>::fmt(&self, f)
137    }
138}
139
140impl Matcher {
141    pub fn get<T: ToString>(&self, src: T) -> Option<String> {
142        self.inner.get(&src.to_string())
143    }
144    pub fn get_at<T: ToString>(&self, src: T, index: usize) -> Option<String> {
145        self.inner.get_at(&src.to_string(), index)
146    }
147}
148
149impl<T: MatcherTrait + Any> From<T> for Matcher {
150    fn from(value: T) -> Self {
151        Self {
152            inner: Box::new(value),
153        }
154    }
155}
great_tokenizer/matcher.rs

great_tokenizer/
matcher.rs
