1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139

extern crate regex;
use regex::Regex;
use std::cmp::max;
use std::collections::HashMap;
use std::str;

// The LanguageModel holds the path to the dictionary (english
// by default), the maximum word cost and the maximum word length
pub struct LanguageModel {
    word_cost: HashMap<String, f64>,
    max_wlen: u8,
}

impl LanguageModel {
    // Initialize the LanguageModel using the words in the dictionary to
    // get the maximum word length and calculate word cost uzing Zipf's law
    pub fn new() -> LanguageModel {
        let dict = include_str!("dicts/english.txt");

        let word_count: f64 = dict.len() as f64;
        let mut max_wlen: u8 = 0;
        let mut word_cost: HashMap<String, f64> = HashMap::new();

        for (index, line) in dict.lines().enumerate() {
            let word = line; // Ignore errors.
            let word_len: u8 = word.chars().count() as u8;
            max_wlen = max(word_len, max_wlen);

            word_cost.insert(
                String::from(word),
                ((index + 1) as f64 * word_count.ln()).ln(),
            );
        }

        LanguageModel {
            word_cost,
            max_wlen,
        }
    }

    // Split the input into alphanumerical substrings and feed the results
    // into the internal splitter
    pub fn untangle(&self, s: &str) -> Vec<String> {
        let re = Regex::new(r"[^a-zA-Z0-9']+").unwrap();
        re.split(s).flat_map(|x| self.split(x.into())).collect()
    }

    // Takes as input a string and returns a vector of the substrings that
    // match the frequency dictionary
    fn split(&self, s: String) -> Vec<String> {
        let mut cost: Vec<f64> = vec![0.0];
        for i in 1..s.len() + 1 {
            let (_, k) = self.best_match(i, &cost, &s);
            cost.push(k);
        }

        let mut out: Vec<String> = Vec::new();
        let mut i = s.len();
        while i > 0 {
            let (k, c) = self.best_match(i, &cost, &s);
            assert!(c == cost[i]);
            let mut new_token = true;
            let z: usize = 0;
            let outlen = out.len();
            let idx = if (i as i8 - k as i8) < 0 {
                0
            } else {
                i - k as usize
            };
            if &s[idx..i] != "'" {
                if &out.len() > &z {
                    let o = &out[out.len() - 1];
                    if out[out.len() - 1] == "'s"
                        || (s.as_bytes()[i - 1].is_ascii_digit()
                            && o.as_bytes()[0].is_ascii_digit())
                    {
                        out[outlen - 1] =
                            format!("{}{}", &s[i - k as usize..i], &out[out.len() - 1]);
                        new_token = false;
                    }
                }
            }

            if new_token {
                out.push(s[i - k as usize..i].to_string());
            }
            i -= k as usize;
        }
        out.reverse();
        out
    }

    fn best_match(&self, i: usize, cost: &[f64], s: &str) -> (u8, f64) {
        let candidates = &cost[(max(0, i as i64 - self.max_wlen as i64) as usize)..i];
        let mut storedmin: (u8, f64) = (0, 9e99);

        for (index, &candidate) in candidates.into_iter().rev().enumerate() {
            let current_word_cost = match self
                .word_cost
                .get(&s[i - index - 1..i].to_ascii_lowercase())
            {
                Some(value) => value,
                _ => &(9e99 as f64),
            };
            if candidate + current_word_cost < storedmin.1 {
                storedmin.1 = candidate + *current_word_cost;
                storedmin.0 = index as u8 + 1;
            }
        }

        storedmin
    }
}


// Unit tests 
#[cfg(test)]
mod tests {
    use super::*;
    #[test]
    fn basic_split() {
        let no_spaces = "thequickbrownfoxjumpedoverthelazydog";
        let with_spaces = "the quick brown fox jumped over the lazy dog";
        let lm = LanguageModel::new();
        let correct: Vec<&str> = with_spaces.split_whitespace().collect();

        assert_eq!(lm.untangle(no_spaces), correct);
    }

    #[test]
    fn split_with_punctuation() {
        let no_spaces = "thequick!brownfox.jumpedoverthe,lazydog?";
        let with_spaces = "the quick brown fox jumped over the lazy dog";
        let lm = LanguageModel::new();
        let correct: Vec<&str> = with_spaces.split_whitespace().collect();

        assert_eq!(lm.untangle(no_spaces), correct);
    }
}