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
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
|
use std::fmt;
use std::fs;
use crate::config::*;
extern crate json;
const fn build_lookup_table() -> [usize; 128] {
let mut result = [0; 128];
let mut i = 0;
while i < NUM_KEYS {
result[KEY_CHARS[i] as usize] = i;
i += 1;
}
return result;
}
const CHAR_TO_INDEX: [usize; 128] = build_lookup_table();
const NUM_BIGRAMS: usize = NUM_KEYS * NUM_KEYS;
pub struct Corpus {
bigram_percs: [f32; NUM_BIGRAMS],
skipgram_percs: [f32; NUM_BIGRAMS],
character_percs: [f32; 128],
}
fn pair_to_index(x: u8, y: u8) -> usize {
let xi = CHAR_TO_INDEX[x as usize];
let yi = CHAR_TO_INDEX[y as usize];
xi + NUM_KEYS * yi
}
impl Corpus {
pub fn get_character_perc(&self, c: u8) -> f32 {
self.character_percs[c as usize]
}
pub fn get_bigram_perc(&self, x: u8, y: u8) -> f32 {
self.bigram_percs[pair_to_index(x, y)]
}
pub fn get_skipgram_perc(&self, x: u8, y: u8) -> f32 {
self.skipgram_percs[pair_to_index(x, y)]
}
pub fn load_from_json_file(path: &str) -> Result<Corpus, std::io::Error> {
let contents = fs::read_to_string(path)?;
let json = json::parse(&contents).unwrap();
let mut character_percs = [0.0; 128];
let mut bigram_percs = [0.0; NUM_BIGRAMS];
let mut skipgram_percs = [0.0; NUM_BIGRAMS];
for (key, value) in json.entries() {
if key == "characters" {
for (c, p) in value.entries() {
if let Some(c) = canonicalise(c.as_bytes()[0] as char) {
character_percs[c as usize] = p.as_f32().unwrap();
}
}
} else if key == "bigrams" {
for (b, p) in value.entries() {
let canon: Vec<Option<u8>> = b
.as_bytes()
.iter()
.map(|&b| canonicalise(b as char))
.collect();
if let Some(c1) = canon[0] {
if let Some(c2) = canon[1] {
bigram_percs[pair_to_index(c1, c2)] += p.as_f32().unwrap();
bigram_percs[pair_to_index(c2, c1)] += p.as_f32().unwrap();
}
}
}
} else if key == "skipgrams" {
for (t, p) in value.entries() {
let canon: Vec<Option<u8>> = t
.as_bytes()
.iter()
.map(|&b| canonicalise(b as char))
.collect();
if let Some(c1) = canon[0] {
if let Some(c3) = canon[1] {
skipgram_percs[pair_to_index(c1, c3)] += p.as_f32().unwrap();
skipgram_percs[pair_to_index(c3, c1)] += p.as_f32().unwrap();
}
}
}
}
}
println!("Chars: {:?}\nBigs:{:?}", character_percs, bigram_percs);
return Ok(Corpus {
skipgram_percs,
bigram_percs,
character_percs,
});
}
pub fn load_from_text_file(path: &str) -> Result<Corpus, std::io::Error> {
let contents = fs::read_to_string(path)?;
let mut bigram_count = [0; NUM_BIGRAMS];
let mut skipgram_count = [0; NUM_BIGRAMS];
let mut character_count = [0; 128];
let mut total_count = 0;
let mut total_bigrams = 0;
let mut total_trigrams = 0;
let mut last_char: Option<u8> = None;
let mut last_last_char: Option<u8> = None;
for c in contents.chars() {
let current_char = canonicalise(c);
if let Some(c) = current_char {
if let Some(lc) = last_char {
if let Some(llc) = last_last_char {
if llc != lc && lc != c && llc != c {
skipgram_count[pair_to_index(llc, c)] += 1;
skipgram_count[pair_to_index(c, llc)] += 1;
}
total_trigrams += 1;
}
if lc != c {
bigram_count[pair_to_index(c, lc)] += 1;
bigram_count[pair_to_index(lc, c)] += 1;
}
total_bigrams += 1;
}
character_count[c as usize] += 1;
total_count += 1;
}
last_last_char = last_char;
last_char = current_char;
}
let mut bigram_percs = [0.0; NUM_BIGRAMS];
let mut skipgram_percs = [0.0; NUM_BIGRAMS];
let mut character_percs = [0.0; 128];
for i in 0..NUM_BIGRAMS {
bigram_percs[i] = bigram_count[i] as f32 / total_bigrams as f32;
skipgram_percs[i] = skipgram_count[i] as f32 / total_trigrams as f32;
}
for i in 0..128 {
character_percs[i] = character_count[i] as f32 / total_count as f32;
}
return Ok(Corpus {
skipgram_percs,
bigram_percs,
character_percs,
});
}
}
fn dump_bigrams(corpus: &Corpus) -> Vec<(String, f32)> {
let mut result: Vec<(String, f32)> = Vec::new();
for &x in KEY_CHARS.iter() {
for &y in KEY_CHARS.iter() {
let mut pair = String::from(x);
pair.push(y);
let perc = corpus.bigram_percs[pair_to_index(x as u8, y as u8)];
if perc > 0.0 {
result.push((pair, perc));
}
}
}
result.sort_by(|(_, c1), (_, c2)| {
if c2 < c1 {
std::cmp::Ordering::Less
} else {
std::cmp::Ordering::Greater
}
});
return result;
}
impl fmt::Display for Corpus {
fn fmt(&self, formatter: &mut fmt::Formatter) -> fmt::Result {
formatter.write_str(&format!(
"Corpus has top 5 bigrams {:?}",
&dump_bigrams(self)[0..5]
))
}
}
|
