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
use std::str::from_utf8;

const MODE_WIDTH: u8 = 10;

lazy_static::lazy_static! {
    static ref WIDTHS: Vec<(char, u8)> = {
        use std::io::Read;
        // Format of this file is documented in character_analyzer.rs
        let mut raw = include_bytes!("character_widths.bin").as_slice();

        // First byte is mode length.
        let mut mode = [0u8];
        raw.read(&mut mode).unwrap();
        let mode = mode[0];

        assert_eq!(mode, MODE_WIDTH);

        let mut widths = Vec::new();

        while !raw.is_empty() {
            // Read one UTF-8 character.
            // TODO: Once stable, use: utf8_char_width(raw[0])
            let s = from_utf8(&raw[..1])
                .or_else(|_| from_utf8(&raw[..2]))
                .or_else(|_| from_utf8(&raw[..3]))
                .or_else(|_| from_utf8(&raw[..4]))
                .unwrap();
            let c = s.chars().next().unwrap();
            raw = &raw[c.len_utf8()..];

            // After character comes a byte of length.
            let mut len = [0u8];
            raw.read(&mut len).unwrap();
            let len = len[0];

            widths.push((c, len));
        }

        widths
    };
}

/// Returns an estimate of the worst-case display width in milli-`m`'s (thousandths of the
/// the width of an `m` character).
///
/// For example, `width('m')` returns 1000 and `width('\u{FDFD}')` returns 10300 (wouldn't you like
/// to know if your user's text is 10.3X longer per character than you might have expected?).
///
/// Precision is not necessarily 1 milli-`m` (currently, it is 100 milli-`m`'s).
#[cfg_attr(doc, doc(cfg(feature = "width")))]
pub fn width(c: char) -> usize {
    let width = match WIDTHS.binary_search_by_key(&c, |&(c, _)| c) {
        Ok(idx) => WIDTHS[idx].1,
        Err(_) => MODE_WIDTH,
    } as usize;

    width * 100
}

/// Convenience method for getting the width, in `m`'s, of an entire string.
///
/// Warning: If the width overflows, the result is undefined (e.g. panic or overflow).
#[cfg_attr(doc, doc(cfg(feature = "width")))]
pub fn width_str(s: &str) -> usize {
    s.chars().map(|c| width(c) / 100).sum::<usize>() / 10
}

/// Trims a string to a maximum number of `m`'s. A budget of 5 would allow five m, or more narrower
/// characters, or fewer wider characters.
pub fn trim_to_width(s: &str, mut budget: usize) -> &str {
    // Convert to milli-`m`'s.
    budget *= 10;
    for (idx, c) in s.char_indices() {
        match budget.checked_sub(width(c) / 100) {
            Some(new_budget) => budget = new_budget,
            None => return &s[..idx],
        }
    }
    return s;
}

#[cfg(test)]
mod test {
    use crate::width::{trim_to_width, width_str};
    use crate::{width, CensorStr};
    use serial_test::serial;

    /*
    #[test]
    pub fn i() {
        assert_eq!(width('i'), 600);
    }
     */

    #[test]
    pub fn m() {
        assert_eq!(width('m'), 1000);
    }

    #[test]
    pub fn fdfd() {
        // https://commons.wikimedia.org/wiki/File:Lateef_unicode_U%2BFDFD_2020-03-09_122519.png
        assert_eq!(width('\u{FDFD}'), 10300)
    }

    #[test]
    pub fn three_em_dash() {
        assert!(width('⸻') >= 2500);
    }

    #[test]
    pub fn lattice() {
        assert!(width('𒐫') >= 3000);
    }

    #[test]
    pub fn cuneiform() {
        assert!(width('𒈙') >= 3000);
    }

    #[test]
    pub fn javanese() {
        assert!(width('꧅') >= 1500);
    }

    #[test]
    pub fn emoji() {
        assert_eq!(width_str("😀🐿"), 4);
    }

    #[test]
    pub fn cjk() {
        assert_eq!(width_str("大はㅂ"), 6)
    }

    #[test]
    pub fn string() {
        //assert_eq!(width_str("abc‱DŽဪ"), 7);
        assert_eq!(width_str("abc‱DŽဪ"), 8);
    }

    #[test]
    #[serial]
    pub fn tall() {
        assert_eq!("a꧁a".censor(), "aa");
    }

    #[test]
    #[serial]
    pub fn trim() {
        assert_eq!(trim_to_width("aa", 0), "");
        assert_eq!(trim_to_width("mmm", 1), "m");
        assert_eq!(trim_to_width("mmm", 2), "mm");
        assert_eq!(trim_to_width("mmm", 3), "mmm");
        assert_eq!(trim_to_width("mmm", 4), "mmm");

        let mut s = String::new();
        for u in 0..10000 {
            if let Some(c) = char::from_u32(u) {
                s.push(c);
            }
        }
        for b in 0..1000 {
            let t = trim_to_width(&s, b);
            let w = width_str(t);
            assert!(w <= b);
            assert!(w + 15 >= b)
        }
    }
}