unicode_normalization/
stream_safe.rs

Help
1use crate::lookups::{
2    canonical_combining_class, canonical_fully_decomposed, compatibility_fully_decomposed,
3    stream_safe_trailing_nonstarters,
4};
5use crate::normalize::{hangul_decomposition_length, is_hangul_syllable};
6use crate::tables::stream_safe_leading_nonstarters;
7
8pub(crate) const MAX_NONSTARTERS: usize = 30;
9const COMBINING_GRAPHEME_JOINER: char = '\u{034F}';
10
11/// UAX15-D4: This iterator keeps track of how many non-starters there have been
12/// since the last starter in *NFKD* and will emit a Combining Grapheme Joiner
13/// (U+034F) if the count exceeds 30.
14pub struct StreamSafe<I> {
15    iter: I,
16    nonstarter_count: usize,
17    buffer: Option<char>,
18}
19
20impl<I> StreamSafe<I> {
21    pub(crate) fn new(iter: I) -> Self {
22        Self {
23            iter,
24            nonstarter_count: 0,
25            buffer: None,
26        }
27    }
28}
29
30impl<I: Iterator<Item = char>> Iterator for StreamSafe<I> {
31    type Item = char;
32
33    #[inline]
34    fn next(&mut self) -> Option<char> {
35        let next_ch = match self.buffer.take().or_else(|| self.iter.next()) {
36            None => return None,
37            Some(c) => c,
38        };
39        let d = classify_nonstarters(next_ch);
40        if self.nonstarter_count + d.leading_nonstarters > MAX_NONSTARTERS {
41            // Since we're emitting a CGJ, the suffix of the emitted string in NFKD has no trailing
42            // nonstarters, so we can reset the counter to zero. Put `next_ch` back into the
43            // iterator (via `self.buffer`), and we'll reclassify it next iteration.
44            self.nonstarter_count = 0;
45            self.buffer = Some(next_ch);
46            return Some(COMBINING_GRAPHEME_JOINER);
47        }
48
49        // Is the character all nonstarters in NFKD? If so, increment our counter of contiguous
50        // nonstarters in NKFD.
51        if d.leading_nonstarters == d.decomposition_len {
52            self.nonstarter_count += d.decomposition_len;
53        }
54        // Otherwise, reset the counter to the decomposition's number of trailing nonstarters.
55        else {
56            self.nonstarter_count = d.trailing_nonstarters;
57        }
58        Some(next_ch)
59    }
60}
61
62#[derive(Debug)]
63pub(crate) struct Decomposition {
64    pub(crate) leading_nonstarters: usize,
65    pub(crate) trailing_nonstarters: usize,
66    pub(crate) decomposition_len: usize,
67}
68
69#[inline]
70pub(crate) fn classify_nonstarters(c: char) -> Decomposition {
71    // As usual, fast path for ASCII (which is always a starter)
72    if c <= '\x7f' {
73        return Decomposition {
74            leading_nonstarters: 0,
75            trailing_nonstarters: 0,
76            decomposition_len: 1,
77        };
78    }
79    // Next, special case Hangul, since it's not handled by our tables.
80    if is_hangul_syllable(c) {
81        return Decomposition {
82            leading_nonstarters: 0,
83            trailing_nonstarters: 0,
84            decomposition_len: hangul_decomposition_length(c),
85        };
86    }
87    let decomp = compatibility_fully_decomposed(c).or_else(|| canonical_fully_decomposed(c));
88    match decomp {
89        Some(decomp) => Decomposition {
90            leading_nonstarters: stream_safe_leading_nonstarters(c),
91            trailing_nonstarters: stream_safe_trailing_nonstarters(c),
92            decomposition_len: decomp.len(),
93        },
94        None => {
95            let is_nonstarter = canonical_combining_class(c) != 0;
96            let nonstarter = if is_nonstarter { 1 } else { 0 };
97            Decomposition {
98                leading_nonstarters: nonstarter,
99                trailing_nonstarters: nonstarter,
100                decomposition_len: 1,
101            }
102        }
103    }
104}
105
106#[cfg(test)]
107mod tests {
108    use super::{classify_nonstarters, StreamSafe};
109    use crate::lookups::canonical_combining_class;
110    use crate::normalize::decompose_compatible;
111
112    #[cfg(not(feature = "std"))]
113    use crate::no_std_prelude::*;
114
115    use core::char;
116
117    fn stream_safe(s: &str) -> String {
118        StreamSafe::new(s.chars()).collect()
119    }
120
121    #[test]
122    fn test_simple() {
123        let technically_okay = "Da\u{0300}\u{0301}\u{0302}\u{0303}\u{0304}\u{0305}\u{0306}\u{0307}\u{0308}\u{0309}\u{030a}\u{030b}\u{030c}\u{030d}\u{030e}\u{030f}\u{0310}\u{0311}\u{0312}\u{0313}\u{0314}\u{0315}\u{0316}\u{0317}\u{0318}\u{0319}\u{031a}\u{031b}\u{031c}\u{031d}ngerzone";
124        assert_eq!(stream_safe(technically_okay), technically_okay);
125
126        let too_much = "Da\u{0300}\u{0301}\u{0302}\u{0303}\u{0304}\u{0305}\u{0306}\u{0307}\u{0308}\u{0309}\u{030a}\u{030b}\u{030c}\u{030d}\u{030e}\u{030f}\u{0310}\u{0311}\u{0312}\u{0313}\u{0314}\u{0315}\u{0316}\u{0317}\u{0318}\u{0319}\u{031a}\u{031b}\u{031c}\u{031d}\u{032e}ngerzone";
127        let fixed_it = "Da\u{0300}\u{0301}\u{0302}\u{0303}\u{0304}\u{0305}\u{0306}\u{0307}\u{0308}\u{0309}\u{030a}\u{030b}\u{030c}\u{030d}\u{030e}\u{030f}\u{0310}\u{0311}\u{0312}\u{0313}\u{0314}\u{0315}\u{0316}\u{0317}\u{0318}\u{0319}\u{031a}\u{031b}\u{031c}\u{031d}\u{034f}\u{032e}ngerzone";
128        assert_eq!(stream_safe(too_much), fixed_it);
129
130        let woah_nelly = "Da\u{0300}\u{0301}\u{0302}\u{0303}\u{0304}\u{0305}\u{0306}\u{0307}\u{0308}\u{0309}\u{030a}\u{030b}\u{030c}\u{030d}\u{030e}\u{030f}\u{0310}\u{0311}\u{0312}\u{0313}\u{0314}\u{0315}\u{0316}\u{0317}\u{0318}\u{0319}\u{031a}\u{031b}\u{031c}\u{031d}\u{032e}\u{0300}\u{0301}\u{0302}\u{0303}\u{0304}\u{0305}\u{0306}\u{0307}\u{0308}\u{0309}\u{030a}\u{030b}\u{030c}\u{030d}\u{030e}\u{030f}\u{0310}\u{0311}\u{0312}\u{0313}\u{0314}\u{0315}\u{0316}\u{0317}\u{0318}\u{0319}\u{031a}\u{031b}\u{031c}\u{031d}\u{032e}ngerzone";
131        let its_cool = "Da\u{0300}\u{0301}\u{0302}\u{0303}\u{0304}\u{0305}\u{0306}\u{0307}\u{0308}\u{0309}\u{030a}\u{030b}\u{030c}\u{030d}\u{030e}\u{030f}\u{0310}\u{0311}\u{0312}\u{0313}\u{0314}\u{0315}\u{0316}\u{0317}\u{0318}\u{0319}\u{031a}\u{031b}\u{031c}\u{031d}\u{034f}\u{032e}\u{0300}\u{0301}\u{0302}\u{0303}\u{0304}\u{0305}\u{0306}\u{0307}\u{0308}\u{0309}\u{030a}\u{030b}\u{030c}\u{030d}\u{030e}\u{030f}\u{0310}\u{0311}\u{0312}\u{0313}\u{0314}\u{0315}\u{0316}\u{0317}\u{0318}\u{0319}\u{031a}\u{031b}\u{031c}\u{034f}\u{031d}\u{032e}ngerzone";
132        assert_eq!(stream_safe(woah_nelly), its_cool);
133    }
134
135    #[test]
136    fn test_all_nonstarters() {
137        let s = "\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}\u{0300}";
138        let expected = "\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{034F}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}\u{300}";
139        assert_eq!(stream_safe(s), expected);
140    }
141
142    #[test]
143    fn test_classify_nonstarters() {
144        // Highest character in the `compat_fully_decomp` table is 2FA1D
145        for ch in 0..0x2FA1E {
146            let ch = match char::from_u32(ch) {
147                Some(c) => c,
148                None => continue,
149            };
150            let c = classify_nonstarters(ch);
151            let mut s = Vec::new();
152            decompose_compatible(ch, |c| s.push(c));
153
154            assert_eq!(s.len(), c.decomposition_len);
155
156            let num_leading = s
157                .iter()
158                .take_while(|&c| canonical_combining_class(*c) != 0)
159                .count();
160            let num_trailing = s
161                .iter()
162                .rev()
163                .take_while(|&c| canonical_combining_class(*c) != 0)
164                .count();
165
166            assert_eq!(num_leading, c.leading_nonstarters);
167            assert_eq!(num_trailing, c.trailing_nonstarters);
168        }
169    }
170}
unicode_normalization/stream_safe.rs

unicode_normalization/
stream_safe.rs
