394 lines
9.5 KiB
Rust
394 lines
9.5 KiB
Rust
/*!
|
|
|
|
This library implements
|
|
[Unicode Bidi Mirroring](https://unicode.org/reports/tr44/#BidiMirroring.txt) property detection.
|
|
|
|
```rust
|
|
use unicode_bidi_mirroring::*;
|
|
|
|
assert_eq!(get_mirrored('A'), None);
|
|
assert_eq!(get_mirrored('\u{2039}'), Some('\u{203A}'));
|
|
assert_eq!(get_mirrored('\u{203A}'), Some('\u{2039}'));
|
|
|
|
assert_eq!(is_mirroring('A'), false);
|
|
assert_eq!(is_mirroring('\u{29C4}'), true);
|
|
assert_eq!(is_mirroring('\u{22FF}'), true);
|
|
```
|
|
|
|
*/
|
|
|
|
#![no_std]
|
|
#![forbid(unsafe_code)]
|
|
|
|
/// The Unicode version.
|
|
pub const UNICODE_VERSION: (u8, u8, u8) = (14, 0, 0);
|
|
|
|
const PAIRS: &[(char, char)] = &[
|
|
('\u{0028}', '\u{0029}'),
|
|
('\u{003C}', '\u{003E}'),
|
|
('\u{005B}', '\u{005D}'),
|
|
('\u{007B}', '\u{007D}'),
|
|
('\u{00AB}', '\u{00BB}'),
|
|
('\u{0F3A}', '\u{0F3B}'),
|
|
('\u{0F3C}', '\u{0F3D}'),
|
|
('\u{169B}', '\u{169C}'),
|
|
('\u{2039}', '\u{203A}'),
|
|
('\u{2045}', '\u{2046}'),
|
|
('\u{207D}', '\u{207E}'),
|
|
('\u{208D}', '\u{208E}'),
|
|
('\u{2208}', '\u{220B}'),
|
|
('\u{2209}', '\u{220C}'),
|
|
('\u{220A}', '\u{220D}'),
|
|
('\u{2215}', '\u{29F5}'),
|
|
('\u{221F}', '\u{2BFE}'),
|
|
('\u{2220}', '\u{29A3}'),
|
|
('\u{2221}', '\u{299B}'),
|
|
('\u{2222}', '\u{29A0}'),
|
|
('\u{2224}', '\u{2AEE}'),
|
|
('\u{223C}', '\u{223D}'),
|
|
('\u{2243}', '\u{22CD}'),
|
|
('\u{2245}', '\u{224C}'),
|
|
('\u{2252}', '\u{2253}'),
|
|
('\u{2254}', '\u{2255}'),
|
|
('\u{2264}', '\u{2265}'),
|
|
('\u{2266}', '\u{2267}'),
|
|
('\u{2268}', '\u{2269}'),
|
|
('\u{226A}', '\u{226B}'),
|
|
('\u{226E}', '\u{226F}'),
|
|
('\u{2270}', '\u{2271}'),
|
|
('\u{2272}', '\u{2273}'),
|
|
('\u{2274}', '\u{2275}'),
|
|
('\u{2276}', '\u{2277}'),
|
|
('\u{2278}', '\u{2279}'),
|
|
('\u{227A}', '\u{227B}'),
|
|
('\u{227C}', '\u{227D}'),
|
|
('\u{227E}', '\u{227F}'),
|
|
('\u{2280}', '\u{2281}'),
|
|
('\u{2282}', '\u{2283}'),
|
|
('\u{2284}', '\u{2285}'),
|
|
('\u{2286}', '\u{2287}'),
|
|
('\u{2288}', '\u{2289}'),
|
|
('\u{228A}', '\u{228B}'),
|
|
('\u{228F}', '\u{2290}'),
|
|
('\u{2291}', '\u{2292}'),
|
|
('\u{2298}', '\u{29B8}'),
|
|
('\u{22A2}', '\u{22A3}'),
|
|
('\u{22A6}', '\u{2ADE}'),
|
|
('\u{22A8}', '\u{2AE4}'),
|
|
('\u{22A9}', '\u{2AE3}'),
|
|
('\u{22AB}', '\u{2AE5}'),
|
|
('\u{22B0}', '\u{22B1}'),
|
|
('\u{22B2}', '\u{22B3}'),
|
|
('\u{22B4}', '\u{22B5}'),
|
|
('\u{22B6}', '\u{22B7}'),
|
|
('\u{22B8}', '\u{27DC}'),
|
|
('\u{22C9}', '\u{22CA}'),
|
|
('\u{22CB}', '\u{22CC}'),
|
|
('\u{22D0}', '\u{22D1}'),
|
|
('\u{22D6}', '\u{22D7}'),
|
|
('\u{22D8}', '\u{22D9}'),
|
|
('\u{22DA}', '\u{22DB}'),
|
|
('\u{22DC}', '\u{22DD}'),
|
|
('\u{22DE}', '\u{22DF}'),
|
|
('\u{22E0}', '\u{22E1}'),
|
|
('\u{22E2}', '\u{22E3}'),
|
|
('\u{22E4}', '\u{22E5}'),
|
|
('\u{22E6}', '\u{22E7}'),
|
|
('\u{22E8}', '\u{22E9}'),
|
|
('\u{22EA}', '\u{22EB}'),
|
|
('\u{22EC}', '\u{22ED}'),
|
|
('\u{22F0}', '\u{22F1}'),
|
|
('\u{22F2}', '\u{22FA}'),
|
|
('\u{22F3}', '\u{22FB}'),
|
|
('\u{22F4}', '\u{22FC}'),
|
|
('\u{22F6}', '\u{22FD}'),
|
|
('\u{22F7}', '\u{22FE}'),
|
|
('\u{2308}', '\u{2309}'),
|
|
('\u{230A}', '\u{230B}'),
|
|
('\u{2329}', '\u{232A}'),
|
|
('\u{2768}', '\u{2769}'),
|
|
('\u{276A}', '\u{276B}'),
|
|
('\u{276C}', '\u{276D}'),
|
|
('\u{276E}', '\u{276F}'),
|
|
('\u{2770}', '\u{2771}'),
|
|
('\u{2772}', '\u{2773}'),
|
|
('\u{2774}', '\u{2775}'),
|
|
('\u{27C3}', '\u{27C4}'),
|
|
('\u{27C5}', '\u{27C6}'),
|
|
('\u{27C8}', '\u{27C9}'),
|
|
('\u{27CB}', '\u{27CD}'),
|
|
('\u{27D5}', '\u{27D6}'),
|
|
('\u{27DD}', '\u{27DE}'),
|
|
('\u{27E2}', '\u{27E3}'),
|
|
('\u{27E4}', '\u{27E5}'),
|
|
('\u{27E6}', '\u{27E7}'),
|
|
('\u{27E8}', '\u{27E9}'),
|
|
('\u{27EA}', '\u{27EB}'),
|
|
('\u{27EC}', '\u{27ED}'),
|
|
('\u{27EE}', '\u{27EF}'),
|
|
('\u{2983}', '\u{2984}'),
|
|
('\u{2985}', '\u{2986}'),
|
|
('\u{2987}', '\u{2988}'),
|
|
('\u{2989}', '\u{298A}'),
|
|
('\u{298B}', '\u{298C}'),
|
|
('\u{298D}', '\u{2990}'),
|
|
('\u{298E}', '\u{298F}'),
|
|
('\u{2991}', '\u{2992}'),
|
|
('\u{2993}', '\u{2994}'),
|
|
('\u{2995}', '\u{2996}'),
|
|
('\u{2997}', '\u{2998}'),
|
|
('\u{29A4}', '\u{29A5}'),
|
|
('\u{29A8}', '\u{29A9}'),
|
|
('\u{29AA}', '\u{29AB}'),
|
|
('\u{29AC}', '\u{29AD}'),
|
|
('\u{29AE}', '\u{29AF}'),
|
|
('\u{29C0}', '\u{29C1}'),
|
|
('\u{29C4}', '\u{29C5}'),
|
|
('\u{29CF}', '\u{29D0}'),
|
|
('\u{29D1}', '\u{29D2}'),
|
|
('\u{29D4}', '\u{29D5}'),
|
|
('\u{29D8}', '\u{29D9}'),
|
|
('\u{29DA}', '\u{29DB}'),
|
|
('\u{29E8}', '\u{29E9}'),
|
|
('\u{29F8}', '\u{29F9}'),
|
|
('\u{29FC}', '\u{29FD}'),
|
|
('\u{2A2B}', '\u{2A2C}'),
|
|
('\u{2A2D}', '\u{2A2E}'),
|
|
('\u{2A34}', '\u{2A35}'),
|
|
('\u{2A3C}', '\u{2A3D}'),
|
|
('\u{2A64}', '\u{2A65}'),
|
|
('\u{2A79}', '\u{2A7A}'),
|
|
('\u{2A7B}', '\u{2A7C}'),
|
|
('\u{2A7D}', '\u{2A7E}'),
|
|
('\u{2A7F}', '\u{2A80}'),
|
|
('\u{2A81}', '\u{2A82}'),
|
|
('\u{2A83}', '\u{2A84}'),
|
|
('\u{2A85}', '\u{2A86}'),
|
|
('\u{2A87}', '\u{2A88}'),
|
|
('\u{2A89}', '\u{2A8A}'),
|
|
('\u{2A8B}', '\u{2A8C}'),
|
|
('\u{2A8D}', '\u{2A8E}'),
|
|
('\u{2A8F}', '\u{2A90}'),
|
|
('\u{2A91}', '\u{2A92}'),
|
|
('\u{2A93}', '\u{2A94}'),
|
|
('\u{2A95}', '\u{2A96}'),
|
|
('\u{2A97}', '\u{2A98}'),
|
|
('\u{2A99}', '\u{2A9A}'),
|
|
('\u{2A9B}', '\u{2A9C}'),
|
|
('\u{2A9D}', '\u{2A9E}'),
|
|
('\u{2A9F}', '\u{2AA0}'),
|
|
('\u{2AA1}', '\u{2AA2}'),
|
|
('\u{2AA6}', '\u{2AA7}'),
|
|
('\u{2AA8}', '\u{2AA9}'),
|
|
('\u{2AAA}', '\u{2AAB}'),
|
|
('\u{2AAC}', '\u{2AAD}'),
|
|
('\u{2AAF}', '\u{2AB0}'),
|
|
('\u{2AB1}', '\u{2AB2}'),
|
|
('\u{2AB3}', '\u{2AB4}'),
|
|
('\u{2AB5}', '\u{2AB6}'),
|
|
('\u{2AB7}', '\u{2AB8}'),
|
|
('\u{2AB9}', '\u{2ABA}'),
|
|
('\u{2ABB}', '\u{2ABC}'),
|
|
('\u{2ABD}', '\u{2ABE}'),
|
|
('\u{2ABF}', '\u{2AC0}'),
|
|
('\u{2AC1}', '\u{2AC2}'),
|
|
('\u{2AC3}', '\u{2AC4}'),
|
|
('\u{2AC5}', '\u{2AC6}'),
|
|
('\u{2AC7}', '\u{2AC8}'),
|
|
('\u{2AC9}', '\u{2ACA}'),
|
|
('\u{2ACB}', '\u{2ACC}'),
|
|
('\u{2ACD}', '\u{2ACE}'),
|
|
('\u{2ACF}', '\u{2AD0}'),
|
|
('\u{2AD1}', '\u{2AD2}'),
|
|
('\u{2AD3}', '\u{2AD4}'),
|
|
('\u{2AD5}', '\u{2AD6}'),
|
|
('\u{2AEC}', '\u{2AED}'),
|
|
('\u{2AF7}', '\u{2AF8}'),
|
|
('\u{2AF9}', '\u{2AFA}'),
|
|
('\u{2E02}', '\u{2E03}'),
|
|
('\u{2E04}', '\u{2E05}'),
|
|
('\u{2E09}', '\u{2E0A}'),
|
|
('\u{2E0C}', '\u{2E0D}'),
|
|
('\u{2E1C}', '\u{2E1D}'),
|
|
('\u{2E20}', '\u{2E21}'),
|
|
('\u{2E22}', '\u{2E23}'),
|
|
('\u{2E24}', '\u{2E25}'),
|
|
('\u{2E26}', '\u{2E27}'),
|
|
('\u{2E28}', '\u{2E29}'),
|
|
('\u{2E55}', '\u{2E56}'),
|
|
('\u{2E57}', '\u{2E58}'),
|
|
('\u{2E59}', '\u{2E5A}'),
|
|
('\u{2E5B}', '\u{2E5C}'),
|
|
('\u{3008}', '\u{3009}'),
|
|
('\u{300A}', '\u{300B}'),
|
|
('\u{300C}', '\u{300D}'),
|
|
('\u{300E}', '\u{300F}'),
|
|
('\u{3010}', '\u{3011}'),
|
|
('\u{3014}', '\u{3015}'),
|
|
('\u{3016}', '\u{3017}'),
|
|
('\u{3018}', '\u{3019}'),
|
|
('\u{301A}', '\u{301B}'),
|
|
('\u{FE59}', '\u{FE5A}'),
|
|
('\u{FE5B}', '\u{FE5C}'),
|
|
('\u{FE5D}', '\u{FE5E}'),
|
|
('\u{FE64}', '\u{FE65}'),
|
|
('\u{FF08}', '\u{FF09}'),
|
|
('\u{FF1C}', '\u{FF1E}'),
|
|
('\u{FF3B}', '\u{FF3D}'),
|
|
('\u{FF5B}', '\u{FF5D}'),
|
|
('\u{FF5F}', '\u{FF60}'),
|
|
('\u{FF62}', '\u{FF63}'),
|
|
];
|
|
|
|
const OTHER: &[char] = &[
|
|
'\u{2140}',
|
|
'\u{2201}',
|
|
'\u{2202}',
|
|
'\u{2203}',
|
|
'\u{2204}',
|
|
'\u{2211}',
|
|
'\u{2216}',
|
|
'\u{221A}',
|
|
'\u{221B}',
|
|
'\u{221C}',
|
|
'\u{221D}',
|
|
'\u{2226}',
|
|
'\u{222B}',
|
|
'\u{222C}',
|
|
'\u{222D}',
|
|
'\u{222E}',
|
|
'\u{222F}',
|
|
'\u{2230}',
|
|
'\u{2231}',
|
|
'\u{2232}',
|
|
'\u{2233}',
|
|
'\u{2239}',
|
|
'\u{223B}',
|
|
'\u{223E}',
|
|
'\u{223F}',
|
|
'\u{2240}',
|
|
'\u{2241}',
|
|
'\u{2242}',
|
|
'\u{2244}',
|
|
'\u{2246}',
|
|
'\u{2247}',
|
|
'\u{2248}',
|
|
'\u{2249}',
|
|
'\u{224A}',
|
|
'\u{224B}',
|
|
'\u{225F}',
|
|
'\u{2260}',
|
|
'\u{2262}',
|
|
'\u{228C}',
|
|
'\u{22A7}',
|
|
'\u{22AA}',
|
|
'\u{22AC}',
|
|
'\u{22AD}',
|
|
'\u{22AE}',
|
|
'\u{22AF}',
|
|
'\u{22BE}',
|
|
'\u{22BF}',
|
|
'\u{22F5}',
|
|
'\u{22F8}',
|
|
'\u{22F9}',
|
|
'\u{22FF}',
|
|
'\u{2320}',
|
|
'\u{2321}',
|
|
'\u{27C0}',
|
|
'\u{27CC}',
|
|
'\u{27D3}',
|
|
'\u{27D4}',
|
|
'\u{299C}',
|
|
'\u{299D}',
|
|
'\u{299E}',
|
|
'\u{299F}',
|
|
'\u{29A2}',
|
|
'\u{29A6}',
|
|
'\u{29A7}',
|
|
'\u{29C2}',
|
|
'\u{29C3}',
|
|
'\u{29C9}',
|
|
'\u{29CE}',
|
|
'\u{29DC}',
|
|
'\u{29E1}',
|
|
'\u{29E3}',
|
|
'\u{29E4}',
|
|
'\u{29E5}',
|
|
'\u{29F4}',
|
|
'\u{29F6}',
|
|
'\u{29F7}',
|
|
'\u{2A0A}',
|
|
'\u{2A0B}',
|
|
'\u{2A0C}',
|
|
'\u{2A0D}',
|
|
'\u{2A0E}',
|
|
'\u{2A0F}',
|
|
'\u{2A10}',
|
|
'\u{2A11}',
|
|
'\u{2A12}',
|
|
'\u{2A13}',
|
|
'\u{2A14}',
|
|
'\u{2A15}',
|
|
'\u{2A16}',
|
|
'\u{2A17}',
|
|
'\u{2A18}',
|
|
'\u{2A19}',
|
|
'\u{2A1A}',
|
|
'\u{2A1B}',
|
|
'\u{2A1C}',
|
|
'\u{2A1E}',
|
|
'\u{2A1F}',
|
|
'\u{2A20}',
|
|
'\u{2A21}',
|
|
'\u{2A24}',
|
|
'\u{2A26}',
|
|
'\u{2A29}',
|
|
'\u{2A3E}',
|
|
'\u{2A57}',
|
|
'\u{2A58}',
|
|
'\u{2A6A}',
|
|
'\u{2A6B}',
|
|
'\u{2A6C}',
|
|
'\u{2A6D}',
|
|
'\u{2A6F}',
|
|
'\u{2A70}',
|
|
'\u{2A73}',
|
|
'\u{2A74}',
|
|
'\u{2AA3}',
|
|
'\u{2ADC}',
|
|
'\u{2AE2}',
|
|
'\u{2AE6}',
|
|
'\u{2AF3}',
|
|
'\u{2AFB}',
|
|
'\u{2AFD}',
|
|
'\u{1D6DB}',
|
|
'\u{1D715}',
|
|
'\u{1D74F}',
|
|
'\u{1D789}',
|
|
'\u{1D7C3}',
|
|
];
|
|
|
|
/// Returns a bidi mirrored character.
|
|
///
|
|
/// Based on <https://www.unicode.org/Public/UNIDATA/BidiMirroring.txt>.
|
|
pub fn get_mirrored(c: char) -> Option<char> {
|
|
if let Ok(idx) = PAIRS.binary_search_by(|v| v.0.cmp(&c)) {
|
|
return Some(PAIRS[idx].1);
|
|
}
|
|
|
|
if let Ok(idx) = PAIRS.binary_search_by(|v| v.1.cmp(&c)) {
|
|
return Some(PAIRS[idx].0);
|
|
}
|
|
|
|
None
|
|
}
|
|
|
|
/// Checks that character is bidi mirrored.
|
|
///
|
|
/// Based on <https://www.unicode.org/Public/UNIDATA/BidiMirroring.txt>.
|
|
pub fn is_mirroring(c: char) -> bool {
|
|
PAIRS.binary_search_by(|v| v.0.cmp(&c)).is_ok() ||
|
|
PAIRS.binary_search_by(|v| v.1.cmp(&c)).is_ok() ||
|
|
OTHER.binary_search_by(|v| v.cmp(&c)).is_ok()
|
|
}
|