|
| 1 | +//! [Mixed-script detection](https://www.unicode.org/reports/tr39/#Mixed_Script_Detection) |
| 2 | +
|
| 3 | +use unicode_script::{Script,ScriptExtension}; |
| 4 | + |
| 5 | +/// An Augmented script set, as defined by UTS 39 |
| 6 | +/// |
| 7 | +/// https://www.unicode.org/reports/tr39/#def-augmented-script-set |
| 8 | +pubstructAugmentedScriptSet{ |
| 9 | +/// The base ScriptExtension value |
| 10 | +pubbase:ScriptExtension, |
| 11 | +/// Han With Bopomofo |
| 12 | +pubhanb:bool, |
| 13 | +/// Japanese |
| 14 | +pubjpan:bool, |
| 15 | +/// Korean |
| 16 | +pubkore:bool, |
| 17 | +} |
| 18 | + |
| 19 | +implFrom<ScriptExtension>forAugmentedScriptSet{ |
| 20 | +fnfrom(ext:ScriptExtension) ->Self{ |
| 21 | +letmut hanb =false; |
| 22 | +letmut jpan =false; |
| 23 | +letmut kore =false; |
| 24 | + |
| 25 | +if ext.contains_script(Script::Han){ |
| 26 | + hanb =true; |
| 27 | + jpan =true; |
| 28 | + kore =true; |
| 29 | +}else{ |
| 30 | +if ext.contains_script(Script::Hiragana) || ext.contains_script(Script::Katakana){ |
| 31 | + jpan =true; |
| 32 | +} |
| 33 | + |
| 34 | +if ext.contains_script(Script::Hangul){ |
| 35 | + kore =true; |
| 36 | +} |
| 37 | + |
| 38 | +if ext.contains_script(Script::Bopomofo){ |
| 39 | + hanb =true; |
| 40 | +} |
| 41 | +} |
| 42 | +Self{ |
| 43 | +base: ext, |
| 44 | + hanb, jpan, kore |
| 45 | +} |
| 46 | +} |
| 47 | +} |
| 48 | + |
| 49 | +implFrom<char>forAugmentedScriptSet{ |
| 50 | +fnfrom(c:char) ->Self{ |
| 51 | +AugmentedScriptSet::for_char(c) |
| 52 | +} |
| 53 | +} |
| 54 | + |
| 55 | +implFrom<&'_str>forAugmentedScriptSet{ |
| 56 | +fnfrom(s:&'_str) ->Self{ |
| 57 | +AugmentedScriptSet::for_str(s) |
| 58 | +} |
| 59 | +} |
| 60 | + |
| 61 | +implDefaultforAugmentedScriptSet{ |
| 62 | +fndefault() ->Self{ |
| 63 | +AugmentedScriptSet{ |
| 64 | +base:ScriptExtension::Single(Script::Common), |
| 65 | +hanb:true, |
| 66 | +jpan:true, |
| 67 | +kore:true, |
| 68 | +} |
| 69 | +} |
| 70 | +} |
| 71 | + |
| 72 | +implAugmentedScriptSet{ |
| 73 | +/// Intersect this set with another |
| 74 | +pubfnintersect(mutself,other:Self) ->Self{ |
| 75 | +self.base =self.base.intersect(other.base); |
| 76 | +self.hanb =self.hanb && other.hanb; |
| 77 | +self.jpan =self.jpan && other.jpan; |
| 78 | +self.kore =self.kore && other.kore; |
| 79 | +self |
| 80 | +} |
| 81 | + |
| 82 | +/// Check if the set is empty |
| 83 | +pubfnis_empty(&self) ->bool{ |
| 84 | +self.base.is_empty() && !self.hanb && !self.jpan && !self.kore |
| 85 | +} |
| 86 | + |
| 87 | +/// Construct an AugmentedScriptSet for a given character |
| 88 | +pubfnfor_char(c:char) ->Self{ |
| 89 | +ScriptExtension::from(c).into() |
| 90 | +} |
| 91 | + |
| 92 | +/// Find the [resolved script set](https://www.unicode.org/reports/tr39/#def-resolved-script-set) of a given string |
| 93 | +pubfnfor_str(s:&str) ->Self{ |
| 94 | +letmut set =AugmentedScriptSet::default(); |
| 95 | +for chin s.chars(){ |
| 96 | + set = set.intersect(ch.into()) |
| 97 | +} |
| 98 | + set |
| 99 | +} |
| 100 | +} |
| 101 | + |
| 102 | +/// Extension trait for [mixed-script detection](https://www.unicode.org/reports/tr39/#Mixed_Script_Detection) |
| 103 | +pubtraitUnicodeMixedScript{ |
| 104 | +/// Check if a string is [single-script](https://www.unicode.org/reports/tr39/#def-single-script) |
| 105 | +/// |
| 106 | +/// Note that a single-script string may still contain multiple Script properties! |
| 107 | +fnis_single_script(self) ->bool; |
| 108 | + |
| 109 | +/// Find the [resolved script set](https://www.unicode.org/reports/tr39/#def-resolved-script-set) of a given string |
| 110 | +fnresolve_script_set(self) ->AugmentedScriptSet; |
| 111 | +} |
| 112 | + |
| 113 | +implUnicodeMixedScriptfor&'_str{ |
| 114 | +fnis_single_script(self) ->bool{ |
| 115 | + !AugmentedScriptSet::for_str(self).is_empty() |
| 116 | +} |
| 117 | + |
| 118 | +fnresolve_script_set(self) ->AugmentedScriptSet{ |
| 119 | +self.into() |
| 120 | +} |
| 121 | +} |