`impl ScriptWithExtensionsBorrowed<'static>`

fn new() -> Self

Creates a new instance of ScriptWithExtensionsBorrowed using compiled data.

✨ Enabled with the compiled_data Cargo feature.

const fn static_to_owned(self: Self) -> ScriptWithExtensions

Cheaply converts a [ScriptWithExtensionsBorrowed<'static>] into a ScriptWithExtensions.

Note: Due to branching and indirection, using ScriptWithExtensions might inhibit some compile-time optimizations that are possible with ScriptWithExtensionsBorrowed.

`impl<'a> ScriptWithExtensionsBorrowed<'a>`

fn get_script_val(self: Self, ch: char) -> Script

Returns the Script property value for this code point.

Examples

use icu::properties::script::ScriptWithExtensions;
use icu::properties::props::Script;

let swe = ScriptWithExtensions::new();

// U+0640 ARABIC TATWEEL
assert_eq!(swe.get_script_val('ـ'), Script::Common); // main Script value
assert_ne!(swe.get_script_val('ـ'), Script::Arabic);
assert_ne!(swe.get_script_val('ـ'), Script::Syriac);
assert_ne!(swe.get_script_val('ـ'), Script::Thaana);

// U+0650 ARABIC KASRA
assert_eq!(swe.get_script_val('\u{0650}'), Script::Inherited); // main Script value
assert_ne!(swe.get_script_val('\u{0650}'), Script::Arabic);
assert_ne!(swe.get_script_val('\u{0650}'), Script::Syriac);
assert_ne!(swe.get_script_val('\u{0650}'), Script::Thaana);

// U+0660 ARABIC-INDIC DIGIT ZERO
assert_ne!(swe.get_script_val('٠'), Script::Common);
assert_eq!(swe.get_script_val('٠'), Script::Arabic); // main Script value
assert_ne!(swe.get_script_val('٠'), Script::Syriac);
assert_ne!(swe.get_script_val('٠'), Script::Thaana);

// U+FDF2 ARABIC LIGATURE ALLAH ISOLATED FORM
assert_ne!(swe.get_script_val('ﷲ'), Script::Common);
assert_eq!(swe.get_script_val('ﷲ'), Script::Arabic); // main Script value
assert_ne!(swe.get_script_val('ﷲ'), Script::Syriac);
assert_ne!(swe.get_script_val('ﷲ'), Script::Thaana);

fn get_script_val32(self: Self, code_point: u32) -> Script

See Self::get_script_val.

fn get_script_extensions_val(self: Self, ch: char) -> ScriptExtensionsSet<'a>

Return the Script_Extensions property value for this code point.

If code_point has Script_Extensions, then return the Script codes in the Script_Extensions. In this case, the Script property value (normally Common or Inherited) is not included in the ScriptExtensionsSet.

If c does not have Script_Extensions, then the one Script code is put into the ScriptExtensionsSet and also returned.

If c is not a valid code point, then return an empty ScriptExtensionsSet.

Examples

use icu::properties::script::ScriptWithExtensions;
use icu::properties::props::Script;

let swe = ScriptWithExtensions::new();

assert_eq!(
    swe.get_script_extensions_val('𐓐') // U+104D0 OSAGE CAPITAL LETTER KHA
        .iter()
        .collect::<Vec<_>>(),
    [Script::Osage]
);
assert_eq!(
    swe.get_script_extensions_val('🥳') // U+1F973 FACE WITH PARTY HORN AND PARTY HAT
        .iter()
        .collect::<Vec<_>>(),
    [Script::Common]
);
assert_eq!(
    swe.get_script_extensions_val('\u{200D}') // ZERO WIDTH JOINER
        .iter()
        .collect::<Vec<_>>(),
    [Script::Inherited]
);
assert_eq!(
    swe.get_script_extensions_val('௫') // U+0BEB TAMIL DIGIT FIVE
        .iter()
        .collect::<Vec<_>>(),
    [Script::Tamil, Script::Grantha]
);

fn get_script_extensions_val32(self: Self, code_point: u32) -> ScriptExtensionsSet<'a>

See Self::get_script_extensions_val.

fn has_script(self: Self, ch: char, script: Script) -> bool

Returns whether script is contained in the Script_Extensions property value if the code_point has Script_Extensions, otherwise if the code point does not have Script_Extensions then returns whether the Script property value matches.

Some characters are commonly used in multiple scripts. For more information, see UAX #24: http://www.unicode.org/reports/tr24/.

Examples

use icu::properties::script::ScriptWithExtensions;
use icu::properties::props::Script;

let swe = ScriptWithExtensions::new();

// U+0650 ARABIC KASRA
assert!(!swe.has_script('\u{0650}', Script::Inherited)); // main Script value
assert!(swe.has_script('\u{0650}', Script::Arabic));
assert!(swe.has_script('\u{0650}', Script::Syriac));
assert!(!swe.has_script('\u{0650}', Script::Thaana));

// U+0660 ARABIC-INDIC DIGIT ZERO
assert!(!swe.has_script('٠', Script::Common)); // main Script value
assert!(swe.has_script('٠', Script::Arabic));
assert!(!swe.has_script('٠', Script::Syriac));
assert!(swe.has_script('٠', Script::Thaana));

// U+FDF2 ARABIC LIGATURE ALLAH ISOLATED FORM
assert!(!swe.has_script('ﷲ', Script::Common));
assert!(swe.has_script('ﷲ', Script::Arabic)); // main Script value
assert!(!swe.has_script('ﷲ', Script::Syriac));
assert!(swe.has_script('ﷲ', Script::Thaana));

fn has_script32(self: Self, code_point: u32, script: Script) -> bool

See Self::has_script.

fn get_script_extensions_ranges(self: Self, script: Script) -> impl Iterator<Item = RangeInclusive<u32>> + 'a

Returns all of the matching CodePointMapRanges for the given Script in which has_script will return true for all of the contained code points.

Examples

use icu::properties::props::Script;
use icu::properties::script::ScriptWithExtensions;

let swe = ScriptWithExtensions::new();

let syriac_script_extensions_ranges =
    swe.get_script_extensions_ranges(Script::Syriac);

let exp_ranges = [
    0x0303..=0x0304, // COMBINING TILDE..COMBINING MACRON
    0x0307..=0x0308, // COMBINING DOT ABOVE..COMBINING DIAERESIS
    0x030A..=0x030A, // COMBINING RING ABOVE
    0x0320..=0x0320, // COMBINING MINUS SIGN BELOW
    0x0323..=0x0325, // COMBINING DOT BELOW..COMBINING RING BELOW
    0x032D..=0x032E, // COMBINING CIRCUMFLEX ACCENT BELOW..COMBINING BREVE BELOW
    0x0330..=0x0330, // COMBINING TILDE BELOW
    0x060C..=0x060C, // ARABIC COMMA
    0x061B..=0x061C, // ARABIC SEMICOLON, ARABIC LETTER MARK
    0x061F..=0x061F, // ARABIC QUESTION MARK
    0x0640..=0x0640, // ARABIC TATWEEL
    0x064B..=0x0655, // ARABIC FATHATAN..ARABIC HAMZA BELOW
    0x0670..=0x0670, // ARABIC LETTER SUPERSCRIPT ALEF
    0x0700..=0x070D, // Syriac block begins at U+0700
    0x070F..=0x074A, // Syriac block
    0x074D..=0x074F, // Syriac block ends at U+074F
    0x0860..=0x086A, // Syriac Supplement block is U+0860..=U+086F
    0x1DF8..=0x1DF8, // COMBINING DOT ABOVE LEFT
    0x1DFA..=0x1DFA, // COMBINING DOT BELOW LEFT
];

assert_eq!(
    syriac_script_extensions_ranges.collect::<Vec<_>>(),
    exp_ranges
);