| 1 | // This file is part of ICU4X. For terms of use, please see the file |
| 2 | // called LICENSE at the top level of the ICU4X source tree |
| 3 | // (online at: https://github.com/unicode-org/icu4x/blob/main/LICENSE ). |
| 4 | |
| 5 | //! Other Use Extensions is a list of extensions other than unicode, |
| 6 | //! transform or private. |
| 7 | //! |
| 8 | //! Those extensions are treated as a pass-through, and no Unicode related |
| 9 | //! behavior depends on them. |
| 10 | //! |
| 11 | //! The main struct for this extension is [`Other`] which is a list of [`Subtag`]s. |
| 12 | //! |
| 13 | //! # Examples |
| 14 | //! |
| 15 | //! ``` |
| 16 | //! use icu::locid::extensions::other::Other; |
| 17 | //! use icu::locid::Locale; |
| 18 | //! |
| 19 | //! let mut loc: Locale = "en-US-a-foo-faa" .parse().expect("Parsing failed." ); |
| 20 | //! ``` |
| 21 | |
| 22 | mod subtag; |
| 23 | |
| 24 | use crate::parser::ParserError; |
| 25 | use crate::parser::SubtagIterator; |
| 26 | use crate::shortvec::ShortBoxSlice; |
| 27 | use alloc::vec::Vec; |
| 28 | #[doc (inline)] |
| 29 | pub use subtag::{subtag, Subtag}; |
| 30 | |
| 31 | /// A list of [`Other Use Extensions`] as defined in [`Unicode Locale |
| 32 | /// Identifier`] specification. |
| 33 | /// |
| 34 | /// Those extensions are treated as a pass-through, and no Unicode related |
| 35 | /// behavior depends on them. |
| 36 | /// |
| 37 | /// # Examples |
| 38 | /// |
| 39 | /// ``` |
| 40 | /// use icu::locid::extensions::other::{Other, Subtag}; |
| 41 | /// |
| 42 | /// let subtag1: Subtag = "foo" .parse().expect("Failed to parse a Subtag." ); |
| 43 | /// let subtag2: Subtag = "bar" .parse().expect("Failed to parse a Subtag." ); |
| 44 | /// |
| 45 | /// let other = Other::from_vec_unchecked(b'a' , vec![subtag1, subtag2]); |
| 46 | /// assert_eq!(&other.to_string(), "a-foo-bar" ); |
| 47 | /// ``` |
| 48 | /// |
| 49 | /// [`Other Use Extensions`]: https://unicode.org/reports/tr35/#other_extensions |
| 50 | /// [`Unicode Locale Identifier`]: https://unicode.org/reports/tr35/#Unicode_locale_identifier |
| 51 | #[derive (Clone, PartialEq, Eq, Debug, Default, Hash, PartialOrd, Ord)] |
| 52 | pub struct Other { |
| 53 | ext: u8, |
| 54 | keys: ShortBoxSlice<Subtag>, |
| 55 | } |
| 56 | |
| 57 | impl Other { |
| 58 | /// A constructor which takes a pre-sorted list of [`Subtag`]. |
| 59 | /// |
| 60 | /// # Panics |
| 61 | /// |
| 62 | /// Panics if `ext` is not ASCII alphabetic. |
| 63 | /// |
| 64 | /// # Examples |
| 65 | /// |
| 66 | /// ``` |
| 67 | /// use icu::locid::extensions::other::{Other, Subtag}; |
| 68 | /// |
| 69 | /// let subtag1: Subtag = "foo" .parse().expect("Failed to parse a Subtag." ); |
| 70 | /// let subtag2: Subtag = "bar" .parse().expect("Failed to parse a Subtag." ); |
| 71 | /// |
| 72 | /// let other = Other::from_vec_unchecked(b'a' , vec![subtag1, subtag2]); |
| 73 | /// assert_eq!(&other.to_string(), "a-foo-bar" ); |
| 74 | /// ``` |
| 75 | pub fn from_vec_unchecked(ext: u8, keys: Vec<Subtag>) -> Self { |
| 76 | Self::from_short_slice_unchecked(ext, keys.into()) |
| 77 | } |
| 78 | |
| 79 | pub(crate) fn from_short_slice_unchecked(ext: u8, keys: ShortBoxSlice<Subtag>) -> Self { |
| 80 | assert!(ext.is_ascii_alphabetic()); |
| 81 | Self { ext, keys } |
| 82 | } |
| 83 | |
| 84 | pub(crate) fn try_from_iter(ext: u8, iter: &mut SubtagIterator) -> Result<Self, ParserError> { |
| 85 | debug_assert!(ext.is_ascii_alphabetic()); |
| 86 | |
| 87 | let mut keys = ShortBoxSlice::new(); |
| 88 | while let Some(subtag) = iter.peek() { |
| 89 | if !Subtag::valid_key(subtag) { |
| 90 | break; |
| 91 | } |
| 92 | if let Ok(key) = Subtag::try_from_bytes(subtag) { |
| 93 | keys.push(key); |
| 94 | } |
| 95 | iter.next(); |
| 96 | } |
| 97 | |
| 98 | Ok(Self::from_short_slice_unchecked(ext, keys)) |
| 99 | } |
| 100 | |
| 101 | /// Gets the tag character for this extension as a &str. |
| 102 | /// |
| 103 | /// # Examples |
| 104 | /// |
| 105 | /// ``` |
| 106 | /// use icu::locid::Locale; |
| 107 | /// |
| 108 | /// let loc: Locale = "und-a-hello-world" .parse().unwrap(); |
| 109 | /// let other_ext = &loc.extensions.other[0]; |
| 110 | /// assert_eq!(other_ext.get_ext_str(), "a" ); |
| 111 | /// ``` |
| 112 | pub fn get_ext_str(&self) -> &str { |
| 113 | debug_assert!(self.ext.is_ascii_alphabetic()); |
| 114 | unsafe { core::str::from_utf8_unchecked(core::slice::from_ref(&self.ext)) } |
| 115 | } |
| 116 | |
| 117 | /// Gets the tag character for this extension as a char. |
| 118 | /// |
| 119 | /// # Examples |
| 120 | /// |
| 121 | /// ``` |
| 122 | /// use icu::locid::Locale; |
| 123 | /// |
| 124 | /// let loc: Locale = "und-a-hello-world" .parse().unwrap(); |
| 125 | /// let other_ext = &loc.extensions.other[0]; |
| 126 | /// assert_eq!(other_ext.get_ext(), 'a' ); |
| 127 | /// ``` |
| 128 | pub fn get_ext(&self) -> char { |
| 129 | self.ext as char |
| 130 | } |
| 131 | |
| 132 | /// Gets the tag character for this extension as a byte. |
| 133 | /// |
| 134 | /// # Examples |
| 135 | /// |
| 136 | /// ``` |
| 137 | /// use icu::locid::Locale; |
| 138 | /// |
| 139 | /// let loc: Locale = "und-a-hello-world" .parse().unwrap(); |
| 140 | /// let other_ext = &loc.extensions.other[0]; |
| 141 | /// assert_eq!(other_ext.get_ext_byte(), b'a' ); |
| 142 | /// ``` |
| 143 | pub fn get_ext_byte(&self) -> u8 { |
| 144 | self.ext |
| 145 | } |
| 146 | |
| 147 | pub(crate) fn for_each_subtag_str<E, F>(&self, f: &mut F) -> Result<(), E> |
| 148 | where |
| 149 | F: FnMut(&str) -> Result<(), E>, |
| 150 | { |
| 151 | f(self.get_ext_str())?; |
| 152 | self.keys.iter().map(|t| t.as_str()).try_for_each(f) |
| 153 | } |
| 154 | } |
| 155 | |
| 156 | writeable::impl_display_with_writeable!(Other); |
| 157 | |
| 158 | impl writeable::Writeable for Other { |
| 159 | fn write_to<W: core::fmt::Write + ?Sized>(&self, sink: &mut W) -> core::fmt::Result { |
| 160 | sink.write_str(self.get_ext_str())?; |
| 161 | for key in self.keys.iter() { |
| 162 | sink.write_char('-' )?; |
| 163 | writeable::Writeable::write_to(key, sink)?; |
| 164 | } |
| 165 | |
| 166 | Ok(()) |
| 167 | } |
| 168 | |
| 169 | fn writeable_length_hint(&self) -> writeable::LengthHint { |
| 170 | let mut result = writeable::LengthHint::exact(1); |
| 171 | for key in self.keys.iter() { |
| 172 | result += writeable::Writeable::writeable_length_hint(key) + 1; |
| 173 | } |
| 174 | result |
| 175 | } |
| 176 | |
| 177 | fn write_to_string(&self) -> alloc::borrow::Cow<str> { |
| 178 | if self.keys.is_empty() { |
| 179 | return alloc::borrow::Cow::Borrowed(self.get_ext_str()); |
| 180 | } |
| 181 | let mut string = |
| 182 | alloc::string::String::with_capacity(self.writeable_length_hint().capacity()); |
| 183 | let _ = self.write_to(&mut string); |
| 184 | alloc::borrow::Cow::Owned(string) |
| 185 | } |
| 186 | } |
| 187 | |