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::locale::extensions::other::Other; 17 //! use icu::locale::Locale; 18 //! 19 //! let mut loc: Locale = "en-US-a-foo-faa".parse().expect("Parsing failed."); 20 //! ``` 21 22 #[cfg(feature = "alloc")] 23 use core::str::FromStr; 24 25 #[cfg(feature = "alloc")] 26 use super::ExtensionType; 27 #[cfg(feature = "alloc")] 28 use crate::parser::ParseError; 29 #[cfg(feature = "alloc")] 30 use crate::parser::SubtagIterator; 31 use crate::shortvec::ShortBoxSlice; 32 use crate::subtags::Subtag; 33 #[cfg(feature = "alloc")] 34 use alloc::vec::Vec; 35 36 /// A list of [`Other Use Extensions`] as defined in [`Unicode Locale 37 /// Identifier`] specification. 38 /// 39 /// Those extensions are treated as a pass-through, and no Unicode related 40 /// behavior depends on them. 41 /// 42 /// # Examples 43 /// 44 /// ``` 45 /// use icu::locale::extensions::other::Other; 46 /// use icu::locale::subtags::Subtag; 47 /// 48 /// let subtag1: Subtag = "foo".parse().expect("Failed to parse a Subtag."); 49 /// let subtag2: Subtag = "bar".parse().expect("Failed to parse a Subtag."); 50 /// 51 /// let other = Other::from_vec_unchecked(b'a', vec![subtag1, subtag2]); 52 /// assert_eq!(&other.to_string(), "a-foo-bar"); 53 /// ``` 54 /// 55 /// [`Other Use Extensions`]: https://unicode.org/reports/tr35/#other_extensions 56 /// [`Unicode Locale Identifier`]: https://unicode.org/reports/tr35/#Unicode_locale_identifier 57 #[derive(Clone, PartialEq, Eq, Debug, Default, Hash, PartialOrd, Ord)] 58 pub struct Other { 59 // Safety invariant: must be ASCII 60 ext: u8, 61 keys: ShortBoxSlice<Subtag>, 62 } 63 64 impl Other { 65 /// A constructor which takes a str slice, parses it and 66 /// produces a well-formed [`Other`]. 67 #[inline] 68 #[cfg(feature = "alloc")] try_from_str(s: &str) -> Result<Self, ParseError>69 pub fn try_from_str(s: &str) -> Result<Self, ParseError> { 70 Self::try_from_utf8(s.as_bytes()) 71 } 72 73 /// See [`Self::try_from_str`] 74 #[cfg(feature = "alloc")] try_from_utf8(code_units: &[u8]) -> Result<Self, ParseError>75 pub fn try_from_utf8(code_units: &[u8]) -> Result<Self, ParseError> { 76 let mut iter = SubtagIterator::new(code_units); 77 78 let ext = iter.next().ok_or(ParseError::InvalidExtension)?; 79 if let ExtensionType::Other(b) = ExtensionType::try_from_byte_slice(ext)? { 80 return Self::try_from_iter(b, &mut iter); 81 } 82 83 Err(ParseError::InvalidExtension) 84 } 85 86 /// A constructor which takes a pre-sorted list of [`Subtag`]. 87 /// 88 /// # Panics 89 /// 90 /// Panics if `ext` is not ASCII alphabetic. 91 /// 92 /// # Examples 93 /// 94 /// ``` 95 /// use icu::locale::extensions::other::Other; 96 /// use icu::locale::subtags::Subtag; 97 /// 98 /// let subtag1: Subtag = "foo".parse().expect("Failed to parse a Subtag."); 99 /// let subtag2: Subtag = "bar".parse().expect("Failed to parse a Subtag."); 100 /// 101 /// let other = Other::from_vec_unchecked(b'a', vec![subtag1, subtag2]); 102 /// assert_eq!(&other.to_string(), "a-foo-bar"); 103 /// ``` 104 #[cfg(feature = "alloc")] from_vec_unchecked(ext: u8, keys: Vec<Subtag>) -> Self105 pub fn from_vec_unchecked(ext: u8, keys: Vec<Subtag>) -> Self { 106 Self::from_short_slice_unchecked(ext, keys.into()) 107 } 108 109 #[allow(dead_code)] from_short_slice_unchecked(ext: u8, keys: ShortBoxSlice<Subtag>) -> Self110 pub(crate) fn from_short_slice_unchecked(ext: u8, keys: ShortBoxSlice<Subtag>) -> Self { 111 assert!(ext.is_ascii_alphabetic()); 112 // Safety invariant upheld here: ext checked as ASCII above 113 Self { ext, keys } 114 } 115 116 #[cfg(feature = "alloc")] try_from_iter(ext: u8, iter: &mut SubtagIterator) -> Result<Self, ParseError>117 pub(crate) fn try_from_iter(ext: u8, iter: &mut SubtagIterator) -> Result<Self, ParseError> { 118 debug_assert!(matches!( 119 ExtensionType::try_from_byte(ext), 120 Ok(ExtensionType::Other(_)), 121 )); 122 123 let mut keys = ShortBoxSlice::new(); 124 while let Some(subtag) = iter.peek() { 125 if !Subtag::valid_key(subtag) { 126 break; 127 } 128 if let Ok(key) = Subtag::try_from_utf8(subtag) { 129 keys.push(key); 130 } 131 iter.next(); 132 } 133 134 if keys.is_empty() { 135 Err(ParseError::InvalidExtension) 136 } else { 137 Ok(Self::from_short_slice_unchecked(ext, keys)) 138 } 139 } 140 141 /// Gets the tag character for this extension as a &str. 142 /// 143 /// # Examples 144 /// 145 /// ``` 146 /// use icu::locale::Locale; 147 /// 148 /// let loc: Locale = "und-a-hello-world".parse().unwrap(); 149 /// let other_ext = &loc.extensions.other[0]; 150 /// assert_eq!(other_ext.get_ext_str(), "a"); 151 /// ``` get_ext_str(&self) -> &str152 pub fn get_ext_str(&self) -> &str { 153 debug_assert!(self.ext.is_ascii_alphabetic()); 154 // Safety: from safety invariant on self.ext (that it is ASCII) 155 unsafe { core::str::from_utf8_unchecked(core::slice::from_ref(&self.ext)) } 156 } 157 158 /// Gets the tag character for this extension as a char. 159 /// 160 /// # Examples 161 /// 162 /// ``` 163 /// use icu::locale::Locale; 164 /// 165 /// let loc: Locale = "und-a-hello-world".parse().unwrap(); 166 /// let other_ext = &loc.extensions.other[0]; 167 /// assert_eq!(other_ext.get_ext(), 'a'); 168 /// ``` get_ext(&self) -> char169 pub fn get_ext(&self) -> char { 170 self.ext as char 171 } 172 173 /// Gets the tag character for this extension as a byte. 174 /// 175 /// # Examples 176 /// 177 /// ``` 178 /// use icu::locale::Locale; 179 /// 180 /// let loc: Locale = "und-a-hello-world".parse().unwrap(); 181 /// let other_ext = &loc.extensions.other[0]; 182 /// assert_eq!(other_ext.get_ext_byte(), b'a'); 183 /// ``` get_ext_byte(&self) -> u8184 pub fn get_ext_byte(&self) -> u8 { 185 self.ext 186 } 187 for_each_subtag_str<E, F>(&self, f: &mut F, with_ext: bool) -> Result<(), E> where F: FnMut(&str) -> Result<(), E>,188 pub(crate) fn for_each_subtag_str<E, F>(&self, f: &mut F, with_ext: bool) -> Result<(), E> 189 where 190 F: FnMut(&str) -> Result<(), E>, 191 { 192 if self.keys.is_empty() { 193 return Ok(()); 194 } 195 196 if with_ext { 197 f(self.get_ext_str())?; 198 } 199 self.keys.iter().map(|t| t.as_str()).try_for_each(f) 200 } 201 } 202 203 #[cfg(feature = "alloc")] 204 impl FromStr for Other { 205 type Err = ParseError; 206 207 #[inline] from_str(s: &str) -> Result<Self, Self::Err>208 fn from_str(s: &str) -> Result<Self, Self::Err> { 209 Self::try_from_str(s) 210 } 211 } 212 213 writeable::impl_display_with_writeable!(Other); 214 215 impl writeable::Writeable for Other { write_to<W: core::fmt::Write + ?Sized>(&self, sink: &mut W) -> core::fmt::Result216 fn write_to<W: core::fmt::Write + ?Sized>(&self, sink: &mut W) -> core::fmt::Result { 217 if self.keys.is_empty() { 218 return Ok(()); 219 } 220 sink.write_str(self.get_ext_str())?; 221 for key in self.keys.iter() { 222 sink.write_char('-')?; 223 writeable::Writeable::write_to(key, sink)?; 224 } 225 226 Ok(()) 227 } 228 writeable_length_hint(&self) -> writeable::LengthHint229 fn writeable_length_hint(&self) -> writeable::LengthHint { 230 if self.keys.is_empty() { 231 return writeable::LengthHint::exact(0); 232 }; 233 let mut result = writeable::LengthHint::exact(1); 234 for key in self.keys.iter() { 235 result += writeable::Writeable::writeable_length_hint(key) + 1; 236 } 237 result 238 } 239 240 #[cfg(feature = "alloc")] write_to_string(&self) -> alloc::borrow::Cow<str>241 fn write_to_string(&self) -> alloc::borrow::Cow<str> { 242 if self.keys.is_empty() { 243 return alloc::borrow::Cow::Borrowed(""); 244 } 245 let mut string = 246 alloc::string::String::with_capacity(self.writeable_length_hint().capacity()); 247 let _ = self.write_to(&mut string); 248 alloc::borrow::Cow::Owned(string) 249 } 250 } 251 252 #[cfg(test)] 253 mod tests { 254 use super::*; 255 256 #[test] test_other_extension_fromstr()257 fn test_other_extension_fromstr() { 258 let oe: Other = "o-foo-bar".parse().expect("Failed to parse Other"); 259 assert_eq!(oe.to_string(), "o-foo-bar"); 260 261 let oe: Result<Other, _> = "o".parse(); 262 assert!(oe.is_err()); 263 } 264 } 265