diff options
author | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-17 12:06:37 +0000 |
---|---|---|
committer | Daniel Baumann <daniel.baumann@progress-linux.org> | 2024-04-17 12:06:37 +0000 |
commit | 246f239d9f40f633160f0c18f87a20922d4e77bb (patch) | |
tree | 5a88572663584b3d4d28e5a20e10abab1be40884 /vendor/regex | |
parent | Releasing progress-linux version 1.64.0+dfsg1-1~progress7.99u1. (diff) | |
download | rustc-246f239d9f40f633160f0c18f87a20922d4e77bb.tar.xz rustc-246f239d9f40f633160f0c18f87a20922d4e77bb.zip |
Merging debian version 1.65.0+dfsg1-2.
Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>
Diffstat (limited to '')
49 files changed, 2317 insertions, 1405 deletions
diff --git a/vendor/regex-syntax/.cargo-checksum.json b/vendor/regex-syntax/.cargo-checksum.json index 127f562f7..881437810 100644 --- a/vendor/regex-syntax/.cargo-checksum.json +++ b/vendor/regex-syntax/.cargo-checksum.json @@ -1 +1 @@ -{"files":{"Cargo.toml":"e6d037be81ec5e78c7cce40650b1082f07e43882ce6ad7c360b5d7f804003dbb","LICENSE-APACHE":"a60eea817514531668d7e00765731449fe14d059d3249e0bc93b36de45f759f2","LICENSE-MIT":"6485b8ed310d3f0340bf1ad1f47645069ce4069dcc6bb46c7d5c6faf41de1fdb","README.md":"67a3e673a9da6826fd4db5be6902841c821b52b98dc22c300f6e327872392b0a","benches/bench.rs":"d2b6ae5b939abd6093064f144b981b7739d7f474ec0698a1268052fc92406635","src/ast/mod.rs":"b0fe9af7ae15d9448246c204977634e5827bbae247bd59ab2e61411996fc68d6","src/ast/parse.rs":"ba7b2c46d18ada21fae986e2202d638515c8be3e275657a6b87b4d86d5054cd3","src/ast/print.rs":"d12f2cc75cd62f35623e1eb7a77ab8ac804b971752082700d2c4f550f834b249","src/ast/visitor.rs":"8ffcad13eb2c2a2f745f7bc8d823bd2f0bb728bd150f439455be5a245731f1d2","src/either.rs":"1758e3edd056884eccadd995708d1e374ba9aa65846bd0e13b1aae852607c560","src/error.rs":"cc99a11392b52f7665ff5ee8ea350f7386ed7c6c6bedd46e216b2f396785317f","src/hir/interval.rs":"2ffab258f204fe47bc5fe9ca84376fcd9ecb4929649f683a9412f2e382e908dc","src/hir/literal/mod.rs":"f3e9b291af6a143ab693fe0b63a7c070a2a41935c15b38ecf2cfa8ab0ab2e89c","src/hir/mod.rs":"51f1642b75e298b0847855d7f490eca2e8f445c9039dd3aa91491f533ec83f15","src/hir/print.rs":"651b5d9776532a78612a5f9081372a57bad693890639ac19e3128b4defa96662","src/hir/translate.rs":"de10ee2928001567f80c6ab602de0e280a8e3a1e615cf73def91d2dcc4f9f199","src/hir/visitor.rs":"e5bf7f8c09f6155e59c9d676fe25437f7e3700f9bf5d91101d7e246a64c11d5a","src/lib.rs":"0fc94332a971691862ca17531881302b10ef6fa4aba65c123f0b69ffb14b989a","src/parser.rs":"ce2392d748d909c2eb92f3bf7ad50065c59a615e6c46f81e87f48e9d87817061","src/unicode.rs":"3b486b36e2ffcae306cb6d7387a82069163c7269597ff2b50589a05462464c36","src/unicode_tables/LICENSE-UNICODE":"74db5baf44a41b1000312c673544b3374e4198af5605c7f9080a402cec42cfa3","src/unicode_tables/age.rs":"b0932a020d3386478dd2f4839c59e30c525e8591735052b9e791e1ce3a2e2b72","src/unicode_tables/case_folding_simple.rs":"6d1f3d095132639228faf4806d05308c70ce2baa68cce69dca01ea159c4eaa15","src/unicode_tables/general_category.rs":"d21877600d387b8a0c5fbb0942458d0330c69aad6565e28134b8a1a371d2f4f4","src/unicode_tables/grapheme_cluster_break.rs":"f03a8be4a00265b568ca2a41d99f66a5d0f5fb423cb4113494153423a1123cda","src/unicode_tables/mod.rs":"26c837099cd934c8062e24bc9a0aaecf15fe1de03f9c6da3f3e1e5ac3ca24bee","src/unicode_tables/perl_decimal.rs":"e39a5934b504eb3282ccb26bbf50ecd764e720120eb7cf6c43662a2321665ab5","src/unicode_tables/perl_space.rs":"014e5d92b66730557e408c2d5c9b2f46d3d288aa85400ab9193c218c7b98ad21","src/unicode_tables/perl_word.rs":"ddf126f39171776ef83151d7a0dbc41da8dd09186723211fb966c4b304247a5e","src/unicode_tables/property_bool.rs":"21f72bd9f3955e3443549ef6609418817ae6df3c81fb5be90a0ceee9d7d3002d","src/unicode_tables/property_names.rs":"504ea44604cd15a7e827a89066bb81a847dd5c57cef360d9f4a914cf22afcf36","src/unicode_tables/property_values.rs":"4d793ad1b664c1913db146897c8eb4fa29d181b821f096de90dc889b738edb88","src/unicode_tables/script.rs":"5a7d2a958b93056081b8b2eb87c3a5609579ad791ad5b0c42959362ce6ea5b31","src/unicode_tables/script_extension.rs":"1d5f1985f7dcae833e78c3858231666b535bf60e032cfacc09d014c22bda6690","src/unicode_tables/sentence_break.rs":"cd5f0eb7ab6b0ec1c1fb4d78496dfecd691d0d0b76495538b9f376645a753deb","src/unicode_tables/word_break.rs":"eabeacfde7558cfe7b1556b0221f09c65f049de0b08c7cd464c1669040610a6b","src/utf8.rs":"f145b2cb0324e6a39260db685fdf2d88675dead54c5b808fb1b7f73a4b530d66","test":"8a9bd1bd9fb389e08288f951319a9bbb0d4c5284a2ba63cbdab7f6afa2c2f76e"},"package":"49b3de9ec5dc0a3417da371aab17d729997c15010e7fd24ff707773a33bddb64"}
\ No newline at end of file +{"files":{"Cargo.toml":"91aed5795d8faeb9a9f43face557622d92417a17bb453b5f12ee87073040fb7d","LICENSE-APACHE":"a60eea817514531668d7e00765731449fe14d059d3249e0bc93b36de45f759f2","LICENSE-MIT":"6485b8ed310d3f0340bf1ad1f47645069ce4069dcc6bb46c7d5c6faf41de1fdb","README.md":"67a3e673a9da6826fd4db5be6902841c821b52b98dc22c300f6e327872392b0a","benches/bench.rs":"d2b6ae5b939abd6093064f144b981b7739d7f474ec0698a1268052fc92406635","src/ast/mod.rs":"91b277a9bb979f85a44a67e39f17f77bde033764eea1f1a93aad1b61f1250089","src/ast/parse.rs":"150b42e944f766fdca70d654dbe32f8a17498432729c78b9eb50b73ae7f91f86","src/ast/print.rs":"d12f2cc75cd62f35623e1eb7a77ab8ac804b971752082700d2c4f550f834b249","src/ast/visitor.rs":"1a7b473147e4f6b89623ef1744a9e87f665bcf160fe08a33ce8e35011811ba71","src/either.rs":"1758e3edd056884eccadd995708d1e374ba9aa65846bd0e13b1aae852607c560","src/error.rs":"b3c5903a8937d2aff229a3ec65d4571d01ec4d9874c9a242ed6562c32702bcbd","src/hir/interval.rs":"e767fed363bebe4bbda0d78b8f07e73f321eaf4f837e2d7bd14a1617387e9a89","src/hir/literal/mod.rs":"ffe9a0aff7827f97bffd29eb2f4ba96627b16953161dce6c50a2f760e76bbd98","src/hir/mod.rs":"7f83c828223a54d236d348e48d5cedf015c904812110b6c38e9d52039c2b1572","src/hir/print.rs":"651b5d9776532a78612a5f9081372a57bad693890639ac19e3128b4defa96662","src/hir/translate.rs":"c7cd9693f73760263fd49a968714d27e7985ebe840211b2d83bca6686b0602a8","src/hir/visitor.rs":"e5bf7f8c09f6155e59c9d676fe25437f7e3700f9bf5d91101d7e246a64c11d5a","src/lib.rs":"a004f65196dd5745b3112e4acc8c467b18495cecac64a58d6608b35de67371cb","src/parser.rs":"0dfb553a152e008b2755f115663e553ed99c4b8e6a4dcbcad1662737534de49d","src/unicode.rs":"2b575c75dcb8fd6becb06f2a8faa33d6f54779708bc6b103070b8acb2b3323bb","src/unicode_tables/LICENSE-UNICODE":"74db5baf44a41b1000312c673544b3374e4198af5605c7f9080a402cec42cfa3","src/unicode_tables/age.rs":"9b36dd7d359d2fa21e6ea9734a37415ba0ba9469e27b8536308f5b79139d4191","src/unicode_tables/case_folding_simple.rs":"52b74161fa1e8f2db66737156e081cece82c1f503dee604f901b2df095beb90d","src/unicode_tables/general_category.rs":"c10beb78cdab6ec14846573bfb7965b5b3a4c114d20352c21a666d8de740a049","src/unicode_tables/grapheme_cluster_break.rs":"10ba739e06f880570eaf90bddd78ec468d939c0be7cd6e39f52cfa68371a9885","src/unicode_tables/mod.rs":"26c837099cd934c8062e24bc9a0aaecf15fe1de03f9c6da3f3e1e5ac3ca24bee","src/unicode_tables/perl_decimal.rs":"2084b101c615ff368a47af72df760003a76e869dfc191e0224cd024bb88165ed","src/unicode_tables/perl_space.rs":"75b9f3c9e302fd0994ee6e9ee21ee0ab36efc513cfd083647aed9854b977a33d","src/unicode_tables/perl_word.rs":"c588d6b29b98c1160452b54e9275d43583bc3454e29aee1c07e6a18389011a9c","src/unicode_tables/property_bool.rs":"319740ac6074b2d4e6e22bf4dde7db2feb569f9b71467c893fb2553d149b9f1d","src/unicode_tables/property_names.rs":"f33b186d7d8ee5342d74ef214f7b2cfbb24345233fa49a2abf2578cae61fbdd3","src/unicode_tables/property_values.rs":"716eb87716a9a4b2a9ef2c2242d6831692564875c8e218ffa758266ca33c88b6","src/unicode_tables/script.rs":"f6020589e33bd3a058468a22ce51391f6e512f9eb88c8ac60635fae7cd641ee2","src/unicode_tables/script_extension.rs":"2ba03d13813161a064b11e9f87b87685fbf29699ef553acc6112606b7ca98169","src/unicode_tables/sentence_break.rs":"bf7635623e4dbe0195789ed8b21f83ce3394ed2c445a3005f929f4f75b2a83c3","src/unicode_tables/word_break.rs":"8f7e261a67d1adb32ea627e43cb61d6566c33a4e229113d911e86941e6997a41","src/utf8.rs":"de854b3bfb3f7dbefc422f6a25935aaeef55ead2c35386c712a1fe9bf81a7b6f","test":"8a9bd1bd9fb389e08288f951319a9bbb0d4c5284a2ba63cbdab7f6afa2c2f76e"},"package":"a3f87b73ce11b1619a3c6332f45341e0047173771e8b8b73f87bfeefb7b56244"}
\ No newline at end of file diff --git a/vendor/regex-syntax/Cargo.toml b/vendor/regex-syntax/Cargo.toml index 16ac50e74..10e6bbf53 100644 --- a/vendor/regex-syntax/Cargo.toml +++ b/vendor/regex-syntax/Cargo.toml @@ -12,7 +12,7 @@ [package] edition = "2018" name = "regex-syntax" -version = "0.6.26" +version = "0.6.27" authors = ["The Rust Project Developers"] description = "A regular expression parser." homepage = "https://github.com/rust-lang/regex" diff --git a/vendor/regex-syntax/src/ast/mod.rs b/vendor/regex-syntax/src/ast/mod.rs index 9b9127b1f..387ea3a69 100644 --- a/vendor/regex-syntax/src/ast/mod.rs +++ b/vendor/regex-syntax/src/ast/mod.rs @@ -15,7 +15,7 @@ mod visitor; /// An error that occurred while parsing a regular expression into an abstract /// syntax tree. /// -/// Note that note all ASTs represents a valid regular expression. For example, +/// Note that not all ASTs represents a valid regular expression. For example, /// an AST is constructed without error for `\p{Quux}`, but `Quux` is not a /// valid Unicode property name. That particular error is reported when /// translating an AST to the high-level intermediate representation (`HIR`). @@ -385,7 +385,7 @@ impl PartialOrd for Position { impl Span { /// Create a new span with the given positions. pub fn new(start: Position, end: Position) -> Span { - Span { start: start, end: end } + Span { start, end } } /// Create a new span using the given position as the start and end. @@ -427,7 +427,7 @@ impl Position { /// /// `column` is the approximate column number, starting at `1`. pub fn new(offset: usize, line: usize, column: usize) -> Position { - Position { offset: offset, line: line, column: column } + Position { offset, line, column } } } diff --git a/vendor/regex-syntax/src/ast/parse.rs b/vendor/regex-syntax/src/ast/parse.rs index 9824661c9..6e9c9aca0 100644 --- a/vendor/regex-syntax/src/ast/parse.rs +++ b/vendor/regex-syntax/src/ast/parse.rs @@ -202,7 +202,7 @@ impl ParserBuilder { /// Enable verbose mode in the regular expression. /// - /// When enabled, verbose mode permits insigificant whitespace in many + /// When enabled, verbose mode permits insignificant whitespace in many /// places in the regular expression, as well as comments. Comments are /// started using `#` and continue until the end of the line. /// @@ -366,7 +366,7 @@ impl Parser { impl<'s, P: Borrow<Parser>> ParserI<'s, P> { /// Build an internal parser from a parser configuration and a pattern. fn new(parser: P, pattern: &'s str) -> ParserI<'s, P> { - ParserI { parser: parser, pattern: pattern } + ParserI { parser, pattern } } /// Return a reference to the parser state. @@ -381,11 +381,7 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { /// Create a new error with the given span and error type. fn error(&self, span: Span, kind: ast::ErrorKind) -> ast::Error { - ast::Error { - kind: kind, - pattern: self.pattern().to_string(), - span: span, - } + ast::Error { kind, pattern: self.pattern().to_string(), span } } /// Return the current offset of the parser. @@ -481,11 +477,7 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { column = column.checked_add(1).unwrap(); } offset += self.char().len_utf8(); - self.parser().pos.set(Position { - offset: offset, - line: line, - column: column, - }); + self.parser().pos.set(Position { offset, line, column }); self.pattern()[self.offset()..].chars().next().is_some() } @@ -703,8 +695,8 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { .unwrap_or(old_ignore_whitespace); self.parser().stack_group.borrow_mut().push( GroupState::Group { - concat: concat, - group: group, + concat, + group, ignore_whitespace: old_ignore_whitespace, }, ); @@ -899,12 +891,8 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { #[inline(never)] fn unclosed_class_error(&self) -> ast::Error { for state in self.parser().stack_class.borrow().iter().rev() { - match *state { - ClassState::Open { ref set, .. } => { - return self - .error(set.span, ast::ErrorKind::ClassUnclosed); - } - _ => {} + if let ClassState::Open { ref set, .. } = *state { + return self.error(set.span, ast::ErrorKind::ClassUnclosed); } } // We are guaranteed to have a non-empty stack with at least @@ -950,8 +938,8 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { }; let span = Span::new(lhs.span().start, rhs.span().end); ast::ClassSet::BinaryOp(ast::ClassSetBinaryOp { - span: span, - kind: kind, + span, + kind, lhs: Box::new(lhs), rhs: Box::new(rhs), }) @@ -1010,7 +998,7 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { let ast = self.pop_group_end(concat)?; NestLimiter::new(self).check(&ast)?; Ok(ast::WithComments { - ast: ast, + ast, comments: mem::replace( &mut *self.parser().comments.borrow_mut(), vec![], @@ -1066,9 +1054,9 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { span: ast.span().with_end(self.pos()), op: ast::RepetitionOp { span: Span::new(op_start, self.pos()), - kind: kind, + kind, }, - greedy: greedy, + greedy, ast: Box::new(ast), })); Ok(concat) @@ -1170,7 +1158,7 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { span: op_span, kind: ast::RepetitionKind::Range(range), }, - greedy: greedy, + greedy, ast: Box::new(ast), })); Ok(concat) @@ -1235,7 +1223,7 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { } Ok(Either::Left(ast::SetFlags { span: Span { end: self.pos(), ..open_span }, - flags: flags, + flags, })) } else { assert_eq!(char_end, ':'); @@ -1428,7 +1416,7 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { let ast = Primitive::Literal(ast::Literal { span: self.span_char(), kind: ast::LiteralKind::Verbatim, - c: c, + c, }); self.bump(); Ok(ast) @@ -1494,16 +1482,16 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { let span = Span::new(start, self.pos()); if is_meta_character(c) { return Ok(Primitive::Literal(ast::Literal { - span: span, + span, kind: ast::LiteralKind::Punctuation, - c: c, + c, })); } let special = |kind, c| { Ok(Primitive::Literal(ast::Literal { - span: span, + span, kind: ast::LiteralKind::Special(kind), - c: c, + c, })) }; match c { @@ -1517,19 +1505,19 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { special(ast::SpecialLiteralKind::Space, ' ') } 'A' => Ok(Primitive::Assertion(ast::Assertion { - span: span, + span, kind: ast::AssertionKind::StartText, })), 'z' => Ok(Primitive::Assertion(ast::Assertion { - span: span, + span, kind: ast::AssertionKind::EndText, })), 'b' => Ok(Primitive::Assertion(ast::Assertion { - span: span, + span, kind: ast::AssertionKind::WordBoundary, })), 'B' => Ok(Primitive::Assertion(ast::Assertion { - span: span, + span, kind: ast::AssertionKind::NotWordBoundary, })), _ => Err(self.error(span, ast::ErrorKind::EscapeUnrecognized)), @@ -1569,7 +1557,7 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { ast::Literal { span: Span::new(start, end), kind: ast::LiteralKind::Octal, - c: c, + c, } } @@ -1645,7 +1633,7 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { Some(c) => Ok(ast::Literal { span: Span::new(start, end), kind: ast::LiteralKind::HexFixed(kind), - c: c, + c, }), } } @@ -1700,7 +1688,7 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { Some(c) => Ok(ast::Literal { span: Span::new(start, self.pos()), kind: ast::LiteralKind::HexBrace(kind), - c: c, + c, }), } } @@ -1927,7 +1915,7 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { })); if !self.bump_and_bump_space() { return Err(self.error( - Span::new(start, self.pos()), + Span::new(start, start), ast::ErrorKind::ClassUnclosed, )); } @@ -1949,7 +1937,7 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { } let set = ast::ClassBracketed { span: Span::new(start, self.pos()), - negated: negated, + negated, kind: ast::ClassSet::union(ast::ClassSetUnion { span: Span::new(union.span.start, union.span.start), items: vec![], @@ -2026,8 +2014,8 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { }; Some(ast::ClassAscii { span: Span::new(start, self.pos()), - kind: kind, - negated: negated, + kind, + negated, }) } @@ -2108,8 +2096,8 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { }; Ok(ast::ClassUnicode { span: Span::new(start, self.pos()), - negated: negated, - kind: kind, + negated, + kind, }) } @@ -2130,7 +2118,7 @@ impl<'s, P: Borrow<Parser>> ParserI<'s, P> { 'W' => (true, ast::ClassPerlKind::Word), c => panic!("expected valid Perl class but got '{}'", c), }; - ast::ClassPerl { span: span, kind: kind, negated: negated } + ast::ClassPerl { span, kind, negated } } } @@ -2146,7 +2134,7 @@ struct NestLimiter<'p, 's, P> { impl<'p, 's, P: Borrow<Parser>> NestLimiter<'p, 's, P> { fn new(p: &'p ParserI<'s, P>) -> NestLimiter<'p, 's, P> { - NestLimiter { p: p, depth: 0 } + NestLimiter { p, depth: 0 } } #[inline(never)] @@ -2429,18 +2417,18 @@ mod tests { /// Create a punctuation literal starting at the given position. fn punct_lit(c: char, span: Span) -> Ast { Ast::Literal(ast::Literal { - span: span, + span, kind: ast::LiteralKind::Punctuation, - c: c, + c, }) } /// Create a verbatim literal with the given span. fn lit_with(c: char, span: Span) -> Ast { Ast::Literal(ast::Literal { - span: span, + span, kind: ast::LiteralKind::Verbatim, - c: c, + c, }) } @@ -2451,12 +2439,12 @@ mod tests { /// Create a concatenation with the given span. fn concat_with(span: Span, asts: Vec<Ast>) -> Ast { - Ast::Concat(ast::Concat { span: span, asts: asts }) + Ast::Concat(ast::Concat { span, asts }) } /// Create an alternation with the given span. fn alt(range: Range<usize>, asts: Vec<Ast>) -> Ast { - Ast::Alternation(ast::Alternation { span: span(range), asts: asts }) + Ast::Alternation(ast::Alternation { span: span(range), asts }) } /// Create a capturing group with the given span. @@ -2498,7 +2486,7 @@ mod tests { span: span_range(pat, range.clone()), flags: ast::Flags { span: span_range(pat, (range.start + 2)..(range.end - 1)), - items: items, + items, }, }) } @@ -4208,7 +4196,7 @@ bar Ok(Primitive::Literal(ast::Literal { span: span(0..2), kind: ast::LiteralKind::Special(kind.clone()), - c: c, + c, })) ); } @@ -4402,7 +4390,7 @@ bar kind: ast::LiteralKind::HexFixed( ast::HexLiteralKind::UnicodeShort ), - c: c, + c, })) ); } @@ -4466,7 +4454,7 @@ bar kind: ast::LiteralKind::HexFixed( ast::HexLiteralKind::UnicodeLong ), - c: c, + c, })) ); } @@ -4667,10 +4655,7 @@ bar #[test] fn parse_set_class() { fn union(span: Span, items: Vec<ast::ClassSetItem>) -> ast::ClassSet { - ast::ClassSet::union(ast::ClassSetUnion { - span: span, - items: items, - }) + ast::ClassSet::union(ast::ClassSetUnion { span, items }) } fn intersection( @@ -4679,7 +4664,7 @@ bar rhs: ast::ClassSet, ) -> ast::ClassSet { ast::ClassSet::BinaryOp(ast::ClassSetBinaryOp { - span: span, + span, kind: ast::ClassSetBinaryOpKind::Intersection, lhs: Box::new(lhs), rhs: Box::new(rhs), @@ -4692,7 +4677,7 @@ bar rhs: ast::ClassSet, ) -> ast::ClassSet { ast::ClassSet::BinaryOp(ast::ClassSetBinaryOp { - span: span, + span, kind: ast::ClassSetBinaryOpKind::Difference, lhs: Box::new(lhs), rhs: Box::new(rhs), @@ -4705,7 +4690,7 @@ bar rhs: ast::ClassSet, ) -> ast::ClassSet { ast::ClassSet::BinaryOp(ast::ClassSetBinaryOp { - span: span, + span, kind: ast::ClassSetBinaryOpKind::SymmetricDifference, lhs: Box::new(lhs), rhs: Box::new(rhs), @@ -4734,9 +4719,9 @@ bar fn lit(span: Span, c: char) -> ast::ClassSetItem { ast::ClassSetItem::Literal(ast::Literal { - span: span, + span, kind: ast::LiteralKind::Verbatim, - c: c, + c, }) } @@ -4756,7 +4741,7 @@ bar ..span.end }; ast::ClassSetItem::Range(ast::ClassSetRange { - span: span, + span, start: ast::Literal { span: Span { end: pos1, ..span }, kind: ast::LiteralKind::Verbatim, @@ -4771,19 +4756,11 @@ bar } fn alnum(span: Span, negated: bool) -> ast::ClassAscii { - ast::ClassAscii { - span: span, - kind: ast::ClassAsciiKind::Alnum, - negated: negated, - } + ast::ClassAscii { span, kind: ast::ClassAsciiKind::Alnum, negated } } fn lower(span: Span, negated: bool) -> ast::ClassAscii { - ast::ClassAscii { - span: span, - kind: ast::ClassAsciiKind::Lower, - negated: negated, - } + ast::ClassAscii { span, kind: ast::ClassAsciiKind::Lower, negated } } assert_eq!( @@ -5515,14 +5492,23 @@ bar assert_eq!( parser("[-").parse_set_class_open().unwrap_err(), TestError { - span: span(0..2), + span: span(0..0), kind: ast::ErrorKind::ClassUnclosed, } ); assert_eq!( parser("[--").parse_set_class_open().unwrap_err(), TestError { - span: span(0..3), + span: span(0..0), + kind: ast::ErrorKind::ClassUnclosed, + } + ); + + // See: https://github.com/rust-lang/regex/issues/792 + assert_eq!( + parser("(?x)[-#]").parse_with_comments().unwrap_err(), + TestError { + span: span(4..4), kind: ast::ErrorKind::ClassUnclosed, } ); diff --git a/vendor/regex-syntax/src/ast/visitor.rs b/vendor/regex-syntax/src/ast/visitor.rs index a0d1e7dd5..78ee487cf 100644 --- a/vendor/regex-syntax/src/ast/visitor.rs +++ b/vendor/regex-syntax/src/ast/visitor.rs @@ -388,7 +388,7 @@ impl<'a> HeapVisitor<'a> { Some(ClassFrame::Union { head: item, tail: &[] }) } ast::ClassSet::BinaryOp(ref op) => { - Some(ClassFrame::Binary { op: op }) + Some(ClassFrame::Binary { op }) } } } @@ -402,11 +402,9 @@ impl<'a> HeapVisitor<'a> { }) } } - ClassInduct::BinaryOp(op) => Some(ClassFrame::BinaryLHS { - op: op, - lhs: &op.lhs, - rhs: &op.rhs, - }), + ClassInduct::BinaryOp(op) => { + Some(ClassFrame::BinaryLHS { op, lhs: &op.lhs, rhs: &op.rhs }) + } _ => None, } } @@ -427,7 +425,7 @@ impl<'a> HeapVisitor<'a> { } ClassFrame::Binary { .. } => None, ClassFrame::BinaryLHS { op, rhs, .. } => { - Some(ClassFrame::BinaryRHS { op: op, rhs: rhs }) + Some(ClassFrame::BinaryRHS { op, rhs }) } ClassFrame::BinaryRHS { .. } => None, } diff --git a/vendor/regex-syntax/src/error.rs b/vendor/regex-syntax/src/error.rs index 71cfa426a..1230d2fc5 100644 --- a/vendor/regex-syntax/src/error.rs +++ b/vendor/regex-syntax/src/error.rs @@ -182,7 +182,7 @@ impl<'p> Spans<'p> { if line_count <= 1 { 0 } else { line_count.to_string().len() }; let mut spans = Spans { pattern: &fmter.pattern, - line_number_width: line_number_width, + line_number_width, by_line: vec![vec![]; line_count], multi_line: vec![], }; @@ -288,7 +288,7 @@ fn repeat_char(c: char, count: usize) -> String { mod tests { use crate::ast::parse::Parser; - fn assert_panic_message(pattern: &str, expected_msg: &str) -> () { + fn assert_panic_message(pattern: &str, expected_msg: &str) { let result = Parser::new().parse(pattern); match result { Ok(_) => { diff --git a/vendor/regex-syntax/src/hir/interval.rs b/vendor/regex-syntax/src/hir/interval.rs index cfaa2cb45..56698c53a 100644 --- a/vendor/regex-syntax/src/hir/interval.rs +++ b/vendor/regex-syntax/src/hir/interval.rs @@ -114,8 +114,8 @@ impl<I: Interval> IntervalSet<I> { // we're done. let drain_end = self.ranges.len(); - let mut ita = (0..drain_end).into_iter(); - let mut itb = (0..other.ranges.len()).into_iter(); + let mut ita = 0..drain_end; + let mut itb = 0..other.ranges.len(); let mut a = ita.next().unwrap(); let mut b = itb.next().unwrap(); loop { diff --git a/vendor/regex-syntax/src/hir/literal/mod.rs b/vendor/regex-syntax/src/hir/literal/mod.rs index 1e66d2cc3..fbc5d3c97 100644 --- a/vendor/regex-syntax/src/hir/literal/mod.rs +++ b/vendor/regex-syntax/src/hir/literal/mod.rs @@ -225,7 +225,7 @@ impl Literals { if self.lits.is_empty() { return self.to_empty(); } - let mut old: Vec<Literal> = self.lits.iter().cloned().collect(); + let mut old = self.lits.to_vec(); let mut new = self.to_empty(); 'OUTER: while let Some(mut candidate) = old.pop() { if candidate.is_empty() { @@ -256,15 +256,13 @@ impl Literals { old.push(lit3); lit2.clear(); } - } else { - if let Some(i) = position(&lit2, &candidate) { - lit2.cut(); - let mut new_candidate = candidate.clone(); - new_candidate.truncate(i); - new_candidate.cut(); - old.push(new_candidate); - candidate.clear(); - } + } else if let Some(i) = position(&lit2, &candidate) { + lit2.cut(); + let mut new_candidate = candidate.clone(); + new_candidate.truncate(i); + new_candidate.cut(); + old.push(new_candidate); + candidate.clear(); } // Oops, the candidate is already represented in the set. if candidate.is_empty() { @@ -793,7 +791,7 @@ fn repeat_range_literals<F: FnMut(&Hir, &mut Literals)>( f( &Hir::repetition(hir::Repetition { kind: hir::RepetitionKind::ZeroOrMore, - greedy: greedy, + greedy, hir: Box::new(e.clone()), }), lits, @@ -932,12 +930,10 @@ fn escape_unicode(bytes: &[u8]) -> String { if c.is_whitespace() { let escaped = if c as u32 <= 0x7F { escape_byte(c as u8) + } else if c as u32 <= 0xFFFF { + format!(r"\u{{{:04x}}}", c as u32) } else { - if c as u32 <= 0xFFFF { - format!(r"\u{{{:04x}}}", c as u32) - } else { - format!(r"\U{{{:08x}}}", c as u32) - } + format!(r"\U{{{:08x}}}", c as u32) }; space_escaped.push_str(&escaped); } else { diff --git a/vendor/regex-syntax/src/hir/mod.rs b/vendor/regex-syntax/src/hir/mod.rs index f5cf992e5..1096e9f05 100644 --- a/vendor/regex-syntax/src/hir/mod.rs +++ b/vendor/regex-syntax/src/hir/mod.rs @@ -243,7 +243,7 @@ impl Hir { info.set_match_empty(true); info.set_literal(false); info.set_alternation_literal(false); - Hir { kind: HirKind::Empty, info: info } + Hir { kind: HirKind::Empty, info } } /// Creates a literal HIR expression. @@ -268,7 +268,7 @@ impl Hir { info.set_match_empty(false); info.set_literal(true); info.set_alternation_literal(true); - Hir { kind: HirKind::Literal(lit), info: info } + Hir { kind: HirKind::Literal(lit), info } } /// Creates a class HIR expression. @@ -285,7 +285,7 @@ impl Hir { info.set_match_empty(false); info.set_literal(false); info.set_alternation_literal(false); - Hir { kind: HirKind::Class(class), info: info } + Hir { kind: HirKind::Class(class), info } } /// Creates an anchor assertion HIR expression. @@ -318,7 +318,7 @@ impl Hir { if let Anchor::EndLine = anchor { info.set_line_anchored_end(true); } - Hir { kind: HirKind::Anchor(anchor), info: info } + Hir { kind: HirKind::Anchor(anchor), info } } /// Creates a word boundary assertion HIR expression. @@ -345,7 +345,7 @@ impl Hir { if let WordBoundary::AsciiNegate = word_boundary { info.set_always_utf8(false); } - Hir { kind: HirKind::WordBoundary(word_boundary), info: info } + Hir { kind: HirKind::WordBoundary(word_boundary), info } } /// Creates a repetition HIR expression. @@ -372,7 +372,7 @@ impl Hir { info.set_match_empty(rep.is_match_empty() || rep.hir.is_match_empty()); info.set_literal(false); info.set_alternation_literal(false); - Hir { kind: HirKind::Repetition(rep), info: info } + Hir { kind: HirKind::Repetition(rep), info } } /// Creates a group HIR expression. @@ -389,7 +389,7 @@ impl Hir { info.set_match_empty(group.hir.is_match_empty()); info.set_literal(false); info.set_alternation_literal(false); - Hir { kind: HirKind::Group(group), info: info } + Hir { kind: HirKind::Group(group), info } } /// Returns the concatenation of the given expressions. @@ -480,7 +480,7 @@ impl Hir { }) .any(|e| e.is_line_anchored_end()), ); - Hir { kind: HirKind::Concat(exprs), info: info } + Hir { kind: HirKind::Concat(exprs), info } } } } @@ -542,7 +542,7 @@ impl Hir { let x = info.is_alternation_literal() && e.is_literal(); info.set_alternation_literal(x); } - Hir { kind: HirKind::Alternation(exprs), info: info } + Hir { kind: HirKind::Alternation(exprs), info } } } } diff --git a/vendor/regex-syntax/src/hir/translate.rs b/vendor/regex-syntax/src/hir/translate.rs index 56afbbed8..890e1608b 100644 --- a/vendor/regex-syntax/src/hir/translate.rs +++ b/vendor/regex-syntax/src/hir/translate.rs @@ -589,7 +589,7 @@ struct TranslatorI<'t, 'p> { impl<'t, 'p> TranslatorI<'t, 'p> { /// Build a new internal translator. fn new(trans: &'t Translator, pattern: &'p str) -> TranslatorI<'t, 'p> { - TranslatorI { trans: trans, pattern: pattern } + TranslatorI { trans, pattern } } /// Return a reference to the underlying translator. @@ -609,7 +609,7 @@ impl<'t, 'p> TranslatorI<'t, 'p> { /// Create a new error with the given span and error type. fn error(&self, span: Span, kind: ErrorKind) -> Error { - Error { kind: kind, pattern: self.pattern.to_string(), span: span } + Error { kind, pattern: self.pattern.to_string(), span } } /// Return a copy of the active flags. @@ -779,7 +779,7 @@ impl<'t, 'p> TranslatorI<'t, 'p> { } ast::GroupKind::NonCapturing(_) => hir::GroupKind::NonCapturing, }; - Hir::group(hir::Group { kind: kind, hir: Box::new(expr) }) + Hir::group(hir::Group { kind, hir: Box::new(expr) }) } fn hir_repetition(&self, rep: &ast::Repetition, expr: Hir) -> Hir { @@ -802,11 +802,7 @@ impl<'t, 'p> TranslatorI<'t, 'p> { }; let greedy = if self.flags().swap_greed() { !rep.greedy } else { rep.greedy }; - Hir::repetition(hir::Repetition { - kind: kind, - greedy: greedy, - hir: Box::new(expr), - }) + Hir::repetition(hir::Repetition { kind, greedy, hir: Box::new(expr) }) } fn hir_unicode_class( @@ -1238,7 +1234,7 @@ mod tests { fn hir_quest(greedy: bool, expr: Hir) -> Hir { Hir::repetition(hir::Repetition { kind: hir::RepetitionKind::ZeroOrOne, - greedy: greedy, + greedy, hir: Box::new(expr), }) } @@ -1246,7 +1242,7 @@ mod tests { fn hir_star(greedy: bool, expr: Hir) -> Hir { Hir::repetition(hir::Repetition { kind: hir::RepetitionKind::ZeroOrMore, - greedy: greedy, + greedy, hir: Box::new(expr), }) } @@ -1254,7 +1250,7 @@ mod tests { fn hir_plus(greedy: bool, expr: Hir) -> Hir { Hir::repetition(hir::Repetition { kind: hir::RepetitionKind::OneOrMore, - greedy: greedy, + greedy, hir: Box::new(expr), }) } @@ -1262,7 +1258,7 @@ mod tests { fn hir_range(greedy: bool, range: hir::RepetitionRange, expr: Hir) -> Hir { Hir::repetition(hir::Repetition { kind: hir::RepetitionKind::Range(range), - greedy: greedy, + greedy, hir: Box::new(expr), }) } diff --git a/vendor/regex-syntax/src/lib.rs b/vendor/regex-syntax/src/lib.rs index 9e9af756a..1dfb38af3 100644 --- a/vendor/regex-syntax/src/lib.rs +++ b/vendor/regex-syntax/src/lib.rs @@ -195,7 +195,7 @@ pub fn escape_into(text: &str, buf: &mut String) { } } -/// Returns true if the give character has significance in a regex. +/// Returns true if the given character has significance in a regex. /// /// These are the only characters that are allowed to be escaped, with one /// exception: an ASCII space character may be escaped when extended mode (with diff --git a/vendor/regex-syntax/src/parser.rs b/vendor/regex-syntax/src/parser.rs index a5ee524a8..ded95b280 100644 --- a/vendor/regex-syntax/src/parser.rs +++ b/vendor/regex-syntax/src/parser.rs @@ -96,7 +96,7 @@ impl ParserBuilder { /// Enable verbose mode in the regular expression. /// - /// When enabled, verbose mode permits insigificant whitespace in many + /// When enabled, verbose mode permits insignificant whitespace in many /// places in the regular expression, as well as comments. Comments are /// started using `#` and continue until the end of the line. /// diff --git a/vendor/regex-syntax/src/unicode.rs b/vendor/regex-syntax/src/unicode.rs index b894c7db2..70d5954b7 100644 --- a/vendor/regex-syntax/src/unicode.rs +++ b/vendor/regex-syntax/src/unicode.rs @@ -99,7 +99,7 @@ pub fn simple_fold( Ok(CASE_FOLDING_SIMPLE .binary_search_by_key(&c, |&(c1, _)| c1) - .map(|i| CASE_FOLDING_SIMPLE[i].1.iter().map(|&c| c)) + .map(|i| CASE_FOLDING_SIMPLE[i].1.iter().copied()) .map_err(|i| { if i >= CASE_FOLDING_SIMPLE.len() { None @@ -580,7 +580,7 @@ fn ages(canonical_age: &str) -> Result<impl Iterator<Item = Range>> { fn imp(canonical_age: &str) -> Result<impl Iterator<Item = Range>> { use crate::unicode_tables::age; - const AGES: &'static [(&'static str, Range)] = &[ + const AGES: &[(&str, Range)] = &[ ("V1_1", age::V1_1), ("V2_0", age::V2_0), ("V2_1", age::V2_1), @@ -604,13 +604,14 @@ fn ages(canonical_age: &str) -> Result<impl Iterator<Item = Range>> { ("V12_0", age::V12_0), ("V12_1", age::V12_1), ("V13_0", age::V13_0), + ("V14_0", age::V14_0), ]; assert_eq!(AGES.len(), age::BY_NAME.len(), "ages are out of sync"); let pos = AGES.iter().position(|&(age, _)| canonical_age == age); match pos { None => Err(Error::PropertyValueNotFound), - Some(i) => Ok(AGES[..i + 1].iter().map(|&(_, classes)| classes)), + Some(i) => Ok(AGES[..=i].iter().map(|&(_, classes)| classes)), } } diff --git a/vendor/regex-syntax/src/unicode_tables/age.rs b/vendor/regex-syntax/src/unicode_tables/age.rs index 7772919eb..ffdfef316 100644 --- a/vendor/regex-syntax/src/unicode_tables/age.rs +++ b/vendor/regex-syntax/src/unicode_tables/age.rs @@ -1,10 +1,10 @@ // DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY: // -// ucd-generate age ucd-13.0.0 --chars +// ucd-generate age /tmp/ucd --chars // -// Unicode version: 13.0.0. +// Unicode version: 14.0.0. // -// ucd-generate 0.2.8 is available on crates.io. +// ucd-generate 0.2.11 is available on crates.io. pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("V10_0", V10_0), @@ -12,6 +12,7 @@ pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("V12_0", V12_0), ("V12_1", V12_1), ("V13_0", V13_0), + ("V14_0", V14_0), ("V1_1", V1_1), ("V2_0", V2_0), ("V2_1", V2_1), @@ -203,69 +204,150 @@ pub const V12_0: &'static [(char, char)] = &[ pub const V12_1: &'static [(char, char)] = &[('ã‹¿', 'ã‹¿')]; pub const V13_0: &'static [(char, char)] = &[ - ('\u{8be}', '\u{8c7}'), + ('ࢾ', 'ࣇ'), ('\u{b55}', '\u{b55}'), - ('\u{d04}', '\u{d04}'), + ('à´„', 'à´„'), ('\u{d81}', '\u{d81}'), ('\u{1abf}', '\u{1ac0}'), - ('\u{2b97}', '\u{2b97}'), - ('\u{2e50}', '\u{2e52}'), - ('\u{31bb}', '\u{31bf}'), - ('\u{4db6}', '\u{4dbf}'), - ('\u{9ff0}', '\u{9ffc}'), - ('\u{a7c7}', '\u{a7ca}'), - ('\u{a7f5}', '\u{a7f6}'), + ('â®—', 'â®—'), + ('â¹', 'â¹’'), + ('ㆻ', 'ㆿ'), + ('䶶', '䶿'), + ('é¿°', '鿼'), + ('Ꟈ', 'ꟊ'), + ('Ꟶ', 'ꟶ'), ('\u{a82c}', '\u{a82c}'), - ('\u{ab68}', '\u{ab6b}'), - ('\u{1019c}', '\u{1019c}'), - ('\u{10e80}', '\u{10ea9}'), - ('\u{10eab}', '\u{10ead}'), - ('\u{10eb0}', '\u{10eb1}'), - ('\u{10fb0}', '\u{10fcb}'), - ('\u{11147}', '\u{11147}'), - ('\u{111ce}', '\u{111cf}'), - ('\u{1145a}', '\u{1145a}'), - ('\u{11460}', '\u{11461}'), - ('\u{11900}', '\u{11906}'), - ('\u{11909}', '\u{11909}'), - ('\u{1190c}', '\u{11913}'), - ('\u{11915}', '\u{11916}'), - ('\u{11918}', '\u{11935}'), - ('\u{11937}', '\u{11938}'), - ('\u{1193b}', '\u{11946}'), - ('\u{11950}', '\u{11959}'), - ('\u{11fb0}', '\u{11fb0}'), + ('ê¨', 'ê«'), + ('ð†œ', 'ð†œ'), + ('ðº€', 'ðº©'), + ('\u{10eab}', 'ðº'), + ('ðº°', 'ðº±'), + ('ð¾°', 'ð¿‹'), + ('ð‘…‡', 'ð‘…‡'), + ('𑇎', '\u{111cf}'), + ('ð‘‘š', 'ð‘‘š'), + ('ð‘‘ ', 'ð‘‘¡'), + ('𑤀', '𑤆'), + ('𑤉', '𑤉'), + ('𑤌', '𑤓'), + ('𑤕', '𑤖'), + ('𑤘', '𑤵'), + ('𑤷', '𑤸'), + ('\u{1193b}', '𑥆'), + ('ð‘¥', 'ð‘¥™'), + ('𑾰', '𑾰'), ('\u{16fe4}', '\u{16fe4}'), - ('\u{16ff0}', '\u{16ff1}'), - ('\u{18af3}', '\u{18cd5}'), - ('\u{18d00}', '\u{18d08}'), - ('\u{1f10d}', '\u{1f10f}'), - ('\u{1f16d}', '\u{1f16f}'), - ('\u{1f1ad}', '\u{1f1ad}'), - ('\u{1f6d6}', '\u{1f6d7}'), - ('\u{1f6fb}', '\u{1f6fc}'), - ('\u{1f8b0}', '\u{1f8b1}'), - ('\u{1f90c}', '\u{1f90c}'), - ('\u{1f972}', '\u{1f972}'), - ('\u{1f977}', '\u{1f978}'), - ('\u{1f9a3}', '\u{1f9a4}'), - ('\u{1f9ab}', '\u{1f9ad}'), - ('\u{1f9cb}', '\u{1f9cb}'), - ('\u{1fa74}', '\u{1fa74}'), - ('\u{1fa83}', '\u{1fa86}'), - ('\u{1fa96}', '\u{1faa8}'), - ('\u{1fab0}', '\u{1fab6}'), - ('\u{1fac0}', '\u{1fac2}'), - ('\u{1fad0}', '\u{1fad6}'), - ('\u{1fb00}', '\u{1fb92}'), - ('\u{1fb94}', '\u{1fbca}'), - ('\u{1fbf0}', '\u{1fbf9}'), - ('\u{2a6d7}', '\u{2a6dd}'), - ('\u{30000}', '\u{3134a}'), + ('ð–¿°', 'ð–¿±'), + ('𘫳', '𘳕'), + ('𘴀', '𘴈'), + ('ðŸ„', 'ðŸ„'), + ('ðŸ…', '🅯'), + ('ðŸ†', 'ðŸ†'), + ('🛖', '🛗'), + ('🛻', '🛼'), + ('🢰', '🢱'), + ('🤌', '🤌'), + ('🥲', '🥲'), + ('🥷', '🥸'), + ('🦣', '🦤'), + ('🦫', 'ðŸ¦'), + ('🧋', '🧋'), + ('🩴', '🩴'), + ('🪃', '🪆'), + ('🪖', '🪨'), + ('🪰', '🪶'), + ('🫀', '🫂'), + ('ðŸ«', '🫖'), + ('🬀', '🮒'), + ('🮔', '🯊'), + ('🯰', '🯹'), + ('𪛗', 'ðª›'), + ('ð°€€', 'ð±Š'), +]; + +pub const V14_0: &'static [(char, char)] = &[ + ('Ø', 'Ø'), + ('à¡°', 'ࢎ'), + ('\u{890}', '\u{891}'), + ('\u{898}', '\u{89f}'), + ('ࢵ', 'ࢵ'), + ('ࣈ', '\u{8d2}'), + ('\u{c3c}', '\u{c3c}'), + ('à±', 'à±'), + ('à³', 'à³'), + ('áœ', 'áœ'), + ('᜕', '᜕'), + ('ᜟ', 'ᜟ'), + ('\u{180f}', '\u{180f}'), + ('\u{1ac1}', '\u{1ace}'), + ('áŒ', 'áŒ'), + ('á½', 'á¾'), + ('\u{1dfa}', '\u{1dfa}'), + ('⃀', '⃀'), + ('â°¯', 'â°¯'), + ('ⱟ', 'ⱟ'), + ('⹓', 'â¹'), + ('鿽', 'é¿¿'), + ('Ꟁ', 'êŸ'), + ('êŸ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟙ'), + ('ꟲ', 'ꟴ'), + ('﯂', '﯂'), + ('ïµ€', 'ïµ'), + ('ï·', 'ï·'), + ('ï·¾', 'ï·¿'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), + ('ðž€', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), + ('ð½°', 'ð¾‰'), + ('\u{11070}', 'ð‘µ'), + ('\u{110c2}', '\u{110c2}'), + ('ð‘š¹', 'ð‘š¹'), + ('ð‘€', 'ð‘†'), + ('𑪰', '𑪿'), + ('ð’¾', 'ð’¿²'), + ('ð–©°', '𖪾'), + ('ð–«€', '𖫉'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), + ('𛄟', '𛄢'), + ('\u{1cf00}', '\u{1cf2d}'), + ('\u{1cf30}', '\u{1cf46}'), + ('ðœ½', '𜿃'), + ('ð‡©', 'ð‡ª'), + ('ð¼€', 'ð¼ž'), + ('ðžŠ', '\u{1e2ae}'), + ('𞟠', '𞟦'), + ('𞟨', '𞟫'), + ('ðžŸ', '𞟮'), + ('𞟰', '𞟾'), + ('ðŸ›', '🛟'), + ('🟰', '🟰'), + ('🥹', '🥹'), + ('🧌', '🧌'), + ('🩻', '🩼'), + ('🪩', '🪬'), + ('🪷', '🪺'), + ('🫃', '🫅'), + ('🫗', '🫙'), + ('🫠', '🫧'), + ('🫰', '🫶'), + ('𪛞', '𪛟'), + ('𫜵', '𫜸'), ]; pub const V1_1: &'static [(char, char)] = &[ - ('\u{0}', 'ǵ'), + ('\0', 'ǵ'), ('Ǻ', 'È—'), ('É', 'ʨ'), ('Ê°', 'Ëž'), diff --git a/vendor/regex-syntax/src/unicode_tables/case_folding_simple.rs b/vendor/regex-syntax/src/unicode_tables/case_folding_simple.rs index cfb83f363..766d21b48 100644 --- a/vendor/regex-syntax/src/unicode_tables/case_folding_simple.rs +++ b/vendor/regex-syntax/src/unicode_tables/case_folding_simple.rs @@ -1,10 +1,10 @@ // DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY: // -// ucd-generate case-folding-simple ucd-13.0.0 --chars --all-pairs +// ucd-generate case-folding-simple /tmp/ucd --chars --all-pairs // -// Unicode version: 13.0.0. +// Unicode version: 14.0.0. // -// ucd-generate 0.2.8 is available on crates.io. +// ucd-generate 0.2.11 is available on crates.io. pub const CASE_FOLDING_SIMPLE: &'static [(char, &'static [char])] = &[ ('A', &['a']), @@ -1781,6 +1781,7 @@ pub const CASE_FOLDING_SIMPLE: &'static [(char, &'static [char])] = &[ ('â°¬', &['ⱜ']), ('â°', &['â±']), ('â°®', &['ⱞ']), + ('â°¯', &['ⱟ']), ('â°°', &['â°€']), ('â°±', &['â°']), ('â°²', &['â°‚']), @@ -1828,6 +1829,7 @@ pub const CASE_FOLDING_SIMPLE: &'static [(char, &'static [char])] = &[ ('ⱜ', &['â°¬']), ('â±', &['â°']), ('ⱞ', &['â°®']), + ('ⱟ', &['â°¯']), ('â± ', &['ⱡ']), ('ⱡ', &['â± ']), ('â±¢', &['É«']), @@ -2211,17 +2213,25 @@ pub const CASE_FOLDING_SIMPLE: &'static [(char, &'static [char])] = &[ ('êž½', &['êž¼']), ('êž¾', &['êž¿']), ('êž¿', &['êž¾']), + ('Ꟁ', &['êŸ']), + ('êŸ', &['Ꟁ']), ('Ꟃ', &['ꟃ']), ('ꟃ', &['Ꟃ']), ('Ꞔ', &['êž”']), ('Ʂ', &['Ê‚']), ('Ᶎ', &['ᶎ']), - ('\u{a7c7}', &['\u{a7c8}']), - ('\u{a7c8}', &['\u{a7c7}']), - ('\u{a7c9}', &['\u{a7ca}']), - ('\u{a7ca}', &['\u{a7c9}']), - ('\u{a7f5}', &['\u{a7f6}']), - ('\u{a7f6}', &['\u{a7f5}']), + ('Ꟈ', &['ꟈ']), + ('ꟈ', &['Ꟈ']), + ('Ꟊ', &['ꟊ']), + ('ꟊ', &['Ꟊ']), + ('êŸ', &['ꟑ']), + ('ꟑ', &['êŸ']), + ('Ꟗ', &['ꟗ']), + ('ꟗ', &['Ꟗ']), + ('Ꟙ', &['ꟙ']), + ('ꟙ', &['Ꟙ']), + ('Ꟶ', &['ꟶ']), + ('ꟶ', &['Ꟶ']), ('ê“', &['êž³']), ('ê°', &['Ꭰ']), ('ê±', &['Ꭱ']), @@ -2507,6 +2517,76 @@ pub const CASE_FOLDING_SIMPLE: &'static [(char, &'static [char])] = &[ ('ð“¹', &['ð“‘']), ('ð“º', &['ð“’']), ('ð“»', &['ð““']), + ('ð•°', &['ð–—']), + ('ð•±', &['ð–˜']), + ('ð•²', &['ð–™']), + ('ð•³', &['ð–š']), + ('ð•´', &['ð–›']), + ('ð•µ', &['ð–œ']), + ('ð•¶', &['ð–']), + ('ð•·', &['ð–ž']), + ('ð•¸', &['ð–Ÿ']), + ('ð•¹', &['ð– ']), + ('ð•º', &['ð–¡']), + ('ð•¼', &['ð–£']), + ('ð•½', &['ð–¤']), + ('ð•¾', &['ð–¥']), + ('ð•¿', &['ð–¦']), + ('ð–€', &['ð–§']), + ('ð–', &['ð–¨']), + ('ð–‚', &['ð–©']), + ('ð–ƒ', &['ð–ª']), + ('ð–„', &['ð–«']), + ('ð–…', &['ð–¬']), + ('ð–†', &['ð–']), + ('ð–‡', &['ð–®']), + ('ð–ˆ', &['ð–¯']), + ('ð–‰', &['ð–°']), + ('ð–Š', &['ð–±']), + ('ð–Œ', &['ð–³']), + ('ð–', &['ð–´']), + ('ð–Ž', &['ð–µ']), + ('ð–', &['ð–¶']), + ('ð–', &['ð–·']), + ('ð–‘', &['ð–¸']), + ('ð–’', &['ð–¹']), + ('ð–”', &['ð–»']), + ('ð–•', &['ð–¼']), + ('ð–—', &['ð•°']), + ('ð–˜', &['ð•±']), + ('ð–™', &['ð•²']), + ('ð–š', &['ð•³']), + ('ð–›', &['ð•´']), + ('ð–œ', &['ð•µ']), + ('ð–', &['ð•¶']), + ('ð–ž', &['ð•·']), + ('ð–Ÿ', &['ð•¸']), + ('ð– ', &['ð•¹']), + ('ð–¡', &['ð•º']), + ('ð–£', &['ð•¼']), + ('ð–¤', &['ð•½']), + ('ð–¥', &['ð•¾']), + ('ð–¦', &['ð•¿']), + ('ð–§', &['ð–€']), + ('ð–¨', &['ð–']), + ('ð–©', &['ð–‚']), + ('ð–ª', &['ð–ƒ']), + ('ð–«', &['ð–„']), + ('ð–¬', &['ð–…']), + ('ð–', &['ð–†']), + ('ð–®', &['ð–‡']), + ('ð–¯', &['ð–ˆ']), + ('ð–°', &['ð–‰']), + ('ð–±', &['ð–Š']), + ('ð–³', &['ð–Œ']), + ('ð–´', &['ð–']), + ('ð–µ', &['ð–Ž']), + ('ð–¶', &['ð–']), + ('ð–·', &['ð–']), + ('ð–¸', &['ð–‘']), + ('ð–¹', &['ð–’']), + ('ð–»', &['ð–”']), + ('ð–¼', &['ð–•']), ('ð²€', &['ð³€']), ('ð²', &['ð³']), ('ð²‚', &['ð³‚']), diff --git a/vendor/regex-syntax/src/unicode_tables/general_category.rs b/vendor/regex-syntax/src/unicode_tables/general_category.rs index 33b7b7e6e..8aa6b0078 100644 --- a/vendor/regex-syntax/src/unicode_tables/general_category.rs +++ b/vendor/regex-syntax/src/unicode_tables/general_category.rs @@ -1,10 +1,10 @@ // DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY: // -// ucd-generate general-category ucd-13.0.0 --chars --exclude surrogate +// ucd-generate general-category /tmp/ucd --chars --exclude surrogate // -// Unicode version: 13.0.0. +// Unicode version: 14.0.0. // -// ucd-generate 0.2.8 is available on crates.io. +// ucd-generate 0.2.11 is available on crates.io. pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("Cased_Letter", CASED_LETTER), @@ -116,9 +116,7 @@ pub const CASED_LETTER: &'static [(char, char)] = &[ ('â……', 'â…‰'), ('â…Ž', 'â…Ž'), ('Ↄ', 'ↄ'), - ('â°€', 'â°®'), - ('â°°', 'ⱞ'), - ('â± ', 'â±»'), + ('â°€', 'â±»'), ('â±¾', 'ⳤ'), ('Ⳬ', 'â³®'), ('â³²', 'â³³'), @@ -130,12 +128,14 @@ pub const CASED_LETTER: &'static [(char, char)] = &[ ('Ꜣ', 'ê¯'), ('ê±', 'ꞇ'), ('êž‹', 'ꞎ'), - ('êž', 'êž¿'), - ('Ꟃ', '\u{a7ca}'), - ('\u{a7f5}', '\u{a7f6}'), + ('êž', 'ꟊ'), + ('êŸ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟙ'), + ('Ꟶ', 'ꟶ'), ('ꟺ', 'ꟺ'), ('ꬰ', 'êš'), - ('ê ', '\u{ab68}'), + ('ê ', 'ê¨'), ('ê°', 'ꮿ'), ('ff', 'st'), ('ﬓ', 'ﬗ'), @@ -144,6 +144,14 @@ pub const CASED_LETTER: &'static [(char, char)] = &[ ('ð€', 'ð‘'), ('ð’°', 'ð““'), ('ð“˜', 'ð“»'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), ('ð²€', 'ð²²'), ('ð³€', 'ð³²'), ('ð‘¢ ', '𑣟'), @@ -178,6 +186,8 @@ pub const CASED_LETTER: &'static [(char, char)] = &[ ('ðžŠ', 'ðž¨'), ('ðžª', 'ðŸ‚'), ('ðŸ„', 'ðŸ‹'), + ('ð¼€', 'ð¼‰'), + ('ð¼‹', 'ð¼ž'), ('𞤀', '𞥃'), ]; @@ -225,6 +235,10 @@ pub const CLOSE_PUNCTUATION: &'static [(char, char)] = &[ ('⸥', '⸥'), ('⸧', '⸧'), ('⸩', '⸩'), + ('â¹–', 'â¹–'), + ('⹘', '⹘'), + ('⹚', '⹚'), + ('⹜', '⹜'), ('〉', '〉'), ('》', '》'), ('ã€', 'ã€'), @@ -266,7 +280,7 @@ pub const CONNECTOR_PUNCTUATION: &'static [(char, char)] = &[ ]; pub const CONTROL: &'static [(char, char)] = - &[('\u{0}', '\u{1f}'), ('\u{7f}', '\u{9f}')]; + &[('\0', '\u{1f}'), ('\u{7f}', '\u{9f}')]; pub const CURRENCY_SYMBOL: &'static [(char, char)] = &[ ('$', '$'), @@ -280,7 +294,7 @@ pub const CURRENCY_SYMBOL: &'static [(char, char)] = &[ ('௹', '௹'), ('฿', '฿'), ('៛', '៛'), - ('â‚ ', 'â‚¿'), + ('â‚ ', '⃀'), ('ê ¸', 'ê ¸'), ('ï·¼', 'ï·¼'), ('﹩', '﹩'), @@ -303,6 +317,7 @@ pub const DASH_PUNCTUATION: &'static [(char, char)] = &[ ('⸚', '⸚'), ('⸺', '⸻'), ('â¹€', 'â¹€'), + ('â¹', 'â¹'), ('〜', '〜'), ('〰', '〰'), ('ã‚ ', 'ã‚ '), @@ -310,7 +325,7 @@ pub const DASH_PUNCTUATION: &'static [(char, char)] = &[ ('﹘', '﹘'), ('ï¹£', 'ï¹£'), ('ï¼', 'ï¼'), - ('\u{10ead}', '\u{10ead}'), + ('ðº', 'ðº'), ]; pub const DECIMAL_NUMBER: &'static [(char, char)] = &[ @@ -364,17 +379,18 @@ pub const DECIMAL_NUMBER: &'static [(char, char)] = &[ ('𑛀', '𑛉'), ('𑜰', '𑜹'), ('ð‘£ ', '𑣩'), - ('\u{11950}', '\u{11959}'), + ('ð‘¥', 'ð‘¥™'), ('ð‘±', '𑱙'), ('ð‘µ', '𑵙'), ('𑶠', '𑶩'), ('ð–© ', 'ð–©©'), + ('ð–«€', '𖫉'), ('ð–', 'ð–™'), ('ðŸŽ', 'ðŸ¿'), ('ðž…€', 'ðž…‰'), ('ðž‹°', '𞋹'), ('ðž¥', '𞥙'), - ('\u{1fbf0}', '\u{1fbf9}'), + ('🯰', '🯹'), ]; pub const ENCLOSING_MARK: &'static [(char, char)] = &[ @@ -404,6 +420,7 @@ pub const FORMAT: &'static [(char, char)] = &[ ('\u{61c}', '\u{61c}'), ('\u{6dd}', '\u{6dd}'), ('\u{70f}', '\u{70f}'), + ('\u{890}', '\u{891}'), ('\u{8e2}', '\u{8e2}'), ('\u{180e}', '\u{180e}'), ('\u{200b}', '\u{200f}'), @@ -485,8 +502,9 @@ pub const LETTER: &'static [(char, char)] = &[ ('à ¨', 'à ¨'), ('à¡€', 'ࡘ'), ('à¡ ', 'ࡪ'), - ('ࢠ', 'ࢴ'), - ('ࢶ', '\u{8c7}'), + ('à¡°', 'ࢇ'), + ('ࢉ', 'ࢎ'), + ('ࢠ', 'ࣉ'), ('ऄ', 'ह'), ('ऽ', 'ऽ'), ('à¥', 'à¥'), @@ -551,6 +569,7 @@ pub const LETTER: &'static [(char, char)] = &[ ('à°ª', 'à°¹'), ('à°½', 'à°½'), ('ౘ', 'ౚ'), + ('à±', 'à±'), ('à± ', 'ౡ'), ('ಀ', 'ಀ'), ('ಅ', 'ಌ'), @@ -559,10 +578,10 @@ pub const LETTER: &'static [(char, char)] = &[ ('ಪ', 'ಳ'), ('ವ', 'ಹ'), ('ಽ', 'ಽ'), - ('ೞ', 'ೞ'), + ('à³', 'ೞ'), ('à³ ', 'ೡ'), ('à³±', 'à³²'), - ('\u{d04}', 'à´Œ'), + ('à´„', 'à´Œ'), ('à´Ž', 'à´'), ('à´’', 'à´º'), ('à´½', 'à´½'), @@ -630,9 +649,8 @@ pub const LETTER: &'static [(char, char)] = &[ ('áš', 'ášš'), ('áš ', 'ᛪ'), ('á›±', 'ᛸ'), - ('ᜀ', 'ᜌ'), - ('ᜎ', 'ᜑ'), - ('ᜠ', 'ᜱ'), + ('ᜀ', 'ᜑ'), + ('ᜟ', 'ᜱ'), ('á€', 'á‘'), ('á ', 'á¬'), ('á®', 'á°'), @@ -653,7 +671,7 @@ pub const LETTER: &'static [(char, char)] = &[ ('ᨠ', 'á©”'), ('ᪧ', 'ᪧ'), ('ᬅ', 'ᬳ'), - ('á…', 'á‹'), + ('á…', 'áŒ'), ('ᮃ', 'á® '), ('á®®', 'ᮯ'), ('ᮺ', 'ᯥ'), @@ -704,9 +722,7 @@ pub const LETTER: &'static [(char, char)] = &[ ('â……', 'â…‰'), ('â…Ž', 'â…Ž'), ('Ↄ', 'ↄ'), - ('â°€', 'â°®'), - ('â°°', 'ⱞ'), - ('â± ', 'ⳤ'), + ('â°€', 'ⳤ'), ('Ⳬ', 'â³®'), ('â³²', 'â³³'), ('â´€', 'â´¥'), @@ -733,11 +749,10 @@ pub const LETTER: &'static [(char, char)] = &[ ('ー', 'ヿ'), ('ã„…', 'ㄯ'), ('ㄱ', 'ㆎ'), - ('ㆠ', '\u{31bf}'), + ('ㆠ', 'ㆿ'), ('ㇰ', 'ㇿ'), - ('ã€', '\u{4dbf}'), - ('一', '\u{9ffc}'), - ('ꀀ', 'ê’Œ'), + ('ã€', '䶿'), + ('一', 'ê’Œ'), ('ê“', 'ꓽ'), ('ꔀ', 'ꘌ'), ('ê˜', 'ꘟ'), @@ -747,9 +762,11 @@ pub const LETTER: &'static [(char, char)] = &[ ('êš ', 'ꛥ'), ('ꜗ', 'ꜟ'), ('Ꜣ', 'ꞈ'), - ('êž‹', 'êž¿'), - ('Ꟃ', '\u{a7ca}'), - ('\u{a7f5}', 'ê '), + ('êž‹', 'ꟊ'), + ('êŸ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟙ'), + ('ꟲ', 'ê '), ('ê ƒ', 'ê …'), ('ê ‡', 'ê Š'), ('ê Œ', 'ê ¢'), @@ -786,7 +803,7 @@ pub const LETTER: &'static [(char, char)] = &[ ('ê¬ ', 'ꬦ'), ('ꬨ', 'ꬮ'), ('ꬰ', 'êš'), - ('êœ', '\u{ab69}'), + ('êœ', 'ê©'), ('ê°', 'ꯢ'), ('ê°€', '힣'), ('íž°', 'ퟆ'), @@ -837,9 +854,20 @@ pub const LETTER: &'static [(char, char)] = &[ ('ð“˜', 'ð“»'), ('ð”€', 'ð”§'), ('ð”°', 'ð•£'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), ('ð˜€', 'ðœ¶'), ('ð€', 'ð•'), ('ð ', 'ð§'), + ('ðž€', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ('ð €', 'ð …'), ('ð ˆ', 'ð ˆ'), ('ð Š', 'ð µ'), @@ -870,19 +898,22 @@ pub const LETTER: &'static [(char, char)] = &[ ('ð²€', 'ð²²'), ('ð³€', 'ð³²'), ('ð´€', 'ð´£'), - ('\u{10e80}', '\u{10ea9}'), - ('\u{10eb0}', '\u{10eb1}'), + ('ðº€', 'ðº©'), + ('ðº°', 'ðº±'), ('ð¼€', 'ð¼œ'), ('ð¼§', 'ð¼§'), ('ð¼°', 'ð½…'), - ('\u{10fb0}', '\u{10fc4}'), + ('ð½°', 'ð¾'), + ('ð¾°', 'ð¿„'), ('ð¿ ', 'ð¿¶'), ('𑀃', 'ð‘€·'), + ('ð‘±', 'ð‘²'), + ('ð‘µ', 'ð‘µ'), ('𑂃', '𑂯'), ('ð‘ƒ', '𑃨'), ('𑄃', '𑄦'), ('ð‘…„', 'ð‘…„'), - ('\u{11147}', '\u{11147}'), + ('ð‘…‡', 'ð‘…‡'), ('ð‘…', 'ð‘…²'), ('ð‘…¶', 'ð‘…¶'), ('𑆃', '𑆲'), @@ -908,7 +939,7 @@ pub const LETTER: &'static [(char, char)] = &[ ('ð‘', 'ð‘¡'), ('ð‘€', 'ð‘´'), ('𑑇', 'ð‘‘Š'), - ('ð‘‘Ÿ', '\u{11461}'), + ('ð‘‘Ÿ', 'ð‘‘¡'), ('ð‘’€', 'ð‘’¯'), ('ð‘“„', 'ð‘“…'), ('𑓇', '𑓇'), @@ -919,15 +950,16 @@ pub const LETTER: &'static [(char, char)] = &[ ('𑚀', '𑚪'), ('𑚸', '𑚸'), ('𑜀', '𑜚'), + ('ð‘€', 'ð‘†'), ('ð‘ €', 'ð‘ «'), ('ð‘¢ ', '𑣟'), - ('𑣿', '\u{11906}'), - ('\u{11909}', '\u{11909}'), - ('\u{1190c}', '\u{11913}'), - ('\u{11915}', '\u{11916}'), - ('\u{11918}', '\u{1192f}'), - ('\u{1193f}', '\u{1193f}'), - ('\u{11941}', '\u{11941}'), + ('𑣿', '𑤆'), + ('𑤉', '𑤉'), + ('𑤌', '𑤓'), + ('𑤕', '𑤖'), + ('𑤘', '𑤯'), + ('𑤿', '𑤿'), + ('ð‘¥', 'ð‘¥'), ('𑦠', '𑦧'), ('𑦪', 'ð‘§'), ('𑧡', '𑧡'), @@ -938,7 +970,7 @@ pub const LETTER: &'static [(char, char)] = &[ ('ð‘©', 'ð‘©'), ('ð‘©œ', '𑪉'), ('ð‘ª', 'ð‘ª'), - ('ð‘«€', '𑫸'), + ('𑪰', '𑫸'), ('ð‘°€', 'ð‘°ˆ'), ('ð‘°Š', 'ð‘°®'), ('𑱀', '𑱀'), @@ -952,13 +984,15 @@ pub const LETTER: &'static [(char, char)] = &[ ('𑵪', '𑶉'), ('𑶘', '𑶘'), ('ð‘» ', 'ð‘»²'), - ('\u{11fb0}', '\u{11fb0}'), + ('𑾰', '𑾰'), ('ð’€€', 'ð’Ž™'), ('ð’’€', '𒕃'), + ('ð’¾', 'ð’¿°'), ('ð“€€', 'ð“®'), ('ð”€', '𔙆'), ('ð– €', '𖨸'), ('ð–©€', 'ð–©ž'), + ('ð–©°', '𖪾'), ('ð–«', 'ð–«'), ('𖬀', '𖬯'), ('ð–€', 'ð–ƒ'), @@ -971,9 +1005,12 @@ pub const LETTER: &'static [(char, char)] = &[ ('ð–¿ ', 'ð–¿¡'), ('ð–¿£', 'ð–¿£'), ('ð—€€', '𘟷'), - ('𘠀', '\u{18cd5}'), - ('\u{18d00}', '\u{18d08}'), - ('𛀀', '𛄞'), + ('𘠀', '𘳕'), + ('𘴀', '𘴈'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), + ('𛀀', '𛄢'), ('ð›…', 'ð›…’'), ('ð›…¤', 'ð›…§'), ('ð›…°', '𛋻'), @@ -1011,10 +1048,16 @@ pub const LETTER: &'static [(char, char)] = &[ ('ðžŠ', 'ðž¨'), ('ðžª', 'ðŸ‚'), ('ðŸ„', 'ðŸ‹'), + ('ð¼€', 'ð¼ž'), ('ðž„€', '𞄬'), ('ðž„·', '𞄽'), ('ðž…Ž', 'ðž…Ž'), + ('ðžŠ', 'ðžŠ'), ('ðž‹€', 'ðž‹«'), + ('𞟠', '𞟦'), + ('𞟨', '𞟫'), + ('ðžŸ', '𞟮'), + ('𞟰', '𞟾'), ('ðž €', '𞣄'), ('𞤀', '𞥃'), ('𞥋', '𞥋'), @@ -1051,13 +1094,13 @@ pub const LETTER: &'static [(char, char)] = &[ ('𞺡', '𞺣'), ('𞺥', '𞺩'), ('𞺫', '𞺻'), - ('ð €€', '\u{2a6dd}'), - ('𪜀', '𫜴'), + ('ð €€', '𪛟'), + ('𪜀', '𫜸'), ('ð«€', 'ð« '), ('ð« ', '𬺡'), ('𬺰', '𮯠'), ('丽', 'ð¯¨'), - ('\u{30000}', '\u{3134a}'), + ('ð°€€', 'ð±Š'), ]; pub const LETTER_NUMBER: &'static [(char, char)] = &[ @@ -1510,7 +1553,7 @@ pub const LOWERCASE_LETTER: &'static [(char, char)] = &[ ('â…†', 'â…‰'), ('â…Ž', 'â…Ž'), ('ↄ', 'ↄ'), - ('â°°', 'ⱞ'), + ('â°°', 'ⱟ'), ('ⱡ', 'ⱡ'), ('â±¥', 'ⱦ'), ('ⱨ', 'ⱨ'), @@ -1679,19 +1722,29 @@ pub const LOWERCASE_LETTER: &'static [(char, char)] = &[ ('êž»', 'êž»'), ('êž½', 'êž½'), ('êž¿', 'êž¿'), + ('êŸ', 'êŸ'), ('ꟃ', 'ꟃ'), - ('\u{a7c8}', '\u{a7c8}'), - ('\u{a7ca}', '\u{a7ca}'), - ('\u{a7f6}', '\u{a7f6}'), + ('ꟈ', 'ꟈ'), + ('ꟊ', 'ꟊ'), + ('ꟑ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟕ'), + ('ꟗ', 'ꟗ'), + ('ꟙ', 'ꟙ'), + ('ꟶ', 'ꟶ'), ('ꟺ', 'ꟺ'), ('ꬰ', 'êš'), - ('ê ', '\u{ab68}'), + ('ê ', 'ê¨'), ('ê°', 'ꮿ'), ('ff', 'st'), ('ﬓ', 'ﬗ'), ('ï½', 'z'), ('ð¨', 'ð‘'), ('ð“˜', 'ð“»'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), ('ð³€', 'ð³²'), ('ð‘£€', '𑣟'), ('ð–¹ ', '𖹿'), @@ -1723,6 +1776,8 @@ pub const LOWERCASE_LETTER: &'static [(char, char)] = &[ ('ðžª', 'ðŸ‚'), ('ðŸ„', 'ðŸ‰'), ('ðŸ‹', 'ðŸ‹'), + ('ð¼€', 'ð¼‰'), + ('ð¼‹', 'ð¼ž'), ('𞤢', '𞥃'), ]; @@ -1751,7 +1806,8 @@ pub const MARK: &'static [(char, char)] = &[ ('\u{825}', '\u{827}'), ('\u{829}', '\u{82d}'), ('\u{859}', '\u{85b}'), - ('\u{8d3}', '\u{8e1}'), + ('\u{898}', '\u{89f}'), + ('\u{8ca}', '\u{8e1}'), ('\u{8e3}', 'ः'), ('\u{93a}', '\u{93c}'), ('ा', 'à¥'), @@ -1793,6 +1849,7 @@ pub const MARK: &'static [(char, char)] = &[ ('ொ', '\u{bcd}'), ('\u{bd7}', '\u{bd7}'), ('\u{c00}', '\u{c04}'), + ('\u{c3c}', '\u{c3c}'), ('\u{c3e}', 'ౄ'), ('\u{c46}', '\u{c48}'), ('\u{c4a}', '\u{c4d}'), @@ -1844,13 +1901,14 @@ pub const MARK: &'static [(char, char)] = &[ ('á‚', 'á‚'), ('á‚š', '\u{109d}'), ('\u{135d}', '\u{135f}'), - ('\u{1712}', '\u{1714}'), - ('\u{1732}', '\u{1734}'), + ('\u{1712}', '᜕'), + ('\u{1732}', '᜴'), ('\u{1752}', '\u{1753}'), ('\u{1772}', '\u{1773}'), ('\u{17b4}', '\u{17d3}'), ('\u{17dd}', '\u{17dd}'), ('\u{180b}', '\u{180d}'), + ('\u{180f}', '\u{180f}'), ('\u{1885}', '\u{1886}'), ('\u{18a9}', '\u{18a9}'), ('\u{1920}', 'ᤫ'), @@ -1859,7 +1917,7 @@ pub const MARK: &'static [(char, char)] = &[ ('á©•', '\u{1a5e}'), ('\u{1a60}', '\u{1a7c}'), ('\u{1a7f}', '\u{1a7f}'), - ('\u{1ab0}', '\u{1ac0}'), + ('\u{1ab0}', '\u{1ace}'), ('\u{1b00}', 'ᬄ'), ('\u{1b34}', 'á„'), ('\u{1b6b}', '\u{1b73}'), @@ -1872,8 +1930,7 @@ pub const MARK: &'static [(char, char)] = &[ ('\u{1ced}', '\u{1ced}'), ('\u{1cf4}', '\u{1cf4}'), ('á³·', '\u{1cf9}'), - ('\u{1dc0}', '\u{1df9}'), - ('\u{1dfb}', '\u{1dff}'), + ('\u{1dc0}', '\u{1dff}'), ('\u{20d0}', '\u{20f0}'), ('\u{2cef}', '\u{2cf1}'), ('\u{2d7f}', '\u{2d7f}'), @@ -1926,10 +1983,14 @@ pub const MARK: &'static [(char, char)] = &[ ('\u{10d24}', '\u{10d27}'), ('\u{10eab}', '\u{10eac}'), ('\u{10f46}', '\u{10f50}'), + ('\u{10f82}', '\u{10f85}'), ('ð‘€€', '𑀂'), ('\u{11038}', '\u{11046}'), + ('\u{11070}', '\u{11070}'), + ('\u{11073}', '\u{11074}'), ('\u{1107f}', 'ð‘‚‚'), ('ð‘‚°', '\u{110ba}'), + ('\u{110c2}', '\u{110c2}'), ('\u{11100}', '\u{11102}'), ('\u{11127}', '\u{11134}'), ('ð‘……', 'ð‘…†'), @@ -1937,7 +1998,7 @@ pub const MARK: &'static [(char, char)] = &[ ('\u{11180}', '𑆂'), ('𑆳', '𑇀'), ('\u{111c9}', '\u{111cc}'), - ('\u{111ce}', '\u{111cf}'), + ('𑇎', '\u{111cf}'), ('𑈬', '\u{11237}'), ('\u{1123e}', '\u{1123e}'), ('\u{112df}', '\u{112ea}'), @@ -1960,11 +2021,11 @@ pub const MARK: &'static [(char, char)] = &[ ('\u{116ab}', '\u{116b7}'), ('\u{1171d}', '\u{1172b}'), ('ð‘ ¬', '\u{1183a}'), - ('\u{11930}', '\u{11935}'), - ('\u{11937}', '\u{11938}'), + ('\u{11930}', '𑤵'), + ('𑤷', '𑤸'), ('\u{1193b}', '\u{1193e}'), - ('\u{11940}', '\u{11940}'), - ('\u{11942}', '\u{11943}'), + ('ð‘¥€', 'ð‘¥€'), + ('𑥂', '\u{11943}'), ('𑧑', '\u{119d7}'), ('\u{119da}', '\u{119e0}'), ('𑧤', '𑧤'), @@ -1993,8 +2054,10 @@ pub const MARK: &'static [(char, char)] = &[ ('𖽑', '𖾇'), ('\u{16f8f}', '\u{16f92}'), ('\u{16fe4}', '\u{16fe4}'), - ('\u{16ff0}', '\u{16ff1}'), + ('ð–¿°', 'ð–¿±'), ('\u{1bc9d}', '\u{1bc9e}'), + ('\u{1cf00}', '\u{1cf2d}'), + ('\u{1cf30}', '\u{1cf46}'), ('\u{1d165}', '\u{1d169}'), ('ð…', '\u{1d172}'), ('\u{1d17b}', '\u{1d182}'), @@ -2013,6 +2076,7 @@ pub const MARK: &'static [(char, char)] = &[ ('\u{1e023}', '\u{1e024}'), ('\u{1e026}', '\u{1e02a}'), ('\u{1e130}', '\u{1e136}'), + ('\u{1e2ae}', '\u{1e2ae}'), ('\u{1e2ec}', '\u{1e2ef}'), ('\u{1e8d0}', '\u{1e8d6}'), ('\u{1e944}', '\u{1e94a}'), @@ -2102,6 +2166,7 @@ pub const MODIFIER_LETTER: &'static [(char, char)] = &[ ('à š', 'à š'), ('à ¤', 'à ¤'), ('à ¨', 'à ¨'), + ('ࣉ', 'ࣉ'), ('ॱ', 'ॱ'), ('ๆ', 'ๆ'), ('ໆ', 'ໆ'), @@ -2132,6 +2197,7 @@ pub const MODIFIER_LETTER: &'static [(char, char)] = &[ ('ꜗ', 'ꜟ'), ('ê°', 'ê°'), ('ꞈ', 'ꞈ'), + ('ꟲ', 'ꟴ'), ('ꟸ', 'ꟹ'), ('ê§', 'ê§'), ('ꧦ', 'ꧦ'), @@ -2139,13 +2205,19 @@ pub const MODIFIER_LETTER: &'static [(char, char)] = &[ ('ê«', 'ê«'), ('ꫳ', 'ê«´'), ('êœ', 'êŸ'), - ('\u{ab69}', '\u{ab69}'), + ('ê©', 'ê©'), ('ï½°', 'ï½°'), ('\u{ff9e}', '\u{ff9f}'), + ('ðž€', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ('ð–€', 'ð–ƒ'), ('𖾓', '𖾟'), ('ð–¿ ', 'ð–¿¡'), ('ð–¿£', 'ð–¿£'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), ('ðž„·', '𞄽'), ('𞥋', '𞥋'), ]; @@ -2164,6 +2236,7 @@ pub const MODIFIER_SYMBOL: &'static [(char, char)] = &[ ('˯', 'Ë¿'), ('͵', '͵'), ('΄', 'Î…'), + ('࢈', '࢈'), ('á¾½', 'á¾½'), ('᾿', 'á¿'), ('á¿', 'á¿'), @@ -2175,8 +2248,8 @@ pub const MODIFIER_SYMBOL: &'static [(char, char)] = &[ ('êœ ', '꜡'), ('꞉', '꞊'), ('ê›', 'ê›'), - ('\u{ab6a}', '\u{ab6b}'), - ('﮲', 'ï¯'), + ('êª', 'ê«'), + ('﮲', '﯂'), ('ï¼¾', 'ï¼¾'), ('ï½€', 'ï½€'), ('ï¿£', 'ï¿£'), @@ -2208,7 +2281,8 @@ pub const NONSPACING_MARK: &'static [(char, char)] = &[ ('\u{825}', '\u{827}'), ('\u{829}', '\u{82d}'), ('\u{859}', '\u{85b}'), - ('\u{8d3}', '\u{8e1}'), + ('\u{898}', '\u{89f}'), + ('\u{8ca}', '\u{8e1}'), ('\u{8e3}', '\u{902}'), ('\u{93a}', '\u{93a}'), ('\u{93c}', '\u{93c}'), @@ -2249,6 +2323,7 @@ pub const NONSPACING_MARK: &'static [(char, char)] = &[ ('\u{bcd}', '\u{bcd}'), ('\u{c00}', '\u{c00}'), ('\u{c04}', '\u{c04}'), + ('\u{c3c}', '\u{c3c}'), ('\u{c3e}', '\u{c40}'), ('\u{c46}', '\u{c48}'), ('\u{c4a}', '\u{c4d}'), @@ -2298,7 +2373,7 @@ pub const NONSPACING_MARK: &'static [(char, char)] = &[ ('\u{109d}', '\u{109d}'), ('\u{135d}', '\u{135f}'), ('\u{1712}', '\u{1714}'), - ('\u{1732}', '\u{1734}'), + ('\u{1732}', '\u{1733}'), ('\u{1752}', '\u{1753}'), ('\u{1772}', '\u{1773}'), ('\u{17b4}', '\u{17b5}'), @@ -2307,6 +2382,7 @@ pub const NONSPACING_MARK: &'static [(char, char)] = &[ ('\u{17c9}', '\u{17d3}'), ('\u{17dd}', '\u{17dd}'), ('\u{180b}', '\u{180d}'), + ('\u{180f}', '\u{180f}'), ('\u{1885}', '\u{1886}'), ('\u{18a9}', '\u{18a9}'), ('\u{1920}', '\u{1922}'), @@ -2323,7 +2399,7 @@ pub const NONSPACING_MARK: &'static [(char, char)] = &[ ('\u{1a73}', '\u{1a7c}'), ('\u{1a7f}', '\u{1a7f}'), ('\u{1ab0}', '\u{1abd}'), - ('\u{1abf}', '\u{1ac0}'), + ('\u{1abf}', '\u{1ace}'), ('\u{1b00}', '\u{1b03}'), ('\u{1b34}', '\u{1b34}'), ('\u{1b36}', '\u{1b3a}'), @@ -2346,8 +2422,7 @@ pub const NONSPACING_MARK: &'static [(char, char)] = &[ ('\u{1ced}', '\u{1ced}'), ('\u{1cf4}', '\u{1cf4}'), ('\u{1cf8}', '\u{1cf9}'), - ('\u{1dc0}', '\u{1df9}'), - ('\u{1dfb}', '\u{1dff}'), + ('\u{1dc0}', '\u{1dff}'), ('\u{20d0}', '\u{20dc}'), ('\u{20e1}', '\u{20e1}'), ('\u{20e5}', '\u{20f0}'), @@ -2406,11 +2481,15 @@ pub const NONSPACING_MARK: &'static [(char, char)] = &[ ('\u{10d24}', '\u{10d27}'), ('\u{10eab}', '\u{10eac}'), ('\u{10f46}', '\u{10f50}'), + ('\u{10f82}', '\u{10f85}'), ('\u{11001}', '\u{11001}'), ('\u{11038}', '\u{11046}'), + ('\u{11070}', '\u{11070}'), + ('\u{11073}', '\u{11074}'), ('\u{1107f}', '\u{11081}'), ('\u{110b3}', '\u{110b6}'), ('\u{110b9}', '\u{110ba}'), + ('\u{110c2}', '\u{110c2}'), ('\u{11100}', '\u{11102}'), ('\u{11127}', '\u{1112b}'), ('\u{1112d}', '\u{11134}'), @@ -2490,6 +2569,8 @@ pub const NONSPACING_MARK: &'static [(char, char)] = &[ ('\u{16f8f}', '\u{16f92}'), ('\u{16fe4}', '\u{16fe4}'), ('\u{1bc9d}', '\u{1bc9e}'), + ('\u{1cf00}', '\u{1cf2d}'), + ('\u{1cf30}', '\u{1cf46}'), ('\u{1d167}', '\u{1d169}'), ('\u{1d17b}', '\u{1d182}'), ('\u{1d185}', '\u{1d18b}'), @@ -2507,6 +2588,7 @@ pub const NONSPACING_MARK: &'static [(char, char)] = &[ ('\u{1e023}', '\u{1e024}'), ('\u{1e026}', '\u{1e02a}'), ('\u{1e130}', '\u{1e136}'), + ('\u{1e2ae}', '\u{1e2ae}'), ('\u{1e2ec}', '\u{1e2ef}'), ('\u{1e8d0}', '\u{1e8d6}'), ('\u{1e944}', '\u{1e94a}'), @@ -2610,7 +2692,7 @@ pub const NUMBER: &'static [(char, char)] = &[ ('ð¹ ', 'ð¹¾'), ('ð¼', 'ð¼¦'), ('ð½‘', 'ð½”'), - ('\u{10fc5}', '\u{10fcb}'), + ('ð¿…', 'ð¿‹'), ('ð‘’', 'ð‘¯'), ('𑃰', '𑃹'), ('𑄶', 'ð‘„¿'), @@ -2623,13 +2705,14 @@ pub const NUMBER: &'static [(char, char)] = &[ ('𑛀', '𑛉'), ('𑜰', '𑜻'), ('ð‘£ ', 'ð‘£²'), - ('\u{11950}', '\u{11959}'), + ('ð‘¥', 'ð‘¥™'), ('ð‘±', '𑱬'), ('ð‘µ', '𑵙'), ('𑶠', '𑶩'), ('ð‘¿€', 'ð‘¿”'), ('ð’€', 'ð’‘®'), ('ð–© ', 'ð–©©'), + ('ð–«€', '𖫉'), ('ð–', 'ð–™'), ('ð–›', 'ð–¡'), ('𖺀', 'ð–º–'), @@ -2646,7 +2729,7 @@ pub const NUMBER: &'static [(char, char)] = &[ ('ðž´', 'ðž´'), ('ðž´¯', 'ðž´½'), ('🄀', '🄌'), - ('\u{1fbf0}', '\u{1fbf9}'), + ('🯰', '🯹'), ]; pub const OPEN_PUNCTUATION: &'static [(char, char)] = &[ @@ -2696,6 +2779,10 @@ pub const OPEN_PUNCTUATION: &'static [(char, char)] = &[ ('⸦', '⸦'), ('⸨', '⸨'), ('⹂', '⹂'), + ('⹕', '⹕'), + ('â¹—', 'â¹—'), + ('â¹™', 'â¹™'), + ('â¹›', 'â¹›'), ('〈', '〈'), ('《', '《'), ('「', '「'), @@ -2728,7 +2815,7 @@ pub const OPEN_PUNCTUATION: &'static [(char, char)] = &[ ]; pub const OTHER: &'static [(char, char)] = &[ - ('\u{0}', '\u{1f}'), + ('\0', '\u{1f}'), ('\u{7f}', '\u{9f}'), ('\u{ad}', '\u{ad}'), ('\u{378}', '\u{379}'), @@ -2743,7 +2830,7 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{5c8}', '\u{5cf}'), ('\u{5eb}', '\u{5ee}'), ('\u{5f5}', '\u{605}'), - ('\u{61c}', '\u{61d}'), + ('\u{61c}', '\u{61c}'), ('\u{6dd}', '\u{6dd}'), ('\u{70e}', '\u{70f}'), ('\u{74b}', '\u{74c}'), @@ -2753,9 +2840,8 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{83f}', '\u{83f}'), ('\u{85c}', '\u{85d}'), ('\u{85f}', '\u{85f}'), - ('\u{86b}', '\u{89f}'), - ('\u{8b5}', '\u{8b5}'), - ('\u{8c8}', '\u{8d2}'), + ('\u{86b}', '\u{86f}'), + ('\u{88f}', '\u{897}'), ('\u{8e2}', '\u{8e2}'), ('\u{984}', '\u{984}'), ('\u{98d}', '\u{98e}'), @@ -2834,12 +2920,13 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{c0d}', '\u{c0d}'), ('\u{c11}', '\u{c11}'), ('\u{c29}', '\u{c29}'), - ('\u{c3a}', '\u{c3c}'), + ('\u{c3a}', '\u{c3b}'), ('\u{c45}', '\u{c45}'), ('\u{c49}', '\u{c49}'), ('\u{c4e}', '\u{c54}'), ('\u{c57}', '\u{c57}'), - ('\u{c5b}', '\u{c5f}'), + ('\u{c5b}', '\u{c5c}'), + ('\u{c5e}', '\u{c5f}'), ('\u{c64}', '\u{c65}'), ('\u{c70}', '\u{c76}'), ('\u{c8d}', '\u{c8d}'), @@ -2850,7 +2937,7 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{cc5}', '\u{cc5}'), ('\u{cc9}', '\u{cc9}'), ('\u{cce}', '\u{cd4}'), - ('\u{cd7}', '\u{cdd}'), + ('\u{cd7}', '\u{cdc}'), ('\u{cdf}', '\u{cdf}'), ('\u{ce4}', '\u{ce5}'), ('\u{cf0}', '\u{cf0}'), @@ -2918,8 +3005,7 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{13fe}', '\u{13ff}'), ('\u{169d}', '\u{169f}'), ('\u{16f9}', '\u{16ff}'), - ('\u{170d}', '\u{170d}'), - ('\u{1715}', '\u{171f}'), + ('\u{1716}', '\u{171e}'), ('\u{1737}', '\u{173f}'), ('\u{1754}', '\u{175f}'), ('\u{176d}', '\u{176d}'), @@ -2928,7 +3014,7 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{17de}', '\u{17df}'), ('\u{17ea}', '\u{17ef}'), ('\u{17fa}', '\u{17ff}'), - ('\u{180e}', '\u{180f}'), + ('\u{180e}', '\u{180e}'), ('\u{181a}', '\u{181f}'), ('\u{1879}', '\u{187f}'), ('\u{18ab}', '\u{18af}'), @@ -2948,9 +3034,9 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{1a8a}', '\u{1a8f}'), ('\u{1a9a}', '\u{1a9f}'), ('\u{1aae}', '\u{1aaf}'), - ('\u{1ac1}', '\u{1aff}'), - ('\u{1b4c}', '\u{1b4f}'), - ('\u{1b7d}', '\u{1b7f}'), + ('\u{1acf}', '\u{1aff}'), + ('\u{1b4d}', '\u{1b4f}'), + ('\u{1b7f}', '\u{1b7f}'), ('\u{1bf4}', '\u{1bfb}'), ('\u{1c38}', '\u{1c3a}'), ('\u{1c4a}', '\u{1c4c}'), @@ -2958,7 +3044,6 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{1cbb}', '\u{1cbc}'), ('\u{1cc8}', '\u{1ccf}'), ('\u{1cfb}', '\u{1cff}'), - ('\u{1dfa}', '\u{1dfa}'), ('\u{1f16}', '\u{1f17}'), ('\u{1f1e}', '\u{1f1f}'), ('\u{1f46}', '\u{1f47}'), @@ -2981,15 +3066,13 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{2072}', '\u{2073}'), ('\u{208f}', '\u{208f}'), ('\u{209d}', '\u{209f}'), - ('\u{20c0}', '\u{20cf}'), + ('\u{20c1}', '\u{20cf}'), ('\u{20f1}', '\u{20ff}'), ('\u{218c}', '\u{218f}'), ('\u{2427}', '\u{243f}'), ('\u{244b}', '\u{245f}'), ('\u{2b74}', '\u{2b75}'), ('\u{2b96}', '\u{2b96}'), - ('\u{2c2f}', '\u{2c2f}'), - ('\u{2c5f}', '\u{2c5f}'), ('\u{2cf4}', '\u{2cf8}'), ('\u{2d26}', '\u{2d26}'), ('\u{2d28}', '\u{2d2c}'), @@ -3005,7 +3088,7 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{2dcf}', '\u{2dcf}'), ('\u{2dd7}', '\u{2dd7}'), ('\u{2ddf}', '\u{2ddf}'), - ('\u{2e53}', '\u{2e7f}'), + ('\u{2e5e}', '\u{2e7f}'), ('\u{2e9a}', '\u{2e9a}'), ('\u{2ef4}', '\u{2eff}'), ('\u{2fd6}', '\u{2fef}'), @@ -3017,13 +3100,14 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{318f}', '\u{318f}'), ('\u{31e4}', '\u{31ef}'), ('\u{321f}', '\u{321f}'), - ('\u{9ffd}', '\u{9fff}'), ('\u{a48d}', '\u{a48f}'), ('\u{a4c7}', '\u{a4cf}'), ('\u{a62c}', '\u{a63f}'), ('\u{a6f8}', '\u{a6ff}'), - ('\u{a7c0}', '\u{a7c1}'), - ('\u{a7cb}', '\u{a7f4}'), + ('\u{a7cb}', '\u{a7cf}'), + ('\u{a7d2}', '\u{a7d2}'), + ('\u{a7d4}', '\u{a7d4}'), + ('\u{a7da}', '\u{a7f1}'), ('\u{a82d}', '\u{a82f}'), ('\u{a83a}', '\u{a83f}'), ('\u{a878}', '\u{a87f}'), @@ -3059,11 +3143,10 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{fb3f}', '\u{fb3f}'), ('\u{fb42}', '\u{fb42}'), ('\u{fb45}', '\u{fb45}'), - ('\u{fbc2}', '\u{fbd2}'), - ('\u{fd40}', '\u{fd4f}'), + ('\u{fbc3}', '\u{fbd2}'), ('\u{fd90}', '\u{fd91}'), - ('\u{fdc8}', '\u{fdef}'), - ('\u{fdfe}', '\u{fdff}'), + ('\u{fdc8}', '\u{fdce}'), + ('\u{fdd0}', '\u{fdef}'), ('\u{fe1a}', '\u{fe1f}'), ('\u{fe53}', '\u{fe53}'), ('\u{fe67}', '\u{fe67}'), @@ -3106,10 +3189,20 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{104fc}', '\u{104ff}'), ('\u{10528}', '\u{1052f}'), ('\u{10564}', '\u{1056e}'), - ('\u{10570}', '\u{105ff}'), + ('\u{1057b}', '\u{1057b}'), + ('\u{1058b}', '\u{1058b}'), + ('\u{10593}', '\u{10593}'), + ('\u{10596}', '\u{10596}'), + ('\u{105a2}', '\u{105a2}'), + ('\u{105b2}', '\u{105b2}'), + ('\u{105ba}', '\u{105ba}'), + ('\u{105bd}', '\u{105ff}'), ('\u{10737}', '\u{1073f}'), ('\u{10756}', '\u{1075f}'), - ('\u{10768}', '\u{107ff}'), + ('\u{10768}', '\u{1077f}'), + ('\u{10786}', '\u{10786}'), + ('\u{107b1}', '\u{107b1}'), + ('\u{107bb}', '\u{107ff}'), ('\u{10806}', '\u{10807}'), ('\u{10809}', '\u{10809}'), ('\u{10836}', '\u{10836}'), @@ -3152,13 +3245,14 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{10eae}', '\u{10eaf}'), ('\u{10eb2}', '\u{10eff}'), ('\u{10f28}', '\u{10f2f}'), - ('\u{10f5a}', '\u{10faf}'), + ('\u{10f5a}', '\u{10f6f}'), + ('\u{10f8a}', '\u{10faf}'), ('\u{10fcc}', '\u{10fdf}'), ('\u{10ff7}', '\u{10fff}'), ('\u{1104e}', '\u{11051}'), - ('\u{11070}', '\u{1107e}'), + ('\u{11076}', '\u{1107e}'), ('\u{110bd}', '\u{110bd}'), - ('\u{110c2}', '\u{110cf}'), + ('\u{110c3}', '\u{110cf}'), ('\u{110e9}', '\u{110ef}'), ('\u{110fa}', '\u{110ff}'), ('\u{11135}', '\u{11135}'), @@ -3199,11 +3293,11 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{11645}', '\u{1164f}'), ('\u{1165a}', '\u{1165f}'), ('\u{1166d}', '\u{1167f}'), - ('\u{116b9}', '\u{116bf}'), + ('\u{116ba}', '\u{116bf}'), ('\u{116ca}', '\u{116ff}'), ('\u{1171b}', '\u{1171c}'), ('\u{1172c}', '\u{1172f}'), - ('\u{11740}', '\u{117ff}'), + ('\u{11747}', '\u{117ff}'), ('\u{1183c}', '\u{1189f}'), ('\u{118f3}', '\u{118fe}'), ('\u{11907}', '\u{11908}'), @@ -3218,7 +3312,7 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{119d8}', '\u{119d9}'), ('\u{119e5}', '\u{119ff}'), ('\u{11a48}', '\u{11a4f}'), - ('\u{11aa3}', '\u{11abf}'), + ('\u{11aa3}', '\u{11aaf}'), ('\u{11af9}', '\u{11bff}'), ('\u{11c09}', '\u{11c09}'), ('\u{11c37}', '\u{11c37}'), @@ -3246,13 +3340,15 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{1239a}', '\u{123ff}'), ('\u{1246f}', '\u{1246f}'), ('\u{12475}', '\u{1247f}'), - ('\u{12544}', '\u{12fff}'), + ('\u{12544}', '\u{12f8f}'), + ('\u{12ff3}', '\u{12fff}'), ('\u{1342f}', '\u{143ff}'), ('\u{14647}', '\u{167ff}'), ('\u{16a39}', '\u{16a3f}'), ('\u{16a5f}', '\u{16a5f}'), ('\u{16a6a}', '\u{16a6d}'), - ('\u{16a70}', '\u{16acf}'), + ('\u{16abf}', '\u{16abf}'), + ('\u{16aca}', '\u{16acf}'), ('\u{16aee}', '\u{16aef}'), ('\u{16af6}', '\u{16aff}'), ('\u{16b46}', '\u{16b4f}'), @@ -3268,8 +3364,11 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{16ff2}', '\u{16fff}'), ('\u{187f8}', '\u{187ff}'), ('\u{18cd6}', '\u{18cff}'), - ('\u{18d09}', '\u{1afff}'), - ('\u{1b11f}', '\u{1b14f}'), + ('\u{18d09}', '\u{1afef}'), + ('\u{1aff4}', '\u{1aff4}'), + ('\u{1affc}', '\u{1affc}'), + ('\u{1afff}', '\u{1afff}'), + ('\u{1b123}', '\u{1b14f}'), ('\u{1b153}', '\u{1b163}'), ('\u{1b168}', '\u{1b16f}'), ('\u{1b2fc}', '\u{1bbff}'), @@ -3277,11 +3376,14 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{1bc7d}', '\u{1bc7f}'), ('\u{1bc89}', '\u{1bc8f}'), ('\u{1bc9a}', '\u{1bc9b}'), - ('\u{1bca0}', '\u{1cfff}'), + ('\u{1bca0}', '\u{1ceff}'), + ('\u{1cf2e}', '\u{1cf2f}'), + ('\u{1cf47}', '\u{1cf4f}'), + ('\u{1cfc4}', '\u{1cfff}'), ('\u{1d0f6}', '\u{1d0ff}'), ('\u{1d127}', '\u{1d128}'), ('\u{1d173}', '\u{1d17a}'), - ('\u{1d1e9}', '\u{1d1ff}'), + ('\u{1d1eb}', '\u{1d1ff}'), ('\u{1d246}', '\u{1d2df}'), ('\u{1d2f4}', '\u{1d2ff}'), ('\u{1d357}', '\u{1d35f}'), @@ -3308,7 +3410,8 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{1d7cc}', '\u{1d7cd}'), ('\u{1da8c}', '\u{1da9a}'), ('\u{1daa0}', '\u{1daa0}'), - ('\u{1dab0}', '\u{1dfff}'), + ('\u{1dab0}', '\u{1deff}'), + ('\u{1df1f}', '\u{1dfff}'), ('\u{1e007}', '\u{1e007}'), ('\u{1e019}', '\u{1e01a}'), ('\u{1e022}', '\u{1e022}'), @@ -3317,9 +3420,14 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{1e12d}', '\u{1e12f}'), ('\u{1e13e}', '\u{1e13f}'), ('\u{1e14a}', '\u{1e14d}'), - ('\u{1e150}', '\u{1e2bf}'), + ('\u{1e150}', '\u{1e28f}'), + ('\u{1e2af}', '\u{1e2bf}'), ('\u{1e2fa}', '\u{1e2fe}'), - ('\u{1e300}', '\u{1e7ff}'), + ('\u{1e300}', '\u{1e7df}'), + ('\u{1e7e7}', '\u{1e7e7}'), + ('\u{1e7ec}', '\u{1e7ec}'), + ('\u{1e7ef}', '\u{1e7ef}'), + ('\u{1e7ff}', '\u{1e7ff}'), ('\u{1e8c5}', '\u{1e8c6}'), ('\u{1e8d7}', '\u{1e8ff}'), ('\u{1e94c}', '\u{1e94f}'), @@ -3373,34 +3481,35 @@ pub const OTHER: &'static [(char, char)] = &[ ('\u{1f249}', '\u{1f24f}'), ('\u{1f252}', '\u{1f25f}'), ('\u{1f266}', '\u{1f2ff}'), - ('\u{1f6d8}', '\u{1f6df}'), + ('\u{1f6d8}', '\u{1f6dc}'), ('\u{1f6ed}', '\u{1f6ef}'), ('\u{1f6fd}', '\u{1f6ff}'), ('\u{1f774}', '\u{1f77f}'), ('\u{1f7d9}', '\u{1f7df}'), - ('\u{1f7ec}', '\u{1f7ff}'), + ('\u{1f7ec}', '\u{1f7ef}'), + ('\u{1f7f1}', '\u{1f7ff}'), ('\u{1f80c}', '\u{1f80f}'), ('\u{1f848}', '\u{1f84f}'), ('\u{1f85a}', '\u{1f85f}'), ('\u{1f888}', '\u{1f88f}'), ('\u{1f8ae}', '\u{1f8af}'), ('\u{1f8b2}', '\u{1f8ff}'), - ('\u{1f979}', '\u{1f979}'), - ('\u{1f9cc}', '\u{1f9cc}'), ('\u{1fa54}', '\u{1fa5f}'), ('\u{1fa6e}', '\u{1fa6f}'), ('\u{1fa75}', '\u{1fa77}'), - ('\u{1fa7b}', '\u{1fa7f}'), + ('\u{1fa7d}', '\u{1fa7f}'), ('\u{1fa87}', '\u{1fa8f}'), - ('\u{1faa9}', '\u{1faaf}'), - ('\u{1fab7}', '\u{1fabf}'), - ('\u{1fac3}', '\u{1facf}'), - ('\u{1fad7}', '\u{1faff}'), + ('\u{1faad}', '\u{1faaf}'), + ('\u{1fabb}', '\u{1fabf}'), + ('\u{1fac6}', '\u{1facf}'), + ('\u{1fada}', '\u{1fadf}'), + ('\u{1fae8}', '\u{1faef}'), + ('\u{1faf7}', '\u{1faff}'), ('\u{1fb93}', '\u{1fb93}'), ('\u{1fbcb}', '\u{1fbef}'), ('\u{1fbfa}', '\u{1ffff}'), - ('\u{2a6de}', '\u{2a6ff}'), - ('\u{2b735}', '\u{2b73f}'), + ('\u{2a6e0}', '\u{2a6ff}'), + ('\u{2b739}', '\u{2b73f}'), ('\u{2b81e}', '\u{2b81f}'), ('\u{2cea2}', '\u{2ceaf}'), ('\u{2ebe1}', '\u{2f7ff}'), @@ -3433,8 +3542,9 @@ pub const OTHER_LETTER: &'static [(char, char)] = &[ ('à €', 'à •'), ('à¡€', 'ࡘ'), ('à¡ ', 'ࡪ'), - ('ࢠ', 'ࢴ'), - ('ࢶ', '\u{8c7}'), + ('à¡°', 'ࢇ'), + ('ࢉ', 'ࢎ'), + ('ࢠ', 'ࣈ'), ('ऄ', 'ह'), ('ऽ', 'ऽ'), ('à¥', 'à¥'), @@ -3499,6 +3609,7 @@ pub const OTHER_LETTER: &'static [(char, char)] = &[ ('à°ª', 'à°¹'), ('à°½', 'à°½'), ('ౘ', 'ౚ'), + ('à±', 'à±'), ('à± ', 'ౡ'), ('ಀ', 'ಀ'), ('ಅ', 'ಌ'), @@ -3507,10 +3618,10 @@ pub const OTHER_LETTER: &'static [(char, char)] = &[ ('ಪ', 'ಳ'), ('ವ', 'ಹ'), ('ಽ', 'ಽ'), - ('ೞ', 'ೞ'), + ('à³', 'ೞ'), ('à³ ', 'ೡ'), ('à³±', 'à³²'), - ('\u{d04}', 'à´Œ'), + ('à´„', 'à´Œ'), ('à´Ž', 'à´'), ('à´’', 'à´º'), ('à´½', 'à´½'), @@ -3571,9 +3682,8 @@ pub const OTHER_LETTER: &'static [(char, char)] = &[ ('áš', 'ášš'), ('áš ', 'ᛪ'), ('á›±', 'ᛸ'), - ('ᜀ', 'ᜌ'), - ('ᜎ', 'ᜑ'), - ('ᜠ', 'ᜱ'), + ('ᜀ', 'ᜑ'), + ('ᜟ', 'ᜱ'), ('á€', 'á‘'), ('á ', 'á¬'), ('á®', 'á°'), @@ -3593,7 +3703,7 @@ pub const OTHER_LETTER: &'static [(char, char)] = &[ ('ᨀ', 'ᨖ'), ('ᨠ', 'á©”'), ('ᬅ', 'ᬳ'), - ('á…', 'á‹'), + ('á…', 'áŒ'), ('ᮃ', 'á® '), ('á®®', 'ᮯ'), ('ᮺ', 'ᯥ'), @@ -3623,11 +3733,10 @@ pub const OTHER_LETTER: &'static [(char, char)] = &[ ('ヿ', 'ヿ'), ('ã„…', 'ㄯ'), ('ㄱ', 'ㆎ'), - ('ㆠ', '\u{31bf}'), + ('ㆠ', 'ㆿ'), ('ㇰ', 'ㇿ'), - ('ã€', '\u{4dbf}'), - ('一', '\u{9ffc}'), - ('ꀀ', 'ꀔ'), + ('ã€', '䶿'), + ('一', 'ꀔ'), ('ꀖ', 'ê’Œ'), ('ê“', 'ê“·'), ('ꔀ', 'ꘋ'), @@ -3750,19 +3859,22 @@ pub const OTHER_LETTER: &'static [(char, char)] = &[ ('ð®€', 'ð®‘'), ('ð°€', 'ð±ˆ'), ('ð´€', 'ð´£'), - ('\u{10e80}', '\u{10ea9}'), - ('\u{10eb0}', '\u{10eb1}'), + ('ðº€', 'ðº©'), + ('ðº°', 'ðº±'), ('ð¼€', 'ð¼œ'), ('ð¼§', 'ð¼§'), ('ð¼°', 'ð½…'), - ('\u{10fb0}', '\u{10fc4}'), + ('ð½°', 'ð¾'), + ('ð¾°', 'ð¿„'), ('ð¿ ', 'ð¿¶'), ('𑀃', 'ð‘€·'), + ('ð‘±', 'ð‘²'), + ('ð‘µ', 'ð‘µ'), ('𑂃', '𑂯'), ('ð‘ƒ', '𑃨'), ('𑄃', '𑄦'), ('ð‘…„', 'ð‘…„'), - ('\u{11147}', '\u{11147}'), + ('ð‘…‡', 'ð‘…‡'), ('ð‘…', 'ð‘…²'), ('ð‘…¶', 'ð‘…¶'), ('𑆃', '𑆲'), @@ -3788,7 +3900,7 @@ pub const OTHER_LETTER: &'static [(char, char)] = &[ ('ð‘', 'ð‘¡'), ('ð‘€', 'ð‘´'), ('𑑇', 'ð‘‘Š'), - ('ð‘‘Ÿ', '\u{11461}'), + ('ð‘‘Ÿ', 'ð‘‘¡'), ('ð‘’€', 'ð‘’¯'), ('ð‘“„', 'ð‘“…'), ('𑓇', '𑓇'), @@ -3799,14 +3911,15 @@ pub const OTHER_LETTER: &'static [(char, char)] = &[ ('𑚀', '𑚪'), ('𑚸', '𑚸'), ('𑜀', '𑜚'), + ('ð‘€', 'ð‘†'), ('ð‘ €', 'ð‘ «'), - ('𑣿', '\u{11906}'), - ('\u{11909}', '\u{11909}'), - ('\u{1190c}', '\u{11913}'), - ('\u{11915}', '\u{11916}'), - ('\u{11918}', '\u{1192f}'), - ('\u{1193f}', '\u{1193f}'), - ('\u{11941}', '\u{11941}'), + ('𑣿', '𑤆'), + ('𑤉', '𑤉'), + ('𑤌', '𑤓'), + ('𑤕', '𑤖'), + ('𑤘', '𑤯'), + ('𑤿', '𑤿'), + ('ð‘¥', 'ð‘¥'), ('𑦠', '𑦧'), ('𑦪', 'ð‘§'), ('𑧡', '𑧡'), @@ -3817,7 +3930,7 @@ pub const OTHER_LETTER: &'static [(char, char)] = &[ ('ð‘©', 'ð‘©'), ('ð‘©œ', '𑪉'), ('ð‘ª', 'ð‘ª'), - ('ð‘«€', '𑫸'), + ('𑪰', '𑫸'), ('ð‘°€', 'ð‘°ˆ'), ('ð‘°Š', 'ð‘°®'), ('𑱀', '𑱀'), @@ -3831,13 +3944,15 @@ pub const OTHER_LETTER: &'static [(char, char)] = &[ ('𑵪', '𑶉'), ('𑶘', '𑶘'), ('ð‘» ', 'ð‘»²'), - ('\u{11fb0}', '\u{11fb0}'), + ('𑾰', '𑾰'), ('ð’€€', 'ð’Ž™'), ('ð’’€', '𒕃'), + ('ð’¾', 'ð’¿°'), ('ð“€€', 'ð“®'), ('ð”€', '𔙆'), ('ð– €', '𖨸'), ('ð–©€', 'ð–©ž'), + ('ð–©°', '𖪾'), ('ð–«', 'ð–«'), ('𖬀', '𖬯'), ('ð–£', 'ð–·'), @@ -3845,9 +3960,9 @@ pub const OTHER_LETTER: &'static [(char, char)] = &[ ('ð–¼€', '𖽊'), ('ð–½', 'ð–½'), ('ð—€€', '𘟷'), - ('𘠀', '\u{18cd5}'), - ('\u{18d00}', '\u{18d08}'), - ('𛀀', '𛄞'), + ('𘠀', '𘳕'), + ('𘴀', '𘴈'), + ('𛀀', '𛄢'), ('ð›…', 'ð›…’'), ('ð›…¤', 'ð›…§'), ('ð›…°', '𛋻'), @@ -3855,9 +3970,15 @@ pub const OTHER_LETTER: &'static [(char, char)] = &[ ('ð›±°', 'ð›±¼'), ('𛲀', '𛲈'), ('ð›²', '𛲙'), + ('ð¼Š', 'ð¼Š'), ('ðž„€', '𞄬'), ('ðž…Ž', 'ðž…Ž'), + ('ðžŠ', 'ðžŠ'), ('ðž‹€', 'ðž‹«'), + ('𞟠', '𞟦'), + ('𞟨', '𞟫'), + ('ðžŸ', '𞟮'), + ('𞟰', '𞟾'), ('ðž €', '𞣄'), ('𞸀', '𞸃'), ('𞸅', '𞸟'), @@ -3892,13 +4013,13 @@ pub const OTHER_LETTER: &'static [(char, char)] = &[ ('𞺡', '𞺣'), ('𞺥', '𞺩'), ('𞺫', '𞺻'), - ('ð €€', '\u{2a6dd}'), - ('𪜀', '𫜴'), + ('ð €€', '𪛟'), + ('𪜀', '𫜸'), ('ð«€', 'ð« '), ('ð« ', '𬺡'), ('𬺰', '𮯠'), ('丽', 'ð¯¨'), - ('\u{30000}', '\u{3134a}'), + ('ð°€€', 'ð±Š'), ]; pub const OTHER_NUMBER: &'static [(char, char)] = &[ @@ -3955,7 +4076,7 @@ pub const OTHER_NUMBER: &'static [(char, char)] = &[ ('ð¹ ', 'ð¹¾'), ('ð¼', 'ð¼¦'), ('ð½‘', 'ð½”'), - ('\u{10fc5}', '\u{10fcb}'), + ('ð¿…', 'ð¿‹'), ('ð‘’', 'ð‘¥'), ('𑇡', '𑇴'), ('𑜺', '𑜻'), @@ -3999,7 +4120,7 @@ pub const OTHER_PUNCTUATION: &'static [(char, char)] = &[ ('؉', 'ØŠ'), ('ØŒ', 'Ø'), ('Ø›', 'Ø›'), - ('Øž', 'ØŸ'), + ('Ø', 'ØŸ'), ('Ùª', 'Ù'), ('Û”', 'Û”'), ('Ü€', 'Ü'), @@ -4036,6 +4157,7 @@ pub const OTHER_PUNCTUATION: &'static [(char, char)] = &[ ('᪠', '᪦'), ('᪨', 'áª'), ('áš', 'á '), + ('á½', 'á¾'), ('᯼', '᯿'), ('á°»', 'á°¿'), ('á±¾', '᱿'), @@ -4064,7 +4186,7 @@ pub const OTHER_PUNCTUATION: &'static [(char, char)] = &[ ('⸼', '⸿'), ('â¹', 'â¹'), ('⹃', 'â¹'), - ('\u{2e52}', '\u{2e52}'), + ('â¹’', 'â¹”'), ('ã€', '〃'), ('〽', '〽'), ('・', '・'), @@ -4118,6 +4240,7 @@ pub const OTHER_PUNCTUATION: &'static [(char, char)] = &[ ('ð¬¹', 'ð¬¿'), ('ð®™', 'ð®œ'), ('ð½•', 'ð½™'), + ('ð¾†', 'ð¾‰'), ('ð‘‡', 'ð‘'), ('ð‘‚»', 'ð‘‚¼'), ('ð‘‚¾', 'ð‘ƒ'), @@ -4130,15 +4253,16 @@ pub const OTHER_PUNCTUATION: &'static [(char, char)] = &[ ('𑈸', '𑈽'), ('ð‘Š©', 'ð‘Š©'), ('ð‘‘‹', 'ð‘‘'), - ('\u{1145a}', 'ð‘‘›'), + ('ð‘‘š', 'ð‘‘›'), ('ð‘‘', 'ð‘‘'), ('𑓆', '𑓆'), ('ð‘—', 'ð‘——'), ('ð‘™', '𑙃'), ('ð‘™ ', '𑙬'), + ('ð‘š¹', 'ð‘š¹'), ('𑜼', '𑜾'), ('ð‘ »', 'ð‘ »'), - ('\u{11944}', '\u{11946}'), + ('𑥄', '𑥆'), ('𑧢', '𑧢'), ('𑨿', '𑩆'), ('𑪚', '𑪜'), @@ -4148,6 +4272,7 @@ pub const OTHER_PUNCTUATION: &'static [(char, char)] = &[ ('ð‘»·', '𑻸'), ('ð‘¿¿', 'ð‘¿¿'), ('ð’‘°', 'ð’‘´'), + ('ð’¿±', 'ð’¿²'), ('ð–©®', '𖩯'), ('ð–«µ', 'ð–«µ'), ('ð–¬·', '𖬻'), @@ -4240,9 +4365,9 @@ pub const OTHER_SYMBOL: &'static [(char, char)] = &[ ('â…', 'â†'), ('â', 'â³'), ('â¶', '⮕'), - ('\u{2b97}', '⯿'), + ('â®—', '⯿'), ('â³¥', '⳪'), - ('\u{2e50}', '\u{2e51}'), + ('â¹', '⹑'), ('⺀', '⺙'), ('⺛', '⻳'), ('â¼€', 'â¿•'), @@ -4267,7 +4392,9 @@ pub const OTHER_SYMBOL: &'static [(char, char)] = &[ ('ê ¶', 'ê ·'), ('ê ¹', 'ê ¹'), ('ê©·', '꩹'), - ('ï·½', 'ï·½'), + ('ïµ€', 'ïµ'), + ('ï·', 'ï·'), + ('ï·½', 'ï·¿'), ('¦', '¦'), ('│', '│'), ('ï¿', 'ï¿®'), @@ -4275,7 +4402,7 @@ pub const OTHER_SYMBOL: &'static [(char, char)] = &[ ('ð„·', 'ð„¿'), ('ð…¹', 'ð†‰'), ('ð†Œ', 'ð†Ž'), - ('ð†', '\u{1019c}'), + ('ð†', 'ð†œ'), ('ð† ', 'ð† '), ('ð‡', 'ð‡¼'), ('ð¡·', 'ð¡¸'), @@ -4286,13 +4413,14 @@ pub const OTHER_SYMBOL: &'static [(char, char)] = &[ ('𖬼', '𖬿'), ('ð–…', 'ð–…'), ('𛲜', '𛲜'), + ('ðœ½', '𜿃'), ('ð€€', 'ðƒµ'), ('ð„€', 'ð„¦'), ('ð„©', 'ð…¤'), ('ð…ª', 'ð…¬'), ('ð†ƒ', 'ð†„'), ('ð†Œ', 'ð†©'), - ('ð†®', 'ð‡¨'), + ('ð†®', 'ð‡ª'), ('ðˆ€', 'ð‰'), ('ð‰…', 'ð‰…'), ('ðŒ€', 'ð–'), @@ -4310,38 +4438,39 @@ pub const OTHER_SYMBOL: &'static [(char, char)] = &[ ('🂱', '🂿'), ('ðŸƒ', 'ðŸƒ'), ('🃑', '🃵'), - ('\u{1f10d}', '\u{1f1ad}'), + ('ðŸ„', 'ðŸ†'), ('🇦', '🈂'), ('ðŸˆ', '🈻'), ('🉀', '🉈'), ('ðŸ‰', '🉑'), ('🉠', '🉥'), ('🌀', 'ðŸº'), - ('ðŸ€', '\u{1f6d7}'), - ('🛠', '🛬'), - ('🛰', '\u{1f6fc}'), + ('ðŸ€', '🛗'), + ('ðŸ›', '🛬'), + ('🛰', '🛼'), ('🜀', 'ðŸ³'), ('🞀', '🟘'), ('🟠', '🟫'), + ('🟰', '🟰'), ('🠀', '🠋'), ('ðŸ ', '🡇'), ('ðŸ¡', '🡙'), ('🡠', '🢇'), ('ðŸ¢', 'ðŸ¢'), - ('\u{1f8b0}', '\u{1f8b1}'), - ('🤀', '\u{1f978}'), - ('🥺', '\u{1f9cb}'), - ('ðŸ§', '🩓'), + ('🢰', '🢱'), + ('🤀', '🩓'), ('🩠', 'ðŸ©'), - ('🩰', '\u{1fa74}'), - ('🩸', '🩺'), - ('🪀', '\u{1fa86}'), - ('ðŸª', '\u{1faa8}'), - ('\u{1fab0}', '\u{1fab6}'), - ('\u{1fac0}', '\u{1fac2}'), - ('\u{1fad0}', '\u{1fad6}'), - ('\u{1fb00}', '\u{1fb92}'), - ('\u{1fb94}', '\u{1fbca}'), + ('🩰', '🩴'), + ('🩸', '🩼'), + ('🪀', '🪆'), + ('ðŸª', '🪬'), + ('🪰', '🪺'), + ('🫀', '🫅'), + ('ðŸ«', '🫙'), + ('🫠', '🫧'), + ('🫰', '🫶'), + ('🬀', '🮒'), + ('🮔', '🯊'), ]; pub const PARAGRAPH_SEPARATOR: &'static [(char, char)] = @@ -4381,7 +4510,7 @@ pub const PUNCTUATION: &'static [(char, char)] = &[ ('؉', 'ØŠ'), ('ØŒ', 'Ø'), ('Ø›', 'Ø›'), - ('Øž', 'ØŸ'), + ('Ø', 'ØŸ'), ('Ùª', 'Ù'), ('Û”', 'Û”'), ('Ü€', 'Ü'), @@ -4420,6 +4549,7 @@ pub const PUNCTUATION: &'static [(char, char)] = &[ ('᪠', '᪦'), ('᪨', 'áª'), ('áš', 'á '), + ('á½', 'á¾'), ('᯼', '᯿'), ('á°»', 'á°¿'), ('á±¾', '᱿'), @@ -4444,7 +4574,7 @@ pub const PUNCTUATION: &'static [(char, char)] = &[ ('âµ°', 'âµ°'), ('⸀', '⸮'), ('⸰', 'â¹'), - ('\u{2e52}', '\u{2e52}'), + ('â¹’', 'â¹'), ('ã€', '〃'), ('〈', '】'), ('〔', '〟'), @@ -4498,8 +4628,9 @@ pub const PUNCTUATION: &'static [(char, char)] = &[ ('ð«°', 'ð«¶'), ('ð¬¹', 'ð¬¿'), ('ð®™', 'ð®œ'), - ('\u{10ead}', '\u{10ead}'), + ('ðº', 'ðº'), ('ð½•', 'ð½™'), + ('ð¾†', 'ð¾‰'), ('ð‘‡', 'ð‘'), ('ð‘‚»', 'ð‘‚¼'), ('ð‘‚¾', 'ð‘ƒ'), @@ -4512,15 +4643,16 @@ pub const PUNCTUATION: &'static [(char, char)] = &[ ('𑈸', '𑈽'), ('ð‘Š©', 'ð‘Š©'), ('ð‘‘‹', 'ð‘‘'), - ('\u{1145a}', 'ð‘‘›'), + ('ð‘‘š', 'ð‘‘›'), ('ð‘‘', 'ð‘‘'), ('𑓆', '𑓆'), ('ð‘—', 'ð‘——'), ('ð‘™', '𑙃'), ('ð‘™ ', '𑙬'), + ('ð‘š¹', 'ð‘š¹'), ('𑜼', '𑜾'), ('ð‘ »', 'ð‘ »'), - ('\u{11944}', '\u{11946}'), + ('𑥄', '𑥆'), ('𑧢', '𑧢'), ('𑨿', '𑩆'), ('𑪚', '𑪜'), @@ -4530,6 +4662,7 @@ pub const PUNCTUATION: &'static [(char, char)] = &[ ('ð‘»·', '𑻸'), ('ð‘¿¿', 'ð‘¿¿'), ('ð’‘°', 'ð’‘´'), + ('ð’¿±', 'ð’¿²'), ('ð–©®', '𖩯'), ('ð–«µ', 'ð–«µ'), ('ð–¬·', '𖬻'), @@ -4620,6 +4753,8 @@ pub const SPACING_MARK: &'static [(char, char)] = &[ ('ႇ', 'á‚Œ'), ('á‚', 'á‚'), ('á‚š', 'á‚œ'), + ('᜕', '᜕'), + ('᜴', '᜴'), ('ា', 'ា'), ('áž¾', 'ៅ'), ('ះ', 'ៈ'), @@ -4682,7 +4817,7 @@ pub const SPACING_MARK: &'static [(char, char)] = &[ ('𑆂', '𑆂'), ('𑆳', '𑆵'), ('𑆿', '𑇀'), - ('\u{111ce}', '\u{111ce}'), + ('𑇎', '𑇎'), ('𑈬', '𑈮'), ('𑈲', '𑈳'), ('𑈵', '𑈵'), @@ -4714,11 +4849,11 @@ pub const SPACING_MARK: &'static [(char, char)] = &[ ('𑜦', '𑜦'), ('ð‘ ¬', 'ð‘ ®'), ('ð‘ ¸', 'ð‘ ¸'), - ('\u{11930}', '\u{11935}'), - ('\u{11937}', '\u{11938}'), - ('\u{1193d}', '\u{1193d}'), - ('\u{11940}', '\u{11940}'), - ('\u{11942}', '\u{11942}'), + ('\u{11930}', '𑤵'), + ('𑤷', '𑤸'), + ('𑤽', '𑤽'), + ('ð‘¥€', 'ð‘¥€'), + ('𑥂', '𑥂'), ('𑧑', '𑧓'), ('𑧜', '𑧟'), ('𑧤', '𑧤'), @@ -4735,7 +4870,7 @@ pub const SPACING_MARK: &'static [(char, char)] = &[ ('𑶖', '𑶖'), ('ð‘»µ', '𑻶'), ('𖽑', '𖾇'), - ('\u{16ff0}', '\u{16ff1}'), + ('ð–¿°', 'ð–¿±'), ('\u{1d165}', 'ð…¦'), ('ð…', '\u{1d172}'), ]; @@ -4774,6 +4909,7 @@ pub const SYMBOL: &'static [(char, char)] = &[ ('Û½', 'Û¾'), ('߶', '߶'), ('ß¾', 'ß¿'), + ('࢈', '࢈'), ('৲', '৳'), ('৺', '৻'), ('૱', '૱'), @@ -4812,7 +4948,7 @@ pub const SYMBOL: &'static [(char, char)] = &[ ('â’', 'â’'), ('âº', 'â¼'), ('â‚Š', 'â‚Œ'), - ('â‚ ', 'â‚¿'), + ('â‚ ', '⃀'), ('â„€', 'â„'), ('℃', '℆'), ('℈', '℉'), @@ -4841,9 +4977,9 @@ pub const SYMBOL: &'static [(char, char)] = &[ ('⧜', '⧻'), ('⧾', 'â³'), ('â¶', '⮕'), - ('\u{2b97}', '⯿'), + ('â®—', '⯿'), ('â³¥', '⳪'), - ('\u{2e50}', '\u{2e51}'), + ('â¹', '⹑'), ('⺀', '⺙'), ('⺛', '⻳'), ('â¼€', 'â¿•'), @@ -4872,10 +5008,12 @@ pub const SYMBOL: &'static [(char, char)] = &[ ('ê ¶', 'ê ¹'), ('ê©·', '꩹'), ('ê›', 'ê›'), - ('\u{ab6a}', '\u{ab6b}'), + ('êª', 'ê«'), ('﬩', '﬩'), - ('﮲', 'ï¯'), - ('ï·¼', 'ï·½'), + ('﮲', '﯂'), + ('ïµ€', 'ïµ'), + ('ï·', 'ï·'), + ('ï·¼', 'ï·¿'), ('ï¹¢', 'ï¹¢'), ('﹤', '﹦'), ('﹩', '﹩'), @@ -4892,7 +5030,7 @@ pub const SYMBOL: &'static [(char, char)] = &[ ('ð„·', 'ð„¿'), ('ð…¹', 'ð†‰'), ('ð†Œ', 'ð†Ž'), - ('ð†', '\u{1019c}'), + ('ð†', 'ð†œ'), ('ð† ', 'ð† '), ('ð‡', 'ð‡¼'), ('ð¡·', 'ð¡¸'), @@ -4902,13 +5040,14 @@ pub const SYMBOL: &'static [(char, char)] = &[ ('𖬼', '𖬿'), ('ð–…', 'ð–…'), ('𛲜', '𛲜'), + ('ðœ½', '𜿃'), ('ð€€', 'ðƒµ'), ('ð„€', 'ð„¦'), ('ð„©', 'ð…¤'), ('ð…ª', 'ð…¬'), ('ð†ƒ', 'ð†„'), ('ð†Œ', 'ð†©'), - ('ð†®', 'ð‡¨'), + ('ð†®', 'ð‡ª'), ('ðˆ€', 'ð‰'), ('ð‰…', 'ð‰…'), ('ðŒ€', 'ð–'), @@ -4939,37 +5078,38 @@ pub const SYMBOL: &'static [(char, char)] = &[ ('🂱', '🂿'), ('ðŸƒ', 'ðŸƒ'), ('🃑', '🃵'), - ('\u{1f10d}', '\u{1f1ad}'), + ('ðŸ„', 'ðŸ†'), ('🇦', '🈂'), ('ðŸˆ', '🈻'), ('🉀', '🉈'), ('ðŸ‰', '🉑'), ('🉠', '🉥'), - ('🌀', '\u{1f6d7}'), - ('🛠', '🛬'), - ('🛰', '\u{1f6fc}'), + ('🌀', '🛗'), + ('ðŸ›', '🛬'), + ('🛰', '🛼'), ('🜀', 'ðŸ³'), ('🞀', '🟘'), ('🟠', '🟫'), + ('🟰', '🟰'), ('🠀', '🠋'), ('ðŸ ', '🡇'), ('ðŸ¡', '🡙'), ('🡠', '🢇'), ('ðŸ¢', 'ðŸ¢'), - ('\u{1f8b0}', '\u{1f8b1}'), - ('🤀', '\u{1f978}'), - ('🥺', '\u{1f9cb}'), - ('ðŸ§', '🩓'), + ('🢰', '🢱'), + ('🤀', '🩓'), ('🩠', 'ðŸ©'), - ('🩰', '\u{1fa74}'), - ('🩸', '🩺'), - ('🪀', '\u{1fa86}'), - ('ðŸª', '\u{1faa8}'), - ('\u{1fab0}', '\u{1fab6}'), - ('\u{1fac0}', '\u{1fac2}'), - ('\u{1fad0}', '\u{1fad6}'), - ('\u{1fb00}', '\u{1fb92}'), - ('\u{1fb94}', '\u{1fbca}'), + ('🩰', '🩴'), + ('🩸', '🩼'), + ('🪀', '🪆'), + ('ðŸª', '🪬'), + ('🪰', '🪺'), + ('🫀', '🫅'), + ('ðŸ«', '🫙'), + ('🫠', '🫧'), + ('🫰', '🫶'), + ('🬀', '🮒'), + ('🮔', '🯊'), ]; pub const TITLECASE_LETTER: &'static [(char, char)] = &[ @@ -4998,7 +5138,6 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{5c8}', '\u{5cf}'), ('\u{5eb}', '\u{5ee}'), ('\u{5f5}', '\u{5ff}'), - ('\u{61d}', '\u{61d}'), ('\u{70e}', '\u{70e}'), ('\u{74b}', '\u{74c}'), ('\u{7b2}', '\u{7bf}'), @@ -5007,9 +5146,9 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{83f}', '\u{83f}'), ('\u{85c}', '\u{85d}'), ('\u{85f}', '\u{85f}'), - ('\u{86b}', '\u{89f}'), - ('\u{8b5}', '\u{8b5}'), - ('\u{8c8}', '\u{8d2}'), + ('\u{86b}', '\u{86f}'), + ('\u{88f}', '\u{88f}'), + ('\u{892}', '\u{897}'), ('\u{984}', '\u{984}'), ('\u{98d}', '\u{98e}'), ('\u{991}', '\u{992}'), @@ -5087,12 +5226,13 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{c0d}', '\u{c0d}'), ('\u{c11}', '\u{c11}'), ('\u{c29}', '\u{c29}'), - ('\u{c3a}', '\u{c3c}'), + ('\u{c3a}', '\u{c3b}'), ('\u{c45}', '\u{c45}'), ('\u{c49}', '\u{c49}'), ('\u{c4e}', '\u{c54}'), ('\u{c57}', '\u{c57}'), - ('\u{c5b}', '\u{c5f}'), + ('\u{c5b}', '\u{c5c}'), + ('\u{c5e}', '\u{c5f}'), ('\u{c64}', '\u{c65}'), ('\u{c70}', '\u{c76}'), ('\u{c8d}', '\u{c8d}'), @@ -5103,7 +5243,7 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{cc5}', '\u{cc5}'), ('\u{cc9}', '\u{cc9}'), ('\u{cce}', '\u{cd4}'), - ('\u{cd7}', '\u{cdd}'), + ('\u{cd7}', '\u{cdc}'), ('\u{cdf}', '\u{cdf}'), ('\u{ce4}', '\u{ce5}'), ('\u{cf0}', '\u{cf0}'), @@ -5171,8 +5311,7 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{13fe}', '\u{13ff}'), ('\u{169d}', '\u{169f}'), ('\u{16f9}', '\u{16ff}'), - ('\u{170d}', '\u{170d}'), - ('\u{1715}', '\u{171f}'), + ('\u{1716}', '\u{171e}'), ('\u{1737}', '\u{173f}'), ('\u{1754}', '\u{175f}'), ('\u{176d}', '\u{176d}'), @@ -5181,7 +5320,6 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{17de}', '\u{17df}'), ('\u{17ea}', '\u{17ef}'), ('\u{17fa}', '\u{17ff}'), - ('\u{180f}', '\u{180f}'), ('\u{181a}', '\u{181f}'), ('\u{1879}', '\u{187f}'), ('\u{18ab}', '\u{18af}'), @@ -5201,9 +5339,9 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{1a8a}', '\u{1a8f}'), ('\u{1a9a}', '\u{1a9f}'), ('\u{1aae}', '\u{1aaf}'), - ('\u{1ac1}', '\u{1aff}'), - ('\u{1b4c}', '\u{1b4f}'), - ('\u{1b7d}', '\u{1b7f}'), + ('\u{1acf}', '\u{1aff}'), + ('\u{1b4d}', '\u{1b4f}'), + ('\u{1b7f}', '\u{1b7f}'), ('\u{1bf4}', '\u{1bfb}'), ('\u{1c38}', '\u{1c3a}'), ('\u{1c4a}', '\u{1c4c}'), @@ -5211,7 +5349,6 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{1cbb}', '\u{1cbc}'), ('\u{1cc8}', '\u{1ccf}'), ('\u{1cfb}', '\u{1cff}'), - ('\u{1dfa}', '\u{1dfa}'), ('\u{1f16}', '\u{1f17}'), ('\u{1f1e}', '\u{1f1f}'), ('\u{1f46}', '\u{1f47}'), @@ -5232,15 +5369,13 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{2072}', '\u{2073}'), ('\u{208f}', '\u{208f}'), ('\u{209d}', '\u{209f}'), - ('\u{20c0}', '\u{20cf}'), + ('\u{20c1}', '\u{20cf}'), ('\u{20f1}', '\u{20ff}'), ('\u{218c}', '\u{218f}'), ('\u{2427}', '\u{243f}'), ('\u{244b}', '\u{245f}'), ('\u{2b74}', '\u{2b75}'), ('\u{2b96}', '\u{2b96}'), - ('\u{2c2f}', '\u{2c2f}'), - ('\u{2c5f}', '\u{2c5f}'), ('\u{2cf4}', '\u{2cf8}'), ('\u{2d26}', '\u{2d26}'), ('\u{2d28}', '\u{2d2c}'), @@ -5256,7 +5391,7 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{2dcf}', '\u{2dcf}'), ('\u{2dd7}', '\u{2dd7}'), ('\u{2ddf}', '\u{2ddf}'), - ('\u{2e53}', '\u{2e7f}'), + ('\u{2e5e}', '\u{2e7f}'), ('\u{2e9a}', '\u{2e9a}'), ('\u{2ef4}', '\u{2eff}'), ('\u{2fd6}', '\u{2fef}'), @@ -5268,13 +5403,14 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{318f}', '\u{318f}'), ('\u{31e4}', '\u{31ef}'), ('\u{321f}', '\u{321f}'), - ('\u{9ffd}', '\u{9fff}'), ('\u{a48d}', '\u{a48f}'), ('\u{a4c7}', '\u{a4cf}'), ('\u{a62c}', '\u{a63f}'), ('\u{a6f8}', '\u{a6ff}'), - ('\u{a7c0}', '\u{a7c1}'), - ('\u{a7cb}', '\u{a7f4}'), + ('\u{a7cb}', '\u{a7cf}'), + ('\u{a7d2}', '\u{a7d2}'), + ('\u{a7d4}', '\u{a7d4}'), + ('\u{a7da}', '\u{a7f1}'), ('\u{a82d}', '\u{a82f}'), ('\u{a83a}', '\u{a83f}'), ('\u{a878}', '\u{a87f}'), @@ -5310,11 +5446,10 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{fb3f}', '\u{fb3f}'), ('\u{fb42}', '\u{fb42}'), ('\u{fb45}', '\u{fb45}'), - ('\u{fbc2}', '\u{fbd2}'), - ('\u{fd40}', '\u{fd4f}'), + ('\u{fbc3}', '\u{fbd2}'), ('\u{fd90}', '\u{fd91}'), - ('\u{fdc8}', '\u{fdef}'), - ('\u{fdfe}', '\u{fdff}'), + ('\u{fdc8}', '\u{fdce}'), + ('\u{fdd0}', '\u{fdef}'), ('\u{fe1a}', '\u{fe1f}'), ('\u{fe53}', '\u{fe53}'), ('\u{fe67}', '\u{fe67}'), @@ -5358,10 +5493,20 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{104fc}', '\u{104ff}'), ('\u{10528}', '\u{1052f}'), ('\u{10564}', '\u{1056e}'), - ('\u{10570}', '\u{105ff}'), + ('\u{1057b}', '\u{1057b}'), + ('\u{1058b}', '\u{1058b}'), + ('\u{10593}', '\u{10593}'), + ('\u{10596}', '\u{10596}'), + ('\u{105a2}', '\u{105a2}'), + ('\u{105b2}', '\u{105b2}'), + ('\u{105ba}', '\u{105ba}'), + ('\u{105bd}', '\u{105ff}'), ('\u{10737}', '\u{1073f}'), ('\u{10756}', '\u{1075f}'), - ('\u{10768}', '\u{107ff}'), + ('\u{10768}', '\u{1077f}'), + ('\u{10786}', '\u{10786}'), + ('\u{107b1}', '\u{107b1}'), + ('\u{107bb}', '\u{107ff}'), ('\u{10806}', '\u{10807}'), ('\u{10809}', '\u{10809}'), ('\u{10836}', '\u{10836}'), @@ -5404,12 +5549,13 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{10eae}', '\u{10eaf}'), ('\u{10eb2}', '\u{10eff}'), ('\u{10f28}', '\u{10f2f}'), - ('\u{10f5a}', '\u{10faf}'), + ('\u{10f5a}', '\u{10f6f}'), + ('\u{10f8a}', '\u{10faf}'), ('\u{10fcc}', '\u{10fdf}'), ('\u{10ff7}', '\u{10fff}'), ('\u{1104e}', '\u{11051}'), - ('\u{11070}', '\u{1107e}'), - ('\u{110c2}', '\u{110cc}'), + ('\u{11076}', '\u{1107e}'), + ('\u{110c3}', '\u{110cc}'), ('\u{110ce}', '\u{110cf}'), ('\u{110e9}', '\u{110ef}'), ('\u{110fa}', '\u{110ff}'), @@ -5451,11 +5597,11 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{11645}', '\u{1164f}'), ('\u{1165a}', '\u{1165f}'), ('\u{1166d}', '\u{1167f}'), - ('\u{116b9}', '\u{116bf}'), + ('\u{116ba}', '\u{116bf}'), ('\u{116ca}', '\u{116ff}'), ('\u{1171b}', '\u{1171c}'), ('\u{1172c}', '\u{1172f}'), - ('\u{11740}', '\u{117ff}'), + ('\u{11747}', '\u{117ff}'), ('\u{1183c}', '\u{1189f}'), ('\u{118f3}', '\u{118fe}'), ('\u{11907}', '\u{11908}'), @@ -5470,7 +5616,7 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{119d8}', '\u{119d9}'), ('\u{119e5}', '\u{119ff}'), ('\u{11a48}', '\u{11a4f}'), - ('\u{11aa3}', '\u{11abf}'), + ('\u{11aa3}', '\u{11aaf}'), ('\u{11af9}', '\u{11bff}'), ('\u{11c09}', '\u{11c09}'), ('\u{11c37}', '\u{11c37}'), @@ -5498,14 +5644,16 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{1239a}', '\u{123ff}'), ('\u{1246f}', '\u{1246f}'), ('\u{12475}', '\u{1247f}'), - ('\u{12544}', '\u{12fff}'), + ('\u{12544}', '\u{12f8f}'), + ('\u{12ff3}', '\u{12fff}'), ('\u{1342f}', '\u{1342f}'), ('\u{13439}', '\u{143ff}'), ('\u{14647}', '\u{167ff}'), ('\u{16a39}', '\u{16a3f}'), ('\u{16a5f}', '\u{16a5f}'), ('\u{16a6a}', '\u{16a6d}'), - ('\u{16a70}', '\u{16acf}'), + ('\u{16abf}', '\u{16abf}'), + ('\u{16aca}', '\u{16acf}'), ('\u{16aee}', '\u{16aef}'), ('\u{16af6}', '\u{16aff}'), ('\u{16b46}', '\u{16b4f}'), @@ -5521,8 +5669,11 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{16ff2}', '\u{16fff}'), ('\u{187f8}', '\u{187ff}'), ('\u{18cd6}', '\u{18cff}'), - ('\u{18d09}', '\u{1afff}'), - ('\u{1b11f}', '\u{1b14f}'), + ('\u{18d09}', '\u{1afef}'), + ('\u{1aff4}', '\u{1aff4}'), + ('\u{1affc}', '\u{1affc}'), + ('\u{1afff}', '\u{1afff}'), + ('\u{1b123}', '\u{1b14f}'), ('\u{1b153}', '\u{1b163}'), ('\u{1b168}', '\u{1b16f}'), ('\u{1b2fc}', '\u{1bbff}'), @@ -5530,10 +5681,13 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{1bc7d}', '\u{1bc7f}'), ('\u{1bc89}', '\u{1bc8f}'), ('\u{1bc9a}', '\u{1bc9b}'), - ('\u{1bca4}', '\u{1cfff}'), + ('\u{1bca4}', '\u{1ceff}'), + ('\u{1cf2e}', '\u{1cf2f}'), + ('\u{1cf47}', '\u{1cf4f}'), + ('\u{1cfc4}', '\u{1cfff}'), ('\u{1d0f6}', '\u{1d0ff}'), ('\u{1d127}', '\u{1d128}'), - ('\u{1d1e9}', '\u{1d1ff}'), + ('\u{1d1eb}', '\u{1d1ff}'), ('\u{1d246}', '\u{1d2df}'), ('\u{1d2f4}', '\u{1d2ff}'), ('\u{1d357}', '\u{1d35f}'), @@ -5560,7 +5714,8 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{1d7cc}', '\u{1d7cd}'), ('\u{1da8c}', '\u{1da9a}'), ('\u{1daa0}', '\u{1daa0}'), - ('\u{1dab0}', '\u{1dfff}'), + ('\u{1dab0}', '\u{1deff}'), + ('\u{1df1f}', '\u{1dfff}'), ('\u{1e007}', '\u{1e007}'), ('\u{1e019}', '\u{1e01a}'), ('\u{1e022}', '\u{1e022}'), @@ -5569,9 +5724,14 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{1e12d}', '\u{1e12f}'), ('\u{1e13e}', '\u{1e13f}'), ('\u{1e14a}', '\u{1e14d}'), - ('\u{1e150}', '\u{1e2bf}'), + ('\u{1e150}', '\u{1e28f}'), + ('\u{1e2af}', '\u{1e2bf}'), ('\u{1e2fa}', '\u{1e2fe}'), - ('\u{1e300}', '\u{1e7ff}'), + ('\u{1e300}', '\u{1e7df}'), + ('\u{1e7e7}', '\u{1e7e7}'), + ('\u{1e7ec}', '\u{1e7ec}'), + ('\u{1e7ef}', '\u{1e7ef}'), + ('\u{1e7ff}', '\u{1e7ff}'), ('\u{1e8c5}', '\u{1e8c6}'), ('\u{1e8d7}', '\u{1e8ff}'), ('\u{1e94c}', '\u{1e94f}'), @@ -5625,34 +5785,35 @@ pub const UNASSIGNED: &'static [(char, char)] = &[ ('\u{1f249}', '\u{1f24f}'), ('\u{1f252}', '\u{1f25f}'), ('\u{1f266}', '\u{1f2ff}'), - ('\u{1f6d8}', '\u{1f6df}'), + ('\u{1f6d8}', '\u{1f6dc}'), ('\u{1f6ed}', '\u{1f6ef}'), ('\u{1f6fd}', '\u{1f6ff}'), ('\u{1f774}', '\u{1f77f}'), ('\u{1f7d9}', '\u{1f7df}'), - ('\u{1f7ec}', '\u{1f7ff}'), + ('\u{1f7ec}', '\u{1f7ef}'), + ('\u{1f7f1}', '\u{1f7ff}'), ('\u{1f80c}', '\u{1f80f}'), ('\u{1f848}', '\u{1f84f}'), ('\u{1f85a}', '\u{1f85f}'), ('\u{1f888}', '\u{1f88f}'), ('\u{1f8ae}', '\u{1f8af}'), ('\u{1f8b2}', '\u{1f8ff}'), - ('\u{1f979}', '\u{1f979}'), - ('\u{1f9cc}', '\u{1f9cc}'), ('\u{1fa54}', '\u{1fa5f}'), ('\u{1fa6e}', '\u{1fa6f}'), ('\u{1fa75}', '\u{1fa77}'), - ('\u{1fa7b}', '\u{1fa7f}'), + ('\u{1fa7d}', '\u{1fa7f}'), ('\u{1fa87}', '\u{1fa8f}'), - ('\u{1faa9}', '\u{1faaf}'), - ('\u{1fab7}', '\u{1fabf}'), - ('\u{1fac3}', '\u{1facf}'), - ('\u{1fad7}', '\u{1faff}'), + ('\u{1faad}', '\u{1faaf}'), + ('\u{1fabb}', '\u{1fabf}'), + ('\u{1fac6}', '\u{1facf}'), + ('\u{1fada}', '\u{1fadf}'), + ('\u{1fae8}', '\u{1faef}'), + ('\u{1faf7}', '\u{1faff}'), ('\u{1fb93}', '\u{1fb93}'), ('\u{1fbcb}', '\u{1fbef}'), ('\u{1fbfa}', '\u{1ffff}'), - ('\u{2a6de}', '\u{2a6ff}'), - ('\u{2b735}', '\u{2b73f}'), + ('\u{2a6e0}', '\u{2a6ff}'), + ('\u{2b739}', '\u{2b73f}'), ('\u{2b81e}', '\u{2b81f}'), ('\u{2cea2}', '\u{2ceaf}'), ('\u{2ebe1}', '\u{2f7ff}'), @@ -6097,7 +6258,7 @@ pub const UPPERCASE_LETTER: &'static [(char, char)] = &[ ('ℾ', 'â„¿'), ('â……', 'â……'), ('Ↄ', 'Ↄ'), - ('â°€', 'â°®'), + ('â°€', 'â°¯'), ('â± ', 'â± '), ('â±¢', 'Ɽ'), ('Ⱨ', 'Ⱨ'), @@ -6262,13 +6423,21 @@ pub const UPPERCASE_LETTER: &'static [(char, char)] = &[ ('Ꞻ', 'Ꞻ'), ('êž¼', 'êž¼'), ('êž¾', 'êž¾'), + ('Ꟁ', 'Ꟁ'), ('Ꟃ', 'Ꟃ'), - ('Ꞔ', '\u{a7c7}'), - ('\u{a7c9}', '\u{a7c9}'), - ('\u{a7f5}', '\u{a7f5}'), + ('Ꞔ', 'Ꟈ'), + ('Ꟊ', 'Ꟊ'), + ('êŸ', 'êŸ'), + ('Ꟗ', 'Ꟗ'), + ('Ꟙ', 'Ꟙ'), + ('Ꟶ', 'Ꟶ'), ('A', 'Z'), ('ð€', 'ð§'), ('ð’°', 'ð““'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), ('ð²€', 'ð²²'), ('ð‘¢ ', '𑢿'), ('ð–¹€', '𖹟'), diff --git a/vendor/regex-syntax/src/unicode_tables/grapheme_cluster_break.rs b/vendor/regex-syntax/src/unicode_tables/grapheme_cluster_break.rs index 7df9d2b93..38cfc73af 100644 --- a/vendor/regex-syntax/src/unicode_tables/grapheme_cluster_break.rs +++ b/vendor/regex-syntax/src/unicode_tables/grapheme_cluster_break.rs @@ -1,10 +1,10 @@ // DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY: // -// ucd-generate grapheme-cluster-break ucd-13.0.0 --chars +// ucd-generate grapheme-cluster-break /tmp/ucd --chars // -// Unicode version: 13.0.0. +// Unicode version: 14.0.0. // -// ucd-generate 0.2.8 is available on crates.io. +// ucd-generate 0.2.11 is available on crates.io. pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("CR", CR), @@ -25,7 +25,7 @@ pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ pub const CR: &'static [(char, char)] = &[('\r', '\r')]; pub const CONTROL: &'static [(char, char)] = &[ - ('\u{0}', '\t'), + ('\0', '\t'), ('\u{b}', '\u{c}'), ('\u{e}', '\u{1f}'), ('\u{7f}', '\u{9f}'), @@ -71,7 +71,8 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{825}', '\u{827}'), ('\u{829}', '\u{82d}'), ('\u{859}', '\u{85b}'), - ('\u{8d3}', '\u{8e1}'), + ('\u{898}', '\u{89f}'), + ('\u{8ca}', '\u{8e1}'), ('\u{8e3}', '\u{902}'), ('\u{93a}', '\u{93a}'), ('\u{93c}', '\u{93c}'), @@ -116,6 +117,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{bd7}', '\u{bd7}'), ('\u{c00}', '\u{c00}'), ('\u{c04}', '\u{c04}'), + ('\u{c3c}', '\u{c3c}'), ('\u{c3e}', '\u{c40}'), ('\u{c46}', '\u{c48}'), ('\u{c4a}', '\u{c4d}'), @@ -171,7 +173,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{109d}', '\u{109d}'), ('\u{135d}', '\u{135f}'), ('\u{1712}', '\u{1714}'), - ('\u{1732}', '\u{1734}'), + ('\u{1732}', '\u{1733}'), ('\u{1752}', '\u{1753}'), ('\u{1772}', '\u{1773}'), ('\u{17b4}', '\u{17b5}'), @@ -180,6 +182,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{17c9}', '\u{17d3}'), ('\u{17dd}', '\u{17dd}'), ('\u{180b}', '\u{180d}'), + ('\u{180f}', '\u{180f}'), ('\u{1885}', '\u{1886}'), ('\u{18a9}', '\u{18a9}'), ('\u{1920}', '\u{1922}'), @@ -195,7 +198,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{1a65}', '\u{1a6c}'), ('\u{1a73}', '\u{1a7c}'), ('\u{1a7f}', '\u{1a7f}'), - ('\u{1ab0}', '\u{1ac0}'), + ('\u{1ab0}', '\u{1ace}'), ('\u{1b00}', '\u{1b03}'), ('\u{1b34}', '\u{1b3a}'), ('\u{1b3c}', '\u{1b3c}'), @@ -217,8 +220,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{1ced}', '\u{1ced}'), ('\u{1cf4}', '\u{1cf4}'), ('\u{1cf8}', '\u{1cf9}'), - ('\u{1dc0}', '\u{1df9}'), - ('\u{1dfb}', '\u{1dff}'), + ('\u{1dc0}', '\u{1dff}'), ('\u{200c}', '\u{200c}'), ('\u{20d0}', '\u{20f0}'), ('\u{2cef}', '\u{2cf1}'), @@ -277,11 +279,15 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{10d24}', '\u{10d27}'), ('\u{10eab}', '\u{10eac}'), ('\u{10f46}', '\u{10f50}'), + ('\u{10f82}', '\u{10f85}'), ('\u{11001}', '\u{11001}'), ('\u{11038}', '\u{11046}'), + ('\u{11070}', '\u{11070}'), + ('\u{11073}', '\u{11074}'), ('\u{1107f}', '\u{11081}'), ('\u{110b3}', '\u{110b6}'), ('\u{110b9}', '\u{110ba}'), + ('\u{110c2}', '\u{110c2}'), ('\u{11100}', '\u{11102}'), ('\u{11127}', '\u{1112b}'), ('\u{1112d}', '\u{11134}'), @@ -367,6 +373,8 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{16f8f}', '\u{16f92}'), ('\u{16fe4}', '\u{16fe4}'), ('\u{1bc9d}', '\u{1bc9e}'), + ('\u{1cf00}', '\u{1cf2d}'), + ('\u{1cf30}', '\u{1cf46}'), ('\u{1d165}', '\u{1d165}'), ('\u{1d167}', '\u{1d169}'), ('\u{1d16e}', '\u{1d172}'), @@ -386,6 +394,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{1e023}', '\u{1e024}'), ('\u{1e026}', '\u{1e02a}'), ('\u{1e130}', '\u{1e136}'), + ('\u{1e2ae}', '\u{1e2ae}'), ('\u{1e2ec}', '\u{1e2ef}'), ('\u{1e8d0}', '\u{1e8d6}'), ('\u{1e944}', '\u{1e94a}'), @@ -1206,13 +1215,14 @@ pub const PREPEND: &'static [(char, char)] = &[ ('\u{600}', '\u{605}'), ('\u{6dd}', '\u{6dd}'), ('\u{70f}', '\u{70f}'), + ('\u{890}', '\u{891}'), ('\u{8e2}', '\u{8e2}'), ('ൎ', 'ൎ'), ('\u{110bd}', '\u{110bd}'), ('\u{110cd}', '\u{110cd}'), ('𑇂', '𑇃'), - ('\u{1193f}', '\u{1193f}'), - ('\u{11941}', '\u{11941}'), + ('𑤿', '𑤿'), + ('ð‘¥', 'ð‘¥'), ('𑨺', '𑨺'), ('𑪄', '𑪉'), ('𑵆', '𑵆'), @@ -1268,6 +1278,8 @@ pub const SPACINGMARK: &'static [(char, char)] = &[ ('ျ', 'ြ'), ('á–', 'á—'), ('á‚„', 'á‚„'), + ('᜕', '᜕'), + ('᜴', '᜴'), ('ា', 'ា'), ('áž¾', 'ៅ'), ('ះ', 'ៈ'), @@ -1324,7 +1336,7 @@ pub const SPACINGMARK: &'static [(char, char)] = &[ ('𑆂', '𑆂'), ('𑆳', '𑆵'), ('𑆿', '𑇀'), - ('\u{111ce}', '\u{111ce}'), + ('𑇎', '𑇎'), ('𑈬', '𑈮'), ('𑈲', '𑈳'), ('𑈵', '𑈵'), @@ -1352,15 +1364,14 @@ pub const SPACINGMARK: &'static [(char, char)] = &[ ('𑚬', '𑚬'), ('ð‘š®', '𑚯'), ('𑚶', '𑚶'), - ('𑜠', '𑜡'), ('𑜦', '𑜦'), ('ð‘ ¬', 'ð‘ ®'), ('ð‘ ¸', 'ð‘ ¸'), - ('\u{11931}', '\u{11935}'), - ('\u{11937}', '\u{11938}'), - ('\u{1193d}', '\u{1193d}'), - ('\u{11940}', '\u{11940}'), - ('\u{11942}', '\u{11942}'), + ('𑤱', '𑤵'), + ('𑤷', '𑤸'), + ('𑤽', '𑤽'), + ('ð‘¥€', 'ð‘¥€'), + ('𑥂', '𑥂'), ('𑧑', '𑧓'), ('𑧜', '𑧟'), ('𑧤', '𑧤'), @@ -1377,7 +1388,7 @@ pub const SPACINGMARK: &'static [(char, char)] = &[ ('𑶖', '𑶖'), ('ð‘»µ', '𑻶'), ('𖽑', '𖾇'), - ('\u{16ff0}', '\u{16ff1}'), + ('ð–¿°', 'ð–¿±'), ('ð…¦', 'ð…¦'), ('ð…', 'ð…'), ]; diff --git a/vendor/regex-syntax/src/unicode_tables/perl_decimal.rs b/vendor/regex-syntax/src/unicode_tables/perl_decimal.rs index 2a09259fc..9a14e4395 100644 --- a/vendor/regex-syntax/src/unicode_tables/perl_decimal.rs +++ b/vendor/regex-syntax/src/unicode_tables/perl_decimal.rs @@ -1,10 +1,10 @@ // DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY: // -// ucd-generate general-category ucd-13.0.0 --chars --include decimalnumber +// ucd-generate general-category /tmp/ucd --chars --include decimalnumber // -// Unicode version: 13.0.0. +// Unicode version: 14.0.0. // -// ucd-generate 0.2.8 is available on crates.io. +// ucd-generate 0.2.11 is available on crates.io. pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[("Decimal_Number", DECIMAL_NUMBER)]; @@ -60,15 +60,16 @@ pub const DECIMAL_NUMBER: &'static [(char, char)] = &[ ('𑛀', '𑛉'), ('𑜰', '𑜹'), ('ð‘£ ', '𑣩'), - ('\u{11950}', '\u{11959}'), + ('ð‘¥', 'ð‘¥™'), ('ð‘±', '𑱙'), ('ð‘µ', '𑵙'), ('𑶠', '𑶩'), ('ð–© ', 'ð–©©'), + ('ð–«€', '𖫉'), ('ð–', 'ð–™'), ('ðŸŽ', 'ðŸ¿'), ('ðž…€', 'ðž…‰'), ('ðž‹°', '𞋹'), ('ðž¥', '𞥙'), - ('\u{1fbf0}', '\u{1fbf9}'), + ('🯰', '🯹'), ]; diff --git a/vendor/regex-syntax/src/unicode_tables/perl_space.rs b/vendor/regex-syntax/src/unicode_tables/perl_space.rs index c112dd126..bb69ce1ba 100644 --- a/vendor/regex-syntax/src/unicode_tables/perl_space.rs +++ b/vendor/regex-syntax/src/unicode_tables/perl_space.rs @@ -1,10 +1,10 @@ // DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY: // -// ucd-generate property-bool ucd-13.0.0 --chars --include whitespace +// ucd-generate property-bool /tmp/ucd --chars --include whitespace // -// Unicode version: 13.0.0. +// Unicode version: 14.0.0. // -// ucd-generate 0.2.8 is available on crates.io. +// ucd-generate 0.2.11 is available on crates.io. pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[("White_Space", WHITE_SPACE)]; diff --git a/vendor/regex-syntax/src/unicode_tables/perl_word.rs b/vendor/regex-syntax/src/unicode_tables/perl_word.rs index df9eac7d7..2c8171b2b 100644 --- a/vendor/regex-syntax/src/unicode_tables/perl_word.rs +++ b/vendor/regex-syntax/src/unicode_tables/perl_word.rs @@ -1,10 +1,10 @@ // DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY: // -// ucd-generate perl-word ucd-13.0.0 --chars +// ucd-generate perl-word /tmp/ucd --chars // -// Unicode version: 13.0.0. +// Unicode version: 14.0.0. // -// ucd-generate 0.2.8 is available on crates.io. +// ucd-generate 0.2.11 is available on crates.io. pub const PERL_WORD: &'static [(char, char)] = &[ ('0', '9'), @@ -57,9 +57,9 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('à €', '\u{82d}'), ('à¡€', '\u{85b}'), ('à¡ ', 'ࡪ'), - ('ࢠ', 'ࢴ'), - ('ࢶ', '\u{8c7}'), - ('\u{8d3}', '\u{8e1}'), + ('à¡°', 'ࢇ'), + ('ࢉ', 'ࢎ'), + ('\u{898}', '\u{8e1}'), ('\u{8e3}', '\u{963}'), ('०', '९'), ('ॱ', 'ঃ'), @@ -143,11 +143,12 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('à°Ž', 'à°'), ('à°’', 'à°¨'), ('à°ª', 'à°¹'), - ('à°½', 'ౄ'), + ('\u{c3c}', 'ౄ'), ('\u{c46}', '\u{c48}'), ('\u{c4a}', '\u{c4d}'), ('\u{c55}', '\u{c56}'), ('ౘ', 'ౚ'), + ('à±', 'à±'), ('à± ', '\u{c63}'), ('౦', '౯'), ('ಀ', 'ಃ'), @@ -160,7 +161,7 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('\u{cc6}', 'ೈ'), ('ೊ', '\u{ccd}'), ('\u{cd5}', '\u{cd6}'), - ('ೞ', 'ೞ'), + ('à³', 'ೞ'), ('à³ ', '\u{ce3}'), ('೦', '೯'), ('à³±', 'à³²'), @@ -242,9 +243,8 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('áš', 'ášš'), ('áš ', 'ᛪ'), ('á›®', 'ᛸ'), - ('ᜀ', 'ᜌ'), - ('ᜎ', '\u{1714}'), - ('ᜠ', '\u{1734}'), + ('ᜀ', '᜕'), + ('ᜟ', '᜴'), ('á€', '\u{1753}'), ('á ', 'á¬'), ('á®', 'á°'), @@ -254,7 +254,7 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('ៜ', '\u{17dd}'), ('០', '៩'), ('\u{180b}', '\u{180d}'), - ('á ', 'á ™'), + ('\u{180f}', 'á ™'), ('á ', 'ᡸ'), ('ᢀ', 'ᢪ'), ('ᢰ', 'ᣵ'), @@ -272,8 +272,8 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('\u{1a7f}', '᪉'), ('áª', '᪙'), ('ᪧ', 'ᪧ'), - ('\u{1ab0}', '\u{1ac0}'), - ('\u{1b00}', 'á‹'), + ('\u{1ab0}', '\u{1ace}'), + ('\u{1b00}', 'áŒ'), ('á', 'á™'), ('\u{1b6b}', '\u{1b73}'), ('\u{1b80}', '᯳'), @@ -285,8 +285,7 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('á²½', 'Ჿ'), ('\u{1cd0}', '\u{1cd2}'), ('\u{1cd4}', 'ᳺ'), - ('á´€', '\u{1df9}'), - ('\u{1dfb}', 'ἕ'), + ('á´€', 'ἕ'), ('Ἐ', 'á¼'), ('á¼ ', 'á½…'), ('Ὀ', 'á½'), @@ -327,9 +326,7 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('â…Ž', 'â…Ž'), ('â… ', 'ↈ'), ('â’¶', 'â“©'), - ('â°€', 'â°®'), - ('â°°', 'ⱞ'), - ('â± ', 'ⳤ'), + ('â°€', 'ⳤ'), ('Ⳬ', 'â³³'), ('â´€', 'â´¥'), ('â´§', 'â´§'), @@ -358,11 +355,10 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('ー', 'ヿ'), ('ã„…', 'ㄯ'), ('ㄱ', 'ㆎ'), - ('ㆠ', '\u{31bf}'), + ('ㆠ', 'ㆿ'), ('ㇰ', 'ㇿ'), - ('ã€', '\u{4dbf}'), - ('一', '\u{9ffc}'), - ('ꀀ', 'ê’Œ'), + ('ã€', '䶿'), + ('一', 'ê’Œ'), ('ê“', 'ꓽ'), ('ꔀ', 'ꘌ'), ('ê˜', 'ꘫ'), @@ -371,9 +367,11 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('ꙿ', '\u{a6f1}'), ('ꜗ', 'ꜟ'), ('Ꜣ', 'ꞈ'), - ('êž‹', 'êž¿'), - ('Ꟃ', '\u{a7ca}'), - ('\u{a7f5}', 'ê §'), + ('êž‹', 'ꟊ'), + ('êŸ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟙ'), + ('ꟲ', 'ê §'), ('\u{a82c}', '\u{a82c}'), ('ê¡€', 'ꡳ'), ('ꢀ', '\u{a8c5}'), @@ -400,7 +398,7 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('ê¬ ', 'ꬦ'), ('ꬨ', 'ꬮ'), ('ꬰ', 'êš'), - ('êœ', '\u{ab69}'), + ('êœ', 'ê©'), ('ê°', 'ꯪ'), ('꯬', '\u{abed}'), ('꯰', '꯹'), @@ -462,9 +460,20 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('ð“˜', 'ð“»'), ('ð”€', 'ð”§'), ('ð”°', 'ð•£'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), ('ð˜€', 'ðœ¶'), ('ð€', 'ð•'), ('ð ', 'ð§'), + ('ðž€', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ('ð €', 'ð …'), ('ð ˆ', 'ð ˆ'), ('ð Š', 'ð µ'), @@ -499,27 +508,29 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('ð³€', 'ð³²'), ('ð´€', '\u{10d27}'), ('ð´°', 'ð´¹'), - ('\u{10e80}', '\u{10ea9}'), + ('ðº€', 'ðº©'), ('\u{10eab}', '\u{10eac}'), - ('\u{10eb0}', '\u{10eb1}'), + ('ðº°', 'ðº±'), ('ð¼€', 'ð¼œ'), ('ð¼§', 'ð¼§'), ('ð¼°', '\u{10f50}'), - ('\u{10fb0}', '\u{10fc4}'), + ('ð½°', '\u{10f85}'), + ('ð¾°', 'ð¿„'), ('ð¿ ', 'ð¿¶'), ('ð‘€€', '\u{11046}'), - ('ð‘¦', 'ð‘¯'), + ('ð‘¦', 'ð‘µ'), ('\u{1107f}', '\u{110ba}'), + ('\u{110c2}', '\u{110c2}'), ('ð‘ƒ', '𑃨'), ('𑃰', '𑃹'), ('\u{11100}', '\u{11134}'), ('𑄶', 'ð‘„¿'), - ('ð‘…„', '\u{11147}'), + ('ð‘…„', 'ð‘…‡'), ('ð‘…', '\u{11173}'), ('ð‘…¶', 'ð‘…¶'), ('\u{11180}', '𑇄'), ('\u{111c9}', '\u{111cc}'), - ('\u{111ce}', '𑇚'), + ('𑇎', '𑇚'), ('𑇜', '𑇜'), ('𑈀', '𑈑'), ('𑈓', '\u{11237}'), @@ -548,7 +559,7 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('\u{11370}', '\u{11374}'), ('ð‘€', 'ð‘‘Š'), ('ð‘‘', 'ð‘‘™'), - ('\u{1145e}', '\u{11461}'), + ('\u{1145e}', 'ð‘‘¡'), ('ð‘’€', 'ð‘“…'), ('𑓇', '𑓇'), ('ð‘“', 'ð‘“™'), @@ -563,16 +574,17 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('𑜀', '𑜚'), ('\u{1171d}', '\u{1172b}'), ('𑜰', '𑜹'), + ('ð‘€', 'ð‘†'), ('ð‘ €', '\u{1183a}'), ('ð‘¢ ', '𑣩'), - ('𑣿', '\u{11906}'), - ('\u{11909}', '\u{11909}'), - ('\u{1190c}', '\u{11913}'), - ('\u{11915}', '\u{11916}'), - ('\u{11918}', '\u{11935}'), - ('\u{11937}', '\u{11938}'), + ('𑣿', '𑤆'), + ('𑤉', '𑤉'), + ('𑤌', '𑤓'), + ('𑤕', '𑤖'), + ('𑤘', '𑤵'), + ('𑤷', '𑤸'), ('\u{1193b}', '\u{11943}'), - ('\u{11950}', '\u{11959}'), + ('ð‘¥', 'ð‘¥™'), ('𑦠', '𑦧'), ('𑦪', '\u{119d7}'), ('\u{119da}', '𑧡'), @@ -581,7 +593,7 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('\u{11a47}', '\u{11a47}'), ('ð‘©', '\u{11a99}'), ('ð‘ª', 'ð‘ª'), - ('ð‘«€', '𑫸'), + ('𑪰', '𑫸'), ('ð‘°€', 'ð‘°ˆ'), ('ð‘°Š', '\u{11c36}'), ('\u{11c38}', '𑱀'), @@ -603,15 +615,18 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('𑶓', '𑶘'), ('𑶠', '𑶩'), ('ð‘» ', '𑻶'), - ('\u{11fb0}', '\u{11fb0}'), + ('𑾰', '𑾰'), ('ð’€€', 'ð’Ž™'), ('ð’€', 'ð’‘®'), ('ð’’€', '𒕃'), + ('ð’¾', 'ð’¿°'), ('ð“€€', 'ð“®'), ('ð”€', '𔙆'), ('ð– €', '𖨸'), ('ð–©€', 'ð–©ž'), ('ð–© ', 'ð–©©'), + ('ð–©°', '𖪾'), + ('ð–«€', '𖫉'), ('ð–«', 'ð–«'), ('\u{16af0}', '\u{16af4}'), ('𖬀', '\u{16b36}'), @@ -625,11 +640,14 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('\u{16f8f}', '𖾟'), ('ð–¿ ', 'ð–¿¡'), ('ð–¿£', '\u{16fe4}'), - ('\u{16ff0}', '\u{16ff1}'), + ('ð–¿°', 'ð–¿±'), ('ð—€€', '𘟷'), - ('𘠀', '\u{18cd5}'), - ('\u{18d00}', '\u{18d08}'), - ('𛀀', '𛄞'), + ('𘠀', '𘳕'), + ('𘴀', '𘴈'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), + ('𛀀', '𛄢'), ('ð›…', 'ð›…’'), ('ð›…¤', 'ð›…§'), ('ð›…°', '𛋻'), @@ -638,6 +656,8 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('𛲀', '𛲈'), ('ð›²', '𛲙'), ('\u{1bc9d}', '\u{1bc9e}'), + ('\u{1cf00}', '\u{1cf2d}'), + ('\u{1cf30}', '\u{1cf46}'), ('\u{1d165}', '\u{1d169}'), ('ð…', '\u{1d172}'), ('\u{1d17b}', '\u{1d182}'), @@ -681,6 +701,7 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('\u{1da84}', '\u{1da84}'), ('\u{1da9b}', '\u{1da9f}'), ('\u{1daa1}', '\u{1daaf}'), + ('ð¼€', 'ð¼ž'), ('\u{1e000}', '\u{1e006}'), ('\u{1e008}', '\u{1e018}'), ('\u{1e01b}', '\u{1e021}'), @@ -690,7 +711,12 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('\u{1e130}', '𞄽'), ('ðž…€', 'ðž…‰'), ('ðž…Ž', 'ðž…Ž'), + ('ðžŠ', '\u{1e2ae}'), ('ðž‹€', '𞋹'), + ('𞟠', '𞟦'), + ('𞟨', '𞟫'), + ('ðžŸ', '𞟮'), + ('𞟰', '𞟾'), ('ðž €', '𞣄'), ('\u{1e8d0}', '\u{1e8d6}'), ('𞤀', '𞥋'), @@ -731,13 +757,13 @@ pub const PERL_WORD: &'static [(char, char)] = &[ ('🄰', '🅉'), ('ðŸ…', '🅩'), ('🅰', '🆉'), - ('\u{1fbf0}', '\u{1fbf9}'), - ('ð €€', '\u{2a6dd}'), - ('𪜀', '𫜴'), + ('🯰', '🯹'), + ('ð €€', '𪛟'), + ('𪜀', '𫜸'), ('ð«€', 'ð« '), ('ð« ', '𬺡'), ('𬺰', '𮯠'), ('丽', 'ð¯¨'), - ('\u{30000}', '\u{3134a}'), + ('ð°€€', 'ð±Š'), ('\u{e0100}', '\u{e01ef}'), ]; diff --git a/vendor/regex-syntax/src/unicode_tables/property_bool.rs b/vendor/regex-syntax/src/unicode_tables/property_bool.rs index 21cbaf9ae..8fb211030 100644 --- a/vendor/regex-syntax/src/unicode_tables/property_bool.rs +++ b/vendor/regex-syntax/src/unicode_tables/property_bool.rs @@ -1,10 +1,10 @@ // DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY: // -// ucd-generate property-bool ucd-13.0.0 --chars +// ucd-generate property-bool /tmp/ucd --chars // -// Unicode version: 13.0.0. +// Unicode version: 14.0.0. // -// ucd-generate 0.2.8 is available on crates.io. +// ucd-generate 0.2.11 is available on crates.io. pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("ASCII_Hex_Digit", ASCII_HEX_DIGIT), @@ -125,8 +125,9 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('à š', '\u{82c}'), ('à¡€', 'ࡘ'), ('à¡ ', 'ࡪ'), - ('ࢠ', 'ࢴ'), - ('ࢶ', '\u{8c7}'), + ('à¡°', 'ࢇ'), + ('ࢉ', 'ࢎ'), + ('ࢠ', 'ࣉ'), ('\u{8d4}', '\u{8df}'), ('\u{8e3}', '\u{8e9}'), ('\u{8f0}', 'ऻ'), @@ -216,6 +217,7 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('\u{c4a}', '\u{c4c}'), ('\u{c55}', '\u{c56}'), ('ౘ', 'ౚ'), + ('à±', 'à±'), ('à± ', '\u{c63}'), ('ಀ', 'ಃ'), ('ಅ', 'ಌ'), @@ -227,7 +229,7 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('\u{cc6}', 'ೈ'), ('ೊ', '\u{ccc}'), ('\u{cd5}', '\u{cd6}'), - ('ೞ', 'ೞ'), + ('à³', 'ೞ'), ('à³ ', '\u{ce3}'), ('à³±', 'à³²'), ('\u{d00}', 'à´Œ'), @@ -303,9 +305,8 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('áš', 'ášš'), ('áš ', 'ᛪ'), ('á›®', 'ᛸ'), - ('ᜀ', 'ᜌ'), - ('ᜎ', '\u{1713}'), - ('ᜠ', '\u{1733}'), + ('ᜀ', '\u{1713}'), + ('ᜟ', '\u{1733}'), ('á€', '\u{1753}'), ('á ', 'á¬'), ('á®', 'á°'), @@ -329,9 +330,10 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('á©¡', '\u{1a74}'), ('ᪧ', 'ᪧ'), ('\u{1abf}', '\u{1ac0}'), + ('\u{1acc}', '\u{1ace}'), ('\u{1b00}', 'ᬳ'), ('\u{1b35}', 'áƒ'), - ('á…', 'á‹'), + ('á…', 'áŒ'), ('\u{1b80}', '\u{1ba9}'), ('\u{1bac}', 'ᮯ'), ('ᮺ', 'ᯥ'), @@ -385,9 +387,7 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('â…Ž', 'â…Ž'), ('â… ', 'ↈ'), ('â’¶', 'â“©'), - ('â°€', 'â°®'), - ('â°°', 'ⱞ'), - ('â± ', 'ⳤ'), + ('â°€', 'ⳤ'), ('Ⳬ', 'â³®'), ('â³²', 'â³³'), ('â´€', 'â´¥'), @@ -416,11 +416,10 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('ー', 'ヿ'), ('ã„…', 'ㄯ'), ('ㄱ', 'ㆎ'), - ('ㆠ', '\u{31bf}'), + ('ㆠ', 'ㆿ'), ('ㇰ', 'ㇿ'), - ('ã€', '\u{4dbf}'), - ('一', '\u{9ffc}'), - ('ꀀ', 'ê’Œ'), + ('ã€', '䶿'), + ('一', 'ê’Œ'), ('ê“', 'ꓽ'), ('ꔀ', 'ꘌ'), ('ê˜', 'ꘟ'), @@ -430,9 +429,11 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('ꙿ', 'ꛯ'), ('ꜗ', 'ꜟ'), ('Ꜣ', 'ꞈ'), - ('êž‹', 'êž¿'), - ('Ꟃ', '\u{a7ca}'), - ('\u{a7f5}', 'ê …'), + ('êž‹', 'ꟊ'), + ('êŸ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟙ'), + ('ꟲ', 'ê …'), ('ê ‡', 'ê §'), ('ê¡€', 'ꡳ'), ('ꢀ', 'ꣃ'), @@ -463,7 +464,7 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('ê¬ ', 'ꬦ'), ('ꬨ', 'ꬮ'), ('ꬰ', 'êš'), - ('êœ', '\u{ab69}'), + ('êœ', 'ê©'), ('ê°', 'ꯪ'), ('ê°€', '힣'), ('íž°', 'ퟆ'), @@ -514,9 +515,20 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('ð“˜', 'ð“»'), ('ð”€', 'ð”§'), ('ð”°', 'ð•£'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), ('ð˜€', 'ðœ¶'), ('ð€', 'ð•'), ('ð ', 'ð§'), + ('ðž€', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ('ð €', 'ð …'), ('ð ˆ', 'ð ˆ'), ('ð Š', 'ð µ'), @@ -548,24 +560,27 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('ð²€', 'ð²²'), ('ð³€', 'ð³²'), ('ð´€', '\u{10d27}'), - ('\u{10e80}', '\u{10ea9}'), + ('ðº€', 'ðº©'), ('\u{10eab}', '\u{10eac}'), - ('\u{10eb0}', '\u{10eb1}'), + ('ðº°', 'ðº±'), ('ð¼€', 'ð¼œ'), ('ð¼§', 'ð¼§'), ('ð¼°', 'ð½…'), - ('\u{10fb0}', '\u{10fc4}'), + ('ð½°', 'ð¾'), + ('ð¾°', 'ð¿„'), ('ð¿ ', 'ð¿¶'), ('ð‘€€', '\u{11045}'), + ('ð‘±', 'ð‘µ'), ('ð‘‚‚', '𑂸'), + ('\u{110c2}', '\u{110c2}'), ('ð‘ƒ', '𑃨'), ('\u{11100}', '\u{11132}'), - ('ð‘…„', '\u{11147}'), + ('ð‘…„', 'ð‘…‡'), ('ð‘…', 'ð‘…²'), ('ð‘…¶', 'ð‘…¶'), ('\u{11180}', '𑆿'), ('ð‘‡', '𑇄'), - ('\u{111ce}', '\u{111cf}'), + ('𑇎', '\u{111cf}'), ('𑇚', '𑇚'), ('𑇜', '𑇜'), ('𑈀', '𑈑'), @@ -594,7 +609,7 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('ð‘€', 'ð‘‘'), ('\u{11443}', 'ð‘‘…'), ('𑑇', 'ð‘‘Š'), - ('ð‘‘Ÿ', '\u{11461}'), + ('ð‘‘Ÿ', 'ð‘‘¡'), ('ð‘’€', 'ð‘“'), ('ð‘“„', 'ð‘“…'), ('𑓇', '𑓇'), @@ -608,16 +623,17 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('𑚸', '𑚸'), ('𑜀', '𑜚'), ('\u{1171d}', '\u{1172a}'), + ('ð‘€', 'ð‘†'), ('ð‘ €', 'ð‘ ¸'), ('ð‘¢ ', '𑣟'), - ('𑣿', '\u{11906}'), - ('\u{11909}', '\u{11909}'), - ('\u{1190c}', '\u{11913}'), - ('\u{11915}', '\u{11916}'), - ('\u{11918}', '\u{11935}'), - ('\u{11937}', '\u{11938}'), + ('𑣿', '𑤆'), + ('𑤉', '𑤉'), + ('𑤌', '𑤓'), + ('𑤕', '𑤖'), + ('𑤘', '𑤵'), + ('𑤷', '𑤸'), ('\u{1193b}', '\u{1193c}'), - ('\u{1193f}', '\u{11942}'), + ('𑤿', '𑥂'), ('𑦠', '𑦧'), ('𑦪', '\u{119d7}'), ('\u{119da}', '𑧟'), @@ -627,7 +643,7 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('\u{11a35}', '\u{11a3e}'), ('ð‘©', '𑪗'), ('ð‘ª', 'ð‘ª'), - ('ð‘«€', '𑫸'), + ('𑪰', '𑫸'), ('ð‘°€', 'ð‘°ˆ'), ('ð‘°Š', '\u{11c36}'), ('\u{11c38}', 'ð‘°¾'), @@ -650,14 +666,16 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('𑶓', '𑶖'), ('𑶘', '𑶘'), ('ð‘» ', '𑻶'), - ('\u{11fb0}', '\u{11fb0}'), + ('𑾰', '𑾰'), ('ð’€€', 'ð’Ž™'), ('ð’€', 'ð’‘®'), ('ð’’€', '𒕃'), + ('ð’¾', 'ð’¿°'), ('ð“€€', 'ð“®'), ('ð”€', '𔙆'), ('ð– €', '𖨸'), ('ð–©€', 'ð–©ž'), + ('ð–©°', '𖪾'), ('ð–«', 'ð–«'), ('𖬀', '𖬯'), ('ð–€', 'ð–ƒ'), @@ -669,11 +687,14 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('\u{16f8f}', '𖾟'), ('ð–¿ ', 'ð–¿¡'), ('ð–¿£', 'ð–¿£'), - ('\u{16ff0}', '\u{16ff1}'), + ('ð–¿°', 'ð–¿±'), ('ð—€€', '𘟷'), - ('𘠀', '\u{18cd5}'), - ('\u{18d00}', '\u{18d08}'), - ('𛀀', '𛄞'), + ('𘠀', '𘳕'), + ('𘴀', '𘴈'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), + ('𛀀', '𛄢'), ('ð›…', 'ð›…’'), ('ð›…¤', 'ð›…§'), ('ð›…°', '𛋻'), @@ -712,6 +733,7 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('ðžŠ', 'ðž¨'), ('ðžª', 'ðŸ‚'), ('ðŸ„', 'ðŸ‹'), + ('ð¼€', 'ð¼ž'), ('\u{1e000}', '\u{1e006}'), ('\u{1e008}', '\u{1e018}'), ('\u{1e01b}', '\u{1e021}'), @@ -720,7 +742,12 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('ðž„€', '𞄬'), ('ðž„·', '𞄽'), ('ðž…Ž', 'ðž…Ž'), + ('ðžŠ', 'ðžŠ'), ('ðž‹€', 'ðž‹«'), + ('𞟠', '𞟦'), + ('𞟨', '𞟫'), + ('ðžŸ', '𞟮'), + ('𞟰', '𞟾'), ('ðž €', '𞣄'), ('𞤀', '𞥃'), ('\u{1e947}', '\u{1e947}'), @@ -761,13 +788,13 @@ pub const ALPHABETIC: &'static [(char, char)] = &[ ('🄰', '🅉'), ('ðŸ…', '🅩'), ('🅰', '🆉'), - ('ð €€', '\u{2a6dd}'), - ('𪜀', '𫜴'), + ('ð €€', '𪛟'), + ('𪜀', '𫜸'), ('ð«€', 'ð« '), ('ð« ', '𬺡'), ('𬺰', '𮯠'), ('丽', 'ð¯¨'), - ('\u{30000}', '\u{3134a}'), + ('ð°€€', 'ð±Š'), ]; pub const BIDI_CONTROL: &'static [(char, char)] = &[ @@ -873,6 +900,7 @@ pub const BIDI_MIRRORED: &'static [(char, char)] = &[ ('⸌', 'â¸'), ('⸜', 'â¸'), ('⸠', '⸩'), + ('⹕', '⹜'), ('〈', '】'), ('〔', '〛'), ('ï¹™', '﹞'), @@ -936,7 +964,10 @@ pub const CASE_IGNORABLE: &'static [(char, char)] = &[ ('\u{7fd}', '\u{7fd}'), ('\u{816}', '\u{82d}'), ('\u{859}', '\u{85b}'), - ('\u{8d3}', '\u{902}'), + ('࢈', '࢈'), + ('\u{890}', '\u{891}'), + ('\u{898}', '\u{89f}'), + ('ࣉ', '\u{902}'), ('\u{93a}', '\u{93a}'), ('\u{93c}', '\u{93c}'), ('\u{941}', '\u{948}'), @@ -977,6 +1008,7 @@ pub const CASE_IGNORABLE: &'static [(char, char)] = &[ ('\u{bcd}', '\u{bcd}'), ('\u{c00}', '\u{c00}'), ('\u{c04}', '\u{c04}'), + ('\u{c3c}', '\u{c3c}'), ('\u{c3e}', '\u{c40}'), ('\u{c46}', '\u{c48}'), ('\u{c4a}', '\u{c4d}'), @@ -1028,7 +1060,7 @@ pub const CASE_IGNORABLE: &'static [(char, char)] = &[ ('ჼ', 'ჼ'), ('\u{135d}', '\u{135f}'), ('\u{1712}', '\u{1714}'), - ('\u{1732}', '\u{1734}'), + ('\u{1732}', '\u{1733}'), ('\u{1752}', '\u{1753}'), ('\u{1772}', '\u{1773}'), ('\u{17b4}', '\u{17b5}'), @@ -1037,7 +1069,7 @@ pub const CASE_IGNORABLE: &'static [(char, char)] = &[ ('\u{17c9}', '\u{17d3}'), ('ៗ', 'ៗ'), ('\u{17dd}', '\u{17dd}'), - ('\u{180b}', '\u{180e}'), + ('\u{180b}', '\u{180f}'), ('ᡃ', 'ᡃ'), ('\u{1885}', '\u{1886}'), ('\u{18a9}', '\u{18a9}'), @@ -1055,7 +1087,7 @@ pub const CASE_IGNORABLE: &'static [(char, char)] = &[ ('\u{1a73}', '\u{1a7c}'), ('\u{1a7f}', '\u{1a7f}'), ('ᪧ', 'ᪧ'), - ('\u{1ab0}', '\u{1ac0}'), + ('\u{1ab0}', '\u{1ace}'), ('\u{1b00}', '\u{1b03}'), ('\u{1b34}', '\u{1b34}'), ('\u{1b36}', '\u{1b3a}'), @@ -1081,8 +1113,7 @@ pub const CASE_IGNORABLE: &'static [(char, char)] = &[ ('\u{1cf8}', '\u{1cf9}'), ('á´¬', 'ᵪ'), ('ᵸ', 'ᵸ'), - ('ᶛ', '\u{1df9}'), - ('\u{1dfb}', '\u{1dff}'), + ('ᶛ', '\u{1dff}'), ('á¾½', 'á¾½'), ('᾿', 'á¿'), ('á¿', 'á¿'), @@ -1123,6 +1154,7 @@ pub const CASE_IGNORABLE: &'static [(char, char)] = &[ ('꜀', '꜡'), ('ê°', 'ê°'), ('ꞈ', '꞊'), + ('ꟲ', 'ꟴ'), ('ꟸ', 'ꟹ'), ('\u{a802}', '\u{a802}'), ('\u{a806}', '\u{a806}'), @@ -1157,12 +1189,12 @@ pub const CASE_IGNORABLE: &'static [(char, char)] = &[ ('ꫳ', 'ê«´'), ('\u{aaf6}', '\u{aaf6}'), ('ê›', 'êŸ'), - ('\u{ab69}', '\u{ab6b}'), + ('ê©', 'ê«'), ('\u{abe5}', '\u{abe5}'), ('\u{abe8}', '\u{abe8}'), ('\u{abed}', '\u{abed}'), ('\u{fb1e}', '\u{fb1e}'), - ('﮲', 'ï¯'), + ('﮲', '﯂'), ('\u{fe00}', '\u{fe0f}'), ('︓', '︓'), ('\u{fe20}', '\u{fe2f}'), @@ -1181,6 +1213,9 @@ pub const CASE_IGNORABLE: &'static [(char, char)] = &[ ('\u{101fd}', '\u{101fd}'), ('\u{102e0}', '\u{102e0}'), ('\u{10376}', '\u{1037a}'), + ('ðž€', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ('\u{10a01}', '\u{10a03}'), ('\u{10a05}', '\u{10a06}'), ('\u{10a0c}', '\u{10a0f}'), @@ -1190,12 +1225,16 @@ pub const CASE_IGNORABLE: &'static [(char, char)] = &[ ('\u{10d24}', '\u{10d27}'), ('\u{10eab}', '\u{10eac}'), ('\u{10f46}', '\u{10f50}'), + ('\u{10f82}', '\u{10f85}'), ('\u{11001}', '\u{11001}'), ('\u{11038}', '\u{11046}'), + ('\u{11070}', '\u{11070}'), + ('\u{11073}', '\u{11074}'), ('\u{1107f}', '\u{11081}'), ('\u{110b3}', '\u{110b6}'), ('\u{110b9}', '\u{110ba}'), ('\u{110bd}', '\u{110bd}'), + ('\u{110c2}', '\u{110c2}'), ('\u{110cd}', '\u{110cd}'), ('\u{11100}', '\u{11102}'), ('\u{11127}', '\u{1112b}'), @@ -1278,8 +1317,13 @@ pub const CASE_IGNORABLE: &'static [(char, char)] = &[ ('\u{16f8f}', '𖾟'), ('ð–¿ ', 'ð–¿¡'), ('ð–¿£', '\u{16fe4}'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), ('\u{1bc9d}', '\u{1bc9e}'), ('\u{1bca0}', '\u{1bca3}'), + ('\u{1cf00}', '\u{1cf2d}'), + ('\u{1cf30}', '\u{1cf46}'), ('\u{1d167}', '\u{1d169}'), ('\u{1d173}', '\u{1d182}'), ('\u{1d185}', '\u{1d18b}'), @@ -1297,6 +1341,7 @@ pub const CASE_IGNORABLE: &'static [(char, char)] = &[ ('\u{1e023}', '\u{1e024}'), ('\u{1e026}', '\u{1e02a}'), ('\u{1e130}', '𞄽'), + ('\u{1e2ae}', '\u{1e2ae}'), ('\u{1e2ec}', '\u{1e2ef}'), ('\u{1e8d0}', '\u{1e8d6}'), ('\u{1e944}', '𞥋'), @@ -1384,9 +1429,7 @@ pub const CASED: &'static [(char, char)] = &[ ('â… ', 'â…¿'), ('Ↄ', 'ↄ'), ('â’¶', 'â“©'), - ('â°€', 'â°®'), - ('â°°', 'ⱞ'), - ('â± ', 'ⳤ'), + ('â°€', 'ⳤ'), ('Ⳬ', 'â³®'), ('â³²', 'â³³'), ('â´€', 'â´¥'), @@ -1396,12 +1439,14 @@ pub const CASED: &'static [(char, char)] = &[ ('Ꚁ', 'êš'), ('Ꜣ', 'ꞇ'), ('êž‹', 'ꞎ'), - ('êž', 'êž¿'), - ('Ꟃ', '\u{a7ca}'), - ('\u{a7f5}', '\u{a7f6}'), + ('êž', 'ꟊ'), + ('êŸ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟙ'), + ('Ꟶ', 'ꟶ'), ('ꟸ', 'ꟺ'), ('ꬰ', 'êš'), - ('êœ', '\u{ab68}'), + ('êœ', 'ê¨'), ('ê°', 'ꮿ'), ('ff', 'st'), ('ﬓ', 'ﬗ'), @@ -1410,6 +1455,18 @@ pub const CASED: &'static [(char, char)] = &[ ('ð€', 'ð‘'), ('ð’°', 'ð““'), ('ð“˜', 'ð“»'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), + ('ðž€', 'ðž€'), + ('ðžƒ', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ('ð²€', 'ð²²'), ('ð³€', 'ð³²'), ('ð‘¢ ', '𑣟'), @@ -1444,6 +1501,8 @@ pub const CASED: &'static [(char, char)] = &[ ('ðžŠ', 'ðž¨'), ('ðžª', 'ðŸ‚'), ('ðŸ„', 'ðŸ‹'), + ('ð¼€', 'ð¼‰'), + ('ð¼‹', 'ð¼ž'), ('𞤀', '𞥃'), ('🄰', '🅉'), ('ðŸ…', '🅩'), @@ -1886,7 +1945,7 @@ pub const CHANGES_WHEN_CASEFOLDED: &'static [(char, char)] = &[ ('â… ', 'â…¯'), ('Ↄ', 'Ↄ'), ('â’¶', 'â“'), - ('â°€', 'â°®'), + ('â°€', 'â°¯'), ('â± ', 'â± '), ('â±¢', 'Ɽ'), ('Ⱨ', 'Ⱨ'), @@ -2051,16 +2110,24 @@ pub const CHANGES_WHEN_CASEFOLDED: &'static [(char, char)] = &[ ('Ꞻ', 'Ꞻ'), ('êž¼', 'êž¼'), ('êž¾', 'êž¾'), + ('Ꟁ', 'Ꟁ'), ('Ꟃ', 'Ꟃ'), - ('Ꞔ', '\u{a7c7}'), - ('\u{a7c9}', '\u{a7c9}'), - ('\u{a7f5}', '\u{a7f5}'), + ('Ꞔ', 'Ꟈ'), + ('Ꟊ', 'Ꟊ'), + ('êŸ', 'êŸ'), + ('Ꟗ', 'Ꟗ'), + ('Ꟙ', 'Ꟙ'), + ('Ꟶ', 'Ꟶ'), ('ê°', 'ꮿ'), ('ff', 'st'), ('ﬓ', 'ﬗ'), ('A', 'Z'), ('ð€', 'ð§'), ('ð’°', 'ð““'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), ('ð²€', 'ð²²'), ('ð‘¢ ', '𑢿'), ('ð–¹€', '𖹟'), @@ -2156,9 +2223,7 @@ pub const CHANGES_WHEN_CASEMAPPED: &'static [(char, char)] = &[ ('â… ', 'â…¿'), ('Ↄ', 'ↄ'), ('â’¶', 'â“©'), - ('â°€', 'â°®'), - ('â°°', 'ⱞ'), - ('â± ', 'â±°'), + ('â°€', 'â±°'), ('â±²', 'â±³'), ('â±µ', 'ⱶ'), ('â±¾', 'â³£'), @@ -2175,9 +2240,10 @@ pub const CHANGES_WHEN_CASEMAPPED: &'static [(char, char)] = &[ ('êž‹', 'êž'), ('êž', 'êž”'), ('êž–', 'êž®'), - ('êž°', 'êž¿'), - ('Ꟃ', '\u{a7ca}'), - ('\u{a7f5}', '\u{a7f6}'), + ('êž°', 'ꟊ'), + ('êŸ', 'ꟑ'), + ('Ꟗ', 'ꟙ'), + ('Ꟶ', 'ꟶ'), ('ê“', 'ê“'), ('ê°', 'ꮿ'), ('ff', 'st'), @@ -2187,6 +2253,14 @@ pub const CHANGES_WHEN_CASEMAPPED: &'static [(char, char)] = &[ ('ð€', 'ð‘'), ('ð’°', 'ð““'), ('ð“˜', 'ð“»'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), ('ð²€', 'ð²²'), ('ð³€', 'ð³²'), ('ð‘¢ ', '𑣟'), @@ -2620,7 +2694,7 @@ pub const CHANGES_WHEN_LOWERCASED: &'static [(char, char)] = &[ ('â… ', 'â…¯'), ('Ↄ', 'Ↄ'), ('â’¶', 'â“'), - ('â°€', 'â°®'), + ('â°€', 'â°¯'), ('â± ', 'â± '), ('â±¢', 'Ɽ'), ('Ⱨ', 'Ⱨ'), @@ -2785,13 +2859,21 @@ pub const CHANGES_WHEN_LOWERCASED: &'static [(char, char)] = &[ ('Ꞻ', 'Ꞻ'), ('êž¼', 'êž¼'), ('êž¾', 'êž¾'), + ('Ꟁ', 'Ꟁ'), ('Ꟃ', 'Ꟃ'), - ('Ꞔ', '\u{a7c7}'), - ('\u{a7c9}', '\u{a7c9}'), - ('\u{a7f5}', '\u{a7f5}'), + ('Ꞔ', 'Ꟈ'), + ('Ꟊ', 'Ꟊ'), + ('êŸ', 'êŸ'), + ('Ꟗ', 'Ꟗ'), + ('Ꟙ', 'Ꟙ'), + ('Ꟶ', 'Ꟶ'), ('A', 'Z'), ('ð€', 'ð§'), ('ð’°', 'ð““'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), ('ð²€', 'ð²²'), ('ð‘¢ ', '𑢿'), ('ð–¹€', '𖹟'), @@ -3237,7 +3319,7 @@ pub const CHANGES_WHEN_TITLECASED: &'static [(char, char)] = &[ ('â…°', 'â…¿'), ('ↄ', 'ↄ'), ('â“', 'â“©'), - ('â°°', 'ⱞ'), + ('â°°', 'ⱟ'), ('ⱡ', 'ⱡ'), ('â±¥', 'ⱦ'), ('ⱨ', 'ⱨ'), @@ -3402,10 +3484,14 @@ pub const CHANGES_WHEN_TITLECASED: &'static [(char, char)] = &[ ('êž»', 'êž»'), ('êž½', 'êž½'), ('êž¿', 'êž¿'), + ('êŸ', 'êŸ'), ('ꟃ', 'ꟃ'), - ('\u{a7c8}', '\u{a7c8}'), - ('\u{a7ca}', '\u{a7ca}'), - ('\u{a7f6}', '\u{a7f6}'), + ('ꟈ', 'ꟈ'), + ('ꟊ', 'ꟊ'), + ('ꟑ', 'ꟑ'), + ('ꟗ', 'ꟗ'), + ('ꟙ', 'ꟙ'), + ('ꟶ', 'ꟶ'), ('ê“', 'ê“'), ('ê°', 'ꮿ'), ('ff', 'st'), @@ -3413,6 +3499,10 @@ pub const CHANGES_WHEN_TITLECASED: &'static [(char, char)] = &[ ('ï½', 'z'), ('ð¨', 'ð‘'), ('ð“˜', 'ð“»'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), ('ð³€', 'ð³²'), ('ð‘£€', '𑣟'), ('ð–¹ ', '𖹿'), @@ -3859,7 +3949,7 @@ pub const CHANGES_WHEN_UPPERCASED: &'static [(char, char)] = &[ ('â…°', 'â…¿'), ('ↄ', 'ↄ'), ('â“', 'â“©'), - ('â°°', 'ⱞ'), + ('â°°', 'ⱟ'), ('ⱡ', 'ⱡ'), ('â±¥', 'ⱦ'), ('ⱨ', 'ⱨ'), @@ -4024,10 +4114,14 @@ pub const CHANGES_WHEN_UPPERCASED: &'static [(char, char)] = &[ ('êž»', 'êž»'), ('êž½', 'êž½'), ('êž¿', 'êž¿'), + ('êŸ', 'êŸ'), ('ꟃ', 'ꟃ'), - ('\u{a7c8}', '\u{a7c8}'), - ('\u{a7ca}', '\u{a7ca}'), - ('\u{a7f6}', '\u{a7f6}'), + ('ꟈ', 'ꟈ'), + ('ꟊ', 'ꟊ'), + ('ꟑ', 'ꟑ'), + ('ꟗ', 'ꟗ'), + ('ꟙ', 'ꟙ'), + ('ꟶ', 'ꟶ'), ('ê“', 'ê“'), ('ê°', 'ꮿ'), ('ff', 'st'), @@ -4035,6 +4129,10 @@ pub const CHANGES_WHEN_UPPERCASED: &'static [(char, char)] = &[ ('ï½', 'z'), ('ð¨', 'ð‘'), ('ð“˜', 'ð“»'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), ('ð³€', 'ð³²'), ('ð‘£€', '𑣟'), ('ð–¹ ', '𖹿'), @@ -4056,6 +4154,7 @@ pub const DASH: &'static [(char, char)] = &[ ('⸚', '⸚'), ('⸺', '⸻'), ('â¹€', 'â¹€'), + ('â¹', 'â¹'), ('〜', '〜'), ('〰', '〰'), ('ã‚ ', 'ã‚ '), @@ -4063,7 +4162,7 @@ pub const DASH: &'static [(char, char)] = &[ ('﹘', '﹘'), ('ï¹£', 'ï¹£'), ('ï¼', 'ï¼'), - ('\u{10ead}', '\u{10ead}'), + ('ðº', 'ðº'), ]; pub const DEFAULT_IGNORABLE_CODE_POINT: &'static [(char, char)] = &[ @@ -4072,7 +4171,7 @@ pub const DEFAULT_IGNORABLE_CODE_POINT: &'static [(char, char)] = &[ ('\u{61c}', '\u{61c}'), ('á…Ÿ', 'á… '), ('\u{17b4}', '\u{17b5}'), - ('\u{180b}', '\u{180e}'), + ('\u{180b}', '\u{180f}'), ('\u{200b}', '\u{200f}'), ('\u{202a}', '\u{202e}'), ('\u{2060}', '\u{206f}'), @@ -4126,6 +4225,8 @@ pub const DIACRITIC: &'static [(char, char)] = &[ ('\u{7a6}', '\u{7b0}'), ('\u{7eb}', 'ßµ'), ('\u{818}', '\u{819}'), + ('\u{898}', '\u{89f}'), + ('ࣉ', '\u{8d2}'), ('\u{8e3}', '\u{8fe}'), ('\u{93c}', '\u{93c}'), ('\u{94d}', '\u{94d}'), @@ -4142,6 +4243,7 @@ pub const DIACRITIC: &'static [(char, char)] = &[ ('\u{b4d}', '\u{b4d}'), ('\u{b55}', '\u{b55}'), ('\u{bcd}', '\u{bcd}'), + ('\u{c3c}', '\u{c3c}'), ('\u{c4d}', '\u{c4d}'), ('\u{cbc}', '\u{cbc}'), ('\u{ccd}', '\u{ccd}'), @@ -4168,12 +4270,14 @@ pub const DIACRITIC: &'static [(char, char)] = &[ ('á‚', 'á‚'), ('á‚š', 'á‚›'), ('\u{135d}', '\u{135f}'), + ('\u{1714}', '᜕'), ('\u{17c9}', '\u{17d3}'), ('\u{17dd}', '\u{17dd}'), ('\u{1939}', '\u{193b}'), ('\u{1a75}', '\u{1a7c}'), ('\u{1a7f}', '\u{1a7f}'), - ('\u{1ab0}', '\u{1abd}'), + ('\u{1ab0}', '\u{1abe}'), + ('\u{1ac1}', '\u{1acb}'), ('\u{1b34}', '\u{1b34}'), ('á„', 'á„'), ('\u{1b6b}', '\u{1b73}'), @@ -4186,8 +4290,7 @@ pub const DIACRITIC: &'static [(char, char)] = &[ ('á³·', '\u{1cf9}'), ('á´¬', 'ᵪ'), ('\u{1dc4}', '\u{1dcf}'), - ('\u{1df5}', '\u{1df9}'), - ('\u{1dfd}', '\u{1dff}'), + ('\u{1df5}', '\u{1dff}'), ('á¾½', 'á¾½'), ('᾿', 'á¿'), ('á¿', 'á¿'), @@ -4218,7 +4321,7 @@ pub const DIACRITIC: &'static [(char, char)] = &[ ('\u{aabf}', 'ê«‚'), ('\u{aaf6}', '\u{aaf6}'), ('ê›', 'êŸ'), - ('\u{ab69}', '\u{ab6b}'), + ('ê©', 'ê«'), ('꯬', '\u{abed}'), ('\u{fb1e}', '\u{fb1e}'), ('\u{fe20}', '\u{fe2f}'), @@ -4228,9 +4331,15 @@ pub const DIACRITIC: &'static [(char, char)] = &[ ('\u{ff9e}', '\u{ff9f}'), ('ï¿£', 'ï¿£'), ('\u{102e0}', '\u{102e0}'), + ('ðž€', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ('\u{10ae5}', '\u{10ae6}'), ('ð´¢', '\u{10d27}'), ('\u{10f46}', '\u{10f50}'), + ('\u{10f82}', '\u{10f85}'), + ('\u{11046}', '\u{11046}'), + ('\u{11070}', '\u{11070}'), ('\u{110b9}', '\u{110ba}'), ('\u{11133}', '\u{11134}'), ('\u{11173}', '\u{11173}'), @@ -4250,7 +4359,7 @@ pub const DIACRITIC: &'static [(char, char)] = &[ ('𑚶', '\u{116b7}'), ('\u{1172b}', '\u{1172b}'), ('\u{11839}', '\u{1183a}'), - ('\u{1193d}', '\u{1193e}'), + ('𑤽', '\u{1193e}'), ('\u{11943}', '\u{11943}'), ('\u{119e0}', '\u{119e0}'), ('\u{11a34}', '\u{11a34}'), @@ -4263,13 +4372,19 @@ pub const DIACRITIC: &'static [(char, char)] = &[ ('\u{16af0}', '\u{16af4}'), ('\u{16b30}', '\u{16b36}'), ('\u{16f8f}', '𖾟'), - ('\u{16ff0}', '\u{16ff1}'), + ('ð–¿°', 'ð–¿±'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), + ('\u{1cf00}', '\u{1cf2d}'), + ('\u{1cf30}', '\u{1cf46}'), ('\u{1d167}', '\u{1d169}'), ('ð…', '\u{1d172}'), ('\u{1d17b}', '\u{1d182}'), ('\u{1d185}', '\u{1d18b}'), ('\u{1d1aa}', '\u{1d1ad}'), ('\u{1e130}', '\u{1e136}'), + ('\u{1e2ae}', '\u{1e2ae}'), ('\u{1e2ec}', '\u{1e2ef}'), ('\u{1e8d0}', '\u{1e8d6}'), ('\u{1e944}', '\u{1e946}'), @@ -4410,25 +4525,26 @@ pub const EMOJI: &'static [(char, char)] = &[ ('🗺', 'ðŸ™'), ('🚀', '🛅'), ('🛋', '🛒'), - ('🛕', '\u{1f6d7}'), - ('🛠', '🛥'), + ('🛕', '🛗'), + ('ðŸ›', '🛥'), ('🛩', '🛩'), ('🛫', '🛬'), ('🛰', '🛰'), - ('🛳', '\u{1f6fc}'), + ('🛳', '🛼'), ('🟠', '🟫'), - ('\u{1f90c}', '🤺'), + ('🟰', '🟰'), + ('🤌', '🤺'), ('🤼', '🥅'), - ('🥇', '\u{1f978}'), - ('🥺', '\u{1f9cb}'), - ('ðŸ§', '🧿'), - ('🩰', '\u{1fa74}'), - ('🩸', '🩺'), - ('🪀', '\u{1fa86}'), - ('ðŸª', '\u{1faa8}'), - ('\u{1fab0}', '\u{1fab6}'), - ('\u{1fac0}', '\u{1fac2}'), - ('\u{1fad0}', '\u{1fad6}'), + ('🥇', '🧿'), + ('🩰', '🩴'), + ('🩸', '🩼'), + ('🪀', '🪆'), + ('ðŸª', '🪬'), + ('🪰', '🪺'), + ('🫀', '🫅'), + ('ðŸ«', '🫙'), + ('🫠', '🫧'), + ('🫰', '🫶'), ]; pub const EMOJI_COMPONENT: &'static [(char, char)] = &[ @@ -4473,18 +4589,20 @@ pub const EMOJI_MODIFIER_BASE: &'static [(char, char)] = &[ ('🚴', '🚶'), ('🛀', '🛀'), ('🛌', '🛌'), - ('\u{1f90c}', '\u{1f90c}'), + ('🤌', '🤌'), ('ðŸ¤', 'ðŸ¤'), ('🤘', '🤟'), ('🤦', '🤦'), ('🤰', '🤹'), ('🤼', '🤾'), - ('\u{1f977}', '\u{1f977}'), + ('🥷', '🥷'), ('🦵', '🦶'), ('🦸', '🦹'), ('🦻', '🦻'), ('ðŸ§', 'ðŸ§'), ('🧑', 'ðŸ§'), + ('🫃', '🫅'), + ('🫰', '🫶'), ]; pub const EMOJI_PRESENTATION: &'static [(char, char)] = &[ @@ -4553,22 +4671,24 @@ pub const EMOJI_PRESENTATION: &'static [(char, char)] = &[ ('🚀', '🛅'), ('🛌', '🛌'), ('ðŸ›', '🛒'), - ('🛕', '\u{1f6d7}'), + ('🛕', '🛗'), + ('ðŸ›', '🛟'), ('🛫', '🛬'), - ('🛴', '\u{1f6fc}'), + ('🛴', '🛼'), ('🟠', '🟫'), - ('\u{1f90c}', '🤺'), + ('🟰', '🟰'), + ('🤌', '🤺'), ('🤼', '🥅'), - ('🥇', '\u{1f978}'), - ('🥺', '\u{1f9cb}'), - ('ðŸ§', '🧿'), - ('🩰', '\u{1fa74}'), - ('🩸', '🩺'), - ('🪀', '\u{1fa86}'), - ('ðŸª', '\u{1faa8}'), - ('\u{1fab0}', '\u{1fab6}'), - ('\u{1fac0}', '\u{1fac2}'), - ('\u{1fad0}', '\u{1fad6}'), + ('🥇', '🧿'), + ('🩰', '🩴'), + ('🩸', '🩼'), + ('🪀', '🪆'), + ('ðŸª', '🪬'), + ('🪰', '🪺'), + ('🫀', '🫅'), + ('ðŸ«', '🫙'), + ('🫠', '🫧'), + ('🫰', '🫶'), ]; pub const EXTENDED_PICTOGRAPHIC: &'static [(char, char)] = &[ @@ -4623,13 +4743,13 @@ pub const EXTENDED_PICTOGRAPHIC: &'static [(char, char)] = &[ ('㊗', '㊗'), ('㊙', '㊙'), ('🀀', '\u{1f0ff}'), - ('\u{1f10d}', '\u{1f10f}'), + ('ðŸ„', 'ðŸ„'), ('🄯', '🄯'), ('🅬', '🅱'), ('🅾', '🅿'), ('🆎', '🆎'), ('🆑', '🆚'), - ('\u{1f1ad}', '\u{1f1e5}'), + ('ðŸ†', '\u{1f1e5}'), ('ðŸˆ', '\u{1f20f}'), ('🈚', '🈚'), ('🈯', '🈯'), @@ -4646,7 +4766,7 @@ pub const EXTENDED_PICTOGRAPHIC: &'static [(char, char)] = &[ ('\u{1f85a}', '\u{1f85f}'), ('\u{1f888}', '\u{1f88f}'), ('\u{1f8ae}', '\u{1f8ff}'), - ('\u{1f90c}', '🤺'), + ('🤌', '🤺'), ('🤼', '🥅'), ('🥇', '\u{1faff}'), ('\u{1fc00}', '\u{1fffd}'), @@ -4677,6 +4797,7 @@ pub const EXTENDER: &'static [(char, char)] = &[ ('ê«', 'ê«'), ('ꫳ', 'ê«´'), ('ï½°', 'ï½°'), + ('ðž', 'ðž‚'), ('ð‘', 'ð‘'), ('ð‘—†', 'ð‘—ˆ'), ('\u{11a98}', '\u{11a98}'), @@ -4709,7 +4830,7 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('ׯ', '×´'), ('؆', 'Ø'), ('Ø›', 'Ø›'), - ('Øž', 'ÙŠ'), + ('Ø', 'ÙŠ'), ('Ù ', 'Ù¯'), ('Ù±', 'Û•'), ('Ûž', 'Ûž'), @@ -4730,8 +4851,8 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('à¡€', 'ࡘ'), ('à¡ž', 'à¡ž'), ('à¡ ', 'ࡪ'), - ('ࢠ', 'ࢴ'), - ('ࢶ', '\u{8c7}'), + ('à¡°', 'ࢎ'), + ('ࢠ', 'ࣉ'), ('ः', 'ह'), ('ऻ', 'ऻ'), ('ऽ', 'ी'), @@ -4820,6 +4941,7 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('à°½', 'à°½'), ('à±', 'ౄ'), ('ౘ', 'ౚ'), + ('à±', 'à±'), ('à± ', 'ౡ'), ('౦', '౯'), ('à±·', 'ಀ'), @@ -4833,7 +4955,7 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('ೃ', 'ೄ'), ('ೇ', 'ೈ'), ('ೊ', 'ೋ'), - ('ೞ', 'ೞ'), + ('à³', 'ೞ'), ('à³ ', 'ೡ'), ('೦', '೯'), ('à³±', 'à³²'), @@ -4922,10 +5044,10 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('á¸', 'á½'), ('á€', 'ášœ'), ('áš ', 'ᛸ'), - ('ᜀ', 'ᜌ'), - ('ᜎ', 'ᜑ'), - ('ᜠ', 'ᜱ'), - ('᜵', '᜶'), + ('ᜀ', 'ᜑ'), + ('᜕', '᜕'), + ('ᜟ', 'ᜱ'), + ('᜴', '᜶'), ('á€', 'á‘'), ('á ', 'á¬'), ('á®', 'á°'), @@ -4967,9 +5089,9 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('ᬄ', 'ᬳ'), ('ᬻ', 'ᬻ'), ('ᬽ', 'á'), - ('áƒ', 'á‹'), + ('áƒ', 'áŒ'), ('á', 'áª'), - ('á´', 'á¼'), + ('á´', 'á¾'), ('ᮂ', 'ᮡ'), ('ᮦ', 'ᮧ'), ('᮪', '᮪'), @@ -5013,15 +5135,13 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('â°', 'â±'), ('â´', 'â‚Ž'), ('â‚', 'â‚œ'), - ('â‚ ', 'â‚¿'), + ('â‚ ', '⃀'), ('â„€', '↋'), ('â†', 'â¦'), ('â‘€', 'â‘Š'), ('â‘ ', 'â³'), ('â¶', '⮕'), - ('\u{2b97}', 'â°®'), - ('â°°', 'ⱞ'), - ('â± ', 'â³®'), + ('â®—', 'â³®'), ('â³²', 'â³³'), ('â³¹', 'â´¥'), ('â´§', 'â´§'), @@ -5037,7 +5157,7 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('â·ˆ', 'â·Ž'), ('â·', 'â·–'), ('â·˜', 'â·ž'), - ('⸀', '\u{2e52}'), + ('⸀', 'â¹'), ('⺀', '⺙'), ('⺛', '⻳'), ('â¼€', 'â¿•'), @@ -5050,8 +5170,7 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('ㄱ', 'ㆎ'), ('ã†', '㇣'), ('ㇰ', '㈞'), - ('㈠', '\u{9ffc}'), - ('ꀀ', 'ê’Œ'), + ('㈠', 'ê’Œ'), ('ê’', '꓆'), ('ê“', 'ꘫ'), ('Ꙁ', 'ê™®'), @@ -5059,9 +5178,11 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('꙾', 'êš'), ('êš ', 'ꛯ'), ('꛲', 'ê›·'), - ('꜀', 'êž¿'), - ('Ꟃ', '\u{a7ca}'), - ('\u{a7f5}', 'ê '), + ('꜀', 'ꟊ'), + ('êŸ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟙ'), + ('ꟲ', 'ê '), ('ê ƒ', 'ê …'), ('ê ‡', 'ê Š'), ('ê Œ', 'ê ¤'), @@ -5103,7 +5224,7 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('ꬑ', 'ꬖ'), ('ê¬ ', 'ꬦ'), ('ꬨ', 'ꬮ'), - ('ꬰ', '\u{ab6b}'), + ('ꬰ', 'ê«'), ('ê°', 'ꯤ'), ('ꯦ', 'ꯧ'), ('ꯩ', '꯬'), @@ -5121,11 +5242,11 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('מּ', 'מּ'), ('ï€', 'ï'), ('ïƒ', 'ï„'), - ('ï†', 'ï¯'), - ('ﯓ', 'ï´¿'), - ('ïµ', 'ï¶'), + ('ï†', '﯂'), + ('ﯓ', 'ï¶'), ('ﶒ', 'ï·‡'), - ('ï·°', 'ï·½'), + ('ï·', 'ï·'), + ('ï·°', 'ï·¿'), ('ï¸', '︙'), ('︰', 'ï¹’'), ('ï¹”', '﹦'), @@ -5151,7 +5272,7 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('ð„€', 'ð„‚'), ('ð„‡', 'ð„³'), ('ð„·', 'ð†Ž'), - ('ð†', '\u{1019c}'), + ('ð†', 'ð†œ'), ('ð† ', 'ð† '), ('ð‡', 'ð‡¼'), ('ðŠ€', 'ðŠœ'), @@ -5169,10 +5290,20 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('ð“˜', 'ð“»'), ('ð”€', 'ð”§'), ('ð”°', 'ð•£'), - ('ð•¯', 'ð•¯'), + ('ð•¯', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), ('ð˜€', 'ðœ¶'), ('ð€', 'ð•'), ('ð ', 'ð§'), + ('ðž€', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ('ð €', 'ð …'), ('ð ˆ', 'ð ˆ'), ('ð Š', 'ð µ'), @@ -5209,18 +5340,22 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('ð³º', 'ð´£'), ('ð´°', 'ð´¹'), ('ð¹ ', 'ð¹¾'), - ('\u{10e80}', '\u{10ea9}'), - ('\u{10ead}', '\u{10ead}'), - ('\u{10eb0}', '\u{10eb1}'), + ('ðº€', 'ðº©'), + ('ðº', 'ðº'), + ('ðº°', 'ðº±'), ('ð¼€', 'ð¼§'), ('ð¼°', 'ð½…'), ('ð½‘', 'ð½™'), - ('\u{10fb0}', '\u{10fcb}'), + ('ð½°', 'ð¾'), + ('ð¾†', 'ð¾‰'), + ('ð¾°', 'ð¿‹'), ('ð¿ ', 'ð¿¶'), ('ð‘€€', 'ð‘€€'), ('𑀂', 'ð‘€·'), ('ð‘‡', 'ð‘'), ('ð‘’', 'ð‘¯'), + ('ð‘±', 'ð‘²'), + ('ð‘µ', 'ð‘µ'), ('ð‘‚‚', 'ð‘‚²'), ('ð‘‚·', '𑂸'), ('ð‘‚»', 'ð‘‚¼'), @@ -5229,12 +5364,12 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('𑃰', '𑃹'), ('𑄃', '𑄦'), ('𑄬', '𑄬'), - ('𑄶', '\u{11147}'), + ('𑄶', 'ð‘…‡'), ('ð‘…', 'ð‘…²'), ('ð‘…´', 'ð‘…¶'), ('𑆂', '𑆵'), ('𑆿', '𑇈'), - ('ð‘‡', '\u{111ce}'), + ('ð‘‡', '𑇎'), ('ð‘‡', '𑇟'), ('𑇡', '𑇴'), ('𑈀', '𑈑'), @@ -5269,7 +5404,7 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('ð‘‘…', 'ð‘‘…'), ('𑑇', 'ð‘‘›'), ('ð‘‘', 'ð‘‘'), - ('ð‘‘Ÿ', '\u{11461}'), + ('ð‘‘Ÿ', 'ð‘‘¡'), ('ð‘’€', 'ð‘’¯'), ('ð‘’±', 'ð‘’²'), ('ð‘’¹', 'ð‘’¹'), @@ -5293,27 +5428,27 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('𑚬', '𑚬'), ('ð‘š®', '𑚯'), ('𑚶', '𑚶'), - ('𑚸', '𑚸'), + ('𑚸', 'ð‘š¹'), ('𑛀', '𑛉'), ('𑜀', '𑜚'), ('𑜠', '𑜡'), ('𑜦', '𑜦'), - ('𑜰', '𑜿'), + ('𑜰', 'ð‘†'), ('ð‘ €', 'ð‘ ®'), ('ð‘ ¸', 'ð‘ ¸'), ('ð‘ »', 'ð‘ »'), ('ð‘¢ ', 'ð‘£²'), - ('𑣿', '\u{11906}'), - ('\u{11909}', '\u{11909}'), - ('\u{1190c}', '\u{11913}'), - ('\u{11915}', '\u{11916}'), - ('\u{11918}', '\u{1192f}'), - ('\u{11931}', '\u{11935}'), - ('\u{11937}', '\u{11938}'), - ('\u{1193d}', '\u{1193d}'), - ('\u{1193f}', '\u{11942}'), - ('\u{11944}', '\u{11946}'), - ('\u{11950}', '\u{11959}'), + ('𑣿', '𑤆'), + ('𑤉', '𑤉'), + ('𑤌', '𑤓'), + ('𑤕', '𑤖'), + ('𑤘', '𑤯'), + ('𑤱', '𑤵'), + ('𑤷', '𑤸'), + ('𑤽', '𑤽'), + ('𑤿', '𑥂'), + ('𑥄', '𑥆'), + ('ð‘¥', 'ð‘¥™'), ('𑦠', '𑦧'), ('𑦪', '𑧓'), ('𑧜', '𑧟'), @@ -5327,7 +5462,7 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('ð‘©œ', '𑪉'), ('𑪗', '𑪗'), ('𑪚', '𑪢'), - ('ð‘«€', '𑫸'), + ('𑪰', '𑫸'), ('ð‘°€', 'ð‘°ˆ'), ('ð‘°Š', 'ð‘°¯'), ('ð‘°¾', 'ð‘°¾'), @@ -5351,18 +5486,20 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('𑶠', '𑶩'), ('ð‘» ', 'ð‘»²'), ('ð‘»µ', '𑻸'), - ('\u{11fb0}', '\u{11fb0}'), + ('𑾰', '𑾰'), ('ð‘¿€', 'ð‘¿±'), ('ð‘¿¿', 'ð’Ž™'), ('ð’€', 'ð’‘®'), ('ð’‘°', 'ð’‘´'), ('ð’’€', '𒕃'), + ('ð’¾', 'ð’¿²'), ('ð“€€', 'ð“®'), ('ð”€', '𔙆'), ('ð– €', '𖨸'), ('ð–©€', 'ð–©ž'), ('ð–© ', 'ð–©©'), - ('ð–©®', '𖩯'), + ('ð–©®', '𖪾'), + ('ð–«€', '𖫉'), ('ð–«', 'ð–«'), ('ð–«µ', 'ð–«µ'), ('𖬀', '𖬯'), @@ -5376,11 +5513,14 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('ð–½', '𖾇'), ('𖾓', '𖾟'), ('ð–¿ ', 'ð–¿£'), - ('\u{16ff0}', '\u{16ff1}'), + ('ð–¿°', 'ð–¿±'), ('ð—€€', '𘟷'), - ('𘠀', '\u{18cd5}'), - ('\u{18d00}', '\u{18d08}'), - ('𛀀', '𛄞'), + ('𘠀', '𘳕'), + ('𘴀', '𘴈'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), + ('𛀀', '𛄢'), ('ð›…', 'ð›…’'), ('ð›…¤', 'ð›…§'), ('ð›…°', '𛋻'), @@ -5390,6 +5530,7 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('ð›²', '𛲙'), ('𛲜', '𛲜'), ('𛲟', '𛲟'), + ('ðœ½', '𜿃'), ('ð€€', 'ðƒµ'), ('ð„€', 'ð„¦'), ('ð„©', 'ð…¤'), @@ -5397,7 +5538,7 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('ð…ª', 'ð…'), ('ð†ƒ', 'ð†„'), ('ð†Œ', 'ð†©'), - ('ð†®', 'ð‡¨'), + ('ð†®', 'ð‡ª'), ('ðˆ€', 'ð‰'), ('ð‰…', 'ð‰…'), ('ð‹ ', 'ð‹³'), @@ -5428,13 +5569,19 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('ð©', 'ð©´'), ('ð©¶', 'ðªƒ'), ('ðª…', 'ðª‹'), + ('ð¼€', 'ð¼ž'), ('ðž„€', '𞄬'), ('ðž„·', '𞄽'), ('ðž…€', 'ðž…‰'), ('ðž…Ž', 'ðž…'), + ('ðžŠ', 'ðžŠ'), ('ðž‹€', 'ðž‹«'), ('ðž‹°', '𞋹'), ('ðž‹¿', 'ðž‹¿'), + ('𞟠', '𞟦'), + ('𞟨', '𞟫'), + ('ðžŸ', '𞟮'), + ('𞟰', '𞟾'), ('ðž €', '𞣄'), ('𞣇', 'ðž£'), ('𞤀', '𞥃'), @@ -5483,45 +5630,46 @@ pub const GRAPHEME_BASE: &'static [(char, char)] = &[ ('🂱', '🂿'), ('ðŸƒ', 'ðŸƒ'), ('🃑', '🃵'), - ('🄀', '\u{1f1ad}'), + ('🄀', 'ðŸ†'), ('🇦', '🈂'), ('ðŸˆ', '🈻'), ('🉀', '🉈'), ('ðŸ‰', '🉑'), ('🉠', '🉥'), - ('🌀', '\u{1f6d7}'), - ('🛠', '🛬'), - ('🛰', '\u{1f6fc}'), + ('🌀', '🛗'), + ('ðŸ›', '🛬'), + ('🛰', '🛼'), ('🜀', 'ðŸ³'), ('🞀', '🟘'), ('🟠', '🟫'), + ('🟰', '🟰'), ('🠀', '🠋'), ('ðŸ ', '🡇'), ('ðŸ¡', '🡙'), ('🡠', '🢇'), ('ðŸ¢', 'ðŸ¢'), - ('\u{1f8b0}', '\u{1f8b1}'), - ('🤀', '\u{1f978}'), - ('🥺', '\u{1f9cb}'), - ('ðŸ§', '🩓'), + ('🢰', '🢱'), + ('🤀', '🩓'), ('🩠', 'ðŸ©'), - ('🩰', '\u{1fa74}'), - ('🩸', '🩺'), - ('🪀', '\u{1fa86}'), - ('ðŸª', '\u{1faa8}'), - ('\u{1fab0}', '\u{1fab6}'), - ('\u{1fac0}', '\u{1fac2}'), - ('\u{1fad0}', '\u{1fad6}'), - ('\u{1fb00}', '\u{1fb92}'), - ('\u{1fb94}', '\u{1fbca}'), - ('\u{1fbf0}', '\u{1fbf9}'), - ('ð €€', '\u{2a6dd}'), - ('𪜀', '𫜴'), + ('🩰', '🩴'), + ('🩸', '🩼'), + ('🪀', '🪆'), + ('ðŸª', '🪬'), + ('🪰', '🪺'), + ('🫀', '🫅'), + ('ðŸ«', '🫙'), + ('🫠', '🫧'), + ('🫰', '🫶'), + ('🬀', '🮒'), + ('🮔', '🯊'), + ('🯰', '🯹'), + ('ð €€', '𪛟'), + ('𪜀', '𫜸'), ('ð«€', 'ð« '), ('ð« ', '𬺡'), ('𬺰', '𮯠'), ('丽', 'ð¯¨'), - ('\u{30000}', '\u{3134a}'), + ('ð°€€', 'ð±Š'), ]; pub const GRAPHEME_EXTEND: &'static [(char, char)] = &[ @@ -5549,7 +5697,8 @@ pub const GRAPHEME_EXTEND: &'static [(char, char)] = &[ ('\u{825}', '\u{827}'), ('\u{829}', '\u{82d}'), ('\u{859}', '\u{85b}'), - ('\u{8d3}', '\u{8e1}'), + ('\u{898}', '\u{89f}'), + ('\u{8ca}', '\u{8e1}'), ('\u{8e3}', '\u{902}'), ('\u{93a}', '\u{93a}'), ('\u{93c}', '\u{93c}'), @@ -5594,6 +5743,7 @@ pub const GRAPHEME_EXTEND: &'static [(char, char)] = &[ ('\u{bd7}', '\u{bd7}'), ('\u{c00}', '\u{c00}'), ('\u{c04}', '\u{c04}'), + ('\u{c3c}', '\u{c3c}'), ('\u{c3e}', '\u{c40}'), ('\u{c46}', '\u{c48}'), ('\u{c4a}', '\u{c4d}'), @@ -5649,7 +5799,7 @@ pub const GRAPHEME_EXTEND: &'static [(char, char)] = &[ ('\u{109d}', '\u{109d}'), ('\u{135d}', '\u{135f}'), ('\u{1712}', '\u{1714}'), - ('\u{1732}', '\u{1734}'), + ('\u{1732}', '\u{1733}'), ('\u{1752}', '\u{1753}'), ('\u{1772}', '\u{1773}'), ('\u{17b4}', '\u{17b5}'), @@ -5658,6 +5808,7 @@ pub const GRAPHEME_EXTEND: &'static [(char, char)] = &[ ('\u{17c9}', '\u{17d3}'), ('\u{17dd}', '\u{17dd}'), ('\u{180b}', '\u{180d}'), + ('\u{180f}', '\u{180f}'), ('\u{1885}', '\u{1886}'), ('\u{18a9}', '\u{18a9}'), ('\u{1920}', '\u{1922}'), @@ -5673,7 +5824,7 @@ pub const GRAPHEME_EXTEND: &'static [(char, char)] = &[ ('\u{1a65}', '\u{1a6c}'), ('\u{1a73}', '\u{1a7c}'), ('\u{1a7f}', '\u{1a7f}'), - ('\u{1ab0}', '\u{1ac0}'), + ('\u{1ab0}', '\u{1ace}'), ('\u{1b00}', '\u{1b03}'), ('\u{1b34}', '\u{1b3a}'), ('\u{1b3c}', '\u{1b3c}'), @@ -5695,8 +5846,7 @@ pub const GRAPHEME_EXTEND: &'static [(char, char)] = &[ ('\u{1ced}', '\u{1ced}'), ('\u{1cf4}', '\u{1cf4}'), ('\u{1cf8}', '\u{1cf9}'), - ('\u{1dc0}', '\u{1df9}'), - ('\u{1dfb}', '\u{1dff}'), + ('\u{1dc0}', '\u{1dff}'), ('\u{200c}', '\u{200c}'), ('\u{20d0}', '\u{20f0}'), ('\u{2cef}', '\u{2cf1}'), @@ -5755,11 +5905,15 @@ pub const GRAPHEME_EXTEND: &'static [(char, char)] = &[ ('\u{10d24}', '\u{10d27}'), ('\u{10eab}', '\u{10eac}'), ('\u{10f46}', '\u{10f50}'), + ('\u{10f82}', '\u{10f85}'), ('\u{11001}', '\u{11001}'), ('\u{11038}', '\u{11046}'), + ('\u{11070}', '\u{11070}'), + ('\u{11073}', '\u{11074}'), ('\u{1107f}', '\u{11081}'), ('\u{110b3}', '\u{110b6}'), ('\u{110b9}', '\u{110ba}'), + ('\u{110c2}', '\u{110c2}'), ('\u{11100}', '\u{11102}'), ('\u{11127}', '\u{1112b}'), ('\u{1112d}', '\u{11134}'), @@ -5845,6 +5999,8 @@ pub const GRAPHEME_EXTEND: &'static [(char, char)] = &[ ('\u{16f8f}', '\u{16f92}'), ('\u{16fe4}', '\u{16fe4}'), ('\u{1bc9d}', '\u{1bc9e}'), + ('\u{1cf00}', '\u{1cf2d}'), + ('\u{1cf30}', '\u{1cf46}'), ('\u{1d165}', '\u{1d165}'), ('\u{1d167}', '\u{1d169}'), ('\u{1d16e}', '\u{1d172}'), @@ -5864,6 +6020,7 @@ pub const GRAPHEME_EXTEND: &'static [(char, char)] = &[ ('\u{1e023}', '\u{1e024}'), ('\u{1e026}', '\u{1e02a}'), ('\u{1e130}', '\u{1e136}'), + ('\u{1e2ae}', '\u{1e2ae}'), ('\u{1e2ec}', '\u{1e2ef}'), ('\u{1e8d0}', '\u{1e8d6}'), ('\u{1e944}', '\u{1e94a}'), @@ -5887,8 +6044,8 @@ pub const GRAPHEME_LINK: &'static [(char, char)] = &[ ('\u{eba}', '\u{eba}'), ('\u{f84}', '\u{f84}'), ('\u{1039}', '\u{103a}'), - ('\u{1714}', '\u{1714}'), - ('\u{1734}', '\u{1734}'), + ('\u{1714}', '᜕'), + ('᜴', '᜴'), ('\u{17d2}', '\u{17d2}'), ('\u{1a60}', '\u{1a60}'), ('á„', 'á„'), @@ -5904,6 +6061,7 @@ pub const GRAPHEME_LINK: &'static [(char, char)] = &[ ('\u{abed}', '\u{abed}'), ('\u{10a3f}', '\u{10a3f}'), ('\u{11046}', '\u{11046}'), + ('\u{11070}', '\u{11070}'), ('\u{1107f}', '\u{1107f}'), ('\u{110b9}', '\u{110b9}'), ('\u{11133}', '\u{11134}'), @@ -5918,7 +6076,7 @@ pub const GRAPHEME_LINK: &'static [(char, char)] = &[ ('𑚶', '𑚶'), ('\u{1172b}', '\u{1172b}'), ('\u{11839}', '\u{11839}'), - ('\u{1193d}', '\u{1193e}'), + ('𑤽', '\u{1193e}'), ('\u{119e0}', '\u{119e0}'), ('\u{11a34}', '\u{11a34}'), ('\u{11a47}', '\u{11a47}'), @@ -6007,9 +6165,9 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('à €', '\u{82d}'), ('à¡€', '\u{85b}'), ('à¡ ', 'ࡪ'), - ('ࢠ', 'ࢴ'), - ('ࢶ', '\u{8c7}'), - ('\u{8d3}', '\u{8e1}'), + ('à¡°', 'ࢇ'), + ('ࢉ', 'ࢎ'), + ('\u{898}', '\u{8e1}'), ('\u{8e3}', '\u{963}'), ('०', '९'), ('ॱ', 'ঃ'), @@ -6093,11 +6251,12 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('à°Ž', 'à°'), ('à°’', 'à°¨'), ('à°ª', 'à°¹'), - ('à°½', 'ౄ'), + ('\u{c3c}', 'ౄ'), ('\u{c46}', '\u{c48}'), ('\u{c4a}', '\u{c4d}'), ('\u{c55}', '\u{c56}'), ('ౘ', 'ౚ'), + ('à±', 'à±'), ('à± ', '\u{c63}'), ('౦', '౯'), ('ಀ', 'ಃ'), @@ -6110,7 +6269,7 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('\u{cc6}', 'ೈ'), ('ೊ', '\u{ccd}'), ('\u{cd5}', '\u{cd6}'), - ('ೞ', 'ೞ'), + ('à³', 'ೞ'), ('à³ ', '\u{ce3}'), ('೦', '೯'), ('à³±', 'à³²'), @@ -6193,9 +6352,8 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('áš', 'ášš'), ('áš ', 'ᛪ'), ('á›®', 'ᛸ'), - ('ᜀ', 'ᜌ'), - ('ᜎ', '\u{1714}'), - ('ᜠ', '\u{1734}'), + ('ᜀ', '᜕'), + ('ᜟ', '᜴'), ('á€', '\u{1753}'), ('á ', 'á¬'), ('á®', 'á°'), @@ -6205,7 +6363,7 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('ៜ', '\u{17dd}'), ('០', '៩'), ('\u{180b}', '\u{180d}'), - ('á ', 'á ™'), + ('\u{180f}', 'á ™'), ('á ', 'ᡸ'), ('ᢀ', 'ᢪ'), ('ᢰ', 'ᣵ'), @@ -6224,8 +6382,8 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('áª', '᪙'), ('ᪧ', 'ᪧ'), ('\u{1ab0}', '\u{1abd}'), - ('\u{1abf}', '\u{1ac0}'), - ('\u{1b00}', 'á‹'), + ('\u{1abf}', '\u{1ace}'), + ('\u{1b00}', 'áŒ'), ('á', 'á™'), ('\u{1b6b}', '\u{1b73}'), ('\u{1b80}', '᯳'), @@ -6237,8 +6395,7 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('á²½', 'Ჿ'), ('\u{1cd0}', '\u{1cd2}'), ('\u{1cd4}', 'ᳺ'), - ('á´€', '\u{1df9}'), - ('\u{1dfb}', 'ἕ'), + ('á´€', 'ἕ'), ('Ἐ', 'á¼'), ('á¼ ', 'á½…'), ('Ὀ', 'á½'), @@ -6278,9 +6435,7 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('â……', 'â…‰'), ('â…Ž', 'â…Ž'), ('â… ', 'ↈ'), - ('â°€', 'â°®'), - ('â°°', 'ⱞ'), - ('â± ', 'ⳤ'), + ('â°€', 'ⳤ'), ('Ⳬ', 'â³³'), ('â´€', 'â´¥'), ('â´§', 'â´§'), @@ -6307,11 +6462,10 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('ー', 'ヿ'), ('ã„…', 'ㄯ'), ('ㄱ', 'ㆎ'), - ('ㆠ', '\u{31bf}'), + ('ㆠ', 'ㆿ'), ('ㇰ', 'ㇿ'), - ('ã€', '\u{4dbf}'), - ('一', '\u{9ffc}'), - ('ꀀ', 'ê’Œ'), + ('ã€', '䶿'), + ('一', 'ê’Œ'), ('ê“', 'ꓽ'), ('ꔀ', 'ꘌ'), ('ê˜', 'ꘫ'), @@ -6320,9 +6474,11 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('ꙿ', '\u{a6f1}'), ('ꜗ', 'ꜟ'), ('Ꜣ', 'ꞈ'), - ('êž‹', 'êž¿'), - ('Ꟃ', '\u{a7ca}'), - ('\u{a7f5}', 'ê §'), + ('êž‹', 'ꟊ'), + ('êŸ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟙ'), + ('ꟲ', 'ê §'), ('\u{a82c}', '\u{a82c}'), ('ê¡€', 'ꡳ'), ('ꢀ', '\u{a8c5}'), @@ -6349,7 +6505,7 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('ê¬ ', 'ꬦ'), ('ꬨ', 'ꬮ'), ('ꬰ', 'êš'), - ('êœ', '\u{ab69}'), + ('êœ', 'ê©'), ('ê°', 'ꯪ'), ('꯬', '\u{abed}'), ('꯰', '꯹'), @@ -6411,9 +6567,20 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('ð“˜', 'ð“»'), ('ð”€', 'ð”§'), ('ð”°', 'ð•£'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), ('ð˜€', 'ðœ¶'), ('ð€', 'ð•'), ('ð ', 'ð§'), + ('ðž€', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ('ð €', 'ð …'), ('ð ˆ', 'ð ˆ'), ('ð Š', 'ð µ'), @@ -6448,27 +6615,29 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('ð³€', 'ð³²'), ('ð´€', '\u{10d27}'), ('ð´°', 'ð´¹'), - ('\u{10e80}', '\u{10ea9}'), + ('ðº€', 'ðº©'), ('\u{10eab}', '\u{10eac}'), - ('\u{10eb0}', '\u{10eb1}'), + ('ðº°', 'ðº±'), ('ð¼€', 'ð¼œ'), ('ð¼§', 'ð¼§'), ('ð¼°', '\u{10f50}'), - ('\u{10fb0}', '\u{10fc4}'), + ('ð½°', '\u{10f85}'), + ('ð¾°', 'ð¿„'), ('ð¿ ', 'ð¿¶'), ('ð‘€€', '\u{11046}'), - ('ð‘¦', 'ð‘¯'), + ('ð‘¦', 'ð‘µ'), ('\u{1107f}', '\u{110ba}'), + ('\u{110c2}', '\u{110c2}'), ('ð‘ƒ', '𑃨'), ('𑃰', '𑃹'), ('\u{11100}', '\u{11134}'), ('𑄶', 'ð‘„¿'), - ('ð‘…„', '\u{11147}'), + ('ð‘…„', 'ð‘…‡'), ('ð‘…', '\u{11173}'), ('ð‘…¶', 'ð‘…¶'), ('\u{11180}', '𑇄'), ('\u{111c9}', '\u{111cc}'), - ('\u{111ce}', '𑇚'), + ('𑇎', '𑇚'), ('𑇜', '𑇜'), ('𑈀', '𑈑'), ('𑈓', '\u{11237}'), @@ -6497,7 +6666,7 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('\u{11370}', '\u{11374}'), ('ð‘€', 'ð‘‘Š'), ('ð‘‘', 'ð‘‘™'), - ('\u{1145e}', '\u{11461}'), + ('\u{1145e}', 'ð‘‘¡'), ('ð‘’€', 'ð‘“…'), ('𑓇', '𑓇'), ('ð‘“', 'ð‘“™'), @@ -6512,16 +6681,17 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('𑜀', '𑜚'), ('\u{1171d}', '\u{1172b}'), ('𑜰', '𑜹'), + ('ð‘€', 'ð‘†'), ('ð‘ €', '\u{1183a}'), ('ð‘¢ ', '𑣩'), - ('𑣿', '\u{11906}'), - ('\u{11909}', '\u{11909}'), - ('\u{1190c}', '\u{11913}'), - ('\u{11915}', '\u{11916}'), - ('\u{11918}', '\u{11935}'), - ('\u{11937}', '\u{11938}'), + ('𑣿', '𑤆'), + ('𑤉', '𑤉'), + ('𑤌', '𑤓'), + ('𑤕', '𑤖'), + ('𑤘', '𑤵'), + ('𑤷', '𑤸'), ('\u{1193b}', '\u{11943}'), - ('\u{11950}', '\u{11959}'), + ('ð‘¥', 'ð‘¥™'), ('𑦠', '𑦧'), ('𑦪', '\u{119d7}'), ('\u{119da}', '𑧡'), @@ -6530,7 +6700,7 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('\u{11a47}', '\u{11a47}'), ('ð‘©', '\u{11a99}'), ('ð‘ª', 'ð‘ª'), - ('ð‘«€', '𑫸'), + ('𑪰', '𑫸'), ('ð‘°€', 'ð‘°ˆ'), ('ð‘°Š', '\u{11c36}'), ('\u{11c38}', '𑱀'), @@ -6552,15 +6722,18 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('𑶓', '𑶘'), ('𑶠', '𑶩'), ('ð‘» ', '𑻶'), - ('\u{11fb0}', '\u{11fb0}'), + ('𑾰', '𑾰'), ('ð’€€', 'ð’Ž™'), ('ð’€', 'ð’‘®'), ('ð’’€', '𒕃'), + ('ð’¾', 'ð’¿°'), ('ð“€€', 'ð“®'), ('ð”€', '𔙆'), ('ð– €', '𖨸'), ('ð–©€', 'ð–©ž'), ('ð–© ', 'ð–©©'), + ('ð–©°', '𖪾'), + ('ð–«€', '𖫉'), ('ð–«', 'ð–«'), ('\u{16af0}', '\u{16af4}'), ('𖬀', '\u{16b36}'), @@ -6574,11 +6747,14 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('\u{16f8f}', '𖾟'), ('ð–¿ ', 'ð–¿¡'), ('ð–¿£', '\u{16fe4}'), - ('\u{16ff0}', '\u{16ff1}'), + ('ð–¿°', 'ð–¿±'), ('ð—€€', '𘟷'), - ('𘠀', '\u{18cd5}'), - ('\u{18d00}', '\u{18d08}'), - ('𛀀', '𛄞'), + ('𘠀', '𘳕'), + ('𘴀', '𘴈'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), + ('𛀀', '𛄢'), ('ð›…', 'ð›…’'), ('ð›…¤', 'ð›…§'), ('ð›…°', '𛋻'), @@ -6587,6 +6763,8 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('𛲀', '𛲈'), ('ð›²', '𛲙'), ('\u{1bc9d}', '\u{1bc9e}'), + ('\u{1cf00}', '\u{1cf2d}'), + ('\u{1cf30}', '\u{1cf46}'), ('\u{1d165}', '\u{1d169}'), ('ð…', '\u{1d172}'), ('\u{1d17b}', '\u{1d182}'), @@ -6630,6 +6808,7 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('\u{1da84}', '\u{1da84}'), ('\u{1da9b}', '\u{1da9f}'), ('\u{1daa1}', '\u{1daaf}'), + ('ð¼€', 'ð¼ž'), ('\u{1e000}', '\u{1e006}'), ('\u{1e008}', '\u{1e018}'), ('\u{1e01b}', '\u{1e021}'), @@ -6639,7 +6818,12 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('\u{1e130}', '𞄽'), ('ðž…€', 'ðž…‰'), ('ðž…Ž', 'ðž…Ž'), + ('ðžŠ', '\u{1e2ae}'), ('ðž‹€', '𞋹'), + ('𞟠', '𞟦'), + ('𞟨', '𞟫'), + ('ðžŸ', '𞟮'), + ('𞟰', '𞟾'), ('ðž €', '𞣄'), ('\u{1e8d0}', '\u{1e8d6}'), ('𞤀', '𞥋'), @@ -6677,14 +6861,14 @@ pub const ID_CONTINUE: &'static [(char, char)] = &[ ('𞺡', '𞺣'), ('𞺥', '𞺩'), ('𞺫', '𞺻'), - ('\u{1fbf0}', '\u{1fbf9}'), - ('ð €€', '\u{2a6dd}'), - ('𪜀', '𫜴'), + ('🯰', '🯹'), + ('ð €€', '𪛟'), + ('𪜀', '𫜸'), ('ð«€', 'ð« '), ('ð« ', '𬺡'), ('𬺰', '𮯠'), ('丽', 'ð¯¨'), - ('\u{30000}', '\u{3134a}'), + ('ð°€€', 'ð±Š'), ('\u{e0100}', '\u{e01ef}'), ]; @@ -6738,8 +6922,9 @@ pub const ID_START: &'static [(char, char)] = &[ ('à ¨', 'à ¨'), ('à¡€', 'ࡘ'), ('à¡ ', 'ࡪ'), - ('ࢠ', 'ࢴ'), - ('ࢶ', '\u{8c7}'), + ('à¡°', 'ࢇ'), + ('ࢉ', 'ࢎ'), + ('ࢠ', 'ࣉ'), ('ऄ', 'ह'), ('ऽ', 'ऽ'), ('à¥', 'à¥'), @@ -6804,6 +6989,7 @@ pub const ID_START: &'static [(char, char)] = &[ ('à°ª', 'à°¹'), ('à°½', 'à°½'), ('ౘ', 'ౚ'), + ('à±', 'à±'), ('à± ', 'ౡ'), ('ಀ', 'ಀ'), ('ಅ', 'ಌ'), @@ -6812,10 +6998,10 @@ pub const ID_START: &'static [(char, char)] = &[ ('ಪ', 'ಳ'), ('ವ', 'ಹ'), ('ಽ', 'ಽ'), - ('ೞ', 'ೞ'), + ('à³', 'ೞ'), ('à³ ', 'ೡ'), ('à³±', 'à³²'), - ('\u{d04}', 'à´Œ'), + ('à´„', 'à´Œ'), ('à´Ž', 'à´'), ('à´’', 'à´º'), ('à´½', 'à´½'), @@ -6883,9 +7069,8 @@ pub const ID_START: &'static [(char, char)] = &[ ('áš', 'ášš'), ('áš ', 'ᛪ'), ('á›®', 'ᛸ'), - ('ᜀ', 'ᜌ'), - ('ᜎ', 'ᜑ'), - ('ᜠ', 'ᜱ'), + ('ᜀ', 'ᜑ'), + ('ᜟ', 'ᜱ'), ('á€', 'á‘'), ('á ', 'á¬'), ('á®', 'á°'), @@ -6905,7 +7090,7 @@ pub const ID_START: &'static [(char, char)] = &[ ('ᨠ', 'á©”'), ('ᪧ', 'ᪧ'), ('ᬅ', 'ᬳ'), - ('á…', 'á‹'), + ('á…', 'áŒ'), ('ᮃ', 'á® '), ('á®®', 'ᮯ'), ('ᮺ', 'ᯥ'), @@ -6955,9 +7140,7 @@ pub const ID_START: &'static [(char, char)] = &[ ('â……', 'â…‰'), ('â…Ž', 'â…Ž'), ('â… ', 'ↈ'), - ('â°€', 'â°®'), - ('â°°', 'ⱞ'), - ('â± ', 'ⳤ'), + ('â°€', 'ⳤ'), ('Ⳬ', 'â³®'), ('â³²', 'â³³'), ('â´€', 'â´¥'), @@ -6984,11 +7167,10 @@ pub const ID_START: &'static [(char, char)] = &[ ('ー', 'ヿ'), ('ã„…', 'ㄯ'), ('ㄱ', 'ㆎ'), - ('ㆠ', '\u{31bf}'), + ('ㆠ', 'ㆿ'), ('ㇰ', 'ㇿ'), - ('ã€', '\u{4dbf}'), - ('一', '\u{9ffc}'), - ('ꀀ', 'ê’Œ'), + ('ã€', '䶿'), + ('一', 'ê’Œ'), ('ê“', 'ꓽ'), ('ꔀ', 'ꘌ'), ('ê˜', 'ꘟ'), @@ -6998,9 +7180,11 @@ pub const ID_START: &'static [(char, char)] = &[ ('êš ', 'ꛯ'), ('ꜗ', 'ꜟ'), ('Ꜣ', 'ꞈ'), - ('êž‹', 'êž¿'), - ('Ꟃ', '\u{a7ca}'), - ('\u{a7f5}', 'ê '), + ('êž‹', 'ꟊ'), + ('êŸ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟙ'), + ('ꟲ', 'ê '), ('ê ƒ', 'ê …'), ('ê ‡', 'ê Š'), ('ê Œ', 'ê ¢'), @@ -7037,7 +7221,7 @@ pub const ID_START: &'static [(char, char)] = &[ ('ê¬ ', 'ꬦ'), ('ꬨ', 'ꬮ'), ('ꬰ', 'êš'), - ('êœ', '\u{ab69}'), + ('êœ', 'ê©'), ('ê°', 'ꯢ'), ('ê°€', '힣'), ('íž°', 'ퟆ'), @@ -7089,9 +7273,20 @@ pub const ID_START: &'static [(char, char)] = &[ ('ð“˜', 'ð“»'), ('ð”€', 'ð”§'), ('ð”°', 'ð•£'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), ('ð˜€', 'ðœ¶'), ('ð€', 'ð•'), ('ð ', 'ð§'), + ('ðž€', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ('ð €', 'ð …'), ('ð ˆ', 'ð ˆ'), ('ð Š', 'ð µ'), @@ -7122,19 +7317,22 @@ pub const ID_START: &'static [(char, char)] = &[ ('ð²€', 'ð²²'), ('ð³€', 'ð³²'), ('ð´€', 'ð´£'), - ('\u{10e80}', '\u{10ea9}'), - ('\u{10eb0}', '\u{10eb1}'), + ('ðº€', 'ðº©'), + ('ðº°', 'ðº±'), ('ð¼€', 'ð¼œ'), ('ð¼§', 'ð¼§'), ('ð¼°', 'ð½…'), - ('\u{10fb0}', '\u{10fc4}'), + ('ð½°', 'ð¾'), + ('ð¾°', 'ð¿„'), ('ð¿ ', 'ð¿¶'), ('𑀃', 'ð‘€·'), + ('ð‘±', 'ð‘²'), + ('ð‘µ', 'ð‘µ'), ('𑂃', '𑂯'), ('ð‘ƒ', '𑃨'), ('𑄃', '𑄦'), ('ð‘…„', 'ð‘…„'), - ('\u{11147}', '\u{11147}'), + ('ð‘…‡', 'ð‘…‡'), ('ð‘…', 'ð‘…²'), ('ð‘…¶', 'ð‘…¶'), ('𑆃', '𑆲'), @@ -7160,7 +7358,7 @@ pub const ID_START: &'static [(char, char)] = &[ ('ð‘', 'ð‘¡'), ('ð‘€', 'ð‘´'), ('𑑇', 'ð‘‘Š'), - ('ð‘‘Ÿ', '\u{11461}'), + ('ð‘‘Ÿ', 'ð‘‘¡'), ('ð‘’€', 'ð‘’¯'), ('ð‘“„', 'ð‘“…'), ('𑓇', '𑓇'), @@ -7171,15 +7369,16 @@ pub const ID_START: &'static [(char, char)] = &[ ('𑚀', '𑚪'), ('𑚸', '𑚸'), ('𑜀', '𑜚'), + ('ð‘€', 'ð‘†'), ('ð‘ €', 'ð‘ «'), ('ð‘¢ ', '𑣟'), - ('𑣿', '\u{11906}'), - ('\u{11909}', '\u{11909}'), - ('\u{1190c}', '\u{11913}'), - ('\u{11915}', '\u{11916}'), - ('\u{11918}', '\u{1192f}'), - ('\u{1193f}', '\u{1193f}'), - ('\u{11941}', '\u{11941}'), + ('𑣿', '𑤆'), + ('𑤉', '𑤉'), + ('𑤌', '𑤓'), + ('𑤕', '𑤖'), + ('𑤘', '𑤯'), + ('𑤿', '𑤿'), + ('ð‘¥', 'ð‘¥'), ('𑦠', '𑦧'), ('𑦪', 'ð‘§'), ('𑧡', '𑧡'), @@ -7190,7 +7389,7 @@ pub const ID_START: &'static [(char, char)] = &[ ('ð‘©', 'ð‘©'), ('ð‘©œ', '𑪉'), ('ð‘ª', 'ð‘ª'), - ('ð‘«€', '𑫸'), + ('𑪰', '𑫸'), ('ð‘°€', 'ð‘°ˆ'), ('ð‘°Š', 'ð‘°®'), ('𑱀', '𑱀'), @@ -7204,14 +7403,16 @@ pub const ID_START: &'static [(char, char)] = &[ ('𑵪', '𑶉'), ('𑶘', '𑶘'), ('ð‘» ', 'ð‘»²'), - ('\u{11fb0}', '\u{11fb0}'), + ('𑾰', '𑾰'), ('ð’€€', 'ð’Ž™'), ('ð’€', 'ð’‘®'), ('ð’’€', '𒕃'), + ('ð’¾', 'ð’¿°'), ('ð“€€', 'ð“®'), ('ð”€', '𔙆'), ('ð– €', '𖨸'), ('ð–©€', 'ð–©ž'), + ('ð–©°', '𖪾'), ('ð–«', 'ð–«'), ('𖬀', '𖬯'), ('ð–€', 'ð–ƒ'), @@ -7224,9 +7425,12 @@ pub const ID_START: &'static [(char, char)] = &[ ('ð–¿ ', 'ð–¿¡'), ('ð–¿£', 'ð–¿£'), ('ð—€€', '𘟷'), - ('𘠀', '\u{18cd5}'), - ('\u{18d00}', '\u{18d08}'), - ('𛀀', '𛄞'), + ('𘠀', '𘳕'), + ('𘴀', '𘴈'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), + ('𛀀', '𛄢'), ('ð›…', 'ð›…’'), ('ð›…¤', 'ð›…§'), ('ð›…°', '𛋻'), @@ -7264,10 +7468,16 @@ pub const ID_START: &'static [(char, char)] = &[ ('ðžŠ', 'ðž¨'), ('ðžª', 'ðŸ‚'), ('ðŸ„', 'ðŸ‹'), + ('ð¼€', 'ð¼ž'), ('ðž„€', '𞄬'), ('ðž„·', '𞄽'), ('ðž…Ž', 'ðž…Ž'), + ('ðžŠ', 'ðžŠ'), ('ðž‹€', 'ðž‹«'), + ('𞟠', '𞟦'), + ('𞟨', '𞟫'), + ('ðžŸ', '𞟮'), + ('𞟰', '𞟾'), ('ðž €', '𞣄'), ('𞤀', '𞥃'), ('𞥋', '𞥋'), @@ -7304,35 +7514,35 @@ pub const ID_START: &'static [(char, char)] = &[ ('𞺡', '𞺣'), ('𞺥', '𞺩'), ('𞺫', '𞺻'), - ('ð €€', '\u{2a6dd}'), - ('𪜀', '𫜴'), + ('ð €€', '𪛟'), + ('𪜀', '𫜸'), ('ð«€', 'ð« '), ('ð« ', '𬺡'), ('𬺰', '𮯠'), ('丽', 'ð¯¨'), - ('\u{30000}', '\u{3134a}'), + ('ð°€€', 'ð±Š'), ]; pub const IDEOGRAPHIC: &'static [(char, char)] = &[ ('〆', '〇'), ('〡', '〩'), ('〸', '〺'), - ('ã€', '\u{4dbf}'), - ('一', '\u{9ffc}'), + ('ã€', '䶿'), + ('一', 'é¿¿'), ('豈', 'ï©'), ('ï©°', 'ï«™'), ('\u{16fe4}', '\u{16fe4}'), ('ð—€€', '𘟷'), - ('𘠀', '\u{18cd5}'), - ('\u{18d00}', '\u{18d08}'), + ('𘠀', '𘳕'), + ('𘴀', '𘴈'), ('ð›…°', '𛋻'), - ('ð €€', '\u{2a6dd}'), - ('𪜀', '𫜴'), + ('ð €€', '𪛟'), + ('𪜀', '𫜸'), ('ð«€', 'ð« '), ('ð« ', '𬺡'), ('𬺰', '𮯠'), ('丽', 'ð¯¨'), - ('\u{30000}', '\u{3134a}'), + ('ð°€€', 'ð±Š'), ]; pub const JOIN_CONTROL: &'static [(char, char)] = &[('\u{200c}', '\u{200d}')]; @@ -7787,7 +7997,7 @@ pub const LOWERCASE: &'static [(char, char)] = &[ ('â…°', 'â…¿'), ('ↄ', 'ↄ'), ('â“', 'â“©'), - ('â°°', 'ⱞ'), + ('â°°', 'ⱟ'), ('ⱡ', 'ⱡ'), ('â±¥', 'ⱦ'), ('ⱨ', 'ⱨ'), @@ -7955,19 +8165,33 @@ pub const LOWERCASE: &'static [(char, char)] = &[ ('êž»', 'êž»'), ('êž½', 'êž½'), ('êž¿', 'êž¿'), + ('êŸ', 'êŸ'), ('ꟃ', 'ꟃ'), - ('\u{a7c8}', '\u{a7c8}'), - ('\u{a7ca}', '\u{a7ca}'), - ('\u{a7f6}', '\u{a7f6}'), + ('ꟈ', 'ꟈ'), + ('ꟊ', 'ꟊ'), + ('ꟑ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟕ'), + ('ꟗ', 'ꟗ'), + ('ꟙ', 'ꟙ'), + ('ꟶ', 'ꟶ'), ('ꟸ', 'ꟺ'), ('ꬰ', 'êš'), - ('êœ', '\u{ab68}'), + ('êœ', 'ê¨'), ('ê°', 'ꮿ'), ('ff', 'st'), ('ﬓ', 'ﬗ'), ('ï½', 'z'), ('ð¨', 'ð‘'), ('ð“˜', 'ð“»'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), + ('ðž€', 'ðž€'), + ('ðžƒ', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ('ð³€', 'ð³²'), ('ð‘£€', '𑣟'), ('ð–¹ ', '𖹿'), @@ -7999,6 +8223,8 @@ pub const LOWERCASE: &'static [(char, char)] = &[ ('ðžª', 'ðŸ‚'), ('ðŸ„', 'ðŸ‰'), ('ðŸ‹', 'ðŸ‹'), + ('ð¼€', 'ð¼‰'), + ('ð¼‹', 'ð¼ž'), ('𞤢', '𞥃'), ]; @@ -8281,6 +8507,7 @@ pub const OTHER_ALPHABETIC: &'static [(char, char)] = &[ ('á©•', '\u{1a5e}'), ('á©¡', '\u{1a74}'), ('\u{1abf}', '\u{1ac0}'), + ('\u{1acc}', '\u{1ace}'), ('\u{1b00}', 'ᬄ'), ('\u{1b35}', 'áƒ'), ('\u{1b80}', 'ᮂ'), @@ -8325,14 +8552,16 @@ pub const OTHER_ALPHABETIC: &'static [(char, char)] = &[ ('\u{10eab}', '\u{10eac}'), ('ð‘€€', '𑀂'), ('\u{11038}', '\u{11045}'), + ('\u{11073}', '\u{11074}'), ('ð‘‚‚', 'ð‘‚‚'), ('ð‘‚°', '𑂸'), + ('\u{110c2}', '\u{110c2}'), ('\u{11100}', '\u{11102}'), ('\u{11127}', '\u{11132}'), ('ð‘……', 'ð‘…†'), ('\u{11180}', '𑆂'), ('𑆳', '𑆿'), - ('\u{111ce}', '\u{111cf}'), + ('𑇎', '\u{111cf}'), ('𑈬', '\u{11234}'), ('\u{11237}', '\u{11237}'), ('\u{1123e}', '\u{1123e}'), @@ -8354,11 +8583,11 @@ pub const OTHER_ALPHABETIC: &'static [(char, char)] = &[ ('\u{116ab}', '\u{116b5}'), ('\u{1171d}', '\u{1172a}'), ('ð‘ ¬', 'ð‘ ¸'), - ('\u{11930}', '\u{11935}'), - ('\u{11937}', '\u{11938}'), + ('\u{11930}', '𑤵'), + ('𑤷', '𑤸'), ('\u{1193b}', '\u{1193c}'), - ('\u{11940}', '\u{11940}'), - ('\u{11942}', '\u{11942}'), + ('ð‘¥€', 'ð‘¥€'), + ('𑥂', '𑥂'), ('𑧑', '\u{119d7}'), ('\u{119da}', '𑧟'), ('𑧤', '𑧤'), @@ -8384,7 +8613,7 @@ pub const OTHER_ALPHABETIC: &'static [(char, char)] = &[ ('\u{16f4f}', '\u{16f4f}'), ('𖽑', '𖾇'), ('\u{16f8f}', '\u{16f92}'), - ('\u{16ff0}', '\u{16ff1}'), + ('ð–¿°', 'ð–¿±'), ('\u{1bc9e}', '\u{1bc9e}'), ('\u{1e000}', '\u{1e006}'), ('\u{1e008}', '\u{1e018}'), @@ -8466,6 +8695,10 @@ pub const OTHER_LOWERCASE: &'static [(char, char)] = &[ ('ê°', 'ê°'), ('ꟸ', 'ꟹ'), ('êœ', 'êŸ'), + ('ðž€', 'ðž€'), + ('ðžƒ', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ]; pub const OTHER_MATH: &'static [(char, char)] = &[ @@ -8651,13 +8884,14 @@ pub const PREPENDED_CONCATENATION_MARK: &'static [(char, char)] = &[ ('\u{600}', '\u{605}'), ('\u{6dd}', '\u{6dd}'), ('\u{70f}', '\u{70f}'), + ('\u{890}', '\u{891}'), ('\u{8e2}', '\u{8e2}'), ('\u{110bd}', '\u{110bd}'), ('\u{110cd}', '\u{110cd}'), ]; pub const QUOTATION_MARK: &'static [(char, char)] = &[ - ('\"', '\"'), + ('"', '"'), ('\'', '\''), ('«', '«'), ('»', '»'), @@ -8682,7 +8916,7 @@ pub const SENTENCE_TERMINAL: &'static [(char, char)] = &[ ('.', '.'), ('?', '?'), ('Ö‰', 'Ö‰'), - ('Øž', 'ØŸ'), + ('Ø', 'ØŸ'), ('Û”', 'Û”'), ('Ü€', 'Ü‚'), ('ß¹', 'ß¹'), @@ -8701,12 +8935,14 @@ pub const SENTENCE_TERMINAL: &'static [(char, char)] = &[ ('᪨', '᪫'), ('áš', 'á›'), ('áž', 'áŸ'), + ('á½', 'á¾'), ('á°»', 'á°¼'), ('á±¾', '᱿'), ('‼', '‽'), ('â‡', 'â‰'), ('⸮', '⸮'), ('⸼', '⸼'), + ('⹓', 'â¹”'), ('。', '。'), ('ê“¿', 'ê“¿'), ('꘎', 'ê˜'), @@ -8727,6 +8963,7 @@ pub const SENTENCE_TERMINAL: &'static [(char, char)] = &[ ('。', '。'), ('ð©–', 'ð©—'), ('ð½•', 'ð½™'), + ('ð¾†', 'ð¾‰'), ('ð‘‡', 'ð‘ˆ'), ('ð‘‚¾', 'ð‘ƒ'), ('ð‘…', 'ð‘…ƒ'), @@ -8741,8 +8978,8 @@ pub const SENTENCE_TERMINAL: &'static [(char, char)] = &[ ('ð‘—‰', 'ð‘——'), ('ð‘™', 'ð‘™‚'), ('𑜼', '𑜾'), - ('\u{11944}', '\u{11944}'), - ('\u{11946}', '\u{11946}'), + ('𑥄', '𑥄'), + ('𑥆', '𑥆'), ('ð‘©‚', '𑩃'), ('𑪛', '𑪜'), ('ð‘±', '𑱂'), @@ -8788,6 +9025,7 @@ pub const SOFT_DOTTED: &'static [(char, char)] = &[ ('ð˜ª', 'ð˜«'), ('ð™ž', 'ð™Ÿ'), ('ðš’', 'ðš“'), + ('ð¼š', 'ð¼š'), ]; pub const TERMINAL_PUNCTUATION: &'static [(char, char)] = &[ @@ -8802,7 +9040,7 @@ pub const TERMINAL_PUNCTUATION: &'static [(char, char)] = &[ ('׃', '׃'), ('ØŒ', 'ØŒ'), ('Ø›', 'Ø›'), - ('Øž', 'ØŸ'), + ('Ø', 'ØŸ'), ('Û”', 'Û”'), ('Ü€', 'ÜŠ'), ('ÜŒ', 'ÜŒ'), @@ -8826,6 +9064,7 @@ pub const TERMINAL_PUNCTUATION: &'static [(char, char)] = &[ ('᪨', '᪫'), ('áš', 'á›'), ('á', 'áŸ'), + ('á½', 'á¾'), ('á°»', 'á°¿'), ('á±¾', '᱿'), ('‼', '‽'), @@ -8835,6 +9074,7 @@ pub const TERMINAL_PUNCTUATION: &'static [(char, char)] = &[ ('â¹', 'â¹'), ('⹌', '⹌'), ('⹎', 'â¹'), + ('⹓', 'â¹”'), ('ã€', '。'), ('꓾', 'ê“¿'), ('ê˜', 'ê˜'), @@ -8865,6 +9105,7 @@ pub const TERMINAL_PUNCTUATION: &'static [(char, char)] = &[ ('ð¬º', 'ð¬¿'), ('ð®™', 'ð®œ'), ('ð½•', 'ð½™'), + ('ð¾†', 'ð¾‰'), ('ð‘‡', 'ð‘'), ('ð‘‚¾', 'ð‘ƒ'), ('ð‘…', 'ð‘…ƒ'), @@ -8874,13 +9115,13 @@ pub const TERMINAL_PUNCTUATION: &'static [(char, char)] = &[ ('𑈸', '𑈼'), ('ð‘Š©', 'ð‘Š©'), ('ð‘‘‹', 'ð‘‘'), - ('\u{1145a}', 'ð‘‘›'), + ('ð‘‘š', 'ð‘‘›'), ('ð‘—‚', 'ð‘—…'), ('ð‘—‰', 'ð‘——'), ('ð‘™', 'ð‘™‚'), ('𑜼', '𑜾'), - ('\u{11944}', '\u{11944}'), - ('\u{11946}', '\u{11946}'), + ('𑥄', '𑥄'), + ('𑥆', '𑥆'), ('ð‘©‚', '𑩃'), ('𑪛', '𑪜'), ('𑪡', '𑪢'), @@ -8898,8 +9139,8 @@ pub const TERMINAL_PUNCTUATION: &'static [(char, char)] = &[ ]; pub const UNIFIED_IDEOGRAPH: &'static [(char, char)] = &[ - ('ã€', '\u{4dbf}'), - ('一', '\u{9ffc}'), + ('ã€', '䶿'), + ('一', 'é¿¿'), ('﨎', 'ï¨'), ('﨑', '﨑'), ('﨓', '﨔'), @@ -8907,12 +9148,12 @@ pub const UNIFIED_IDEOGRAPH: &'static [(char, char)] = &[ ('﨡', '﨡'), ('﨣', '﨤'), ('﨧', '﨩'), - ('ð €€', '\u{2a6dd}'), - ('𪜀', '𫜴'), + ('ð €€', '𪛟'), + ('𪜀', '𫜸'), ('ð«€', 'ð« '), ('ð« ', '𬺡'), ('𬺰', '𮯠'), - ('\u{30000}', '\u{3134a}'), + ('ð°€€', 'ð±Š'), ]; pub const UPPERCASE: &'static [(char, char)] = &[ @@ -9349,7 +9590,7 @@ pub const UPPERCASE: &'static [(char, char)] = &[ ('â… ', 'â…¯'), ('Ↄ', 'Ↄ'), ('â’¶', 'â“'), - ('â°€', 'â°®'), + ('â°€', 'â°¯'), ('â± ', 'â± '), ('â±¢', 'Ɽ'), ('Ⱨ', 'Ⱨ'), @@ -9514,13 +9755,21 @@ pub const UPPERCASE: &'static [(char, char)] = &[ ('Ꞻ', 'Ꞻ'), ('êž¼', 'êž¼'), ('êž¾', 'êž¾'), + ('Ꟁ', 'Ꟁ'), ('Ꟃ', 'Ꟃ'), - ('Ꞔ', '\u{a7c7}'), - ('\u{a7c9}', '\u{a7c9}'), - ('\u{a7f5}', '\u{a7f5}'), + ('Ꞔ', 'Ꟈ'), + ('Ꟊ', 'Ꟊ'), + ('êŸ', 'êŸ'), + ('Ꟗ', 'Ꟗ'), + ('Ꟙ', 'Ꟙ'), + ('Ꟶ', 'Ꟶ'), ('A', 'Z'), ('ð€', 'ð§'), ('ð’°', 'ð““'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), ('ð²€', 'ð²²'), ('ð‘¢ ', '𑢿'), ('ð–¹€', '𖹟'), @@ -9563,6 +9812,7 @@ pub const UPPERCASE: &'static [(char, char)] = &[ pub const VARIATION_SELECTOR: &'static [(char, char)] = &[ ('\u{180b}', '\u{180d}'), + ('\u{180f}', '\u{180f}'), ('\u{fe00}', '\u{fe0f}'), ('\u{e0100}', '\u{e01ef}'), ]; @@ -9632,9 +9882,9 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('à €', '\u{82d}'), ('à¡€', '\u{85b}'), ('à¡ ', 'ࡪ'), - ('ࢠ', 'ࢴ'), - ('ࢶ', '\u{8c7}'), - ('\u{8d3}', '\u{8e1}'), + ('à¡°', 'ࢇ'), + ('ࢉ', 'ࢎ'), + ('\u{898}', '\u{8e1}'), ('\u{8e3}', '\u{963}'), ('०', '९'), ('ॱ', 'ঃ'), @@ -9718,11 +9968,12 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('à°Ž', 'à°'), ('à°’', 'à°¨'), ('à°ª', 'à°¹'), - ('à°½', 'ౄ'), + ('\u{c3c}', 'ౄ'), ('\u{c46}', '\u{c48}'), ('\u{c4a}', '\u{c4d}'), ('\u{c55}', '\u{c56}'), ('ౘ', 'ౚ'), + ('à±', 'à±'), ('à± ', '\u{c63}'), ('౦', '౯'), ('ಀ', 'ಃ'), @@ -9735,7 +9986,7 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('\u{cc6}', 'ೈ'), ('ೊ', '\u{ccd}'), ('\u{cd5}', '\u{cd6}'), - ('ೞ', 'ೞ'), + ('à³', 'ೞ'), ('à³ ', '\u{ce3}'), ('೦', '೯'), ('à³±', 'à³²'), @@ -9818,9 +10069,8 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('áš', 'ášš'), ('áš ', 'ᛪ'), ('á›®', 'ᛸ'), - ('ᜀ', 'ᜌ'), - ('ᜎ', '\u{1714}'), - ('ᜠ', '\u{1734}'), + ('ᜀ', '᜕'), + ('ᜟ', '᜴'), ('á€', '\u{1753}'), ('á ', 'á¬'), ('á®', 'á°'), @@ -9830,7 +10080,7 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('ៜ', '\u{17dd}'), ('០', '៩'), ('\u{180b}', '\u{180d}'), - ('á ', 'á ™'), + ('\u{180f}', 'á ™'), ('á ', 'ᡸ'), ('ᢀ', 'ᢪ'), ('ᢰ', 'ᣵ'), @@ -9849,8 +10099,8 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('áª', '᪙'), ('ᪧ', 'ᪧ'), ('\u{1ab0}', '\u{1abd}'), - ('\u{1abf}', '\u{1ac0}'), - ('\u{1b00}', 'á‹'), + ('\u{1abf}', '\u{1ace}'), + ('\u{1b00}', 'áŒ'), ('á', 'á™'), ('\u{1b6b}', '\u{1b73}'), ('\u{1b80}', '᯳'), @@ -9862,8 +10112,7 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('á²½', 'Ჿ'), ('\u{1cd0}', '\u{1cd2}'), ('\u{1cd4}', 'ᳺ'), - ('á´€', '\u{1df9}'), - ('\u{1dfb}', 'ἕ'), + ('á´€', 'ἕ'), ('Ἐ', 'á¼'), ('á¼ ', 'á½…'), ('Ὀ', 'á½'), @@ -9903,9 +10152,7 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('â……', 'â…‰'), ('â…Ž', 'â…Ž'), ('â… ', 'ↈ'), - ('â°€', 'â°®'), - ('â°°', 'ⱞ'), - ('â± ', 'ⳤ'), + ('â°€', 'ⳤ'), ('Ⳬ', 'â³³'), ('â´€', 'â´¥'), ('â´§', 'â´§'), @@ -9933,11 +10180,10 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('ー', 'ヿ'), ('ã„…', 'ㄯ'), ('ㄱ', 'ㆎ'), - ('ㆠ', '\u{31bf}'), + ('ㆠ', 'ㆿ'), ('ㇰ', 'ㇿ'), - ('ã€', '\u{4dbf}'), - ('一', '\u{9ffc}'), - ('ꀀ', 'ê’Œ'), + ('ã€', '䶿'), + ('一', 'ê’Œ'), ('ê“', 'ꓽ'), ('ꔀ', 'ꘌ'), ('ê˜', 'ꘫ'), @@ -9946,9 +10192,11 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('ꙿ', '\u{a6f1}'), ('ꜗ', 'ꜟ'), ('Ꜣ', 'ꞈ'), - ('êž‹', 'êž¿'), - ('Ꟃ', '\u{a7ca}'), - ('\u{a7f5}', 'ê §'), + ('êž‹', 'ꟊ'), + ('êŸ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟙ'), + ('ꟲ', 'ê §'), ('\u{a82c}', '\u{a82c}'), ('ê¡€', 'ꡳ'), ('ꢀ', '\u{a8c5}'), @@ -9975,7 +10223,7 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('ê¬ ', 'ꬦ'), ('ꬨ', 'ꬮ'), ('ꬰ', 'êš'), - ('êœ', '\u{ab69}'), + ('êœ', 'ê©'), ('ê°', 'ꯪ'), ('꯬', '\u{abed}'), ('꯰', '꯹'), @@ -10043,9 +10291,20 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('ð“˜', 'ð“»'), ('ð”€', 'ð”§'), ('ð”°', 'ð•£'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), ('ð˜€', 'ðœ¶'), ('ð€', 'ð•'), ('ð ', 'ð§'), + ('ðž€', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ('ð €', 'ð …'), ('ð ˆ', 'ð ˆ'), ('ð Š', 'ð µ'), @@ -10080,27 +10339,29 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('ð³€', 'ð³²'), ('ð´€', '\u{10d27}'), ('ð´°', 'ð´¹'), - ('\u{10e80}', '\u{10ea9}'), + ('ðº€', 'ðº©'), ('\u{10eab}', '\u{10eac}'), - ('\u{10eb0}', '\u{10eb1}'), + ('ðº°', 'ðº±'), ('ð¼€', 'ð¼œ'), ('ð¼§', 'ð¼§'), ('ð¼°', '\u{10f50}'), - ('\u{10fb0}', '\u{10fc4}'), + ('ð½°', '\u{10f85}'), + ('ð¾°', 'ð¿„'), ('ð¿ ', 'ð¿¶'), ('ð‘€€', '\u{11046}'), - ('ð‘¦', 'ð‘¯'), + ('ð‘¦', 'ð‘µ'), ('\u{1107f}', '\u{110ba}'), + ('\u{110c2}', '\u{110c2}'), ('ð‘ƒ', '𑃨'), ('𑃰', '𑃹'), ('\u{11100}', '\u{11134}'), ('𑄶', 'ð‘„¿'), - ('ð‘…„', '\u{11147}'), + ('ð‘…„', 'ð‘…‡'), ('ð‘…', '\u{11173}'), ('ð‘…¶', 'ð‘…¶'), ('\u{11180}', '𑇄'), ('\u{111c9}', '\u{111cc}'), - ('\u{111ce}', '𑇚'), + ('𑇎', '𑇚'), ('𑇜', '𑇜'), ('𑈀', '𑈑'), ('𑈓', '\u{11237}'), @@ -10129,7 +10390,7 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('\u{11370}', '\u{11374}'), ('ð‘€', 'ð‘‘Š'), ('ð‘‘', 'ð‘‘™'), - ('\u{1145e}', '\u{11461}'), + ('\u{1145e}', 'ð‘‘¡'), ('ð‘’€', 'ð‘“…'), ('𑓇', '𑓇'), ('ð‘“', 'ð‘“™'), @@ -10144,16 +10405,17 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('𑜀', '𑜚'), ('\u{1171d}', '\u{1172b}'), ('𑜰', '𑜹'), + ('ð‘€', 'ð‘†'), ('ð‘ €', '\u{1183a}'), ('ð‘¢ ', '𑣩'), - ('𑣿', '\u{11906}'), - ('\u{11909}', '\u{11909}'), - ('\u{1190c}', '\u{11913}'), - ('\u{11915}', '\u{11916}'), - ('\u{11918}', '\u{11935}'), - ('\u{11937}', '\u{11938}'), + ('𑣿', '𑤆'), + ('𑤉', '𑤉'), + ('𑤌', '𑤓'), + ('𑤕', '𑤖'), + ('𑤘', '𑤵'), + ('𑤷', '𑤸'), ('\u{1193b}', '\u{11943}'), - ('\u{11950}', '\u{11959}'), + ('ð‘¥', 'ð‘¥™'), ('𑦠', '𑦧'), ('𑦪', '\u{119d7}'), ('\u{119da}', '𑧡'), @@ -10162,7 +10424,7 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('\u{11a47}', '\u{11a47}'), ('ð‘©', '\u{11a99}'), ('ð‘ª', 'ð‘ª'), - ('ð‘«€', '𑫸'), + ('𑪰', '𑫸'), ('ð‘°€', 'ð‘°ˆ'), ('ð‘°Š', '\u{11c36}'), ('\u{11c38}', '𑱀'), @@ -10184,15 +10446,18 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('𑶓', '𑶘'), ('𑶠', '𑶩'), ('ð‘» ', '𑻶'), - ('\u{11fb0}', '\u{11fb0}'), + ('𑾰', '𑾰'), ('ð’€€', 'ð’Ž™'), ('ð’€', 'ð’‘®'), ('ð’’€', '𒕃'), + ('ð’¾', 'ð’¿°'), ('ð“€€', 'ð“®'), ('ð”€', '𔙆'), ('ð– €', '𖨸'), ('ð–©€', 'ð–©ž'), ('ð–© ', 'ð–©©'), + ('ð–©°', '𖪾'), + ('ð–«€', '𖫉'), ('ð–«', 'ð–«'), ('\u{16af0}', '\u{16af4}'), ('𖬀', '\u{16b36}'), @@ -10206,11 +10471,14 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('\u{16f8f}', '𖾟'), ('ð–¿ ', 'ð–¿¡'), ('ð–¿£', '\u{16fe4}'), - ('\u{16ff0}', '\u{16ff1}'), + ('ð–¿°', 'ð–¿±'), ('ð—€€', '𘟷'), - ('𘠀', '\u{18cd5}'), - ('\u{18d00}', '\u{18d08}'), - ('𛀀', '𛄞'), + ('𘠀', '𘳕'), + ('𘴀', '𘴈'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), + ('𛀀', '𛄢'), ('ð›…', 'ð›…’'), ('ð›…¤', 'ð›…§'), ('ð›…°', '𛋻'), @@ -10219,6 +10487,8 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('𛲀', '𛲈'), ('ð›²', '𛲙'), ('\u{1bc9d}', '\u{1bc9e}'), + ('\u{1cf00}', '\u{1cf2d}'), + ('\u{1cf30}', '\u{1cf46}'), ('\u{1d165}', '\u{1d169}'), ('ð…', '\u{1d172}'), ('\u{1d17b}', '\u{1d182}'), @@ -10262,6 +10532,7 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('\u{1da84}', '\u{1da84}'), ('\u{1da9b}', '\u{1da9f}'), ('\u{1daa1}', '\u{1daaf}'), + ('ð¼€', 'ð¼ž'), ('\u{1e000}', '\u{1e006}'), ('\u{1e008}', '\u{1e018}'), ('\u{1e01b}', '\u{1e021}'), @@ -10271,7 +10542,12 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('\u{1e130}', '𞄽'), ('ðž…€', 'ðž…‰'), ('ðž…Ž', 'ðž…Ž'), + ('ðžŠ', '\u{1e2ae}'), ('ðž‹€', '𞋹'), + ('𞟠', '𞟦'), + ('𞟨', '𞟫'), + ('ðžŸ', '𞟮'), + ('𞟰', '𞟾'), ('ðž €', '𞣄'), ('\u{1e8d0}', '\u{1e8d6}'), ('𞤀', '𞥋'), @@ -10309,14 +10585,14 @@ pub const XID_CONTINUE: &'static [(char, char)] = &[ ('𞺡', '𞺣'), ('𞺥', '𞺩'), ('𞺫', '𞺻'), - ('\u{1fbf0}', '\u{1fbf9}'), - ('ð €€', '\u{2a6dd}'), - ('𪜀', '𫜴'), + ('🯰', '🯹'), + ('ð €€', '𪛟'), + ('𪜀', '𫜸'), ('ð«€', 'ð« '), ('ð« ', '𬺡'), ('𬺰', '𮯠'), ('丽', 'ð¯¨'), - ('\u{30000}', '\u{3134a}'), + ('ð°€€', 'ð±Š'), ('\u{e0100}', '\u{e01ef}'), ]; @@ -10370,8 +10646,9 @@ pub const XID_START: &'static [(char, char)] = &[ ('à ¨', 'à ¨'), ('à¡€', 'ࡘ'), ('à¡ ', 'ࡪ'), - ('ࢠ', 'ࢴ'), - ('ࢶ', '\u{8c7}'), + ('à¡°', 'ࢇ'), + ('ࢉ', 'ࢎ'), + ('ࢠ', 'ࣉ'), ('ऄ', 'ह'), ('ऽ', 'ऽ'), ('à¥', 'à¥'), @@ -10436,6 +10713,7 @@ pub const XID_START: &'static [(char, char)] = &[ ('à°ª', 'à°¹'), ('à°½', 'à°½'), ('ౘ', 'ౚ'), + ('à±', 'à±'), ('à± ', 'ౡ'), ('ಀ', 'ಀ'), ('ಅ', 'ಌ'), @@ -10444,10 +10722,10 @@ pub const XID_START: &'static [(char, char)] = &[ ('ಪ', 'ಳ'), ('ವ', 'ಹ'), ('ಽ', 'ಽ'), - ('ೞ', 'ೞ'), + ('à³', 'ೞ'), ('à³ ', 'ೡ'), ('à³±', 'à³²'), - ('\u{d04}', 'à´Œ'), + ('à´„', 'à´Œ'), ('à´Ž', 'à´'), ('à´’', 'à´º'), ('à´½', 'à´½'), @@ -10515,9 +10793,8 @@ pub const XID_START: &'static [(char, char)] = &[ ('áš', 'ášš'), ('áš ', 'ᛪ'), ('á›®', 'ᛸ'), - ('ᜀ', 'ᜌ'), - ('ᜎ', 'ᜑ'), - ('ᜠ', 'ᜱ'), + ('ᜀ', 'ᜑ'), + ('ᜟ', 'ᜱ'), ('á€', 'á‘'), ('á ', 'á¬'), ('á®', 'á°'), @@ -10537,7 +10814,7 @@ pub const XID_START: &'static [(char, char)] = &[ ('ᨠ', 'á©”'), ('ᪧ', 'ᪧ'), ('ᬅ', 'ᬳ'), - ('á…', 'á‹'), + ('á…', 'áŒ'), ('ᮃ', 'á® '), ('á®®', 'ᮯ'), ('ᮺ', 'ᯥ'), @@ -10587,9 +10864,7 @@ pub const XID_START: &'static [(char, char)] = &[ ('â……', 'â…‰'), ('â…Ž', 'â…Ž'), ('â… ', 'ↈ'), - ('â°€', 'â°®'), - ('â°°', 'ⱞ'), - ('â± ', 'ⳤ'), + ('â°€', 'ⳤ'), ('Ⳬ', 'â³®'), ('â³²', 'â³³'), ('â´€', 'â´¥'), @@ -10616,11 +10891,10 @@ pub const XID_START: &'static [(char, char)] = &[ ('ー', 'ヿ'), ('ã„…', 'ㄯ'), ('ㄱ', 'ㆎ'), - ('ㆠ', '\u{31bf}'), + ('ㆠ', 'ㆿ'), ('ㇰ', 'ㇿ'), - ('ã€', '\u{4dbf}'), - ('一', '\u{9ffc}'), - ('ꀀ', 'ê’Œ'), + ('ã€', '䶿'), + ('一', 'ê’Œ'), ('ê“', 'ꓽ'), ('ꔀ', 'ꘌ'), ('ê˜', 'ꘟ'), @@ -10630,9 +10904,11 @@ pub const XID_START: &'static [(char, char)] = &[ ('êš ', 'ꛯ'), ('ꜗ', 'ꜟ'), ('Ꜣ', 'ꞈ'), - ('êž‹', 'êž¿'), - ('Ꟃ', '\u{a7ca}'), - ('\u{a7f5}', 'ê '), + ('êž‹', 'ꟊ'), + ('êŸ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟙ'), + ('ꟲ', 'ê '), ('ê ƒ', 'ê …'), ('ê ‡', 'ê Š'), ('ê Œ', 'ê ¢'), @@ -10669,7 +10945,7 @@ pub const XID_START: &'static [(char, char)] = &[ ('ê¬ ', 'ꬦ'), ('ꬨ', 'ꬮ'), ('ꬰ', 'êš'), - ('êœ', '\u{ab69}'), + ('êœ', 'ê©'), ('ê°', 'ꯢ'), ('ê°€', '힣'), ('íž°', 'ퟆ'), @@ -10728,9 +11004,20 @@ pub const XID_START: &'static [(char, char)] = &[ ('ð“˜', 'ð“»'), ('ð”€', 'ð”§'), ('ð”°', 'ð•£'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), ('ð˜€', 'ðœ¶'), ('ð€', 'ð•'), ('ð ', 'ð§'), + ('ðž€', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ('ð €', 'ð …'), ('ð ˆ', 'ð ˆ'), ('ð Š', 'ð µ'), @@ -10761,19 +11048,22 @@ pub const XID_START: &'static [(char, char)] = &[ ('ð²€', 'ð²²'), ('ð³€', 'ð³²'), ('ð´€', 'ð´£'), - ('\u{10e80}', '\u{10ea9}'), - ('\u{10eb0}', '\u{10eb1}'), + ('ðº€', 'ðº©'), + ('ðº°', 'ðº±'), ('ð¼€', 'ð¼œ'), ('ð¼§', 'ð¼§'), ('ð¼°', 'ð½…'), - ('\u{10fb0}', '\u{10fc4}'), + ('ð½°', 'ð¾'), + ('ð¾°', 'ð¿„'), ('ð¿ ', 'ð¿¶'), ('𑀃', 'ð‘€·'), + ('ð‘±', 'ð‘²'), + ('ð‘µ', 'ð‘µ'), ('𑂃', '𑂯'), ('ð‘ƒ', '𑃨'), ('𑄃', '𑄦'), ('ð‘…„', 'ð‘…„'), - ('\u{11147}', '\u{11147}'), + ('ð‘…‡', 'ð‘…‡'), ('ð‘…', 'ð‘…²'), ('ð‘…¶', 'ð‘…¶'), ('𑆃', '𑆲'), @@ -10799,7 +11089,7 @@ pub const XID_START: &'static [(char, char)] = &[ ('ð‘', 'ð‘¡'), ('ð‘€', 'ð‘´'), ('𑑇', 'ð‘‘Š'), - ('ð‘‘Ÿ', '\u{11461}'), + ('ð‘‘Ÿ', 'ð‘‘¡'), ('ð‘’€', 'ð‘’¯'), ('ð‘“„', 'ð‘“…'), ('𑓇', '𑓇'), @@ -10810,15 +11100,16 @@ pub const XID_START: &'static [(char, char)] = &[ ('𑚀', '𑚪'), ('𑚸', '𑚸'), ('𑜀', '𑜚'), + ('ð‘€', 'ð‘†'), ('ð‘ €', 'ð‘ «'), ('ð‘¢ ', '𑣟'), - ('𑣿', '\u{11906}'), - ('\u{11909}', '\u{11909}'), - ('\u{1190c}', '\u{11913}'), - ('\u{11915}', '\u{11916}'), - ('\u{11918}', '\u{1192f}'), - ('\u{1193f}', '\u{1193f}'), - ('\u{11941}', '\u{11941}'), + ('𑣿', '𑤆'), + ('𑤉', '𑤉'), + ('𑤌', '𑤓'), + ('𑤕', '𑤖'), + ('𑤘', '𑤯'), + ('𑤿', '𑤿'), + ('ð‘¥', 'ð‘¥'), ('𑦠', '𑦧'), ('𑦪', 'ð‘§'), ('𑧡', '𑧡'), @@ -10829,7 +11120,7 @@ pub const XID_START: &'static [(char, char)] = &[ ('ð‘©', 'ð‘©'), ('ð‘©œ', '𑪉'), ('ð‘ª', 'ð‘ª'), - ('ð‘«€', '𑫸'), + ('𑪰', '𑫸'), ('ð‘°€', 'ð‘°ˆ'), ('ð‘°Š', 'ð‘°®'), ('𑱀', '𑱀'), @@ -10843,14 +11134,16 @@ pub const XID_START: &'static [(char, char)] = &[ ('𑵪', '𑶉'), ('𑶘', '𑶘'), ('ð‘» ', 'ð‘»²'), - ('\u{11fb0}', '\u{11fb0}'), + ('𑾰', '𑾰'), ('ð’€€', 'ð’Ž™'), ('ð’€', 'ð’‘®'), ('ð’’€', '𒕃'), + ('ð’¾', 'ð’¿°'), ('ð“€€', 'ð“®'), ('ð”€', '𔙆'), ('ð– €', '𖨸'), ('ð–©€', 'ð–©ž'), + ('ð–©°', '𖪾'), ('ð–«', 'ð–«'), ('𖬀', '𖬯'), ('ð–€', 'ð–ƒ'), @@ -10863,9 +11156,12 @@ pub const XID_START: &'static [(char, char)] = &[ ('ð–¿ ', 'ð–¿¡'), ('ð–¿£', 'ð–¿£'), ('ð—€€', '𘟷'), - ('𘠀', '\u{18cd5}'), - ('\u{18d00}', '\u{18d08}'), - ('𛀀', '𛄞'), + ('𘠀', '𘳕'), + ('𘴀', '𘴈'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), + ('𛀀', '𛄢'), ('ð›…', 'ð›…’'), ('ð›…¤', 'ð›…§'), ('ð›…°', '𛋻'), @@ -10903,10 +11199,16 @@ pub const XID_START: &'static [(char, char)] = &[ ('ðžŠ', 'ðž¨'), ('ðžª', 'ðŸ‚'), ('ðŸ„', 'ðŸ‹'), + ('ð¼€', 'ð¼ž'), ('ðž„€', '𞄬'), ('ðž„·', '𞄽'), ('ðž…Ž', 'ðž…Ž'), + ('ðžŠ', 'ðžŠ'), ('ðž‹€', 'ðž‹«'), + ('𞟠', '𞟦'), + ('𞟨', '𞟫'), + ('ðžŸ', '𞟮'), + ('𞟰', '𞟾'), ('ðž €', '𞣄'), ('𞤀', '𞥃'), ('𞥋', '𞥋'), @@ -10943,11 +11245,11 @@ pub const XID_START: &'static [(char, char)] = &[ ('𞺡', '𞺣'), ('𞺥', '𞺩'), ('𞺫', '𞺻'), - ('ð €€', '\u{2a6dd}'), - ('𪜀', '𫜴'), + ('ð €€', '𪛟'), + ('𪜀', '𫜸'), ('ð«€', 'ð« '), ('ð« ', '𬺡'), ('𬺰', '𮯠'), ('丽', 'ð¯¨'), - ('\u{30000}', '\u{3134a}'), + ('ð°€€', 'ð±Š'), ]; diff --git a/vendor/regex-syntax/src/unicode_tables/property_names.rs b/vendor/regex-syntax/src/unicode_tables/property_names.rs index 6393df2f8..3fce1bf03 100644 --- a/vendor/regex-syntax/src/unicode_tables/property_names.rs +++ b/vendor/regex-syntax/src/unicode_tables/property_names.rs @@ -1,10 +1,10 @@ // DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY: // -// ucd-generate property-names ucd-13.0.0 +// ucd-generate property-names /tmp/ucd // -// Unicode version: 13.0.0. +// Unicode version: 14.0.0. // -// ucd-generate 0.2.8 is available on crates.io. +// ucd-generate 0.2.11 is available on crates.io. pub const PROPERTY_NAMES: &'static [(&'static str, &'static str)] = &[ ("age", "Age"), diff --git a/vendor/regex-syntax/src/unicode_tables/property_values.rs b/vendor/regex-syntax/src/unicode_tables/property_values.rs index c46653a7b..08a91477b 100644 --- a/vendor/regex-syntax/src/unicode_tables/property_values.rs +++ b/vendor/regex-syntax/src/unicode_tables/property_values.rs @@ -1,10 +1,10 @@ // DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY: // -// ucd-generate property-values ucd-13.0.0 --include gc,script,scx,age,gcb,wb,sb +// ucd-generate property-values /tmp/ucd --include gc,script,scx,age,gcb,wb,sb // -// Unicode version: 13.0.0. +// Unicode version: 14.0.0. // -// ucd-generate 0.2.8 is available on crates.io. +// ucd-generate 0.2.11 is available on crates.io. pub const PROPERTY_VALUES: &'static [( &'static str, @@ -19,6 +19,7 @@ pub const PROPERTY_VALUES: &'static [( ("12.0", "V12_0"), ("12.1", "V12_1"), ("13.0", "V13_0"), + ("14.0", "V14_0"), ("2.0", "V2_0"), ("2.1", "V2_1"), ("3.0", "V3_0"), @@ -44,6 +45,7 @@ pub const PROPERTY_VALUES: &'static [( ("v120", "V12_0"), ("v121", "V12_1"), ("v130", "V13_0"), + ("v140", "V14_0"), ("v20", "V2_0"), ("v21", "V2_1"), ("v30", "V3_0"), @@ -233,9 +235,11 @@ pub const PROPERTY_VALUES: &'static [( ("common", "Common"), ("copt", "Coptic"), ("coptic", "Coptic"), + ("cpmn", "Cypro_Minoan"), ("cprt", "Cypriot"), ("cuneiform", "Cuneiform"), ("cypriot", "Cypriot"), + ("cyprominoan", "Cypro_Minoan"), ("cyrillic", "Cyrillic"), ("cyrl", "Cyrillic"), ("deseret", "Deseret"), @@ -391,6 +395,7 @@ pub const PROPERTY_VALUES: &'static [( ("oldsogdian", "Old_Sogdian"), ("oldsoutharabian", "Old_South_Arabian"), ("oldturkic", "Old_Turkic"), + ("olduyghur", "Old_Uyghur"), ("oriya", "Oriya"), ("orkh", "Old_Turkic"), ("orya", "Oriya"), @@ -398,6 +403,7 @@ pub const PROPERTY_VALUES: &'static [( ("osge", "Osage"), ("osma", "Osmanya"), ("osmanya", "Osmanya"), + ("ougr", "Old_Uyghur"), ("pahawhhmong", "Pahawh_Hmong"), ("palm", "Palmyrene"), ("palmyrene", "Palmyrene"), @@ -462,6 +468,7 @@ pub const PROPERTY_VALUES: &'static [( ("tamil", "Tamil"), ("taml", "Tamil"), ("tang", "Tangut"), + ("tangsa", "Tangsa"), ("tangut", "Tangut"), ("tavt", "Tai_Viet"), ("telu", "Telugu"), @@ -476,11 +483,15 @@ pub const PROPERTY_VALUES: &'static [( ("tifinagh", "Tifinagh"), ("tirh", "Tirhuta"), ("tirhuta", "Tirhuta"), + ("tnsa", "Tangsa"), + ("toto", "Toto"), ("ugar", "Ugaritic"), ("ugaritic", "Ugaritic"), ("unknown", "Unknown"), ("vai", "Vai"), ("vaii", "Vai"), + ("vith", "Vithkuqi"), + ("vithkuqi", "Vithkuqi"), ("wancho", "Wancho"), ("wara", "Warang_Citi"), ("warangciti", "Warang_Citi"), @@ -550,9 +561,11 @@ pub const PROPERTY_VALUES: &'static [( ("common", "Common"), ("copt", "Coptic"), ("coptic", "Coptic"), + ("cpmn", "Cypro_Minoan"), ("cprt", "Cypriot"), ("cuneiform", "Cuneiform"), ("cypriot", "Cypriot"), + ("cyprominoan", "Cypro_Minoan"), ("cyrillic", "Cyrillic"), ("cyrl", "Cyrillic"), ("deseret", "Deseret"), @@ -708,6 +721,7 @@ pub const PROPERTY_VALUES: &'static [( ("oldsogdian", "Old_Sogdian"), ("oldsoutharabian", "Old_South_Arabian"), ("oldturkic", "Old_Turkic"), + ("olduyghur", "Old_Uyghur"), ("oriya", "Oriya"), ("orkh", "Old_Turkic"), ("orya", "Oriya"), @@ -715,6 +729,7 @@ pub const PROPERTY_VALUES: &'static [( ("osge", "Osage"), ("osma", "Osmanya"), ("osmanya", "Osmanya"), + ("ougr", "Old_Uyghur"), ("pahawhhmong", "Pahawh_Hmong"), ("palm", "Palmyrene"), ("palmyrene", "Palmyrene"), @@ -779,6 +794,7 @@ pub const PROPERTY_VALUES: &'static [( ("tamil", "Tamil"), ("taml", "Tamil"), ("tang", "Tangut"), + ("tangsa", "Tangsa"), ("tangut", "Tangut"), ("tavt", "Tai_Viet"), ("telu", "Telugu"), @@ -793,11 +809,15 @@ pub const PROPERTY_VALUES: &'static [( ("tifinagh", "Tifinagh"), ("tirh", "Tirhuta"), ("tirhuta", "Tirhuta"), + ("tnsa", "Tangsa"), + ("toto", "Toto"), ("ugar", "Ugaritic"), ("ugaritic", "Ugaritic"), ("unknown", "Unknown"), ("vai", "Vai"), ("vaii", "Vai"), + ("vith", "Vithkuqi"), + ("vithkuqi", "Vithkuqi"), ("wancho", "Wancho"), ("wara", "Warang_Citi"), ("warangciti", "Warang_Citi"), diff --git a/vendor/regex-syntax/src/unicode_tables/script.rs b/vendor/regex-syntax/src/unicode_tables/script.rs index cd86cba0d..3327b76ae 100644 --- a/vendor/regex-syntax/src/unicode_tables/script.rs +++ b/vendor/regex-syntax/src/unicode_tables/script.rs @@ -1,10 +1,10 @@ // DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY: // -// ucd-generate script ucd-13.0.0 --chars +// ucd-generate script /tmp/ucd --chars // -// Unicode version: 13.0.0. +// Unicode version: 14.0.0. // -// ucd-generate 0.2.8 is available on crates.io. +// ucd-generate 0.2.11 is available on crates.io. pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("Adlam", ADLAM), @@ -35,6 +35,7 @@ pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("Coptic", COPTIC), ("Cuneiform", CUNEIFORM), ("Cypriot", CYPRIOT), + ("Cypro_Minoan", CYPRO_MINOAN), ("Cyrillic", CYRILLIC), ("Deseret", DESERET), ("Devanagari", DEVANAGARI), @@ -118,6 +119,7 @@ pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("Old_Sogdian", OLD_SOGDIAN), ("Old_South_Arabian", OLD_SOUTH_ARABIAN), ("Old_Turkic", OLD_TURKIC), + ("Old_Uyghur", OLD_UYGHUR), ("Oriya", ORIYA), ("Osage", OSAGE), ("Osmanya", OSMANYA), @@ -149,6 +151,7 @@ pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("Tai_Viet", TAI_VIET), ("Takri", TAKRI), ("Tamil", TAMIL), + ("Tangsa", TANGSA), ("Tangut", TANGUT), ("Telugu", TELUGU), ("Thaana", THAANA), @@ -156,8 +159,10 @@ pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("Tibetan", TIBETAN), ("Tifinagh", TIFINAGH), ("Tirhuta", TIRHUTA), + ("Toto", TOTO), ("Ugaritic", UGARITIC), ("Vai", VAI), + ("Vithkuqi", VITHKUQI), ("Wancho", WANCHO), ("Warang_Citi", WARANG_CITI), ("Yezidi", YEZIDI), @@ -169,7 +174,7 @@ pub const ADLAM: &'static [(char, char)] = &[('𞤀', '𞥋'), ('ðž¥', '𞥙'), ('𞥞', '𞥟')]; pub const AHOM: &'static [(char, char)] = - &[('𑜀', '𑜚'), ('\u{1171d}', '\u{1172b}'), ('𑜰', '𑜿')]; + &[('𑜀', '𑜚'), ('\u{1171d}', '\u{1172b}'), ('𑜰', 'ð‘†')]; pub const ANATOLIAN_HIEROGLYPHS: &'static [(char, char)] = &[('ð”€', '𔙆')]; @@ -177,23 +182,23 @@ pub const ARABIC: &'static [(char, char)] = &[ ('\u{600}', '\u{604}'), ('؆', 'Ø‹'), ('Ø', '\u{61a}'), - ('\u{61c}', '\u{61c}'), - ('Øž', 'Øž'), + ('\u{61c}', 'Øž'), ('Ø ', 'Ø¿'), ('Ù', 'ÙŠ'), ('\u{656}', 'Ù¯'), ('Ù±', '\u{6dc}'), ('Ûž', 'Û¿'), ('Ý', 'Ý¿'), - ('ࢠ', 'ࢴ'), - ('ࢶ', '\u{8c7}'), - ('\u{8d3}', '\u{8e1}'), + ('à¡°', 'ࢎ'), + ('\u{890}', '\u{891}'), + ('\u{898}', '\u{8e1}'), ('\u{8e3}', '\u{8ff}'), - ('ï', 'ï¯'), + ('ï', '﯂'), ('ﯓ', 'ï´½'), - ('ïµ', 'ï¶'), + ('ïµ€', 'ï¶'), ('ﶒ', 'ï·‡'), - ('ï·°', 'ï·½'), + ('ï·', 'ï·'), + ('ï·°', 'ï·¿'), ('ï¹°', 'ï¹´'), ('ﹶ', 'ﻼ'), ('ð¹ ', 'ð¹¾'), @@ -238,7 +243,7 @@ pub const ARMENIAN: &'static [(char, char)] = pub const AVESTAN: &'static [(char, char)] = &[('ð¬€', 'ð¬µ'), ('ð¬¹', 'ð¬¿')]; -pub const BALINESE: &'static [(char, char)] = &[('\u{1b00}', 'á‹'), ('á', 'á¼')]; +pub const BALINESE: &'static [(char, char)] = &[('\u{1b00}', 'áŒ'), ('á', 'á¾')]; pub const BAMUM: &'static [(char, char)] = &[('êš ', 'ê›·'), ('ð– €', '𖨸')]; @@ -268,10 +273,10 @@ pub const BHAIKSUKI: &'static [(char, char)] = &[('ð‘°€', 'ð‘°ˆ'), ('ð‘°Š', '\u{11c36}'), ('\u{11c38}', '𑱅'), ('ð‘±', '𑱬')]; pub const BOPOMOFO: &'static [(char, char)] = - &[('˪', 'Ë«'), ('ã„…', 'ㄯ'), ('ㆠ', '\u{31bf}')]; + &[('˪', 'Ë«'), ('ã„…', 'ㄯ'), ('ㆠ', 'ㆿ')]; pub const BRAHMI: &'static [(char, char)] = - &[('ð‘€€', 'ð‘'), ('ð‘’', 'ð‘¯'), ('\u{1107f}', '\u{1107f}')]; + &[('ð‘€€', 'ð‘'), ('ð‘’', 'ð‘µ'), ('\u{1107f}', '\u{1107f}')]; pub const BRAILLE: &'static [(char, char)] = &[('â €', '⣿')]; @@ -280,7 +285,7 @@ pub const BUGINESE: &'static [(char, char)] = &[('ᨀ', '\u{1a1b}'), ('᨞', 'ᨠpub const BUHID: &'static [(char, char)] = &[('á€', '\u{1753}')]; pub const CANADIAN_ABORIGINAL: &'static [(char, char)] = - &[('á€', 'ᙿ'), ('ᢰ', 'ᣵ')]; + &[('á€', 'ᙿ'), ('ᢰ', 'ᣵ'), ('𑪰', '𑪿')]; pub const CARIAN: &'static [(char, char)] = &[('ðŠ ', 'ð‹')]; @@ -288,7 +293,7 @@ pub const CAUCASIAN_ALBANIAN: &'static [(char, char)] = &[('ð”°', 'ð•£'), ('ð•¯', 'ð•¯')]; pub const CHAKMA: &'static [(char, char)] = - &[('\u{11100}', '\u{11134}'), ('𑄶', '\u{11147}')]; + &[('\u{11100}', '\u{11134}'), ('𑄶', 'ð‘…‡')]; pub const CHAM: &'static [(char, char)] = &[('ꨀ', '\u{aa36}'), ('ê©€', 'ê©'), ('ê©', 'ê©™'), ('ê©œ', 'ê©Ÿ')]; @@ -296,10 +301,10 @@ pub const CHAM: &'static [(char, char)] = pub const CHEROKEE: &'static [(char, char)] = &[('Ꭰ', 'áµ'), ('á¸', 'á½'), ('ê°', 'ꮿ')]; -pub const CHORASMIAN: &'static [(char, char)] = &[('\u{10fb0}', '\u{10fcb}')]; +pub const CHORASMIAN: &'static [(char, char)] = &[('ð¾°', 'ð¿‹')]; pub const COMMON: &'static [(char, char)] = &[ - ('\u{0}', '@'), + ('\0', '@'), ('[', '`'), ('{', '©'), ('«', '¹'), @@ -339,7 +344,7 @@ pub const COMMON: &'static [(char, char)] = &[ ('\u{2066}', 'â°'), ('â´', 'â¾'), ('â‚€', 'â‚Ž'), - ('â‚ ', 'â‚¿'), + ('â‚ ', '⃀'), ('â„€', 'â„¥'), ('℧', 'â„©'), ('ℬ', 'ℱ'), @@ -351,8 +356,8 @@ pub const COMMON: &'static [(char, char)] = &[ ('â‘ ', '⟿'), ('⤀', 'â³'), ('â¶', '⮕'), - ('\u{2b97}', '⯿'), - ('⸀', '\u{2e52}'), + ('â®—', '⯿'), + ('⸀', 'â¹'), ('â¿°', 'â¿»'), ('\u{3000}', '〄'), ('〆', '〆'), @@ -375,7 +380,7 @@ pub const COMMON: &'static [(char, char)] = &[ ('꤮', '꤮'), ('ê§', 'ê§'), ('ê›', 'ê›'), - ('\u{ab6a}', '\u{ab6b}'), + ('êª', 'ê«'), ('ï´¾', 'ï´¿'), ('ï¸', '︙'), ('︰', 'ï¹’'), @@ -393,18 +398,18 @@ pub const COMMON: &'static [(char, char)] = &[ ('ð„€', 'ð„‚'), ('ð„‡', 'ð„³'), ('ð„·', 'ð„¿'), - ('ð†', '\u{1019c}'), + ('ð†', 'ð†œ'), ('ð‡', 'ð‡¼'), ('ð‹¡', 'ð‹»'), - ('ð–¿¢', 'ð–¿£'), ('\u{1bca0}', '\u{1bca3}'), + ('ðœ½', '𜿃'), ('ð€€', 'ðƒµ'), ('ð„€', 'ð„¦'), ('ð„©', 'ð…¦'), ('ð…ª', '\u{1d17a}'), ('ð†ƒ', 'ð†„'), ('ð†Œ', 'ð†©'), - ('ð†®', 'ð‡¨'), + ('ð†®', 'ð‡ª'), ('ð‹ ', 'ð‹³'), ('ðŒ€', 'ð–'), ('ð ', 'ð¸'), @@ -437,39 +442,40 @@ pub const COMMON: &'static [(char, char)] = &[ ('🂱', '🂿'), ('ðŸƒ', 'ðŸƒ'), ('🃑', '🃵'), - ('🄀', '\u{1f1ad}'), + ('🄀', 'ðŸ†'), ('🇦', '🇿'), ('ðŸˆ', '🈂'), ('ðŸˆ', '🈻'), ('🉀', '🉈'), ('ðŸ‰', '🉑'), ('🉠', '🉥'), - ('🌀', '\u{1f6d7}'), - ('🛠', '🛬'), - ('🛰', '\u{1f6fc}'), + ('🌀', '🛗'), + ('ðŸ›', '🛬'), + ('🛰', '🛼'), ('🜀', 'ðŸ³'), ('🞀', '🟘'), ('🟠', '🟫'), + ('🟰', '🟰'), ('🠀', '🠋'), ('ðŸ ', '🡇'), ('ðŸ¡', '🡙'), ('🡠', '🢇'), ('ðŸ¢', 'ðŸ¢'), - ('\u{1f8b0}', '\u{1f8b1}'), - ('🤀', '\u{1f978}'), - ('🥺', '\u{1f9cb}'), - ('ðŸ§', '🩓'), + ('🢰', '🢱'), + ('🤀', '🩓'), ('🩠', 'ðŸ©'), - ('🩰', '\u{1fa74}'), - ('🩸', '🩺'), - ('🪀', '\u{1fa86}'), - ('ðŸª', '\u{1faa8}'), - ('\u{1fab0}', '\u{1fab6}'), - ('\u{1fac0}', '\u{1fac2}'), - ('\u{1fad0}', '\u{1fad6}'), - ('\u{1fb00}', '\u{1fb92}'), - ('\u{1fb94}', '\u{1fbca}'), - ('\u{1fbf0}', '\u{1fbf9}'), + ('🩰', '🩴'), + ('🩸', '🩼'), + ('🪀', '🪆'), + ('ðŸª', '🪬'), + ('🪰', '🪺'), + ('🫀', '🫅'), + ('ðŸ«', '🫙'), + ('🫠', '🫧'), + ('🫰', '🫶'), + ('🬀', '🮒'), + ('🮔', '🯊'), + ('🯰', '🯹'), ('\u{e0001}', '\u{e0001}'), ('\u{e0020}', '\u{e007f}'), ]; @@ -483,6 +489,8 @@ pub const CUNEIFORM: &'static [(char, char)] = pub const CYPRIOT: &'static [(char, char)] = &[('ð €', 'ð …'), ('ð ˆ', 'ð ˆ'), ('ð Š', 'ð µ'), ('ð ·', 'ð ¸'), ('ð ¼', 'ð ¼'), ('ð ¿', 'ð ¿')]; +pub const CYPRO_MINOAN: &'static [(char, char)] = &[('ð’¾', 'ð’¿²')]; + pub const CYRILLIC: &'static [(char, char)] = &[ ('Ѐ', '\u{484}'), ('\u{487}', 'Ô¯'), @@ -504,14 +512,14 @@ pub const DEVANAGARI: &'static [(char, char)] = &[ ]; pub const DIVES_AKURU: &'static [(char, char)] = &[ - ('\u{11900}', '\u{11906}'), - ('\u{11909}', '\u{11909}'), - ('\u{1190c}', '\u{11913}'), - ('\u{11915}', '\u{11916}'), - ('\u{11918}', '\u{11935}'), - ('\u{11937}', '\u{11938}'), - ('\u{1193b}', '\u{11946}'), - ('\u{11950}', '\u{11959}'), + ('𑤀', '𑤆'), + ('𑤉', '𑤉'), + ('𑤌', '𑤓'), + ('𑤕', '𑤖'), + ('𑤘', '𑤵'), + ('𑤷', '𑤸'), + ('\u{1193b}', '𑥆'), + ('ð‘¥', 'ð‘¥™'), ]; pub const DOGRA: &'static [(char, char)] = &[('ð‘ €', 'ð‘ »')]; @@ -559,6 +567,10 @@ pub const ETHIOPIC: &'static [(char, char)] = &[ ('ꬑ', 'ꬖ'), ('ê¬ ', 'ꬦ'), ('ꬨ', 'ꬮ'), + ('𞟠', '𞟦'), + ('𞟨', '𞟫'), + ('ðžŸ', '𞟮'), + ('𞟰', '𞟾'), ]; pub const GEORGIAN: &'static [(char, char)] = &[ @@ -575,8 +587,7 @@ pub const GEORGIAN: &'static [(char, char)] = &[ ]; pub const GLAGOLITIC: &'static [(char, char)] = &[ - ('â°€', 'â°®'), - ('â°°', 'ⱞ'), + ('â°€', 'ⱟ'), ('\u{1e000}', '\u{1e006}'), ('\u{1e008}', '\u{1e018}'), ('\u{1e01b}', '\u{1e021}'), @@ -696,18 +707,19 @@ pub const HAN: &'static [(char, char)] = &[ ('〇', '〇'), ('〡', '〩'), ('〸', '〻'), - ('ã€', '\u{4dbf}'), - ('一', '\u{9ffc}'), + ('ã€', '䶿'), + ('一', 'é¿¿'), ('豈', 'ï©'), ('ï©°', 'ï«™'), - ('\u{16ff0}', '\u{16ff1}'), - ('ð €€', '\u{2a6dd}'), - ('𪜀', '𫜴'), + ('ð–¿¢', 'ð–¿£'), + ('ð–¿°', 'ð–¿±'), + ('ð €€', '𪛟'), + ('𪜀', '𫜸'), ('ð«€', 'ð« '), ('ð« ', '𬺡'), ('𬺰', '𮯠'), ('丽', 'ð¯¨'), - ('\u{30000}', '\u{3134a}'), + ('ð°€€', 'ð±Š'), ]; pub const HANGUL: &'static [(char, char)] = &[ @@ -730,7 +742,7 @@ pub const HANGUL: &'static [(char, char)] = &[ pub const HANIFI_ROHINGYA: &'static [(char, char)] = &[('ð´€', '\u{10d27}'), ('ð´°', 'ð´¹')]; -pub const HANUNOO: &'static [(char, char)] = &[('ᜠ', '\u{1734}')]; +pub const HANUNOO: &'static [(char, char)] = &[('ᜠ', '᜴')]; pub const HATRAN: &'static [(char, char)] = &[('ð£ ', 'ð£²'), ('ð£´', 'ð£µ'), ('ð£»', 'ð£¿')]; @@ -748,7 +760,7 @@ pub const HEBREW: &'static [(char, char)] = &[ ]; pub const HIRAGANA: &'static [(char, char)] = - &[('ã', 'ã‚–'), ('ã‚', 'ã‚Ÿ'), ('ð›€', '𛄞'), ('ð›…', 'ð›…’'), ('🈀', '🈀')]; + &[('ã', 'ã‚–'), ('ã‚', 'ã‚Ÿ'), ('ð›€', '𛄟'), ('ð›…', 'ð›…’'), ('🈀', '🈀')]; pub const IMPERIAL_ARAMAIC: &'static [(char, char)] = &[('ð¡€', 'ð¡•'), ('ð¡—', 'ð¡Ÿ')]; @@ -759,15 +771,14 @@ pub const INHERITED: &'static [(char, char)] = &[ ('\u{64b}', '\u{655}'), ('\u{670}', '\u{670}'), ('\u{951}', '\u{954}'), - ('\u{1ab0}', '\u{1ac0}'), + ('\u{1ab0}', '\u{1ace}'), ('\u{1cd0}', '\u{1cd2}'), ('\u{1cd4}', '\u{1ce0}'), ('\u{1ce2}', '\u{1ce8}'), ('\u{1ced}', '\u{1ced}'), ('\u{1cf4}', '\u{1cf4}'), ('\u{1cf8}', '\u{1cf9}'), - ('\u{1dc0}', '\u{1df9}'), - ('\u{1dfb}', '\u{1dff}'), + ('\u{1dc0}', '\u{1dff}'), ('\u{200c}', '\u{200d}'), ('\u{20d0}', '\u{20f0}'), ('\u{302a}', '\u{302d}'), @@ -777,6 +788,8 @@ pub const INHERITED: &'static [(char, char)] = &[ ('\u{101fd}', '\u{101fd}'), ('\u{102e0}', '\u{102e0}'), ('\u{1133b}', '\u{1133b}'), + ('\u{1cf00}', '\u{1cf2d}'), + ('\u{1cf30}', '\u{1cf46}'), ('\u{1d167}', '\u{1d169}'), ('\u{1d17b}', '\u{1d182}'), ('\u{1d185}', '\u{1d18b}'), @@ -794,7 +807,7 @@ pub const JAVANESE: &'static [(char, char)] = &[('\u{a980}', 'ê§'), ('ê§', '꧙'), ('꧞', '꧟')]; pub const KAITHI: &'static [(char, char)] = - &[('\u{11080}', 'ð‘ƒ'), ('\u{110cd}', '\u{110cd}')]; + &[('\u{11080}', '\u{110c2}'), ('\u{110cd}', '\u{110cd}')]; pub const KANNADA: &'static [(char, char)] = &[ ('ಀ', 'ಌ'), @@ -806,7 +819,7 @@ pub const KANNADA: &'static [(char, char)] = &[ ('\u{cc6}', 'ೈ'), ('ೊ', '\u{ccd}'), ('\u{cd5}', '\u{cd6}'), - ('ೞ', 'ೞ'), + ('à³', 'ೞ'), ('à³ ', '\u{ce3}'), ('೦', '೯'), ('à³±', 'à³²'), @@ -820,7 +833,11 @@ pub const KATAKANA: &'static [(char, char)] = &[ ('㌀', 'ã—'), ('ヲ', 'ッ'), ('ï½±', 'ï¾'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), ('𛀀', '𛀀'), + ('𛄠', '𛄢'), ('ð›…¤', 'ð›…§'), ]; @@ -838,7 +855,7 @@ pub const KHAROSHTHI: &'static [(char, char)] = &[ ]; pub const KHITAN_SMALL_SCRIPT: &'static [(char, char)] = - &[('\u{16fe4}', '\u{16fe4}'), ('\u{18b00}', '\u{18cd5}')]; + &[('\u{16fe4}', '\u{16fe4}'), ('𘬀', '𘳕')]; pub const KHMER: &'static [(char, char)] = &[('ក', '\u{17dd}'), ('០', '៩'), ('៰', '៹'), ('᧠', '᧿')]; @@ -886,15 +903,21 @@ pub const LATIN: &'static [(char, char)] = &[ ('â… ', 'ↈ'), ('â± ', 'Ɀ'), ('Ꜣ', 'ꞇ'), - ('êž‹', 'êž¿'), - ('Ꟃ', '\u{a7ca}'), - ('\u{a7f5}', 'ꟿ'), + ('êž‹', 'ꟊ'), + ('êŸ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟙ'), + ('ꟲ', 'ꟿ'), ('ꬰ', 'êš'), ('êœ', 'ê¤'), - ('ê¦', '\u{ab69}'), + ('ê¦', 'ê©'), ('ff', 'st'), ('A', 'Z'), ('ï½', 'z'), + ('ðž€', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), + ('ð¼€', 'ð¼ž'), ]; pub const LEPCHA: &'static [(char, char)] = @@ -921,8 +944,7 @@ pub const LINEAR_B: &'static [(char, char)] = &[ ('ð‚€', 'ðƒº'), ]; -pub const LISU: &'static [(char, char)] = - &[('ê“', 'ê“¿'), ('\u{11fb0}', '\u{11fb0}')]; +pub const LISU: &'static [(char, char)] = &[('ê“', 'ê“¿'), ('𑾰', '𑾰')]; pub const LYCIAN: &'static [(char, char)] = &[('ðŠ€', 'ðŠœ')]; @@ -978,15 +1000,8 @@ pub const MIAO: &'static [(char, char)] = pub const MODI: &'static [(char, char)] = &[('𑘀', 'ð‘™„'), ('ð‘™', 'ð‘™™')]; -pub const MONGOLIAN: &'static [(char, char)] = &[ - ('á €', 'á '), - ('á „', 'á „'), - ('á †', '\u{180e}'), - ('á ', 'á ™'), - ('á ', 'ᡸ'), - ('ᢀ', 'ᢪ'), - ('ð‘™ ', '𑙬'), -]; +pub const MONGOLIAN: &'static [(char, char)] = + &[('á €', 'á '), ('á „', 'á „'), ('á †', 'á ™'), ('á ', 'ᡸ'), ('ᢀ', 'ᢪ'), ('ð‘™ ', '𑙬')]; pub const MRO: &'static [(char, char)] = &[('ð–©€', 'ð–©ž'), ('ð–© ', 'ð–©©'), ('ð–©®', '𖩯')]; @@ -1004,7 +1019,7 @@ pub const NANDINAGARI: &'static [(char, char)] = pub const NEW_TAI_LUE: &'static [(char, char)] = &[('ᦀ', 'ᦫ'), ('ᦰ', 'ᧉ'), ('á§', '᧚'), ('᧞', '᧟')]; -pub const NEWA: &'static [(char, char)] = &[('ð‘€', 'ð‘‘›'), ('ð‘‘', '\u{11461}')]; +pub const NEWA: &'static [(char, char)] = &[('ð‘€', 'ð‘‘›'), ('ð‘‘', 'ð‘‘¡')]; pub const NKO: &'static [(char, char)] = &[('߀', 'ߺ'), ('\u{7fd}', 'ß¿')]; @@ -1034,6 +1049,8 @@ pub const OLD_SOUTH_ARABIAN: &'static [(char, char)] = &[('ð© ', 'ð©¿')]; pub const OLD_TURKIC: &'static [(char, char)] = &[('ð°€', 'ð±ˆ')]; +pub const OLD_UYGHUR: &'static [(char, char)] = &[('ð½°', 'ð¾‰')]; + pub const ORIYA: &'static [(char, char)] = &[ ('\u{b01}', 'ଃ'), ('ଅ', 'ଌ'), @@ -1118,7 +1135,7 @@ pub const SYLOTI_NAGRI: &'static [(char, char)] = &[('ê €', '\u{a82c}')]; pub const SYRIAC: &'static [(char, char)] = &[('Ü€', 'Ü'), ('\u{70f}', '\u{74a}'), ('Ý', 'Ý'), ('à¡ ', 'ࡪ')]; -pub const TAGALOG: &'static [(char, char)] = &[('ᜀ', 'ᜌ'), ('ᜎ', '\u{1714}')]; +pub const TAGALOG: &'static [(char, char)] = &[('ᜀ', '᜕'), ('ᜟ', 'ᜟ')]; pub const TAGBANWA: &'static [(char, char)] = &[('á ', 'á¬'), ('á®', 'á°'), ('\u{1772}', '\u{1773}')]; @@ -1135,7 +1152,7 @@ pub const TAI_THAM: &'static [(char, char)] = &[ pub const TAI_VIET: &'static [(char, char)] = &[('ꪀ', 'ê«‚'), ('ê«›', 'ê«Ÿ')]; -pub const TAKRI: &'static [(char, char)] = &[('𑚀', '𑚸'), ('𑛀', '𑛉')]; +pub const TAKRI: &'static [(char, char)] = &[('𑚀', 'ð‘š¹'), ('𑛀', '𑛉')]; pub const TAMIL: &'static [(char, char)] = &[ ('\u{b82}', 'ஃ'), @@ -1158,23 +1175,22 @@ pub const TAMIL: &'static [(char, char)] = &[ ('ð‘¿¿', 'ð‘¿¿'), ]; -pub const TANGUT: &'static [(char, char)] = &[ - ('ð–¿ ', 'ð–¿ '), - ('ð—€€', '𘟷'), - ('𘠀', '\u{18aff}'), - ('\u{18d00}', '\u{18d08}'), -]; +pub const TANGSA: &'static [(char, char)] = &[('ð–©°', '𖪾'), ('ð–«€', '𖫉')]; + +pub const TANGUT: &'static [(char, char)] = + &[('ð–¿ ', 'ð–¿ '), ('ð—€€', '𘟷'), ('𘠀', '𘫿'), ('𘴀', '𘴈')]; pub const TELUGU: &'static [(char, char)] = &[ ('\u{c00}', 'à°Œ'), ('à°Ž', 'à°'), ('à°’', 'à°¨'), ('à°ª', 'à°¹'), - ('à°½', 'ౄ'), + ('\u{c3c}', 'ౄ'), ('\u{c46}', '\u{c48}'), ('\u{c4a}', '\u{c4d}'), ('\u{c55}', '\u{c56}'), ('ౘ', 'ౚ'), + ('à±', 'à±'), ('à± ', '\u{c63}'), ('౦', '౯'), ('à±·', '౿'), @@ -1199,19 +1215,29 @@ pub const TIFINAGH: &'static [(char, char)] = pub const TIRHUTA: &'static [(char, char)] = &[('ð‘’€', '𑓇'), ('ð‘“', 'ð‘“™')]; +pub const TOTO: &'static [(char, char)] = &[('ðžŠ', '\u{1e2ae}')]; + pub const UGARITIC: &'static [(char, char)] = &[('ðŽ€', 'ðŽ'), ('ðŽŸ', 'ðŽŸ')]; pub const VAI: &'static [(char, char)] = &[('ꔀ', 'ꘫ')]; +pub const VITHKUQI: &'static [(char, char)] = &[ + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), +]; + pub const WANCHO: &'static [(char, char)] = &[('ðž‹€', '𞋹'), ('ðž‹¿', 'ðž‹¿')]; pub const WARANG_CITI: &'static [(char, char)] = &[('ð‘¢ ', 'ð‘£²'), ('𑣿', '𑣿')]; -pub const YEZIDI: &'static [(char, char)] = &[ - ('\u{10e80}', '\u{10ea9}'), - ('\u{10eab}', '\u{10ead}'), - ('\u{10eb0}', '\u{10eb1}'), -]; +pub const YEZIDI: &'static [(char, char)] = + &[('ðº€', 'ðº©'), ('\u{10eab}', 'ðº'), ('ðº°', 'ðº±')]; pub const YI: &'static [(char, char)] = &[('ꀀ', 'ê’Œ'), ('ê’', '꓆')]; diff --git a/vendor/regex-syntax/src/unicode_tables/script_extension.rs b/vendor/regex-syntax/src/unicode_tables/script_extension.rs index 7fca2af9d..c970e0104 100644 --- a/vendor/regex-syntax/src/unicode_tables/script_extension.rs +++ b/vendor/regex-syntax/src/unicode_tables/script_extension.rs @@ -1,10 +1,10 @@ // DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY: // -// ucd-generate script-extension ucd-13.0.0 --chars +// ucd-generate script-extension /tmp/ucd --chars // -// Unicode version: 13.0.0. +// Unicode version: 14.0.0. // -// ucd-generate 0.2.8 is available on crates.io. +// ucd-generate 0.2.11 is available on crates.io. pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("Adlam", ADLAM), @@ -35,6 +35,7 @@ pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("Coptic", COPTIC), ("Cuneiform", CUNEIFORM), ("Cypriot", CYPRIOT), + ("Cypro_Minoan", CYPRO_MINOAN), ("Cyrillic", CYRILLIC), ("Deseret", DESERET), ("Devanagari", DEVANAGARI), @@ -118,6 +119,7 @@ pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("Old_Sogdian", OLD_SOGDIAN), ("Old_South_Arabian", OLD_SOUTH_ARABIAN), ("Old_Turkic", OLD_TURKIC), + ("Old_Uyghur", OLD_UYGHUR), ("Oriya", ORIYA), ("Osage", OSAGE), ("Osmanya", OSMANYA), @@ -149,6 +151,7 @@ pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("Tai_Viet", TAI_VIET), ("Takri", TAKRI), ("Tamil", TAMIL), + ("Tangsa", TANGSA), ("Tangut", TANGUT), ("Telugu", TELUGU), ("Thaana", THAANA), @@ -156,8 +159,10 @@ pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("Tibetan", TIBETAN), ("Tifinagh", TIFINAGH), ("Tirhuta", TIRHUTA), + ("Toto", TOTO), ("Ugaritic", UGARITIC), ("Vai", VAI), + ("Vithkuqi", VITHKUQI), ("Wancho", WANCHO), ("Warang_Citi", WARANG_CITI), ("Yezidi", YEZIDI), @@ -166,28 +171,27 @@ pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ]; pub const ADLAM: &'static [(char, char)] = - &[('Ù€', 'Ù€'), ('𞤀', '𞥋'), ('ðž¥', '𞥙'), ('𞥞', '𞥟')]; + &[('ØŸ', 'ØŸ'), ('Ù€', 'Ù€'), ('𞤀', '𞥋'), ('ðž¥', '𞥙'), ('𞥞', '𞥟')]; pub const AHOM: &'static [(char, char)] = - &[('𑜀', '𑜚'), ('\u{1171d}', '\u{1172b}'), ('𑜰', '𑜿')]; + &[('𑜀', '𑜚'), ('\u{1171d}', '\u{1172b}'), ('𑜰', 'ð‘†')]; pub const ANATOLIAN_HIEROGLYPHS: &'static [(char, char)] = &[('ð”€', '𔙆')]; pub const ARABIC: &'static [(char, char)] = &[ ('\u{600}', '\u{604}'), - ('؆', '\u{61c}'), - ('Øž', '\u{6dc}'), + ('؆', '\u{6dc}'), ('Ûž', 'Û¿'), ('Ý', 'Ý¿'), - ('ࢠ', 'ࢴ'), - ('ࢶ', '\u{8c7}'), - ('\u{8d3}', '\u{8e1}'), + ('à¡°', 'ࢎ'), + ('\u{890}', '\u{891}'), + ('\u{898}', '\u{8e1}'), ('\u{8e3}', '\u{8ff}'), - ('ï', 'ï¯'), - ('ﯓ', 'ï´½'), - ('ïµ', 'ï¶'), + ('ï', '﯂'), + ('ﯓ', 'ï¶'), ('ﶒ', 'ï·‡'), - ('ï·°', 'ï·½'), + ('ï·', 'ï·'), + ('ï·°', 'ï·¿'), ('ï¹°', 'ï¹´'), ('ﹶ', 'ﻼ'), ('\u{102e0}', 'ð‹»'), @@ -233,7 +237,7 @@ pub const ARMENIAN: &'static [(char, char)] = pub const AVESTAN: &'static [(char, char)] = &[('ð¬€', 'ð¬µ'), ('ð¬¹', 'ð¬¿')]; -pub const BALINESE: &'static [(char, char)] = &[('\u{1b00}', 'á‹'), ('á', 'á¼')]; +pub const BALINESE: &'static [(char, char)] = &[('\u{1b00}', 'áŒ'), ('á', 'á¾')]; pub const BAMUM: &'static [(char, char)] = &[('êš ', 'ê›·'), ('ð– €', '𖨸')]; @@ -284,13 +288,13 @@ pub const BOPOMOFO: &'static [(char, char)] = &[ ('〷', '〷'), ('・', '・'), ('ã„…', 'ㄯ'), - ('ㆠ', '\u{31bf}'), + ('ㆠ', 'ㆿ'), ('ï¹…', '﹆'), ('。', 'ï½¥'), ]; pub const BRAHMI: &'static [(char, char)] = - &[('ð‘€€', 'ð‘'), ('ð‘’', 'ð‘¯'), ('\u{1107f}', '\u{1107f}')]; + &[('ð‘€€', 'ð‘'), ('ð‘’', 'ð‘µ'), ('\u{1107f}', '\u{1107f}')]; pub const BRAILLE: &'static [(char, char)] = &[('â €', '⣿')]; @@ -300,7 +304,7 @@ pub const BUGINESE: &'static [(char, char)] = pub const BUHID: &'static [(char, char)] = &[('᜵', '᜶'), ('á€', '\u{1753}')]; pub const CANADIAN_ABORIGINAL: &'static [(char, char)] = - &[('á€', 'ᙿ'), ('ᢰ', 'ᣵ')]; + &[('á€', 'ᙿ'), ('ᢰ', 'ᣵ'), ('𑪰', '𑪿')]; pub const CARIAN: &'static [(char, char)] = &[('ðŠ ', 'ð‹')]; @@ -308,7 +312,7 @@ pub const CAUCASIAN_ALBANIAN: &'static [(char, char)] = &[('ð”°', 'ð•£'), ('ð•¯', 'ð•¯')]; pub const CHAKMA: &'static [(char, char)] = - &[('০', '৯'), ('á€', 'á‰'), ('\u{11100}', '\u{11134}'), ('𑄶', '\u{11147}')]; + &[('০', '৯'), ('á€', 'á‰'), ('\u{11100}', '\u{11134}'), ('𑄶', 'ð‘…‡')]; pub const CHAM: &'static [(char, char)] = &[('ꨀ', '\u{aa36}'), ('ê©€', 'ê©'), ('ê©', 'ê©™'), ('ê©œ', 'ê©Ÿ')]; @@ -316,10 +320,10 @@ pub const CHAM: &'static [(char, char)] = pub const CHEROKEE: &'static [(char, char)] = &[('Ꭰ', 'áµ'), ('á¸', 'á½'), ('ê°', 'ꮿ')]; -pub const CHORASMIAN: &'static [(char, char)] = &[('\u{10fb0}', '\u{10fcb}')]; +pub const CHORASMIAN: &'static [(char, char)] = &[('ð¾°', 'ð¿‹')]; pub const COMMON: &'static [(char, char)] = &[ - ('\u{0}', '@'), + ('\0', '@'), ('[', '`'), ('{', '©'), ('«', '¹'), @@ -345,7 +349,7 @@ pub const COMMON: &'static [(char, char)] = &[ ('\u{2066}', 'â°'), ('â´', 'â¾'), ('â‚€', 'â‚Ž'), - ('â‚ ', 'â‚¿'), + ('â‚ ', '⃀'), ('â„€', 'â„¥'), ('℧', 'â„©'), ('ℬ', 'ℱ'), @@ -357,9 +361,9 @@ pub const COMMON: &'static [(char, char)] = &[ ('â‘ ', '⟿'), ('⤀', 'â³'), ('â¶', '⮕'), - ('\u{2b97}', '⯿'), + ('â®—', '⯿'), ('⸀', '⹂'), - ('⹄', '\u{2e52}'), + ('⹄', 'â¹'), ('â¿°', 'â¿»'), ('\u{3000}', '\u{3000}'), ('〄', '〄'), @@ -377,8 +381,7 @@ pub const COMMON: &'static [(char, char)] = &[ ('꜈', '꜡'), ('ꞈ', '꞊'), ('ê›', 'ê›'), - ('\u{ab6a}', '\u{ab6b}'), - ('ï´¾', 'ï´¿'), + ('êª', 'ê«'), ('ï¸', '︙'), ('︰', '﹄'), ('﹇', 'ï¹’'), @@ -391,16 +394,16 @@ pub const COMMON: &'static [(char, char)] = &[ ('ï¿ ', '₩'), ('│', 'ï¿®'), ('\u{fff9}', '�'), - ('ð†', '\u{1019c}'), + ('ð†', 'ð†œ'), ('ð‡', 'ð‡¼'), - ('ð–¿¢', 'ð–¿£'), + ('ðœ½', '𜿃'), ('ð€€', 'ðƒµ'), ('ð„€', 'ð„¦'), ('ð„©', 'ð…¦'), ('ð…ª', '\u{1d17a}'), ('ð†ƒ', 'ð†„'), ('ð†Œ', 'ð†©'), - ('ð†®', 'ð‡¨'), + ('ð†®', 'ð‡ª'), ('ð‹ ', 'ð‹³'), ('ðŒ€', 'ð–'), ('ð²', 'ð¸'), @@ -433,38 +436,39 @@ pub const COMMON: &'static [(char, char)] = &[ ('🂱', '🂿'), ('ðŸƒ', 'ðŸƒ'), ('🃑', '🃵'), - ('🄀', '\u{1f1ad}'), + ('🄀', 'ðŸ†'), ('🇦', '🇿'), ('ðŸˆ', '🈂'), ('ðŸˆ', '🈻'), ('🉀', '🉈'), ('🉠', '🉥'), - ('🌀', '\u{1f6d7}'), - ('🛠', '🛬'), - ('🛰', '\u{1f6fc}'), + ('🌀', '🛗'), + ('ðŸ›', '🛬'), + ('🛰', '🛼'), ('🜀', 'ðŸ³'), ('🞀', '🟘'), ('🟠', '🟫'), + ('🟰', '🟰'), ('🠀', '🠋'), ('ðŸ ', '🡇'), ('ðŸ¡', '🡙'), ('🡠', '🢇'), ('ðŸ¢', 'ðŸ¢'), - ('\u{1f8b0}', '\u{1f8b1}'), - ('🤀', '\u{1f978}'), - ('🥺', '\u{1f9cb}'), - ('ðŸ§', '🩓'), + ('🢰', '🢱'), + ('🤀', '🩓'), ('🩠', 'ðŸ©'), - ('🩰', '\u{1fa74}'), - ('🩸', '🩺'), - ('🪀', '\u{1fa86}'), - ('ðŸª', '\u{1faa8}'), - ('\u{1fab0}', '\u{1fab6}'), - ('\u{1fac0}', '\u{1fac2}'), - ('\u{1fad0}', '\u{1fad6}'), - ('\u{1fb00}', '\u{1fb92}'), - ('\u{1fb94}', '\u{1fbca}'), - ('\u{1fbf0}', '\u{1fbf9}'), + ('🩰', '🩴'), + ('🩸', '🩼'), + ('🪀', '🪆'), + ('ðŸª', '🪬'), + ('🪰', '🪺'), + ('🫀', '🫅'), + ('ðŸ«', '🫙'), + ('🫠', '🫧'), + ('🫰', '🫶'), + ('🬀', '🮒'), + ('🮔', '🯊'), + ('🯰', '🯹'), ('\u{e0001}', '\u{e0001}'), ('\u{e0020}', '\u{e007f}'), ]; @@ -487,6 +491,8 @@ pub const CYPRIOT: &'static [(char, char)] = &[ ('ð ¿', 'ð ¿'), ]; +pub const CYPRO_MINOAN: &'static [(char, char)] = &[('ð„€', 'ð„'), ('ð’¾', 'ð’¿²')]; + pub const CYRILLIC: &'static [(char, char)] = &[ ('Ѐ', 'Ô¯'), ('á²€', 'ᲈ'), @@ -512,14 +518,14 @@ pub const DEVANAGARI: &'static [(char, char)] = &[ ]; pub const DIVES_AKURU: &'static [(char, char)] = &[ - ('\u{11900}', '\u{11906}'), - ('\u{11909}', '\u{11909}'), - ('\u{1190c}', '\u{11913}'), - ('\u{11915}', '\u{11916}'), - ('\u{11918}', '\u{11935}'), - ('\u{11937}', '\u{11938}'), - ('\u{1193b}', '\u{11946}'), - ('\u{11950}', '\u{11959}'), + ('𑤀', '𑤆'), + ('𑤉', '𑤉'), + ('𑤌', '𑤓'), + ('𑤕', '𑤖'), + ('𑤘', '𑤵'), + ('𑤷', '𑤸'), + ('\u{1193b}', '𑥆'), + ('ð‘¥', 'ð‘¥™'), ]; pub const DOGRA: &'static [(char, char)] = @@ -568,6 +574,10 @@ pub const ETHIOPIC: &'static [(char, char)] = &[ ('ꬑ', 'ꬖ'), ('ê¬ ', 'ꬦ'), ('ꬨ', 'ꬮ'), + ('𞟠', '𞟦'), + ('𞟨', '𞟫'), + ('ðžŸ', '𞟮'), + ('𞟰', '𞟾'), ]; pub const GEORGIAN: &'static [(char, char)] = &[ @@ -585,8 +595,7 @@ pub const GEORGIAN: &'static [(char, char)] = &[ pub const GLAGOLITIC: &'static [(char, char)] = &[ ('\u{484}', '\u{484}'), ('\u{487}', '\u{487}'), - ('â°€', 'â°®'), - ('â°°', 'ⱞ'), + ('â°€', 'ⱟ'), ('⹃', '⹃'), ('\u{a66f}', '\u{a66f}'), ('\u{1e000}', '\u{1e006}'), @@ -739,23 +748,24 @@ pub const HAN: &'static [(char, char)] = &[ ('ã˜', 'ã°'), ('ã»', 'ã¿'), ('ã ', 'ã¾'), - ('ã€', '\u{4dbf}'), - ('一', '\u{9ffc}'), + ('ã€', '䶿'), + ('一', 'é¿¿'), ('꜀', '꜇'), ('豈', 'ï©'), ('ï©°', 'ï«™'), ('ï¹…', '﹆'), ('。', 'ï½¥'), - ('\u{16ff0}', '\u{16ff1}'), + ('ð–¿¢', 'ð–¿£'), + ('ð–¿°', 'ð–¿±'), ('ð ', 'ð±'), ('ðŸ‰', '🉑'), - ('ð €€', '\u{2a6dd}'), - ('𪜀', '𫜴'), + ('ð €€', '𪛟'), + ('𪜀', '𫜸'), ('ð«€', 'ð« '), ('ð« ', '𬺡'), ('𬺰', '𮯠'), ('丽', 'ð¯¨'), - ('\u{30000}', '\u{3134a}'), + ('ð°€€', 'ð±Š'), ]; pub const HANGUL: &'static [(char, char)] = &[ @@ -823,7 +833,7 @@ pub const HIRAGANA: &'static [(char, char)] = &[ ('。', 'ï½¥'), ('ï½°', 'ï½°'), ('\u{ff9e}', '\u{ff9f}'), - ('ð›€', '𛄞'), + ('ð›€', '𛄟'), ('ð›…', 'ð›…’'), ('🈀', '🈀'), ]; @@ -836,7 +846,7 @@ pub const INHERITED: &'static [(char, char)] = &[ ('\u{343}', '\u{344}'), ('\u{346}', '\u{362}'), ('\u{953}', '\u{954}'), - ('\u{1ab0}', '\u{1ac0}'), + ('\u{1ab0}', '\u{1ace}'), ('\u{1dc2}', '\u{1df7}'), ('\u{1df9}', '\u{1df9}'), ('\u{1dfb}', '\u{1dff}'), @@ -845,6 +855,8 @@ pub const INHERITED: &'static [(char, char)] = &[ ('\u{fe00}', '\u{fe0f}'), ('\u{fe20}', '\u{fe2d}'), ('\u{101fd}', '\u{101fd}'), + ('\u{1cf00}', '\u{1cf2d}'), + ('\u{1cf30}', '\u{1cf46}'), ('\u{1d167}', '\u{1d169}'), ('\u{1d17b}', '\u{1d182}'), ('\u{1d185}', '\u{1d18b}'), @@ -861,8 +873,12 @@ pub const INSCRIPTIONAL_PARTHIAN: &'static [(char, char)] = pub const JAVANESE: &'static [(char, char)] = &[('\u{a980}', 'ê§'), ('ê§', '꧙'), ('꧞', '꧟')]; -pub const KAITHI: &'static [(char, char)] = - &[('०', '९'), ('ê °', 'ê ¹'), ('\u{11080}', 'ð‘ƒ'), ('\u{110cd}', '\u{110cd}')]; +pub const KAITHI: &'static [(char, char)] = &[ + ('०', '९'), + ('ê °', 'ê ¹'), + ('\u{11080}', '\u{110c2}'), + ('\u{110cd}', '\u{110cd}'), +]; pub const KANNADA: &'static [(char, char)] = &[ ('\u{951}', '\u{952}'), @@ -876,7 +892,7 @@ pub const KANNADA: &'static [(char, char)] = &[ ('\u{cc6}', 'ೈ'), ('ೊ', '\u{ccd}'), ('\u{cd5}', '\u{cd6}'), - ('ೞ', 'ೞ'), + ('à³', 'ೞ'), ('à³ ', '\u{ce3}'), ('೦', '೯'), ('à³±', 'à³²'), @@ -902,7 +918,11 @@ pub const KATAKANA: &'static [(char, char)] = &[ ('㌀', 'ã—'), ('ï¹…', '﹆'), ('。', '\u{ff9f}'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), ('𛀀', '𛀀'), + ('𛄠', '𛄢'), ('ð›…¤', 'ð›…§'), ]; @@ -920,7 +940,7 @@ pub const KHAROSHTHI: &'static [(char, char)] = &[ ]; pub const KHITAN_SMALL_SCRIPT: &'static [(char, char)] = - &[('\u{16fe4}', '\u{16fe4}'), ('\u{18b00}', '\u{18cd5}')]; + &[('\u{16fe4}', '\u{16fe4}'), ('𘬀', '𘳕')]; pub const KHMER: &'static [(char, char)] = &[('ក', '\u{17dd}'), ('០', '៩'), ('៰', '៹'), ('᧠', '᧿')]; @@ -976,16 +996,22 @@ pub const LATIN: &'static [(char, char)] = &[ ('â± ', 'Ɀ'), ('꜀', '꜇'), ('Ꜣ', 'ꞇ'), - ('êž‹', 'êž¿'), - ('Ꟃ', '\u{a7ca}'), - ('\u{a7f5}', 'ꟿ'), + ('êž‹', 'ꟊ'), + ('êŸ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟙ'), + ('ꟲ', 'ꟿ'), ('꤮', '꤮'), ('ꬰ', 'êš'), ('êœ', 'ê¤'), - ('ê¦', '\u{ab69}'), + ('ê¦', 'ê©'), ('ff', 'st'), ('A', 'Z'), ('ï½', 'z'), + ('ðž€', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), + ('ð¼€', 'ð¼ž'), ]; pub const LEPCHA: &'static [(char, char)] = @@ -1016,8 +1042,7 @@ pub const LINEAR_B: &'static [(char, char)] = &[ ('ð„·', 'ð„¿'), ]; -pub const LISU: &'static [(char, char)] = - &[('ê“', 'ê“¿'), ('\u{11fb0}', '\u{11fb0}')]; +pub const LISU: &'static [(char, char)] = &[('ê“', 'ê“¿'), ('𑾰', '𑾰')]; pub const LYCIAN: &'static [(char, char)] = &[('ðŠ€', 'ðŠœ')]; @@ -1082,8 +1107,7 @@ pub const MODI: &'static [(char, char)] = &[('ê °', 'ê ¹'), ('𑘀', 'ð‘™„'), ('ð‘™', 'ð‘™™')]; pub const MONGOLIAN: &'static [(char, char)] = &[ - ('á €', '\u{180e}'), - ('á ', 'á ™'), + ('á €', 'á ™'), ('á ', 'ᡸ'), ('ᢀ', 'ᢪ'), ('\u{202f}', '\u{202f}'), @@ -1115,9 +1139,16 @@ pub const NANDINAGARI: &'static [(char, char)] = &[ pub const NEW_TAI_LUE: &'static [(char, char)] = &[('ᦀ', 'ᦫ'), ('ᦰ', 'ᧉ'), ('á§', '᧚'), ('᧞', '᧟')]; -pub const NEWA: &'static [(char, char)] = &[('ð‘€', 'ð‘‘›'), ('ð‘‘', '\u{11461}')]; +pub const NEWA: &'static [(char, char)] = &[('ð‘€', 'ð‘‘›'), ('ð‘‘', 'ð‘‘¡')]; -pub const NKO: &'static [(char, char)] = &[('߀', 'ߺ'), ('\u{7fd}', 'ß¿')]; +pub const NKO: &'static [(char, char)] = &[ + ('ØŒ', 'ØŒ'), + ('Ø›', 'Ø›'), + ('ØŸ', 'ØŸ'), + ('߀', 'ߺ'), + ('\u{7fd}', 'ß¿'), + ('ï´¾', 'ï´¿'), +]; pub const NUSHU: &'static [(char, char)] = &[('ð–¿¡', 'ð–¿¡'), ('ð›…°', '𛋻')]; @@ -1146,6 +1177,9 @@ pub const OLD_SOUTH_ARABIAN: &'static [(char, char)] = &[('ð© ', 'ð©¿')]; pub const OLD_TURKIC: &'static [(char, char)] = &[('ð°€', 'ð±ˆ')]; +pub const OLD_UYGHUR: &'static [(char, char)] = + &[('Ù€', 'Ù€'), ('ð«²', 'ð«²'), ('ð½°', 'ð¾‰')]; + pub const ORIYA: &'static [(char, char)] = &[ ('\u{951}', '\u{952}'), ('।', '॥'), @@ -1253,10 +1287,11 @@ pub const SYRIAC: &'static [(char, char)] = &[ ('Ý', 'Ý'), ('à¡ ', 'ࡪ'), ('\u{1df8}', '\u{1df8}'), + ('\u{1dfa}', '\u{1dfa}'), ]; pub const TAGALOG: &'static [(char, char)] = - &[('ᜀ', 'ᜌ'), ('ᜎ', '\u{1714}'), ('᜵', '᜶')]; + &[('ᜀ', '᜕'), ('ᜟ', 'ᜟ'), ('᜵', '᜶')]; pub const TAGBANWA: &'static [(char, char)] = &[('᜵', '᜶'), ('á ', 'á¬'), ('á®', 'á°'), ('\u{1772}', '\u{1773}')]; @@ -1275,7 +1310,7 @@ pub const TAI_THAM: &'static [(char, char)] = &[ pub const TAI_VIET: &'static [(char, char)] = &[('ꪀ', 'ê«‚'), ('ê«›', 'ê«Ÿ')]; pub const TAKRI: &'static [(char, char)] = - &[('।', '॥'), ('ê °', 'ê ¹'), ('𑚀', '𑚸'), ('𑛀', '𑛉')]; + &[('।', '॥'), ('ê °', 'ê ¹'), ('𑚀', 'ð‘š¹'), ('𑛀', '𑛉')]; pub const TAMIL: &'static [(char, char)] = &[ ('\u{951}', '\u{952}'), @@ -1305,12 +1340,10 @@ pub const TAMIL: &'static [(char, char)] = &[ ('ð‘¿¿', 'ð‘¿¿'), ]; -pub const TANGUT: &'static [(char, char)] = &[ - ('ð–¿ ', 'ð–¿ '), - ('ð—€€', '𘟷'), - ('𘠀', '\u{18aff}'), - ('\u{18d00}', '\u{18d08}'), -]; +pub const TANGSA: &'static [(char, char)] = &[('ð–©°', '𖪾'), ('ð–«€', '𖫉')]; + +pub const TANGUT: &'static [(char, char)] = + &[('ð–¿ ', 'ð–¿ '), ('ð—€€', '𘟷'), ('𘠀', '𘫿'), ('𘴀', '𘴈')]; pub const TELUGU: &'static [(char, char)] = &[ ('\u{951}', '\u{952}'), @@ -1319,11 +1352,12 @@ pub const TELUGU: &'static [(char, char)] = &[ ('à°Ž', 'à°'), ('à°’', 'à°¨'), ('à°ª', 'à°¹'), - ('à°½', 'ౄ'), + ('\u{c3c}', 'ౄ'), ('\u{c46}', '\u{c48}'), ('\u{c4a}', '\u{c4d}'), ('\u{c55}', '\u{c56}'), ('ౘ', 'ౚ'), + ('à±', 'à±'), ('à± ', '\u{c63}'), ('౦', '౯'), ('à±·', '౿'), @@ -1365,10 +1399,23 @@ pub const TIRHUTA: &'static [(char, char)] = &[ ('ð‘“', 'ð‘“™'), ]; +pub const TOTO: &'static [(char, char)] = &[('ðžŠ', '\u{1e2ae}')]; + pub const UGARITIC: &'static [(char, char)] = &[('ðŽ€', 'ðŽ'), ('ðŽŸ', 'ðŽŸ')]; pub const VAI: &'static [(char, char)] = &[('ꔀ', 'ꘫ')]; +pub const VITHKUQI: &'static [(char, char)] = &[ + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), +]; + pub const WANCHO: &'static [(char, char)] = &[('ðž‹€', '𞋹'), ('ðž‹¿', 'ðž‹¿')]; pub const WARANG_CITI: &'static [(char, char)] = &[('ð‘¢ ', 'ð‘£²'), ('𑣿', '𑣿')]; @@ -1378,9 +1425,9 @@ pub const YEZIDI: &'static [(char, char)] = &[ ('Ø›', 'Ø›'), ('ØŸ', 'ØŸ'), ('Ù ', 'Ù©'), - ('\u{10e80}', '\u{10ea9}'), - ('\u{10eab}', '\u{10ead}'), - ('\u{10eb0}', '\u{10eb1}'), + ('ðº€', 'ðº©'), + ('\u{10eab}', 'ðº'), + ('ðº°', 'ðº±'), ]; pub const YI: &'static [(char, char)] = &[ diff --git a/vendor/regex-syntax/src/unicode_tables/sentence_break.rs b/vendor/regex-syntax/src/unicode_tables/sentence_break.rs index 67d830f74..db8ad282b 100644 --- a/vendor/regex-syntax/src/unicode_tables/sentence_break.rs +++ b/vendor/regex-syntax/src/unicode_tables/sentence_break.rs @@ -1,10 +1,10 @@ // DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY: // -// ucd-generate sentence-break ucd-13.0.0 --chars +// ucd-generate sentence-break /tmp/ucd --chars // -// Unicode version: 13.0.0. +// Unicode version: 14.0.0. // -// ucd-generate 0.2.8 is available on crates.io. +// ucd-generate 0.2.11 is available on crates.io. pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("ATerm", ATERM), @@ -29,7 +29,7 @@ pub const ATERM: &'static [(char, char)] = pub const CR: &'static [(char, char)] = &[('\r', '\r')]; pub const CLOSE: &'static [(char, char)] = &[ - ('\"', '\"'), + ('"', '"'), ('\'', ')'), ('[', '['), (']', ']'), @@ -57,6 +57,7 @@ pub const CLOSE: &'static [(char, char)] = &[ ('⸜', 'â¸'), ('⸠', '⸩'), ('⹂', '⹂'), + ('⹕', '⹜'), ('〈', '】'), ('〔', '〛'), ('ã€', '〟'), @@ -100,7 +101,8 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{825}', '\u{827}'), ('\u{829}', '\u{82d}'), ('\u{859}', '\u{85b}'), - ('\u{8d3}', '\u{8e1}'), + ('\u{898}', '\u{89f}'), + ('\u{8ca}', '\u{8e1}'), ('\u{8e3}', 'ः'), ('\u{93a}', '\u{93c}'), ('ा', 'à¥'), @@ -142,6 +144,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('ொ', '\u{bcd}'), ('\u{bd7}', '\u{bd7}'), ('\u{c00}', '\u{c04}'), + ('\u{c3c}', '\u{c3c}'), ('\u{c3e}', 'ౄ'), ('\u{c46}', '\u{c48}'), ('\u{c4a}', '\u{c4d}'), @@ -193,13 +196,14 @@ pub const EXTEND: &'static [(char, char)] = &[ ('á‚', 'á‚'), ('á‚š', '\u{109d}'), ('\u{135d}', '\u{135f}'), - ('\u{1712}', '\u{1714}'), - ('\u{1732}', '\u{1734}'), + ('\u{1712}', '᜕'), + ('\u{1732}', '᜴'), ('\u{1752}', '\u{1753}'), ('\u{1772}', '\u{1773}'), ('\u{17b4}', '\u{17d3}'), ('\u{17dd}', '\u{17dd}'), ('\u{180b}', '\u{180d}'), + ('\u{180f}', '\u{180f}'), ('\u{1885}', '\u{1886}'), ('\u{18a9}', '\u{18a9}'), ('\u{1920}', 'ᤫ'), @@ -208,7 +212,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('á©•', '\u{1a5e}'), ('\u{1a60}', '\u{1a7c}'), ('\u{1a7f}', '\u{1a7f}'), - ('\u{1ab0}', '\u{1ac0}'), + ('\u{1ab0}', '\u{1ace}'), ('\u{1b00}', 'ᬄ'), ('\u{1b34}', 'á„'), ('\u{1b6b}', '\u{1b73}'), @@ -221,8 +225,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{1ced}', '\u{1ced}'), ('\u{1cf4}', '\u{1cf4}'), ('á³·', '\u{1cf9}'), - ('\u{1dc0}', '\u{1df9}'), - ('\u{1dfb}', '\u{1dff}'), + ('\u{1dc0}', '\u{1dff}'), ('\u{200c}', '\u{200d}'), ('\u{20d0}', '\u{20f0}'), ('\u{2cef}', '\u{2cf1}'), @@ -277,10 +280,14 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{10d24}', '\u{10d27}'), ('\u{10eab}', '\u{10eac}'), ('\u{10f46}', '\u{10f50}'), + ('\u{10f82}', '\u{10f85}'), ('ð‘€€', '𑀂'), ('\u{11038}', '\u{11046}'), + ('\u{11070}', '\u{11070}'), + ('\u{11073}', '\u{11074}'), ('\u{1107f}', 'ð‘‚‚'), ('ð‘‚°', '\u{110ba}'), + ('\u{110c2}', '\u{110c2}'), ('\u{11100}', '\u{11102}'), ('\u{11127}', '\u{11134}'), ('ð‘……', 'ð‘…†'), @@ -288,7 +295,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{11180}', '𑆂'), ('𑆳', '𑇀'), ('\u{111c9}', '\u{111cc}'), - ('\u{111ce}', '\u{111cf}'), + ('𑇎', '\u{111cf}'), ('𑈬', '\u{11237}'), ('\u{1123e}', '\u{1123e}'), ('\u{112df}', '\u{112ea}'), @@ -311,11 +318,11 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{116ab}', '\u{116b7}'), ('\u{1171d}', '\u{1172b}'), ('ð‘ ¬', '\u{1183a}'), - ('\u{11930}', '\u{11935}'), - ('\u{11937}', '\u{11938}'), + ('\u{11930}', '𑤵'), + ('𑤷', '𑤸'), ('\u{1193b}', '\u{1193e}'), - ('\u{11940}', '\u{11940}'), - ('\u{11942}', '\u{11943}'), + ('ð‘¥€', 'ð‘¥€'), + ('𑥂', '\u{11943}'), ('𑧑', '\u{119d7}'), ('\u{119da}', '\u{119e0}'), ('𑧤', '𑧤'), @@ -344,8 +351,10 @@ pub const EXTEND: &'static [(char, char)] = &[ ('𖽑', '𖾇'), ('\u{16f8f}', '\u{16f92}'), ('\u{16fe4}', '\u{16fe4}'), - ('\u{16ff0}', '\u{16ff1}'), + ('ð–¿°', 'ð–¿±'), ('\u{1bc9d}', '\u{1bc9e}'), + ('\u{1cf00}', '\u{1cf2d}'), + ('\u{1cf30}', '\u{1cf46}'), ('\u{1d165}', '\u{1d169}'), ('ð…', '\u{1d172}'), ('\u{1d17b}', '\u{1d182}'), @@ -364,6 +373,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{1e023}', '\u{1e024}'), ('\u{1e026}', '\u{1e02a}'), ('\u{1e130}', '\u{1e136}'), + ('\u{1e2ae}', '\u{1e2ae}'), ('\u{1e2ec}', '\u{1e2ef}'), ('\u{1e8d0}', '\u{1e8d6}'), ('\u{1e944}', '\u{1e94a}'), @@ -377,6 +387,7 @@ pub const FORMAT: &'static [(char, char)] = &[ ('\u{61c}', '\u{61c}'), ('\u{6dd}', '\u{6dd}'), ('\u{70f}', '\u{70f}'), + ('\u{890}', '\u{891}'), ('\u{8e2}', '\u{8e2}'), ('\u{180e}', '\u{180e}'), ('\u{200b}', '\u{200b}'), @@ -833,7 +844,7 @@ pub const LOWER: &'static [(char, char)] = &[ ('â…°', 'â…¿'), ('ↄ', 'ↄ'), ('â“', 'â“©'), - ('â°°', 'ⱞ'), + ('â°°', 'ⱟ'), ('ⱡ', 'ⱡ'), ('â±¥', 'ⱦ'), ('ⱨ', 'ⱨ'), @@ -1001,19 +1012,33 @@ pub const LOWER: &'static [(char, char)] = &[ ('êž»', 'êž»'), ('êž½', 'êž½'), ('êž¿', 'êž¿'), + ('êŸ', 'êŸ'), ('ꟃ', 'ꟃ'), - ('\u{a7c8}', '\u{a7c8}'), - ('\u{a7ca}', '\u{a7ca}'), - ('\u{a7f6}', '\u{a7f6}'), + ('ꟈ', 'ꟈ'), + ('ꟊ', 'ꟊ'), + ('ꟑ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟕ'), + ('ꟗ', 'ꟗ'), + ('ꟙ', 'ꟙ'), + ('ꟶ', 'ꟶ'), ('ꟸ', 'ꟺ'), ('ꬰ', 'êš'), - ('êœ', '\u{ab68}'), + ('êœ', 'ê¨'), ('ê°', 'ꮿ'), ('ff', 'st'), ('ﬓ', 'ﬗ'), ('ï½', 'z'), ('ð¨', 'ð‘'), ('ð“˜', 'ð“»'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), + ('ðž€', 'ðž€'), + ('ðžƒ', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ('ð³€', 'ð³²'), ('ð‘£€', '𑣟'), ('ð–¹ ', '𖹿'), @@ -1045,6 +1070,8 @@ pub const LOWER: &'static [(char, char)] = &[ ('ðžª', 'ðŸ‚'), ('ðŸ„', 'ðŸ‰'), ('ðŸ‹', 'ðŸ‹'), + ('ð¼€', 'ð¼‰'), + ('ð¼‹', 'ð¼ž'), ('𞤢', '𞥃'), ]; @@ -1100,17 +1127,18 @@ pub const NUMERIC: &'static [(char, char)] = &[ ('𑛀', '𑛉'), ('𑜰', '𑜹'), ('ð‘£ ', '𑣩'), - ('\u{11950}', '\u{11959}'), + ('ð‘¥', 'ð‘¥™'), ('ð‘±', '𑱙'), ('ð‘µ', '𑵙'), ('𑶠', '𑶩'), ('ð–© ', 'ð–©©'), + ('ð–«€', '𖫉'), ('ð–', 'ð–™'), ('ðŸŽ', 'ðŸ¿'), ('ðž…€', 'ðž…‰'), ('ðž‹°', '𞋹'), ('ðž¥', '𞥙'), - ('\u{1fbf0}', '\u{1fbf9}'), + ('🯰', '🯹'), ]; pub const OLETTER: &'static [(char, char)] = &[ @@ -1146,8 +1174,9 @@ pub const OLETTER: &'static [(char, char)] = &[ ('à ¨', 'à ¨'), ('à¡€', 'ࡘ'), ('à¡ ', 'ࡪ'), - ('ࢠ', 'ࢴ'), - ('ࢶ', '\u{8c7}'), + ('à¡°', 'ࢇ'), + ('ࢉ', 'ࢎ'), + ('ࢠ', 'ࣉ'), ('ऄ', 'ह'), ('ऽ', 'ऽ'), ('à¥', 'à¥'), @@ -1212,6 +1241,7 @@ pub const OLETTER: &'static [(char, char)] = &[ ('à°ª', 'à°¹'), ('à°½', 'à°½'), ('ౘ', 'ౚ'), + ('à±', 'à±'), ('à± ', 'ౡ'), ('ಀ', 'ಀ'), ('ಅ', 'ಌ'), @@ -1220,10 +1250,10 @@ pub const OLETTER: &'static [(char, char)] = &[ ('ಪ', 'ಳ'), ('ವ', 'ಹ'), ('ಽ', 'ಽ'), - ('ೞ', 'ೞ'), + ('à³', 'ೞ'), ('à³ ', 'ೡ'), ('à³±', 'à³²'), - ('\u{d04}', 'à´Œ'), + ('à´„', 'à´Œ'), ('à´Ž', 'à´'), ('à´’', 'à´º'), ('à´½', 'à´½'), @@ -1286,9 +1316,8 @@ pub const OLETTER: &'static [(char, char)] = &[ ('áš', 'ášš'), ('áš ', 'ᛪ'), ('á›®', 'ᛸ'), - ('ᜀ', 'ᜌ'), - ('ᜎ', 'ᜑ'), - ('ᜠ', 'ᜱ'), + ('ᜀ', 'ᜑ'), + ('ᜟ', 'ᜱ'), ('á€', 'á‘'), ('á ', 'á¬'), ('á®', 'á°'), @@ -1309,7 +1338,7 @@ pub const OLETTER: &'static [(char, char)] = &[ ('ᨠ', 'á©”'), ('ᪧ', 'ᪧ'), ('ᬅ', 'ᬳ'), - ('á…', 'á‹'), + ('á…', 'áŒ'), ('ᮃ', 'á® '), ('á®®', 'ᮯ'), ('ᮺ', 'ᯥ'), @@ -1347,11 +1376,10 @@ pub const OLETTER: &'static [(char, char)] = &[ ('ー', 'ヿ'), ('ã„…', 'ㄯ'), ('ㄱ', 'ㆎ'), - ('ㆠ', '\u{31bf}'), + ('ㆠ', 'ㆿ'), ('ㇰ', 'ㇿ'), - ('ã€', '\u{4dbf}'), - ('一', '\u{9ffc}'), - ('ꀀ', 'ê’Œ'), + ('ã€', '䶿'), + ('一', 'ê’Œ'), ('ê“', 'ꓽ'), ('ꔀ', 'ꘌ'), ('ê˜', 'ꘟ'), @@ -1362,6 +1390,7 @@ pub const OLETTER: &'static [(char, char)] = &[ ('ꜗ', 'ꜟ'), ('ꞈ', 'ꞈ'), ('êž', 'êž'), + ('ꟲ', 'ꟴ'), ('ꟷ', 'ꟷ'), ('ꟻ', 'ê '), ('ê ƒ', 'ê …'), @@ -1399,7 +1428,7 @@ pub const OLETTER: &'static [(char, char)] = &[ ('ꬑ', 'ꬖ'), ('ê¬ ', 'ꬦ'), ('ꬨ', 'ꬮ'), - ('\u{ab69}', '\u{ab69}'), + ('ê©', 'ê©'), ('ꯀ', 'ꯢ'), ('ê°€', '힣'), ('íž°', 'ퟆ'), @@ -1449,6 +1478,7 @@ pub const OLETTER: &'static [(char, char)] = &[ ('ð˜€', 'ðœ¶'), ('ð€', 'ð•'), ('ð ', 'ð§'), + ('ðž', 'ðž‚'), ('ð €', 'ð …'), ('ð ˆ', 'ð ˆ'), ('ð Š', 'ð µ'), @@ -1477,19 +1507,22 @@ pub const OLETTER: &'static [(char, char)] = &[ ('ð®€', 'ð®‘'), ('ð°€', 'ð±ˆ'), ('ð´€', 'ð´£'), - ('\u{10e80}', '\u{10ea9}'), - ('\u{10eb0}', '\u{10eb1}'), + ('ðº€', 'ðº©'), + ('ðº°', 'ðº±'), ('ð¼€', 'ð¼œ'), ('ð¼§', 'ð¼§'), ('ð¼°', 'ð½…'), - ('\u{10fb0}', '\u{10fc4}'), + ('ð½°', 'ð¾'), + ('ð¾°', 'ð¿„'), ('ð¿ ', 'ð¿¶'), ('𑀃', 'ð‘€·'), + ('ð‘±', 'ð‘²'), + ('ð‘µ', 'ð‘µ'), ('𑂃', '𑂯'), ('ð‘ƒ', '𑃨'), ('𑄃', '𑄦'), ('ð‘…„', 'ð‘…„'), - ('\u{11147}', '\u{11147}'), + ('ð‘…‡', 'ð‘…‡'), ('ð‘…', 'ð‘…²'), ('ð‘…¶', 'ð‘…¶'), ('𑆃', '𑆲'), @@ -1515,7 +1548,7 @@ pub const OLETTER: &'static [(char, char)] = &[ ('ð‘', 'ð‘¡'), ('ð‘€', 'ð‘´'), ('𑑇', 'ð‘‘Š'), - ('ð‘‘Ÿ', '\u{11461}'), + ('ð‘‘Ÿ', 'ð‘‘¡'), ('ð‘’€', 'ð‘’¯'), ('ð‘“„', 'ð‘“…'), ('𑓇', '𑓇'), @@ -1526,14 +1559,15 @@ pub const OLETTER: &'static [(char, char)] = &[ ('𑚀', '𑚪'), ('𑚸', '𑚸'), ('𑜀', '𑜚'), + ('ð‘€', 'ð‘†'), ('ð‘ €', 'ð‘ «'), - ('𑣿', '\u{11906}'), - ('\u{11909}', '\u{11909}'), - ('\u{1190c}', '\u{11913}'), - ('\u{11915}', '\u{11916}'), - ('\u{11918}', '\u{1192f}'), - ('\u{1193f}', '\u{1193f}'), - ('\u{11941}', '\u{11941}'), + ('𑣿', '𑤆'), + ('𑤉', '𑤉'), + ('𑤌', '𑤓'), + ('𑤕', '𑤖'), + ('𑤘', '𑤯'), + ('𑤿', '𑤿'), + ('ð‘¥', 'ð‘¥'), ('𑦠', '𑦧'), ('𑦪', 'ð‘§'), ('𑧡', '𑧡'), @@ -1544,7 +1578,7 @@ pub const OLETTER: &'static [(char, char)] = &[ ('ð‘©', 'ð‘©'), ('ð‘©œ', '𑪉'), ('ð‘ª', 'ð‘ª'), - ('ð‘«€', '𑫸'), + ('𑪰', '𑫸'), ('ð‘°€', 'ð‘°ˆ'), ('ð‘°Š', 'ð‘°®'), ('𑱀', '𑱀'), @@ -1558,14 +1592,16 @@ pub const OLETTER: &'static [(char, char)] = &[ ('𑵪', '𑶉'), ('𑶘', '𑶘'), ('ð‘» ', 'ð‘»²'), - ('\u{11fb0}', '\u{11fb0}'), + ('𑾰', '𑾰'), ('ð’€€', 'ð’Ž™'), ('ð’€', 'ð’‘®'), ('ð’’€', '𒕃'), + ('ð’¾', 'ð’¿°'), ('ð“€€', 'ð“®'), ('ð”€', '𔙆'), ('ð– €', '𖨸'), ('ð–©€', 'ð–©ž'), + ('ð–©°', '𖪾'), ('ð–«', 'ð–«'), ('𖬀', '𖬯'), ('ð–€', 'ð–ƒ'), @@ -1577,9 +1613,12 @@ pub const OLETTER: &'static [(char, char)] = &[ ('ð–¿ ', 'ð–¿¡'), ('ð–¿£', 'ð–¿£'), ('ð—€€', '𘟷'), - ('𘠀', '\u{18cd5}'), - ('\u{18d00}', '\u{18d08}'), - ('𛀀', '𛄞'), + ('𘠀', '𘳕'), + ('𘴀', '𘴈'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), + ('𛀀', '𛄢'), ('ð›…', 'ð›…’'), ('ð›…¤', 'ð›…§'), ('ð›…°', '𛋻'), @@ -1587,10 +1626,16 @@ pub const OLETTER: &'static [(char, char)] = &[ ('ð›±°', 'ð›±¼'), ('𛲀', '𛲈'), ('ð›²', '𛲙'), + ('ð¼Š', 'ð¼Š'), ('ðž„€', '𞄬'), ('ðž„·', '𞄽'), ('ðž…Ž', 'ðž…Ž'), + ('ðžŠ', 'ðžŠ'), ('ðž‹€', 'ðž‹«'), + ('𞟠', '𞟦'), + ('𞟨', '𞟫'), + ('ðžŸ', '𞟮'), + ('𞟰', '𞟾'), ('ðž €', '𞣄'), ('𞥋', '𞥋'), ('𞸀', '𞸃'), @@ -1626,13 +1671,13 @@ pub const OLETTER: &'static [(char, char)] = &[ ('𞺡', '𞺣'), ('𞺥', '𞺩'), ('𞺫', '𞺻'), - ('ð €€', '\u{2a6dd}'), - ('𪜀', '𫜴'), + ('ð €€', '𪛟'), + ('𪜀', '𫜸'), ('ð«€', 'ð« '), ('ð« ', '𬺡'), ('𬺰', '𮯠'), ('丽', 'ð¯¨'), - ('\u{30000}', '\u{3134a}'), + ('ð°€€', 'ð±Š'), ]; pub const SCONTINUE: &'static [(char, char)] = &[ @@ -1661,7 +1706,7 @@ pub const STERM: &'static [(char, char)] = &[ ('!', '!'), ('?', '?'), ('Ö‰', 'Ö‰'), - ('Øž', 'ØŸ'), + ('Ø', 'ØŸ'), ('Û”', 'Û”'), ('Ü€', 'Ü‚'), ('ß¹', 'ß¹'), @@ -1680,12 +1725,14 @@ pub const STERM: &'static [(char, char)] = &[ ('᪨', '᪫'), ('áš', 'á›'), ('áž', 'áŸ'), + ('á½', 'á¾'), ('á°»', 'á°¼'), ('á±¾', '᱿'), ('‼', '‽'), ('â‡', 'â‰'), ('⸮', '⸮'), ('⸼', '⸼'), + ('⹓', 'â¹”'), ('。', '。'), ('ê“¿', 'ê“¿'), ('꘎', 'ê˜'), @@ -1704,6 +1751,7 @@ pub const STERM: &'static [(char, char)] = &[ ('。', '。'), ('ð©–', 'ð©—'), ('ð½•', 'ð½™'), + ('ð¾†', 'ð¾‰'), ('ð‘‡', 'ð‘ˆ'), ('ð‘‚¾', 'ð‘ƒ'), ('ð‘…', 'ð‘…ƒ'), @@ -1718,8 +1766,8 @@ pub const STERM: &'static [(char, char)] = &[ ('ð‘—‰', 'ð‘——'), ('ð‘™', 'ð‘™‚'), ('𑜼', '𑜾'), - ('\u{11944}', '\u{11944}'), - ('\u{11946}', '\u{11946}'), + ('𑥄', '𑥄'), + ('𑥆', '𑥆'), ('ð‘©‚', '𑩃'), ('𑪛', '𑪜'), ('ð‘±', '𑱂'), @@ -2183,7 +2231,7 @@ pub const UPPER: &'static [(char, char)] = &[ ('â… ', 'â…¯'), ('Ↄ', 'Ↄ'), ('â’¶', 'â“'), - ('â°€', 'â°®'), + ('â°€', 'â°¯'), ('â± ', 'â± '), ('â±¢', 'Ɽ'), ('Ⱨ', 'Ⱨ'), @@ -2348,13 +2396,21 @@ pub const UPPER: &'static [(char, char)] = &[ ('Ꞻ', 'Ꞻ'), ('êž¼', 'êž¼'), ('êž¾', 'êž¾'), + ('Ꟁ', 'Ꟁ'), ('Ꟃ', 'Ꟃ'), - ('Ꞔ', '\u{a7c7}'), - ('\u{a7c9}', '\u{a7c9}'), - ('\u{a7f5}', '\u{a7f5}'), + ('Ꞔ', 'Ꟈ'), + ('Ꟊ', 'Ꟊ'), + ('êŸ', 'êŸ'), + ('Ꟗ', 'Ꟗ'), + ('Ꟙ', 'Ꟙ'), + ('Ꟶ', 'Ꟶ'), ('A', 'Z'), ('ð€', 'ð§'), ('ð’°', 'ð““'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), ('ð²€', 'ð²²'), ('ð‘¢ ', '𑢿'), ('ð–¹€', '𖹟'), diff --git a/vendor/regex-syntax/src/unicode_tables/word_break.rs b/vendor/regex-syntax/src/unicode_tables/word_break.rs index bd23e00a8..19b2a1c64 100644 --- a/vendor/regex-syntax/src/unicode_tables/word_break.rs +++ b/vendor/regex-syntax/src/unicode_tables/word_break.rs @@ -1,10 +1,10 @@ // DO NOT EDIT THIS FILE. IT WAS AUTOMATICALLY GENERATED BY: // -// ucd-generate word-break ucd-13.0.0 --chars +// ucd-generate word-break /tmp/ucd --chars // -// Unicode version: 13.0.0. +// Unicode version: 14.0.0. // -// ucd-generate 0.2.8 is available on crates.io. +// ucd-generate 0.2.11 is available on crates.io. pub const BY_NAME: &'static [(&'static str, &'static [(char, char)])] = &[ ("ALetter", ALETTER), @@ -75,8 +75,9 @@ pub const ALETTER: &'static [(char, char)] = &[ ('à ¨', 'à ¨'), ('à¡€', 'ࡘ'), ('à¡ ', 'ࡪ'), - ('ࢠ', 'ࢴ'), - ('ࢶ', '\u{8c7}'), + ('à¡°', 'ࢇ'), + ('ࢉ', 'ࢎ'), + ('ࢠ', 'ࣉ'), ('ऄ', 'ह'), ('ऽ', 'ऽ'), ('à¥', 'à¥'), @@ -141,6 +142,7 @@ pub const ALETTER: &'static [(char, char)] = &[ ('à°ª', 'à°¹'), ('à°½', 'à°½'), ('ౘ', 'ౚ'), + ('à±', 'à±'), ('à± ', 'ౡ'), ('ಀ', 'ಀ'), ('ಅ', 'ಌ'), @@ -149,10 +151,10 @@ pub const ALETTER: &'static [(char, char)] = &[ ('ಪ', 'ಳ'), ('ವ', 'ಹ'), ('ಽ', 'ಽ'), - ('ೞ', 'ೞ'), + ('à³', 'ೞ'), ('à³ ', 'ೡ'), ('à³±', 'à³²'), - ('\u{d04}', 'à´Œ'), + ('à´„', 'à´Œ'), ('à´Ž', 'à´'), ('à´’', 'à´º'), ('à´½', 'à´½'), @@ -197,9 +199,8 @@ pub const ALETTER: &'static [(char, char)] = &[ ('áš', 'ášš'), ('áš ', 'ᛪ'), ('á›®', 'ᛸ'), - ('ᜀ', 'ᜌ'), - ('ᜎ', 'ᜑ'), - ('ᜠ', 'ᜱ'), + ('ᜀ', 'ᜑ'), + ('ᜟ', 'ᜱ'), ('á€', 'á‘'), ('á ', 'á¬'), ('á®', 'á°'), @@ -211,7 +212,7 @@ pub const ALETTER: &'static [(char, char)] = &[ ('ᤀ', 'ᤞ'), ('ᨀ', 'ᨖ'), ('ᬅ', 'ᬳ'), - ('á…', 'á‹'), + ('á…', 'áŒ'), ('ᮃ', 'á® '), ('á®®', 'ᮯ'), ('ᮺ', 'ᯥ'), @@ -263,9 +264,7 @@ pub const ALETTER: &'static [(char, char)] = &[ ('â…Ž', 'â…Ž'), ('â… ', 'ↈ'), ('â’¶', 'â“©'), - ('â°€', 'â°®'), - ('â°°', 'ⱞ'), - ('â± ', 'ⳤ'), + ('â°€', 'ⳤ'), ('Ⳬ', 'â³®'), ('â³²', 'â³³'), ('â´€', 'â´¥'), @@ -287,7 +286,7 @@ pub const ALETTER: &'static [(char, char)] = &[ ('〻', '〼'), ('ã„…', 'ㄯ'), ('ㄱ', 'ㆎ'), - ('ㆠ', '\u{31bf}'), + ('ㆠ', 'ㆿ'), ('ꀀ', 'ê’Œ'), ('ê“', 'ꓽ'), ('ꔀ', 'ꘌ'), @@ -296,9 +295,11 @@ pub const ALETTER: &'static [(char, char)] = &[ ('Ꙁ', 'ê™®'), ('ꙿ', 'êš'), ('êš ', 'ꛯ'), - ('꜈', 'êž¿'), - ('Ꟃ', '\u{a7ca}'), - ('\u{a7f5}', 'ê '), + ('꜈', 'ꟊ'), + ('êŸ', 'ꟑ'), + ('ꟓ', 'ꟓ'), + ('ꟕ', 'ꟙ'), + ('ꟲ', 'ê '), ('ê ƒ', 'ê …'), ('ê ‡', 'ê Š'), ('ê Œ', 'ê ¢'), @@ -322,7 +323,7 @@ pub const ALETTER: &'static [(char, char)] = &[ ('ꬑ', 'ꬖ'), ('ê¬ ', 'ꬦ'), ('ꬨ', 'ꬮ'), - ('ꬰ', '\u{ab69}'), + ('ꬰ', 'ê©'), ('ê°', 'ꯢ'), ('ê°€', '힣'), ('íž°', 'ퟆ'), @@ -365,9 +366,20 @@ pub const ALETTER: &'static [(char, char)] = &[ ('ð“˜', 'ð“»'), ('ð”€', 'ð”§'), ('ð”°', 'ð•£'), + ('ð•°', 'ð•º'), + ('ð•¼', 'ð–Š'), + ('ð–Œ', 'ð–’'), + ('ð–”', 'ð–•'), + ('ð–—', 'ð–¡'), + ('ð–£', 'ð–±'), + ('ð–³', 'ð–¹'), + ('ð–»', 'ð–¼'), ('ð˜€', 'ðœ¶'), ('ð€', 'ð•'), ('ð ', 'ð§'), + ('ðž€', 'ðž…'), + ('ðž‡', 'ðž°'), + ('ðž²', 'ðžº'), ('ð €', 'ð …'), ('ð ˆ', 'ð ˆ'), ('ð Š', 'ð µ'), @@ -398,19 +410,22 @@ pub const ALETTER: &'static [(char, char)] = &[ ('ð²€', 'ð²²'), ('ð³€', 'ð³²'), ('ð´€', 'ð´£'), - ('\u{10e80}', '\u{10ea9}'), - ('\u{10eb0}', '\u{10eb1}'), + ('ðº€', 'ðº©'), + ('ðº°', 'ðº±'), ('ð¼€', 'ð¼œ'), ('ð¼§', 'ð¼§'), ('ð¼°', 'ð½…'), - ('\u{10fb0}', '\u{10fc4}'), + ('ð½°', 'ð¾'), + ('ð¾°', 'ð¿„'), ('ð¿ ', 'ð¿¶'), ('𑀃', 'ð‘€·'), + ('ð‘±', 'ð‘²'), + ('ð‘µ', 'ð‘µ'), ('𑂃', '𑂯'), ('ð‘ƒ', '𑃨'), ('𑄃', '𑄦'), ('ð‘…„', 'ð‘…„'), - ('\u{11147}', '\u{11147}'), + ('ð‘…‡', 'ð‘…‡'), ('ð‘…', 'ð‘…²'), ('ð‘…¶', 'ð‘…¶'), ('𑆃', '𑆲'), @@ -436,7 +451,7 @@ pub const ALETTER: &'static [(char, char)] = &[ ('ð‘', 'ð‘¡'), ('ð‘€', 'ð‘´'), ('𑑇', 'ð‘‘Š'), - ('ð‘‘Ÿ', '\u{11461}'), + ('ð‘‘Ÿ', 'ð‘‘¡'), ('ð‘’€', 'ð‘’¯'), ('ð‘“„', 'ð‘“…'), ('𑓇', '𑓇'), @@ -448,13 +463,13 @@ pub const ALETTER: &'static [(char, char)] = &[ ('𑚸', '𑚸'), ('ð‘ €', 'ð‘ «'), ('ð‘¢ ', '𑣟'), - ('𑣿', '\u{11906}'), - ('\u{11909}', '\u{11909}'), - ('\u{1190c}', '\u{11913}'), - ('\u{11915}', '\u{11916}'), - ('\u{11918}', '\u{1192f}'), - ('\u{1193f}', '\u{1193f}'), - ('\u{11941}', '\u{11941}'), + ('𑣿', '𑤆'), + ('𑤉', '𑤉'), + ('𑤌', '𑤓'), + ('𑤕', '𑤖'), + ('𑤘', '𑤯'), + ('𑤿', '𑤿'), + ('ð‘¥', 'ð‘¥'), ('𑦠', '𑦧'), ('𑦪', 'ð‘§'), ('𑧡', '𑧡'), @@ -465,7 +480,7 @@ pub const ALETTER: &'static [(char, char)] = &[ ('ð‘©', 'ð‘©'), ('ð‘©œ', '𑪉'), ('ð‘ª', 'ð‘ª'), - ('ð‘«€', '𑫸'), + ('𑪰', '𑫸'), ('ð‘°€', 'ð‘°ˆ'), ('ð‘°Š', 'ð‘°®'), ('𑱀', '𑱀'), @@ -479,14 +494,16 @@ pub const ALETTER: &'static [(char, char)] = &[ ('𑵪', '𑶉'), ('𑶘', '𑶘'), ('ð‘» ', 'ð‘»²'), - ('\u{11fb0}', '\u{11fb0}'), + ('𑾰', '𑾰'), ('ð’€€', 'ð’Ž™'), ('ð’€', 'ð’‘®'), ('ð’’€', '𒕃'), + ('ð’¾', 'ð’¿°'), ('ð“€€', 'ð“®'), ('ð”€', '𔙆'), ('ð– €', '𖨸'), ('ð–©€', 'ð–©ž'), + ('ð–©°', '𖪾'), ('ð–«', 'ð–«'), ('𖬀', '𖬯'), ('ð–€', 'ð–ƒ'), @@ -532,10 +549,16 @@ pub const ALETTER: &'static [(char, char)] = &[ ('ðžŠ', 'ðž¨'), ('ðžª', 'ðŸ‚'), ('ðŸ„', 'ðŸ‹'), + ('ð¼€', 'ð¼ž'), ('ðž„€', '𞄬'), ('ðž„·', '𞄽'), ('ðž…Ž', 'ðž…Ž'), + ('ðžŠ', 'ðžŠ'), ('ðž‹€', 'ðž‹«'), + ('𞟠', '𞟦'), + ('𞟨', '𞟫'), + ('ðžŸ', '𞟮'), + ('𞟰', '𞟾'), ('ðž €', '𞣄'), ('𞤀', '𞥃'), ('𞥋', '𞥋'), @@ -579,7 +602,7 @@ pub const ALETTER: &'static [(char, char)] = &[ pub const CR: &'static [(char, char)] = &[('\r', '\r')]; -pub const DOUBLE_QUOTE: &'static [(char, char)] = &[('\"', '\"')]; +pub const DOUBLE_QUOTE: &'static [(char, char)] = &[('"', '"')]; pub const EXTEND: &'static [(char, char)] = &[ ('\u{300}', '\u{36f}'), @@ -606,7 +629,8 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{825}', '\u{827}'), ('\u{829}', '\u{82d}'), ('\u{859}', '\u{85b}'), - ('\u{8d3}', '\u{8e1}'), + ('\u{898}', '\u{89f}'), + ('\u{8ca}', '\u{8e1}'), ('\u{8e3}', 'ः'), ('\u{93a}', '\u{93c}'), ('ा', 'à¥'), @@ -648,6 +672,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('ொ', '\u{bcd}'), ('\u{bd7}', '\u{bd7}'), ('\u{c00}', '\u{c04}'), + ('\u{c3c}', '\u{c3c}'), ('\u{c3e}', 'ౄ'), ('\u{c46}', '\u{c48}'), ('\u{c4a}', '\u{c4d}'), @@ -699,13 +724,14 @@ pub const EXTEND: &'static [(char, char)] = &[ ('á‚', 'á‚'), ('á‚š', '\u{109d}'), ('\u{135d}', '\u{135f}'), - ('\u{1712}', '\u{1714}'), - ('\u{1732}', '\u{1734}'), + ('\u{1712}', '᜕'), + ('\u{1732}', '᜴'), ('\u{1752}', '\u{1753}'), ('\u{1772}', '\u{1773}'), ('\u{17b4}', '\u{17d3}'), ('\u{17dd}', '\u{17dd}'), ('\u{180b}', '\u{180d}'), + ('\u{180f}', '\u{180f}'), ('\u{1885}', '\u{1886}'), ('\u{18a9}', '\u{18a9}'), ('\u{1920}', 'ᤫ'), @@ -714,7 +740,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('á©•', '\u{1a5e}'), ('\u{1a60}', '\u{1a7c}'), ('\u{1a7f}', '\u{1a7f}'), - ('\u{1ab0}', '\u{1ac0}'), + ('\u{1ab0}', '\u{1ace}'), ('\u{1b00}', 'ᬄ'), ('\u{1b34}', 'á„'), ('\u{1b6b}', '\u{1b73}'), @@ -727,8 +753,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{1ced}', '\u{1ced}'), ('\u{1cf4}', '\u{1cf4}'), ('á³·', '\u{1cf9}'), - ('\u{1dc0}', '\u{1df9}'), - ('\u{1dfb}', '\u{1dff}'), + ('\u{1dc0}', '\u{1dff}'), ('\u{200c}', '\u{200c}'), ('\u{20d0}', '\u{20f0}'), ('\u{2cef}', '\u{2cf1}'), @@ -783,10 +808,14 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{10d24}', '\u{10d27}'), ('\u{10eab}', '\u{10eac}'), ('\u{10f46}', '\u{10f50}'), + ('\u{10f82}', '\u{10f85}'), ('ð‘€€', '𑀂'), ('\u{11038}', '\u{11046}'), + ('\u{11070}', '\u{11070}'), + ('\u{11073}', '\u{11074}'), ('\u{1107f}', 'ð‘‚‚'), ('ð‘‚°', '\u{110ba}'), + ('\u{110c2}', '\u{110c2}'), ('\u{11100}', '\u{11102}'), ('\u{11127}', '\u{11134}'), ('ð‘……', 'ð‘…†'), @@ -794,7 +823,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{11180}', '𑆂'), ('𑆳', '𑇀'), ('\u{111c9}', '\u{111cc}'), - ('\u{111ce}', '\u{111cf}'), + ('𑇎', '\u{111cf}'), ('𑈬', '\u{11237}'), ('\u{1123e}', '\u{1123e}'), ('\u{112df}', '\u{112ea}'), @@ -817,11 +846,11 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{116ab}', '\u{116b7}'), ('\u{1171d}', '\u{1172b}'), ('ð‘ ¬', '\u{1183a}'), - ('\u{11930}', '\u{11935}'), - ('\u{11937}', '\u{11938}'), + ('\u{11930}', '𑤵'), + ('𑤷', '𑤸'), ('\u{1193b}', '\u{1193e}'), - ('\u{11940}', '\u{11940}'), - ('\u{11942}', '\u{11943}'), + ('ð‘¥€', 'ð‘¥€'), + ('𑥂', '\u{11943}'), ('𑧑', '\u{119d7}'), ('\u{119da}', '\u{119e0}'), ('𑧤', '𑧤'), @@ -850,8 +879,10 @@ pub const EXTEND: &'static [(char, char)] = &[ ('𖽑', '𖾇'), ('\u{16f8f}', '\u{16f92}'), ('\u{16fe4}', '\u{16fe4}'), - ('\u{16ff0}', '\u{16ff1}'), + ('ð–¿°', 'ð–¿±'), ('\u{1bc9d}', '\u{1bc9e}'), + ('\u{1cf00}', '\u{1cf2d}'), + ('\u{1cf30}', '\u{1cf46}'), ('\u{1d165}', '\u{1d169}'), ('ð…', '\u{1d172}'), ('\u{1d17b}', '\u{1d182}'), @@ -870,6 +901,7 @@ pub const EXTEND: &'static [(char, char)] = &[ ('\u{1e023}', '\u{1e024}'), ('\u{1e026}', '\u{1e02a}'), ('\u{1e130}', '\u{1e136}'), + ('\u{1e2ae}', '\u{1e2ae}'), ('\u{1e2ec}', '\u{1e2ef}'), ('\u{1e8d0}', '\u{1e8d6}'), ('\u{1e944}', '\u{1e94a}'), @@ -894,6 +926,7 @@ pub const FORMAT: &'static [(char, char)] = &[ ('\u{61c}', '\u{61c}'), ('\u{6dd}', '\u{6dd}'), ('\u{70f}', '\u{70f}'), + ('\u{890}', '\u{891}'), ('\u{8e2}', '\u{8e2}'), ('\u{180e}', '\u{180e}'), ('\u{200e}', '\u{200f}'), @@ -932,7 +965,11 @@ pub const KATAKANA: &'static [(char, char)] = &[ ('ã‹', '㋾'), ('㌀', 'ã—'), ('ヲ', 'ï¾'), + ('ðš¿°', '𚿳'), + ('𚿵', 'ðš¿»'), + ('𚿽', '𚿾'), ('𛀀', '𛀀'), + ('𛄠', '𛄢'), ('ð›…¤', 'ð›…§'), ]; @@ -1031,17 +1068,18 @@ pub const NUMERIC: &'static [(char, char)] = &[ ('𑛀', '𑛉'), ('𑜰', '𑜹'), ('ð‘£ ', '𑣩'), - ('\u{11950}', '\u{11959}'), + ('ð‘¥', 'ð‘¥™'), ('ð‘±', '𑱙'), ('ð‘µ', '𑵙'), ('𑶠', '𑶩'), ('ð–© ', 'ð–©©'), + ('ð–«€', '𖫉'), ('ð–', 'ð–™'), ('ðŸŽ', 'ðŸ¿'), ('ðž…€', 'ðž…‰'), ('ðž‹°', '𞋹'), ('ðž¥', '𞥙'), - ('\u{1fbf0}', '\u{1fbf9}'), + ('🯰', '🯹'), ]; pub const REGIONAL_INDICATOR: &'static [(char, char)] = &[('🇦', '🇿')]; diff --git a/vendor/regex-syntax/src/utf8.rs b/vendor/regex-syntax/src/utf8.rs index dc055033e..b9c865532 100644 --- a/vendor/regex-syntax/src/utf8.rs +++ b/vendor/regex-syntax/src/utf8.rs @@ -198,7 +198,7 @@ impl<'a> IntoIterator for &'a Utf8Sequence { type Item = &'a Utf8Range; fn into_iter(self) -> Self::IntoIter { - self.as_slice().into_iter() + self.as_slice().iter() } } @@ -448,7 +448,7 @@ fn max_scalar_value(nbytes: usize) -> u32 { 1 => 0x007F, 2 => 0x07FF, 3 => 0xFFFF, - 4 => 0x10FFFF, + 4 => 0x0010_FFFF, _ => unreachable!("invalid UTF-8 byte sequence size"), } } @@ -492,7 +492,7 @@ mod tests { fn single_codepoint_one_sequence() { // Tests that every range of scalar values that contains a single // scalar value is recognized by one sequence of byte ranges. - for i in 0x0..(0x10FFFF + 1) { + for i in 0x0..=0x0010_FFFF { let c = match char::from_u32(i) { None => continue, Some(c) => c, diff --git a/vendor/regex/.cargo-checksum.json b/vendor/regex/.cargo-checksum.json index f05551b4f..463368b15 100644 --- a/vendor/regex/.cargo-checksum.json +++ b/vendor/regex/.cargo-checksum.json @@ -1 +1 @@ -{"files":{"CHANGELOG.md":"34a39916f7d2c49f8e779257a906bab6500dda007dbb42188f31ad72c8ba9bef","Cargo.lock":"5cacc641e7eae4988d6f45ad69dee331358bfd7bc7c1180c0e1ff5fae6d4a969","Cargo.toml":"1f360fc3a09b65707c76b5c8409733718abd5e46a970b33e5d0dc09b9078c12a","HACKING.md":"17818f7a17723608f6bdbe6388ad0a913d4f96f76a16649aaf4e274b1fa0ea97","LICENSE-APACHE":"a60eea817514531668d7e00765731449fe14d059d3249e0bc93b36de45f759f2","LICENSE-MIT":"6485b8ed310d3f0340bf1ad1f47645069ce4069dcc6bb46c7d5c6faf41de1fdb","PERFORMANCE.md":"0d5ef3866386918dfdefb1aa9a28cfe33cb3c8ceeb79f3f8ba5b88253dd95991","README.md":"6125b1c70b9b560412529c54dc6aacdfd39cc82f69b5ad7776fa86e4cb720347","UNICODE.md":"a8a8399540eed000d19420135a527f400247a04572e44d124c786b870f518776","examples/regexdna-input.txt":"156a49710bb3e1ed4bc2bbb0af0f383b747b3d0281453cfff39c296124c598f8","examples/regexdna-output.txt":"35e85b19b70a893d752fd43e54e1e9da08bac43559191cea85b33387c24c4cc1","examples/shootout-regex-dna-bytes.rs":"fa2daedb4e0a05f64f33f4af62fbb0176db998e3676f8637ab684b725367a7b4","examples/shootout-regex-dna-cheat.rs":"1f871a6eaaf8372299fa3c762051112fa89a14235b03f734fc50ebd51ecaee72","examples/shootout-regex-dna-replace.rs":"32ffdf13ac6c4ce3fc32116a048e9cc682aa34cdb8e5beaf565a22addbdcd9ab","examples/shootout-regex-dna-single-cheat.rs":"809f75bf1e1917a53623eb6f1a3ce3b7d2ed98a6a1dbc0bd4853bec49a0c6f94","examples/shootout-regex-dna-single.rs":"1ab14f5703cd4be2e75a2e792e0ba1d322b9e4b14535d396805a4316d577f5bb","examples/shootout-regex-dna.rs":"20ea46ab63f91e3ac6a64e997eadd436a9cbc2f1bdade28e4512052f0e25bc34","rustfmt.toml":"1ca600239a27401c4a43f363cf3f38183a212affc1f31bff3ae93234bbaec228","src/backtrack.rs":"9018950f86564184a5dafd869a6c2cb3c4538ff302adbeccac33015f6e88f904","src/compile.rs":"5c834c2125abc10d0de3c377e34b1f2e2c10b837809631b1c688d3b536368cfd","src/dfa.rs":"2dfed56e1bd3506d328a3d2031aaa15dd87a7561f92f1a9fc1946a14cfbb64ae","src/error.rs":"71c85db839514f26ee024a689061743ea94a34eb7a3291e6c2b69b45a9682d09","src/exec.rs":"88b49098761db539e717c2965d2f56908c81aa10d91873d29a08484be7a11218","src/expand.rs":"ccba4798bd48ce40c34e91318f12258f46d8382e178a2c2b31c0d03b1ab51c5b","src/find_byte.rs":"b387247b77e3269f057c3399aefe5a815032c3af918c876f80eb4b282e4eb95e","src/freqs.rs":"255555f3d95b08a5bb3bc2f38d5a06cc100a39c0f0127fe4f50c33afa1cadc65","src/input.rs":"69595d1ea8d35351f5065ffdbf5965427d2e3fb5160a37008fa7e21d0eaa7720","src/lib.rs":"20bc28509e1853faea9581d43b21bc3ab144bb776e47fda4560082c4673854a6","src/literal/imp.rs":"5f73e0bcbee70c11041deca5ae84a8d30995963f452b29dd2fe5ab46d4978c12","src/literal/mod.rs":"533f1d68af088e9485170145e27518368e541a0337fdb44f63249ebf97310300","src/pattern.rs":"6f5909315f4542a989d9b1cee0bb3bc5b6749a665d73e32ec1c8ec4d71b78fac","src/pikevm.rs":"83423e5a94ea36e99dc6f69891ab200c1d0dadd3389ee296a816ec8d68bf556f","src/pool.rs":"942e991ae31ef349bd76efd78b2a712c01166dec965bf93742977ed0870d5a10","src/prog.rs":"78a02dcc1fc7b1d4f37a4a4eeb075eb5cc84aea1736e4de3a2cc7449a9ce5103","src/re_builder.rs":"943344bf6e2fc90902ee04b11b741c32418ac6814b21b7982cc0a3a817713f3e","src/re_bytes.rs":"c19bf2df00024e91f6f2b68be2d8971e847d6f16d7b949a04d2569736b1cdadb","src/re_set.rs":"a0cb76fafe7e33ea8c7b65aae53fa3432fc1651be186218b2284cb3c002ea966","src/re_trait.rs":"1c209fe30392b957f1bdcacdb900f222fc761a2e1634ab1c3f4ee97f315a0c22","src/re_unicode.rs":"122df2eecd8727a4169fe7efb0925a1767fdfa2e290d9b90ca0beaa137b5abfa","src/sparse.rs":"0da3ddb7972109869248a764dbb10254555f4bb51c375e89fb3fab9cafa47320","src/testdata/LICENSE":"58cf078acc03da3e280a938c2bd9943f554fc9b6ced89ad93ba35ca436872899","src/testdata/README":"45f869e37f798905c773bfbe0ef19a5fb7e585cbf0b7c21b5b5a784e8cec3c14","src/testdata/basic.dat":"b5b33aa89d48a61cd67cb1fbfd8f70e62c83e30b86256f9f915a5190dd38ff06","src/testdata/nullsubexpr.dat":"496ac0278eec3b6d9170faace14554569032dd3d909618364d9326156de39ecf","src/testdata/repetition.dat":"1f7959063015b284b18a4a2c1c8b416d438a2d6c4b1a362da43406b865f50e69","src/utf8.rs":"708615a4859110cc9766b342a9c1da6c5c4a8a04ad239046b2725385db977efe","test":"0d62fdca7da12fc19ea5306b5de1d83e68d9365a029c043d524334da138b0304","tests/api.rs":"7b2a0ef75e99b9776094967bd66e9cdeaa8e11359f5f0a12bd08ef0e8d0c11fc","tests/api_str.rs":"2ae38c04e7e8fac008b609a820d0b1561ba75f39b0edc0987d6d3d06132da77f","tests/bytes.rs":"edc50f526c5fee43df89d639ef18b237e4eb91e9d533bfc43f3cbab7417d38ba","tests/consistent.rs":"d69435154c09478076497216e43081a835ac65147181a4fbddad7bff469605b2","tests/crates_regex.rs":"91a59d470e0700b4bcb3ff735d06799f3107b8ef4875a2e9904607b164be0326","tests/crazy.rs":"c0d56380dff19bdd5d7a3eb731d0e2dc564e169a1b73c81e1879b1e87f5f5f77","tests/flags.rs":"05caace2c81a99d2168037f3a38035d4dffe9f85ef3ebd7ef18b1bc6612f1ea8","tests/fowler.rs":"d78cf914de40b1e125cc92b65ccb444d462586bd07b5e05de4e4a1b5de16aa76","tests/macros.rs":"6db70c16fc90df13e6b30d2b606f8b6dd4dc976697967f6ee001b15aab6d0b19","tests/macros_bytes.rs":"a049f528a93173a1bb176cd46932dce1880679f4a1752e099be920f0e4546fd0","tests/macros_str.rs":"e585b1461374c45a2eca44ca045bc3c1fe984b2b4212e432b0c695b420e708b7","tests/misc.rs":"395f52793fa022e4cdda78675b6a6fba1a3106b4b99c834c39f7801574054bd1","tests/multiline.rs":"1b1a3326ed976437c1357f01d81833ece7ea244f38826246eab55cacd5d0862a","tests/noparse.rs":"12b6be0eff3d80779d33c6459396c74c0f6ebf4ddc9f1d33c3e747ea9e3bf268","tests/regression.rs":"1c965fefb8c7a2b1dfdab3e3fdeebaf47846555c50c8005e5537f96a52a3e252","tests/regression_fuzz.rs":"a504ec563e0d23bd2039493b7b1767fe1f831d7d668f6f4b2ecd124fc7899bcd","tests/replace.rs":"0efa042c0d531911e8ac41ce98a6b60236cbf40954102c59f9f6dea78d9d74dd","tests/searcher.rs":"ce35e47b0a276a7e8c9060c6a0b225ffba163aebc61fbc15555a6897fa0e552c","tests/set.rs":"f1e2af6baeeaed3cc99ed347ff516fe7b2eb0027ef64b891502e1486598eaf8a","tests/shortest_match.rs":"a2c94390c0d61bc24796b4c1288c924e90c8c9c6156fdebb858175177a194a42","tests/suffix_reverse.rs":"b95f89397404871227d9efe6df23b9ded147f183db81597e608f693955c668b5","tests/test_backtrack.rs":"b70c5e5f1241efd76dd9f9dd4a4df8a7b38113bd407d1f5f56867f1176177a59","tests/test_backtrack_bytes.rs":"b8a111d4b4109c8bba7e2afb650572c495a14d357fb1f743c1076fb001f704b5","tests/test_backtrack_utf8bytes.rs":"c0c279785d18beac2b4e178e7bf6c14ed235d65f00ca467cfd9c333d79487649","tests/test_crates_regex.rs":"fd9525c2eef0e2f8cb7f787bc2b721bcd0b5d84f3bca49adfe48d657a99c721a","tests/test_default.rs":"c2dfa0298896f86f1be2abf6b0c347a7ca12f95aeac92bf614dc3b86bdfff269","tests/test_default_bytes.rs":"831d3e6bfb882feb15f700e30304bd34328f888fb4c15c7169371e25024ce9a7","tests/test_nfa.rs":"f119fc43a018249c39c813d57096b0654ff69f337345f2bbd9b0e61cc9137285","tests/test_nfa_bytes.rs":"89eae3bef6a1d0bcea6b5de5be35ad72f613f2ceb8b58fe82a6c6ef2ccdc07d0","tests/test_nfa_utf8bytes.rs":"7d830b4aa401887d7cf098b62fed4cd8017ef8b61f625c7c9a2159a6b4cfeb71","tests/unicode.rs":"4bf85f5c3d547fa8b5623194a09b6413067499dfbe7c1d29d8b50bf1cddacf6b","tests/word_boundary.rs":"7081317ddcec1e82dd4a2090a571c6abf2ff4bbfa8cd10395e1eb3f386157fae","tests/word_boundary_ascii.rs":"cd0be5b5b485de0ba7994b42e2864585556c3d2d8bf5eab05b58931d9aaf4b87","tests/word_boundary_unicode.rs":"75dbcc35d3abc0f9795c2ea99e216dc227b0a5b58e9ca5eef767815ff0513921"},"package":"d83f127d94bdbcda4c8cc2e50f6f84f4b611f69c902699ca385a39c3a75f9ff1"}
\ No newline at end of file +{"files":{"CHANGELOG.md":"1034edbdcb6cbd83b9e9a1dc3dfcdfd6c852112ab402a1cc83734191ea7b3a29","Cargo.lock":"27cdf849031e8c491581b3a436b119e2cd0106fcd94e159fc84751d3337b1684","Cargo.toml":"4a10b9986b5b858cb6ff3a8537492f4b6f7394289b991e182bfc10ad29c3323b","HACKING.md":"17818f7a17723608f6bdbe6388ad0a913d4f96f76a16649aaf4e274b1fa0ea97","LICENSE-APACHE":"a60eea817514531668d7e00765731449fe14d059d3249e0bc93b36de45f759f2","LICENSE-MIT":"6485b8ed310d3f0340bf1ad1f47645069ce4069dcc6bb46c7d5c6faf41de1fdb","PERFORMANCE.md":"0d5ef3866386918dfdefb1aa9a28cfe33cb3c8ceeb79f3f8ba5b88253dd95991","README.md":"6125b1c70b9b560412529c54dc6aacdfd39cc82f69b5ad7776fa86e4cb720347","UNICODE.md":"a8a8399540eed000d19420135a527f400247a04572e44d124c786b870f518776","examples/regexdna-input.txt":"156a49710bb3e1ed4bc2bbb0af0f383b747b3d0281453cfff39c296124c598f8","examples/regexdna-output.txt":"35e85b19b70a893d752fd43e54e1e9da08bac43559191cea85b33387c24c4cc1","examples/shootout-regex-dna-bytes.rs":"fa2daedb4e0a05f64f33f4af62fbb0176db998e3676f8637ab684b725367a7b4","examples/shootout-regex-dna-cheat.rs":"1f871a6eaaf8372299fa3c762051112fa89a14235b03f734fc50ebd51ecaee72","examples/shootout-regex-dna-replace.rs":"32ffdf13ac6c4ce3fc32116a048e9cc682aa34cdb8e5beaf565a22addbdcd9ab","examples/shootout-regex-dna-single-cheat.rs":"809f75bf1e1917a53623eb6f1a3ce3b7d2ed98a6a1dbc0bd4853bec49a0c6f94","examples/shootout-regex-dna-single.rs":"1ab14f5703cd4be2e75a2e792e0ba1d322b9e4b14535d396805a4316d577f5bb","examples/shootout-regex-dna.rs":"20ea46ab63f91e3ac6a64e997eadd436a9cbc2f1bdade28e4512052f0e25bc34","rustfmt.toml":"1ca600239a27401c4a43f363cf3f38183a212affc1f31bff3ae93234bbaec228","src/backtrack.rs":"52987d80448f3d7f5d4e3545ddfc09f1f30de7602d9b5489961db4b215a377fd","src/compile.rs":"79a59be2d2db650b5a322e15e9bf1d3227944410bc780fc6089da8f4d2609b77","src/dfa.rs":"10273980d1f08aaff495e11efa240249a2b2c08a4db7c49c8d6759bc65a3b174","src/error.rs":"71c85db839514f26ee024a689061743ea94a34eb7a3291e6c2b69b45a9682d09","src/exec.rs":"21495ab6813598204a444aeea3a0121674081389fd0f07fc3443eb8858b1c677","src/expand.rs":"71220309a3bac797f55129f49e79c03e96efec894ea338c735b78695367e04ca","src/find_byte.rs":"b387247b77e3269f057c3399aefe5a815032c3af918c876f80eb4b282e4eb95e","src/freqs.rs":"255555f3d95b08a5bb3bc2f38d5a06cc100a39c0f0127fe4f50c33afa1cadc65","src/input.rs":"13f49c1bce2fadd04a45b421d374cd0f8b72bef83f7e8fda958962aaccbe799a","src/lib.rs":"de28e1ad68d4b35750667c7fbb47915e6c159ef04b148f16c3507a0a7a682f96","src/literal/imp.rs":"b7f63a861c299bea4baaab17353a420ee339c2cf76d3858c95f39342bd4463e7","src/literal/mod.rs":"533f1d68af088e9485170145e27518368e541a0337fdb44f63249ebf97310300","src/pattern.rs":"993d8b6b4bcea5e02bee3c76e17c356a5a47f8fc53c5555edfd1ebb71c0878bf","src/pikevm.rs":"6c0eaa7e878c945ac4c3c545c98f5706ad04846fc432a5086c8ee78eb030dfa7","src/pool.rs":"942e991ae31ef349bd76efd78b2a712c01166dec965bf93742977ed0870d5a10","src/prog.rs":"bebb3e50745bbc05d6c8240d972ba55a1818c51b1161dc1c21f3fe13c11d4884","src/re_builder.rs":"943344bf6e2fc90902ee04b11b741c32418ac6814b21b7982cc0a3a817713f3e","src/re_bytes.rs":"e2eddc896cea1e878716e77798a8146a67d1d8d9bcf4d053155c1caf3b8f5518","src/re_set.rs":"7921ac4a919b7a5deffe82d099a9ccaf5487aebd890dfb7a661e602c6ad3f1a9","src/re_trait.rs":"d237121b6f6b606836c72305cbcb3bbdbc54d1f6827d19a19cd0fbb4372e0145","src/re_unicode.rs":"ba4d793ff194bfd33a3735e3664c7590f5f166c452e7632a25e4558ffba14e5a","src/sparse.rs":"0da3ddb7972109869248a764dbb10254555f4bb51c375e89fb3fab9cafa47320","src/testdata/LICENSE":"58cf078acc03da3e280a938c2bd9943f554fc9b6ced89ad93ba35ca436872899","src/testdata/README":"45f869e37f798905c773bfbe0ef19a5fb7e585cbf0b7c21b5b5a784e8cec3c14","src/testdata/basic.dat":"b5b33aa89d48a61cd67cb1fbfd8f70e62c83e30b86256f9f915a5190dd38ff06","src/testdata/nullsubexpr.dat":"496ac0278eec3b6d9170faace14554569032dd3d909618364d9326156de39ecf","src/testdata/repetition.dat":"1f7959063015b284b18a4a2c1c8b416d438a2d6c4b1a362da43406b865f50e69","src/utf8.rs":"f85a356ff5d5b19e417b73ce1dd84581b21d283f6dddd195547c30af9c60bd1a","test":"0d62fdca7da12fc19ea5306b5de1d83e68d9365a029c043d524334da138b0304","tests/api.rs":"7b2a0ef75e99b9776094967bd66e9cdeaa8e11359f5f0a12bd08ef0e8d0c11fc","tests/api_str.rs":"2ae38c04e7e8fac008b609a820d0b1561ba75f39b0edc0987d6d3d06132da77f","tests/bytes.rs":"edc50f526c5fee43df89d639ef18b237e4eb91e9d533bfc43f3cbab7417d38ba","tests/consistent.rs":"d69435154c09478076497216e43081a835ac65147181a4fbddad7bff469605b2","tests/crates_regex.rs":"91a59d470e0700b4bcb3ff735d06799f3107b8ef4875a2e9904607b164be0326","tests/crazy.rs":"c0d56380dff19bdd5d7a3eb731d0e2dc564e169a1b73c81e1879b1e87f5f5f77","tests/flags.rs":"05caace2c81a99d2168037f3a38035d4dffe9f85ef3ebd7ef18b1bc6612f1ea8","tests/fowler.rs":"d78cf914de40b1e125cc92b65ccb444d462586bd07b5e05de4e4a1b5de16aa76","tests/macros.rs":"6db70c16fc90df13e6b30d2b606f8b6dd4dc976697967f6ee001b15aab6d0b19","tests/macros_bytes.rs":"a049f528a93173a1bb176cd46932dce1880679f4a1752e099be920f0e4546fd0","tests/macros_str.rs":"e585b1461374c45a2eca44ca045bc3c1fe984b2b4212e432b0c695b420e708b7","tests/misc.rs":"395f52793fa022e4cdda78675b6a6fba1a3106b4b99c834c39f7801574054bd1","tests/multiline.rs":"1b1a3326ed976437c1357f01d81833ece7ea244f38826246eab55cacd5d0862a","tests/noparse.rs":"12b6be0eff3d80779d33c6459396c74c0f6ebf4ddc9f1d33c3e747ea9e3bf268","tests/regression.rs":"1c965fefb8c7a2b1dfdab3e3fdeebaf47846555c50c8005e5537f96a52a3e252","tests/regression_fuzz.rs":"a504ec563e0d23bd2039493b7b1767fe1f831d7d668f6f4b2ecd124fc7899bcd","tests/replace.rs":"0efa042c0d531911e8ac41ce98a6b60236cbf40954102c59f9f6dea78d9d74dd","tests/searcher.rs":"ce35e47b0a276a7e8c9060c6a0b225ffba163aebc61fbc15555a6897fa0e552c","tests/set.rs":"f1e2af6baeeaed3cc99ed347ff516fe7b2eb0027ef64b891502e1486598eaf8a","tests/shortest_match.rs":"a2c94390c0d61bc24796b4c1288c924e90c8c9c6156fdebb858175177a194a42","tests/suffix_reverse.rs":"b95f89397404871227d9efe6df23b9ded147f183db81597e608f693955c668b5","tests/test_backtrack.rs":"b70c5e5f1241efd76dd9f9dd4a4df8a7b38113bd407d1f5f56867f1176177a59","tests/test_backtrack_bytes.rs":"b8a111d4b4109c8bba7e2afb650572c495a14d357fb1f743c1076fb001f704b5","tests/test_backtrack_utf8bytes.rs":"c0c279785d18beac2b4e178e7bf6c14ed235d65f00ca467cfd9c333d79487649","tests/test_crates_regex.rs":"fd9525c2eef0e2f8cb7f787bc2b721bcd0b5d84f3bca49adfe48d657a99c721a","tests/test_default.rs":"c2dfa0298896f86f1be2abf6b0c347a7ca12f95aeac92bf614dc3b86bdfff269","tests/test_default_bytes.rs":"831d3e6bfb882feb15f700e30304bd34328f888fb4c15c7169371e25024ce9a7","tests/test_nfa.rs":"f119fc43a018249c39c813d57096b0654ff69f337345f2bbd9b0e61cc9137285","tests/test_nfa_bytes.rs":"89eae3bef6a1d0bcea6b5de5be35ad72f613f2ceb8b58fe82a6c6ef2ccdc07d0","tests/test_nfa_utf8bytes.rs":"7d830b4aa401887d7cf098b62fed4cd8017ef8b61f625c7c9a2159a6b4cfeb71","tests/unicode.rs":"1af9db7f09a6b0113b8a64733e06c8415fef720b2fdef227ae398d94332287cd","tests/word_boundary.rs":"7081317ddcec1e82dd4a2090a571c6abf2ff4bbfa8cd10395e1eb3f386157fae","tests/word_boundary_ascii.rs":"cd0be5b5b485de0ba7994b42e2864585556c3d2d8bf5eab05b58931d9aaf4b87","tests/word_boundary_unicode.rs":"75dbcc35d3abc0f9795c2ea99e216dc227b0a5b58e9ca5eef767815ff0513921"},"package":"4c4eb3267174b8c6c2f654116623910a0fef09c4753f8dd83db29c48a0df988b"}
\ No newline at end of file diff --git a/vendor/regex/CHANGELOG.md b/vendor/regex/CHANGELOG.md index 26f9955b8..f2d45ed73 100644 --- a/vendor/regex/CHANGELOG.md +++ b/vendor/regex/CHANGELOG.md @@ -1,3 +1,24 @@ +1.6.0 (2022-07-05) +================== +This release principally includes an upgrade to Unicode 14. + +New features: + +* [FEATURE #832](https://github.com/rust-lang/regex/pull/832): + Clarify that `Captures::len` includes all groups, not just matching groups. +* [FEATURE #857](https://github.com/rust-lang/regex/pull/857): + Add an `ExactSizeIterator` impl for `SubCaptureMatches`. +* [FEATURE #861](https://github.com/rust-lang/regex/pull/861): + Improve `RegexSet` documentation examples. +* [FEATURE #877](https://github.com/rust-lang/regex/issues/877): + Upgrade to Unicode 14. + +Bug fixes: + +* [BUG #792](https://github.com/rust-lang/regex/issues/792): + Fix error message rendering bug. + + 1.5.6 (2022-05-20) ================== This release includes a few bug fixes, including a bug that produced incorrect diff --git a/vendor/regex/Cargo.lock b/vendor/regex/Cargo.lock index 897a4a32f..fd2c1101d 100644 --- a/vendor/regex/Cargo.lock +++ b/vendor/regex/Cargo.lock @@ -19,9 +19,9 @@ checksum = "baf1de4339761588bc0619e3cbc0120ee582ebb74b53b4efbf79117bd2da40fd" [[package]] name = "getrandom" -version = "0.2.6" +version = "0.2.7" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "9be70c98951c83b8d2f8f60d7065fa6d5146873094452a1008da8c2f1e4205ad" +checksum = "4eb1a864a501629691edf6c15a593b7a51eebaa1e8468e9ddc623de7c9b58ec6" dependencies = [ "cfg-if", "libc", @@ -36,9 +36,9 @@ checksum = "e2abad23fbc42b3700f2f279844dc832adb2b2eb069b2df918f455c4e18cc646" [[package]] name = "libc" -version = "0.2.125" +version = "0.2.126" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "5916d2ae698f6de9bfb891ad7a8d65c09d232dc58cc4ac433c7da3b2fd84bc2b" +checksum = "349d5a591cd28b49e1d1037471617a32ddcda5731b99419008085f72d5a53836" [[package]] name = "memchr" @@ -75,7 +75,7 @@ dependencies = [ [[package]] name = "regex" -version = "1.5.6" +version = "1.6.0" dependencies = [ "aho-corasick", "lazy_static", @@ -87,12 +87,12 @@ dependencies = [ [[package]] name = "regex-syntax" -version = "0.6.26" +version = "0.6.27" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "49b3de9ec5dc0a3417da371aab17d729997c15010e7fd24ff707773a33bddb64" +checksum = "a3f87b73ce11b1619a3c6332f45341e0047173771e8b8b73f87bfeefb7b56244" [[package]] name = "wasi" -version = "0.10.2+wasi-snapshot-preview1" +version = "0.11.0+wasi-snapshot-preview1" source = "registry+https://github.com/rust-lang/crates.io-index" -checksum = "fd6fbd9a79829dd1ad0cc20627bf1ed606756a7f77edff7b66b7064f9cb327c6" +checksum = "9c8d87e72b64a3b4db28d11ce29237c246188f4f51057d65a7eab63b7987e423" diff --git a/vendor/regex/Cargo.toml b/vendor/regex/Cargo.toml index 772bdb372..abe6ac033 100644 --- a/vendor/regex/Cargo.toml +++ b/vendor/regex/Cargo.toml @@ -12,7 +12,7 @@ [package] edition = "2018" name = "regex" -version = "1.5.6" +version = "1.6.0" authors = ["The Rust Project Developers"] exclude = [ "/scripts/*", @@ -88,7 +88,7 @@ version = "2.4.0" optional = true [dependencies.regex-syntax] -version = "0.6.26" +version = "0.6.27" default-features = false [dev-dependencies.lazy_static] diff --git a/vendor/regex/src/backtrack.rs b/vendor/regex/src/backtrack.rs index a3d25d662..4d83856ca 100644 --- a/vendor/regex/src/backtrack.rs +++ b/vendor/regex/src/backtrack.rs @@ -93,13 +93,7 @@ impl<'a, 'm, 'r, 's, I: Input> Bounded<'a, 'm, 'r, 's, I> { let mut cache = cache.borrow_mut(); let cache = &mut cache.backtrack; let start = input.at(start); - let mut b = Bounded { - prog: prog, - input: input, - matches: matches, - slots: slots, - m: cache, - }; + let mut b = Bounded { prog, input, matches, slots, m: cache }; b.exec_(start, end) } @@ -220,14 +214,14 @@ impl<'a, 'm, 'r, 's, I: Input> Bounded<'a, 'm, 'r, 's, I> { // job is popped and the old capture index is restored. self.m.jobs.push(Job::SaveRestore { slot: inst.slot, - old_pos: old_pos, + old_pos, }); self.slots[inst.slot] = Some(at.pos()); } ip = inst.goto; } Split(ref inst) => { - self.m.jobs.push(Job::Inst { ip: inst.goto2, at: at }); + self.m.jobs.push(Job::Inst { ip: inst.goto2, at }); ip = inst.goto1; } EmptyLook(ref inst) => { diff --git a/vendor/regex/src/compile.rs b/vendor/regex/src/compile.rs index 069f445c8..90ca25015 100644 --- a/vendor/regex/src/compile.rs +++ b/vendor/regex/src/compile.rs @@ -149,7 +149,8 @@ impl Compiler { self.compiled.start = dotstar_patch.entry; } self.compiled.captures = vec![None]; - let patch = self.c_capture(0, expr)?.unwrap_or(self.next_inst()); + let patch = + self.c_capture(0, expr)?.unwrap_or_else(|| self.next_inst()); if self.compiled.needs_dotstar() { self.fill(dotstar_patch.hole, patch.entry); } else { @@ -185,7 +186,7 @@ impl Compiler { self.fill_to_next(prev_hole); let split = self.push_split_hole(); let Patch { hole, entry } = - self.c_capture(0, expr)?.unwrap_or(self.next_inst()); + self.c_capture(0, expr)?.unwrap_or_else(|| self.next_inst()); self.fill_to_next(hole); self.compiled.matches.push(self.insts.len()); self.push_compiled(Inst::Match(i)); @@ -193,7 +194,7 @@ impl Compiler { } let i = exprs.len() - 1; let Patch { hole, entry } = - self.c_capture(0, &exprs[i])?.unwrap_or(self.next_inst()); + self.c_capture(0, &exprs[i])?.unwrap_or_else(|| self.next_inst()); self.fill(prev_hole, entry); self.fill_to_next(hole); self.compiled.matches.push(self.insts.len()); @@ -410,11 +411,11 @@ impl Compiler { } else { let entry = self.insts.len(); let hole = self.push_hole(InstHole::Save { slot: first_slot }); - let patch = self.c(expr)?.unwrap_or(self.next_inst()); + let patch = self.c(expr)?.unwrap_or_else(|| self.next_inst()); self.fill(hole, patch.entry); self.fill_to_next(patch.hole); let hole = self.push_hole(InstHole::Save { slot: first_slot + 1 }); - Ok(Some(Patch { hole: hole, entry: entry })) + Ok(Some(Patch { hole, entry })) } } @@ -448,7 +449,7 @@ impl Compiler { self.c_class(&[hir::ClassUnicodeRange::new(c, c)]) } } else { - let hole = self.push_hole(InstHole::Char { c: c }); + let hole = self.push_hole(InstHole::Char { c }); Ok(Some(Patch { hole, entry: self.insts.len() - 1 })) } } @@ -458,7 +459,7 @@ impl Compiler { assert!(!ranges.is_empty()); if self.compiled.uses_bytes() { - Ok(Some(CompileClass { c: self, ranges: ranges }.compile()?)) + Ok(Some(CompileClass { c: self, ranges }.compile()?)) } else { let ranges: Vec<(char, char)> = ranges.iter().map(|r| (r.start(), r.end())).collect(); @@ -467,9 +468,9 @@ impl Compiler { } else { self.extra_inst_bytes += ranges.len() * (size_of::<char>() * 2); - self.push_hole(InstHole::Ranges { ranges: ranges }) + self.push_hole(InstHole::Ranges { ranges }) }; - Ok(Some(Patch { hole: hole, entry: self.insts.len() - 1 })) + Ok(Some(Patch { hole, entry: self.insts.len() - 1 })) } } @@ -508,8 +509,8 @@ impl Compiler { } fn c_empty_look(&mut self, look: EmptyLook) -> ResultOrEmpty { - let hole = self.push_hole(InstHole::EmptyLook { look: look }); - Ok(Some(Patch { hole: hole, entry: self.insts.len() - 1 })) + let hole = self.push_hole(InstHole::EmptyLook { look }); + Ok(Some(Patch { hole, entry: self.insts.len() - 1 })) } fn c_concat<'a, I>(&mut self, exprs: I) -> ResultOrEmpty @@ -533,7 +534,7 @@ impl Compiler { hole = p.hole; } } - Ok(Some(Patch { hole: hole, entry: entry })) + Ok(Some(Patch { hole, entry })) } fn c_alternate(&mut self, exprs: &[Hir]) -> ResultOrEmpty { @@ -676,7 +677,7 @@ impl Compiler { // None). let patch_concat = self .c_concat(iter::repeat(expr).take(min))? - .unwrap_or(self.next_inst()); + .unwrap_or_else(|| self.next_inst()); if let Some(patch_rep) = self.c_repeat_zero_or_more(expr, greedy)? { self.fill(patch_concat.hole, patch_rep.entry); Ok(Some(Patch { hole: patch_rep.hole, entry: patch_concat.entry })) @@ -700,7 +701,7 @@ impl Compiler { } // Same reasoning as in c_repeat_range_min_or_more (we know that min < // max at this point). - let patch_concat = patch_concat.unwrap_or(self.next_inst()); + let patch_concat = patch_concat.unwrap_or_else(|| self.next_inst()); let initial_entry = patch_concat.entry; // It is much simpler to compile, e.g., `a{2,5}` as: // @@ -879,14 +880,14 @@ impl MaybeInst { } MaybeInst::Split1(goto1) => { MaybeInst::Compiled(Inst::Split(InstSplit { - goto1: goto1, + goto1, goto2: goto, })) } MaybeInst::Split2(goto2) => { MaybeInst::Compiled(Inst::Split(InstSplit { goto1: goto, - goto2: goto2, + goto2, })) } _ => unreachable!( @@ -900,9 +901,7 @@ impl MaybeInst { fn fill_split(&mut self, goto1: InstPtr, goto2: InstPtr) { let filled = match *self { - MaybeInst::Split => { - Inst::Split(InstSplit { goto1: goto1, goto2: goto2 }) - } + MaybeInst::Split => Inst::Split(InstSplit { goto1, goto2 }), _ => unreachable!( "must be called on Split instruction, \ instead it was called on: {:?}", @@ -960,19 +959,17 @@ enum InstHole { impl InstHole { fn fill(&self, goto: InstPtr) -> Inst { match *self { - InstHole::Save { slot } => { - Inst::Save(InstSave { goto: goto, slot: slot }) - } + InstHole::Save { slot } => Inst::Save(InstSave { goto, slot }), InstHole::EmptyLook { look } => { - Inst::EmptyLook(InstEmptyLook { goto: goto, look: look }) + Inst::EmptyLook(InstEmptyLook { goto, look }) } - InstHole::Char { c } => Inst::Char(InstChar { goto: goto, c: c }), + InstHole::Char { c } => Inst::Char(InstChar { goto, c }), InstHole::Ranges { ref ranges } => Inst::Ranges(InstRanges { - goto: goto, + goto, ranges: ranges.clone().into_boxed_slice(), }), InstHole::Bytes { start, end } => { - Inst::Bytes(InstBytes { goto: goto, start: start, end: end }) + Inst::Bytes(InstBytes { goto, start, end }) } } } @@ -1042,7 +1039,7 @@ impl<'a, 'b> CompileClass<'a, 'b> { let mut last_hole = Hole::None; for byte_range in seq { let key = SuffixCacheKey { - from_inst: from_inst, + from_inst, start: byte_range.start, end: byte_range.end, }; @@ -1132,7 +1129,7 @@ impl SuffixCache { } } *pos = self.dense.len(); - self.dense.push(SuffixCacheEntry { key: key, pc: pc }); + self.dense.push(SuffixCacheEntry { key, pc }); None } @@ -1143,8 +1140,8 @@ impl SuffixCache { fn hash(&self, suffix: &SuffixCacheKey) -> usize { // Basic FNV-1a hash as described: // https://en.wikipedia.org/wiki/Fowler%E2%80%93Noll%E2%80%93Vo_hash_function - const FNV_PRIME: u64 = 1099511628211; - let mut h = 14695981039346656037; + const FNV_PRIME: u64 = 1_099_511_628_211; + let mut h = 14_695_981_039_346_656_037; h = (h ^ (suffix.from_inst as u64)).wrapping_mul(FNV_PRIME); h = (h ^ (suffix.start as u64)).wrapping_mul(FNV_PRIME); h = (h ^ (suffix.end as u64)).wrapping_mul(FNV_PRIME); diff --git a/vendor/regex/src/dfa.rs b/vendor/regex/src/dfa.rs index 4aee8039c..dc9952120 100644 --- a/vendor/regex/src/dfa.rs +++ b/vendor/regex/src/dfa.rs @@ -31,7 +31,7 @@ considerably more complex than one might expect out of a DFA. A number of tricks are employed to make it fast. Tread carefully. N.B. While this implementation is heavily commented, Russ Cox's series of -articles on regexes is strongly recommended: https://swtch.com/~rsc/regexp/ +articles on regexes is strongly recommended: <https://swtch.com/~rsc/regexp/> (As is the DFA implementation in RE2, which heavily influenced this implementation.) */ @@ -454,10 +454,10 @@ impl<'a> Fsm<'a> { let mut cache = cache.borrow_mut(); let cache = &mut cache.dfa; let mut dfa = Fsm { - prog: prog, + prog, start: 0, // filled in below - at: at, - quit_after_match: quit_after_match, + at, + quit_after_match, last_match_si: STATE_UNKNOWN, last_cache_flush: at, cache: &mut cache.inner, @@ -484,10 +484,10 @@ impl<'a> Fsm<'a> { let mut cache = cache.borrow_mut(); let cache = &mut cache.dfa_reverse; let mut dfa = Fsm { - prog: prog, + prog, start: 0, // filled in below - at: at, - quit_after_match: quit_after_match, + at, + quit_after_match, last_match_si: STATE_UNKNOWN, last_cache_flush: at, cache: &mut cache.inner, @@ -515,9 +515,9 @@ impl<'a> Fsm<'a> { let mut cache = cache.borrow_mut(); let cache = &mut cache.dfa; let mut dfa = Fsm { - prog: prog, + prog, start: 0, // filled in below - at: at, + at, quit_after_match: false, last_match_si: STATE_UNKNOWN, last_cache_flush: at, @@ -1606,11 +1606,7 @@ struct StateMap { impl StateMap { fn new(num_byte_classes: usize) -> StateMap { - StateMap { - map: HashMap::new(), - states: vec![], - num_byte_classes: num_byte_classes, - } + StateMap { map: HashMap::new(), states: vec![], num_byte_classes } } fn len(&self) -> usize { @@ -1646,7 +1642,7 @@ impl Transitions { /// The number of byte classes corresponds to the stride. Every state will /// have `num_byte_classes` slots for transitions. fn new(num_byte_classes: usize) -> Transitions { - Transitions { table: vec![], num_byte_classes: num_byte_classes } + Transitions { table: vec![], num_byte_classes } } /// Returns the total number of states currently in this table. @@ -1696,27 +1692,27 @@ impl Transitions { impl StateFlags { fn is_match(&self) -> bool { - self.0 & 0b0000000_1 > 0 + self.0 & 0b0000_0001 > 0 } fn set_match(&mut self) { - self.0 |= 0b0000000_1; + self.0 |= 0b0000_0001; } fn is_word(&self) -> bool { - self.0 & 0b000000_1_0 > 0 + self.0 & 0b0000_0010 > 0 } fn set_word(&mut self) { - self.0 |= 0b000000_1_0; + self.0 |= 0b0000_0010; } fn has_empty(&self) -> bool { - self.0 & 0b00000_1_00 > 0 + self.0 & 0b0000_0100 > 0 } fn set_empty(&mut self) { - self.0 |= 0b00000_1_00; + self.0 |= 0b0000_0100; } } diff --git a/vendor/regex/src/exec.rs b/vendor/regex/src/exec.rs index d5fad1c0e..e75ca083a 100644 --- a/vendor/regex/src/exec.rs +++ b/vendor/regex/src/exec.rs @@ -288,10 +288,10 @@ impl ExecBuilder { exprs.push(expr); } Ok(Parsed { - exprs: exprs, + exprs, prefixes: prefixes.unwrap_or_else(Literals::empty), suffixes: suffixes.unwrap_or_else(Literals::empty), - bytes: bytes, + bytes, }) } @@ -311,7 +311,7 @@ impl ExecBuilder { match_type: MatchType::Nothing, }); let pool = ExecReadOnly::new_pool(&ro); - return Ok(Exec { ro: ro, pool }); + return Ok(Exec { ro, pool }); } let parsed = self.parse()?; let mut nfa = Compiler::new() @@ -340,12 +340,12 @@ impl ExecBuilder { let mut ro = ExecReadOnly { res: self.options.pats, - nfa: nfa, - dfa: dfa, - dfa_reverse: dfa_reverse, + nfa, + dfa, + dfa_reverse, suffixes: LiteralSearcher::suffixes(parsed.suffixes), #[cfg(feature = "perf-literal")] - ac: ac, + ac, match_type: MatchType::Nothing, }; ro.match_type = ro.choose_match_type(self.match_type); diff --git a/vendor/regex/src/expand.rs b/vendor/regex/src/expand.rs index fd9c2d05d..67b514926 100644 --- a/vendor/regex/src/expand.rs +++ b/vendor/regex/src/expand.rs @@ -127,7 +127,7 @@ impl From<usize> for Ref<'static> { /// If no such valid reference could be found, None is returned. fn find_cap_ref(replacement: &[u8]) -> Option<CaptureRef<'_>> { let mut i = 0; - let rep: &[u8] = replacement.as_ref(); + let rep: &[u8] = replacement; if rep.len() <= 1 || rep[0] != b'$' { return None; } @@ -136,7 +136,7 @@ fn find_cap_ref(replacement: &[u8]) -> Option<CaptureRef<'_>> { return find_cap_ref_braced(rep, i + 1); } let mut cap_end = i; - while rep.get(cap_end).map_or(false, is_valid_cap_letter) { + while rep.get(cap_end).copied().map_or(false, is_valid_cap_letter) { cap_end += 1; } if cap_end == i { @@ -183,8 +183,8 @@ fn find_cap_ref_braced(rep: &[u8], mut i: usize) -> Option<CaptureRef<'_>> { } /// Returns true if and only if the given byte is allowed in a capture name. -fn is_valid_cap_letter(b: &u8) -> bool { - match *b { +fn is_valid_cap_letter(b: u8) -> bool { + match b { b'0'..=b'9' | b'a'..=b'z' | b'A'..=b'Z' | b'_' => true, _ => false, } diff --git a/vendor/regex/src/input.rs b/vendor/regex/src/input.rs index 5d50ee340..df6c3e0c9 100644 --- a/vendor/regex/src/input.rs +++ b/vendor/regex/src/input.rs @@ -160,7 +160,7 @@ impl<'t> Input for CharInput<'t> { InputAt { pos: self.len(), c: None.into(), byte: None, len: 0 } } else { let c = decode_utf8(&self[i..]).map(|(c, _)| c).into(); - InputAt { pos: i, c: c, byte: None, len: c.len_utf8() } + InputAt { pos: i, c, byte: None, len: c.len_utf8() } } } @@ -231,7 +231,7 @@ pub struct ByteInput<'t> { impl<'t> ByteInput<'t> { /// Return a new byte-based input reader for the given string. pub fn new(text: &'t [u8], only_utf8: bool) -> ByteInput<'t> { - ByteInput { text: text, only_utf8: only_utf8 } + ByteInput { text, only_utf8 } } } diff --git a/vendor/regex/src/lib.rs b/vendor/regex/src/lib.rs index 7f2dec815..3e3b0a007 100644 --- a/vendor/regex/src/lib.rs +++ b/vendor/regex/src/lib.rs @@ -628,7 +628,6 @@ pub use crate::re_builder::unicode::*; #[cfg(feature = "std")] pub use crate::re_set::unicode::*; #[cfg(feature = "std")] -#[cfg(feature = "std")] pub use crate::re_unicode::{ escape, CaptureLocations, CaptureMatches, CaptureNames, Captures, Locations, Match, Matches, NoExpand, Regex, Replacer, ReplacerRef, Split, diff --git a/vendor/regex/src/literal/imp.rs b/vendor/regex/src/literal/imp.rs index 82f050a0d..90b2f1160 100644 --- a/vendor/regex/src/literal/imp.rs +++ b/vendor/regex/src/literal/imp.rs @@ -57,10 +57,10 @@ impl LiteralSearcher { fn new(lits: Literals, matcher: Matcher) -> Self { let complete = lits.all_complete(); LiteralSearcher { - complete: complete, + complete, lcp: Memmem::new(lits.longest_common_prefix()), lcs: Memmem::new(lits.longest_common_suffix()), - matcher: matcher, + matcher, } } diff --git a/vendor/regex/src/pattern.rs b/vendor/regex/src/pattern.rs index b4ffd8e16..00549e510 100644 --- a/vendor/regex/src/pattern.rs +++ b/vendor/regex/src/pattern.rs @@ -15,7 +15,7 @@ impl<'r, 't> Pattern<'t> for &'r Regex { fn into_searcher(self, haystack: &'t str) -> RegexSearcher<'r, 't> { RegexSearcher { - haystack: haystack, + haystack, it: self.find_iter(haystack), last_step_end: 0, next_match: None, diff --git a/vendor/regex/src/pikevm.rs b/vendor/regex/src/pikevm.rs index 9a1424086..8c9eac2d3 100644 --- a/vendor/regex/src/pikevm.rs +++ b/vendor/regex/src/pikevm.rs @@ -100,7 +100,7 @@ impl<'r, I: Input> Fsm<'r, I> { cache.clist.resize(prog.len(), prog.captures.len()); cache.nlist.resize(prog.len(), prog.captures.len()); let at = input.at(start); - Fsm { prog: prog, stack: &mut cache.stack, input: input }.exec_( + Fsm { prog, stack: &mut cache.stack, input }.exec_( &mut cache.clist, &mut cache.nlist, matches, diff --git a/vendor/regex/src/prog.rs b/vendor/regex/src/prog.rs index 475a8112a..c211f71d8 100644 --- a/vendor/regex/src/prog.rs +++ b/vendor/regex/src/prog.rs @@ -233,7 +233,7 @@ impl fmt::Debug for Program { if pc == self.start { write!(f, " (start)")?; } - write!(f, "\n")?; + writeln!(f)?; } Ok(()) } diff --git a/vendor/regex/src/re_bytes.rs b/vendor/regex/src/re_bytes.rs index ae55d6d25..d71969257 100644 --- a/vendor/regex/src/re_bytes.rs +++ b/vendor/regex/src/re_bytes.rs @@ -53,7 +53,7 @@ impl<'t> Match<'t> { /// Creates a new match from the given haystack and byte offsets. #[inline] fn new(haystack: &'t [u8], start: usize, end: usize) -> Match<'t> { - Match { text: haystack, start: start, end: end } + Match { text: haystack, start, end } } } @@ -255,7 +255,7 @@ impl Regex { pub fn captures<'t>(&self, text: &'t [u8]) -> Option<Captures<'t>> { let mut locs = self.capture_locations(); self.captures_read_at(&mut locs, text, 0).map(move |_| Captures { - text: text, + text, locs: locs.0, named_groups: self.0.capture_name_idx().clone(), }) @@ -578,7 +578,7 @@ impl Regex { /// context into consideration. For example, the `\A` anchor can only /// match when `start == 0`. pub fn is_match_at(&self, text: &[u8], start: usize) -> bool { - self.shortest_match_at(text, start).is_some() + self.0.searcher().is_match_at(text, start) } /// Returns the same as find, but starts the search at the given @@ -723,7 +723,7 @@ impl<'r, 't> Iterator for CaptureMatches<'r, 't> { fn next(&mut self) -> Option<Captures<'t>> { self.0.next().map(|locs| Captures { text: self.0.text(), - locs: locs, + locs, named_groups: self.0.regex().capture_name_idx().clone(), }) } @@ -877,7 +877,7 @@ impl CaptureLocations { self.0.pos(i) } - /// Returns the total number of capturing groups. + /// Returns the total number of capture groups (even if they didn't match). /// /// This is always at least `1` since every regex has at least `1` /// capturing group that corresponds to the entire match. @@ -979,7 +979,7 @@ impl<'t> Captures<'t> { expand_bytes(self, replacement, dst) } - /// Returns the number of captured groups. + /// Returns the total number of capture groups (even if they didn't match). /// /// This is always at least `1`, since every regex has at least one capture /// group that corresponds to the full match. diff --git a/vendor/regex/src/re_set.rs b/vendor/regex/src/re_set.rs index 73d59532e..a6d886d76 100644 --- a/vendor/regex/src/re_set.rs +++ b/vendor/regex/src/re_set.rs @@ -59,13 +59,45 @@ $(#[$doc_regexset_example])* /// 1. Does any regex in the set match? /// 2. If so, which regexes in the set match? /// -/// As with the main `Regex` type, it is cheaper to ask (1) instead of (2) -/// since the matching engines can stop after the first match is found. +/// As with the main [`Regex`][crate::Regex] type, it is cheaper to ask (1) +/// instead of (2) since the matching engines can stop after the first match +/// is found. /// -/// Other features like finding the location of successive matches or their -/// sub-captures aren't supported. If you need this functionality, the -/// recommended approach is to compile each regex in the set independently and -/// selectively match them based on which regexes in the set matched. +/// You cannot directly extract [`Match`][crate::Match] or +/// [`Captures`][crate::Captures] objects from a regex set. If you need these +/// operations, the recommended approach is to compile each pattern in the set +/// independently and scan the exact same input a second time with those +/// independently compiled patterns: +/// +/// ```rust +/// use regex::{Regex, RegexSet}; +/// +/// let patterns = ["foo", "bar"]; +/// // Both patterns will match different ranges of this string. +/// let text = "barfoo"; +/// +/// // Compile a set matching any of our patterns. +/// let set = RegexSet::new(&patterns).unwrap(); +/// // Compile each pattern independently. +/// let regexes: Vec<_> = set.patterns().iter() +/// .map(|pat| Regex::new(pat).unwrap()) +/// .collect(); +/// +/// // Match against the whole set first and identify the individual +/// // matching patterns. +/// let matches: Vec<&str> = set.matches(text).into_iter() +/// // Dereference the match index to get the corresponding +/// // compiled pattern. +/// .map(|match_idx| ®exes[match_idx]) +/// // To get match locations or any other info, we then have to search +/// // the exact same text again, using our separately-compiled pattern. +/// .map(|pat| pat.find(text).unwrap().as_str()) +/// .collect(); +/// +/// // Matches arrive in the order the constituent patterns were declared, +/// // not the order they appear in the input. +/// assert_eq!(vec!["foo", "bar"], matches); +/// ``` /// /// # Performance /// diff --git a/vendor/regex/src/re_trait.rs b/vendor/regex/src/re_trait.rs index 680aa5459..d0c717df5 100644 --- a/vendor/regex/src/re_trait.rs +++ b/vendor/regex/src/re_trait.rs @@ -74,8 +74,19 @@ impl<'c> Iterator for SubCapturesPosIter<'c> { self.idx += 1; x } + + fn size_hint(&self) -> (usize, Option<usize>) { + let len = self.locs.len() - self.idx; + (len, Some(len)) + } + + fn count(self) -> usize { + self.len() + } } +impl<'c> ExactSizeIterator for SubCapturesPosIter<'c> {} + impl<'c> FusedIterator for SubCapturesPosIter<'c> {} /// `RegularExpression` describes types that can implement regex searching. @@ -139,7 +150,7 @@ pub trait RegularExpression: Sized + fmt::Debug { /// Returns an iterator over all non-overlapping successive leftmost-first /// matches. fn find_iter(self, text: &Self::Text) -> Matches<'_, Self> { - Matches { re: self, text: text, last_end: 0, last_match: None } + Matches { re: self, text, last_end: 0, last_match: None } } /// Returns an iterator over all non-overlapping successive leftmost-first diff --git a/vendor/regex/src/re_unicode.rs b/vendor/regex/src/re_unicode.rs index e4871a621..60d81a7d9 100644 --- a/vendor/regex/src/re_unicode.rs +++ b/vendor/regex/src/re_unicode.rs @@ -61,7 +61,7 @@ impl<'t> Match<'t> { /// Creates a new match from the given haystack and byte offsets. #[inline] fn new(haystack: &'t str, start: usize, end: usize) -> Match<'t> { - Match { text: haystack, start: start, end: end } + Match { text: haystack, start, end } } } @@ -129,7 +129,7 @@ impl<'t> From<Match<'t>> for Range<usize> { /// assert!(haystack.contains(&re)); /// assert_eq!(haystack.find(&re), Some(1)); /// assert_eq!(haystack.match_indices(&re).collect::<Vec<_>>(), -/// vec![(1, 4), (5, 8)]); +/// vec![(1, "111"), (5, "222")]); /// assert_eq!(haystack.split(&re).collect::<Vec<_>>(), vec!["a", "b", "c"]); /// ``` #[derive(Clone)] @@ -311,7 +311,7 @@ impl Regex { pub fn captures<'t>(&self, text: &'t str) -> Option<Captures<'t>> { let mut locs = self.capture_locations(); self.captures_read_at(&mut locs, text, 0).map(move |_| Captures { - text: text, + text, locs: locs.0, named_groups: self.0.capture_name_idx().clone(), }) @@ -636,7 +636,7 @@ impl Regex { /// context into consideration. For example, the `\A` anchor can only /// match when `start == 0`. pub fn is_match_at(&self, text: &str, start: usize) -> bool { - self.shortest_match_at(text, start).is_some() + self.0.searcher_str().is_match_at(text, start) } /// Returns the same as find, but starts the search at the given @@ -887,7 +887,7 @@ impl CaptureLocations { self.0.pos(i) } - /// Returns the total number of capturing groups. + /// Returns the total number of capture groups (even if they didn't match). /// /// This is always at least `1` since every regex has at least `1` /// capturing group that corresponds to the entire match. @@ -989,7 +989,7 @@ impl<'t> Captures<'t> { expand_str(self, replacement, dst) } - /// Returns the number of captured groups. + /// Returns the total number of capture groups (even if they didn't match). /// /// This is always at least `1`, since every regex has at least one capture /// group that corresponds to the full match. @@ -1092,8 +1092,18 @@ impl<'c, 't> Iterator for SubCaptureMatches<'c, 't> { .next() .map(|cap| cap.map(|(s, e)| Match::new(self.caps.text, s, e))) } + + fn size_hint(&self) -> (usize, Option<usize>) { + self.it.size_hint() + } + + fn count(self) -> usize { + self.it.count() + } } +impl<'c, 't> ExactSizeIterator for SubCaptureMatches<'c, 't> {} + impl<'c, 't> FusedIterator for SubCaptureMatches<'c, 't> {} /// An iterator that yields all non-overlapping capture groups matching a @@ -1114,7 +1124,7 @@ impl<'r, 't> Iterator for CaptureMatches<'r, 't> { fn next(&mut self) -> Option<Captures<'t>> { self.0.next().map(|locs| Captures { text: self.0.text(), - locs: locs, + locs, named_groups: self.0.regex().capture_name_idx().clone(), }) } diff --git a/vendor/regex/src/utf8.rs b/vendor/regex/src/utf8.rs index 6e0608fdb..2dfd2c0d1 100644 --- a/vendor/regex/src/utf8.rs +++ b/vendor/regex/src/utf8.rs @@ -108,7 +108,7 @@ pub fn decode_utf8(src: &[u8]) -> Option<(char, usize)> { | ((b2 & !TAG_CONT) as u32) << 6 | ((b3 & !TAG_CONT) as u32); match cp { - 0x10000..=0x10FFFF => char::from_u32(cp).map(|cp| (cp, 4)), + 0x10000..=0x0010_FFFF => char::from_u32(cp).map(|cp| (cp, 4)), _ => None, } } diff --git a/vendor/regex/tests/unicode.rs b/vendor/regex/tests/unicode.rs index 9f1cd0c01..9b3228624 100644 --- a/vendor/regex/tests/unicode.rs +++ b/vendor/regex/tests/unicode.rs @@ -232,3 +232,20 @@ mat!(uni_class_sb2, r"\p{sb=lower}", "\u{0469}", Some((0, 2))); mat!(uni_class_sb3, r"\p{sb=Close}", "\u{FF60}", Some((0, 3))); mat!(uni_class_sb4, r"\p{sb=Close}", "\u{1F677}", Some((0, 4))); mat!(uni_class_sb5, r"\p{sb=SContinue}", "\u{FF64}", Some((0, 3))); + +// Test 'Vithkuqi' support, which was added in Unicode 14. +// See: https://github.com/rust-lang/regex/issues/877 +mat!( + uni_vithkuqi_literal_upper, + r"(?i)^\u{10570}$", + "\u{10570}", + Some((0, 4)) +); +mat!( + uni_vithkuqi_literal_lower, + r"(?i)^\u{10570}$", + "\u{10597}", + Some((0, 4)) +); +mat!(uni_vithkuqi_word_upper, r"^\w$", "\u{10570}", Some((0, 4))); +mat!(uni_vithkuqi_word_lower, r"^\w$", "\u{10597}", Some((0, 4))); |