Upgrade rust/crates/unicode-segmentation to 1.7.1
Test: m libunicode_segmentation
Change-Id: I427b3e0bc92b8a73b85f7a010ff757bdee9fc62c
diff --git a/.cargo_vcs_info.json b/.cargo_vcs_info.json
index 1397729..22f2cec 100644
--- a/.cargo_vcs_info.json
+++ b/.cargo_vcs_info.json
@@ -1,5 +1,5 @@
{
"git": {
- "sha1": "fc6c486b2fe51adb3552f23036e9b692d7b3ac96"
+ "sha1": "3b75ee19b3c0ddacaeec03be688a7b8766833728"
}
}
diff --git a/Cargo.toml b/Cargo.toml
index 9f48df3..0f21309 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -12,9 +12,9 @@
[package]
name = "unicode-segmentation"
-version = "1.6.0"
+version = "1.7.1"
authors = ["kwantam <kwantam@gmail.com>", "Manish Goregaokar <manishsmail@gmail.com>"]
-exclude = ["target/*", "Cargo.lock", "scripts/tmp", "*.txt"]
+exclude = ["target/*", "Cargo.lock", "scripts/tmp", "benches/texts/*", "*.txt"]
description = "This crate provides Grapheme Cluster, Word and Sentence boundaries\naccording to Unicode Standard Annex #29 rules.\n"
homepage = "https://github.com/unicode-rs/unicode-segmentation"
documentation = "https://unicode-rs.github.io/unicode-segmentation"
@@ -22,6 +22,13 @@
keywords = ["text", "unicode", "grapheme", "word", "boundary"]
license = "MIT/Apache-2.0"
repository = "https://github.com/unicode-rs/unicode-segmentation"
+
+[[bench]]
+name = "graphemes"
+harness = false
+[dev-dependencies.bencher]
+version = "0.1"
+
[dev-dependencies.quickcheck]
version = "0.7"
diff --git a/Cargo.toml.orig b/Cargo.toml.orig
index c513e11..3f55167 100644
--- a/Cargo.toml.orig
+++ b/Cargo.toml.orig
@@ -1,7 +1,7 @@
[package]
name = "unicode-segmentation"
-version = "1.6.0"
+version = "1.7.1"
authors = ["kwantam <kwantam@gmail.com>", "Manish Goregaokar <manishsmail@gmail.com>"]
homepage = "https://github.com/unicode-rs/unicode-segmentation"
@@ -16,10 +16,15 @@
according to Unicode Standard Annex #29 rules.
"""
-exclude = [ "target/*", "Cargo.lock", "scripts/tmp", "*.txt" ]
+exclude = [ "target/*", "Cargo.lock", "scripts/tmp", "benches/texts/*", "*.txt", ]
[features]
no_std = [] # This is a no-op, preserved for backward compatibility only.
[dev-dependencies]
quickcheck = "0.7"
+bencher = "0.1"
+
+[[bench]]
+name = "graphemes"
+harness = false
\ No newline at end of file
diff --git a/METADATA b/METADATA
index df8c5ad..bf983b5 100644
--- a/METADATA
+++ b/METADATA
@@ -7,13 +7,13 @@
}
url {
type: ARCHIVE
- value: "https://static.crates.io/crates/unicode-segmentation/unicode-segmentation-1.6.0.crate"
+ value: "https://static.crates.io/crates/unicode-segmentation/unicode-segmentation-1.7.1.crate"
}
- version: "1.6.0"
+ version: "1.7.1"
license_type: NOTICE
last_upgrade_date {
- year: 2020
- month: 5
- day: 6
+ year: 2021
+ month: 1
+ day: 12
}
}
diff --git a/README.md b/README.md
index 08fe6b8..2f3bdca 100644
--- a/README.md
+++ b/README.md
@@ -10,7 +10,7 @@
fn main() {
let s = "a̐éö̲\r\n";
- let g = UnicodeSegmentation::graphemes(s, true).collect::<Vec<&str>>();
+ let g = s.graphemes(true).collect::<Vec<&str>>();
let b: &[_] = &["a̐", "é", "ö̲", "\r\n"];
assert_eq!(g, b);
@@ -38,11 +38,21 @@
```toml
[dependencies]
-unicode-segmentation = "1.3.0"
+unicode-segmentation = "1.7.1"
```
# Change Log
+## 1.7.1
+
+* Update docs on version number
+
+## 1.7.0
+
+* [#87](https://github.com/unicode-rs/unicode-segmentation/pull/87) Upgrade to Unicode 13
+* [#79](https://github.com/unicode-rs/unicode-segmentation/pull/79) Implement a special-case lookup for ascii grapheme categories
+* [#77](https://github.com/unicode-rs/unicode-segmentation/pull/77) Optimization for grapheme iteration
+
## 1.6.0
* [#72](https://github.com/unicode-rs/unicode-segmentation/pull/72) Upgrade to Unicode 12
diff --git a/benches/graphemes.rs b/benches/graphemes.rs
new file mode 100644
index 0000000..5f14352
--- /dev/null
+++ b/benches/graphemes.rs
@@ -0,0 +1,64 @@
+#[macro_use]
+extern crate bencher;
+extern crate unicode_segmentation;
+
+use bencher::Bencher;
+use unicode_segmentation::UnicodeSegmentation;
+use std::fs;
+
+fn graphemes(bench: &mut Bencher, path: &str) {
+ let text = fs::read_to_string(path).unwrap();
+ bench.iter(|| {
+ for g in UnicodeSegmentation::graphemes(&*text, true) {
+ bencher::black_box(g);
+ }
+ });
+
+ bench.bytes = text.len() as u64;
+}
+
+fn graphemes_arabic(bench: &mut Bencher) {
+ graphemes(bench, "benches/texts/arabic.txt");
+}
+
+fn graphemes_english(bench: &mut Bencher) {
+ graphemes(bench, "benches/texts/english.txt");
+}
+
+fn graphemes_hindi(bench: &mut Bencher) {
+ graphemes(bench, "benches/texts/hindi.txt");
+}
+
+fn graphemes_japanese(bench: &mut Bencher) {
+ graphemes(bench, "benches/texts/japanese.txt");
+}
+
+fn graphemes_korean(bench: &mut Bencher) {
+ graphemes(bench, "benches/texts/korean.txt");
+}
+
+fn graphemes_mandarin(bench: &mut Bencher) {
+ graphemes(bench, "benches/texts/mandarin.txt");
+}
+
+fn graphemes_russian(bench: &mut Bencher) {
+ graphemes(bench, "benches/texts/russian.txt");
+}
+
+fn graphemes_source_code(bench: &mut Bencher) {
+ graphemes(bench, "benches/texts/source_code.txt");
+}
+
+benchmark_group!(
+ benches,
+ graphemes_arabic,
+ graphemes_english,
+ graphemes_hindi,
+ graphemes_japanese,
+ graphemes_korean,
+ graphemes_mandarin,
+ graphemes_russian,
+ graphemes_source_code,
+);
+
+benchmark_main!(benches);
diff --git a/scripts/unicode.py b/scripts/unicode.py
old mode 100755
new mode 100644
index 580b3c0..1841e35
--- a/scripts/unicode.py
+++ b/scripts/unicode.py
@@ -54,7 +54,7 @@
# these are the surrogate codepoints, which are not valid rust characters
surrogate_codepoints = (0xd800, 0xdfff)
-UNICODE_VERSION = (12, 0, 0)
+UNICODE_VERSION = (13, 0, 0)
UNICODE_VERSION_NUMBER = "%s.%s.%s" %UNICODE_VERSION
@@ -64,10 +64,10 @@
def fetch(f):
if not os.path.exists(os.path.basename(f)):
if "emoji" in f:
- os.system("curl -O https://www.unicode.org/Public/emoji/%s.%s/%s"
- % (UNICODE_VERSION[0], UNICODE_VERSION[1], f))
+ os.system("curl -O https://www.unicode.org/Public/%s/ucd/emoji/%s"
+ % (UNICODE_VERSION_NUMBER, f))
else:
- os.system("curl -O http://www.unicode.org/Public/%s/ucd/%s"
+ os.system("curl -O https://www.unicode.org/Public/%s/ucd/%s"
% (UNICODE_VERSION_NUMBER, f))
if not os.path.exists(os.path.basename(f)):
@@ -280,7 +280,8 @@
f.write((" %sC_" % Name[0]) + cat + ",\n")
f.write(""" }
- fn bsearch_range_value_table(c: char, r: &'static [(char, char, %sCat)]) -> %sCat {
+ fn bsearch_range_value_table(c: char, r: &'static [(char, char, %sCat)]) -> (u32, u32, %sCat) {
+ use core;
use core::cmp::Ordering::{Equal, Less, Greater};
match r.binary_search_by(|&(lo, hi, _)| {
if lo <= c && c <= hi { Equal }
@@ -288,14 +289,20 @@
else { Greater }
}) {
Ok(idx) => {
- let (_, _, cat) = r[idx];
- cat
+ let (lower, upper, cat) = r[idx];
+ (lower as u32, upper as u32, cat)
}
- Err(_) => %sC_Any
+ Err(idx) => {
+ (
+ if idx > 0 { r[idx-1].1 as u32 + 1 } else { 0 },
+ r.get(idx).map(|c|c.0 as u32 - 1).unwrap_or(core::u32::MAX),
+ %sC_Any,
+ )
+ }
}
}
- pub fn %s_category(c: char) -> %sCat {
+ pub fn %s_category(c: char) -> (u32, u32, %sCat) {
bsearch_range_value_table(c, %s_cat_table)
}
diff --git a/scripts/unicode_gen_breaktests.py b/scripts/unicode_gen_breaktests.py
old mode 100755
new mode 100644
diff --git a/src/grapheme.rs b/src/grapheme.rs
index cde6526..e95d478 100644
--- a/src/grapheme.rs
+++ b/src/grapheme.rs
@@ -13,6 +13,12 @@
use tables::grapheme::GraphemeCat;
/// External iterator for grapheme clusters and byte offsets.
+///
+/// This struct is created by the [`grapheme_indices`] method on the [`UnicodeSegmentation`]
+/// trait. See its documentation for more.
+///
+/// [`grapheme_indices`]: trait.UnicodeSegmentation.html#tymethod.grapheme_indices
+/// [`UnicodeSegmentation`]: trait.UnicodeSegmentation.html
#[derive(Clone)]
pub struct GraphemeIndices<'a> {
start_offset: usize,
@@ -61,6 +67,12 @@
/// External iterator for a string's
/// [grapheme clusters](http://www.unicode.org/reports/tr29/#Grapheme_Cluster_Boundaries).
+///
+/// This struct is created by the [`graphemes`] method on the [`UnicodeSegmentation`] trait. See its
+/// documentation for more.
+///
+/// [`graphemes`]: trait.UnicodeSegmentation.html#tymethod.graphemes
+/// [`UnicodeSegmentation`]: trait.UnicodeSegmentation.html
#[derive(Clone)]
pub struct Graphemes<'a> {
string: &'a str,
@@ -178,6 +190,8 @@
// Set if a call to `prev_boundary` or `next_boundary` was suspended due
// to needing more input.
resuming: bool,
+ // Cached grapheme category and associated scalar value range.
+ grapheme_cat_cache: (u32, u32, GraphemeCat),
}
/// An error return indicating that not enough content was available in the
@@ -276,6 +290,35 @@
pre_context_offset: None,
ris_count: None,
resuming: false,
+ grapheme_cat_cache: (0, 0, GraphemeCat::GC_Control),
+ }
+ }
+
+ fn grapheme_category(&mut self, ch: char) -> GraphemeCat {
+ use tables::grapheme as gr;
+ use tables::grapheme::GraphemeCat::*;
+
+ if ch <= '\u{7e}' {
+ // Special-case optimization for ascii, except U+007F. This
+ // improves performance even for many primarily non-ascii texts,
+ // due to use of punctuation and white space characters from the
+ // ascii range.
+ if ch >= '\u{20}' {
+ GC_Any
+ } else if ch == '\n' {
+ GC_LF
+ } else if ch == '\r' {
+ GC_CR
+ } else {
+ GC_Control
+ }
+ } else {
+ // If this char isn't within the cached range, update the cache to the
+ // range that includes it.
+ if (ch as u32) < self.grapheme_cat_cache.0 || (ch as u32) > self.grapheme_cat_cache.1 {
+ self.grapheme_cat_cache = gr::grapheme_category(ch);
+ }
+ self.grapheme_cat_cache.2
}
}
@@ -349,7 +392,7 @@
self.pre_context_offset = None;
if self.is_extended && chunk_start + chunk.len() == self.offset {
let ch = chunk.chars().rev().next().unwrap();
- if gr::grapheme_category(ch) == gr::GC_Prepend {
+ if self.grapheme_category(ch) == gr::GC_Prepend {
self.decide(false); // GB9b
return;
}
@@ -359,7 +402,7 @@
GraphemeState::Emoji => self.handle_emoji(chunk, chunk_start),
_ => if self.cat_before.is_none() && self.offset == chunk.len() + chunk_start {
let ch = chunk.chars().rev().next().unwrap();
- self.cat_before = Some(gr::grapheme_category(ch));
+ self.cat_before = Some(self.grapheme_category(ch));
},
}
}
@@ -393,7 +436,7 @@
use tables::grapheme as gr;
let mut ris_count = self.ris_count.unwrap_or(0);
for ch in chunk.chars().rev() {
- if gr::grapheme_category(ch) != gr::GC_Regional_Indicator {
+ if self.grapheme_category(ch) != gr::GC_Regional_Indicator {
self.ris_count = Some(ris_count);
self.decide((ris_count % 2) == 0);
return;
@@ -413,13 +456,13 @@
use tables::grapheme as gr;
let mut iter = chunk.chars().rev();
if let Some(ch) = iter.next() {
- if gr::grapheme_category(ch) != gr::GC_ZWJ {
+ if self.grapheme_category(ch) != gr::GC_ZWJ {
self.decide(true);
return;
}
}
for ch in iter {
- match gr::grapheme_category(ch) {
+ match self.grapheme_category(ch) {
gr::GC_Extend => (),
gr::GC_Extended_Pictographic => {
self.decide(false);
@@ -481,7 +524,7 @@
let offset_in_chunk = self.offset - chunk_start;
if self.cat_after.is_none() {
let ch = chunk[offset_in_chunk..].chars().next().unwrap();
- self.cat_after = Some(gr::grapheme_category(ch));
+ self.cat_after = Some(self.grapheme_category(ch));
}
if self.offset == chunk_start {
let mut need_pre_context = true;
@@ -497,7 +540,7 @@
}
if self.cat_before.is_none() {
let ch = chunk[..offset_in_chunk].chars().rev().next().unwrap();
- self.cat_before = Some(gr::grapheme_category(ch));
+ self.cat_before = Some(self.grapheme_category(ch));
}
match check_pair(self.cat_before.unwrap(), self.cat_after.unwrap()) {
PairResult::NotBreak => return self.decision(false),
@@ -553,7 +596,6 @@
/// assert_eq!(cursor.next_boundary(&s[2..4], 2), Ok(None));
/// ```
pub fn next_boundary(&mut self, chunk: &str, chunk_start: usize) -> Result<Option<usize>, GraphemeIncomplete> {
- use tables::grapheme as gr;
if self.offset == self.len {
return Ok(None);
}
@@ -562,14 +604,14 @@
loop {
if self.resuming {
if self.cat_after.is_none() {
- self.cat_after = Some(gr::grapheme_category(ch));
+ self.cat_after = Some(self.grapheme_category(ch));
}
} else {
self.offset += ch.len_utf8();
self.state = GraphemeState::Unknown;
self.cat_before = self.cat_after.take();
if self.cat_before.is_none() {
- self.cat_before = Some(gr::grapheme_category(ch));
+ self.cat_before = Some(self.grapheme_category(ch));
}
if self.cat_before.unwrap() == GraphemeCat::GC_Regional_Indicator {
self.ris_count = self.ris_count.map(|c| c + 1);
@@ -578,7 +620,7 @@
}
if let Some(next_ch) = iter.next() {
ch = next_ch;
- self.cat_after = Some(gr::grapheme_category(ch));
+ self.cat_after = Some(self.grapheme_category(ch));
} else if self.offset == self.len {
self.decide(true);
} else {
@@ -629,7 +671,6 @@
/// assert_eq!(cursor.prev_boundary(&s[0..2], 0), Ok(None));
/// ```
pub fn prev_boundary(&mut self, chunk: &str, chunk_start: usize) -> Result<Option<usize>, GraphemeIncomplete> {
- use tables::grapheme as gr;
if self.offset == 0 {
return Ok(None);
}
@@ -644,7 +685,7 @@
return Err(GraphemeIncomplete::PrevChunk);
}
if self.resuming {
- self.cat_before = Some(gr::grapheme_category(ch));
+ self.cat_before = Some(self.grapheme_category(ch));
} else {
self.offset -= ch.len_utf8();
self.cat_after = self.cat_before.take();
@@ -654,12 +695,12 @@
}
if let Some(prev_ch) = iter.next() {
ch = prev_ch;
- self.cat_before = Some(gr::grapheme_category(ch));
+ self.cat_before = Some(self.grapheme_category(ch));
} else if self.offset == 0 {
self.decide(true);
} else {
self.resuming = true;
- self.cat_after = Some(gr::grapheme_category(ch));
+ self.cat_after = Some(self.grapheme_category(ch));
return Err(GraphemeIncomplete::PrevChunk);
}
}
diff --git a/src/lib.rs b/src/lib.rs
index 88aa97e..571e33a 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -46,7 +46,7 @@
//!
//! ```toml
//! [dependencies]
-//! unicode-segmentation = "1.3.0"
+//! unicode-segmentation = "1.7.1"
//! ```
#![deny(missing_docs, unsafe_code)]
@@ -180,22 +180,56 @@
/// Returns an iterator over substrings of `self` separated on
/// [UAX#29 sentence boundaries](http://www.unicode.org/reports/tr29/#Sentence_Boundaries).
///
- /// The concatenation of the substrings returned by this function is just the original string.
- fn unicode_sentences<'a>(&'a self) -> UnicodeSentences<'a>;
-
- /// Returns an iterator over substrings of `self` separated on
- /// [UAX#29 sentence boundaries](http://www.unicode.org/reports/tr29/#Sentence_Boundaries).
- ///
/// Here, "sentences" are just those substrings which, after splitting on
/// UAX#29 sentence boundaries, contain any alphanumeric characters. That is, the
/// substring must contain at least one character with the
/// [Alphabetic](http://unicode.org/reports/tr44/#Alphabetic)
/// property, or with
/// [General_Category=Number](http://unicode.org/reports/tr44/#General_Category_Values).
+ ///
+ /// # Example
+ ///
+ /// ```
+ /// # use self::unicode_segmentation::UnicodeSegmentation;
+ /// let uss = "Mr. Fox jumped. [...] The dog was too lazy.";
+ /// let us1 = uss.unicode_sentences().collect::<Vec<&str>>();
+ /// let b: &[_] = &["Mr. ", "Fox jumped. ", "The dog was too lazy."];
+ ///
+ /// assert_eq!(&us1[..], b);
+ /// ```
+ fn unicode_sentences<'a>(&'a self) -> UnicodeSentences<'a>;
+
+ /// Returns an iterator over substrings of `self` separated on
+ /// [UAX#29 sentence boundaries](http://www.unicode.org/reports/tr29/#Sentence_Boundaries).
+ ///
+ /// The concatenation of the substrings returned by this function is just the original string.
+ ///
+ /// # Example
+ ///
+ /// ```
+ /// # use self::unicode_segmentation::UnicodeSegmentation;
+ /// let ssbs = "Mr. Fox jumped. [...] The dog was too lazy.";
+ /// let ssb1 = ssbs.split_sentence_bounds().collect::<Vec<&str>>();
+ /// let b: &[_] = &["Mr. ", "Fox jumped. ", "[...] ", "The dog was too lazy."];
+ ///
+ /// assert_eq!(&ssb1[..], b);
+ /// ```
fn split_sentence_bounds<'a>(&'a self) -> USentenceBounds<'a>;
/// Returns an iterator over substrings of `self`, split on UAX#29 sentence boundaries,
/// and their offsets. See `split_sentence_bounds()` for more information.
+ ///
+ /// # Example
+ ///
+ /// ```
+ /// # use self::unicode_segmentation::UnicodeSegmentation;
+ /// let ssis = "Mr. Fox jumped. [...] The dog was too lazy.";
+ /// let ssi1 = ssis.split_sentence_bound_indices().collect::<Vec<(usize, &str)>>();
+ /// let b: &[_] = &[(0, "Mr. "), (4, "Fox jumped. "), (16, "[...] "),
+ /// (22, "The dog was too lazy.")];
+ ///
+ /// assert_eq!(&ssi1[..], b);
+ /// ```
fn split_sentence_bound_indices<'a>(&'a self) -> USentenceBoundIndices<'a>;
}
diff --git a/src/sentence.rs b/src/sentence.rs
index c16c927..275da52 100644
--- a/src/sentence.rs
+++ b/src/sentence.rs
@@ -115,7 +115,7 @@
for next_char in ahead.chars() {
//( ¬(OLetter | Upper | Lower | ParaSep | SATerm) )* Lower
- match se::sentence_category(next_char) {
+ match se::sentence_category(next_char).2 {
se::SC_Lower => return true,
se::SC_OLetter |
se::SC_Upper |
@@ -182,7 +182,7 @@
let position_before = self.pos;
let state_before = self.state.clone();
- let next_cat = se::sentence_category(next_char);
+ let next_cat = se::sentence_category(next_char).2;
self.pos += next_char.len_utf8();
self.state = self.state.next(next_cat);
@@ -274,6 +274,12 @@
/// [Alphabetic](http://unicode.org/reports/tr44/#Alphabetic)
/// property, or with
/// [General_Category=Number](http://unicode.org/reports/tr44/#General_Category_Values).
+///
+/// This struct is created by the [`unicode_sentences`] method on the [`UnicodeSegmentation`]
+/// trait. See its documentation for more.
+///
+/// [`unicode_sentences`]: trait.UnicodeSegmentation.html#tymethod.unicode_sentences
+/// [`UnicodeSegmentation`]: trait.UnicodeSegmentation.html
#[derive(Clone)]
pub struct UnicodeSentences<'a> {
inner: Filter<USentenceBounds<'a>, fn(&&str) -> bool>,
@@ -281,6 +287,12 @@
/// External iterator for a string's
/// [sentence boundaries](http://www.unicode.org/reports/tr29/#Sentence_Boundaries).
+///
+/// This struct is created by the [`split_sentence_bounds`] method on the [`UnicodeSegmentation`]
+/// trait. See its documentation for more.
+///
+/// [`split_sentence_bounds`]: trait.UnicodeSegmentation.html#tymethod.split_sentence_bounds
+/// [`UnicodeSegmentation`]: trait.UnicodeSegmentation.html
#[derive(Clone)]
pub struct USentenceBounds<'a> {
iter: fwd::SentenceBreaks<'a>,
@@ -288,6 +300,12 @@
}
/// External iterator for sentence boundaries and byte offsets.
+///
+/// This struct is created by the [`split_sentence_bound_indices`] method on the
+/// [`UnicodeSegmentation`] trait. See its documentation for more.
+///
+/// [`split_sentence_bound_indices`]: trait.UnicodeSegmentation.html#tymethod.split_sentence_bound_indices
+/// [`UnicodeSegmentation`]: trait.UnicodeSegmentation.html
#[derive(Clone)]
pub struct USentenceBoundIndices<'a> {
start_offset: usize,
diff --git a/src/tables.rs b/src/tables.rs
index bfd7290..7062e36 100644
--- a/src/tables.rs
+++ b/src/tables.rs
@@ -14,7 +14,7 @@
/// The version of [Unicode](http://www.unicode.org/)
/// that this version of unicode-segmentation is based on.
-pub const UNICODE_VERSION: (u64, u64, u64) = (12, 0, 0);
+pub const UNICODE_VERSION: (u64, u64, u64) = (13, 0, 0);
pub mod util {
#[inline]
@@ -81,18 +81,19 @@
('\u{10a7d}', '\u{10a7e}'), ('\u{10a9d}', '\u{10a9f}'), ('\u{10aeb}', '\u{10aef}'),
('\u{10b58}', '\u{10b5f}'), ('\u{10b78}', '\u{10b7f}'), ('\u{10ba9}', '\u{10baf}'),
('\u{10cfa}', '\u{10cff}'), ('\u{10d30}', '\u{10d39}'), ('\u{10e60}', '\u{10e7e}'),
- ('\u{10f1d}', '\u{10f26}'), ('\u{10f51}', '\u{10f54}'), ('\u{11052}', '\u{1106f}'),
- ('\u{110f0}', '\u{110f9}'), ('\u{11136}', '\u{1113f}'), ('\u{111d0}', '\u{111d9}'),
- ('\u{111e1}', '\u{111f4}'), ('\u{112f0}', '\u{112f9}'), ('\u{11450}', '\u{11459}'),
- ('\u{114d0}', '\u{114d9}'), ('\u{11650}', '\u{11659}'), ('\u{116c0}', '\u{116c9}'),
- ('\u{11730}', '\u{1173b}'), ('\u{118e0}', '\u{118f2}'), ('\u{11c50}', '\u{11c6c}'),
- ('\u{11d50}', '\u{11d59}'), ('\u{11da0}', '\u{11da9}'), ('\u{11fc0}', '\u{11fd4}'),
- ('\u{12400}', '\u{1246e}'), ('\u{16a60}', '\u{16a69}'), ('\u{16b50}', '\u{16b59}'),
- ('\u{16b5b}', '\u{16b61}'), ('\u{16e80}', '\u{16e96}'), ('\u{1d2e0}', '\u{1d2f3}'),
- ('\u{1d360}', '\u{1d378}'), ('\u{1d7ce}', '\u{1d7ff}'), ('\u{1e140}', '\u{1e149}'),
- ('\u{1e2f0}', '\u{1e2f9}'), ('\u{1e8c7}', '\u{1e8cf}'), ('\u{1e950}', '\u{1e959}'),
- ('\u{1ec71}', '\u{1ecab}'), ('\u{1ecad}', '\u{1ecaf}'), ('\u{1ecb1}', '\u{1ecb4}'),
- ('\u{1ed01}', '\u{1ed2d}'), ('\u{1ed2f}', '\u{1ed3d}'), ('\u{1f100}', '\u{1f10c}')
+ ('\u{10f1d}', '\u{10f26}'), ('\u{10f51}', '\u{10f54}'), ('\u{10fc5}', '\u{10fcb}'),
+ ('\u{11052}', '\u{1106f}'), ('\u{110f0}', '\u{110f9}'), ('\u{11136}', '\u{1113f}'),
+ ('\u{111d0}', '\u{111d9}'), ('\u{111e1}', '\u{111f4}'), ('\u{112f0}', '\u{112f9}'),
+ ('\u{11450}', '\u{11459}'), ('\u{114d0}', '\u{114d9}'), ('\u{11650}', '\u{11659}'),
+ ('\u{116c0}', '\u{116c9}'), ('\u{11730}', '\u{1173b}'), ('\u{118e0}', '\u{118f2}'),
+ ('\u{11950}', '\u{11959}'), ('\u{11c50}', '\u{11c6c}'), ('\u{11d50}', '\u{11d59}'),
+ ('\u{11da0}', '\u{11da9}'), ('\u{11fc0}', '\u{11fd4}'), ('\u{12400}', '\u{1246e}'),
+ ('\u{16a60}', '\u{16a69}'), ('\u{16b50}', '\u{16b59}'), ('\u{16b5b}', '\u{16b61}'),
+ ('\u{16e80}', '\u{16e96}'), ('\u{1d2e0}', '\u{1d2f3}'), ('\u{1d360}', '\u{1d378}'),
+ ('\u{1d7ce}', '\u{1d7ff}'), ('\u{1e140}', '\u{1e149}'), ('\u{1e2f0}', '\u{1e2f9}'),
+ ('\u{1e8c7}', '\u{1e8cf}'), ('\u{1e950}', '\u{1e959}'), ('\u{1ec71}', '\u{1ecab}'),
+ ('\u{1ecad}', '\u{1ecaf}'), ('\u{1ecb1}', '\u{1ecb4}'), ('\u{1ed01}', '\u{1ed2d}'),
+ ('\u{1ed2f}', '\u{1ed3d}'), ('\u{1f100}', '\u{1f10c}'), ('\u{1fbf0}', '\u{1fbf9}')
];
#[inline]
@@ -119,7 +120,7 @@
('\u{710}', '\u{73f}'), ('\u{74d}', '\u{7b1}'), ('\u{7ca}', '\u{7ea}'), ('\u{7f4}',
'\u{7f5}'), ('\u{7fa}', '\u{7fa}'), ('\u{800}', '\u{817}'), ('\u{81a}', '\u{82c}'),
('\u{840}', '\u{858}'), ('\u{860}', '\u{86a}'), ('\u{8a0}', '\u{8b4}'), ('\u{8b6}',
- '\u{8bd}'), ('\u{8d4}', '\u{8df}'), ('\u{8e3}', '\u{8e9}'), ('\u{8f0}', '\u{93b}'),
+ '\u{8c7}'), ('\u{8d4}', '\u{8df}'), ('\u{8e3}', '\u{8e9}'), ('\u{8f0}', '\u{93b}'),
('\u{93d}', '\u{94c}'), ('\u{94e}', '\u{950}'), ('\u{955}', '\u{963}'), ('\u{971}',
'\u{983}'), ('\u{985}', '\u{98c}'), ('\u{98f}', '\u{990}'), ('\u{993}', '\u{9a8}'),
('\u{9aa}', '\u{9b0}'), ('\u{9b2}', '\u{9b2}'), ('\u{9b6}', '\u{9b9}'), ('\u{9bd}',
@@ -148,35 +149,35 @@
'\u{c83}'), ('\u{c85}', '\u{c8c}'), ('\u{c8e}', '\u{c90}'), ('\u{c92}', '\u{ca8}'),
('\u{caa}', '\u{cb3}'), ('\u{cb5}', '\u{cb9}'), ('\u{cbd}', '\u{cc4}'), ('\u{cc6}',
'\u{cc8}'), ('\u{cca}', '\u{ccc}'), ('\u{cd5}', '\u{cd6}'), ('\u{cde}', '\u{cde}'),
- ('\u{ce0}', '\u{ce3}'), ('\u{cf1}', '\u{cf2}'), ('\u{d00}', '\u{d03}'), ('\u{d05}',
- '\u{d0c}'), ('\u{d0e}', '\u{d10}'), ('\u{d12}', '\u{d3a}'), ('\u{d3d}', '\u{d44}'),
- ('\u{d46}', '\u{d48}'), ('\u{d4a}', '\u{d4c}'), ('\u{d4e}', '\u{d4e}'), ('\u{d54}',
- '\u{d57}'), ('\u{d5f}', '\u{d63}'), ('\u{d7a}', '\u{d7f}'), ('\u{d82}', '\u{d83}'),
- ('\u{d85}', '\u{d96}'), ('\u{d9a}', '\u{db1}'), ('\u{db3}', '\u{dbb}'), ('\u{dbd}',
- '\u{dbd}'), ('\u{dc0}', '\u{dc6}'), ('\u{dcf}', '\u{dd4}'), ('\u{dd6}', '\u{dd6}'),
- ('\u{dd8}', '\u{ddf}'), ('\u{df2}', '\u{df3}'), ('\u{e01}', '\u{e3a}'), ('\u{e40}',
- '\u{e46}'), ('\u{e4d}', '\u{e4d}'), ('\u{e81}', '\u{e82}'), ('\u{e84}', '\u{e84}'),
- ('\u{e86}', '\u{e8a}'), ('\u{e8c}', '\u{ea3}'), ('\u{ea5}', '\u{ea5}'), ('\u{ea7}',
- '\u{eb9}'), ('\u{ebb}', '\u{ebd}'), ('\u{ec0}', '\u{ec4}'), ('\u{ec6}', '\u{ec6}'),
- ('\u{ecd}', '\u{ecd}'), ('\u{edc}', '\u{edf}'), ('\u{f00}', '\u{f00}'), ('\u{f40}',
- '\u{f47}'), ('\u{f49}', '\u{f6c}'), ('\u{f71}', '\u{f81}'), ('\u{f88}', '\u{f97}'),
- ('\u{f99}', '\u{fbc}'), ('\u{1000}', '\u{1036}'), ('\u{1038}', '\u{1038}'), ('\u{103b}',
- '\u{103f}'), ('\u{1050}', '\u{108f}'), ('\u{109a}', '\u{109d}'), ('\u{10a0}', '\u{10c5}'),
- ('\u{10c7}', '\u{10c7}'), ('\u{10cd}', '\u{10cd}'), ('\u{10d0}', '\u{10fa}'), ('\u{10fc}',
- '\u{1248}'), ('\u{124a}', '\u{124d}'), ('\u{1250}', '\u{1256}'), ('\u{1258}', '\u{1258}'),
- ('\u{125a}', '\u{125d}'), ('\u{1260}', '\u{1288}'), ('\u{128a}', '\u{128d}'), ('\u{1290}',
- '\u{12b0}'), ('\u{12b2}', '\u{12b5}'), ('\u{12b8}', '\u{12be}'), ('\u{12c0}', '\u{12c0}'),
- ('\u{12c2}', '\u{12c5}'), ('\u{12c8}', '\u{12d6}'), ('\u{12d8}', '\u{1310}'), ('\u{1312}',
- '\u{1315}'), ('\u{1318}', '\u{135a}'), ('\u{1380}', '\u{138f}'), ('\u{13a0}', '\u{13f5}'),
- ('\u{13f8}', '\u{13fd}'), ('\u{1401}', '\u{166c}'), ('\u{166f}', '\u{167f}'), ('\u{1681}',
- '\u{169a}'), ('\u{16a0}', '\u{16ea}'), ('\u{16ee}', '\u{16f8}'), ('\u{1700}', '\u{170c}'),
- ('\u{170e}', '\u{1713}'), ('\u{1720}', '\u{1733}'), ('\u{1740}', '\u{1753}'), ('\u{1760}',
- '\u{176c}'), ('\u{176e}', '\u{1770}'), ('\u{1772}', '\u{1773}'), ('\u{1780}', '\u{17b3}'),
- ('\u{17b6}', '\u{17c8}'), ('\u{17d7}', '\u{17d7}'), ('\u{17dc}', '\u{17dc}'), ('\u{1820}',
- '\u{1878}'), ('\u{1880}', '\u{18aa}'), ('\u{18b0}', '\u{18f5}'), ('\u{1900}', '\u{191e}'),
- ('\u{1920}', '\u{192b}'), ('\u{1930}', '\u{1938}'), ('\u{1950}', '\u{196d}'), ('\u{1970}',
- '\u{1974}'), ('\u{1980}', '\u{19ab}'), ('\u{19b0}', '\u{19c9}'), ('\u{1a00}', '\u{1a1b}'),
- ('\u{1a20}', '\u{1a5e}'), ('\u{1a61}', '\u{1a74}'), ('\u{1aa7}', '\u{1aa7}'), ('\u{1b00}',
+ ('\u{ce0}', '\u{ce3}'), ('\u{cf1}', '\u{cf2}'), ('\u{d00}', '\u{d0c}'), ('\u{d0e}',
+ '\u{d10}'), ('\u{d12}', '\u{d3a}'), ('\u{d3d}', '\u{d44}'), ('\u{d46}', '\u{d48}'),
+ ('\u{d4a}', '\u{d4c}'), ('\u{d4e}', '\u{d4e}'), ('\u{d54}', '\u{d57}'), ('\u{d5f}',
+ '\u{d63}'), ('\u{d7a}', '\u{d7f}'), ('\u{d81}', '\u{d83}'), ('\u{d85}', '\u{d96}'),
+ ('\u{d9a}', '\u{db1}'), ('\u{db3}', '\u{dbb}'), ('\u{dbd}', '\u{dbd}'), ('\u{dc0}',
+ '\u{dc6}'), ('\u{dcf}', '\u{dd4}'), ('\u{dd6}', '\u{dd6}'), ('\u{dd8}', '\u{ddf}'),
+ ('\u{df2}', '\u{df3}'), ('\u{e01}', '\u{e3a}'), ('\u{e40}', '\u{e46}'), ('\u{e4d}',
+ '\u{e4d}'), ('\u{e81}', '\u{e82}'), ('\u{e84}', '\u{e84}'), ('\u{e86}', '\u{e8a}'),
+ ('\u{e8c}', '\u{ea3}'), ('\u{ea5}', '\u{ea5}'), ('\u{ea7}', '\u{eb9}'), ('\u{ebb}',
+ '\u{ebd}'), ('\u{ec0}', '\u{ec4}'), ('\u{ec6}', '\u{ec6}'), ('\u{ecd}', '\u{ecd}'),
+ ('\u{edc}', '\u{edf}'), ('\u{f00}', '\u{f00}'), ('\u{f40}', '\u{f47}'), ('\u{f49}',
+ '\u{f6c}'), ('\u{f71}', '\u{f81}'), ('\u{f88}', '\u{f97}'), ('\u{f99}', '\u{fbc}'),
+ ('\u{1000}', '\u{1036}'), ('\u{1038}', '\u{1038}'), ('\u{103b}', '\u{103f}'), ('\u{1050}',
+ '\u{108f}'), ('\u{109a}', '\u{109d}'), ('\u{10a0}', '\u{10c5}'), ('\u{10c7}', '\u{10c7}'),
+ ('\u{10cd}', '\u{10cd}'), ('\u{10d0}', '\u{10fa}'), ('\u{10fc}', '\u{1248}'), ('\u{124a}',
+ '\u{124d}'), ('\u{1250}', '\u{1256}'), ('\u{1258}', '\u{1258}'), ('\u{125a}', '\u{125d}'),
+ ('\u{1260}', '\u{1288}'), ('\u{128a}', '\u{128d}'), ('\u{1290}', '\u{12b0}'), ('\u{12b2}',
+ '\u{12b5}'), ('\u{12b8}', '\u{12be}'), ('\u{12c0}', '\u{12c0}'), ('\u{12c2}', '\u{12c5}'),
+ ('\u{12c8}', '\u{12d6}'), ('\u{12d8}', '\u{1310}'), ('\u{1312}', '\u{1315}'), ('\u{1318}',
+ '\u{135a}'), ('\u{1380}', '\u{138f}'), ('\u{13a0}', '\u{13f5}'), ('\u{13f8}', '\u{13fd}'),
+ ('\u{1401}', '\u{166c}'), ('\u{166f}', '\u{167f}'), ('\u{1681}', '\u{169a}'), ('\u{16a0}',
+ '\u{16ea}'), ('\u{16ee}', '\u{16f8}'), ('\u{1700}', '\u{170c}'), ('\u{170e}', '\u{1713}'),
+ ('\u{1720}', '\u{1733}'), ('\u{1740}', '\u{1753}'), ('\u{1760}', '\u{176c}'), ('\u{176e}',
+ '\u{1770}'), ('\u{1772}', '\u{1773}'), ('\u{1780}', '\u{17b3}'), ('\u{17b6}', '\u{17c8}'),
+ ('\u{17d7}', '\u{17d7}'), ('\u{17dc}', '\u{17dc}'), ('\u{1820}', '\u{1878}'), ('\u{1880}',
+ '\u{18aa}'), ('\u{18b0}', '\u{18f5}'), ('\u{1900}', '\u{191e}'), ('\u{1920}', '\u{192b}'),
+ ('\u{1930}', '\u{1938}'), ('\u{1950}', '\u{196d}'), ('\u{1970}', '\u{1974}'), ('\u{1980}',
+ '\u{19ab}'), ('\u{19b0}', '\u{19c9}'), ('\u{1a00}', '\u{1a1b}'), ('\u{1a20}', '\u{1a5e}'),
+ ('\u{1a61}', '\u{1a74}'), ('\u{1aa7}', '\u{1aa7}'), ('\u{1abf}', '\u{1ac0}'), ('\u{1b00}',
'\u{1b33}'), ('\u{1b35}', '\u{1b43}'), ('\u{1b45}', '\u{1b4b}'), ('\u{1b80}', '\u{1ba9}'),
('\u{1bac}', '\u{1baf}'), ('\u{1bba}', '\u{1be5}'), ('\u{1be7}', '\u{1bf1}'), ('\u{1c00}',
'\u{1c36}'), ('\u{1c4d}', '\u{1c4f}'), ('\u{1c5a}', '\u{1c7d}'), ('\u{1c80}', '\u{1c88}'),
@@ -201,12 +202,12 @@
('\u{2dd8}', '\u{2dde}'), ('\u{2de0}', '\u{2dff}'), ('\u{2e2f}', '\u{2e2f}'), ('\u{3005}',
'\u{3007}'), ('\u{3021}', '\u{3029}'), ('\u{3031}', '\u{3035}'), ('\u{3038}', '\u{303c}'),
('\u{3041}', '\u{3096}'), ('\u{309d}', '\u{309f}'), ('\u{30a1}', '\u{30fa}'), ('\u{30fc}',
- '\u{30ff}'), ('\u{3105}', '\u{312f}'), ('\u{3131}', '\u{318e}'), ('\u{31a0}', '\u{31ba}'),
- ('\u{31f0}', '\u{31ff}'), ('\u{3400}', '\u{4db5}'), ('\u{4e00}', '\u{9fef}'), ('\u{a000}',
+ '\u{30ff}'), ('\u{3105}', '\u{312f}'), ('\u{3131}', '\u{318e}'), ('\u{31a0}', '\u{31bf}'),
+ ('\u{31f0}', '\u{31ff}'), ('\u{3400}', '\u{4dbf}'), ('\u{4e00}', '\u{9ffc}'), ('\u{a000}',
'\u{a48c}'), ('\u{a4d0}', '\u{a4fd}'), ('\u{a500}', '\u{a60c}'), ('\u{a610}', '\u{a61f}'),
('\u{a62a}', '\u{a62b}'), ('\u{a640}', '\u{a66e}'), ('\u{a674}', '\u{a67b}'), ('\u{a67f}',
'\u{a6ef}'), ('\u{a717}', '\u{a71f}'), ('\u{a722}', '\u{a788}'), ('\u{a78b}', '\u{a7bf}'),
- ('\u{a7c2}', '\u{a7c6}'), ('\u{a7f7}', '\u{a805}'), ('\u{a807}', '\u{a827}'), ('\u{a840}',
+ ('\u{a7c2}', '\u{a7ca}'), ('\u{a7f5}', '\u{a805}'), ('\u{a807}', '\u{a827}'), ('\u{a840}',
'\u{a873}'), ('\u{a880}', '\u{a8c3}'), ('\u{a8c5}', '\u{a8c5}'), ('\u{a8f2}', '\u{a8f7}'),
('\u{a8fb}', '\u{a8fb}'), ('\u{a8fd}', '\u{a8ff}'), ('\u{a90a}', '\u{a92a}'), ('\u{a930}',
'\u{a952}'), ('\u{a960}', '\u{a97c}'), ('\u{a980}', '\u{a9b2}'), ('\u{a9b4}', '\u{a9bf}'),
@@ -215,7 +216,7 @@
('\u{aac0}', '\u{aac0}'), ('\u{aac2}', '\u{aac2}'), ('\u{aadb}', '\u{aadd}'), ('\u{aae0}',
'\u{aaef}'), ('\u{aaf2}', '\u{aaf5}'), ('\u{ab01}', '\u{ab06}'), ('\u{ab09}', '\u{ab0e}'),
('\u{ab11}', '\u{ab16}'), ('\u{ab20}', '\u{ab26}'), ('\u{ab28}', '\u{ab2e}'), ('\u{ab30}',
- '\u{ab5a}'), ('\u{ab5c}', '\u{ab67}'), ('\u{ab70}', '\u{abea}'), ('\u{ac00}', '\u{d7a3}'),
+ '\u{ab5a}'), ('\u{ab5c}', '\u{ab69}'), ('\u{ab70}', '\u{abea}'), ('\u{ac00}', '\u{d7a3}'),
('\u{d7b0}', '\u{d7c6}'), ('\u{d7cb}', '\u{d7fb}'), ('\u{f900}', '\u{fa6d}'), ('\u{fa70}',
'\u{fad9}'), ('\u{fb00}', '\u{fb06}'), ('\u{fb13}', '\u{fb17}'), ('\u{fb1d}', '\u{fb28}'),
('\u{fb2a}', '\u{fb36}'), ('\u{fb38}', '\u{fb3c}'), ('\u{fb3e}', '\u{fb3e}'), ('\u{fb40}',
@@ -242,26 +243,30 @@
'\u{10ac7}'), ('\u{10ac9}', '\u{10ae4}'), ('\u{10b00}', '\u{10b35}'), ('\u{10b40}',
'\u{10b55}'), ('\u{10b60}', '\u{10b72}'), ('\u{10b80}', '\u{10b91}'), ('\u{10c00}',
'\u{10c48}'), ('\u{10c80}', '\u{10cb2}'), ('\u{10cc0}', '\u{10cf2}'), ('\u{10d00}',
- '\u{10d27}'), ('\u{10f00}', '\u{10f1c}'), ('\u{10f27}', '\u{10f27}'), ('\u{10f30}',
- '\u{10f45}'), ('\u{10fe0}', '\u{10ff6}'), ('\u{11000}', '\u{11045}'), ('\u{11082}',
- '\u{110b8}'), ('\u{110d0}', '\u{110e8}'), ('\u{11100}', '\u{11132}'), ('\u{11144}',
- '\u{11146}'), ('\u{11150}', '\u{11172}'), ('\u{11176}', '\u{11176}'), ('\u{11180}',
- '\u{111bf}'), ('\u{111c1}', '\u{111c4}'), ('\u{111da}', '\u{111da}'), ('\u{111dc}',
- '\u{111dc}'), ('\u{11200}', '\u{11211}'), ('\u{11213}', '\u{11234}'), ('\u{11237}',
- '\u{11237}'), ('\u{1123e}', '\u{1123e}'), ('\u{11280}', '\u{11286}'), ('\u{11288}',
- '\u{11288}'), ('\u{1128a}', '\u{1128d}'), ('\u{1128f}', '\u{1129d}'), ('\u{1129f}',
- '\u{112a8}'), ('\u{112b0}', '\u{112e8}'), ('\u{11300}', '\u{11303}'), ('\u{11305}',
- '\u{1130c}'), ('\u{1130f}', '\u{11310}'), ('\u{11313}', '\u{11328}'), ('\u{1132a}',
- '\u{11330}'), ('\u{11332}', '\u{11333}'), ('\u{11335}', '\u{11339}'), ('\u{1133d}',
- '\u{11344}'), ('\u{11347}', '\u{11348}'), ('\u{1134b}', '\u{1134c}'), ('\u{11350}',
- '\u{11350}'), ('\u{11357}', '\u{11357}'), ('\u{1135d}', '\u{11363}'), ('\u{11400}',
- '\u{11441}'), ('\u{11443}', '\u{11445}'), ('\u{11447}', '\u{1144a}'), ('\u{1145f}',
- '\u{1145f}'), ('\u{11480}', '\u{114c1}'), ('\u{114c4}', '\u{114c5}'), ('\u{114c7}',
- '\u{114c7}'), ('\u{11580}', '\u{115b5}'), ('\u{115b8}', '\u{115be}'), ('\u{115d8}',
- '\u{115dd}'), ('\u{11600}', '\u{1163e}'), ('\u{11640}', '\u{11640}'), ('\u{11644}',
- '\u{11644}'), ('\u{11680}', '\u{116b5}'), ('\u{116b8}', '\u{116b8}'), ('\u{11700}',
- '\u{1171a}'), ('\u{1171d}', '\u{1172a}'), ('\u{11800}', '\u{11838}'), ('\u{118a0}',
- '\u{118df}'), ('\u{118ff}', '\u{118ff}'), ('\u{119a0}', '\u{119a7}'), ('\u{119aa}',
+ '\u{10d27}'), ('\u{10e80}', '\u{10ea9}'), ('\u{10eab}', '\u{10eac}'), ('\u{10eb0}',
+ '\u{10eb1}'), ('\u{10f00}', '\u{10f1c}'), ('\u{10f27}', '\u{10f27}'), ('\u{10f30}',
+ '\u{10f45}'), ('\u{10fb0}', '\u{10fc4}'), ('\u{10fe0}', '\u{10ff6}'), ('\u{11000}',
+ '\u{11045}'), ('\u{11082}', '\u{110b8}'), ('\u{110d0}', '\u{110e8}'), ('\u{11100}',
+ '\u{11132}'), ('\u{11144}', '\u{11147}'), ('\u{11150}', '\u{11172}'), ('\u{11176}',
+ '\u{11176}'), ('\u{11180}', '\u{111bf}'), ('\u{111c1}', '\u{111c4}'), ('\u{111ce}',
+ '\u{111cf}'), ('\u{111da}', '\u{111da}'), ('\u{111dc}', '\u{111dc}'), ('\u{11200}',
+ '\u{11211}'), ('\u{11213}', '\u{11234}'), ('\u{11237}', '\u{11237}'), ('\u{1123e}',
+ '\u{1123e}'), ('\u{11280}', '\u{11286}'), ('\u{11288}', '\u{11288}'), ('\u{1128a}',
+ '\u{1128d}'), ('\u{1128f}', '\u{1129d}'), ('\u{1129f}', '\u{112a8}'), ('\u{112b0}',
+ '\u{112e8}'), ('\u{11300}', '\u{11303}'), ('\u{11305}', '\u{1130c}'), ('\u{1130f}',
+ '\u{11310}'), ('\u{11313}', '\u{11328}'), ('\u{1132a}', '\u{11330}'), ('\u{11332}',
+ '\u{11333}'), ('\u{11335}', '\u{11339}'), ('\u{1133d}', '\u{11344}'), ('\u{11347}',
+ '\u{11348}'), ('\u{1134b}', '\u{1134c}'), ('\u{11350}', '\u{11350}'), ('\u{11357}',
+ '\u{11357}'), ('\u{1135d}', '\u{11363}'), ('\u{11400}', '\u{11441}'), ('\u{11443}',
+ '\u{11445}'), ('\u{11447}', '\u{1144a}'), ('\u{1145f}', '\u{11461}'), ('\u{11480}',
+ '\u{114c1}'), ('\u{114c4}', '\u{114c5}'), ('\u{114c7}', '\u{114c7}'), ('\u{11580}',
+ '\u{115b5}'), ('\u{115b8}', '\u{115be}'), ('\u{115d8}', '\u{115dd}'), ('\u{11600}',
+ '\u{1163e}'), ('\u{11640}', '\u{11640}'), ('\u{11644}', '\u{11644}'), ('\u{11680}',
+ '\u{116b5}'), ('\u{116b8}', '\u{116b8}'), ('\u{11700}', '\u{1171a}'), ('\u{1171d}',
+ '\u{1172a}'), ('\u{11800}', '\u{11838}'), ('\u{118a0}', '\u{118df}'), ('\u{118ff}',
+ '\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{1193c}'), ('\u{1193f}', '\u{11942}'), ('\u{119a0}', '\u{119a7}'), ('\u{119aa}',
'\u{119d7}'), ('\u{119da}', '\u{119df}'), ('\u{119e1}', '\u{119e1}'), ('\u{119e3}',
'\u{119e4}'), ('\u{11a00}', '\u{11a32}'), ('\u{11a35}', '\u{11a3e}'), ('\u{11a50}',
'\u{11a97}'), ('\u{11a9d}', '\u{11a9d}'), ('\u{11ac0}', '\u{11af8}'), ('\u{11c00}',
@@ -272,13 +277,14 @@
'\u{11d41}'), ('\u{11d43}', '\u{11d43}'), ('\u{11d46}', '\u{11d47}'), ('\u{11d60}',
'\u{11d65}'), ('\u{11d67}', '\u{11d68}'), ('\u{11d6a}', '\u{11d8e}'), ('\u{11d90}',
'\u{11d91}'), ('\u{11d93}', '\u{11d96}'), ('\u{11d98}', '\u{11d98}'), ('\u{11ee0}',
- '\u{11ef6}'), ('\u{12000}', '\u{12399}'), ('\u{12400}', '\u{1246e}'), ('\u{12480}',
- '\u{12543}'), ('\u{13000}', '\u{1342e}'), ('\u{14400}', '\u{14646}'), ('\u{16800}',
- '\u{16a38}'), ('\u{16a40}', '\u{16a5e}'), ('\u{16ad0}', '\u{16aed}'), ('\u{16b00}',
- '\u{16b2f}'), ('\u{16b40}', '\u{16b43}'), ('\u{16b63}', '\u{16b77}'), ('\u{16b7d}',
- '\u{16b8f}'), ('\u{16e40}', '\u{16e7f}'), ('\u{16f00}', '\u{16f4a}'), ('\u{16f4f}',
- '\u{16f87}'), ('\u{16f8f}', '\u{16f9f}'), ('\u{16fe0}', '\u{16fe1}'), ('\u{16fe3}',
- '\u{16fe3}'), ('\u{17000}', '\u{187f7}'), ('\u{18800}', '\u{18af2}'), ('\u{1b000}',
+ '\u{11ef6}'), ('\u{11fb0}', '\u{11fb0}'), ('\u{12000}', '\u{12399}'), ('\u{12400}',
+ '\u{1246e}'), ('\u{12480}', '\u{12543}'), ('\u{13000}', '\u{1342e}'), ('\u{14400}',
+ '\u{14646}'), ('\u{16800}', '\u{16a38}'), ('\u{16a40}', '\u{16a5e}'), ('\u{16ad0}',
+ '\u{16aed}'), ('\u{16b00}', '\u{16b2f}'), ('\u{16b40}', '\u{16b43}'), ('\u{16b63}',
+ '\u{16b77}'), ('\u{16b7d}', '\u{16b8f}'), ('\u{16e40}', '\u{16e7f}'), ('\u{16f00}',
+ '\u{16f4a}'), ('\u{16f4f}', '\u{16f87}'), ('\u{16f8f}', '\u{16f9f}'), ('\u{16fe0}',
+ '\u{16fe1}'), ('\u{16fe3}', '\u{16fe3}'), ('\u{16ff0}', '\u{16ff1}'), ('\u{17000}',
+ '\u{187f7}'), ('\u{18800}', '\u{18cd5}'), ('\u{18d00}', '\u{18d08}'), ('\u{1b000}',
'\u{1b11e}'), ('\u{1b150}', '\u{1b152}'), ('\u{1b164}', '\u{1b167}'), ('\u{1b170}',
'\u{1b2fb}'), ('\u{1bc00}', '\u{1bc6a}'), ('\u{1bc70}', '\u{1bc7c}'), ('\u{1bc80}',
'\u{1bc88}'), ('\u{1bc90}', '\u{1bc99}'), ('\u{1bc9e}', '\u{1bc9e}'), ('\u{1d400}',
@@ -308,9 +314,9 @@
'\u{1ee7c}'), ('\u{1ee7e}', '\u{1ee7e}'), ('\u{1ee80}', '\u{1ee89}'), ('\u{1ee8b}',
'\u{1ee9b}'), ('\u{1eea1}', '\u{1eea3}'), ('\u{1eea5}', '\u{1eea9}'), ('\u{1eeab}',
'\u{1eebb}'), ('\u{1f130}', '\u{1f149}'), ('\u{1f150}', '\u{1f169}'), ('\u{1f170}',
- '\u{1f189}'), ('\u{20000}', '\u{2a6d6}'), ('\u{2a700}', '\u{2b734}'), ('\u{2b740}',
+ '\u{1f189}'), ('\u{20000}', '\u{2a6dd}'), ('\u{2a700}', '\u{2b734}'), ('\u{2b740}',
'\u{2b81d}'), ('\u{2b820}', '\u{2cea1}'), ('\u{2ceb0}', '\u{2ebe0}'), ('\u{2f800}',
- '\u{2fa1d}')
+ '\u{2fa1d}'), ('\u{30000}', '\u{3134a}')
];
#[inline]
@@ -345,7 +351,8 @@
GC_ZWJ,
}
- fn bsearch_range_value_table(c: char, r: &'static [(char, char, GraphemeCat)]) -> GraphemeCat {
+ fn bsearch_range_value_table(c: char, r: &'static [(char, char, GraphemeCat)]) -> (u32, u32, GraphemeCat) {
+ use core;
use core::cmp::Ordering::{Equal, Less, Greater};
match r.binary_search_by(|&(lo, hi, _)| {
if lo <= c && c <= hi { Equal }
@@ -353,14 +360,20 @@
else { Greater }
}) {
Ok(idx) => {
- let (_, _, cat) = r[idx];
- cat
+ let (lower, upper, cat) = r[idx];
+ (lower as u32, upper as u32, cat)
}
- Err(_) => GC_Any
+ Err(idx) => {
+ (
+ if idx > 0 { r[idx-1].1 as u32 + 1 } else { 0 },
+ r.get(idx).map(|c|c.0 as u32 - 1).unwrap_or(core::u32::MAX),
+ GC_Any,
+ )
+ }
}
}
- pub fn grapheme_category(c: char) -> GraphemeCat {
+ pub fn grapheme_category(c: char) -> (u32, u32, GraphemeCat) {
bsearch_range_value_table(c, grapheme_cat_table)
}
@@ -403,7 +416,7 @@
GC_Extend), ('\u{b02}', '\u{b03}', GC_SpacingMark), ('\u{b3c}', '\u{b3c}', GC_Extend),
('\u{b3e}', '\u{b3f}', GC_Extend), ('\u{b40}', '\u{b40}', GC_SpacingMark), ('\u{b41}',
'\u{b44}', GC_Extend), ('\u{b47}', '\u{b48}', GC_SpacingMark), ('\u{b4b}', '\u{b4c}',
- GC_SpacingMark), ('\u{b4d}', '\u{b4d}', GC_Extend), ('\u{b56}', '\u{b57}', GC_Extend),
+ GC_SpacingMark), ('\u{b4d}', '\u{b4d}', GC_Extend), ('\u{b55}', '\u{b57}', GC_Extend),
('\u{b62}', '\u{b63}', GC_Extend), ('\u{b82}', '\u{b82}', GC_Extend), ('\u{bbe}', '\u{bbe}',
GC_Extend), ('\u{bbf}', '\u{bbf}', GC_SpacingMark), ('\u{bc0}', '\u{bc0}', GC_Extend),
('\u{bc1}', '\u{bc2}', GC_SpacingMark), ('\u{bc6}', '\u{bc8}', GC_SpacingMark), ('\u{bca}',
@@ -422,44 +435,44 @@
'\u{d3e}', GC_Extend), ('\u{d3f}', '\u{d40}', GC_SpacingMark), ('\u{d41}', '\u{d44}',
GC_Extend), ('\u{d46}', '\u{d48}', GC_SpacingMark), ('\u{d4a}', '\u{d4c}', GC_SpacingMark),
('\u{d4d}', '\u{d4d}', GC_Extend), ('\u{d4e}', '\u{d4e}', GC_Prepend), ('\u{d57}',
- '\u{d57}', GC_Extend), ('\u{d62}', '\u{d63}', GC_Extend), ('\u{d82}', '\u{d83}',
- GC_SpacingMark), ('\u{dca}', '\u{dca}', GC_Extend), ('\u{dcf}', '\u{dcf}', GC_Extend),
- ('\u{dd0}', '\u{dd1}', GC_SpacingMark), ('\u{dd2}', '\u{dd4}', GC_Extend), ('\u{dd6}',
- '\u{dd6}', GC_Extend), ('\u{dd8}', '\u{dde}', GC_SpacingMark), ('\u{ddf}', '\u{ddf}',
- GC_Extend), ('\u{df2}', '\u{df3}', GC_SpacingMark), ('\u{e31}', '\u{e31}', GC_Extend),
- ('\u{e33}', '\u{e33}', GC_SpacingMark), ('\u{e34}', '\u{e3a}', GC_Extend), ('\u{e47}',
- '\u{e4e}', GC_Extend), ('\u{eb1}', '\u{eb1}', GC_Extend), ('\u{eb3}', '\u{eb3}',
- GC_SpacingMark), ('\u{eb4}', '\u{ebc}', GC_Extend), ('\u{ec8}', '\u{ecd}', GC_Extend),
- ('\u{f18}', '\u{f19}', GC_Extend), ('\u{f35}', '\u{f35}', GC_Extend), ('\u{f37}', '\u{f37}',
- GC_Extend), ('\u{f39}', '\u{f39}', GC_Extend), ('\u{f3e}', '\u{f3f}', GC_SpacingMark),
- ('\u{f71}', '\u{f7e}', GC_Extend), ('\u{f7f}', '\u{f7f}', GC_SpacingMark), ('\u{f80}',
- '\u{f84}', GC_Extend), ('\u{f86}', '\u{f87}', GC_Extend), ('\u{f8d}', '\u{f97}', GC_Extend),
- ('\u{f99}', '\u{fbc}', GC_Extend), ('\u{fc6}', '\u{fc6}', GC_Extend), ('\u{102d}',
- '\u{1030}', GC_Extend), ('\u{1031}', '\u{1031}', GC_SpacingMark), ('\u{1032}', '\u{1037}',
- GC_Extend), ('\u{1039}', '\u{103a}', GC_Extend), ('\u{103b}', '\u{103c}', GC_SpacingMark),
- ('\u{103d}', '\u{103e}', GC_Extend), ('\u{1056}', '\u{1057}', GC_SpacingMark), ('\u{1058}',
- '\u{1059}', GC_Extend), ('\u{105e}', '\u{1060}', GC_Extend), ('\u{1071}', '\u{1074}',
- GC_Extend), ('\u{1082}', '\u{1082}', GC_Extend), ('\u{1084}', '\u{1084}', GC_SpacingMark),
- ('\u{1085}', '\u{1086}', GC_Extend), ('\u{108d}', '\u{108d}', GC_Extend), ('\u{109d}',
- '\u{109d}', GC_Extend), ('\u{1100}', '\u{115f}', GC_L), ('\u{1160}', '\u{11a7}', GC_V),
- ('\u{11a8}', '\u{11ff}', GC_T), ('\u{135d}', '\u{135f}', GC_Extend), ('\u{1712}',
- '\u{1714}', GC_Extend), ('\u{1732}', '\u{1734}', GC_Extend), ('\u{1752}', '\u{1753}',
- GC_Extend), ('\u{1772}', '\u{1773}', GC_Extend), ('\u{17b4}', '\u{17b5}', GC_Extend),
- ('\u{17b6}', '\u{17b6}', GC_SpacingMark), ('\u{17b7}', '\u{17bd}', GC_Extend), ('\u{17be}',
- '\u{17c5}', GC_SpacingMark), ('\u{17c6}', '\u{17c6}', GC_Extend), ('\u{17c7}', '\u{17c8}',
- GC_SpacingMark), ('\u{17c9}', '\u{17d3}', GC_Extend), ('\u{17dd}', '\u{17dd}', GC_Extend),
- ('\u{180b}', '\u{180d}', GC_Extend), ('\u{180e}', '\u{180e}', GC_Control), ('\u{1885}',
- '\u{1886}', GC_Extend), ('\u{18a9}', '\u{18a9}', GC_Extend), ('\u{1920}', '\u{1922}',
- GC_Extend), ('\u{1923}', '\u{1926}', GC_SpacingMark), ('\u{1927}', '\u{1928}', GC_Extend),
- ('\u{1929}', '\u{192b}', GC_SpacingMark), ('\u{1930}', '\u{1931}', GC_SpacingMark),
- ('\u{1932}', '\u{1932}', GC_Extend), ('\u{1933}', '\u{1938}', GC_SpacingMark), ('\u{1939}',
- '\u{193b}', GC_Extend), ('\u{1a17}', '\u{1a18}', GC_Extend), ('\u{1a19}', '\u{1a1a}',
- GC_SpacingMark), ('\u{1a1b}', '\u{1a1b}', GC_Extend), ('\u{1a55}', '\u{1a55}',
- GC_SpacingMark), ('\u{1a56}', '\u{1a56}', GC_Extend), ('\u{1a57}', '\u{1a57}',
+ '\u{d57}', GC_Extend), ('\u{d62}', '\u{d63}', GC_Extend), ('\u{d81}', '\u{d81}', GC_Extend),
+ ('\u{d82}', '\u{d83}', GC_SpacingMark), ('\u{dca}', '\u{dca}', GC_Extend), ('\u{dcf}',
+ '\u{dcf}', GC_Extend), ('\u{dd0}', '\u{dd1}', GC_SpacingMark), ('\u{dd2}', '\u{dd4}',
+ GC_Extend), ('\u{dd6}', '\u{dd6}', GC_Extend), ('\u{dd8}', '\u{dde}', GC_SpacingMark),
+ ('\u{ddf}', '\u{ddf}', GC_Extend), ('\u{df2}', '\u{df3}', GC_SpacingMark), ('\u{e31}',
+ '\u{e31}', GC_Extend), ('\u{e33}', '\u{e33}', GC_SpacingMark), ('\u{e34}', '\u{e3a}',
+ GC_Extend), ('\u{e47}', '\u{e4e}', GC_Extend), ('\u{eb1}', '\u{eb1}', GC_Extend),
+ ('\u{eb3}', '\u{eb3}', GC_SpacingMark), ('\u{eb4}', '\u{ebc}', GC_Extend), ('\u{ec8}',
+ '\u{ecd}', GC_Extend), ('\u{f18}', '\u{f19}', GC_Extend), ('\u{f35}', '\u{f35}', GC_Extend),
+ ('\u{f37}', '\u{f37}', GC_Extend), ('\u{f39}', '\u{f39}', GC_Extend), ('\u{f3e}', '\u{f3f}',
+ GC_SpacingMark), ('\u{f71}', '\u{f7e}', GC_Extend), ('\u{f7f}', '\u{f7f}', GC_SpacingMark),
+ ('\u{f80}', '\u{f84}', GC_Extend), ('\u{f86}', '\u{f87}', GC_Extend), ('\u{f8d}', '\u{f97}',
+ GC_Extend), ('\u{f99}', '\u{fbc}', GC_Extend), ('\u{fc6}', '\u{fc6}', GC_Extend),
+ ('\u{102d}', '\u{1030}', GC_Extend), ('\u{1031}', '\u{1031}', GC_SpacingMark), ('\u{1032}',
+ '\u{1037}', GC_Extend), ('\u{1039}', '\u{103a}', GC_Extend), ('\u{103b}', '\u{103c}',
+ GC_SpacingMark), ('\u{103d}', '\u{103e}', GC_Extend), ('\u{1056}', '\u{1057}',
+ GC_SpacingMark), ('\u{1058}', '\u{1059}', GC_Extend), ('\u{105e}', '\u{1060}', GC_Extend),
+ ('\u{1071}', '\u{1074}', GC_Extend), ('\u{1082}', '\u{1082}', GC_Extend), ('\u{1084}',
+ '\u{1084}', GC_SpacingMark), ('\u{1085}', '\u{1086}', GC_Extend), ('\u{108d}', '\u{108d}',
+ GC_Extend), ('\u{109d}', '\u{109d}', GC_Extend), ('\u{1100}', '\u{115f}', GC_L),
+ ('\u{1160}', '\u{11a7}', GC_V), ('\u{11a8}', '\u{11ff}', GC_T), ('\u{135d}', '\u{135f}',
+ GC_Extend), ('\u{1712}', '\u{1714}', GC_Extend), ('\u{1732}', '\u{1734}', GC_Extend),
+ ('\u{1752}', '\u{1753}', GC_Extend), ('\u{1772}', '\u{1773}', GC_Extend), ('\u{17b4}',
+ '\u{17b5}', GC_Extend), ('\u{17b6}', '\u{17b6}', GC_SpacingMark), ('\u{17b7}', '\u{17bd}',
+ GC_Extend), ('\u{17be}', '\u{17c5}', GC_SpacingMark), ('\u{17c6}', '\u{17c6}', GC_Extend),
+ ('\u{17c7}', '\u{17c8}', GC_SpacingMark), ('\u{17c9}', '\u{17d3}', GC_Extend), ('\u{17dd}',
+ '\u{17dd}', GC_Extend), ('\u{180b}', '\u{180d}', GC_Extend), ('\u{180e}', '\u{180e}',
+ GC_Control), ('\u{1885}', '\u{1886}', GC_Extend), ('\u{18a9}', '\u{18a9}', GC_Extend),
+ ('\u{1920}', '\u{1922}', GC_Extend), ('\u{1923}', '\u{1926}', GC_SpacingMark), ('\u{1927}',
+ '\u{1928}', GC_Extend), ('\u{1929}', '\u{192b}', GC_SpacingMark), ('\u{1930}', '\u{1931}',
+ GC_SpacingMark), ('\u{1932}', '\u{1932}', GC_Extend), ('\u{1933}', '\u{1938}',
+ GC_SpacingMark), ('\u{1939}', '\u{193b}', GC_Extend), ('\u{1a17}', '\u{1a18}', GC_Extend),
+ ('\u{1a19}', '\u{1a1a}', GC_SpacingMark), ('\u{1a1b}', '\u{1a1b}', GC_Extend), ('\u{1a55}',
+ '\u{1a55}', GC_SpacingMark), ('\u{1a56}', '\u{1a56}', GC_Extend), ('\u{1a57}', '\u{1a57}',
GC_SpacingMark), ('\u{1a58}', '\u{1a5e}', GC_Extend), ('\u{1a60}', '\u{1a60}', GC_Extend),
('\u{1a62}', '\u{1a62}', GC_Extend), ('\u{1a65}', '\u{1a6c}', GC_Extend), ('\u{1a6d}',
'\u{1a72}', GC_SpacingMark), ('\u{1a73}', '\u{1a7c}', GC_Extend), ('\u{1a7f}', '\u{1a7f}',
- GC_Extend), ('\u{1ab0}', '\u{1abe}', GC_Extend), ('\u{1b00}', '\u{1b03}', GC_Extend),
+ GC_Extend), ('\u{1ab0}', '\u{1ac0}', GC_Extend), ('\u{1b00}', '\u{1b03}', GC_Extend),
('\u{1b04}', '\u{1b04}', GC_SpacingMark), ('\u{1b34}', '\u{1b3a}', GC_Extend), ('\u{1b3b}',
'\u{1b3b}', GC_SpacingMark), ('\u{1b3c}', '\u{1b3c}', GC_Extend), ('\u{1b3d}', '\u{1b41}',
GC_SpacingMark), ('\u{1b42}', '\u{1b42}', GC_Extend), ('\u{1b43}', '\u{1b44}',
@@ -515,380 +528,388 @@
GC_Extend), ('\u{a6f0}', '\u{a6f1}', GC_Extend), ('\u{a802}', '\u{a802}', GC_Extend),
('\u{a806}', '\u{a806}', GC_Extend), ('\u{a80b}', '\u{a80b}', GC_Extend), ('\u{a823}',
'\u{a824}', GC_SpacingMark), ('\u{a825}', '\u{a826}', GC_Extend), ('\u{a827}', '\u{a827}',
- GC_SpacingMark), ('\u{a880}', '\u{a881}', GC_SpacingMark), ('\u{a8b4}', '\u{a8c3}',
- GC_SpacingMark), ('\u{a8c4}', '\u{a8c5}', GC_Extend), ('\u{a8e0}', '\u{a8f1}', GC_Extend),
- ('\u{a8ff}', '\u{a8ff}', GC_Extend), ('\u{a926}', '\u{a92d}', GC_Extend), ('\u{a947}',
- '\u{a951}', GC_Extend), ('\u{a952}', '\u{a953}', GC_SpacingMark), ('\u{a960}', '\u{a97c}',
- GC_L), ('\u{a980}', '\u{a982}', GC_Extend), ('\u{a983}', '\u{a983}', GC_SpacingMark),
- ('\u{a9b3}', '\u{a9b3}', GC_Extend), ('\u{a9b4}', '\u{a9b5}', GC_SpacingMark), ('\u{a9b6}',
- '\u{a9b9}', GC_Extend), ('\u{a9ba}', '\u{a9bb}', GC_SpacingMark), ('\u{a9bc}', '\u{a9bd}',
- GC_Extend), ('\u{a9be}', '\u{a9c0}', GC_SpacingMark), ('\u{a9e5}', '\u{a9e5}', GC_Extend),
- ('\u{aa29}', '\u{aa2e}', GC_Extend), ('\u{aa2f}', '\u{aa30}', GC_SpacingMark), ('\u{aa31}',
- '\u{aa32}', GC_Extend), ('\u{aa33}', '\u{aa34}', GC_SpacingMark), ('\u{aa35}', '\u{aa36}',
- GC_Extend), ('\u{aa43}', '\u{aa43}', GC_Extend), ('\u{aa4c}', '\u{aa4c}', GC_Extend),
- ('\u{aa4d}', '\u{aa4d}', GC_SpacingMark), ('\u{aa7c}', '\u{aa7c}', GC_Extend), ('\u{aab0}',
- '\u{aab0}', GC_Extend), ('\u{aab2}', '\u{aab4}', GC_Extend), ('\u{aab7}', '\u{aab8}',
- GC_Extend), ('\u{aabe}', '\u{aabf}', GC_Extend), ('\u{aac1}', '\u{aac1}', GC_Extend),
- ('\u{aaeb}', '\u{aaeb}', GC_SpacingMark), ('\u{aaec}', '\u{aaed}', GC_Extend), ('\u{aaee}',
- '\u{aaef}', GC_SpacingMark), ('\u{aaf5}', '\u{aaf5}', GC_SpacingMark), ('\u{aaf6}',
- '\u{aaf6}', GC_Extend), ('\u{abe3}', '\u{abe4}', GC_SpacingMark), ('\u{abe5}', '\u{abe5}',
- GC_Extend), ('\u{abe6}', '\u{abe7}', GC_SpacingMark), ('\u{abe8}', '\u{abe8}', GC_Extend),
- ('\u{abe9}', '\u{abea}', GC_SpacingMark), ('\u{abec}', '\u{abec}', GC_SpacingMark),
- ('\u{abed}', '\u{abed}', GC_Extend), ('\u{ac00}', '\u{ac00}', GC_LV), ('\u{ac01}',
- '\u{ac1b}', GC_LVT), ('\u{ac1c}', '\u{ac1c}', GC_LV), ('\u{ac1d}', '\u{ac37}', GC_LVT),
- ('\u{ac38}', '\u{ac38}', GC_LV), ('\u{ac39}', '\u{ac53}', GC_LVT), ('\u{ac54}', '\u{ac54}',
- GC_LV), ('\u{ac55}', '\u{ac6f}', GC_LVT), ('\u{ac70}', '\u{ac70}', GC_LV), ('\u{ac71}',
- '\u{ac8b}', GC_LVT), ('\u{ac8c}', '\u{ac8c}', GC_LV), ('\u{ac8d}', '\u{aca7}', GC_LVT),
- ('\u{aca8}', '\u{aca8}', GC_LV), ('\u{aca9}', '\u{acc3}', GC_LVT), ('\u{acc4}', '\u{acc4}',
- GC_LV), ('\u{acc5}', '\u{acdf}', GC_LVT), ('\u{ace0}', '\u{ace0}', GC_LV), ('\u{ace1}',
- '\u{acfb}', GC_LVT), ('\u{acfc}', '\u{acfc}', GC_LV), ('\u{acfd}', '\u{ad17}', GC_LVT),
- ('\u{ad18}', '\u{ad18}', GC_LV), ('\u{ad19}', '\u{ad33}', GC_LVT), ('\u{ad34}', '\u{ad34}',
- GC_LV), ('\u{ad35}', '\u{ad4f}', GC_LVT), ('\u{ad50}', '\u{ad50}', GC_LV), ('\u{ad51}',
- '\u{ad6b}', GC_LVT), ('\u{ad6c}', '\u{ad6c}', GC_LV), ('\u{ad6d}', '\u{ad87}', GC_LVT),
- ('\u{ad88}', '\u{ad88}', GC_LV), ('\u{ad89}', '\u{ada3}', GC_LVT), ('\u{ada4}', '\u{ada4}',
- GC_LV), ('\u{ada5}', '\u{adbf}', GC_LVT), ('\u{adc0}', '\u{adc0}', GC_LV), ('\u{adc1}',
- '\u{addb}', GC_LVT), ('\u{addc}', '\u{addc}', GC_LV), ('\u{addd}', '\u{adf7}', GC_LVT),
- ('\u{adf8}', '\u{adf8}', GC_LV), ('\u{adf9}', '\u{ae13}', GC_LVT), ('\u{ae14}', '\u{ae14}',
- GC_LV), ('\u{ae15}', '\u{ae2f}', GC_LVT), ('\u{ae30}', '\u{ae30}', GC_LV), ('\u{ae31}',
- '\u{ae4b}', GC_LVT), ('\u{ae4c}', '\u{ae4c}', GC_LV), ('\u{ae4d}', '\u{ae67}', GC_LVT),
- ('\u{ae68}', '\u{ae68}', GC_LV), ('\u{ae69}', '\u{ae83}', GC_LVT), ('\u{ae84}', '\u{ae84}',
- GC_LV), ('\u{ae85}', '\u{ae9f}', GC_LVT), ('\u{aea0}', '\u{aea0}', GC_LV), ('\u{aea1}',
- '\u{aebb}', GC_LVT), ('\u{aebc}', '\u{aebc}', GC_LV), ('\u{aebd}', '\u{aed7}', GC_LVT),
- ('\u{aed8}', '\u{aed8}', GC_LV), ('\u{aed9}', '\u{aef3}', GC_LVT), ('\u{aef4}', '\u{aef4}',
- GC_LV), ('\u{aef5}', '\u{af0f}', GC_LVT), ('\u{af10}', '\u{af10}', GC_LV), ('\u{af11}',
- '\u{af2b}', GC_LVT), ('\u{af2c}', '\u{af2c}', GC_LV), ('\u{af2d}', '\u{af47}', GC_LVT),
- ('\u{af48}', '\u{af48}', GC_LV), ('\u{af49}', '\u{af63}', GC_LVT), ('\u{af64}', '\u{af64}',
- GC_LV), ('\u{af65}', '\u{af7f}', GC_LVT), ('\u{af80}', '\u{af80}', GC_LV), ('\u{af81}',
- '\u{af9b}', GC_LVT), ('\u{af9c}', '\u{af9c}', GC_LV), ('\u{af9d}', '\u{afb7}', GC_LVT),
- ('\u{afb8}', '\u{afb8}', GC_LV), ('\u{afb9}', '\u{afd3}', GC_LVT), ('\u{afd4}', '\u{afd4}',
- GC_LV), ('\u{afd5}', '\u{afef}', GC_LVT), ('\u{aff0}', '\u{aff0}', GC_LV), ('\u{aff1}',
- '\u{b00b}', GC_LVT), ('\u{b00c}', '\u{b00c}', GC_LV), ('\u{b00d}', '\u{b027}', GC_LVT),
- ('\u{b028}', '\u{b028}', GC_LV), ('\u{b029}', '\u{b043}', GC_LVT), ('\u{b044}', '\u{b044}',
- GC_LV), ('\u{b045}', '\u{b05f}', GC_LVT), ('\u{b060}', '\u{b060}', GC_LV), ('\u{b061}',
- '\u{b07b}', GC_LVT), ('\u{b07c}', '\u{b07c}', GC_LV), ('\u{b07d}', '\u{b097}', GC_LVT),
- ('\u{b098}', '\u{b098}', GC_LV), ('\u{b099}', '\u{b0b3}', GC_LVT), ('\u{b0b4}', '\u{b0b4}',
- GC_LV), ('\u{b0b5}', '\u{b0cf}', GC_LVT), ('\u{b0d0}', '\u{b0d0}', GC_LV), ('\u{b0d1}',
- '\u{b0eb}', GC_LVT), ('\u{b0ec}', '\u{b0ec}', GC_LV), ('\u{b0ed}', '\u{b107}', GC_LVT),
- ('\u{b108}', '\u{b108}', GC_LV), ('\u{b109}', '\u{b123}', GC_LVT), ('\u{b124}', '\u{b124}',
- GC_LV), ('\u{b125}', '\u{b13f}', GC_LVT), ('\u{b140}', '\u{b140}', GC_LV), ('\u{b141}',
- '\u{b15b}', GC_LVT), ('\u{b15c}', '\u{b15c}', GC_LV), ('\u{b15d}', '\u{b177}', GC_LVT),
- ('\u{b178}', '\u{b178}', GC_LV), ('\u{b179}', '\u{b193}', GC_LVT), ('\u{b194}', '\u{b194}',
- GC_LV), ('\u{b195}', '\u{b1af}', GC_LVT), ('\u{b1b0}', '\u{b1b0}', GC_LV), ('\u{b1b1}',
- '\u{b1cb}', GC_LVT), ('\u{b1cc}', '\u{b1cc}', GC_LV), ('\u{b1cd}', '\u{b1e7}', GC_LVT),
- ('\u{b1e8}', '\u{b1e8}', GC_LV), ('\u{b1e9}', '\u{b203}', GC_LVT), ('\u{b204}', '\u{b204}',
- GC_LV), ('\u{b205}', '\u{b21f}', GC_LVT), ('\u{b220}', '\u{b220}', GC_LV), ('\u{b221}',
- '\u{b23b}', GC_LVT), ('\u{b23c}', '\u{b23c}', GC_LV), ('\u{b23d}', '\u{b257}', GC_LVT),
- ('\u{b258}', '\u{b258}', GC_LV), ('\u{b259}', '\u{b273}', GC_LVT), ('\u{b274}', '\u{b274}',
- GC_LV), ('\u{b275}', '\u{b28f}', GC_LVT), ('\u{b290}', '\u{b290}', GC_LV), ('\u{b291}',
- '\u{b2ab}', GC_LVT), ('\u{b2ac}', '\u{b2ac}', GC_LV), ('\u{b2ad}', '\u{b2c7}', GC_LVT),
- ('\u{b2c8}', '\u{b2c8}', GC_LV), ('\u{b2c9}', '\u{b2e3}', GC_LVT), ('\u{b2e4}', '\u{b2e4}',
- GC_LV), ('\u{b2e5}', '\u{b2ff}', GC_LVT), ('\u{b300}', '\u{b300}', GC_LV), ('\u{b301}',
- '\u{b31b}', GC_LVT), ('\u{b31c}', '\u{b31c}', GC_LV), ('\u{b31d}', '\u{b337}', GC_LVT),
- ('\u{b338}', '\u{b338}', GC_LV), ('\u{b339}', '\u{b353}', GC_LVT), ('\u{b354}', '\u{b354}',
- GC_LV), ('\u{b355}', '\u{b36f}', GC_LVT), ('\u{b370}', '\u{b370}', GC_LV), ('\u{b371}',
- '\u{b38b}', GC_LVT), ('\u{b38c}', '\u{b38c}', GC_LV), ('\u{b38d}', '\u{b3a7}', GC_LVT),
- ('\u{b3a8}', '\u{b3a8}', GC_LV), ('\u{b3a9}', '\u{b3c3}', GC_LVT), ('\u{b3c4}', '\u{b3c4}',
- GC_LV), ('\u{b3c5}', '\u{b3df}', GC_LVT), ('\u{b3e0}', '\u{b3e0}', GC_LV), ('\u{b3e1}',
- '\u{b3fb}', GC_LVT), ('\u{b3fc}', '\u{b3fc}', GC_LV), ('\u{b3fd}', '\u{b417}', GC_LVT),
- ('\u{b418}', '\u{b418}', GC_LV), ('\u{b419}', '\u{b433}', GC_LVT), ('\u{b434}', '\u{b434}',
- GC_LV), ('\u{b435}', '\u{b44f}', GC_LVT), ('\u{b450}', '\u{b450}', GC_LV), ('\u{b451}',
- '\u{b46b}', GC_LVT), ('\u{b46c}', '\u{b46c}', GC_LV), ('\u{b46d}', '\u{b487}', GC_LVT),
- ('\u{b488}', '\u{b488}', GC_LV), ('\u{b489}', '\u{b4a3}', GC_LVT), ('\u{b4a4}', '\u{b4a4}',
- GC_LV), ('\u{b4a5}', '\u{b4bf}', GC_LVT), ('\u{b4c0}', '\u{b4c0}', GC_LV), ('\u{b4c1}',
- '\u{b4db}', GC_LVT), ('\u{b4dc}', '\u{b4dc}', GC_LV), ('\u{b4dd}', '\u{b4f7}', GC_LVT),
- ('\u{b4f8}', '\u{b4f8}', GC_LV), ('\u{b4f9}', '\u{b513}', GC_LVT), ('\u{b514}', '\u{b514}',
- GC_LV), ('\u{b515}', '\u{b52f}', GC_LVT), ('\u{b530}', '\u{b530}', GC_LV), ('\u{b531}',
- '\u{b54b}', GC_LVT), ('\u{b54c}', '\u{b54c}', GC_LV), ('\u{b54d}', '\u{b567}', GC_LVT),
- ('\u{b568}', '\u{b568}', GC_LV), ('\u{b569}', '\u{b583}', GC_LVT), ('\u{b584}', '\u{b584}',
- GC_LV), ('\u{b585}', '\u{b59f}', GC_LVT), ('\u{b5a0}', '\u{b5a0}', GC_LV), ('\u{b5a1}',
- '\u{b5bb}', GC_LVT), ('\u{b5bc}', '\u{b5bc}', GC_LV), ('\u{b5bd}', '\u{b5d7}', GC_LVT),
- ('\u{b5d8}', '\u{b5d8}', GC_LV), ('\u{b5d9}', '\u{b5f3}', GC_LVT), ('\u{b5f4}', '\u{b5f4}',
- GC_LV), ('\u{b5f5}', '\u{b60f}', GC_LVT), ('\u{b610}', '\u{b610}', GC_LV), ('\u{b611}',
- '\u{b62b}', GC_LVT), ('\u{b62c}', '\u{b62c}', GC_LV), ('\u{b62d}', '\u{b647}', GC_LVT),
- ('\u{b648}', '\u{b648}', GC_LV), ('\u{b649}', '\u{b663}', GC_LVT), ('\u{b664}', '\u{b664}',
- GC_LV), ('\u{b665}', '\u{b67f}', GC_LVT), ('\u{b680}', '\u{b680}', GC_LV), ('\u{b681}',
- '\u{b69b}', GC_LVT), ('\u{b69c}', '\u{b69c}', GC_LV), ('\u{b69d}', '\u{b6b7}', GC_LVT),
- ('\u{b6b8}', '\u{b6b8}', GC_LV), ('\u{b6b9}', '\u{b6d3}', GC_LVT), ('\u{b6d4}', '\u{b6d4}',
- GC_LV), ('\u{b6d5}', '\u{b6ef}', GC_LVT), ('\u{b6f0}', '\u{b6f0}', GC_LV), ('\u{b6f1}',
- '\u{b70b}', GC_LVT), ('\u{b70c}', '\u{b70c}', GC_LV), ('\u{b70d}', '\u{b727}', GC_LVT),
- ('\u{b728}', '\u{b728}', GC_LV), ('\u{b729}', '\u{b743}', GC_LVT), ('\u{b744}', '\u{b744}',
- GC_LV), ('\u{b745}', '\u{b75f}', GC_LVT), ('\u{b760}', '\u{b760}', GC_LV), ('\u{b761}',
- '\u{b77b}', GC_LVT), ('\u{b77c}', '\u{b77c}', GC_LV), ('\u{b77d}', '\u{b797}', GC_LVT),
- ('\u{b798}', '\u{b798}', GC_LV), ('\u{b799}', '\u{b7b3}', GC_LVT), ('\u{b7b4}', '\u{b7b4}',
- GC_LV), ('\u{b7b5}', '\u{b7cf}', GC_LVT), ('\u{b7d0}', '\u{b7d0}', GC_LV), ('\u{b7d1}',
- '\u{b7eb}', GC_LVT), ('\u{b7ec}', '\u{b7ec}', GC_LV), ('\u{b7ed}', '\u{b807}', GC_LVT),
- ('\u{b808}', '\u{b808}', GC_LV), ('\u{b809}', '\u{b823}', GC_LVT), ('\u{b824}', '\u{b824}',
- GC_LV), ('\u{b825}', '\u{b83f}', GC_LVT), ('\u{b840}', '\u{b840}', GC_LV), ('\u{b841}',
- '\u{b85b}', GC_LVT), ('\u{b85c}', '\u{b85c}', GC_LV), ('\u{b85d}', '\u{b877}', GC_LVT),
- ('\u{b878}', '\u{b878}', GC_LV), ('\u{b879}', '\u{b893}', GC_LVT), ('\u{b894}', '\u{b894}',
- GC_LV), ('\u{b895}', '\u{b8af}', GC_LVT), ('\u{b8b0}', '\u{b8b0}', GC_LV), ('\u{b8b1}',
- '\u{b8cb}', GC_LVT), ('\u{b8cc}', '\u{b8cc}', GC_LV), ('\u{b8cd}', '\u{b8e7}', GC_LVT),
- ('\u{b8e8}', '\u{b8e8}', GC_LV), ('\u{b8e9}', '\u{b903}', GC_LVT), ('\u{b904}', '\u{b904}',
- GC_LV), ('\u{b905}', '\u{b91f}', GC_LVT), ('\u{b920}', '\u{b920}', GC_LV), ('\u{b921}',
- '\u{b93b}', GC_LVT), ('\u{b93c}', '\u{b93c}', GC_LV), ('\u{b93d}', '\u{b957}', GC_LVT),
- ('\u{b958}', '\u{b958}', GC_LV), ('\u{b959}', '\u{b973}', GC_LVT), ('\u{b974}', '\u{b974}',
- GC_LV), ('\u{b975}', '\u{b98f}', GC_LVT), ('\u{b990}', '\u{b990}', GC_LV), ('\u{b991}',
- '\u{b9ab}', GC_LVT), ('\u{b9ac}', '\u{b9ac}', GC_LV), ('\u{b9ad}', '\u{b9c7}', GC_LVT),
- ('\u{b9c8}', '\u{b9c8}', GC_LV), ('\u{b9c9}', '\u{b9e3}', GC_LVT), ('\u{b9e4}', '\u{b9e4}',
- GC_LV), ('\u{b9e5}', '\u{b9ff}', GC_LVT), ('\u{ba00}', '\u{ba00}', GC_LV), ('\u{ba01}',
- '\u{ba1b}', GC_LVT), ('\u{ba1c}', '\u{ba1c}', GC_LV), ('\u{ba1d}', '\u{ba37}', GC_LVT),
- ('\u{ba38}', '\u{ba38}', GC_LV), ('\u{ba39}', '\u{ba53}', GC_LVT), ('\u{ba54}', '\u{ba54}',
- GC_LV), ('\u{ba55}', '\u{ba6f}', GC_LVT), ('\u{ba70}', '\u{ba70}', GC_LV), ('\u{ba71}',
- '\u{ba8b}', GC_LVT), ('\u{ba8c}', '\u{ba8c}', GC_LV), ('\u{ba8d}', '\u{baa7}', GC_LVT),
- ('\u{baa8}', '\u{baa8}', GC_LV), ('\u{baa9}', '\u{bac3}', GC_LVT), ('\u{bac4}', '\u{bac4}',
- GC_LV), ('\u{bac5}', '\u{badf}', GC_LVT), ('\u{bae0}', '\u{bae0}', GC_LV), ('\u{bae1}',
- '\u{bafb}', GC_LVT), ('\u{bafc}', '\u{bafc}', GC_LV), ('\u{bafd}', '\u{bb17}', GC_LVT),
- ('\u{bb18}', '\u{bb18}', GC_LV), ('\u{bb19}', '\u{bb33}', GC_LVT), ('\u{bb34}', '\u{bb34}',
- GC_LV), ('\u{bb35}', '\u{bb4f}', GC_LVT), ('\u{bb50}', '\u{bb50}', GC_LV), ('\u{bb51}',
- '\u{bb6b}', GC_LVT), ('\u{bb6c}', '\u{bb6c}', GC_LV), ('\u{bb6d}', '\u{bb87}', GC_LVT),
- ('\u{bb88}', '\u{bb88}', GC_LV), ('\u{bb89}', '\u{bba3}', GC_LVT), ('\u{bba4}', '\u{bba4}',
- GC_LV), ('\u{bba5}', '\u{bbbf}', GC_LVT), ('\u{bbc0}', '\u{bbc0}', GC_LV), ('\u{bbc1}',
- '\u{bbdb}', GC_LVT), ('\u{bbdc}', '\u{bbdc}', GC_LV), ('\u{bbdd}', '\u{bbf7}', GC_LVT),
- ('\u{bbf8}', '\u{bbf8}', GC_LV), ('\u{bbf9}', '\u{bc13}', GC_LVT), ('\u{bc14}', '\u{bc14}',
- GC_LV), ('\u{bc15}', '\u{bc2f}', GC_LVT), ('\u{bc30}', '\u{bc30}', GC_LV), ('\u{bc31}',
- '\u{bc4b}', GC_LVT), ('\u{bc4c}', '\u{bc4c}', GC_LV), ('\u{bc4d}', '\u{bc67}', GC_LVT),
- ('\u{bc68}', '\u{bc68}', GC_LV), ('\u{bc69}', '\u{bc83}', GC_LVT), ('\u{bc84}', '\u{bc84}',
- GC_LV), ('\u{bc85}', '\u{bc9f}', GC_LVT), ('\u{bca0}', '\u{bca0}', GC_LV), ('\u{bca1}',
- '\u{bcbb}', GC_LVT), ('\u{bcbc}', '\u{bcbc}', GC_LV), ('\u{bcbd}', '\u{bcd7}', GC_LVT),
- ('\u{bcd8}', '\u{bcd8}', GC_LV), ('\u{bcd9}', '\u{bcf3}', GC_LVT), ('\u{bcf4}', '\u{bcf4}',
- GC_LV), ('\u{bcf5}', '\u{bd0f}', GC_LVT), ('\u{bd10}', '\u{bd10}', GC_LV), ('\u{bd11}',
- '\u{bd2b}', GC_LVT), ('\u{bd2c}', '\u{bd2c}', GC_LV), ('\u{bd2d}', '\u{bd47}', GC_LVT),
- ('\u{bd48}', '\u{bd48}', GC_LV), ('\u{bd49}', '\u{bd63}', GC_LVT), ('\u{bd64}', '\u{bd64}',
- GC_LV), ('\u{bd65}', '\u{bd7f}', GC_LVT), ('\u{bd80}', '\u{bd80}', GC_LV), ('\u{bd81}',
- '\u{bd9b}', GC_LVT), ('\u{bd9c}', '\u{bd9c}', GC_LV), ('\u{bd9d}', '\u{bdb7}', GC_LVT),
- ('\u{bdb8}', '\u{bdb8}', GC_LV), ('\u{bdb9}', '\u{bdd3}', GC_LVT), ('\u{bdd4}', '\u{bdd4}',
- GC_LV), ('\u{bdd5}', '\u{bdef}', GC_LVT), ('\u{bdf0}', '\u{bdf0}', GC_LV), ('\u{bdf1}',
- '\u{be0b}', GC_LVT), ('\u{be0c}', '\u{be0c}', GC_LV), ('\u{be0d}', '\u{be27}', GC_LVT),
- ('\u{be28}', '\u{be28}', GC_LV), ('\u{be29}', '\u{be43}', GC_LVT), ('\u{be44}', '\u{be44}',
- GC_LV), ('\u{be45}', '\u{be5f}', GC_LVT), ('\u{be60}', '\u{be60}', GC_LV), ('\u{be61}',
- '\u{be7b}', GC_LVT), ('\u{be7c}', '\u{be7c}', GC_LV), ('\u{be7d}', '\u{be97}', GC_LVT),
- ('\u{be98}', '\u{be98}', GC_LV), ('\u{be99}', '\u{beb3}', GC_LVT), ('\u{beb4}', '\u{beb4}',
- GC_LV), ('\u{beb5}', '\u{becf}', GC_LVT), ('\u{bed0}', '\u{bed0}', GC_LV), ('\u{bed1}',
- '\u{beeb}', GC_LVT), ('\u{beec}', '\u{beec}', GC_LV), ('\u{beed}', '\u{bf07}', GC_LVT),
- ('\u{bf08}', '\u{bf08}', GC_LV), ('\u{bf09}', '\u{bf23}', GC_LVT), ('\u{bf24}', '\u{bf24}',
- GC_LV), ('\u{bf25}', '\u{bf3f}', GC_LVT), ('\u{bf40}', '\u{bf40}', GC_LV), ('\u{bf41}',
- '\u{bf5b}', GC_LVT), ('\u{bf5c}', '\u{bf5c}', GC_LV), ('\u{bf5d}', '\u{bf77}', GC_LVT),
- ('\u{bf78}', '\u{bf78}', GC_LV), ('\u{bf79}', '\u{bf93}', GC_LVT), ('\u{bf94}', '\u{bf94}',
- GC_LV), ('\u{bf95}', '\u{bfaf}', GC_LVT), ('\u{bfb0}', '\u{bfb0}', GC_LV), ('\u{bfb1}',
- '\u{bfcb}', GC_LVT), ('\u{bfcc}', '\u{bfcc}', GC_LV), ('\u{bfcd}', '\u{bfe7}', GC_LVT),
- ('\u{bfe8}', '\u{bfe8}', GC_LV), ('\u{bfe9}', '\u{c003}', GC_LVT), ('\u{c004}', '\u{c004}',
- GC_LV), ('\u{c005}', '\u{c01f}', GC_LVT), ('\u{c020}', '\u{c020}', GC_LV), ('\u{c021}',
- '\u{c03b}', GC_LVT), ('\u{c03c}', '\u{c03c}', GC_LV), ('\u{c03d}', '\u{c057}', GC_LVT),
- ('\u{c058}', '\u{c058}', GC_LV), ('\u{c059}', '\u{c073}', GC_LVT), ('\u{c074}', '\u{c074}',
- GC_LV), ('\u{c075}', '\u{c08f}', GC_LVT), ('\u{c090}', '\u{c090}', GC_LV), ('\u{c091}',
- '\u{c0ab}', GC_LVT), ('\u{c0ac}', '\u{c0ac}', GC_LV), ('\u{c0ad}', '\u{c0c7}', GC_LVT),
- ('\u{c0c8}', '\u{c0c8}', GC_LV), ('\u{c0c9}', '\u{c0e3}', GC_LVT), ('\u{c0e4}', '\u{c0e4}',
- GC_LV), ('\u{c0e5}', '\u{c0ff}', GC_LVT), ('\u{c100}', '\u{c100}', GC_LV), ('\u{c101}',
- '\u{c11b}', GC_LVT), ('\u{c11c}', '\u{c11c}', GC_LV), ('\u{c11d}', '\u{c137}', GC_LVT),
- ('\u{c138}', '\u{c138}', GC_LV), ('\u{c139}', '\u{c153}', GC_LVT), ('\u{c154}', '\u{c154}',
- GC_LV), ('\u{c155}', '\u{c16f}', GC_LVT), ('\u{c170}', '\u{c170}', GC_LV), ('\u{c171}',
- '\u{c18b}', GC_LVT), ('\u{c18c}', '\u{c18c}', GC_LV), ('\u{c18d}', '\u{c1a7}', GC_LVT),
- ('\u{c1a8}', '\u{c1a8}', GC_LV), ('\u{c1a9}', '\u{c1c3}', GC_LVT), ('\u{c1c4}', '\u{c1c4}',
- GC_LV), ('\u{c1c5}', '\u{c1df}', GC_LVT), ('\u{c1e0}', '\u{c1e0}', GC_LV), ('\u{c1e1}',
- '\u{c1fb}', GC_LVT), ('\u{c1fc}', '\u{c1fc}', GC_LV), ('\u{c1fd}', '\u{c217}', GC_LVT),
- ('\u{c218}', '\u{c218}', GC_LV), ('\u{c219}', '\u{c233}', GC_LVT), ('\u{c234}', '\u{c234}',
- GC_LV), ('\u{c235}', '\u{c24f}', GC_LVT), ('\u{c250}', '\u{c250}', GC_LV), ('\u{c251}',
- '\u{c26b}', GC_LVT), ('\u{c26c}', '\u{c26c}', GC_LV), ('\u{c26d}', '\u{c287}', GC_LVT),
- ('\u{c288}', '\u{c288}', GC_LV), ('\u{c289}', '\u{c2a3}', GC_LVT), ('\u{c2a4}', '\u{c2a4}',
- GC_LV), ('\u{c2a5}', '\u{c2bf}', GC_LVT), ('\u{c2c0}', '\u{c2c0}', GC_LV), ('\u{c2c1}',
- '\u{c2db}', GC_LVT), ('\u{c2dc}', '\u{c2dc}', GC_LV), ('\u{c2dd}', '\u{c2f7}', GC_LVT),
- ('\u{c2f8}', '\u{c2f8}', GC_LV), ('\u{c2f9}', '\u{c313}', GC_LVT), ('\u{c314}', '\u{c314}',
- GC_LV), ('\u{c315}', '\u{c32f}', GC_LVT), ('\u{c330}', '\u{c330}', GC_LV), ('\u{c331}',
- '\u{c34b}', GC_LVT), ('\u{c34c}', '\u{c34c}', GC_LV), ('\u{c34d}', '\u{c367}', GC_LVT),
- ('\u{c368}', '\u{c368}', GC_LV), ('\u{c369}', '\u{c383}', GC_LVT), ('\u{c384}', '\u{c384}',
- GC_LV), ('\u{c385}', '\u{c39f}', GC_LVT), ('\u{c3a0}', '\u{c3a0}', GC_LV), ('\u{c3a1}',
- '\u{c3bb}', GC_LVT), ('\u{c3bc}', '\u{c3bc}', GC_LV), ('\u{c3bd}', '\u{c3d7}', GC_LVT),
- ('\u{c3d8}', '\u{c3d8}', GC_LV), ('\u{c3d9}', '\u{c3f3}', GC_LVT), ('\u{c3f4}', '\u{c3f4}',
- GC_LV), ('\u{c3f5}', '\u{c40f}', GC_LVT), ('\u{c410}', '\u{c410}', GC_LV), ('\u{c411}',
- '\u{c42b}', GC_LVT), ('\u{c42c}', '\u{c42c}', GC_LV), ('\u{c42d}', '\u{c447}', GC_LVT),
- ('\u{c448}', '\u{c448}', GC_LV), ('\u{c449}', '\u{c463}', GC_LVT), ('\u{c464}', '\u{c464}',
- GC_LV), ('\u{c465}', '\u{c47f}', GC_LVT), ('\u{c480}', '\u{c480}', GC_LV), ('\u{c481}',
- '\u{c49b}', GC_LVT), ('\u{c49c}', '\u{c49c}', GC_LV), ('\u{c49d}', '\u{c4b7}', GC_LVT),
- ('\u{c4b8}', '\u{c4b8}', GC_LV), ('\u{c4b9}', '\u{c4d3}', GC_LVT), ('\u{c4d4}', '\u{c4d4}',
- GC_LV), ('\u{c4d5}', '\u{c4ef}', GC_LVT), ('\u{c4f0}', '\u{c4f0}', GC_LV), ('\u{c4f1}',
- '\u{c50b}', GC_LVT), ('\u{c50c}', '\u{c50c}', GC_LV), ('\u{c50d}', '\u{c527}', GC_LVT),
- ('\u{c528}', '\u{c528}', GC_LV), ('\u{c529}', '\u{c543}', GC_LVT), ('\u{c544}', '\u{c544}',
- GC_LV), ('\u{c545}', '\u{c55f}', GC_LVT), ('\u{c560}', '\u{c560}', GC_LV), ('\u{c561}',
- '\u{c57b}', GC_LVT), ('\u{c57c}', '\u{c57c}', GC_LV), ('\u{c57d}', '\u{c597}', GC_LVT),
- ('\u{c598}', '\u{c598}', GC_LV), ('\u{c599}', '\u{c5b3}', GC_LVT), ('\u{c5b4}', '\u{c5b4}',
- GC_LV), ('\u{c5b5}', '\u{c5cf}', GC_LVT), ('\u{c5d0}', '\u{c5d0}', GC_LV), ('\u{c5d1}',
- '\u{c5eb}', GC_LVT), ('\u{c5ec}', '\u{c5ec}', GC_LV), ('\u{c5ed}', '\u{c607}', GC_LVT),
- ('\u{c608}', '\u{c608}', GC_LV), ('\u{c609}', '\u{c623}', GC_LVT), ('\u{c624}', '\u{c624}',
- GC_LV), ('\u{c625}', '\u{c63f}', GC_LVT), ('\u{c640}', '\u{c640}', GC_LV), ('\u{c641}',
- '\u{c65b}', GC_LVT), ('\u{c65c}', '\u{c65c}', GC_LV), ('\u{c65d}', '\u{c677}', GC_LVT),
- ('\u{c678}', '\u{c678}', GC_LV), ('\u{c679}', '\u{c693}', GC_LVT), ('\u{c694}', '\u{c694}',
- GC_LV), ('\u{c695}', '\u{c6af}', GC_LVT), ('\u{c6b0}', '\u{c6b0}', GC_LV), ('\u{c6b1}',
- '\u{c6cb}', GC_LVT), ('\u{c6cc}', '\u{c6cc}', GC_LV), ('\u{c6cd}', '\u{c6e7}', GC_LVT),
- ('\u{c6e8}', '\u{c6e8}', GC_LV), ('\u{c6e9}', '\u{c703}', GC_LVT), ('\u{c704}', '\u{c704}',
- GC_LV), ('\u{c705}', '\u{c71f}', GC_LVT), ('\u{c720}', '\u{c720}', GC_LV), ('\u{c721}',
- '\u{c73b}', GC_LVT), ('\u{c73c}', '\u{c73c}', GC_LV), ('\u{c73d}', '\u{c757}', GC_LVT),
- ('\u{c758}', '\u{c758}', GC_LV), ('\u{c759}', '\u{c773}', GC_LVT), ('\u{c774}', '\u{c774}',
- GC_LV), ('\u{c775}', '\u{c78f}', GC_LVT), ('\u{c790}', '\u{c790}', GC_LV), ('\u{c791}',
- '\u{c7ab}', GC_LVT), ('\u{c7ac}', '\u{c7ac}', GC_LV), ('\u{c7ad}', '\u{c7c7}', GC_LVT),
- ('\u{c7c8}', '\u{c7c8}', GC_LV), ('\u{c7c9}', '\u{c7e3}', GC_LVT), ('\u{c7e4}', '\u{c7e4}',
- GC_LV), ('\u{c7e5}', '\u{c7ff}', GC_LVT), ('\u{c800}', '\u{c800}', GC_LV), ('\u{c801}',
- '\u{c81b}', GC_LVT), ('\u{c81c}', '\u{c81c}', GC_LV), ('\u{c81d}', '\u{c837}', GC_LVT),
- ('\u{c838}', '\u{c838}', GC_LV), ('\u{c839}', '\u{c853}', GC_LVT), ('\u{c854}', '\u{c854}',
- GC_LV), ('\u{c855}', '\u{c86f}', GC_LVT), ('\u{c870}', '\u{c870}', GC_LV), ('\u{c871}',
- '\u{c88b}', GC_LVT), ('\u{c88c}', '\u{c88c}', GC_LV), ('\u{c88d}', '\u{c8a7}', GC_LVT),
- ('\u{c8a8}', '\u{c8a8}', GC_LV), ('\u{c8a9}', '\u{c8c3}', GC_LVT), ('\u{c8c4}', '\u{c8c4}',
- GC_LV), ('\u{c8c5}', '\u{c8df}', GC_LVT), ('\u{c8e0}', '\u{c8e0}', GC_LV), ('\u{c8e1}',
- '\u{c8fb}', GC_LVT), ('\u{c8fc}', '\u{c8fc}', GC_LV), ('\u{c8fd}', '\u{c917}', GC_LVT),
- ('\u{c918}', '\u{c918}', GC_LV), ('\u{c919}', '\u{c933}', GC_LVT), ('\u{c934}', '\u{c934}',
- GC_LV), ('\u{c935}', '\u{c94f}', GC_LVT), ('\u{c950}', '\u{c950}', GC_LV), ('\u{c951}',
- '\u{c96b}', GC_LVT), ('\u{c96c}', '\u{c96c}', GC_LV), ('\u{c96d}', '\u{c987}', GC_LVT),
- ('\u{c988}', '\u{c988}', GC_LV), ('\u{c989}', '\u{c9a3}', GC_LVT), ('\u{c9a4}', '\u{c9a4}',
- GC_LV), ('\u{c9a5}', '\u{c9bf}', GC_LVT), ('\u{c9c0}', '\u{c9c0}', GC_LV), ('\u{c9c1}',
- '\u{c9db}', GC_LVT), ('\u{c9dc}', '\u{c9dc}', GC_LV), ('\u{c9dd}', '\u{c9f7}', GC_LVT),
- ('\u{c9f8}', '\u{c9f8}', GC_LV), ('\u{c9f9}', '\u{ca13}', GC_LVT), ('\u{ca14}', '\u{ca14}',
- GC_LV), ('\u{ca15}', '\u{ca2f}', GC_LVT), ('\u{ca30}', '\u{ca30}', GC_LV), ('\u{ca31}',
- '\u{ca4b}', GC_LVT), ('\u{ca4c}', '\u{ca4c}', GC_LV), ('\u{ca4d}', '\u{ca67}', GC_LVT),
- ('\u{ca68}', '\u{ca68}', GC_LV), ('\u{ca69}', '\u{ca83}', GC_LVT), ('\u{ca84}', '\u{ca84}',
- GC_LV), ('\u{ca85}', '\u{ca9f}', GC_LVT), ('\u{caa0}', '\u{caa0}', GC_LV), ('\u{caa1}',
- '\u{cabb}', GC_LVT), ('\u{cabc}', '\u{cabc}', GC_LV), ('\u{cabd}', '\u{cad7}', GC_LVT),
- ('\u{cad8}', '\u{cad8}', GC_LV), ('\u{cad9}', '\u{caf3}', GC_LVT), ('\u{caf4}', '\u{caf4}',
- GC_LV), ('\u{caf5}', '\u{cb0f}', GC_LVT), ('\u{cb10}', '\u{cb10}', GC_LV), ('\u{cb11}',
- '\u{cb2b}', GC_LVT), ('\u{cb2c}', '\u{cb2c}', GC_LV), ('\u{cb2d}', '\u{cb47}', GC_LVT),
- ('\u{cb48}', '\u{cb48}', GC_LV), ('\u{cb49}', '\u{cb63}', GC_LVT), ('\u{cb64}', '\u{cb64}',
- GC_LV), ('\u{cb65}', '\u{cb7f}', GC_LVT), ('\u{cb80}', '\u{cb80}', GC_LV), ('\u{cb81}',
- '\u{cb9b}', GC_LVT), ('\u{cb9c}', '\u{cb9c}', GC_LV), ('\u{cb9d}', '\u{cbb7}', GC_LVT),
- ('\u{cbb8}', '\u{cbb8}', GC_LV), ('\u{cbb9}', '\u{cbd3}', GC_LVT), ('\u{cbd4}', '\u{cbd4}',
- GC_LV), ('\u{cbd5}', '\u{cbef}', GC_LVT), ('\u{cbf0}', '\u{cbf0}', GC_LV), ('\u{cbf1}',
- '\u{cc0b}', GC_LVT), ('\u{cc0c}', '\u{cc0c}', GC_LV), ('\u{cc0d}', '\u{cc27}', GC_LVT),
- ('\u{cc28}', '\u{cc28}', GC_LV), ('\u{cc29}', '\u{cc43}', GC_LVT), ('\u{cc44}', '\u{cc44}',
- GC_LV), ('\u{cc45}', '\u{cc5f}', GC_LVT), ('\u{cc60}', '\u{cc60}', GC_LV), ('\u{cc61}',
- '\u{cc7b}', GC_LVT), ('\u{cc7c}', '\u{cc7c}', GC_LV), ('\u{cc7d}', '\u{cc97}', GC_LVT),
- ('\u{cc98}', '\u{cc98}', GC_LV), ('\u{cc99}', '\u{ccb3}', GC_LVT), ('\u{ccb4}', '\u{ccb4}',
- GC_LV), ('\u{ccb5}', '\u{cccf}', GC_LVT), ('\u{ccd0}', '\u{ccd0}', GC_LV), ('\u{ccd1}',
- '\u{cceb}', GC_LVT), ('\u{ccec}', '\u{ccec}', GC_LV), ('\u{cced}', '\u{cd07}', GC_LVT),
- ('\u{cd08}', '\u{cd08}', GC_LV), ('\u{cd09}', '\u{cd23}', GC_LVT), ('\u{cd24}', '\u{cd24}',
- GC_LV), ('\u{cd25}', '\u{cd3f}', GC_LVT), ('\u{cd40}', '\u{cd40}', GC_LV), ('\u{cd41}',
- '\u{cd5b}', GC_LVT), ('\u{cd5c}', '\u{cd5c}', GC_LV), ('\u{cd5d}', '\u{cd77}', GC_LVT),
- ('\u{cd78}', '\u{cd78}', GC_LV), ('\u{cd79}', '\u{cd93}', GC_LVT), ('\u{cd94}', '\u{cd94}',
- GC_LV), ('\u{cd95}', '\u{cdaf}', GC_LVT), ('\u{cdb0}', '\u{cdb0}', GC_LV), ('\u{cdb1}',
- '\u{cdcb}', GC_LVT), ('\u{cdcc}', '\u{cdcc}', GC_LV), ('\u{cdcd}', '\u{cde7}', GC_LVT),
- ('\u{cde8}', '\u{cde8}', GC_LV), ('\u{cde9}', '\u{ce03}', GC_LVT), ('\u{ce04}', '\u{ce04}',
- GC_LV), ('\u{ce05}', '\u{ce1f}', GC_LVT), ('\u{ce20}', '\u{ce20}', GC_LV), ('\u{ce21}',
- '\u{ce3b}', GC_LVT), ('\u{ce3c}', '\u{ce3c}', GC_LV), ('\u{ce3d}', '\u{ce57}', GC_LVT),
- ('\u{ce58}', '\u{ce58}', GC_LV), ('\u{ce59}', '\u{ce73}', GC_LVT), ('\u{ce74}', '\u{ce74}',
- GC_LV), ('\u{ce75}', '\u{ce8f}', GC_LVT), ('\u{ce90}', '\u{ce90}', GC_LV), ('\u{ce91}',
- '\u{ceab}', GC_LVT), ('\u{ceac}', '\u{ceac}', GC_LV), ('\u{cead}', '\u{cec7}', GC_LVT),
- ('\u{cec8}', '\u{cec8}', GC_LV), ('\u{cec9}', '\u{cee3}', GC_LVT), ('\u{cee4}', '\u{cee4}',
- GC_LV), ('\u{cee5}', '\u{ceff}', GC_LVT), ('\u{cf00}', '\u{cf00}', GC_LV), ('\u{cf01}',
- '\u{cf1b}', GC_LVT), ('\u{cf1c}', '\u{cf1c}', GC_LV), ('\u{cf1d}', '\u{cf37}', GC_LVT),
- ('\u{cf38}', '\u{cf38}', GC_LV), ('\u{cf39}', '\u{cf53}', GC_LVT), ('\u{cf54}', '\u{cf54}',
- GC_LV), ('\u{cf55}', '\u{cf6f}', GC_LVT), ('\u{cf70}', '\u{cf70}', GC_LV), ('\u{cf71}',
- '\u{cf8b}', GC_LVT), ('\u{cf8c}', '\u{cf8c}', GC_LV), ('\u{cf8d}', '\u{cfa7}', GC_LVT),
- ('\u{cfa8}', '\u{cfa8}', GC_LV), ('\u{cfa9}', '\u{cfc3}', GC_LVT), ('\u{cfc4}', '\u{cfc4}',
- GC_LV), ('\u{cfc5}', '\u{cfdf}', GC_LVT), ('\u{cfe0}', '\u{cfe0}', GC_LV), ('\u{cfe1}',
- '\u{cffb}', GC_LVT), ('\u{cffc}', '\u{cffc}', GC_LV), ('\u{cffd}', '\u{d017}', GC_LVT),
- ('\u{d018}', '\u{d018}', GC_LV), ('\u{d019}', '\u{d033}', GC_LVT), ('\u{d034}', '\u{d034}',
- GC_LV), ('\u{d035}', '\u{d04f}', GC_LVT), ('\u{d050}', '\u{d050}', GC_LV), ('\u{d051}',
- '\u{d06b}', GC_LVT), ('\u{d06c}', '\u{d06c}', GC_LV), ('\u{d06d}', '\u{d087}', GC_LVT),
- ('\u{d088}', '\u{d088}', GC_LV), ('\u{d089}', '\u{d0a3}', GC_LVT), ('\u{d0a4}', '\u{d0a4}',
- GC_LV), ('\u{d0a5}', '\u{d0bf}', GC_LVT), ('\u{d0c0}', '\u{d0c0}', GC_LV), ('\u{d0c1}',
- '\u{d0db}', GC_LVT), ('\u{d0dc}', '\u{d0dc}', GC_LV), ('\u{d0dd}', '\u{d0f7}', GC_LVT),
- ('\u{d0f8}', '\u{d0f8}', GC_LV), ('\u{d0f9}', '\u{d113}', GC_LVT), ('\u{d114}', '\u{d114}',
- GC_LV), ('\u{d115}', '\u{d12f}', GC_LVT), ('\u{d130}', '\u{d130}', GC_LV), ('\u{d131}',
- '\u{d14b}', GC_LVT), ('\u{d14c}', '\u{d14c}', GC_LV), ('\u{d14d}', '\u{d167}', GC_LVT),
- ('\u{d168}', '\u{d168}', GC_LV), ('\u{d169}', '\u{d183}', GC_LVT), ('\u{d184}', '\u{d184}',
- GC_LV), ('\u{d185}', '\u{d19f}', GC_LVT), ('\u{d1a0}', '\u{d1a0}', GC_LV), ('\u{d1a1}',
- '\u{d1bb}', GC_LVT), ('\u{d1bc}', '\u{d1bc}', GC_LV), ('\u{d1bd}', '\u{d1d7}', GC_LVT),
- ('\u{d1d8}', '\u{d1d8}', GC_LV), ('\u{d1d9}', '\u{d1f3}', GC_LVT), ('\u{d1f4}', '\u{d1f4}',
- GC_LV), ('\u{d1f5}', '\u{d20f}', GC_LVT), ('\u{d210}', '\u{d210}', GC_LV), ('\u{d211}',
- '\u{d22b}', GC_LVT), ('\u{d22c}', '\u{d22c}', GC_LV), ('\u{d22d}', '\u{d247}', GC_LVT),
- ('\u{d248}', '\u{d248}', GC_LV), ('\u{d249}', '\u{d263}', GC_LVT), ('\u{d264}', '\u{d264}',
- GC_LV), ('\u{d265}', '\u{d27f}', GC_LVT), ('\u{d280}', '\u{d280}', GC_LV), ('\u{d281}',
- '\u{d29b}', GC_LVT), ('\u{d29c}', '\u{d29c}', GC_LV), ('\u{d29d}', '\u{d2b7}', GC_LVT),
- ('\u{d2b8}', '\u{d2b8}', GC_LV), ('\u{d2b9}', '\u{d2d3}', GC_LVT), ('\u{d2d4}', '\u{d2d4}',
- GC_LV), ('\u{d2d5}', '\u{d2ef}', GC_LVT), ('\u{d2f0}', '\u{d2f0}', GC_LV), ('\u{d2f1}',
- '\u{d30b}', GC_LVT), ('\u{d30c}', '\u{d30c}', GC_LV), ('\u{d30d}', '\u{d327}', GC_LVT),
- ('\u{d328}', '\u{d328}', GC_LV), ('\u{d329}', '\u{d343}', GC_LVT), ('\u{d344}', '\u{d344}',
- GC_LV), ('\u{d345}', '\u{d35f}', GC_LVT), ('\u{d360}', '\u{d360}', GC_LV), ('\u{d361}',
- '\u{d37b}', GC_LVT), ('\u{d37c}', '\u{d37c}', GC_LV), ('\u{d37d}', '\u{d397}', GC_LVT),
- ('\u{d398}', '\u{d398}', GC_LV), ('\u{d399}', '\u{d3b3}', GC_LVT), ('\u{d3b4}', '\u{d3b4}',
- GC_LV), ('\u{d3b5}', '\u{d3cf}', GC_LVT), ('\u{d3d0}', '\u{d3d0}', GC_LV), ('\u{d3d1}',
- '\u{d3eb}', GC_LVT), ('\u{d3ec}', '\u{d3ec}', GC_LV), ('\u{d3ed}', '\u{d407}', GC_LVT),
- ('\u{d408}', '\u{d408}', GC_LV), ('\u{d409}', '\u{d423}', GC_LVT), ('\u{d424}', '\u{d424}',
- GC_LV), ('\u{d425}', '\u{d43f}', GC_LVT), ('\u{d440}', '\u{d440}', GC_LV), ('\u{d441}',
- '\u{d45b}', GC_LVT), ('\u{d45c}', '\u{d45c}', GC_LV), ('\u{d45d}', '\u{d477}', GC_LVT),
- ('\u{d478}', '\u{d478}', GC_LV), ('\u{d479}', '\u{d493}', GC_LVT), ('\u{d494}', '\u{d494}',
- GC_LV), ('\u{d495}', '\u{d4af}', GC_LVT), ('\u{d4b0}', '\u{d4b0}', GC_LV), ('\u{d4b1}',
- '\u{d4cb}', GC_LVT), ('\u{d4cc}', '\u{d4cc}', GC_LV), ('\u{d4cd}', '\u{d4e7}', GC_LVT),
- ('\u{d4e8}', '\u{d4e8}', GC_LV), ('\u{d4e9}', '\u{d503}', GC_LVT), ('\u{d504}', '\u{d504}',
- GC_LV), ('\u{d505}', '\u{d51f}', GC_LVT), ('\u{d520}', '\u{d520}', GC_LV), ('\u{d521}',
- '\u{d53b}', GC_LVT), ('\u{d53c}', '\u{d53c}', GC_LV), ('\u{d53d}', '\u{d557}', GC_LVT),
- ('\u{d558}', '\u{d558}', GC_LV), ('\u{d559}', '\u{d573}', GC_LVT), ('\u{d574}', '\u{d574}',
- GC_LV), ('\u{d575}', '\u{d58f}', GC_LVT), ('\u{d590}', '\u{d590}', GC_LV), ('\u{d591}',
- '\u{d5ab}', GC_LVT), ('\u{d5ac}', '\u{d5ac}', GC_LV), ('\u{d5ad}', '\u{d5c7}', GC_LVT),
- ('\u{d5c8}', '\u{d5c8}', GC_LV), ('\u{d5c9}', '\u{d5e3}', GC_LVT), ('\u{d5e4}', '\u{d5e4}',
- GC_LV), ('\u{d5e5}', '\u{d5ff}', GC_LVT), ('\u{d600}', '\u{d600}', GC_LV), ('\u{d601}',
- '\u{d61b}', GC_LVT), ('\u{d61c}', '\u{d61c}', GC_LV), ('\u{d61d}', '\u{d637}', GC_LVT),
- ('\u{d638}', '\u{d638}', GC_LV), ('\u{d639}', '\u{d653}', GC_LVT), ('\u{d654}', '\u{d654}',
- GC_LV), ('\u{d655}', '\u{d66f}', GC_LVT), ('\u{d670}', '\u{d670}', GC_LV), ('\u{d671}',
- '\u{d68b}', GC_LVT), ('\u{d68c}', '\u{d68c}', GC_LV), ('\u{d68d}', '\u{d6a7}', GC_LVT),
- ('\u{d6a8}', '\u{d6a8}', GC_LV), ('\u{d6a9}', '\u{d6c3}', GC_LVT), ('\u{d6c4}', '\u{d6c4}',
- GC_LV), ('\u{d6c5}', '\u{d6df}', GC_LVT), ('\u{d6e0}', '\u{d6e0}', GC_LV), ('\u{d6e1}',
- '\u{d6fb}', GC_LVT), ('\u{d6fc}', '\u{d6fc}', GC_LV), ('\u{d6fd}', '\u{d717}', GC_LVT),
- ('\u{d718}', '\u{d718}', GC_LV), ('\u{d719}', '\u{d733}', GC_LVT), ('\u{d734}', '\u{d734}',
- GC_LV), ('\u{d735}', '\u{d74f}', GC_LVT), ('\u{d750}', '\u{d750}', GC_LV), ('\u{d751}',
- '\u{d76b}', GC_LVT), ('\u{d76c}', '\u{d76c}', GC_LV), ('\u{d76d}', '\u{d787}', GC_LVT),
- ('\u{d788}', '\u{d788}', GC_LV), ('\u{d789}', '\u{d7a3}', GC_LVT), ('\u{d7b0}', '\u{d7c6}',
- GC_V), ('\u{d7cb}', '\u{d7fb}', GC_T), ('\u{fb1e}', '\u{fb1e}', GC_Extend), ('\u{fe00}',
- '\u{fe0f}', GC_Extend), ('\u{fe20}', '\u{fe2f}', GC_Extend), ('\u{feff}', '\u{feff}',
- GC_Control), ('\u{ff9e}', '\u{ff9f}', GC_Extend), ('\u{fff0}', '\u{fffb}', GC_Control),
- ('\u{101fd}', '\u{101fd}', GC_Extend), ('\u{102e0}', '\u{102e0}', GC_Extend), ('\u{10376}',
- '\u{1037a}', GC_Extend), ('\u{10a01}', '\u{10a03}', GC_Extend), ('\u{10a05}', '\u{10a06}',
- GC_Extend), ('\u{10a0c}', '\u{10a0f}', GC_Extend), ('\u{10a38}', '\u{10a3a}', GC_Extend),
- ('\u{10a3f}', '\u{10a3f}', GC_Extend), ('\u{10ae5}', '\u{10ae6}', GC_Extend), ('\u{10d24}',
- '\u{10d27}', GC_Extend), ('\u{10f46}', '\u{10f50}', GC_Extend), ('\u{11000}', '\u{11000}',
- GC_SpacingMark), ('\u{11001}', '\u{11001}', GC_Extend), ('\u{11002}', '\u{11002}',
- GC_SpacingMark), ('\u{11038}', '\u{11046}', GC_Extend), ('\u{1107f}', '\u{11081}',
- GC_Extend), ('\u{11082}', '\u{11082}', GC_SpacingMark), ('\u{110b0}', '\u{110b2}',
- GC_SpacingMark), ('\u{110b3}', '\u{110b6}', GC_Extend), ('\u{110b7}', '\u{110b8}',
- GC_SpacingMark), ('\u{110b9}', '\u{110ba}', GC_Extend), ('\u{110bd}', '\u{110bd}',
- GC_Prepend), ('\u{110cd}', '\u{110cd}', GC_Prepend), ('\u{11100}', '\u{11102}', GC_Extend),
- ('\u{11127}', '\u{1112b}', GC_Extend), ('\u{1112c}', '\u{1112c}', GC_SpacingMark),
- ('\u{1112d}', '\u{11134}', GC_Extend), ('\u{11145}', '\u{11146}', GC_SpacingMark),
- ('\u{11173}', '\u{11173}', GC_Extend), ('\u{11180}', '\u{11181}', GC_Extend), ('\u{11182}',
- '\u{11182}', GC_SpacingMark), ('\u{111b3}', '\u{111b5}', GC_SpacingMark), ('\u{111b6}',
- '\u{111be}', GC_Extend), ('\u{111bf}', '\u{111c0}', GC_SpacingMark), ('\u{111c2}',
- '\u{111c3}', GC_Prepend), ('\u{111c9}', '\u{111cc}', GC_Extend), ('\u{1122c}', '\u{1122e}',
- GC_SpacingMark), ('\u{1122f}', '\u{11231}', GC_Extend), ('\u{11232}', '\u{11233}',
- GC_SpacingMark), ('\u{11234}', '\u{11234}', GC_Extend), ('\u{11235}', '\u{11235}',
- GC_SpacingMark), ('\u{11236}', '\u{11237}', GC_Extend), ('\u{1123e}', '\u{1123e}',
- GC_Extend), ('\u{112df}', '\u{112df}', GC_Extend), ('\u{112e0}', '\u{112e2}',
- GC_SpacingMark), ('\u{112e3}', '\u{112ea}', GC_Extend), ('\u{11300}', '\u{11301}',
- GC_Extend), ('\u{11302}', '\u{11303}', GC_SpacingMark), ('\u{1133b}', '\u{1133c}',
- GC_Extend), ('\u{1133e}', '\u{1133e}', GC_Extend), ('\u{1133f}', '\u{1133f}',
- GC_SpacingMark), ('\u{11340}', '\u{11340}', GC_Extend), ('\u{11341}', '\u{11344}',
- GC_SpacingMark), ('\u{11347}', '\u{11348}', GC_SpacingMark), ('\u{1134b}', '\u{1134d}',
- GC_SpacingMark), ('\u{11357}', '\u{11357}', GC_Extend), ('\u{11362}', '\u{11363}',
- GC_SpacingMark), ('\u{11366}', '\u{1136c}', GC_Extend), ('\u{11370}', '\u{11374}',
- GC_Extend), ('\u{11435}', '\u{11437}', GC_SpacingMark), ('\u{11438}', '\u{1143f}',
- GC_Extend), ('\u{11440}', '\u{11441}', GC_SpacingMark), ('\u{11442}', '\u{11444}',
- GC_Extend), ('\u{11445}', '\u{11445}', GC_SpacingMark), ('\u{11446}', '\u{11446}',
- GC_Extend), ('\u{1145e}', '\u{1145e}', GC_Extend), ('\u{114b0}', '\u{114b0}', GC_Extend),
- ('\u{114b1}', '\u{114b2}', GC_SpacingMark), ('\u{114b3}', '\u{114b8}', GC_Extend),
- ('\u{114b9}', '\u{114b9}', GC_SpacingMark), ('\u{114ba}', '\u{114ba}', GC_Extend),
- ('\u{114bb}', '\u{114bc}', GC_SpacingMark), ('\u{114bd}', '\u{114bd}', GC_Extend),
- ('\u{114be}', '\u{114be}', GC_SpacingMark), ('\u{114bf}', '\u{114c0}', GC_Extend),
- ('\u{114c1}', '\u{114c1}', GC_SpacingMark), ('\u{114c2}', '\u{114c3}', GC_Extend),
- ('\u{115af}', '\u{115af}', GC_Extend), ('\u{115b0}', '\u{115b1}', GC_SpacingMark),
- ('\u{115b2}', '\u{115b5}', GC_Extend), ('\u{115b8}', '\u{115bb}', GC_SpacingMark),
- ('\u{115bc}', '\u{115bd}', GC_Extend), ('\u{115be}', '\u{115be}', GC_SpacingMark),
- ('\u{115bf}', '\u{115c0}', GC_Extend), ('\u{115dc}', '\u{115dd}', GC_Extend), ('\u{11630}',
- '\u{11632}', GC_SpacingMark), ('\u{11633}', '\u{1163a}', GC_Extend), ('\u{1163b}',
- '\u{1163c}', GC_SpacingMark), ('\u{1163d}', '\u{1163d}', GC_Extend), ('\u{1163e}',
- '\u{1163e}', GC_SpacingMark), ('\u{1163f}', '\u{11640}', GC_Extend), ('\u{116ab}',
- '\u{116ab}', GC_Extend), ('\u{116ac}', '\u{116ac}', GC_SpacingMark), ('\u{116ad}',
- '\u{116ad}', GC_Extend), ('\u{116ae}', '\u{116af}', GC_SpacingMark), ('\u{116b0}',
- '\u{116b5}', GC_Extend), ('\u{116b6}', '\u{116b6}', GC_SpacingMark), ('\u{116b7}',
- '\u{116b7}', GC_Extend), ('\u{1171d}', '\u{1171f}', GC_Extend), ('\u{11720}', '\u{11721}',
- GC_SpacingMark), ('\u{11722}', '\u{11725}', GC_Extend), ('\u{11726}', '\u{11726}',
- GC_SpacingMark), ('\u{11727}', '\u{1172b}', GC_Extend), ('\u{1182c}', '\u{1182e}',
- GC_SpacingMark), ('\u{1182f}', '\u{11837}', GC_Extend), ('\u{11838}', '\u{11838}',
- GC_SpacingMark), ('\u{11839}', '\u{1183a}', GC_Extend), ('\u{119d1}', '\u{119d3}',
+ GC_SpacingMark), ('\u{a82c}', '\u{a82c}', GC_Extend), ('\u{a880}', '\u{a881}',
+ GC_SpacingMark), ('\u{a8b4}', '\u{a8c3}', GC_SpacingMark), ('\u{a8c4}', '\u{a8c5}',
+ GC_Extend), ('\u{a8e0}', '\u{a8f1}', GC_Extend), ('\u{a8ff}', '\u{a8ff}', GC_Extend),
+ ('\u{a926}', '\u{a92d}', GC_Extend), ('\u{a947}', '\u{a951}', GC_Extend), ('\u{a952}',
+ '\u{a953}', GC_SpacingMark), ('\u{a960}', '\u{a97c}', GC_L), ('\u{a980}', '\u{a982}',
+ GC_Extend), ('\u{a983}', '\u{a983}', GC_SpacingMark), ('\u{a9b3}', '\u{a9b3}', GC_Extend),
+ ('\u{a9b4}', '\u{a9b5}', GC_SpacingMark), ('\u{a9b6}', '\u{a9b9}', GC_Extend), ('\u{a9ba}',
+ '\u{a9bb}', GC_SpacingMark), ('\u{a9bc}', '\u{a9bd}', GC_Extend), ('\u{a9be}', '\u{a9c0}',
+ GC_SpacingMark), ('\u{a9e5}', '\u{a9e5}', GC_Extend), ('\u{aa29}', '\u{aa2e}', GC_Extend),
+ ('\u{aa2f}', '\u{aa30}', GC_SpacingMark), ('\u{aa31}', '\u{aa32}', GC_Extend), ('\u{aa33}',
+ '\u{aa34}', GC_SpacingMark), ('\u{aa35}', '\u{aa36}', GC_Extend), ('\u{aa43}', '\u{aa43}',
+ GC_Extend), ('\u{aa4c}', '\u{aa4c}', GC_Extend), ('\u{aa4d}', '\u{aa4d}', GC_SpacingMark),
+ ('\u{aa7c}', '\u{aa7c}', GC_Extend), ('\u{aab0}', '\u{aab0}', GC_Extend), ('\u{aab2}',
+ '\u{aab4}', GC_Extend), ('\u{aab7}', '\u{aab8}', GC_Extend), ('\u{aabe}', '\u{aabf}',
+ GC_Extend), ('\u{aac1}', '\u{aac1}', GC_Extend), ('\u{aaeb}', '\u{aaeb}', GC_SpacingMark),
+ ('\u{aaec}', '\u{aaed}', GC_Extend), ('\u{aaee}', '\u{aaef}', GC_SpacingMark), ('\u{aaf5}',
+ '\u{aaf5}', GC_SpacingMark), ('\u{aaf6}', '\u{aaf6}', GC_Extend), ('\u{abe3}', '\u{abe4}',
+ GC_SpacingMark), ('\u{abe5}', '\u{abe5}', GC_Extend), ('\u{abe6}', '\u{abe7}',
+ GC_SpacingMark), ('\u{abe8}', '\u{abe8}', GC_Extend), ('\u{abe9}', '\u{abea}',
+ GC_SpacingMark), ('\u{abec}', '\u{abec}', GC_SpacingMark), ('\u{abed}', '\u{abed}',
+ GC_Extend), ('\u{ac00}', '\u{ac00}', GC_LV), ('\u{ac01}', '\u{ac1b}', GC_LVT), ('\u{ac1c}',
+ '\u{ac1c}', GC_LV), ('\u{ac1d}', '\u{ac37}', GC_LVT), ('\u{ac38}', '\u{ac38}', GC_LV),
+ ('\u{ac39}', '\u{ac53}', GC_LVT), ('\u{ac54}', '\u{ac54}', GC_LV), ('\u{ac55}', '\u{ac6f}',
+ GC_LVT), ('\u{ac70}', '\u{ac70}', GC_LV), ('\u{ac71}', '\u{ac8b}', GC_LVT), ('\u{ac8c}',
+ '\u{ac8c}', GC_LV), ('\u{ac8d}', '\u{aca7}', GC_LVT), ('\u{aca8}', '\u{aca8}', GC_LV),
+ ('\u{aca9}', '\u{acc3}', GC_LVT), ('\u{acc4}', '\u{acc4}', GC_LV), ('\u{acc5}', '\u{acdf}',
+ GC_LVT), ('\u{ace0}', '\u{ace0}', GC_LV), ('\u{ace1}', '\u{acfb}', GC_LVT), ('\u{acfc}',
+ '\u{acfc}', GC_LV), ('\u{acfd}', '\u{ad17}', GC_LVT), ('\u{ad18}', '\u{ad18}', GC_LV),
+ ('\u{ad19}', '\u{ad33}', GC_LVT), ('\u{ad34}', '\u{ad34}', GC_LV), ('\u{ad35}', '\u{ad4f}',
+ GC_LVT), ('\u{ad50}', '\u{ad50}', GC_LV), ('\u{ad51}', '\u{ad6b}', GC_LVT), ('\u{ad6c}',
+ '\u{ad6c}', GC_LV), ('\u{ad6d}', '\u{ad87}', GC_LVT), ('\u{ad88}', '\u{ad88}', GC_LV),
+ ('\u{ad89}', '\u{ada3}', GC_LVT), ('\u{ada4}', '\u{ada4}', GC_LV), ('\u{ada5}', '\u{adbf}',
+ GC_LVT), ('\u{adc0}', '\u{adc0}', GC_LV), ('\u{adc1}', '\u{addb}', GC_LVT), ('\u{addc}',
+ '\u{addc}', GC_LV), ('\u{addd}', '\u{adf7}', GC_LVT), ('\u{adf8}', '\u{adf8}', GC_LV),
+ ('\u{adf9}', '\u{ae13}', GC_LVT), ('\u{ae14}', '\u{ae14}', GC_LV), ('\u{ae15}', '\u{ae2f}',
+ GC_LVT), ('\u{ae30}', '\u{ae30}', GC_LV), ('\u{ae31}', '\u{ae4b}', GC_LVT), ('\u{ae4c}',
+ '\u{ae4c}', GC_LV), ('\u{ae4d}', '\u{ae67}', GC_LVT), ('\u{ae68}', '\u{ae68}', GC_LV),
+ ('\u{ae69}', '\u{ae83}', GC_LVT), ('\u{ae84}', '\u{ae84}', GC_LV), ('\u{ae85}', '\u{ae9f}',
+ GC_LVT), ('\u{aea0}', '\u{aea0}', GC_LV), ('\u{aea1}', '\u{aebb}', GC_LVT), ('\u{aebc}',
+ '\u{aebc}', GC_LV), ('\u{aebd}', '\u{aed7}', GC_LVT), ('\u{aed8}', '\u{aed8}', GC_LV),
+ ('\u{aed9}', '\u{aef3}', GC_LVT), ('\u{aef4}', '\u{aef4}', GC_LV), ('\u{aef5}', '\u{af0f}',
+ GC_LVT), ('\u{af10}', '\u{af10}', GC_LV), ('\u{af11}', '\u{af2b}', GC_LVT), ('\u{af2c}',
+ '\u{af2c}', GC_LV), ('\u{af2d}', '\u{af47}', GC_LVT), ('\u{af48}', '\u{af48}', GC_LV),
+ ('\u{af49}', '\u{af63}', GC_LVT), ('\u{af64}', '\u{af64}', GC_LV), ('\u{af65}', '\u{af7f}',
+ GC_LVT), ('\u{af80}', '\u{af80}', GC_LV), ('\u{af81}', '\u{af9b}', GC_LVT), ('\u{af9c}',
+ '\u{af9c}', GC_LV), ('\u{af9d}', '\u{afb7}', GC_LVT), ('\u{afb8}', '\u{afb8}', GC_LV),
+ ('\u{afb9}', '\u{afd3}', GC_LVT), ('\u{afd4}', '\u{afd4}', GC_LV), ('\u{afd5}', '\u{afef}',
+ GC_LVT), ('\u{aff0}', '\u{aff0}', GC_LV), ('\u{aff1}', '\u{b00b}', GC_LVT), ('\u{b00c}',
+ '\u{b00c}', GC_LV), ('\u{b00d}', '\u{b027}', GC_LVT), ('\u{b028}', '\u{b028}', GC_LV),
+ ('\u{b029}', '\u{b043}', GC_LVT), ('\u{b044}', '\u{b044}', GC_LV), ('\u{b045}', '\u{b05f}',
+ GC_LVT), ('\u{b060}', '\u{b060}', GC_LV), ('\u{b061}', '\u{b07b}', GC_LVT), ('\u{b07c}',
+ '\u{b07c}', GC_LV), ('\u{b07d}', '\u{b097}', GC_LVT), ('\u{b098}', '\u{b098}', GC_LV),
+ ('\u{b099}', '\u{b0b3}', GC_LVT), ('\u{b0b4}', '\u{b0b4}', GC_LV), ('\u{b0b5}', '\u{b0cf}',
+ GC_LVT), ('\u{b0d0}', '\u{b0d0}', GC_LV), ('\u{b0d1}', '\u{b0eb}', GC_LVT), ('\u{b0ec}',
+ '\u{b0ec}', GC_LV), ('\u{b0ed}', '\u{b107}', GC_LVT), ('\u{b108}', '\u{b108}', GC_LV),
+ ('\u{b109}', '\u{b123}', GC_LVT), ('\u{b124}', '\u{b124}', GC_LV), ('\u{b125}', '\u{b13f}',
+ GC_LVT), ('\u{b140}', '\u{b140}', GC_LV), ('\u{b141}', '\u{b15b}', GC_LVT), ('\u{b15c}',
+ '\u{b15c}', GC_LV), ('\u{b15d}', '\u{b177}', GC_LVT), ('\u{b178}', '\u{b178}', GC_LV),
+ ('\u{b179}', '\u{b193}', GC_LVT), ('\u{b194}', '\u{b194}', GC_LV), ('\u{b195}', '\u{b1af}',
+ GC_LVT), ('\u{b1b0}', '\u{b1b0}', GC_LV), ('\u{b1b1}', '\u{b1cb}', GC_LVT), ('\u{b1cc}',
+ '\u{b1cc}', GC_LV), ('\u{b1cd}', '\u{b1e7}', GC_LVT), ('\u{b1e8}', '\u{b1e8}', GC_LV),
+ ('\u{b1e9}', '\u{b203}', GC_LVT), ('\u{b204}', '\u{b204}', GC_LV), ('\u{b205}', '\u{b21f}',
+ GC_LVT), ('\u{b220}', '\u{b220}', GC_LV), ('\u{b221}', '\u{b23b}', GC_LVT), ('\u{b23c}',
+ '\u{b23c}', GC_LV), ('\u{b23d}', '\u{b257}', GC_LVT), ('\u{b258}', '\u{b258}', GC_LV),
+ ('\u{b259}', '\u{b273}', GC_LVT), ('\u{b274}', '\u{b274}', GC_LV), ('\u{b275}', '\u{b28f}',
+ GC_LVT), ('\u{b290}', '\u{b290}', GC_LV), ('\u{b291}', '\u{b2ab}', GC_LVT), ('\u{b2ac}',
+ '\u{b2ac}', GC_LV), ('\u{b2ad}', '\u{b2c7}', GC_LVT), ('\u{b2c8}', '\u{b2c8}', GC_LV),
+ ('\u{b2c9}', '\u{b2e3}', GC_LVT), ('\u{b2e4}', '\u{b2e4}', GC_LV), ('\u{b2e5}', '\u{b2ff}',
+ GC_LVT), ('\u{b300}', '\u{b300}', GC_LV), ('\u{b301}', '\u{b31b}', GC_LVT), ('\u{b31c}',
+ '\u{b31c}', GC_LV), ('\u{b31d}', '\u{b337}', GC_LVT), ('\u{b338}', '\u{b338}', GC_LV),
+ ('\u{b339}', '\u{b353}', GC_LVT), ('\u{b354}', '\u{b354}', GC_LV), ('\u{b355}', '\u{b36f}',
+ GC_LVT), ('\u{b370}', '\u{b370}', GC_LV), ('\u{b371}', '\u{b38b}', GC_LVT), ('\u{b38c}',
+ '\u{b38c}', GC_LV), ('\u{b38d}', '\u{b3a7}', GC_LVT), ('\u{b3a8}', '\u{b3a8}', GC_LV),
+ ('\u{b3a9}', '\u{b3c3}', GC_LVT), ('\u{b3c4}', '\u{b3c4}', GC_LV), ('\u{b3c5}', '\u{b3df}',
+ GC_LVT), ('\u{b3e0}', '\u{b3e0}', GC_LV), ('\u{b3e1}', '\u{b3fb}', GC_LVT), ('\u{b3fc}',
+ '\u{b3fc}', GC_LV), ('\u{b3fd}', '\u{b417}', GC_LVT), ('\u{b418}', '\u{b418}', GC_LV),
+ ('\u{b419}', '\u{b433}', GC_LVT), ('\u{b434}', '\u{b434}', GC_LV), ('\u{b435}', '\u{b44f}',
+ GC_LVT), ('\u{b450}', '\u{b450}', GC_LV), ('\u{b451}', '\u{b46b}', GC_LVT), ('\u{b46c}',
+ '\u{b46c}', GC_LV), ('\u{b46d}', '\u{b487}', GC_LVT), ('\u{b488}', '\u{b488}', GC_LV),
+ ('\u{b489}', '\u{b4a3}', GC_LVT), ('\u{b4a4}', '\u{b4a4}', GC_LV), ('\u{b4a5}', '\u{b4bf}',
+ GC_LVT), ('\u{b4c0}', '\u{b4c0}', GC_LV), ('\u{b4c1}', '\u{b4db}', GC_LVT), ('\u{b4dc}',
+ '\u{b4dc}', GC_LV), ('\u{b4dd}', '\u{b4f7}', GC_LVT), ('\u{b4f8}', '\u{b4f8}', GC_LV),
+ ('\u{b4f9}', '\u{b513}', GC_LVT), ('\u{b514}', '\u{b514}', GC_LV), ('\u{b515}', '\u{b52f}',
+ GC_LVT), ('\u{b530}', '\u{b530}', GC_LV), ('\u{b531}', '\u{b54b}', GC_LVT), ('\u{b54c}',
+ '\u{b54c}', GC_LV), ('\u{b54d}', '\u{b567}', GC_LVT), ('\u{b568}', '\u{b568}', GC_LV),
+ ('\u{b569}', '\u{b583}', GC_LVT), ('\u{b584}', '\u{b584}', GC_LV), ('\u{b585}', '\u{b59f}',
+ GC_LVT), ('\u{b5a0}', '\u{b5a0}', GC_LV), ('\u{b5a1}', '\u{b5bb}', GC_LVT), ('\u{b5bc}',
+ '\u{b5bc}', GC_LV), ('\u{b5bd}', '\u{b5d7}', GC_LVT), ('\u{b5d8}', '\u{b5d8}', GC_LV),
+ ('\u{b5d9}', '\u{b5f3}', GC_LVT), ('\u{b5f4}', '\u{b5f4}', GC_LV), ('\u{b5f5}', '\u{b60f}',
+ GC_LVT), ('\u{b610}', '\u{b610}', GC_LV), ('\u{b611}', '\u{b62b}', GC_LVT), ('\u{b62c}',
+ '\u{b62c}', GC_LV), ('\u{b62d}', '\u{b647}', GC_LVT), ('\u{b648}', '\u{b648}', GC_LV),
+ ('\u{b649}', '\u{b663}', GC_LVT), ('\u{b664}', '\u{b664}', GC_LV), ('\u{b665}', '\u{b67f}',
+ GC_LVT), ('\u{b680}', '\u{b680}', GC_LV), ('\u{b681}', '\u{b69b}', GC_LVT), ('\u{b69c}',
+ '\u{b69c}', GC_LV), ('\u{b69d}', '\u{b6b7}', GC_LVT), ('\u{b6b8}', '\u{b6b8}', GC_LV),
+ ('\u{b6b9}', '\u{b6d3}', GC_LVT), ('\u{b6d4}', '\u{b6d4}', GC_LV), ('\u{b6d5}', '\u{b6ef}',
+ GC_LVT), ('\u{b6f0}', '\u{b6f0}', GC_LV), ('\u{b6f1}', '\u{b70b}', GC_LVT), ('\u{b70c}',
+ '\u{b70c}', GC_LV), ('\u{b70d}', '\u{b727}', GC_LVT), ('\u{b728}', '\u{b728}', GC_LV),
+ ('\u{b729}', '\u{b743}', GC_LVT), ('\u{b744}', '\u{b744}', GC_LV), ('\u{b745}', '\u{b75f}',
+ GC_LVT), ('\u{b760}', '\u{b760}', GC_LV), ('\u{b761}', '\u{b77b}', GC_LVT), ('\u{b77c}',
+ '\u{b77c}', GC_LV), ('\u{b77d}', '\u{b797}', GC_LVT), ('\u{b798}', '\u{b798}', GC_LV),
+ ('\u{b799}', '\u{b7b3}', GC_LVT), ('\u{b7b4}', '\u{b7b4}', GC_LV), ('\u{b7b5}', '\u{b7cf}',
+ GC_LVT), ('\u{b7d0}', '\u{b7d0}', GC_LV), ('\u{b7d1}', '\u{b7eb}', GC_LVT), ('\u{b7ec}',
+ '\u{b7ec}', GC_LV), ('\u{b7ed}', '\u{b807}', GC_LVT), ('\u{b808}', '\u{b808}', GC_LV),
+ ('\u{b809}', '\u{b823}', GC_LVT), ('\u{b824}', '\u{b824}', GC_LV), ('\u{b825}', '\u{b83f}',
+ GC_LVT), ('\u{b840}', '\u{b840}', GC_LV), ('\u{b841}', '\u{b85b}', GC_LVT), ('\u{b85c}',
+ '\u{b85c}', GC_LV), ('\u{b85d}', '\u{b877}', GC_LVT), ('\u{b878}', '\u{b878}', GC_LV),
+ ('\u{b879}', '\u{b893}', GC_LVT), ('\u{b894}', '\u{b894}', GC_LV), ('\u{b895}', '\u{b8af}',
+ GC_LVT), ('\u{b8b0}', '\u{b8b0}', GC_LV), ('\u{b8b1}', '\u{b8cb}', GC_LVT), ('\u{b8cc}',
+ '\u{b8cc}', GC_LV), ('\u{b8cd}', '\u{b8e7}', GC_LVT), ('\u{b8e8}', '\u{b8e8}', GC_LV),
+ ('\u{b8e9}', '\u{b903}', GC_LVT), ('\u{b904}', '\u{b904}', GC_LV), ('\u{b905}', '\u{b91f}',
+ GC_LVT), ('\u{b920}', '\u{b920}', GC_LV), ('\u{b921}', '\u{b93b}', GC_LVT), ('\u{b93c}',
+ '\u{b93c}', GC_LV), ('\u{b93d}', '\u{b957}', GC_LVT), ('\u{b958}', '\u{b958}', GC_LV),
+ ('\u{b959}', '\u{b973}', GC_LVT), ('\u{b974}', '\u{b974}', GC_LV), ('\u{b975}', '\u{b98f}',
+ GC_LVT), ('\u{b990}', '\u{b990}', GC_LV), ('\u{b991}', '\u{b9ab}', GC_LVT), ('\u{b9ac}',
+ '\u{b9ac}', GC_LV), ('\u{b9ad}', '\u{b9c7}', GC_LVT), ('\u{b9c8}', '\u{b9c8}', GC_LV),
+ ('\u{b9c9}', '\u{b9e3}', GC_LVT), ('\u{b9e4}', '\u{b9e4}', GC_LV), ('\u{b9e5}', '\u{b9ff}',
+ GC_LVT), ('\u{ba00}', '\u{ba00}', GC_LV), ('\u{ba01}', '\u{ba1b}', GC_LVT), ('\u{ba1c}',
+ '\u{ba1c}', GC_LV), ('\u{ba1d}', '\u{ba37}', GC_LVT), ('\u{ba38}', '\u{ba38}', GC_LV),
+ ('\u{ba39}', '\u{ba53}', GC_LVT), ('\u{ba54}', '\u{ba54}', GC_LV), ('\u{ba55}', '\u{ba6f}',
+ GC_LVT), ('\u{ba70}', '\u{ba70}', GC_LV), ('\u{ba71}', '\u{ba8b}', GC_LVT), ('\u{ba8c}',
+ '\u{ba8c}', GC_LV), ('\u{ba8d}', '\u{baa7}', GC_LVT), ('\u{baa8}', '\u{baa8}', GC_LV),
+ ('\u{baa9}', '\u{bac3}', GC_LVT), ('\u{bac4}', '\u{bac4}', GC_LV), ('\u{bac5}', '\u{badf}',
+ GC_LVT), ('\u{bae0}', '\u{bae0}', GC_LV), ('\u{bae1}', '\u{bafb}', GC_LVT), ('\u{bafc}',
+ '\u{bafc}', GC_LV), ('\u{bafd}', '\u{bb17}', GC_LVT), ('\u{bb18}', '\u{bb18}', GC_LV),
+ ('\u{bb19}', '\u{bb33}', GC_LVT), ('\u{bb34}', '\u{bb34}', GC_LV), ('\u{bb35}', '\u{bb4f}',
+ GC_LVT), ('\u{bb50}', '\u{bb50}', GC_LV), ('\u{bb51}', '\u{bb6b}', GC_LVT), ('\u{bb6c}',
+ '\u{bb6c}', GC_LV), ('\u{bb6d}', '\u{bb87}', GC_LVT), ('\u{bb88}', '\u{bb88}', GC_LV),
+ ('\u{bb89}', '\u{bba3}', GC_LVT), ('\u{bba4}', '\u{bba4}', GC_LV), ('\u{bba5}', '\u{bbbf}',
+ GC_LVT), ('\u{bbc0}', '\u{bbc0}', GC_LV), ('\u{bbc1}', '\u{bbdb}', GC_LVT), ('\u{bbdc}',
+ '\u{bbdc}', GC_LV), ('\u{bbdd}', '\u{bbf7}', GC_LVT), ('\u{bbf8}', '\u{bbf8}', GC_LV),
+ ('\u{bbf9}', '\u{bc13}', GC_LVT), ('\u{bc14}', '\u{bc14}', GC_LV), ('\u{bc15}', '\u{bc2f}',
+ GC_LVT), ('\u{bc30}', '\u{bc30}', GC_LV), ('\u{bc31}', '\u{bc4b}', GC_LVT), ('\u{bc4c}',
+ '\u{bc4c}', GC_LV), ('\u{bc4d}', '\u{bc67}', GC_LVT), ('\u{bc68}', '\u{bc68}', GC_LV),
+ ('\u{bc69}', '\u{bc83}', GC_LVT), ('\u{bc84}', '\u{bc84}', GC_LV), ('\u{bc85}', '\u{bc9f}',
+ GC_LVT), ('\u{bca0}', '\u{bca0}', GC_LV), ('\u{bca1}', '\u{bcbb}', GC_LVT), ('\u{bcbc}',
+ '\u{bcbc}', GC_LV), ('\u{bcbd}', '\u{bcd7}', GC_LVT), ('\u{bcd8}', '\u{bcd8}', GC_LV),
+ ('\u{bcd9}', '\u{bcf3}', GC_LVT), ('\u{bcf4}', '\u{bcf4}', GC_LV), ('\u{bcf5}', '\u{bd0f}',
+ GC_LVT), ('\u{bd10}', '\u{bd10}', GC_LV), ('\u{bd11}', '\u{bd2b}', GC_LVT), ('\u{bd2c}',
+ '\u{bd2c}', GC_LV), ('\u{bd2d}', '\u{bd47}', GC_LVT), ('\u{bd48}', '\u{bd48}', GC_LV),
+ ('\u{bd49}', '\u{bd63}', GC_LVT), ('\u{bd64}', '\u{bd64}', GC_LV), ('\u{bd65}', '\u{bd7f}',
+ GC_LVT), ('\u{bd80}', '\u{bd80}', GC_LV), ('\u{bd81}', '\u{bd9b}', GC_LVT), ('\u{bd9c}',
+ '\u{bd9c}', GC_LV), ('\u{bd9d}', '\u{bdb7}', GC_LVT), ('\u{bdb8}', '\u{bdb8}', GC_LV),
+ ('\u{bdb9}', '\u{bdd3}', GC_LVT), ('\u{bdd4}', '\u{bdd4}', GC_LV), ('\u{bdd5}', '\u{bdef}',
+ GC_LVT), ('\u{bdf0}', '\u{bdf0}', GC_LV), ('\u{bdf1}', '\u{be0b}', GC_LVT), ('\u{be0c}',
+ '\u{be0c}', GC_LV), ('\u{be0d}', '\u{be27}', GC_LVT), ('\u{be28}', '\u{be28}', GC_LV),
+ ('\u{be29}', '\u{be43}', GC_LVT), ('\u{be44}', '\u{be44}', GC_LV), ('\u{be45}', '\u{be5f}',
+ GC_LVT), ('\u{be60}', '\u{be60}', GC_LV), ('\u{be61}', '\u{be7b}', GC_LVT), ('\u{be7c}',
+ '\u{be7c}', GC_LV), ('\u{be7d}', '\u{be97}', GC_LVT), ('\u{be98}', '\u{be98}', GC_LV),
+ ('\u{be99}', '\u{beb3}', GC_LVT), ('\u{beb4}', '\u{beb4}', GC_LV), ('\u{beb5}', '\u{becf}',
+ GC_LVT), ('\u{bed0}', '\u{bed0}', GC_LV), ('\u{bed1}', '\u{beeb}', GC_LVT), ('\u{beec}',
+ '\u{beec}', GC_LV), ('\u{beed}', '\u{bf07}', GC_LVT), ('\u{bf08}', '\u{bf08}', GC_LV),
+ ('\u{bf09}', '\u{bf23}', GC_LVT), ('\u{bf24}', '\u{bf24}', GC_LV), ('\u{bf25}', '\u{bf3f}',
+ GC_LVT), ('\u{bf40}', '\u{bf40}', GC_LV), ('\u{bf41}', '\u{bf5b}', GC_LVT), ('\u{bf5c}',
+ '\u{bf5c}', GC_LV), ('\u{bf5d}', '\u{bf77}', GC_LVT), ('\u{bf78}', '\u{bf78}', GC_LV),
+ ('\u{bf79}', '\u{bf93}', GC_LVT), ('\u{bf94}', '\u{bf94}', GC_LV), ('\u{bf95}', '\u{bfaf}',
+ GC_LVT), ('\u{bfb0}', '\u{bfb0}', GC_LV), ('\u{bfb1}', '\u{bfcb}', GC_LVT), ('\u{bfcc}',
+ '\u{bfcc}', GC_LV), ('\u{bfcd}', '\u{bfe7}', GC_LVT), ('\u{bfe8}', '\u{bfe8}', GC_LV),
+ ('\u{bfe9}', '\u{c003}', GC_LVT), ('\u{c004}', '\u{c004}', GC_LV), ('\u{c005}', '\u{c01f}',
+ GC_LVT), ('\u{c020}', '\u{c020}', GC_LV), ('\u{c021}', '\u{c03b}', GC_LVT), ('\u{c03c}',
+ '\u{c03c}', GC_LV), ('\u{c03d}', '\u{c057}', GC_LVT), ('\u{c058}', '\u{c058}', GC_LV),
+ ('\u{c059}', '\u{c073}', GC_LVT), ('\u{c074}', '\u{c074}', GC_LV), ('\u{c075}', '\u{c08f}',
+ GC_LVT), ('\u{c090}', '\u{c090}', GC_LV), ('\u{c091}', '\u{c0ab}', GC_LVT), ('\u{c0ac}',
+ '\u{c0ac}', GC_LV), ('\u{c0ad}', '\u{c0c7}', GC_LVT), ('\u{c0c8}', '\u{c0c8}', GC_LV),
+ ('\u{c0c9}', '\u{c0e3}', GC_LVT), ('\u{c0e4}', '\u{c0e4}', GC_LV), ('\u{c0e5}', '\u{c0ff}',
+ GC_LVT), ('\u{c100}', '\u{c100}', GC_LV), ('\u{c101}', '\u{c11b}', GC_LVT), ('\u{c11c}',
+ '\u{c11c}', GC_LV), ('\u{c11d}', '\u{c137}', GC_LVT), ('\u{c138}', '\u{c138}', GC_LV),
+ ('\u{c139}', '\u{c153}', GC_LVT), ('\u{c154}', '\u{c154}', GC_LV), ('\u{c155}', '\u{c16f}',
+ GC_LVT), ('\u{c170}', '\u{c170}', GC_LV), ('\u{c171}', '\u{c18b}', GC_LVT), ('\u{c18c}',
+ '\u{c18c}', GC_LV), ('\u{c18d}', '\u{c1a7}', GC_LVT), ('\u{c1a8}', '\u{c1a8}', GC_LV),
+ ('\u{c1a9}', '\u{c1c3}', GC_LVT), ('\u{c1c4}', '\u{c1c4}', GC_LV), ('\u{c1c5}', '\u{c1df}',
+ GC_LVT), ('\u{c1e0}', '\u{c1e0}', GC_LV), ('\u{c1e1}', '\u{c1fb}', GC_LVT), ('\u{c1fc}',
+ '\u{c1fc}', GC_LV), ('\u{c1fd}', '\u{c217}', GC_LVT), ('\u{c218}', '\u{c218}', GC_LV),
+ ('\u{c219}', '\u{c233}', GC_LVT), ('\u{c234}', '\u{c234}', GC_LV), ('\u{c235}', '\u{c24f}',
+ GC_LVT), ('\u{c250}', '\u{c250}', GC_LV), ('\u{c251}', '\u{c26b}', GC_LVT), ('\u{c26c}',
+ '\u{c26c}', GC_LV), ('\u{c26d}', '\u{c287}', GC_LVT), ('\u{c288}', '\u{c288}', GC_LV),
+ ('\u{c289}', '\u{c2a3}', GC_LVT), ('\u{c2a4}', '\u{c2a4}', GC_LV), ('\u{c2a5}', '\u{c2bf}',
+ GC_LVT), ('\u{c2c0}', '\u{c2c0}', GC_LV), ('\u{c2c1}', '\u{c2db}', GC_LVT), ('\u{c2dc}',
+ '\u{c2dc}', GC_LV), ('\u{c2dd}', '\u{c2f7}', GC_LVT), ('\u{c2f8}', '\u{c2f8}', GC_LV),
+ ('\u{c2f9}', '\u{c313}', GC_LVT), ('\u{c314}', '\u{c314}', GC_LV), ('\u{c315}', '\u{c32f}',
+ GC_LVT), ('\u{c330}', '\u{c330}', GC_LV), ('\u{c331}', '\u{c34b}', GC_LVT), ('\u{c34c}',
+ '\u{c34c}', GC_LV), ('\u{c34d}', '\u{c367}', GC_LVT), ('\u{c368}', '\u{c368}', GC_LV),
+ ('\u{c369}', '\u{c383}', GC_LVT), ('\u{c384}', '\u{c384}', GC_LV), ('\u{c385}', '\u{c39f}',
+ GC_LVT), ('\u{c3a0}', '\u{c3a0}', GC_LV), ('\u{c3a1}', '\u{c3bb}', GC_LVT), ('\u{c3bc}',
+ '\u{c3bc}', GC_LV), ('\u{c3bd}', '\u{c3d7}', GC_LVT), ('\u{c3d8}', '\u{c3d8}', GC_LV),
+ ('\u{c3d9}', '\u{c3f3}', GC_LVT), ('\u{c3f4}', '\u{c3f4}', GC_LV), ('\u{c3f5}', '\u{c40f}',
+ GC_LVT), ('\u{c410}', '\u{c410}', GC_LV), ('\u{c411}', '\u{c42b}', GC_LVT), ('\u{c42c}',
+ '\u{c42c}', GC_LV), ('\u{c42d}', '\u{c447}', GC_LVT), ('\u{c448}', '\u{c448}', GC_LV),
+ ('\u{c449}', '\u{c463}', GC_LVT), ('\u{c464}', '\u{c464}', GC_LV), ('\u{c465}', '\u{c47f}',
+ GC_LVT), ('\u{c480}', '\u{c480}', GC_LV), ('\u{c481}', '\u{c49b}', GC_LVT), ('\u{c49c}',
+ '\u{c49c}', GC_LV), ('\u{c49d}', '\u{c4b7}', GC_LVT), ('\u{c4b8}', '\u{c4b8}', GC_LV),
+ ('\u{c4b9}', '\u{c4d3}', GC_LVT), ('\u{c4d4}', '\u{c4d4}', GC_LV), ('\u{c4d5}', '\u{c4ef}',
+ GC_LVT), ('\u{c4f0}', '\u{c4f0}', GC_LV), ('\u{c4f1}', '\u{c50b}', GC_LVT), ('\u{c50c}',
+ '\u{c50c}', GC_LV), ('\u{c50d}', '\u{c527}', GC_LVT), ('\u{c528}', '\u{c528}', GC_LV),
+ ('\u{c529}', '\u{c543}', GC_LVT), ('\u{c544}', '\u{c544}', GC_LV), ('\u{c545}', '\u{c55f}',
+ GC_LVT), ('\u{c560}', '\u{c560}', GC_LV), ('\u{c561}', '\u{c57b}', GC_LVT), ('\u{c57c}',
+ '\u{c57c}', GC_LV), ('\u{c57d}', '\u{c597}', GC_LVT), ('\u{c598}', '\u{c598}', GC_LV),
+ ('\u{c599}', '\u{c5b3}', GC_LVT), ('\u{c5b4}', '\u{c5b4}', GC_LV), ('\u{c5b5}', '\u{c5cf}',
+ GC_LVT), ('\u{c5d0}', '\u{c5d0}', GC_LV), ('\u{c5d1}', '\u{c5eb}', GC_LVT), ('\u{c5ec}',
+ '\u{c5ec}', GC_LV), ('\u{c5ed}', '\u{c607}', GC_LVT), ('\u{c608}', '\u{c608}', GC_LV),
+ ('\u{c609}', '\u{c623}', GC_LVT), ('\u{c624}', '\u{c624}', GC_LV), ('\u{c625}', '\u{c63f}',
+ GC_LVT), ('\u{c640}', '\u{c640}', GC_LV), ('\u{c641}', '\u{c65b}', GC_LVT), ('\u{c65c}',
+ '\u{c65c}', GC_LV), ('\u{c65d}', '\u{c677}', GC_LVT), ('\u{c678}', '\u{c678}', GC_LV),
+ ('\u{c679}', '\u{c693}', GC_LVT), ('\u{c694}', '\u{c694}', GC_LV), ('\u{c695}', '\u{c6af}',
+ GC_LVT), ('\u{c6b0}', '\u{c6b0}', GC_LV), ('\u{c6b1}', '\u{c6cb}', GC_LVT), ('\u{c6cc}',
+ '\u{c6cc}', GC_LV), ('\u{c6cd}', '\u{c6e7}', GC_LVT), ('\u{c6e8}', '\u{c6e8}', GC_LV),
+ ('\u{c6e9}', '\u{c703}', GC_LVT), ('\u{c704}', '\u{c704}', GC_LV), ('\u{c705}', '\u{c71f}',
+ GC_LVT), ('\u{c720}', '\u{c720}', GC_LV), ('\u{c721}', '\u{c73b}', GC_LVT), ('\u{c73c}',
+ '\u{c73c}', GC_LV), ('\u{c73d}', '\u{c757}', GC_LVT), ('\u{c758}', '\u{c758}', GC_LV),
+ ('\u{c759}', '\u{c773}', GC_LVT), ('\u{c774}', '\u{c774}', GC_LV), ('\u{c775}', '\u{c78f}',
+ GC_LVT), ('\u{c790}', '\u{c790}', GC_LV), ('\u{c791}', '\u{c7ab}', GC_LVT), ('\u{c7ac}',
+ '\u{c7ac}', GC_LV), ('\u{c7ad}', '\u{c7c7}', GC_LVT), ('\u{c7c8}', '\u{c7c8}', GC_LV),
+ ('\u{c7c9}', '\u{c7e3}', GC_LVT), ('\u{c7e4}', '\u{c7e4}', GC_LV), ('\u{c7e5}', '\u{c7ff}',
+ GC_LVT), ('\u{c800}', '\u{c800}', GC_LV), ('\u{c801}', '\u{c81b}', GC_LVT), ('\u{c81c}',
+ '\u{c81c}', GC_LV), ('\u{c81d}', '\u{c837}', GC_LVT), ('\u{c838}', '\u{c838}', GC_LV),
+ ('\u{c839}', '\u{c853}', GC_LVT), ('\u{c854}', '\u{c854}', GC_LV), ('\u{c855}', '\u{c86f}',
+ GC_LVT), ('\u{c870}', '\u{c870}', GC_LV), ('\u{c871}', '\u{c88b}', GC_LVT), ('\u{c88c}',
+ '\u{c88c}', GC_LV), ('\u{c88d}', '\u{c8a7}', GC_LVT), ('\u{c8a8}', '\u{c8a8}', GC_LV),
+ ('\u{c8a9}', '\u{c8c3}', GC_LVT), ('\u{c8c4}', '\u{c8c4}', GC_LV), ('\u{c8c5}', '\u{c8df}',
+ GC_LVT), ('\u{c8e0}', '\u{c8e0}', GC_LV), ('\u{c8e1}', '\u{c8fb}', GC_LVT), ('\u{c8fc}',
+ '\u{c8fc}', GC_LV), ('\u{c8fd}', '\u{c917}', GC_LVT), ('\u{c918}', '\u{c918}', GC_LV),
+ ('\u{c919}', '\u{c933}', GC_LVT), ('\u{c934}', '\u{c934}', GC_LV), ('\u{c935}', '\u{c94f}',
+ GC_LVT), ('\u{c950}', '\u{c950}', GC_LV), ('\u{c951}', '\u{c96b}', GC_LVT), ('\u{c96c}',
+ '\u{c96c}', GC_LV), ('\u{c96d}', '\u{c987}', GC_LVT), ('\u{c988}', '\u{c988}', GC_LV),
+ ('\u{c989}', '\u{c9a3}', GC_LVT), ('\u{c9a4}', '\u{c9a4}', GC_LV), ('\u{c9a5}', '\u{c9bf}',
+ GC_LVT), ('\u{c9c0}', '\u{c9c0}', GC_LV), ('\u{c9c1}', '\u{c9db}', GC_LVT), ('\u{c9dc}',
+ '\u{c9dc}', GC_LV), ('\u{c9dd}', '\u{c9f7}', GC_LVT), ('\u{c9f8}', '\u{c9f8}', GC_LV),
+ ('\u{c9f9}', '\u{ca13}', GC_LVT), ('\u{ca14}', '\u{ca14}', GC_LV), ('\u{ca15}', '\u{ca2f}',
+ GC_LVT), ('\u{ca30}', '\u{ca30}', GC_LV), ('\u{ca31}', '\u{ca4b}', GC_LVT), ('\u{ca4c}',
+ '\u{ca4c}', GC_LV), ('\u{ca4d}', '\u{ca67}', GC_LVT), ('\u{ca68}', '\u{ca68}', GC_LV),
+ ('\u{ca69}', '\u{ca83}', GC_LVT), ('\u{ca84}', '\u{ca84}', GC_LV), ('\u{ca85}', '\u{ca9f}',
+ GC_LVT), ('\u{caa0}', '\u{caa0}', GC_LV), ('\u{caa1}', '\u{cabb}', GC_LVT), ('\u{cabc}',
+ '\u{cabc}', GC_LV), ('\u{cabd}', '\u{cad7}', GC_LVT), ('\u{cad8}', '\u{cad8}', GC_LV),
+ ('\u{cad9}', '\u{caf3}', GC_LVT), ('\u{caf4}', '\u{caf4}', GC_LV), ('\u{caf5}', '\u{cb0f}',
+ GC_LVT), ('\u{cb10}', '\u{cb10}', GC_LV), ('\u{cb11}', '\u{cb2b}', GC_LVT), ('\u{cb2c}',
+ '\u{cb2c}', GC_LV), ('\u{cb2d}', '\u{cb47}', GC_LVT), ('\u{cb48}', '\u{cb48}', GC_LV),
+ ('\u{cb49}', '\u{cb63}', GC_LVT), ('\u{cb64}', '\u{cb64}', GC_LV), ('\u{cb65}', '\u{cb7f}',
+ GC_LVT), ('\u{cb80}', '\u{cb80}', GC_LV), ('\u{cb81}', '\u{cb9b}', GC_LVT), ('\u{cb9c}',
+ '\u{cb9c}', GC_LV), ('\u{cb9d}', '\u{cbb7}', GC_LVT), ('\u{cbb8}', '\u{cbb8}', GC_LV),
+ ('\u{cbb9}', '\u{cbd3}', GC_LVT), ('\u{cbd4}', '\u{cbd4}', GC_LV), ('\u{cbd5}', '\u{cbef}',
+ GC_LVT), ('\u{cbf0}', '\u{cbf0}', GC_LV), ('\u{cbf1}', '\u{cc0b}', GC_LVT), ('\u{cc0c}',
+ '\u{cc0c}', GC_LV), ('\u{cc0d}', '\u{cc27}', GC_LVT), ('\u{cc28}', '\u{cc28}', GC_LV),
+ ('\u{cc29}', '\u{cc43}', GC_LVT), ('\u{cc44}', '\u{cc44}', GC_LV), ('\u{cc45}', '\u{cc5f}',
+ GC_LVT), ('\u{cc60}', '\u{cc60}', GC_LV), ('\u{cc61}', '\u{cc7b}', GC_LVT), ('\u{cc7c}',
+ '\u{cc7c}', GC_LV), ('\u{cc7d}', '\u{cc97}', GC_LVT), ('\u{cc98}', '\u{cc98}', GC_LV),
+ ('\u{cc99}', '\u{ccb3}', GC_LVT), ('\u{ccb4}', '\u{ccb4}', GC_LV), ('\u{ccb5}', '\u{cccf}',
+ GC_LVT), ('\u{ccd0}', '\u{ccd0}', GC_LV), ('\u{ccd1}', '\u{cceb}', GC_LVT), ('\u{ccec}',
+ '\u{ccec}', GC_LV), ('\u{cced}', '\u{cd07}', GC_LVT), ('\u{cd08}', '\u{cd08}', GC_LV),
+ ('\u{cd09}', '\u{cd23}', GC_LVT), ('\u{cd24}', '\u{cd24}', GC_LV), ('\u{cd25}', '\u{cd3f}',
+ GC_LVT), ('\u{cd40}', '\u{cd40}', GC_LV), ('\u{cd41}', '\u{cd5b}', GC_LVT), ('\u{cd5c}',
+ '\u{cd5c}', GC_LV), ('\u{cd5d}', '\u{cd77}', GC_LVT), ('\u{cd78}', '\u{cd78}', GC_LV),
+ ('\u{cd79}', '\u{cd93}', GC_LVT), ('\u{cd94}', '\u{cd94}', GC_LV), ('\u{cd95}', '\u{cdaf}',
+ GC_LVT), ('\u{cdb0}', '\u{cdb0}', GC_LV), ('\u{cdb1}', '\u{cdcb}', GC_LVT), ('\u{cdcc}',
+ '\u{cdcc}', GC_LV), ('\u{cdcd}', '\u{cde7}', GC_LVT), ('\u{cde8}', '\u{cde8}', GC_LV),
+ ('\u{cde9}', '\u{ce03}', GC_LVT), ('\u{ce04}', '\u{ce04}', GC_LV), ('\u{ce05}', '\u{ce1f}',
+ GC_LVT), ('\u{ce20}', '\u{ce20}', GC_LV), ('\u{ce21}', '\u{ce3b}', GC_LVT), ('\u{ce3c}',
+ '\u{ce3c}', GC_LV), ('\u{ce3d}', '\u{ce57}', GC_LVT), ('\u{ce58}', '\u{ce58}', GC_LV),
+ ('\u{ce59}', '\u{ce73}', GC_LVT), ('\u{ce74}', '\u{ce74}', GC_LV), ('\u{ce75}', '\u{ce8f}',
+ GC_LVT), ('\u{ce90}', '\u{ce90}', GC_LV), ('\u{ce91}', '\u{ceab}', GC_LVT), ('\u{ceac}',
+ '\u{ceac}', GC_LV), ('\u{cead}', '\u{cec7}', GC_LVT), ('\u{cec8}', '\u{cec8}', GC_LV),
+ ('\u{cec9}', '\u{cee3}', GC_LVT), ('\u{cee4}', '\u{cee4}', GC_LV), ('\u{cee5}', '\u{ceff}',
+ GC_LVT), ('\u{cf00}', '\u{cf00}', GC_LV), ('\u{cf01}', '\u{cf1b}', GC_LVT), ('\u{cf1c}',
+ '\u{cf1c}', GC_LV), ('\u{cf1d}', '\u{cf37}', GC_LVT), ('\u{cf38}', '\u{cf38}', GC_LV),
+ ('\u{cf39}', '\u{cf53}', GC_LVT), ('\u{cf54}', '\u{cf54}', GC_LV), ('\u{cf55}', '\u{cf6f}',
+ GC_LVT), ('\u{cf70}', '\u{cf70}', GC_LV), ('\u{cf71}', '\u{cf8b}', GC_LVT), ('\u{cf8c}',
+ '\u{cf8c}', GC_LV), ('\u{cf8d}', '\u{cfa7}', GC_LVT), ('\u{cfa8}', '\u{cfa8}', GC_LV),
+ ('\u{cfa9}', '\u{cfc3}', GC_LVT), ('\u{cfc4}', '\u{cfc4}', GC_LV), ('\u{cfc5}', '\u{cfdf}',
+ GC_LVT), ('\u{cfe0}', '\u{cfe0}', GC_LV), ('\u{cfe1}', '\u{cffb}', GC_LVT), ('\u{cffc}',
+ '\u{cffc}', GC_LV), ('\u{cffd}', '\u{d017}', GC_LVT), ('\u{d018}', '\u{d018}', GC_LV),
+ ('\u{d019}', '\u{d033}', GC_LVT), ('\u{d034}', '\u{d034}', GC_LV), ('\u{d035}', '\u{d04f}',
+ GC_LVT), ('\u{d050}', '\u{d050}', GC_LV), ('\u{d051}', '\u{d06b}', GC_LVT), ('\u{d06c}',
+ '\u{d06c}', GC_LV), ('\u{d06d}', '\u{d087}', GC_LVT), ('\u{d088}', '\u{d088}', GC_LV),
+ ('\u{d089}', '\u{d0a3}', GC_LVT), ('\u{d0a4}', '\u{d0a4}', GC_LV), ('\u{d0a5}', '\u{d0bf}',
+ GC_LVT), ('\u{d0c0}', '\u{d0c0}', GC_LV), ('\u{d0c1}', '\u{d0db}', GC_LVT), ('\u{d0dc}',
+ '\u{d0dc}', GC_LV), ('\u{d0dd}', '\u{d0f7}', GC_LVT), ('\u{d0f8}', '\u{d0f8}', GC_LV),
+ ('\u{d0f9}', '\u{d113}', GC_LVT), ('\u{d114}', '\u{d114}', GC_LV), ('\u{d115}', '\u{d12f}',
+ GC_LVT), ('\u{d130}', '\u{d130}', GC_LV), ('\u{d131}', '\u{d14b}', GC_LVT), ('\u{d14c}',
+ '\u{d14c}', GC_LV), ('\u{d14d}', '\u{d167}', GC_LVT), ('\u{d168}', '\u{d168}', GC_LV),
+ ('\u{d169}', '\u{d183}', GC_LVT), ('\u{d184}', '\u{d184}', GC_LV), ('\u{d185}', '\u{d19f}',
+ GC_LVT), ('\u{d1a0}', '\u{d1a0}', GC_LV), ('\u{d1a1}', '\u{d1bb}', GC_LVT), ('\u{d1bc}',
+ '\u{d1bc}', GC_LV), ('\u{d1bd}', '\u{d1d7}', GC_LVT), ('\u{d1d8}', '\u{d1d8}', GC_LV),
+ ('\u{d1d9}', '\u{d1f3}', GC_LVT), ('\u{d1f4}', '\u{d1f4}', GC_LV), ('\u{d1f5}', '\u{d20f}',
+ GC_LVT), ('\u{d210}', '\u{d210}', GC_LV), ('\u{d211}', '\u{d22b}', GC_LVT), ('\u{d22c}',
+ '\u{d22c}', GC_LV), ('\u{d22d}', '\u{d247}', GC_LVT), ('\u{d248}', '\u{d248}', GC_LV),
+ ('\u{d249}', '\u{d263}', GC_LVT), ('\u{d264}', '\u{d264}', GC_LV), ('\u{d265}', '\u{d27f}',
+ GC_LVT), ('\u{d280}', '\u{d280}', GC_LV), ('\u{d281}', '\u{d29b}', GC_LVT), ('\u{d29c}',
+ '\u{d29c}', GC_LV), ('\u{d29d}', '\u{d2b7}', GC_LVT), ('\u{d2b8}', '\u{d2b8}', GC_LV),
+ ('\u{d2b9}', '\u{d2d3}', GC_LVT), ('\u{d2d4}', '\u{d2d4}', GC_LV), ('\u{d2d5}', '\u{d2ef}',
+ GC_LVT), ('\u{d2f0}', '\u{d2f0}', GC_LV), ('\u{d2f1}', '\u{d30b}', GC_LVT), ('\u{d30c}',
+ '\u{d30c}', GC_LV), ('\u{d30d}', '\u{d327}', GC_LVT), ('\u{d328}', '\u{d328}', GC_LV),
+ ('\u{d329}', '\u{d343}', GC_LVT), ('\u{d344}', '\u{d344}', GC_LV), ('\u{d345}', '\u{d35f}',
+ GC_LVT), ('\u{d360}', '\u{d360}', GC_LV), ('\u{d361}', '\u{d37b}', GC_LVT), ('\u{d37c}',
+ '\u{d37c}', GC_LV), ('\u{d37d}', '\u{d397}', GC_LVT), ('\u{d398}', '\u{d398}', GC_LV),
+ ('\u{d399}', '\u{d3b3}', GC_LVT), ('\u{d3b4}', '\u{d3b4}', GC_LV), ('\u{d3b5}', '\u{d3cf}',
+ GC_LVT), ('\u{d3d0}', '\u{d3d0}', GC_LV), ('\u{d3d1}', '\u{d3eb}', GC_LVT), ('\u{d3ec}',
+ '\u{d3ec}', GC_LV), ('\u{d3ed}', '\u{d407}', GC_LVT), ('\u{d408}', '\u{d408}', GC_LV),
+ ('\u{d409}', '\u{d423}', GC_LVT), ('\u{d424}', '\u{d424}', GC_LV), ('\u{d425}', '\u{d43f}',
+ GC_LVT), ('\u{d440}', '\u{d440}', GC_LV), ('\u{d441}', '\u{d45b}', GC_LVT), ('\u{d45c}',
+ '\u{d45c}', GC_LV), ('\u{d45d}', '\u{d477}', GC_LVT), ('\u{d478}', '\u{d478}', GC_LV),
+ ('\u{d479}', '\u{d493}', GC_LVT), ('\u{d494}', '\u{d494}', GC_LV), ('\u{d495}', '\u{d4af}',
+ GC_LVT), ('\u{d4b0}', '\u{d4b0}', GC_LV), ('\u{d4b1}', '\u{d4cb}', GC_LVT), ('\u{d4cc}',
+ '\u{d4cc}', GC_LV), ('\u{d4cd}', '\u{d4e7}', GC_LVT), ('\u{d4e8}', '\u{d4e8}', GC_LV),
+ ('\u{d4e9}', '\u{d503}', GC_LVT), ('\u{d504}', '\u{d504}', GC_LV), ('\u{d505}', '\u{d51f}',
+ GC_LVT), ('\u{d520}', '\u{d520}', GC_LV), ('\u{d521}', '\u{d53b}', GC_LVT), ('\u{d53c}',
+ '\u{d53c}', GC_LV), ('\u{d53d}', '\u{d557}', GC_LVT), ('\u{d558}', '\u{d558}', GC_LV),
+ ('\u{d559}', '\u{d573}', GC_LVT), ('\u{d574}', '\u{d574}', GC_LV), ('\u{d575}', '\u{d58f}',
+ GC_LVT), ('\u{d590}', '\u{d590}', GC_LV), ('\u{d591}', '\u{d5ab}', GC_LVT), ('\u{d5ac}',
+ '\u{d5ac}', GC_LV), ('\u{d5ad}', '\u{d5c7}', GC_LVT), ('\u{d5c8}', '\u{d5c8}', GC_LV),
+ ('\u{d5c9}', '\u{d5e3}', GC_LVT), ('\u{d5e4}', '\u{d5e4}', GC_LV), ('\u{d5e5}', '\u{d5ff}',
+ GC_LVT), ('\u{d600}', '\u{d600}', GC_LV), ('\u{d601}', '\u{d61b}', GC_LVT), ('\u{d61c}',
+ '\u{d61c}', GC_LV), ('\u{d61d}', '\u{d637}', GC_LVT), ('\u{d638}', '\u{d638}', GC_LV),
+ ('\u{d639}', '\u{d653}', GC_LVT), ('\u{d654}', '\u{d654}', GC_LV), ('\u{d655}', '\u{d66f}',
+ GC_LVT), ('\u{d670}', '\u{d670}', GC_LV), ('\u{d671}', '\u{d68b}', GC_LVT), ('\u{d68c}',
+ '\u{d68c}', GC_LV), ('\u{d68d}', '\u{d6a7}', GC_LVT), ('\u{d6a8}', '\u{d6a8}', GC_LV),
+ ('\u{d6a9}', '\u{d6c3}', GC_LVT), ('\u{d6c4}', '\u{d6c4}', GC_LV), ('\u{d6c5}', '\u{d6df}',
+ GC_LVT), ('\u{d6e0}', '\u{d6e0}', GC_LV), ('\u{d6e1}', '\u{d6fb}', GC_LVT), ('\u{d6fc}',
+ '\u{d6fc}', GC_LV), ('\u{d6fd}', '\u{d717}', GC_LVT), ('\u{d718}', '\u{d718}', GC_LV),
+ ('\u{d719}', '\u{d733}', GC_LVT), ('\u{d734}', '\u{d734}', GC_LV), ('\u{d735}', '\u{d74f}',
+ GC_LVT), ('\u{d750}', '\u{d750}', GC_LV), ('\u{d751}', '\u{d76b}', GC_LVT), ('\u{d76c}',
+ '\u{d76c}', GC_LV), ('\u{d76d}', '\u{d787}', GC_LVT), ('\u{d788}', '\u{d788}', GC_LV),
+ ('\u{d789}', '\u{d7a3}', GC_LVT), ('\u{d7b0}', '\u{d7c6}', GC_V), ('\u{d7cb}', '\u{d7fb}',
+ GC_T), ('\u{fb1e}', '\u{fb1e}', GC_Extend), ('\u{fe00}', '\u{fe0f}', GC_Extend),
+ ('\u{fe20}', '\u{fe2f}', GC_Extend), ('\u{feff}', '\u{feff}', GC_Control), ('\u{ff9e}',
+ '\u{ff9f}', GC_Extend), ('\u{fff0}', '\u{fffb}', GC_Control), ('\u{101fd}', '\u{101fd}',
+ GC_Extend), ('\u{102e0}', '\u{102e0}', GC_Extend), ('\u{10376}', '\u{1037a}', GC_Extend),
+ ('\u{10a01}', '\u{10a03}', GC_Extend), ('\u{10a05}', '\u{10a06}', GC_Extend), ('\u{10a0c}',
+ '\u{10a0f}', GC_Extend), ('\u{10a38}', '\u{10a3a}', GC_Extend), ('\u{10a3f}', '\u{10a3f}',
+ GC_Extend), ('\u{10ae5}', '\u{10ae6}', GC_Extend), ('\u{10d24}', '\u{10d27}', GC_Extend),
+ ('\u{10eab}', '\u{10eac}', GC_Extend), ('\u{10f46}', '\u{10f50}', GC_Extend), ('\u{11000}',
+ '\u{11000}', GC_SpacingMark), ('\u{11001}', '\u{11001}', GC_Extend), ('\u{11002}',
+ '\u{11002}', GC_SpacingMark), ('\u{11038}', '\u{11046}', GC_Extend), ('\u{1107f}',
+ '\u{11081}', GC_Extend), ('\u{11082}', '\u{11082}', GC_SpacingMark), ('\u{110b0}',
+ '\u{110b2}', GC_SpacingMark), ('\u{110b3}', '\u{110b6}', GC_Extend), ('\u{110b7}',
+ '\u{110b8}', GC_SpacingMark), ('\u{110b9}', '\u{110ba}', GC_Extend), ('\u{110bd}',
+ '\u{110bd}', GC_Prepend), ('\u{110cd}', '\u{110cd}', GC_Prepend), ('\u{11100}', '\u{11102}',
+ GC_Extend), ('\u{11127}', '\u{1112b}', GC_Extend), ('\u{1112c}', '\u{1112c}',
+ GC_SpacingMark), ('\u{1112d}', '\u{11134}', GC_Extend), ('\u{11145}', '\u{11146}',
+ GC_SpacingMark), ('\u{11173}', '\u{11173}', GC_Extend), ('\u{11180}', '\u{11181}',
+ GC_Extend), ('\u{11182}', '\u{11182}', GC_SpacingMark), ('\u{111b3}', '\u{111b5}',
+ GC_SpacingMark), ('\u{111b6}', '\u{111be}', GC_Extend), ('\u{111bf}', '\u{111c0}',
+ GC_SpacingMark), ('\u{111c2}', '\u{111c3}', GC_Prepend), ('\u{111c9}', '\u{111cc}',
+ GC_Extend), ('\u{111ce}', '\u{111ce}', GC_SpacingMark), ('\u{111cf}', '\u{111cf}',
+ GC_Extend), ('\u{1122c}', '\u{1122e}', GC_SpacingMark), ('\u{1122f}', '\u{11231}',
+ GC_Extend), ('\u{11232}', '\u{11233}', GC_SpacingMark), ('\u{11234}', '\u{11234}',
+ GC_Extend), ('\u{11235}', '\u{11235}', GC_SpacingMark), ('\u{11236}', '\u{11237}',
+ GC_Extend), ('\u{1123e}', '\u{1123e}', GC_Extend), ('\u{112df}', '\u{112df}', GC_Extend),
+ ('\u{112e0}', '\u{112e2}', GC_SpacingMark), ('\u{112e3}', '\u{112ea}', GC_Extend),
+ ('\u{11300}', '\u{11301}', GC_Extend), ('\u{11302}', '\u{11303}', GC_SpacingMark),
+ ('\u{1133b}', '\u{1133c}', GC_Extend), ('\u{1133e}', '\u{1133e}', GC_Extend), ('\u{1133f}',
+ '\u{1133f}', GC_SpacingMark), ('\u{11340}', '\u{11340}', GC_Extend), ('\u{11341}',
+ '\u{11344}', GC_SpacingMark), ('\u{11347}', '\u{11348}', GC_SpacingMark), ('\u{1134b}',
+ '\u{1134d}', GC_SpacingMark), ('\u{11357}', '\u{11357}', GC_Extend), ('\u{11362}',
+ '\u{11363}', GC_SpacingMark), ('\u{11366}', '\u{1136c}', GC_Extend), ('\u{11370}',
+ '\u{11374}', GC_Extend), ('\u{11435}', '\u{11437}', GC_SpacingMark), ('\u{11438}',
+ '\u{1143f}', GC_Extend), ('\u{11440}', '\u{11441}', GC_SpacingMark), ('\u{11442}',
+ '\u{11444}', GC_Extend), ('\u{11445}', '\u{11445}', GC_SpacingMark), ('\u{11446}',
+ '\u{11446}', GC_Extend), ('\u{1145e}', '\u{1145e}', GC_Extend), ('\u{114b0}', '\u{114b0}',
+ GC_Extend), ('\u{114b1}', '\u{114b2}', GC_SpacingMark), ('\u{114b3}', '\u{114b8}',
+ GC_Extend), ('\u{114b9}', '\u{114b9}', GC_SpacingMark), ('\u{114ba}', '\u{114ba}',
+ GC_Extend), ('\u{114bb}', '\u{114bc}', GC_SpacingMark), ('\u{114bd}', '\u{114bd}',
+ GC_Extend), ('\u{114be}', '\u{114be}', GC_SpacingMark), ('\u{114bf}', '\u{114c0}',
+ GC_Extend), ('\u{114c1}', '\u{114c1}', GC_SpacingMark), ('\u{114c2}', '\u{114c3}',
+ GC_Extend), ('\u{115af}', '\u{115af}', GC_Extend), ('\u{115b0}', '\u{115b1}',
+ GC_SpacingMark), ('\u{115b2}', '\u{115b5}', GC_Extend), ('\u{115b8}', '\u{115bb}',
+ GC_SpacingMark), ('\u{115bc}', '\u{115bd}', GC_Extend), ('\u{115be}', '\u{115be}',
+ GC_SpacingMark), ('\u{115bf}', '\u{115c0}', GC_Extend), ('\u{115dc}', '\u{115dd}',
+ GC_Extend), ('\u{11630}', '\u{11632}', GC_SpacingMark), ('\u{11633}', '\u{1163a}',
+ GC_Extend), ('\u{1163b}', '\u{1163c}', GC_SpacingMark), ('\u{1163d}', '\u{1163d}',
+ GC_Extend), ('\u{1163e}', '\u{1163e}', GC_SpacingMark), ('\u{1163f}', '\u{11640}',
+ GC_Extend), ('\u{116ab}', '\u{116ab}', GC_Extend), ('\u{116ac}', '\u{116ac}',
+ GC_SpacingMark), ('\u{116ad}', '\u{116ad}', GC_Extend), ('\u{116ae}', '\u{116af}',
+ GC_SpacingMark), ('\u{116b0}', '\u{116b5}', GC_Extend), ('\u{116b6}', '\u{116b6}',
+ GC_SpacingMark), ('\u{116b7}', '\u{116b7}', GC_Extend), ('\u{1171d}', '\u{1171f}',
+ GC_Extend), ('\u{11720}', '\u{11721}', GC_SpacingMark), ('\u{11722}', '\u{11725}',
+ GC_Extend), ('\u{11726}', '\u{11726}', GC_SpacingMark), ('\u{11727}', '\u{1172b}',
+ GC_Extend), ('\u{1182c}', '\u{1182e}', GC_SpacingMark), ('\u{1182f}', '\u{11837}',
+ GC_Extend), ('\u{11838}', '\u{11838}', GC_SpacingMark), ('\u{11839}', '\u{1183a}',
+ GC_Extend), ('\u{11930}', '\u{11930}', GC_Extend), ('\u{11931}', '\u{11935}',
+ GC_SpacingMark), ('\u{11937}', '\u{11938}', GC_SpacingMark), ('\u{1193b}', '\u{1193c}',
+ GC_Extend), ('\u{1193d}', '\u{1193d}', GC_SpacingMark), ('\u{1193e}', '\u{1193e}',
+ GC_Extend), ('\u{1193f}', '\u{1193f}', GC_Prepend), ('\u{11940}', '\u{11940}',
+ GC_SpacingMark), ('\u{11941}', '\u{11941}', GC_Prepend), ('\u{11942}', '\u{11942}',
+ GC_SpacingMark), ('\u{11943}', '\u{11943}', GC_Extend), ('\u{119d1}', '\u{119d3}',
GC_SpacingMark), ('\u{119d4}', '\u{119d7}', GC_Extend), ('\u{119da}', '\u{119db}',
GC_Extend), ('\u{119dc}', '\u{119df}', GC_SpacingMark), ('\u{119e0}', '\u{119e0}',
GC_Extend), ('\u{119e4}', '\u{119e4}', GC_SpacingMark), ('\u{11a01}', '\u{11a0a}',
@@ -913,6 +934,7 @@
GC_SpacingMark), ('\u{13430}', '\u{13438}', GC_Control), ('\u{16af0}', '\u{16af4}',
GC_Extend), ('\u{16b30}', '\u{16b36}', GC_Extend), ('\u{16f4f}', '\u{16f4f}', GC_Extend),
('\u{16f51}', '\u{16f87}', GC_SpacingMark), ('\u{16f8f}', '\u{16f92}', GC_Extend),
+ ('\u{16fe4}', '\u{16fe4}', GC_Extend), ('\u{16ff0}', '\u{16ff1}', GC_SpacingMark),
('\u{1bc9d}', '\u{1bc9e}', GC_Extend), ('\u{1bca0}', '\u{1bca3}', GC_Control), ('\u{1d165}',
'\u{1d165}', GC_Extend), ('\u{1d166}', '\u{1d166}', GC_SpacingMark), ('\u{1d167}',
'\u{1d169}', GC_Extend), ('\u{1d16d}', '\u{1d16d}', GC_SpacingMark), ('\u{1d16e}',
@@ -943,10 +965,10 @@
('\u{1f85a}', '\u{1f85f}', GC_Extended_Pictographic), ('\u{1f888}', '\u{1f88f}',
GC_Extended_Pictographic), ('\u{1f8ae}', '\u{1f8ff}', GC_Extended_Pictographic),
('\u{1f90c}', '\u{1f93a}', GC_Extended_Pictographic), ('\u{1f93c}', '\u{1f945}',
- GC_Extended_Pictographic), ('\u{1f947}', '\u{1fffd}', GC_Extended_Pictographic),
- ('\u{e0000}', '\u{e001f}', GC_Control), ('\u{e0020}', '\u{e007f}', GC_Extend), ('\u{e0080}',
- '\u{e00ff}', GC_Control), ('\u{e0100}', '\u{e01ef}', GC_Extend), ('\u{e01f0}', '\u{e0fff}',
- GC_Control)
+ GC_Extended_Pictographic), ('\u{1f947}', '\u{1faff}', GC_Extended_Pictographic),
+ ('\u{1fc00}', '\u{1fffd}', GC_Extended_Pictographic), ('\u{e0000}', '\u{e001f}',
+ GC_Control), ('\u{e0020}', '\u{e007f}', GC_Extend), ('\u{e0080}', '\u{e00ff}', GC_Control),
+ ('\u{e0100}', '\u{e01ef}', GC_Extend), ('\u{e01f0}', '\u{e0fff}', GC_Control)
];
}
@@ -980,7 +1002,8 @@
WC_ZWJ,
}
- fn bsearch_range_value_table(c: char, r: &'static [(char, char, WordCat)]) -> WordCat {
+ fn bsearch_range_value_table(c: char, r: &'static [(char, char, WordCat)]) -> (u32, u32, WordCat) {
+ use core;
use core::cmp::Ordering::{Equal, Less, Greater};
match r.binary_search_by(|&(lo, hi, _)| {
if lo <= c && c <= hi { Equal }
@@ -988,14 +1011,20 @@
else { Greater }
}) {
Ok(idx) => {
- let (_, _, cat) = r[idx];
- cat
+ let (lower, upper, cat) = r[idx];
+ (lower as u32, upper as u32, cat)
}
- Err(_) => WC_Any
+ Err(idx) => {
+ (
+ if idx > 0 { r[idx-1].1 as u32 + 1 } else { 0 },
+ r.get(idx).map(|c|c.0 as u32 - 1).unwrap_or(core::u32::MAX),
+ WC_Any,
+ )
+ }
}
}
- pub fn word_category(c: char) -> WordCat {
+ pub fn word_category(c: char) -> (u32, u32, WordCat) {
bsearch_range_value_table(c, word_cat_table)
}
@@ -1009,16 +1038,16 @@
('\u{aa}', '\u{aa}', WC_ALetter), ('\u{ad}', '\u{ad}', WC_Format), ('\u{b5}', '\u{b5}',
WC_ALetter), ('\u{b7}', '\u{b7}', WC_MidLetter), ('\u{ba}', '\u{ba}', WC_ALetter),
('\u{c0}', '\u{d6}', WC_ALetter), ('\u{d8}', '\u{f6}', WC_ALetter), ('\u{f8}', '\u{2d7}',
- WC_ALetter), ('\u{2de}', '\u{2e4}', WC_ALetter), ('\u{2ec}', '\u{2ff}', WC_ALetter),
- ('\u{300}', '\u{36f}', WC_Extend), ('\u{370}', '\u{374}', WC_ALetter), ('\u{376}',
- '\u{377}', WC_ALetter), ('\u{37a}', '\u{37d}', WC_ALetter), ('\u{37e}', '\u{37e}',
- WC_MidNum), ('\u{37f}', '\u{37f}', WC_ALetter), ('\u{386}', '\u{386}', WC_ALetter),
- ('\u{387}', '\u{387}', WC_MidLetter), ('\u{388}', '\u{38a}', WC_ALetter), ('\u{38c}',
- '\u{38c}', WC_ALetter), ('\u{38e}', '\u{3a1}', WC_ALetter), ('\u{3a3}', '\u{3f5}',
- WC_ALetter), ('\u{3f7}', '\u{481}', WC_ALetter), ('\u{483}', '\u{489}', WC_Extend),
- ('\u{48a}', '\u{52f}', WC_ALetter), ('\u{531}', '\u{556}', WC_ALetter), ('\u{559}',
- '\u{559}', WC_ALetter), ('\u{55b}', '\u{55c}', WC_ALetter), ('\u{55e}', '\u{55e}',
- WC_ALetter), ('\u{560}', '\u{588}', WC_ALetter), ('\u{589}', '\u{589}', WC_MidNum),
+ WC_ALetter), ('\u{2de}', '\u{2ff}', WC_ALetter), ('\u{300}', '\u{36f}', WC_Extend),
+ ('\u{370}', '\u{374}', WC_ALetter), ('\u{376}', '\u{377}', WC_ALetter), ('\u{37a}',
+ '\u{37d}', WC_ALetter), ('\u{37e}', '\u{37e}', WC_MidNum), ('\u{37f}', '\u{37f}',
+ WC_ALetter), ('\u{386}', '\u{386}', WC_ALetter), ('\u{387}', '\u{387}', WC_MidLetter),
+ ('\u{388}', '\u{38a}', WC_ALetter), ('\u{38c}', '\u{38c}', WC_ALetter), ('\u{38e}',
+ '\u{3a1}', WC_ALetter), ('\u{3a3}', '\u{3f5}', WC_ALetter), ('\u{3f7}', '\u{481}',
+ WC_ALetter), ('\u{483}', '\u{489}', WC_Extend), ('\u{48a}', '\u{52f}', WC_ALetter),
+ ('\u{531}', '\u{556}', WC_ALetter), ('\u{559}', '\u{55c}', WC_ALetter), ('\u{55e}',
+ '\u{55e}', WC_ALetter), ('\u{55f}', '\u{55f}', WC_MidLetter), ('\u{560}', '\u{588}',
+ WC_ALetter), ('\u{589}', '\u{589}', WC_MidNum), ('\u{58a}', '\u{58a}', WC_ALetter),
('\u{591}', '\u{5bd}', WC_Extend), ('\u{5bf}', '\u{5bf}', WC_Extend), ('\u{5c1}', '\u{5c2}',
WC_Extend), ('\u{5c4}', '\u{5c5}', WC_Extend), ('\u{5c7}', '\u{5c7}', WC_Extend),
('\u{5d0}', '\u{5ea}', WC_Hebrew_Letter), ('\u{5ef}', '\u{5f2}', WC_Hebrew_Letter),
@@ -1044,7 +1073,7 @@
('\u{825}', '\u{827}', WC_Extend), ('\u{828}', '\u{828}', WC_ALetter), ('\u{829}',
'\u{82d}', WC_Extend), ('\u{840}', '\u{858}', WC_ALetter), ('\u{859}', '\u{85b}',
WC_Extend), ('\u{860}', '\u{86a}', WC_ALetter), ('\u{8a0}', '\u{8b4}', WC_ALetter),
- ('\u{8b6}', '\u{8bd}', WC_ALetter), ('\u{8d3}', '\u{8e1}', WC_Extend), ('\u{8e2}',
+ ('\u{8b6}', '\u{8c7}', WC_ALetter), ('\u{8d3}', '\u{8e1}', WC_Extend), ('\u{8e2}',
'\u{8e2}', WC_Format), ('\u{8e3}', '\u{903}', WC_Extend), ('\u{904}', '\u{939}',
WC_ALetter), ('\u{93a}', '\u{93c}', WC_Extend), ('\u{93d}', '\u{93d}', WC_ALetter),
('\u{93e}', '\u{94f}', WC_Extend), ('\u{950}', '\u{950}', WC_ALetter), ('\u{951}',
@@ -1079,7 +1108,7 @@
'\u{b28}', WC_ALetter), ('\u{b2a}', '\u{b30}', WC_ALetter), ('\u{b32}', '\u{b33}',
WC_ALetter), ('\u{b35}', '\u{b39}', WC_ALetter), ('\u{b3c}', '\u{b3c}', WC_Extend),
('\u{b3d}', '\u{b3d}', WC_ALetter), ('\u{b3e}', '\u{b44}', WC_Extend), ('\u{b47}',
- '\u{b48}', WC_Extend), ('\u{b4b}', '\u{b4d}', WC_Extend), ('\u{b56}', '\u{b57}', WC_Extend),
+ '\u{b48}', WC_Extend), ('\u{b4b}', '\u{b4d}', WC_Extend), ('\u{b55}', '\u{b57}', WC_Extend),
('\u{b5c}', '\u{b5d}', WC_ALetter), ('\u{b5f}', '\u{b61}', WC_ALetter), ('\u{b62}',
'\u{b63}', WC_Extend), ('\u{b66}', '\u{b6f}', WC_Numeric), ('\u{b71}', '\u{b71}',
WC_ALetter), ('\u{b82}', '\u{b82}', WC_Extend), ('\u{b83}', '\u{b83}', WC_ALetter),
@@ -1103,13 +1132,13 @@
WC_Extend), ('\u{cd5}', '\u{cd6}', WC_Extend), ('\u{cde}', '\u{cde}', WC_ALetter),
('\u{ce0}', '\u{ce1}', WC_ALetter), ('\u{ce2}', '\u{ce3}', WC_Extend), ('\u{ce6}',
'\u{cef}', WC_Numeric), ('\u{cf1}', '\u{cf2}', WC_ALetter), ('\u{d00}', '\u{d03}',
- WC_Extend), ('\u{d05}', '\u{d0c}', WC_ALetter), ('\u{d0e}', '\u{d10}', WC_ALetter),
+ WC_Extend), ('\u{d04}', '\u{d0c}', WC_ALetter), ('\u{d0e}', '\u{d10}', WC_ALetter),
('\u{d12}', '\u{d3a}', WC_ALetter), ('\u{d3b}', '\u{d3c}', WC_Extend), ('\u{d3d}',
'\u{d3d}', WC_ALetter), ('\u{d3e}', '\u{d44}', WC_Extend), ('\u{d46}', '\u{d48}',
WC_Extend), ('\u{d4a}', '\u{d4d}', WC_Extend), ('\u{d4e}', '\u{d4e}', WC_ALetter),
('\u{d54}', '\u{d56}', WC_ALetter), ('\u{d57}', '\u{d57}', WC_Extend), ('\u{d5f}',
'\u{d61}', WC_ALetter), ('\u{d62}', '\u{d63}', WC_Extend), ('\u{d66}', '\u{d6f}',
- WC_Numeric), ('\u{d7a}', '\u{d7f}', WC_ALetter), ('\u{d82}', '\u{d83}', WC_Extend),
+ WC_Numeric), ('\u{d7a}', '\u{d7f}', WC_ALetter), ('\u{d81}', '\u{d83}', WC_Extend),
('\u{d85}', '\u{d96}', WC_ALetter), ('\u{d9a}', '\u{db1}', WC_ALetter), ('\u{db3}',
'\u{dbb}', WC_ALetter), ('\u{dbd}', '\u{dbd}', WC_ALetter), ('\u{dc0}', '\u{dc6}',
WC_ALetter), ('\u{dca}', '\u{dca}', WC_Extend), ('\u{dcf}', '\u{dd4}', WC_Extend),
@@ -1156,7 +1185,7 @@
WC_Numeric), ('\u{19d0}', '\u{19d9}', WC_Numeric), ('\u{1a00}', '\u{1a16}', WC_ALetter),
('\u{1a17}', '\u{1a1b}', WC_Extend), ('\u{1a55}', '\u{1a5e}', WC_Extend), ('\u{1a60}',
'\u{1a7c}', WC_Extend), ('\u{1a7f}', '\u{1a7f}', WC_Extend), ('\u{1a80}', '\u{1a89}',
- WC_Numeric), ('\u{1a90}', '\u{1a99}', WC_Numeric), ('\u{1ab0}', '\u{1abe}', WC_Extend),
+ WC_Numeric), ('\u{1a90}', '\u{1a99}', WC_Numeric), ('\u{1ab0}', '\u{1ac0}', WC_Extend),
('\u{1b00}', '\u{1b04}', WC_Extend), ('\u{1b05}', '\u{1b33}', WC_ALetter), ('\u{1b34}',
'\u{1b44}', WC_Extend), ('\u{1b45}', '\u{1b4b}', WC_ALetter), ('\u{1b50}', '\u{1b59}',
WC_Numeric), ('\u{1b6b}', '\u{1b73}', WC_Extend), ('\u{1b80}', '\u{1b82}', WC_Extend),
@@ -1210,111 +1239,114 @@
'\u{303c}', WC_ALetter), ('\u{3099}', '\u{309a}', WC_Extend), ('\u{309b}', '\u{309c}',
WC_Katakana), ('\u{30a0}', '\u{30fa}', WC_Katakana), ('\u{30fc}', '\u{30ff}', WC_Katakana),
('\u{3105}', '\u{312f}', WC_ALetter), ('\u{3131}', '\u{318e}', WC_ALetter), ('\u{31a0}',
- '\u{31ba}', WC_ALetter), ('\u{31f0}', '\u{31ff}', WC_Katakana), ('\u{32d0}', '\u{32fe}',
+ '\u{31bf}', WC_ALetter), ('\u{31f0}', '\u{31ff}', WC_Katakana), ('\u{32d0}', '\u{32fe}',
WC_Katakana), ('\u{3300}', '\u{3357}', WC_Katakana), ('\u{a000}', '\u{a48c}', WC_ALetter),
('\u{a4d0}', '\u{a4fd}', WC_ALetter), ('\u{a500}', '\u{a60c}', WC_ALetter), ('\u{a610}',
'\u{a61f}', WC_ALetter), ('\u{a620}', '\u{a629}', WC_Numeric), ('\u{a62a}', '\u{a62b}',
WC_ALetter), ('\u{a640}', '\u{a66e}', WC_ALetter), ('\u{a66f}', '\u{a672}', WC_Extend),
('\u{a674}', '\u{a67d}', WC_Extend), ('\u{a67f}', '\u{a69d}', WC_ALetter), ('\u{a69e}',
'\u{a69f}', WC_Extend), ('\u{a6a0}', '\u{a6ef}', WC_ALetter), ('\u{a6f0}', '\u{a6f1}',
- WC_Extend), ('\u{a717}', '\u{a7bf}', WC_ALetter), ('\u{a7c2}', '\u{a7c6}', WC_ALetter),
- ('\u{a7f7}', '\u{a801}', WC_ALetter), ('\u{a802}', '\u{a802}', WC_Extend), ('\u{a803}',
+ WC_Extend), ('\u{a708}', '\u{a7bf}', WC_ALetter), ('\u{a7c2}', '\u{a7ca}', WC_ALetter),
+ ('\u{a7f5}', '\u{a801}', WC_ALetter), ('\u{a802}', '\u{a802}', WC_Extend), ('\u{a803}',
'\u{a805}', WC_ALetter), ('\u{a806}', '\u{a806}', WC_Extend), ('\u{a807}', '\u{a80a}',
WC_ALetter), ('\u{a80b}', '\u{a80b}', WC_Extend), ('\u{a80c}', '\u{a822}', WC_ALetter),
- ('\u{a823}', '\u{a827}', WC_Extend), ('\u{a840}', '\u{a873}', WC_ALetter), ('\u{a880}',
- '\u{a881}', WC_Extend), ('\u{a882}', '\u{a8b3}', WC_ALetter), ('\u{a8b4}', '\u{a8c5}',
- WC_Extend), ('\u{a8d0}', '\u{a8d9}', WC_Numeric), ('\u{a8e0}', '\u{a8f1}', WC_Extend),
- ('\u{a8f2}', '\u{a8f7}', WC_ALetter), ('\u{a8fb}', '\u{a8fb}', WC_ALetter), ('\u{a8fd}',
- '\u{a8fe}', WC_ALetter), ('\u{a8ff}', '\u{a8ff}', WC_Extend), ('\u{a900}', '\u{a909}',
- WC_Numeric), ('\u{a90a}', '\u{a925}', WC_ALetter), ('\u{a926}', '\u{a92d}', WC_Extend),
- ('\u{a930}', '\u{a946}', WC_ALetter), ('\u{a947}', '\u{a953}', WC_Extend), ('\u{a960}',
- '\u{a97c}', WC_ALetter), ('\u{a980}', '\u{a983}', WC_Extend), ('\u{a984}', '\u{a9b2}',
- WC_ALetter), ('\u{a9b3}', '\u{a9c0}', WC_Extend), ('\u{a9cf}', '\u{a9cf}', WC_ALetter),
- ('\u{a9d0}', '\u{a9d9}', WC_Numeric), ('\u{a9e5}', '\u{a9e5}', WC_Extend), ('\u{a9f0}',
- '\u{a9f9}', WC_Numeric), ('\u{aa00}', '\u{aa28}', WC_ALetter), ('\u{aa29}', '\u{aa36}',
- WC_Extend), ('\u{aa40}', '\u{aa42}', WC_ALetter), ('\u{aa43}', '\u{aa43}', WC_Extend),
- ('\u{aa44}', '\u{aa4b}', WC_ALetter), ('\u{aa4c}', '\u{aa4d}', WC_Extend), ('\u{aa50}',
- '\u{aa59}', WC_Numeric), ('\u{aa7b}', '\u{aa7d}', WC_Extend), ('\u{aab0}', '\u{aab0}',
- WC_Extend), ('\u{aab2}', '\u{aab4}', WC_Extend), ('\u{aab7}', '\u{aab8}', WC_Extend),
- ('\u{aabe}', '\u{aabf}', WC_Extend), ('\u{aac1}', '\u{aac1}', WC_Extend), ('\u{aae0}',
- '\u{aaea}', WC_ALetter), ('\u{aaeb}', '\u{aaef}', WC_Extend), ('\u{aaf2}', '\u{aaf4}',
- WC_ALetter), ('\u{aaf5}', '\u{aaf6}', WC_Extend), ('\u{ab01}', '\u{ab06}', WC_ALetter),
- ('\u{ab09}', '\u{ab0e}', WC_ALetter), ('\u{ab11}', '\u{ab16}', WC_ALetter), ('\u{ab20}',
- '\u{ab26}', WC_ALetter), ('\u{ab28}', '\u{ab2e}', WC_ALetter), ('\u{ab30}', '\u{ab67}',
- WC_ALetter), ('\u{ab70}', '\u{abe2}', WC_ALetter), ('\u{abe3}', '\u{abea}', WC_Extend),
- ('\u{abec}', '\u{abed}', WC_Extend), ('\u{abf0}', '\u{abf9}', WC_Numeric), ('\u{ac00}',
- '\u{d7a3}', WC_ALetter), ('\u{d7b0}', '\u{d7c6}', WC_ALetter), ('\u{d7cb}', '\u{d7fb}',
- WC_ALetter), ('\u{fb00}', '\u{fb06}', WC_ALetter), ('\u{fb13}', '\u{fb17}', WC_ALetter),
- ('\u{fb1d}', '\u{fb1d}', WC_Hebrew_Letter), ('\u{fb1e}', '\u{fb1e}', WC_Extend),
- ('\u{fb1f}', '\u{fb28}', WC_Hebrew_Letter), ('\u{fb2a}', '\u{fb36}', WC_Hebrew_Letter),
- ('\u{fb38}', '\u{fb3c}', WC_Hebrew_Letter), ('\u{fb3e}', '\u{fb3e}', WC_Hebrew_Letter),
- ('\u{fb40}', '\u{fb41}', WC_Hebrew_Letter), ('\u{fb43}', '\u{fb44}', WC_Hebrew_Letter),
- ('\u{fb46}', '\u{fb4f}', WC_Hebrew_Letter), ('\u{fb50}', '\u{fbb1}', WC_ALetter),
- ('\u{fbd3}', '\u{fd3d}', WC_ALetter), ('\u{fd50}', '\u{fd8f}', WC_ALetter), ('\u{fd92}',
- '\u{fdc7}', WC_ALetter), ('\u{fdf0}', '\u{fdfb}', WC_ALetter), ('\u{fe00}', '\u{fe0f}',
- WC_Extend), ('\u{fe10}', '\u{fe10}', WC_MidNum), ('\u{fe13}', '\u{fe13}', WC_MidLetter),
- ('\u{fe14}', '\u{fe14}', WC_MidNum), ('\u{fe20}', '\u{fe2f}', WC_Extend), ('\u{fe33}',
- '\u{fe34}', WC_ExtendNumLet), ('\u{fe4d}', '\u{fe4f}', WC_ExtendNumLet), ('\u{fe50}',
- '\u{fe50}', WC_MidNum), ('\u{fe52}', '\u{fe52}', WC_MidNumLet), ('\u{fe54}', '\u{fe54}',
- WC_MidNum), ('\u{fe55}', '\u{fe55}', WC_MidLetter), ('\u{fe70}', '\u{fe74}', WC_ALetter),
- ('\u{fe76}', '\u{fefc}', WC_ALetter), ('\u{feff}', '\u{feff}', WC_Format), ('\u{ff07}',
- '\u{ff07}', WC_MidNumLet), ('\u{ff0c}', '\u{ff0c}', WC_MidNum), ('\u{ff0e}', '\u{ff0e}',
- WC_MidNumLet), ('\u{ff10}', '\u{ff19}', WC_Numeric), ('\u{ff1a}', '\u{ff1a}', WC_MidLetter),
- ('\u{ff1b}', '\u{ff1b}', WC_MidNum), ('\u{ff21}', '\u{ff3a}', WC_ALetter), ('\u{ff3f}',
- '\u{ff3f}', WC_ExtendNumLet), ('\u{ff41}', '\u{ff5a}', WC_ALetter), ('\u{ff66}', '\u{ff9d}',
- WC_Katakana), ('\u{ff9e}', '\u{ff9f}', WC_Extend), ('\u{ffa0}', '\u{ffbe}', WC_ALetter),
- ('\u{ffc2}', '\u{ffc7}', WC_ALetter), ('\u{ffca}', '\u{ffcf}', WC_ALetter), ('\u{ffd2}',
- '\u{ffd7}', WC_ALetter), ('\u{ffda}', '\u{ffdc}', WC_ALetter), ('\u{fff9}', '\u{fffb}',
- WC_Format), ('\u{10000}', '\u{1000b}', WC_ALetter), ('\u{1000d}', '\u{10026}', WC_ALetter),
- ('\u{10028}', '\u{1003a}', WC_ALetter), ('\u{1003c}', '\u{1003d}', WC_ALetter),
- ('\u{1003f}', '\u{1004d}', WC_ALetter), ('\u{10050}', '\u{1005d}', WC_ALetter),
- ('\u{10080}', '\u{100fa}', WC_ALetter), ('\u{10140}', '\u{10174}', WC_ALetter),
- ('\u{101fd}', '\u{101fd}', WC_Extend), ('\u{10280}', '\u{1029c}', WC_ALetter), ('\u{102a0}',
- '\u{102d0}', WC_ALetter), ('\u{102e0}', '\u{102e0}', WC_Extend), ('\u{10300}', '\u{1031f}',
- WC_ALetter), ('\u{1032d}', '\u{1034a}', WC_ALetter), ('\u{10350}', '\u{10375}', WC_ALetter),
- ('\u{10376}', '\u{1037a}', WC_Extend), ('\u{10380}', '\u{1039d}', WC_ALetter), ('\u{103a0}',
- '\u{103c3}', WC_ALetter), ('\u{103c8}', '\u{103cf}', WC_ALetter), ('\u{103d1}', '\u{103d5}',
- WC_ALetter), ('\u{10400}', '\u{1049d}', WC_ALetter), ('\u{104a0}', '\u{104a9}', WC_Numeric),
- ('\u{104b0}', '\u{104d3}', WC_ALetter), ('\u{104d8}', '\u{104fb}', WC_ALetter),
- ('\u{10500}', '\u{10527}', WC_ALetter), ('\u{10530}', '\u{10563}', WC_ALetter),
- ('\u{10600}', '\u{10736}', WC_ALetter), ('\u{10740}', '\u{10755}', WC_ALetter),
- ('\u{10760}', '\u{10767}', WC_ALetter), ('\u{10800}', '\u{10805}', WC_ALetter),
- ('\u{10808}', '\u{10808}', WC_ALetter), ('\u{1080a}', '\u{10835}', WC_ALetter),
- ('\u{10837}', '\u{10838}', WC_ALetter), ('\u{1083c}', '\u{1083c}', WC_ALetter),
- ('\u{1083f}', '\u{10855}', WC_ALetter), ('\u{10860}', '\u{10876}', WC_ALetter),
- ('\u{10880}', '\u{1089e}', WC_ALetter), ('\u{108e0}', '\u{108f2}', WC_ALetter),
- ('\u{108f4}', '\u{108f5}', WC_ALetter), ('\u{10900}', '\u{10915}', WC_ALetter),
- ('\u{10920}', '\u{10939}', WC_ALetter), ('\u{10980}', '\u{109b7}', WC_ALetter),
- ('\u{109be}', '\u{109bf}', WC_ALetter), ('\u{10a00}', '\u{10a00}', WC_ALetter),
- ('\u{10a01}', '\u{10a03}', WC_Extend), ('\u{10a05}', '\u{10a06}', WC_Extend), ('\u{10a0c}',
- '\u{10a0f}', WC_Extend), ('\u{10a10}', '\u{10a13}', WC_ALetter), ('\u{10a15}', '\u{10a17}',
- WC_ALetter), ('\u{10a19}', '\u{10a35}', WC_ALetter), ('\u{10a38}', '\u{10a3a}', WC_Extend),
- ('\u{10a3f}', '\u{10a3f}', WC_Extend), ('\u{10a60}', '\u{10a7c}', WC_ALetter), ('\u{10a80}',
- '\u{10a9c}', WC_ALetter), ('\u{10ac0}', '\u{10ac7}', WC_ALetter), ('\u{10ac9}', '\u{10ae4}',
- WC_ALetter), ('\u{10ae5}', '\u{10ae6}', WC_Extend), ('\u{10b00}', '\u{10b35}', WC_ALetter),
- ('\u{10b40}', '\u{10b55}', WC_ALetter), ('\u{10b60}', '\u{10b72}', WC_ALetter),
- ('\u{10b80}', '\u{10b91}', WC_ALetter), ('\u{10c00}', '\u{10c48}', WC_ALetter),
- ('\u{10c80}', '\u{10cb2}', WC_ALetter), ('\u{10cc0}', '\u{10cf2}', WC_ALetter),
- ('\u{10d00}', '\u{10d23}', WC_ALetter), ('\u{10d24}', '\u{10d27}', WC_Extend), ('\u{10d30}',
- '\u{10d39}', WC_Numeric), ('\u{10f00}', '\u{10f1c}', WC_ALetter), ('\u{10f27}', '\u{10f27}',
- WC_ALetter), ('\u{10f30}', '\u{10f45}', WC_ALetter), ('\u{10f46}', '\u{10f50}', WC_Extend),
- ('\u{10fe0}', '\u{10ff6}', WC_ALetter), ('\u{11000}', '\u{11002}', WC_Extend), ('\u{11003}',
- '\u{11037}', WC_ALetter), ('\u{11038}', '\u{11046}', WC_Extend), ('\u{11066}', '\u{1106f}',
- WC_Numeric), ('\u{1107f}', '\u{11082}', WC_Extend), ('\u{11083}', '\u{110af}', WC_ALetter),
- ('\u{110b0}', '\u{110ba}', WC_Extend), ('\u{110bd}', '\u{110bd}', WC_Format), ('\u{110cd}',
- '\u{110cd}', WC_Format), ('\u{110d0}', '\u{110e8}', WC_ALetter), ('\u{110f0}', '\u{110f9}',
- WC_Numeric), ('\u{11100}', '\u{11102}', WC_Extend), ('\u{11103}', '\u{11126}', WC_ALetter),
- ('\u{11127}', '\u{11134}', WC_Extend), ('\u{11136}', '\u{1113f}', WC_Numeric), ('\u{11144}',
- '\u{11144}', WC_ALetter), ('\u{11145}', '\u{11146}', WC_Extend), ('\u{11150}', '\u{11172}',
- WC_ALetter), ('\u{11173}', '\u{11173}', WC_Extend), ('\u{11176}', '\u{11176}', WC_ALetter),
- ('\u{11180}', '\u{11182}', WC_Extend), ('\u{11183}', '\u{111b2}', WC_ALetter), ('\u{111b3}',
- '\u{111c0}', WC_Extend), ('\u{111c1}', '\u{111c4}', WC_ALetter), ('\u{111c9}', '\u{111cc}',
- WC_Extend), ('\u{111d0}', '\u{111d9}', WC_Numeric), ('\u{111da}', '\u{111da}', WC_ALetter),
- ('\u{111dc}', '\u{111dc}', WC_ALetter), ('\u{11200}', '\u{11211}', WC_ALetter),
- ('\u{11213}', '\u{1122b}', WC_ALetter), ('\u{1122c}', '\u{11237}', WC_Extend), ('\u{1123e}',
- '\u{1123e}', WC_Extend), ('\u{11280}', '\u{11286}', WC_ALetter), ('\u{11288}', '\u{11288}',
- WC_ALetter), ('\u{1128a}', '\u{1128d}', WC_ALetter), ('\u{1128f}', '\u{1129d}', WC_ALetter),
- ('\u{1129f}', '\u{112a8}', WC_ALetter), ('\u{112b0}', '\u{112de}', WC_ALetter),
+ ('\u{a823}', '\u{a827}', WC_Extend), ('\u{a82c}', '\u{a82c}', WC_Extend), ('\u{a840}',
+ '\u{a873}', WC_ALetter), ('\u{a880}', '\u{a881}', WC_Extend), ('\u{a882}', '\u{a8b3}',
+ WC_ALetter), ('\u{a8b4}', '\u{a8c5}', WC_Extend), ('\u{a8d0}', '\u{a8d9}', WC_Numeric),
+ ('\u{a8e0}', '\u{a8f1}', WC_Extend), ('\u{a8f2}', '\u{a8f7}', WC_ALetter), ('\u{a8fb}',
+ '\u{a8fb}', WC_ALetter), ('\u{a8fd}', '\u{a8fe}', WC_ALetter), ('\u{a8ff}', '\u{a8ff}',
+ WC_Extend), ('\u{a900}', '\u{a909}', WC_Numeric), ('\u{a90a}', '\u{a925}', WC_ALetter),
+ ('\u{a926}', '\u{a92d}', WC_Extend), ('\u{a930}', '\u{a946}', WC_ALetter), ('\u{a947}',
+ '\u{a953}', WC_Extend), ('\u{a960}', '\u{a97c}', WC_ALetter), ('\u{a980}', '\u{a983}',
+ WC_Extend), ('\u{a984}', '\u{a9b2}', WC_ALetter), ('\u{a9b3}', '\u{a9c0}', WC_Extend),
+ ('\u{a9cf}', '\u{a9cf}', WC_ALetter), ('\u{a9d0}', '\u{a9d9}', WC_Numeric), ('\u{a9e5}',
+ '\u{a9e5}', WC_Extend), ('\u{a9f0}', '\u{a9f9}', WC_Numeric), ('\u{aa00}', '\u{aa28}',
+ WC_ALetter), ('\u{aa29}', '\u{aa36}', WC_Extend), ('\u{aa40}', '\u{aa42}', WC_ALetter),
+ ('\u{aa43}', '\u{aa43}', WC_Extend), ('\u{aa44}', '\u{aa4b}', WC_ALetter), ('\u{aa4c}',
+ '\u{aa4d}', WC_Extend), ('\u{aa50}', '\u{aa59}', WC_Numeric), ('\u{aa7b}', '\u{aa7d}',
+ WC_Extend), ('\u{aab0}', '\u{aab0}', WC_Extend), ('\u{aab2}', '\u{aab4}', WC_Extend),
+ ('\u{aab7}', '\u{aab8}', WC_Extend), ('\u{aabe}', '\u{aabf}', WC_Extend), ('\u{aac1}',
+ '\u{aac1}', WC_Extend), ('\u{aae0}', '\u{aaea}', WC_ALetter), ('\u{aaeb}', '\u{aaef}',
+ WC_Extend), ('\u{aaf2}', '\u{aaf4}', WC_ALetter), ('\u{aaf5}', '\u{aaf6}', WC_Extend),
+ ('\u{ab01}', '\u{ab06}', WC_ALetter), ('\u{ab09}', '\u{ab0e}', WC_ALetter), ('\u{ab11}',
+ '\u{ab16}', WC_ALetter), ('\u{ab20}', '\u{ab26}', WC_ALetter), ('\u{ab28}', '\u{ab2e}',
+ WC_ALetter), ('\u{ab30}', '\u{ab69}', WC_ALetter), ('\u{ab70}', '\u{abe2}', WC_ALetter),
+ ('\u{abe3}', '\u{abea}', WC_Extend), ('\u{abec}', '\u{abed}', WC_Extend), ('\u{abf0}',
+ '\u{abf9}', WC_Numeric), ('\u{ac00}', '\u{d7a3}', WC_ALetter), ('\u{d7b0}', '\u{d7c6}',
+ WC_ALetter), ('\u{d7cb}', '\u{d7fb}', WC_ALetter), ('\u{fb00}', '\u{fb06}', WC_ALetter),
+ ('\u{fb13}', '\u{fb17}', WC_ALetter), ('\u{fb1d}', '\u{fb1d}', WC_Hebrew_Letter),
+ ('\u{fb1e}', '\u{fb1e}', WC_Extend), ('\u{fb1f}', '\u{fb28}', WC_Hebrew_Letter),
+ ('\u{fb2a}', '\u{fb36}', WC_Hebrew_Letter), ('\u{fb38}', '\u{fb3c}', WC_Hebrew_Letter),
+ ('\u{fb3e}', '\u{fb3e}', WC_Hebrew_Letter), ('\u{fb40}', '\u{fb41}', WC_Hebrew_Letter),
+ ('\u{fb43}', '\u{fb44}', WC_Hebrew_Letter), ('\u{fb46}', '\u{fb4f}', WC_Hebrew_Letter),
+ ('\u{fb50}', '\u{fbb1}', WC_ALetter), ('\u{fbd3}', '\u{fd3d}', WC_ALetter), ('\u{fd50}',
+ '\u{fd8f}', WC_ALetter), ('\u{fd92}', '\u{fdc7}', WC_ALetter), ('\u{fdf0}', '\u{fdfb}',
+ WC_ALetter), ('\u{fe00}', '\u{fe0f}', WC_Extend), ('\u{fe10}', '\u{fe10}', WC_MidNum),
+ ('\u{fe13}', '\u{fe13}', WC_MidLetter), ('\u{fe14}', '\u{fe14}', WC_MidNum), ('\u{fe20}',
+ '\u{fe2f}', WC_Extend), ('\u{fe33}', '\u{fe34}', WC_ExtendNumLet), ('\u{fe4d}', '\u{fe4f}',
+ WC_ExtendNumLet), ('\u{fe50}', '\u{fe50}', WC_MidNum), ('\u{fe52}', '\u{fe52}',
+ WC_MidNumLet), ('\u{fe54}', '\u{fe54}', WC_MidNum), ('\u{fe55}', '\u{fe55}', WC_MidLetter),
+ ('\u{fe70}', '\u{fe74}', WC_ALetter), ('\u{fe76}', '\u{fefc}', WC_ALetter), ('\u{feff}',
+ '\u{feff}', WC_Format), ('\u{ff07}', '\u{ff07}', WC_MidNumLet), ('\u{ff0c}', '\u{ff0c}',
+ WC_MidNum), ('\u{ff0e}', '\u{ff0e}', WC_MidNumLet), ('\u{ff10}', '\u{ff19}', WC_Numeric),
+ ('\u{ff1a}', '\u{ff1a}', WC_MidLetter), ('\u{ff1b}', '\u{ff1b}', WC_MidNum), ('\u{ff21}',
+ '\u{ff3a}', WC_ALetter), ('\u{ff3f}', '\u{ff3f}', WC_ExtendNumLet), ('\u{ff41}', '\u{ff5a}',
+ WC_ALetter), ('\u{ff66}', '\u{ff9d}', WC_Katakana), ('\u{ff9e}', '\u{ff9f}', WC_Extend),
+ ('\u{ffa0}', '\u{ffbe}', WC_ALetter), ('\u{ffc2}', '\u{ffc7}', WC_ALetter), ('\u{ffca}',
+ '\u{ffcf}', WC_ALetter), ('\u{ffd2}', '\u{ffd7}', WC_ALetter), ('\u{ffda}', '\u{ffdc}',
+ WC_ALetter), ('\u{fff9}', '\u{fffb}', WC_Format), ('\u{10000}', '\u{1000b}', WC_ALetter),
+ ('\u{1000d}', '\u{10026}', WC_ALetter), ('\u{10028}', '\u{1003a}', WC_ALetter),
+ ('\u{1003c}', '\u{1003d}', WC_ALetter), ('\u{1003f}', '\u{1004d}', WC_ALetter),
+ ('\u{10050}', '\u{1005d}', WC_ALetter), ('\u{10080}', '\u{100fa}', WC_ALetter),
+ ('\u{10140}', '\u{10174}', WC_ALetter), ('\u{101fd}', '\u{101fd}', WC_Extend), ('\u{10280}',
+ '\u{1029c}', WC_ALetter), ('\u{102a0}', '\u{102d0}', WC_ALetter), ('\u{102e0}', '\u{102e0}',
+ WC_Extend), ('\u{10300}', '\u{1031f}', WC_ALetter), ('\u{1032d}', '\u{1034a}', WC_ALetter),
+ ('\u{10350}', '\u{10375}', WC_ALetter), ('\u{10376}', '\u{1037a}', WC_Extend), ('\u{10380}',
+ '\u{1039d}', WC_ALetter), ('\u{103a0}', '\u{103c3}', WC_ALetter), ('\u{103c8}', '\u{103cf}',
+ WC_ALetter), ('\u{103d1}', '\u{103d5}', WC_ALetter), ('\u{10400}', '\u{1049d}', WC_ALetter),
+ ('\u{104a0}', '\u{104a9}', WC_Numeric), ('\u{104b0}', '\u{104d3}', WC_ALetter),
+ ('\u{104d8}', '\u{104fb}', WC_ALetter), ('\u{10500}', '\u{10527}', WC_ALetter),
+ ('\u{10530}', '\u{10563}', WC_ALetter), ('\u{10600}', '\u{10736}', WC_ALetter),
+ ('\u{10740}', '\u{10755}', WC_ALetter), ('\u{10760}', '\u{10767}', WC_ALetter),
+ ('\u{10800}', '\u{10805}', WC_ALetter), ('\u{10808}', '\u{10808}', WC_ALetter),
+ ('\u{1080a}', '\u{10835}', WC_ALetter), ('\u{10837}', '\u{10838}', WC_ALetter),
+ ('\u{1083c}', '\u{1083c}', WC_ALetter), ('\u{1083f}', '\u{10855}', WC_ALetter),
+ ('\u{10860}', '\u{10876}', WC_ALetter), ('\u{10880}', '\u{1089e}', WC_ALetter),
+ ('\u{108e0}', '\u{108f2}', WC_ALetter), ('\u{108f4}', '\u{108f5}', WC_ALetter),
+ ('\u{10900}', '\u{10915}', WC_ALetter), ('\u{10920}', '\u{10939}', WC_ALetter),
+ ('\u{10980}', '\u{109b7}', WC_ALetter), ('\u{109be}', '\u{109bf}', WC_ALetter),
+ ('\u{10a00}', '\u{10a00}', WC_ALetter), ('\u{10a01}', '\u{10a03}', WC_Extend), ('\u{10a05}',
+ '\u{10a06}', WC_Extend), ('\u{10a0c}', '\u{10a0f}', WC_Extend), ('\u{10a10}', '\u{10a13}',
+ WC_ALetter), ('\u{10a15}', '\u{10a17}', WC_ALetter), ('\u{10a19}', '\u{10a35}', WC_ALetter),
+ ('\u{10a38}', '\u{10a3a}', WC_Extend), ('\u{10a3f}', '\u{10a3f}', WC_Extend), ('\u{10a60}',
+ '\u{10a7c}', WC_ALetter), ('\u{10a80}', '\u{10a9c}', WC_ALetter), ('\u{10ac0}', '\u{10ac7}',
+ WC_ALetter), ('\u{10ac9}', '\u{10ae4}', WC_ALetter), ('\u{10ae5}', '\u{10ae6}', WC_Extend),
+ ('\u{10b00}', '\u{10b35}', WC_ALetter), ('\u{10b40}', '\u{10b55}', WC_ALetter),
+ ('\u{10b60}', '\u{10b72}', WC_ALetter), ('\u{10b80}', '\u{10b91}', WC_ALetter),
+ ('\u{10c00}', '\u{10c48}', WC_ALetter), ('\u{10c80}', '\u{10cb2}', WC_ALetter),
+ ('\u{10cc0}', '\u{10cf2}', WC_ALetter), ('\u{10d00}', '\u{10d23}', WC_ALetter),
+ ('\u{10d24}', '\u{10d27}', WC_Extend), ('\u{10d30}', '\u{10d39}', WC_Numeric), ('\u{10e80}',
+ '\u{10ea9}', WC_ALetter), ('\u{10eab}', '\u{10eac}', WC_Extend), ('\u{10eb0}', '\u{10eb1}',
+ WC_ALetter), ('\u{10f00}', '\u{10f1c}', WC_ALetter), ('\u{10f27}', '\u{10f27}', WC_ALetter),
+ ('\u{10f30}', '\u{10f45}', WC_ALetter), ('\u{10f46}', '\u{10f50}', WC_Extend), ('\u{10fb0}',
+ '\u{10fc4}', WC_ALetter), ('\u{10fe0}', '\u{10ff6}', WC_ALetter), ('\u{11000}', '\u{11002}',
+ WC_Extend), ('\u{11003}', '\u{11037}', WC_ALetter), ('\u{11038}', '\u{11046}', WC_Extend),
+ ('\u{11066}', '\u{1106f}', WC_Numeric), ('\u{1107f}', '\u{11082}', WC_Extend), ('\u{11083}',
+ '\u{110af}', WC_ALetter), ('\u{110b0}', '\u{110ba}', WC_Extend), ('\u{110bd}', '\u{110bd}',
+ WC_Format), ('\u{110cd}', '\u{110cd}', WC_Format), ('\u{110d0}', '\u{110e8}', WC_ALetter),
+ ('\u{110f0}', '\u{110f9}', WC_Numeric), ('\u{11100}', '\u{11102}', WC_Extend), ('\u{11103}',
+ '\u{11126}', WC_ALetter), ('\u{11127}', '\u{11134}', WC_Extend), ('\u{11136}', '\u{1113f}',
+ WC_Numeric), ('\u{11144}', '\u{11144}', WC_ALetter), ('\u{11145}', '\u{11146}', WC_Extend),
+ ('\u{11147}', '\u{11147}', WC_ALetter), ('\u{11150}', '\u{11172}', WC_ALetter),
+ ('\u{11173}', '\u{11173}', WC_Extend), ('\u{11176}', '\u{11176}', WC_ALetter), ('\u{11180}',
+ '\u{11182}', WC_Extend), ('\u{11183}', '\u{111b2}', WC_ALetter), ('\u{111b3}', '\u{111c0}',
+ WC_Extend), ('\u{111c1}', '\u{111c4}', WC_ALetter), ('\u{111c9}', '\u{111cc}', WC_Extend),
+ ('\u{111ce}', '\u{111cf}', WC_Extend), ('\u{111d0}', '\u{111d9}', WC_Numeric), ('\u{111da}',
+ '\u{111da}', WC_ALetter), ('\u{111dc}', '\u{111dc}', WC_ALetter), ('\u{11200}', '\u{11211}',
+ WC_ALetter), ('\u{11213}', '\u{1122b}', WC_ALetter), ('\u{1122c}', '\u{11237}', WC_Extend),
+ ('\u{1123e}', '\u{1123e}', WC_Extend), ('\u{11280}', '\u{11286}', WC_ALetter), ('\u{11288}',
+ '\u{11288}', WC_ALetter), ('\u{1128a}', '\u{1128d}', WC_ALetter), ('\u{1128f}', '\u{1129d}',
+ WC_ALetter), ('\u{1129f}', '\u{112a8}', WC_ALetter), ('\u{112b0}', '\u{112de}', WC_ALetter),
('\u{112df}', '\u{112ea}', WC_Extend), ('\u{112f0}', '\u{112f9}', WC_Numeric), ('\u{11300}',
'\u{11303}', WC_Extend), ('\u{11305}', '\u{1130c}', WC_ALetter), ('\u{1130f}', '\u{11310}',
WC_ALetter), ('\u{11313}', '\u{11328}', WC_ALetter), ('\u{1132a}', '\u{11330}', WC_ALetter),
@@ -1326,7 +1358,7 @@
'\u{1136c}', WC_Extend), ('\u{11370}', '\u{11374}', WC_Extend), ('\u{11400}', '\u{11434}',
WC_ALetter), ('\u{11435}', '\u{11446}', WC_Extend), ('\u{11447}', '\u{1144a}', WC_ALetter),
('\u{11450}', '\u{11459}', WC_Numeric), ('\u{1145e}', '\u{1145e}', WC_Extend), ('\u{1145f}',
- '\u{1145f}', WC_ALetter), ('\u{11480}', '\u{114af}', WC_ALetter), ('\u{114b0}', '\u{114c3}',
+ '\u{11461}', WC_ALetter), ('\u{11480}', '\u{114af}', WC_ALetter), ('\u{114b0}', '\u{114c3}',
WC_Extend), ('\u{114c4}', '\u{114c5}', WC_ALetter), ('\u{114c7}', '\u{114c7}', WC_ALetter),
('\u{114d0}', '\u{114d9}', WC_Numeric), ('\u{11580}', '\u{115ae}', WC_ALetter),
('\u{115af}', '\u{115b5}', WC_Extend), ('\u{115b8}', '\u{115c0}', WC_Extend), ('\u{115d8}',
@@ -1337,49 +1369,55 @@
'\u{116c9}', WC_Numeric), ('\u{1171d}', '\u{1172b}', WC_Extend), ('\u{11730}', '\u{11739}',
WC_Numeric), ('\u{11800}', '\u{1182b}', WC_ALetter), ('\u{1182c}', '\u{1183a}', WC_Extend),
('\u{118a0}', '\u{118df}', WC_ALetter), ('\u{118e0}', '\u{118e9}', WC_Numeric),
- ('\u{118ff}', '\u{118ff}', WC_ALetter), ('\u{119a0}', '\u{119a7}', WC_ALetter),
- ('\u{119aa}', '\u{119d0}', WC_ALetter), ('\u{119d1}', '\u{119d7}', WC_Extend), ('\u{119da}',
- '\u{119e0}', WC_Extend), ('\u{119e1}', '\u{119e1}', WC_ALetter), ('\u{119e3}', '\u{119e3}',
- WC_ALetter), ('\u{119e4}', '\u{119e4}', WC_Extend), ('\u{11a00}', '\u{11a00}', WC_ALetter),
- ('\u{11a01}', '\u{11a0a}', WC_Extend), ('\u{11a0b}', '\u{11a32}', WC_ALetter), ('\u{11a33}',
- '\u{11a39}', WC_Extend), ('\u{11a3a}', '\u{11a3a}', WC_ALetter), ('\u{11a3b}', '\u{11a3e}',
- WC_Extend), ('\u{11a47}', '\u{11a47}', WC_Extend), ('\u{11a50}', '\u{11a50}', WC_ALetter),
- ('\u{11a51}', '\u{11a5b}', WC_Extend), ('\u{11a5c}', '\u{11a89}', WC_ALetter), ('\u{11a8a}',
- '\u{11a99}', WC_Extend), ('\u{11a9d}', '\u{11a9d}', WC_ALetter), ('\u{11ac0}', '\u{11af8}',
- WC_ALetter), ('\u{11c00}', '\u{11c08}', WC_ALetter), ('\u{11c0a}', '\u{11c2e}', WC_ALetter),
- ('\u{11c2f}', '\u{11c36}', WC_Extend), ('\u{11c38}', '\u{11c3f}', WC_Extend), ('\u{11c40}',
- '\u{11c40}', WC_ALetter), ('\u{11c50}', '\u{11c59}', WC_Numeric), ('\u{11c72}', '\u{11c8f}',
- WC_ALetter), ('\u{11c92}', '\u{11ca7}', WC_Extend), ('\u{11ca9}', '\u{11cb6}', WC_Extend),
- ('\u{11d00}', '\u{11d06}', WC_ALetter), ('\u{11d08}', '\u{11d09}', WC_ALetter),
- ('\u{11d0b}', '\u{11d30}', WC_ALetter), ('\u{11d31}', '\u{11d36}', WC_Extend), ('\u{11d3a}',
- '\u{11d3a}', WC_Extend), ('\u{11d3c}', '\u{11d3d}', WC_Extend), ('\u{11d3f}', '\u{11d45}',
- WC_Extend), ('\u{11d46}', '\u{11d46}', WC_ALetter), ('\u{11d47}', '\u{11d47}', WC_Extend),
- ('\u{11d50}', '\u{11d59}', WC_Numeric), ('\u{11d60}', '\u{11d65}', WC_ALetter),
- ('\u{11d67}', '\u{11d68}', WC_ALetter), ('\u{11d6a}', '\u{11d89}', WC_ALetter),
+ ('\u{118ff}', '\u{11906}', WC_ALetter), ('\u{11909}', '\u{11909}', WC_ALetter),
+ ('\u{1190c}', '\u{11913}', WC_ALetter), ('\u{11915}', '\u{11916}', WC_ALetter),
+ ('\u{11918}', '\u{1192f}', WC_ALetter), ('\u{11930}', '\u{11935}', WC_Extend), ('\u{11937}',
+ '\u{11938}', WC_Extend), ('\u{1193b}', '\u{1193e}', WC_Extend), ('\u{1193f}', '\u{1193f}',
+ WC_ALetter), ('\u{11940}', '\u{11940}', WC_Extend), ('\u{11941}', '\u{11941}', WC_ALetter),
+ ('\u{11942}', '\u{11943}', WC_Extend), ('\u{11950}', '\u{11959}', WC_Numeric), ('\u{119a0}',
+ '\u{119a7}', WC_ALetter), ('\u{119aa}', '\u{119d0}', WC_ALetter), ('\u{119d1}', '\u{119d7}',
+ WC_Extend), ('\u{119da}', '\u{119e0}', WC_Extend), ('\u{119e1}', '\u{119e1}', WC_ALetter),
+ ('\u{119e3}', '\u{119e3}', WC_ALetter), ('\u{119e4}', '\u{119e4}', WC_Extend), ('\u{11a00}',
+ '\u{11a00}', WC_ALetter), ('\u{11a01}', '\u{11a0a}', WC_Extend), ('\u{11a0b}', '\u{11a32}',
+ WC_ALetter), ('\u{11a33}', '\u{11a39}', WC_Extend), ('\u{11a3a}', '\u{11a3a}', WC_ALetter),
+ ('\u{11a3b}', '\u{11a3e}', WC_Extend), ('\u{11a47}', '\u{11a47}', WC_Extend), ('\u{11a50}',
+ '\u{11a50}', WC_ALetter), ('\u{11a51}', '\u{11a5b}', WC_Extend), ('\u{11a5c}', '\u{11a89}',
+ WC_ALetter), ('\u{11a8a}', '\u{11a99}', WC_Extend), ('\u{11a9d}', '\u{11a9d}', WC_ALetter),
+ ('\u{11ac0}', '\u{11af8}', WC_ALetter), ('\u{11c00}', '\u{11c08}', WC_ALetter),
+ ('\u{11c0a}', '\u{11c2e}', WC_ALetter), ('\u{11c2f}', '\u{11c36}', WC_Extend), ('\u{11c38}',
+ '\u{11c3f}', WC_Extend), ('\u{11c40}', '\u{11c40}', WC_ALetter), ('\u{11c50}', '\u{11c59}',
+ WC_Numeric), ('\u{11c72}', '\u{11c8f}', WC_ALetter), ('\u{11c92}', '\u{11ca7}', WC_Extend),
+ ('\u{11ca9}', '\u{11cb6}', WC_Extend), ('\u{11d00}', '\u{11d06}', WC_ALetter), ('\u{11d08}',
+ '\u{11d09}', WC_ALetter), ('\u{11d0b}', '\u{11d30}', WC_ALetter), ('\u{11d31}', '\u{11d36}',
+ WC_Extend), ('\u{11d3a}', '\u{11d3a}', WC_Extend), ('\u{11d3c}', '\u{11d3d}', WC_Extend),
+ ('\u{11d3f}', '\u{11d45}', WC_Extend), ('\u{11d46}', '\u{11d46}', WC_ALetter), ('\u{11d47}',
+ '\u{11d47}', WC_Extend), ('\u{11d50}', '\u{11d59}', WC_Numeric), ('\u{11d60}', '\u{11d65}',
+ WC_ALetter), ('\u{11d67}', '\u{11d68}', WC_ALetter), ('\u{11d6a}', '\u{11d89}', WC_ALetter),
('\u{11d8a}', '\u{11d8e}', WC_Extend), ('\u{11d90}', '\u{11d91}', WC_Extend), ('\u{11d93}',
'\u{11d97}', WC_Extend), ('\u{11d98}', '\u{11d98}', WC_ALetter), ('\u{11da0}', '\u{11da9}',
WC_Numeric), ('\u{11ee0}', '\u{11ef2}', WC_ALetter), ('\u{11ef3}', '\u{11ef6}', WC_Extend),
- ('\u{12000}', '\u{12399}', WC_ALetter), ('\u{12400}', '\u{1246e}', WC_ALetter),
- ('\u{12480}', '\u{12543}', WC_ALetter), ('\u{13000}', '\u{1342e}', WC_ALetter),
- ('\u{13430}', '\u{13438}', WC_Format), ('\u{14400}', '\u{14646}', WC_ALetter), ('\u{16800}',
- '\u{16a38}', WC_ALetter), ('\u{16a40}', '\u{16a5e}', WC_ALetter), ('\u{16a60}', '\u{16a69}',
- WC_Numeric), ('\u{16ad0}', '\u{16aed}', WC_ALetter), ('\u{16af0}', '\u{16af4}', WC_Extend),
- ('\u{16b00}', '\u{16b2f}', WC_ALetter), ('\u{16b30}', '\u{16b36}', WC_Extend), ('\u{16b40}',
- '\u{16b43}', WC_ALetter), ('\u{16b50}', '\u{16b59}', WC_Numeric), ('\u{16b63}', '\u{16b77}',
- WC_ALetter), ('\u{16b7d}', '\u{16b8f}', WC_ALetter), ('\u{16e40}', '\u{16e7f}', WC_ALetter),
- ('\u{16f00}', '\u{16f4a}', WC_ALetter), ('\u{16f4f}', '\u{16f4f}', WC_Extend), ('\u{16f50}',
- '\u{16f50}', WC_ALetter), ('\u{16f51}', '\u{16f87}', WC_Extend), ('\u{16f8f}', '\u{16f92}',
- WC_Extend), ('\u{16f93}', '\u{16f9f}', WC_ALetter), ('\u{16fe0}', '\u{16fe1}', WC_ALetter),
- ('\u{16fe3}', '\u{16fe3}', WC_ALetter), ('\u{1b000}', '\u{1b000}', WC_Katakana),
- ('\u{1b164}', '\u{1b167}', WC_Katakana), ('\u{1bc00}', '\u{1bc6a}', WC_ALetter),
- ('\u{1bc70}', '\u{1bc7c}', WC_ALetter), ('\u{1bc80}', '\u{1bc88}', WC_ALetter),
- ('\u{1bc90}', '\u{1bc99}', WC_ALetter), ('\u{1bc9d}', '\u{1bc9e}', WC_Extend), ('\u{1bca0}',
- '\u{1bca3}', WC_Format), ('\u{1d165}', '\u{1d169}', WC_Extend), ('\u{1d16d}', '\u{1d172}',
- WC_Extend), ('\u{1d173}', '\u{1d17a}', WC_Format), ('\u{1d17b}', '\u{1d182}', WC_Extend),
- ('\u{1d185}', '\u{1d18b}', WC_Extend), ('\u{1d1aa}', '\u{1d1ad}', WC_Extend), ('\u{1d242}',
- '\u{1d244}', WC_Extend), ('\u{1d400}', '\u{1d454}', WC_ALetter), ('\u{1d456}', '\u{1d49c}',
- WC_ALetter), ('\u{1d49e}', '\u{1d49f}', WC_ALetter), ('\u{1d4a2}', '\u{1d4a2}', WC_ALetter),
- ('\u{1d4a5}', '\u{1d4a6}', WC_ALetter), ('\u{1d4a9}', '\u{1d4ac}', WC_ALetter),
+ ('\u{11fb0}', '\u{11fb0}', WC_ALetter), ('\u{12000}', '\u{12399}', WC_ALetter),
+ ('\u{12400}', '\u{1246e}', WC_ALetter), ('\u{12480}', '\u{12543}', WC_ALetter),
+ ('\u{13000}', '\u{1342e}', WC_ALetter), ('\u{13430}', '\u{13438}', WC_Format), ('\u{14400}',
+ '\u{14646}', WC_ALetter), ('\u{16800}', '\u{16a38}', WC_ALetter), ('\u{16a40}', '\u{16a5e}',
+ WC_ALetter), ('\u{16a60}', '\u{16a69}', WC_Numeric), ('\u{16ad0}', '\u{16aed}', WC_ALetter),
+ ('\u{16af0}', '\u{16af4}', WC_Extend), ('\u{16b00}', '\u{16b2f}', WC_ALetter), ('\u{16b30}',
+ '\u{16b36}', WC_Extend), ('\u{16b40}', '\u{16b43}', WC_ALetter), ('\u{16b50}', '\u{16b59}',
+ WC_Numeric), ('\u{16b63}', '\u{16b77}', WC_ALetter), ('\u{16b7d}', '\u{16b8f}', WC_ALetter),
+ ('\u{16e40}', '\u{16e7f}', WC_ALetter), ('\u{16f00}', '\u{16f4a}', WC_ALetter),
+ ('\u{16f4f}', '\u{16f4f}', WC_Extend), ('\u{16f50}', '\u{16f50}', WC_ALetter), ('\u{16f51}',
+ '\u{16f87}', WC_Extend), ('\u{16f8f}', '\u{16f92}', WC_Extend), ('\u{16f93}', '\u{16f9f}',
+ WC_ALetter), ('\u{16fe0}', '\u{16fe1}', WC_ALetter), ('\u{16fe3}', '\u{16fe3}', WC_ALetter),
+ ('\u{16fe4}', '\u{16fe4}', WC_Extend), ('\u{16ff0}', '\u{16ff1}', WC_Extend), ('\u{1b000}',
+ '\u{1b000}', WC_Katakana), ('\u{1b164}', '\u{1b167}', WC_Katakana), ('\u{1bc00}',
+ '\u{1bc6a}', WC_ALetter), ('\u{1bc70}', '\u{1bc7c}', WC_ALetter), ('\u{1bc80}', '\u{1bc88}',
+ WC_ALetter), ('\u{1bc90}', '\u{1bc99}', WC_ALetter), ('\u{1bc9d}', '\u{1bc9e}', WC_Extend),
+ ('\u{1bca0}', '\u{1bca3}', WC_Format), ('\u{1d165}', '\u{1d169}', WC_Extend), ('\u{1d16d}',
+ '\u{1d172}', WC_Extend), ('\u{1d173}', '\u{1d17a}', WC_Format), ('\u{1d17b}', '\u{1d182}',
+ WC_Extend), ('\u{1d185}', '\u{1d18b}', WC_Extend), ('\u{1d1aa}', '\u{1d1ad}', WC_Extend),
+ ('\u{1d242}', '\u{1d244}', WC_Extend), ('\u{1d400}', '\u{1d454}', WC_ALetter), ('\u{1d456}',
+ '\u{1d49c}', WC_ALetter), ('\u{1d49e}', '\u{1d49f}', WC_ALetter), ('\u{1d4a2}', '\u{1d4a2}',
+ WC_ALetter), ('\u{1d4a5}', '\u{1d4a6}', WC_ALetter), ('\u{1d4a9}', '\u{1d4ac}', WC_ALetter),
('\u{1d4ae}', '\u{1d4b9}', WC_ALetter), ('\u{1d4bb}', '\u{1d4bb}', WC_ALetter),
('\u{1d4bd}', '\u{1d4c3}', WC_ALetter), ('\u{1d4c5}', '\u{1d505}', WC_ALetter),
('\u{1d507}', '\u{1d50a}', WC_ALetter), ('\u{1d50d}', '\u{1d514}', WC_ALetter),
@@ -1422,8 +1460,9 @@
('\u{1eea5}', '\u{1eea9}', WC_ALetter), ('\u{1eeab}', '\u{1eebb}', WC_ALetter),
('\u{1f130}', '\u{1f149}', WC_ALetter), ('\u{1f150}', '\u{1f169}', WC_ALetter),
('\u{1f170}', '\u{1f189}', WC_ALetter), ('\u{1f1e6}', '\u{1f1ff}', WC_Regional_Indicator),
- ('\u{1f3fb}', '\u{1f3ff}', WC_Extend), ('\u{e0001}', '\u{e0001}', WC_Format), ('\u{e0020}',
- '\u{e007f}', WC_Extend), ('\u{e0100}', '\u{e01ef}', WC_Extend)
+ ('\u{1f3fb}', '\u{1f3ff}', WC_Extend), ('\u{1fbf0}', '\u{1fbf9}', WC_Numeric), ('\u{e0001}',
+ '\u{e0001}', WC_Format), ('\u{e0020}', '\u{e007f}', WC_Extend), ('\u{e0100}', '\u{e01ef}',
+ WC_Extend)
];
}
@@ -1439,7 +1478,8 @@
EC_Extended_Pictographic,
}
- fn bsearch_range_value_table(c: char, r: &'static [(char, char, EmojiCat)]) -> EmojiCat {
+ fn bsearch_range_value_table(c: char, r: &'static [(char, char, EmojiCat)]) -> (u32, u32, EmojiCat) {
+ use core;
use core::cmp::Ordering::{Equal, Less, Greater};
match r.binary_search_by(|&(lo, hi, _)| {
if lo <= c && c <= hi { Equal }
@@ -1447,14 +1487,20 @@
else { Greater }
}) {
Ok(idx) => {
- let (_, _, cat) = r[idx];
- cat
+ let (lower, upper, cat) = r[idx];
+ (lower as u32, upper as u32, cat)
}
- Err(_) => EC_Any
+ Err(idx) => {
+ (
+ if idx > 0 { r[idx-1].1 as u32 + 1 } else { 0 },
+ r.get(idx).map(|c|c.0 as u32 - 1).unwrap_or(core::u32::MAX),
+ EC_Any,
+ )
+ }
}
}
- pub fn emoji_category(c: char) -> EmojiCat {
+ pub fn emoji_category(c: char) -> (u32, u32, EmojiCat) {
bsearch_range_value_table(c, emoji_cat_table)
}
@@ -1506,7 +1552,8 @@
('\u{1f85a}', '\u{1f85f}', EC_Extended_Pictographic), ('\u{1f888}', '\u{1f88f}',
EC_Extended_Pictographic), ('\u{1f8ae}', '\u{1f8ff}', EC_Extended_Pictographic),
('\u{1f90c}', '\u{1f93a}', EC_Extended_Pictographic), ('\u{1f93c}', '\u{1f945}',
- EC_Extended_Pictographic), ('\u{1f947}', '\u{1fffd}', EC_Extended_Pictographic)
+ EC_Extended_Pictographic), ('\u{1f947}', '\u{1faff}', EC_Extended_Pictographic),
+ ('\u{1fc00}', '\u{1fffd}', EC_Extended_Pictographic)
];
}
@@ -1535,7 +1582,8 @@
SC_Upper,
}
- fn bsearch_range_value_table(c: char, r: &'static [(char, char, SentenceCat)]) -> SentenceCat {
+ fn bsearch_range_value_table(c: char, r: &'static [(char, char, SentenceCat)]) -> (u32, u32, SentenceCat) {
+ use core;
use core::cmp::Ordering::{Equal, Less, Greater};
match r.binary_search_by(|&(lo, hi, _)| {
if lo <= c && c <= hi { Equal }
@@ -1543,14 +1591,20 @@
else { Greater }
}) {
Ok(idx) => {
- let (_, _, cat) = r[idx];
- cat
+ let (lower, upper, cat) = r[idx];
+ (lower as u32, upper as u32, cat)
}
- Err(_) => SC_Any
+ Err(idx) => {
+ (
+ if idx > 0 { r[idx-1].1 as u32 + 1 } else { 0 },
+ r.get(idx).map(|c|c.0 as u32 - 1).unwrap_or(core::u32::MAX),
+ SC_Any,
+ )
+ }
}
}
- pub fn sentence_category(c: char) -> SentenceCat {
+ pub fn sentence_category(c: char) -> (u32, u32, SentenceCat) {
bsearch_range_value_table(c, sentence_cat_table)
}
@@ -1799,7 +1853,7 @@
('\u{829}', '\u{82d}', SC_Extend), ('\u{837}', '\u{837}', SC_STerm), ('\u{839}', '\u{839}',
SC_STerm), ('\u{83d}', '\u{83e}', SC_STerm), ('\u{840}', '\u{858}', SC_OLetter), ('\u{859}',
'\u{85b}', SC_Extend), ('\u{860}', '\u{86a}', SC_OLetter), ('\u{8a0}', '\u{8b4}',
- SC_OLetter), ('\u{8b6}', '\u{8bd}', SC_OLetter), ('\u{8d3}', '\u{8e1}', SC_Extend),
+ SC_OLetter), ('\u{8b6}', '\u{8c7}', SC_OLetter), ('\u{8d3}', '\u{8e1}', SC_Extend),
('\u{8e2}', '\u{8e2}', SC_Format), ('\u{8e3}', '\u{903}', SC_Extend), ('\u{904}', '\u{939}',
SC_OLetter), ('\u{93a}', '\u{93c}', SC_Extend), ('\u{93d}', '\u{93d}', SC_OLetter),
('\u{93e}', '\u{94f}', SC_Extend), ('\u{950}', '\u{950}', SC_OLetter), ('\u{951}',
@@ -1835,7 +1889,7 @@
SC_OLetter), ('\u{b32}', '\u{b33}', SC_OLetter), ('\u{b35}', '\u{b39}', SC_OLetter),
('\u{b3c}', '\u{b3c}', SC_Extend), ('\u{b3d}', '\u{b3d}', SC_OLetter), ('\u{b3e}',
'\u{b44}', SC_Extend), ('\u{b47}', '\u{b48}', SC_Extend), ('\u{b4b}', '\u{b4d}', SC_Extend),
- ('\u{b56}', '\u{b57}', SC_Extend), ('\u{b5c}', '\u{b5d}', SC_OLetter), ('\u{b5f}',
+ ('\u{b55}', '\u{b57}', SC_Extend), ('\u{b5c}', '\u{b5d}', SC_OLetter), ('\u{b5f}',
'\u{b61}', SC_OLetter), ('\u{b62}', '\u{b63}', SC_Extend), ('\u{b66}', '\u{b6f}',
SC_Numeric), ('\u{b71}', '\u{b71}', SC_OLetter), ('\u{b82}', '\u{b82}', SC_Extend),
('\u{b83}', '\u{b83}', SC_OLetter), ('\u{b85}', '\u{b8a}', SC_OLetter), ('\u{b8e}',
@@ -1858,14 +1912,14 @@
('\u{cc6}', '\u{cc8}', SC_Extend), ('\u{cca}', '\u{ccd}', SC_Extend), ('\u{cd5}', '\u{cd6}',
SC_Extend), ('\u{cde}', '\u{cde}', SC_OLetter), ('\u{ce0}', '\u{ce1}', SC_OLetter),
('\u{ce2}', '\u{ce3}', SC_Extend), ('\u{ce6}', '\u{cef}', SC_Numeric), ('\u{cf1}',
- '\u{cf2}', SC_OLetter), ('\u{d00}', '\u{d03}', SC_Extend), ('\u{d05}', '\u{d0c}',
+ '\u{cf2}', SC_OLetter), ('\u{d00}', '\u{d03}', SC_Extend), ('\u{d04}', '\u{d0c}',
SC_OLetter), ('\u{d0e}', '\u{d10}', SC_OLetter), ('\u{d12}', '\u{d3a}', SC_OLetter),
('\u{d3b}', '\u{d3c}', SC_Extend), ('\u{d3d}', '\u{d3d}', SC_OLetter), ('\u{d3e}',
'\u{d44}', SC_Extend), ('\u{d46}', '\u{d48}', SC_Extend), ('\u{d4a}', '\u{d4d}', SC_Extend),
('\u{d4e}', '\u{d4e}', SC_OLetter), ('\u{d54}', '\u{d56}', SC_OLetter), ('\u{d57}',
'\u{d57}', SC_Extend), ('\u{d5f}', '\u{d61}', SC_OLetter), ('\u{d62}', '\u{d63}',
SC_Extend), ('\u{d66}', '\u{d6f}', SC_Numeric), ('\u{d7a}', '\u{d7f}', SC_OLetter),
- ('\u{d82}', '\u{d83}', SC_Extend), ('\u{d85}', '\u{d96}', SC_OLetter), ('\u{d9a}',
+ ('\u{d81}', '\u{d83}', SC_Extend), ('\u{d85}', '\u{d96}', SC_OLetter), ('\u{d9a}',
'\u{db1}', SC_OLetter), ('\u{db3}', '\u{dbb}', SC_OLetter), ('\u{dbd}', '\u{dbd}',
SC_OLetter), ('\u{dc0}', '\u{dc6}', SC_OLetter), ('\u{dca}', '\u{dca}', SC_Extend),
('\u{dcf}', '\u{dd4}', SC_Extend), ('\u{dd6}', '\u{dd6}', SC_Extend), ('\u{dd8}', '\u{ddf}',
@@ -1931,7 +1985,7 @@
('\u{1a20}', '\u{1a54}', SC_OLetter), ('\u{1a55}', '\u{1a5e}', SC_Extend), ('\u{1a60}',
'\u{1a7c}', SC_Extend), ('\u{1a7f}', '\u{1a7f}', SC_Extend), ('\u{1a80}', '\u{1a89}',
SC_Numeric), ('\u{1a90}', '\u{1a99}', SC_Numeric), ('\u{1aa7}', '\u{1aa7}', SC_OLetter),
- ('\u{1aa8}', '\u{1aab}', SC_STerm), ('\u{1ab0}', '\u{1abe}', SC_Extend), ('\u{1b00}',
+ ('\u{1aa8}', '\u{1aab}', SC_STerm), ('\u{1ab0}', '\u{1ac0}', SC_Extend), ('\u{1b00}',
'\u{1b04}', SC_Extend), ('\u{1b05}', '\u{1b33}', SC_OLetter), ('\u{1b34}', '\u{1b44}',
SC_Extend), ('\u{1b45}', '\u{1b4b}', SC_OLetter), ('\u{1b50}', '\u{1b59}', SC_Numeric),
('\u{1b5a}', '\u{1b5b}', SC_STerm), ('\u{1b5e}', '\u{1b5f}', SC_STerm), ('\u{1b6b}',
@@ -2168,9 +2222,9 @@
SC_OLetter), ('\u{3038}', '\u{303c}', SC_OLetter), ('\u{3041}', '\u{3096}', SC_OLetter),
('\u{3099}', '\u{309a}', SC_Extend), ('\u{309d}', '\u{309f}', SC_OLetter), ('\u{30a1}',
'\u{30fa}', SC_OLetter), ('\u{30fc}', '\u{30ff}', SC_OLetter), ('\u{3105}', '\u{312f}',
- SC_OLetter), ('\u{3131}', '\u{318e}', SC_OLetter), ('\u{31a0}', '\u{31ba}', SC_OLetter),
- ('\u{31f0}', '\u{31ff}', SC_OLetter), ('\u{3400}', '\u{4db5}', SC_OLetter), ('\u{4e00}',
- '\u{9fef}', SC_OLetter), ('\u{a000}', '\u{a48c}', SC_OLetter), ('\u{a4d0}', '\u{a4fd}',
+ SC_OLetter), ('\u{3131}', '\u{318e}', SC_OLetter), ('\u{31a0}', '\u{31bf}', SC_OLetter),
+ ('\u{31f0}', '\u{31ff}', SC_OLetter), ('\u{3400}', '\u{4dbf}', SC_OLetter), ('\u{4e00}',
+ '\u{9ffc}', SC_OLetter), ('\u{a000}', '\u{a48c}', SC_OLetter), ('\u{a4d0}', '\u{a4fd}',
SC_OLetter), ('\u{a4ff}', '\u{a4ff}', SC_STerm), ('\u{a500}', '\u{a60c}', SC_OLetter),
('\u{a60e}', '\u{a60f}', SC_STerm), ('\u{a610}', '\u{a61f}', SC_OLetter), ('\u{a620}',
'\u{a629}', SC_Numeric), ('\u{a62a}', '\u{a62b}', SC_OLetter), ('\u{a640}', '\u{a640}',
@@ -2268,39 +2322,42 @@
('\u{a7bb}', '\u{a7bb}', SC_Lower), ('\u{a7bc}', '\u{a7bc}', SC_Upper), ('\u{a7bd}',
'\u{a7bd}', SC_Lower), ('\u{a7be}', '\u{a7be}', SC_Upper), ('\u{a7bf}', '\u{a7bf}',
SC_Lower), ('\u{a7c2}', '\u{a7c2}', SC_Upper), ('\u{a7c3}', '\u{a7c3}', SC_Lower),
- ('\u{a7c4}', '\u{a7c6}', SC_Upper), ('\u{a7f7}', '\u{a7f7}', SC_OLetter), ('\u{a7f8}',
- '\u{a7fa}', SC_Lower), ('\u{a7fb}', '\u{a801}', SC_OLetter), ('\u{a802}', '\u{a802}',
- SC_Extend), ('\u{a803}', '\u{a805}', SC_OLetter), ('\u{a806}', '\u{a806}', SC_Extend),
- ('\u{a807}', '\u{a80a}', SC_OLetter), ('\u{a80b}', '\u{a80b}', SC_Extend), ('\u{a80c}',
- '\u{a822}', SC_OLetter), ('\u{a823}', '\u{a827}', SC_Extend), ('\u{a840}', '\u{a873}',
- SC_OLetter), ('\u{a876}', '\u{a877}', SC_STerm), ('\u{a880}', '\u{a881}', SC_Extend),
- ('\u{a882}', '\u{a8b3}', SC_OLetter), ('\u{a8b4}', '\u{a8c5}', SC_Extend), ('\u{a8ce}',
- '\u{a8cf}', SC_STerm), ('\u{a8d0}', '\u{a8d9}', SC_Numeric), ('\u{a8e0}', '\u{a8f1}',
- SC_Extend), ('\u{a8f2}', '\u{a8f7}', SC_OLetter), ('\u{a8fb}', '\u{a8fb}', SC_OLetter),
- ('\u{a8fd}', '\u{a8fe}', SC_OLetter), ('\u{a8ff}', '\u{a8ff}', SC_Extend), ('\u{a900}',
- '\u{a909}', SC_Numeric), ('\u{a90a}', '\u{a925}', SC_OLetter), ('\u{a926}', '\u{a92d}',
- SC_Extend), ('\u{a92f}', '\u{a92f}', SC_STerm), ('\u{a930}', '\u{a946}', SC_OLetter),
- ('\u{a947}', '\u{a953}', SC_Extend), ('\u{a960}', '\u{a97c}', SC_OLetter), ('\u{a980}',
- '\u{a983}', SC_Extend), ('\u{a984}', '\u{a9b2}', SC_OLetter), ('\u{a9b3}', '\u{a9c0}',
- SC_Extend), ('\u{a9c8}', '\u{a9c9}', SC_STerm), ('\u{a9cf}', '\u{a9cf}', SC_OLetter),
- ('\u{a9d0}', '\u{a9d9}', SC_Numeric), ('\u{a9e0}', '\u{a9e4}', SC_OLetter), ('\u{a9e5}',
- '\u{a9e5}', SC_Extend), ('\u{a9e6}', '\u{a9ef}', SC_OLetter), ('\u{a9f0}', '\u{a9f9}',
- SC_Numeric), ('\u{a9fa}', '\u{a9fe}', SC_OLetter), ('\u{aa00}', '\u{aa28}', SC_OLetter),
- ('\u{aa29}', '\u{aa36}', SC_Extend), ('\u{aa40}', '\u{aa42}', SC_OLetter), ('\u{aa43}',
- '\u{aa43}', SC_Extend), ('\u{aa44}', '\u{aa4b}', SC_OLetter), ('\u{aa4c}', '\u{aa4d}',
- SC_Extend), ('\u{aa50}', '\u{aa59}', SC_Numeric), ('\u{aa5d}', '\u{aa5f}', SC_STerm),
- ('\u{aa60}', '\u{aa76}', SC_OLetter), ('\u{aa7a}', '\u{aa7a}', SC_OLetter), ('\u{aa7b}',
- '\u{aa7d}', SC_Extend), ('\u{aa7e}', '\u{aaaf}', SC_OLetter), ('\u{aab0}', '\u{aab0}',
- SC_Extend), ('\u{aab1}', '\u{aab1}', SC_OLetter), ('\u{aab2}', '\u{aab4}', SC_Extend),
- ('\u{aab5}', '\u{aab6}', SC_OLetter), ('\u{aab7}', '\u{aab8}', SC_Extend), ('\u{aab9}',
- '\u{aabd}', SC_OLetter), ('\u{aabe}', '\u{aabf}', SC_Extend), ('\u{aac0}', '\u{aac0}',
- SC_OLetter), ('\u{aac1}', '\u{aac1}', SC_Extend), ('\u{aac2}', '\u{aac2}', SC_OLetter),
- ('\u{aadb}', '\u{aadd}', SC_OLetter), ('\u{aae0}', '\u{aaea}', SC_OLetter), ('\u{aaeb}',
- '\u{aaef}', SC_Extend), ('\u{aaf0}', '\u{aaf1}', SC_STerm), ('\u{aaf2}', '\u{aaf4}',
- SC_OLetter), ('\u{aaf5}', '\u{aaf6}', SC_Extend), ('\u{ab01}', '\u{ab06}', SC_OLetter),
- ('\u{ab09}', '\u{ab0e}', SC_OLetter), ('\u{ab11}', '\u{ab16}', SC_OLetter), ('\u{ab20}',
- '\u{ab26}', SC_OLetter), ('\u{ab28}', '\u{ab2e}', SC_OLetter), ('\u{ab30}', '\u{ab5a}',
- SC_Lower), ('\u{ab5c}', '\u{ab67}', SC_Lower), ('\u{ab70}', '\u{abbf}', SC_Lower),
+ ('\u{a7c4}', '\u{a7c7}', SC_Upper), ('\u{a7c8}', '\u{a7c8}', SC_Lower), ('\u{a7c9}',
+ '\u{a7c9}', SC_Upper), ('\u{a7ca}', '\u{a7ca}', SC_Lower), ('\u{a7f5}', '\u{a7f5}',
+ SC_Upper), ('\u{a7f6}', '\u{a7f6}', SC_Lower), ('\u{a7f7}', '\u{a7f7}', SC_OLetter),
+ ('\u{a7f8}', '\u{a7fa}', SC_Lower), ('\u{a7fb}', '\u{a801}', SC_OLetter), ('\u{a802}',
+ '\u{a802}', SC_Extend), ('\u{a803}', '\u{a805}', SC_OLetter), ('\u{a806}', '\u{a806}',
+ SC_Extend), ('\u{a807}', '\u{a80a}', SC_OLetter), ('\u{a80b}', '\u{a80b}', SC_Extend),
+ ('\u{a80c}', '\u{a822}', SC_OLetter), ('\u{a823}', '\u{a827}', SC_Extend), ('\u{a82c}',
+ '\u{a82c}', SC_Extend), ('\u{a840}', '\u{a873}', SC_OLetter), ('\u{a876}', '\u{a877}',
+ SC_STerm), ('\u{a880}', '\u{a881}', SC_Extend), ('\u{a882}', '\u{a8b3}', SC_OLetter),
+ ('\u{a8b4}', '\u{a8c5}', SC_Extend), ('\u{a8ce}', '\u{a8cf}', SC_STerm), ('\u{a8d0}',
+ '\u{a8d9}', SC_Numeric), ('\u{a8e0}', '\u{a8f1}', SC_Extend), ('\u{a8f2}', '\u{a8f7}',
+ SC_OLetter), ('\u{a8fb}', '\u{a8fb}', SC_OLetter), ('\u{a8fd}', '\u{a8fe}', SC_OLetter),
+ ('\u{a8ff}', '\u{a8ff}', SC_Extend), ('\u{a900}', '\u{a909}', SC_Numeric), ('\u{a90a}',
+ '\u{a925}', SC_OLetter), ('\u{a926}', '\u{a92d}', SC_Extend), ('\u{a92f}', '\u{a92f}',
+ SC_STerm), ('\u{a930}', '\u{a946}', SC_OLetter), ('\u{a947}', '\u{a953}', SC_Extend),
+ ('\u{a960}', '\u{a97c}', SC_OLetter), ('\u{a980}', '\u{a983}', SC_Extend), ('\u{a984}',
+ '\u{a9b2}', SC_OLetter), ('\u{a9b3}', '\u{a9c0}', SC_Extend), ('\u{a9c8}', '\u{a9c9}',
+ SC_STerm), ('\u{a9cf}', '\u{a9cf}', SC_OLetter), ('\u{a9d0}', '\u{a9d9}', SC_Numeric),
+ ('\u{a9e0}', '\u{a9e4}', SC_OLetter), ('\u{a9e5}', '\u{a9e5}', SC_Extend), ('\u{a9e6}',
+ '\u{a9ef}', SC_OLetter), ('\u{a9f0}', '\u{a9f9}', SC_Numeric), ('\u{a9fa}', '\u{a9fe}',
+ SC_OLetter), ('\u{aa00}', '\u{aa28}', SC_OLetter), ('\u{aa29}', '\u{aa36}', SC_Extend),
+ ('\u{aa40}', '\u{aa42}', SC_OLetter), ('\u{aa43}', '\u{aa43}', SC_Extend), ('\u{aa44}',
+ '\u{aa4b}', SC_OLetter), ('\u{aa4c}', '\u{aa4d}', SC_Extend), ('\u{aa50}', '\u{aa59}',
+ SC_Numeric), ('\u{aa5d}', '\u{aa5f}', SC_STerm), ('\u{aa60}', '\u{aa76}', SC_OLetter),
+ ('\u{aa7a}', '\u{aa7a}', SC_OLetter), ('\u{aa7b}', '\u{aa7d}', SC_Extend), ('\u{aa7e}',
+ '\u{aaaf}', SC_OLetter), ('\u{aab0}', '\u{aab0}', SC_Extend), ('\u{aab1}', '\u{aab1}',
+ SC_OLetter), ('\u{aab2}', '\u{aab4}', SC_Extend), ('\u{aab5}', '\u{aab6}', SC_OLetter),
+ ('\u{aab7}', '\u{aab8}', SC_Extend), ('\u{aab9}', '\u{aabd}', SC_OLetter), ('\u{aabe}',
+ '\u{aabf}', SC_Extend), ('\u{aac0}', '\u{aac0}', SC_OLetter), ('\u{aac1}', '\u{aac1}',
+ SC_Extend), ('\u{aac2}', '\u{aac2}', SC_OLetter), ('\u{aadb}', '\u{aadd}', SC_OLetter),
+ ('\u{aae0}', '\u{aaea}', SC_OLetter), ('\u{aaeb}', '\u{aaef}', SC_Extend), ('\u{aaf0}',
+ '\u{aaf1}', SC_STerm), ('\u{aaf2}', '\u{aaf4}', SC_OLetter), ('\u{aaf5}', '\u{aaf6}',
+ SC_Extend), ('\u{ab01}', '\u{ab06}', SC_OLetter), ('\u{ab09}', '\u{ab0e}', SC_OLetter),
+ ('\u{ab11}', '\u{ab16}', SC_OLetter), ('\u{ab20}', '\u{ab26}', SC_OLetter), ('\u{ab28}',
+ '\u{ab2e}', SC_OLetter), ('\u{ab30}', '\u{ab5a}', SC_Lower), ('\u{ab5c}', '\u{ab68}',
+ SC_Lower), ('\u{ab69}', '\u{ab69}', SC_OLetter), ('\u{ab70}', '\u{abbf}', SC_Lower),
('\u{abc0}', '\u{abe2}', SC_OLetter), ('\u{abe3}', '\u{abea}', SC_Extend), ('\u{abeb}',
'\u{abeb}', SC_STerm), ('\u{abec}', '\u{abed}', SC_Extend), ('\u{abf0}', '\u{abf9}',
SC_Numeric), ('\u{ac00}', '\u{d7a3}', SC_OLetter), ('\u{d7b0}', '\u{d7c6}', SC_OLetter),
@@ -2362,56 +2419,65 @@
('\u{10b60}', '\u{10b72}', SC_OLetter), ('\u{10b80}', '\u{10b91}', SC_OLetter),
('\u{10c00}', '\u{10c48}', SC_OLetter), ('\u{10c80}', '\u{10cb2}', SC_Upper), ('\u{10cc0}',
'\u{10cf2}', SC_Lower), ('\u{10d00}', '\u{10d23}', SC_OLetter), ('\u{10d24}', '\u{10d27}',
- SC_Extend), ('\u{10d30}', '\u{10d39}', SC_Numeric), ('\u{10f00}', '\u{10f1c}', SC_OLetter),
- ('\u{10f27}', '\u{10f27}', SC_OLetter), ('\u{10f30}', '\u{10f45}', SC_OLetter),
- ('\u{10f46}', '\u{10f50}', SC_Extend), ('\u{10f55}', '\u{10f59}', SC_STerm), ('\u{10fe0}',
- '\u{10ff6}', SC_OLetter), ('\u{11000}', '\u{11002}', SC_Extend), ('\u{11003}', '\u{11037}',
- SC_OLetter), ('\u{11038}', '\u{11046}', SC_Extend), ('\u{11047}', '\u{11048}', SC_STerm),
- ('\u{11066}', '\u{1106f}', SC_Numeric), ('\u{1107f}', '\u{11082}', SC_Extend), ('\u{11083}',
- '\u{110af}', SC_OLetter), ('\u{110b0}', '\u{110ba}', SC_Extend), ('\u{110bd}', '\u{110bd}',
- SC_Format), ('\u{110be}', '\u{110c1}', SC_STerm), ('\u{110cd}', '\u{110cd}', SC_Format),
- ('\u{110d0}', '\u{110e8}', SC_OLetter), ('\u{110f0}', '\u{110f9}', SC_Numeric),
- ('\u{11100}', '\u{11102}', SC_Extend), ('\u{11103}', '\u{11126}', SC_OLetter), ('\u{11127}',
- '\u{11134}', SC_Extend), ('\u{11136}', '\u{1113f}', SC_Numeric), ('\u{11141}', '\u{11143}',
- SC_STerm), ('\u{11144}', '\u{11144}', SC_OLetter), ('\u{11145}', '\u{11146}', SC_Extend),
+ SC_Extend), ('\u{10d30}', '\u{10d39}', SC_Numeric), ('\u{10e80}', '\u{10ea9}', SC_OLetter),
+ ('\u{10eab}', '\u{10eac}', SC_Extend), ('\u{10eb0}', '\u{10eb1}', SC_OLetter), ('\u{10f00}',
+ '\u{10f1c}', SC_OLetter), ('\u{10f27}', '\u{10f27}', SC_OLetter), ('\u{10f30}', '\u{10f45}',
+ SC_OLetter), ('\u{10f46}', '\u{10f50}', SC_Extend), ('\u{10f55}', '\u{10f59}', SC_STerm),
+ ('\u{10fb0}', '\u{10fc4}', SC_OLetter), ('\u{10fe0}', '\u{10ff6}', SC_OLetter),
+ ('\u{11000}', '\u{11002}', SC_Extend), ('\u{11003}', '\u{11037}', SC_OLetter), ('\u{11038}',
+ '\u{11046}', SC_Extend), ('\u{11047}', '\u{11048}', SC_STerm), ('\u{11066}', '\u{1106f}',
+ SC_Numeric), ('\u{1107f}', '\u{11082}', SC_Extend), ('\u{11083}', '\u{110af}', SC_OLetter),
+ ('\u{110b0}', '\u{110ba}', SC_Extend), ('\u{110bd}', '\u{110bd}', SC_Format), ('\u{110be}',
+ '\u{110c1}', SC_STerm), ('\u{110cd}', '\u{110cd}', SC_Format), ('\u{110d0}', '\u{110e8}',
+ SC_OLetter), ('\u{110f0}', '\u{110f9}', SC_Numeric), ('\u{11100}', '\u{11102}', SC_Extend),
+ ('\u{11103}', '\u{11126}', SC_OLetter), ('\u{11127}', '\u{11134}', SC_Extend), ('\u{11136}',
+ '\u{1113f}', SC_Numeric), ('\u{11141}', '\u{11143}', SC_STerm), ('\u{11144}', '\u{11144}',
+ SC_OLetter), ('\u{11145}', '\u{11146}', SC_Extend), ('\u{11147}', '\u{11147}', SC_OLetter),
('\u{11150}', '\u{11172}', SC_OLetter), ('\u{11173}', '\u{11173}', SC_Extend), ('\u{11176}',
'\u{11176}', SC_OLetter), ('\u{11180}', '\u{11182}', SC_Extend), ('\u{11183}', '\u{111b2}',
SC_OLetter), ('\u{111b3}', '\u{111c0}', SC_Extend), ('\u{111c1}', '\u{111c4}', SC_OLetter),
('\u{111c5}', '\u{111c6}', SC_STerm), ('\u{111c9}', '\u{111cc}', SC_Extend), ('\u{111cd}',
- '\u{111cd}', SC_STerm), ('\u{111d0}', '\u{111d9}', SC_Numeric), ('\u{111da}', '\u{111da}',
- SC_OLetter), ('\u{111dc}', '\u{111dc}', SC_OLetter), ('\u{111de}', '\u{111df}', SC_STerm),
- ('\u{11200}', '\u{11211}', SC_OLetter), ('\u{11213}', '\u{1122b}', SC_OLetter),
- ('\u{1122c}', '\u{11237}', SC_Extend), ('\u{11238}', '\u{11239}', SC_STerm), ('\u{1123b}',
- '\u{1123c}', SC_STerm), ('\u{1123e}', '\u{1123e}', SC_Extend), ('\u{11280}', '\u{11286}',
- SC_OLetter), ('\u{11288}', '\u{11288}', SC_OLetter), ('\u{1128a}', '\u{1128d}', SC_OLetter),
- ('\u{1128f}', '\u{1129d}', SC_OLetter), ('\u{1129f}', '\u{112a8}', SC_OLetter),
- ('\u{112a9}', '\u{112a9}', SC_STerm), ('\u{112b0}', '\u{112de}', SC_OLetter), ('\u{112df}',
- '\u{112ea}', SC_Extend), ('\u{112f0}', '\u{112f9}', SC_Numeric), ('\u{11300}', '\u{11303}',
- SC_Extend), ('\u{11305}', '\u{1130c}', SC_OLetter), ('\u{1130f}', '\u{11310}', SC_OLetter),
- ('\u{11313}', '\u{11328}', SC_OLetter), ('\u{1132a}', '\u{11330}', SC_OLetter),
- ('\u{11332}', '\u{11333}', SC_OLetter), ('\u{11335}', '\u{11339}', SC_OLetter),
- ('\u{1133b}', '\u{1133c}', SC_Extend), ('\u{1133d}', '\u{1133d}', SC_OLetter), ('\u{1133e}',
- '\u{11344}', SC_Extend), ('\u{11347}', '\u{11348}', SC_Extend), ('\u{1134b}', '\u{1134d}',
- SC_Extend), ('\u{11350}', '\u{11350}', SC_OLetter), ('\u{11357}', '\u{11357}', SC_Extend),
- ('\u{1135d}', '\u{11361}', SC_OLetter), ('\u{11362}', '\u{11363}', SC_Extend), ('\u{11366}',
- '\u{1136c}', SC_Extend), ('\u{11370}', '\u{11374}', SC_Extend), ('\u{11400}', '\u{11434}',
- SC_OLetter), ('\u{11435}', '\u{11446}', SC_Extend), ('\u{11447}', '\u{1144a}', SC_OLetter),
- ('\u{1144b}', '\u{1144c}', SC_STerm), ('\u{11450}', '\u{11459}', SC_Numeric), ('\u{1145e}',
- '\u{1145e}', SC_Extend), ('\u{1145f}', '\u{1145f}', SC_OLetter), ('\u{11480}', '\u{114af}',
- SC_OLetter), ('\u{114b0}', '\u{114c3}', SC_Extend), ('\u{114c4}', '\u{114c5}', SC_OLetter),
- ('\u{114c7}', '\u{114c7}', SC_OLetter), ('\u{114d0}', '\u{114d9}', SC_Numeric),
- ('\u{11580}', '\u{115ae}', SC_OLetter), ('\u{115af}', '\u{115b5}', SC_Extend), ('\u{115b8}',
- '\u{115c0}', SC_Extend), ('\u{115c2}', '\u{115c3}', SC_STerm), ('\u{115c9}', '\u{115d7}',
- SC_STerm), ('\u{115d8}', '\u{115db}', SC_OLetter), ('\u{115dc}', '\u{115dd}', SC_Extend),
- ('\u{11600}', '\u{1162f}', SC_OLetter), ('\u{11630}', '\u{11640}', SC_Extend), ('\u{11641}',
- '\u{11642}', SC_STerm), ('\u{11644}', '\u{11644}', SC_OLetter), ('\u{11650}', '\u{11659}',
- SC_Numeric), ('\u{11680}', '\u{116aa}', SC_OLetter), ('\u{116ab}', '\u{116b7}', SC_Extend),
- ('\u{116b8}', '\u{116b8}', SC_OLetter), ('\u{116c0}', '\u{116c9}', SC_Numeric),
- ('\u{11700}', '\u{1171a}', SC_OLetter), ('\u{1171d}', '\u{1172b}', SC_Extend), ('\u{11730}',
- '\u{11739}', SC_Numeric), ('\u{1173c}', '\u{1173e}', SC_STerm), ('\u{11800}', '\u{1182b}',
- SC_OLetter), ('\u{1182c}', '\u{1183a}', SC_Extend), ('\u{118a0}', '\u{118bf}', SC_Upper),
- ('\u{118c0}', '\u{118df}', SC_Lower), ('\u{118e0}', '\u{118e9}', SC_Numeric), ('\u{118ff}',
- '\u{118ff}', SC_OLetter), ('\u{119a0}', '\u{119a7}', SC_OLetter), ('\u{119aa}', '\u{119d0}',
+ '\u{111cd}', SC_STerm), ('\u{111ce}', '\u{111cf}', SC_Extend), ('\u{111d0}', '\u{111d9}',
+ SC_Numeric), ('\u{111da}', '\u{111da}', SC_OLetter), ('\u{111dc}', '\u{111dc}', SC_OLetter),
+ ('\u{111de}', '\u{111df}', SC_STerm), ('\u{11200}', '\u{11211}', SC_OLetter), ('\u{11213}',
+ '\u{1122b}', SC_OLetter), ('\u{1122c}', '\u{11237}', SC_Extend), ('\u{11238}', '\u{11239}',
+ SC_STerm), ('\u{1123b}', '\u{1123c}', SC_STerm), ('\u{1123e}', '\u{1123e}', SC_Extend),
+ ('\u{11280}', '\u{11286}', SC_OLetter), ('\u{11288}', '\u{11288}', SC_OLetter),
+ ('\u{1128a}', '\u{1128d}', SC_OLetter), ('\u{1128f}', '\u{1129d}', SC_OLetter),
+ ('\u{1129f}', '\u{112a8}', SC_OLetter), ('\u{112a9}', '\u{112a9}', SC_STerm), ('\u{112b0}',
+ '\u{112de}', SC_OLetter), ('\u{112df}', '\u{112ea}', SC_Extend), ('\u{112f0}', '\u{112f9}',
+ SC_Numeric), ('\u{11300}', '\u{11303}', SC_Extend), ('\u{11305}', '\u{1130c}', SC_OLetter),
+ ('\u{1130f}', '\u{11310}', SC_OLetter), ('\u{11313}', '\u{11328}', SC_OLetter),
+ ('\u{1132a}', '\u{11330}', SC_OLetter), ('\u{11332}', '\u{11333}', SC_OLetter),
+ ('\u{11335}', '\u{11339}', SC_OLetter), ('\u{1133b}', '\u{1133c}', SC_Extend), ('\u{1133d}',
+ '\u{1133d}', SC_OLetter), ('\u{1133e}', '\u{11344}', SC_Extend), ('\u{11347}', '\u{11348}',
+ SC_Extend), ('\u{1134b}', '\u{1134d}', SC_Extend), ('\u{11350}', '\u{11350}', SC_OLetter),
+ ('\u{11357}', '\u{11357}', SC_Extend), ('\u{1135d}', '\u{11361}', SC_OLetter), ('\u{11362}',
+ '\u{11363}', SC_Extend), ('\u{11366}', '\u{1136c}', SC_Extend), ('\u{11370}', '\u{11374}',
+ SC_Extend), ('\u{11400}', '\u{11434}', SC_OLetter), ('\u{11435}', '\u{11446}', SC_Extend),
+ ('\u{11447}', '\u{1144a}', SC_OLetter), ('\u{1144b}', '\u{1144c}', SC_STerm), ('\u{11450}',
+ '\u{11459}', SC_Numeric), ('\u{1145e}', '\u{1145e}', SC_Extend), ('\u{1145f}', '\u{11461}',
+ SC_OLetter), ('\u{11480}', '\u{114af}', SC_OLetter), ('\u{114b0}', '\u{114c3}', SC_Extend),
+ ('\u{114c4}', '\u{114c5}', SC_OLetter), ('\u{114c7}', '\u{114c7}', SC_OLetter),
+ ('\u{114d0}', '\u{114d9}', SC_Numeric), ('\u{11580}', '\u{115ae}', SC_OLetter),
+ ('\u{115af}', '\u{115b5}', SC_Extend), ('\u{115b8}', '\u{115c0}', SC_Extend), ('\u{115c2}',
+ '\u{115c3}', SC_STerm), ('\u{115c9}', '\u{115d7}', SC_STerm), ('\u{115d8}', '\u{115db}',
+ SC_OLetter), ('\u{115dc}', '\u{115dd}', SC_Extend), ('\u{11600}', '\u{1162f}', SC_OLetter),
+ ('\u{11630}', '\u{11640}', SC_Extend), ('\u{11641}', '\u{11642}', SC_STerm), ('\u{11644}',
+ '\u{11644}', SC_OLetter), ('\u{11650}', '\u{11659}', SC_Numeric), ('\u{11680}', '\u{116aa}',
+ SC_OLetter), ('\u{116ab}', '\u{116b7}', SC_Extend), ('\u{116b8}', '\u{116b8}', SC_OLetter),
+ ('\u{116c0}', '\u{116c9}', SC_Numeric), ('\u{11700}', '\u{1171a}', SC_OLetter),
+ ('\u{1171d}', '\u{1172b}', SC_Extend), ('\u{11730}', '\u{11739}', SC_Numeric), ('\u{1173c}',
+ '\u{1173e}', SC_STerm), ('\u{11800}', '\u{1182b}', SC_OLetter), ('\u{1182c}', '\u{1183a}',
+ SC_Extend), ('\u{118a0}', '\u{118bf}', SC_Upper), ('\u{118c0}', '\u{118df}', SC_Lower),
+ ('\u{118e0}', '\u{118e9}', SC_Numeric), ('\u{118ff}', '\u{11906}', SC_OLetter),
+ ('\u{11909}', '\u{11909}', SC_OLetter), ('\u{1190c}', '\u{11913}', SC_OLetter),
+ ('\u{11915}', '\u{11916}', SC_OLetter), ('\u{11918}', '\u{1192f}', SC_OLetter),
+ ('\u{11930}', '\u{11935}', SC_Extend), ('\u{11937}', '\u{11938}', SC_Extend), ('\u{1193b}',
+ '\u{1193e}', SC_Extend), ('\u{1193f}', '\u{1193f}', SC_OLetter), ('\u{11940}', '\u{11940}',
+ SC_Extend), ('\u{11941}', '\u{11941}', SC_OLetter), ('\u{11942}', '\u{11943}', SC_Extend),
+ ('\u{11944}', '\u{11944}', SC_STerm), ('\u{11946}', '\u{11946}', SC_STerm), ('\u{11950}',
+ '\u{11959}', SC_Numeric), ('\u{119a0}', '\u{119a7}', SC_OLetter), ('\u{119aa}', '\u{119d0}',
SC_OLetter), ('\u{119d1}', '\u{119d7}', SC_Extend), ('\u{119da}', '\u{119e0}', SC_Extend),
('\u{119e1}', '\u{119e1}', SC_OLetter), ('\u{119e3}', '\u{119e3}', SC_OLetter),
('\u{119e4}', '\u{119e4}', SC_Extend), ('\u{11a00}', '\u{11a00}', SC_OLetter), ('\u{11a01}',
@@ -2434,67 +2500,69 @@
('\u{11d8a}', '\u{11d8e}', SC_Extend), ('\u{11d90}', '\u{11d91}', SC_Extend), ('\u{11d93}',
'\u{11d97}', SC_Extend), ('\u{11d98}', '\u{11d98}', SC_OLetter), ('\u{11da0}', '\u{11da9}',
SC_Numeric), ('\u{11ee0}', '\u{11ef2}', SC_OLetter), ('\u{11ef3}', '\u{11ef6}', SC_Extend),
- ('\u{11ef7}', '\u{11ef8}', SC_STerm), ('\u{12000}', '\u{12399}', SC_OLetter), ('\u{12400}',
- '\u{1246e}', SC_OLetter), ('\u{12480}', '\u{12543}', SC_OLetter), ('\u{13000}', '\u{1342e}',
- SC_OLetter), ('\u{13430}', '\u{13438}', SC_Format), ('\u{14400}', '\u{14646}', SC_OLetter),
- ('\u{16800}', '\u{16a38}', SC_OLetter), ('\u{16a40}', '\u{16a5e}', SC_OLetter),
- ('\u{16a60}', '\u{16a69}', SC_Numeric), ('\u{16a6e}', '\u{16a6f}', SC_STerm), ('\u{16ad0}',
- '\u{16aed}', SC_OLetter), ('\u{16af0}', '\u{16af4}', SC_Extend), ('\u{16af5}', '\u{16af5}',
- SC_STerm), ('\u{16b00}', '\u{16b2f}', SC_OLetter), ('\u{16b30}', '\u{16b36}', SC_Extend),
- ('\u{16b37}', '\u{16b38}', SC_STerm), ('\u{16b40}', '\u{16b43}', SC_OLetter), ('\u{16b44}',
- '\u{16b44}', SC_STerm), ('\u{16b50}', '\u{16b59}', SC_Numeric), ('\u{16b63}', '\u{16b77}',
- SC_OLetter), ('\u{16b7d}', '\u{16b8f}', SC_OLetter), ('\u{16e40}', '\u{16e5f}', SC_Upper),
- ('\u{16e60}', '\u{16e7f}', SC_Lower), ('\u{16e98}', '\u{16e98}', SC_STerm), ('\u{16f00}',
- '\u{16f4a}', SC_OLetter), ('\u{16f4f}', '\u{16f4f}', SC_Extend), ('\u{16f50}', '\u{16f50}',
- SC_OLetter), ('\u{16f51}', '\u{16f87}', SC_Extend), ('\u{16f8f}', '\u{16f92}', SC_Extend),
- ('\u{16f93}', '\u{16f9f}', SC_OLetter), ('\u{16fe0}', '\u{16fe1}', SC_OLetter),
- ('\u{16fe3}', '\u{16fe3}', SC_OLetter), ('\u{17000}', '\u{187f7}', SC_OLetter),
- ('\u{18800}', '\u{18af2}', SC_OLetter), ('\u{1b000}', '\u{1b11e}', SC_OLetter),
- ('\u{1b150}', '\u{1b152}', SC_OLetter), ('\u{1b164}', '\u{1b167}', SC_OLetter),
- ('\u{1b170}', '\u{1b2fb}', SC_OLetter), ('\u{1bc00}', '\u{1bc6a}', SC_OLetter),
- ('\u{1bc70}', '\u{1bc7c}', SC_OLetter), ('\u{1bc80}', '\u{1bc88}', SC_OLetter),
- ('\u{1bc90}', '\u{1bc99}', SC_OLetter), ('\u{1bc9d}', '\u{1bc9e}', SC_Extend), ('\u{1bc9f}',
- '\u{1bc9f}', SC_STerm), ('\u{1bca0}', '\u{1bca3}', SC_Format), ('\u{1d165}', '\u{1d169}',
- SC_Extend), ('\u{1d16d}', '\u{1d172}', SC_Extend), ('\u{1d173}', '\u{1d17a}', SC_Format),
- ('\u{1d17b}', '\u{1d182}', SC_Extend), ('\u{1d185}', '\u{1d18b}', SC_Extend), ('\u{1d1aa}',
- '\u{1d1ad}', SC_Extend), ('\u{1d242}', '\u{1d244}', SC_Extend), ('\u{1d400}', '\u{1d419}',
- SC_Upper), ('\u{1d41a}', '\u{1d433}', SC_Lower), ('\u{1d434}', '\u{1d44d}', SC_Upper),
- ('\u{1d44e}', '\u{1d454}', SC_Lower), ('\u{1d456}', '\u{1d467}', SC_Lower), ('\u{1d468}',
- '\u{1d481}', SC_Upper), ('\u{1d482}', '\u{1d49b}', SC_Lower), ('\u{1d49c}', '\u{1d49c}',
- SC_Upper), ('\u{1d49e}', '\u{1d49f}', SC_Upper), ('\u{1d4a2}', '\u{1d4a2}', SC_Upper),
- ('\u{1d4a5}', '\u{1d4a6}', SC_Upper), ('\u{1d4a9}', '\u{1d4ac}', SC_Upper), ('\u{1d4ae}',
- '\u{1d4b5}', SC_Upper), ('\u{1d4b6}', '\u{1d4b9}', SC_Lower), ('\u{1d4bb}', '\u{1d4bb}',
- SC_Lower), ('\u{1d4bd}', '\u{1d4c3}', SC_Lower), ('\u{1d4c5}', '\u{1d4cf}', SC_Lower),
- ('\u{1d4d0}', '\u{1d4e9}', SC_Upper), ('\u{1d4ea}', '\u{1d503}', SC_Lower), ('\u{1d504}',
- '\u{1d505}', SC_Upper), ('\u{1d507}', '\u{1d50a}', SC_Upper), ('\u{1d50d}', '\u{1d514}',
- SC_Upper), ('\u{1d516}', '\u{1d51c}', SC_Upper), ('\u{1d51e}', '\u{1d537}', SC_Lower),
- ('\u{1d538}', '\u{1d539}', SC_Upper), ('\u{1d53b}', '\u{1d53e}', SC_Upper), ('\u{1d540}',
- '\u{1d544}', SC_Upper), ('\u{1d546}', '\u{1d546}', SC_Upper), ('\u{1d54a}', '\u{1d550}',
- SC_Upper), ('\u{1d552}', '\u{1d56b}', SC_Lower), ('\u{1d56c}', '\u{1d585}', SC_Upper),
- ('\u{1d586}', '\u{1d59f}', SC_Lower), ('\u{1d5a0}', '\u{1d5b9}', SC_Upper), ('\u{1d5ba}',
- '\u{1d5d3}', SC_Lower), ('\u{1d5d4}', '\u{1d5ed}', SC_Upper), ('\u{1d5ee}', '\u{1d607}',
- SC_Lower), ('\u{1d608}', '\u{1d621}', SC_Upper), ('\u{1d622}', '\u{1d63b}', SC_Lower),
- ('\u{1d63c}', '\u{1d655}', SC_Upper), ('\u{1d656}', '\u{1d66f}', SC_Lower), ('\u{1d670}',
- '\u{1d689}', SC_Upper), ('\u{1d68a}', '\u{1d6a5}', SC_Lower), ('\u{1d6a8}', '\u{1d6c0}',
- SC_Upper), ('\u{1d6c2}', '\u{1d6da}', SC_Lower), ('\u{1d6dc}', '\u{1d6e1}', SC_Lower),
- ('\u{1d6e2}', '\u{1d6fa}', SC_Upper), ('\u{1d6fc}', '\u{1d714}', SC_Lower), ('\u{1d716}',
- '\u{1d71b}', SC_Lower), ('\u{1d71c}', '\u{1d734}', SC_Upper), ('\u{1d736}', '\u{1d74e}',
- SC_Lower), ('\u{1d750}', '\u{1d755}', SC_Lower), ('\u{1d756}', '\u{1d76e}', SC_Upper),
- ('\u{1d770}', '\u{1d788}', SC_Lower), ('\u{1d78a}', '\u{1d78f}', SC_Lower), ('\u{1d790}',
- '\u{1d7a8}', SC_Upper), ('\u{1d7aa}', '\u{1d7c2}', SC_Lower), ('\u{1d7c4}', '\u{1d7c9}',
- SC_Lower), ('\u{1d7ca}', '\u{1d7ca}', SC_Upper), ('\u{1d7cb}', '\u{1d7cb}', SC_Lower),
- ('\u{1d7ce}', '\u{1d7ff}', SC_Numeric), ('\u{1da00}', '\u{1da36}', SC_Extend), ('\u{1da3b}',
- '\u{1da6c}', SC_Extend), ('\u{1da75}', '\u{1da75}', SC_Extend), ('\u{1da84}', '\u{1da84}',
- SC_Extend), ('\u{1da88}', '\u{1da88}', SC_STerm), ('\u{1da9b}', '\u{1da9f}', SC_Extend),
- ('\u{1daa1}', '\u{1daaf}', SC_Extend), ('\u{1e000}', '\u{1e006}', SC_Extend), ('\u{1e008}',
- '\u{1e018}', SC_Extend), ('\u{1e01b}', '\u{1e021}', SC_Extend), ('\u{1e023}', '\u{1e024}',
- SC_Extend), ('\u{1e026}', '\u{1e02a}', SC_Extend), ('\u{1e100}', '\u{1e12c}', SC_OLetter),
- ('\u{1e130}', '\u{1e136}', SC_Extend), ('\u{1e137}', '\u{1e13d}', SC_OLetter), ('\u{1e140}',
- '\u{1e149}', SC_Numeric), ('\u{1e14e}', '\u{1e14e}', SC_OLetter), ('\u{1e2c0}', '\u{1e2eb}',
- SC_OLetter), ('\u{1e2ec}', '\u{1e2ef}', SC_Extend), ('\u{1e2f0}', '\u{1e2f9}', SC_Numeric),
- ('\u{1e800}', '\u{1e8c4}', SC_OLetter), ('\u{1e8d0}', '\u{1e8d6}', SC_Extend), ('\u{1e900}',
- '\u{1e921}', SC_Upper), ('\u{1e922}', '\u{1e943}', SC_Lower), ('\u{1e944}', '\u{1e94a}',
- SC_Extend), ('\u{1e94b}', '\u{1e94b}', SC_OLetter), ('\u{1e950}', '\u{1e959}', SC_Numeric),
+ ('\u{11ef7}', '\u{11ef8}', SC_STerm), ('\u{11fb0}', '\u{11fb0}', SC_OLetter), ('\u{12000}',
+ '\u{12399}', SC_OLetter), ('\u{12400}', '\u{1246e}', SC_OLetter), ('\u{12480}', '\u{12543}',
+ SC_OLetter), ('\u{13000}', '\u{1342e}', SC_OLetter), ('\u{13430}', '\u{13438}', SC_Format),
+ ('\u{14400}', '\u{14646}', SC_OLetter), ('\u{16800}', '\u{16a38}', SC_OLetter),
+ ('\u{16a40}', '\u{16a5e}', SC_OLetter), ('\u{16a60}', '\u{16a69}', SC_Numeric),
+ ('\u{16a6e}', '\u{16a6f}', SC_STerm), ('\u{16ad0}', '\u{16aed}', SC_OLetter), ('\u{16af0}',
+ '\u{16af4}', SC_Extend), ('\u{16af5}', '\u{16af5}', SC_STerm), ('\u{16b00}', '\u{16b2f}',
+ SC_OLetter), ('\u{16b30}', '\u{16b36}', SC_Extend), ('\u{16b37}', '\u{16b38}', SC_STerm),
+ ('\u{16b40}', '\u{16b43}', SC_OLetter), ('\u{16b44}', '\u{16b44}', SC_STerm), ('\u{16b50}',
+ '\u{16b59}', SC_Numeric), ('\u{16b63}', '\u{16b77}', SC_OLetter), ('\u{16b7d}', '\u{16b8f}',
+ SC_OLetter), ('\u{16e40}', '\u{16e5f}', SC_Upper), ('\u{16e60}', '\u{16e7f}', SC_Lower),
+ ('\u{16e98}', '\u{16e98}', SC_STerm), ('\u{16f00}', '\u{16f4a}', SC_OLetter), ('\u{16f4f}',
+ '\u{16f4f}', SC_Extend), ('\u{16f50}', '\u{16f50}', SC_OLetter), ('\u{16f51}', '\u{16f87}',
+ SC_Extend), ('\u{16f8f}', '\u{16f92}', SC_Extend), ('\u{16f93}', '\u{16f9f}', SC_OLetter),
+ ('\u{16fe0}', '\u{16fe1}', SC_OLetter), ('\u{16fe3}', '\u{16fe3}', SC_OLetter),
+ ('\u{16fe4}', '\u{16fe4}', SC_Extend), ('\u{16ff0}', '\u{16ff1}', SC_Extend), ('\u{17000}',
+ '\u{187f7}', SC_OLetter), ('\u{18800}', '\u{18cd5}', SC_OLetter), ('\u{18d00}', '\u{18d08}',
+ SC_OLetter), ('\u{1b000}', '\u{1b11e}', SC_OLetter), ('\u{1b150}', '\u{1b152}', SC_OLetter),
+ ('\u{1b164}', '\u{1b167}', SC_OLetter), ('\u{1b170}', '\u{1b2fb}', SC_OLetter),
+ ('\u{1bc00}', '\u{1bc6a}', SC_OLetter), ('\u{1bc70}', '\u{1bc7c}', SC_OLetter),
+ ('\u{1bc80}', '\u{1bc88}', SC_OLetter), ('\u{1bc90}', '\u{1bc99}', SC_OLetter),
+ ('\u{1bc9d}', '\u{1bc9e}', SC_Extend), ('\u{1bc9f}', '\u{1bc9f}', SC_STerm), ('\u{1bca0}',
+ '\u{1bca3}', SC_Format), ('\u{1d165}', '\u{1d169}', SC_Extend), ('\u{1d16d}', '\u{1d172}',
+ SC_Extend), ('\u{1d173}', '\u{1d17a}', SC_Format), ('\u{1d17b}', '\u{1d182}', SC_Extend),
+ ('\u{1d185}', '\u{1d18b}', SC_Extend), ('\u{1d1aa}', '\u{1d1ad}', SC_Extend), ('\u{1d242}',
+ '\u{1d244}', SC_Extend), ('\u{1d400}', '\u{1d419}', SC_Upper), ('\u{1d41a}', '\u{1d433}',
+ SC_Lower), ('\u{1d434}', '\u{1d44d}', SC_Upper), ('\u{1d44e}', '\u{1d454}', SC_Lower),
+ ('\u{1d456}', '\u{1d467}', SC_Lower), ('\u{1d468}', '\u{1d481}', SC_Upper), ('\u{1d482}',
+ '\u{1d49b}', SC_Lower), ('\u{1d49c}', '\u{1d49c}', SC_Upper), ('\u{1d49e}', '\u{1d49f}',
+ SC_Upper), ('\u{1d4a2}', '\u{1d4a2}', SC_Upper), ('\u{1d4a5}', '\u{1d4a6}', SC_Upper),
+ ('\u{1d4a9}', '\u{1d4ac}', SC_Upper), ('\u{1d4ae}', '\u{1d4b5}', SC_Upper), ('\u{1d4b6}',
+ '\u{1d4b9}', SC_Lower), ('\u{1d4bb}', '\u{1d4bb}', SC_Lower), ('\u{1d4bd}', '\u{1d4c3}',
+ SC_Lower), ('\u{1d4c5}', '\u{1d4cf}', SC_Lower), ('\u{1d4d0}', '\u{1d4e9}', SC_Upper),
+ ('\u{1d4ea}', '\u{1d503}', SC_Lower), ('\u{1d504}', '\u{1d505}', SC_Upper), ('\u{1d507}',
+ '\u{1d50a}', SC_Upper), ('\u{1d50d}', '\u{1d514}', SC_Upper), ('\u{1d516}', '\u{1d51c}',
+ SC_Upper), ('\u{1d51e}', '\u{1d537}', SC_Lower), ('\u{1d538}', '\u{1d539}', SC_Upper),
+ ('\u{1d53b}', '\u{1d53e}', SC_Upper), ('\u{1d540}', '\u{1d544}', SC_Upper), ('\u{1d546}',
+ '\u{1d546}', SC_Upper), ('\u{1d54a}', '\u{1d550}', SC_Upper), ('\u{1d552}', '\u{1d56b}',
+ SC_Lower), ('\u{1d56c}', '\u{1d585}', SC_Upper), ('\u{1d586}', '\u{1d59f}', SC_Lower),
+ ('\u{1d5a0}', '\u{1d5b9}', SC_Upper), ('\u{1d5ba}', '\u{1d5d3}', SC_Lower), ('\u{1d5d4}',
+ '\u{1d5ed}', SC_Upper), ('\u{1d5ee}', '\u{1d607}', SC_Lower), ('\u{1d608}', '\u{1d621}',
+ SC_Upper), ('\u{1d622}', '\u{1d63b}', SC_Lower), ('\u{1d63c}', '\u{1d655}', SC_Upper),
+ ('\u{1d656}', '\u{1d66f}', SC_Lower), ('\u{1d670}', '\u{1d689}', SC_Upper), ('\u{1d68a}',
+ '\u{1d6a5}', SC_Lower), ('\u{1d6a8}', '\u{1d6c0}', SC_Upper), ('\u{1d6c2}', '\u{1d6da}',
+ SC_Lower), ('\u{1d6dc}', '\u{1d6e1}', SC_Lower), ('\u{1d6e2}', '\u{1d6fa}', SC_Upper),
+ ('\u{1d6fc}', '\u{1d714}', SC_Lower), ('\u{1d716}', '\u{1d71b}', SC_Lower), ('\u{1d71c}',
+ '\u{1d734}', SC_Upper), ('\u{1d736}', '\u{1d74e}', SC_Lower), ('\u{1d750}', '\u{1d755}',
+ SC_Lower), ('\u{1d756}', '\u{1d76e}', SC_Upper), ('\u{1d770}', '\u{1d788}', SC_Lower),
+ ('\u{1d78a}', '\u{1d78f}', SC_Lower), ('\u{1d790}', '\u{1d7a8}', SC_Upper), ('\u{1d7aa}',
+ '\u{1d7c2}', SC_Lower), ('\u{1d7c4}', '\u{1d7c9}', SC_Lower), ('\u{1d7ca}', '\u{1d7ca}',
+ SC_Upper), ('\u{1d7cb}', '\u{1d7cb}', SC_Lower), ('\u{1d7ce}', '\u{1d7ff}', SC_Numeric),
+ ('\u{1da00}', '\u{1da36}', SC_Extend), ('\u{1da3b}', '\u{1da6c}', SC_Extend), ('\u{1da75}',
+ '\u{1da75}', SC_Extend), ('\u{1da84}', '\u{1da84}', SC_Extend), ('\u{1da88}', '\u{1da88}',
+ SC_STerm), ('\u{1da9b}', '\u{1da9f}', SC_Extend), ('\u{1daa1}', '\u{1daaf}', SC_Extend),
+ ('\u{1e000}', '\u{1e006}', SC_Extend), ('\u{1e008}', '\u{1e018}', SC_Extend), ('\u{1e01b}',
+ '\u{1e021}', SC_Extend), ('\u{1e023}', '\u{1e024}', SC_Extend), ('\u{1e026}', '\u{1e02a}',
+ SC_Extend), ('\u{1e100}', '\u{1e12c}', SC_OLetter), ('\u{1e130}', '\u{1e136}', SC_Extend),
+ ('\u{1e137}', '\u{1e13d}', SC_OLetter), ('\u{1e140}', '\u{1e149}', SC_Numeric),
+ ('\u{1e14e}', '\u{1e14e}', SC_OLetter), ('\u{1e2c0}', '\u{1e2eb}', SC_OLetter),
+ ('\u{1e2ec}', '\u{1e2ef}', SC_Extend), ('\u{1e2f0}', '\u{1e2f9}', SC_Numeric), ('\u{1e800}',
+ '\u{1e8c4}', SC_OLetter), ('\u{1e8d0}', '\u{1e8d6}', SC_Extend), ('\u{1e900}', '\u{1e921}',
+ SC_Upper), ('\u{1e922}', '\u{1e943}', SC_Lower), ('\u{1e944}', '\u{1e94a}', SC_Extend),
+ ('\u{1e94b}', '\u{1e94b}', SC_OLetter), ('\u{1e950}', '\u{1e959}', SC_Numeric),
('\u{1ee00}', '\u{1ee03}', SC_OLetter), ('\u{1ee05}', '\u{1ee1f}', SC_OLetter),
('\u{1ee21}', '\u{1ee22}', SC_OLetter), ('\u{1ee24}', '\u{1ee24}', SC_OLetter),
('\u{1ee27}', '\u{1ee27}', SC_OLetter), ('\u{1ee29}', '\u{1ee32}', SC_OLetter),
@@ -2513,9 +2581,10 @@
('\u{1eea1}', '\u{1eea3}', SC_OLetter), ('\u{1eea5}', '\u{1eea9}', SC_OLetter),
('\u{1eeab}', '\u{1eebb}', SC_OLetter), ('\u{1f130}', '\u{1f149}', SC_Upper), ('\u{1f150}',
'\u{1f169}', SC_Upper), ('\u{1f170}', '\u{1f189}', SC_Upper), ('\u{1f676}', '\u{1f678}',
- SC_Close), ('\u{20000}', '\u{2a6d6}', SC_OLetter), ('\u{2a700}', '\u{2b734}', SC_OLetter),
- ('\u{2b740}', '\u{2b81d}', SC_OLetter), ('\u{2b820}', '\u{2cea1}', SC_OLetter),
- ('\u{2ceb0}', '\u{2ebe0}', SC_OLetter), ('\u{2f800}', '\u{2fa1d}', SC_OLetter),
+ SC_Close), ('\u{1fbf0}', '\u{1fbf9}', SC_Numeric), ('\u{20000}', '\u{2a6dd}', SC_OLetter),
+ ('\u{2a700}', '\u{2b734}', SC_OLetter), ('\u{2b740}', '\u{2b81d}', SC_OLetter),
+ ('\u{2b820}', '\u{2cea1}', SC_OLetter), ('\u{2ceb0}', '\u{2ebe0}', SC_OLetter),
+ ('\u{2f800}', '\u{2fa1d}', SC_OLetter), ('\u{30000}', '\u{3134a}', SC_OLetter),
('\u{e0001}', '\u{e0001}', SC_Format), ('\u{e0020}', '\u{e007f}', SC_Extend), ('\u{e0100}',
'\u{e01ef}', SC_Extend)
];
diff --git a/src/testdata.rs b/src/testdata.rs
index b0402fa..7ecceba 100644
--- a/src/testdata.rs
+++ b/src/testdata.rs
@@ -12,7 +12,7 @@
#![allow(missing_docs, non_upper_case_globals, non_snake_case)]
// official Unicode test data
- // http://www.unicode.org/Public/12.0.0/ucd/auxiliary/GraphemeBreakTest.txt
+ // http://www.unicode.org/Public/13.0.0/ucd/auxiliary/GraphemeBreakTest.txt
pub const TEST_SAME: &'static [(&'static str, &'static [&'static str])] = &[
("\u{20}\u{20}", &["\u{20}", "\u{20}"]), ("\u{20}\u{308}\u{20}", &["\u{20}\u{308}",
"\u{20}"]), ("\u{20}\u{d}", &["\u{20}", "\u{d}"]), ("\u{20}\u{308}\u{d}", &["\u{20}\u{308}",
@@ -421,7 +421,7 @@
];
// official Unicode test data
- // http://www.unicode.org/Public/12.0.0/ucd/auxiliary/WordBreakTest.txt
+ // http://www.unicode.org/Public/13.0.0/ucd/auxiliary/WordBreakTest.txt
pub const TEST_WORD: &'static [(&'static str, &'static [&'static str])] = &[
("\u{1}\u{1}", &["\u{1}", "\u{1}"]), ("\u{1}\u{308}\u{1}", &["\u{1}\u{308}", "\u{1}"]),
("\u{1}\u{d}", &["\u{1}", "\u{d}"]), ("\u{1}\u{308}\u{d}", &["\u{1}\u{308}", "\u{d}"]),
@@ -1788,7 +1788,7 @@
];
// official Unicode test data
- // http://www.unicode.org/Public/12.0.0/ucd/auxiliary/SentenceBreakTest.txt
+ // http://www.unicode.org/Public/13.0.0/ucd/auxiliary/SentenceBreakTest.txt
pub const TEST_SENTENCE: &'static [(&'static str, &'static [&'static str])] = &[
("\u{1}\u{1}", &["\u{1}\u{1}"]), ("\u{1}\u{308}\u{1}", &["\u{1}\u{308}\u{1}"]),
("\u{1}\u{d}", &["\u{1}\u{d}"]), ("\u{1}\u{308}\u{d}", &["\u{1}\u{308}\u{d}"]),
diff --git a/src/word.rs b/src/word.rs
index 6e9c049..179d122 100644
--- a/src/word.rs
+++ b/src/word.rs
@@ -19,6 +19,12 @@
/// [Alphabetic](http://unicode.org/reports/tr44/#Alphabetic)
/// property, or with
/// [General_Category=Number](http://unicode.org/reports/tr44/#General_Category_Values).
+///
+/// This struct is created by the [`unicode_words`] method on the [`UnicodeSegmentation`] trait. See
+/// its documentation for more.
+///
+/// [`unicode_words`]: trait.UnicodeSegmentation.html#tymethod.unicode_words
+/// [`UnicodeSegmentation`]: trait.UnicodeSegmentation.html
pub struct UnicodeWords<'a> {
inner: Filter<UWordBounds<'a>, fn(&&str) -> bool>,
}
@@ -36,6 +42,12 @@
/// External iterator for a string's
/// [word boundaries](http://www.unicode.org/reports/tr29/#Word_Boundaries).
+///
+/// This struct is created by the [`split_word_bounds`] method on the [`UnicodeSegmentation`]
+/// trait. See its documentation for more.
+///
+/// [`split_word_bounds`]: trait.UnicodeSegmentation.html#tymethod.split_word_bounds
+/// [`UnicodeSegmentation`]: trait.UnicodeSegmentation.html
#[derive(Clone)]
pub struct UWordBounds<'a> {
string: &'a str,
@@ -44,6 +56,12 @@
}
/// External iterator for word boundaries and byte offsets.
+///
+/// This struct is created by the [`split_word_bound_indices`] method on the
+/// [`UnicodeSegmentation`] trait. See its documentation for more.
+///
+/// [`split_word_bound_indices`]: trait.UnicodeSegmentation.html#tymethod.split_word_bound_indices
+/// [`UnicodeSegmentation`]: trait.UnicodeSegmentation.html
#[derive(Clone)]
pub struct UWordBoundIndices<'a> {
start_offset: usize,
@@ -125,7 +143,7 @@
fn is_emoji(ch: char) -> bool {
use tables::emoji;
- emoji::emoji_category(ch) == emoji::EmojiCat::EC_Extended_Pictographic
+ emoji::emoji_category(ch).2 == emoji::EmojiCat::EC_Extended_Pictographic
}
impl<'a> Iterator for UWordBounds<'a> {
@@ -164,7 +182,7 @@
prev_zwj = cat == wd::WC_ZWJ;
// if there's a category cached, grab it
cat = match self.cat {
- None => wd::word_category(ch),
+ None => wd::word_category(ch).2,
_ => self.cat.take().unwrap()
};
take_cat = true;
@@ -391,7 +409,7 @@
// if there's a category cached, grab it
cat = match self.catb {
- None => wd::word_category(ch),
+ None => wd::word_category(ch).2,
_ => self.catb.take().unwrap()
};
take_cat = true;
@@ -533,7 +551,7 @@
if regional_state == RegionalState::Unknown {
let count = self.string[..previdx]
.chars().rev()
- .map(|c| wd::word_category(c))
+ .map(|c| wd::word_category(c).2)
.filter(|&c| ! (c == wd::WC_ZWJ || c == wd::WC_Extend || c == wd::WC_Format))
.take_while(|&c| c == wd::WC_Regional_Indicator)
.count();
@@ -624,7 +642,7 @@
let nidx = idx + self.string[idx..].chars().next().unwrap().len_utf8();
if nidx < self.string.len() {
let nch = self.string[nidx..].chars().next().unwrap();
- Some(wd::word_category(nch))
+ Some(wd::word_category(nch).2)
} else {
None
}
@@ -635,7 +653,7 @@
use tables::word as wd;
if idx > 0 {
let nch = self.string[..idx].chars().next_back().unwrap();
- Some(wd::word_category(nch))
+ Some(wd::word_category(nch).2)
} else {
None
}
