sonicop 26.8.111 → 26.8.113

data/src/display_width.rs

@@ -4,55 +4,95 @@
 //! CJK text is wider than its character count and the two disagree about where a caret belongs.
 //! Both the alignment cops and the `clang`-style formatters need the same answer, which is why this
 //! lives outside either of them.
+//!
+//! The answer comes from [`crate::display_width_table`], generated from the gem itself. A hand
+//! written table stood here before and drifted: it counted the combining marks U+3099 and U+309A
+//! two columns each, because the blanket wide range `0x3041..=0x33FF` swallowed them, so a source
+//! file holding decomposed Japanese -- which is what macOS hands over routinely -- drew ten carets
+//! where RuboCop draws six. Reproducing an exception table by hand is what went wrong; the ranges
+//! are now taken from the reference implementation rather than restated.
+
+use crate::display_width_table::WIDTHS;
 
-/// `Unicode::DisplayWidth.of`, restricted to what a source file can hold: East Asian Wide and
-/// Fullwidth characters take two columns, combining marks take none, everything else takes one.
+/// `Unicode::DisplayWidth.of`, called the way RuboCop calls it: no options, one character at a
+/// time.
 pub fn display_width(text: &str) -> i64 {
-    text.chars().map(character_width).sum()
+    // `of` clamps its own result at zero, which is observable because BACKSPACE is worth -1: the
+    // gem answers 0 for `"ab\b\b\b\b"` rather than -2. Nothing is narrower than nothing.
+    text.chars().map(character_width).sum::<i64>().max(0)
 }
 
 fn character_width(character: char) -> i64 {
     let code = character as u32;
-    if is_zero_width(code) {
-        return 0;
-    }
-    if is_wide(code) { 2 } else { 1 }
+    // The table holds only the ranges whose width is not 1, sorted by their first code point and
+    // non-overlapping, so the range that could contain `code` is the last one starting at or below
+    // it. Anything else is one column wide.
+    let candidate = WIDTHS.partition_point(|(first, _, _)| *first <= code);
+    candidate
+        .checked_sub(1)
+        .and_then(|index| WIDTHS.get(index))
+        .filter(|(_, last, _)| code <= *last)
+        .map_or(1, |(_, _, width)| i64::from(*width))
 }
 
-fn is_zero_width(code: u32) -> bool {
-    matches!(code,
-        0x0300..=0x036F | 0x0483..=0x0489 | 0x0591..=0x05BD | 0x0610..=0x061A
-        | 0x064B..=0x065F | 0x0670 | 0x06D6..=0x06DC | 0x0730..=0x074A
-        | 0x07A6..=0x07B0 | 0x0816..=0x0819 | 0x081B..=0x0823 | 0x0825..=0x0827
-        | 0x0829..=0x082D | 0x0900..=0x0902 | 0x093A | 0x093C | 0x0941..=0x0948
-        | 0x094D | 0x0951..=0x0957 | 0x0962..=0x0963 | 0x0E31 | 0x0E34..=0x0E3A
-        | 0x0E47..=0x0E4E | 0x200B..=0x200F | 0x202A..=0x202E | 0x2060..=0x2064
-        | 0x20D0..=0x20F0 | 0xFE00..=0xFE0F | 0xFE20..=0xFE2F | 0xFEFF
-        | 0x1AB0..=0x1AFF | 0x1DC0..=0x1DFF)
-}
+#[cfg(test)]
+mod tests {
+    use super::{WIDTHS, display_width};
+
+    #[test]
+    fn ascii_is_one_column_per_character() {
+        assert_eq!(display_width(""), 0);
+        assert_eq!(display_width("x = 1"), 5);
+    }
+
+    #[test]
+    fn east_asian_wide_and_fullwidth_characters_take_two_columns() {
+        assert_eq!(display_width("日本語"), 6);
+        assert_eq!(display_width("ａ"), 2); // U+FF41 FULLWIDTH LATIN SMALL LETTER A
+        assert_eq!(display_width("ｱ"), 1); // U+FF71 halfwidth katakana stays narrow
+        assert_eq!(display_width("😀"), 2);
+    }
 
-/// The East Asian Wide and Fullwidth ranges, which is what makes a character two columns wide.
-fn is_wide(code: u32) -> bool {
-    matches!(code,
-        0x1100..=0x115F | 0x231A..=0x231B | 0x2329..=0x232A | 0x23E9..=0x23EC
-        | 0x23F0 | 0x23F3 | 0x25FD..=0x25FE | 0x2614..=0x2615 | 0x2648..=0x2653
-        | 0x267F | 0x2693 | 0x26A1 | 0x26AA..=0x26AB | 0x26BD..=0x26BE
-        | 0x26C4..=0x26C5 | 0x26CE | 0x26D4 | 0x26EA | 0x26F2..=0x26F3 | 0x26F5
-        | 0x26FA | 0x26FD | 0x2705 | 0x270A..=0x270B | 0x2728 | 0x274C | 0x274E
-        | 0x2753..=0x2755 | 0x2757 | 0x2795..=0x2797 | 0x27B0 | 0x27BF
-        | 0x2B1B..=0x2B1C | 0x2B50 | 0x2B55 | 0x2E80..=0x303E | 0x3041..=0x33FF
-        | 0x3400..=0x4DBF | 0x4E00..=0x9FFF | 0xA000..=0xA4CF | 0xA960..=0xA97F
-        | 0xAC00..=0xD7A3 | 0xF900..=0xFAFF | 0xFE10..=0xFE19 | 0xFE30..=0xFE6F
-        | 0xFF00..=0xFF60 | 0xFFE0..=0xFFE6 | 0x16FE0..=0x16FE4 | 0x17000..=0x18AFF
-        | 0x1B000..=0x1B2FF | 0x1F004 | 0x1F0CF | 0x1F18E | 0x1F191..=0x1F19A
-        | 0x1F200..=0x1F320 | 0x1F32D..=0x1F335 | 0x1F337..=0x1F37C
-        | 0x1F37E..=0x1F393 | 0x1F3A0..=0x1F3CA | 0x1F3CF..=0x1F3D3
-        | 0x1F3E0..=0x1F3F0 | 0x1F3F4 | 0x1F3F8..=0x1F43E | 0x1F440
-        | 0x1F442..=0x1F4FC | 0x1F4FF..=0x1F53D | 0x1F54B..=0x1F54E
-        | 0x1F550..=0x1F567 | 0x1F57A | 0x1F595..=0x1F596 | 0x1F5A4
-        | 0x1F5FB..=0x1F64F | 0x1F680..=0x1F6C5 | 0x1F6CC | 0x1F6D0..=0x1F6D2
-        | 0x1F6EB..=0x1F6EC | 0x1F6F4..=0x1F6FC | 0x1F7E0..=0x1F7EB
-        | 0x1F90C..=0x1F93A | 0x1F93C..=0x1F945 | 0x1F947..=0x1F978
-        | 0x1F97A..=0x1F9CB | 0x1F9CD..=0x1F9FF | 0x1FA70..=0x1FAF6
-        | 0x20000..=0x2FFFD | 0x30000..=0x3FFFD)
+    /// The regression this table was generated for. macOS hands over decomposed Japanese, and
+    /// counting the voiced sound mark as a character of its own made every one of them two columns
+    /// too wide: `x = "がが"` drew ten carets where RuboCop draws six.
+    #[test]
+    fn a_combining_mark_takes_no_columns_of_its_own() {
+        assert_eq!(display_width("\u{304B}\u{3099}"), 2); // か + ◌゙ = が
+        assert_eq!(display_width("\u{304C}"), 2); // the composed form agrees
+        assert_eq!(display_width("e\u{0301}"), 1); // e + ◌́
+        assert_eq!(display_width("\u{115F}"), 0); // HANGUL CHOSEONG FILLER
+    }
+
+    /// `of` never answers with less than zero, however many backspaces the text holds.
+    #[test]
+    fn backspace_takes_a_column_back_but_never_past_zero() {
+        assert_eq!(display_width("a\u{0008}"), 0);
+        assert_eq!(display_width("ab\u{0008}\u{0008}\u{0008}\u{0008}"), 0);
+        assert_eq!(display_width("\u{0008}"), 0);
+    }
+
+    /// Not every character is one or two columns wide, which is why the table stores a width rather
+    /// than a flag.
+    #[test]
+    fn the_two_em_dash_takes_three_columns() {
+        assert_eq!(display_width("\u{2E3B}"), 3);
+        assert_eq!(display_width("\u{2E3A}"), 2); // THREE-EM DASH's narrower sibling
+    }
+
+    /// The lookup is a binary search, so a regenerated table that is unsorted or overlapping would
+    /// answer wrongly rather than fail to build.
+    #[test]
+    fn the_generated_table_is_sorted_and_disjoint() {
+        for window in WIDTHS.windows(2) {
+            let (first, last, width) = window[0];
+            let (next_first, _, _) = window[1];
+            assert!(first <= last, "range 0x{first:04X} ends before it starts");
+            assert!(
+                last < next_first,
+                "range 0x{first:04X}..=0x{last:04X} overlaps 0x{next_first:04X}"
+            );
+            assert_ne!(width, 1, "the table only holds widths other than 1");
+        }
+    }
 }

data/src/display_width_table.rs

@@ -0,0 +1,506 @@
+//! Character widths as `Unicode::DisplayWidth` reports them. **Generated; do not edit.**
+//!
+//! Regenerate with `ruby scripts/dump_display_width.rb > src/display_width_table.rs`, which
+//! documents why the gem rather than the Unicode data files is the source.
+//!
+//! Generated from unicode-display_width 3.2.0 (Unicode 17.0.0).
+//!
+//! RuboCop's gemspec allows `>= 2.4.0, < 4.0`, so a user's own RuboCop may measure with a
+//! different release of the gem and disagree about a handful of code points. Pinning to one
+//! release is as close as a table can get; the version above is what the table reproduces.
+
+/// `(first, last, width)` for every inclusive range whose width is not 1, sorted by `first` and
+/// non-overlapping. A code point matching no range is one column wide.
+pub(crate) static WIDTHS: &[(u32, u32, i8)] = &[
+    (0x0000, 0x0000, 0),
+    (0x0005, 0x0005, 0),
+    (0x0007, 0x0007, 0),
+    (0x0008, 0x0008, -1),
+    (0x000A, 0x000F, 0),
+    (0x0300, 0x036F, 0),
+    (0x0483, 0x0489, 0),
+    (0x0591, 0x05BD, 0),
+    (0x05BF, 0x05BF, 0),
+    (0x05C1, 0x05C2, 0),
+    (0x05C4, 0x05C5, 0),
+    (0x05C7, 0x05C7, 0),
+    (0x0605, 0x0605, 0),
+    (0x0610, 0x061A, 0),
+    (0x061C, 0x061C, 0),
+    (0x064B, 0x065F, 0),
+    (0x0670, 0x0670, 0),
+    (0x06D6, 0x06DD, 0),
+    (0x06DF, 0x06E4, 0),
+    (0x06E7, 0x06E8, 0),
+    (0x06EA, 0x06ED, 0),
+    (0x070F, 0x070F, 0),
+    (0x0711, 0x0711, 0),
+    (0x0730, 0x074A, 0),
+    (0x07A6, 0x07B0, 0),
+    (0x07EB, 0x07F3, 0),
+    (0x07FD, 0x07FD, 0),
+    (0x0816, 0x0819, 0),
+    (0x081B, 0x0823, 0),
+    (0x0825, 0x0827, 0),
+    (0x0829, 0x082D, 0),
+    (0x0859, 0x085B, 0),
+    (0x0897, 0x089F, 0),
+    (0x08CA, 0x0902, 0),
+    (0x093A, 0x093A, 0),
+    (0x093C, 0x093C, 0),
+    (0x0941, 0x0948, 0),
+    (0x094D, 0x094D, 0),
+    (0x0951, 0x0957, 0),
+    (0x0962, 0x0963, 0),
+    (0x0981, 0x0981, 0),
+    (0x09BC, 0x09BC, 0),
+    (0x09C1, 0x09C4, 0),
+    (0x09CD, 0x09CD, 0),
+    (0x09E2, 0x09E3, 0),
+    (0x09FE, 0x09FE, 0),
+    (0x0A01, 0x0A02, 0),
+    (0x0A3C, 0x0A3C, 0),
+    (0x0A41, 0x0A42, 0),
+    (0x0A47, 0x0A48, 0),
+    (0x0A4B, 0x0A4D, 0),
+    (0x0A51, 0x0A51, 0),
+    (0x0A70, 0x0A71, 0),
+    (0x0A75, 0x0A75, 0),
+    (0x0A81, 0x0A82, 0),
+    (0x0ABC, 0x0ABC, 0),
+    (0x0AC1, 0x0AC5, 0),
+    (0x0AC7, 0x0AC8, 0),
+    (0x0ACD, 0x0ACD, 0),
+    (0x0AE2, 0x0AE3, 0),
+    (0x0AFA, 0x0AFF, 0),
+    (0x0B01, 0x0B01, 0),
+    (0x0B3C, 0x0B3C, 0),
+    (0x0B3F, 0x0B3F, 0),
+    (0x0B41, 0x0B44, 0),
+    (0x0B4D, 0x0B4D, 0),
+    (0x0B55, 0x0B56, 0),
+    (0x0B62, 0x0B63, 0),
+    (0x0B82, 0x0B82, 0),
+    (0x0BC0, 0x0BC0, 0),
+    (0x0BCD, 0x0BCD, 0),
+    (0x0C00, 0x0C00, 0),
+    (0x0C04, 0x0C04, 0),
+    (0x0C3C, 0x0C3C, 0),
+    (0x0C3E, 0x0C40, 0),
+    (0x0C46, 0x0C48, 0),
+    (0x0C4A, 0x0C4D, 0),
+    (0x0C55, 0x0C56, 0),
+    (0x0C62, 0x0C63, 0),
+    (0x0C81, 0x0C81, 0),
+    (0x0CBC, 0x0CBC, 0),
+    (0x0CBF, 0x0CBF, 0),
+    (0x0CC6, 0x0CC6, 0),
+    (0x0CCC, 0x0CCD, 0),
+    (0x0CE2, 0x0CE3, 0),
+    (0x0D00, 0x0D01, 0),
+    (0x0D3B, 0x0D3C, 0),
+    (0x0D41, 0x0D44, 0),
+    (0x0D4D, 0x0D4D, 0),
+    (0x0D62, 0x0D63, 0),
+    (0x0D81, 0x0D81, 0),
+    (0x0DCA, 0x0DCA, 0),
+    (0x0DD2, 0x0DD4, 0),
+    (0x0DD6, 0x0DD6, 0),
+    (0x0E31, 0x0E31, 0),
+    (0x0E34, 0x0E3A, 0),
+    (0x0E47, 0x0E4E, 0),
+    (0x0EB1, 0x0EB1, 0),
+    (0x0EB4, 0x0EBC, 0),
+    (0x0EC8, 0x0ECE, 0),
+    (0x0F18, 0x0F19, 0),
+    (0x0F35, 0x0F35, 0),
+    (0x0F37, 0x0F37, 0),
+    (0x0F39, 0x0F39, 0),
+    (0x0F71, 0x0F7E, 0),
+    (0x0F80, 0x0F84, 0),
+    (0x0F86, 0x0F87, 0),
+    (0x0F8D, 0x0F97, 0),
+    (0x0F99, 0x0FBC, 0),
+    (0x0FC6, 0x0FC6, 0),
+    (0x102D, 0x1030, 0),
+    (0x1032, 0x1037, 0),
+    (0x1039, 0x103A, 0),
+    (0x103D, 0x103E, 0),
+    (0x1058, 0x1059, 0),
+    (0x105E, 0x1060, 0),
+    (0x1071, 0x1074, 0),
+    (0x1082, 0x1082, 0),
+    (0x1085, 0x1086, 0),
+    (0x108D, 0x108D, 0),
+    (0x109D, 0x109D, 0),
+    (0x1100, 0x115E, 2),
+    (0x115F, 0x11FF, 0),
+    (0x135D, 0x135F, 0),
+    (0x1712, 0x1714, 0),
+    (0x1732, 0x1733, 0),
+    (0x1752, 0x1753, 0),
+    (0x1772, 0x1773, 0),
+    (0x17B4, 0x17B5, 0),
+    (0x17B7, 0x17BD, 0),
+    (0x17C6, 0x17C6, 0),
+    (0x17C9, 0x17D3, 0),
+    (0x17DD, 0x17DD, 0),
+    (0x180B, 0x180F, 0),
+    (0x1885, 0x1886, 0),
+    (0x18A9, 0x18A9, 0),
+    (0x1920, 0x1922, 0),
+    (0x1927, 0x1928, 0),
+    (0x1932, 0x1932, 0),
+    (0x1939, 0x193B, 0),
+    (0x1A17, 0x1A18, 0),
+    (0x1A1B, 0x1A1B, 0),
+    (0x1A56, 0x1A56, 0),
+    (0x1A58, 0x1A5E, 0),
+    (0x1A60, 0x1A60, 0),
+    (0x1A62, 0x1A62, 0),
+    (0x1A65, 0x1A6C, 0),
+    (0x1A73, 0x1A7C, 0),
+    (0x1A7F, 0x1A7F, 0),
+    (0x1AB0, 0x1AEF, 0),
+    (0x1B00, 0x1B03, 0),
+    (0x1B34, 0x1B34, 0),
+    (0x1B36, 0x1B3A, 0),
+    (0x1B3C, 0x1B3C, 0),
+    (0x1B42, 0x1B42, 0),
+    (0x1B6B, 0x1B73, 0),
+    (0x1B80, 0x1B81, 0),
+    (0x1BA2, 0x1BA5, 0),
+    (0x1BA8, 0x1BA9, 0),
+    (0x1BAB, 0x1BAD, 0),
+    (0x1BE6, 0x1BE6, 0),
+    (0x1BE8, 0x1BE9, 0),
+    (0x1BED, 0x1BED, 0),
+    (0x1BEF, 0x1BF1, 0),
+    (0x1C2C, 0x1C33, 0),
+    (0x1C36, 0x1C37, 0),
+    (0x1CD0, 0x1CD2, 0),
+    (0x1CD4, 0x1CE0, 0),
+    (0x1CE2, 0x1CE8, 0),
+    (0x1CED, 0x1CED, 0),
+    (0x1CF4, 0x1CF4, 0),
+    (0x1CF8, 0x1CF9, 0),
+    (0x1DC0, 0x1DFF, 0),
+    (0x200B, 0x200F, 0),
+    (0x2028, 0x202E, 0),
+    (0x2060, 0x206F, 0),
+    (0x20D0, 0x20FF, 0),
+    (0x231A, 0x231B, 2),
+    (0x2329, 0x232A, 2),
+    (0x23E9, 0x23EC, 2),
+    (0x23F0, 0x23F0, 2),
+    (0x23F3, 0x23F3, 2),
+    (0x25FD, 0x25FE, 2),
+    (0x2614, 0x2615, 2),
+    (0x2630, 0x2637, 2),
+    (0x2648, 0x2653, 2),
+    (0x267F, 0x267F, 2),
+    (0x268A, 0x268F, 2),
+    (0x2693, 0x2693, 2),
+    (0x26A1, 0x26A1, 2),
+    (0x26AA, 0x26AB, 2),
+    (0x26BD, 0x26BE, 2),
+    (0x26C4, 0x26C5, 2),
+    (0x26CE, 0x26CE, 2),
+    (0x26D4, 0x26D4, 2),
+    (0x26EA, 0x26EA, 2),
+    (0x26F2, 0x26F3, 2),
+    (0x26F5, 0x26F5, 2),
+    (0x26FA, 0x26FA, 2),
+    (0x26FD, 0x26FD, 2),
+    (0x2705, 0x2705, 2),
+    (0x270A, 0x270B, 2),
+    (0x2728, 0x2728, 2),
+    (0x274C, 0x274C, 2),
+    (0x274E, 0x274E, 2),
+    (0x2753, 0x2755, 2),
+    (0x2757, 0x2757, 2),
+    (0x2795, 0x2797, 2),
+    (0x27B0, 0x27B0, 2),
+    (0x27BF, 0x27BF, 2),
+    (0x2B1B, 0x2B1C, 2),
+    (0x2B50, 0x2B50, 2),
+    (0x2B55, 0x2B55, 2),
+    (0x2CEF, 0x2CF1, 0),
+    (0x2D7F, 0x2D7F, 0),
+    (0x2DE0, 0x2DFF, 0),
+    (0x2E3A, 0x2E3A, 2),
+    (0x2E3B, 0x2E3B, 3),
+    (0x2E80, 0x2E99, 2),
+    (0x2E9B, 0x2FDF, 2),
+    (0x2FF0, 0x3029, 2),
+    (0x302A, 0x302D, 0),
+    (0x302E, 0x303E, 2),
+    (0x3041, 0x3096, 2),
+    (0x3099, 0x309A, 0),
+    (0x309B, 0x30FF, 2),
+    (0x3105, 0x312F, 2),
+    (0x3131, 0x3163, 2),
+    (0x3164, 0x3164, 0),
+    (0x3165, 0x31E5, 2),
+    (0x31EF, 0x3247, 2),
+    (0x3250, 0xA4CF, 2),
+    (0xA66F, 0xA672, 0),
+    (0xA674, 0xA67D, 0),
+    (0xA69E, 0xA69F, 0),
+    (0xA6F0, 0xA6F1, 0),
+    (0xA802, 0xA802, 0),
+    (0xA806, 0xA806, 0),
+    (0xA80B, 0xA80B, 0),
+    (0xA825, 0xA826, 0),
+    (0xA82C, 0xA82C, 0),
+    (0xA8C4, 0xA8C5, 0),
+    (0xA8E0, 0xA8F1, 0),
+    (0xA8FF, 0xA8FF, 0),
+    (0xA926, 0xA92D, 0),
+    (0xA947, 0xA951, 0),
+    (0xA960, 0xA97F, 2),
+    (0xA980, 0xA982, 0),
+    (0xA9B3, 0xA9B3, 0),
+    (0xA9B6, 0xA9B9, 0),
+    (0xA9BC, 0xA9BD, 0),
+    (0xA9E5, 0xA9E5, 0),
+    (0xAA29, 0xAA2E, 0),
+    (0xAA31, 0xAA32, 0),
+    (0xAA35, 0xAA36, 0),
+    (0xAA43, 0xAA43, 0),
+    (0xAA4C, 0xAA4C, 0),
+    (0xAA7C, 0xAA7C, 0),
+    (0xAAB0, 0xAAB0, 0),
+    (0xAAB2, 0xAAB4, 0),
+    (0xAAB7, 0xAAB8, 0),
+    (0xAABE, 0xAABF, 0),
+    (0xAAC1, 0xAAC1, 0),
+    (0xAAEC, 0xAAED, 0),
+    (0xAAF6, 0xAAF6, 0),
+    (0xABE5, 0xABE5, 0),
+    (0xABE8, 0xABE8, 0),
+    (0xABED, 0xABED, 0),
+    (0xAC00, 0xD7AF, 2),
+    (0xD7B0, 0xD7FF, 0),
+    (0xF900, 0xFAFF, 2),
+    (0xFB1E, 0xFB1E, 0),
+    (0xFE00, 0xFE0F, 0),
+    (0xFE10, 0xFE1F, 2),
+    (0xFE20, 0xFE2F, 0),
+    (0xFE30, 0xFE52, 2),
+    (0xFE54, 0xFE66, 2),
+    (0xFE68, 0xFE6B, 2),
+    (0xFEFF, 0xFEFF, 0),
+    (0xFF01, 0xFF60, 2),
+    (0xFFA0, 0xFFA0, 0),
+    (0xFFE0, 0xFFE6, 2),
+    (0xFFF0, 0xFFFB, 0),
+    (0x101FD, 0x101FD, 0),
+    (0x102E0, 0x102E0, 0),
+    (0x10376, 0x1037A, 0),
+    (0x10A01, 0x10A03, 0),
+    (0x10A05, 0x10A06, 0),
+    (0x10A0C, 0x10A0F, 0),
+    (0x10A38, 0x10A3A, 0),
+    (0x10A3F, 0x10A3F, 0),
+    (0x10AE5, 0x10AE6, 0),
+    (0x10D24, 0x10D27, 0),
+    (0x10D69, 0x10D6D, 0),
+    (0x10EAB, 0x10EAC, 0),
+    (0x10EFA, 0x10EFF, 0),
+    (0x10F46, 0x10F50, 0),
+    (0x10F82, 0x10F85, 0),
+    (0x11001, 0x11001, 0),
+    (0x11038, 0x11046, 0),
+    (0x11070, 0x11070, 0),
+    (0x11073, 0x11074, 0),
+    (0x1107F, 0x11081, 0),
+    (0x110B3, 0x110B6, 0),
+    (0x110B9, 0x110BA, 0),
+    (0x110BD, 0x110BD, 0),
+    (0x110C2, 0x110C2, 0),
+    (0x110CD, 0x110CD, 0),
+    (0x11100, 0x11102, 0),
+    (0x11127, 0x1112B, 0),
+    (0x1112D, 0x11134, 0),
+    (0x11173, 0x11173, 0),
+    (0x11180, 0x11181, 0),
+    (0x111B6, 0x111BE, 0),
+    (0x111C9, 0x111CC, 0),
+    (0x111CF, 0x111CF, 0),
+    (0x1122F, 0x11231, 0),
+    (0x11234, 0x11234, 0),
+    (0x11236, 0x11237, 0),
+    (0x1123E, 0x1123E, 0),
+    (0x11241, 0x11241, 0),
+    (0x112DF, 0x112DF, 0),
+    (0x112E3, 0x112EA, 0),
+    (0x11300, 0x11301, 0),
+    (0x1133B, 0x1133C, 0),
+    (0x11340, 0x11340, 0),
+    (0x11366, 0x1136C, 0),
+    (0x11370, 0x1137F, 0),
+    (0x113BB, 0x113C0, 0),
+    (0x113CE, 0x113CE, 0),
+    (0x113D0, 0x113D0, 0),
+    (0x113D2, 0x113D2, 0),
+    (0x113E1, 0x113E2, 0),
+    (0x11438, 0x1143F, 0),
+    (0x11442, 0x11444, 0),
+    (0x11446, 0x11446, 0),
+    (0x1145E, 0x1145E, 0),
+    (0x114B3, 0x114B8, 0),
+    (0x114BA, 0x114BA, 0),
+    (0x114BF, 0x114C0, 0),
+    (0x114C2, 0x114C3, 0),
+    (0x115B2, 0x115B5, 0),
+    (0x115BC, 0x115BD, 0),
+    (0x115BF, 0x115C0, 0),
+    (0x115DC, 0x115DD, 0),
+    (0x11633, 0x1163A, 0),
+    (0x1163D, 0x1163D, 0),
+    (0x1163F, 0x11640, 0),
+    (0x116AB, 0x116AB, 0),
+    (0x116AD, 0x116AD, 0),
+    (0x116B0, 0x116B5, 0),
+    (0x116B7, 0x116B7, 0),
+    (0x1171D, 0x1171D, 0),
+    (0x1171F, 0x1171F, 0),
+    (0x11722, 0x11725, 0),
+    (0x11727, 0x1172B, 0),
+    (0x1182F, 0x11837, 0),
+    (0x11839, 0x1183A, 0),
+    (0x1193B, 0x1193C, 0),
+    (0x1193E, 0x1193E, 0),
+    (0x11943, 0x11943, 0),
+    (0x119D4, 0x119D7, 0),
+    (0x119DA, 0x119DB, 0),
+    (0x119E0, 0x119E0, 0),
+    (0x11A01, 0x11A0A, 0),
+    (0x11A33, 0x11A38, 0),
+    (0x11A3B, 0x11A3E, 0),
+    (0x11A47, 0x11A47, 0),
+    (0x11A51, 0x11A56, 0),
+    (0x11A59, 0x11A5B, 0),
+    (0x11A8A, 0x11A96, 0),
+    (0x11A98, 0x11A99, 0),
+    (0x11B60, 0x11B60, 0),
+    (0x11B62, 0x11B64, 0),
+    (0x11B66, 0x11B66, 0),
+    (0x11C30, 0x11C36, 0),
+    (0x11C38, 0x11C3D, 0),
+    (0x11C3F, 0x11C3F, 0),
+    (0x11C92, 0x11CA7, 0),
+    (0x11CAA, 0x11CB0, 0),
+    (0x11CB2, 0x11CB3, 0),
+    (0x11CB5, 0x11CB6, 0),
+    (0x11D31, 0x11D36, 0),
+    (0x11D3A, 0x11D3A, 0),
+    (0x11D3C, 0x11D3D, 0),
+    (0x11D3F, 0x11D45, 0),
+    (0x11D47, 0x11D47, 0),
+    (0x11D90, 0x11D91, 0),
+    (0x11D95, 0x11D95, 0),
+    (0x11D97, 0x11D97, 0),
+    (0x11EF3, 0x11EF4, 0),
+    (0x11F00, 0x11F01, 0),
+    (0x11F36, 0x11F3A, 0),
+    (0x11F40, 0x11F40, 0),
+    (0x11F42, 0x11F42, 0),
+    (0x11F5A, 0x11F5A, 0),
+    (0x13430, 0x13440, 0),
+    (0x13447, 0x1345F, 0),
+    (0x1611E, 0x16129, 0),
+    (0x1612D, 0x1612F, 0),
+    (0x16AF0, 0x16AF4, 0),
+    (0x16B30, 0x16B36, 0),
+    (0x16F4F, 0x16F4F, 0),
+    (0x16F8F, 0x16F92, 0),
+    (0x16FE0, 0x16FE3, 2),
+    (0x16FE4, 0x16FE4, 0),
+    (0x16FF0, 0x18CDF, 2),
+    (0x18CFF, 0x18D1F, 2),
+    (0x18D80, 0x18DFF, 2),
+    (0x1AFF0, 0x1AFF3, 2),
+    (0x1AFF5, 0x1AFFB, 2),
+    (0x1AFFD, 0x1AFFE, 2),
+    (0x1B000, 0x1B12F, 2),
+    (0x1B132, 0x1B132, 2),
+    (0x1B150, 0x1B152, 2),
+    (0x1B155, 0x1B155, 2),
+    (0x1B164, 0x1B167, 2),
+    (0x1B170, 0x1B2FF, 2),
+    (0x1BC9D, 0x1BC9E, 0),
+    (0x1BCA0, 0x1BCAF, 0),
+    (0x1CF00, 0x1CF4F, 0),
+    (0x1D167, 0x1D169, 0),
+    (0x1D173, 0x1D182, 0),
+    (0x1D185, 0x1D18B, 0),
+    (0x1D1AA, 0x1D1AD, 0),
+    (0x1D242, 0x1D244, 0),
+    (0x1D300, 0x1D376, 2),
+    (0x1DA00, 0x1DA36, 0),
+    (0x1DA3B, 0x1DA6C, 0),
+    (0x1DA75, 0x1DA75, 0),
+    (0x1DA84, 0x1DA84, 0),
+    (0x1DA9B, 0x1DA9F, 0),
+    (0x1DAA1, 0x1DAAF, 0),
+    (0x1E000, 0x1E006, 0),
+    (0x1E008, 0x1E018, 0),
+    (0x1E01B, 0x1E021, 0),
+    (0x1E023, 0x1E024, 0),
+    (0x1E026, 0x1E02A, 0),
+    (0x1E08F, 0x1E08F, 0),
+    (0x1E130, 0x1E136, 0),
+    (0x1E2AE, 0x1E2AE, 0),
+    (0x1E2EC, 0x1E2EF, 0),
+    (0x1E4EC, 0x1E4EF, 0),
+    (0x1E5EE, 0x1E5EF, 0),
+    (0x1E6E3, 0x1E6E3, 0),
+    (0x1E6E6, 0x1E6E6, 0),
+    (0x1E6EE, 0x1E6EF, 0),
+    (0x1E6F5, 0x1E6F5, 0),
+    (0x1E8D0, 0x1E8DF, 0),
+    (0x1E944, 0x1E94A, 0),
+    (0x1F004, 0x1F004, 2),
+    (0x1F0CF, 0x1F0CF, 2),
+    (0x1F18E, 0x1F18E, 2),
+    (0x1F191, 0x1F19A, 2),
+    (0x1F200, 0x1F320, 2),
+    (0x1F32D, 0x1F335, 2),
+    (0x1F337, 0x1F37C, 2),
+    (0x1F37E, 0x1F393, 2),
+    (0x1F3A0, 0x1F3CA, 2),
+    (0x1F3CF, 0x1F3D3, 2),
+    (0x1F3E0, 0x1F3F0, 2),
+    (0x1F3F4, 0x1F3F4, 2),
+    (0x1F3F8, 0x1F43E, 2),
+    (0x1F440, 0x1F440, 2),
+    (0x1F442, 0x1F4FC, 2),
+    (0x1F4FF, 0x1F53D, 2),
+    (0x1F54B, 0x1F54E, 2),
+    (0x1F550, 0x1F567, 2),
+    (0x1F57A, 0x1F57A, 2),
+    (0x1F595, 0x1F596, 2),
+    (0x1F5A4, 0x1F5A4, 2),
+    (0x1F5FB, 0x1F64F, 2),
+    (0x1F680, 0x1F6C5, 2),
+    (0x1F6CC, 0x1F6CC, 2),
+    (0x1F6D0, 0x1F6D2, 2),
+    (0x1F6D5, 0x1F6D8, 2),
+    (0x1F6DC, 0x1F6DF, 2),
+    (0x1F6EB, 0x1F6EC, 2),
+    (0x1F6F4, 0x1F6FC, 2),
+    (0x1F7E0, 0x1F7FF, 2),
+    (0x1F90C, 0x1F93A, 2),
+    (0x1F93C, 0x1F945, 2),
+    (0x1F947, 0x1F9FF, 2),
+    (0x1FA70, 0x1FA8A, 2),
+    (0x1FA8E, 0x1FAC6, 2),
+    (0x1FAC8, 0x1FAC8, 2),
+    (0x1FACD, 0x1FADC, 2),
+    (0x1FADF, 0x1FAEA, 2),
+    (0x1FAEF, 0x1FAFF, 2),
+    (0x20000, 0x3FFFF, 2),
+    (0xE0000, 0xEFFFF, 0),
+];