ferray-strings 0.4.9

// ferray-strings: String classification functions (is* family)
//
// Elementwise boolean classification matching NumPy's
// `numpy.strings.isalpha`, `numpy.strings.isdigit`, etc.
//
// ## REQ status
// All classification predicates below match CPython `str.is*()` element
// by element, which is exactly what `numpy.strings`/`numpy.char` delegate
// to. The non-ASCII / category-sensitive predicates (`isalpha`,
// `isdecimal`, `isdigit`, `isupper`, `islower`, `istitle`, `isspace`)
// are backed by sorted, non-overlapping codepoint-range tables derived
// LIVE from the oracle CPython 3.13 / Unicode 15.1.0 (the version backing
// the installed numpy 2.4 build this crate is verified against), because
// Rust `std` and third-party Unicode crates track a *newer* Unicode
// revision and would classify ~9000 codepoints (assigned post-15.1.0)
// differently from the numpy oracle. `isnumeric` is likewise backed by a
// 15.1.0-locked table (`NUMERIC_TYPE_RANGES`): Rust `char::is_numeric`
// tests the general categories `Nd|Nl|No`, which does NOT coincide with
// the Unicode `Numeric_Type` property CPython `str.isnumeric()` uses — it
// MISSES ideographic numerals like `一` (U+4E00, category `Lo`,
// `Numeric_Type=Numeric`) and ADDS codepoints assigned after 15.1.0.

use ferray_core::Array;
use ferray_core::dimension::Dimension;
use ferray_core::error::FerrayResult;

use crate::string_array::StringArray;

/// Binary-search a sorted, non-overlapping list of inclusive codepoint
/// ranges for `c`.
fn in_ranges(table: &[(char, char)], c: char) -> bool {
    table
        .binary_search_by(|&(lo, hi)| {
            if c < lo {
                core::cmp::Ordering::Greater
            } else if c > hi {
                core::cmp::Ordering::Less
            } else {
                core::cmp::Ordering::Equal
            }
        })
        .is_ok()
}

/// Inclusive codepoint ranges of characters with Unicode
/// `Numeric_Type` of Decimal **or** Digit — i.e. exactly the set for
/// which CPython's `str.isdigit()` returns `True` for a single
/// character (`Py_UNICODE_ISDIGIT`). This is broader than the Decimal-
/// only `isdecimal` (it adds superscripts/subscripts ²³⁴, circled ①,
/// parenthesized ⑴, etc.) and narrower than `isnumeric` (it excludes
/// fractions ½ and letter-numerals like Ⅻ).
///
/// Derived live from the oracle CPython 3.13 / Unicode 15.1.0 (the
/// version backing the installed `numpy` 2.4 build that this crate is
/// verified against):
/// `[c for c in range(0x110000) if chr(c).isdigit()]`, run-length
/// compressed to 83 ranges. Rust's `std` exposes no `Numeric_Type`
/// predicate (`char::is_numeric` covers `Nd|Nl|No` — too broad — and
/// `char::to_digit(10)` is ASCII-only — too narrow), so the table is
/// embedded directly.
const DIGIT_RANGES: &[(char, char)] = &[
    ('\u{0030}', '\u{0039}'),
    ('\u{00B2}', '\u{00B3}'),
    ('\u{00B9}', '\u{00B9}'),
    ('\u{0660}', '\u{0669}'),
    ('\u{06F0}', '\u{06F9}'),
    ('\u{07C0}', '\u{07C9}'),
    ('\u{0966}', '\u{096F}'),
    ('\u{09E6}', '\u{09EF}'),
    ('\u{0A66}', '\u{0A6F}'),
    ('\u{0AE6}', '\u{0AEF}'),
    ('\u{0B66}', '\u{0B6F}'),
    ('\u{0BE6}', '\u{0BEF}'),
    ('\u{0C66}', '\u{0C6F}'),
    ('\u{0CE6}', '\u{0CEF}'),
    ('\u{0D66}', '\u{0D6F}'),
    ('\u{0DE6}', '\u{0DEF}'),
    ('\u{0E50}', '\u{0E59}'),
    ('\u{0ED0}', '\u{0ED9}'),
    ('\u{0F20}', '\u{0F29}'),
    ('\u{1040}', '\u{1049}'),
    ('\u{1090}', '\u{1099}'),
    ('\u{1369}', '\u{1371}'),
    ('\u{17E0}', '\u{17E9}'),
    ('\u{1810}', '\u{1819}'),
    ('\u{1946}', '\u{194F}'),
    ('\u{19D0}', '\u{19DA}'),
    ('\u{1A80}', '\u{1A89}'),
    ('\u{1A90}', '\u{1A99}'),
    ('\u{1B50}', '\u{1B59}'),
    ('\u{1BB0}', '\u{1BB9}'),
    ('\u{1C40}', '\u{1C49}'),
    ('\u{1C50}', '\u{1C59}'),
    ('\u{2070}', '\u{2070}'),
    ('\u{2074}', '\u{2079}'),
    ('\u{2080}', '\u{2089}'),
    ('\u{2460}', '\u{2468}'),
    ('\u{2474}', '\u{247C}'),
    ('\u{2488}', '\u{2490}'),
    ('\u{24EA}', '\u{24EA}'),
    ('\u{24F5}', '\u{24FD}'),
    ('\u{24FF}', '\u{24FF}'),
    ('\u{2776}', '\u{277E}'),
    ('\u{2780}', '\u{2788}'),
    ('\u{278A}', '\u{2792}'),
    ('\u{A620}', '\u{A629}'),
    ('\u{A8D0}', '\u{A8D9}'),
    ('\u{A900}', '\u{A909}'),
    ('\u{A9D0}', '\u{A9D9}'),
    ('\u{A9F0}', '\u{A9F9}'),
    ('\u{AA50}', '\u{AA59}'),
    ('\u{ABF0}', '\u{ABF9}'),
    ('\u{FF10}', '\u{FF19}'),
    ('\u{104A0}', '\u{104A9}'),
    ('\u{10A40}', '\u{10A43}'),
    ('\u{10D30}', '\u{10D39}'),
    ('\u{10E60}', '\u{10E68}'),
    ('\u{11052}', '\u{1105A}'),
    ('\u{11066}', '\u{1106F}'),
    ('\u{110F0}', '\u{110F9}'),
    ('\u{11136}', '\u{1113F}'),
    ('\u{111D0}', '\u{111D9}'),
    ('\u{112F0}', '\u{112F9}'),
    ('\u{11450}', '\u{11459}'),
    ('\u{114D0}', '\u{114D9}'),
    ('\u{11650}', '\u{11659}'),
    ('\u{116C0}', '\u{116C9}'),
    ('\u{11730}', '\u{11739}'),
    ('\u{118E0}', '\u{118E9}'),
    ('\u{11950}', '\u{11959}'),
    ('\u{11C50}', '\u{11C59}'),
    ('\u{11D50}', '\u{11D59}'),
    ('\u{11DA0}', '\u{11DA9}'),
    ('\u{11F50}', '\u{11F59}'),
    ('\u{16A60}', '\u{16A69}'),
    ('\u{16AC0}', '\u{16AC9}'),
    ('\u{16B50}', '\u{16B59}'),
    ('\u{1D7CE}', '\u{1D7FF}'),
    ('\u{1E140}', '\u{1E149}'),
    ('\u{1E2F0}', '\u{1E2F9}'),
    ('\u{1E4F0}', '\u{1E4F9}'),
    ('\u{1E950}', '\u{1E959}'),
    ('\u{1F100}', '\u{1F10A}'),
    ('\u{1FBF0}', '\u{1FBF9}'),
];

/// Inclusive codepoint ranges of characters with Unicode
/// `Numeric_Type` of Decimal — i.e. the set for which CPython's
/// `str.isdecimal()` returns `True` for a single character
/// (`Py_UNICODE_ISDECIMAL`). This is the Decimal-only subset of
/// [`DIGIT_RANGES`]: it drops superscripts/subscripts (`²³⁴`), circled
/// (`①`) and parenthesized (`⑴`) digits, keeping only true positional
/// decimal digits (ASCII `0-9`, fullwidth `０-９`, Arabic-Indic `٠-٩`,
/// Devanagari, etc.).
///
/// Derived live from the oracle CPython 3.13 / Unicode 15.1.0:
/// `[c for c in range(0x110000) if chr(c).isdecimal()]`, run-length
/// compressed to 64 ranges.
const DECIMAL_RANGES: &[(char, char)] = &[
    ('\u{0030}', '\u{0039}'),
    ('\u{0660}', '\u{0669}'),
    ('\u{06F0}', '\u{06F9}'),
    ('\u{07C0}', '\u{07C9}'),
    ('\u{0966}', '\u{096F}'),
    ('\u{09E6}', '\u{09EF}'),
    ('\u{0A66}', '\u{0A6F}'),
    ('\u{0AE6}', '\u{0AEF}'),
    ('\u{0B66}', '\u{0B6F}'),
    ('\u{0BE6}', '\u{0BEF}'),
    ('\u{0C66}', '\u{0C6F}'),
    ('\u{0CE6}', '\u{0CEF}'),
    ('\u{0D66}', '\u{0D6F}'),
    ('\u{0DE6}', '\u{0DEF}'),
    ('\u{0E50}', '\u{0E59}'),
    ('\u{0ED0}', '\u{0ED9}'),
    ('\u{0F20}', '\u{0F29}'),
    ('\u{1040}', '\u{1049}'),
    ('\u{1090}', '\u{1099}'),
    ('\u{17E0}', '\u{17E9}'),
    ('\u{1810}', '\u{1819}'),
    ('\u{1946}', '\u{194F}'),
    ('\u{19D0}', '\u{19D9}'),
    ('\u{1A80}', '\u{1A89}'),
    ('\u{1A90}', '\u{1A99}'),
    ('\u{1B50}', '\u{1B59}'),
    ('\u{1BB0}', '\u{1BB9}'),
    ('\u{1C40}', '\u{1C49}'),
    ('\u{1C50}', '\u{1C59}'),
    ('\u{A620}', '\u{A629}'),
    ('\u{A8D0}', '\u{A8D9}'),
    ('\u{A900}', '\u{A909}'),
    ('\u{A9D0}', '\u{A9D9}'),
    ('\u{A9F0}', '\u{A9F9}'),
    ('\u{AA50}', '\u{AA59}'),
    ('\u{ABF0}', '\u{ABF9}'),
    ('\u{FF10}', '\u{FF19}'),
    ('\u{104A0}', '\u{104A9}'),
    ('\u{10D30}', '\u{10D39}'),
    ('\u{11066}', '\u{1106F}'),
    ('\u{110F0}', '\u{110F9}'),
    ('\u{11136}', '\u{1113F}'),
    ('\u{111D0}', '\u{111D9}'),
    ('\u{112F0}', '\u{112F9}'),
    ('\u{11450}', '\u{11459}'),
    ('\u{114D0}', '\u{114D9}'),
    ('\u{11650}', '\u{11659}'),
    ('\u{116C0}', '\u{116C9}'),
    ('\u{11730}', '\u{11739}'),
    ('\u{118E0}', '\u{118E9}'),
    ('\u{11950}', '\u{11959}'),
    ('\u{11C50}', '\u{11C59}'),
    ('\u{11D50}', '\u{11D59}'),
    ('\u{11DA0}', '\u{11DA9}'),
    ('\u{11F50}', '\u{11F59}'),
    ('\u{16A60}', '\u{16A69}'),
    ('\u{16AC0}', '\u{16AC9}'),
    ('\u{16B50}', '\u{16B59}'),
    ('\u{1D7CE}', '\u{1D7FF}'),
    ('\u{1E140}', '\u{1E149}'),
    ('\u{1E2F0}', '\u{1E2F9}'),
    ('\u{1E4F0}', '\u{1E4F9}'),
    ('\u{1E950}', '\u{1E959}'),
    ('\u{1FBF0}', '\u{1FBF9}'),
];

/// Inclusive codepoint ranges of characters in Unicode general category
/// `L*` (`Lu | Ll | Lt | Lm | Lo`), i.e. exactly the set for which
/// CPython's `str.isalpha()` returns `True` for a single character. This
/// is NARROWER than Rust's `char::is_alphabetic` (the `Alphabetic`
/// derived property, which also includes combining marks `Mn`, letter-
/// numbers `Nl` such as roman numerals `Ⅻ`, etc.).
///
/// Derived live from the oracle CPython 3.13 / Unicode 15.1.0:
/// `[c for c in range(0x110000) if chr(c).isalpha()]`, run-length
/// compressed to 660 ranges.
const ALPHA_RANGES: &[(char, char)] = &[
    ('\u{0041}', '\u{005A}'),
    ('\u{0061}', '\u{007A}'),
    ('\u{00AA}', '\u{00AA}'),
    ('\u{00B5}', '\u{00B5}'),
    ('\u{00BA}', '\u{00BA}'),
    ('\u{00C0}', '\u{00D6}'),
    ('\u{00D8}', '\u{00F6}'),
    ('\u{00F8}', '\u{02C1}'),
    ('\u{02C6}', '\u{02D1}'),
    ('\u{02E0}', '\u{02E4}'),
    ('\u{02EC}', '\u{02EC}'),
    ('\u{02EE}', '\u{02EE}'),
    ('\u{0370}', '\u{0374}'),
    ('\u{0376}', '\u{0377}'),
    ('\u{037A}', '\u{037D}'),
    ('\u{037F}', '\u{037F}'),
    ('\u{0386}', '\u{0386}'),
    ('\u{0388}', '\u{038A}'),
    ('\u{038C}', '\u{038C}'),
    ('\u{038E}', '\u{03A1}'),
    ('\u{03A3}', '\u{03F5}'),
    ('\u{03F7}', '\u{0481}'),
    ('\u{048A}', '\u{052F}'),
    ('\u{0531}', '\u{0556}'),
    ('\u{0559}', '\u{0559}'),
    ('\u{0560}', '\u{0588}'),
    ('\u{05D0}', '\u{05EA}'),
    ('\u{05EF}', '\u{05F2}'),
    ('\u{0620}', '\u{064A}'),
    ('\u{066E}', '\u{066F}'),
    ('\u{0671}', '\u{06D3}'),
    ('\u{06D5}', '\u{06D5}'),
    ('\u{06E5}', '\u{06E6}'),
    ('\u{06EE}', '\u{06EF}'),
    ('\u{06FA}', '\u{06FC}'),
    ('\u{06FF}', '\u{06FF}'),
    ('\u{0710}', '\u{0710}'),
    ('\u{0712}', '\u{072F}'),
    ('\u{074D}', '\u{07A5}'),
    ('\u{07B1}', '\u{07B1}'),
    ('\u{07CA}', '\u{07EA}'),
    ('\u{07F4}', '\u{07F5}'),
    ('\u{07FA}', '\u{07FA}'),
    ('\u{0800}', '\u{0815}'),
    ('\u{081A}', '\u{081A}'),
    ('\u{0824}', '\u{0824}'),
    ('\u{0828}', '\u{0828}'),
    ('\u{0840}', '\u{0858}'),
    ('\u{0860}', '\u{086A}'),
    ('\u{0870}', '\u{0887}'),
    ('\u{0889}', '\u{088E}'),
    ('\u{08A0}', '\u{08C9}'),
    ('\u{0904}', '\u{0939}'),
    ('\u{093D}', '\u{093D}'),
    ('\u{0950}', '\u{0950}'),
    ('\u{0958}', '\u{0961}'),
    ('\u{0971}', '\u{0980}'),
    ('\u{0985}', '\u{098C}'),
    ('\u{098F}', '\u{0990}'),
    ('\u{0993}', '\u{09A8}'),
    ('\u{09AA}', '\u{09B0}'),
    ('\u{09B2}', '\u{09B2}'),
    ('\u{09B6}', '\u{09B9}'),
    ('\u{09BD}', '\u{09BD}'),
    ('\u{09CE}', '\u{09CE}'),
    ('\u{09DC}', '\u{09DD}'),
    ('\u{09DF}', '\u{09E1}'),
    ('\u{09F0}', '\u{09F1}'),
    ('\u{09FC}', '\u{09FC}'),
    ('\u{0A05}', '\u{0A0A}'),
    ('\u{0A0F}', '\u{0A10}'),
    ('\u{0A13}', '\u{0A28}'),
    ('\u{0A2A}', '\u{0A30}'),
    ('\u{0A32}', '\u{0A33}'),
    ('\u{0A35}', '\u{0A36}'),
    ('\u{0A38}', '\u{0A39}'),
    ('\u{0A59}', '\u{0A5C}'),
    ('\u{0A5E}', '\u{0A5E}'),
    ('\u{0A72}', '\u{0A74}'),
    ('\u{0A85}', '\u{0A8D}'),
    ('\u{0A8F}', '\u{0A91}'),
    ('\u{0A93}', '\u{0AA8}'),
    ('\u{0AAA}', '\u{0AB0}'),
    ('\u{0AB2}', '\u{0AB3}'),
    ('\u{0AB5}', '\u{0AB9}'),
    ('\u{0ABD}', '\u{0ABD}'),
    ('\u{0AD0}', '\u{0AD0}'),
    ('\u{0AE0}', '\u{0AE1}'),
    ('\u{0AF9}', '\u{0AF9}'),
    ('\u{0B05}', '\u{0B0C}'),
    ('\u{0B0F}', '\u{0B10}'),
    ('\u{0B13}', '\u{0B28}'),
    ('\u{0B2A}', '\u{0B30}'),
    ('\u{0B32}', '\u{0B33}'),
    ('\u{0B35}', '\u{0B39}'),
    ('\u{0B3D}', '\u{0B3D}'),
    ('\u{0B5C}', '\u{0B5D}'),
    ('\u{0B5F}', '\u{0B61}'),
    ('\u{0B71}', '\u{0B71}'),
    ('\u{0B83}', '\u{0B83}'),
    ('\u{0B85}', '\u{0B8A}'),
    ('\u{0B8E}', '\u{0B90}'),
    ('\u{0B92}', '\u{0B95}'),
    ('\u{0B99}', '\u{0B9A}'),
    ('\u{0B9C}', '\u{0B9C}'),
    ('\u{0B9E}', '\u{0B9F}'),
    ('\u{0BA3}', '\u{0BA4}'),
    ('\u{0BA8}', '\u{0BAA}'),
    ('\u{0BAE}', '\u{0BB9}'),
    ('\u{0BD0}', '\u{0BD0}'),
    ('\u{0C05}', '\u{0C0C}'),
    ('\u{0C0E}', '\u{0C10}'),
    ('\u{0C12}', '\u{0C28}'),
    ('\u{0C2A}', '\u{0C39}'),
    ('\u{0C3D}', '\u{0C3D}'),
    ('\u{0C58}', '\u{0C5A}'),
    ('\u{0C5D}', '\u{0C5D}'),
    ('\u{0C60}', '\u{0C61}'),
    ('\u{0C80}', '\u{0C80}'),
    ('\u{0C85}', '\u{0C8C}'),
    ('\u{0C8E}', '\u{0C90}'),
    ('\u{0C92}', '\u{0CA8}'),
    ('\u{0CAA}', '\u{0CB3}'),
    ('\u{0CB5}', '\u{0CB9}'),
    ('\u{0CBD}', '\u{0CBD}'),
    ('\u{0CDD}', '\u{0CDE}'),
    ('\u{0CE0}', '\u{0CE1}'),
    ('\u{0CF1}', '\u{0CF2}'),
    ('\u{0D04}', '\u{0D0C}'),
    ('\u{0D0E}', '\u{0D10}'),
    ('\u{0D12}', '\u{0D3A}'),
    ('\u{0D3D}', '\u{0D3D}'),
    ('\u{0D4E}', '\u{0D4E}'),
    ('\u{0D54}', '\u{0D56}'),
    ('\u{0D5F}', '\u{0D61}'),
    ('\u{0D7A}', '\u{0D7F}'),
    ('\u{0D85}', '\u{0D96}'),
    ('\u{0D9A}', '\u{0DB1}'),
    ('\u{0DB3}', '\u{0DBB}'),
    ('\u{0DBD}', '\u{0DBD}'),
    ('\u{0DC0}', '\u{0DC6}'),
    ('\u{0E01}', '\u{0E30}'),
    ('\u{0E32}', '\u{0E33}'),
    ('\u{0E40}', '\u{0E46}'),
    ('\u{0E81}', '\u{0E82}'),
    ('\u{0E84}', '\u{0E84}'),
    ('\u{0E86}', '\u{0E8A}'),
    ('\u{0E8C}', '\u{0EA3}'),
    ('\u{0EA5}', '\u{0EA5}'),
    ('\u{0EA7}', '\u{0EB0}'),
    ('\u{0EB2}', '\u{0EB3}'),
    ('\u{0EBD}', '\u{0EBD}'),
    ('\u{0EC0}', '\u{0EC4}'),
    ('\u{0EC6}', '\u{0EC6}'),
    ('\u{0EDC}', '\u{0EDF}'),
    ('\u{0F00}', '\u{0F00}'),
    ('\u{0F40}', '\u{0F47}'),
    ('\u{0F49}', '\u{0F6C}'),
    ('\u{0F88}', '\u{0F8C}'),
    ('\u{1000}', '\u{102A}'),
    ('\u{103F}', '\u{103F}'),
    ('\u{1050}', '\u{1055}'),
    ('\u{105A}', '\u{105D}'),
    ('\u{1061}', '\u{1061}'),
    ('\u{1065}', '\u{1066}'),
    ('\u{106E}', '\u{1070}'),
    ('\u{1075}', '\u{1081}'),
    ('\u{108E}', '\u{108E}'),
    ('\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{16F1}', '\u{16F8}'),
    ('\u{1700}', '\u{1711}'),
    ('\u{171F}', '\u{1731}'),
    ('\u{1740}', '\u{1751}'),
    ('\u{1760}', '\u{176C}'),
    ('\u{176E}', '\u{1770}'),
    ('\u{1780}', '\u{17B3}'),
    ('\u{17D7}', '\u{17D7}'),
    ('\u{17DC}', '\u{17DC}'),
    ('\u{1820}', '\u{1878}'),
    ('\u{1880}', '\u{1884}'),
    ('\u{1887}', '\u{18A8}'),
    ('\u{18AA}', '\u{18AA}'),
    ('\u{18B0}', '\u{18F5}'),
    ('\u{1900}', '\u{191E}'),
    ('\u{1950}', '\u{196D}'),
    ('\u{1970}', '\u{1974}'),
    ('\u{1980}', '\u{19AB}'),
    ('\u{19B0}', '\u{19C9}'),
    ('\u{1A00}', '\u{1A16}'),
    ('\u{1A20}', '\u{1A54}'),
    ('\u{1AA7}', '\u{1AA7}'),
    ('\u{1B05}', '\u{1B33}'),
    ('\u{1B45}', '\u{1B4C}'),
    ('\u{1B83}', '\u{1BA0}'),
    ('\u{1BAE}', '\u{1BAF}'),
    ('\u{1BBA}', '\u{1BE5}'),
    ('\u{1C00}', '\u{1C23}'),
    ('\u{1C4D}', '\u{1C4F}'),
    ('\u{1C5A}', '\u{1C7D}'),
    ('\u{1C80}', '\u{1C88}'),
    ('\u{1C90}', '\u{1CBA}'),
    ('\u{1CBD}', '\u{1CBF}'),
    ('\u{1CE9}', '\u{1CEC}'),
    ('\u{1CEE}', '\u{1CF3}'),
    ('\u{1CF5}', '\u{1CF6}'),
    ('\u{1CFA}', '\u{1CFA}'),
    ('\u{1D00}', '\u{1DBF}'),
    ('\u{1E00}', '\u{1F15}'),
    ('\u{1F18}', '\u{1F1D}'),
    ('\u{1F20}', '\u{1F45}'),
    ('\u{1F48}', '\u{1F4D}'),
    ('\u{1F50}', '\u{1F57}'),
    ('\u{1F59}', '\u{1F59}'),
    ('\u{1F5B}', '\u{1F5B}'),
    ('\u{1F5D}', '\u{1F5D}'),
    ('\u{1F5F}', '\u{1F7D}'),
    ('\u{1F80}', '\u{1FB4}'),
    ('\u{1FB6}', '\u{1FBC}'),
    ('\u{1FBE}', '\u{1FBE}'),
    ('\u{1FC2}', '\u{1FC4}'),
    ('\u{1FC6}', '\u{1FCC}'),
    ('\u{1FD0}', '\u{1FD3}'),
    ('\u{1FD6}', '\u{1FDB}'),
    ('\u{1FE0}', '\u{1FEC}'),
    ('\u{1FF2}', '\u{1FF4}'),
    ('\u{1FF6}', '\u{1FFC}'),
    ('\u{2071}', '\u{2071}'),
    ('\u{207F}', '\u{207F}'),
    ('\u{2090}', '\u{209C}'),
    ('\u{2102}', '\u{2102}'),
    ('\u{2107}', '\u{2107}'),
    ('\u{210A}', '\u{2113}'),
    ('\u{2115}', '\u{2115}'),
    ('\u{2119}', '\u{211D}'),
    ('\u{2124}', '\u{2124}'),
    ('\u{2126}', '\u{2126}'),
    ('\u{2128}', '\u{2128}'),
    ('\u{212A}', '\u{212D}'),
    ('\u{212F}', '\u{2139}'),
    ('\u{213C}', '\u{213F}'),
    ('\u{2145}', '\u{2149}'),
    ('\u{214E}', '\u{214E}'),
    ('\u{2183}', '\u{2184}'),
    ('\u{2C00}', '\u{2CE4}'),
    ('\u{2CEB}', '\u{2CEE}'),
    ('\u{2CF2}', '\u{2CF3}'),
    ('\u{2D00}', '\u{2D25}'),
    ('\u{2D27}', '\u{2D27}'),
    ('\u{2D2D}', '\u{2D2D}'),
    ('\u{2D30}', '\u{2D67}'),
    ('\u{2D6F}', '\u{2D6F}'),
    ('\u{2D80}', '\u{2D96}'),
    ('\u{2DA0}', '\u{2DA6}'),
    ('\u{2DA8}', '\u{2DAE}'),
    ('\u{2DB0}', '\u{2DB6}'),
    ('\u{2DB8}', '\u{2DBE}'),
    ('\u{2DC0}', '\u{2DC6}'),
    ('\u{2DC8}', '\u{2DCE}'),
    ('\u{2DD0}', '\u{2DD6}'),
    ('\u{2DD8}', '\u{2DDE}'),
    ('\u{2E2F}', '\u{2E2F}'),
    ('\u{3005}', '\u{3006}'),
    ('\u{3031}', '\u{3035}'),
    ('\u{303B}', '\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{31BF}'),
    ('\u{31F0}', '\u{31FF}'),
    ('\u{3400}', '\u{4DBF}'),
    ('\u{4E00}', '\u{A48C}'),
    ('\u{A4D0}', '\u{A4FD}'),
    ('\u{A500}', '\u{A60C}'),
    ('\u{A610}', '\u{A61F}'),
    ('\u{A62A}', '\u{A62B}'),
    ('\u{A640}', '\u{A66E}'),
    ('\u{A67F}', '\u{A69D}'),
    ('\u{A6A0}', '\u{A6E5}'),
    ('\u{A717}', '\u{A71F}'),
    ('\u{A722}', '\u{A788}'),
    ('\u{A78B}', '\u{A7CA}'),
    ('\u{A7D0}', '\u{A7D1}'),
    ('\u{A7D3}', '\u{A7D3}'),
    ('\u{A7D5}', '\u{A7D9}'),
    ('\u{A7F2}', '\u{A801}'),
    ('\u{A803}', '\u{A805}'),
    ('\u{A807}', '\u{A80A}'),
    ('\u{A80C}', '\u{A822}'),
    ('\u{A840}', '\u{A873}'),
    ('\u{A882}', '\u{A8B3}'),
    ('\u{A8F2}', '\u{A8F7}'),
    ('\u{A8FB}', '\u{A8FB}'),
    ('\u{A8FD}', '\u{A8FE}'),
    ('\u{A90A}', '\u{A925}'),
    ('\u{A930}', '\u{A946}'),
    ('\u{A960}', '\u{A97C}'),
    ('\u{A984}', '\u{A9B2}'),
    ('\u{A9CF}', '\u{A9CF}'),
    ('\u{A9E0}', '\u{A9E4}'),
    ('\u{A9E6}', '\u{A9EF}'),
    ('\u{A9FA}', '\u{A9FE}'),
    ('\u{AA00}', '\u{AA28}'),
    ('\u{AA40}', '\u{AA42}'),
    ('\u{AA44}', '\u{AA4B}'),
    ('\u{AA60}', '\u{AA76}'),
    ('\u{AA7A}', '\u{AA7A}'),
    ('\u{AA7E}', '\u{AAAF}'),
    ('\u{AAB1}', '\u{AAB1}'),
    ('\u{AAB5}', '\u{AAB6}'),
    ('\u{AAB9}', '\u{AABD}'),
    ('\u{AAC0}', '\u{AAC0}'),
    ('\u{AAC2}', '\u{AAC2}'),
    ('\u{AADB}', '\u{AADD}'),
    ('\u{AAE0}', '\u{AAEA}'),
    ('\u{AAF2}', '\u{AAF4}'),
    ('\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{AB69}'),
    ('\u{AB70}', '\u{ABE2}'),
    ('\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{FB1D}'),
    ('\u{FB1F}', '\u{FB28}'),
    ('\u{FB2A}', '\u{FB36}'),
    ('\u{FB38}', '\u{FB3C}'),
    ('\u{FB3E}', '\u{FB3E}'),
    ('\u{FB40}', '\u{FB41}'),
    ('\u{FB43}', '\u{FB44}'),
    ('\u{FB46}', '\u{FBB1}'),
    ('\u{FBD3}', '\u{FD3D}'),
    ('\u{FD50}', '\u{FD8F}'),
    ('\u{FD92}', '\u{FDC7}'),
    ('\u{FDF0}', '\u{FDFB}'),
    ('\u{FE70}', '\u{FE74}'),
    ('\u{FE76}', '\u{FEFC}'),
    ('\u{FF21}', '\u{FF3A}'),
    ('\u{FF41}', '\u{FF5A}'),
    ('\u{FF66}', '\u{FFBE}'),
    ('\u{FFC2}', '\u{FFC7}'),
    ('\u{FFCA}', '\u{FFCF}'),
    ('\u{FFD2}', '\u{FFD7}'),
    ('\u{FFDA}', '\u{FFDC}'),
    ('\u{10000}', '\u{1000B}'),
    ('\u{1000D}', '\u{10026}'),
    ('\u{10028}', '\u{1003A}'),
    ('\u{1003C}', '\u{1003D}'),
    ('\u{1003F}', '\u{1004D}'),
    ('\u{10050}', '\u{1005D}'),
    ('\u{10080}', '\u{100FA}'),
    ('\u{10280}', '\u{1029C}'),
    ('\u{102A0}', '\u{102D0}'),
    ('\u{10300}', '\u{1031F}'),
    ('\u{1032D}', '\u{10340}'),
    ('\u{10342}', '\u{10349}'),
    ('\u{10350}', '\u{10375}'),
    ('\u{10380}', '\u{1039D}'),
    ('\u{103A0}', '\u{103C3}'),
    ('\u{103C8}', '\u{103CF}'),
    ('\u{10400}', '\u{1049D}'),
    ('\u{104B0}', '\u{104D3}'),
    ('\u{104D8}', '\u{104FB}'),
    ('\u{10500}', '\u{10527}'),
    ('\u{10530}', '\u{10563}'),
    ('\u{10570}', '\u{1057A}'),
    ('\u{1057C}', '\u{1058A}'),
    ('\u{1058C}', '\u{10592}'),
    ('\u{10594}', '\u{10595}'),
    ('\u{10597}', '\u{105A1}'),
    ('\u{105A3}', '\u{105B1}'),
    ('\u{105B3}', '\u{105B9}'),
    ('\u{105BB}', '\u{105BC}'),
    ('\u{10600}', '\u{10736}'),
    ('\u{10740}', '\u{10755}'),
    ('\u{10760}', '\u{10767}'),
    ('\u{10780}', '\u{10785}'),
    ('\u{10787}', '\u{107B0}'),
    ('\u{107B2}', '\u{107BA}'),
    ('\u{10800}', '\u{10805}'),
    ('\u{10808}', '\u{10808}'),
    ('\u{1080A}', '\u{10835}'),
    ('\u{10837}', '\u{10838}'),
    ('\u{1083C}', '\u{1083C}'),
    ('\u{1083F}', '\u{10855}'),
    ('\u{10860}', '\u{10876}'),
    ('\u{10880}', '\u{1089E}'),
    ('\u{108E0}', '\u{108F2}'),
    ('\u{108F4}', '\u{108F5}'),
    ('\u{10900}', '\u{10915}'),
    ('\u{10920}', '\u{10939}'),
    ('\u{10980}', '\u{109B7}'),
    ('\u{109BE}', '\u{109BF}'),
    ('\u{10A00}', '\u{10A00}'),
    ('\u{10A10}', '\u{10A13}'),
    ('\u{10A15}', '\u{10A17}'),
    ('\u{10A19}', '\u{10A35}'),
    ('\u{10A60}', '\u{10A7C}'),
    ('\u{10A80}', '\u{10A9C}'),
    ('\u{10AC0}', '\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{10D23}'),
    ('\u{10E80}', '\u{10EA9}'),
    ('\u{10EB0}', '\u{10EB1}'),
    ('\u{10F00}', '\u{10F1C}'),
    ('\u{10F27}', '\u{10F27}'),
    ('\u{10F30}', '\u{10F45}'),
    ('\u{10F70}', '\u{10F81}'),
    ('\u{10FB0}', '\u{10FC4}'),
    ('\u{10FE0}', '\u{10FF6}'),
    ('\u{11003}', '\u{11037}'),
    ('\u{11071}', '\u{11072}'),
    ('\u{11075}', '\u{11075}'),
    ('\u{11083}', '\u{110AF}'),
    ('\u{110D0}', '\u{110E8}'),
    ('\u{11103}', '\u{11126}'),
    ('\u{11144}', '\u{11144}'),
    ('\u{11147}', '\u{11147}'),
    ('\u{11150}', '\u{11172}'),
    ('\u{11176}', '\u{11176}'),
    ('\u{11183}', '\u{111B2}'),
    ('\u{111C1}', '\u{111C4}'),
    ('\u{111DA}', '\u{111DA}'),
    ('\u{111DC}', '\u{111DC}'),
    ('\u{11200}', '\u{11211}'),
    ('\u{11213}', '\u{1122B}'),
    ('\u{1123F}', '\u{11240}'),
    ('\u{11280}', '\u{11286}'),
    ('\u{11288}', '\u{11288}'),
    ('\u{1128A}', '\u{1128D}'),
    ('\u{1128F}', '\u{1129D}'),
    ('\u{1129F}', '\u{112A8}'),
    ('\u{112B0}', '\u{112DE}'),
    ('\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{1133D}'),
    ('\u{11350}', '\u{11350}'),
    ('\u{1135D}', '\u{11361}'),
    ('\u{11400}', '\u{11434}'),
    ('\u{11447}', '\u{1144A}'),
    ('\u{1145F}', '\u{11461}'),
    ('\u{11480}', '\u{114AF}'),
    ('\u{114C4}', '\u{114C5}'),
    ('\u{114C7}', '\u{114C7}'),
    ('\u{11580}', '\u{115AE}'),
    ('\u{115D8}', '\u{115DB}'),
    ('\u{11600}', '\u{1162F}'),
    ('\u{11644}', '\u{11644}'),
    ('\u{11680}', '\u{116AA}'),
    ('\u{116B8}', '\u{116B8}'),
    ('\u{11700}', '\u{1171A}'),
    ('\u{11740}', '\u{11746}'),
    ('\u{11800}', '\u{1182B}'),
    ('\u{118A0}', '\u{118DF}'),
    ('\u{118FF}', '\u{11906}'),
    ('\u{11909}', '\u{11909}'),
    ('\u{1190C}', '\u{11913}'),
    ('\u{11915}', '\u{11916}'),
    ('\u{11918}', '\u{1192F}'),
    ('\u{1193F}', '\u{1193F}'),
    ('\u{11941}', '\u{11941}'),
    ('\u{119A0}', '\u{119A7}'),
    ('\u{119AA}', '\u{119D0}'),
    ('\u{119E1}', '\u{119E1}'),
    ('\u{119E3}', '\u{119E3}'),
    ('\u{11A00}', '\u{11A00}'),
    ('\u{11A0B}', '\u{11A32}'),
    ('\u{11A3A}', '\u{11A3A}'),
    ('\u{11A50}', '\u{11A50}'),
    ('\u{11A5C}', '\u{11A89}'),
    ('\u{11A9D}', '\u{11A9D}'),
    ('\u{11AB0}', '\u{11AF8}'),
    ('\u{11C00}', '\u{11C08}'),
    ('\u{11C0A}', '\u{11C2E}'),
    ('\u{11C40}', '\u{11C40}'),
    ('\u{11C72}', '\u{11C8F}'),
    ('\u{11D00}', '\u{11D06}'),
    ('\u{11D08}', '\u{11D09}'),
    ('\u{11D0B}', '\u{11D30}'),
    ('\u{11D46}', '\u{11D46}'),
    ('\u{11D60}', '\u{11D65}'),
    ('\u{11D67}', '\u{11D68}'),
    ('\u{11D6A}', '\u{11D89}'),
    ('\u{11D98}', '\u{11D98}'),
    ('\u{11EE0}', '\u{11EF2}'),
    ('\u{11F02}', '\u{11F02}'),
    ('\u{11F04}', '\u{11F10}'),
    ('\u{11F12}', '\u{11F33}'),
    ('\u{11FB0}', '\u{11FB0}'),
    ('\u{12000}', '\u{12399}'),
    ('\u{12480}', '\u{12543}'),
    ('\u{12F90}', '\u{12FF0}'),
    ('\u{13000}', '\u{1342F}'),
    ('\u{13441}', '\u{13446}'),
    ('\u{14400}', '\u{14646}'),
    ('\u{16800}', '\u{16A38}'),
    ('\u{16A40}', '\u{16A5E}'),
    ('\u{16A70}', '\u{16ABE}'),
    ('\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{16F50}', '\u{16F50}'),
    ('\u{16F93}', '\u{16F9F}'),
    ('\u{16FE0}', '\u{16FE1}'),
    ('\u{16FE3}', '\u{16FE3}'),
    ('\u{17000}', '\u{187F7}'),
    ('\u{18800}', '\u{18CD5}'),
    ('\u{18D00}', '\u{18D08}'),
    ('\u{1AFF0}', '\u{1AFF3}'),
    ('\u{1AFF5}', '\u{1AFFB}'),
    ('\u{1AFFD}', '\u{1AFFE}'),
    ('\u{1B000}', '\u{1B122}'),
    ('\u{1B132}', '\u{1B132}'),
    ('\u{1B150}', '\u{1B152}'),
    ('\u{1B155}', '\u{1B155}'),
    ('\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{1D400}', '\u{1D454}'),
    ('\u{1D456}', '\u{1D49C}'),
    ('\u{1D49E}', '\u{1D49F}'),
    ('\u{1D4A2}', '\u{1D4A2}'),
    ('\u{1D4A5}', '\u{1D4A6}'),
    ('\u{1D4A9}', '\u{1D4AC}'),
    ('\u{1D4AE}', '\u{1D4B9}'),
    ('\u{1D4BB}', '\u{1D4BB}'),
    ('\u{1D4BD}', '\u{1D4C3}'),
    ('\u{1D4C5}', '\u{1D505}'),
    ('\u{1D507}', '\u{1D50A}'),
    ('\u{1D50D}', '\u{1D514}'),
    ('\u{1D516}', '\u{1D51C}'),
    ('\u{1D51E}', '\u{1D539}'),
    ('\u{1D53B}', '\u{1D53E}'),
    ('\u{1D540}', '\u{1D544}'),
    ('\u{1D546}', '\u{1D546}'),
    ('\u{1D54A}', '\u{1D550}'),
    ('\u{1D552}', '\u{1D6A5}'),
    ('\u{1D6A8}', '\u{1D6C0}'),
    ('\u{1D6C2}', '\u{1D6DA}'),
    ('\u{1D6DC}', '\u{1D6FA}'),
    ('\u{1D6FC}', '\u{1D714}'),
    ('\u{1D716}', '\u{1D734}'),
    ('\u{1D736}', '\u{1D74E}'),
    ('\u{1D750}', '\u{1D76E}'),
    ('\u{1D770}', '\u{1D788}'),
    ('\u{1D78A}', '\u{1D7A8}'),
    ('\u{1D7AA}', '\u{1D7C2}'),
    ('\u{1D7C4}', '\u{1D7CB}'),
    ('\u{1DF00}', '\u{1DF1E}'),
    ('\u{1DF25}', '\u{1DF2A}'),
    ('\u{1E030}', '\u{1E06D}'),
    ('\u{1E100}', '\u{1E12C}'),
    ('\u{1E137}', '\u{1E13D}'),
    ('\u{1E14E}', '\u{1E14E}'),
    ('\u{1E290}', '\u{1E2AD}'),
    ('\u{1E2C0}', '\u{1E2EB}'),
    ('\u{1E4D0}', '\u{1E4EB}'),
    ('\u{1E7E0}', '\u{1E7E6}'),
    ('\u{1E7E8}', '\u{1E7EB}'),
    ('\u{1E7ED}', '\u{1E7EE}'),
    ('\u{1E7F0}', '\u{1E7FE}'),
    ('\u{1E800}', '\u{1E8C4}'),
    ('\u{1E900}', '\u{1E943}'),
    ('\u{1E94B}', '\u{1E94B}'),
    ('\u{1EE00}', '\u{1EE03}'),
    ('\u{1EE05}', '\u{1EE1F}'),
    ('\u{1EE21}', '\u{1EE22}'),
    ('\u{1EE24}', '\u{1EE24}'),
    ('\u{1EE27}', '\u{1EE27}'),
    ('\u{1EE29}', '\u{1EE32}'),
    ('\u{1EE34}', '\u{1EE37}'),
    ('\u{1EE39}', '\u{1EE39}'),
    ('\u{1EE3B}', '\u{1EE3B}'),
    ('\u{1EE42}', '\u{1EE42}'),
    ('\u{1EE47}', '\u{1EE47}'),
    ('\u{1EE49}', '\u{1EE49}'),
    ('\u{1EE4B}', '\u{1EE4B}'),
    ('\u{1EE4D}', '\u{1EE4F}'),
    ('\u{1EE51}', '\u{1EE52}'),
    ('\u{1EE54}', '\u{1EE54}'),
    ('\u{1EE57}', '\u{1EE57}'),
    ('\u{1EE59}', '\u{1EE59}'),
    ('\u{1EE5B}', '\u{1EE5B}'),
    ('\u{1EE5D}', '\u{1EE5D}'),
    ('\u{1EE5F}', '\u{1EE5F}'),
    ('\u{1EE61}', '\u{1EE62}'),
    ('\u{1EE64}', '\u{1EE64}'),
    ('\u{1EE67}', '\u{1EE6A}'),
    ('\u{1EE6C}', '\u{1EE72}'),
    ('\u{1EE74}', '\u{1EE77}'),
    ('\u{1EE79}', '\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{20000}', '\u{2A6DF}'),
    ('\u{2A700}', '\u{2B739}'),
    ('\u{2B740}', '\u{2B81D}'),
    ('\u{2B820}', '\u{2CEA1}'),
    ('\u{2CEB0}', '\u{2EBE0}'),
    ('\u{2EBF0}', '\u{2EE5D}'),
    ('\u{2F800}', '\u{2FA1D}'),
    ('\u{30000}', '\u{3134A}'),
    ('\u{31350}', '\u{323AF}'),
];

/// Inclusive codepoint ranges of characters with the Unicode `Uppercase`
/// property (general category `Lu` plus `Other_Uppercase`, e.g. roman
/// numerals `Ⅻ` and circled letters `Ⓐ`). Mirrors CPython's
/// `Py_UNICODE_ISUPPER`. Titlecase (`Lt`) characters are excluded (they
/// live in [`TITLE_RANGES`]).
///
/// Derived live from the oracle CPython 3.13 / Unicode 15.1.0,
/// run-length compressed to 651 ranges.
const UPPER_RANGES: &[(char, char)] = &[
    ('\u{0041}', '\u{005A}'),
    ('\u{00C0}', '\u{00D6}'),
    ('\u{00D8}', '\u{00DE}'),
    ('\u{0100}', '\u{0100}'),
    ('\u{0102}', '\u{0102}'),
    ('\u{0104}', '\u{0104}'),
    ('\u{0106}', '\u{0106}'),
    ('\u{0108}', '\u{0108}'),
    ('\u{010A}', '\u{010A}'),
    ('\u{010C}', '\u{010C}'),
    ('\u{010E}', '\u{010E}'),
    ('\u{0110}', '\u{0110}'),
    ('\u{0112}', '\u{0112}'),
    ('\u{0114}', '\u{0114}'),
    ('\u{0116}', '\u{0116}'),
    ('\u{0118}', '\u{0118}'),
    ('\u{011A}', '\u{011A}'),
    ('\u{011C}', '\u{011C}'),
    ('\u{011E}', '\u{011E}'),
    ('\u{0120}', '\u{0120}'),
    ('\u{0122}', '\u{0122}'),
    ('\u{0124}', '\u{0124}'),
    ('\u{0126}', '\u{0126}'),
    ('\u{0128}', '\u{0128}'),
    ('\u{012A}', '\u{012A}'),
    ('\u{012C}', '\u{012C}'),
    ('\u{012E}', '\u{012E}'),
    ('\u{0130}', '\u{0130}'),
    ('\u{0132}', '\u{0132}'),
    ('\u{0134}', '\u{0134}'),
    ('\u{0136}', '\u{0136}'),
    ('\u{0139}', '\u{0139}'),
    ('\u{013B}', '\u{013B}'),
    ('\u{013D}', '\u{013D}'),
    ('\u{013F}', '\u{013F}'),
    ('\u{0141}', '\u{0141}'),
    ('\u{0143}', '\u{0143}'),
    ('\u{0145}', '\u{0145}'),
    ('\u{0147}', '\u{0147}'),
    ('\u{014A}', '\u{014A}'),
    ('\u{014C}', '\u{014C}'),
    ('\u{014E}', '\u{014E}'),
    ('\u{0150}', '\u{0150}'),
    ('\u{0152}', '\u{0152}'),
    ('\u{0154}', '\u{0154}'),
    ('\u{0156}', '\u{0156}'),
    ('\u{0158}', '\u{0158}'),
    ('\u{015A}', '\u{015A}'),
    ('\u{015C}', '\u{015C}'),
    ('\u{015E}', '\u{015E}'),
    ('\u{0160}', '\u{0160}'),
    ('\u{0162}', '\u{0162}'),
    ('\u{0164}', '\u{0164}'),
    ('\u{0166}', '\u{0166}'),
    ('\u{0168}', '\u{0168}'),
    ('\u{016A}', '\u{016A}'),
    ('\u{016C}', '\u{016C}'),
    ('\u{016E}', '\u{016E}'),
    ('\u{0170}', '\u{0170}'),
    ('\u{0172}', '\u{0172}'),
    ('\u{0174}', '\u{0174}'),
    ('\u{0176}', '\u{0176}'),
    ('\u{0178}', '\u{0179}'),
    ('\u{017B}', '\u{017B}'),
    ('\u{017D}', '\u{017D}'),
    ('\u{0181}', '\u{0182}'),
    ('\u{0184}', '\u{0184}'),
    ('\u{0186}', '\u{0187}'),
    ('\u{0189}', '\u{018B}'),
    ('\u{018E}', '\u{0191}'),
    ('\u{0193}', '\u{0194}'),
    ('\u{0196}', '\u{0198}'),
    ('\u{019C}', '\u{019D}'),
    ('\u{019F}', '\u{01A0}'),
    ('\u{01A2}', '\u{01A2}'),
    ('\u{01A4}', '\u{01A4}'),
    ('\u{01A6}', '\u{01A7}'),
    ('\u{01A9}', '\u{01A9}'),
    ('\u{01AC}', '\u{01AC}'),
    ('\u{01AE}', '\u{01AF}'),
    ('\u{01B1}', '\u{01B3}'),
    ('\u{01B5}', '\u{01B5}'),
    ('\u{01B7}', '\u{01B8}'),
    ('\u{01BC}', '\u{01BC}'),
    ('\u{01C4}', '\u{01C4}'),
    ('\u{01C7}', '\u{01C7}'),
    ('\u{01CA}', '\u{01CA}'),
    ('\u{01CD}', '\u{01CD}'),
    ('\u{01CF}', '\u{01CF}'),
    ('\u{01D1}', '\u{01D1}'),
    ('\u{01D3}', '\u{01D3}'),
    ('\u{01D5}', '\u{01D5}'),
    ('\u{01D7}', '\u{01D7}'),
    ('\u{01D9}', '\u{01D9}'),
    ('\u{01DB}', '\u{01DB}'),
    ('\u{01DE}', '\u{01DE}'),
    ('\u{01E0}', '\u{01E0}'),
    ('\u{01E2}', '\u{01E2}'),
    ('\u{01E4}', '\u{01E4}'),
    ('\u{01E6}', '\u{01E6}'),
    ('\u{01E8}', '\u{01E8}'),
    ('\u{01EA}', '\u{01EA}'),
    ('\u{01EC}', '\u{01EC}'),
    ('\u{01EE}', '\u{01EE}'),
    ('\u{01F1}', '\u{01F1}'),
    ('\u{01F4}', '\u{01F4}'),
    ('\u{01F6}', '\u{01F8}'),
    ('\u{01FA}', '\u{01FA}'),
    ('\u{01FC}', '\u{01FC}'),
    ('\u{01FE}', '\u{01FE}'),
    ('\u{0200}', '\u{0200}'),
    ('\u{0202}', '\u{0202}'),
    ('\u{0204}', '\u{0204}'),
    ('\u{0206}', '\u{0206}'),
    ('\u{0208}', '\u{0208}'),
    ('\u{020A}', '\u{020A}'),
    ('\u{020C}', '\u{020C}'),
    ('\u{020E}', '\u{020E}'),
    ('\u{0210}', '\u{0210}'),
    ('\u{0212}', '\u{0212}'),
    ('\u{0214}', '\u{0214}'),
    ('\u{0216}', '\u{0216}'),
    ('\u{0218}', '\u{0218}'),
    ('\u{021A}', '\u{021A}'),
    ('\u{021C}', '\u{021C}'),
    ('\u{021E}', '\u{021E}'),
    ('\u{0220}', '\u{0220}'),
    ('\u{0222}', '\u{0222}'),
    ('\u{0224}', '\u{0224}'),
    ('\u{0226}', '\u{0226}'),
    ('\u{0228}', '\u{0228}'),
    ('\u{022A}', '\u{022A}'),
    ('\u{022C}', '\u{022C}'),
    ('\u{022E}', '\u{022E}'),
    ('\u{0230}', '\u{0230}'),
    ('\u{0232}', '\u{0232}'),
    ('\u{023A}', '\u{023B}'),
    ('\u{023D}', '\u{023E}'),
    ('\u{0241}', '\u{0241}'),
    ('\u{0243}', '\u{0246}'),
    ('\u{0248}', '\u{0248}'),
    ('\u{024A}', '\u{024A}'),
    ('\u{024C}', '\u{024C}'),
    ('\u{024E}', '\u{024E}'),
    ('\u{0370}', '\u{0370}'),
    ('\u{0372}', '\u{0372}'),
    ('\u{0376}', '\u{0376}'),
    ('\u{037F}', '\u{037F}'),
    ('\u{0386}', '\u{0386}'),
    ('\u{0388}', '\u{038A}'),
    ('\u{038C}', '\u{038C}'),
    ('\u{038E}', '\u{038F}'),
    ('\u{0391}', '\u{03A1}'),
    ('\u{03A3}', '\u{03AB}'),
    ('\u{03CF}', '\u{03CF}'),
    ('\u{03D2}', '\u{03D4}'),
    ('\u{03D8}', '\u{03D8}'),
    ('\u{03DA}', '\u{03DA}'),
    ('\u{03DC}', '\u{03DC}'),
    ('\u{03DE}', '\u{03DE}'),
    ('\u{03E0}', '\u{03E0}'),
    ('\u{03E2}', '\u{03E2}'),
    ('\u{03E4}', '\u{03E4}'),
    ('\u{03E6}', '\u{03E6}'),
    ('\u{03E8}', '\u{03E8}'),
    ('\u{03EA}', '\u{03EA}'),
    ('\u{03EC}', '\u{03EC}'),
    ('\u{03EE}', '\u{03EE}'),
    ('\u{03F4}', '\u{03F4}'),
    ('\u{03F7}', '\u{03F7}'),
    ('\u{03F9}', '\u{03FA}'),
    ('\u{03FD}', '\u{042F}'),
    ('\u{0460}', '\u{0460}'),
    ('\u{0462}', '\u{0462}'),
    ('\u{0464}', '\u{0464}'),
    ('\u{0466}', '\u{0466}'),
    ('\u{0468}', '\u{0468}'),
    ('\u{046A}', '\u{046A}'),
    ('\u{046C}', '\u{046C}'),
    ('\u{046E}', '\u{046E}'),
    ('\u{0470}', '\u{0470}'),
    ('\u{0472}', '\u{0472}'),
    ('\u{0474}', '\u{0474}'),
    ('\u{0476}', '\u{0476}'),
    ('\u{0478}', '\u{0478}'),
    ('\u{047A}', '\u{047A}'),
    ('\u{047C}', '\u{047C}'),
    ('\u{047E}', '\u{047E}'),
    ('\u{0480}', '\u{0480}'),
    ('\u{048A}', '\u{048A}'),
    ('\u{048C}', '\u{048C}'),
    ('\u{048E}', '\u{048E}'),
    ('\u{0490}', '\u{0490}'),
    ('\u{0492}', '\u{0492}'),
    ('\u{0494}', '\u{0494}'),
    ('\u{0496}', '\u{0496}'),
    ('\u{0498}', '\u{0498}'),
    ('\u{049A}', '\u{049A}'),
    ('\u{049C}', '\u{049C}'),
    ('\u{049E}', '\u{049E}'),
    ('\u{04A0}', '\u{04A0}'),
    ('\u{04A2}', '\u{04A2}'),
    ('\u{04A4}', '\u{04A4}'),
    ('\u{04A6}', '\u{04A6}'),
    ('\u{04A8}', '\u{04A8}'),
    ('\u{04AA}', '\u{04AA}'),
    ('\u{04AC}', '\u{04AC}'),
    ('\u{04AE}', '\u{04AE}'),
    ('\u{04B0}', '\u{04B0}'),
    ('\u{04B2}', '\u{04B2}'),
    ('\u{04B4}', '\u{04B4}'),
    ('\u{04B6}', '\u{04B6}'),
    ('\u{04B8}', '\u{04B8}'),
    ('\u{04BA}', '\u{04BA}'),
    ('\u{04BC}', '\u{04BC}'),
    ('\u{04BE}', '\u{04BE}'),
    ('\u{04C0}', '\u{04C1}'),
    ('\u{04C3}', '\u{04C3}'),
    ('\u{04C5}', '\u{04C5}'),
    ('\u{04C7}', '\u{04C7}'),
    ('\u{04C9}', '\u{04C9}'),
    ('\u{04CB}', '\u{04CB}'),
    ('\u{04CD}', '\u{04CD}'),
    ('\u{04D0}', '\u{04D0}'),
    ('\u{04D2}', '\u{04D2}'),
    ('\u{04D4}', '\u{04D4}'),
    ('\u{04D6}', '\u{04D6}'),
    ('\u{04D8}', '\u{04D8}'),
    ('\u{04DA}', '\u{04DA}'),
    ('\u{04DC}', '\u{04DC}'),
    ('\u{04DE}', '\u{04DE}'),
    ('\u{04E0}', '\u{04E0}'),
    ('\u{04E2}', '\u{04E2}'),
    ('\u{04E4}', '\u{04E4}'),
    ('\u{04E6}', '\u{04E6}'),
    ('\u{04E8}', '\u{04E8}'),
    ('\u{04EA}', '\u{04EA}'),
    ('\u{04EC}', '\u{04EC}'),
    ('\u{04EE}', '\u{04EE}'),
    ('\u{04F0}', '\u{04F0}'),
    ('\u{04F2}', '\u{04F2}'),
    ('\u{04F4}', '\u{04F4}'),
    ('\u{04F6}', '\u{04F6}'),
    ('\u{04F8}', '\u{04F8}'),
    ('\u{04FA}', '\u{04FA}'),
    ('\u{04FC}', '\u{04FC}'),
    ('\u{04FE}', '\u{04FE}'),
    ('\u{0500}', '\u{0500}'),
    ('\u{0502}', '\u{0502}'),
    ('\u{0504}', '\u{0504}'),
    ('\u{0506}', '\u{0506}'),
    ('\u{0508}', '\u{0508}'),
    ('\u{050A}', '\u{050A}'),
    ('\u{050C}', '\u{050C}'),
    ('\u{050E}', '\u{050E}'),
    ('\u{0510}', '\u{0510}'),
    ('\u{0512}', '\u{0512}'),
    ('\u{0514}', '\u{0514}'),
    ('\u{0516}', '\u{0516}'),
    ('\u{0518}', '\u{0518}'),
    ('\u{051A}', '\u{051A}'),
    ('\u{051C}', '\u{051C}'),
    ('\u{051E}', '\u{051E}'),
    ('\u{0520}', '\u{0520}'),
    ('\u{0522}', '\u{0522}'),
    ('\u{0524}', '\u{0524}'),
    ('\u{0526}', '\u{0526}'),
    ('\u{0528}', '\u{0528}'),
    ('\u{052A}', '\u{052A}'),
    ('\u{052C}', '\u{052C}'),
    ('\u{052E}', '\u{052E}'),
    ('\u{0531}', '\u{0556}'),
    ('\u{10A0}', '\u{10C5}'),
    ('\u{10C7}', '\u{10C7}'),
    ('\u{10CD}', '\u{10CD}'),
    ('\u{13A0}', '\u{13F5}'),
    ('\u{1C90}', '\u{1CBA}'),
    ('\u{1CBD}', '\u{1CBF}'),
    ('\u{1E00}', '\u{1E00}'),
    ('\u{1E02}', '\u{1E02}'),
    ('\u{1E04}', '\u{1E04}'),
    ('\u{1E06}', '\u{1E06}'),
    ('\u{1E08}', '\u{1E08}'),
    ('\u{1E0A}', '\u{1E0A}'),
    ('\u{1E0C}', '\u{1E0C}'),
    ('\u{1E0E}', '\u{1E0E}'),
    ('\u{1E10}', '\u{1E10}'),
    ('\u{1E12}', '\u{1E12}'),
    ('\u{1E14}', '\u{1E14}'),
    ('\u{1E16}', '\u{1E16}'),
    ('\u{1E18}', '\u{1E18}'),
    ('\u{1E1A}', '\u{1E1A}'),
    ('\u{1E1C}', '\u{1E1C}'),
    ('\u{1E1E}', '\u{1E1E}'),
    ('\u{1E20}', '\u{1E20}'),
    ('\u{1E22}', '\u{1E22}'),
    ('\u{1E24}', '\u{1E24}'),
    ('\u{1E26}', '\u{1E26}'),
    ('\u{1E28}', '\u{1E28}'),
    ('\u{1E2A}', '\u{1E2A}'),
    ('\u{1E2C}', '\u{1E2C}'),
    ('\u{1E2E}', '\u{1E2E}'),
    ('\u{1E30}', '\u{1E30}'),
    ('\u{1E32}', '\u{1E32}'),
    ('\u{1E34}', '\u{1E34}'),
    ('\u{1E36}', '\u{1E36}'),
    ('\u{1E38}', '\u{1E38}'),
    ('\u{1E3A}', '\u{1E3A}'),
    ('\u{1E3C}', '\u{1E3C}'),
    ('\u{1E3E}', '\u{1E3E}'),
    ('\u{1E40}', '\u{1E40}'),
    ('\u{1E42}', '\u{1E42}'),
    ('\u{1E44}', '\u{1E44}'),
    ('\u{1E46}', '\u{1E46}'),
    ('\u{1E48}', '\u{1E48}'),
    ('\u{1E4A}', '\u{1E4A}'),
    ('\u{1E4C}', '\u{1E4C}'),
    ('\u{1E4E}', '\u{1E4E}'),
    ('\u{1E50}', '\u{1E50}'),
    ('\u{1E52}', '\u{1E52}'),
    ('\u{1E54}', '\u{1E54}'),
    ('\u{1E56}', '\u{1E56}'),
    ('\u{1E58}', '\u{1E58}'),
    ('\u{1E5A}', '\u{1E5A}'),
    ('\u{1E5C}', '\u{1E5C}'),
    ('\u{1E5E}', '\u{1E5E}'),
    ('\u{1E60}', '\u{1E60}'),
    ('\u{1E62}', '\u{1E62}'),
    ('\u{1E64}', '\u{1E64}'),
    ('\u{1E66}', '\u{1E66}'),
    ('\u{1E68}', '\u{1E68}'),
    ('\u{1E6A}', '\u{1E6A}'),
    ('\u{1E6C}', '\u{1E6C}'),
    ('\u{1E6E}', '\u{1E6E}'),
    ('\u{1E70}', '\u{1E70}'),
    ('\u{1E72}', '\u{1E72}'),
    ('\u{1E74}', '\u{1E74}'),
    ('\u{1E76}', '\u{1E76}'),
    ('\u{1E78}', '\u{1E78}'),
    ('\u{1E7A}', '\u{1E7A}'),
    ('\u{1E7C}', '\u{1E7C}'),
    ('\u{1E7E}', '\u{1E7E}'),
    ('\u{1E80}', '\u{1E80}'),
    ('\u{1E82}', '\u{1E82}'),
    ('\u{1E84}', '\u{1E84}'),
    ('\u{1E86}', '\u{1E86}'),
    ('\u{1E88}', '\u{1E88}'),
    ('\u{1E8A}', '\u{1E8A}'),
    ('\u{1E8C}', '\u{1E8C}'),
    ('\u{1E8E}', '\u{1E8E}'),
    ('\u{1E90}', '\u{1E90}'),
    ('\u{1E92}', '\u{1E92}'),
    ('\u{1E94}', '\u{1E94}'),
    ('\u{1E9E}', '\u{1E9E}'),
    ('\u{1EA0}', '\u{1EA0}'),
    ('\u{1EA2}', '\u{1EA2}'),
    ('\u{1EA4}', '\u{1EA4}'),
    ('\u{1EA6}', '\u{1EA6}'),
    ('\u{1EA8}', '\u{1EA8}'),
    ('\u{1EAA}', '\u{1EAA}'),
    ('\u{1EAC}', '\u{1EAC}'),
    ('\u{1EAE}', '\u{1EAE}'),
    ('\u{1EB0}', '\u{1EB0}'),
    ('\u{1EB2}', '\u{1EB2}'),
    ('\u{1EB4}', '\u{1EB4}'),
    ('\u{1EB6}', '\u{1EB6}'),
    ('\u{1EB8}', '\u{1EB8}'),
    ('\u{1EBA}', '\u{1EBA}'),
    ('\u{1EBC}', '\u{1EBC}'),
    ('\u{1EBE}', '\u{1EBE}'),
    ('\u{1EC0}', '\u{1EC0}'),
    ('\u{1EC2}', '\u{1EC2}'),
    ('\u{1EC4}', '\u{1EC4}'),
    ('\u{1EC6}', '\u{1EC6}'),
    ('\u{1EC8}', '\u{1EC8}'),
    ('\u{1ECA}', '\u{1ECA}'),
    ('\u{1ECC}', '\u{1ECC}'),
    ('\u{1ECE}', '\u{1ECE}'),
    ('\u{1ED0}', '\u{1ED0}'),
    ('\u{1ED2}', '\u{1ED2}'),
    ('\u{1ED4}', '\u{1ED4}'),
    ('\u{1ED6}', '\u{1ED6}'),
    ('\u{1ED8}', '\u{1ED8}'),
    ('\u{1EDA}', '\u{1EDA}'),
    ('\u{1EDC}', '\u{1EDC}'),
    ('\u{1EDE}', '\u{1EDE}'),
    ('\u{1EE0}', '\u{1EE0}'),
    ('\u{1EE2}', '\u{1EE2}'),
    ('\u{1EE4}', '\u{1EE4}'),
    ('\u{1EE6}', '\u{1EE6}'),
    ('\u{1EE8}', '\u{1EE8}'),
    ('\u{1EEA}', '\u{1EEA}'),
    ('\u{1EEC}', '\u{1EEC}'),
    ('\u{1EEE}', '\u{1EEE}'),
    ('\u{1EF0}', '\u{1EF0}'),
    ('\u{1EF2}', '\u{1EF2}'),
    ('\u{1EF4}', '\u{1EF4}'),
    ('\u{1EF6}', '\u{1EF6}'),
    ('\u{1EF8}', '\u{1EF8}'),
    ('\u{1EFA}', '\u{1EFA}'),
    ('\u{1EFC}', '\u{1EFC}'),
    ('\u{1EFE}', '\u{1EFE}'),
    ('\u{1F08}', '\u{1F0F}'),
    ('\u{1F18}', '\u{1F1D}'),
    ('\u{1F28}', '\u{1F2F}'),
    ('\u{1F38}', '\u{1F3F}'),
    ('\u{1F48}', '\u{1F4D}'),
    ('\u{1F59}', '\u{1F59}'),
    ('\u{1F5B}', '\u{1F5B}'),
    ('\u{1F5D}', '\u{1F5D}'),
    ('\u{1F5F}', '\u{1F5F}'),
    ('\u{1F68}', '\u{1F6F}'),
    ('\u{1FB8}', '\u{1FBB}'),
    ('\u{1FC8}', '\u{1FCB}'),
    ('\u{1FD8}', '\u{1FDB}'),
    ('\u{1FE8}', '\u{1FEC}'),
    ('\u{1FF8}', '\u{1FFB}'),
    ('\u{2102}', '\u{2102}'),
    ('\u{2107}', '\u{2107}'),
    ('\u{210B}', '\u{210D}'),
    ('\u{2110}', '\u{2112}'),
    ('\u{2115}', '\u{2115}'),
    ('\u{2119}', '\u{211D}'),
    ('\u{2124}', '\u{2124}'),
    ('\u{2126}', '\u{2126}'),
    ('\u{2128}', '\u{2128}'),
    ('\u{212A}', '\u{212D}'),
    ('\u{2130}', '\u{2133}'),
    ('\u{213E}', '\u{213F}'),
    ('\u{2145}', '\u{2145}'),
    ('\u{2160}', '\u{216F}'),
    ('\u{2183}', '\u{2183}'),
    ('\u{24B6}', '\u{24CF}'),
    ('\u{2C00}', '\u{2C2F}'),
    ('\u{2C60}', '\u{2C60}'),
    ('\u{2C62}', '\u{2C64}'),
    ('\u{2C67}', '\u{2C67}'),
    ('\u{2C69}', '\u{2C69}'),
    ('\u{2C6B}', '\u{2C6B}'),
    ('\u{2C6D}', '\u{2C70}'),
    ('\u{2C72}', '\u{2C72}'),
    ('\u{2C75}', '\u{2C75}'),
    ('\u{2C7E}', '\u{2C80}'),
    ('\u{2C82}', '\u{2C82}'),
    ('\u{2C84}', '\u{2C84}'),
    ('\u{2C86}', '\u{2C86}'),
    ('\u{2C88}', '\u{2C88}'),
    ('\u{2C8A}', '\u{2C8A}'),
    ('\u{2C8C}', '\u{2C8C}'),
    ('\u{2C8E}', '\u{2C8E}'),
    ('\u{2C90}', '\u{2C90}'),
    ('\u{2C92}', '\u{2C92}'),
    ('\u{2C94}', '\u{2C94}'),
    ('\u{2C96}', '\u{2C96}'),
    ('\u{2C98}', '\u{2C98}'),
    ('\u{2C9A}', '\u{2C9A}'),
    ('\u{2C9C}', '\u{2C9C}'),
    ('\u{2C9E}', '\u{2C9E}'),
    ('\u{2CA0}', '\u{2CA0}'),
    ('\u{2CA2}', '\u{2CA2}'),
    ('\u{2CA4}', '\u{2CA4}'),
    ('\u{2CA6}', '\u{2CA6}'),
    ('\u{2CA8}', '\u{2CA8}'),
    ('\u{2CAA}', '\u{2CAA}'),
    ('\u{2CAC}', '\u{2CAC}'),
    ('\u{2CAE}', '\u{2CAE}'),
    ('\u{2CB0}', '\u{2CB0}'),
    ('\u{2CB2}', '\u{2CB2}'),
    ('\u{2CB4}', '\u{2CB4}'),
    ('\u{2CB6}', '\u{2CB6}'),
    ('\u{2CB8}', '\u{2CB8}'),
    ('\u{2CBA}', '\u{2CBA}'),
    ('\u{2CBC}', '\u{2CBC}'),
    ('\u{2CBE}', '\u{2CBE}'),
    ('\u{2CC0}', '\u{2CC0}'),
    ('\u{2CC2}', '\u{2CC2}'),
    ('\u{2CC4}', '\u{2CC4}'),
    ('\u{2CC6}', '\u{2CC6}'),
    ('\u{2CC8}', '\u{2CC8}'),
    ('\u{2CCA}', '\u{2CCA}'),
    ('\u{2CCC}', '\u{2CCC}'),
    ('\u{2CCE}', '\u{2CCE}'),
    ('\u{2CD0}', '\u{2CD0}'),
    ('\u{2CD2}', '\u{2CD2}'),
    ('\u{2CD4}', '\u{2CD4}'),
    ('\u{2CD6}', '\u{2CD6}'),
    ('\u{2CD8}', '\u{2CD8}'),
    ('\u{2CDA}', '\u{2CDA}'),
    ('\u{2CDC}', '\u{2CDC}'),
    ('\u{2CDE}', '\u{2CDE}'),
    ('\u{2CE0}', '\u{2CE0}'),
    ('\u{2CE2}', '\u{2CE2}'),
    ('\u{2CEB}', '\u{2CEB}'),
    ('\u{2CED}', '\u{2CED}'),
    ('\u{2CF2}', '\u{2CF2}'),
    ('\u{A640}', '\u{A640}'),
    ('\u{A642}', '\u{A642}'),
    ('\u{A644}', '\u{A644}'),
    ('\u{A646}', '\u{A646}'),
    ('\u{A648}', '\u{A648}'),
    ('\u{A64A}', '\u{A64A}'),
    ('\u{A64C}', '\u{A64C}'),
    ('\u{A64E}', '\u{A64E}'),
    ('\u{A650}', '\u{A650}'),
    ('\u{A652}', '\u{A652}'),
    ('\u{A654}', '\u{A654}'),
    ('\u{A656}', '\u{A656}'),
    ('\u{A658}', '\u{A658}'),
    ('\u{A65A}', '\u{A65A}'),
    ('\u{A65C}', '\u{A65C}'),
    ('\u{A65E}', '\u{A65E}'),
    ('\u{A660}', '\u{A660}'),
    ('\u{A662}', '\u{A662}'),
    ('\u{A664}', '\u{A664}'),
    ('\u{A666}', '\u{A666}'),
    ('\u{A668}', '\u{A668}'),
    ('\u{A66A}', '\u{A66A}'),
    ('\u{A66C}', '\u{A66C}'),
    ('\u{A680}', '\u{A680}'),
    ('\u{A682}', '\u{A682}'),
    ('\u{A684}', '\u{A684}'),
    ('\u{A686}', '\u{A686}'),
    ('\u{A688}', '\u{A688}'),
    ('\u{A68A}', '\u{A68A}'),
    ('\u{A68C}', '\u{A68C}'),
    ('\u{A68E}', '\u{A68E}'),
    ('\u{A690}', '\u{A690}'),
    ('\u{A692}', '\u{A692}'),
    ('\u{A694}', '\u{A694}'),
    ('\u{A696}', '\u{A696}'),
    ('\u{A698}', '\u{A698}'),
    ('\u{A69A}', '\u{A69A}'),
    ('\u{A722}', '\u{A722}'),
    ('\u{A724}', '\u{A724}'),
    ('\u{A726}', '\u{A726}'),
    ('\u{A728}', '\u{A728}'),
    ('\u{A72A}', '\u{A72A}'),
    ('\u{A72C}', '\u{A72C}'),
    ('\u{A72E}', '\u{A72E}'),
    ('\u{A732}', '\u{A732}'),
    ('\u{A734}', '\u{A734}'),
    ('\u{A736}', '\u{A736}'),
    ('\u{A738}', '\u{A738}'),
    ('\u{A73A}', '\u{A73A}'),
    ('\u{A73C}', '\u{A73C}'),
    ('\u{A73E}', '\u{A73E}'),
    ('\u{A740}', '\u{A740}'),
    ('\u{A742}', '\u{A742}'),
    ('\u{A744}', '\u{A744}'),
    ('\u{A746}', '\u{A746}'),
    ('\u{A748}', '\u{A748}'),
    ('\u{A74A}', '\u{A74A}'),
    ('\u{A74C}', '\u{A74C}'),
    ('\u{A74E}', '\u{A74E}'),
    ('\u{A750}', '\u{A750}'),
    ('\u{A752}', '\u{A752}'),
    ('\u{A754}', '\u{A754}'),
    ('\u{A756}', '\u{A756}'),
    ('\u{A758}', '\u{A758}'),
    ('\u{A75A}', '\u{A75A}'),
    ('\u{A75C}', '\u{A75C}'),
    ('\u{A75E}', '\u{A75E}'),
    ('\u{A760}', '\u{A760}'),
    ('\u{A762}', '\u{A762}'),
    ('\u{A764}', '\u{A764}'),
    ('\u{A766}', '\u{A766}'),
    ('\u{A768}', '\u{A768}'),
    ('\u{A76A}', '\u{A76A}'),
    ('\u{A76C}', '\u{A76C}'),
    ('\u{A76E}', '\u{A76E}'),
    ('\u{A779}', '\u{A779}'),
    ('\u{A77B}', '\u{A77B}'),
    ('\u{A77D}', '\u{A77E}'),
    ('\u{A780}', '\u{A780}'),
    ('\u{A782}', '\u{A782}'),
    ('\u{A784}', '\u{A784}'),
    ('\u{A786}', '\u{A786}'),
    ('\u{A78B}', '\u{A78B}'),
    ('\u{A78D}', '\u{A78D}'),
    ('\u{A790}', '\u{A790}'),
    ('\u{A792}', '\u{A792}'),
    ('\u{A796}', '\u{A796}'),
    ('\u{A798}', '\u{A798}'),
    ('\u{A79A}', '\u{A79A}'),
    ('\u{A79C}', '\u{A79C}'),
    ('\u{A79E}', '\u{A79E}'),
    ('\u{A7A0}', '\u{A7A0}'),
    ('\u{A7A2}', '\u{A7A2}'),
    ('\u{A7A4}', '\u{A7A4}'),
    ('\u{A7A6}', '\u{A7A6}'),
    ('\u{A7A8}', '\u{A7A8}'),
    ('\u{A7AA}', '\u{A7AE}'),
    ('\u{A7B0}', '\u{A7B4}'),
    ('\u{A7B6}', '\u{A7B6}'),
    ('\u{A7B8}', '\u{A7B8}'),
    ('\u{A7BA}', '\u{A7BA}'),
    ('\u{A7BC}', '\u{A7BC}'),
    ('\u{A7BE}', '\u{A7BE}'),
    ('\u{A7C0}', '\u{A7C0}'),
    ('\u{A7C2}', '\u{A7C2}'),
    ('\u{A7C4}', '\u{A7C7}'),
    ('\u{A7C9}', '\u{A7C9}'),
    ('\u{A7D0}', '\u{A7D0}'),
    ('\u{A7D6}', '\u{A7D6}'),
    ('\u{A7D8}', '\u{A7D8}'),
    ('\u{A7F5}', '\u{A7F5}'),
    ('\u{FF21}', '\u{FF3A}'),
    ('\u{10400}', '\u{10427}'),
    ('\u{104B0}', '\u{104D3}'),
    ('\u{10570}', '\u{1057A}'),
    ('\u{1057C}', '\u{1058A}'),
    ('\u{1058C}', '\u{10592}'),
    ('\u{10594}', '\u{10595}'),
    ('\u{10C80}', '\u{10CB2}'),
    ('\u{118A0}', '\u{118BF}'),
    ('\u{16E40}', '\u{16E5F}'),
    ('\u{1D400}', '\u{1D419}'),
    ('\u{1D434}', '\u{1D44D}'),
    ('\u{1D468}', '\u{1D481}'),
    ('\u{1D49C}', '\u{1D49C}'),
    ('\u{1D49E}', '\u{1D49F}'),
    ('\u{1D4A2}', '\u{1D4A2}'),
    ('\u{1D4A5}', '\u{1D4A6}'),
    ('\u{1D4A9}', '\u{1D4AC}'),
    ('\u{1D4AE}', '\u{1D4B5}'),
    ('\u{1D4D0}', '\u{1D4E9}'),
    ('\u{1D504}', '\u{1D505}'),
    ('\u{1D507}', '\u{1D50A}'),
    ('\u{1D50D}', '\u{1D514}'),
    ('\u{1D516}', '\u{1D51C}'),
    ('\u{1D538}', '\u{1D539}'),
    ('\u{1D53B}', '\u{1D53E}'),
    ('\u{1D540}', '\u{1D544}'),
    ('\u{1D546}', '\u{1D546}'),
    ('\u{1D54A}', '\u{1D550}'),
    ('\u{1D56C}', '\u{1D585}'),
    ('\u{1D5A0}', '\u{1D5B9}'),
    ('\u{1D5D4}', '\u{1D5ED}'),
    ('\u{1D608}', '\u{1D621}'),
    ('\u{1D63C}', '\u{1D655}'),
    ('\u{1D670}', '\u{1D689}'),
    ('\u{1D6A8}', '\u{1D6C0}'),
    ('\u{1D6E2}', '\u{1D6FA}'),
    ('\u{1D71C}', '\u{1D734}'),
    ('\u{1D756}', '\u{1D76E}'),
    ('\u{1D790}', '\u{1D7A8}'),
    ('\u{1D7CA}', '\u{1D7CA}'),
    ('\u{1E900}', '\u{1E921}'),
    ('\u{1F130}', '\u{1F149}'),
    ('\u{1F150}', '\u{1F169}'),
    ('\u{1F170}', '\u{1F189}'),
];

/// Inclusive codepoint ranges of characters with the Unicode `Lowercase`
/// property (general category `Ll` plus `Other_Lowercase`, e.g. roman
/// numerals `ⅻ`). Mirrors CPython's `Py_UNICODE_ISLOWER`.
///
/// Derived live from the oracle CPython 3.13 / Unicode 15.1.0,
/// run-length compressed to 671 ranges.
const LOWER_RANGES: &[(char, char)] = &[
    ('\u{0061}', '\u{007A}'),
    ('\u{00AA}', '\u{00AA}'),
    ('\u{00B5}', '\u{00B5}'),
    ('\u{00BA}', '\u{00BA}'),
    ('\u{00DF}', '\u{00F6}'),
    ('\u{00F8}', '\u{00FF}'),
    ('\u{0101}', '\u{0101}'),
    ('\u{0103}', '\u{0103}'),
    ('\u{0105}', '\u{0105}'),
    ('\u{0107}', '\u{0107}'),
    ('\u{0109}', '\u{0109}'),
    ('\u{010B}', '\u{010B}'),
    ('\u{010D}', '\u{010D}'),
    ('\u{010F}', '\u{010F}'),
    ('\u{0111}', '\u{0111}'),
    ('\u{0113}', '\u{0113}'),
    ('\u{0115}', '\u{0115}'),
    ('\u{0117}', '\u{0117}'),
    ('\u{0119}', '\u{0119}'),
    ('\u{011B}', '\u{011B}'),
    ('\u{011D}', '\u{011D}'),
    ('\u{011F}', '\u{011F}'),
    ('\u{0121}', '\u{0121}'),
    ('\u{0123}', '\u{0123}'),
    ('\u{0125}', '\u{0125}'),
    ('\u{0127}', '\u{0127}'),
    ('\u{0129}', '\u{0129}'),
    ('\u{012B}', '\u{012B}'),
    ('\u{012D}', '\u{012D}'),
    ('\u{012F}', '\u{012F}'),
    ('\u{0131}', '\u{0131}'),
    ('\u{0133}', '\u{0133}'),
    ('\u{0135}', '\u{0135}'),
    ('\u{0137}', '\u{0138}'),
    ('\u{013A}', '\u{013A}'),
    ('\u{013C}', '\u{013C}'),
    ('\u{013E}', '\u{013E}'),
    ('\u{0140}', '\u{0140}'),
    ('\u{0142}', '\u{0142}'),
    ('\u{0144}', '\u{0144}'),
    ('\u{0146}', '\u{0146}'),
    ('\u{0148}', '\u{0149}'),
    ('\u{014B}', '\u{014B}'),
    ('\u{014D}', '\u{014D}'),
    ('\u{014F}', '\u{014F}'),
    ('\u{0151}', '\u{0151}'),
    ('\u{0153}', '\u{0153}'),
    ('\u{0155}', '\u{0155}'),
    ('\u{0157}', '\u{0157}'),
    ('\u{0159}', '\u{0159}'),
    ('\u{015B}', '\u{015B}'),
    ('\u{015D}', '\u{015D}'),
    ('\u{015F}', '\u{015F}'),
    ('\u{0161}', '\u{0161}'),
    ('\u{0163}', '\u{0163}'),
    ('\u{0165}', '\u{0165}'),
    ('\u{0167}', '\u{0167}'),
    ('\u{0169}', '\u{0169}'),
    ('\u{016B}', '\u{016B}'),
    ('\u{016D}', '\u{016D}'),
    ('\u{016F}', '\u{016F}'),
    ('\u{0171}', '\u{0171}'),
    ('\u{0173}', '\u{0173}'),
    ('\u{0175}', '\u{0175}'),
    ('\u{0177}', '\u{0177}'),
    ('\u{017A}', '\u{017A}'),
    ('\u{017C}', '\u{017C}'),
    ('\u{017E}', '\u{0180}'),
    ('\u{0183}', '\u{0183}'),
    ('\u{0185}', '\u{0185}'),
    ('\u{0188}', '\u{0188}'),
    ('\u{018C}', '\u{018D}'),
    ('\u{0192}', '\u{0192}'),
    ('\u{0195}', '\u{0195}'),
    ('\u{0199}', '\u{019B}'),
    ('\u{019E}', '\u{019E}'),
    ('\u{01A1}', '\u{01A1}'),
    ('\u{01A3}', '\u{01A3}'),
    ('\u{01A5}', '\u{01A5}'),
    ('\u{01A8}', '\u{01A8}'),
    ('\u{01AA}', '\u{01AB}'),
    ('\u{01AD}', '\u{01AD}'),
    ('\u{01B0}', '\u{01B0}'),
    ('\u{01B4}', '\u{01B4}'),
    ('\u{01B6}', '\u{01B6}'),
    ('\u{01B9}', '\u{01BA}'),
    ('\u{01BD}', '\u{01BF}'),
    ('\u{01C6}', '\u{01C6}'),
    ('\u{01C9}', '\u{01C9}'),
    ('\u{01CC}', '\u{01CC}'),
    ('\u{01CE}', '\u{01CE}'),
    ('\u{01D0}', '\u{01D0}'),
    ('\u{01D2}', '\u{01D2}'),
    ('\u{01D4}', '\u{01D4}'),
    ('\u{01D6}', '\u{01D6}'),
    ('\u{01D8}', '\u{01D8}'),
    ('\u{01DA}', '\u{01DA}'),
    ('\u{01DC}', '\u{01DD}'),
    ('\u{01DF}', '\u{01DF}'),
    ('\u{01E1}', '\u{01E1}'),
    ('\u{01E3}', '\u{01E3}'),
    ('\u{01E5}', '\u{01E5}'),
    ('\u{01E7}', '\u{01E7}'),
    ('\u{01E9}', '\u{01E9}'),
    ('\u{01EB}', '\u{01EB}'),
    ('\u{01ED}', '\u{01ED}'),
    ('\u{01EF}', '\u{01F0}'),
    ('\u{01F3}', '\u{01F3}'),
    ('\u{01F5}', '\u{01F5}'),
    ('\u{01F9}', '\u{01F9}'),
    ('\u{01FB}', '\u{01FB}'),
    ('\u{01FD}', '\u{01FD}'),
    ('\u{01FF}', '\u{01FF}'),
    ('\u{0201}', '\u{0201}'),
    ('\u{0203}', '\u{0203}'),
    ('\u{0205}', '\u{0205}'),
    ('\u{0207}', '\u{0207}'),
    ('\u{0209}', '\u{0209}'),
    ('\u{020B}', '\u{020B}'),
    ('\u{020D}', '\u{020D}'),
    ('\u{020F}', '\u{020F}'),
    ('\u{0211}', '\u{0211}'),
    ('\u{0213}', '\u{0213}'),
    ('\u{0215}', '\u{0215}'),
    ('\u{0217}', '\u{0217}'),
    ('\u{0219}', '\u{0219}'),
    ('\u{021B}', '\u{021B}'),
    ('\u{021D}', '\u{021D}'),
    ('\u{021F}', '\u{021F}'),
    ('\u{0221}', '\u{0221}'),
    ('\u{0223}', '\u{0223}'),
    ('\u{0225}', '\u{0225}'),
    ('\u{0227}', '\u{0227}'),
    ('\u{0229}', '\u{0229}'),
    ('\u{022B}', '\u{022B}'),
    ('\u{022D}', '\u{022D}'),
    ('\u{022F}', '\u{022F}'),
    ('\u{0231}', '\u{0231}'),
    ('\u{0233}', '\u{0239}'),
    ('\u{023C}', '\u{023C}'),
    ('\u{023F}', '\u{0240}'),
    ('\u{0242}', '\u{0242}'),
    ('\u{0247}', '\u{0247}'),
    ('\u{0249}', '\u{0249}'),
    ('\u{024B}', '\u{024B}'),
    ('\u{024D}', '\u{024D}'),
    ('\u{024F}', '\u{0293}'),
    ('\u{0295}', '\u{02B8}'),
    ('\u{02C0}', '\u{02C1}'),
    ('\u{02E0}', '\u{02E4}'),
    ('\u{0345}', '\u{0345}'),
    ('\u{0371}', '\u{0371}'),
    ('\u{0373}', '\u{0373}'),
    ('\u{0377}', '\u{0377}'),
    ('\u{037A}', '\u{037D}'),
    ('\u{0390}', '\u{0390}'),
    ('\u{03AC}', '\u{03CE}'),
    ('\u{03D0}', '\u{03D1}'),
    ('\u{03D5}', '\u{03D7}'),
    ('\u{03D9}', '\u{03D9}'),
    ('\u{03DB}', '\u{03DB}'),
    ('\u{03DD}', '\u{03DD}'),
    ('\u{03DF}', '\u{03DF}'),
    ('\u{03E1}', '\u{03E1}'),
    ('\u{03E3}', '\u{03E3}'),
    ('\u{03E5}', '\u{03E5}'),
    ('\u{03E7}', '\u{03E7}'),
    ('\u{03E9}', '\u{03E9}'),
    ('\u{03EB}', '\u{03EB}'),
    ('\u{03ED}', '\u{03ED}'),
    ('\u{03EF}', '\u{03F3}'),
    ('\u{03F5}', '\u{03F5}'),
    ('\u{03F8}', '\u{03F8}'),
    ('\u{03FB}', '\u{03FC}'),
    ('\u{0430}', '\u{045F}'),
    ('\u{0461}', '\u{0461}'),
    ('\u{0463}', '\u{0463}'),
    ('\u{0465}', '\u{0465}'),
    ('\u{0467}', '\u{0467}'),
    ('\u{0469}', '\u{0469}'),
    ('\u{046B}', '\u{046B}'),
    ('\u{046D}', '\u{046D}'),
    ('\u{046F}', '\u{046F}'),
    ('\u{0471}', '\u{0471}'),
    ('\u{0473}', '\u{0473}'),
    ('\u{0475}', '\u{0475}'),
    ('\u{0477}', '\u{0477}'),
    ('\u{0479}', '\u{0479}'),
    ('\u{047B}', '\u{047B}'),
    ('\u{047D}', '\u{047D}'),
    ('\u{047F}', '\u{047F}'),
    ('\u{0481}', '\u{0481}'),
    ('\u{048B}', '\u{048B}'),
    ('\u{048D}', '\u{048D}'),
    ('\u{048F}', '\u{048F}'),
    ('\u{0491}', '\u{0491}'),
    ('\u{0493}', '\u{0493}'),
    ('\u{0495}', '\u{0495}'),
    ('\u{0497}', '\u{0497}'),
    ('\u{0499}', '\u{0499}'),
    ('\u{049B}', '\u{049B}'),
    ('\u{049D}', '\u{049D}'),
    ('\u{049F}', '\u{049F}'),
    ('\u{04A1}', '\u{04A1}'),
    ('\u{04A3}', '\u{04A3}'),
    ('\u{04A5}', '\u{04A5}'),
    ('\u{04A7}', '\u{04A7}'),
    ('\u{04A9}', '\u{04A9}'),
    ('\u{04AB}', '\u{04AB}'),
    ('\u{04AD}', '\u{04AD}'),
    ('\u{04AF}', '\u{04AF}'),
    ('\u{04B1}', '\u{04B1}'),
    ('\u{04B3}', '\u{04B3}'),
    ('\u{04B5}', '\u{04B5}'),
    ('\u{04B7}', '\u{04B7}'),
    ('\u{04B9}', '\u{04B9}'),
    ('\u{04BB}', '\u{04BB}'),
    ('\u{04BD}', '\u{04BD}'),
    ('\u{04BF}', '\u{04BF}'),
    ('\u{04C2}', '\u{04C2}'),
    ('\u{04C4}', '\u{04C4}'),
    ('\u{04C6}', '\u{04C6}'),
    ('\u{04C8}', '\u{04C8}'),
    ('\u{04CA}', '\u{04CA}'),
    ('\u{04CC}', '\u{04CC}'),
    ('\u{04CE}', '\u{04CF}'),
    ('\u{04D1}', '\u{04D1}'),
    ('\u{04D3}', '\u{04D3}'),
    ('\u{04D5}', '\u{04D5}'),
    ('\u{04D7}', '\u{04D7}'),
    ('\u{04D9}', '\u{04D9}'),
    ('\u{04DB}', '\u{04DB}'),
    ('\u{04DD}', '\u{04DD}'),
    ('\u{04DF}', '\u{04DF}'),
    ('\u{04E1}', '\u{04E1}'),
    ('\u{04E3}', '\u{04E3}'),
    ('\u{04E5}', '\u{04E5}'),
    ('\u{04E7}', '\u{04E7}'),
    ('\u{04E9}', '\u{04E9}'),
    ('\u{04EB}', '\u{04EB}'),
    ('\u{04ED}', '\u{04ED}'),
    ('\u{04EF}', '\u{04EF}'),
    ('\u{04F1}', '\u{04F1}'),
    ('\u{04F3}', '\u{04F3}'),
    ('\u{04F5}', '\u{04F5}'),
    ('\u{04F7}', '\u{04F7}'),
    ('\u{04F9}', '\u{04F9}'),
    ('\u{04FB}', '\u{04FB}'),
    ('\u{04FD}', '\u{04FD}'),
    ('\u{04FF}', '\u{04FF}'),
    ('\u{0501}', '\u{0501}'),
    ('\u{0503}', '\u{0503}'),
    ('\u{0505}', '\u{0505}'),
    ('\u{0507}', '\u{0507}'),
    ('\u{0509}', '\u{0509}'),
    ('\u{050B}', '\u{050B}'),
    ('\u{050D}', '\u{050D}'),
    ('\u{050F}', '\u{050F}'),
    ('\u{0511}', '\u{0511}'),
    ('\u{0513}', '\u{0513}'),
    ('\u{0515}', '\u{0515}'),
    ('\u{0517}', '\u{0517}'),
    ('\u{0519}', '\u{0519}'),
    ('\u{051B}', '\u{051B}'),
    ('\u{051D}', '\u{051D}'),
    ('\u{051F}', '\u{051F}'),
    ('\u{0521}', '\u{0521}'),
    ('\u{0523}', '\u{0523}'),
    ('\u{0525}', '\u{0525}'),
    ('\u{0527}', '\u{0527}'),
    ('\u{0529}', '\u{0529}'),
    ('\u{052B}', '\u{052B}'),
    ('\u{052D}', '\u{052D}'),
    ('\u{052F}', '\u{052F}'),
    ('\u{0560}', '\u{0588}'),
    ('\u{10D0}', '\u{10FA}'),
    ('\u{10FC}', '\u{10FF}'),
    ('\u{13F8}', '\u{13FD}'),
    ('\u{1C80}', '\u{1C88}'),
    ('\u{1D00}', '\u{1DBF}'),
    ('\u{1E01}', '\u{1E01}'),
    ('\u{1E03}', '\u{1E03}'),
    ('\u{1E05}', '\u{1E05}'),
    ('\u{1E07}', '\u{1E07}'),
    ('\u{1E09}', '\u{1E09}'),
    ('\u{1E0B}', '\u{1E0B}'),
    ('\u{1E0D}', '\u{1E0D}'),
    ('\u{1E0F}', '\u{1E0F}'),
    ('\u{1E11}', '\u{1E11}'),
    ('\u{1E13}', '\u{1E13}'),
    ('\u{1E15}', '\u{1E15}'),
    ('\u{1E17}', '\u{1E17}'),
    ('\u{1E19}', '\u{1E19}'),
    ('\u{1E1B}', '\u{1E1B}'),
    ('\u{1E1D}', '\u{1E1D}'),
    ('\u{1E1F}', '\u{1E1F}'),
    ('\u{1E21}', '\u{1E21}'),
    ('\u{1E23}', '\u{1E23}'),
    ('\u{1E25}', '\u{1E25}'),
    ('\u{1E27}', '\u{1E27}'),
    ('\u{1E29}', '\u{1E29}'),
    ('\u{1E2B}', '\u{1E2B}'),
    ('\u{1E2D}', '\u{1E2D}'),
    ('\u{1E2F}', '\u{1E2F}'),
    ('\u{1E31}', '\u{1E31}'),
    ('\u{1E33}', '\u{1E33}'),
    ('\u{1E35}', '\u{1E35}'),
    ('\u{1E37}', '\u{1E37}'),
    ('\u{1E39}', '\u{1E39}'),
    ('\u{1E3B}', '\u{1E3B}'),
    ('\u{1E3D}', '\u{1E3D}'),
    ('\u{1E3F}', '\u{1E3F}'),
    ('\u{1E41}', '\u{1E41}'),
    ('\u{1E43}', '\u{1E43}'),
    ('\u{1E45}', '\u{1E45}'),
    ('\u{1E47}', '\u{1E47}'),
    ('\u{1E49}', '\u{1E49}'),
    ('\u{1E4B}', '\u{1E4B}'),
    ('\u{1E4D}', '\u{1E4D}'),
    ('\u{1E4F}', '\u{1E4F}'),
    ('\u{1E51}', '\u{1E51}'),
    ('\u{1E53}', '\u{1E53}'),
    ('\u{1E55}', '\u{1E55}'),
    ('\u{1E57}', '\u{1E57}'),
    ('\u{1E59}', '\u{1E59}'),
    ('\u{1E5B}', '\u{1E5B}'),
    ('\u{1E5D}', '\u{1E5D}'),
    ('\u{1E5F}', '\u{1E5F}'),
    ('\u{1E61}', '\u{1E61}'),
    ('\u{1E63}', '\u{1E63}'),
    ('\u{1E65}', '\u{1E65}'),
    ('\u{1E67}', '\u{1E67}'),
    ('\u{1E69}', '\u{1E69}'),
    ('\u{1E6B}', '\u{1E6B}'),
    ('\u{1E6D}', '\u{1E6D}'),
    ('\u{1E6F}', '\u{1E6F}'),
    ('\u{1E71}', '\u{1E71}'),
    ('\u{1E73}', '\u{1E73}'),
    ('\u{1E75}', '\u{1E75}'),
    ('\u{1E77}', '\u{1E77}'),
    ('\u{1E79}', '\u{1E79}'),
    ('\u{1E7B}', '\u{1E7B}'),
    ('\u{1E7D}', '\u{1E7D}'),
    ('\u{1E7F}', '\u{1E7F}'),
    ('\u{1E81}', '\u{1E81}'),
    ('\u{1E83}', '\u{1E83}'),
    ('\u{1E85}', '\u{1E85}'),
    ('\u{1E87}', '\u{1E87}'),
    ('\u{1E89}', '\u{1E89}'),
    ('\u{1E8B}', '\u{1E8B}'),
    ('\u{1E8D}', '\u{1E8D}'),
    ('\u{1E8F}', '\u{1E8F}'),
    ('\u{1E91}', '\u{1E91}'),
    ('\u{1E93}', '\u{1E93}'),
    ('\u{1E95}', '\u{1E9D}'),
    ('\u{1E9F}', '\u{1E9F}'),
    ('\u{1EA1}', '\u{1EA1}'),
    ('\u{1EA3}', '\u{1EA3}'),
    ('\u{1EA5}', '\u{1EA5}'),
    ('\u{1EA7}', '\u{1EA7}'),
    ('\u{1EA9}', '\u{1EA9}'),
    ('\u{1EAB}', '\u{1EAB}'),
    ('\u{1EAD}', '\u{1EAD}'),
    ('\u{1EAF}', '\u{1EAF}'),
    ('\u{1EB1}', '\u{1EB1}'),
    ('\u{1EB3}', '\u{1EB3}'),
    ('\u{1EB5}', '\u{1EB5}'),
    ('\u{1EB7}', '\u{1EB7}'),
    ('\u{1EB9}', '\u{1EB9}'),
    ('\u{1EBB}', '\u{1EBB}'),
    ('\u{1EBD}', '\u{1EBD}'),
    ('\u{1EBF}', '\u{1EBF}'),
    ('\u{1EC1}', '\u{1EC1}'),
    ('\u{1EC3}', '\u{1EC3}'),
    ('\u{1EC5}', '\u{1EC5}'),
    ('\u{1EC7}', '\u{1EC7}'),
    ('\u{1EC9}', '\u{1EC9}'),
    ('\u{1ECB}', '\u{1ECB}'),
    ('\u{1ECD}', '\u{1ECD}'),
    ('\u{1ECF}', '\u{1ECF}'),
    ('\u{1ED1}', '\u{1ED1}'),
    ('\u{1ED3}', '\u{1ED3}'),
    ('\u{1ED5}', '\u{1ED5}'),
    ('\u{1ED7}', '\u{1ED7}'),
    ('\u{1ED9}', '\u{1ED9}'),
    ('\u{1EDB}', '\u{1EDB}'),
    ('\u{1EDD}', '\u{1EDD}'),
    ('\u{1EDF}', '\u{1EDF}'),
    ('\u{1EE1}', '\u{1EE1}'),
    ('\u{1EE3}', '\u{1EE3}'),
    ('\u{1EE5}', '\u{1EE5}'),
    ('\u{1EE7}', '\u{1EE7}'),
    ('\u{1EE9}', '\u{1EE9}'),
    ('\u{1EEB}', '\u{1EEB}'),
    ('\u{1EED}', '\u{1EED}'),
    ('\u{1EEF}', '\u{1EEF}'),
    ('\u{1EF1}', '\u{1EF1}'),
    ('\u{1EF3}', '\u{1EF3}'),
    ('\u{1EF5}', '\u{1EF5}'),
    ('\u{1EF7}', '\u{1EF7}'),
    ('\u{1EF9}', '\u{1EF9}'),
    ('\u{1EFB}', '\u{1EFB}'),
    ('\u{1EFD}', '\u{1EFD}'),
    ('\u{1EFF}', '\u{1F07}'),
    ('\u{1F10}', '\u{1F15}'),
    ('\u{1F20}', '\u{1F27}'),
    ('\u{1F30}', '\u{1F37}'),
    ('\u{1F40}', '\u{1F45}'),
    ('\u{1F50}', '\u{1F57}'),
    ('\u{1F60}', '\u{1F67}'),
    ('\u{1F70}', '\u{1F7D}'),
    ('\u{1F80}', '\u{1F87}'),
    ('\u{1F90}', '\u{1F97}'),
    ('\u{1FA0}', '\u{1FA7}'),
    ('\u{1FB0}', '\u{1FB4}'),
    ('\u{1FB6}', '\u{1FB7}'),
    ('\u{1FBE}', '\u{1FBE}'),
    ('\u{1FC2}', '\u{1FC4}'),
    ('\u{1FC6}', '\u{1FC7}'),
    ('\u{1FD0}', '\u{1FD3}'),
    ('\u{1FD6}', '\u{1FD7}'),
    ('\u{1FE0}', '\u{1FE7}'),
    ('\u{1FF2}', '\u{1FF4}'),
    ('\u{1FF6}', '\u{1FF7}'),
    ('\u{2071}', '\u{2071}'),
    ('\u{207F}', '\u{207F}'),
    ('\u{2090}', '\u{209C}'),
    ('\u{210A}', '\u{210A}'),
    ('\u{210E}', '\u{210F}'),
    ('\u{2113}', '\u{2113}'),
    ('\u{212F}', '\u{212F}'),
    ('\u{2134}', '\u{2134}'),
    ('\u{2139}', '\u{2139}'),
    ('\u{213C}', '\u{213D}'),
    ('\u{2146}', '\u{2149}'),
    ('\u{214E}', '\u{214E}'),
    ('\u{2170}', '\u{217F}'),
    ('\u{2184}', '\u{2184}'),
    ('\u{24D0}', '\u{24E9}'),
    ('\u{2C30}', '\u{2C5F}'),
    ('\u{2C61}', '\u{2C61}'),
    ('\u{2C65}', '\u{2C66}'),
    ('\u{2C68}', '\u{2C68}'),
    ('\u{2C6A}', '\u{2C6A}'),
    ('\u{2C6C}', '\u{2C6C}'),
    ('\u{2C71}', '\u{2C71}'),
    ('\u{2C73}', '\u{2C74}'),
    ('\u{2C76}', '\u{2C7D}'),
    ('\u{2C81}', '\u{2C81}'),
    ('\u{2C83}', '\u{2C83}'),
    ('\u{2C85}', '\u{2C85}'),
    ('\u{2C87}', '\u{2C87}'),
    ('\u{2C89}', '\u{2C89}'),
    ('\u{2C8B}', '\u{2C8B}'),
    ('\u{2C8D}', '\u{2C8D}'),
    ('\u{2C8F}', '\u{2C8F}'),
    ('\u{2C91}', '\u{2C91}'),
    ('\u{2C93}', '\u{2C93}'),
    ('\u{2C95}', '\u{2C95}'),
    ('\u{2C97}', '\u{2C97}'),
    ('\u{2C99}', '\u{2C99}'),
    ('\u{2C9B}', '\u{2C9B}'),
    ('\u{2C9D}', '\u{2C9D}'),
    ('\u{2C9F}', '\u{2C9F}'),
    ('\u{2CA1}', '\u{2CA1}'),
    ('\u{2CA3}', '\u{2CA3}'),
    ('\u{2CA5}', '\u{2CA5}'),
    ('\u{2CA7}', '\u{2CA7}'),
    ('\u{2CA9}', '\u{2CA9}'),
    ('\u{2CAB}', '\u{2CAB}'),
    ('\u{2CAD}', '\u{2CAD}'),
    ('\u{2CAF}', '\u{2CAF}'),
    ('\u{2CB1}', '\u{2CB1}'),
    ('\u{2CB3}', '\u{2CB3}'),
    ('\u{2CB5}', '\u{2CB5}'),
    ('\u{2CB7}', '\u{2CB7}'),
    ('\u{2CB9}', '\u{2CB9}'),
    ('\u{2CBB}', '\u{2CBB}'),
    ('\u{2CBD}', '\u{2CBD}'),
    ('\u{2CBF}', '\u{2CBF}'),
    ('\u{2CC1}', '\u{2CC1}'),
    ('\u{2CC3}', '\u{2CC3}'),
    ('\u{2CC5}', '\u{2CC5}'),
    ('\u{2CC7}', '\u{2CC7}'),
    ('\u{2CC9}', '\u{2CC9}'),
    ('\u{2CCB}', '\u{2CCB}'),
    ('\u{2CCD}', '\u{2CCD}'),
    ('\u{2CCF}', '\u{2CCF}'),
    ('\u{2CD1}', '\u{2CD1}'),
    ('\u{2CD3}', '\u{2CD3}'),
    ('\u{2CD5}', '\u{2CD5}'),
    ('\u{2CD7}', '\u{2CD7}'),
    ('\u{2CD9}', '\u{2CD9}'),
    ('\u{2CDB}', '\u{2CDB}'),
    ('\u{2CDD}', '\u{2CDD}'),
    ('\u{2CDF}', '\u{2CDF}'),
    ('\u{2CE1}', '\u{2CE1}'),
    ('\u{2CE3}', '\u{2CE4}'),
    ('\u{2CEC}', '\u{2CEC}'),
    ('\u{2CEE}', '\u{2CEE}'),
    ('\u{2CF3}', '\u{2CF3}'),
    ('\u{2D00}', '\u{2D25}'),
    ('\u{2D27}', '\u{2D27}'),
    ('\u{2D2D}', '\u{2D2D}'),
    ('\u{A641}', '\u{A641}'),
    ('\u{A643}', '\u{A643}'),
    ('\u{A645}', '\u{A645}'),
    ('\u{A647}', '\u{A647}'),
    ('\u{A649}', '\u{A649}'),
    ('\u{A64B}', '\u{A64B}'),
    ('\u{A64D}', '\u{A64D}'),
    ('\u{A64F}', '\u{A64F}'),
    ('\u{A651}', '\u{A651}'),
    ('\u{A653}', '\u{A653}'),
    ('\u{A655}', '\u{A655}'),
    ('\u{A657}', '\u{A657}'),
    ('\u{A659}', '\u{A659}'),
    ('\u{A65B}', '\u{A65B}'),
    ('\u{A65D}', '\u{A65D}'),
    ('\u{A65F}', '\u{A65F}'),
    ('\u{A661}', '\u{A661}'),
    ('\u{A663}', '\u{A663}'),
    ('\u{A665}', '\u{A665}'),
    ('\u{A667}', '\u{A667}'),
    ('\u{A669}', '\u{A669}'),
    ('\u{A66B}', '\u{A66B}'),
    ('\u{A66D}', '\u{A66D}'),
    ('\u{A681}', '\u{A681}'),
    ('\u{A683}', '\u{A683}'),
    ('\u{A685}', '\u{A685}'),
    ('\u{A687}', '\u{A687}'),
    ('\u{A689}', '\u{A689}'),
    ('\u{A68B}', '\u{A68B}'),
    ('\u{A68D}', '\u{A68D}'),
    ('\u{A68F}', '\u{A68F}'),
    ('\u{A691}', '\u{A691}'),
    ('\u{A693}', '\u{A693}'),
    ('\u{A695}', '\u{A695}'),
    ('\u{A697}', '\u{A697}'),
    ('\u{A699}', '\u{A699}'),
    ('\u{A69B}', '\u{A69D}'),
    ('\u{A723}', '\u{A723}'),
    ('\u{A725}', '\u{A725}'),
    ('\u{A727}', '\u{A727}'),
    ('\u{A729}', '\u{A729}'),
    ('\u{A72B}', '\u{A72B}'),
    ('\u{A72D}', '\u{A72D}'),
    ('\u{A72F}', '\u{A731}'),
    ('\u{A733}', '\u{A733}'),
    ('\u{A735}', '\u{A735}'),
    ('\u{A737}', '\u{A737}'),
    ('\u{A739}', '\u{A739}'),
    ('\u{A73B}', '\u{A73B}'),
    ('\u{A73D}', '\u{A73D}'),
    ('\u{A73F}', '\u{A73F}'),
    ('\u{A741}', '\u{A741}'),
    ('\u{A743}', '\u{A743}'),
    ('\u{A745}', '\u{A745}'),
    ('\u{A747}', '\u{A747}'),
    ('\u{A749}', '\u{A749}'),
    ('\u{A74B}', '\u{A74B}'),
    ('\u{A74D}', '\u{A74D}'),
    ('\u{A74F}', '\u{A74F}'),
    ('\u{A751}', '\u{A751}'),
    ('\u{A753}', '\u{A753}'),
    ('\u{A755}', '\u{A755}'),
    ('\u{A757}', '\u{A757}'),
    ('\u{A759}', '\u{A759}'),
    ('\u{A75B}', '\u{A75B}'),
    ('\u{A75D}', '\u{A75D}'),
    ('\u{A75F}', '\u{A75F}'),
    ('\u{A761}', '\u{A761}'),
    ('\u{A763}', '\u{A763}'),
    ('\u{A765}', '\u{A765}'),
    ('\u{A767}', '\u{A767}'),
    ('\u{A769}', '\u{A769}'),
    ('\u{A76B}', '\u{A76B}'),
    ('\u{A76D}', '\u{A76D}'),
    ('\u{A76F}', '\u{A778}'),
    ('\u{A77A}', '\u{A77A}'),
    ('\u{A77C}', '\u{A77C}'),
    ('\u{A77F}', '\u{A77F}'),
    ('\u{A781}', '\u{A781}'),
    ('\u{A783}', '\u{A783}'),
    ('\u{A785}', '\u{A785}'),
    ('\u{A787}', '\u{A787}'),
    ('\u{A78C}', '\u{A78C}'),
    ('\u{A78E}', '\u{A78E}'),
    ('\u{A791}', '\u{A791}'),
    ('\u{A793}', '\u{A795}'),
    ('\u{A797}', '\u{A797}'),
    ('\u{A799}', '\u{A799}'),
    ('\u{A79B}', '\u{A79B}'),
    ('\u{A79D}', '\u{A79D}'),
    ('\u{A79F}', '\u{A79F}'),
    ('\u{A7A1}', '\u{A7A1}'),
    ('\u{A7A3}', '\u{A7A3}'),
    ('\u{A7A5}', '\u{A7A5}'),
    ('\u{A7A7}', '\u{A7A7}'),
    ('\u{A7A9}', '\u{A7A9}'),
    ('\u{A7AF}', '\u{A7AF}'),
    ('\u{A7B5}', '\u{A7B5}'),
    ('\u{A7B7}', '\u{A7B7}'),
    ('\u{A7B9}', '\u{A7B9}'),
    ('\u{A7BB}', '\u{A7BB}'),
    ('\u{A7BD}', '\u{A7BD}'),
    ('\u{A7BF}', '\u{A7BF}'),
    ('\u{A7C1}', '\u{A7C1}'),
    ('\u{A7C3}', '\u{A7C3}'),
    ('\u{A7C8}', '\u{A7C8}'),
    ('\u{A7CA}', '\u{A7CA}'),
    ('\u{A7D1}', '\u{A7D1}'),
    ('\u{A7D3}', '\u{A7D3}'),
    ('\u{A7D5}', '\u{A7D5}'),
    ('\u{A7D7}', '\u{A7D7}'),
    ('\u{A7D9}', '\u{A7D9}'),
    ('\u{A7F2}', '\u{A7F4}'),
    ('\u{A7F6}', '\u{A7F6}'),
    ('\u{A7F8}', '\u{A7FA}'),
    ('\u{AB30}', '\u{AB5A}'),
    ('\u{AB5C}', '\u{AB69}'),
    ('\u{AB70}', '\u{ABBF}'),
    ('\u{FB00}', '\u{FB06}'),
    ('\u{FB13}', '\u{FB17}'),
    ('\u{FF41}', '\u{FF5A}'),
    ('\u{10428}', '\u{1044F}'),
    ('\u{104D8}', '\u{104FB}'),
    ('\u{10597}', '\u{105A1}'),
    ('\u{105A3}', '\u{105B1}'),
    ('\u{105B3}', '\u{105B9}'),
    ('\u{105BB}', '\u{105BC}'),
    ('\u{10780}', '\u{10780}'),
    ('\u{10783}', '\u{10785}'),
    ('\u{10787}', '\u{107B0}'),
    ('\u{107B2}', '\u{107BA}'),
    ('\u{10CC0}', '\u{10CF2}'),
    ('\u{118C0}', '\u{118DF}'),
    ('\u{16E60}', '\u{16E7F}'),
    ('\u{1D41A}', '\u{1D433}'),
    ('\u{1D44E}', '\u{1D454}'),
    ('\u{1D456}', '\u{1D467}'),
    ('\u{1D482}', '\u{1D49B}'),
    ('\u{1D4B6}', '\u{1D4B9}'),
    ('\u{1D4BB}', '\u{1D4BB}'),
    ('\u{1D4BD}', '\u{1D4C3}'),
    ('\u{1D4C5}', '\u{1D4CF}'),
    ('\u{1D4EA}', '\u{1D503}'),
    ('\u{1D51E}', '\u{1D537}'),
    ('\u{1D552}', '\u{1D56B}'),
    ('\u{1D586}', '\u{1D59F}'),
    ('\u{1D5BA}', '\u{1D5D3}'),
    ('\u{1D5EE}', '\u{1D607}'),
    ('\u{1D622}', '\u{1D63B}'),
    ('\u{1D656}', '\u{1D66F}'),
    ('\u{1D68A}', '\u{1D6A5}'),
    ('\u{1D6C2}', '\u{1D6DA}'),
    ('\u{1D6DC}', '\u{1D6E1}'),
    ('\u{1D6FC}', '\u{1D714}'),
    ('\u{1D716}', '\u{1D71B}'),
    ('\u{1D736}', '\u{1D74E}'),
    ('\u{1D750}', '\u{1D755}'),
    ('\u{1D770}', '\u{1D788}'),
    ('\u{1D78A}', '\u{1D78F}'),
    ('\u{1D7AA}', '\u{1D7C2}'),
    ('\u{1D7C4}', '\u{1D7C9}'),
    ('\u{1D7CB}', '\u{1D7CB}'),
    ('\u{1DF00}', '\u{1DF09}'),
    ('\u{1DF0B}', '\u{1DF1E}'),
    ('\u{1DF25}', '\u{1DF2A}'),
    ('\u{1E030}', '\u{1E06D}'),
    ('\u{1E922}', '\u{1E943}'),
];

/// Inclusive codepoint ranges of characters in Unicode general category
/// `Lt` (titlecase letters such as `Dž`, `Lj`, `Nj`, `Dz`). Mirrors
/// CPython's `Py_UNICODE_ISTITLE`. These count as neither upper nor
/// lower but ARE cased, and open a title-word.
///
/// Derived live from the oracle CPython 3.13 / Unicode 15.1.0.
const TITLE_RANGES: &[(char, char)] = &[
    ('\u{01C5}', '\u{01C5}'),
    ('\u{01C8}', '\u{01C8}'),
    ('\u{01CB}', '\u{01CB}'),
    ('\u{01F2}', '\u{01F2}'),
    ('\u{1F88}', '\u{1F8F}'),
    ('\u{1F98}', '\u{1F9F}'),
    ('\u{1FA8}', '\u{1FAF}'),
    ('\u{1FBC}', '\u{1FBC}'),
    ('\u{1FCC}', '\u{1FCC}'),
    ('\u{1FFC}', '\u{1FFC}'),
];

/// Inclusive codepoint ranges of characters for which CPython's
/// `str.isspace()` returns `True` (Unicode `White_Space` plus the bidi
/// `WS`/`B`/`S` C0 controls `U+001C`..`U+001F`). BROADER than Rust's
/// `char::is_whitespace` (the `White_Space` property only), which omits
/// the file/group/record/unit separators.
///
/// Derived live from the oracle CPython 3.13 / Unicode 15.1.0.
const WHITESPACE_RANGES: &[(char, char)] = &[
    ('\u{0009}', '\u{000D}'),
    ('\u{001C}', '\u{0020}'),
    ('\u{0085}', '\u{0085}'),
    ('\u{00A0}', '\u{00A0}'),
    ('\u{1680}', '\u{1680}'),
    ('\u{2000}', '\u{200A}'),
    ('\u{2028}', '\u{2029}'),
    ('\u{202F}', '\u{202F}'),
    ('\u{205F}', '\u{205F}'),
    ('\u{3000}', '\u{3000}'),
];

/// Inclusive codepoint ranges of characters with Unicode `Numeric_Type`
/// of Decimal, Digit, **or** Numeric — i.e. exactly the set for which
/// CPython's `str.isnumeric()` returns `True` for a single character
/// (`Py_UNICODE_ISNUMERIC`). This is the BROADEST of the numeric
/// predicates: it is a superset of [`DIGIT_RANGES`] (which is itself a
/// superset of [`DECIMAL_RANGES`]), additionally accepting fractions
/// (`½`), letter-numerals such as roman `Ⅻ` (category `Nl`), and
/// ideographic numerals such as `一` (U+4E00, category `Lo`).
///
/// Derived live from the oracle CPython 3.13 / Unicode 15.1.0 (the
/// version backing the installed `numpy` 2.4 build that this crate is
/// verified against):
/// `[c for c in range(0x110000) if chr(c).isnumeric()]`, run-length
/// compressed to 219 ranges. Rust's `char::is_numeric` tests the
/// general categories `Nd|Nl|No` instead of the `Numeric_Type`
/// property, so it MISSES ideographic numerals (category `Lo`, e.g.
/// `一`) and INCLUDES codepoints assigned after Unicode 15.1.0; the
/// table is therefore embedded directly rather than computed from
/// `std` or a Unicode crate.
const NUMERIC_TYPE_RANGES: &[(char, char)] = &[
    ('\u{0030}', '\u{0039}'),
    ('\u{00B2}', '\u{00B3}'),
    ('\u{00B9}', '\u{00B9}'),
    ('\u{00BC}', '\u{00BE}'),
    ('\u{0660}', '\u{0669}'),
    ('\u{06F0}', '\u{06F9}'),
    ('\u{07C0}', '\u{07C9}'),
    ('\u{0966}', '\u{096F}'),
    ('\u{09E6}', '\u{09EF}'),
    ('\u{09F4}', '\u{09F9}'),
    ('\u{0A66}', '\u{0A6F}'),
    ('\u{0AE6}', '\u{0AEF}'),
    ('\u{0B66}', '\u{0B6F}'),
    ('\u{0B72}', '\u{0B77}'),
    ('\u{0BE6}', '\u{0BF2}'),
    ('\u{0C66}', '\u{0C6F}'),
    ('\u{0C78}', '\u{0C7E}'),
    ('\u{0CE6}', '\u{0CEF}'),
    ('\u{0D58}', '\u{0D5E}'),
    ('\u{0D66}', '\u{0D78}'),
    ('\u{0DE6}', '\u{0DEF}'),
    ('\u{0E50}', '\u{0E59}'),
    ('\u{0ED0}', '\u{0ED9}'),
    ('\u{0F20}', '\u{0F33}'),
    ('\u{1040}', '\u{1049}'),
    ('\u{1090}', '\u{1099}'),
    ('\u{1369}', '\u{137C}'),
    ('\u{16EE}', '\u{16F0}'),
    ('\u{17E0}', '\u{17E9}'),
    ('\u{17F0}', '\u{17F9}'),
    ('\u{1810}', '\u{1819}'),
    ('\u{1946}', '\u{194F}'),
    ('\u{19D0}', '\u{19DA}'),
    ('\u{1A80}', '\u{1A89}'),
    ('\u{1A90}', '\u{1A99}'),
    ('\u{1B50}', '\u{1B59}'),
    ('\u{1BB0}', '\u{1BB9}'),
    ('\u{1C40}', '\u{1C49}'),
    ('\u{1C50}', '\u{1C59}'),
    ('\u{2070}', '\u{2070}'),
    ('\u{2074}', '\u{2079}'),
    ('\u{2080}', '\u{2089}'),
    ('\u{2150}', '\u{2182}'),
    ('\u{2185}', '\u{2189}'),
    ('\u{2460}', '\u{249B}'),
    ('\u{24EA}', '\u{24FF}'),
    ('\u{2776}', '\u{2793}'),
    ('\u{2CFD}', '\u{2CFD}'),
    ('\u{3007}', '\u{3007}'),
    ('\u{3021}', '\u{3029}'),
    ('\u{3038}', '\u{303A}'),
    ('\u{3192}', '\u{3195}'),
    ('\u{3220}', '\u{3229}'),
    ('\u{3248}', '\u{324F}'),
    ('\u{3251}', '\u{325F}'),
    ('\u{3280}', '\u{3289}'),
    ('\u{32B1}', '\u{32BF}'),
    ('\u{3405}', '\u{3405}'),
    ('\u{3483}', '\u{3483}'),
    ('\u{382A}', '\u{382A}'),
    ('\u{3B4D}', '\u{3B4D}'),
    ('\u{4E00}', '\u{4E00}'),
    ('\u{4E03}', '\u{4E03}'),
    ('\u{4E07}', '\u{4E07}'),
    ('\u{4E09}', '\u{4E09}'),
    ('\u{4E24}', '\u{4E24}'),
    ('\u{4E5D}', '\u{4E5D}'),
    ('\u{4E8C}', '\u{4E8C}'),
    ('\u{4E94}', '\u{4E94}'),
    ('\u{4E96}', '\u{4E96}'),
    ('\u{4EAC}', '\u{4EAC}'),
    ('\u{4EBF}', '\u{4EC0}'),
    ('\u{4EDF}', '\u{4EDF}'),
    ('\u{4EE8}', '\u{4EE8}'),
    ('\u{4F0D}', '\u{4F0D}'),
    ('\u{4F70}', '\u{4F70}'),
    ('\u{4FE9}', '\u{4FE9}'),
    ('\u{5006}', '\u{5006}'),
    ('\u{5104}', '\u{5104}'),
    ('\u{5146}', '\u{5146}'),
    ('\u{5169}', '\u{5169}'),
    ('\u{516B}', '\u{516B}'),
    ('\u{516D}', '\u{516D}'),
    ('\u{5341}', '\u{5341}'),
    ('\u{5343}', '\u{5345}'),
    ('\u{534C}', '\u{534C}'),
    ('\u{53C1}', '\u{53C4}'),
    ('\u{56DB}', '\u{56DB}'),
    ('\u{58F1}', '\u{58F1}'),
    ('\u{58F9}', '\u{58F9}'),
    ('\u{5E7A}', '\u{5E7A}'),
    ('\u{5EFE}', '\u{5EFF}'),
    ('\u{5F0C}', '\u{5F0E}'),
    ('\u{5F10}', '\u{5F10}'),
    ('\u{62D0}', '\u{62D0}'),
    ('\u{62FE}', '\u{62FE}'),
    ('\u{634C}', '\u{634C}'),
    ('\u{67D2}', '\u{67D2}'),
    ('\u{6D1E}', '\u{6D1E}'),
    ('\u{6F06}', '\u{6F06}'),
    ('\u{7396}', '\u{7396}'),
    ('\u{767E}', '\u{767E}'),
    ('\u{7695}', '\u{7695}'),
    ('\u{79ED}', '\u{79ED}'),
    ('\u{8086}', '\u{8086}'),
    ('\u{842C}', '\u{842C}'),
    ('\u{8CAE}', '\u{8CAE}'),
    ('\u{8CB3}', '\u{8CB3}'),
    ('\u{8D30}', '\u{8D30}'),
    ('\u{920E}', '\u{920E}'),
    ('\u{94A9}', '\u{94A9}'),
    ('\u{9621}', '\u{9621}'),
    ('\u{9646}', '\u{9646}'),
    ('\u{964C}', '\u{964C}'),
    ('\u{9678}', '\u{9678}'),
    ('\u{96F6}', '\u{96F6}'),
    ('\u{A620}', '\u{A629}'),
    ('\u{A6E6}', '\u{A6EF}'),
    ('\u{A830}', '\u{A835}'),
    ('\u{A8D0}', '\u{A8D9}'),
    ('\u{A900}', '\u{A909}'),
    ('\u{A9D0}', '\u{A9D9}'),
    ('\u{A9F0}', '\u{A9F9}'),
    ('\u{AA50}', '\u{AA59}'),
    ('\u{ABF0}', '\u{ABF9}'),
    ('\u{F96B}', '\u{F96B}'),
    ('\u{F973}', '\u{F973}'),
    ('\u{F978}', '\u{F978}'),
    ('\u{F9B2}', '\u{F9B2}'),
    ('\u{F9D1}', '\u{F9D1}'),
    ('\u{F9D3}', '\u{F9D3}'),
    ('\u{F9FD}', '\u{F9FD}'),
    ('\u{FF10}', '\u{FF19}'),
    ('\u{10107}', '\u{10133}'),
    ('\u{10140}', '\u{10178}'),
    ('\u{1018A}', '\u{1018B}'),
    ('\u{102E1}', '\u{102FB}'),
    ('\u{10320}', '\u{10323}'),
    ('\u{10341}', '\u{10341}'),
    ('\u{1034A}', '\u{1034A}'),
    ('\u{103D1}', '\u{103D5}'),
    ('\u{104A0}', '\u{104A9}'),
    ('\u{10858}', '\u{1085F}'),
    ('\u{10879}', '\u{1087F}'),
    ('\u{108A7}', '\u{108AF}'),
    ('\u{108FB}', '\u{108FF}'),
    ('\u{10916}', '\u{1091B}'),
    ('\u{109BC}', '\u{109BD}'),
    ('\u{109C0}', '\u{109CF}'),
    ('\u{109D2}', '\u{109FF}'),
    ('\u{10A40}', '\u{10A48}'),
    ('\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{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{11F50}', '\u{11F59}'),
    ('\u{11FC0}', '\u{11FD4}'),
    ('\u{12400}', '\u{1246E}'),
    ('\u{16A60}', '\u{16A69}'),
    ('\u{16AC0}', '\u{16AC9}'),
    ('\u{16B50}', '\u{16B59}'),
    ('\u{16B5B}', '\u{16B61}'),
    ('\u{16E80}', '\u{16E96}'),
    ('\u{1D2C0}', '\u{1D2D3}'),
    ('\u{1D2E0}', '\u{1D2F3}'),
    ('\u{1D360}', '\u{1D378}'),
    ('\u{1D7CE}', '\u{1D7FF}'),
    ('\u{1E140}', '\u{1E149}'),
    ('\u{1E2F0}', '\u{1E2F9}'),
    ('\u{1E4F0}', '\u{1E4F9}'),
    ('\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}'),
    ('\u{20001}', '\u{20001}'),
    ('\u{20064}', '\u{20064}'),
    ('\u{200E2}', '\u{200E2}'),
    ('\u{20121}', '\u{20121}'),
    ('\u{2092A}', '\u{2092A}'),
    ('\u{20983}', '\u{20983}'),
    ('\u{2098C}', '\u{2098C}'),
    ('\u{2099C}', '\u{2099C}'),
    ('\u{20AEA}', '\u{20AEA}'),
    ('\u{20AFD}', '\u{20AFD}'),
    ('\u{20B19}', '\u{20B19}'),
    ('\u{22390}', '\u{22390}'),
    ('\u{22998}', '\u{22998}'),
    ('\u{23B1B}', '\u{23B1B}'),
    ('\u{2626D}', '\u{2626D}'),
    ('\u{2F890}', '\u{2F890}'),
];

/// `true` if `c` has Unicode `Numeric_Type` Decimal or Digit
/// (`Py_UNICODE_ISDIGIT`).
fn is_digit_char(c: char) -> bool {
    in_ranges(DIGIT_RANGES, c)
}

/// `true` if `c` has Unicode `Numeric_Type` Decimal (`Py_UNICODE_ISDECIMAL`).
fn is_decimal_char(c: char) -> bool {
    in_ranges(DECIMAL_RANGES, c)
}

/// `true` if `c` has Unicode `Numeric_Type` Decimal, Digit, or Numeric
/// (`Py_UNICODE_ISNUMERIC`). Backed by the 15.1.0-locked
/// [`NUMERIC_TYPE_RANGES`] table rather than Rust's `char::is_numeric`,
/// which tracks the `Nd|Nl|No` general categories against a newer
/// Unicode revision and so both misses ideographic numerals (`Lo`) and
/// includes post-15.1.0 codepoints.
fn is_numeric_char(c: char) -> bool {
    in_ranges(NUMERIC_TYPE_RANGES, c)
}

/// `true` if `c` is in Unicode general category `L*` (`str.isalpha`).
fn is_alpha_char(c: char) -> bool {
    in_ranges(ALPHA_RANGES, c)
}

/// `true` if `c` has the Unicode `Uppercase` property (`Py_UNICODE_ISUPPER`).
fn is_upper_char(c: char) -> bool {
    in_ranges(UPPER_RANGES, c)
}

/// `true` if `c` has the Unicode `Lowercase` property (`Py_UNICODE_ISLOWER`).
fn is_lower_char(c: char) -> bool {
    in_ranges(LOWER_RANGES, c)
}

/// `true` if `c` is a titlecase letter, category `Lt` (`Py_UNICODE_ISTITLE`).
fn is_title_char(c: char) -> bool {
    in_ranges(TITLE_RANGES, c)
}

/// `true` if `c` is whitespace per `str.isspace` (`Py_UNICODE_ISSPACE`).
fn is_space_char(c: char) -> bool {
    in_ranges(WHITESPACE_RANGES, c)
}

/// Classify each element by applying `pred` to the string content.
fn classify<D: Dimension>(
    a: &StringArray<D>,
    pred: impl Fn(&str) -> bool,
) -> FerrayResult<Array<bool, D>> {
    let data: Vec<bool> = a.iter().map(|s| pred(s)).collect();
    Array::from_vec(a.dim().clone(), data)
}

/// Return `true` where every character is alphabetic (Unicode general
/// category `L*`) and the string is non-empty. Matches
/// `numpy.strings.isalpha`, which delegates per element to CPython's
/// `str.isalpha()`. NARROWER than Rust's `char::is_alphabetic`: roman
/// numerals like `Ⅻ` (category `Nl`) and combining marks return `false`.
/// See [`is_alpha_char`]/[`ALPHA_RANGES`].
pub fn isalpha<D: Dimension>(a: &StringArray<D>) -> FerrayResult<Array<bool, D>> {
    classify(a, |s| !s.is_empty() && s.chars().all(is_alpha_char))
}

/// Return `true` where every character is a digit and the string is
/// non-empty. Matches `numpy.strings.isdigit`, which delegates per
/// element to CPython's `str.isdigit()` (`Py_UNICODE_ISDIGIT`):
/// Unicode `Numeric_Type` of Decimal **or** Digit. This is broader than
/// [`isdecimal`] (it accepts superscripts/subscripts like `²³`, circled
/// digits like `①`, and parenthesized digits) and narrower than
/// [`isnumeric`] (it rejects fractions like `½` and letter-numerals like
/// `Ⅻ`). See [`is_digit_char`]/[`DIGIT_RANGES`].
pub fn isdigit<D: Dimension>(a: &StringArray<D>) -> FerrayResult<Array<bool, D>> {
    classify(a, |s| !s.is_empty() && s.chars().all(is_digit_char))
}

/// Return `true` where every character is whitespace and the string is
/// non-empty. Matches `numpy.strings.isspace`, which delegates per
/// element to CPython's `str.isspace()` (`Py_UNICODE_ISSPACE`): the
/// Unicode `White_Space` property plus the bidi separators
/// `U+001C`..`U+001F`. See [`is_space_char`]/[`WHITESPACE_RANGES`].
pub fn isspace<D: Dimension>(a: &StringArray<D>) -> FerrayResult<Array<bool, D>> {
    classify(a, |s| !s.is_empty() && s.chars().all(is_space_char))
}

/// Return `true` where the string is uppercased and the string is
/// non-empty. Matches `numpy.strings.isupper` / CPython `str.isupper()`
/// (`do_isupper`): there is at least one cased character and no cased
/// character is lowercase or titlecase. Uncased characters (digits,
/// punctuation, roman numerals like `Ⅻ` which has the `Uppercase`
/// property) do not disqualify the string.
pub fn isupper<D: Dimension>(a: &StringArray<D>) -> FerrayResult<Array<bool, D>> {
    classify(a, |s| {
        let mut cased = false;
        for c in s.chars() {
            if is_lower_char(c) || is_title_char(c) {
                return false;
            }
            if is_upper_char(c) {
                cased = true;
            }
        }
        cased
    })
}

/// Return `true` where the string is lowercased and the string is
/// non-empty. Matches `numpy.strings.islower` / CPython `str.islower()`
/// (`do_islower`): at least one cased character and no cased character is
/// uppercase or titlecase.
pub fn islower<D: Dimension>(a: &StringArray<D>) -> FerrayResult<Array<bool, D>> {
    classify(a, |s| {
        let mut cased = false;
        for c in s.chars() {
            if is_upper_char(c) || is_title_char(c) {
                return false;
            }
            if is_lower_char(c) {
                cased = true;
            }
        }
        cased
    })
}

/// Return `true` where every character is alphanumeric and the string is
/// non-empty. Matches `numpy.strings.isalnum` / CPython `str.isalnum()`:
/// every character is alphabetic (`isalpha`) OR a decimal/digit/numeric
/// character (`isdecimal`/`isdigit`/`isnumeric`). Composes the corrected
/// predicates: `½` (numeric) and `②` (digit) are alphanumeric.
pub fn isalnum<D: Dimension>(a: &StringArray<D>) -> FerrayResult<Array<bool, D>> {
    classify(a, |s| {
        !s.is_empty()
            && s.chars().all(|c| {
                is_alpha_char(c) || is_decimal_char(c) || is_digit_char(c) || is_numeric_char(c)
            })
    })
}

/// Return `true` where every character is numeric and the string is
/// non-empty. Matches `numpy.strings.isnumeric`, which delegates per
/// element to CPython's `str.isnumeric()` (Unicode `Numeric_Type` =
/// Decimal, Digit, or Numeric): superscripts/fractions like `²` and `½`,
/// roman numerals like `Ⅻ`, and ideographic numerals like `一` are
/// numeric, while `.`, `+`, and `-` are not. Backed by the 15.1.0-locked
/// [`NUMERIC_TYPE_RANGES`] table via [`is_numeric_char`], NOT Rust's
/// `char::is_numeric` (which tests the `Nd|Nl|No` categories against a
/// newer Unicode revision: it misses the `Lo` ideographic numerals and
/// wrongly accepts codepoints assigned after Unicode 15.1.0).
pub fn isnumeric<D: Dimension>(a: &StringArray<D>) -> FerrayResult<Array<bool, D>> {
    classify(a, |s| !s.is_empty() && s.chars().all(is_numeric_char))
}

/// Return `true` where every character is a Unicode decimal digit
/// (`Numeric_Type` = Decimal) and the string is non-empty. Matches
/// `numpy.strings.isdecimal` / CPython `str.isdecimal()`
/// (`Py_UNICODE_ISDECIMAL`). Stricter than [`isdigit`]: superscripts
/// (`²`), circled digits (`①`) and fractions (`½`) return `false`, while
/// non-ASCII positional decimals such as fullwidth `０` and Arabic-Indic
/// `٣` return `true`. See [`is_decimal_char`]/[`DECIMAL_RANGES`].
pub fn isdecimal<D: Dimension>(a: &StringArray<D>) -> FerrayResult<Array<bool, D>> {
    classify(a, |s| !s.is_empty() && s.chars().all(is_decimal_char))
}

/// Return `true` where the string is titlecased and non-empty. Matches
/// `numpy.strings.istitle` / CPython `str.istitle()` (`do_istitle`):
/// uppercase/titlecase characters may only follow an uncased character
/// (they open a word), lowercase characters may only follow a cased
/// character, and at least one cased character is present. Titlecase
/// (`Lt`) characters such as `Dž` count as an opening (uppercase-position)
/// character, so `"Dž"` and `"Džx"` are titlecased.
pub fn istitle<D: Dimension>(a: &StringArray<D>) -> FerrayResult<Array<bool, D>> {
    classify(a, |s| {
        let mut cased = false;
        let mut previous_is_cased = false;
        for c in s.chars() {
            if is_upper_char(c) || is_title_char(c) {
                if previous_is_cased {
                    return false;
                }
                previous_is_cased = true;
                cased = true;
            } else if is_lower_char(c) {
                if !previous_is_cased {
                    return false;
                }
                cased = true;
            } else {
                previous_is_cased = false;
            }
        }
        cased
    })
}

#[cfg(test)]
mod tests {
    use super::*;
    use crate::string_array::array;

    #[test]
    fn test_isalpha() {
        let a = array(&["hello", "hello123", "", "HELLO"]).unwrap();
        let r = isalpha(&a).unwrap();
        assert_eq!(r.as_slice().unwrap(), &[true, false, false, true]);
    }

    #[test]
    fn test_isdigit() {
        let a = array(&["123", "12.3", "", "abc"]).unwrap();
        let r = isdigit(&a).unwrap();
        assert_eq!(r.as_slice().unwrap(), &[true, false, false, false]);
    }

    #[test]
    fn test_isspace() {
        let a = array(&["  ", "\t\n", "", "a b"]).unwrap();
        let r = isspace(&a).unwrap();
        assert_eq!(r.as_slice().unwrap(), &[true, true, false, false]);
    }

    #[test]
    fn test_isupper() {
        let a = array(&["HELLO", "Hello", "hello", ""]).unwrap();
        let r = isupper(&a).unwrap();
        assert_eq!(r.as_slice().unwrap(), &[true, false, false, false]);
    }

    #[test]
    fn test_islower() {
        let a = array(&["hello", "Hello", "HELLO", ""]).unwrap();
        let r = islower(&a).unwrap();
        assert_eq!(r.as_slice().unwrap(), &[true, false, false, false]);
    }

    #[test]
    fn test_isalnum() {
        let a = array(&["abc123", "abc 123", "", "abc"]).unwrap();
        let r = isalnum(&a).unwrap();
        assert_eq!(r.as_slice().unwrap(), &[true, false, false, true]);
    }

    #[test]
    fn test_istitle() {
        let a = array(&["Hello World", "hello world", "HELLO WORLD", ""]).unwrap();
        let r = istitle(&a).unwrap();
        assert_eq!(r.as_slice().unwrap(), &[true, false, false, false]);
    }

    // --- Divergence pins for crosslink #823/#911 (isnumeric/isdecimal/
    // isalnum/isdigit). Expected values are the live results of CPython
    // 3.13 / numpy 2.4 (Unicode 15.1.0) per-character `str.is*()`. ---

    #[test]
    fn test_isnumeric_python_semantics() {
        let a = array(&["12.3", "²", "½", "123", ""]).unwrap();
        let r = isnumeric(&a).unwrap();
        assert_eq!(r.as_slice().unwrap(), &[false, true, true, true, false]);
    }

    #[test]
    fn test_isdigit_python_semantics() {
        let a = array(&[
            "²³", "⁴", "₀", "①", "⓪", "⑴", "０９", "٣", "½", "Ⅻ", "123", "12a", "",
        ])
        .unwrap();
        let r = isdigit(&a).unwrap();
        assert_eq!(
            r.as_slice().unwrap(),
            &[
                true, true, true, true, true, true, true, true, false, false, true, false, false,
            ]
        );
    }

    #[test]
    fn test_isdigit_does_not_regress_isdecimal_isnumeric() {
        let a = array(&["²", "½", "①", "Ⅻ", "0123456789"]).unwrap();
        assert_eq!(
            isdecimal(&a).unwrap().as_slice().unwrap(),
            &[false, false, false, false, true]
        );
        assert_eq!(
            isnumeric(&a).unwrap().as_slice().unwrap(),
            &[true, true, true, true, true]
        );
        assert_eq!(
            isdigit(&a).unwrap().as_slice().unwrap(),
            &[true, false, true, false, true]
        );
    }

    // --- Divergence pins for crosslink #912 (refs #836). Expected values
    // are the live results of CPython 3.13 / numpy.char (Unicode 15.1.0),
    // confirmed against `numpy.char.is*` element by element. ---

    // isdecimal: `Numeric_Type=Decimal`. Non-ASCII positional decimals
    // (fullwidth '０', Arabic-Indic '٣') are True; superscript '²' and
    // fraction '½' are False; ASCII '123' True. The PRE-FIX impl used
    // `is_ascii_digit()` and returned False for '０'/'٣' (the divergence).
    #[test]
    fn test_isdecimal_python_semantics_912() {
        let a = array(&["０", "٣", "²", "½", "①", "123", "12.3", ""]).unwrap();
        let r = isdecimal(&a).unwrap();
        assert_eq!(
            r.as_slice().unwrap(),
            &[true, true, false, false, false, true, false, false]
        );
    }

    // isalpha: Unicode general category L* ONLY. Roman numeral 'Ⅻ' (Nl)
    // and circled 'Ⓐ' (So) are NOT alpha; accented 'café', CJK '日本' and
    // titlecase digraph 'Dž' (Lt) ARE. The PRE-FIX impl used
    // `char::is_alphabetic` (the Alphabetic property) and returned True
    // for 'Ⅻ' (the divergence).
    #[test]
    fn test_isalpha_python_semantics_912() {
        let a = array(&["café", "日本", "Dž", "Ⅻ", "Ⓐ", "abc123", "123", ""]).unwrap();
        let r = isalpha(&a).unwrap();
        assert_eq!(
            r.as_slice().unwrap(),
            &[true, true, true, false, false, false, false, false]
        );
    }

    // istitle: titlecase Lt char counts as an opening (uppercase-position)
    // character. 'Dž' and 'Džx' are titlecased; 'DžX' is not (two cased in a
    // row). 'Ⅻ' (Nl, Uppercase property) opens a title word -> True. The
    // PRE-FIX impl used `char::is_uppercase` and missed Lt (returned False
    // for 'Dž', the divergence).
    #[test]
    fn test_istitle_python_semantics_912() {
        let a = array(&["Hello", "Dž", "Džx", "DžX", "Ⅻ", "hELLO", "Hello World", ""]).unwrap();
        let r = istitle(&a).unwrap();
        assert_eq!(
            r.as_slice().unwrap(),
            &[true, true, true, false, true, false, true, false]
        );
    }

    // isupper/islower: Unicode Uppercase/Lowercase properties (incl.
    // Other_Uppercase/Lowercase). Roman 'Ⅻ' (Nl, Uppercase prop) is upper;
    // 'ⅻ' (lowercase prop) is lower; titlecase 'Dž' is neither. Digraph
    // 'dž' (Ll) is lower.
    #[test]
    fn test_isupper_islower_python_semantics_912() {
        let up = array(&["ABC", "Ⅻ", "DŽ", "aBc", "Dž", "A1", "123", ""]).unwrap();
        assert_eq!(
            isupper(&up).unwrap().as_slice().unwrap(),
            &[true, true, true, false, false, true, false, false]
        );
        let lo = array(&["abc", "ⅻ", "dž", "Hello", "Dž", "a1", "café", ""]).unwrap();
        assert_eq!(
            islower(&lo).unwrap().as_slice().unwrap(),
            &[true, true, true, false, false, true, true, false]
        );
    }

    // isalnum composes the corrected predicates: alpha OR decimal OR digit
    // OR numeric. '½' (numeric) and '②' (digit) are alnum; whitespace is
    // not. 'Ⅻ' is numeric -> alnum.
    #[test]
    fn test_isalnum_python_semantics_912() {
        let a = array(&["a1", "②", "½", "Ⅻ", "café", "ab c", "abc", ""]).unwrap();
        let r = isalnum(&a).unwrap();
        assert_eq!(
            r.as_slice().unwrap(),
            &[true, true, true, true, true, false, true, false]
        );
    }

    // isspace: Unicode White_Space + bidi separators U+001C..U+001F. The
    // non-ASCII spaces NBSP (U+00A0) and line/para separator (U+2028/9)
    // are space; ZWSP (U+200B) is NOT.
    #[test]
    fn test_isspace_python_semantics_912() {
        let a = array(&["\u{00A0}", "\u{2028}", "\u{1C}", "\u{200B}", " ", "a b", ""]).unwrap();
        let r = isspace(&a).unwrap();
        assert_eq!(
            r.as_slice().unwrap(),
            &[true, true, true, false, true, false, false]
        );
    }

    // --- Divergence pins for crosslink #1044/#1045 (isnumeric/isalnum
    // backed by a Unicode-15.1.0 `Numeric_Type` table instead of Rust's
    // `char::is_numeric`). Expected values are the live results of CPython
    // 3.13 / Unicode 15.1.0 `str.isnumeric()`/`str.isalnum()`. ---

    // isnumeric: CJK ideographic numerals have `Numeric_Type=Numeric` but
    // general category `Lo`, so Rust's `char::is_numeric` (`Nd|Nl|No`)
    // MISSES them. The oracle CPython 15.1.0 `chr(0x4E00).isnumeric()` is
    // True. '一' (U+4E00), '九' (U+4E5D), '〇' (U+3007, Nl) and roman 'Ⅻ'
    // (U+216B, Nl) are numeric; the letter 'A' is not.
    #[test]
    fn test_isnumeric_cjk_ideographic_numerals_1044() {
        let a = array(&["\u{4E00}", "\u{4E5D}", "\u{3007}", "\u{216B}", "A", ""]).unwrap();
        let r = isnumeric(&a).unwrap();
        assert_eq!(
            r.as_slice().unwrap(),
            &[true, true, true, true, false, false]
        );
    }

    // isnumeric: U+10D40 (Garay digit) was assigned AFTER Unicode 15.1.0,
    // so the oracle CPython 15.1.0 `chr(0x10D40).isnumeric()` is False,
    // but Rust's newer `char::is_numeric` tables return True (the
    // false-positive divergence). The 15.1.0-locked table must say False.
    #[test]
    fn test_isnumeric_post_15_1_codepoint_false_1044() {
        let a = array(&["\u{10D40}"]).unwrap();
        let r = isnumeric(&a).unwrap();
        assert_eq!(r.as_slice().unwrap(), &[false]);
    }

    // isalnum inherits the isnumeric fix: U+10D40 is no longer wrongly
    // numeric, so it is no longer wrongly alphanumeric. '一' (numeric) and
    // 'a' (alpha) ARE alnum; "a一" mixing alpha+numeric is alnum, but any
    // post-15.1.0 codepoint disqualifies the string.
    #[test]
    fn test_isalnum_drops_post_15_1_false_positive_1045() {
        let a = array(&["\u{10D40}", "\u{4E00}", "a", "a\u{10D40}"]).unwrap();
        let r = isalnum(&a).unwrap();
        assert_eq!(r.as_slice().unwrap(), &[false, true, true, false]);
    }

    // The NUMERIC_TYPE_RANGES table must DIVERGE from Rust's
    // `char::is_numeric` at both extremes of the divergence, proving it is
    // a real 15.1.0-locked table and not a delegation to std: an `Lo`
    // ideographic numeral std misses, and a post-15.1.0 codepoint std
    // wrongly includes.
    #[test]
    fn test_numeric_table_diverges_from_char_is_numeric() {
        // U+4E00 一: Numeric_Type=Numeric (Lo) -> table True, std False.
        assert!(is_numeric_char('\u{4E00}'));
        assert!(!'\u{4E00}'.is_numeric());
        // U+10D40: assigned after 15.1.0 -> table False, std True.
        assert!(!is_numeric_char('\u{10D40}'));
        assert!('\u{10D40}'.is_numeric());
    }

    // isalnum == isalpha || isdecimal || isdigit || isnumeric, per CPython
    // `str.isalnum`. Sample the composition across each contributing class:
    // alpha 'a', ideographic-numeric '一', fraction '½' (No/Numeric),
    // circled digit '②' (No/Digit), fullwidth decimal '０', and '.' (none).
    #[test]
    fn test_isalnum_composition_1045() {
        let samples = ["a", "\u{4E00}", "\u{00BD}", "\u{2461}", "\u{FF10}", "."];
        for s in samples {
            let c = s.chars().next().unwrap();
            let composed =
                is_alpha_char(c) || is_decimal_char(c) || is_digit_char(c) || is_numeric_char(c);
            let a = array(&[s]).unwrap();
            let got = isalnum(&a).unwrap();
            assert_eq!(got.as_slice().unwrap(), &[composed], "isalnum({s:?})");
        }
    }
}