271 lines
8.5 KiB
Python
271 lines
8.5 KiB
Python
from .util import UnicodeMixin
|
|
|
|
class Category:
|
|
LETTER: str
|
|
UPPERCASE_LETTER: str
|
|
LOWERCASE_LETTER: str
|
|
TITLECASE_LETTER: str
|
|
MODIFIER_LETTER: str
|
|
OTHER_LETTER: str
|
|
MARK: str
|
|
NON_SPACING_MARK: str
|
|
SPACING_COMBINING_MARK: str
|
|
ENCLOSING_MARK: str
|
|
NUMBER: str
|
|
DECIMAL_DIGIT_NUMBER: str
|
|
LETTER_NUMBER: str
|
|
OTHER_NUMBER: str
|
|
SYMBOL: str
|
|
MATH_SYMBOL: str
|
|
CURRENCY_SYMBOL: str
|
|
MODIFIER_SYMBOL: str
|
|
OTHER_SYMBOL: str
|
|
PUNCTUATION: str
|
|
CONNECTOR_PUNCTUATION: str
|
|
DASH_PUNCTUATION: str
|
|
OPEN_PUNCTUATION: str
|
|
CLOSE_PUNCTUATION: str
|
|
INITIAL_PUNCTUATION: str
|
|
FINAL_PUNCTUATION: str
|
|
OTHER_PUNCTUATION: str
|
|
SEPARATOR: str
|
|
SPACE_SEPARATOR: str
|
|
LINE_SEPARATOR: str
|
|
PARAGRAPH_SEPARATOR: str
|
|
OTHER: str
|
|
CONTROL: str
|
|
FORMAT: str
|
|
SURROGATE: str
|
|
PRIVATE_USE: str
|
|
NOT_ASSIGNED: str
|
|
@classmethod
|
|
def get(cls, uni_char: str) -> str: ...
|
|
|
|
def is_letter(uni_char: str) -> bool: ...
|
|
|
|
class _BlockRange(UnicodeMixin):
|
|
start: int
|
|
end: int
|
|
def __init__(self, start: int, end: int, regdict: dict[int, _BlockRange] | None = ...) -> None: ...
|
|
def __eq__(self, other: object) -> bool: ...
|
|
def __ne__(self, other: object) -> bool: ...
|
|
def __hash__(self) -> int: ...
|
|
def __unicode__(self) -> str: ...
|
|
|
|
class Block:
|
|
_RANGES: dict[int, _BlockRange]
|
|
_RANGE_KEYS: list[int] | None
|
|
BASIC_LATIN: _BlockRange
|
|
LATIN_1_SUPPLEMENT: _BlockRange
|
|
LATIN_EXTENDED_A: _BlockRange
|
|
LATIN_EXTENDED_B: _BlockRange
|
|
IPA_EXTENSIONS: _BlockRange
|
|
SPACING_MODIFIER_LETTERS: _BlockRange
|
|
COMBINING_DIACRITICAL_MARKS: _BlockRange
|
|
GREEK_AND_COPTIC: _BlockRange
|
|
CYRILLIC: _BlockRange
|
|
CYRILLIC_SUPPLEMENT: _BlockRange
|
|
ARMENIAN: _BlockRange
|
|
HEBREW: _BlockRange
|
|
ARABIC: _BlockRange
|
|
SYRIAC: _BlockRange
|
|
ARABIC_SUPPLEMENT: _BlockRange
|
|
THAANA: _BlockRange
|
|
NKO: _BlockRange
|
|
SAMARITAN: _BlockRange
|
|
MANDAIC: _BlockRange
|
|
DEVANAGARI: _BlockRange
|
|
BENGALI: _BlockRange
|
|
GURMUKHI: _BlockRange
|
|
GUJARATI: _BlockRange
|
|
ORIYA: _BlockRange
|
|
TAMIL: _BlockRange
|
|
TELUGU: _BlockRange
|
|
KANNADA: _BlockRange
|
|
MALAYALAM: _BlockRange
|
|
SINHALA: _BlockRange
|
|
THAI: _BlockRange
|
|
LAO: _BlockRange
|
|
TIBETAN: _BlockRange
|
|
MYANMAR: _BlockRange
|
|
GEORGIAN: _BlockRange
|
|
HANGUL_JAMO: _BlockRange
|
|
ETHIOPIC: _BlockRange
|
|
ETHIOPIC_SUPPLEMENT: _BlockRange
|
|
CHEROKEE: _BlockRange
|
|
UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS: _BlockRange
|
|
OGHAM: _BlockRange
|
|
RUNIC: _BlockRange
|
|
TAGALOG: _BlockRange
|
|
HANUNOO: _BlockRange
|
|
BUHID: _BlockRange
|
|
TAGBANWA: _BlockRange
|
|
KHMER: _BlockRange
|
|
MONGOLIAN: _BlockRange
|
|
UNIFIED_CANADIAN_ABORIGINAL_SYLLABICS_EXTENDED: _BlockRange
|
|
LIMBU: _BlockRange
|
|
TAI_LE: _BlockRange
|
|
NEW_TAI_LUE: _BlockRange
|
|
KHMER_SYMBOLS: _BlockRange
|
|
BUGINESE: _BlockRange
|
|
TAI_THAM: _BlockRange
|
|
BALINESE: _BlockRange
|
|
SUNDANESE: _BlockRange
|
|
BATAK: _BlockRange
|
|
LEPCHA: _BlockRange
|
|
OL_CHIKI: _BlockRange
|
|
VEDIC_EXTENSIONS: _BlockRange
|
|
PHONETIC_EXTENSIONS: _BlockRange
|
|
PHONETIC_EXTENSIONS_SUPPLEMENT: _BlockRange
|
|
COMBINING_DIACRITICAL_MARKS_SUPPLEMENT: _BlockRange
|
|
LATIN_EXTENDED_ADDITIONAL: _BlockRange
|
|
GREEK_EXTENDED: _BlockRange
|
|
GENERAL_PUNCTUATION: _BlockRange
|
|
SUPERSCRIPTS_AND_SUBSCRIPTS: _BlockRange
|
|
CURRENCY_SYMBOLS: _BlockRange
|
|
COMBINING_DIACRITICAL_MARKS_FOR_SYMBOLS: _BlockRange
|
|
LETTERLIKE_SYMBOLS: _BlockRange
|
|
NUMBER_FORMS: _BlockRange
|
|
ARROWS: _BlockRange
|
|
MATHEMATICAL_OPERATORS: _BlockRange
|
|
MISCELLANEOUS_TECHNICAL: _BlockRange
|
|
CONTROL_PICTURES: _BlockRange
|
|
OPTICAL_CHARACTER_RECOGNITION: _BlockRange
|
|
ENCLOSED_ALPHANUMERICS: _BlockRange
|
|
BOX_DRAWING: _BlockRange
|
|
BLOCK_ELEMENTS: _BlockRange
|
|
GEOMETRIC_SHAPES: _BlockRange
|
|
MISCELLANEOUS_SYMBOLS: _BlockRange
|
|
DINGBATS: _BlockRange
|
|
MISCELLANEOUS_MATHEMATICAL_SYMBOLS_A: _BlockRange
|
|
SUPPLEMENTAL_ARROWS_A: _BlockRange
|
|
BRAILLE_PATTERNS: _BlockRange
|
|
SUPPLEMENTAL_ARROWS_B: _BlockRange
|
|
MISCELLANEOUS_MATHEMATICAL_SYMBOLS_B: _BlockRange
|
|
SUPPLEMENTAL_MATHEMATICAL_OPERATORS: _BlockRange
|
|
MISCELLANEOUS_SYMBOLS_AND_ARROWS: _BlockRange
|
|
GLAGOLITIC: _BlockRange
|
|
LATIN_EXTENDED_C: _BlockRange
|
|
COPTIC: _BlockRange
|
|
GEORGIAN_SUPPLEMENT: _BlockRange
|
|
TIFINAGH: _BlockRange
|
|
ETHIOPIC_EXTENDED: _BlockRange
|
|
CYRILLIC_EXTENDED_A: _BlockRange
|
|
SUPPLEMENTAL_PUNCTUATION: _BlockRange
|
|
CJK_RADICALS_SUPPLEMENT: _BlockRange
|
|
KANGXI_RADICALS: _BlockRange
|
|
IDEOGRAPHIC_DESCRIPTION_CHARACTERS: _BlockRange
|
|
CJK_SYMBOLS_AND_PUNCTUATION: _BlockRange
|
|
HIRAGANA: _BlockRange
|
|
KATAKANA: _BlockRange
|
|
BOPOMOFO: _BlockRange
|
|
HANGUL_COMPATIBILITY_JAMO: _BlockRange
|
|
KANBUN: _BlockRange
|
|
BOPOMOFO_EXTENDED: _BlockRange
|
|
CJK_STROKES: _BlockRange
|
|
KATAKANA_PHONETIC_EXTENSIONS: _BlockRange
|
|
ENCLOSED_CJK_LETTERS_AND_MONTHS: _BlockRange
|
|
CJK_COMPATIBILITY: _BlockRange
|
|
CJK_UNIFIED_IDEOGRAPHS_EXTENSION_A: _BlockRange
|
|
YIJING_HEXAGRAM_SYMBOLS: _BlockRange
|
|
CJK_UNIFIED_IDEOGRAPHS: _BlockRange
|
|
YI_SYLLABLES: _BlockRange
|
|
YI_RADICALS: _BlockRange
|
|
LISU: _BlockRange
|
|
VAI: _BlockRange
|
|
CYRILLIC_EXTENDED_B: _BlockRange
|
|
BAMUM: _BlockRange
|
|
MODIFIER_TONE_LETTERS: _BlockRange
|
|
LATIN_EXTENDED_D: _BlockRange
|
|
SYLOTI_NAGRI: _BlockRange
|
|
COMMON_INDIC_NUMBER_FORMS: _BlockRange
|
|
PHAGS_PA: _BlockRange
|
|
SAURASHTRA: _BlockRange
|
|
DEVANAGARI_EXTENDED: _BlockRange
|
|
KAYAH_LI: _BlockRange
|
|
REJANG: _BlockRange
|
|
HANGUL_JAMO_EXTENDED_A: _BlockRange
|
|
JAVANESE: _BlockRange
|
|
CHAM: _BlockRange
|
|
MYANMAR_EXTENDED_A: _BlockRange
|
|
TAI_VIET: _BlockRange
|
|
ETHIOPIC_EXTENDED_A: _BlockRange
|
|
MEETEI_MAYEK: _BlockRange
|
|
HANGUL_SYLLABLES: _BlockRange
|
|
HANGUL_JAMO_EXTENDED_B: _BlockRange
|
|
HIGH_SURROGATES: _BlockRange
|
|
HIGH_PRIVATE_USE_SURROGATES: _BlockRange
|
|
LOW_SURROGATES: _BlockRange
|
|
PRIVATE_USE_AREA: _BlockRange
|
|
CJK_COMPATIBILITY_IDEOGRAPHS: _BlockRange
|
|
ALPHABETIC_PRESENTATION_FORMS: _BlockRange
|
|
ARABIC_PRESENTATION_FORMS_A: _BlockRange
|
|
VARIATION_SELECTORS: _BlockRange
|
|
VERTICAL_FORMS: _BlockRange
|
|
COMBINING_HALF_MARKS: _BlockRange
|
|
CJK_COMPATIBILITY_FORMS: _BlockRange
|
|
SMALL_FORM_VARIANTS: _BlockRange
|
|
ARABIC_PRESENTATION_FORMS_B: _BlockRange
|
|
HALFWIDTH_AND_FULLWIDTH_FORMS: _BlockRange
|
|
SPECIALS: _BlockRange
|
|
LINEAR_B_SYLLABARY: _BlockRange
|
|
LINEAR_B_IDEOGRAMS: _BlockRange
|
|
AEGEAN_NUMBERS: _BlockRange
|
|
ANCIENT_GREEK_NUMBERS: _BlockRange
|
|
ANCIENT_SYMBOLS: _BlockRange
|
|
PHAISTOS_DISC: _BlockRange
|
|
LYCIAN: _BlockRange
|
|
CARIAN: _BlockRange
|
|
OLD_ITALIC: _BlockRange
|
|
GOTHIC: _BlockRange
|
|
UGARITIC: _BlockRange
|
|
OLD_PERSIAN: _BlockRange
|
|
DESERET: _BlockRange
|
|
SHAVIAN: _BlockRange
|
|
OSMANYA: _BlockRange
|
|
CYPRIOT_SYLLABARY: _BlockRange
|
|
IMPERIAL_ARAMAIC: _BlockRange
|
|
PHOENICIAN: _BlockRange
|
|
LYDIAN: _BlockRange
|
|
KHAROSHTHI: _BlockRange
|
|
OLD_SOUTH_ARABIAN: _BlockRange
|
|
AVESTAN: _BlockRange
|
|
INSCRIPTIONAL_PARTHIAN: _BlockRange
|
|
INSCRIPTIONAL_PAHLAVI: _BlockRange
|
|
OLD_TURKIC: _BlockRange
|
|
RUMI_NUMERAL_SYMBOLS: _BlockRange
|
|
BRAHMI: _BlockRange
|
|
KAITHI: _BlockRange
|
|
CUNEIFORM: _BlockRange
|
|
CUNEIFORM_NUMBERS_AND_PUNCTUATION: _BlockRange
|
|
EGYPTIAN_HIEROGLYPHS: _BlockRange
|
|
BAMUM_SUPPLEMENT: _BlockRange
|
|
KANA_SUPPLEMENT: _BlockRange
|
|
BYZANTINE_MUSICAL_SYMBOLS: _BlockRange
|
|
MUSICAL_SYMBOLS: _BlockRange
|
|
ANCIENT_GREEK_MUSICAL_NOTATION: _BlockRange
|
|
TAI_XUAN_JING_SYMBOLS: _BlockRange
|
|
COUNTING_ROD_NUMERALS: _BlockRange
|
|
MATHEMATICAL_ALPHANUMERIC_SYMBOLS: _BlockRange
|
|
MAHJONG_TILES: _BlockRange
|
|
DOMINO_TILES: _BlockRange
|
|
PLAYING_CARDS: _BlockRange
|
|
ENCLOSED_ALPHANUMERIC_SUPPLEMENT: _BlockRange
|
|
ENCLOSED_IDEOGRAPHIC_SUPPLEMENT: _BlockRange
|
|
MISCELLANEOUS_SYMBOLS_AND_PICTOGRAPHS: _BlockRange
|
|
EMOTICONS: _BlockRange
|
|
TRANSPORT_AND_MAP_SYMBOLS: _BlockRange
|
|
ALCHEMICAL_SYMBOLS: _BlockRange
|
|
CJK_UNIFIED_IDEOGRAPHS_EXTENSION_B: _BlockRange
|
|
CJK_UNIFIED_IDEOGRAPHS_EXTENSION_C: _BlockRange
|
|
CJK_UNIFIED_IDEOGRAPHS_EXTENSION_D: _BlockRange
|
|
CJK_COMPATIBILITY_IDEOGRAPHS_SUPPLEMENT: _BlockRange
|
|
TAGS: _BlockRange
|
|
VARIATION_SELECTORS_SUPPLEMENT: _BlockRange
|
|
SUPPLEMENTARY_PRIVATE_USE_AREA_A: _BlockRange
|
|
SUPPLEMENTARY_PRIVATE_USE_AREA_B: _BlockRange
|
|
UNKNOWN: _BlockRange
|
|
@classmethod
|
|
def get(cls, uni_char: str) -> _BlockRange: ...
|
|
|
|
def digit(uni_char: str, default_value: int | None = ...) -> int: ...
|