ucs_width_table.h_shipped 32 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453
  1. /* SPDX-License-Identifier: GPL-2.0 */
  2. /*
  3. * ucs_width_table.h - Unicode character width
  4. *
  5. * Auto-generated by gen_ucs_width_table.py
  6. *
  7. * Unicode Version: 16.0.0
  8. */
  9. /* Zero-width character ranges (BMP - Basic Multilingual Plane, U+0000 to U+FFFF) */
  10. static const struct ucs_interval16 ucs_zero_width_bmp_ranges[] = {
  11. { 0x00AD, 0x00AD }, /* SOFT HYPHEN */
  12. { 0x0300, 0x036F }, /* COMBINING GRAVE ACCENT - COMBINING LATIN SMALL LETTER X */
  13. { 0x0483, 0x0489 }, /* COMBINING CYRILLIC TITLO - COMBINING CYRILLIC MILLIONS SIGN */
  14. { 0x0591, 0x05BD }, /* HEBREW ACCENT ETNAHTA - HEBREW POINT METEG */
  15. { 0x05BF, 0x05BF }, /* HEBREW POINT RAFE */
  16. { 0x05C1, 0x05C2 }, /* HEBREW POINT SHIN DOT - HEBREW POINT SIN DOT */
  17. { 0x05C4, 0x05C5 }, /* HEBREW MARK UPPER DOT - HEBREW MARK LOWER DOT */
  18. { 0x05C7, 0x05C7 }, /* HEBREW POINT QAMATS QATAN */
  19. { 0x0600, 0x0605 }, /* ARABIC NUMBER SIGN - ARABIC NUMBER MARK ABOVE */
  20. { 0x0610, 0x061A }, /* ARABIC SIGN SALLALLAHOU ALAYHE WASSALLAM - ARABIC SMALL KASRA */
  21. { 0x061C, 0x061C }, /* ARABIC LETTER MARK */
  22. { 0x064B, 0x065F }, /* ARABIC FATHATAN - ARABIC WAVY HAMZA BELOW */
  23. { 0x0670, 0x0670 }, /* ARABIC LETTER SUPERSCRIPT ALEF */
  24. { 0x06D6, 0x06DD }, /* ARABIC SMALL HIGH LIGATURE SAD WITH LAM WITH ALEF MAKSURA - ARABIC END OF AYAH */
  25. { 0x06DF, 0x06E4 }, /* ARABIC SMALL HIGH ROUNDED ZERO - ARABIC SMALL HIGH MADDA */
  26. { 0x06E7, 0x06E8 }, /* ARABIC SMALL HIGH YEH - ARABIC SMALL HIGH NOON */
  27. { 0x06EA, 0x06ED }, /* ARABIC EMPTY CENTRE LOW STOP - ARABIC SMALL LOW MEEM */
  28. { 0x070F, 0x070F }, /* SYRIAC ABBREVIATION MARK */
  29. { 0x0711, 0x0711 }, /* SYRIAC LETTER SUPERSCRIPT ALAPH */
  30. { 0x0730, 0x074A }, /* SYRIAC PTHAHA ABOVE - SYRIAC BARREKH */
  31. { 0x07A6, 0x07B0 }, /* THAANA ABAFILI - THAANA SUKUN */
  32. { 0x07EB, 0x07F3 }, /* NKO COMBINING SHORT HIGH TONE - NKO COMBINING DOUBLE DOT ABOVE */
  33. { 0x07FD, 0x07FD }, /* NKO DANTAYALAN */
  34. { 0x0816, 0x0819 }, /* SAMARITAN MARK IN - SAMARITAN MARK DAGESH */
  35. { 0x081B, 0x0823 }, /* SAMARITAN MARK EPENTHETIC YUT - SAMARITAN VOWEL SIGN A */
  36. { 0x0825, 0x0827 }, /* SAMARITAN VOWEL SIGN SHORT A - SAMARITAN VOWEL SIGN U */
  37. { 0x0829, 0x082D }, /* SAMARITAN VOWEL SIGN LONG I - SAMARITAN MARK NEQUDAA */
  38. { 0x0859, 0x085B }, /* MANDAIC AFFRICATION MARK - MANDAIC GEMINATION MARK */
  39. { 0x0890, 0x0891 }, /* ARABIC POUND MARK ABOVE - ARABIC PIASTRE MARK ABOVE */
  40. { 0x0897, 0x089F }, /* ARABIC PEPET - ARABIC HALF MADDA OVER MADDA */
  41. { 0x08CA, 0x0903 }, /* ARABIC SMALL HIGH FARSI YEH - DEVANAGARI SIGN VISARGA */
  42. { 0x093A, 0x093C }, /* DEVANAGARI VOWEL SIGN OE - DEVANAGARI SIGN NUKTA */
  43. { 0x093E, 0x094F }, /* DEVANAGARI VOWEL SIGN AA - DEVANAGARI VOWEL SIGN AW */
  44. { 0x0951, 0x0957 }, /* DEVANAGARI STRESS SIGN UDATTA - DEVANAGARI VOWEL SIGN UUE */
  45. { 0x0962, 0x0963 }, /* DEVANAGARI VOWEL SIGN VOCALIC L - DEVANAGARI VOWEL SIGN VOCALIC LL */
  46. { 0x0981, 0x0983 }, /* BENGALI SIGN CANDRABINDU - BENGALI SIGN VISARGA */
  47. { 0x09BC, 0x09BC }, /* BENGALI SIGN NUKTA */
  48. { 0x09BE, 0x09C4 }, /* BENGALI VOWEL SIGN AA - BENGALI VOWEL SIGN VOCALIC RR */
  49. { 0x09C7, 0x09C8 }, /* BENGALI VOWEL SIGN E - BENGALI VOWEL SIGN AI */
  50. { 0x09CB, 0x09CD }, /* BENGALI VOWEL SIGN O - BENGALI SIGN VIRAMA */
  51. { 0x09D7, 0x09D7 }, /* BENGALI AU LENGTH MARK */
  52. { 0x09E2, 0x09E3 }, /* BENGALI VOWEL SIGN VOCALIC L - BENGALI VOWEL SIGN VOCALIC LL */
  53. { 0x09FE, 0x09FE }, /* BENGALI SANDHI MARK */
  54. { 0x0A01, 0x0A03 }, /* GURMUKHI SIGN ADAK BINDI - GURMUKHI SIGN VISARGA */
  55. { 0x0A3C, 0x0A3C }, /* GURMUKHI SIGN NUKTA */
  56. { 0x0A3E, 0x0A42 }, /* GURMUKHI VOWEL SIGN AA - GURMUKHI VOWEL SIGN UU */
  57. { 0x0A47, 0x0A48 }, /* GURMUKHI VOWEL SIGN EE - GURMUKHI VOWEL SIGN AI */
  58. { 0x0A4B, 0x0A4D }, /* GURMUKHI VOWEL SIGN OO - GURMUKHI SIGN VIRAMA */
  59. { 0x0A51, 0x0A51 }, /* GURMUKHI SIGN UDAAT */
  60. { 0x0A70, 0x0A71 }, /* GURMUKHI TIPPI - GURMUKHI ADDAK */
  61. { 0x0A75, 0x0A75 }, /* GURMUKHI SIGN YAKASH */
  62. { 0x0A81, 0x0A83 }, /* GUJARATI SIGN CANDRABINDU - GUJARATI SIGN VISARGA */
  63. { 0x0ABC, 0x0ABC }, /* GUJARATI SIGN NUKTA */
  64. { 0x0ABE, 0x0AC5 }, /* GUJARATI VOWEL SIGN AA - GUJARATI VOWEL SIGN CANDRA E */
  65. { 0x0AC7, 0x0AC9 }, /* GUJARATI VOWEL SIGN E - GUJARATI VOWEL SIGN CANDRA O */
  66. { 0x0ACB, 0x0ACD }, /* GUJARATI VOWEL SIGN O - GUJARATI SIGN VIRAMA */
  67. { 0x0AE2, 0x0AE3 }, /* GUJARATI VOWEL SIGN VOCALIC L - GUJARATI VOWEL SIGN VOCALIC LL */
  68. { 0x0AFA, 0x0AFF }, /* GUJARATI SIGN SUKUN - GUJARATI SIGN TWO-CIRCLE NUKTA ABOVE */
  69. { 0x0B01, 0x0B03 }, /* ORIYA SIGN CANDRABINDU - ORIYA SIGN VISARGA */
  70. { 0x0B3C, 0x0B3C }, /* ORIYA SIGN NUKTA */
  71. { 0x0B3E, 0x0B44 }, /* ORIYA VOWEL SIGN AA - ORIYA VOWEL SIGN VOCALIC RR */
  72. { 0x0B47, 0x0B48 }, /* ORIYA VOWEL SIGN E - ORIYA VOWEL SIGN AI */
  73. { 0x0B4B, 0x0B4D }, /* ORIYA VOWEL SIGN O - ORIYA SIGN VIRAMA */
  74. { 0x0B55, 0x0B57 }, /* ORIYA SIGN OVERLINE - ORIYA AU LENGTH MARK */
  75. { 0x0B62, 0x0B63 }, /* ORIYA VOWEL SIGN VOCALIC L - ORIYA VOWEL SIGN VOCALIC LL */
  76. { 0x0B82, 0x0B82 }, /* TAMIL SIGN ANUSVARA */
  77. { 0x0BBE, 0x0BC2 }, /* TAMIL VOWEL SIGN AA - TAMIL VOWEL SIGN UU */
  78. { 0x0BC6, 0x0BC8 }, /* TAMIL VOWEL SIGN E - TAMIL VOWEL SIGN AI */
  79. { 0x0BCA, 0x0BCD }, /* TAMIL VOWEL SIGN O - TAMIL SIGN VIRAMA */
  80. { 0x0BD7, 0x0BD7 }, /* TAMIL AU LENGTH MARK */
  81. { 0x0C00, 0x0C04 }, /* TELUGU SIGN COMBINING CANDRABINDU ABOVE - TELUGU SIGN COMBINING ANUSVARA ABOVE */
  82. { 0x0C3C, 0x0C3C }, /* TELUGU SIGN NUKTA */
  83. { 0x0C3E, 0x0C44 }, /* TELUGU VOWEL SIGN AA - TELUGU VOWEL SIGN VOCALIC RR */
  84. { 0x0C46, 0x0C48 }, /* TELUGU VOWEL SIGN E - TELUGU VOWEL SIGN AI */
  85. { 0x0C4A, 0x0C4D }, /* TELUGU VOWEL SIGN O - TELUGU SIGN VIRAMA */
  86. { 0x0C55, 0x0C56 }, /* TELUGU LENGTH MARK - TELUGU AI LENGTH MARK */
  87. { 0x0C62, 0x0C63 }, /* TELUGU VOWEL SIGN VOCALIC L - TELUGU VOWEL SIGN VOCALIC LL */
  88. { 0x0C81, 0x0C83 }, /* KANNADA SIGN CANDRABINDU - KANNADA SIGN VISARGA */
  89. { 0x0CBC, 0x0CBC }, /* KANNADA SIGN NUKTA */
  90. { 0x0CBE, 0x0CC4 }, /* KANNADA VOWEL SIGN AA - KANNADA VOWEL SIGN VOCALIC RR */
  91. { 0x0CC6, 0x0CC8 }, /* KANNADA VOWEL SIGN E - KANNADA VOWEL SIGN AI */
  92. { 0x0CCA, 0x0CCD }, /* KANNADA VOWEL SIGN O - KANNADA SIGN VIRAMA */
  93. { 0x0CD5, 0x0CD6 }, /* KANNADA LENGTH MARK - KANNADA AI LENGTH MARK */
  94. { 0x0CE2, 0x0CE3 }, /* KANNADA VOWEL SIGN VOCALIC L - KANNADA VOWEL SIGN VOCALIC LL */
  95. { 0x0CF3, 0x0CF3 }, /* KANNADA SIGN COMBINING ANUSVARA ABOVE RIGHT */
  96. { 0x0D00, 0x0D03 }, /* MALAYALAM SIGN COMBINING ANUSVARA ABOVE - MALAYALAM SIGN VISARGA */
  97. { 0x0D3B, 0x0D3C }, /* MALAYALAM SIGN VERTICAL BAR VIRAMA - MALAYALAM SIGN CIRCULAR VIRAMA */
  98. { 0x0D3E, 0x0D44 }, /* MALAYALAM VOWEL SIGN AA - MALAYALAM VOWEL SIGN VOCALIC RR */
  99. { 0x0D46, 0x0D48 }, /* MALAYALAM VOWEL SIGN E - MALAYALAM VOWEL SIGN AI */
  100. { 0x0D4A, 0x0D4D }, /* MALAYALAM VOWEL SIGN O - MALAYALAM SIGN VIRAMA */
  101. { 0x0D57, 0x0D57 }, /* MALAYALAM AU LENGTH MARK */
  102. { 0x0D62, 0x0D63 }, /* MALAYALAM VOWEL SIGN VOCALIC L - MALAYALAM VOWEL SIGN VOCALIC LL */
  103. { 0x0D81, 0x0D83 }, /* SINHALA SIGN CANDRABINDU - SINHALA SIGN VISARGAYA */
  104. { 0x0DCA, 0x0DCA }, /* SINHALA SIGN AL-LAKUNA */
  105. { 0x0DCF, 0x0DD4 }, /* SINHALA VOWEL SIGN AELA-PILLA - SINHALA VOWEL SIGN KETTI PAA-PILLA */
  106. { 0x0DD6, 0x0DD6 }, /* SINHALA VOWEL SIGN DIGA PAA-PILLA */
  107. { 0x0DD8, 0x0DDF }, /* SINHALA VOWEL SIGN GAETTA-PILLA - SINHALA VOWEL SIGN GAYANUKITTA */
  108. { 0x0DF2, 0x0DF3 }, /* SINHALA VOWEL SIGN DIGA GAETTA-PILLA - SINHALA VOWEL SIGN DIGA GAYANUKITTA */
  109. { 0x0E31, 0x0E31 }, /* THAI CHARACTER MAI HAN-AKAT */
  110. { 0x0E34, 0x0E3A }, /* THAI CHARACTER SARA I - THAI CHARACTER PHINTHU */
  111. { 0x0E47, 0x0E4E }, /* THAI CHARACTER MAITAIKHU - THAI CHARACTER YAMAKKAN */
  112. { 0x0EB1, 0x0EB1 }, /* LAO VOWEL SIGN MAI KAN */
  113. { 0x0EB4, 0x0EBC }, /* LAO VOWEL SIGN I - LAO SEMIVOWEL SIGN LO */
  114. { 0x0EC8, 0x0ECE }, /* LAO TONE MAI EK - LAO YAMAKKAN */
  115. { 0x0F18, 0x0F19 }, /* TIBETAN ASTROLOGICAL SIGN -KHYUD PA - TIBETAN ASTROLOGICAL SIGN SDONG TSHUGS */
  116. { 0x0F35, 0x0F35 }, /* TIBETAN MARK NGAS BZUNG NYI ZLA */
  117. { 0x0F37, 0x0F37 }, /* TIBETAN MARK NGAS BZUNG SGOR RTAGS */
  118. { 0x0F39, 0x0F39 }, /* TIBETAN MARK TSA -PHRU */
  119. { 0x0F3E, 0x0F3F }, /* TIBETAN SIGN YAR TSHES - TIBETAN SIGN MAR TSHES */
  120. { 0x0F71, 0x0F84 }, /* TIBETAN VOWEL SIGN AA - TIBETAN MARK HALANTA */
  121. { 0x0F86, 0x0F87 }, /* TIBETAN SIGN LCI RTAGS - TIBETAN SIGN YANG RTAGS */
  122. { 0x0F8D, 0x0F97 }, /* TIBETAN SUBJOINED SIGN LCE TSA CAN - TIBETAN SUBJOINED LETTER JA */
  123. { 0x0F99, 0x0FBC }, /* TIBETAN SUBJOINED LETTER NYA - TIBETAN SUBJOINED LETTER FIXED-FORM RA */
  124. { 0x0FC6, 0x0FC6 }, /* TIBETAN SYMBOL PADMA GDAN */
  125. { 0x102B, 0x103E }, /* MYANMAR VOWEL SIGN TALL AA - MYANMAR CONSONANT SIGN MEDIAL HA */
  126. { 0x1056, 0x1059 }, /* MYANMAR VOWEL SIGN VOCALIC R - MYANMAR VOWEL SIGN VOCALIC LL */
  127. { 0x105E, 0x1060 }, /* MYANMAR CONSONANT SIGN MON MEDIAL NA - MYANMAR CONSONANT SIGN MON MEDIAL LA */
  128. { 0x1062, 0x1064 }, /* MYANMAR VOWEL SIGN SGAW KAREN EU - MYANMAR TONE MARK SGAW KAREN KE PHO */
  129. { 0x1067, 0x106D }, /* MYANMAR VOWEL SIGN WESTERN PWO KAREN EU - MYANMAR SIGN WESTERN PWO KAREN TONE-5 */
  130. { 0x1071, 0x1074 }, /* MYANMAR VOWEL SIGN GEBA KAREN I - MYANMAR VOWEL SIGN KAYAH EE */
  131. { 0x1082, 0x108D }, /* MYANMAR CONSONANT SIGN SHAN MEDIAL WA - MYANMAR SIGN SHAN COUNCIL EMPHATIC TONE */
  132. { 0x108F, 0x108F }, /* MYANMAR SIGN RUMAI PALAUNG TONE-5 */
  133. { 0x109A, 0x109D }, /* MYANMAR SIGN KHAMTI TONE-1 - MYANMAR VOWEL SIGN AITON AI */
  134. { 0x135D, 0x135F }, /* ETHIOPIC COMBINING GEMINATION AND VOWEL LENGTH MARK - ETHIOPIC COMBINING GEMINATION MARK */
  135. { 0x1712, 0x1715 }, /* TAGALOG VOWEL SIGN I - TAGALOG SIGN PAMUDPOD */
  136. { 0x1732, 0x1734 }, /* HANUNOO VOWEL SIGN I - HANUNOO SIGN PAMUDPOD */
  137. { 0x1752, 0x1753 }, /* BUHID VOWEL SIGN I - BUHID VOWEL SIGN U */
  138. { 0x1772, 0x1773 }, /* TAGBANWA VOWEL SIGN I - TAGBANWA VOWEL SIGN U */
  139. { 0x17B4, 0x17D3 }, /* KHMER VOWEL INHERENT AQ - KHMER SIGN BATHAMASAT */
  140. { 0x17DD, 0x17DD }, /* KHMER SIGN ATTHACAN */
  141. { 0x180B, 0x180F }, /* MONGOLIAN FREE VARIATION SELECTOR ONE - MONGOLIAN FREE VARIATION SELECTOR FOUR */
  142. { 0x1885, 0x1886 }, /* MONGOLIAN LETTER ALI GALI BALUDA - MONGOLIAN LETTER ALI GALI THREE BALUDA */
  143. { 0x18A9, 0x18A9 }, /* MONGOLIAN LETTER ALI GALI DAGALGA */
  144. { 0x1920, 0x192B }, /* LIMBU VOWEL SIGN A - LIMBU SUBJOINED LETTER WA */
  145. { 0x1930, 0x193B }, /* LIMBU SMALL LETTER KA - LIMBU SIGN SA-I */
  146. { 0x1A17, 0x1A1B }, /* BUGINESE VOWEL SIGN I - BUGINESE VOWEL SIGN AE */
  147. { 0x1A55, 0x1A5E }, /* TAI THAM CONSONANT SIGN MEDIAL RA - TAI THAM CONSONANT SIGN SA */
  148. { 0x1A60, 0x1A7C }, /* TAI THAM SIGN SAKOT - TAI THAM SIGN KHUEN-LUE KARAN */
  149. { 0x1A7F, 0x1A7F }, /* TAI THAM COMBINING CRYPTOGRAMMIC DOT */
  150. { 0x1AB0, 0x1ACE }, /* COMBINING DOUBLED CIRCUMFLEX ACCENT - COMBINING LATIN SMALL LETTER INSULAR T */
  151. { 0x1B00, 0x1B04 }, /* BALINESE SIGN ULU RICEM - BALINESE SIGN BISAH */
  152. { 0x1B34, 0x1B44 }, /* BALINESE SIGN REREKAN - BALINESE ADEG ADEG */
  153. { 0x1B6B, 0x1B73 }, /* BALINESE MUSICAL SYMBOL COMBINING TEGEH - BALINESE MUSICAL SYMBOL COMBINING GONG */
  154. { 0x1B80, 0x1B82 }, /* SUNDANESE SIGN PANYECEK - SUNDANESE SIGN PANGWISAD */
  155. { 0x1BA1, 0x1BAD }, /* SUNDANESE CONSONANT SIGN PAMINGKAL - SUNDANESE CONSONANT SIGN PASANGAN WA */
  156. { 0x1BE6, 0x1BF3 }, /* BATAK SIGN TOMPI - BATAK PANONGONAN */
  157. { 0x1C24, 0x1C37 }, /* LEPCHA SUBJOINED LETTER YA - LEPCHA SIGN NUKTA */
  158. { 0x1CD0, 0x1CD2 }, /* VEDIC TONE KARSHANA - VEDIC TONE PRENKHA */
  159. { 0x1CD4, 0x1CE8 }, /* VEDIC SIGN YAJURVEDIC MIDLINE SVARITA - VEDIC SIGN VISARGA ANUDATTA WITH TAIL */
  160. { 0x1CED, 0x1CED }, /* VEDIC SIGN TIRYAK */
  161. { 0x1CF4, 0x1CF4 }, /* VEDIC TONE CANDRA ABOVE */
  162. { 0x1CF7, 0x1CF9 }, /* VEDIC SIGN ATIKRAMA - VEDIC TONE DOUBLE RING ABOVE */
  163. { 0x1DC0, 0x1DFF }, /* COMBINING DOTTED GRAVE ACCENT - COMBINING RIGHT ARROWHEAD AND DOWN ARROWHEAD BELOW */
  164. { 0x200B, 0x200F }, /* ZERO WIDTH SPACE - RIGHT-TO-LEFT MARK */
  165. { 0x202A, 0x202E }, /* LEFT-TO-RIGHT EMBEDDING - RIGHT-TO-LEFT OVERRIDE */
  166. { 0x2060, 0x2064 }, /* WORD JOINER - INVISIBLE PLUS */
  167. { 0x2066, 0x206F }, /* LEFT-TO-RIGHT ISOLATE - NOMINAL DIGIT SHAPES */
  168. { 0x20D0, 0x20F0 }, /* COMBINING LEFT HARPOON ABOVE - COMBINING ASTERISK ABOVE */
  169. { 0x2640, 0x2640 }, /* FEMALE SIGN */
  170. { 0x2642, 0x2642 }, /* MALE SIGN */
  171. { 0x26A7, 0x26A7 }, /* MALE WITH STROKE AND MALE AND FEMALE SIGN */
  172. { 0x2CEF, 0x2CF1 }, /* COPTIC COMBINING NI ABOVE - COPTIC COMBINING SPIRITUS LENIS */
  173. { 0x2D7F, 0x2D7F }, /* TIFINAGH CONSONANT JOINER */
  174. { 0x2DE0, 0x2DFF }, /* COMBINING CYRILLIC LETTER BE - COMBINING CYRILLIC LETTER IOTIFIED BIG YUS */
  175. { 0x302A, 0x302F }, /* IDEOGRAPHIC LEVEL TONE MARK - HANGUL DOUBLE DOT TONE MARK */
  176. { 0x3099, 0x309A }, /* COMBINING KATAKANA-HIRAGANA VOICED SOUND MARK - COMBINING KATAKANA-HIRAGANA SEMI-VOICED SOUND MARK */
  177. { 0xA66F, 0xA672 }, /* COMBINING CYRILLIC VZMET - COMBINING CYRILLIC THOUSAND MILLIONS SIGN */
  178. { 0xA674, 0xA67D }, /* COMBINING CYRILLIC LETTER UKRAINIAN IE - COMBINING CYRILLIC PAYEROK */
  179. { 0xA69E, 0xA69F }, /* COMBINING CYRILLIC LETTER EF - COMBINING CYRILLIC LETTER IOTIFIED E */
  180. { 0xA6F0, 0xA6F1 }, /* BAMUM COMBINING MARK KOQNDON - BAMUM COMBINING MARK TUKWENTIS */
  181. { 0xA802, 0xA802 }, /* SYLOTI NAGRI SIGN DVISVARA */
  182. { 0xA806, 0xA806 }, /* SYLOTI NAGRI SIGN HASANTA */
  183. { 0xA80B, 0xA80B }, /* SYLOTI NAGRI SIGN ANUSVARA */
  184. { 0xA823, 0xA827 }, /* SYLOTI NAGRI VOWEL SIGN A - SYLOTI NAGRI VOWEL SIGN OO */
  185. { 0xA82C, 0xA82C }, /* SYLOTI NAGRI SIGN ALTERNATE HASANTA */
  186. { 0xA880, 0xA881 }, /* SAURASHTRA SIGN ANUSVARA - SAURASHTRA SIGN VISARGA */
  187. { 0xA8B4, 0xA8C5 }, /* SAURASHTRA CONSONANT SIGN HAARU - SAURASHTRA SIGN CANDRABINDU */
  188. { 0xA8E0, 0xA8F1 }, /* COMBINING DEVANAGARI DIGIT ZERO - COMBINING DEVANAGARI SIGN AVAGRAHA */
  189. { 0xA8FF, 0xA8FF }, /* DEVANAGARI VOWEL SIGN AY */
  190. { 0xA926, 0xA92D }, /* KAYAH LI VOWEL UE - KAYAH LI TONE CALYA PLOPHU */
  191. { 0xA947, 0xA953 }, /* REJANG VOWEL SIGN I - REJANG VIRAMA */
  192. { 0xA980, 0xA983 }, /* JAVANESE SIGN PANYANGGA - JAVANESE SIGN WIGNYAN */
  193. { 0xA9B3, 0xA9C0 }, /* JAVANESE SIGN CECAK TELU - JAVANESE PANGKON */
  194. { 0xA9E5, 0xA9E5 }, /* MYANMAR SIGN SHAN SAW */
  195. { 0xAA29, 0xAA36 }, /* CHAM VOWEL SIGN AA - CHAM CONSONANT SIGN WA */
  196. { 0xAA43, 0xAA43 }, /* CHAM CONSONANT SIGN FINAL NG */
  197. { 0xAA4C, 0xAA4D }, /* CHAM CONSONANT SIGN FINAL M - CHAM CONSONANT SIGN FINAL H */
  198. { 0xAA7B, 0xAA7D }, /* MYANMAR SIGN PAO KAREN TONE - MYANMAR SIGN TAI LAING TONE-5 */
  199. { 0xAAB0, 0xAAB0 }, /* TAI VIET MAI KANG */
  200. { 0xAAB2, 0xAAB4 }, /* TAI VIET VOWEL I - TAI VIET VOWEL U */
  201. { 0xAAB7, 0xAAB8 }, /* TAI VIET MAI KHIT - TAI VIET VOWEL IA */
  202. { 0xAABE, 0xAABF }, /* TAI VIET VOWEL AM - TAI VIET TONE MAI EK */
  203. { 0xAAC1, 0xAAC1 }, /* TAI VIET TONE MAI THO */
  204. { 0xAAEB, 0xAAEF }, /* MEETEI MAYEK VOWEL SIGN II - MEETEI MAYEK VOWEL SIGN AAU */
  205. { 0xAAF5, 0xAAF6 }, /* MEETEI MAYEK VOWEL SIGN VISARGA - MEETEI MAYEK VIRAMA */
  206. { 0xABE3, 0xABEA }, /* MEETEI MAYEK VOWEL SIGN ONAP - MEETEI MAYEK VOWEL SIGN NUNG */
  207. { 0xABEC, 0xABED }, /* MEETEI MAYEK LUM IYEK - MEETEI MAYEK APUN IYEK */
  208. { 0xFB1E, 0xFB1E }, /* HEBREW POINT JUDEO-SPANISH VARIKA */
  209. { 0xFE00, 0xFE0F }, /* VARIATION SELECTOR-1 - VARIATION SELECTOR-16 */
  210. { 0xFE20, 0xFE2F }, /* COMBINING LIGATURE LEFT HALF - COMBINING CYRILLIC TITLO RIGHT HALF */
  211. { 0xFEFF, 0xFEFF }, /* ZERO WIDTH NO-BREAK SPACE */
  212. { 0xFFF9, 0xFFFB }, /* INTERLINEAR ANNOTATION ANCHOR - INTERLINEAR ANNOTATION TERMINATOR */
  213. };
  214. /* Zero-width character ranges (non-BMP, U+10000 and above) */
  215. static const struct ucs_interval32 ucs_zero_width_non_bmp_ranges[] = {
  216. { 0x101FD, 0x101FD }, /* PHAISTOS DISC SIGN COMBINING OBLIQUE STROKE */
  217. { 0x102E0, 0x102E0 }, /* COPTIC EPACT THOUSANDS MARK */
  218. { 0x10376, 0x1037A }, /* COMBINING OLD PERMIC LETTER AN - COMBINING OLD PERMIC LETTER SII */
  219. { 0x10A01, 0x10A03 }, /* KHAROSHTHI VOWEL SIGN I - KHAROSHTHI VOWEL SIGN VOCALIC R */
  220. { 0x10A05, 0x10A06 }, /* KHAROSHTHI VOWEL SIGN E - KHAROSHTHI VOWEL SIGN O */
  221. { 0x10A0C, 0x10A0F }, /* KHAROSHTHI VOWEL LENGTH MARK - KHAROSHTHI SIGN VISARGA */
  222. { 0x10A38, 0x10A3A }, /* KHAROSHTHI SIGN BAR ABOVE - KHAROSHTHI SIGN DOT BELOW */
  223. { 0x10A3F, 0x10A3F }, /* KHAROSHTHI VIRAMA */
  224. { 0x10AE5, 0x10AE6 }, /* MANICHAEAN ABBREVIATION MARK ABOVE - MANICHAEAN ABBREVIATION MARK BELOW */
  225. { 0x10D24, 0x10D27 }, /* HANIFI ROHINGYA SIGN HARBAHAY - HANIFI ROHINGYA SIGN TASSI */
  226. { 0x10D69, 0x10D6D }, /* GARAY VOWEL SIGN E - GARAY CONSONANT NASALIZATION MARK */
  227. { 0x10EAB, 0x10EAC }, /* YEZIDI COMBINING HAMZA MARK - YEZIDI COMBINING MADDA MARK */
  228. { 0x10EFC, 0x10EFF }, /* ARABIC COMBINING ALEF OVERLAY - ARABIC SMALL LOW WORD MADDA */
  229. { 0x10F46, 0x10F50 }, /* SOGDIAN COMBINING DOT BELOW - SOGDIAN COMBINING STROKE BELOW */
  230. { 0x10F82, 0x10F85 }, /* OLD UYGHUR COMBINING DOT ABOVE - OLD UYGHUR COMBINING TWO DOTS BELOW */
  231. { 0x11000, 0x11002 }, /* BRAHMI SIGN CANDRABINDU - BRAHMI SIGN VISARGA */
  232. { 0x11038, 0x11046 }, /* BRAHMI VOWEL SIGN AA - BRAHMI VIRAMA */
  233. { 0x11070, 0x11070 }, /* BRAHMI SIGN OLD TAMIL VIRAMA */
  234. { 0x11073, 0x11074 }, /* BRAHMI VOWEL SIGN OLD TAMIL SHORT E - BRAHMI VOWEL SIGN OLD TAMIL SHORT O */
  235. { 0x1107F, 0x11082 }, /* BRAHMI NUMBER JOINER - KAITHI SIGN VISARGA */
  236. { 0x110B0, 0x110BA }, /* KAITHI VOWEL SIGN AA - KAITHI SIGN NUKTA */
  237. { 0x110BD, 0x110BD }, /* KAITHI NUMBER SIGN */
  238. { 0x110C2, 0x110C2 }, /* KAITHI VOWEL SIGN VOCALIC R */
  239. { 0x110CD, 0x110CD }, /* KAITHI NUMBER SIGN ABOVE */
  240. { 0x11100, 0x11102 }, /* CHAKMA SIGN CANDRABINDU - CHAKMA SIGN VISARGA */
  241. { 0x11127, 0x11134 }, /* CHAKMA VOWEL SIGN A - CHAKMA MAAYYAA */
  242. { 0x11145, 0x11146 }, /* CHAKMA VOWEL SIGN AA - CHAKMA VOWEL SIGN EI */
  243. { 0x11173, 0x11173 }, /* MAHAJANI SIGN NUKTA */
  244. { 0x11180, 0x11182 }, /* SHARADA SIGN CANDRABINDU - SHARADA SIGN VISARGA */
  245. { 0x111B3, 0x111C0 }, /* SHARADA VOWEL SIGN AA - SHARADA SIGN VIRAMA */
  246. { 0x111C9, 0x111CC }, /* SHARADA SANDHI MARK - SHARADA EXTRA SHORT VOWEL MARK */
  247. { 0x111CE, 0x111CF }, /* SHARADA VOWEL SIGN PRISHTHAMATRA E - SHARADA SIGN INVERTED CANDRABINDU */
  248. { 0x1122C, 0x11237 }, /* KHOJKI VOWEL SIGN AA - KHOJKI SIGN SHADDA */
  249. { 0x1123E, 0x1123E }, /* KHOJKI SIGN SUKUN */
  250. { 0x11241, 0x11241 }, /* KHOJKI VOWEL SIGN VOCALIC R */
  251. { 0x112DF, 0x112EA }, /* KHUDAWADI SIGN ANUSVARA - KHUDAWADI SIGN VIRAMA */
  252. { 0x11300, 0x11303 }, /* GRANTHA SIGN COMBINING ANUSVARA ABOVE - GRANTHA SIGN VISARGA */
  253. { 0x1133B, 0x1133C }, /* COMBINING BINDU BELOW - GRANTHA SIGN NUKTA */
  254. { 0x1133E, 0x11344 }, /* GRANTHA VOWEL SIGN AA - GRANTHA VOWEL SIGN VOCALIC RR */
  255. { 0x11347, 0x11348 }, /* GRANTHA VOWEL SIGN EE - GRANTHA VOWEL SIGN AI */
  256. { 0x1134B, 0x1134D }, /* GRANTHA VOWEL SIGN OO - GRANTHA SIGN VIRAMA */
  257. { 0x11357, 0x11357 }, /* GRANTHA AU LENGTH MARK */
  258. { 0x11362, 0x11363 }, /* GRANTHA VOWEL SIGN VOCALIC L - GRANTHA VOWEL SIGN VOCALIC LL */
  259. { 0x11366, 0x1136C }, /* COMBINING GRANTHA DIGIT ZERO - COMBINING GRANTHA DIGIT SIX */
  260. { 0x11370, 0x11374 }, /* COMBINING GRANTHA LETTER A - COMBINING GRANTHA LETTER PA */
  261. { 0x113B8, 0x113C0 }, /* TULU-TIGALARI VOWEL SIGN AA - TULU-TIGALARI VOWEL SIGN VOCALIC LL */
  262. { 0x113C2, 0x113C2 }, /* TULU-TIGALARI VOWEL SIGN EE */
  263. { 0x113C5, 0x113C5 }, /* TULU-TIGALARI VOWEL SIGN AI */
  264. { 0x113C7, 0x113CA }, /* TULU-TIGALARI VOWEL SIGN OO - TULU-TIGALARI SIGN CANDRA ANUNASIKA */
  265. { 0x113CC, 0x113D0 }, /* TULU-TIGALARI SIGN ANUSVARA - TULU-TIGALARI CONJOINER */
  266. { 0x113D2, 0x113D2 }, /* TULU-TIGALARI GEMINATION MARK */
  267. { 0x113E1, 0x113E2 }, /* TULU-TIGALARI VEDIC TONE SVARITA - TULU-TIGALARI VEDIC TONE ANUDATTA */
  268. { 0x11435, 0x11446 }, /* NEWA VOWEL SIGN AA - NEWA SIGN NUKTA */
  269. { 0x1145E, 0x1145E }, /* NEWA SANDHI MARK */
  270. { 0x114B0, 0x114C3 }, /* TIRHUTA VOWEL SIGN AA - TIRHUTA SIGN NUKTA */
  271. { 0x115AF, 0x115B5 }, /* SIDDHAM VOWEL SIGN AA - SIDDHAM VOWEL SIGN VOCALIC RR */
  272. { 0x115B8, 0x115C0 }, /* SIDDHAM VOWEL SIGN E - SIDDHAM SIGN NUKTA */
  273. { 0x115DC, 0x115DD }, /* SIDDHAM VOWEL SIGN ALTERNATE U - SIDDHAM VOWEL SIGN ALTERNATE UU */
  274. { 0x11630, 0x11640 }, /* MODI VOWEL SIGN AA - MODI SIGN ARDHACANDRA */
  275. { 0x116AB, 0x116B7 }, /* TAKRI SIGN ANUSVARA - TAKRI SIGN NUKTA */
  276. { 0x1171D, 0x1172B }, /* AHOM CONSONANT SIGN MEDIAL LA - AHOM SIGN KILLER */
  277. { 0x1182C, 0x1183A }, /* DOGRA VOWEL SIGN AA - DOGRA SIGN NUKTA */
  278. { 0x11930, 0x11935 }, /* DIVES AKURU VOWEL SIGN AA - DIVES AKURU VOWEL SIGN E */
  279. { 0x11937, 0x11938 }, /* DIVES AKURU VOWEL SIGN AI - DIVES AKURU VOWEL SIGN O */
  280. { 0x1193B, 0x1193E }, /* DIVES AKURU SIGN ANUSVARA - DIVES AKURU VIRAMA */
  281. { 0x11940, 0x11940 }, /* DIVES AKURU MEDIAL YA */
  282. { 0x11942, 0x11943 }, /* DIVES AKURU MEDIAL RA - DIVES AKURU SIGN NUKTA */
  283. { 0x119D1, 0x119D7 }, /* NANDINAGARI VOWEL SIGN AA - NANDINAGARI VOWEL SIGN VOCALIC RR */
  284. { 0x119DA, 0x119E0 }, /* NANDINAGARI VOWEL SIGN E - NANDINAGARI SIGN VIRAMA */
  285. { 0x119E4, 0x119E4 }, /* NANDINAGARI VOWEL SIGN PRISHTHAMATRA E */
  286. { 0x11A01, 0x11A0A }, /* ZANABAZAR SQUARE VOWEL SIGN I - ZANABAZAR SQUARE VOWEL LENGTH MARK */
  287. { 0x11A33, 0x11A39 }, /* ZANABAZAR SQUARE FINAL CONSONANT MARK - ZANABAZAR SQUARE SIGN VISARGA */
  288. { 0x11A3B, 0x11A3E }, /* ZANABAZAR SQUARE CLUSTER-FINAL LETTER YA - ZANABAZAR SQUARE CLUSTER-FINAL LETTER VA */
  289. { 0x11A47, 0x11A47 }, /* ZANABAZAR SQUARE SUBJOINER */
  290. { 0x11A51, 0x11A5B }, /* SOYOMBO VOWEL SIGN I - SOYOMBO VOWEL LENGTH MARK */
  291. { 0x11A8A, 0x11A99 }, /* SOYOMBO FINAL CONSONANT SIGN G - SOYOMBO SUBJOINER */
  292. { 0x11C2F, 0x11C36 }, /* BHAIKSUKI VOWEL SIGN AA - BHAIKSUKI VOWEL SIGN VOCALIC L */
  293. { 0x11C38, 0x11C3F }, /* BHAIKSUKI VOWEL SIGN E - BHAIKSUKI SIGN VIRAMA */
  294. { 0x11C92, 0x11CA7 }, /* MARCHEN SUBJOINED LETTER KA - MARCHEN SUBJOINED LETTER ZA */
  295. { 0x11CA9, 0x11CB6 }, /* MARCHEN SUBJOINED LETTER YA - MARCHEN SIGN CANDRABINDU */
  296. { 0x11D31, 0x11D36 }, /* MASARAM GONDI VOWEL SIGN AA - MASARAM GONDI VOWEL SIGN VOCALIC R */
  297. { 0x11D3A, 0x11D3A }, /* MASARAM GONDI VOWEL SIGN E */
  298. { 0x11D3C, 0x11D3D }, /* MASARAM GONDI VOWEL SIGN AI - MASARAM GONDI VOWEL SIGN O */
  299. { 0x11D3F, 0x11D45 }, /* MASARAM GONDI VOWEL SIGN AU - MASARAM GONDI VIRAMA */
  300. { 0x11D47, 0x11D47 }, /* MASARAM GONDI RA-KARA */
  301. { 0x11D8A, 0x11D8E }, /* GUNJALA GONDI VOWEL SIGN AA - GUNJALA GONDI VOWEL SIGN UU */
  302. { 0x11D90, 0x11D91 }, /* GUNJALA GONDI VOWEL SIGN EE - GUNJALA GONDI VOWEL SIGN AI */
  303. { 0x11D93, 0x11D97 }, /* GUNJALA GONDI VOWEL SIGN OO - GUNJALA GONDI VIRAMA */
  304. { 0x11EF3, 0x11EF6 }, /* MAKASAR VOWEL SIGN I - MAKASAR VOWEL SIGN O */
  305. { 0x11F00, 0x11F01 }, /* KAWI SIGN CANDRABINDU - KAWI SIGN ANUSVARA */
  306. { 0x11F03, 0x11F03 }, /* KAWI SIGN VISARGA */
  307. { 0x11F34, 0x11F3A }, /* KAWI VOWEL SIGN AA - KAWI VOWEL SIGN VOCALIC R */
  308. { 0x11F3E, 0x11F42 }, /* KAWI VOWEL SIGN E - KAWI CONJOINER */
  309. { 0x11F5A, 0x11F5A }, /* KAWI SIGN NUKTA */
  310. { 0x13430, 0x13440 }, /* EGYPTIAN HIEROGLYPH VERTICAL JOINER - EGYPTIAN HIEROGLYPH MIRROR HORIZONTALLY */
  311. { 0x13447, 0x13455 }, /* EGYPTIAN HIEROGLYPH MODIFIER DAMAGED AT TOP START - EGYPTIAN HIEROGLYPH MODIFIER DAMAGED */
  312. { 0x1611E, 0x1612F }, /* GURUNG KHEMA VOWEL SIGN AA - GURUNG KHEMA SIGN THOLHOMA */
  313. { 0x16AF0, 0x16AF4 }, /* BASSA VAH COMBINING HIGH TONE - BASSA VAH COMBINING HIGH-LOW TONE */
  314. { 0x16B30, 0x16B36 }, /* PAHAWH HMONG MARK CIM TUB - PAHAWH HMONG MARK CIM TAUM */
  315. { 0x16F4F, 0x16F4F }, /* MIAO SIGN CONSONANT MODIFIER BAR */
  316. { 0x16F51, 0x16F87 }, /* MIAO SIGN ASPIRATION - MIAO VOWEL SIGN UI */
  317. { 0x16F8F, 0x16F92 }, /* MIAO TONE RIGHT - MIAO TONE BELOW */
  318. { 0x16FE4, 0x16FE4 }, /* KHITAN SMALL SCRIPT FILLER */
  319. { 0x16FF0, 0x16FF1 }, /* VIETNAMESE ALTERNATE READING MARK CA - VIETNAMESE ALTERNATE READING MARK NHAY */
  320. { 0x1BC9D, 0x1BC9E }, /* DUPLOYAN THICK LETTER SELECTOR - DUPLOYAN DOUBLE MARK */
  321. { 0x1BCA0, 0x1BCA3 }, /* SHORTHAND FORMAT LETTER OVERLAP - SHORTHAND FORMAT UP STEP */
  322. { 0x1CF00, 0x1CF2D }, /* ZNAMENNY COMBINING MARK GORAZDO NIZKO S KRYZHEM ON LEFT - ZNAMENNY COMBINING MARK KRYZH ON LEFT */
  323. { 0x1CF30, 0x1CF46 }, /* ZNAMENNY COMBINING TONAL RANGE MARK MRACHNO - ZNAMENNY PRIZNAK MODIFIER ROG */
  324. { 0x1D165, 0x1D169 }, /* MUSICAL SYMBOL COMBINING STEM - MUSICAL SYMBOL COMBINING TREMOLO-3 */
  325. { 0x1D16D, 0x1D182 }, /* MUSICAL SYMBOL COMBINING AUGMENTATION DOT - MUSICAL SYMBOL COMBINING LOURE */
  326. { 0x1D185, 0x1D18B }, /* MUSICAL SYMBOL COMBINING DOIT - MUSICAL SYMBOL COMBINING TRIPLE TONGUE */
  327. { 0x1D1AA, 0x1D1AD }, /* MUSICAL SYMBOL COMBINING DOWN BOW - MUSICAL SYMBOL COMBINING SNAP PIZZICATO */
  328. { 0x1D242, 0x1D244 }, /* COMBINING GREEK MUSICAL TRISEME - COMBINING GREEK MUSICAL PENTASEME */
  329. { 0x1DA00, 0x1DA36 }, /* SIGNWRITING HEAD RIM - SIGNWRITING AIR SUCKING IN */
  330. { 0x1DA3B, 0x1DA6C }, /* SIGNWRITING MOUTH CLOSED NEUTRAL - SIGNWRITING EXCITEMENT */
  331. { 0x1DA75, 0x1DA75 }, /* SIGNWRITING UPPER BODY TILTING FROM HIP JOINTS */
  332. { 0x1DA84, 0x1DA84 }, /* SIGNWRITING LOCATION HEAD NECK */
  333. { 0x1DA9B, 0x1DA9F }, /* SIGNWRITING FILL MODIFIER-2 - SIGNWRITING FILL MODIFIER-6 */
  334. { 0x1DAA1, 0x1DAAF }, /* SIGNWRITING ROTATION MODIFIER-2 - SIGNWRITING ROTATION MODIFIER-16 */
  335. { 0x1E000, 0x1E006 }, /* COMBINING GLAGOLITIC LETTER AZU - COMBINING GLAGOLITIC LETTER ZHIVETE */
  336. { 0x1E008, 0x1E018 }, /* COMBINING GLAGOLITIC LETTER ZEMLJA - COMBINING GLAGOLITIC LETTER HERU */
  337. { 0x1E01B, 0x1E021 }, /* COMBINING GLAGOLITIC LETTER SHTA - COMBINING GLAGOLITIC LETTER YATI */
  338. { 0x1E023, 0x1E024 }, /* COMBINING GLAGOLITIC LETTER YU - COMBINING GLAGOLITIC LETTER SMALL YUS */
  339. { 0x1E026, 0x1E02A }, /* COMBINING GLAGOLITIC LETTER YO - COMBINING GLAGOLITIC LETTER FITA */
  340. { 0x1E08F, 0x1E08F }, /* COMBINING CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I */
  341. { 0x1E130, 0x1E136 }, /* NYIAKENG PUACHUE HMONG TONE-B - NYIAKENG PUACHUE HMONG TONE-D */
  342. { 0x1E2AE, 0x1E2AE }, /* TOTO SIGN RISING TONE */
  343. { 0x1E2EC, 0x1E2EF }, /* WANCHO TONE TUP - WANCHO TONE KOINI */
  344. { 0x1E4EC, 0x1E4EF }, /* NAG MUNDARI SIGN MUHOR - NAG MUNDARI SIGN SUTUH */
  345. { 0x1E5EE, 0x1E5EF }, /* OL ONAL SIGN MU - OL ONAL SIGN IKIR */
  346. { 0x1E8D0, 0x1E8D6 }, /* MENDE KIKAKUI COMBINING NUMBER TEENS - MENDE KIKAKUI COMBINING NUMBER MILLIONS */
  347. { 0x1E944, 0x1E94A }, /* ADLAM ALIF LENGTHENER - ADLAM NUKTA */
  348. { 0x1F3FB, 0x1F3FF }, /* EMOJI MODIFIER FITZPATRICK TYPE-1-2 - EMOJI MODIFIER FITZPATRICK TYPE-6 */
  349. { 0x1F9B0, 0x1F9B3 }, /* EMOJI COMPONENT RED HAIR - EMOJI COMPONENT WHITE HAIR */
  350. { 0xE0001, 0xE0001 }, /* LANGUAGE TAG */
  351. { 0xE0020, 0xE007F }, /* TAG SPACE - CANCEL TAG */
  352. { 0xE0100, 0xE01EF }, /* VARIATION SELECTOR-17 - VARIATION SELECTOR-256 */
  353. };
  354. /* Double-width character ranges (BMP - Basic Multilingual Plane, U+0000 to U+FFFF) */
  355. static const struct ucs_interval16 ucs_double_width_bmp_ranges[] = {
  356. { 0x1100, 0x115F }, /* HANGUL CHOSEONG KIYEOK - HANGUL CHOSEONG FILLER */
  357. { 0x231A, 0x231B }, /* WATCH - HOURGLASS */
  358. { 0x2329, 0x232A }, /* LEFT-POINTING ANGLE BRACKET - RIGHT-POINTING ANGLE BRACKET */
  359. { 0x23E9, 0x23EC }, /* BLACK RIGHT-POINTING DOUBLE TRIANGLE - BLACK DOWN-POINTING DOUBLE TRIANGLE */
  360. { 0x23F0, 0x23F0 }, /* ALARM CLOCK */
  361. { 0x23F3, 0x23F3 }, /* HOURGLASS WITH FLOWING SAND */
  362. { 0x25FD, 0x25FE }, /* WHITE MEDIUM SMALL SQUARE - BLACK MEDIUM SMALL SQUARE */
  363. { 0x2614, 0x2615 }, /* UMBRELLA WITH RAIN DROPS - HOT BEVERAGE */
  364. { 0x2630, 0x2637 }, /* TRIGRAM FOR HEAVEN - TRIGRAM FOR EARTH */
  365. { 0x2648, 0x2653 }, /* ARIES - PISCES */
  366. { 0x267F, 0x267F }, /* WHEELCHAIR SYMBOL */
  367. { 0x268A, 0x268F }, /* MONOGRAM FOR YANG - DIGRAM FOR GREATER YIN */
  368. { 0x2693, 0x2693 }, /* ANCHOR */
  369. { 0x26A1, 0x26A1 }, /* HIGH VOLTAGE SIGN */
  370. { 0x26AA, 0x26AB }, /* MEDIUM WHITE CIRCLE - MEDIUM BLACK CIRCLE */
  371. { 0x26BD, 0x26BE }, /* SOCCER BALL - BASEBALL */
  372. { 0x26C4, 0x26C5 }, /* SNOWMAN WITHOUT SNOW - SUN BEHIND CLOUD */
  373. { 0x26CE, 0x26CE }, /* OPHIUCHUS */
  374. { 0x26D4, 0x26D4 }, /* NO ENTRY */
  375. { 0x26EA, 0x26EA }, /* CHURCH */
  376. { 0x26F2, 0x26F3 }, /* FOUNTAIN - FLAG IN HOLE */
  377. { 0x26F5, 0x26F5 }, /* SAILBOAT */
  378. { 0x26FA, 0x26FA }, /* TENT */
  379. { 0x26FD, 0x26FD }, /* FUEL PUMP */
  380. { 0x2705, 0x2705 }, /* WHITE HEAVY CHECK MARK */
  381. { 0x270A, 0x270B }, /* RAISED FIST - RAISED HAND */
  382. { 0x2728, 0x2728 }, /* SPARKLES */
  383. { 0x274C, 0x274C }, /* CROSS MARK */
  384. { 0x274E, 0x274E }, /* NEGATIVE SQUARED CROSS MARK */
  385. { 0x2753, 0x2755 }, /* BLACK QUESTION MARK ORNAMENT - WHITE EXCLAMATION MARK ORNAMENT */
  386. { 0x2757, 0x2757 }, /* HEAVY EXCLAMATION MARK SYMBOL */
  387. { 0x2795, 0x2797 }, /* HEAVY PLUS SIGN - HEAVY DIVISION SIGN */
  388. { 0x27B0, 0x27B0 }, /* CURLY LOOP */
  389. { 0x27BF, 0x27BF }, /* DOUBLE CURLY LOOP */
  390. { 0x2B1B, 0x2B1C }, /* BLACK LARGE SQUARE - WHITE LARGE SQUARE */
  391. { 0x2B50, 0x2B50 }, /* WHITE MEDIUM STAR */
  392. { 0x2B55, 0x2B55 }, /* HEAVY LARGE CIRCLE */
  393. { 0x2E80, 0x2E99 }, /* CJK RADICAL REPEAT - CJK RADICAL RAP */
  394. { 0x2E9B, 0x2EF3 }, /* CJK RADICAL CHOKE - CJK RADICAL C-SIMPLIFIED TURTLE */
  395. { 0x2F00, 0x2FD5 }, /* KANGXI RADICAL ONE - KANGXI RADICAL FLUTE */
  396. { 0x2FF0, 0x3029 }, /* IDEOGRAPHIC DESCRIPTION CHARACTER LEFT TO RIGHT - HANGZHOU NUMERAL NINE */
  397. { 0x3030, 0x303E }, /* WAVY DASH - IDEOGRAPHIC VARIATION INDICATOR */
  398. { 0x3041, 0x3096 }, /* HIRAGANA LETTER SMALL A - HIRAGANA LETTER SMALL KE */
  399. { 0x309B, 0x30FF }, /* KATAKANA-HIRAGANA VOICED SOUND MARK - KATAKANA DIGRAPH KOTO */
  400. { 0x3105, 0x312F }, /* BOPOMOFO LETTER B - BOPOMOFO LETTER NN */
  401. { 0x3131, 0x318E }, /* HANGUL LETTER KIYEOK - HANGUL LETTER ARAEAE */
  402. { 0x3190, 0x31E5 }, /* IDEOGRAPHIC ANNOTATION LINKING MARK - CJK STROKE SZP */
  403. { 0x31EF, 0x321E }, /* IDEOGRAPHIC DESCRIPTION CHARACTER SUBTRACTION - PARENTHESIZED KOREAN CHARACTER O HU */
  404. { 0x3220, 0x3247 }, /* PARENTHESIZED IDEOGRAPH ONE - CIRCLED IDEOGRAPH KOTO */
  405. { 0x3250, 0xA48C }, /* PARTNERSHIP SIGN - YI SYLLABLE YYR */
  406. { 0xA490, 0xA4C6 }, /* YI RADICAL QOT - YI RADICAL KE */
  407. { 0xA960, 0xA97C }, /* HANGUL CHOSEONG TIKEUT-MIEUM - HANGUL CHOSEONG SSANGYEORINHIEUH */
  408. { 0xAC00, 0xD7A3 }, /* HANGUL SYLLABLE GA - HANGUL SYLLABLE HIH */
  409. { 0xF900, 0xFAFF }, /* U+F900 - U+FAFF */
  410. { 0xFE10, 0xFE19 }, /* PRESENTATION FORM FOR VERTICAL COMMA - PRESENTATION FORM FOR VERTICAL HORIZONTAL ELLIPSIS */
  411. { 0xFE30, 0xFE52 }, /* PRESENTATION FORM FOR VERTICAL TWO DOT LEADER - SMALL FULL STOP */
  412. { 0xFE54, 0xFE66 }, /* SMALL SEMICOLON - SMALL EQUALS SIGN */
  413. { 0xFE68, 0xFE6B }, /* SMALL REVERSE SOLIDUS - SMALL COMMERCIAL AT */
  414. { 0xFF01, 0xFF60 }, /* FULLWIDTH EXCLAMATION MARK - FULLWIDTH RIGHT WHITE PARENTHESIS */
  415. { 0xFFE0, 0xFFE6 }, /* FULLWIDTH CENT SIGN - FULLWIDTH WON SIGN */
  416. };
  417. /* Double-width character ranges (non-BMP, U+10000 and above) */
  418. static const struct ucs_interval32 ucs_double_width_non_bmp_ranges[] = {
  419. { 0x16FE0, 0x16FE3 }, /* TANGUT ITERATION MARK - OLD CHINESE ITERATION MARK */
  420. { 0x17000, 0x187F7 }, /* U+17000 - U+187F7 */
  421. { 0x18800, 0x18CD5 }, /* TANGUT COMPONENT-001 - KHITAN SMALL SCRIPT CHARACTER-18CD5 */
  422. { 0x18CFF, 0x18D08 }, /* U+18CFF - U+18D08 */
  423. { 0x1AFF0, 0x1AFF3 }, /* KATAKANA LETTER MINNAN TONE-2 - KATAKANA LETTER MINNAN TONE-5 */
  424. { 0x1AFF5, 0x1AFFB }, /* KATAKANA LETTER MINNAN TONE-7 - KATAKANA LETTER MINNAN NASALIZED TONE-5 */
  425. { 0x1AFFD, 0x1AFFE }, /* KATAKANA LETTER MINNAN NASALIZED TONE-7 - KATAKANA LETTER MINNAN NASALIZED TONE-8 */
  426. { 0x1B000, 0x1B122 }, /* KATAKANA LETTER ARCHAIC E - KATAKANA LETTER ARCHAIC WU */
  427. { 0x1B132, 0x1B132 }, /* HIRAGANA LETTER SMALL KO */
  428. { 0x1B150, 0x1B152 }, /* HIRAGANA LETTER SMALL WI - HIRAGANA LETTER SMALL WO */
  429. { 0x1B155, 0x1B155 }, /* KATAKANA LETTER SMALL KO */
  430. { 0x1B164, 0x1B167 }, /* KATAKANA LETTER SMALL WI - KATAKANA LETTER SMALL N */
  431. { 0x1B170, 0x1B2FB }, /* NUSHU CHARACTER-1B170 - NUSHU CHARACTER-1B2FB */
  432. { 0x1D300, 0x1D356 }, /* MONOGRAM FOR EARTH - TETRAGRAM FOR FOSTERING */
  433. { 0x1D360, 0x1D376 }, /* COUNTING ROD UNIT DIGIT ONE - IDEOGRAPHIC TALLY MARK FIVE */
  434. { 0x1F000, 0x1F02F }, /* U+1F000 - U+1F02F */
  435. { 0x1F0A0, 0x1F0FF }, /* U+1F0A0 - U+1F0FF */
  436. { 0x1F18E, 0x1F18E }, /* NEGATIVE SQUARED AB */
  437. { 0x1F191, 0x1F19A }, /* SQUARED CL - SQUARED VS */
  438. { 0x1F200, 0x1F202 }, /* SQUARE HIRAGANA HOKA - SQUARED KATAKANA SA */
  439. { 0x1F210, 0x1F23B }, /* SQUARED CJK UNIFIED IDEOGRAPH-624B - SQUARED CJK UNIFIED IDEOGRAPH-914D */
  440. { 0x1F240, 0x1F248 }, /* TORTOISE SHELL BRACKETED CJK UNIFIED IDEOGRAPH-672C - TORTOISE SHELL BRACKETED CJK UNIFIED IDEOGRAPH-6557 */
  441. { 0x1F250, 0x1F251 }, /* CIRCLED IDEOGRAPH ADVANTAGE - CIRCLED IDEOGRAPH ACCEPT */
  442. { 0x1F260, 0x1F265 }, /* ROUNDED SYMBOL FOR FU - ROUNDED SYMBOL FOR CAI */
  443. { 0x1F300, 0x1F3FA }, /* CYCLONE - AMPHORA */
  444. { 0x1F400, 0x1F64F }, /* RAT - PERSON WITH FOLDED HANDS */
  445. { 0x1F680, 0x1F9AF }, /* ROCKET - PROBING CANE */
  446. { 0x1F9B4, 0x1FAFF }, /* U+1F9B4 - U+1FAFF */
  447. { 0x20000, 0x2FFFD }, /* U+20000 - U+2FFFD */
  448. { 0x30000, 0x3FFFD }, /* U+30000 - U+3FFFD */
  449. };