31 constexpr char32 kOpeningBrackets[] = {
43 constexpr char32 kClosingBrackets[] = {
54 constexpr char32 kWhitespaces[] = {
65 constexpr char32 kBidirectional[] = {0x061C, 0x200E, 0x200F, 0x202A,
73 constexpr char32 kDecimalDigitRangesEnd[] = {
88 constexpr char32 kLatinLettersRangesStart[] = {0x0041, 0x0061, 0x00C0, 0x00D8,
92 constexpr char32 kLatinLettersRangesEnd[] = {0x005A, 0x007A, 0x00D6, 0x00F7,
98 constexpr char32 kArabicLettersRangesStart[] = {
102 constexpr char32 kArabicLettersRangesEnd[] = {
107 constexpr char32 kCyrillicLettersRangesStart[] = {0x0400, 0x1C80, 0x2DE0,
111 constexpr char32 kCyrillicLettersRangesEnd[] = {0x052F, 0x1C88, 0x2DFF,
116 constexpr char32 kChineseLettersRangesStart[] = {
121 constexpr char32 kChineseLettersRangesEnd[] = {
128 constexpr char32 kJapaneseLettersRangesStart[] = {0x3041, 0x30A1, 0x31F0,
132 constexpr char32 kJapaneseLettersRangesEnd[] = {0x3096, 0x30FA, 0x31FF, 0xFF9D};
138 constexpr char32 kKoreanLettersRangesStart[] = {0x3131, 0xFFA1};
141 constexpr char32 kKoreanLettersRangesEnd[] = {0x318E, 0xFFDC};
145 constexpr char32 kThaiLettersRangesStart[] = {0x0E01};
147 constexpr char32 kThaiLettersRangesEnd[] = {0x0E2E};
151 constexpr char32 kPunctuationRangesStart[] = {
176 constexpr char32 kPunctuationRangesEnd[] = {
207 constexpr char32 kUpperSingles[] = {
212 constexpr char32 kUpperRanges1Start[] = {
220 constexpr char32 kUpperRanges1End[] = {
228 constexpr char32 kUpperRanges2Start[] = {
235 constexpr char32 kUpperRanges2End[] = {
244 constexpr char32 kLowerSingles[] = {
250 constexpr char32 kLowerRanges1Start[] = {
264 constexpr char32 kLowerRanges1End[] = {
278 constexpr char32 kLowerRanges2Start[] = {
285 constexpr char32 kLowerRanges2End[] = {
379 constexpr char32 kPercentages[] = {0x0025, 0x066A, 0xFE6A, 0xFF05};
383 constexpr char32 kSlashes[] = {0x002f, 0x0337, 0x0338, 0x2044, 0x2215, 0xff0f};
387 constexpr char32 kMinuses[] = {0x002d, 0x02d7, 0x2212, 0xff0d};
391 constexpr char32 kNumberSign[] = {0x0023, 0xfe5f, 0xff03};
395 constexpr char32 kDots[] = {0x002e, 0xfe52, 0xff0e};
399 constexpr char32 kApostrophe[] = {0x0027, 0x02BC, 0x02EE, 0x055A,
404 constexpr char32 kQuotation[] = {
411 constexpr char32 kAmpersand[] = {0x0026, 0xFE60, 0xFF06, 0x1F674, 0x1F675};
460 int GetMatchIndex(const char32* array, int array_length, char32 c) {  in GetMatchIndex()
461   const char32* end = array + array_length;  in GetMatchIndex()
473 int GetOverlappingRangeIndex(const char32* arr, int arr_length,  in GetOverlappingRangeIndex()
474                              int range_length, char32 c) {  in GetOverlappingRangeIndex()
475   const char32* end = arr + arr_length;  in GetOverlappingRangeIndex()
481   const char32 range_end = *find_it;  in GetOverlappingRangeIndex()
482   const char32 range_start = range_end - (range_length - 1);  in GetOverlappingRangeIndex()
492 int GetOverlappingRangeIndex(const char32* start_arr, const char32* end_arr,  in GetOverlappingRangeIndex()
493                              int arr_length, int stride, char32 c) {  in GetOverlappingRangeIndex()
494   const char32* end_arr_end = end_arr + arr_length;  in GetOverlappingRangeIndex()
501   const char32 range_start = start_arr[range_index];  in GetOverlappingRangeIndex()
502   const char32 range_end = *find_it;  in GetOverlappingRangeIndex()
515 bool IsOpeningBracket(char32 codepoint) {  in IsOpeningBracket()
519 bool IsClosingBracket(char32 codepoint) {  in IsClosingBracket()
523 bool IsWhitespace(char32 codepoint) {  in IsWhitespace()
527 bool IsBidirectional(char32 codepoint) {  in IsBidirectional()
531 bool IsDigit(char32 codepoint) {  in IsDigit()
537 bool IsLower(char32 codepoint) {  in IsLower()
553 bool IsUpper(char32 codepoint) {  in IsUpper()
569 bool IsPunctuation(char32 codepoint) {  in IsPunctuation()
575 bool IsPercentage(char32 codepoint) {  in IsPercentage()
579 bool IsSlash(char32 codepoint) {  in IsSlash()
583 bool IsMinus(char32 codepoint) {  in IsMinus()
587 bool IsNumberSign(char32 codepoint) {  in IsNumberSign()
591 bool IsDot(char32 codepoint) {  in IsDot()
595 bool IsApostrophe(char32 codepoint) {  in IsApostrophe()
599 bool IsQuotation(char32 codepoint) {  in IsQuotation()
603 bool IsAmpersand(char32 codepoint) {  in IsAmpersand()
607 bool IsLatinLetter(char32 codepoint) {  in IsLatinLetter()
613 bool IsArabicLetter(char32 codepoint) {  in IsArabicLetter()
619 bool IsCyrillicLetter(char32 codepoint) {  in IsCyrillicLetter()
625 bool IsChineseLetter(char32 codepoint) {  in IsChineseLetter()
631 bool IsJapaneseLetter(char32 codepoint) {  in IsJapaneseLetter()
637 bool IsKoreanLetter(char32 codepoint) {  in IsKoreanLetter()
643 bool IsThaiLetter(char32 codepoint) {  in IsThaiLetter()
649 bool IsCJTletter(char32 codepoint) {  in IsCJTletter()
654 bool IsLetter(char32 codepoint) {  in IsLetter()
661 char32 ToLower(char32 codepoint) {  in ToLower()
681 char32 ToUpper(char32 codepoint) {  in ToUpper()
701 char32 GetPairedBracket(char32 codepoint) {  in GetPairedBracket()