clerk/internal/lsp/transliteration.go (view raw)
Oleksandr Smirnov
Oleksandr Smirnov
olexsmir@gmail.com lsp: latin to cyrillic transliteration, 1 month ago
olexsmir@gmail.com lsp: latin to cyrillic transliteration, 1 month ago
| 1 | package lsp |
| 2 | |
| 3 | import "strings" |
| 4 | |
| 5 | // latinToCyrillicTable maps latin input to Ukrainian Cyrillic. |
| 6 | // Ordered longest-first so greedy scanning matches "shch" as щ before it can match "sh"+"ch". |
| 7 | var latinToCyrillicTable = [...]struct{ latin, cyrillic string }{ |
| 8 | {"shch", "щ"}, |
| 9 | {"sch", "щ"}, |
| 10 | {"dzh", "дж"}, |
| 11 | {"zh", "ж"}, |
| 12 | {"kh", "х"}, |
| 13 | {"ts", "ц"}, |
| 14 | {"ch", "ч"}, |
| 15 | {"sh", "ш"}, |
| 16 | {"dz", "дз"}, |
| 17 | {"yu", "ю"}, |
| 18 | {"ya", "я"}, |
| 19 | {"ye", "є"}, |
| 20 | {"yi", "ї"}, |
| 21 | {"ji", "ї"}, |
| 22 | {"a", "а"}, |
| 23 | {"b", "б"}, |
| 24 | {"v", "в"}, |
| 25 | {"g", "г"}, |
| 26 | {"d", "д"}, |
| 27 | {"e", "е"}, |
| 28 | {"z", "з"}, |
| 29 | {"i", "і"}, |
| 30 | {"j", "й"}, |
| 31 | {"k", "к"}, |
| 32 | {"l", "л"}, |
| 33 | {"m", "м"}, |
| 34 | {"n", "н"}, |
| 35 | {"o", "о"}, |
| 36 | {"p", "п"}, |
| 37 | {"r", "р"}, |
| 38 | {"s", "с"}, |
| 39 | {"t", "т"}, |
| 40 | {"u", "у"}, |
| 41 | {"f", "ф"}, |
| 42 | {"h", "х"}, |
| 43 | {"c", "ц"}, |
| 44 | {"q", "к"}, |
| 45 | {"x", "кс"}, |
| 46 | {"w", "в"}, |
| 47 | {"y", "и"}, |
| 48 | } |
| 49 | |
| 50 | func latinToCyrillic(pattern string) string { |
| 51 | p := strings.ToLower(pattern) |
| 52 | var b strings.Builder |
| 53 | b.Grow(len(p)) |
| 54 | for i := 0; i < len(p); { |
| 55 | for _, kv := range latinToCyrillicTable { |
| 56 | if strings.HasPrefix(p[i:], kv.latin) { |
| 57 | b.WriteString(kv.cyrillic) |
| 58 | i += len(kv.latin) |
| 59 | goto next |
| 60 | } |
| 61 | } |
| 62 | b.WriteByte(p[i]) |
| 63 | i++ |
| 64 | next: |
| 65 | } |
| 66 | return b.String() |
| 67 | } |