Anand S Katti | 0954135 | 2020-01-29 15:54:01 +0530 | [diff] [blame] | 1 | package runewidth |
| 2 | |
| 3 | import ( |
| 4 | "os" |
| 5 | ) |
| 6 | |
| 7 | //go:generate go run script/generate.go |
| 8 | |
| 9 | var ( |
| 10 | // EastAsianWidth will be set true if the current locale is CJK |
| 11 | EastAsianWidth bool |
| 12 | |
| 13 | // ZeroWidthJoiner is flag to set to use UTR#51 ZWJ |
| 14 | ZeroWidthJoiner bool |
| 15 | |
| 16 | // DefaultCondition is a condition in current locale |
| 17 | DefaultCondition = &Condition{} |
| 18 | ) |
| 19 | |
| 20 | func init() { |
| 21 | handleEnv() |
| 22 | } |
| 23 | |
| 24 | func handleEnv() { |
| 25 | env := os.Getenv("RUNEWIDTH_EASTASIAN") |
| 26 | if env == "" { |
| 27 | EastAsianWidth = IsEastAsian() |
| 28 | } else { |
| 29 | EastAsianWidth = env == "1" |
| 30 | } |
| 31 | // update DefaultCondition |
| 32 | DefaultCondition.EastAsianWidth = EastAsianWidth |
| 33 | DefaultCondition.ZeroWidthJoiner = ZeroWidthJoiner |
| 34 | } |
| 35 | |
| 36 | type interval struct { |
| 37 | first rune |
| 38 | last rune |
| 39 | } |
| 40 | |
| 41 | type table []interval |
| 42 | |
| 43 | func inTables(r rune, ts ...table) bool { |
| 44 | for _, t := range ts { |
| 45 | if inTable(r, t) { |
| 46 | return true |
| 47 | } |
| 48 | } |
| 49 | return false |
| 50 | } |
| 51 | |
| 52 | func inTable(r rune, t table) bool { |
| 53 | // func (t table) IncludesRune(r rune) bool { |
| 54 | if r < t[0].first { |
| 55 | return false |
| 56 | } |
| 57 | |
| 58 | bot := 0 |
| 59 | top := len(t) - 1 |
| 60 | for top >= bot { |
| 61 | mid := (bot + top) >> 1 |
| 62 | |
| 63 | switch { |
| 64 | case t[mid].last < r: |
| 65 | bot = mid + 1 |
| 66 | case t[mid].first > r: |
| 67 | top = mid - 1 |
| 68 | default: |
| 69 | return true |
| 70 | } |
| 71 | } |
| 72 | |
| 73 | return false |
| 74 | } |
| 75 | |
| 76 | var private = table{ |
| 77 | {0x00E000, 0x00F8FF}, {0x0F0000, 0x0FFFFD}, {0x100000, 0x10FFFD}, |
| 78 | } |
| 79 | |
| 80 | var nonprint = table{ |
| 81 | {0x0000, 0x001F}, {0x007F, 0x009F}, {0x00AD, 0x00AD}, |
| 82 | {0x070F, 0x070F}, {0x180B, 0x180E}, {0x200B, 0x200F}, |
| 83 | {0x2028, 0x202E}, {0x206A, 0x206F}, {0xD800, 0xDFFF}, |
| 84 | {0xFEFF, 0xFEFF}, {0xFFF9, 0xFFFB}, {0xFFFE, 0xFFFF}, |
| 85 | } |
| 86 | |
| 87 | // Condition have flag EastAsianWidth whether the current locale is CJK or not. |
| 88 | type Condition struct { |
| 89 | EastAsianWidth bool |
| 90 | ZeroWidthJoiner bool |
| 91 | } |
| 92 | |
| 93 | // NewCondition return new instance of Condition which is current locale. |
| 94 | func NewCondition() *Condition { |
| 95 | return &Condition{ |
| 96 | EastAsianWidth: EastAsianWidth, |
| 97 | ZeroWidthJoiner: ZeroWidthJoiner, |
| 98 | } |
| 99 | } |
| 100 | |
| 101 | // RuneWidth returns the number of cells in r. |
| 102 | // See http://www.unicode.org/reports/tr11/ |
| 103 | func (c *Condition) RuneWidth(r rune) int { |
| 104 | switch { |
| 105 | case r < 0 || r > 0x10FFFF || inTables(r, nonprint, combining, notassigned): |
| 106 | return 0 |
| 107 | case (c.EastAsianWidth && IsAmbiguousWidth(r)) || inTables(r, doublewidth): |
| 108 | return 2 |
| 109 | default: |
| 110 | return 1 |
| 111 | } |
| 112 | } |
| 113 | |
| 114 | func (c *Condition) stringWidth(s string) (width int) { |
| 115 | for _, r := range []rune(s) { |
| 116 | width += c.RuneWidth(r) |
| 117 | } |
| 118 | return width |
| 119 | } |
| 120 | |
| 121 | func (c *Condition) stringWidthZeroJoiner(s string) (width int) { |
| 122 | r1, r2 := rune(0), rune(0) |
| 123 | for _, r := range []rune(s) { |
| 124 | if r == 0xFE0E || r == 0xFE0F { |
| 125 | continue |
| 126 | } |
| 127 | w := c.RuneWidth(r) |
| 128 | if r2 == 0x200D && inTables(r, emoji) && inTables(r1, emoji) { |
| 129 | if width < w { |
| 130 | width = w |
| 131 | } |
| 132 | } else { |
| 133 | width += w |
| 134 | } |
| 135 | r1, r2 = r2, r |
| 136 | } |
| 137 | return width |
| 138 | } |
| 139 | |
| 140 | // StringWidth return width as you can see |
| 141 | func (c *Condition) StringWidth(s string) (width int) { |
| 142 | if c.ZeroWidthJoiner { |
| 143 | return c.stringWidthZeroJoiner(s) |
| 144 | } |
| 145 | return c.stringWidth(s) |
| 146 | } |
| 147 | |
| 148 | // Truncate return string truncated with w cells |
| 149 | func (c *Condition) Truncate(s string, w int, tail string) string { |
| 150 | if c.StringWidth(s) <= w { |
| 151 | return s |
| 152 | } |
| 153 | r := []rune(s) |
| 154 | tw := c.StringWidth(tail) |
| 155 | w -= tw |
| 156 | width := 0 |
| 157 | i := 0 |
| 158 | for ; i < len(r); i++ { |
| 159 | cw := c.RuneWidth(r[i]) |
| 160 | if width+cw > w { |
| 161 | break |
| 162 | } |
| 163 | width += cw |
| 164 | } |
| 165 | return string(r[0:i]) + tail |
| 166 | } |
| 167 | |
| 168 | // Wrap return string wrapped with w cells |
| 169 | func (c *Condition) Wrap(s string, w int) string { |
| 170 | width := 0 |
| 171 | out := "" |
| 172 | for _, r := range []rune(s) { |
| 173 | cw := RuneWidth(r) |
| 174 | if r == '\n' { |
| 175 | out += string(r) |
| 176 | width = 0 |
| 177 | continue |
| 178 | } else if width+cw > w { |
| 179 | out += "\n" |
| 180 | width = 0 |
| 181 | out += string(r) |
| 182 | width += cw |
| 183 | continue |
| 184 | } |
| 185 | out += string(r) |
| 186 | width += cw |
| 187 | } |
| 188 | return out |
| 189 | } |
| 190 | |
| 191 | // FillLeft return string filled in left by spaces in w cells |
| 192 | func (c *Condition) FillLeft(s string, w int) string { |
| 193 | width := c.StringWidth(s) |
| 194 | count := w - width |
| 195 | if count > 0 { |
| 196 | b := make([]byte, count) |
| 197 | for i := range b { |
| 198 | b[i] = ' ' |
| 199 | } |
| 200 | return string(b) + s |
| 201 | } |
| 202 | return s |
| 203 | } |
| 204 | |
| 205 | // FillRight return string filled in left by spaces in w cells |
| 206 | func (c *Condition) FillRight(s string, w int) string { |
| 207 | width := c.StringWidth(s) |
| 208 | count := w - width |
| 209 | if count > 0 { |
| 210 | b := make([]byte, count) |
| 211 | for i := range b { |
| 212 | b[i] = ' ' |
| 213 | } |
| 214 | return s + string(b) |
| 215 | } |
| 216 | return s |
| 217 | } |
| 218 | |
| 219 | // RuneWidth returns the number of cells in r. |
| 220 | // See http://www.unicode.org/reports/tr11/ |
| 221 | func RuneWidth(r rune) int { |
| 222 | return DefaultCondition.RuneWidth(r) |
| 223 | } |
| 224 | |
| 225 | // IsAmbiguousWidth returns whether is ambiguous width or not. |
| 226 | func IsAmbiguousWidth(r rune) bool { |
| 227 | return inTables(r, private, ambiguous) |
| 228 | } |
| 229 | |
| 230 | // IsNeutralWidth returns whether is neutral width or not. |
| 231 | func IsNeutralWidth(r rune) bool { |
| 232 | return inTable(r, neutral) |
| 233 | } |
| 234 | |
| 235 | // StringWidth return width as you can see |
| 236 | func StringWidth(s string) (width int) { |
| 237 | return DefaultCondition.StringWidth(s) |
| 238 | } |
| 239 | |
| 240 | // Truncate return string truncated with w cells |
| 241 | func Truncate(s string, w int, tail string) string { |
| 242 | return DefaultCondition.Truncate(s, w, tail) |
| 243 | } |
| 244 | |
| 245 | // Wrap return string wrapped with w cells |
| 246 | func Wrap(s string, w int) string { |
| 247 | return DefaultCondition.Wrap(s, w) |
| 248 | } |
| 249 | |
| 250 | // FillLeft return string filled in left by spaces in w cells |
| 251 | func FillLeft(s string, w int) string { |
| 252 | return DefaultCondition.FillLeft(s, w) |
| 253 | } |
| 254 | |
| 255 | // FillRight return string filled in left by spaces in w cells |
| 256 | func FillRight(s string, w int) string { |
| 257 | return DefaultCondition.FillRight(s, w) |
| 258 | } |