Serge Bazanski | cc25bdf | 2018-10-25 14:02:58 +0200 | [diff] [blame] | 1 | package sortorder |
| 2 | |
| 3 | // Natural implements sort.Interface to sort strings in natural order. This |
| 4 | // means that e.g. "abc2" < "abc12". |
| 5 | // |
| 6 | // Non-digit sequences and numbers are compared separately. The former are |
| 7 | // compared bytewise, while the latter are compared numerically (except that |
| 8 | // the number of leading zeros is used as a tie-breaker, so e.g. "2" < "02") |
| 9 | // |
| 10 | // Limitation: only ASCII digits (0-9) are considered. |
| 11 | type Natural []string |
| 12 | |
| 13 | func (n Natural) Len() int { return len(n) } |
| 14 | func (n Natural) Swap(i, j int) { n[i], n[j] = n[j], n[i] } |
| 15 | func (n Natural) Less(i, j int) bool { return NaturalLess(n[i], n[j]) } |
| 16 | |
| 17 | func isdigit(b byte) bool { return '0' <= b && b <= '9' } |
| 18 | |
| 19 | // NaturalLess compares two strings using natural ordering. This means that e.g. |
| 20 | // "abc2" < "abc12". |
| 21 | // |
| 22 | // Non-digit sequences and numbers are compared separately. The former are |
| 23 | // compared bytewise, while the latter are compared numerically (except that |
| 24 | // the number of leading zeros is used as a tie-breaker, so e.g. "2" < "02") |
| 25 | // |
| 26 | // Limitation: only ASCII digits (0-9) are considered. |
| 27 | func NaturalLess(str1, str2 string) bool { |
| 28 | idx1, idx2 := 0, 0 |
| 29 | for idx1 < len(str1) && idx2 < len(str2) { |
| 30 | c1, c2 := str1[idx1], str2[idx2] |
| 31 | dig1, dig2 := isdigit(c1), isdigit(c2) |
| 32 | switch { |
| 33 | case dig1 != dig2: // Digits before other characters. |
| 34 | return dig1 // True if LHS is a digit, false if the RHS is one. |
| 35 | case !dig1: // && !dig2, because dig1 == dig2 |
| 36 | // UTF-8 compares bytewise-lexicographically, no need to decode |
| 37 | // codepoints. |
| 38 | if c1 != c2 { |
| 39 | return c1 < c2 |
| 40 | } |
| 41 | idx1++ |
| 42 | idx2++ |
| 43 | default: // Digits |
| 44 | // Eat zeros. |
| 45 | for ; idx1 < len(str1) && str1[idx1] == '0'; idx1++ { |
| 46 | } |
| 47 | for ; idx2 < len(str2) && str2[idx2] == '0'; idx2++ { |
| 48 | } |
| 49 | // Eat all digits. |
| 50 | nonZero1, nonZero2 := idx1, idx2 |
| 51 | for ; idx1 < len(str1) && isdigit(str1[idx1]); idx1++ { |
| 52 | } |
| 53 | for ; idx2 < len(str2) && isdigit(str2[idx2]); idx2++ { |
| 54 | } |
| 55 | // If lengths of numbers with non-zero prefix differ, the shorter |
| 56 | // one is less. |
| 57 | if len1, len2 := idx1-nonZero1, idx2-nonZero2; len1 != len2 { |
| 58 | return len1 < len2 |
| 59 | } |
| 60 | // If they're equal, string comparison is correct. |
| 61 | if nr1, nr2 := str1[nonZero1:idx1], str2[nonZero2:idx2]; nr1 != nr2 { |
| 62 | return nr1 < nr2 |
| 63 | } |
| 64 | // Otherwise, the one with less zeros is less. |
| 65 | // Because everything up to the number is equal, comparing the index |
| 66 | // after the zeros is sufficient. |
| 67 | if nonZero1 != nonZero2 { |
| 68 | return nonZero1 < nonZero2 |
| 69 | } |
| 70 | } |
| 71 | // They're identical so far, so continue comparing. |
| 72 | } |
| 73 | // So far they are identical. At least one is ended. If the other continues, |
| 74 | // it sorts last. |
| 75 | return len(str1) < len(str2) |
| 76 | } |