Improve natural sort (#29611)

Hugely simplify the code, and add more tests (only new approach could
pass)
This commit is contained in:
wxiaoguang 2024-03-05 23:13:35 +08:00 committed by GitHub
parent 3f3335ae51
commit ebff37ae09
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 12 additions and 78 deletions

View File

@ -4,85 +4,12 @@
package base package base
import ( import (
"math/big" "golang.org/x/text/collate"
"unicode/utf8" "golang.org/x/text/language"
) )
// NaturalSortLess compares two strings so that they could be sorted in natural order // NaturalSortLess compares two strings so that they could be sorted in natural order
func NaturalSortLess(s1, s2 string) bool { func NaturalSortLess(s1, s2 string) bool {
var i1, i2 int c := collate.New(language.English, collate.Numeric)
for { return c.CompareString(s1, s2) < 0
rune1, j1, end1 := getNextRune(s1, i1)
rune2, j2, end2 := getNextRune(s2, i2)
if end1 || end2 {
return end1 != end2 && end1
}
dec1 := isDecimal(rune1)
dec2 := isDecimal(rune2)
var less, equal bool
if dec1 && dec2 {
i1, i2, less, equal = compareByNumbers(s1, i1, s2, i2)
} else if !dec1 && !dec2 {
equal = rune1 == rune2
less = rune1 < rune2
i1 = j1
i2 = j2
} else {
return rune1 < rune2
}
if !equal {
return less
}
}
}
func getNextRune(str string, pos int) (rune, int, bool) {
if pos < len(str) {
r, w := utf8.DecodeRuneInString(str[pos:])
// Fallback to ascii
if r == utf8.RuneError {
r = rune(str[pos])
w = 1
}
return r, pos + w, false
}
return 0, pos, true
}
func isDecimal(r rune) bool {
return '0' <= r && r <= '9'
}
func compareByNumbers(str1 string, pos1 int, str2 string, pos2 int) (i1, i2 int, less, equal bool) {
d1, d2 := true, true
var dec1, dec2 string
for d1 || d2 {
if d1 {
r, j, end := getNextRune(str1, pos1)
if !end && isDecimal(r) {
dec1 += string(r)
pos1 = j
} else {
d1 = false
}
}
if d2 {
r, j, end := getNextRune(str2, pos2)
if !end && isDecimal(r) {
dec2 += string(r)
pos2 = j
} else {
d2 = false
}
}
}
less, equal = compareBigNumbers(dec1, dec2)
return pos1, pos2, less, equal
}
func compareBigNumbers(dec1, dec2 string) (less, equal bool) {
d1, _ := big.NewInt(0).SetString(dec1, 10)
d2, _ := big.NewInt(0).SetString(dec2, 10)
cmp := d1.Cmp(d2)
return cmp < 0, cmp == 0
} }

View File

@ -11,7 +11,7 @@ import (
func TestNaturalSortLess(t *testing.T) { func TestNaturalSortLess(t *testing.T) {
test := func(s1, s2 string, less bool) { test := func(s1, s2 string, less bool) {
assert.Equal(t, less, NaturalSortLess(s1, s2)) assert.Equal(t, less, NaturalSortLess(s1, s2), "s1=%q, s2=%q", s1, s2)
} }
test("v1.20.0", "v1.2.0", false) test("v1.20.0", "v1.2.0", false)
test("v1.20.0", "v1.29.0", true) test("v1.20.0", "v1.29.0", true)
@ -20,4 +20,11 @@ func TestNaturalSortLess(t *testing.T) {
test("a-1-a", "a-1-b", true) test("a-1-a", "a-1-b", true)
test("2", "12", true) test("2", "12", true)
test("a", "ab", true) test("a", "ab", true)
test("A", "b", true)
test("a", "B", true)
test("cafe", "café", true)
test("café", "cafe", false)
test("caff", "café", false)
} }