x/text/internal/colltab: Sort numbers with leading zeros

Makes numeric sorting deterministic. Particularly useful for
  approximating how an OS file manager will sort files.

collate/collate_test.go

@@ -474,15 +474,15 @@ func TestNumeric(t *testing.T) {
 		{"A-1", "A-2", -1},
 		{"A-2", "A-12", -1},
 		{"A-12", "A-2", 1},
-		{"A-0001", "A-1", 0},
+		{"A-0001", "A-1", 1},
 		{"0000-", "1-", -1},
-		{"00001", "1", 0},
+		{"00001", "1", 1},
 		{"00", "00", 0},
-		{"0", "00", 0},
-		{"00", "0", 0},
-		{"01", "001", 0},
-		{"01", "1", 0},
-		{"1", "01", 0},
+		{"0", "00", -1},
+		{"00", "0", 1},
+		{"01", "001", -1},
+		{"01", "1", 1},
+		{"1", "01", -1},
 		{"9-A", "0-A", 1},
 		{"99-A", "0-A", 1},
 		{"9-A", "1-A", 1},

internal/colltab/numeric.go

@@ -80,14 +80,20 @@ func (nw *numericWeighter) AppendNext(buf []Elem, s []byte) (ce []Elem, n int) {
 	}
 	// ce might have been grown already, so take it instead of buf.
 	nc.init(ce, len(buf), isZero)
+	oldIndex := len(nc.elems)
 	for n < len(s) {
 		ce, sz := nw.Weighter.AppendNext(nc.elems, s[n:])
 		nc.b = s
 		n += sz
 		if !nc.update(ce) {
 			break
 		}
+		oldIndex = len(nc.elems)
 	}
+	nc.elems = append(nc.elems, 0)
+	copy(nc.elems[oldIndex+1:], nc.elems[oldIndex:])
+	nc.elems[oldIndex], _ = MakeElem(nc.zero+1, defaultSecondary, defaultTertiary, 0)
+
 	return nc.result(), n
 }
 
@@ -105,14 +111,20 @@ func (nw *numericWeighter) AppendNextString(buf []Elem, s string) (ce []Elem, n
 		return ce, n
 	}
 	nc.init(ce, len(buf), isZero)
+	oldIndex := len(nc.elems)
 	for n < len(s) {
 		ce, sz := nw.Weighter.AppendNextString(nc.elems, s[n:])
 		nc.s = s
 		n += sz
 		if !nc.update(ce) {
 			break
 		}
+		oldIndex = len(nc.elems)
 	}
+	nc.elems = append(nc.elems, 0)
+	copy(nc.elems[oldIndex+1:], nc.elems[oldIndex:])
+	nc.elems[oldIndex], _ = MakeElem(nc.zero+1, defaultSecondary, defaultTertiary, 0)
+
 	return nc.result(), n
 }
 
@@ -122,6 +134,7 @@ type numberConverter struct {
 	elems    []Elem
 	nDigits  int
 	lenIndex int
+	zero     int
 
 	s string // set if the input was of type string
 	b []byte // set if the input was of type []byte
@@ -133,6 +146,7 @@ func (nc *numberConverter) init(elems []Elem, oldLen int, isZero bool) {
 	// Insert a marker indicating the start of a number and a placeholder
 	// for the number of digits.
 	if isZero {
+		nc.zero++
 		elems = append(elems[:oldLen], nc.w.numberStart, 0)
 	} else {
 		elems = append(elems, 0, 0)
@@ -217,6 +231,9 @@ const maxDigits = 1<<maxPrimaryBits - 1
 func (nc *numberConverter) update(elems []Elem) bool {
 	isZero, ok := nc.checkNextDigit(elems)
 	if nc.nDigits == 0 && isZero {
+		if nc.zero+1 < maxDigits {
+			nc.zero++
+		}
 		return true
 	}
 	nc.elems = elems

internal/colltab/numeric_test.go

@@ -78,36 +78,37 @@ func TestNumericAppendNext(t *testing.T) {
 		{"a", p(5)},
 		{"klm", p(99)},
 		{"aa", p(5, 5)},
-		{"1", p(120, 2, 101)},
-		{"0", p(120, 1)},
-		{"00", p(120, 1)},
-		{"01", p(120, 2, 101)},
-		{"0001", p(120, 2, 101)},
-		{"02", p(120, 2, 102)},
-		{"10", p(120, 3, 101, 100)},
-		{"99", p(120, 3, 119, 119)},
-		{"9999", p(120, 5, 119, 119, 119, 119)},
-		{"1a", p(120, 2, 101, 5)},
-		{"0b", p(120, 1, 6)},
-		{"01c", p(120, 2, 101, 8, 2)},
-		{"10x", p(120, 3, 101, 100, 200)},
-		{"99y", p(120, 3, 119, 119, 201)},
-		{"9999nop", p(120, 5, 119, 119, 119, 119, 121)},
+		{"1", p(120, 2, 101, 1)},
+		{"0", p(120, 1, 2)},
+		{"00", p(120, 1, 3)},
+		{"01", p(120, 2, 101, 2)},
+		{"0001", p(120, 2, 101, 4)},
+		{"02", p(120, 2, 102, 2)},
+		{"10", p(120, 3, 101, 100, 1)},
+		{"99", p(120, 3, 119, 119, 1)},
+		{"9999", p(120, 5, 119, 119, 119, 119, 1)},
+		{"1a", p(120, 2, 101, 1, 5)},
+		{"0b", p(120, 1, 2, 6)},
+		{"01c", p(120, 2, 101, 2, 8, 2)},
+		{"10x", p(120, 3, 101, 100, 1, 200)},
+		{"99y", p(120, 3, 119, 119, 1, 201)},
+		{"9999nop", p(120, 5, 119, 119, 119, 119, 1, 121)},
 
 		// Allow follow-up collation elements if they have a zero non-primary.
-		{"١٢٩", []Elem{e(120), e(4), e(101), tPlus3, e(102), tPlus3, e(119), tPlus3}},
+		{"١٢٩", []Elem{e(120), e(4), e(101), tPlus3, e(102), tPlus3, e(119), tPlus3, e(1)}},
 		{
 			"１２９",
 			[]Elem{
 				e(120), e(4),
 				e(101, digSec, digTert+1),
 				e(102, digSec, digTert+3),
 				e(119, digSec, digTert+1),
+				e(1),
 			},
 		},
 
 		// Ensure AppendNext* adds to the given buffer.
-		{"a10", p(5, 120, 3, 101, 100)},
+		{"a10", p(5, 120, 3, 101, 100, 1)},
 	} {
 		nw := NewNumericWeighter(numWeighter)
 
@@ -148,6 +149,22 @@ func TestNumericOverflow(t *testing.T) {
 	}
 }
 
+func TestNumericZeroOverflow(t *testing.T) {
+	manyDigits := strings.Repeat("0", maxDigits+1) + "a"
+
+	nw := NewNumericWeighter(numWeighter)
+
+	got, n := nw.AppendNextString(nil, manyDigits)
+
+	if n != maxDigits+2 { // Zeros after maxDigits-1 are ignored but are still consumed so that a number with leading zeros is ordered after a number with less leading zeros
+		t.Errorf("n: got %d; want %d", n, maxDigits+2)
+	}
+
+	if got[2].Primary() != maxDigits {
+		t.Errorf("primary(e[2]): got %d; want %d", got[1].Primary(), maxDigits)
+	}
+}
+
 func TestNumericWeighterAlloc(t *testing.T) {
 	buf := make([]Elem, 100)
 	w := NewNumericWeighter(numWeighter)