internal/set: add set package for set data structures

Package set provides simple set data structures for uint64 and string types.

High-level API:
	type Set(T {}) xxx
	func (Set) Len() int
	func (Set) Has(T) bool
	func (Set) Set(T)
	func (Set) Clear(T)

These data structures are useful for implementing required fields efficiently
or ensuring that protobuf identifiers do not conflict.

Change-Id: If846630a9034909a43121b3e0f6720275f4b7aaf
Reviewed-on: https://go-review.googlesource.com/128898
Reviewed-by: Chris Manghane <cmang@golang.org>

internal/set/doc.go

@@ -0,0 +1,21 @@
+// Copyright 2018 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+// Package set provides simple set data structures for uint64 and string types.
+//
+// The API for every set is:
+//	type Set(T {}) opaque
+//
+//	// Len reports the number of elements in the set.
+//	func (Set) Len() int
+//
+//	// Has reports whether an item is in the set.
+//	func (Set) Has(T) bool
+//
+//	// Set inserts the item into the set.
+//	func (Set) Set(T)
+//
+//	// Clear removes the item from the set.
+//	func (Set) Clear(T)
+package set

internal/set/ints.go

@@ -0,0 +1,73 @@
+// Copyright 2018 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package set
+
+import "math/bits"
+
+// Int32s represents a set of integers within the range of 0..31.
+type Int32s uint32
+
+func (bs *Int32s) Len() int {
+	return bits.OnesCount32(uint32(*bs))
+}
+func (bs *Int32s) Has(n uint64) bool {
+	return uint32(*bs)&(uint32(1)<<n) > 0
+}
+func (bs *Int32s) Set(n uint64) {
+	*(*uint32)(bs) |= uint32(1) << n
+}
+func (bs *Int32s) Clear(n uint64) {
+	*(*uint32)(bs) &^= uint32(1) << n
+}
+
+// Int64s represents a set of integers within the range of 0..63.
+type Int64s uint64
+
+func (bs *Int64s) Len() int {
+	return bits.OnesCount64(uint64(*bs))
+}
+func (bs *Int64s) Has(n uint64) bool {
+	return uint64(*bs)&(uint64(1)<<n) > 0
+}
+func (bs *Int64s) Set(n uint64) {
+	*(*uint64)(bs) |= uint64(1) << n
+}
+func (bs *Int64s) Clear(n uint64) {
+	*(*uint64)(bs) &^= uint64(1) << n
+}
+
+// Ints represents a set of integers within the range of 0..math.MaxUint64.
+type Ints struct {
+	lo Int64s
+	hi map[uint64]struct{}
+}
+
+func (bs *Ints) Len() int {
+	return bs.lo.Len() + len(bs.hi)
+}
+func (bs *Ints) Has(n uint64) bool {
+	if n < 64 {
+		return bs.lo.Has(n)
+	}
+	_, ok := bs.hi[n]
+	return ok
+}
+func (bs *Ints) Set(n uint64) {
+	if n < 64 {
+		bs.lo.Set(n)
+		return
+	}
+	if bs.hi == nil {
+		bs.hi = make(map[uint64]struct{})
+	}
+	bs.hi[n] = struct{}{}
+}
+func (bs *Ints) Clear(n uint64) {
+	if n < 64 {
+		bs.lo.Clear(n)
+		return
+	}
+	delete(bs.hi, n)
+}

internal/set/ints_test.go

@@ -0,0 +1,96 @@
+// Copyright 2018 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package set
+
+import (
+	"math/rand"
+	"testing"
+)
+
+const maxLimit = 1024
+
+var toSet, toClear [maxLimit]bool
+
+func init() {
+	r := rand.New(rand.NewSource(0))
+	for i := 0; i < maxLimit; i++ {
+		toSet[i] = r.Intn(2) == 0
+		toClear[i] = r.Intn(2) == 0
+	}
+}
+
+func TestInts(t *testing.T) {
+	type set interface {
+		Len() int
+		Has(n uint64) bool
+		Set(n uint64)
+		Clear(n uint64)
+	}
+
+	tests := []struct {
+		label   string
+		makeSet func() set
+		limit   int
+	}{
+		{label: "Int32s", makeSet: func() set { return new(Int32s) }, limit: 32},
+		{label: "Int64s", makeSet: func() set { return new(Int64s) }, limit: 64},
+		{label: "Ints", makeSet: func() set { return new(Ints) }, limit: maxLimit},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.label, func(t *testing.T) {
+			ns := tt.makeSet()
+
+			// Check that set starts empty.
+			wantLen := 0
+			if ns.Len() != wantLen {
+				t.Errorf("init: Len() = %d, want %d", ns.Len(), wantLen)
+			}
+			for i := 0; i < tt.limit; i++ {
+				if ns.Has(uint64(i)) {
+					t.Errorf("init: Has(%d) = true, want false", i)
+				}
+			}
+
+			// Set some numbers.
+			for i, b := range toSet[:tt.limit] {
+				if b {
+					ns.Set(uint64(i))
+					wantLen++
+				}
+			}
+
+			// Check that integers were set.
+			if ns.Len() != wantLen {
+				t.Errorf("after Set: Len() = %d, want %d", ns.Len(), wantLen)
+			}
+			for i := 0; i < tt.limit; i++ {
+				if got := ns.Has(uint64(i)); got != toSet[i] {
+					t.Errorf("after Set: Has(%d) = %v, want %v", i, got, !got)
+				}
+			}
+
+			// Clear some numbers.
+			for i, b := range toClear[:tt.limit] {
+				if b {
+					ns.Clear(uint64(i))
+					if toSet[i] {
+						wantLen--
+					}
+				}
+			}
+
+			// Check that integers were cleared.
+			if ns.Len() != wantLen {
+				t.Errorf("after Clear: Len() = %d, want %d", ns.Len(), wantLen)
+			}
+			for i := 0; i < tt.limit; i++ {
+				if got := ns.Has(uint64(i)); got != toSet[i] && !toClear[i] {
+					t.Errorf("after Clear: Has(%d) = %v, want %v", i, got, !got)
+				}
+			}
+		})
+	}
+}

internal/set/strings.go

@@ -0,0 +1,25 @@
+// Copyright 2018 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package set
+
+// Strings represents a set of strings.
+type Strings map[string]struct{}
+
+func (ss *Strings) Len() int {
+	return len(*ss)
+}
+func (ss *Strings) Has(s string) bool {
+	_, ok := (*ss)[s]
+	return ok
+}
+func (ss *Strings) Set(s string) {
+	if *ss == nil {
+		*ss = make(map[string]struct{})
+	}
+	(*ss)[s] = struct{}{}
+}
+func (ss *Strings) Clear(s string) {
+	delete(*ss, s)
+}

internal/set/strings_test.go

@@ -0,0 +1,63 @@
+// Copyright 2018 The Go Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package set
+
+import (
+	"strconv"
+	"testing"
+)
+
+func TestStrings(t *testing.T) {
+	var ss Strings
+
+	// Check that set starts empty.
+	wantLen := 0
+	if ss.Len() != wantLen {
+		t.Errorf("init: Len() = %d, want %d", ss.Len(), wantLen)
+	}
+	for i := 0; i < maxLimit; i++ {
+		if ss.Has(strconv.Itoa(i)) {
+			t.Errorf("init: Has(%d) = true, want false", i)
+		}
+	}
+
+	// Set some strings.
+	for i, b := range toSet[:maxLimit] {
+		if b {
+			ss.Set(strconv.Itoa(i))
+			wantLen++
+		}
+	}
+
+	// Check that strings were set.
+	if ss.Len() != wantLen {
+		t.Errorf("after Set: Len() = %d, want %d", ss.Len(), wantLen)
+	}
+	for i := 0; i < maxLimit; i++ {
+		if got := ss.Has(strconv.Itoa(i)); got != toSet[i] {
+			t.Errorf("after Set: Has(%d) = %v, want %v", i, got, !got)
+		}
+	}
+
+	// Clear some strings.
+	for i, b := range toClear[:maxLimit] {
+		if b {
+			ss.Clear(strconv.Itoa(i))
+			if toSet[i] {
+				wantLen--
+			}
+		}
+	}
+
+	// Check that strings were cleared.
+	if ss.Len() != wantLen {
+		t.Errorf("after Clear: Len() = %d, want %d", ss.Len(), wantLen)
+	}
+	for i := 0; i < maxLimit; i++ {
+		if got := ss.Has(strconv.Itoa(i)); got != toSet[i] && !toClear[i] {
+			t.Errorf("after Clear: Has(%d) = %v, want %v", i, got, !got)
+		}
+	}
+}