Simplify the data structure

cmd/routesum/main.go

@@ -4,21 +4,66 @@ package main
 import (
 	"bufio"
 	"bytes"
+	"flag"
 	"fmt"
 	"io"
 	"os"
+	"runtime/pprof"
 
 	"github.com/PatrickCronin/routesum/pkg/routesum"
+	"github.com/pkg/errors"
 )
 
 func main() {
-	if err := summarize(os.Stdin, os.Stdout); err != nil {
+	cpuProfile := flag.String("cpuprofile", "", "write cpu profile to file")
+	memProfile := flag.String("memprofile", "", "write mem profile to file")
+	showMemStats := flag.Bool(
+		"show-mem-stats",
+		false,
+		"Whether or not to write memory usage stats to STDERR. (This functionity requires the use of `unsafe`, so may not be perfect.)", //nolint: lll
+	)
+	flag.Parse()
+
+	var cpuProfOut io.Writer
+	if *cpuProfile != "" {
+		var err error
+		if cpuProfOut, err = os.Create(*cpuProfile); err != nil {
+			fmt.Fprint(os.Stderr, errors.Wrap(err, "create cpu profile output file").Error())
+			os.Exit(1)
+		}
+	}
+
+	var memProfOut io.WriteCloser
+	if *memProfile != "" {
+		var err error
+		if memProfOut, err = os.Create(*memProfile); err != nil {
+			fmt.Fprint(os.Stderr, errors.Wrap(err, "create mem profile output file").Error())
+		}
+	}
+
+	var memStatsOut io.Writer
+	if *showMemStats {
+		memStatsOut = os.Stderr
+	}
+
+	if err := summarize(os.Stdin, os.Stdout, memStatsOut, cpuProfOut, memProfOut); err != nil {
 		fmt.Fprintf(os.Stderr, "summarize: %s\n", err.Error())
 		os.Exit(1)
 	}
 }
 
-func summarize(in io.Reader, out io.Writer) error {
+func summarize(
+	in io.Reader,
+	out, memStatsOut, cpuProfOut io.Writer,
+	memProfOut io.WriteCloser,
+) error {
+	if cpuProfOut != nil {
+		if err := pprof.StartCPUProfile(cpuProfOut); err != nil {
+			return errors.Wrap(err, "start cpu profiling")
+		}
+		defer pprof.StopCPUProfile()
+	}
+
 	rs := routesum.NewRouteSum()
 	scanner := bufio.NewScanner(in)
 	for scanner.Scan() {
@@ -32,6 +77,32 @@ func summarize(in io.Reader, out io.Writer) error {
 		}
 	}
 
+	if memProfOut != nil {
+		if err := pprof.WriteHeapProfile(memProfOut); err != nil {
+			return errors.Wrap(err, "write mem profile")
+		}
+		if err := memProfOut.Close(); err != nil {
+			return errors.Wrap(err, "close mem profile")
+		}
+	}
+
+	if memStatsOut != nil {
+		numInternalNodes, numLeafNodes, internalNodesTotalSize, leafNodesTotalSize := rs.MemUsage()
+		fmt.Fprintf(memStatsOut,
+			`Num internal nodes:           %d
+Num leaf nodes:               %d
+Size of all internal nodes:   %d
+Size of all leaf nodes:       %d
+Total size of data structure: %d
+`,
+			numInternalNodes,
+			numLeafNodes,
+			internalNodesTotalSize,
+			leafNodesTotalSize,
+			internalNodesTotalSize+leafNodesTotalSize,
+		)
+	}
+
 	for _, s := range rs.SummaryStrings() {
 		if _, err := out.Write([]byte(s + "\n")); err != nil {
 			return fmt.Errorf("write output: %w", err)

cmd/routesum/main_test.go

@@ -1,6 +1,8 @@
 package main
 
 import (
+	"io"
+	"regexp"
 	"strings"
 	"testing"
 
@@ -9,12 +11,42 @@ import (
 )
 
 func TestSummarize(t *testing.T) {
+	tests := []struct {
+		name         string
+		showMemStats bool
+		expected     *regexp.Regexp
+	}{
+		{
+			name:         "without memory statistics",
+			showMemStats: false,
+			expected:     regexp.MustCompile(`^$`),
+		},
+		{
+			name:         "with memory statistics",
+			showMemStats: true,
+			expected:     regexp.MustCompile(`Num internal nodes`),
+		},
+	}
+
 	inStr := "\n192.0.2.0\n192.0.2.1\n"
-	in := strings.NewReader(inStr)
-	var out strings.Builder
 
-	err := summarize(in, &out)
-	require.NoError(t, err, "summarize does not throw an error")
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			in := strings.NewReader(inStr)
+			var out strings.Builder
+
+			var memStatsBuilder strings.Builder
+			var memStatsOut io.Writer
+
+			if test.showMemStats {
+				memStatsOut = &memStatsBuilder
+			}
+
+			err := summarize(in, &out, memStatsOut, nil, nil)
+			require.NoError(t, err, "summarize does not throw an error")
 
-	assert.Equal(t, "192.0.2.0/31\n", out.String(), "read expected output")
+			assert.Equal(t, "192.0.2.0/31\n", out.String(), "read expected output")
+			assert.Regexp(t, test.expected, memStatsBuilder.String(), "read expected memory stats")
+		})
+	}
 }

pkg/routesum/routesum.go

@@ -7,21 +7,21 @@ import (
 	"strings"
 
 	"github.com/PatrickCronin/routesum/pkg/routesum/bitslice"
-	"github.com/PatrickCronin/routesum/pkg/routesum/rstrie"
+	"github.com/PatrickCronin/routesum/pkg/routesum/rstree"
 	"github.com/pkg/errors"
 	"inet.af/netaddr"
 )
 
 // RouteSum has methods supporting route summarization of networks and hosts
 type RouteSum struct {
-	ipv4, ipv6 *rstrie.RSTrie
+	ipv4, ipv6 *rstree.RSTree
 }
 
 // NewRouteSum returns an initialized RouteSum object
 func NewRouteSum() *RouteSum {
 	rs := new(RouteSum)
-	rs.ipv4 = rstrie.NewRSTrie()
-	rs.ipv6 = rstrie.NewRSTrie()
+	rs.ipv4 = rstree.NewRSTree()
+	rs.ipv6 = rstree.NewRSTree()
 
 	return rs
 }
@@ -142,3 +142,13 @@ func ipv6FromBits(bits bitslice.BitSlice) netaddr.IP {
 	copy(byteArray[:], bytes[0:16])
 	return netaddr.IPv6Raw(byteArray)
 }
+
+// MemUsage provides information about memory usage.
+func (rs *RouteSum) MemUsage() (uint, uint, uintptr, uintptr) {
+	ipv4NumInternalNodes, ipv4NumLeafNodes, ipv4InternalNodesTotalSize, ipv4LeafNodesTotalSize := rs.ipv4.MemUsage()
+	ipv6NumInternalNodes, ipv6NumLeafNodes, ipv6InternalNodesTotalSize, ipv6LeafNodesTotalSize := rs.ipv6.MemUsage()
+	return ipv4NumInternalNodes + ipv6NumInternalNodes,
+		ipv4NumLeafNodes + ipv6NumLeafNodes,
+		ipv4InternalNodesTotalSize + ipv6InternalNodesTotalSize,
+		ipv4LeafNodesTotalSize + ipv6LeafNodesTotalSize
+}

pkg/routesum/routesum_test.go

@@ -455,3 +455,76 @@ func TestSummarize(t *testing.T) { //nolint: funlen
 		})
 	}
 }
+
+func TestMemUsage(t *testing.T) { //nolint: funlen
+	tests := []struct {
+		name                     string
+		entries                  []string
+		expectedNumInternalNodes uint
+		expectedNumLeafNodes     uint
+	}{
+		{
+			name:                     "new trie",
+			expectedNumInternalNodes: 0,
+			expectedNumLeafNodes:     0,
+		},
+		{
+			name: "one item, IPv4",
+			entries: []string{
+				"192.0.2.1",
+			},
+			expectedNumInternalNodes: 32,
+			expectedNumLeafNodes:     1,
+		},
+		{
+			name: "two items, IPv4, summarized",
+			entries: []string{
+				"192.0.2.1",
+				"192.0.2.0",
+			},
+			expectedNumInternalNodes: 31,
+			expectedNumLeafNodes:     1,
+		},
+		{
+			name: "two items, IPv4, unsummarized",
+			entries: []string{
+				"192.0.2.1",
+				"192.0.2.2",
+			},
+			expectedNumInternalNodes: 33,
+			expectedNumLeafNodes:     2,
+		},
+		{
+			name: "one item, IPv6",
+			entries: []string{
+				"2001:db8::1",
+			},
+			expectedNumInternalNodes: 128,
+			expectedNumLeafNodes:     1,
+		},
+		{
+			name: "one IPv4 address, one IPv6 address",
+			entries: []string{
+				"192.0.2.0",
+				"2001:db8::1",
+			},
+			expectedNumInternalNodes: 160,
+			expectedNumLeafNodes:     2,
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.name, func(t *testing.T) {
+			rs := NewRouteSum()
+
+			for _, entry := range test.entries {
+				err := rs.InsertFromString(entry)
+				require.NoError(t, err)
+			}
+
+			numInternalNodes, numLeafNodes, _, _ := rs.MemUsage()
+			assert.Equal(t, test.expectedNumInternalNodes, numInternalNodes, "num internal nodes")
+			assert.Equal(t, test.expectedNumLeafNodes, numLeafNodes, "num leaf nodes")
+		})
+	}
+}