add tests

weed/filer2/filechunks2_test.go

@@ -0,0 +1,52 @@
+package filer2
+
+import (
+	"sort"
+	"testing"
+
+	"github.com/chrislusf/seaweedfs/weed/glog"
+	"github.com/chrislusf/seaweedfs/weed/pb/filer_pb"
+)
+
+func TestCompactFileChunksRealCase(t *testing.T) {
+
+	chunks := []*filer_pb.FileChunk{
+		{FileId:"2,512f31f2c0700a", Offset:       0,   Size:      25-      0, Mtime: 5320497},
+		{FileId:"6,512f2c2e24e9e8", Offset:  868352,   Size:  917585- 868352, Mtime: 5320492},
+		{FileId:"7,514468dd5954ca", Offset:  884736,   Size:  901120- 884736, Mtime: 5325928},
+		{FileId:"5,5144463173fe77", Offset:  917504,   Size: 2297856- 917504, Mtime: 5325894},
+		{FileId:"4,51444c7ab54e2d", Offset: 2301952,   Size: 2367488-2301952, Mtime: 5325900},
+		{FileId:"4,514450e643ad22", Offset: 2371584,   Size: 2420736-2371584, Mtime: 5325904},
+		{FileId:"1,51446489116dbb", Offset: 2424832,   Size: 2445312-2424832, Mtime: 5325924},
+		{FileId:"6,514456a5e9e4d7", Offset: 2449408,   Size: 2490368-2449408, Mtime: 5325910},
+		{FileId:"3,51444f8d53eebe", Offset: 2494464,   Size: 2555904-2494464, Mtime: 5325903},
+		{FileId:"4,5144578b097c7e", Offset: 2560000,   Size: 2596864-2560000, Mtime: 5325911},
+		{FileId:"4,51446563a1d91a", Offset: 2600960,   Size: 2617344-2600960, Mtime: 5325925},
+		{FileId:"7,514471a77e7b61", Offset: 2621440,   Size: 2633728-2621440, Mtime: 5325937},
+		{FileId:"3,51445500b6b4ac", Offset: 2637824,   Size: 2678784-2637824, Mtime: 5325909},
+		{FileId:"2,514476409139dc", Offset: 2682880,   Size: 2691072-2682880, Mtime: 5325942},
+		{FileId:"1,51446285e52a61", Offset: 2695168,   Size: 2715648-2695168, Mtime: 5325922},
+		{FileId:"6,51447234faf940", Offset: 2719744,   Size: 2730531-2719744, Mtime: 5325938},
+		{FileId:"5,514473558a90ed", Offset: 2736128,   Size: 2744590-2736128, Mtime: 5325939},
+	}
+
+	printChunks("before", chunks)
+
+	compacted, garbage := CompactFileChunks(nil, chunks)
+
+	printChunks("compacted", compacted)
+	printChunks("garbage", garbage)
+
+}
+
+func printChunks(name string, chunks []*filer_pb.FileChunk) {
+	sort.Slice(chunks, func(i, j int) bool {
+		if chunks[i].Offset == chunks[j].Offset {
+			return chunks[i].Mtime < chunks[j].Mtime
+		}
+		return chunks[i].Offset < chunks[j].Offset
+	})
+	for _, chunk := range chunks {
+		glog.V(0).Infof("%s chunk %s [%10d,%10d)", name, chunk.GetFileIdString(), chunk.Offset, chunk.Offset+int64(chunk.Size))
+	}
+}

weed/filer2/filechunks_test.go

@@ -4,6 +4,8 @@ import (
 	"fmt"
 	"log"
 	"math"
+	"math/rand"
+	"strconv"
 	"testing"
 
 	"github.com/stretchr/testify/assert"
@@ -62,6 +64,42 @@ func TestCompactFileChunks2(t *testing.T) {
 	}
 }
 
+func TestRandomFileChunksCompact(t *testing.T) {
+
+	data := make([]byte, 1024)
+
+	var chunks []*filer_pb.FileChunk
+	for i := 0; i < 15; i++ {
+		start, stop := rand.Intn(len(data)), rand.Intn(len(data))
+		if start > stop {
+			start, stop = stop, start
+		}
+		if start + 16 < stop {
+			stop = start + 16
+		}
+		chunk := &filer_pb.FileChunk{
+			FileId: strconv.Itoa(i),
+			Offset: int64(start),
+			Size:   uint64(stop - start),
+			Mtime:  int64(i),
+			Fid:    &filer_pb.FileId{FileKey: uint64(i)},
+		}
+		chunks = append(chunks, chunk)
+		for x := start; x < stop; x++ {
+			data[x] = byte(i)
+		}
+	}
+
+	visibles, _ := NonOverlappingVisibleIntervals(nil, chunks)
+
+	for _, v := range visibles {
+		for x := v.start; x < v.stop; x++ {
+			assert.Equal(t, strconv.Itoa(int(data[x])), v.fileId)
+		}
+	}
+
+}
+
 func TestIntervalMerging(t *testing.T) {
 
 	testcases := []struct {
@@ -142,12 +180,12 @@ func TestIntervalMerging(t *testing.T) {
 		// case 6: same updates
 		{
 			Chunks: []*filer_pb.FileChunk{
-				{Offset: 0, Size: 100, FileId: "abc", Mtime: 123},
-				{Offset: 0, Size: 100, FileId: "abc", Mtime: 123},
-				{Offset: 0, Size: 100, FileId: "abc", Mtime: 123},
+				{Offset: 0, Size: 100, FileId: "abc", Fid: &filer_pb.FileId{FileKey: 1}, Mtime: 123},
+				{Offset: 0, Size: 100, FileId: "axf", Fid: &filer_pb.FileId{FileKey: 2}, Mtime: 123},
+				{Offset: 0, Size: 100, FileId: "xyz", Fid: &filer_pb.FileId{FileKey: 3}, Mtime: 123},
 			},
 			Expected: []*VisibleInterval{
-				{start: 0, stop: 100, fileId: "abc"},
+				{start: 0, stop: 100, fileId: "xyz"},
 			},
 		},
 		// case 7: real updates