prepare to read from multiple file chunks

weed/filer2/filechunks.go

@@ -52,6 +52,27 @@ func FindUnusedFileChunks(oldChunks, newChunks []*filer_pb.FileChunk) (unused []
 	return
 }
 
+func ReadFromChunks(chunks []*filer_pb.FileChunk, offset int64, size int) (views []*filer_pb.FileChunk) {
+
+	visibles := nonOverlappingVisibleIntervals(chunks)
+
+	stop := offset + int64(size)
+
+	for _, chunk := range visibles {
+		if chunk.start <= offset && offset < chunk.stop {
+			views = append(views, &filer_pb.FileChunk{
+				FileId: chunk.fileId,
+				Offset: offset - chunk.start, // offset is the data starting location in this file id
+				Size:   uint64(min(chunk.stop, stop) - offset),
+			})
+			offset = min(chunk.stop, stop)
+		}
+	}
+
+	return views
+
+}
+
 func logPrintf(name string, visibles []*visibleInterval) {
 
 	return

weed/filer2/filechunks_test.go

@@ -9,10 +9,10 @@ import (
 
 func TestCompactFileChunks(t *testing.T) {
 	chunks := []*filer_pb.FileChunk{
-		{Offset:10, Size:100, FileId:"abc", Mtime:50},
-		{Offset:100, Size:100, FileId:"def", Mtime:100},
-		{Offset:200, Size:100, FileId:"ghi", Mtime:200},
-		{Offset:110, Size:200, FileId:"jkl", Mtime:300},
+		{Offset: 10, Size: 100, FileId: "abc", Mtime: 50},
+		{Offset: 100, Size: 100, FileId: "def", Mtime: 100},
+		{Offset: 200, Size: 100, FileId: "ghi", Mtime: 200},
+		{Offset: 110, Size: 200, FileId: "jkl", Mtime: 300},
 	}
 
 	compacted, garbarge := CompactFileChunks(chunks)
@@ -144,3 +144,135 @@ func TestIntervalMerging(t *testing.T) {
 	}
 
 }
+
+func TestChunksReading(t *testing.T) {
+
+	testcases := []struct {
+		Chunks   []*filer_pb.FileChunk
+		Offset   int64
+		Size     int
+		Expected []*filer_pb.FileChunk
+	}{
+		// case 0: normal
+		{
+			Chunks: []*filer_pb.FileChunk{
+				{Offset: 0, Size: 100, FileId: "abc", Mtime: 123},
+				{Offset: 100, Size: 100, FileId: "asdf", Mtime: 134},
+				{Offset: 200, Size: 100, FileId: "fsad", Mtime: 353},
+			},
+			Offset: 0,
+			Size:   250,
+			Expected: []*filer_pb.FileChunk{
+				{Offset: 0, Size: 100, FileId: "abc"},
+				{Offset: 0, Size: 100, FileId: "asdf"},
+				{Offset: 0, Size: 50, FileId: "fsad"},
+			},
+		},
+		// case 1: updates overwrite full chunks
+		{
+			Chunks: []*filer_pb.FileChunk{
+				{Offset: 0, Size: 100, FileId: "abc", Mtime: 123},
+				{Offset: 0, Size: 200, FileId: "asdf", Mtime: 134},
+			},
+			Offset: 50,
+			Size:   100,
+			Expected: []*filer_pb.FileChunk{
+				{Offset: 50, Size: 100, FileId: "asdf"},
+			},
+		},
+		// case 2: updates overwrite part of previous chunks
+		{
+			Chunks: []*filer_pb.FileChunk{
+				{Offset: 0, Size: 100, FileId: "abc", Mtime: 123},
+				{Offset: 0, Size: 50, FileId: "asdf", Mtime: 134},
+			},
+			Offset: 25,
+			Size:   50,
+			Expected: []*filer_pb.FileChunk{
+				{Offset: 25, Size: 25, FileId: "asdf"},
+				{Offset: 0, Size: 25, FileId: "abc"},
+			},
+		},
+		// case 3: updates overwrite full chunks
+		{
+			Chunks: []*filer_pb.FileChunk{
+				{Offset: 0, Size: 100, FileId: "abc", Mtime: 123},
+				{Offset: 0, Size: 200, FileId: "asdf", Mtime: 134},
+				{Offset: 50, Size: 250, FileId: "xxxx", Mtime: 154},
+			},
+			Offset: 0,
+			Size:   200,
+			Expected: []*filer_pb.FileChunk{
+				{Offset: 0, Size: 50, FileId: "asdf"},
+				{Offset: 0, Size: 150, FileId: "xxxx"},
+			},
+		},
+		// case 4: updates far away from prev chunks
+		{
+			Chunks: []*filer_pb.FileChunk{
+				{Offset: 0, Size: 100, FileId: "abc", Mtime: 123},
+				{Offset: 0, Size: 200, FileId: "asdf", Mtime: 134},
+				{Offset: 250, Size: 250, FileId: "xxxx", Mtime: 154},
+			},
+			Offset: 0,
+			Size:   400,
+			Expected: []*filer_pb.FileChunk{
+				{Offset: 0, Size: 200, FileId: "asdf"},
+				// {Offset: 0, Size: 150, FileId: "xxxx"}, // missing intervals should not happen
+			},
+		},
+		// case 5: updates overwrite full chunks
+		{
+			Chunks: []*filer_pb.FileChunk{
+				{Offset: 0, Size: 100, FileId: "abc", Mtime: 123},
+				{Offset: 0, Size: 200, FileId: "asdf", Mtime: 184},
+				{Offset: 70, Size: 150, FileId: "abc", Mtime: 143},
+				{Offset: 80, Size: 100, FileId: "xxxx", Mtime: 134},
+			},
+			Offset: 0,
+			Size:   220,
+			Expected: []*filer_pb.FileChunk{
+				{Offset: 0, Size: 200, FileId: "asdf"},
+				{Offset: 0, Size: 20, FileId: "abc"},
+			},
+		},
+		// 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,
+			Expected: []*filer_pb.FileChunk{
+				{Offset: 0, Size: 100, FileId: "abc"},
+			},
+		},
+	}
+
+	for i, testcase := range testcases {
+		log.Printf("++++++++++ read test case %d ++++++++++++++++++++", i)
+		chunks := ReadFromChunks(testcase.Chunks, testcase.Offset, testcase.Size)
+		for x, chunk := range chunks {
+			log.Printf("read case %d, chunk %d, offset=%d, size=%d, fileId=%s",
+				i, x, chunk.Offset, chunk.Size, chunk.FileId)
+			if chunk.Offset != testcase.Expected[x].Offset {
+				t.Fatalf("failed on read case %d, chunk %d, Offset %d, expect %d",
+					i, x, chunk.Offset, testcase.Expected[x].Offset)
+			}
+			if chunk.Size != testcase.Expected[x].Size {
+				t.Fatalf("failed on read case %d, chunk %d, Size %d, expect %d",
+					i, x, chunk.Size, testcase.Expected[x].Size)
+			}
+			if chunk.FileId != testcase.Expected[x].FileId {
+				t.Fatalf("failed on read case %d, chunk %d, FileId %s, expect %s",
+					i, x, chunk.FileId, testcase.Expected[x].FileId)
+			}
+		}
+		if len(chunks) != len(testcase.Expected) {
+			t.Fatalf("failed to read test case %d, len %d expected %d", i, len(chunks), len(testcase.Expected))
+		}
+	}
+
+}