|
| 1 | +// +build gitlabnvidia |
| 2 | + |
| 3 | +package dpipes |
| 4 | + |
| 5 | +import ( |
| 6 | + "encoding/json" |
| 7 | + "fmt" |
| 8 | + "io" |
| 9 | + "io/ioutil" |
| 10 | + "os" |
| 11 | + "testing" |
| 12 | + |
| 13 | + "github.com/NVIDIA/go-tfdata/tfdata/core" |
| 14 | + "github.com/NVIDIA/go-tfdata/tfdata/transform" |
| 15 | + "github.com/stretchr/testify/assert" |
| 16 | +) |
| 17 | + |
| 18 | +type ( |
| 19 | + SamplesReader struct { |
| 20 | + pipe Pipe |
| 21 | + } |
| 22 | +) |
| 23 | + |
| 24 | +func (r *SamplesReader) Read() (sample *core.Sample, err error) { |
| 25 | + s, ok := <-r.pipe |
| 26 | + if !ok { |
| 27 | + return nil, io.EOF |
| 28 | + } |
| 29 | + |
| 30 | + return tarpSampleToTfDataSample(s), nil |
| 31 | +} |
| 32 | + |
| 33 | +func TFRecordSink(t *testing.T, writer io.Writer) Sink { |
| 34 | + return func(pipe Pipe) { |
| 35 | + w := core.NewTFRecordWriter(writer) |
| 36 | + samplesReader := &SamplesReader{pipe} |
| 37 | + tfExamplesReader := transform.NewSamplesToTFExample(samplesReader) |
| 38 | + err := w.WriteMessages(tfExamplesReader) |
| 39 | + |
| 40 | + assert.NoError(t, err) |
| 41 | + } |
| 42 | +} |
| 43 | + |
| 44 | +func TFRecordSource(t *testing.T, reader io.Reader) Source { |
| 45 | + return func(pipe Pipe) { |
| 46 | + defer close(pipe) |
| 47 | + var ( |
| 48 | + ex *core.TFExample |
| 49 | + err error |
| 50 | + r core.TFExampleReader |
| 51 | + ) |
| 52 | + r = core.NewTFRecordReader(reader) |
| 53 | + for ex, err = r.Read(); err == nil; ex, err = r.Read() { |
| 54 | + pipe <- tfExampleTarpSample(ex) |
| 55 | + } |
| 56 | + if err != io.EOF { |
| 57 | + assert.Fail(t, "expected to get io.EOF, got %v instead", err) |
| 58 | + } |
| 59 | + } |
| 60 | +} |
| 61 | + |
| 62 | +func SamplesChecker(t *testing.T, target int) Process { |
| 63 | + return func(in, out Pipe) { |
| 64 | + total := 0 |
| 65 | + for s := range in { |
| 66 | + assert.Equal(t, s["txt"], Bytes(fmt.Sprintf("%d", total))) |
| 67 | + assert.Equal(t, s["__key__"], Bytes(fmt.Sprintf("%06d", total))) |
| 68 | + total++ |
| 69 | + out <- s |
| 70 | + } |
| 71 | + close(out) |
| 72 | + assert.Equal(t, target, total) |
| 73 | + } |
| 74 | +} |
| 75 | + |
| 76 | +func tarpSampleToTfDataSample(sample Sample) *core.Sample { |
| 77 | + s := core.NewSample() |
| 78 | + for k, v := range sample { |
| 79 | + s.Entries[k] = v |
| 80 | + } |
| 81 | + return s |
| 82 | +} |
| 83 | + |
| 84 | +func tfExampleTarpSample(example *core.TFExample) Sample { |
| 85 | + s := make(map[string]Bytes, len(example.GetFeatures().Feature)) |
| 86 | + for k, v := range example.GetFeatures().Feature { |
| 87 | + var b Bytes |
| 88 | + err := json.Unmarshal(v.GetBytesList().Value[0], &b) |
| 89 | + if err != nil { |
| 90 | + panic(err) |
| 91 | + } |
| 92 | + s[k] = b // assume that all TFExample features are just a list of bytes |
| 93 | + } |
| 94 | + return s |
| 95 | +} |
| 96 | + |
| 97 | +func PrepareTarSource() Source { |
| 98 | + return func(pipe Pipe) { |
| 99 | + for i := 0; i < 1; i++ { |
| 100 | + pipe <- Sample{ |
| 101 | + "__key__": Bytes(fmt.Sprintf("%06d", i)), |
| 102 | + "txt": Bytes(fmt.Sprintf("%d", i)), |
| 103 | + } |
| 104 | + } |
| 105 | + close(pipe) |
| 106 | + } |
| 107 | +} |
| 108 | + |
| 109 | +func prepareTar(t *testing.T) *os.File { |
| 110 | + var ( |
| 111 | + sinkFd *os.File |
| 112 | + err error |
| 113 | + ) |
| 114 | + sinkFd, err = ioutil.TempFile("", "go-tfdata-*.tar") |
| 115 | + assert.NoError(t, err) |
| 116 | + |
| 117 | + sink := TarSink(sinkFd) |
| 118 | + Processing(PrepareTarSource(), nil, sink) |
| 119 | + return sinkFd |
| 120 | +} |
| 121 | + |
| 122 | +func TestGoTfData(t *testing.T) { |
| 123 | + var ( |
| 124 | + sourceFd = prepareTar(t) |
| 125 | + sinkFd *os.File |
| 126 | + err error |
| 127 | + ) |
| 128 | + |
| 129 | + defer os.RemoveAll(sourceFd.Name()) |
| 130 | + sourceFd, err = os.Open(sourceFd.Name()) |
| 131 | + assert.NoError(t, err) |
| 132 | + |
| 133 | + sinkFd, err = ioutil.TempFile("", "go-tfdata-*.tfrecord") |
| 134 | + assert.NoError(t, err) |
| 135 | + defer os.RemoveAll(sinkFd.Name()) |
| 136 | + |
| 137 | + Processing(TarSource(sourceFd), nil, TFRecordSink(t, sinkFd)) |
| 138 | + sinkFd.Close() |
| 139 | + sourceFd, err = os.Open(sinkFd.Name()) |
| 140 | + assert.NoError(t, err) |
| 141 | + sinkFd, err = os.OpenFile(os.DevNull, os.O_RDWR, os.ModeAppend) |
| 142 | + assert.NoError(t, err) |
| 143 | + |
| 144 | + Processing(TFRecordSource(t, sourceFd), SamplesChecker(t, 1), TFRecordSink(t, sinkFd)) |
| 145 | +} |
0 commit comments