// WriteByteOffs generates a kv file with key-value pairs represented as a
// slice of buffer and some int slices of key offsets, key ends, value offsets,
// and value ends.
func WriteByteOffs(fp sophie.FsPath, buffer []byte,
keyOffs, keyEnds, valOffs, valEnds []int) error {
writer, err := fp.Create()
if err != nil {
return err
}
defer writer.Close()
for i, keyOff := range keyOffs {
keyEnd, valOff, valEnd := keyEnds[i], valOffs[i], valEnds[i]
if err := sophie.VInt(keyEnd - keyOff).WriteTo(writer); err != nil {
return err
}
if _, err := writer.Write(buffer[keyOff:keyEnd]); err != nil {
return err
}
if err := sophie.VInt(valEnd - valOff).WriteTo(writer); err != nil {
return err
}
if _, err := writer.Write(buffer[valOff:valEnd]); err != nil {
return err
}
}
return nil
}
func TestReduceValues(t *testing.T) {
/*
* Source are of two parts with nothing in each, but at each mapend, a pair
* of <"part", <part>> is collected. So the reducer will check whether a key
* of "part" with two different values are reduced.
*/
job := MrJob{
Source: []Input{
&InputStruct{
PartCountF: func() (int, error) {
return 2, nil
},
},
},
NewMapperF: func(src, part int) Mapper {
return &MapperStruct{
MapEndF: func(c PartCollector) error {
return c.CollectTo(0, sophie.RawString("part"),
sophie.VInt(part))
},
}
},
NewReducerF: func(part int) Reducer {
st := make(map[sophie.VInt]bool)
return &ReducerStruct{
NewKeyF: sophie.NewRawString,
NewValF: sophie.NewVInt,
ReduceF: func(key sophie.SophieWriter,
nextVal SophierIterator, c []sophie.Collector) error {
keyStr := string(*key.(*sophie.RawString))
if keyStr != "part" {
return errors.New(`Key should be "part"`)
}
for {
val, err := nextVal()
if err == sophie.EOF {
break
}
if err != nil {
return err
}
part := *val.(*sophie.VInt)
if st[part] {
t.Errorf("Duplicated value: %v", part)
}
st[part] = true
}
return nil
},
}
},
}
assert.NoErrorf(t, "job.Run failed: %v", job.Run())
}
func (c *CrawlingEntry) WriteTo(w sophie.Writer) error {
if err := sophie.Time(c.ScheduleTime).WriteTo(w); err != nil {
return err
}
if err := sophie.VInt(c.Version).WriteTo(w); err != nil {
return err
}
if err := sophie.String(c.Etag).WriteTo(w); err != nil {
return err
}
return nil
}
// sophie.CollectCloser interface
func (kvw *Writer) Collect(key, val sophie.SophieWriter) error {
// write key
kvw.objBuf.Reset()
key.WriteTo(&kvw.objBuf)
if err := sophie.VInt(kvw.objBuf.Len()).WriteTo(kvw.writer); err != nil {
return err
}
if _, err := kvw.writer.Write(kvw.objBuf.Bytes()); err != nil {
return err
}
// write val
kvw.objBuf.Reset()
val.WriteTo(&kvw.objBuf)
if err := sophie.VInt(kvw.objBuf.Len()).WriteTo(kvw.writer); err != nil {
return err
}
if _, err := kvw.writer.Write(kvw.objBuf.Bytes()); err != nil {
return err
}
// success
return nil
}