// KeysInRange returns a range of type-specific key components spanning (TkBeg, TkEnd).
func (db *BigTable) KeysInRange(ctx storage.Context, TkBeg, TkEnd storage.TKey) ([]storage.TKey, error) {
if db == nil {
return nil, fmt.Errorf("Can't call KeysInRange() on nil BigTable")
}
if ctx == nil {
return nil, fmt.Errorf("Received nil context in KeysInRange()")
}
tKeys := make([]storage.TKey, 0)
unvKeyBeg, _, err := ctx.SplitKey(TkBeg)
if err != nil {
dvid.Errorf("Error in KeysInRange(): %v\n", err)
}
unvKeyEnd, _, err := ctx.SplitKey(TkEnd)
if err != nil {
dvid.Errorf("Error in KeysInRange(): %v\n", err)
}
rr := api.NewRange(encodeKey(unvKeyBeg), encodeKey(unvKeyEnd))
err = tbl.ReadRows(db.ctx, rr, func(r api.Row) bool {
if len(r[familyName]) == 0 {
dvid.Errorf("Error in KeysInRange(): row has no columns")
return false
}
unvKeyRow, err := decodeKey(r.Key())
if err != nil {
dvid.Errorf("Error in KeysInRange(): %v\n", err)
return false
}
verKeyRow, err := decodeKey(r[familyName][0].Column)
if err != nil {
dvid.Errorf("Error in KeysInRange(): %v\n", err)
return false
}
fullKey := storage.MergeKey(unvKeyRow, verKeyRow)
tkey, err := storage.TKeyFromKey(fullKey)
if err != nil {
dvid.Errorf("Error in KeysInRange(): %v\n", err)
return false
}
tKeys = append(tKeys, tkey)
return true // keep going
}, api.RowFilter(api.StripValueFilter()))
return tKeys, err
}
func doRead(ctx context.Context, args ...string) {
if len(args) < 1 {
log.Fatalf("usage: cbt read <table> [args ...]")
}
tbl := getClient().Open(args[0])
parsed := make(map[string]string)
for _, arg := range args[1:] {
i := strings.Index(arg, "=")
if i < 0 {
log.Fatalf("Bad arg %q", arg)
}
key, val := arg[:i], arg[i+1:]
switch key {
default:
log.Fatalf("Unknown arg key %q", key)
case "limit":
// Be nicer; we used to support this, but renamed it to "end".
log.Fatalf("Unknown arg key %q; did you mean %q?", key, "end")
case "start", "end", "prefix", "count":
parsed[key] = val
}
}
if (parsed["start"] != "" || parsed["end"] != "") && parsed["prefix"] != "" {
log.Fatal(`"start"/"end" may not be mixed with "prefix"`)
}
var rr bigtable.RowRange
if start, end := parsed["start"], parsed["end"]; end != "" {
rr = bigtable.NewRange(start, end)
} else if start != "" {
rr = bigtable.InfiniteRange(start)
}
if prefix := parsed["prefix"]; prefix != "" {
rr = bigtable.PrefixRange(prefix)
}
var opts []bigtable.ReadOption
if count := parsed["count"]; count != "" {
n, err := strconv.ParseInt(count, 0, 64)
if err != nil {
log.Fatalf("Bad count %q: %v", count, err)
}
opts = append(opts, bigtable.LimitRows(n))
}
// TODO(dsymonds): Support filters.
err := tbl.ReadRows(ctx, rr, func(r bigtable.Row) bool {
printRow(r)
return true
}, opts...)
if err != nil {
log.Fatalf("Reading rows: %v", err)
}
}
// GetRange returns a range of values spanning (TkBeg, kEnd) keys.
func (db *BigTable) GetRange(ctx storage.Context, TkBeg, TkEnd storage.TKey) ([]*storage.TKeyValue, error) {
if db == nil {
return nil, fmt.Errorf("Can't call GetRange() on nil BigTable")
}
if ctx == nil {
return nil, fmt.Errorf("Received nil context in GetRange()")
}
unvKeyBeg, _, err := ctx.SplitKey(TkBeg)
if err != nil {
dvid.Errorf("Error in GetRange(): %v\n", err)
}
unvKeyEnd, _, err := ctx.SplitKey(TkEnd)
if err != nil {
dvid.Errorf("Error in GetRange(): %v\n", err)
}
tKeyValues := make([]*storage.TKeyValue, 0)
rr := api.NewRange(encodeKey(unvKeyBeg), encodeKey(unvKeyEnd))
err = tbl.ReadRows(db.ctx, rr, func(r api.Row) bool {
unvKeyRow, err := decodeKey(r.Key())
if err != nil {
dvid.Errorf("Error in GetRange() decodeKey(r.Key()): %v\n", err)
return false
}
// dvid.Infof("GetRange() with row key= %v", r.Key())
for _, readItem := range r[familyName] {
verKey, err := decodeKey(readItem.Column)
if err != nil {
dvid.Errorf("Error in GetRange() decodeKey(readItem.Column): %v\n", err)
return false
}
fullKey := storage.MergeKey(unvKeyRow, verKey)
// dvid.Infof("colum key= %v , timestamp = %v", verKey, readItem.Timestamp)
tkey, err := storage.TKeyFromKey(fullKey)
if err != nil {
dvid.Errorf("Error in GetRange() storage.TKeyFromKey(fullKey): %v\n", err)
return false
}
kv := storage.TKeyValue{tkey, readItem.Value}
tKeyValues = append(tKeyValues, &kv)
}
return true // keep going
})
return tKeyValues, err
}
// SendKeysInRange sends a range of full keys down a key channel.
func (db *BigTable) SendKeysInRange(ctx storage.Context, TkBeg, TkEnd storage.TKey, ch storage.KeyChan) error {
if db == nil {
return fmt.Errorf("Can't call SendKeysInRange() on nil BigTable")
}
if ctx == nil {
return fmt.Errorf("Received nil context in SendKeysInRange()")
}
unvKeyBeg, _, err := ctx.SplitKey(TkBeg)
if err != nil {
dvid.Errorf("Error in SendKeysInRange(): %v\n", err)
}
unvKeyEnd, _, err := ctx.SplitKey(TkEnd)
if err != nil {
dvid.Errorf("Error in SendKeysInRange(): %v\n", err)
}
rr := api.NewRange(encodeKey(unvKeyBeg), encodeKey(unvKeyEnd))
err = tbl.ReadRows(db.ctx, rr, func(r api.Row) bool {
unvKeyRow, err := decodeKey(r.Key())
if err != nil {
dvid.Errorf("Error in SendKeysInRange(): %v\n", err)
return false
}
//I need the versioned key to merged it with the unversioned
// and send it throu the channel
for _, readItem := range r[familyName] {
verKey, err := decodeKey(readItem.Column)
if err != nil {
dvid.Errorf("Error in SendKeysInRange(): %v\n", err)
return false
}
fullKey := storage.MergeKey(unvKeyRow, verKey)
ch <- fullKey
}
return true // keep going
}, api.RowFilter(api.StripValueFilter()))
return err
}
func (db *BigTable) metadataExists() (bool, error) {
if db == nil {
return false, fmt.Errorf("Can't call metadataExists() on nil BigTable")
}
var ctx storage.MetadataContext
unvKeyBeg, unvKeyEnd := ctx.KeyRange()
rr := api.NewRange(encodeKey(unvKeyBeg), encodeKey(unvKeyEnd))
var found bool
err := tbl.ReadRows(db.ctx, rr, func(r api.Row) bool {
found = true
return false
})
return found, err
}
// DeleteRange removes all key-value pairs with keys in the given range.
// For all versions
func (db *BigTable) DeleteRange(ctx storage.Context, TkBeg, TkEnd storage.TKey) error {
if db == nil {
return fmt.Errorf("Can't call DeleteRange() on nil BigTable")
}
if ctx == nil {
return fmt.Errorf("Received nil context in DeleteRange()")
}
unvKeyBeg, _, err := ctx.SplitKey(TkBeg)
if err != nil {
return fmt.Errorf("Error in DeleteRange(): %v\n", err)
}
unvKeyEnd, _, err := ctx.SplitKey(TkEnd)
if err != nil {
return fmt.Errorf("Error in DeleteRange(): %v\n", err)
}
rr := api.NewRange(encodeKey(unvKeyBeg), encodeKey(unvKeyEnd))
err = tbl.ReadRows(db.ctx, rr, func(r api.Row) bool {
unvKeyRow, err := decodeKey(r.Key())
if err != nil {
dvid.Errorf("Error in DeleteRange(): %v\n", err)
return false
}
mut := api.NewMutation()
mut.DeleteRow()
err = tbl.Apply(db.ctx, encodeKey(unvKeyRow), mut)
if err != nil {
dvid.Errorf("Failed to delete row in DeleteRange()")
}
return true // keep going
}, api.RowFilter(api.StripValueFilter()))
return err
}
func main() {
flag.Usage = func() {
fmt.Printf("Usage: scantest [options] <table_name>\n\n")
flag.PrintDefaults()
}
var err error
config, err = cbtconfig.Load()
if err != nil {
log.Fatal(err)
}
config.RegisterFlags()
flag.Parse()
if err := config.CheckFlags(cbtconfig.ProjectAndInstanceRequired); err != nil {
log.Fatal(err)
}
if config.Creds != "" {
os.Setenv("GOOGLE_APPLICATION_CREDENTIALS", config.Creds)
}
if flag.NArg() != 1 {
flag.Usage()
os.Exit(1)
}
table := flag.Arg(0)
log.Printf("Dialing connections...")
client, err = bigtable.NewClient(context.Background(), config.Project, config.Instance)
if err != nil {
log.Fatalf("Making bigtable.Client: %v", err)
}
defer client.Close()
log.Printf("Starting scan test... (run for %v)", *runFor)
tbl := client.Open(table)
sem := make(chan int, *numScans) // limit the number of requests happening at once
var scans stats
stopTime := time.Now().Add(*runFor)
var wg sync.WaitGroup
for time.Now().Before(stopTime) {
sem <- 1
wg.Add(1)
go func() {
defer wg.Done()
defer func() { <-sem }()
ok := true
opStart := time.Now()
defer func() {
scans.Record(ok, time.Since(opStart))
}()
// Start at a random row key
key := fmt.Sprintf("user%d", rand.Int63())
limit := bigtable.LimitRows(int64(*rowLimit))
noop := func(bigtable.Row) bool { return true }
if err := tbl.ReadRows(context.Background(), bigtable.NewRange(key, ""), noop, limit); err != nil {
log.Printf("Error during scan: %v", err)
ok = false
}
}()
}
wg.Wait()
agg := stat.NewAggregate("scans", scans.ds, scans.tries-scans.ok)
log.Printf("Scans (%d ok / %d tries):\nscan times:\n%v\nthroughput (rows/second):\n%v",
scans.ok, scans.tries, agg, throughputString(agg))
}
// DeleteAll removes all key-value pairs for the context. If allVersions is true,
// then all versions of the data instance are deleted.
func (db *BigTable) DeleteAll(ctx storage.Context, allVersions bool) error {
if db == nil {
return fmt.Errorf("Can't call DeleteAll() on nil BigTable")
}
if ctx == nil {
return fmt.Errorf("Received nil context in DeleteAll()")
}
//Row range corresponde to all keys corresponding to this data instace.
min, max := ctx.KeyRange()
rr := api.NewRange(encodeKey(min), encodeKey(max))
err := tbl.ReadRows(db.ctx, rr, func(r api.Row) bool {
unvKeyRow, err := decodeKey(r.Key())
if err != nil {
dvid.Errorf("Error in DeleteAll(): %v\n", err)
return false
}
if allVersions {
mut := api.NewMutation()
mut.DeleteRow()
err := tbl.Apply(db.ctx, encodeKey(unvKeyRow), mut)
if err != nil {
dvid.Errorf("Failed to delete row")
}
} else {
emptyTkey := make([]byte, 0)
_, versionToDelete, err := ctx.SplitKey(emptyTkey)
if err != nil {
dvid.Errorf("Error in DeleteAll(): %v\n", err)
return false
}
for _, readItem := range r[familyName] {
verKey, err := decodeKey(readItem.Column)
if err != nil {
dvid.Errorf("Error in DeleteAll(): %v\n", err)
return false
}
if bytes.Equal(verKey, versionToDelete) {
mut := api.NewMutation()
mut.DeleteCellsInColumn(familyName, encodeKey(verKey))
err := tbl.Apply(db.ctx, encodeKey(unvKeyRow), mut)
if err != nil {
dvid.Errorf("Failed to DeleteCellsInColumn in DeleteAll()")
}
return true // One I found the version I don't have to keep serching for it.
}
}
}
return true // keep going
}, api.RowFilter(api.StripValueFilter()))
return err
}
// RawRangeQuery sends a range of full keys. This is to be used for low-level data
// retrieval like DVID-to-DVID communication and should not be used by data type
// implementations if possible because each version's key-value pairs are sent
// without filtering by the current version and its ancestor graph. A nil is sent
// down the channel when the range is complete.
func (db *BigTable) RawRangeQuery(kStart, kEnd storage.Key, keysOnly bool, out chan *storage.KeyValue, cancel <-chan struct{}) error {
if db == nil {
return fmt.Errorf("Can't call RawRangeQuery() on nil BigTable")
}
unvKeyBeg, verKeyBeg, err := storage.SplitKey(kStart)
if err != nil {
dvid.Errorf("Error in RawRangeQuery(): %v\n", err)
return err
}
unvKeyEnd, verKeyEnd, err := storage.SplitKey(kEnd)
if err != nil {
dvid.Errorf("Error in RawRangeQuery(): %v\n", err)
return err
}
rr := api.NewRange(encodeKey(unvKeyBeg), encodeKey(unvKeyEnd))
err = tbl.ReadRows(db.ctx, rr, func(r api.Row) bool {
if cancel != nil {
select {
case <-cancel:
out <- nil
return nil
default:
}
}
unvKeyRow, err := decodeKey(r.Key())
if err != nil {
dvid.Errorf("Error in RawRangeQuery(): %v\n", err)
return false
}
//I need the versioned key to merged it with the unversioned
// and send it throu the channel
for _, readItem := range r[familyName] {
verKey, err := decodeKey(readItem.Column)
if err != nil {
dvid.Errorf("Error in RawRangeQuery(): %v\n", err)
return false
}
lowerLimit := bytes.Equal(unvKeyBeg, unvKeyRow) && bytes.Compare(verKey, verKeyBeg) == -1
upperLimit := bytes.Equal(unvKeyEnd, unvKeyRow) && bytes.Compare(verKey, verKeyEnd) >= 0
if lowerLimit || upperLimit {
continue
}
fullKey := storage.MergeKey(unvKeyRow, verKey)
kv := storage.KeyValue{fullKey, readItem.Value}
out <- &kv
}
return true // keep going
})
return nil
}
// ProcessRange sends a range of type key-value pairs to type-specific chunk handlers,
// allowing chunk processing to be concurrent with key-value sequential reads.
// Since the chunks are typically sent during sequential read iteration, the
// receiving function can be organized as a pool of chunk handling goroutines.
// See datatype/imageblk.ProcessChunk() for an example.
func (db *BigTable) ProcessRange(ctx storage.Context, TkBeg, TkEnd storage.TKey, op *storage.ChunkOp, f storage.ChunkFunc) error {
if db == nil {
return fmt.Errorf("Can't call ProcessRange() on nil BigTable")
}
if ctx == nil {
return fmt.Errorf("Received nil context in ProcessRange()")
}
unvKeyBeg, verKey, err := ctx.SplitKey(TkBeg)
if err != nil {
dvid.Errorf("Error in ProcessRange(): %v\n", err)
}
unvKeyEnd, _, err := ctx.SplitKey(TkEnd)
if err != nil {
dvid.Errorf("Error in ProcessRange(): %v\n", err)
}
rr := api.NewRange(encodeKey(unvKeyBeg), encodeKey(unvKeyEnd))
err = tbl.ReadRows(db.ctx, rr, func(r api.Row) bool {
if len(r[familyName]) == 0 {
dvid.Errorf("Error in KeysInRange(): row has no columns")
return false
}
unvKeyRow, err := decodeKey(r.Key())
if err != nil {
dvid.Errorf("Error in ProcessRange(): %v\n", err)
return false
}
verKeyRow, err := decodeKey(r[familyName][0].Column)
if err != nil {
dvid.Errorf("Error in ProcessRange(): %v\n", err)
return false
}
fullKey := storage.MergeKey(unvKeyRow, verKeyRow)
tkey, err := storage.TKeyFromKey(fullKey)
if err != nil {
dvid.Errorf("Error in ProcessRange(): %v\n", err)
return false
}
if op.Wg != nil {
op.Wg.Add(1)
}
value, err := getValue(r, verKey)
if err != nil {
dvid.Errorf("Error in ProcessRange(): %v\n", err)
return false
}
tkv := storage.TKeyValue{tkey, value}
chunk := &storage.Chunk{op, &tkv}
if err := f(chunk); err != nil {
dvid.Errorf("Error in ProcessRange(): %v\n", err)
return false
}
return true // keep going
})
return err
}
