projectUUID string
runLocal bool
skipOOO bool
+ outputTiles bool
encoder *gob.Encoder
}
flags.StringVar(&cmd.projectUUID, "project", "", "project `UUID` for output data")
flags.BoolVar(&cmd.runLocal, "local", false, "run on local host (default: run in an arvados container)")
flags.BoolVar(&cmd.skipOOO, "skip-ooo", false, "skip out-of-order tags")
+ flags.BoolVar(&cmd.outputTiles, "output-tiles", false, "include tile variant sequences in output file")
priority := flags.Int("priority", 500, "container request priority")
pprof := flags.String("pprof", "", "serve Go profile data at http://`[addr]:port`")
loglevel := flags.String("loglevel", "info", "logging threshold (trace, debug, info, warn, error, fatal, or panic)")
err = errors.New("cannot specify output file in container mode: not implemented")
return 1
}
- runner.Args = append([]string{"import", "-local=true", "-loglevel=" + *loglevel, fmt.Sprintf("-skip-ooo=%v", cmd.skipOOO), "-tag-library", cmd.tagLibraryFile, "-ref", cmd.refFile, "-o", cmd.outputFile}, inputs...)
+ runner.Args = append([]string{"import", "-local=true", "-loglevel=" + *loglevel, fmt.Sprintf("-skip-ooo=%v", cmd.skipOOO), "-tag-library", cmd.tagLibraryFile, "-ref", cmd.refFile, fmt.Sprintf("-output-tiles=%v", cmd.outputTiles), "-o", cmd.outputFile}, inputs...)
var output string
output, err = runner.Run()
if err != nil {
return 1
}
- tilelib, err := cmd.loadTileLibrary()
+ taglib, err := cmd.loadTagLibrary()
if err != nil {
return 1
}
- go func() {
- for range time.Tick(10 * time.Minute) {
- log.Printf("tilelib.Len() == %d", tilelib.Len())
- }
- }()
var output io.WriteCloser
if cmd.outputFile == "-" {
bufw := bufio.NewWriter(output)
cmd.encoder = gob.NewEncoder(bufw)
+ tilelib := &tileLibrary{taglib: taglib, skipOOO: cmd.skipOOO}
+ if cmd.outputTiles {
+ tilelib.encoder = cmd.encoder
+ }
+ go func() {
+ for range time.Tick(10 * time.Minute) {
+ log.Printf("tilelib.Len() == %d", tilelib.Len())
+ }
+ }()
+
err = cmd.tileInputs(tilelib, infiles)
if err != nil {
return 1
return tilelib.TileFasta(infile, input)
}
-func (cmd *importer) loadTileLibrary() (*tileLibrary, error) {
+func (cmd *importer) loadTagLibrary() (*tagLibrary, error) {
log.Printf("tag library %s load starting", cmd.tagLibraryFile)
f, err := os.Open(cmd.tagLibraryFile)
if err != nil {
return nil, fmt.Errorf("cannot tile: tag library is empty")
}
log.Printf("tag library %s load done", cmd.tagLibraryFile)
- return &tileLibrary{taglib: &taglib, skipOOO: cmd.skipOOO}, nil
+ return &taglib, nil
}
func listInputFiles(paths []string) (files []string, err error) {
import (
"bufio"
"bytes"
+ "encoding/gob"
"io"
"strings"
"sync"
// count [][]int
// seq map[[blake2b.Size]byte][]byte
variants int
+ // if non-nil, write out any tile variants added while tiling
+ encoder *gob.Encoder
mtx sync.Mutex
}
}
}
tilelib.mtx.Lock()
- defer tilelib.mtx.Unlock()
// if tilelib.seq == nil {
// tilelib.seq = map[[blake2b.Size]byte][]byte{}
// }
seqhash := blake2b.Sum256(seq)
for i, varhash := range tilelib.variant[tag] {
if varhash == seqhash {
+ tilelib.mtx.Unlock()
return tileLibRef{tag: tag, variant: tileVariantID(i + 1)}
}
}
tilelib.variants++
tilelib.variant[tag] = append(tilelib.variant[tag], seqhash)
// tilelib.seq[seqhash] = append([]byte(nil), seq...)
- return tileLibRef{tag: tag, variant: tileVariantID(len(tilelib.variant[tag]))}
+ ret := tileLibRef{tag: tag, variant: tileVariantID(len(tilelib.variant[tag]))}
+ tilelib.mtx.Unlock()
+
+ if tilelib.encoder != nil {
+ tilelib.encoder.Encode(LibraryEntry{
+ TileVariants: []TileVariant{{
+ Tag: tag,
+ Blake2b: seqhash,
+ Sequence: seq,
+ }},
+ })
+ }
+ return ret
}