1 // Copyright (C) The Lightning Authors. All rights reserved.
3 // SPDX-License-Identifier: AGPL-3.0
20 "git.arvados.org/arvados.git/sdk/go/arvados"
21 "github.com/klauspost/pgzip"
22 log "github.com/sirupsen/logrus"
25 type slicecmd struct{}
27 func (cmd *slicecmd) RunCommand(prog string, args []string, stdin io.Reader, stdout, stderr io.Writer) int {
31 fmt.Fprintf(stderr, "%s\n", err)
34 flags := flag.NewFlagSet("", flag.ContinueOnError)
35 flags.SetOutput(stderr)
36 pprof := flags.String("pprof", "", "serve Go profile data at http://`[addr]:port`")
37 runlocal := flags.Bool("local", false, "run on local host (default: run in an arvados container)")
38 projectUUID := flags.String("project", "", "project `UUID` for output data")
39 priority := flags.Int("priority", 500, "container request priority")
40 inputDir := flags.String("input-dir", "./in", "input `directory`")
41 outputDir := flags.String("output-dir", "./out", "output `directory`")
42 tagsPerFile := flags.Int("tags-per-file", 50000, "tags per file (nfiles will be ~10M÷x)")
43 err = flags.Parse(args)
44 if err == flag.ErrHelp {
47 } else if err != nil {
53 log.Println(http.ListenAndServe(*pprof, nil))
58 runner := arvadosContainerRunner{
59 Name: "lightning slice",
60 Client: arvados.NewClientFromEnv(),
61 ProjectUUID: *projectUUID,
68 err = runner.TranslatePaths(inputDir)
72 runner.Args = []string{"slice", "-local=true",
74 "-input-dir", *inputDir,
75 "-output-dir", "/mnt/output",
78 output, err = runner.Run()
82 fmt.Fprintln(stdout, output)
86 err = Slice(*outputDir, *inputDir, *tagsPerFile)
93 // Read tags+tiles+genomes from srcdir, write to dstdir with (up to)
94 // the specified number of tags per file.
95 func Slice(dstdir, srcdir string, tagsPerFile int) error {
96 infiles, err := allGobFiles(srcdir)
105 bufws []*bufio.Writer
110 throttle := throttle{Max: runtime.GOMAXPROCS(0)}
111 for _, path := range infiles {
115 defer throttle.Release()
122 log.Printf("reading %s", path)
123 err = DecodeLibrary(f, strings.HasSuffix(path, ".gz"), func(ent *LibraryEntry) error {
124 if err := throttle.Err(); err != nil {
127 if len(ent.TagSet) > 0 {
128 tagsetOnce.Do(func() {
131 fs, bufws, gzws, encs, err = openOutFiles(dstdir, len(ent.TagSet), tagsPerFile)
136 for _, enc := range encs {
137 err = enc.Encode(LibraryEntry{TagSet: tagset})
145 if err := throttle.Err(); err != nil {
148 for _, tv := range ent.TileVariants {
149 err := encs[int(tv.Tag)/tagsPerFile].Encode(LibraryEntry{
150 TileVariants: []TileVariant{tv},
156 // Here, each output file gets a
157 // CompactGenome entry for each
158 // genome, even if there are no
159 // variants in the relevant range.
160 // Easier for downstream code.
161 for _, cg := range ent.CompactGenomes {
162 for i, enc := range encs {
163 start := i * tagsPerFile
164 end := start + tagsPerFile
165 if max := len(cg.Variants)/2 + int(cg.StartTag); end > max {
168 if start < int(cg.StartTag) {
169 start = int(cg.StartTag)
171 var variants []tileVariantID
173 variants = cg.Variants[(start-int(cg.StartTag))*2 : (end-int(cg.StartTag))*2]
175 err := enc.Encode(LibraryEntry{CompactGenomes: []CompactGenome{{
178 StartTag: tagID(start),
179 EndTag: tagID(start + tagsPerFile),
186 // Write all ref seqs to the first
187 // slice. Easier for downstream code.
188 if len(ent.CompactSequences) > 0 {
189 err := encs[0].Encode(LibraryEntry{CompactSequences: ent.CompactSequences})
200 if throttle.Err() != nil {
201 closeOutFiles(fs, bufws, gzws, encs)
202 return throttle.Err()
204 return closeOutFiles(fs, bufws, gzws, encs)
207 func openOutFiles(dstdir string, tags, tagsPerFile int) (fs []*os.File, bufws []*bufio.Writer, gzws []*pgzip.Writer, encs []*gob.Encoder, err error) {
208 nfiles := (tags + tagsPerFile - 1) / tagsPerFile
209 fs = make([]*os.File, nfiles)
210 bufws = make([]*bufio.Writer, nfiles)
211 gzws = make([]*pgzip.Writer, nfiles)
212 encs = make([]*gob.Encoder, nfiles)
213 for i := 0; i*tagsPerFile < tags; i++ {
214 fs[i], err = os.Create(dstdir + fmt.Sprintf("/library%04d.gob.gz", i))
218 bufws[i] = bufio.NewWriterSize(fs[i], 1<<26)
219 gzws[i] = pgzip.NewWriter(bufws[i])
220 encs[i] = gob.NewEncoder(gzws[i])
225 func closeOutFiles(fs []*os.File, bufws []*bufio.Writer, gzws []*pgzip.Writer, encs []*gob.Encoder) error {
227 for _, gzw := range gzws {
230 if err != nil && firstErr == nil {
235 for _, bufw := range bufws {
238 if err != nil && firstErr == nil {
243 for _, f := range fs {
246 if err != nil && firstErr == nil {