From ac8e070a93cc50cbb3a74695b5d45426c4808eb0 Mon Sep 17 00:00:00 2001
From: Tom Clegg <tom@tomclegg.ca>
Date: Fri, 25 Sep 2020 16:29:23 -0400
Subject: [PATCH] Option to treat tiles with no-calls as regular tiles.

Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@tomclegg.ca>
---
 import.go  | 15 +++++++++++++--
 tilelib.go | 19 +++++++++++--------
 2 files changed, 24 insertions(+), 10 deletions(-)

diff --git a/import.go b/import.go
index 3ce00ce19c..e24b25dfac 100644
--- a/import.go
+++ b/import.go
@@ -33,6 +33,7 @@ type importer struct {
 	runLocal       bool
 	skipOOO        bool
 	outputTiles    bool
+	includeNoCalls bool
 	encoder        *gob.Encoder
 }
 
@@ -52,6 +53,7 @@ func (cmd *importer) RunCommand(prog string, args []string, stdin io.Reader, std
 	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")
+	flags.BoolVar(&cmd.includeNoCalls, "include-no-calls", false, "treat tiles with no-calls as regular tiles")
 	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)")
@@ -110,7 +112,16 @@ func (cmd *importer) RunCommand(prog string, args []string, stdin io.Reader, std
 			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, fmt.Sprintf("-output-tiles=%v", cmd.outputTiles), "-o", cmd.outputFile}, inputs...)
+		runner.Args = append([]string{"import",
+			"-local=true",
+			"-loglevel=" + *loglevel,
+			fmt.Sprintf("-skip-ooo=%v", cmd.skipOOO),
+			fmt.Sprintf("-output-tiles=%v", cmd.outputTiles),
+			fmt.Sprintf("-include-no-calls=%v", cmd.includeNoCalls),
+			"-tag-library", cmd.tagLibraryFile,
+			"-ref", cmd.refFile,
+			"-o", cmd.outputFile,
+		}, inputs...)
 		var output string
 		output, err = runner.Run()
 		if err != nil {
@@ -143,7 +154,7 @@ func (cmd *importer) RunCommand(prog string, args []string, stdin io.Reader, std
 	bufw := bufio.NewWriter(output)
 	cmd.encoder = gob.NewEncoder(bufw)
 
-	tilelib := &tileLibrary{taglib: taglib, skipOOO: cmd.skipOOO}
+	tilelib := &tileLibrary{taglib: taglib, includeNoCalls: cmd.includeNoCalls, skipOOO: cmd.skipOOO}
 	if cmd.outputTiles {
 		tilelib.encoder = cmd.encoder
 	}
diff --git a/tilelib.go b/tilelib.go
index ca3857566f..47de7b048f 100644
--- a/tilelib.go
+++ b/tilelib.go
@@ -46,9 +46,10 @@ func (tseq tileSeq) Variants() ([]tileVariantID, int, int) {
 }
 
 type tileLibrary struct {
-	skipOOO bool
-	taglib  *tagLibrary
-	variant [][][blake2b.Size256]byte
+	includeNoCalls bool
+	skipOOO        bool
+	taglib         *tagLibrary
+	variant        [][][blake2b.Size256]byte
 	// count [][]int
 	// seq map[[blake2b.Size]byte][]byte
 	variants int
@@ -148,11 +149,13 @@ func (tilelib *tileLibrary) Len() int {
 // Return a tileLibRef for a tile with the given tag and sequence,
 // adding the sequence to the library if needed.
 func (tilelib *tileLibrary) getRef(tag tagID, seq []byte) tileLibRef {
-	for _, b := range seq {
-		if b != 'a' && b != 'c' && b != 'g' && b != 't' {
-			// return "tile not found" if seq has any
-			// no-calls
-			return tileLibRef{tag: tag}
+	if !tilelib.includeNoCalls {
+		for _, b := range seq {
+			if b != 'a' && b != 'c' && b != 'g' && b != 't' {
+				// return "tile not found" if seq has any
+				// no-calls
+				return tileLibRef{tag: tag}
+			}
 		}
 	}
 	tilelib.mtx.Lock()
-- 
2.30.2