From 2d1b6eb5c7adf27029ad3f900b9e357c2e66e105 Mon Sep 17 00:00:00 2001
From: Tom Clegg <tom@tomclegg.ca>
Date: Tue, 21 Sep 2021 09:59:03 -0400
Subject: [PATCH] Log total variant/genome/ref counts.

refs #17966

Arvados-DCO-1.1-Signed-off-by: Tom Clegg <tom@curii.com>
---
 slice.go | 9 +++++++++
 1 file changed, 9 insertions(+)

diff --git a/slice.go b/slice.go
index 50f219fb3c..10e5e01bc2 100644
--- a/slice.go
+++ b/slice.go
@@ -16,6 +16,7 @@ import (
 	"runtime"
 	"strings"
 	"sync"
+	"sync/atomic"
 
 	"git.arvados.org/arvados.git/sdk/go/arvados"
 	"github.com/klauspost/pgzip"
@@ -105,6 +106,10 @@ func Slice(dstdir, srcdir string, tagsPerFile int) error {
 		bufws      []*bufio.Writer
 		gzws       []*pgzip.Writer
 		encs       []*gob.Encoder
+
+		countTileVariants int64
+		countGenomes      int64
+		countReferences   int64
 	)
 
 	throttle := throttle{Max: runtime.GOMAXPROCS(0)}
@@ -145,6 +150,7 @@ func Slice(dstdir, srcdir string, tagsPerFile int) error {
 				if err := throttle.Err(); err != nil {
 					return err
 				}
+				atomic.AddInt64(&countTileVariants, int64(len(ent.TileVariants)))
 				for _, tv := range ent.TileVariants {
 					err := encs[int(tv.Tag)/tagsPerFile].Encode(LibraryEntry{
 						TileVariants: []TileVariant{tv},
@@ -158,6 +164,7 @@ func Slice(dstdir, srcdir string, tagsPerFile int) error {
 				// genome, even if there are no
 				// variants in the relevant range.
 				// Easier for downstream code.
+				atomic.AddInt64(&countGenomes, int64(len(ent.CompactGenomes)))
 				for _, cg := range ent.CompactGenomes {
 					for i, enc := range encs {
 						start := i * tagsPerFile
@@ -185,6 +192,7 @@ func Slice(dstdir, srcdir string, tagsPerFile int) error {
 				}
 				// Write all ref seqs to the first
 				// slice. Easier for downstream code.
+				atomic.AddInt64(&countReferences, int64(len(ent.CompactSequences)))
 				if len(ent.CompactSequences) > 0 {
 					err := encs[0].Encode(LibraryEntry{CompactSequences: ent.CompactSequences})
 					if err != nil {
@@ -201,6 +209,7 @@ func Slice(dstdir, srcdir string, tagsPerFile int) error {
 		closeOutFiles(fs, bufws, gzws, encs)
 		return throttle.Err()
 	}
+	defer log.Printf("Total %d tile variants, %d genomes, %d reference sequences", countTileVariants, countGenomes, countReferences)
 	return closeOutFiles(fs, bufws, gzws, encs)
 }
 
-- 
2.30.2