From c9784fb76a27f0a2a4c0bfb303ab25da49ed8846 Mon Sep 17 00:00:00 2001
From: Erich Eckner <git@eckner.net>
Date: Tue, 25 Oct 2016 13:34:04 +0200
Subject: split stage 3 into two (generate; sort) in backupStatistics

---
 backupStatistics.in | 22 +++++++++++++++-------
 1 file changed, 15 insertions(+), 7 deletions(-)

(limited to 'backupStatistics.in')

diff --git a/backupStatistics.in b/backupStatistics.in
index 1f3c8f7..f910db4 100644
--- a/backupStatistics.in
+++ b/backupStatistics.in
@@ -51,25 +51,33 @@ do_stage()
         echo 'generate lists $inode -> $contentHash'
         return 0
       fi
-      tmpDirA="$(mktemp -d)"
-      tmpDirB="$(mktemp -d "${cacheDir}/tmp.XXXXXX")"
       uniq -m1 "${cacheDir}/${backupID}.inodes.sorted" | \
         parallel \
           sha512sum {=s/^[[:digit:]]\+ //=} \| \
             sed "\"s|^\([0-9a-f]\{128\}\)  .*\$|\1 "{=s/^\([[:digit:]]\+\) .*/\\1/=}"|\"" \
-          \; | \
-        sort -T "${tmpDirA}" -T "${tmpDirB}" > \
+          \; > \
         "${cacheDir}/${backupID}.content"
-      rmdir "${tmpDirA}" "${tmpDirB}"
     ;;
     4)
+      if [ "$2" == '##DESCRIBE##' ]
+      then
+        echo 'sort previous lists by $contentHash'
+        return 0
+      fi
+      tmpDirA="$(mktemp -d)"
+      tmpDirB="$(mktemp -d "${cacheDir}/tmp.XXXXXX")"
+      sort -T "${tmpDirA}" -T "${tmpDirB}" "${cacheDir}/${backupID}.content" > \
+        "${cacheDir}/${backupID}.content.sorted"
+      rmdir "${tmpDirA}" "${tmpDirB}"
+    ;;
+    5)
       if [ "$2" == '##DESCRIBE##' ]
       then
         echo 'find duplicate hashes'
         return 0
       fi
       (
-        uniq -m1 --all-repeated=separate "${cacheDir}/${backupID}.content"
+        uniq -m1 --all-repeated=separate "${cacheDir}/${backupID}.content.sorted"
         echo ""
       ) | \
         sed 's|^\S\+ ||' | \
@@ -85,7 +93,7 @@ do_stage()
         sed 's| $||' > \
         "${cacheDir}/${backupID}.duplicates"
     ;;
-    5)
+    6)
       if [ "$2" == '##DESCRIBE##' ]
       then
         echo 'remove inodes with duplicate hashes'
-- 
cgit v1.2.3-70-g09d2