X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=build-book.sh;h=94f3a298ef6c655f7c296fe24d7b3f1b4da7c462;hb=74f1261e4067cbc1b743b329ed2d419aee735cbf;hp=67181c68167bdf555bb4157d05fc5c2aa829dbc8;hpb=756a91677c1565b1bd555722ea795f911106104f;p=remoteglot-book

diff --git a/build-book.sh b/build-book.sh
index 67181c6..94f3a29 100755
--- a/build-book.sh
+++ b/build-book.sh
@@ -2,23 +2,31 @@
 
 set -e
 
+export MAPPERS=40
+export REDUCERS=40
 export SHARDS=40
-export PARALLEL_MERGES=40
 
-rm -f pgnnames.txt tmp.mtbl* open.mtbl.new open.mtbl.part???? open.mtbl.part????.new 2>/dev/null
+rm -f pgnnames.txt tmp.mtbl* open.mtbl.part????.new 2>/dev/null
 
 for FILE in "$@"; do
 	echo "$FILE" >> pgnnames.txt
 done
 
-for X in $( seq 0 39 ); do
+TS_BEGIN=$( date +%s%3N )
+
+for X in $( seq 0 $(( MAPPERS - 1 )) ); do
 	(
 		(
 			PGNNUM=0
 			for FILE in "$@"; do
-				START=$( ./find-pgn-split-point.sh "$FILE" $X 40 )
-				END=$( ./find-pgn-split-point.sh "$FILE" $(( X + 1 )) 40 )
-				~/nmu/pgn-extract/pgn-extract --startpos $START --endpos $END --startfilenum $PGNNUM -e -Wsessebin "$FILE" 2>/dev/null
+				COMP=0
+				if echo "$FILE" | grep -q ^comp:; then
+					COMP=1
+					FILE=$( echo "$FILE" | sed s/comp:// )
+				fi
+				START=$( ./find-pgn-split-point.sh "$FILE" $X $MAPPERS )
+				END=$( ./find-pgn-split-point.sh "$FILE" $(( X + 1 )) $MAPPERS )
+				~/nmu/pgn-extract/pgn-extract --computerflag $COMP --startpos $START --endpos $END --startfilenum $PGNNUM -e -Wsessebin "$FILE" 2>/dev/null
 				PGNNUM=$(( PGNNUM + 1 ))
 			done
 		) | ./binloader - tmp.mtbl.mapper$X $SHARDS 500000
@@ -26,9 +34,28 @@ for X in $( seq 0 39 ); do
 done
 wait
 
-parallel -i -j ${PARALLEL_MERGES} bash -c 'X={}; set -x; ./binmerger tmp.mtbl.mapper*.part$( printf %04d $X ).subshard* open.mtbl.part$( printf %04d $X ).new' -- $( seq 0 $(( SHARDS - 1 )) )
+TS_MAPDONE=$( date +%s%3N )
+MAPOUTPUT=$( du -sch tmp.mtbl.mapper* | tail -n 1 | cut -f1 | sed 's/\([^0-9]\)$/ \1/' )
+
+echo "Reducing..."
+
+parallel -i -j $REDUCERS bash -c 'X={}; ./binmerger tmp.mtbl.mapper*.part$( printf %04d $X ).subshard* open.mtbl.part$( printf %04d $X ).new' -- $( seq 0 $(( SHARDS - 1 )) )
+
+TS_REDUCEDONE=$( date +%s%3N )
+REDUCEOUTPUT=$( du -sch open.mtbl.part*.new | tail -n 1 | cut -f1 | sed 's/\([^0-9]\)$/ \1/' )
 
 for X in $( seq 0 $(( SHARDS - 1 )) ); do
 	mv open.mtbl.part$( printf %04d $X ).new open.mtbl.part$( printf %04d $X )
 done
 rm -f tmp.mtbl*
+
+TS_END=$( date +%s%3N )
+
+printf "\n"
+printf "  Map time:       %7d ms\n" $(( TS_MAPDONE - TS_BEGIN ))
+printf "  Reduce time:    %7d ms\n" $(( TS_REDUCEDONE - TS_MAPDONE ))
+printf "  Total time:     %7d ms\n" $(( TS_END - TS_BEGIN ))
+printf "\n"
+printf "  Map output:       %7sB\n" "$MAPOUTPUT"
+printf "  Reduce output:    %7sB\n" "$REDUCEOUTPUT"
+printf "\n"