X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=build-book.sh;h=772f3fb567d5f67e65c0d8593e3d2c65239d2d0e;hb=f95dfda2d14615a9816ba1abb32bd240a470e228;hp=7d49663f5fb6f2c90a2bcff821287c6d130b0db2;hpb=4a7c263734f23fc6b2ef0e6f8adb1ce344a22053;p=remoteglot-book

diff --git a/build-book.sh b/build-book.sh
index 7d49663..772f3fb 100755
--- a/build-book.sh
+++ b/build-book.sh
@@ -3,25 +3,53 @@
 set -e
 
 export SHARDS=40
-export PARALLEL_LOADS=20  # Reduce if you have problems with OOM
+export PARALLEL_MERGES=40
 
-rm -f part-*.bin part-*.mtbl part-*.mtbl.part???? open.mtbl.new open.mtbl.part???? open.mtbl.part????.new 2>/dev/null
+rm -f pgnnames.txt tmp.mtbl* open.mtbl.new open.mtbl.part???? open.mtbl.part????.new 2>/dev/null
 
 for FILE in "$@"; do
-	date | tr -d "\n"
-	echo "  $FILE"
-	./parallel-parse-pgn.sh "$FILE"
+	echo "$FILE" >> pgnnames.txt
 done
-date 
 
-echo "Loading..."
-parallel -i -j ${PARALLEL_LOADS} bash -c 'FILE="{}"; set -x; ./binloader "$FILE" "${FILE/bin/mtbl}" $SHARDS' -- part-*.bin
+TS_BEGIN=$( date +%s%3N )
+
+for X in $( seq 0 39 ); do
+	(
+		(
+			PGNNUM=0
+			for FILE in "$@"; 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
+				PGNNUM=$(( PGNNUM + 1 ))
+			done
+		) | ./binloader - tmp.mtbl.mapper$X $SHARDS 500000
+	) &
+done
+wait
+
+TS_MAPDONE=$( date +%s%3N )
+MAPOUTPUT=$( du -sch tmp.mtbl.mapper* | tail -n 1 | cut -f1 | sed 's/\([^0-9]\)$/ \1/' )
+
+echo "Reducing..."
 
-rm -f part-*.bin
+parallel -i -j ${PARALLEL_MERGES} bash -c 'X={}; ./binmerger tmp.mtbl.mapper*.part$( printf %04d $X ).subshard* open.mtbl.part$( printf %04d $X ).new' -- $( seq 0 $(( SHARDS - 1 )) )
 
-parallel -i -j ${PARALLEL_LOADS} bash -c 'X={}; set -x; ./binmerger part-*.mtbl.part$( printf %04d $X ) open.mtbl.part$( 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 $X )
 done
-rm -f part-*.mtbl.part????
+rm -f tmp.mtbl*
+
+TS_END=$( 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"