set -e
-rm -f part-*.bin part-*.mtbl open.mtbl.new 2>/dev/null
+export SHARDS=40
+export PARALLEL_MERGES=40
-for FILE in $@; do
- date | tr -d "\n"
- echo " $FILE"
- ./parallel-parse-pgn.sh $FILE
+rm -f pgnnames.txt tmp.mtbl* open.mtbl.new open.mtbl.part???? open.mtbl.part????.new 2>/dev/null
+
+for FILE in "$@"; do
+ echo "$FILE" >> pgnnames.txt
done
-date
-for FILE in part-*.bin; do
- ( ./binloader $FILE ${FILE/bin/mtbl} ) &
+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
-rm -f part-*.bin
+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 )) )
-./binmerger part-*.mtbl open.mtbl.new
-mv open.mtbl.new open.mtbl
-rm -f part-*.mtbl
+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*