X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=build-book.sh;h=67181c68167bdf555bb4157d05fc5c2aa829dbc8;hb=fc7cab1217dc904a888240e577f4da5001cf4bdb;hp=7d49663f5fb6f2c90a2bcff821287c6d130b0db2;hpb=4a7c263734f23fc6b2ef0e6f8adb1ce344a22053;p=remoteglot-book diff --git a/build-book.sh b/build-book.sh index 7d49663..67181c6 100755 --- a/build-book.sh +++ b/build-book.sh @@ -3,25 +3,32 @@ 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 - -rm -f part-*.bin +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 -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 )) ) +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 )) ) 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*