Store and handle the PGN start position.

[remoteglot-book] / build-book.sh

diff --git a/build-book.sh b/build-book.sh
index 1d5e3b50990c6b3a71567712754e9e5665af4650..7d49663f5fb6f2c90a2bcff821287c6d130b0db2 100755 (executable)
--- a/build-book.sh
+++ b/build-book.sh
@@ -2,22 +2,26 @@
 
 set -e
 
-rm -f part-*.bin part-*.mtbl open.mtbl.new 2>/dev/null
+export SHARDS=40
+export PARALLEL_LOADS=20  # Reduce if you have problems with OOM
 
-for FILE in $@; do
+rm -f part-*.bin part-*.mtbl part-*.mtbl.part???? open.mtbl.new open.mtbl.part???? open.mtbl.part????.new 2>/dev/null
+
+for FILE in "$@"; do
        date | tr -d "\n"
        echo "  $FILE"
-       ./parallel-parse-pgn.sh $FILE
+       ./parallel-parse-pgn.sh "$FILE"
 done
 date 
 
-for FILE in part-*.bin; do
-       ( ./binloader $FILE ${FILE/bin/mtbl} ) &
-done
-wait
+echo "Loading..."
+parallel -i -j ${PARALLEL_LOADS} bash -c 'FILE="{}"; set -x; ./binloader "$FILE" "${FILE/bin/mtbl}" $SHARDS' -- part-*.bin
 
 rm -f part-*.bin
 
-./binmerger part-*.mtbl open.mtbl.new
-mv open.mtbl.new open.mtbl
-rm -f part-*.mtbl 
+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 )) )
+
+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????