Small cleanup.

[remoteglot-book] / build-book.sh

diff --git a/build-book.sh b/build-book.sh
index 01c51c7da920e5fb606b2e637f80ec06392b082e..772f3fb567d5f67e65c0d8593e3d2c65239d2d0e 100755 (executable)
--- a/build-book.sh
+++ b/build-book.sh
@@ -2,28 +2,54 @@
 
 set -e
 
-rm -f part-*.bin part-*.mtbl part-*.mtbl.part???? open.mtbl.new open.mtbl.part???? open.mtbl.part????.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
-done
-date 
+rm -f pgnnames.txt tmp.mtbl* open.mtbl.new open.mtbl.part???? open.mtbl.part????.new 2>/dev/null
 
-for FILE in part-*.bin; do
-       ( ./binloader $FILE ${FILE/bin/mtbl} 40 ) &
+for FILE in "$@"; do
+       echo "$FILE" >> pgnnames.txt
 done
-wait
 
-rm -f part-*.bin
+TS_BEGIN=$( date +%s%3N )
 
 for X in $( seq 0 39 ); do
-       ( ./binmerger part-*.mtbl.part$( printf %04d $X ) open.mtbl.part$( printf %04d $X ).new ) &
+       (
+               (
+                       PGNNUM=0
+                       for FILE in "$@"; 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
 
-for X in $( seq 0 39 ); do
-       mv open.mtbl.part$( printf %04d $X ).new open.mtbl.part$( printf %04d $X)
+TS_MAPDONE=$( date +%s%3N )
+MAPOUTPUT=$( du -sch tmp.mtbl.mapper* | tail -n 1 | cut -f1 | sed 's/\([^0-9]\)$/ \1/' )
+
+echo "Reducing..."
+
+parallel -i -j ${PARALLEL_MERGES} bash -c 'X={}; ./binmerger tmp.mtbl.mapper*.part$( printf %04d $X ).subshard* open.mtbl.part$( printf %04d $X ).new' -- $( seq 0 $(( SHARDS - 1 )) )
+
+TS_REDUCEDONE=$( 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"