git.sesse.net Git - remoteglot-book/blob - build-book.sh

   1 #! /bin/bash
   2
   3 set -e
   4
   5 export SHARDS=40
   6 export PARALLEL_MERGES=40
   7
   8 rm -f pgnnames.txt tmp.mtbl* open.mtbl.new open.mtbl.part???? open.mtbl.part????.new 2>/dev/null
   9
  10 for FILE in "$@"; do
  11         echo "$FILE" >> pgnnames.txt
  12 done
  13
  14 for X in $( seq 0 39 ); do
  15         (
  16                 (
  17                         PGNNUM=0
  18                         for FILE in "$@"; do
  19                                 START=$( ./find-pgn-split-point.sh "$FILE" $X 40 )
  20                                 END=$( ./find-pgn-split-point.sh "$FILE" $(( X + 1 )) 40 )
  21                                 ~/nmu/pgn-extract/pgn-extract --startpos $START --endpos $END --startfilenum $PGNNUM -e -Wsessebin "$FILE" 2>/dev/null
  22                                 PGNNUM=$(( PGNNUM + 1 ))
  23                         done
  24                 ) | ./binloader - tmp.mtbl.mapper$X $SHARDS 500000
  25         ) &
  26 done
  27 wait
  28
  29 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 )) )
  30
  31 for X in $( seq 0 $(( SHARDS - 1 )) ); do
  32         mv open.mtbl.part$( printf %04d $X ).new open.mtbl.part$( printf %04d $X )
  33 done
  34 rm -f tmp.mtbl*