]> git.sesse.net Git - remoteglot-book/blob - build-book.sh
Indent fix.
[remoteglot-book] / build-book.sh
1 #! /bin/bash
2
3 set -e
4
5 export MAPPERS=40
6 export REDUCERS=40
7 export SHARDS=40
8
9 rm -f pgnnames.txt tmp.mtbl* open.mtbl.part????.new 2>/dev/null
10
11 for FILE in "$@"; do
12         echo "$FILE" >> pgnnames.txt
13 done
14
15 TS_BEGIN=$( date +%s%3N )
16
17 for X in $( seq 0 $(( MAPPERS - 1 )) ); do
18         (
19                 (
20                         PGNNUM=0
21                         for FILE in "$@"; do
22                                 COMP=0
23                                 if echo "$FILE" | grep -q ^comp:; then
24                                         COMP=1
25                                         FILE=$( echo "$FILE" | sed s/comp:// )
26                                 fi
27                                 START=$( ./find-pgn-split-point.sh "$FILE" $X $MAPPERS )
28                                 END=$( ./find-pgn-split-point.sh "$FILE" $(( X + 1 )) $MAPPERS )
29                                 ~/nmu/pgn-extract/pgn-extract --computerflag $COMP --startpos $START --endpos $END --startfilenum $PGNNUM -e -Wsessebin "$FILE" 2>/dev/null
30                                 PGNNUM=$(( PGNNUM + 1 ))
31                         done
32                 ) | ./binloader - tmp.mtbl.mapper$X $SHARDS 200000
33         ) &
34 done
35 wait
36
37 TS_MAPDONE=$( date +%s%3N )
38 MAPOUTPUT=$( du -sch tmp.mtbl.mapper* | tail -n 1 | cut -f1 | sed 's/\([^0-9]\)$/ \1/' )
39
40 echo "Reducing..."
41
42 parallel -i -j $REDUCERS bash -c 'X={}; ./binmerger tmp.mtbl.mapper*.part$( printf %04d $X ).subshard* open.mtbl.part$( printf %04d $X ).new' -- $( seq 0 $(( SHARDS - 1 )) )
43
44 TS_REDUCEDONE=$( date +%s%3N )
45 REDUCEOUTPUT=$( du -sch open.mtbl.part*.new | tail -n 1 | cut -f1 | sed 's/\([^0-9]\)$/ \1/' )
46
47 for X in $( seq 0 $(( SHARDS - 1 )) ); do
48         mv open.mtbl.part$( printf %04d $X ).new open.mtbl.part$( printf %04d $X )
49         chmod 0644 open.mtbl.part$( printf %04d $X )
50 done
51 rm -f tmp.mtbl*
52
53 chmod 0644 pgnnames.txt
54
55 TS_END=$( date +%s%3N )
56
57 printf "\n"
58 printf "  Map time:       %7d ms\n" $(( TS_MAPDONE - TS_BEGIN ))
59 printf "  Reduce time:    %7d ms\n" $(( TS_REDUCEDONE - TS_MAPDONE ))
60 printf "  Total time:     %7d ms\n" $(( TS_END - TS_BEGIN ))
61 printf "\n"
62 printf "  Map output:       %7sB\n" "$MAPOUTPUT"
63 printf "  Reduce output:    %7sB\n" "$REDUCEOUTPUT"
64 printf "\n"