pattern_spatial_gen.sh

   1 #!/bin/sh
   2 # pattern_spatial_gen: Initialize spatial dictionary from a SGF collection
   3 #
   4 # We will first gather all spatial features from the SGF collection
   5 # (we take files as arguments) and store these occuring more than 650 times
   6 # in a freshly created spatial dictionary; afterwards, you will probably want
   7 # to do standard pattern-matching.
   8 #
   9 # DO NOT RUN THIS CONCURRENTLY! The dictionary will get corrupted.
  10 #
  11 # To get spatials in competion mode (also get information about unplayed
  12 # spatials present at the board, not just the single played one per move),
  13 # run this script as:
  14 #
  15 #       PATARGS="competition" ./pattern_spatial_gen.sh ...
  16 #
  17 # Similarly, you can set SPATMIN to different number than 650 to include
  18 # spatial features with lower number of occurences.
  19
  20 [ -n "$SPATMIN" ] || SPATMIN=650
  21
  22 rm -f patterns.spat
  23
  24 echo " Gathering population of spatials occuring more than $SPATMIN times..."
  25 (for i in "$@"; do echo $i >&2; ./sgf2gtp.pl $i; done) |
  26         ./zzgo -e patternscan gen_spat_dict,no_pattern_match,spat_threshold=$SPATMIN${PATARGS:+,$PATARGS} >/dev/null
  27
  28 echo " Renumbering patterns.spat..."
  29 perl -i -pe '/^#/ and next; s/^\d+/++$a/e' patterns.spat
  30
  31 echo -n " Counting hash collisions... "
  32 perl -lne 'chomp; my ($id, $d, $p, @h) = split(/ /, $_); foreach (@h) { next if $h{$_} = $id; print "collision $id - $h{$_} ($_)" if $h{$_}; $h{$_}=$id; }' patterns.spat | wc -l