#! /bin/csh -f 
# Last edited on 2001-04-08 21:43:49 by stolfi

set usage = "$0 < INNAME.sep > OUTNAME.dic"

# Converts a lexicon from the Sao Carlos septuple format (INNAME.sep)
# to the new format (INNANE,dic).
#
# The input is assumed to be in the format 
#   (PAL,ORD,CATCODE,ATTRS,CANON,XX,YY) 
# The output will be in the format
#   CATEG(PAL,CANON)
# Single-letter nouns and abbreviations are removed. 
#

cat \
  | tr -d '()' \
  | gawk -v FS=',' \
      ' /./{ \
          if (($3 ~ /^[ac][1-4][n-r]/) && (length($1) == 1)) { next; } \
          printf "%s(%s,%s)\n",$3,$1,$5; \
        } \
      '