parse-g2crossword.sh

   1 #!/bin/bash
   2
   3 crossword=$(w3m -dump_source http://www.guardian.co.uk/crossword/quick/ | sed -e '/FORM action/ { /printer/ { s#^[^"]*"\([^"]*\)".*$#http://www.guardian.co.uk\1#; p; }; d; }; d;' | xargs w3m -dump)
   4 #crossword=$(<crossword.txt)
   5
   6 grid=$(echo "$crossword" | sed -e '/┌/,/┘/p; d;')
   7 clues=$(echo "$crossword" | sed -e '/┘/,/\[sp\]/p; d;' | tail -n +3 - | head -n -1)
   8 across_break=$(($(echo "$clues" | sed -e '1{s#Down.*$##; p; }; d;' | wc -c) - 1))
   9 across=$(echo "$clues" | sed -e '2,${s#^\(.\{'$across_break','$across_break'\}\).*$#\1#; p; }; d;')
  10 down=$(echo "$clues" | sed -e '2,${s#^.\{'$across_break','$across_break'\}##; p; }; d;')
  11
  12 function unformat_clues() {
  13 clues=$1
  14 OLDIFS="$IFS"
  15 IFS="
  16 "
  17 firstline=1
  18 for line in $clues; do
  19     line_formatted=$(echo -n $line | sed -e 's#^[ ]*##; s#[ ]*$##;')
  20     if [ "${line:0:1}" != " " ]; then
  21         if [ $firstline -ne 1 ]; then
  22             echo
  23         else
  24             firstline=0
  25         fi
  26         echo -n $line_formatted
  27     else
  28         if [ "x$line_formatted" != "x" ]; then
  29             echo -n " "$line_formatted
  30         fi
  31     fi
  32 done
  33 IFS="$OLDIFS"
  34 }
  35
  36 function unformat_grid() {
  37     grid=$1
  38     echo "$grid" | sed -e '/───/d; s#[0-9]# #g; s#   # #g; s#● *#x#g; s#│##g;'
  39 }
  40
  41 across_unformatted=$(unformat_clues "$across")
  42 down_unformatted=$(unformat_clues "$down")
  43 grid_unformatted=$(unformat_grid "$grid")
  44
  45 echo GRID
  46 echo "$grid_unformatted"
  47 echo
  48 echo ACROSS
  49 echo "$across_unformatted"
  50 echo
  51 echo DOWN
  52 echo "$down_unformatted"