misc/unictabl/mk_fallbacks.sh

   1 #!/bin/sh
   2
   3
   4 add_fallback()
   5 {
   6     echo "      - for $3..."
   7     cat _tmp3 | grep "$1" | while read i ; do
   8         code=`echo $i | cut -c1-6`
   9         echo "$code     $2" >> _tmp5
  10     done
  11 }
  12
  13
  14 echo "  * getting list of needed unicode characters..."
  15
  16 cat mappings/*.TXT | sed -n '/^0x../p' | \
  17     cut -f2,4 | sort | uniq | sed -n '/^0x/p' > _tmp1
  18 cat _tmp1 | cut -f1 | sort | uniq > _tmp2
  19
  20
  21 echo "  * making unique list of unicode characters meanings..."
  22
  23 rm -f _tmp3
  24 cat _tmp2 | while read i ; do
  25     sed -n "/^$i/p" _tmp1 | (read t ; echo "$t" >> _tmp3)
  26 done
  27
  28 cp _tmp3 UnicodeChars
  29
  30 echo "  * creating one-byte fallback tables..."
  31
  32 rm -f Fallbacks _tmp5
  33
  34 echo "      - for latin capital letters..."
  35
  36 cat _tmp3 | grep 'LATIN CAPITAL LETTER [A-Z]$' > _tmp6
  37 cat _tmp3 | grep 'LATIN CAPITAL LETTER [A-Z] WITH' >> _tmp6
  38 cat _tmp6 | sort +2 > _tmp4
  39
  40 cat _tmp4 | while read i ; do
  41     code=`echo $i | cut -c1-6`
  42     fallb=`echo $i | cut -c8-29`
  43     cat _tmp4 | fgrep "$fallb" | cut -c1-6 | (read i ;
  44         echo "$code     $i" >> _tmp5)
  45 done
  46
  47
  48 echo "      - for latin small letters..."
  49
  50 cat _tmp3 | grep 'LATIN SMALL LETTER [A-Z]$' > _tmp6
  51 cat _tmp3 | grep 'LATIN SMALL LETTER [A-Z] WITH' >> _tmp6
  52 cat _tmp6 | sort +2 > _tmp4
  53
  54 cat _tmp4 | while read i ; do
  55     code=`echo $i | cut -c1-6`
  56     fallb=`echo $i | cut -c8-27`
  57     cat _tmp4 | fgrep "$fallb" | cut -c1-6 | (read i ;
  58         echo "$code     $i" >> _tmp5)
  59 done
  60
  61
  62 add_fallback "DOUBLE .*QUOTATION MARK" "0x0022" "double quotations"
  63 add_fallback "SINGLE .*QUOTATION MARK" "0x0027" "single quotations"
  64 add_fallback "DASH" "0x002D" "dashes"
  65
  66
  67
  68 echo "  * removing infinite loops from fallback tables..."
  69
  70 cat _tmp5 | grep -v '\(0x....\) \1' | sort > Fallbacks
  71
  72 rm -f _tmp1 _tmp2 _tmp3 _tmp4 _tmp5 _tmp6
  73