misc/unictabl/mk_fallbacks.sh

   1 #!/bin/sh
   2 # CVS-ID: $Id$
   3
   4
   5 add_fallback()
   6 {
   7     echo "      - for $3..."
   8     cat _tmp3 | grep "$1" | while read i ; do
   9         code=`echo $i | cut -c1-6`
  10         echo "$code     $2" >> _tmp5
  11     done
  12 }
  13
  14
  15 echo "  * getting list of needed unicode characters..."
  16
  17 cat mappings/*.TXT | sed -n '/^0x../p' | \
  18     cut -f2,4 | sort | uniq | sed -n '/^0x/p' > _tmp1
  19 cat _tmp1 | cut -f1 | sort | uniq > _tmp2
  20
  21
  22 echo "  * making unique list of unicode characters meanings..."
  23
  24 rm -f _tmp3
  25 cat _tmp2 | while read i ; do
  26     sed -n "/^$i/p" _tmp1 | (read t ; echo "$t" >> _tmp3)
  27 done
  28
  29 cp _tmp3 UnicodeChars
  30
  31 echo "  * creating one-byte fallback tables..."
  32
  33 rm -f Fallbacks _tmp5
  34
  35 echo "      - for latin capital letters..."
  36
  37 cat _tmp3 | grep 'LATIN CAPITAL LETTER [A-Z]$' > _tmp6
  38 cat _tmp3 | grep 'LATIN CAPITAL LETTER [A-Z] WITH' >> _tmp6
  39 cat _tmp6 | sort +2 > _tmp4
  40
  41 cat _tmp4 | while read i ; do
  42     code=`echo $i | cut -c1-6`
  43     fallb=`echo $i | cut -c8-29`
  44     cat _tmp4 | fgrep "$fallb" | cut -c1-6 | (read i ;
  45         echo "$code     $i" >> _tmp5)
  46 done
  47
  48
  49 echo "      - for latin small letters..."
  50
  51 cat _tmp3 | grep 'LATIN SMALL LETTER [A-Z]$' > _tmp6
  52 cat _tmp3 | grep 'LATIN SMALL LETTER [A-Z] WITH' >> _tmp6
  53 cat _tmp6 | sort +2 > _tmp4
  54
  55 cat _tmp4 | while read i ; do
  56     code=`echo $i | cut -c1-6`
  57     fallb=`echo $i | cut -c8-27`
  58     cat _tmp4 | fgrep "$fallb" | cut -c1-6 | (read i ;
  59         echo "$code     $i" >> _tmp5)
  60 done
  61
  62
  63 add_fallback "DOUBLE .*QUOTATION MARK" "0x0022" "double quotations"
  64 add_fallback "SINGLE .*QUOTATION MARK" "0x0027" "single quotations"
  65 add_fallback "DASH" "0x002D" "dashes"
  66
  67
  68
  69 echo "  * removing infinite loops from fallback tables..."
  70
  71 cat _tmp5 | grep -v '\(0x....\) \1' | sort > Fallbacks
  72
  73 rm -f _tmp1 _tmp2 _tmp3 _tmp4 _tmp5 _tmp6
  74