contrib/update-unicode/update_unicode.sh

   1 #!/bin/sh
   2 #See http://www.unicode.org/reports/tr44/
   3 #
   4 #Me Enclosing_Mark  an enclosing combining mark
   5 #Mn Nonspacing_Mark a nonspacing combining mark (zero advance width)
   6 #Cf Format          a format control character
   7 #
   8 cd "$(dirname "$0")"
   9 UNICODEWIDTH_H=$(git rev-parse --show-toplevel)/unicode_width.h
  10
  11 if ! test -f UnicodeData.txt; then
  12         wget http://www.unicode.org/Public/UCD/latest/ucd/UnicodeData.txt
  13 fi &&
  14 if ! test -f EastAsianWidth.txt; then
  15         wget http://www.unicode.org/Public/UCD/latest/ucd/EastAsianWidth.txt
  16 fi &&
  17 if ! test -d uniset; then
  18         git clone https://github.com/depp/uniset.git &&
  19         ( cd uniset && git checkout 4b186196dd )
  20 fi &&
  21 (
  22         cd uniset &&
  23         if ! test -x uniset; then
  24                 autoreconf -i &&
  25                 ./configure --enable-warnings=-Werror CFLAGS='-O0 -ggdb'
  26         fi &&
  27         make
  28 ) &&
  29 UNICODE_DIR=. && export UNICODE_DIR &&
  30 cat >$UNICODEWIDTH_H <<-EOF
  31 static const struct interval zero_width[] = {
  32         $(uniset/uniset --32 cat:Me,Mn,Cf + U+1160..U+11FF - U+00AD |
  33           grep -v plane)
  34 };
  35 static const struct interval double_width[] = {
  36         $(uniset/uniset --32 eaw:F,W)
  37 };
  38 EOF