contrib / update-unicode / update_unicode.shon commit Merge branch 'mm/gc-safety-doc' (979b82f)
   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#
   8cd "$(dirname "$0")"
   9UNICODEWIDTH_H=$(git rev-parse --show-toplevel)/unicode_width.h
  10
  11wget -N http://www.unicode.org/Public/UCD/latest/ucd/UnicodeData.txt \
  12        http://www.unicode.org/Public/UCD/latest/ucd/EastAsianWidth.txt &&
  13if ! test -d uniset; then
  14        git clone https://github.com/depp/uniset.git &&
  15        ( cd uniset && git checkout 4b186196dd )
  16fi &&
  17(
  18        cd uniset &&
  19        if ! test -x uniset; then
  20                autoreconf -i &&
  21                ./configure --enable-warnings=-Werror CFLAGS='-O0 -ggdb'
  22        fi &&
  23        make
  24) &&
  25UNICODE_DIR=. && export UNICODE_DIR &&
  26cat >$UNICODEWIDTH_H <<-EOF
  27static const struct interval zero_width[] = {
  28        $(uniset/uniset --32 cat:Me,Mn,Cf + U+1160..U+11FF - U+00AD)
  29};
  30static const struct interval double_width[] = {
  31        $(uniset/uniset --32 eaw:F,W)
  32};
  33EOF