util/toutf8.sh

   1 #! /bin/sh
   2 #
   3 # Very simple script to detect and convert files that we want to re-encode to UTF8
   4
   5 git ls-tree -r --name-only HEAD | \
   6     while read F; do
   7         charset=`file -bi "$F" | sed -e 's|.*charset=||'`
   8         if [ "$charset" != "utf-8" -a "$charset" != "binary" -a "$charset" != "us-ascii" ]; then
   9             iconv -f ISO-8859-1 -t UTF8 < "$F" > "$F.utf8" && \
  10                 ( cmp -s "$F" "$F.utf8" || \
  11                         ( echo "$F"
  12                           mv "$F" "$F.iso-8859-1"
  13                           mv "$F.utf8" "$F"
  14                         )
  15                 )
  16         fi
  17     done