check-isutf8

   1 #!/bin/bash
   2 # Bash needed for the character encodings below.
   3 #
   4 # Run checks for ./isutf8.
   5 #
   6 # Lars Wirzenius <liw@iki.fi>
   7
   8 failed=0
   9
  10 check() {
  11         printf "$2" | ./isutf8 -q
  12         ret=$?
  13         if [ $ret != $1 ]; then
  14                 echo "Failure:"
  15                 echo "  input: $2"
  16                 echo "  expected: $1"
  17                 echo "  got: $ret"
  18                 failed=1
  19         fi
  20         printf "$2" > check-isutf8.tmp.$$
  21         ./isutf8 -q check-isutf8.tmp.$$
  22         ret=$?
  23         rm -f check-isutf8.tmp.$$
  24         if [ $ret != $1 ]; then
  25                 echo "Failure (from file):"
  26                 echo "  input: $2"
  27                 echo "  expected: $1"
  28                 echo "  got: $ret"
  29                 failed=1
  30         fi
  31 }
  32
  33 check 0 ''
  34 check 0 'a'
  35 check 0 'ab'
  36 check 0 '\xc2\xa9'
  37
  38 check 1 '\xc2'
  39 check 1 '\xc2\x20'
  40 check 1 '\x20\xc2'
  41 check 1 '\300\200'
  42 check 1 '\xed\xa0\x88\xed\xbd\x85' # UTF-16 surrogates
  43 check 1 '\xef\xbf\xbe' # 0xFFFE
  44 check 1 '\xef\xbf\xbf' # 0xFFFF
  45
  46 exit $failed