source: trunk/src/grep/tests/multibyte-white-space@ 3670

#! /bin/sh
# Test whether \s matches SP and UTF-8 multi-byte white space characters.
#
# Copyright (C) 2013-2021 Free Software Foundation, Inc.
#
# Copying and distribution of this file, with or without modification,
# are permitted in any medium without royalty provided the copyright
# notice and this notice are preserved.

. "${srcdir=.}/init.sh"; path_prepend_ ../src

require_en_utf8_locale_

LC_ALL=en_US.UTF-8
export LC_ALL

# It would have been nice to be able to use all UTF8 characters
# with the Unicode WSpace=Y character property,
# https://en.wikipedia.org/wiki/Whitespace_character, but that
# would currently cause distracting failures everywhere I've tried.
# Instead, I've listed each with an indicator column, telling what
# this test should do if the system's locale/tools produce the
# wrong answer.

# The values in that column:
# X required on all systems (fail if \s or \S fail to work as expected)
# x required on "modern enough" systems
# O optional: \s or \S misbehavior elicits a warning, but never failure

utf8_space_characters=$(sed 's/.*: *//;s/  */\\x/g' <<\EOF
U+0009 Horizontal Tab:            X 09
U+000A Line feed:                 O 0a
U+000B Vertical Tab:              X 0b
U+000C Form feed:                 X 0c
U+000D Carriage return:           X 0d
U+0020 SPACE:                     X 20
U+0085 Next line:                 O 85
U+00A0 NO-BREAK SPACE:            O c2 a0
U+1680 OGHAM SPACE MARK:          x e1 9a 80
U+2000 EN QUAD:                   x e2 80 80
U+2001 EM QUAD:                   x e2 80 81
U+2002 EN SPACE:                  x e2 80 82
U+2003 EM SPACE:                  x e2 80 83
U+2004 THREE-PER-EM SPACE:        x e2 80 84
U+2005 FOUR-PER-EM SPACE:         x e2 80 85
U+2006 SIX-PER-EM SPACE:          x e2 80 86
U+2007 FIGURE SPACE:              O e2 80 87
U+2008 PUNCTUATION SPACE:         x e2 80 88
U+2009 THIN SPACE:                x e2 80 89
U+200A HAIR SPACE:                x e2 80 8a
U+200B ZERO WIDTH SPACE:          O e2 80 8b
U+202F NARROW NO-BREAK SPACE:     O e2 80 af
U+205F MEDIUM MATHEMATICAL SPACE: x e2 81 9f
U+3000 IDEOGRAPHIC SPACE:         x e3 80 80
EOF
)

fail=0

# On systems that are not "modern enough," simply warn when an "x"-marked
# character is not classified as white space.  Too many systems
# have inadequate UTF-8 tables in this respect, and that lack should not
# discourage/confuse those who consider whether to install grep.

# As for what constitutes "modern enough", I've arbitrarily started
# with "Fedora 20 or newer".  Tested additions welcome.
modern_enough=0
grep -iE 'fedora release [2-9][0-9]+\b' /etc/redhat-release >/dev/null 2>&1 \
  && modern_enough=1

for i in $utf8_space_characters; do
  eval 'fail() { fail=1; }'
  m=ERROR
  case $i in
      X*) ;;
      x*) test $modern_enough = 1 || { eval 'fail() { :; }'; m=warning; } ;;
      O*) m=warning; eval 'fail() { :; }' ;;
      *) warn_ "unexpected prefix: $i"; exit 1 ;;
  esac

  # Strip the prefix byte.
  i=${i#?}

  hex_printf_ "$i" | grep -q '^\s$' \
      || { warn_ " $m: \\s failed to match $i in the $LC_ALL locale"; fail; }
  hex_printf_ "$i" | returns_ 1 grep -q '\S' \
      || { warn_ " $m: \\S mistakenly matched $i in the $LC_ALL locale"; fail; }
done


# This is a separate test, only nominally related to \s.
# It is solely to get coverage of a code path (exercising dfa.c's
# match_mb_charset function) that would have otherwise been untouched.
# However, as of the change-set adding this new test, match_mb_charset
# is unreachable via grep.
printf '\0' | returns_ 1 grep -aE '^\s?$' > out 2>&1 || fail=1
compare /dev/null out

Exit $fail