t/t7812-grep-icase-non-ascii.sh

   1 #!/bin/sh
   2
   3 test_description='grep icase on non-English locales'
   4
   5 . ./lib-gettext.sh
   6
   7 doalarm () {
   8         perl -e 'alarm shift; exec @ARGV' -- "$@"
   9 }
  10
  11 test_expect_success GETTEXT_LOCALE 'setup' '
  12         test_write_lines "TILRAUN: Halló Heimur!" >file &&
  13         git add file &&
  14         LC_ALL="$is_IS_locale" &&
  15         export LC_ALL
  16 '
  17
  18 test_expect_success GETTEXT_LOCALE 'setup REGEX_LOCALE prerequisite' '
  19         # This "test-tool" invocation is identical...
  20         if test-tool regex "HALLÓ" "Halló" ICASE
  21         then
  22                 test_set_prereq REGEX_LOCALE
  23         else
  24
  25                 # ... to this one, but this way "test_must_fail" will
  26                 # tell a segfault or abort() from the regexec() test
  27                 # itself
  28                 test_must_fail test-tool regex "HALLÓ" "Halló" ICASE
  29         fi
  30 '
  31
  32 test_expect_success REGEX_LOCALE 'grep literal string, no -F' '
  33         git grep -i "TILRAUN: Halló Heimur!" &&
  34         git grep -i "TILRAUN: HALLÓ HEIMUR!"
  35 '
  36
  37 test_expect_success GETTEXT_LOCALE,PCRE 'grep pcre utf-8 icase' '
  38         git grep --perl-regexp    "TILRAUN: H.lló Heimur!" &&
  39         git grep --perl-regexp -i "TILRAUN: H.lló Heimur!" &&
  40         git grep --perl-regexp -i "TILRAUN: H.LLÓ HEIMUR!"
  41 '
  42
  43 test_expect_success GETTEXT_LOCALE,PCRE 'grep pcre utf-8 string with "+"' '
  44         test_write_lines "TILRAUN: Hallóó Heimur!" >file2 &&
  45         git add file2 &&
  46         git grep -l --perl-regexp "TILRAUN: H.lló+ Heimur!" >actual &&
  47         echo file >expected &&
  48         echo file2 >>expected &&
  49         test_cmp expected actual
  50 '
  51
  52 test_expect_success REGEX_LOCALE 'grep literal string, with -F' '
  53         git grep -i -F "TILRAUN: Halló Heimur!" &&
  54         git grep -i -F "TILRAUN: HALLÓ HEIMUR!"
  55 '
  56
  57 test_expect_success REGEX_LOCALE 'grep string with regex, with -F' '
  58         test_write_lines "TILRAUN: Halló Heimur [abc]!" >file3 &&
  59         git add file3 &&
  60         git grep -i -F "TILRAUN: Halló Heimur [abc]!" file3
  61 '
  62
  63 test_expect_success REGEX_LOCALE 'pickaxe -i on non-ascii' '
  64         git commit -m first &&
  65         git log --format=%f -i -S"TILRAUN: HALLÓ HEIMUR!" >actual &&
  66         echo first >expected &&
  67         test_cmp expected actual
  68 '
  69
  70 test_expect_success GETTEXT_LOCALE,LIBPCRE2 'PCRE v2: setup invalid UTF-8 data' '
  71         printf "\\200\\n" >invalid-0x80 &&
  72         echo "ævar" >expected &&
  73         cat expected >>invalid-0x80 &&
  74         git add invalid-0x80 &&
  75
  76         # Test for PCRE2_MATCH_INVALID_UTF bug
  77         # https://bugs.exim.org/show_bug.cgi?id=2642
  78         printf "\\345Aæ\\n" >invalid-0xe5 &&
  79         git add invalid-0xe5
  80 '
  81
  82 test_expect_success GETTEXT_LOCALE,LIBPCRE2 'PCRE v2: grep ASCII from invalid UTF-8 data' '
  83         git grep -h "var" invalid-0x80 >actual &&
  84         test_cmp expected actual &&
  85         git grep -h "(*NO_JIT)var" invalid-0x80 >actual &&
  86         test_cmp expected actual
  87 '
  88
  89 test_expect_success GETTEXT_LOCALE,LIBPCRE2 'PCRE v2: grep ASCII from invalid UTF-8 data (PCRE2 bug #2642)' '
  90         git grep -h "Aæ" invalid-0xe5 >actual &&
  91         test_cmp invalid-0xe5 actual &&
  92         git grep -h "(*NO_JIT)Aæ" invalid-0xe5 >actual &&
  93         test_cmp invalid-0xe5 actual
  94 '
  95
  96 test_expect_success GETTEXT_LOCALE,LIBPCRE2 'PCRE v2: grep non-ASCII from invalid UTF-8 data' '
  97         git grep -h "æ" invalid-0x80 >actual &&
  98         test_cmp expected actual &&
  99         git grep -h "(*NO_JIT)æ" invalid-0x80 >actual &&
 100         test_cmp expected actual
 101 '
 102
 103 test_expect_success GETTEXT_LOCALE,LIBPCRE2 'PCRE v2: grep non-ASCII from invalid UTF-8 data (PCRE2 bug #2642)' '
 104         git grep -h "Aæ" invalid-0xe5 >actual &&
 105         test_cmp invalid-0xe5 actual &&
 106         git grep -h "(*NO_JIT)Aæ" invalid-0xe5 >actual &&
 107         test_cmp invalid-0xe5 actual
 108 '
 109
 110 test_lazy_prereq PCRE2_MATCH_INVALID_UTF '
 111         test-tool pcre2-config has-PCRE2_MATCH_INVALID_UTF
 112 '
 113
 114 test_expect_success GETTEXT_LOCALE,LIBPCRE2 'PCRE v2: grep non-ASCII from invalid UTF-8 data with -i' '
 115         test_might_fail git grep -hi "Æ" invalid-0x80 >actual &&
 116         test_might_fail git grep -hi "(*NO_JIT)Æ" invalid-0x80 >actual
 117 '
 118
 119 test_expect_success GETTEXT_LOCALE,LIBPCRE2,PCRE2_MATCH_INVALID_UTF 'PCRE v2: grep non-ASCII from invalid UTF-8 data with -i' '
 120         git grep -hi "Æ" invalid-0x80 >actual &&
 121         test_cmp expected actual &&
 122         git grep -hi "(*NO_JIT)Æ" invalid-0x80 >actual &&
 123         test_cmp expected actual
 124 '
 125
 126 test_expect_success GETTEXT_LOCALE,LIBPCRE2,PCRE2_MATCH_INVALID_UTF 'PCRE v2: grep non-ASCII from invalid UTF-8 data with -i (PCRE2 bug #2642)' '
 127         git grep -hi "Æ" invalid-0xe5 >actual &&
 128         test_cmp invalid-0xe5 actual &&
 129         git grep -hi "(*NO_JIT)Æ" invalid-0xe5 >actual &&
 130         test_cmp invalid-0xe5 actual &&
 131
 132         # Only the case of grepping the ASCII part in a way that
 133         # relies on -i fails
 134         git grep -hi "aÆ" invalid-0xe5 >actual &&
 135         test_cmp invalid-0xe5 actual &&
 136         git grep -hi "(*NO_JIT)aÆ" invalid-0xe5 >actual &&
 137         test_cmp invalid-0xe5 actual
 138 '
 139
 140 test_expect_success GETTEXT_LOCALE,LIBPCRE2 'PCRE v2: grep non-literal ASCII from UTF-8' '
 141         git grep --perl-regexp -h -o -e ll. file >actual &&
 142         echo "lló" >expected &&
 143         test_cmp expected actual
 144 '
 145
 146 test_expect_success GETTEXT_LOCALE,LIBPCRE2 'PCRE v2: grep avoid endless loop bug' '
 147         echo " Halló" >leading-whitespace &&
 148         git add leading-whitespace &&
 149         doalarm 1 git grep --perl-regexp "^\s" leading-whitespace
 150 '
 151
 152 test_done