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