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

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