git.oblomov.eu Git - git/blob - t/t0028-working-tree-encoding.sh

   1 #!/bin/sh
   2
   3 test_description='working-tree-encoding conversion via gitattributes'
   4
   5 . ./test-lib.sh
   6
   7 test_expect_success 'setup test files' '
   8         git config core.eol lf &&
   9
  10         text="hallo there!\ncan you read me?" &&
  11         echo "*.utf16 text working-tree-encoding=utf-16" >.gitattributes &&
  12         printf "$text" >test.utf8.raw &&
  13         printf "$text" | iconv -f UTF-8 -t UTF-16 >test.utf16.raw &&
  14         printf "$text" | iconv -f UTF-8 -t UTF-32 >test.utf32.raw &&
  15
  16         # Line ending tests
  17         printf "one\ntwo\nthree\n" >lf.utf8.raw &&
  18         printf "one\r\ntwo\r\nthree\r\n" >crlf.utf8.raw &&
  19
  20         # BOM tests
  21         printf "\0a\0b\0c"                         >nobom.utf16be.raw &&
  22         printf "a\0b\0c\0"                         >nobom.utf16le.raw &&
  23         printf "\376\777\0a\0b\0c"                 >bebom.utf16be.raw &&
  24         printf "\777\376a\0b\0c\0"                 >lebom.utf16le.raw &&
  25         printf "\0\0\0a\0\0\0b\0\0\0c"             >nobom.utf32be.raw &&
  26         printf "a\0\0\0b\0\0\0c\0\0\0"             >nobom.utf32le.raw &&
  27         printf "\0\0\376\777\0\0\0a\0\0\0b\0\0\0c" >bebom.utf32be.raw &&
  28         printf "\777\376\0\0a\0\0\0b\0\0\0c\0\0\0" >lebom.utf32le.raw &&
  29
  30         # Add only UTF-16 file, we will add the UTF-32 file later
  31         cp test.utf16.raw test.utf16 &&
  32         cp test.utf32.raw test.utf32 &&
  33         git add .gitattributes test.utf16 &&
  34         git commit -m initial
  35 '
  36
  37 test_expect_success 'ensure UTF-8 is stored in Git' '
  38         test_when_finished "rm -f test.utf16.git" &&
  39
  40         git cat-file -p :test.utf16 >test.utf16.git &&
  41         test_cmp_bin test.utf8.raw test.utf16.git
  42 '
  43
  44 test_expect_success 're-encode to UTF-16 on checkout' '
  45         test_when_finished "rm -f test.utf16.raw" &&
  46
  47         rm test.utf16 &&
  48         git checkout test.utf16 &&
  49         test_cmp_bin test.utf16.raw test.utf16
  50 '
  51
  52 test_expect_success 'check $GIT_DIR/info/attributes support' '
  53         test_when_finished "rm -f test.utf32.git" &&
  54         test_when_finished "git reset --hard HEAD" &&
  55
  56         echo "*.utf32 text working-tree-encoding=utf-32" >.git/info/attributes &&
  57         git add test.utf32 &&
  58
  59         git cat-file -p :test.utf32 >test.utf32.git &&
  60         test_cmp_bin test.utf8.raw test.utf32.git
  61 '
  62
  63 for i in 16 32
  64 do
  65         test_expect_success "check prohibited UTF-${i} BOM" '
  66                 test_when_finished "git reset --hard HEAD" &&
  67
  68                 echo "*.utf${i}be text working-tree-encoding=utf-${i}be" >>.gitattributes &&
  69                 echo "*.utf${i}le text working-tree-encoding=utf-${i}LE" >>.gitattributes &&
  70
  71                 # Here we add a UTF-16 (resp. UTF-32) files with BOM (big/little-endian)
  72                 # but we tell Git to treat it as UTF-16BE/UTF-16LE (resp. UTF-32).
  73                 # In these cases the BOM is prohibited.
  74                 cp bebom.utf${i}be.raw bebom.utf${i}be &&
  75                 test_must_fail git add bebom.utf${i}be 2>err.out &&
  76                 test_i18ngrep "fatal: BOM is prohibited .* utf-${i}be" err.out &&
  77                 test_i18ngrep "use UTF-${i} as working-tree-encoding" err.out &&
  78
  79                 cp lebom.utf${i}le.raw lebom.utf${i}be &&
  80                 test_must_fail git add lebom.utf${i}be 2>err.out &&
  81                 test_i18ngrep "fatal: BOM is prohibited .* utf-${i}be" err.out &&
  82                 test_i18ngrep "use UTF-${i} as working-tree-encoding" err.out &&
  83
  84                 cp bebom.utf${i}be.raw bebom.utf${i}le &&
  85                 test_must_fail git add bebom.utf${i}le 2>err.out &&
  86                 test_i18ngrep "fatal: BOM is prohibited .* utf-${i}LE" err.out &&
  87                 test_i18ngrep "use UTF-${i} as working-tree-encoding" err.out &&
  88
  89                 cp lebom.utf${i}le.raw lebom.utf${i}le &&
  90                 test_must_fail git add lebom.utf${i}le 2>err.out &&
  91                 test_i18ngrep "fatal: BOM is prohibited .* utf-${i}LE" err.out &&
  92                 test_i18ngrep "use UTF-${i} as working-tree-encoding" err.out
  93         '
  94
  95         test_expect_success "check required UTF-${i} BOM" '
  96                 test_when_finished "git reset --hard HEAD" &&
  97
  98                 echo "*.utf${i} text working-tree-encoding=utf-${i}" >>.gitattributes &&
  99
 100                 cp nobom.utf${i}be.raw nobom.utf${i} &&
 101                 test_must_fail git add nobom.utf${i} 2>err.out &&
 102                 test_i18ngrep "fatal: BOM is required .* utf-${i}" err.out &&
 103                 test_i18ngrep "use UTF-${i}BE or UTF-${i}LE" err.out &&
 104
 105                 cp nobom.utf${i}le.raw nobom.utf${i} &&
 106                 test_must_fail git add nobom.utf${i} 2>err.out &&
 107                 test_i18ngrep "fatal: BOM is required .* utf-${i}" err.out &&
 108                 test_i18ngrep "use UTF-${i}BE or UTF-${i}LE" err.out
 109         '
 110
 111         test_expect_success "eol conversion for UTF-${i} encoded files on checkout" '
 112                 test_when_finished "rm -f crlf.utf${i}.raw lf.utf${i}.raw" &&
 113                 test_when_finished "git reset --hard HEAD^" &&
 114
 115                 cat lf.utf8.raw | iconv -f UTF-8 -t UTF-${i} >lf.utf${i}.raw &&
 116                 cat crlf.utf8.raw | iconv -f UTF-8 -t UTF-${i} >crlf.utf${i}.raw &&
 117                 cp crlf.utf${i}.raw eol.utf${i} &&
 118
 119                 cat >expectIndexLF <<-EOF &&
 120                         i/lf    w/-text attr/text               eol.utf${i}
 121                 EOF
 122
 123                 git add eol.utf${i} &&
 124                 git commit -m eol &&
 125
 126                 # UTF-${i} with CRLF (Windows line endings)
 127                 rm eol.utf${i} &&
 128                 git -c core.eol=crlf checkout eol.utf${i} &&
 129                 test_cmp_bin crlf.utf${i}.raw eol.utf${i} &&
 130
 131                 # Although the file has CRLF in the working tree,
 132                 # ensure LF in the index
 133                 git ls-files --eol eol.utf${i} >actual &&
 134                 test_cmp expectIndexLF actual &&
 135
 136                 # UTF-${i} with LF (Unix line endings)
 137                 rm eol.utf${i} &&
 138                 git -c core.eol=lf checkout eol.utf${i} &&
 139                 test_cmp_bin lf.utf${i}.raw eol.utf${i} &&
 140
 141                 # The file LF in the working tree, ensure LF in the index
 142                 git ls-files --eol eol.utf${i} >actual &&
 143                 test_cmp expectIndexLF actual
 144         '
 145 done
 146
 147 test_expect_success 'check unsupported encodings' '
 148         test_when_finished "git reset --hard HEAD" &&
 149
 150         echo "*.set text working-tree-encoding" >.gitattributes &&
 151         printf "set" >t.set &&
 152         test_must_fail git add t.set 2>err.out &&
 153         test_i18ngrep "true/false are no valid working-tree-encodings" err.out &&
 154
 155         echo "*.unset text -working-tree-encoding" >.gitattributes &&
 156         printf "unset" >t.unset &&
 157         git add t.unset &&
 158
 159         echo "*.empty text working-tree-encoding=" >.gitattributes &&
 160         printf "empty" >t.empty &&
 161         git add t.empty &&
 162
 163         echo "*.garbage text working-tree-encoding=garbage" >.gitattributes &&
 164         printf "garbage" >t.garbage &&
 165         test_must_fail git add t.garbage 2>err.out &&
 166         test_i18ngrep "failed to encode" err.out
 167 '
 168
 169 test_expect_success 'error if encoding round trip is not the same during refresh' '
 170         BEFORE_STATE=$(git rev-parse HEAD) &&
 171         test_when_finished "git reset --hard $BEFORE_STATE" &&
 172
 173         # Add and commit a UTF-16 file but skip the "working-tree-encoding"
 174         # filter. Consequently, the in-repo representation is UTF-16 and not
 175         # UTF-8. This simulates a Git version that has no working tree encoding
 176         # support.
 177         echo "*.utf16le text working-tree-encoding=utf-16le" >.gitattributes &&
 178         echo "hallo" >nonsense.utf16le &&
 179         TEST_HASH=$(git hash-object --no-filters -w nonsense.utf16le) &&
 180         git update-index --add --cacheinfo 100644 $TEST_HASH nonsense.utf16le &&
 181         COMMIT=$(git commit-tree -p $(git rev-parse HEAD) -m "plain commit" $(git write-tree)) &&
 182         git update-ref refs/heads/master $COMMIT &&
 183
 184         test_must_fail git checkout HEAD^ 2>err.out &&
 185         test_i18ngrep "error: .* overwritten by checkout:" err.out
 186 '
 187
 188 test_expect_success 'error if encoding garbage is already in Git' '
 189         BEFORE_STATE=$(git rev-parse HEAD) &&
 190         test_when_finished "git reset --hard $BEFORE_STATE" &&
 191
 192         # Skip the UTF-16 filter for the added file
 193         # This simulates a Git version that has no checkoutEncoding support
 194         cp nobom.utf16be.raw nonsense.utf16 &&
 195         TEST_HASH=$(git hash-object --no-filters -w nonsense.utf16) &&
 196         git update-index --add --cacheinfo 100644 $TEST_HASH nonsense.utf16 &&
 197         COMMIT=$(git commit-tree -p $(git rev-parse HEAD) -m "plain commit" $(git write-tree)) &&
 198         git update-ref refs/heads/master $COMMIT &&
 199
 200         git diff 2>err.out &&
 201         test_i18ngrep "error: BOM is required" err.out
 202 '
 203
 204 test_done