@matheusbsbs / matheus-bagliano / commits / 6aaf956b08

is_hfs_dotgit: loosen over-eager match of \u{..47}

Our is_hfs_dotgit function relies on the hackily-implemented next_hfs_char to give us the next character that an HFS+ filename comparison would look at. It's hacky because it doesn't implement the full case-folding table of HFS+; it gives us just enough to see if the path matches ".git". At the end of next_hfs_char, we use tolower() to convert our 32-bit code point to lowercase. Our tolower() implementation only takes an 8-bit char, though; it throws away the upper 24 bits. This means we can't have any false negatives for is_hfs_dotgit. We only care about matching 7-bit ASCII characters in ".git", and we will correctly process 'G' or 'g'. However, we _can_ have false positives. Because we throw away the upper bits, code point \u{0147} (for example) will look like 'G' and get downcased to 'g'. It's not known whether a sequence of code points whose truncation ends up as ".git" is meaningful in any language, but it does not hurt to be more accurate here. We can just pass out the full 32-bit code point, and compare it manually to the upper and lowercase characters we care about. Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>

Jeff King committed Dec 23, 2014 at 03:45 UTC 6aaf956b08cfab2dcaa1a1afe4192390d0ef14fd

2 files changed +35 -12

t/t1450-fsck.sh

+15

@@ -273,4 +273,19 @@ dot-backslash-case .\\\\.GIT\\\\foobar

273 dotgit-case-backslash .git\\\\foobar

274 EOF

275

276 +test_expect_success 'fsck allows .Ňit' '

277 + (

278 + git init not-dotgit &&

279 + cd not-dotgit &&

280 + echo content >file &&

281 + git add file &&

282 + git commit -m base &&

283 + blob=$(git rev-parse :file) &&

284 + printf "100644 blob $blob\t.\\305\\207it" >tree &&

285 + tree=$(git mktree <tree) &&

286 + git fsck 2>err &&

287 + test_line_count = 0 err

288 + )

289 +'

290 +

291 test_done

utf8.c

+20 -12

@@ -630,8 +630,8 @@ int mbs_chrlen(const char **text, size_t *remainder_p, const char *encoding)

630 }

631

632 /*

633 - * Pick the next char from the stream, folding as an HFS+ filename comparison

634 - * would. Note that this is _not_ complete by any means. It's just enough

633 + * Pick the next char from the stream, ignoring codepoints an HFS+ would.

634 + * Note that this is _not_ complete by any means. It's just enough

635 * to make is_hfs_dotgit() work, and should not be used otherwise.

636 */

637 static ucs_char_t next_hfs_char(const char **in)

@@ -668,12 +668,7 @@ static ucs_char_t next_hfs_char(const char **in)

668 continue;

669 }

670

671 - /*

672 - * there's a great deal of other case-folding that occurs,

673 - * but this is enough to catch anything that will convert

674 - * to ".git"

675 - */

676 - return tolower(out);

671 + return out;

672 }

673 }

674

@@ -681,10 +676,23 @@ int is_hfs_dotgit(const char *path)

676 {

677 ucs_char_t c;

678

684 - if (next_hfs_char(&path) != '.' ||

685 - next_hfs_char(&path) != 'g' ||

686 - next_hfs_char(&path) != 'i' ||

687 - next_hfs_char(&path) != 't')

679 + c = next_hfs_char(&path);

680 + if (c != '.')

681 + return 0;

682 + c = next_hfs_char(&path);

683 +

684 + /*

685 + * there's a great deal of other case-folding that occurs

686 + * in HFS+, but this is enough to catch anything that will

687 + * convert to ".git"

688 + */

689 + if (c != 'g' && c != 'G')

690 + return 0;

691 + c = next_hfs_char(&path);

692 + if (c != 'i' && c != 'I')

693 + return 0;

694 + c = next_hfs_char(&path);

695 + if (c != 't' && c != 'T')

696 return 0;

697 c = next_hfs_char(&path);

698 if (c && !is_dir_sep(c))