When serializing UTF-16 (and UTF-32), there are three possible ways to
write the stream. One can write the data with a BOM in either big-endian
or little-endian format, or one can write the data without a BOM in
big-endian format.
Most systems' iconv implementations choose to write it with a BOM in
some endianness, since this is the most foolproof, and it is resistant
to misinterpretation on Windows, where UTF-16 and the little-endian
serialization are very common. For compatibility with Windows and to
avoid accidental misuse there, Git always wants to write UTF-16 with a
BOM, and will refuse to read UTF-16 without it.
However, musl's iconv implementation writes UTF-16 without a BOM,
relying on the user to interpret it as big-endian. This causes t0028 and
the related functionality to fail, since Git won't read the file without
a BOM.
Add a Makefile and #define knob, ICONV_OMITS_BOM, that can be set if the
iconv implementation has this behavior. When set, Git will write a BOM
manually for UTF-16 and UTF-32 and then force the data to be written in
UTF-16BE or UTF-32BE. We choose big-endian behavior here because the
tests use the raw "UTF-16" encoding, which will be big-endian when the
implementation requires this knob to be set.
Update the tests to detect this case and write test data with an added
BOM if necessary. Always write the BOM in the tests in big-endian
format, since all iconv implementations that omit a BOM must use
big-endian serialization according to the Unicode standard.
Preserve the existing behavior for systems which do not have this knob
enabled, since they may use optimized implementations, including
defaulting to the native endianness, which may improve performance.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Users who want UTF-16 files in the working tree set the .gitattributes
like this:
test.txt working-tree-encoding=UTF-16
The unicode standard itself defines 3 allowed ways how to encode UTF-16.
The following 3 versions convert all back to 'g' 'i' 't' in UTF-8:
a) UTF-16, without BOM, big endian:
$ printf "\000g\000i\000t" | iconv -f UTF-16 -t UTF-8 | od -c
0000000 g i t
b) UTF-16, with BOM, little endian:
$ printf "\377\376g\000i\000t\000" | iconv -f UTF-16 -t UTF-8 | od -c
0000000 g i t
c) UTF-16, with BOM, big endian:
$ printf "\376\377\000g\000i\000t" | iconv -f UTF-16 -t UTF-8 | od -c
0000000 g i t
Git uses libiconv to convert from UTF-8 in the index into ITF-16 in the
working tree.
After a checkout, the resulting file has a BOM and is encoded in "UTF-16",
in the version (c) above.
This is what iconv generates, more details follow below.
iconv (and libiconv) can generate UTF-16, UTF-16LE or UTF-16BE:
d) UTF-16
$ printf 'git' | iconv -f UTF-8 -t UTF-16 | od -c
0000000 376 377 \0 g \0 i \0 t
e) UTF-16LE
$ printf 'git' | iconv -f UTF-8 -t UTF-16LE | od -c
0000000 g \0 i \0 t \0
f) UTF-16BE
$ printf 'git' | iconv -f UTF-8 -t UTF-16BE | od -c
0000000 \0 g \0 i \0 t
There is no way to generate version (b) from above in a Git working tree,
but that is what some applications need.
(All fully unicode aware applications should be able to read all 3 variants,
but in practise we are not there yet).
When producing UTF-16 as an output, iconv generates the big endian version
with a BOM. (big endian is probably chosen for historical reasons).
iconv can produce UTF-16 files with little endianess by using "UTF-16LE"
as encoding, and that file does not have a BOM.
Not all users (especially under Windows) are happy with this.
Some tools are not fully unicode aware and can only handle version (b).
Today there is no way to produce version (b) with iconv (or libiconv).
Looking into the history of iconv, it seems as if version (c) will
be used in all future iconv versions (for compatibility reasons).
Solve this dilemma and introduce a Git-specific "UTF-16LE-BOM".
libiconv can not handle the encoding, so Git pick it up, handles the BOM
and uses libiconv to convert the rest of the stream.
(UTF-16BE-BOM is added for consistency)
Rported-by: Adrián Gimeno Balaguer <adrigibal@gmail.com>
Signed-off-by: Torsten Bögershausen <tboegi@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Code clean-up to use size_t/ssize_t when they are the right type.
* jk/size-t:
strbuf_humanise: use unsigned variables
pass st.st_size as hint for strbuf_readlink()
strbuf_readlink: use ssize_t
strbuf: use size_t for length in intermediate variables
reencode_string: use size_t for string lengths
reencode_string: use st_add/st_mult helpers
The iconv interface takes a size_t, which is the appropriate
type for an in-memory buffer. But our reencode_string_*
functions use integers, meaning we may get confusing results
when the sizes exceed INT_MAX. Let's use size_t
consistently.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When converting a string with iconv, if the output buffer
isn't big enough, we grow it. But our growth is done without
any concern for integer overflow. So when we add:
outalloc = sofar + insz * 2 + 32;
we may end up wrapping outalloc (which is a size_t), and
allocating a too-small buffer. We then manipulate it
further:
outsz = outalloc - sofar - 1;
and feed outsz back to iconv. If outalloc is wrapped and
smaller than sofar, we'll end up with a small allocation but
feed a very large outsz to iconv, which could result in it
overflowing the buffer.
Can we use this to construct an attack wherein the victim
clones a repository with a very large commit object with an
encoding header, and running "git log" reencodes it into
utf8, causing an overflow?
An attack of this sort is likely impossible in practice.
"sofar" is how many output bytes we've written total, and
"insz" is the number of input bytes remaining. Imagine our
input doubles in size as we output it (which is easy to do
by converting latin1 to utf8, for example), and that we
start with N input bytes. Our initial output buffer also
starts at N bytes, so after the first call we'd have N/2
input bytes remaining (insz), and have written N bytes
(sofar). That means our next allocation will be
(N + N/2 * 2 + 32) bytes, or (2N + 32).
We can therefore overflow a 32-bit size_t with a commit
message that's just under 2^31 bytes, assuming it consists
mostly of "doubling" sequences (e.g., latin1 0xe1 which
becomes utf8 0xc3 0xa1).
But we'll never make it that far with such a message. We'll
be spending 2^31 bytes on the original string. And our
initial output buffer will also be 2^31 bytes. Which is not
going to succeed on a system with a 32-bit size_t, since
there will be other things using the address space, too. The
initial malloc will fail.
If we imagine instead that we can triple the size when
converting, then our second allocation becomes
(N + 2/3N * 2 + 32), or (7/3N + 32). That still requires two
allocations of 3/7 of our address space (6/7 of the total)
to succeed.
If we imagine we can quadruple, it becomes (5/2N + 32); we
need to be able to allocate 4/5 of the address space to
succeed.
This might start to get plausible. But is it possible to get
a 4-to-1 increase in size? Probably if you're converting to
some obscure encoding. But since git defaults to utf8 for
its output, that's the likely destination encoding for an
attack. And while there are 4-character utf8 sequences, it's
unlikely that you'd be able find a single-byte source
sequence in any encoding.
So this is certainly buggy code which should be fixed, but
it is probably not a useful attack vector.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In ISO C, char constants must be in the range -128..127. Change the BOM
constants to char literals to avoid overflow.
Signed-off-by: Beat Bolli <dev+git@drbeat.li>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
* maint-2.14:
Git 2.14.4
Git 2.13.7
verify_path: disallow symlinks in .gitmodules
update-index: stat updated files earlier
verify_dotfile: mention case-insensitivity in comment
verify_path: drop clever fallthrough
skip_prefix: add case-insensitive variant
is_{hfs,ntfs}_dotgitmodules: add tests
is_ntfs_dotgit: match other .git files
is_hfs_dotgit: match other .git files
is_ntfs_dotgit: use a size_t for traversing string
submodule-config: verify submodule names as paths
Both verify_path() and fsck match ".git", ".GIT", and other
variants specific to HFS+. Let's allow matching other
special files like ".gitmodules", which we'll later use to
enforce extra restrictions via verify_path() and fsck.
Signed-off-by: Jeff King <peff@peff.net>
The new "checkout-encoding" attribute can ask Git to convert the
contents to the specified encoding when checking out to the working
tree (and the other way around when checking in).
* ls/checkout-encoding:
convert: add round trip check based on 'core.checkRoundtripEncoding'
convert: add tracing for 'working-tree-encoding' attribute
convert: check for detectable errors in UTF encodings
convert: add 'working-tree-encoding' attribute
utf8: add function to detect a missing UTF-16/32 BOM
utf8: add function to detect prohibited UTF-16/32 BOM
utf8: teach same_encoding() alternative UTF encoding names
strbuf: add a case insensitive starts_with()
strbuf: add xstrdup_toupper()
strbuf: remove unnecessary NUL assignment in xstrdup_tolower()
If the endianness is not defined in the encoding name, then let's
be strict and require a BOM to avoid any encoding confusion. The
is_missing_required_utf_bom() function returns true if a required BOM
is missing.
The Unicode standard instructs to assume big-endian if there in no BOM
for UTF-16/32 [1][2]. However, the W3C/WHATWG encoding standard used
in HTML5 recommends to assume little-endian to "deal with deployed
content" [3]. Strictly requiring a BOM seems to be the safest option
for content in Git.
This function is used in a subsequent commit.
[1] http://unicode.org/faq/utf_bom.html#gen6
[2] http://www.unicode.org/versions/Unicode10.0.0/ch03.pdf
Section 3.10, D98, page 132
[3] https://encoding.spec.whatwg.org/#utf-16le
Signed-off-by: Lars Schneider <larsxschneider@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Whenever a data stream is declared to be UTF-16BE, UTF-16LE, UTF-32BE
or UTF-32LE a BOM must not be used [1]. The function returns true if
this is the case.
This function is used in a subsequent commit.
[1] http://unicode.org/faq/utf_bom.html#bom10
Signed-off-by: Lars Schneider <larsxschneider@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The function same_encoding() could only recognize alternative names for
UTF-8 encodings. Teach it to recognize all kinds of alternative UTF
encoding names (e.g. utf16).
While we are at it, fix a crash that would occur if same_encoding() was
called with a NULL argument and a non-NULL argument.
This function is used in a subsequent commit.
Signed-off-by: Lars Schneider <larsxschneider@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This is more consistent with the project style. The majority of Git's
source files use dashes in preference to underscores in their file names.
Also adjust contrib/update-unicode as well.
Signed-off-by: Stefan Beller <sbeller@google.com>
A common mistake when writing binary search is to allow possible
integer overflow by using the simple average:
mid = (min + max) / 2;
Instead, use the overflow-safe version:
mid = min + (max - min) / 2;
This translation is safe since the operation occurs inside a loop
conditioned on "min < max". The included changes were found using
the following git grep:
git grep '/ *2;' '*.c'
Making this cleanup will prevent future review friction when a new
binary search is contructed based on existing code.
Signed-off-by: Derrick Stolee <dstolee@microsoft.com>
Reviewed-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Even though latin-1 is still seen in e-mail headers, some platforms
only install ISO-8859-1. "iconv -f ISO-8859-1" succeeds, while
"iconv -f latin-1" fails on such a system.
Using the same fallback_encoding() mechanism factored out in the
previous step, teach ourselves that "ISO-8859-1" has a better chance
of being accepted than "latin-1".
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The codepath we use to call iconv_open() has a provision to use a
fallback encoding when it fails, hoping that "UTF-8" being spelled
differently could be the reason why the library function did not
like the encoding names we gave it. Essentially, we turn what we
have observed to be used as variants of "UTF-8" (e.g. "utf8") into
the most official spelling and use that as a fallback.
We do the same thing for input and output encoding. Introduce a
helper function to do just one side and call that twice.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add strbuf_utf8_align() which will align a given string into a strbuf
as per given align_type and width. If the width is greater than the
string length then no alignment is performed.
Helped-by: Eric Sunshine <sunshine@sunshineco.com>
Mentored-by: Christian Couder <christian.couder@gmail.com>
Mentored-by: Matthieu Moy <matthieu.moy@grenoble-inp.fr>
Signed-off-by: Karthik Nayak <karthik.188@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
With the recent change to ignore the UTF8 BOM at the beginning of
.gitignore files, we now have two codepaths that do such a skipping
(the other one is for reading the configuration files).
Introduce utf8_bom[] constant string and skip_utf8_bom() helper
and teach .gitignore code how to use it.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
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>
We do not allow paths with a ".git" component to be added to
the index, as that would mean repository contents could
overwrite our repository files. However, asking "is this
path the same as .git" is not as simple as strcmp() on some
filesystems.
HFS+'s case-folding does more than just fold uppercase into
lowercase (which we already handle with strcasecmp). It may
also skip past certain "ignored" Unicode code points, so
that (for example) ".gi\u200ct" is mapped ot ".git".
The full list of folds can be found in the tables at:
https://www.opensource.apple.com/source/xnu/xnu-1504.15.3/bsd/hfs/hfscommon/Unicode/UCStringCompareData.h
Implementing a full "is this path the same as that path"
comparison would require us importing the whole set of
tables. However, what we want to do is much simpler: we
only care about checking ".git". We know that 'G' is the
only thing that folds to 'g', and so on, so we really only
need to deal with the set of ignored code points, which is
much smaller.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Move strbuf_addchars() to strbuf.c, where it belongs, and make it
available for other callers.
Signed-off-by: Rene Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The main loop in strbuf_utf8_replace() could summed up as:
while ('src' is still valid) {
1) advance 'src' to copy ANSI escape sequences
2) advance 'src' to copy/replace visible characters
}
The problem is after #1, 'src' may have reached the end of the string
(so 'src' points to NUL) and #2 will continue to copy that NUL as if
it's a normal character. Because the output is stored in a strbuf,
this NUL accounted in the 'len' field as well. Check after #1 and
break the loop if necessary.
The test does not look obvious, but the combination of %>>() should
make a call trace like this
show_log()
pretty_print_commit()
format_commit_message()
strbuf_expand()
format_commit_item()
format_and_pad_commit()
strbuf_utf8_replace()
where %C(auto)%d would insert a color reset escape sequence in the end
of the string given to strbuf_utf8_replace() and show_log() uses
fwrite() to send everything to stdout (including the incorrect NUL
inserted by strbuf_utf8_replace)
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Update the logic to compute the display width needed for utf8
strings and allow us to more easily maintain the tables used in
that logic.
We may want to let the users choose if codepoints with ambiguous
widths are treated as a double or single width in a follow-up patch.
* tb/unicode-6.3-zero-width:
utf8: make it easier to auto-update git_wcwidth()
utf8.c: use a table for double_width
The function git_wcwidth() returns for a given unicode code point the
width on the display:
-1 for control characters,
0 for combining or other non-visible code points
1 for e.g. ASCII
2 for double-width code points.
This table had been originally been extracted for one Unicode
version, probably 3.2.
We now use two tables these days, one for zero-width and another for
double-width. Make it easier to update these tables to a later
version of Unicode by factoring out the table from utf8.c into
unicode_width.h and add the script update_unicode.sh to update the
table based on the latest Unicode specification files.
Thanks to Peter Krefting <peter@softwolves.pp.se> and Kevin Bracey
<kevin@bracey.fi> for helping with their Unicode knowledge.
Signed-off-by: Torsten Bögershausen <tboegi@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Refactor git_wcwidth() and replace the if-else-if chain.
Use the table double_width which is scanned by the bisearch() function,
which is already used to find combining code points.
Signed-off-by: Torsten Bögershausen <tboegi@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Unicode 6.3 defines more code points as combining or accents. For
example, the character "ö" could be expressed as an "o" followed by
U+0308 COMBINING DIARESIS (aka umlaut, double-dot-above). We should
consider that such a sequence of two codepoints occupies one display
column for the alignment purposes, and for that, git_wcwidth()
should return 0 for them. Affected codepoints are:
U+0358..U+035C
U+0487
U+05A2, U+05BA, U+05C5, U+05C7
U+0604, U+0616..U+061A, U+0659..U+065F
Earlier unicode standards had defined these as "reserved".
Only the range 0..U+07FF has been checked to see which codepoints
need to be marked as 0-width while preparing for this commit; more
updates may be needed.
Signed-off-by: Torsten Bögershausen <tboegi@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We treat these as unsigned everywhere and compare against unsigned
values, so declare them using the typedef we already have for this.
While we're here, fix the indentation as well.
Signed-off-by: John Keeping <john@keeping.me.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
iconv(3) returns "(size_t) -1" on error. Make sure that we cast the
"-1" properly when checking for this.
Signed-off-by: John Keeping <john@keeping.me.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Some systems experience failures in t4205-*.sh (tests 18-20, 27)
which all relate to the use of truncation with the %< padding
placeholder. This capability was added in the commit a7f01c6b
("pretty: support truncating in %>, %< and %><", 19-04-2013).
The truncation support was implemented with the assistance of a
new strbuf function (strbuf_utf8_replace). This function contains
the following code:
strbuf_attach(sb_src, strbuf_detach(&sb_dst, NULL),
sb_dst.len, sb_dst.alloc);
Unfortunately, this code is subject to unspecified behaviour. In
particular, the order of evaluation of the argument expressions
(along with the associated side effects) is not specified by the
C standard. Note that the second argument expression is a call to
strbuf_detach() which, as a side effect, sets the 'len' and 'alloc'
fields of the sb_dst argument to zero. Depending on the order of
evaluation of the argument expressions to the strbuf_attach call,
this can lead to assigning an empty string to 'sb_src'.
In order to remove the undesired behaviour, we replace the above
line of code with:
strbuf_swap(sb_src, &sb_dst);
strbuf_release(&sb_dst);
which achieves the desired effect without provoking unspecified
behaviour.
Signed-off-by: Ramsay Jones <ramsay@ramsay1.demon.co.uk>
Acked-by: Duy Nguyen <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This is pretty useful in `%<(100)%s%Cred%>(20)% an' where %s does not
use up all 100 columns and %an needs more than 20 columns. By
replacing %>(20) with %>>(20), %an can steal spaces from %s.
%>> understands escape sequences, so %Cred does not stop it from
stealing spaces in %<(100).
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
%>(N,trunc) truncates the right part after N columns and replace the
last two letters with "..". ltrunc does the same on the left. mtrunc
cuts the middle out.
Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When "format-patch" quoted a non-ascii strings on the header files,
it incorrectly applied rfc2047 and chopped a single character in
the middle of it.
* ks/rfc2047-one-char-at-a-time:
format-patch: RFC 2047 says multi-octet character may not be split
Some platforms and users spell UTF-8 differently; retry with the
most official "UTF-8" when the system does not understand the
user-supplied encoding name that are the common alternative
spellings of UTF-8.
* jk/utf-8-can-be-spelled-differently:
utf8: accept alternate spellings of UTF-8
Even though an earlier attempt (bafc478..41dd00bad) cleaned
up RFC 2047 encoding, pretty.c::add_rfc2047() still decides
where to split the output line by going through the input
one byte at a time, and potentially splits a character in
the middle. A subject line may end up showing like this:
".... fö?? bar". (instead of ".... föö bar".)
if split incorrectly.
RFC 2047, section 5 (3) explicitly forbids such beaviour
Each 'encoded-word' MUST represent an integral number of
characters. A multi-octet character may not be split across
adjacent 'encoded- word's.
that means that e.g. for
Subject: .... föö bar
encoding
Subject: =?UTF-8?q?....=20f=C3=B6=C3=B6?=
=?UTF-8?q?=20bar?=
is correct, and
Subject: =?UTF-8?q?....=20f=C3=B6=C3?= <-- NOTE ö is broken here
=?UTF-8?q?=B6=20bar?=
is not, because "ö" character UTF-8 encoding C3 B6 is split here across
adjacent encoded words.
To fix the problem, make the loop grab one _character_ at a time and
determine its output length to see where to break the output line. Note
that this version only knows about UTF-8, but the logic to grab one
character is abstracted out in mbs_chrlen() function to make it possible
to extend it to other encodings with the help of iconv in the future.
Signed-off-by: Kirill Smelkov <kirr@mns.spb.ru>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
