* [PATCH v3 2/2] diff: generate prettier filenames when using GIT_EXTERNAL_DIFF
2009-05-28 9:43 [PATCH v3 1/2] compat: add a mkstemps() compatibility function David Aguilar
@ 2009-05-28 9:43 ` David Aguilar
2009-06-02 12:35 ` [PATCH v3 1/2] compat: add a mkstemps() compatibility function Antriksh Pany
1 sibling, 0 replies; 6+ messages in thread
From: David Aguilar @ 2009-05-28 9:43 UTC (permalink / raw)
To: gitster; +Cc: git, j.sixt, markus.heidelberg, jnareb, David Aguilar
Naturally, prep_temp_blob() did not care about filenames.
As a result, scripts that use GIT_EXTERNAL_DIFF ended up
with filenames such as ".diff_XXXXXX".
This specializes the GIT_EXTERNAL_DIFF code to generate
user-friendly filenames when creating temporary files.
Diffing "name.ext" now generates "XXXXXX_name.ext".
Signed-off-by: David Aguilar <davvid@gmail.com>
Tested-by: Johannes Sixt <j6t@kdbg.org> (Windows)
---
cache.h | 2 ++
diff.c | 45 +++++++++++++++++++++++++++++++++++++--------
path.c | 19 +++++++++++++++++++
t/t4020-diff-external.sh | 9 +++++++++
4 files changed, 67 insertions(+), 8 deletions(-)
diff --git a/cache.h b/cache.h
index b8503ad..871c984 100644
--- a/cache.h
+++ b/cache.h
@@ -614,6 +614,8 @@ extern int is_empty_blob_sha1(const unsigned char *sha1);
int git_mkstemp(char *path, size_t n, const char *template);
+int git_mkstemps(char *path, size_t n, const char *template, int suffix_len);
+
/*
* NOTE NOTE NOTE!!
*
diff --git a/diff.c b/diff.c
index dcfbcb0..60f07a0 100644
--- a/diff.c
+++ b/diff.c
@@ -1960,12 +1960,37 @@ static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
void *blob,
unsigned long size,
const unsigned char *sha1,
- int mode)
+ int mode,
+ int pretty_filename)
{
int fd;
struct strbuf buf = STRBUF_INIT;
- fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
+ if (pretty_filename) {
+ /* Generate "XXXXXX_filename" */
+ struct strbuf pretty_name = STRBUF_INIT;
+ char *basename = ((char*)path) + strlen(path) - 1;
+
+ /* Windows lacks basename() */
+ while(*basename && basename > path) {
+ basename--;
+ if (is_dir_sep(*basename)) {
+ basename++;
+ break;
+ }
+ }
+
+ strbuf_addstr(&pretty_name, "XXXXXX_");
+ strbuf_addstr(&pretty_name, basename);
+
+ fd = git_mkstemps(temp->tmp_path, PATH_MAX,
+ pretty_name.buf, strlen(basename) + 1);
+
+ strbuf_release(&pretty_name);
+ }
+ else {
+ fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
+ }
if (fd < 0)
die("unable to create temp-file: %s", strerror(errno));
if (convert_to_working_tree(path,
@@ -1984,7 +2009,8 @@ static void prep_temp_blob(const char *path, struct diff_tempfile *temp,
}
static struct diff_tempfile *prepare_temp_file(const char *name,
- struct diff_filespec *one)
+ struct diff_filespec *one,
+ int pretty_filename)
{
struct diff_tempfile *temp = claim_diff_tempfile();
@@ -2021,7 +2047,8 @@ static struct diff_tempfile *prepare_temp_file(const char *name,
(one->sha1_valid ?
one->sha1 : null_sha1),
(one->sha1_valid ?
- one->mode : S_IFLNK));
+ one->mode : S_IFLNK),
+ pretty_filename);
strbuf_release(&sb);
}
else {
@@ -2045,7 +2072,7 @@ static struct diff_tempfile *prepare_temp_file(const char *name,
if (diff_populate_filespec(one, 0))
die("cannot read data blob for %s", one->path);
prep_temp_blob(name, temp, one->data, one->size,
- one->sha1, one->mode);
+ one->sha1, one->mode, pretty_filename);
}
return temp;
}
@@ -2071,8 +2098,9 @@ static void run_external_diff(const char *pgm,
if (one && two) {
struct diff_tempfile *temp_one, *temp_two;
const char *othername = (other ? other : name);
- temp_one = prepare_temp_file(name, one);
- temp_two = prepare_temp_file(othername, two);
+ int pretty_filename = 1;
+ temp_one = prepare_temp_file(name, one, pretty_filename);
+ temp_two = prepare_temp_file(othername, two, pretty_filename);
*arg++ = pgm;
*arg++ = name;
*arg++ = temp_one->name;
@@ -3574,8 +3602,9 @@ static char *run_textconv(const char *pgm, struct diff_filespec *spec,
const char **arg = argv;
struct child_process child;
struct strbuf buf = STRBUF_INIT;
+ int pretty_filename = 0;
- temp = prepare_temp_file(spec->path, spec);
+ temp = prepare_temp_file(spec->path, spec, pretty_filename);
*arg++ = pgm;
*arg++ = temp->name;
*arg = NULL;
diff --git a/path.c b/path.c
index 8a0a674..090b490 100644
--- a/path.c
+++ b/path.c
@@ -140,6 +140,25 @@ int git_mkstemp(char *path, size_t len, const char *template)
}
+
+/* git_mkstemps() - create tmp file with suffix honoring TMPDIR variable. */
+int git_mkstemps(char *path, size_t len, const char *template, int suffix_len)
+{
+ const char *tmp;
+ size_t n;
+
+ tmp = getenv("TMPDIR");
+ if (!tmp)
+ tmp = "/tmp";
+ n = snprintf(path, len, "%s/%s", tmp, template);
+ if (len <= n) {
+ errno = ENAMETOOLONG;
+ return -1;
+ }
+ return mkstemps(path, suffix_len);
+}
+
+
int validate_headref(const char *path)
{
struct stat st;
diff --git a/t/t4020-diff-external.sh b/t/t4020-diff-external.sh
index 0720001..4ea42e0 100755
--- a/t/t4020-diff-external.sh
+++ b/t/t4020-diff-external.sh
@@ -136,6 +136,15 @@ test_expect_success 'GIT_EXTERNAL_DIFF with more than one changed files' '
GIT_EXTERNAL_DIFF=echo git diff
'
+test_expect_success 'GIT_EXTERNAL_DIFF generates pretty paths' '
+ touch file.ext &&
+ git add file.ext &&
+ echo with extension > file.ext &&
+ GIT_EXTERNAL_DIFF=echo git diff file.ext | grep ......_file\.ext &&
+ git update-index --force-remove file.ext &&
+ rm file.ext
+'
+
echo "#!$SHELL_PATH" >fake-diff.sh
cat >> fake-diff.sh <<\EOF
cat $2 >> crlfed.txt
--
1.6.3.1.169.g33fd
^ permalink raw reply related [flat|nested] 6+ messages in thread* Re: [PATCH v3 1/2] compat: add a mkstemps() compatibility function
2009-05-28 9:43 [PATCH v3 1/2] compat: add a mkstemps() compatibility function David Aguilar
2009-05-28 9:43 ` [PATCH v3 2/2] diff: generate prettier filenames when using GIT_EXTERNAL_DIFF David Aguilar
@ 2009-06-02 12:35 ` Antriksh Pany
2009-06-02 13:05 ` Johannes Sixt
1 sibling, 1 reply; 6+ messages in thread
From: Antriksh Pany @ 2009-06-02 12:35 UTC (permalink / raw)
To: David Aguilar; +Cc: gitster, git, j.sixt, markus.heidelberg, jnareb
Hi David
I think I see a few minor issues with the code dealing with random six
letters generation. Please correct me if I am wrong.
On Thu, May 28, 2009 at 3:13 PM, David Aguilar <davvid@gmail.com> wrote:
> mkstemps() is a BSD extension so provide an implementation
> for cross-platform use.
>
> Signed-off-by: David Aguilar <davvid@gmail.com>
> Tested-by: Johannes Sixt <j6t@kdbg.org> (Windows)
> ---
> Makefile | 19 +++++++++++++++
> compat/mkstemps.c | 67 +++++++++++++++++++++++++++++++++++++++++++++++++++++
> config.mak.in | 1 +
> configure.ac | 6 ++++
> git-compat-util.h | 5 ++++
> 5 files changed, 98 insertions(+), 0 deletions(-)
> create mode 100644 compat/mkstemps.c
>
> diff --git a/Makefile b/Makefile
> index eaae45d..a70b5f0 100644
> --- a/Makefile
> +++ b/Makefile
> @@ -52,6 +52,8 @@ all::
> #
> # Define NO_MKDTEMP if you don't have mkdtemp in the C library.
> #
> +# Define NO_MKSTEMPS if you don't have mkstemps in the C library.
> +#
> # Define NO_SYS_SELECT_H if you don't have sys/select.h.
> #
> # Define NO_SYMLINK_HEAD if you never want .git/HEAD to be a symbolic link.
> @@ -636,10 +638,12 @@ EXTLIBS =
>
> ifeq ($(uname_S),Linux)
> NO_STRLCPY = YesPlease
> + NO_MKSTEMPS = YesPlease
> THREADED_DELTA_SEARCH = YesPlease
> endif
> ifeq ($(uname_S),GNU/kFreeBSD)
> NO_STRLCPY = YesPlease
> + NO_MKSTEMPS = YesPlease
> THREADED_DELTA_SEARCH = YesPlease
> endif
> ifeq ($(uname_S),UnixWare)
> @@ -651,6 +655,7 @@ ifeq ($(uname_S),UnixWare)
> SHELL_PATH = /usr/local/bin/bash
> NO_IPV6 = YesPlease
> NO_HSTRERROR = YesPlease
> + NO_MKSTEMPS = YesPlease
> BASIC_CFLAGS += -Kthread
> BASIC_CFLAGS += -I/usr/local/include
> BASIC_LDFLAGS += -L/usr/local/lib
> @@ -674,6 +679,7 @@ ifeq ($(uname_S),SCO_SV)
> SHELL_PATH = /usr/bin/bash
> NO_IPV6 = YesPlease
> NO_HSTRERROR = YesPlease
> + NO_MKSTEMPS = YesPlease
> BASIC_CFLAGS += -I/usr/local/include
> BASIC_LDFLAGS += -L/usr/local/lib
> NO_STRCASESTR = YesPlease
> @@ -702,6 +708,7 @@ ifeq ($(uname_S),SunOS)
> NO_MEMMEM = YesPlease
> NO_HSTRERROR = YesPlease
> NO_MKDTEMP = YesPlease
> + NO_MKSTEMPS = YesPlease
> OLD_ICONV = UnfortunatelyYes
> ifeq ($(uname_R),5.8)
> NO_UNSETENV = YesPlease
> @@ -724,6 +731,7 @@ ifeq ($(uname_O),Cygwin)
> NO_D_INO_IN_DIRENT = YesPlease
> NO_STRCASESTR = YesPlease
> NO_MEMMEM = YesPlease
> + NO_MKSTEMPS = YesPlease
> NO_SYMLINK_HEAD = YesPlease
> NEEDS_LIBICONV = YesPlease
> NO_FAST_WORKING_DIRECTORY = UnfortunatelyYes
> @@ -767,11 +775,13 @@ ifeq ($(uname_S),NetBSD)
> BASIC_LDFLAGS += -L/usr/pkg/lib $(CC_LD_DYNPATH)/usr/pkg/lib
> THREADED_DELTA_SEARCH = YesPlease
> USE_ST_TIMESPEC = YesPlease
> + NO_MKSTEMPS = YesPlease
> endif
> ifeq ($(uname_S),AIX)
> NO_STRCASESTR=YesPlease
> NO_MEMMEM = YesPlease
> NO_MKDTEMP = YesPlease
> + NO_MKSTEMPS = YesPlease
> NO_STRLCPY = YesPlease
> NO_NSEC = YesPlease
> FREAD_READS_DIRECTORIES = UnfortunatelyYes
> @@ -787,12 +797,14 @@ endif
> ifeq ($(uname_S),GNU)
> # GNU/Hurd
> NO_STRLCPY=YesPlease
> + NO_MKSTEMPS = YesPlease
> endif
> ifeq ($(uname_S),IRIX64)
> NO_IPV6=YesPlease
> NO_SETENV=YesPlease
> NO_STRCASESTR=YesPlease
> NO_MEMMEM = YesPlease
> + NO_MKSTEMPS = YesPlease
> NO_STRLCPY = YesPlease
> NO_SOCKADDR_STORAGE=YesPlease
> SHELL_PATH=/usr/gnu/bin/bash
> @@ -805,6 +817,7 @@ ifeq ($(uname_S),HP-UX)
> NO_SETENV=YesPlease
> NO_STRCASESTR=YesPlease
> NO_MEMMEM = YesPlease
> + NO_MKSTEMPS = YesPlease
> NO_STRLCPY = YesPlease
> NO_MKDTEMP = YesPlease
> NO_UNSETENV = YesPlease
> @@ -834,6 +847,7 @@ ifneq (,$(findstring MINGW,$(uname_S)))
> NO_C99_FORMAT = YesPlease
> NO_STRTOUMAX = YesPlease
> NO_MKDTEMP = YesPlease
> + NO_MKSTEMPS = YesPlease
> SNPRINTF_RETURNS_BOGUS = YesPlease
> NO_SVN_TESTS = YesPlease
> NO_PERL_MAKEMAKER = YesPlease
> @@ -853,6 +867,7 @@ ifneq (,$(findstring MINGW,$(uname_S)))
> endif
> ifneq (,$(findstring arm,$(uname_M)))
> ARM_SHA1 = YesPlease
> + NO_MKSTEMPS = YesPlease
> endif
>
> -include config.mak.autogen
> @@ -1011,6 +1026,10 @@ ifdef NO_MKDTEMP
> COMPAT_CFLAGS += -DNO_MKDTEMP
> COMPAT_OBJS += compat/mkdtemp.o
> endif
> +ifdef NO_MKSTEMPS
> + COMPAT_CFLAGS += -DNO_MKSTEMPS
> + COMPAT_OBJS += compat/mkstemps.o
> +endif
> ifdef NO_UNSETENV
> COMPAT_CFLAGS += -DNO_UNSETENV
> COMPAT_OBJS += compat/unsetenv.o
> diff --git a/compat/mkstemps.c b/compat/mkstemps.c
> new file mode 100644
> index 0000000..87ebc2a
> --- /dev/null
> +++ b/compat/mkstemps.c
> @@ -0,0 +1,67 @@
> +#include "../git-compat-util.h"
> +
> +#ifndef TMP_MAX
> +#define TMP_MAX 16384
> +#endif
> +
> +/* Adapted from libiberty's mkstemp.c. */
> +int gitmkstemps(char *pattern, int suffix_len)
> +{
> + static const char letters[] =
> + "abcdefghijklmnopqrstuvwxyz"
> + "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
> + "0123456789";
> + static const int num_letters = 62;
> + uint64_t value;
> + struct timeval tv;
> + char *template;
> + size_t len;
> + int fd, count;
> +
> + len = strlen(pattern);
> +
> + if (len < 6 + suffix_len) {
> + errno = EINVAL;
> + return -1;
> + }
> +
> + if (strncmp(&pattern[len - 6 - suffix_len], "XXXXXX", 6)) {
> + errno = EINVAL;
> + return -1;
> + }
> +
> + /* Replace pattern's XXXXXX characters with randomness.
> + * Try TMP_MAX different filenames.
> + */
> + gettimeofday(&tv, NULL);
> + value = ((size_t)(tv.tv_usec << 16)) ^ tv.tv_sec ^ getpid();
> + template = &pattern[len - 6 - suffix_len];
> + for (count = 0; count < TMP_MAX; ++count) {
> + uint64_t v = value;
> + /* Fill in the random bits. */
> + template[0] = letters[v % num_letters]; v/= num_letters;
> + template[1] = letters[v % num_letters]; v/= num_letters;
> + template[2] = letters[v % num_letters]; v/= num_letters;
> + template[3] = letters[v % num_letters]; v/= num_letters;
> + template[4] = letters[v % num_letters]; v/= num_letters;
> + template[5] = letters[v % num_letters]; v/= num_letters;
v is divided 6 times by 62, and a decent amount of randomness shall be
ensured only if 'value' is usually greater than 62^6.
If it is assumed to be ok for 'v' to become zero in the last one odd
steps sometimes (in fact, _frequently_), then you may ignore what I am
pointing out here.
1. On a couple of systems that I checked on, all variables/types on
the right hand side of
value = ((size_t)(tv.tv_usec << 16)) ^ tv.tv_sec ^ getpid();
are 4 bytes long. So 'value' is ultimately going to be assigned a
number that fits in 4 bytes, ie value < 2^32 (at least on a few
systems).
The systems that I checked on have these `uname -a` outputs:
- Linux host 2.6.9-55.ELsmp #1 SMP Fri Apr 20 17:03:35 EDT 2007 i686
i686 i386 GNU/Linux
- Linux host 2.4.21-50.ELsmp #1 SMP Tue May 8 17:18:29 EDT 2007 i686
i686 i386 GNU/Linux
Also, even if size_t were 64 bits, typecasting _after_ the shift does
not help much. And given the uncertainty about sizeof (size_t), we
could use:
((uint64_t) tv.tv_usec) << 16)
2. tv_usec has a decimal value range of 0-999999 (10^6 usec make 1
sec). Which means that tv_usec fits completely in 20 bits (or less).
(tv_usec << 16) yields a number that fits completely in 36 bits (or
less). Max value of this number is 999999 * 2^16, or for convenience,
about M = 10^6 * 2^16. This number (in the range of 0 to M) goes on to
be divided by D=62^6. Also, M > D. Thus, there is about a D / M * 100
~ 87 % probability of the division M / D working out to be zero. i.e,
in 87% cases, the last division (v/= num_letters) will cause 'v' to
become 0.
Therefore, I think an additional shift of 7 or more bits will help in
improving randomness of template[5]. I would suggest left shifting by
24 bits. This seemed ok in some tests I did.
Combining #1 and #2, I guess we could have the computation of 'value' as:
value = (((uint64_t)tv.tv_usec) << 24) ^ tv.tv_sec ^ getpid();
- Antriksh
^ permalink raw reply [flat|nested] 6+ messages in thread