glib/gio/tests/g-file-info.c

1113 lines
44 KiB
C
Raw Normal View History

/* GLib testing framework examples and tests
* Copyright (C) 2008 Red Hat, Inc.
* Authors: Tomas Bzatek <tbzatek@redhat.com>
*
* This work is provided "as is"; redistribution and modification
* in whole or in part, in any medium, physical or electronic is
* permitted without restriction.
*
* This work is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
*
* In no event shall the authors or contributors be liable for any
* direct, indirect, incidental, special, exemplary, or consequential
* damages (including, but not limited to, procurement of substitute
* goods or services; loss of use, data, or profits; or business
* interruption) however caused and on any theory of liability, whether
* in contract, strict liability, or tort (including negligence or
* otherwise) arising in any way out of the use of this software, even
* if advised of the possibility of such damage.
*/
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
#include "config.h"
#include <glib/glib.h>
#include <gio/gio.h>
#include <stdlib.h>
#include <string.h>
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
#ifdef G_OS_WIN32
#include <stdio.h>
#include <glib/gstdio.h>
#include <windows.h>
#include <shlobj.h>
#include <io.h> /* for _get_osfhandle */
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
#endif
#define TEST_NAME "Prilis zlutoucky kun"
#define TEST_DISPLAY_NAME "UTF-8 p\xc5\x99\xc3\xadli\xc5\xa1 \xc5\xbelu\xc5\xa5ou\xc4\x8dk\xc3\xbd k\xc5\xaf\xc5\x88"
#define TEST_SIZE 0xFFFFFFF0
static void
test_assigned_values (GFileInfo *info)
{
const char *name, *display_name, *mistake;
guint64 size;
GFileType type;
/* Test for attributes presence */
g_assert_true (g_file_info_has_attribute (info, G_FILE_ATTRIBUTE_STANDARD_NAME));
g_assert_true (g_file_info_has_attribute (info, G_FILE_ATTRIBUTE_STANDARD_DISPLAY_NAME));
g_assert_true (g_file_info_has_attribute (info, G_FILE_ATTRIBUTE_STANDARD_SIZE));
g_assert_false (g_file_info_has_attribute (info, G_FILE_ATTRIBUTE_STANDARD_COPY_NAME));
/* Retrieve data back and compare */
name = g_file_info_get_attribute_byte_string (info, G_FILE_ATTRIBUTE_STANDARD_NAME);
display_name = g_file_info_get_attribute_string (info, G_FILE_ATTRIBUTE_STANDARD_DISPLAY_NAME);
mistake = g_file_info_get_attribute_string (info, G_FILE_ATTRIBUTE_STANDARD_COPY_NAME);
size = g_file_info_get_attribute_uint64 (info, G_FILE_ATTRIBUTE_STANDARD_SIZE);
type = g_file_info_get_file_type (info);
g_assert_cmpstr (name, ==, TEST_NAME);
g_assert_cmpstr (display_name, ==, TEST_DISPLAY_NAME);
g_assert_null (mistake);
g_assert_cmpint (size, ==, TEST_SIZE);
g_assert_cmpstr (name, ==, g_file_info_get_name (info));
g_assert_cmpstr (display_name, ==, g_file_info_get_display_name (info));
g_assert_cmpint (size, ==, g_file_info_get_size (info) );
g_assert_cmpint (type, ==, G_FILE_TYPE_DIRECTORY);
}
static void
test_g_file_info (void)
{
GFileInfo *info;
GFileInfo *info_dup;
GFileInfo *info_copy;
char **attr_list;
GFileAttributeMatcher *matcher;
info = g_file_info_new ();
/* Test for empty instance */
attr_list = g_file_info_list_attributes (info, NULL);
g_assert_nonnull (attr_list);
g_assert_null (*attr_list);
g_strfreev (attr_list);
g_file_info_set_attribute_byte_string (info, G_FILE_ATTRIBUTE_STANDARD_NAME, TEST_NAME);
g_file_info_set_attribute_string (info, G_FILE_ATTRIBUTE_STANDARD_DISPLAY_NAME, TEST_DISPLAY_NAME);
g_file_info_set_attribute_uint64 (info, G_FILE_ATTRIBUTE_STANDARD_SIZE, TEST_SIZE);
g_file_info_set_file_type (info, G_FILE_TYPE_DIRECTORY);
/* The attr list should not be empty now */
attr_list = g_file_info_list_attributes (info, NULL);
g_assert_nonnull (attr_list);
g_assert_nonnull (*attr_list);
g_strfreev (attr_list);
test_assigned_values (info);
/* Test dups */
info_dup = g_file_info_dup (info);
g_assert_nonnull (info_dup);
test_assigned_values (info_dup);
info_copy = g_file_info_new ();
g_file_info_copy_into (info_dup, info_copy);
g_assert_nonnull (info_copy);
test_assigned_values (info_copy);
/* Test remove attribute */
g_assert_false (g_file_info_has_attribute (info, G_FILE_ATTRIBUTE_STANDARD_SORT_ORDER));
g_file_info_set_attribute_int32 (info, G_FILE_ATTRIBUTE_STANDARD_SORT_ORDER, 10);
g_assert_true (g_file_info_has_attribute (info, G_FILE_ATTRIBUTE_STANDARD_SORT_ORDER));
g_assert_cmpint (g_file_info_get_attribute_type (info, G_FILE_ATTRIBUTE_STANDARD_SORT_ORDER), ==, G_FILE_ATTRIBUTE_TYPE_INT32);
g_assert_cmpint (g_file_info_get_attribute_status (info, G_FILE_ATTRIBUTE_STANDARD_SORT_ORDER), !=, G_FILE_ATTRIBUTE_STATUS_ERROR_SETTING);
g_file_info_remove_attribute (info, G_FILE_ATTRIBUTE_STANDARD_SORT_ORDER);
g_assert_false (g_file_info_has_attribute (info, G_FILE_ATTRIBUTE_STANDARD_SORT_ORDER));
g_assert_cmpint (g_file_info_get_attribute_type (info, G_FILE_ATTRIBUTE_STANDARD_SORT_ORDER), ==, G_FILE_ATTRIBUTE_TYPE_INVALID);
matcher = g_file_attribute_matcher_new (G_FILE_ATTRIBUTE_STANDARD_NAME ","
G_FILE_ATTRIBUTE_STANDARD_DISPLAY_NAME);
g_assert_true (g_file_attribute_matcher_matches (matcher, G_FILE_ATTRIBUTE_STANDARD_NAME));
g_assert_false (g_file_attribute_matcher_matches_only (matcher, G_FILE_ATTRIBUTE_STANDARD_NAME));
g_assert_false (g_file_attribute_matcher_matches (matcher, G_FILE_ATTRIBUTE_STANDARD_SIZE));
g_file_info_set_attribute_mask (info, matcher);
g_file_attribute_matcher_unref (matcher);
g_assert_false (g_file_info_has_attribute (info, G_FILE_ATTRIBUTE_STANDARD_SIZE));
g_assert_true (g_file_info_has_attribute (info, G_FILE_ATTRIBUTE_STANDARD_NAME));
g_object_unref (info);
g_object_unref (info_dup);
g_object_unref (info_copy);
}
static void
test_g_file_info_modification_time (void)
{
GFile *file = NULL;
GFileIOStream *io_stream = NULL;
GFileInfo *info = NULL;
GDateTime *dt = NULL, *dt_usecs = NULL, *dt_new = NULL, *dt_new_usecs = NULL;
GTimeSpan ts;
gboolean nsecs_supported;
gint usecs;
guint32 nsecs;
GError *error = NULL;
g_test_summary ("Test that getting the modification time of a file works.");
file = g_file_new_tmp ("g-file-info-test-XXXXXX", &io_stream, &error);
g_assert_no_error (error);
info = g_file_query_info (file,
G_FILE_ATTRIBUTE_TIME_MODIFIED,
G_FILE_QUERY_INFO_NONE,
NULL, &error);
g_assert_no_error (error);
/* Check the modification time is retrievable. */
dt = g_file_info_get_modification_date_time (info);
g_assert_nonnull (dt);
/* Try again with microsecond precision. */
g_clear_object (&info);
info = g_file_query_info (file,
G_FILE_ATTRIBUTE_TIME_MODIFIED "," G_FILE_ATTRIBUTE_TIME_MODIFIED_USEC,
G_FILE_QUERY_INFO_NONE,
NULL, &error);
g_assert_no_error (error);
dt_usecs = g_file_info_get_modification_date_time (info);
g_assert_nonnull (dt_usecs);
ts = g_date_time_difference (dt_usecs, dt);
g_assert_cmpint (ts, >=, 0);
g_assert_cmpint (ts, <, G_USEC_PER_SEC);
/* Try again with nanosecond precision. */
g_clear_object (&info);
info = g_file_query_info (file,
G_FILE_ATTRIBUTE_TIME_MODIFIED "," G_FILE_ATTRIBUTE_TIME_MODIFIED_USEC "," G_FILE_ATTRIBUTE_TIME_MODIFIED_NSEC,
G_FILE_QUERY_INFO_NONE,
NULL, &error);
g_assert_no_error (error);
nsecs_supported = g_file_info_has_attribute (info, G_FILE_ATTRIBUTE_TIME_MODIFIED_NSEC);
if (nsecs_supported)
{
usecs = g_date_time_get_microsecond (dt_usecs);
nsecs = g_file_info_get_attribute_uint32 (info, G_FILE_ATTRIBUTE_TIME_MODIFIED_NSEC);
g_assert_cmpuint (nsecs, >=, usecs * 1000);
g_assert_cmpuint (nsecs, <, (usecs + 1) * 1000);
}
/* Try round-tripping the modification time. */
dt_new = g_date_time_add (dt_usecs, G_USEC_PER_SEC + 50);
g_file_info_set_modification_date_time (info, dt_new);
dt_new_usecs = g_file_info_get_modification_date_time (info);
ts = g_date_time_difference (dt_new_usecs, dt_new);
g_assert_cmpint (ts, ==, 0);
/* Setting the modification time with usec-precision should have cleared nsecs. */
g_assert_cmpuint (g_file_info_get_attribute_uint32 (info, G_FILE_ATTRIBUTE_TIME_MODIFIED_NSEC), ==, 0);
/* Try setting the modification time with nsec-precision and it should set the
* usecs too. */
if (nsecs_supported)
{
gint new_usecs;
guint32 new_nsecs;
GDateTime *new_dt_usecs = NULL;
g_file_set_attribute_uint32 (file, G_FILE_ATTRIBUTE_TIME_MODIFIED_NSEC, nsecs + 100,
G_FILE_QUERY_INFO_NONE, NULL, &error);
g_assert_no_error (error);
g_clear_object (&info);
info = g_file_query_info (file,
G_FILE_ATTRIBUTE_TIME_MODIFIED "," G_FILE_ATTRIBUTE_TIME_MODIFIED_USEC "," G_FILE_ATTRIBUTE_TIME_MODIFIED_NSEC,
G_FILE_QUERY_INFO_NONE,
NULL, &error);
g_assert_no_error (error);
new_dt_usecs = g_file_info_get_modification_date_time (info);
g_assert_nonnull (new_dt_usecs);
new_usecs = g_date_time_get_microsecond (new_dt_usecs);
new_nsecs = g_file_info_get_attribute_uint32 (info, G_FILE_ATTRIBUTE_TIME_MODIFIED_NSEC);
g_assert_cmpuint (new_nsecs, ==, nsecs + 100);
g_assert_cmpuint (new_nsecs, >=, new_usecs * 1000);
g_assert_cmpuint (new_nsecs, <, (new_usecs + 1) * 1000);
g_date_time_unref (new_dt_usecs);
}
/* Clean up. */
g_clear_object (&io_stream);
g_file_delete (file, NULL, NULL);
g_clear_object (&file);
g_clear_object (&info);
g_date_time_unref (dt);
g_date_time_unref (dt_usecs);
g_date_time_unref (dt_new);
g_date_time_unref (dt_new_usecs);
}
static void
test_g_file_info_access_time (void)
{
GFile *file = NULL;
GFileIOStream *io_stream = NULL;
GFileInfo *info = NULL;
GDateTime *dt = NULL, *dt_usecs = NULL, *dt_new = NULL, *dt_new_usecs = NULL,
*dt_before_epoch = NULL, *dt_before_epoch_returned = NULL;
GTimeSpan ts;
gboolean nsecs_supported;
gint usecs;
guint32 nsecs;
GError *error = NULL;
g_test_summary ("Test that getting the access time of a file works.");
file = g_file_new_tmp ("g-file-info-test-XXXXXX", &io_stream, &error);
g_assert_no_error (error);
info = g_file_query_info (file,
G_FILE_ATTRIBUTE_TIME_ACCESS,
G_FILE_QUERY_INFO_NONE,
NULL, &error);
g_assert_no_error (error);
if (!g_file_info_has_attribute (info, G_FILE_ATTRIBUTE_TIME_ACCESS))
{
g_test_skip ("Skipping testing access time as its not supported by the kernel");
g_file_delete (file, NULL, NULL);
g_clear_object (&file);
g_clear_object (&info);
return;
}
/* Check the access time is retrievable. */
dt = g_file_info_get_access_date_time (info);
g_assert_nonnull (dt);
/* Try again with microsecond precision. */
g_clear_object (&info);
info = g_file_query_info (file,
G_FILE_ATTRIBUTE_TIME_ACCESS "," G_FILE_ATTRIBUTE_TIME_ACCESS_USEC,
G_FILE_QUERY_INFO_NONE,
NULL, &error);
g_assert_no_error (error);
dt_usecs = g_file_info_get_access_date_time (info);
g_assert_nonnull (dt_usecs);
ts = g_date_time_difference (dt_usecs, dt);
g_assert_cmpint (ts, >, 0);
g_assert_cmpint (ts, <, G_USEC_PER_SEC);
/* Try again with nanosecond precision. */
g_clear_object (&info);
info = g_file_query_info (file,
G_FILE_ATTRIBUTE_TIME_ACCESS "," G_FILE_ATTRIBUTE_TIME_ACCESS_USEC "," G_FILE_ATTRIBUTE_TIME_ACCESS_NSEC,
G_FILE_QUERY_INFO_NONE,
NULL, &error);
g_assert_no_error (error);
nsecs_supported = g_file_info_has_attribute (info, G_FILE_ATTRIBUTE_TIME_ACCESS_NSEC);
if (nsecs_supported)
{
usecs = g_date_time_get_microsecond (dt_usecs);
nsecs = g_file_info_get_attribute_uint32 (info, G_FILE_ATTRIBUTE_TIME_ACCESS_NSEC);
g_assert_cmpuint (nsecs, >=, usecs * 1000);
g_assert_cmpuint (nsecs, <, (usecs + 1) * 1000);
}
/* Try round-tripping the access time. */
dt_new = g_date_time_add (dt_usecs, G_USEC_PER_SEC + 50);
g_file_info_set_access_date_time (info, dt_new);
dt_new_usecs = g_file_info_get_access_date_time (info);
ts = g_date_time_difference (dt_new_usecs, dt_new);
g_assert_cmpint (ts, ==, 0);
// try with a negative timestamp
dt_before_epoch = g_date_time_new_from_unix_utc (-10000);
g_file_info_set_access_date_time (info, dt_before_epoch);
dt_before_epoch_returned = g_file_info_get_access_date_time (info);
ts = g_date_time_difference (dt_before_epoch, dt_before_epoch_returned);
g_assert_cmpint (ts, ==, 0);
/* Setting the access time with usec-precision should have cleared nsecs. */
g_assert_cmpuint (g_file_info_get_attribute_uint32 (info, G_FILE_ATTRIBUTE_TIME_ACCESS_NSEC), ==, 0);
/* Try setting the access time with nsec-precision and it should set the
* usecs too. */
if (nsecs_supported)
{
gint new_usecs;
guint32 new_nsecs;
GDateTime *new_dt_usecs = NULL;
g_file_set_attribute_uint32 (file, G_FILE_ATTRIBUTE_TIME_ACCESS_NSEC, nsecs + 100,
G_FILE_QUERY_INFO_NONE, NULL, &error);
g_assert_no_error (error);
g_clear_object (&info);
info = g_file_query_info (file,
G_FILE_ATTRIBUTE_TIME_ACCESS "," G_FILE_ATTRIBUTE_TIME_ACCESS_USEC "," G_FILE_ATTRIBUTE_TIME_ACCESS_NSEC,
G_FILE_QUERY_INFO_NONE,
NULL, &error);
g_assert_no_error (error);
new_dt_usecs = g_file_info_get_access_date_time (info);
g_assert_nonnull (new_dt_usecs);
new_usecs = g_date_time_get_microsecond (new_dt_usecs);
new_nsecs = g_file_info_get_attribute_uint32 (info, G_FILE_ATTRIBUTE_TIME_ACCESS_NSEC);
g_assert_cmpuint (new_nsecs, ==, nsecs + 100);
g_assert_cmpuint (new_nsecs, >=, new_usecs * 1000);
g_assert_cmpuint (new_nsecs, <, (new_usecs + 1) * 1000);
g_date_time_unref (new_dt_usecs);
}
/* Clean up. */
g_clear_object (&io_stream);
g_file_delete (file, NULL, NULL);
g_clear_object (&file);
g_clear_object (&info);
g_date_time_unref (dt);
g_date_time_unref (dt_usecs);
g_date_time_unref (dt_new);
g_date_time_unref (dt_new_usecs);
g_date_time_unref (dt_before_epoch);
g_date_time_unref (dt_before_epoch_returned);
}
static void
test_g_file_info_creation_time (void)
{
GFile *file = NULL;
GFileIOStream *io_stream = NULL;
GFileInfo *info = NULL;
GDateTime *dt = NULL, *dt_usecs = NULL, *dt_new = NULL, *dt_new_usecs = NULL,
*dt_before_epoch = NULL, *dt_before_epoch_returned = NULL;
GTimeSpan ts;
gboolean nsecs_supported;
gint usecs;
guint32 nsecs;
GError *error = NULL;
g_test_summary ("Test that getting the creation time of a file works.");
file = g_file_new_tmp ("g-file-info-test-XXXXXX", &io_stream, &error);
g_assert_no_error (error);
info = g_file_query_info (file,
G_FILE_ATTRIBUTE_TIME_CREATED,
G_FILE_QUERY_INFO_NONE,
NULL, &error);
g_assert_no_error (error);
if (!g_file_info_has_attribute (info, G_FILE_ATTRIBUTE_TIME_CREATED))
{
g_test_skip ("Skipping testing creation time as its not supported by the kernel");
g_clear_object (&io_stream);
g_file_delete (file, NULL, NULL);
g_clear_object (&file);
g_clear_object (&info);
return;
}
/* Check the creation time is retrievable. */
dt = g_file_info_get_creation_date_time (info);
/* Try again with microsecond precision. */
g_clear_object (&info);
info = g_file_query_info (file,
G_FILE_ATTRIBUTE_TIME_CREATED "," G_FILE_ATTRIBUTE_TIME_CREATED_USEC,
G_FILE_QUERY_INFO_NONE,
NULL, &error);
g_assert_no_error (error);
dt_usecs = g_file_info_get_creation_date_time (info);
g_assert_nonnull (dt_usecs);
ts = g_date_time_difference (dt_usecs, dt);
g_assert_cmpint (ts, >, 0);
g_assert_cmpint (ts, <, G_USEC_PER_SEC);
/* Try again with nanosecond precision. */
g_clear_object (&info);
info = g_file_query_info (file,
G_FILE_ATTRIBUTE_TIME_CREATED "," G_FILE_ATTRIBUTE_TIME_CREATED_USEC "," G_FILE_ATTRIBUTE_TIME_CREATED_NSEC,
G_FILE_QUERY_INFO_NONE,
NULL, &error);
g_assert_no_error (error);
nsecs_supported = g_file_info_has_attribute (info, G_FILE_ATTRIBUTE_TIME_CREATED_NSEC);
if (nsecs_supported)
{
usecs = g_date_time_get_microsecond (dt_usecs);
nsecs = g_file_info_get_attribute_uint32 (info, G_FILE_ATTRIBUTE_TIME_CREATED_NSEC);
g_assert_cmpuint (nsecs, >=, usecs * 1000);
g_assert_cmpuint (nsecs, <, (usecs + 1) * 1000);
}
/* Try round-tripping the creation time. */
dt_new = g_date_time_add (dt_usecs, G_USEC_PER_SEC + 50);
g_file_info_set_creation_date_time (info, dt_new);
dt_new_usecs = g_file_info_get_creation_date_time (info);
ts = g_date_time_difference (dt_new_usecs, dt_new);
g_assert_cmpint (ts, ==, 0);
// try with a negative timestamp
dt_before_epoch = g_date_time_new_from_unix_utc (-10000);
g_file_info_set_creation_date_time (info, dt_before_epoch);
dt_before_epoch_returned = g_file_info_get_creation_date_time (info);
ts = g_date_time_difference (dt_before_epoch, dt_before_epoch_returned);
g_assert_cmpint (ts, ==, 0);
/* Setting the creation time with usec-precision should have cleared nsecs. */
g_assert_cmpuint (g_file_info_get_attribute_uint32 (info, G_FILE_ATTRIBUTE_TIME_CREATED_NSEC), ==, 0);
/* Try setting the creation time with nsec-precision and it should set the
* usecs too. */
if (nsecs_supported)
{
gint new_usecs;
guint32 new_nsecs;
GDateTime *new_dt_usecs = NULL;
/* This can fail on some platforms, even if reading CREATED_NSEC works */
g_file_set_attribute_uint32 (file, G_FILE_ATTRIBUTE_TIME_CREATED_NSEC, nsecs + 100,
G_FILE_QUERY_INFO_NONE, NULL, &error);
if (error == NULL)
{
g_clear_object (&info);
info = g_file_query_info (file,
G_FILE_ATTRIBUTE_TIME_CREATED "," G_FILE_ATTRIBUTE_TIME_CREATED_USEC "," G_FILE_ATTRIBUTE_TIME_CREATED_NSEC,
G_FILE_QUERY_INFO_NONE,
NULL, &error);
g_assert_no_error (error);
new_dt_usecs = g_file_info_get_creation_date_time (info);
g_assert_nonnull (new_dt_usecs);
new_usecs = g_date_time_get_microsecond (new_dt_usecs);
new_nsecs = g_file_info_get_attribute_uint32 (info, G_FILE_ATTRIBUTE_TIME_CREATED_NSEC);
g_assert_cmpuint (new_nsecs, ==, nsecs + 100);
g_assert_cmpuint (new_nsecs, >=, new_usecs * 1000);
g_assert_cmpuint (new_nsecs, <, (new_usecs + 1) * 1000);
g_date_time_unref (new_dt_usecs);
}
else
{
if (g_error_matches (error, G_IO_ERROR, G_IO_ERROR_NOT_SUPPORTED))
g_clear_error (&error);
g_assert_no_error (error);
}
}
/* Clean up. */
g_clear_object (&io_stream);
g_file_delete (file, NULL, NULL);
g_clear_object (&file);
g_clear_object (&info);
g_date_time_unref (dt);
g_date_time_unref (dt_usecs);
g_date_time_unref (dt_new);
g_date_time_unref (dt_new_usecs);
g_date_time_unref (dt_before_epoch);
g_date_time_unref (dt_before_epoch_returned);
}
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
#ifdef G_OS_WIN32
static void
test_internal_enhanced_stdio (void)
{
char *p0, *p1, *ps;
gboolean try_sparse;
gchar *tmp_dir_root;
wchar_t *tmp_dir_root_w;
gchar *c;
DWORD fsflags;
FILE *f;
SYSTEMTIME st;
FILETIME ft;
HANDLE h;
GStatBuf statbuf_p0, statbuf_p1, statbuf_ps;
GFile *gf_p0, *gf_p1, *gf_ps;
GFileInfo *fi_p0, *fi_p1, *fi_ps;
guint64 size_p0, alsize_p0, size_ps, alsize_ps;
const gchar *id_p0;
const gchar *id_p1;
guint64 time_p0;
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
gchar *tmp_dir;
wchar_t *programdata_dir_w;
wchar_t *users_dir_w;
static const GUID folder_id_programdata =
{ 0x62AB5D82, 0xFDC1, 0x4DC3, { 0xA9, 0xDD, 0x07, 0x0D, 0x1D, 0x49, 0x5D, 0x97 } };
static const GUID folder_id_users =
{ 0x0762D272, 0xC50A, 0x4BB0, { 0xA3, 0x82, 0x69, 0x7D, 0xCD, 0x72, 0x9B, 0x80 } };
2020-01-19 21:51:59 +01:00
GDateTime *dt = NULL, *dt2 = NULL;
GTimeSpan ts;
/* Just before SYSTEMTIME limit (Jan 1 30827) */
const gint64 one_sec_before_systemtime_limit = 910670515199;
gboolean retval;
GError *local_error = NULL;
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
programdata_dir_w = NULL;
SHGetKnownFolderPath (&folder_id_programdata, 0, NULL, &programdata_dir_w);
users_dir_w = NULL;
SHGetKnownFolderPath (&folder_id_users, 0, NULL, &users_dir_w);
if (programdata_dir_w != NULL && users_dir_w != NULL)
{
gchar *programdata;
gchar *users_dir;
gchar *allusers;
gchar *commondata;
GFile *gf_programdata, *gf_allusers, *gf_commondata;
GFileInfo *fi_programdata, *fi_allusers, *fi_allusers_target, *fi_commondata, *fi_commondata_target;
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
GFileType ft_allusers;
GFileType ft_allusers_target;
GFileType ft_programdata;
GFileType ft_commondata;
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
gboolean allusers_is_symlink;
gboolean commondata_is_symlink;
gboolean commondata_is_mount_point;
guint32 allusers_reparse_tag;
guint32 commondata_reparse_tag;
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
const gchar *id_allusers;
const gchar *id_allusers_target;
const gchar *id_commondata_target;
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
const gchar *id_programdata;
const gchar *allusers_target;
const gchar *commondata_target;
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
/* C:/ProgramData */
programdata = g_utf16_to_utf8 (programdata_dir_w, -1, NULL, NULL, NULL);
g_assert_nonnull (programdata);
/* C:/Users */
users_dir = g_utf16_to_utf8 (users_dir_w, -1, NULL, NULL, NULL);
g_assert_nonnull (users_dir);
/* "C:/Users/All Users" is a known directory symlink
* for "C:/ProgramData".
*/
allusers = g_build_filename (users_dir, "All Users", NULL);
/* "C:/Users/All Users/Application Data" is a known
* junction for "C:/ProgramData"
*/
commondata = g_build_filename (allusers, "Application Data", NULL);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
/* We don't test g_stat() and g_lstat() on these directories,
* because it is pointless - there's no way to tell that these
* functions behave correctly in this case
* (st_ino is useless, so we can't tell apart g_stat() and g_lstat()
* results; st_mode is also useless as it does not support S_ISLNK;
* and these directories have no interesting properties other
* than [not] being symlinks).
*/
gf_programdata = g_file_new_for_path (programdata);
gf_allusers = g_file_new_for_path (allusers);
gf_commondata = g_file_new_for_path (commondata);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
fi_programdata = g_file_query_info (gf_programdata,
G_FILE_ATTRIBUTE_ID_FILE ","
G_FILE_ATTRIBUTE_STANDARD_TYPE,
G_FILE_QUERY_INFO_NONE,
NULL, NULL);
fi_allusers_target = g_file_query_info (gf_allusers,
G_FILE_ATTRIBUTE_ID_FILE ","
G_FILE_ATTRIBUTE_STANDARD_TYPE,
G_FILE_QUERY_INFO_NONE,
NULL, NULL);
fi_allusers = g_file_query_info (gf_allusers,
G_FILE_ATTRIBUTE_STANDARD_SYMLINK_TARGET ","
G_FILE_ATTRIBUTE_STANDARD_IS_SYMLINK ","
G_FILE_ATTRIBUTE_DOS_REPARSE_POINT_TAG ","
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
G_FILE_ATTRIBUTE_ID_FILE ","
G_FILE_ATTRIBUTE_STANDARD_TYPE,
G_FILE_QUERY_INFO_NOFOLLOW_SYMLINKS,
NULL, NULL);
fi_commondata = g_file_query_info (gf_commondata,
G_FILE_ATTRIBUTE_STANDARD_SYMLINK_TARGET ","
G_FILE_ATTRIBUTE_STANDARD_IS_SYMLINK ","
G_FILE_ATTRIBUTE_DOS_IS_MOUNTPOINT ","
G_FILE_ATTRIBUTE_DOS_REPARSE_POINT_TAG ","
G_FILE_ATTRIBUTE_ID_FILE ","
G_FILE_ATTRIBUTE_STANDARD_TYPE,
G_FILE_QUERY_INFO_NOFOLLOW_SYMLINKS,
NULL, NULL);
fi_commondata_target = g_file_query_info (gf_commondata,
G_FILE_ATTRIBUTE_ID_FILE ","
G_FILE_ATTRIBUTE_STANDARD_TYPE,
G_FILE_QUERY_INFO_NONE,
NULL, NULL);
g_assert_true (g_file_info_has_attribute (fi_programdata, G_FILE_ATTRIBUTE_ID_FILE));
g_assert_true (g_file_info_has_attribute (fi_programdata, G_FILE_ATTRIBUTE_STANDARD_TYPE));
g_assert_true (g_file_info_has_attribute (fi_allusers_target, G_FILE_ATTRIBUTE_ID_FILE));
g_assert_true (g_file_info_has_attribute (fi_allusers_target, G_FILE_ATTRIBUTE_STANDARD_TYPE));
g_assert_true (g_file_info_has_attribute (fi_commondata_target, G_FILE_ATTRIBUTE_ID_FILE));
g_assert_true (g_file_info_has_attribute (fi_commondata_target, G_FILE_ATTRIBUTE_STANDARD_TYPE));
g_assert_true (g_file_info_has_attribute (fi_allusers, G_FILE_ATTRIBUTE_ID_FILE));
g_assert_true (g_file_info_has_attribute (fi_allusers, G_FILE_ATTRIBUTE_STANDARD_TYPE));
g_assert_true (g_file_info_has_attribute (fi_allusers, G_FILE_ATTRIBUTE_STANDARD_IS_SYMLINK));
g_assert_true (g_file_info_has_attribute (fi_allusers, G_FILE_ATTRIBUTE_DOS_REPARSE_POINT_TAG));
g_assert_true (g_file_info_has_attribute (fi_allusers, G_FILE_ATTRIBUTE_STANDARD_SYMLINK_TARGET));
g_assert_true (g_file_info_has_attribute (fi_commondata, G_FILE_ATTRIBUTE_ID_FILE));
g_assert_true (g_file_info_has_attribute (fi_commondata, G_FILE_ATTRIBUTE_STANDARD_TYPE));
g_assert_true (g_file_info_has_attribute (fi_commondata, G_FILE_ATTRIBUTE_STANDARD_IS_SYMLINK));
g_assert_true (g_file_info_has_attribute (fi_commondata, G_FILE_ATTRIBUTE_DOS_IS_MOUNTPOINT));
g_assert_true (g_file_info_has_attribute (fi_commondata, G_FILE_ATTRIBUTE_DOS_REPARSE_POINT_TAG));
g_assert_true (g_file_info_has_attribute (fi_commondata, G_FILE_ATTRIBUTE_STANDARD_SYMLINK_TARGET));
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
ft_allusers = g_file_info_get_file_type (fi_allusers);
ft_allusers_target = g_file_info_get_file_type (fi_allusers_target);
ft_programdata = g_file_info_get_file_type (fi_programdata);
ft_commondata = g_file_info_get_file_type (fi_commondata);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
g_assert_cmpint (ft_allusers, ==, G_FILE_TYPE_DIRECTORY);
g_assert_cmpint (ft_allusers_target, ==, G_FILE_TYPE_DIRECTORY);
g_assert_cmpint (ft_programdata, ==, G_FILE_TYPE_DIRECTORY);
g_assert_cmpint (ft_commondata, ==, G_FILE_TYPE_DIRECTORY);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
allusers_is_symlink = g_file_info_get_attribute_boolean (fi_allusers, G_FILE_ATTRIBUTE_STANDARD_IS_SYMLINK);
allusers_reparse_tag = g_file_info_get_attribute_uint32 (fi_allusers, G_FILE_ATTRIBUTE_DOS_REPARSE_POINT_TAG);
commondata_is_symlink = g_file_info_get_attribute_boolean (fi_commondata, G_FILE_ATTRIBUTE_STANDARD_IS_SYMLINK);
commondata_is_mount_point = g_file_info_get_attribute_boolean (fi_commondata, G_FILE_ATTRIBUTE_DOS_IS_MOUNTPOINT);
commondata_reparse_tag = g_file_info_get_attribute_uint32 (fi_commondata, G_FILE_ATTRIBUTE_DOS_REPARSE_POINT_TAG);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
g_assert_true (allusers_is_symlink);
g_assert_cmpuint (allusers_reparse_tag, ==, IO_REPARSE_TAG_SYMLINK);
g_assert_true (commondata_is_symlink);
g_assert_true (commondata_is_mount_point);
g_assert_cmpuint (commondata_reparse_tag, ==, IO_REPARSE_TAG_MOUNT_POINT);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
id_allusers = g_file_info_get_attribute_string (fi_allusers, G_FILE_ATTRIBUTE_ID_FILE);
id_allusers_target = g_file_info_get_attribute_string (fi_allusers_target, G_FILE_ATTRIBUTE_ID_FILE);
id_commondata_target = g_file_info_get_attribute_string (fi_commondata_target, G_FILE_ATTRIBUTE_ID_FILE);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
id_programdata = g_file_info_get_attribute_string (fi_programdata, G_FILE_ATTRIBUTE_ID_FILE);
g_assert_cmpstr (id_allusers_target, ==, id_programdata);
g_assert_cmpstr (id_commondata_target, ==, id_programdata);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
g_assert_cmpstr (id_allusers, !=, id_programdata);
allusers_target = g_file_info_get_symlink_target (fi_allusers);
g_assert_true (g_str_has_suffix (allusers_target, "ProgramData"));
commondata_target = g_file_info_get_symlink_target (fi_commondata);
g_assert_true (g_str_has_suffix (commondata_target, "ProgramData"));
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
g_object_unref (fi_allusers);
g_object_unref (fi_allusers_target);
g_object_unref (fi_commondata);
g_object_unref (fi_commondata_target);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
g_object_unref (fi_programdata);
g_object_unref (gf_allusers);
g_object_unref (gf_commondata);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
g_object_unref (gf_programdata);
g_free (allusers);
g_free (commondata);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
g_free (users_dir);
g_free (programdata);
}
if (programdata_dir_w)
CoTaskMemFree (programdata_dir_w);
if (users_dir_w)
CoTaskMemFree (users_dir_w);
tmp_dir = g_dir_make_tmp ("glib_stdio_testXXXXXX", NULL);
g_assert_nonnull (tmp_dir);
/* Technically, this isn't necessary - we already assume NTFS, because of
* symlink support, and NTFS also supports sparse files. Still, given
* the amount of unusual I/O APIs called in this test, checking for
* sparse file support of the filesystem where temp directory is
* doesn't seem to be out of place.
*/
try_sparse = FALSE;
tmp_dir_root = g_strdup (tmp_dir);
/* We need "C:\\" or "C:/", with a trailing [back]slash */
for (c = tmp_dir_root; c && c[0] && c[1]; c++)
if (c[0] == ':')
{
c[2] = '\0';
break;
}
tmp_dir_root_w = g_utf8_to_utf16 (tmp_dir_root, -1, NULL, NULL, NULL);
g_assert_nonnull (tmp_dir_root_w);
g_free (tmp_dir_root);
g_assert_true (GetVolumeInformationW (tmp_dir_root_w, NULL, 0, NULL, NULL, &fsflags, NULL, 0));
g_free (tmp_dir_root_w);
try_sparse = (fsflags & FILE_SUPPORTS_SPARSE_FILES) == FILE_SUPPORTS_SPARSE_FILES;
p0 = g_build_filename (tmp_dir, "zool", NULL);
p1 = g_build_filename (tmp_dir, "looz", NULL);
ps = g_build_filename (tmp_dir, "sparse", NULL);
if (try_sparse)
{
FILE_SET_SPARSE_BUFFER ssb;
FILE_ZERO_DATA_INFORMATION zdi;
g_remove (ps);
f = g_fopen (ps, "wb");
g_assert_nonnull (f);
h = (HANDLE) _get_osfhandle (fileno (f));
g_assert_cmpuint ((guintptr) h, !=, (guintptr) INVALID_HANDLE_VALUE);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
ssb.SetSparse = TRUE;
g_assert_true (DeviceIoControl (h,
FSCTL_SET_SPARSE,
&ssb, sizeof (ssb),
NULL, 0, NULL, NULL));
/* Make it a sparse file that starts with 4GBs of zeros */
zdi.FileOffset.QuadPart = 0;
zdi.BeyondFinalZero.QuadPart = 0xFFFFFFFFULL + 1;
g_assert_true (DeviceIoControl (h,
FSCTL_SET_ZERO_DATA,
&zdi, sizeof (zdi),
NULL, 0, NULL, NULL));
/* Let's not keep this seemingly 4GB monster around
* longer than we absolutely need to. Do all operations
* without assertions, then remove the file immediately.
*/
_fseeki64 (f, 0xFFFFFFFFULL, SEEK_SET);
fprintf (f, "Hello 4GB World!");
fflush (f);
fclose (f);
memset (&statbuf_ps, 0, sizeof (statbuf_ps));
g_stat (ps, &statbuf_ps);
gf_ps = g_file_new_for_path (ps);
fi_ps = g_file_query_info (gf_ps,
G_FILE_ATTRIBUTE_STANDARD_SIZE ","
G_FILE_ATTRIBUTE_STANDARD_ALLOCATED_SIZE,
G_FILE_QUERY_INFO_NONE,
NULL, NULL);
g_remove (ps);
g_assert_true (g_file_info_has_attribute (fi_ps, G_FILE_ATTRIBUTE_STANDARD_SIZE));
g_assert_true (g_file_info_has_attribute (fi_ps, G_FILE_ATTRIBUTE_STANDARD_ALLOCATED_SIZE));
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
size_ps = g_file_info_get_attribute_uint64 (fi_ps, G_FILE_ATTRIBUTE_STANDARD_SIZE);
alsize_ps = g_file_info_get_attribute_uint64 (fi_ps, G_FILE_ATTRIBUTE_STANDARD_ALLOCATED_SIZE);
/* allocated size should small (usually - size of the FS cluster,
* let's assume it's less than 1 gigabyte),
* size should be more than 4 gigabytes,
* st_size should not exceed its 0xFFFFFFFF 32-bit limit,
* and should be nonzero (this also detects a failed g_stat() earlier).
*/
g_assert_cmpuint (alsize_ps, <, 0x40000000);
g_assert_cmpuint (size_ps, >, G_GUINT64_CONSTANT (0xFFFFFFFF));
g_assert_cmpuint (statbuf_ps.st_size, >, 0);
#if defined(_WIN64)
g_assert_cmpuint (statbuf_ps.st_size, ==, G_GUINT64_CONSTANT (0x10000000f));
#else
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
g_assert_cmpuint (statbuf_ps.st_size, <=, 0xFFFFFFFF);
#endif
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
g_object_unref (fi_ps);
g_object_unref (gf_ps);
}
/* Wa-a-ay past 02/07/2106 @ 6:28am (UTC),
* which is the date corresponding to 0xFFFFFFFF + 1.
* This is easier to check than Y2038 (0x80000000 + 1),
* since there's no need to worry about signedness this way.
*/
st.wYear = 2106;
st.wMonth = 2;
st.wDay = 9;
st.wHour = 0;
st.wMinute = 0;
st.wSecond = 0;
st.wMilliseconds = 0;
g_assert_true (SystemTimeToFileTime (&st, &ft));
f = g_fopen (p0, "w");
g_assert_nonnull (f);
h = (HANDLE) _get_osfhandle (fileno (f));
g_assert_cmpuint ((guintptr) h, !=, (guintptr) INVALID_HANDLE_VALUE);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
fprintf (f, "1");
fflush (f);
g_assert_true (SetFileTime (h, &ft, &ft, &ft));
fclose (f);
f = g_fopen (p1, "w");
g_assert_nonnull (f);
fclose (f);
memset (&statbuf_p0, 0, sizeof (statbuf_p0));
memset (&statbuf_p1, 0, sizeof (statbuf_p1));
g_assert_cmpint (g_stat (p0, &statbuf_p0), ==, 0);
g_assert_cmpint (g_stat (p1, &statbuf_p1), ==, 0);
gf_p0 = g_file_new_for_path (p0);
gf_p1 = g_file_new_for_path (p1);
fi_p0 = g_file_query_info (gf_p0,
G_FILE_ATTRIBUTE_STANDARD_SIZE ","
G_FILE_ATTRIBUTE_STANDARD_ALLOCATED_SIZE ","
G_FILE_ATTRIBUTE_ID_FILE ","
2020-01-19 21:51:59 +01:00
G_FILE_ATTRIBUTE_TIME_MODIFIED ","
G_FILE_ATTRIBUTE_TIME_MODIFIED_USEC ","
G_FILE_ATTRIBUTE_TIME_MODIFIED_NSEC,
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
G_FILE_QUERY_INFO_NONE,
NULL, NULL);
fi_p1 = g_file_query_info (gf_p1,
G_FILE_ATTRIBUTE_STANDARD_SIZE ","
G_FILE_ATTRIBUTE_STANDARD_ALLOCATED_SIZE ","
G_FILE_ATTRIBUTE_ID_FILE ","
2020-01-19 21:51:59 +01:00
G_FILE_ATTRIBUTE_TIME_MODIFIED ","
G_FILE_ATTRIBUTE_TIME_MODIFIED_USEC ","
G_FILE_ATTRIBUTE_TIME_MODIFIED_NSEC,
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
G_FILE_QUERY_INFO_NONE,
NULL, NULL);
g_assert_true (g_file_info_has_attribute (fi_p0, G_FILE_ATTRIBUTE_STANDARD_SIZE));
g_assert_true (g_file_info_has_attribute (fi_p0, G_FILE_ATTRIBUTE_STANDARD_ALLOCATED_SIZE));
g_assert_true (g_file_info_has_attribute (fi_p0, G_FILE_ATTRIBUTE_ID_FILE));
g_assert_true (g_file_info_has_attribute (fi_p0, G_FILE_ATTRIBUTE_TIME_MODIFIED));
2020-01-19 21:51:59 +01:00
g_assert_true (g_file_info_has_attribute (fi_p0, G_FILE_ATTRIBUTE_TIME_MODIFIED_USEC));
g_assert_true (g_file_info_has_attribute (fi_p0, G_FILE_ATTRIBUTE_TIME_MODIFIED_NSEC));
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
g_assert_true (g_file_info_has_attribute (fi_p1, G_FILE_ATTRIBUTE_STANDARD_SIZE));
g_assert_true (g_file_info_has_attribute (fi_p1, G_FILE_ATTRIBUTE_STANDARD_ALLOCATED_SIZE));
g_assert_true (g_file_info_has_attribute (fi_p1, G_FILE_ATTRIBUTE_ID_FILE));
g_assert_true (g_file_info_has_attribute (fi_p1, G_FILE_ATTRIBUTE_TIME_MODIFIED));
2020-01-19 21:51:59 +01:00
g_assert_true (g_file_info_has_attribute (fi_p1, G_FILE_ATTRIBUTE_TIME_MODIFIED_USEC));
g_assert_true (g_file_info_has_attribute (fi_p1, G_FILE_ATTRIBUTE_TIME_MODIFIED_NSEC));
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
size_p0 = g_file_info_get_attribute_uint64 (fi_p0, G_FILE_ATTRIBUTE_STANDARD_SIZE);
alsize_p0 = g_file_info_get_attribute_uint64 (fi_p0, G_FILE_ATTRIBUTE_STANDARD_ALLOCATED_SIZE);
/* size should be 1, allocated size should be something else
* (could be 0 or the size of the FS cluster, but never 1).
*/
g_assert_cmpuint (size_p0, ==, statbuf_p0.st_size);
g_assert_cmpuint (size_p0, ==, 1);
g_assert_cmpuint (alsize_p0, !=, size_p0);
id_p0 = g_file_info_get_attribute_string (fi_p0, G_FILE_ATTRIBUTE_ID_FILE);
id_p1 = g_file_info_get_attribute_string (fi_p1, G_FILE_ATTRIBUTE_ID_FILE);
/* st_ino from W32 stat() is useless for file identification.
* It will be either 0, or it will be the same for both files.
*/
g_assert_cmpint (statbuf_p0.st_ino, ==, statbuf_p1.st_ino);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
g_assert_cmpstr (id_p0, !=, id_p1);
time_p0 = g_file_info_get_attribute_uint64 (fi_p0, G_FILE_ATTRIBUTE_TIME_MODIFIED);
/* Check that GFileInfo doesn't suffer from Y2106 problem.
* Don't check stat(), as its contents may vary depending on
* the host platform architecture
* (time fields are 32-bit on 32-bit Windows,
* and 64-bit on 64-bit Windows, usually),
* so it *can* pass this test in some cases.
*/
g_assert_cmpuint (time_p0, >, G_GUINT64_CONSTANT (0xFFFFFFFF));
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
2020-01-19 21:51:59 +01:00
dt = g_file_info_get_modification_date_time (fi_p0);
g_assert_nonnull (dt);
dt2 = g_date_time_add (dt, G_USEC_PER_SEC / 100 * 200);
g_object_unref (fi_p0);
fi_p0 = g_file_info_new ();
g_file_info_set_modification_date_time (fi_p0, dt2);
g_assert_true (g_file_set_attributes_from_info (gf_p0,
fi_p0,
G_FILE_QUERY_INFO_NONE,
NULL,
NULL));
g_date_time_unref (dt2);
g_object_unref (fi_p0);
fi_p0 = g_file_query_info (gf_p0,
G_FILE_ATTRIBUTE_TIME_MODIFIED ","
G_FILE_ATTRIBUTE_TIME_MODIFIED_USEC ","
G_FILE_ATTRIBUTE_TIME_MODIFIED_NSEC,
2020-01-19 21:51:59 +01:00
G_FILE_QUERY_INFO_NONE,
NULL, NULL);
dt2 = g_file_info_get_modification_date_time (fi_p0);
ts = g_date_time_difference (dt2, dt);
g_assert_cmpint (ts, >, 0);
g_assert_cmpint (ts, <, G_USEC_PER_SEC / 100 * 300);
g_date_time_unref (dt);
g_date_time_unref (dt2);
g_file_info_set_attribute_uint64 (fi_p0,
G_FILE_ATTRIBUTE_TIME_MODIFIED,
one_sec_before_systemtime_limit);
g_file_info_set_attribute_uint32 (fi_p0, G_FILE_ATTRIBUTE_TIME_MODIFIED_USEC, 0);
g_assert_true (g_file_set_attributes_from_info (gf_p0,
fi_p0,
G_FILE_QUERY_INFO_NONE,
NULL,
NULL));
g_file_info_set_attribute_uint64 (fi_p0,
G_FILE_ATTRIBUTE_TIME_MODIFIED,
one_sec_before_systemtime_limit + G_USEC_PER_SEC * 2);
g_file_info_set_attribute_uint32 (fi_p0, G_FILE_ATTRIBUTE_TIME_MODIFIED_USEC, 0);
retval = g_file_set_attributes_from_info (gf_p0,
fi_p0,
G_FILE_QUERY_INFO_NONE,
NULL,
&local_error);
g_assert_error (local_error, G_IO_ERROR, G_IO_ERROR_INVALID_DATA);
g_assert_false (retval);
g_clear_error (&local_error);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
g_object_unref (fi_p0);
g_object_unref (fi_p1);
g_object_unref (gf_p0);
g_object_unref (gf_p1);
g_remove (p0);
g_remove (p1);
g_free (p0);
g_free (p1);
g_rmdir (tmp_dir);
}
#endif
static void
test_xattrs (void)
{
GFile *file = NULL;
GFileIOStream *stream = NULL;
GFileInfo *file_info0 = NULL, *file_info1 = NULL, *file_info2 = NULL;
GError *local_error = NULL;
g_test_summary ("Test setting and getting escaped xattrs");
/* Create a temporary file; no need to write anything to it. */
file = g_file_new_tmp ("g-file-info-test-xattrs-XXXXXX", &stream, &local_error);
g_assert_no_error (local_error);
g_assert_nonnull (file);
g_io_stream_close (G_IO_STREAM (stream), NULL, NULL);
g_object_unref (stream);
/* Check the existing xattrs. */
file_info0 = g_file_query_info (file, "xattr::*", G_FILE_QUERY_INFO_NONE, NULL, &local_error);
g_assert_no_error (local_error);
g_assert_nonnull (file_info0);
/* Set some new xattrs, with escaping and some embedded nuls. */
g_file_info_set_attribute_string (file_info0, "xattr::escaped", "hello\\x82\\x80\\xbd");
g_file_info_set_attribute_string (file_info0, "xattr::string", "hi there");
g_file_info_set_attribute_string (file_info0, "xattr::embedded-nul", "hi\\x00there");
g_file_info_set_attribute_string (file_info0, "xattr::deleteme", "this attribute will be deleted");
g_file_set_attributes_from_info (file, file_info0, G_FILE_QUERY_INFO_NONE, NULL, &local_error);
g_object_unref (file_info0);
if (g_error_matches (local_error, G_IO_ERROR, G_IO_ERROR_NOT_SUPPORTED))
{
g_test_skip ("xattrs not supported on this file system");
g_clear_error (&local_error);
}
else
{
g_assert_no_error (local_error);
/* Check they were set properly. */
file_info1 = g_file_query_info (file, "xattr::*", G_FILE_QUERY_INFO_NONE, NULL, &local_error);
g_assert_no_error (local_error);
g_assert_nonnull (file_info1);
g_assert_true (g_file_info_has_namespace (file_info1, "xattr"));
g_assert_cmpstr (g_file_info_get_attribute_string (file_info1, "xattr::escaped"), ==, "hello\\x82\\x80\\xbd");
g_assert_cmpstr (g_file_info_get_attribute_string (file_info1, "xattr::string"), ==, "hi there");
g_assert_cmpstr (g_file_info_get_attribute_string (file_info1, "xattr::embedded-nul"), ==, "hi\\x00there");
g_assert_cmpstr (g_file_info_get_attribute_string (file_info1, "xattr::deleteme"), ==, "this attribute will be deleted");
g_object_unref (file_info1);
/* Check whether removing extended attributes works. */
g_file_set_attribute (file, "xattr::deleteme", G_FILE_ATTRIBUTE_TYPE_INVALID, NULL, G_FILE_QUERY_INFO_NONE, NULL, &local_error);
g_assert_no_error (local_error);
file_info2 = g_file_query_info (file, "xattr::deleteme", G_FILE_QUERY_INFO_NONE, NULL, &local_error);
g_assert_no_error (local_error);
g_assert_nonnull (file_info2);
g_assert_cmpstr (g_file_info_get_attribute_string (file_info2, "xattr::deleteme"), ==, NULL);
g_object_unref (file_info2);
}
/* Tidy up. */
g_file_delete (file, NULL, NULL);
g_object_unref (file);
}
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
int
main (int argc,
char *argv[])
{
g_test_init (&argc, &argv, NULL);
g_test_add_func ("/g-file-info/test_g_file_info", test_g_file_info);
g_test_add_func ("/g-file-info/test_g_file_info/modification-time", test_g_file_info_modification_time);
g_test_add_func ("/g-file-info/test_g_file_info/access-time", test_g_file_info_access_time);
g_test_add_func ("/g-file-info/test_g_file_info/creation-time", test_g_file_info_creation_time);
W32: Add a stat() implementation for private use This commit adds new W32-only functions to gstdio.c, and a new header file, gstdioprivate.h. These functions are: g_win32_stat_utf8() g_win32_lstat_utf8() g_win32_fstat() and they fill a private structure, GWin32PrivateStat, which has all the fields that normal stat has, as well as some extras. These functions are then used throughout glib and gio to get better data about the system. Specifically: * Full, 64-bit size, guaranteed (g_stat() is forced to use 32-bit st_size) * Full, 64-bit file identifier (st_ino is 0 when normal stat() is used, and still is) * W32 File attributes (which stat() doesn't report); in particular, this allows symlinks to be correctly identified * Full, 64-bit time, guaranteed (g_stat() uses 32-bit st_*time on 32-bit Windows) * Allocated file size (as a W32 replacement for the missing st_blocks) st_mode remains unchanged (thus, no S_ISLNK), so when these are given back to glib users (via g_stat(), for example, which is now implemented by calling g_win32_stat_utf8), this field does not contain anything unexpected. g_lstat() now calls g_win32_lstat_utf8(), which works on symlinks the way it's supposed to. Also adds the g_win32_readlink_utf8() function, which behaves like readlink() (including its inability to return 0-terminated strings and inability to say how large the output buffer should be; these limitations are purely for compatibility with existing glib code). Thus, symlink support should now be much better, although far from being complete. A new W32-only test in gio/tests/file.c highlights the following features: * allocated size * 64-bit time * unique file IDs https://bugzilla.gnome.org/show_bug.cgi?id=788180
2017-09-29 12:14:41 +02:00
#ifdef G_OS_WIN32
g_test_add_func ("/g-file-info/internal-enhanced-stdio", test_internal_enhanced_stdio);
#endif
g_test_add_func ("/g-file-info/xattrs", test_xattrs);
return g_test_run();
}