2017-04-15 01:05:30 +08:00
|
|
|
/*
|
2018-12-07 05:29:57 +08:00
|
|
|
* Copyright 2018, University Corporation for Atmospheric Research
|
2017-04-15 01:05:30 +08:00
|
|
|
* See netcdf/COPYRIGHT file for copying and redistribution conditions.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include "config.h"
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <stdio.h>
|
|
|
|
#include <string.h>
|
|
|
|
#include <assert.h>
|
2019-03-17 16:18:18 +08:00
|
|
|
#include <errno.h>
|
2017-04-15 01:05:30 +08:00
|
|
|
#ifdef HAVE_FCNTL_H
|
|
|
|
#include <fcntl.h>
|
|
|
|
#endif
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
#ifdef HAVE_SYS_STAT_H
|
|
|
|
#include <sys/stat.h>
|
|
|
|
#endif
|
2018-03-21 11:31:31 +08:00
|
|
|
#ifdef HAVE_UNISTD_H
|
|
|
|
#include <unistd.h>
|
|
|
|
#endif
|
This PR adds EXPERIMENTAL support for accessing data in the
cloud using a variant of the Zarr protocol and storage
format. This enhancement is generically referred to as "NCZarr".
The data model supported by NCZarr is netcdf-4 minus the user-defined
types and the String type. In this sense it is similar to the CDF-5
data model.
More detailed information about enabling and using NCZarr is
described in the document NUG/nczarr.md and in a
[Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in).
WARNING: this code has had limited testing, so do use this version
for production work. Also, performance improvements are ongoing.
Note especially the following platform matrix of successful tests:
Platform | Build System | S3 support
------------------------------------
Linux+gcc | Automake | yes
Linux+gcc | CMake | yes
Visual Studio | CMake | no
Additionally, and as a consequence of the addition of NCZarr,
major changes have been made to the Filter API. NOTE: NCZarr
does not yet support filters, but these changes are enablers for
that support in the future. Note that it is possible
(probable?) that there will be some accidental reversions if the
changes here did not correctly mimic the existing filter testing.
In any case, previously filter ids and parameters were of type
unsigned int. In order to support the more general zarr filter
model, this was all converted to char*. The old HDF5-specific,
unsigned int operations are still supported but they are
wrappers around the new, char* based nc_filterx_XXX functions.
This entailed at least the following changes:
1. Added the files libdispatch/dfilterx.c and include/ncfilter.h
2. Some filterx utilities have been moved to libdispatch/daux.c
3. A new entry, "filter_actions" was added to the NCDispatch table
and the version bumped.
4. An overly complex set of structs was created to support funnelling
all of the filterx operations thru a single dispatch
"filter_actions" entry.
5. Move common code to from libhdf5 to libsrc4 so that it is accessible
to nczarr.
Changes directly related to Zarr:
1. Modified CMakeList.txt and configure.ac to support both C and C++
-- this is in support of S3 support via the awd-sdk libraries.
2. Define a size64_t type to support nczarr.
3. More reworking of libdispatch/dinfermodel.c to
support zarr and to regularize the structure of the fragments
section of a URL.
Changes not directly related to Zarr:
1. Make client-side filter registration be conditional, with default off.
2. Hack include/nc4internal.h to make some flags added by Ed be unique:
e.g. NC_CREAT, NC_INDEF, etc.
3. cleanup include/nchttp.h and libdispatch/dhttp.c.
4. Misc. changes to support compiling under Visual Studio including:
* Better testing under windows for dirent.h and opendir and closedir.
5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags
and to centralize error reporting.
6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them.
7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible.
Changes Left TO-DO:
1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
|
|
|
#ifdef HAVE_DIRENT_H
|
|
|
|
#include <dirent.h>
|
|
|
|
#endif
|
|
|
|
#ifdef _WIN32
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
#include <windows.h>
|
2017-04-15 01:05:30 +08:00
|
|
|
#include <io.h>
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
#include <wchar.h>
|
|
|
|
#include <locale.h>
|
2020-12-07 09:19:53 +08:00
|
|
|
#include <direct.h>
|
2017-04-15 01:05:30 +08:00
|
|
|
#endif
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
#ifdef __hpux
|
|
|
|
#include <locale.h>
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#include "netcdf.h"
|
2020-10-14 09:12:15 +08:00
|
|
|
#include "ncpathmgr.h"
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
#include "nclog.h"
|
|
|
|
#include "nclist.h"
|
|
|
|
#include "ncbytes.h"
|
|
|
|
#include "ncuri.h"
|
|
|
|
#include "ncutf8.h"
|
2018-12-06 10:20:43 +08:00
|
|
|
|
2017-07-16 03:03:35 +08:00
|
|
|
#undef PATHFORMAT
|
|
|
|
|
2020-12-07 09:19:53 +08:00
|
|
|
#ifdef _WIN32
|
|
|
|
#define access _access
|
|
|
|
#define mkdir _mkdir
|
|
|
|
#define rmdir _rmdir
|
|
|
|
#define getcwd _getcwd
|
|
|
|
#endif
|
|
|
|
|
2017-04-15 01:05:30 +08:00
|
|
|
/*
|
|
|
|
Code to provide some path conversion code so that
|
|
|
|
cygwin and (some) mingw paths can be passed to open/fopen
|
|
|
|
for Windows. Other cases will be added as needed.
|
|
|
|
Rules:
|
|
|
|
1. a leading single alpha-character path element (e.g. /D/...)
|
|
|
|
will be interpreted as a windows drive letter.
|
|
|
|
2. a leading '/cygdrive/X' will be converted to
|
|
|
|
a drive letter X if X is alpha-char.
|
|
|
|
3. a leading D:/... is treated as a windows drive letter
|
2018-03-21 11:31:31 +08:00
|
|
|
5. If any of the above is encountered, then forward slashes
|
2017-04-15 01:05:30 +08:00
|
|
|
will be converted to backslashes.
|
2018-03-21 11:31:31 +08:00
|
|
|
All other cases are passed thru unchanged
|
2017-04-15 01:05:30 +08:00
|
|
|
*/
|
|
|
|
|
|
|
|
/* Define legal windows drive letters */
|
2018-10-02 05:51:43 +08:00
|
|
|
static const char* windrive = "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ";
|
2017-04-15 01:05:30 +08:00
|
|
|
|
2018-10-02 05:51:43 +08:00
|
|
|
static const size_t cdlen = 10; /* strlen("/cygdrive/") */
|
2017-04-15 01:05:30 +08:00
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
static int pathinitialized = 0;
|
|
|
|
|
2017-07-15 09:37:35 +08:00
|
|
|
static int pathdebug = -1;
|
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
static const struct Path {
|
|
|
|
int kind;
|
|
|
|
int drive;
|
|
|
|
char* path;
|
|
|
|
} empty = {NCPD_UNKNOWN,0,NULL};
|
|
|
|
|
|
|
|
/* Keep the working directory kind and drive */
|
|
|
|
static struct Path wdpath = {NCPD_UNKNOWN,0,NULL};
|
|
|
|
static char wdstaticpath[8192];
|
|
|
|
|
|
|
|
static int parsepath(const char* inpath, struct Path* path);
|
|
|
|
static int unparsepath(struct Path* p, char** pathp);
|
|
|
|
static int getwdpath(struct Path* wd);
|
|
|
|
static char* printPATH(struct Path* p);
|
|
|
|
static int getlocalpathkind(void);
|
|
|
|
static void clearPath(struct Path* path);
|
|
|
|
static void pathinit(void);
|
|
|
|
static int iscygwinspecial(const char* path);
|
|
|
|
static int testurl(const char* path);
|
|
|
|
|
|
|
|
#ifdef WINPATH
|
|
|
|
static int ansi2utf8(const char* local, char** u8p);
|
|
|
|
static int ansi2wide(const char* local, wchar_t** u16p);
|
|
|
|
static int utf82wide(const char* utf8, wchar_t** u16p);
|
|
|
|
static int wide2utf8(const wchar_t* u16, char** u8p);
|
|
|
|
#endif
|
2018-03-21 11:31:31 +08:00
|
|
|
|
2017-04-15 01:05:30 +08:00
|
|
|
EXTERNL
|
|
|
|
char* /* caller frees */
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
NCpathcvt(const char* inpath)
|
2017-04-15 01:05:30 +08:00
|
|
|
{
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
int stat = NC_NOERR;
|
|
|
|
char* tmp1 = NULL;
|
|
|
|
struct Path canon = empty;
|
2017-04-15 01:05:30 +08:00
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
if(inpath == NULL) goto done; /* defensive driving */
|
2017-07-15 09:37:35 +08:00
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
if(!pathinitialized) pathinit();
|
2020-08-18 09:15:47 +08:00
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
if(testurl(inpath)) { /* Pass thru URLs */
|
|
|
|
if((tmp1 = strdup(inpath))==NULL) stat = NC_ENOMEM;
|
|
|
|
goto done;
|
2017-07-15 09:37:35 +08:00
|
|
|
}
|
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
if((stat = parsepath(inpath,&canon))) {goto done;}
|
2017-04-15 01:05:30 +08:00
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
/* Special check for special cygwin paths: /tmp,usr etc */
|
|
|
|
if(getlocalpathkind() == NCPD_CYGWIN
|
|
|
|
&& iscygwinspecial(canon.path)
|
|
|
|
&& canon.kind == NCPD_NIX)
|
|
|
|
canon.kind = NCPD_CYGWIN;
|
2017-04-15 01:05:30 +08:00
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
if(canon.kind != NCPD_REL && wdpath.kind != canon.kind) {
|
|
|
|
nclog(NCLOGWARN,"NCpathcvt: path mismatch: platform=%d inpath=%d\n",
|
|
|
|
wdpath.kind,canon.kind);
|
|
|
|
canon.kind = wdpath.kind; /* override */
|
2017-04-15 01:05:30 +08:00
|
|
|
}
|
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
if((stat = unparsepath(&canon,&tmp1))) {goto done;}
|
|
|
|
done:
|
|
|
|
if(pathdebug) {
|
|
|
|
fprintf(stderr,"xxx: inpath=|%s| outpath=|%s|\n",
|
|
|
|
inpath?inpath:"NULL",tmp1?tmp1:"NULL");
|
|
|
|
fflush(stderr);
|
2017-04-15 01:05:30 +08:00
|
|
|
}
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
if(stat) {
|
|
|
|
nullfree(tmp1); tmp1 = NULL;
|
|
|
|
nclog(NCLOGERR,"NCpathcvt: stat=%d (%s)",
|
|
|
|
stat,nc_strerror(stat));
|
2018-03-21 11:31:31 +08:00
|
|
|
}
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
clearPath(&canon);
|
|
|
|
return tmp1;
|
|
|
|
}
|
2018-03-21 11:31:31 +08:00
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
EXTERNL
|
|
|
|
char* /* caller frees */
|
|
|
|
NCpathabsolute(const char* relpath)
|
|
|
|
{
|
|
|
|
int stat = NC_NOERR;
|
|
|
|
struct Path canon = empty;
|
|
|
|
char* tmp1 = NULL;
|
|
|
|
char* result = NULL;
|
|
|
|
size_t len;
|
|
|
|
|
|
|
|
if(relpath == NULL) goto done; /* defensive driving */
|
|
|
|
|
|
|
|
if(!pathinitialized) pathinit();
|
|
|
|
|
|
|
|
/* Canonicalize relpath */
|
|
|
|
if((stat = parsepath(relpath,&canon))) {goto done;}
|
|
|
|
|
|
|
|
/* See if relative */
|
|
|
|
if(canon.kind == NCPD_REL) {
|
|
|
|
/* prepend the wd path to the inpath, including drive letter, if any */
|
|
|
|
len = strlen(wdpath.path)+strlen(canon.path)+1+1;
|
|
|
|
if((tmp1 = (char*)malloc(len))==NULL)
|
|
|
|
{stat = NC_ENOMEM; {goto done;}}
|
|
|
|
tmp1[0] = '\0';
|
|
|
|
strlcat(tmp1,wdpath.path,len);
|
|
|
|
strlcat(tmp1,"/",len);
|
|
|
|
strlcat(tmp1,canon.path,len);
|
|
|
|
nullfree(canon.path);
|
|
|
|
canon.path = tmp1; tmp1 = NULL;
|
|
|
|
canon.drive = wdpath.drive;
|
|
|
|
canon.kind = wdpath.kind;
|
2017-07-14 00:40:07 +08:00
|
|
|
}
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
/* rebuild */
|
|
|
|
if((stat=unparsepath(&canon,&result))) goto done;
|
2017-04-15 01:05:30 +08:00
|
|
|
done:
|
2017-07-15 09:37:35 +08:00
|
|
|
if(pathdebug) {
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
fprintf(stderr,"xxx: relpath=|%s| result=|%s|\n",
|
|
|
|
relpath?relpath:"NULL",result?result:"NULL");
|
2017-07-15 09:37:35 +08:00
|
|
|
fflush(stderr);
|
|
|
|
}
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
if(stat) {
|
|
|
|
nullfree(tmp1); tmp1 = NULL;
|
|
|
|
nclog(NCLOGERR,"NCpathcvt: stat=%d (%s)",
|
|
|
|
stat,nc_strerror(stat));
|
|
|
|
}
|
|
|
|
clearPath(&canon);
|
|
|
|
nullfree(tmp1);
|
|
|
|
return result;
|
2017-04-15 01:05:30 +08:00
|
|
|
}
|
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
|
|
|
|
/* Testing support */
|
|
|
|
/* Force drive and wd before invoking NCpathcvt
|
|
|
|
and then revert */
|
2020-08-18 09:15:47 +08:00
|
|
|
EXTERNL
|
|
|
|
char* /* caller frees */
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
NCpathcvt_test(const char* inpath, int ukind, int udrive)
|
2020-08-18 09:15:47 +08:00
|
|
|
{
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
char* result = NULL;
|
|
|
|
struct Path oldwd = wdpath;
|
|
|
|
|
|
|
|
if(!pathinitialized) pathinit();
|
|
|
|
/* Override */
|
|
|
|
wdpath.kind = ukind;
|
|
|
|
wdpath.drive = udrive;
|
|
|
|
wdpath.path = strdup("/");
|
|
|
|
if(pathdebug)
|
|
|
|
fprintf(stderr,"xxx: wd=|%s|",printPATH(&wdpath));
|
|
|
|
result = NCpathcvt(inpath);
|
|
|
|
clearPath(&wdpath);
|
|
|
|
wdpath = oldwd;
|
|
|
|
return result;
|
2020-08-18 09:15:47 +08:00
|
|
|
}
|
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
static void
|
|
|
|
pathinit(void)
|
2018-03-21 11:31:31 +08:00
|
|
|
{
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
if(pathinitialized) return;
|
|
|
|
|
|
|
|
/* Check for path debug env vars */
|
|
|
|
if(pathdebug < 0) {
|
|
|
|
const char* s = getenv("NCPATHDEBUG");
|
|
|
|
pathdebug = (s == NULL ? 0 : 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
(void)getwdpath(&wdpath);
|
|
|
|
/* make the path static but remember to never free it (Ugh!) */
|
|
|
|
wdstaticpath[0] = '\0';
|
|
|
|
strlcat(wdstaticpath,wdpath.path,sizeof(wdstaticpath));
|
|
|
|
clearPath(&wdpath);
|
|
|
|
wdpath.path = wdstaticpath;
|
|
|
|
|
|
|
|
pathinitialized = 1;
|
2018-03-21 11:31:31 +08:00
|
|
|
}
|
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
static void
|
|
|
|
clearPath(struct Path* path)
|
This PR adds EXPERIMENTAL support for accessing data in the
cloud using a variant of the Zarr protocol and storage
format. This enhancement is generically referred to as "NCZarr".
The data model supported by NCZarr is netcdf-4 minus the user-defined
types and the String type. In this sense it is similar to the CDF-5
data model.
More detailed information about enabling and using NCZarr is
described in the document NUG/nczarr.md and in a
[Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in).
WARNING: this code has had limited testing, so do use this version
for production work. Also, performance improvements are ongoing.
Note especially the following platform matrix of successful tests:
Platform | Build System | S3 support
------------------------------------
Linux+gcc | Automake | yes
Linux+gcc | CMake | yes
Visual Studio | CMake | no
Additionally, and as a consequence of the addition of NCZarr,
major changes have been made to the Filter API. NOTE: NCZarr
does not yet support filters, but these changes are enablers for
that support in the future. Note that it is possible
(probable?) that there will be some accidental reversions if the
changes here did not correctly mimic the existing filter testing.
In any case, previously filter ids and parameters were of type
unsigned int. In order to support the more general zarr filter
model, this was all converted to char*. The old HDF5-specific,
unsigned int operations are still supported but they are
wrappers around the new, char* based nc_filterx_XXX functions.
This entailed at least the following changes:
1. Added the files libdispatch/dfilterx.c and include/ncfilter.h
2. Some filterx utilities have been moved to libdispatch/daux.c
3. A new entry, "filter_actions" was added to the NCDispatch table
and the version bumped.
4. An overly complex set of structs was created to support funnelling
all of the filterx operations thru a single dispatch
"filter_actions" entry.
5. Move common code to from libhdf5 to libsrc4 so that it is accessible
to nczarr.
Changes directly related to Zarr:
1. Modified CMakeList.txt and configure.ac to support both C and C++
-- this is in support of S3 support via the awd-sdk libraries.
2. Define a size64_t type to support nczarr.
3. More reworking of libdispatch/dinfermodel.c to
support zarr and to regularize the structure of the fragments
section of a URL.
Changes not directly related to Zarr:
1. Make client-side filter registration be conditional, with default off.
2. Hack include/nc4internal.h to make some flags added by Ed be unique:
e.g. NC_CREAT, NC_INDEF, etc.
3. cleanup include/nchttp.h and libdispatch/dhttp.c.
4. Misc. changes to support compiling under Visual Studio including:
* Better testing under windows for dirent.h and opendir and closedir.
5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags
and to centralize error reporting.
6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them.
7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible.
Changes Left TO-DO:
1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
|
|
|
{
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
nullfree(path->path);
|
|
|
|
path->path = NULL;
|
This PR adds EXPERIMENTAL support for accessing data in the
cloud using a variant of the Zarr protocol and storage
format. This enhancement is generically referred to as "NCZarr".
The data model supported by NCZarr is netcdf-4 minus the user-defined
types and the String type. In this sense it is similar to the CDF-5
data model.
More detailed information about enabling and using NCZarr is
described in the document NUG/nczarr.md and in a
[Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in).
WARNING: this code has had limited testing, so do use this version
for production work. Also, performance improvements are ongoing.
Note especially the following platform matrix of successful tests:
Platform | Build System | S3 support
------------------------------------
Linux+gcc | Automake | yes
Linux+gcc | CMake | yes
Visual Studio | CMake | no
Additionally, and as a consequence of the addition of NCZarr,
major changes have been made to the Filter API. NOTE: NCZarr
does not yet support filters, but these changes are enablers for
that support in the future. Note that it is possible
(probable?) that there will be some accidental reversions if the
changes here did not correctly mimic the existing filter testing.
In any case, previously filter ids and parameters were of type
unsigned int. In order to support the more general zarr filter
model, this was all converted to char*. The old HDF5-specific,
unsigned int operations are still supported but they are
wrappers around the new, char* based nc_filterx_XXX functions.
This entailed at least the following changes:
1. Added the files libdispatch/dfilterx.c and include/ncfilter.h
2. Some filterx utilities have been moved to libdispatch/daux.c
3. A new entry, "filter_actions" was added to the NCDispatch table
and the version bumped.
4. An overly complex set of structs was created to support funnelling
all of the filterx operations thru a single dispatch
"filter_actions" entry.
5. Move common code to from libhdf5 to libsrc4 so that it is accessible
to nczarr.
Changes directly related to Zarr:
1. Modified CMakeList.txt and configure.ac to support both C and C++
-- this is in support of S3 support via the awd-sdk libraries.
2. Define a size64_t type to support nczarr.
3. More reworking of libdispatch/dinfermodel.c to
support zarr and to regularize the structure of the fragments
section of a URL.
Changes not directly related to Zarr:
1. Make client-side filter registration be conditional, with default off.
2. Hack include/nc4internal.h to make some flags added by Ed be unique:
e.g. NC_CREAT, NC_INDEF, etc.
3. cleanup include/nchttp.h and libdispatch/dhttp.c.
4. Misc. changes to support compiling under Visual Studio including:
* Better testing under windows for dirent.h and opendir and closedir.
5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags
and to centralize error reporting.
6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them.
7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible.
Changes Left TO-DO:
1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
|
|
|
}
|
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
static const char* cygwinspecial[] =
|
|
|
|
{"/bin/","/dev/","/etc/","/home/",
|
|
|
|
"/lib/","/proc/","/sbin/","/tmp/",
|
|
|
|
"/usr/","/var/",NULL};
|
2020-08-18 09:15:47 +08:00
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
/* Unfortunately, not all cygwin paths start with /cygdrive.
|
|
|
|
So see if the path starts with one of the special paths.
|
|
|
|
*/
|
|
|
|
static int
|
|
|
|
iscygwinspecial(const char* path)
|
2020-08-18 09:15:47 +08:00
|
|
|
{
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
const char** p;
|
|
|
|
if(path == NULL) return 0;
|
|
|
|
for(p=cygwinspecial;*p;p++) {
|
|
|
|
if(strncmp(*p,path,strlen(*p))==0) return 1;
|
|
|
|
}
|
2020-08-18 09:15:47 +08:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
/* return 1 if path looks like a url; 0 otherwise */
|
|
|
|
static int
|
|
|
|
testurl(const char* path)
|
|
|
|
{
|
|
|
|
int isurl = 0;
|
|
|
|
NCURI* tmpurl = NULL;
|
|
|
|
|
|
|
|
if(path == NULL) return 0;
|
|
|
|
|
|
|
|
/* Ok, try to parse as a url */
|
|
|
|
ncuriparse(path,&tmpurl);
|
|
|
|
isurl = (tmpurl == NULL?0:1);
|
|
|
|
ncurifree(tmpurl);
|
|
|
|
return isurl;
|
|
|
|
}
|
|
|
|
|
2017-04-15 01:05:30 +08:00
|
|
|
#ifdef WINPATH
|
|
|
|
|
|
|
|
/*
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
Provide wrappers for Path-related functions
|
2017-04-15 01:05:30 +08:00
|
|
|
*/
|
|
|
|
|
|
|
|
EXTERNL
|
|
|
|
FILE*
|
|
|
|
NCfopen(const char* path, const char* flags)
|
|
|
|
{
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
int stat = NC_NOERR;
|
2017-04-15 01:05:30 +08:00
|
|
|
FILE* f = NULL;
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
char* cvtpath = NULL;
|
|
|
|
wchar_t* wpath = NULL;
|
|
|
|
wchar_t* wflags = NULL;
|
|
|
|
cvtpath = NCpathcvt(path);
|
|
|
|
if(cvtpath == NULL) return NULL;
|
|
|
|
/* Convert from local to wide */
|
|
|
|
if((stat = utf82wide(cvtpath,&wpath))) goto done;
|
|
|
|
if((stat = ansi2wide(flags,&wflags))) goto done;
|
|
|
|
f = _wfopen(wpath,wflags);
|
|
|
|
done:
|
|
|
|
nullfree(cvtpath);
|
|
|
|
nullfree(wpath);
|
|
|
|
nullfree(wflags);
|
2017-04-15 01:05:30 +08:00
|
|
|
return f;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXTERNL
|
|
|
|
int
|
|
|
|
NCopen3(const char* path, int flags, int mode)
|
|
|
|
{
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
int stat = NC_NOERR;
|
2017-04-15 01:05:30 +08:00
|
|
|
int fd = -1;
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
char* cvtpath = NULL;
|
|
|
|
wchar_t* wpath = NULL;
|
|
|
|
cvtpath = NCpathcvt(path);
|
|
|
|
if(cvtpath == NULL) goto done;
|
|
|
|
/* Convert from utf8 to wide */
|
|
|
|
if((stat = utf82wide(cvtpath,&wpath))) goto done;
|
|
|
|
fd = _wopen(wpath,flags,mode);
|
|
|
|
done:
|
|
|
|
nullfree(cvtpath);
|
|
|
|
nullfree(wpath);
|
2017-04-15 01:05:30 +08:00
|
|
|
return fd;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXTERNL
|
|
|
|
int
|
|
|
|
NCopen2(const char *path, int flags)
|
|
|
|
{
|
|
|
|
return NCopen3(path,flags,0);
|
|
|
|
}
|
|
|
|
|
This PR adds EXPERIMENTAL support for accessing data in the
cloud using a variant of the Zarr protocol and storage
format. This enhancement is generically referred to as "NCZarr".
The data model supported by NCZarr is netcdf-4 minus the user-defined
types and the String type. In this sense it is similar to the CDF-5
data model.
More detailed information about enabling and using NCZarr is
described in the document NUG/nczarr.md and in a
[Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in).
WARNING: this code has had limited testing, so do use this version
for production work. Also, performance improvements are ongoing.
Note especially the following platform matrix of successful tests:
Platform | Build System | S3 support
------------------------------------
Linux+gcc | Automake | yes
Linux+gcc | CMake | yes
Visual Studio | CMake | no
Additionally, and as a consequence of the addition of NCZarr,
major changes have been made to the Filter API. NOTE: NCZarr
does not yet support filters, but these changes are enablers for
that support in the future. Note that it is possible
(probable?) that there will be some accidental reversions if the
changes here did not correctly mimic the existing filter testing.
In any case, previously filter ids and parameters were of type
unsigned int. In order to support the more general zarr filter
model, this was all converted to char*. The old HDF5-specific,
unsigned int operations are still supported but they are
wrappers around the new, char* based nc_filterx_XXX functions.
This entailed at least the following changes:
1. Added the files libdispatch/dfilterx.c and include/ncfilter.h
2. Some filterx utilities have been moved to libdispatch/daux.c
3. A new entry, "filter_actions" was added to the NCDispatch table
and the version bumped.
4. An overly complex set of structs was created to support funnelling
all of the filterx operations thru a single dispatch
"filter_actions" entry.
5. Move common code to from libhdf5 to libsrc4 so that it is accessible
to nczarr.
Changes directly related to Zarr:
1. Modified CMakeList.txt and configure.ac to support both C and C++
-- this is in support of S3 support via the awd-sdk libraries.
2. Define a size64_t type to support nczarr.
3. More reworking of libdispatch/dinfermodel.c to
support zarr and to regularize the structure of the fragments
section of a URL.
Changes not directly related to Zarr:
1. Make client-side filter registration be conditional, with default off.
2. Hack include/nc4internal.h to make some flags added by Ed be unique:
e.g. NC_CREAT, NC_INDEF, etc.
3. cleanup include/nchttp.h and libdispatch/dhttp.c.
4. Misc. changes to support compiling under Visual Studio including:
* Better testing under windows for dirent.h and opendir and closedir.
5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags
and to centralize error reporting.
6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them.
7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible.
Changes Left TO-DO:
1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
|
|
|
#ifdef HAVE_DIRENT_H
|
|
|
|
EXTERNL
|
|
|
|
DIR*
|
|
|
|
NCopendir(const char* path)
|
|
|
|
{
|
|
|
|
DIR* ent = NULL;
|
|
|
|
char* cvtname = NCpathcvt(path);
|
2020-12-07 09:19:53 +08:00
|
|
|
if(cvtname == NULL) return NULL;
|
This PR adds EXPERIMENTAL support for accessing data in the
cloud using a variant of the Zarr protocol and storage
format. This enhancement is generically referred to as "NCZarr".
The data model supported by NCZarr is netcdf-4 minus the user-defined
types and the String type. In this sense it is similar to the CDF-5
data model.
More detailed information about enabling and using NCZarr is
described in the document NUG/nczarr.md and in a
[Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in).
WARNING: this code has had limited testing, so do use this version
for production work. Also, performance improvements are ongoing.
Note especially the following platform matrix of successful tests:
Platform | Build System | S3 support
------------------------------------
Linux+gcc | Automake | yes
Linux+gcc | CMake | yes
Visual Studio | CMake | no
Additionally, and as a consequence of the addition of NCZarr,
major changes have been made to the Filter API. NOTE: NCZarr
does not yet support filters, but these changes are enablers for
that support in the future. Note that it is possible
(probable?) that there will be some accidental reversions if the
changes here did not correctly mimic the existing filter testing.
In any case, previously filter ids and parameters were of type
unsigned int. In order to support the more general zarr filter
model, this was all converted to char*. The old HDF5-specific,
unsigned int operations are still supported but they are
wrappers around the new, char* based nc_filterx_XXX functions.
This entailed at least the following changes:
1. Added the files libdispatch/dfilterx.c and include/ncfilter.h
2. Some filterx utilities have been moved to libdispatch/daux.c
3. A new entry, "filter_actions" was added to the NCDispatch table
and the version bumped.
4. An overly complex set of structs was created to support funnelling
all of the filterx operations thru a single dispatch
"filter_actions" entry.
5. Move common code to from libhdf5 to libsrc4 so that it is accessible
to nczarr.
Changes directly related to Zarr:
1. Modified CMakeList.txt and configure.ac to support both C and C++
-- this is in support of S3 support via the awd-sdk libraries.
2. Define a size64_t type to support nczarr.
3. More reworking of libdispatch/dinfermodel.c to
support zarr and to regularize the structure of the fragments
section of a URL.
Changes not directly related to Zarr:
1. Make client-side filter registration be conditional, with default off.
2. Hack include/nc4internal.h to make some flags added by Ed be unique:
e.g. NC_CREAT, NC_INDEF, etc.
3. cleanup include/nchttp.h and libdispatch/dhttp.c.
4. Misc. changes to support compiling under Visual Studio including:
* Better testing under windows for dirent.h and opendir and closedir.
5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags
and to centralize error reporting.
6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them.
7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible.
Changes Left TO-DO:
1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
|
|
|
ent = opendir(cvtname);
|
|
|
|
free(cvtname);
|
|
|
|
return ent;
|
|
|
|
}
|
|
|
|
|
|
|
|
EXTERNL
|
|
|
|
int
|
|
|
|
NCclosedir(DIR* ent)
|
|
|
|
{
|
2020-12-07 09:19:53 +08:00
|
|
|
int stat = NC_NOERR;
|
2021-05-19 04:06:12 +08:00
|
|
|
if(closedir(ent) < 0) stat = errno;
|
This PR adds EXPERIMENTAL support for accessing data in the
cloud using a variant of the Zarr protocol and storage
format. This enhancement is generically referred to as "NCZarr".
The data model supported by NCZarr is netcdf-4 minus the user-defined
types and the String type. In this sense it is similar to the CDF-5
data model.
More detailed information about enabling and using NCZarr is
described in the document NUG/nczarr.md and in a
[Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in).
WARNING: this code has had limited testing, so do use this version
for production work. Also, performance improvements are ongoing.
Note especially the following platform matrix of successful tests:
Platform | Build System | S3 support
------------------------------------
Linux+gcc | Automake | yes
Linux+gcc | CMake | yes
Visual Studio | CMake | no
Additionally, and as a consequence of the addition of NCZarr,
major changes have been made to the Filter API. NOTE: NCZarr
does not yet support filters, but these changes are enablers for
that support in the future. Note that it is possible
(probable?) that there will be some accidental reversions if the
changes here did not correctly mimic the existing filter testing.
In any case, previously filter ids and parameters were of type
unsigned int. In order to support the more general zarr filter
model, this was all converted to char*. The old HDF5-specific,
unsigned int operations are still supported but they are
wrappers around the new, char* based nc_filterx_XXX functions.
This entailed at least the following changes:
1. Added the files libdispatch/dfilterx.c and include/ncfilter.h
2. Some filterx utilities have been moved to libdispatch/daux.c
3. A new entry, "filter_actions" was added to the NCDispatch table
and the version bumped.
4. An overly complex set of structs was created to support funnelling
all of the filterx operations thru a single dispatch
"filter_actions" entry.
5. Move common code to from libhdf5 to libsrc4 so that it is accessible
to nczarr.
Changes directly related to Zarr:
1. Modified CMakeList.txt and configure.ac to support both C and C++
-- this is in support of S3 support via the awd-sdk libraries.
2. Define a size64_t type to support nczarr.
3. More reworking of libdispatch/dinfermodel.c to
support zarr and to regularize the structure of the fragments
section of a URL.
Changes not directly related to Zarr:
1. Make client-side filter registration be conditional, with default off.
2. Hack include/nc4internal.h to make some flags added by Ed be unique:
e.g. NC_CREAT, NC_INDEF, etc.
3. cleanup include/nchttp.h and libdispatch/dhttp.c.
4. Misc. changes to support compiling under Visual Studio including:
* Better testing under windows for dirent.h and opendir and closedir.
5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags
and to centralize error reporting.
6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them.
7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible.
Changes Left TO-DO:
1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
|
|
|
return stat;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
2018-09-05 01:27:47 +08:00
|
|
|
/*
|
|
|
|
Provide wrappers for other file system functions
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* Return access applied to path+mode */
|
|
|
|
EXTERNL
|
|
|
|
int
|
|
|
|
NCaccess(const char* path, int mode)
|
|
|
|
{
|
|
|
|
int status = 0;
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
char* cvtpath = NULL;
|
|
|
|
wchar_t* wpath = NULL;
|
|
|
|
if((cvtpath = NCpathcvt(path)) == NULL)
|
|
|
|
{status = EINVAL; goto done;}
|
|
|
|
if((status = utf82wide(cvtpath,&wpath))) {status = ENOENT; goto done;}
|
|
|
|
if(_waccess(wpath,mode) < 0) {status = errno; goto done;}
|
|
|
|
done:
|
|
|
|
free(cvtpath);
|
|
|
|
free(wpath);
|
|
|
|
errno = status;
|
|
|
|
return (errno?-1:0);
|
2018-09-05 01:27:47 +08:00
|
|
|
}
|
|
|
|
|
|
|
|
EXTERNL
|
|
|
|
int
|
|
|
|
NCremove(const char* path)
|
|
|
|
{
|
|
|
|
int status = 0;
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
char* cvtpath = NULL;
|
|
|
|
wchar_t* wpath = NULL;
|
|
|
|
if((cvtpath = NCpathcvt(path)) == NULL) {status=ENOMEM; goto done;}
|
|
|
|
if((status = utf82wide(cvtpath,&wpath))) {status = ENOENT; goto done;}
|
|
|
|
if(_wremove(wpath) < 0) {status = errno; goto done;}
|
|
|
|
done:
|
|
|
|
free(cvtpath);
|
|
|
|
free(wpath);
|
|
|
|
errno = status;
|
|
|
|
return (errno?-1:0);
|
2018-09-05 01:27:47 +08:00
|
|
|
}
|
|
|
|
|
This PR adds EXPERIMENTAL support for accessing data in the
cloud using a variant of the Zarr protocol and storage
format. This enhancement is generically referred to as "NCZarr".
The data model supported by NCZarr is netcdf-4 minus the user-defined
types and the String type. In this sense it is similar to the CDF-5
data model.
More detailed information about enabling and using NCZarr is
described in the document NUG/nczarr.md and in a
[Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in).
WARNING: this code has had limited testing, so do use this version
for production work. Also, performance improvements are ongoing.
Note especially the following platform matrix of successful tests:
Platform | Build System | S3 support
------------------------------------
Linux+gcc | Automake | yes
Linux+gcc | CMake | yes
Visual Studio | CMake | no
Additionally, and as a consequence of the addition of NCZarr,
major changes have been made to the Filter API. NOTE: NCZarr
does not yet support filters, but these changes are enablers for
that support in the future. Note that it is possible
(probable?) that there will be some accidental reversions if the
changes here did not correctly mimic the existing filter testing.
In any case, previously filter ids and parameters were of type
unsigned int. In order to support the more general zarr filter
model, this was all converted to char*. The old HDF5-specific,
unsigned int operations are still supported but they are
wrappers around the new, char* based nc_filterx_XXX functions.
This entailed at least the following changes:
1. Added the files libdispatch/dfilterx.c and include/ncfilter.h
2. Some filterx utilities have been moved to libdispatch/daux.c
3. A new entry, "filter_actions" was added to the NCDispatch table
and the version bumped.
4. An overly complex set of structs was created to support funnelling
all of the filterx operations thru a single dispatch
"filter_actions" entry.
5. Move common code to from libhdf5 to libsrc4 so that it is accessible
to nczarr.
Changes directly related to Zarr:
1. Modified CMakeList.txt and configure.ac to support both C and C++
-- this is in support of S3 support via the awd-sdk libraries.
2. Define a size64_t type to support nczarr.
3. More reworking of libdispatch/dinfermodel.c to
support zarr and to regularize the structure of the fragments
section of a URL.
Changes not directly related to Zarr:
1. Make client-side filter registration be conditional, with default off.
2. Hack include/nc4internal.h to make some flags added by Ed be unique:
e.g. NC_CREAT, NC_INDEF, etc.
3. cleanup include/nchttp.h and libdispatch/dhttp.c.
4. Misc. changes to support compiling under Visual Studio including:
* Better testing under windows for dirent.h and opendir and closedir.
5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags
and to centralize error reporting.
6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them.
7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible.
Changes Left TO-DO:
1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
|
|
|
EXTERNL
|
|
|
|
int
|
|
|
|
NCmkdir(const char* path, int mode)
|
|
|
|
{
|
|
|
|
int status = 0;
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
char* cvtpath = NULL;
|
|
|
|
wchar_t* wpath = NULL;
|
|
|
|
if((cvtpath = NCpathcvt(path)) == NULL) {status=ENOMEM; goto done;}
|
|
|
|
if((status = utf82wide(cvtpath,&wpath))) {status = ENOENT; goto done;}
|
|
|
|
if(_wmkdir(wpath) < 0) {status = errno; goto done;}
|
|
|
|
done:
|
|
|
|
free(cvtpath);
|
|
|
|
free(wpath);
|
|
|
|
errno = status;
|
|
|
|
return (errno?-1:0);
|
This PR adds EXPERIMENTAL support for accessing data in the
cloud using a variant of the Zarr protocol and storage
format. This enhancement is generically referred to as "NCZarr".
The data model supported by NCZarr is netcdf-4 minus the user-defined
types and the String type. In this sense it is similar to the CDF-5
data model.
More detailed information about enabling and using NCZarr is
described in the document NUG/nczarr.md and in a
[Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in).
WARNING: this code has had limited testing, so do use this version
for production work. Also, performance improvements are ongoing.
Note especially the following platform matrix of successful tests:
Platform | Build System | S3 support
------------------------------------
Linux+gcc | Automake | yes
Linux+gcc | CMake | yes
Visual Studio | CMake | no
Additionally, and as a consequence of the addition of NCZarr,
major changes have been made to the Filter API. NOTE: NCZarr
does not yet support filters, but these changes are enablers for
that support in the future. Note that it is possible
(probable?) that there will be some accidental reversions if the
changes here did not correctly mimic the existing filter testing.
In any case, previously filter ids and parameters were of type
unsigned int. In order to support the more general zarr filter
model, this was all converted to char*. The old HDF5-specific,
unsigned int operations are still supported but they are
wrappers around the new, char* based nc_filterx_XXX functions.
This entailed at least the following changes:
1. Added the files libdispatch/dfilterx.c and include/ncfilter.h
2. Some filterx utilities have been moved to libdispatch/daux.c
3. A new entry, "filter_actions" was added to the NCDispatch table
and the version bumped.
4. An overly complex set of structs was created to support funnelling
all of the filterx operations thru a single dispatch
"filter_actions" entry.
5. Move common code to from libhdf5 to libsrc4 so that it is accessible
to nczarr.
Changes directly related to Zarr:
1. Modified CMakeList.txt and configure.ac to support both C and C++
-- this is in support of S3 support via the awd-sdk libraries.
2. Define a size64_t type to support nczarr.
3. More reworking of libdispatch/dinfermodel.c to
support zarr and to regularize the structure of the fragments
section of a URL.
Changes not directly related to Zarr:
1. Make client-side filter registration be conditional, with default off.
2. Hack include/nc4internal.h to make some flags added by Ed be unique:
e.g. NC_CREAT, NC_INDEF, etc.
3. cleanup include/nchttp.h and libdispatch/dhttp.c.
4. Misc. changes to support compiling under Visual Studio including:
* Better testing under windows for dirent.h and opendir and closedir.
5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags
and to centralize error reporting.
6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them.
7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible.
Changes Left TO-DO:
1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
|
|
|
}
|
|
|
|
|
2020-12-07 09:19:53 +08:00
|
|
|
EXTERNL
|
|
|
|
int
|
|
|
|
NCrmdir(const char* path)
|
|
|
|
{
|
|
|
|
int status = 0;
|
|
|
|
char* cvtname = NCpathcvt(path);
|
|
|
|
if(cvtname == NULL) {errno = ENOENT; return -1;}
|
|
|
|
status = rmdir(cvtname);
|
|
|
|
free(cvtname);
|
|
|
|
return status;
|
|
|
|
}
|
|
|
|
|
This PR adds EXPERIMENTAL support for accessing data in the
cloud using a variant of the Zarr protocol and storage
format. This enhancement is generically referred to as "NCZarr".
The data model supported by NCZarr is netcdf-4 minus the user-defined
types and the String type. In this sense it is similar to the CDF-5
data model.
More detailed information about enabling and using NCZarr is
described in the document NUG/nczarr.md and in a
[Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in).
WARNING: this code has had limited testing, so do use this version
for production work. Also, performance improvements are ongoing.
Note especially the following platform matrix of successful tests:
Platform | Build System | S3 support
------------------------------------
Linux+gcc | Automake | yes
Linux+gcc | CMake | yes
Visual Studio | CMake | no
Additionally, and as a consequence of the addition of NCZarr,
major changes have been made to the Filter API. NOTE: NCZarr
does not yet support filters, but these changes are enablers for
that support in the future. Note that it is possible
(probable?) that there will be some accidental reversions if the
changes here did not correctly mimic the existing filter testing.
In any case, previously filter ids and parameters were of type
unsigned int. In order to support the more general zarr filter
model, this was all converted to char*. The old HDF5-specific,
unsigned int operations are still supported but they are
wrappers around the new, char* based nc_filterx_XXX functions.
This entailed at least the following changes:
1. Added the files libdispatch/dfilterx.c and include/ncfilter.h
2. Some filterx utilities have been moved to libdispatch/daux.c
3. A new entry, "filter_actions" was added to the NCDispatch table
and the version bumped.
4. An overly complex set of structs was created to support funnelling
all of the filterx operations thru a single dispatch
"filter_actions" entry.
5. Move common code to from libhdf5 to libsrc4 so that it is accessible
to nczarr.
Changes directly related to Zarr:
1. Modified CMakeList.txt and configure.ac to support both C and C++
-- this is in support of S3 support via the awd-sdk libraries.
2. Define a size64_t type to support nczarr.
3. More reworking of libdispatch/dinfermodel.c to
support zarr and to regularize the structure of the fragments
section of a URL.
Changes not directly related to Zarr:
1. Make client-side filter registration be conditional, with default off.
2. Hack include/nc4internal.h to make some flags added by Ed be unique:
e.g. NC_CREAT, NC_INDEF, etc.
3. cleanup include/nchttp.h and libdispatch/dhttp.c.
4. Misc. changes to support compiling under Visual Studio including:
* Better testing under windows for dirent.h and opendir and closedir.
5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags
and to centralize error reporting.
6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them.
7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible.
Changes Left TO-DO:
1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
|
|
|
EXTERNL
|
|
|
|
char*
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
NCgetcwd(char* cwdbuf, size_t cwdlen)
|
|
|
|
{
|
|
|
|
int status = NC_NOERR;
|
|
|
|
struct Path wd = empty;
|
|
|
|
char* path = NULL;
|
|
|
|
size_t len;
|
|
|
|
|
|
|
|
errno = 0;
|
|
|
|
if(cwdlen == 0) {status = ENAMETOOLONG; goto done;}
|
|
|
|
if(!pathinitialized) pathinit();
|
|
|
|
if((status = getwdpath(&wd))) {status = ENOENT; goto done;}
|
|
|
|
if((status = unparsepath(&wd,&path))) {status = EINVAL; goto done;}
|
|
|
|
len = strlen(path);
|
|
|
|
if(len >= cwdlen) {status = ENAMETOOLONG; goto done;}
|
|
|
|
if(cwdbuf == NULL) {
|
|
|
|
if((cwdbuf = malloc(cwdlen))==NULL) {status = ENOMEM; goto done;}
|
|
|
|
}
|
|
|
|
memcpy(cwdbuf,path,len+1);
|
This PR adds EXPERIMENTAL support for accessing data in the
cloud using a variant of the Zarr protocol and storage
format. This enhancement is generically referred to as "NCZarr".
The data model supported by NCZarr is netcdf-4 minus the user-defined
types and the String type. In this sense it is similar to the CDF-5
data model.
More detailed information about enabling and using NCZarr is
described in the document NUG/nczarr.md and in a
[Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in).
WARNING: this code has had limited testing, so do use this version
for production work. Also, performance improvements are ongoing.
Note especially the following platform matrix of successful tests:
Platform | Build System | S3 support
------------------------------------
Linux+gcc | Automake | yes
Linux+gcc | CMake | yes
Visual Studio | CMake | no
Additionally, and as a consequence of the addition of NCZarr,
major changes have been made to the Filter API. NOTE: NCZarr
does not yet support filters, but these changes are enablers for
that support in the future. Note that it is possible
(probable?) that there will be some accidental reversions if the
changes here did not correctly mimic the existing filter testing.
In any case, previously filter ids and parameters were of type
unsigned int. In order to support the more general zarr filter
model, this was all converted to char*. The old HDF5-specific,
unsigned int operations are still supported but they are
wrappers around the new, char* based nc_filterx_XXX functions.
This entailed at least the following changes:
1. Added the files libdispatch/dfilterx.c and include/ncfilter.h
2. Some filterx utilities have been moved to libdispatch/daux.c
3. A new entry, "filter_actions" was added to the NCDispatch table
and the version bumped.
4. An overly complex set of structs was created to support funnelling
all of the filterx operations thru a single dispatch
"filter_actions" entry.
5. Move common code to from libhdf5 to libsrc4 so that it is accessible
to nczarr.
Changes directly related to Zarr:
1. Modified CMakeList.txt and configure.ac to support both C and C++
-- this is in support of S3 support via the awd-sdk libraries.
2. Define a size64_t type to support nczarr.
3. More reworking of libdispatch/dinfermodel.c to
support zarr and to regularize the structure of the fragments
section of a URL.
Changes not directly related to Zarr:
1. Make client-side filter registration be conditional, with default off.
2. Hack include/nc4internal.h to make some flags added by Ed be unique:
e.g. NC_CREAT, NC_INDEF, etc.
3. cleanup include/nchttp.h and libdispatch/dhttp.c.
4. Misc. changes to support compiling under Visual Studio including:
* Better testing under windows for dirent.h and opendir and closedir.
5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags
and to centralize error reporting.
6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them.
7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible.
Changes Left TO-DO:
1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
|
|
|
done:
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
clearPath(&wd);
|
|
|
|
nullfree(path);
|
|
|
|
errno = status;
|
This PR adds EXPERIMENTAL support for accessing data in the
cloud using a variant of the Zarr protocol and storage
format. This enhancement is generically referred to as "NCZarr".
The data model supported by NCZarr is netcdf-4 minus the user-defined
types and the String type. In this sense it is similar to the CDF-5
data model.
More detailed information about enabling and using NCZarr is
described in the document NUG/nczarr.md and in a
[Unidata Developer's blog entry](https://www.unidata.ucar.edu/blogs/developer/en/entry/overview-of-zarr-support-in).
WARNING: this code has had limited testing, so do use this version
for production work. Also, performance improvements are ongoing.
Note especially the following platform matrix of successful tests:
Platform | Build System | S3 support
------------------------------------
Linux+gcc | Automake | yes
Linux+gcc | CMake | yes
Visual Studio | CMake | no
Additionally, and as a consequence of the addition of NCZarr,
major changes have been made to the Filter API. NOTE: NCZarr
does not yet support filters, but these changes are enablers for
that support in the future. Note that it is possible
(probable?) that there will be some accidental reversions if the
changes here did not correctly mimic the existing filter testing.
In any case, previously filter ids and parameters were of type
unsigned int. In order to support the more general zarr filter
model, this was all converted to char*. The old HDF5-specific,
unsigned int operations are still supported but they are
wrappers around the new, char* based nc_filterx_XXX functions.
This entailed at least the following changes:
1. Added the files libdispatch/dfilterx.c and include/ncfilter.h
2. Some filterx utilities have been moved to libdispatch/daux.c
3. A new entry, "filter_actions" was added to the NCDispatch table
and the version bumped.
4. An overly complex set of structs was created to support funnelling
all of the filterx operations thru a single dispatch
"filter_actions" entry.
5. Move common code to from libhdf5 to libsrc4 so that it is accessible
to nczarr.
Changes directly related to Zarr:
1. Modified CMakeList.txt and configure.ac to support both C and C++
-- this is in support of S3 support via the awd-sdk libraries.
2. Define a size64_t type to support nczarr.
3. More reworking of libdispatch/dinfermodel.c to
support zarr and to regularize the structure of the fragments
section of a URL.
Changes not directly related to Zarr:
1. Make client-side filter registration be conditional, with default off.
2. Hack include/nc4internal.h to make some flags added by Ed be unique:
e.g. NC_CREAT, NC_INDEF, etc.
3. cleanup include/nchttp.h and libdispatch/dhttp.c.
4. Misc. changes to support compiling under Visual Studio including:
* Better testing under windows for dirent.h and opendir and closedir.
5. Misc. changes to the oc2 code to support various libcurl CURLOPT flags
and to centralize error reporting.
6. By default, suppress the vlen tests that have unfixed memory leaks; add option to enable them.
7. Make part of the nc_test/test_byterange.sh test be contingent on remotetest.unidata.ucar.edu being accessible.
Changes Left TO-DO:
1. fix provenance code, it is too HDF5 specific.
2020-06-29 08:02:47 +08:00
|
|
|
return cwdbuf;
|
|
|
|
}
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
|
2021-05-15 01:33:03 +08:00
|
|
|
EXTERNL
|
|
|
|
int
|
|
|
|
NCmkstemp(char* base)
|
|
|
|
{
|
|
|
|
int stat = 0;
|
|
|
|
int fd, rno;
|
|
|
|
char* tmp = NULL;
|
|
|
|
size_t len;
|
|
|
|
char* xp = NULL;
|
|
|
|
char* cvtpath = NULL;
|
|
|
|
int attempts;
|
|
|
|
|
|
|
|
cvtpath = NCpathcvt(base);
|
|
|
|
len = strlen(cvtpath);
|
|
|
|
xp = cvtpath+(len-6);
|
|
|
|
assert(memcmp(xp,"XXXXXX")==0);
|
|
|
|
for(attempts=10;attempts>0;attempts--) {
|
|
|
|
/* The Windows version of mkstemp does not work right;
|
|
|
|
it only allows for 26 possible XXXXXX values */
|
|
|
|
/* Need to simulate by using some kind of pseudo-random number */
|
|
|
|
rno = rand();
|
|
|
|
if(rno < 0) rno = -rno;
|
|
|
|
snprintf(xp,7,"%06d",rno);
|
|
|
|
fd=NCopen3(cvtpath,O_RDWR|O_BINARY|O_CREAT, _S_IREAD|_S_IWRITE);
|
|
|
|
if(fd >= 0) break;
|
|
|
|
}
|
|
|
|
if(fd < 0) {
|
|
|
|
nclog(NCLOGERR, "Could not create temp file: %s",tmp);
|
|
|
|
stat = EACCES;
|
|
|
|
goto done;
|
|
|
|
}
|
|
|
|
done:
|
|
|
|
nullfree(cvtpath);
|
|
|
|
if(stat && fd >= 0) {close(fd);}
|
|
|
|
return (stat?-1:fd);
|
|
|
|
}
|
|
|
|
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
#ifdef HAVE_SYS_STAT_H
|
|
|
|
EXTERNL
|
|
|
|
int
|
|
|
|
NCstat(char* path, struct stat* buf)
|
|
|
|
{
|
|
|
|
int status = 0;
|
|
|
|
char* cvtpath = NULL;
|
|
|
|
wchar_t* wpath = NULL;
|
|
|
|
if((cvtpath = NCpathcvt(path)) == NULL) {status=ENOMEM; goto done;}
|
|
|
|
if((status = utf82wide(cvtpath,&wpath))) {status = ENOENT; goto done;}
|
|
|
|
if(_wstat(wpath,buf) < 0) {status = errno; goto done;}
|
|
|
|
done:
|
|
|
|
free(cvtpath);
|
|
|
|
free(wpath);
|
|
|
|
errno = status;
|
|
|
|
return (errno?-1:0);
|
|
|
|
}
|
|
|
|
#endif /*HAVE_SYS_STAT_H*/
|
|
|
|
|
|
|
|
int
|
|
|
|
NCpath2utf8(const char* s, char** u8p)
|
|
|
|
{
|
|
|
|
return ansi2utf8(s,u8p);
|
|
|
|
}
|
|
|
|
|
|
|
|
#else /*!WINPATH*/
|
|
|
|
|
|
|
|
int
|
|
|
|
NCpath2utf8(const char* path, char** u8p)
|
|
|
|
{
|
|
|
|
int stat = NC_NOERR;
|
|
|
|
char* u8 = NULL;
|
|
|
|
if(path != NULL) {
|
|
|
|
u8 = strdup(path);
|
|
|
|
if(u8 == NULL) {stat = NC_ENOMEM; goto done;}
|
|
|
|
}
|
|
|
|
if(u8p) {*u8p = u8; u8 = NULL;}
|
|
|
|
done:
|
|
|
|
return stat;
|
|
|
|
}
|
|
|
|
#endif /*!WINPATH*/
|
|
|
|
|
|
|
|
EXTERNL int
|
|
|
|
NChasdriveletter(const char* path)
|
|
|
|
{
|
|
|
|
int stat = NC_NOERR;
|
|
|
|
int hasdl = 0;
|
|
|
|
struct Path canon = empty;
|
|
|
|
|
|
|
|
if(!pathinitialized) pathinit();
|
|
|
|
|
|
|
|
if((stat = parsepath(path,&canon))) goto done;
|
|
|
|
if(canon.kind == NCPD_REL) {
|
|
|
|
clearPath(&canon);
|
|
|
|
/* Get the drive letter (if any) from the local wd */
|
|
|
|
canon.drive = wdpath.drive;
|
|
|
|
}
|
|
|
|
hasdl = (canon.drive != 0);
|
|
|
|
done:
|
|
|
|
clearPath(&canon);
|
|
|
|
return hasdl;
|
|
|
|
}
|
|
|
|
|
|
|
|
/**************************************************/
|
|
|
|
/* Utilities */
|
|
|
|
|
|
|
|
/* Parse a path */
|
|
|
|
static int
|
|
|
|
parsepath(const char* inpath, struct Path* path)
|
|
|
|
{
|
|
|
|
int stat = NC_NOERR;
|
|
|
|
char* tmp1 = NULL;
|
|
|
|
size_t len;
|
|
|
|
char* p;
|
|
|
|
|
|
|
|
assert(path);
|
|
|
|
memset(path,0,sizeof(struct Path));
|
|
|
|
|
|
|
|
if(inpath == NULL) goto done; /* defensive driving */
|
|
|
|
|
|
|
|
/* Convert to UTF8 */
|
|
|
|
#if 0
|
|
|
|
if((stat = NCpath2utf8(inpath,&tmp1))) goto done;
|
|
|
|
#else
|
|
|
|
tmp1 = strdup(inpath);
|
|
|
|
#endif
|
|
|
|
/* Convert to forward slash */
|
|
|
|
for(p=tmp1;*p;p++) {if(*p == '\\') *p = '/';}
|
|
|
|
|
|
|
|
/* parse all paths to 2-parts:
|
|
|
|
1. drive letter (optional)
|
|
|
|
2. path after drive letter
|
|
|
|
*/
|
|
|
|
|
|
|
|
len = strlen(tmp1);
|
|
|
|
|
|
|
|
/* 1. look for MSYS path /D/... */
|
|
|
|
if(len >= 2
|
|
|
|
&& (tmp1[0] == '/')
|
|
|
|
&& strchr(windrive,tmp1[1]) != NULL
|
|
|
|
&& (tmp1[2] == '/' || tmp1[2] == '\0')) {
|
|
|
|
/* Assume this is a mingw path */
|
|
|
|
path->drive = tmp1[1];
|
|
|
|
/* Remainder */
|
|
|
|
if(tmp1[2] == '\0')
|
|
|
|
path->path = NULL;
|
|
|
|
else
|
|
|
|
path->path = strdup(tmp1+2);
|
|
|
|
if(path == NULL)
|
|
|
|
{stat = NC_ENOMEM; goto done;}
|
|
|
|
path->kind = NCPD_MSYS;
|
|
|
|
}
|
|
|
|
/* 2. Look for leading /cygdrive/D where D is a single-char drive letter */
|
|
|
|
else if(len >= (cdlen+1)
|
|
|
|
&& memcmp(tmp1,"/cygdrive/",cdlen)==0
|
|
|
|
&& strchr(windrive,tmp1[cdlen]) != NULL
|
|
|
|
&& (tmp1[cdlen+1] == '/'
|
|
|
|
|| tmp1[cdlen+1] == '\0')) {
|
|
|
|
/* Assume this is a cygwin path */
|
|
|
|
path->drive = tmp1[cdlen];
|
|
|
|
/* Remainder */
|
|
|
|
if(tmp1[cdlen+1] == '\0')
|
|
|
|
path->path = NULL;
|
|
|
|
else
|
|
|
|
path->path = strdup(tmp1+cdlen+1);
|
|
|
|
if(path == NULL)
|
|
|
|
{stat = NC_ENOMEM; goto done;}
|
|
|
|
path->kind = NCPD_CYGWIN;
|
|
|
|
}
|
|
|
|
/* 3. Look for windows path: D:/... where D is a single-char
|
|
|
|
drive letter */
|
|
|
|
else if(len >= 2
|
|
|
|
&& strchr(windrive,tmp1[0]) != NULL
|
|
|
|
&& tmp1[1] == ':'
|
|
|
|
&& (tmp1[2] == '\0' || tmp1[2] == '/')) {
|
|
|
|
/* Assume this is a windows path */
|
|
|
|
path->drive = tmp1[0];
|
|
|
|
/* Remainder */
|
|
|
|
if(tmp1[2] == '\0')
|
|
|
|
path->path = NULL;
|
|
|
|
else
|
|
|
|
path->path = strdup(tmp1+2);
|
|
|
|
if(path == NULL)
|
|
|
|
{stat = NC_ENOMEM; goto done;}
|
|
|
|
path->kind = NCPD_WIN;
|
|
|
|
}
|
|
|
|
/* look for *nix path */
|
|
|
|
else if(len >= 1 && tmp1[0] == '/') {
|
|
|
|
/* Assume this is a *nix path */
|
|
|
|
path->drive = 0; /* no drive letter */
|
|
|
|
/* Remainder */
|
|
|
|
path->path = tmp1; tmp1 = NULL;
|
|
|
|
path->kind = NCPD_NIX;
|
|
|
|
} else {/* Relative path of unknown type */
|
|
|
|
path->kind = NCPD_REL;
|
|
|
|
path->path = tmp1; tmp1 = NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
done:
|
|
|
|
nullfree(tmp1);
|
|
|
|
if(stat) {clearPath(path);}
|
|
|
|
return stat;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
unparsepath(struct Path* xp, char** pathp)
|
|
|
|
{
|
|
|
|
int stat = NC_NOERR;
|
|
|
|
size_t len;
|
|
|
|
char* path = NULL;
|
|
|
|
char sdrive[2] = {'\0','\0'};
|
|
|
|
char* p = NULL;
|
|
|
|
int kind = xp->kind;
|
|
|
|
int cygspecial = 0;
|
|
|
|
|
|
|
|
switch (kind) {
|
|
|
|
case NCPD_NIX:
|
|
|
|
len = nulllen(xp->path);
|
|
|
|
if(xp->drive != 0) {
|
|
|
|
len += 2;
|
|
|
|
sdrive[0] = xp->drive;
|
|
|
|
}
|
|
|
|
len++; /* nul terminate */
|
|
|
|
if((path = (char*)malloc(len))==NULL)
|
|
|
|
{stat = NC_ENOMEM; goto done;}
|
|
|
|
path[0] = '\0';
|
|
|
|
if(xp->drive != 0) {
|
|
|
|
strlcat(path,"/",len);
|
|
|
|
strlcat(path,sdrive,len);
|
|
|
|
}
|
|
|
|
if(xp->path != NULL)
|
|
|
|
strlcat(path,xp->path,len);
|
|
|
|
break;
|
|
|
|
case NCPD_CYGWIN:
|
|
|
|
/* Is this one of the special cygwin paths? */
|
|
|
|
cygspecial = iscygwinspecial(xp->path);
|
|
|
|
if(xp->drive == 0) {xp->drive = wdpath.drive;} /*may require a drive */
|
|
|
|
len = nulllen(xp->path)+cdlen+1+1;
|
|
|
|
if((path = (char*)malloc(len))==NULL)
|
|
|
|
{stat = NC_ENOMEM; goto done;}
|
|
|
|
path[0] = '\0';
|
|
|
|
if(!cygspecial) {
|
|
|
|
strlcat(path,"/cygdrive/",len);
|
|
|
|
sdrive[0] = xp->drive;
|
|
|
|
strlcat(path,sdrive,len);
|
|
|
|
}
|
|
|
|
if(xp->path)
|
|
|
|
strlcat(path,xp->path,len);
|
|
|
|
break;
|
|
|
|
case NCPD_WIN:
|
|
|
|
if(xp->drive == 0) {xp->drive = wdpath.drive;} /*requires a drive */
|
|
|
|
len = nulllen(xp->path)+2+1;
|
|
|
|
if((path = (char*)malloc(len))==NULL)
|
|
|
|
{stat = NC_ENOMEM; goto done;}
|
|
|
|
path[0] = '\0';
|
|
|
|
sdrive[0] = xp->drive;
|
|
|
|
strlcat(path,sdrive,len);
|
|
|
|
strlcat(path,":",len);
|
|
|
|
if(xp->path)
|
|
|
|
strlcat(path,xp->path,len);
|
|
|
|
/* Convert forward to back */
|
|
|
|
for(p=path;*p;p++) {if(*p == '/') *p = '\\';}
|
|
|
|
break;
|
|
|
|
case NCPD_MSYS:
|
|
|
|
if(xp->drive == 0) {xp->drive = wdpath.drive;} /*requires a drive */
|
|
|
|
len = nulllen(xp->path)+2+1;
|
|
|
|
if((path = (char*)malloc(len))==NULL)
|
|
|
|
{stat = NC_ENOMEM; goto done;}
|
|
|
|
path[0] = '\0';
|
|
|
|
sdrive[0] = xp->drive;
|
|
|
|
strlcat(path,"/",len);
|
|
|
|
strlcat(path,sdrive,len);
|
|
|
|
if(xp->path)
|
|
|
|
strlcat(path,xp->path,len);
|
|
|
|
break;
|
|
|
|
case NCPD_REL:
|
|
|
|
path = strdup(xp->path);
|
|
|
|
/* Use wdpath to decide slashing */
|
|
|
|
if(wdpath.kind == NCPD_WIN) {
|
|
|
|
/* Convert forward to back */
|
|
|
|
for(p=path;*p;p++) {if(*p == '/') *p = '\\';}
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
default: stat = NC_EINTERNAL; goto done;
|
|
|
|
}
|
|
|
|
if(pathp) {*pathp = path; path = NULL;}
|
|
|
|
done:
|
|
|
|
nullfree(path);
|
|
|
|
return stat;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
getwdpath(struct Path* wd)
|
|
|
|
{
|
|
|
|
int stat = NC_NOERR;
|
|
|
|
char* path = NULL;
|
|
|
|
if(wd->path != NULL) return stat;
|
|
|
|
memset(wd,0,sizeof(struct Path));
|
|
|
|
{
|
|
|
|
#ifdef _WIN32
|
|
|
|
wchar_t* wpath = NULL;
|
|
|
|
wpath = _wgetcwd(NULL,8192);
|
|
|
|
if((stat = wide2utf8(wpath,&path)))
|
|
|
|
{nullfree(wpath); wpath = NULL; return stat;}
|
|
|
|
#else
|
|
|
|
path = getcwd(NULL,8192);
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
stat = parsepath(path,wd);
|
|
|
|
/* Force the kind */
|
|
|
|
wd->kind = getlocalpathkind();
|
|
|
|
nullfree(path); path = NULL;
|
|
|
|
return stat;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
getlocalpathkind(void)
|
|
|
|
{
|
|
|
|
int kind = NCPD_UNKNOWN;
|
|
|
|
#ifdef __CYGWIN__
|
|
|
|
kind = NCPD_CYGWIN;
|
|
|
|
#elif __MSYS__
|
|
|
|
kind = NCPD_MSYS;
|
|
|
|
#elif _MSC_VER /* not _WIN32 */
|
|
|
|
kind = NCPD_WIN;
|
|
|
|
#else
|
|
|
|
kind = NCPD_NIX;
|
|
|
|
#endif
|
|
|
|
return kind;
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef WINPATH
|
|
|
|
/**
|
|
|
|
* Converts the filename from Locale character set (presumably some
|
|
|
|
* ANSI character set like ISO-Latin-1 or UTF-8 to UTF-8
|
|
|
|
* @param local Pointer to a nul-terminated string in locale char set.
|
|
|
|
* @param u8p Pointer for returning the output utf8 string
|
|
|
|
*
|
|
|
|
* @return NC_NOERR return converted filename
|
|
|
|
* @return NC_EINVAL if conversion fails
|
|
|
|
* @return NC_ENOMEM if no memory available
|
|
|
|
*
|
|
|
|
*/
|
|
|
|
static int
|
|
|
|
ansi2utf8(const char* local, char** u8p)
|
|
|
|
{
|
|
|
|
int stat=NC_NOERR;
|
|
|
|
char* u8 = NULL;
|
|
|
|
int n;
|
|
|
|
wchar_t* u16 = NULL;
|
|
|
|
|
|
|
|
{
|
|
|
|
/* Get length of the converted string */
|
|
|
|
n = MultiByteToWideChar(CP_ACP, 0, local, -1, NULL, 0);
|
|
|
|
if (!n) {stat = NC_EINVAL; goto done;}
|
|
|
|
if((u16 = malloc(sizeof(wchar_t) * n))==NULL)
|
|
|
|
{stat = NC_ENOMEM; goto done;}
|
|
|
|
/* do the conversion */
|
|
|
|
if (!MultiByteToWideChar(CP_ACP, 0, local, -1, u16, n))
|
|
|
|
{stat = NC_EINVAL; goto done;}
|
|
|
|
/* Now reverse the process to produce utf8 */
|
|
|
|
n = WideCharToMultiByte(CP_UTF8, 0, u16, -1, NULL, 0, NULL, NULL);
|
|
|
|
if (!n) {stat = NC_EINVAL; goto done;}
|
|
|
|
if((u8 = malloc(sizeof(char) * n))==NULL)
|
|
|
|
{stat = NC_ENOMEM; goto done;}
|
|
|
|
if (!WideCharToMultiByte(CP_UTF8, 0, u16, -1, u8, n, NULL, NULL))
|
|
|
|
{stat = NC_EINVAL; goto done;}
|
|
|
|
}
|
|
|
|
if(u8p) {*u8p = u8; u8 = NULL;}
|
|
|
|
done:
|
|
|
|
nullfree(u8);
|
|
|
|
return stat;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
ansi2wide(const char* local, wchar_t** u16p)
|
|
|
|
{
|
|
|
|
int stat=NC_NOERR;
|
|
|
|
wchar_t* u16 = NULL;
|
|
|
|
int n;
|
|
|
|
|
|
|
|
/* Get length of the converted string */
|
|
|
|
n = MultiByteToWideChar(CP_ACP, 0, local, -1, NULL, 0);
|
|
|
|
if (!n) {stat = NC_EINVAL; goto done;}
|
|
|
|
if((u16 = malloc(sizeof(wchar_t) * n))==NULL)
|
|
|
|
{stat = NC_ENOMEM; goto done;}
|
|
|
|
/* do the conversion */
|
|
|
|
if (!MultiByteToWideChar(CP_ACP, 0, local, -1, u16, n))
|
|
|
|
{stat = NC_EINVAL; goto done;}
|
|
|
|
if(u16p) {*u16p = u16; u16 = NULL;}
|
|
|
|
done:
|
|
|
|
nullfree(u16);
|
|
|
|
return stat;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
utf82wide(const char* utf8, wchar_t** u16p)
|
|
|
|
{
|
|
|
|
int stat=NC_NOERR;
|
|
|
|
wchar_t* u16 = NULL;
|
|
|
|
int n;
|
|
|
|
|
|
|
|
/* Get length of the converted string */
|
|
|
|
n = MultiByteToWideChar(CP_UTF8, 0, utf8, -1, NULL, 0);
|
|
|
|
if (!n) {stat = NC_EINVAL; goto done;}
|
|
|
|
if((u16 = malloc(sizeof(wchar_t) * n))==NULL)
|
|
|
|
{stat = NC_ENOMEM; goto done;}
|
|
|
|
/* do the conversion */
|
|
|
|
if (!MultiByteToWideChar(CP_UTF8, 0, utf8, -1, u16, n))
|
|
|
|
{stat = NC_EINVAL; goto done;}
|
|
|
|
if(u16p) {*u16p = u16; u16 = NULL;}
|
|
|
|
done:
|
|
|
|
nullfree(u16);
|
|
|
|
return stat;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
wide2utf8(const wchar_t* u16, char** u8p)
|
|
|
|
{
|
|
|
|
int stat=NC_NOERR;
|
|
|
|
char* u8 = NULL;
|
|
|
|
int n;
|
|
|
|
|
|
|
|
/* Get length of the converted string */
|
|
|
|
n = WideCharToMultiByte(CP_UTF8, 0, u16, -1, NULL, 0, NULL, NULL);
|
|
|
|
if (!n) {stat = NC_EINVAL; goto done;}
|
|
|
|
if((u8 = malloc(sizeof(char) * n))==NULL)
|
|
|
|
{stat = NC_ENOMEM; goto done;}
|
|
|
|
/* do the conversion */
|
|
|
|
if (!WideCharToMultiByte(CP_UTF8, 0, u16, -1, u8, n, NULL, NULL))
|
|
|
|
{stat = NC_EINVAL; goto done;}
|
|
|
|
if(u8p) {*u8p = u8; u8 = NULL;}
|
|
|
|
done:
|
|
|
|
nullfree(u8);
|
|
|
|
return stat;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Set locale */
|
|
|
|
void
|
|
|
|
nc_setlocale_utf8(void)
|
|
|
|
{
|
|
|
|
#ifdef __hpux
|
|
|
|
setlocale(LC_CTYPE,"");
|
|
|
|
#else
|
|
|
|
setlocale(LC_ALL,"C.UTF8");
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2017-04-15 01:05:30 +08:00
|
|
|
#endif /*WINPATH*/
|
Codify cross-platform file paths
The netcdf-c code has to deal with a variety of platforms:
Windows, OSX, Linux, Cygwin, MSYS, etc. These platforms differ
significantly in the kind of file paths that they accept. So in
order to handle this, I have created a set of replacements for
the most common file system operations such as _open_ or _fopen_
or _access_ to manage the file path differences correctly.
A more limited version of this idea was already implemented via
the ncwinpath.h and dwinpath.c code. So this can be viewed as a
replacement for that code. And in path in many cases, the only
change that was required was to replace '#include <ncwinpath.h>'
with '#include <ncpathmgt.h>' and then replace file operation
calls with the NCxxx equivalent from ncpathmgr.h Note that
recently, the ncwinpath.h was renamed ncpathmgmt.h, so this pull
request should not require dealing with winpath.
The heart of the change is include/ncpathmgmt.h, which provides
alternate operations such as NCfopen or NCaccess and which properly
parse and rebuild path arguments to work for the platform on which
the code is executing. This mostly matters for Windows because of the
way that it uses backslash and drive letters, as compared to *nix*.
One important feature is that the user can do string manipulations
on a file path without having to worry too much about the platform
because the path management code will properly handle most mixed cases.
So one can for example concatenate a path suffix that uses forward
slashes to a Windows path and have it work correctly.
The conversion code is in libdispatch/dpathmgr.c, and the
important function there is NCpathcvt which does the proper
conversions to the local path format.
As a rule, most code should just replace their file operations with
the corresponding NCxxx ones defined in include/ncpathmgmt.h. These
NCxxx functions all call NCpathcvt on their path arguments before
executing the actual file operation.
In some rare cases, the client may need to directly use NCpathcvt,
but this should be avoided as much as possible. If there is a need
for supporting a new file operation not already in ncpathmgmt.h, then
use the code in dpathmgr.c as a template. Also please notify Unidata
so we can include it as a formal part or our supported operations.
Also, if you see an operation in the library that is not using the
NCxxx form, then please submit an issue so we can fix it.
Misc. Changes:
* Clean up the utf8 testing code; it is impossible to get some
tests to work under windows using shell scripts; the args do
not pass as utf8 but as some other encoding.
* Added an extra utf8 test case: test_unicode_path.sh
* Add a true test for HDF5 1.10.6 or later because as noted in
PR https://github.com/Unidata/netcdf-c/pull/1794,
HDF5 changed its Windows file path handling.
2021-03-05 04:41:31 +08:00
|
|
|
|
|
|
|
static char*
|
|
|
|
printPATH(struct Path* p)
|
|
|
|
{
|
|
|
|
static char buf[4096];
|
|
|
|
buf[0] = '\0';
|
|
|
|
snprintf(buf,sizeof(buf),"Path{kind=%d drive='%c' path=|%s|}",
|
|
|
|
p->kind,(p->drive > 0?p->drive:'0'),p->path);
|
|
|
|
return buf;
|
|
|
|
}
|
|
|
|
|
|
|
|
#if 0
|
|
|
|
static int
|
|
|
|
hexfor(int c)
|
|
|
|
{
|
|
|
|
if(c >= '0' && c <= '9') return c - '0';
|
|
|
|
if(c >= 'a' && c <= 'f') return (c - 'a')+10;
|
|
|
|
if(c >= 'A' && c <= 'F') return (c - 'A')+10;
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
static char hexdigit[] = "0123456789abcdef";
|
|
|
|
|
|
|
|
EXTERNL void
|
|
|
|
printutf8hex(const char* s, char* sx)
|
|
|
|
{
|
|
|
|
const char* p;
|
|
|
|
char* q;
|
|
|
|
for(q=sx,p=s;*p;p++) {
|
|
|
|
unsigned int c = (unsigned char)*p;
|
|
|
|
if(c >= ' ' && c <= 127)
|
|
|
|
*q++ = (char)c;
|
|
|
|
else {
|
|
|
|
*q++ = '\\';
|
|
|
|
*q++ = 'x';
|
|
|
|
*q++ = hexdigit[(c>>4)&0xf];
|
|
|
|
*q++ = hexdigit[(c)&0xf];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
*q = '\0';
|
|
|
|
}
|