netcdf-c/libnczarr/zmap_nzf.c

1273 lines
32 KiB
C
Raw Normal View History

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
/*
* Copyright 2018, University Corporation for Atmospheric Research
* See netcdf/COPYRIGHT file for copying and redistribution conditions.
*/
#undef DEBUG
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
/* Not sure this has any effect */
#define _LARGEFILE_SOURCE 1
#define _LARGEFILE64_SOURCE 1
#include "zincludes.h"
#include <errno.h>
#ifdef HAVE_SYS_TYPES_H
#include <sys/types.h>
#endif
#ifdef HAVE_UNISTD_H
#include <unistd.h>
#endif
#ifdef HAVE_FCNTL_H
#include <fcntl.h>
#endif
#ifdef HAVE_SYS_STAT_H
#include <sys/stat.h>
#endif
#ifdef HAVE_DIRENT_H
#include <dirent.h>
#endif
#ifdef _WIN32
#include <windows.h>
#ifndef S_ISDIR
#define S_ISDIR(mode) ((mode) & _S_IFDIR)
#define S_ISREG(mode) ((mode) & _S_IFREG)
#endif
#if 0
#ifndef __cplusplus
#include <io.h>
#include <iostream>
#endif
#endif
#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
#include "fbits.h"
#include "ncpathmgr.h"
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
#define VERIFY
#ifndef O_DIRECTORY
# define O_DIRECTORY 0200000
#endif
/*Mnemonic*/
#define FLAG_ISDIR 1
#define FLAG_CREATE 1
#define SKIPLAST 1
#define WHOLEPATH 0
#define NCZM_FILE_V1 1
#ifdef S_IRUSR
static int NC_DEFAULT_CREATE_PERMS =
(S_IRUSR|S_IWUSR |S_IRGRP|S_IWGRP);
static int NC_DEFAULT_RWOPEN_PERMS =
(S_IRUSR|S_IWUSR |S_IRGRP|S_IWGRP);
static int NC_DEFAULT_ROPEN_PERMS =
// (S_IRUSR |S_IRGRP);
(S_IRUSR|S_IWUSR |S_IRGRP|S_IWGRP);
static int NC_DEFAULT_DIR_PERMS =
(S_IRUSR|S_IWUSR|S_IXUSR|S_IRGRP|S_IXGRP|S_IWGRP);
#else
static int NC_DEFAULT_CREATE_PERMS = 0660;
static int NC_DEFAULT_RWOPEN_PERMS = 0660;
static int NC_DEFAULT_ROPEN_PERMS = 0660;
static int NC_DEFAULT_DIR_PERMS = 0770;
#endif
/*
Do a simple mapping of our simplified map model
to a file system.
Every dataset is assumed to be rooted at some directory in the
file tree. So, its location is defined by some path to a
directory representing both the dataset and the root group of
that dataset. The root is recognized because it uniquely
contains a "superblock" file name ".nczarr" that provides
general information about a dataset. Nesting a dataset
inside a dataset is prohibited. This can be detected
by looking for an occurrence of a ".nczarr" file in any containing
directory. If such a file is found, then an illegal nested
dataset has been found.
For the object API, the mapping is as follows:
1. Every content-bearing object (e.g. .zgroup or .zarray) is mapped to a file.
The key constraint is that the content bearing objects are files.
This means that if a key points to a content bearing object then
no other key can have that content bearing key as a suffix.
2. The meta data containing files are assumed to contain
UTF-8 character data.
3. The chunk containing files are assumed to contain raw unsigned 8-bit byte data.
*/
/* define the var name containing an objects content */
#define ZCONTENT "data"
typedef struct FD {
int fd;
} FD;
static FD FDNUL = {-1};
/* Define the "subclass" of NCZMAP */
typedef struct ZFMAP {
NCZMAP map;
char* root;
} ZFMAP;
/* Forward */
static NCZMAP_API zapi;
static int zfileclose(NCZMAP* map, int delete);
static int zfcreategroup(ZFMAP*, const char* key, int nskip);
static int zflookupobj(ZFMAP*, const char* key, FD* fd);
static int zfcreateobj(ZFMAP*, const char* key,FD*);
static int zfparseurl(const char* path0, NCURI** urip);
static int zffullpath(ZFMAP* zfmap, const char* key, char**);
static void zfrelease(ZFMAP* zfmap, FD* fd);
static int platformerr(int err);
static int platformcreatefile(ZFMAP* map, const char* truepath,FD*);
static int platformcreatedir(ZFMAP* map, const char* truepath);
static int platformopenfile(ZFMAP* zfmap, const char* truepath, FD* fd);
static int platformopendir(ZFMAP* map, const char* truepath);
static int platformdircontent(ZFMAP* map, const char* path, NClist* contents);
static int platformdelete(ZFMAP* map, const char* path, int delroot);
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
static int platformseek(ZFMAP* map, FD* fd, int pos, size64_t* offset);
static int platformread(ZFMAP* map, FD* fd, size64_t count, void* content);
static int platformwrite(ZFMAP* map, FD* fd, size64_t count, const void* content);
static void platformrelease(ZFMAP* zfmap, FD* fd);
static int platformtestcontentbearing(ZFMAP* zfmap, const char* truepath);
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 VERIFY
static int verify(const char* path, int isdir);
#endif
static int zfinitialized = 0;
static void zfinitialize(void)
{
if(!zfinitialized) {
ZTRACE("","");
const char* env = NULL;
int perms = 0;
env = getenv("NC_DEFAULT_CREATE_PERMS");
if(env != NULL && strlen(env) > 0) {
if(sscanf(env,"%d",&perms) == 1) NC_DEFAULT_CREATE_PERMS = perms;
}
env = getenv("NC_DEFAULT_DIR_PERMS");
if(env != NULL && strlen(env) > 0) {
if(sscanf(env,"%d",&perms) == 1) NC_DEFAULT_DIR_PERMS = perms;
}
zfinitialized = 1;
}
}
/* Define the Dataset level API */
/*
@param datasetpath abs path in the file tree of the root of the dataset'
might be a relative path.
@param mode the netcdf-c mode flags
@param flags extra flags
@param flags extra parameters
@param mapp return the map object in this
*/
static int
zfilecreate(const char *path, int mode, size64_t flags, void* parameters, NCZMAP** mapp)
{
int stat = NC_NOERR;
char* truepath = 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
ZFMAP* zfmap = NULL;
NCURI* url = NULL;
NC_UNUSED(parameters);
ZTRACE("%s %d %llu %p",path,mode,flags,parameters);
if(!zfinitialized) zfinitialize();
/* Fixup mode flags */
mode = (NC_NETCDF4 | NC_WRITE | mode);
if(flags & FLAG_BYTERANGE)
mode &= ~(NC_CLOBBER | NC_WRITE);
if(!(mode & NC_WRITE))
{stat = NC_EPERM; goto done;}
/* path must be a url with file: protocol*/
if((stat=zfparseurl(path,&url)))
goto done;
if(strcasecmp(url->protocol,"file") != 0)
{stat = NC_EURL; goto done;}
/* Canonicalize the root path */
if((stat = nczm_canonicalpath(url->path,&truepath))) goto done;
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 CHECKNESTEDDATASETS
if(isnesteddataset(truepath))
{stat = NC_ENCZARR; goto done;}
#endif
/* Build the z4 state */
if((zfmap = calloc(1,sizeof(ZFMAP))) == NULL)
{stat = NC_ENOMEM; goto done;}
zfmap->map.format = NCZM_FILE;
zfmap->map.url = ncuribuild(url,NULL,NULL,NCURIALL);
zfmap->map.flags = flags;
/* create => NC_WRITE */
zfmap->map.mode = mode;
zfmap->map.api = &zapi;
zfmap->root = truepath;
truepath = NULL;
/* If NC_CLOBBER, then delete below file tree */
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
if(!fIsSet(mode,NC_NOCLOBBER))
platformdelete(zfmap,zfmap->root,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
/* make sure we can access the root directory; create if necessary */
if((stat = platformcreatedir(zfmap, zfmap->root)))
goto done;
/* Dataset superblock will be written by higher layer */
if(mapp) *mapp = (NCZMAP*)zfmap;
done:
ncurifree(url);
nullfree(truepath);
if(stat)
zfileclose((NCZMAP*)zfmap,1);
return (stat);
}
/*
@param datasetpath abs path in the file tree of the root of the dataset'
might be a relative path.
@param mode the netcdf-c mode flags
@param flags extra flags
@param flags extra parameters
@param mapp return the map object in this
*/
static int
zfileopen(const char *path, int mode, size64_t flags, void* parameters, NCZMAP** mapp)
{
int stat = NC_NOERR;
char* truepath = NULL;
ZFMAP* zfmap = NULL;
NCURI*url = NULL;
NC_UNUSED(parameters);
ZTRACE("%s %d %llu %p",path,mode,flags,parameters);
if(!zfinitialized) zfinitialize();
/* Fixup mode flags */
mode = (NC_NETCDF4 | mode);
if(flags & FLAG_BYTERANGE)
mode &= ~(NC_CLOBBER | NC_WRITE);
/* path must be a url with file: protocol*/
if((stat=zfparseurl(path,&url)))
goto done;
if(strcasecmp(url->protocol,"file") != 0)
{stat = NC_EURL; goto done;}
/* Canonicalize the root path */
if((stat = nczm_canonicalpath(url->path,&truepath))) goto done;
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
/* Build the z4 state */
if((zfmap = calloc(1,sizeof(ZFMAP))) == NULL)
{stat = NC_ENOMEM; goto done;}
zfmap->map.format = NCZM_FILE;
zfmap->map.url = ncuribuild(url,NULL,NULL,NCURIALL);
zfmap->map.flags = flags;
zfmap->map.mode = mode;
zfmap->map.api = (NCZMAP_API*)&zapi;
zfmap->root = truepath;
truepath = NULL;
/* Verify root dir exists */
switch (stat = platformopendir(zfmap,zfmap->root)) {
case NC_NOERR: break;
case NC_ENOTFOUND: stat = NC_EEMPTY; /* fall thru */
default:
goto done;
}
/* Dataset superblock will be read by higher layer */
if(mapp) *mapp = (NCZMAP*)zfmap;
done:
ncurifree(url);
nullfree(truepath);
if(stat) zfileclose((NCZMAP*)zfmap,0);
return (stat);
}
/**************************************************/
/* Object API */
static int
zfileexists(NCZMAP* map, const char* key)
{
int stat = NC_NOERR;
ZFMAP* zfmap = (ZFMAP*)map;
FD fd = FDNUL;
ZTRACE("%s",key);
switch(stat=zflookupobj(zfmap,key,&fd)) {
case NC_NOERR: break;
case NC_ENOTFOUND: stat = NC_EEMPTY;
case NC_EEMPTY: break;
default: break;
}
zfrelease(zfmap,&fd);
return (stat);
}
static int
zfilelen(NCZMAP* map, const char* key, size64_t* lenp)
{
int stat = NC_NOERR;
ZFMAP* zfmap = (ZFMAP*)map;
size64_t len = 0;
FD fd = FDNUL;
ZTRACE("%s",key);
switch (stat=zflookupobj(zfmap,key,&fd)) {
case NC_NOERR:
/* Get file size */
if((stat=platformseek(zfmap, &fd, SEEK_END, &len))) goto done;
break;
case NC_ENOTFOUND: stat = NC_EEMPTY;
case NC_EEMPTY: break;
default: break;
}
zfrelease(zfmap,&fd);
if(lenp) *lenp = len;
done:
return THROW(stat);
}
static int
zfiledefineobj(NCZMAP* map, const char* key)
{
int stat = NC_NOERR;
FD fd = FDNUL;
ZFMAP* zfmap = (ZFMAP*)map; /* cast to true type */
ZTRACE("%s",key);
#ifdef VERIFY
if(!verify(key,!FLAG_ISDIR))
assert(!"expected file, have dir");
#endif
/* Create the intermediate groups as directories */
if((stat = zfcreategroup(zfmap,key,SKIPLAST)))
goto done;
stat = zflookupobj(zfmap,key,&fd);
zfrelease(zfmap,&fd);
switch (stat) {
case NC_NOERR: /* Already exists */
goto done;
case NC_ENOTFOUND: stat = NC_EEMPTY; /* file does not exist */
case NC_EEMPTY: /* empty */
if((stat = zfcreateobj(zfmap,key,&fd)))
goto done;
break;
default:
goto done;
}
done:
zfrelease(zfmap,&fd);
return (stat);
}
static int
zfileread(NCZMAP* map, const char* key, size64_t start, size64_t count, void* content)
{
int stat = NC_NOERR;
FD fd = FDNUL;
ZFMAP* zfmap = (ZFMAP*)map; /* cast to true type */
ZTRACE("%s %llu %llu",key,start,count);
#ifdef VERIFY
if(!verify(key,!FLAG_ISDIR))
assert(!"expected file, have dir");
#endif
switch (stat = zflookupobj(zfmap,key,&fd)) {
case NC_NOERR:
if((stat = platformseek(zfmap, &fd, SEEK_SET, &start))) goto done;
if((stat = platformread(zfmap, &fd, count, content))) goto done;
break;
case NC_ENOTFOUND: stat = NC_EEMPTY;
case NC_EEMPTY: break;
default: break;
}
done:
zfrelease(zfmap,&fd);
return (stat);
}
static int
zfilewrite(NCZMAP* map, const char* key, size64_t start, size64_t count, const void* content)
{
int stat = NC_NOERR;
FD fd = FDNUL;
ZFMAP* zfmap = (ZFMAP*)map; /* cast to true type */
ZTRACE("%s %llu %llu",key,start,count);
#ifdef VERIFY
if(!verify(key,!FLAG_ISDIR))
assert(!"expected file, have dir");
#endif
switch (stat = zflookupobj(zfmap,key,&fd)) {
case NC_NOERR:
if((stat = platformseek(zfmap, &fd, SEEK_SET, &start))) goto done;
if((stat = platformwrite(zfmap, &fd, count, content))) goto done;
break;
case NC_ENOTFOUND: stat = NC_EEMPTY;
case NC_EEMPTY: break;
default: break;
}
done:
zfrelease(zfmap,&fd);
return (stat);
}
static int
zfileclose(NCZMAP* map, int delete)
{
int stat = NC_NOERR;
ZFMAP* zfmap = (ZFMAP*)map;
ZTRACE("%d",delete);
if(zfmap == NULL) return NC_NOERR;
/* Delete the subtree below the root and the root */
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
if(delete) {
stat = platformdelete(zfmap,zfmap->root,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
unlink(zfmap->root);
}
nczm_clear(map);
nullfree(zfmap->root);
zfmap->root = NULL;
free(zfmap);
return (stat);
}
/*
Return a list of keys immediately "below" a specified prefix key.
In theory, the returned list should be sorted in lexical order,
but it possible that it is not.
The prefix key is not included.
*/
int
zfilesearch(NCZMAP* map, const char* prefixkey, NClist* matches)
{
int stat = NC_NOERR;
int i;
ZFMAP* zfmap = (ZFMAP*)map;
char* truepath = NULL;
NClist* nextlevel = nclistnew();
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
NCbytes* buf = ncbytesnew();
int trailing;
ZTRACE("%s",prefixkey);
/* Make the root path be true */
if(prefixkey == NULL || strlen(prefixkey)==0 || strcmp(prefixkey,"/")==0)
truepath = strdup(zfmap->root);
else if((stat = nczm_concat(zfmap->root,prefixkey,&truepath))) goto done;
trailing = (prefixkey[strlen(prefixkey)-1] == '/');
/* get names of the next level path entries */
switch (stat = platformdircontent(zfmap, truepath, nextlevel)) {
case NC_NOERR: /* ok */
break;
case NC_EEMPTY: /* not a dir */
stat = NC_NOERR;
goto done;
case NC_ENOTFOUND: /* does not exist */
default:
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
goto done;
}
for(i=0;i<nclistlength(nextlevel);i++) {
const char* segment = nclistget(nextlevel,i);
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
ncbytescat(buf,prefixkey);
if(!trailing) ncbytescat(buf,"/");
ncbytescat(buf,segment);
nclistpush(matches,ncbytesextract(buf));
}
done:
nclistfreeall(nextlevel);
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
ncbytesfree(buf);
nullfree(truepath);
return THROW(stat);
}
/**************************************************/
/* Utilities */
/* Lookup a group by parsed path (segments)*/
/* Return NC_EEMPTY if not found, NC_EINVAL if not a directory; create if create flag is set */
static int
zfcreategroup(ZFMAP* zfmap, const char* key, int nskip)
{
int stat = NC_NOERR;
int i, len;
char* fullpath = NULL;
NCbytes* path = ncbytesnew();
NClist* segments = nclistnew();
ZTRACE("%s %d",key,nskip);
if((stat=nczm_split(key,segments)))
goto done;
len = nclistlength(segments);
len -= nskip; /* leave off last nskip segments */
ncbytescat(path,zfmap->root); /* We need path to be absolute */
for(i=0;i<len;i++) {
const char* seg = nclistget(segments,i);
ncbytescat(path,"/");
ncbytescat(path,seg);
/* open and optionally create the directory */
stat = platformcreatedir(zfmap,ncbytescontents(path));
if(stat) goto done;
}
done:
nullfree(fullpath);
ncbytesfree(path);
nclistfreeall(segments);
return (stat);
}
/* Lookup an object
@return NC_NOERR if found and is a content-bearing object
@return NC_EEMPTY if exists but is not-content-bearing
@return NC_ENOTFOUND if not found
*/
static int
zflookupobj(ZFMAP* zfmap, const char* key, FD* fd)
{
int stat = NC_NOERR;
char* path = NULL;
ZTRACE("%s",key);
if((stat = zffullpath(zfmap,key,&path)))
{goto done;}
/* See if this is content-bearing */
if((stat = platformtestcontentbearing(zfmap,path)))
goto done;
/* Open the file */
if((stat = platformopenfile(zfmap,path,fd)))
goto done;
done:
errno = 0;
nullfree(path);
return (stat);
}
/* When we are finished accessing object */
static void
zfrelease(ZFMAP* zfmap, FD* fd)
{
ZTRACE("",NULL);
platformrelease(zfmap,fd);
}
/* Create an object file corresponding to a key; create any
necessary intermediate groups. Assumed that we actually
want to create this as a file.
*/
static int
zfcreateobj(ZFMAP* zfmap, const char* key, FD* fd)
{
int stat = NC_NOERR;
char* fullpath = NULL;
ZTRACE("%s",key);
#ifdef VERIFY
if(!verify(key,!FLAG_ISDIR))
assert(!"expected file, have dir");
#endif
/* Create all the prefix groups as directories */
if((stat = zfcreategroup(zfmap, key, SKIPLAST))) goto done;
/* Create the final object */
if((stat=zffullpath(zfmap,key,&fullpath))) goto done;
if((stat = platformcreatefile(zfmap,fullpath,fd)))
goto done;
done:
nullfree(fullpath);
return (stat);
}
/**************************************************/
/* External API objects */
NCZMAP_DS_API zmap_nzf = {
NCZM_FILE_V1,
zfilecreate,
zfileopen,
};
static NCZMAP_API zapi = {
NCZM_FILE_V1,
zfileclose,
zfileexists,
zfilelen,
zfiledefineobj,
zfileread,
zfilewrite,
zfilesearch,
};
static int
zffullpath(ZFMAP* zfmap, const char* key, char** pathp)
{
int stat = NC_NOERR;
size_t klen, pxlen, flen;
char* path = NULL;
klen = nulllen(key);
pxlen = strlen(zfmap->root);
flen = klen+pxlen+1+1;
if((path = malloc(flen)) == NULL) {stat = NC_ENOMEM; goto done;}
path[0] = '\0';
strlcat(path,zfmap->root,flen);
/* look for special cases */
if(key != NULL) {
if(key[0] != '/') strlcat(path,"/",flen);
if(strcmp(key,"/") != 0)
strlcat(path,key,flen);
}
if(pathp) {*pathp = path; path = NULL;}
done:
nullfree(path)
return stat;
}
static int
zfparseurl(const char* path0, NCURI** urip)
{
int stat = NC_NOERR;
NCURI* uri = NULL;
ncuriparse(path0,&uri);
if(uri == NULL)
{stat = NC_EURL; goto done;}
if(urip) {*urip = uri; uri = NULL;}
done:
ncurifree(uri);
return stat;
}
/**************************************************/
static int
platformerr(int err)
{
switch (err) {
case ENOENT: err = NC_ENOTFOUND; break; /* File does not exist */
case ENOTDIR: err = NC_EEMPTY; break; /* no content */
case EACCES: err = NC_EAUTH; break; /* file permissions */
case EPERM: err = NC_EAUTH; break; /* ditto */
default: break;
}
return err;
}
/* Test type of the specified file.
@return NC_NOERR if found and is a content-bearing object (file)
@return NC_EEMPTY if exists but is not-content-bearing (a directory)
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 NC_ENOTFOUND if not found
*/
static int
platformtestcontentbearing(ZFMAP* zfmap, const char* truepath)
{
int ret = 0;
struct stat buf;
char* local = 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
/* Localize */
if((ret = nczm_localize(truepath,&local,LOCALIZE))) goto done;
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
errno = 0;
if((ret = stat(local, &buf)) < 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
ret = platformerr(errno);
} else if(S_ISDIR(buf.st_mode)) {
ret = NC_EEMPTY;
} else
ret = NC_NOERR;
done:
nullfree(local);
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
errno = 0;
return ret;
}
/* Create a file */
static int
platformcreatefile(ZFMAP* zfmap, const char* truepath, FD* fd)
{
int stat = NC_NOERR;
int ioflags = 0;
int createflags = 0;
int mode = zfmap->map.mode;
int permissions = NC_DEFAULT_ROPEN_PERMS;
errno = 0;
if(!fIsSet(mode, NC_WRITE))
ioflags |= (O_RDONLY);
else {
ioflags |= (O_RDWR);
permissions = NC_DEFAULT_RWOPEN_PERMS;
}
#ifdef O_BINARY
fSet(ioflags, O_BINARY);
#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
if(fIsSet(mode, NC_NOCLOBBER))
fSet(createflags, O_EXCL);
else
fSet(createflags, O_TRUNC);
if(fIsSet(mode,NC_WRITE))
createflags = (ioflags|O_CREAT);
/* Try to create file (will also localize) */
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
fd->fd = NCopen3(truepath, createflags, permissions);
if(fd->fd < 0) { /* could not create */
stat = platformerr(errno);
goto done; /* could not open */
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:
errno = 0;
return THROW(stat);
}
/* Open a file; fail if it does not exist */
static int
platformopenfile(ZFMAP* zfmap, const char* truepath, FD* fd)
{
int stat = NC_NOERR;
int ioflags = 0;
int mode = zfmap->map.mode;
int permissions = 0;
errno = 0;
if(!fIsSet(mode, NC_WRITE)) {
ioflags |= (O_RDONLY);
permissions = NC_DEFAULT_ROPEN_PERMS;
} else {
ioflags |= (O_RDWR);
permissions = NC_DEFAULT_RWOPEN_PERMS;
}
#ifdef O_BINARY
fSet(ioflags, O_BINARY);
#endif
#ifdef VERIFY
if(!verify(truepath,!FLAG_ISDIR))
assert(!"expected file, have dir");
#endif
/* Try to open file (will localize) */
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
fd->fd = NCopen3(truepath, ioflags, permissions);
if(fd->fd < 0)
{stat = platformerr(errno); goto done;} /* could not open */
done:
errno = 0;
return THROW(stat);
}
/* Create a dir */
static int
platformcreatedir(ZFMAP* zfmap, const char* truepath)
{
int ret = NC_NOERR;
int mode = zfmap->map.mode;
errno = 0;
/* Try to access file as if it exists */
ret = NCaccess(truepath,ACCESS_MODE_EXISTS);
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
if(ret < 0) { /* it does not exist, then it can be anything */
if(fIsSet(mode,NC_WRITE)) {
/* Try to create it */
/* Create the directory using mkdir */
if(NCmkdir(truepath,NC_DEFAULT_DIR_PERMS) < 0)
{ret = platformerr(errno); goto done;}
/* try to access again */
ret = NCaccess(truepath,ACCESS_MODE_EXISTS);
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
if(ret < 0)
{ret = platformerr(errno); goto done;}
} else
{ret = platformerr(errno); goto done;}
}
done:
errno = 0;
return THROW(ret);
}
/* Open a dir; fail if it does not exist */
static int
platformopendir(ZFMAP* zfmap, const char* truepath)
{
int ret = NC_NOERR;
errno = 0;
/* Try to access file as if it exists */
ret = NCaccess(truepath,ACCESS_MODE_EXISTS);
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
if(ret < 0)
{ret = platformerr(errno); goto done;}
done:
errno = 0;
return THROW(ret);
}
/**
Given a path, return the list of all files+dirs immediately below
the specified path: e.g. X s.t. path/X exists.
There are several possibilities:
1. path does not exist => return NC_ENOTFOUND
2. path is not a directory => return NC_EEMPTY and |contents| == 0
3. path is a directory => return NC_NOERR and |contents| >= 0
@return NC_NOERR if path is a directory
@return NC_EEMPTY if path is not a directory
@return NC_ENOTFOUND if path does not exist
*/
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 _WIN32
static int
platformdircontent(ZFMAP* zfmap, const char* truepath, NClist* contents)
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
{
int ret = NC_NOERR;
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
errno = 0;
WIN32_FIND_DATA FindFileData;
HANDLE dir = NULL;
char* ffpath = NULL;
char* lpath = NULL;
size_t len;
char* d = NULL;
switch (ret = platformtestcontentbearing(zfmap, truepath)) {
case NC_EEMPTY: ret = NC_NOERR; break; /* directory */
case NC_NOERR: ret = NC_EEMPTY; goto done;
default: goto done;
}
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
/* We need to process the path to make it work with FindFirstFile */
len = strlen(truepath);
/* Need to terminate path with '/''*' */
ffpath = (char*)malloc(len+2+1);
memcpy(ffpath,truepath,len);
if(truepath[len-1] != '/') {
ffpath[len] = '/';
len++;
}
ffpath[len] = '*'; len++;
ffpath[len] = '\0';
/* localize it */
if((ret = nczm_localize(ffpath,&lpath,LOCALIZE))) goto done;
dir = FindFirstFile(lpath, &FindFileData);
if(dir == INVALID_HANDLE_VALUE) {
/* Distinquish not-a-directory from no-matching-file */
switch (GetLastError()) {
case ERROR_FILE_NOT_FOUND: /* No matching files */ /* fall thru */
ret = NC_NOERR;
goto done;
case ERROR_DIRECTORY: /* not a directory */
default:
ret = NC_EEMPTY;
goto done;
}
}
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
do {
char* p = 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
const char* name = NULL;
name = FindFileData.cFileName;
if(strcmp(name,".")==0 || strcmp(name,"..")==0)
continue;
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
nclistpush(contents,strdup(name));
} while(FindNextFile(dir, &FindFileData));
done:
if(dir) FindClose(dir);
nullfree(lpath);
nullfree(ffpath);
nullfree(d);
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
errno = 0;
return THROW(ret);
}
#else /*!_WIN32*/
static int
platformdircontent(ZFMAP* zfmap, const char* truepath, NClist* contents)
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
{
int ret = NC_NOERR;
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
errno = 0;
DIR* dir = NULL;
switch (ret = platformtestcontentbearing(zfmap, truepath)) {
case NC_EEMPTY: ret = NC_NOERR; break; /* directory */
case NC_NOERR: ret = NC_EEMPTY; goto done;
default: goto done;
}
dir = NCopendir(truepath);
if(dir == NULL)
{ret = platformerr(errno); goto done;}
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
for(;;) {
const char* name = NULL;
struct dirent* de = NULL;
errno = 0;
de = readdir(dir);
if(de == NULL)
{ret = platformerr(errno); goto done;}
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
if(strcmp(de->d_name,".")==0 || strcmp(de->d_name,"..")==0)
continue;
name = de->d_name;
nclistpush(contents,strdup(name));
}
done:
if(dir) NCclosedir(dir);
errno = 0;
return THROW(ret);
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
}
#endif /*_WIN32*/
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
#if 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
static int
platformdeleter(ZFMAP* zfmap, NClist* segments, int depth)
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
{
int ret = NC_NOERR;
struct stat statbuf;
struct dirent* entry = NULL;
DIR* dir = NULL;
char* path = NULL;
char* tmp = 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
if((ret = nczm_join(segments,&path))) goto done;
/* When running on any platform that can accept drive letters */
if((ret = nczm_fixpath(path,&tmp))) goto done;
nullfree(path); path = NULL;
if((path = NCpathcvt(tmp))==NULL) {ret = NC_ENOMEM; goto done;}
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
errno = 0;
ret = stat(path, &statbuf);
if(ret < 0) {
if(errno == ENOENT) {ret = NC_NOERR; goto done;}
else {ret = platformerr(errno); goto done;}
}
/* process this file */
if(S_ISDIR(statbuf.st_mode)) {
if((dir = NCopendir(path)) == NULL)
{ret = platformerr(errno); goto done;}
for(;;) {
char* seg = NULL;
errno = 0;
entry = readdir(dir);
if(entry == NULL) {ret = platformerr(errno); break;}
/* Ignore "." and ".." */
if(strcmp(entry->d_name,".")==0) continue;
if(strcmp(entry->d_name,"..")==0) continue;
/* append name to segments */
if((seg = strdup(entry->d_name)) == NULL)
{ret = NC_ENOMEM; goto done;}
nclistpush(segments,seg);
/* recurse */
if((ret = platformdeleter(zfmap, segments, depth+1))) goto done;
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
/* remove+reclaim last segment */
nclistpop(segments);
nullfree(seg);
}
/* Delete this file */
rmdir(path);
} else {
assert(S_ISREG(statbuf.st_mode));
unlink(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
}
done:
if(dir) NCclosedir(dir);
/* delete this file|dir */
ret = NCremove(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
nullfree(path);
nullfree(tmp);
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 THROW(ret);
}
#endif /*0*/
Fix nczarr-experimental to better support CMake and find AWS libraries The primary fix is to improve CMake build support. Specific changes include: * CMake: Provide a better soln to locating the AWS SDK libraries; the new way is the preferred method as described in the aws-cpp-sdk documentation. * CMake (and Automake): allow -DENABLE_S3_SDK (default off) to suppress looking for AWS libraries. * CMake: add the complete set of nczarr tests * CMake: add EXTERNL as needed to various .h files. * Improve support for windows drive letters in paths. * Add nczarr and s3 flags to nc-config * For VisualStudio X nczarr, cleanup the NAN+INFINITY handling * Convert _MSC_VER -> _WIN32 and vice versa as needed * NCZarr - support multiple platform paths including windows, cygwin. mingw, etc. * NCZarr - sort the test outputs because different platforms produce directory contents in different orders. One big change concerns netcdf-c/CMakeLists.txt and netcdf-c/configure.ac. In the current versions, it was the case that --disable-hdf5 disabled netcdf-4 (libsrc4). With nczarr, this can no longer be the case because nczarr requires libsrc4 even if libhdf5 is disabled. So, I modified the above files to move the format options (HDF5, NCZarr, HDF4, etc) to a single place near the front of the files. Now it is the case that: * Enabling any of the formats that require libsrc4 also does an implicit --enable-netcdf4. * --disable-netcdf4 | --disable-netcdf-4 now becomes and alias for --disable-hdf5. There are probably some bugs in this change in terms of dependencies between format options. Problems: * CMake S3 support is still not working for Visual Studio * A recent issue points out that there is work to do on handling UTF8 filenames, but that will be addressed in a separate fix. Notes: * Consider converting all of our includes/.h files to use EXTERNL
2020-07-13 02:21:56 +08:00
static int
platformdeleter(ZFMAP* zfmap, NCbytes* truepath, int delroot, int depth)
{
int ret = NC_NOERR;
int i;
NClist* contents = nclistnew();
size_t tpathlen = ncbyteslength(truepath);
char* local = NULL;
#ifdef DEBUG
fprintf(stderr,"xxx: platformdeleter: depth=%d delroot=%d truepath=%s\n",depth,delroot,ncbytescontents(truepath));
#endif
ret = platformdircontent(zfmap, ncbytescontents(truepath), contents);
#ifdef DEBUG
{int i;
fprintf(stderr,"xxx: contents:\n");
for(i=0;i<nclistlength(contents);i++)
fprintf(stderr,"\t%s\n",(const char*)nclistget(contents,i));
fprintf(stderr,"xxx: end contents\n");
}
#endif
switch (ret) {
case NC_NOERR: /* recurse to remove levels below */
for(i=0;i<nclistlength(contents);i++) {
const char* name = nclistget(contents,i);
/* append name to current path */
ncbytescat(truepath, "/");
ncbytescat(truepath, name);
/* recurse */
if ((ret = platformdeleter(zfmap, truepath,delroot,depth+1))) goto done;
ncbytessetlength(truepath,tpathlen); /* reset */
ncbytesnull(truepath);
}
if(depth > 0 || delroot) {
/* localize and delete */
if((ret = nczm_localize(ncbytescontents(truepath),&local,LOCALIZE))) goto done;
#ifdef DEBUG
fprintf(stderr,"xxx: remove: %s\n",local);
#endif
if(NCrmdir(local) < 0) { /* kill this dir */
#ifdef DEBUG
fprintf(stderr,"xxx: remove: errno=%d|%s\n",errno,nc_strerror(errno));
#endif
ret = errno;
goto done;
}
}
break;
case NC_EEMPTY: /* Not a directory */
ret = NC_NOERR;
/* localize and delete */
if(local) {nullfree(local); local = NULL;}
if((ret = nczm_localize(ncbytescontents(truepath),&local,LOCALIZE))) goto done;
#ifdef DEBUG
fprintf(stderr,"xxx: remove: %s\n",local);
#endif
if(NCremove(local) < 0) {/* kill this file */
#ifdef DEBUG
fprintf(stderr,"xxx: remove: errno=%d|%s\n",errno,nc_strerror(errno));
#endif
ret = errno;
goto done;
}
break;
case NC_ENOTFOUND:
default:
goto done;
}
done:
nclistfreeall(contents);
nullfree(local);
ncbytessetlength(truepath,tpathlen);
ncbytesnull(truepath);
return THROW(ret);
}
/* Deep file/dir deletion; depth first */
static int
platformdelete(ZFMAP* zfmap, const char* rootpath, int delroot)
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
{
int stat = NC_NOERR;
NCbytes* truepath = ncbytesnew();
if(rootpath == NULL || strlen(rootpath) == 0) goto done;
ncbytescat(truepath,rootpath);
if(rootpath[strlen(rootpath)-1] == '/') /* elide trailing '/' */
ncbytessetlength(truepath,ncbyteslength(truepath)-1);
if((stat = platformdeleter(zfmap,truepath,delroot,0))) goto done;
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:
ncbytesfree(truepath);
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
errno = 0;
return THROW(stat);
}
static int
platformseek(ZFMAP* zfmap, FD* fd, int pos, size64_t* sizep)
{
int ret = NC_NOERR;
off_t size, newsize;
struct stat statbuf;
assert(fd && fd->fd >= 0);
errno = 0;
ret = fstat(fd->fd, &statbuf);
if(ret < 0)
{ret = platformerr(errno); goto done;}
if(sizep) size = *sizep; else size = 0;
newsize = lseek(fd->fd,size,pos);
if(sizep) *sizep = newsize;
done:
errno = 0;
return THROW(ret);
}
static int
platformread(ZFMAP* zfmap, FD* fd, size64_t count, void* content)
{
int stat = NC_NOERR;
size_t need = count;
unsigned char* readpoint = content;
assert(fd && fd->fd >= 0);
while(need > 0) {
ssize_t red;
if((red = read(fd->fd,readpoint,need)) <= 0)
{stat = errno; goto done;}
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
need -= red;
readpoint += red;
}
done:
return THROW(stat);
}
static int
platformwrite(ZFMAP* zfmap, FD* fd, size64_t count, const void* content)
{
int ret = NC_NOERR;
size_t need = count;
unsigned char* writepoint = (unsigned char*)content;
assert(fd && fd->fd >= 0);
while(need > 0) {
ssize_t red = 0;
if((red = write(fd->fd,(void*)writepoint,need)) <= 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
{ret = NC_EACCESS; goto done;}
need -= red;
writepoint += red;
}
done:
return THROW(ret);
}
#if 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
static int
platformcwd(char** cwdp)
{
char buf[4096];
char* cwd = NULL;
cwd = NCcwd(buf,sizeof(buf));
if(cwd == NULL) return errno;
if(cwdp) *cwdp = strdup(buf);
return NC_NOERR;
}
#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
/* When we are finished accessing FD; essentially
equivalent to closing the file descriptor.
*/
static void
platformrelease(ZFMAP* zfmap, FD* fd)
{
if(fd->fd >=0) NCclose(fd->fd);
fd->fd = -1;
}
#if 0
/* Close FD => return typ to FDNONE */
*/
static void
platformclose(ZFMAP* zfmap, FD* fd)
{
if(fd->typ == FDFILE) {
if(fd->fd >=0) close(fd->u,fd);
fd->fd = -1;
} else if(fd->type == FDDIR) {
if(fd->u.dir) NCclosedir(fd->u,dir);
}
fd->typ = FDNONE;
}
#endif
#ifdef VERIFY
static int
verify(const char* path, int isdir)
{
int ret = 0;
struct stat buf;
ret = NCaccess(path,ACCESS_MODE_EXISTS);
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
if(ret < 0)
return 1; /* If it does not exist, then it can be anything */
ret = stat(path,&buf);
if(ret < 0) abort();
if(isdir && S_ISDIR(buf.st_mode)) return 1;
if(!isdir && S_ISREG(buf.st_mode)) return 1;
return 0;
}
#endif
#if 0
/* Return NC_EINVAL if path does not exist; els 1/0 in isdirp and local path in truepathp */
static int
testifdir(const char* path, int* isdirp, char** truepathp)
{
int ret = NC_NOERR;
char* tmp = NULL;
char* truepath = NULL;
struct stat statbuf;
/* Make path be windows compatible */
if((ret = nczm_fixpath(path,&tmp))) goto done;
if((truepath = NCpathcvt(tmp))==NULL) {ret = NC_ENOMEM; goto done;}
errno = 0;
ret = stat(truepath, &statbuf);
if(ret < 0) {
if(errno == ENOENT)
ret = NC_ENOTFOUND; /* path does not exist */
else
ret = platformerr(errno);
goto done;
}
/* Check for being a directory */
if(isdirp) {
if(S_ISDIR(statbuf.st_mode)) {*isdirp = 1;} else {*isdirp = 0;}
}
if(truepathp) {*truepathp = truepath; truepath = NULL;}
done:
errno = 0;
nullfree(tmp);
nullfree(truepath);
return THROW(ret);
}
#endif /* 0 */