netcdf-c/libsrc4/ncindex.c
Dennis Heimbigner 59e04ae071 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.
2020-06-28 18:02:47 -06:00

411 lines
10 KiB
C

/*
Copyright (c) 1998-2018 University Corporation for Atmospheric Research/Unidata
See LICENSE.txt for license information.
*/
/** \file \internal
Internal netcdf-4 functions.
This file contains functions for manipulating ncindex objects.
Warning: This code depends critically on the assumption that
|void*| == |uintptr_t|
*/
/* Define this for debug so that table sizes are small */
#undef SMALLTABLE
#undef NCNOHASH
#include "config.h"
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#ifdef HAVE_STDINT_H
#include <stdint.h>
#endif
#include <assert.h>
#include "nc4internal.h"
#include "ncindex.h"
#ifdef SMALLTABLE
/* Keep the table sizes small initially */
#define DFALTTABLESIZE 7
#else
#define DFALTTABLESIZE 37
#endif
/* Hack to access internal state of a hashmap. Use with care */
/* Convert an entry from ACTIVE to DELETED;
Return 0 if not found.
*/
extern int NC_hashmapdeactivate(NC_hashmap*, uintptr_t data);
#ifndef NCNOHASH
extern void printhashmap(NC_hashmap*);
#endif
/* Locate object by name in an NCindex */
NC_OBJ*
ncindexlookup(NCindex* ncindex, const char* name)
{
NC_OBJ* obj = NULL;
if(ncindex == NULL || name == NULL)
return NULL;
{
#ifndef NCNOHASH
uintptr_t index;
assert(ncindex->map != NULL);
if(!NC_hashmapget(ncindex->map,(void*)name,strlen(name),&index))
return NULL; /* not present */
obj = (NC_OBJ*)nclistget(ncindex->list,(size_t)index);
#else
int i;
for(i=0;i<nclistlength(ncindex->list);i++) {
NC_OBJ* o = (NC_OBJ*)ncindex->list->content[i];
if(strcmp(o->name,name)==0) return o;
}
#endif
}
return obj;
}
/* Get ith object in the vector */
NC_OBJ*
ncindexith(NCindex* index, size_t i)
{
if(index == NULL) return NULL;
assert(index->list != NULL);
return nclistget(index->list,i);
}
/* See if x is contained in the index */
/* Return vector position if in index, otherwise return -1. */
int
ncindexfind(NCindex* index, NC_OBJ* nco)
{
int i;
NClist* list;
if(index == NULL || nco == NULL) return -1;
list = index->list;
for(i=0;i<nclistlength(list);i++) {
NC_OBJ* o = (NC_OBJ*)list->content[i];
if(nco == o) return i;
}
return -1;
}
/* Add object to the end of the vector, also insert into the hashmaps */
/* Return 1 if ok, 0 otherwise.*/
int
ncindexadd(NCindex* ncindex, NC_OBJ* obj)
{
if(ncindex == NULL) return 0;
#ifndef NCNOHASH
{
uintptr_t index; /*Note not the global id */
index = (uintptr_t)nclistlength(ncindex->list);
NC_hashmapadd(ncindex->map,index,(void*)obj->name,strlen(obj->name));
}
#endif
if(!nclistpush(ncindex->list,obj))
return 0;
return 1;
}
/* Insert object at ith position of the vector, also insert into the hashmaps; */
/* Return 1 if ok, 0 otherwise.*/
int
ncindexset(NCindex* ncindex, size_t i, NC_OBJ* obj)
{
if(ncindex == NULL) return 0;
if(!nclistset(ncindex->list,i,obj)) return 0;
#ifndef NCNOHASH
{
uintptr_t index = (uintptr_t)i;
NC_hashmapadd(ncindex->map,index,(void*)obj->name,strlen(obj->name));
}
#endif
return 1;
}
/**
* Remove ith object from the index;
* Return 1 if ok, 0 otherwise.*/
int
ncindexidel(NCindex* index, size_t i)
{
if(index == NULL) return 0;
nclistremove(index->list,i);
#ifndef NCNOHASH
/* Remove from the hash map by deactivating its entry */
if(!NC_hashmapdeactivate(index->map,(uintptr_t)i))
return 0; /* not present */
#endif
return 1;
}
/*Return a duplicate of the index's vector */
/* Return list if ok, NULL otherwise.*/
NC_OBJ**
ncindexdup(NCindex* index)
{
if(index == NULL || nclistlength(index->list) == 0)
return NULL;
return (NC_OBJ**)nclistclone(index->list,0/*!deep*/);
}
/* Count the non-null entries in an NCindex */
int
ncindexcount(NCindex* index)
{
int count = 0;
int i;
for(i=0;i<ncindexsize(index);i++) {
if(ncindexith(index,i) != NULL) count++;
}
return count;
}
/*
Rebuild the list map by rehashing all entries
using their current, possibly changed id and name;
also recompute their hashkey.
*/
/* Return 1 if ok, 0 otherwise.*/
int
ncindexrebuild(NCindex* index)
{
#ifndef NCNOHASH
size_t i;
size_t size = nclistlength(index->list);
NC_OBJ** contents = (NC_OBJ**)nclistextract(index->list);
/* Reset the index map and list*/
nclistfree(index->list);
index->list = nclistnew();
nclistsetalloc(index->list,size);
NC_hashmapfree(index->map);
index->map = NC_hashmapnew(size);
/* Now, reinsert all the attributes except NULLs */
for(i=0;i<size;i++) {
NC_OBJ* tmp = contents[i];
if(tmp == NULL) continue; /* ignore */
if(!ncindexadd(index,tmp))
return 0;
}
#endif
if(contents != NULL) free(contents);
return 1;
}
/* Free a list map */
int
ncindexfree(NCindex* index)
{
if(index == NULL) return 1;
nclistfree(index->list);
NC_hashmapfree(index->map);
free(index);
return 1;
}
/* Create a new index */
NCindex*
ncindexnew(size_t size0)
{
NCindex* index = NULL;
size_t size = (size0 == 0 ? DFALTTABLESIZE : size0);
index = calloc(1,sizeof(NCindex));
if(index == NULL) return NULL;
index->list = nclistnew();
if(index->list == NULL) {ncindexfree(index); return NULL;}
nclistsetalloc(index->list,size);
#ifndef NCNOHASH
index->map = NC_hashmapnew(size);
if(index->map == NULL) {ncindexfree(index); return NULL;}
#endif
return index;
}
#ifndef NCNOHASH
/* Debug: handle the data key part */
static const char*
keystr(NC_hentry* e)
{
if(e->keysize < sizeof(uintptr_t))
return (const char*)(&e->key);
else
return (const char*)(e->key);
}
#endif
int
ncindexverify(NCindex* lm, int dump)
{
size_t i;
NClist* l = lm->list;
int nerrs = 0;
#ifndef NCNOHASH
size_t m;
#endif
if(lm == NULL) {
fprintf(stderr,"index: <empty>\n");
return 1;
}
if(dump) {
fprintf(stderr,"-------------------------\n");
#ifndef NCNOHASH
if(lm->map->active == 0) {
fprintf(stderr,"hash: <empty>\n");
goto next1;
}
for(i=0;i < lm->map->alloc; i++) {
NC_hentry* e = &lm->map->table[i];
if(e->flags != 1) continue;
fprintf(stderr,"hash: %ld: data=%lu key=%s\n",(unsigned long)i,(unsigned long)e->data,keystr(e));
fflush(stderr);
}
next1:
#endif
if(nclistlength(l) == 0) {
fprintf(stderr,"list: <empty>\n");
goto next2;
}
for(i=0;i < nclistlength(l); i++) {
const char** a = (const char**)nclistget(l,i);
fprintf(stderr,"list: %ld: name=%s\n",(unsigned long)i,*a);
fflush(stderr);
}
fprintf(stderr,"-------------------------\n");
fflush(stderr);
}
next2:
#ifndef NCNOHASH
/* Need to verify that every entry in map is also in vector and vice-versa */
/* Verify that map entry points to same-named entry in vector */
for(m=0;m < lm->map->alloc; m++) {
NC_hentry* e = &lm->map->table[m];
char** object = NULL;
char* oname = NULL;
uintptr_t udata = (uintptr_t)e->data;
if((e->flags & 1) == 0) continue;
object = nclistget(l,(size_t)udata);
if(object == NULL) {
fprintf(stderr,"bad data: %d: %lu\n",(int)m,(unsigned long)udata);
nerrs++;
} else {
oname = *object;
if(strcmp(oname,keystr(e)) != 0) {
fprintf(stderr,"name mismatch: %d: %lu: hash=%s list=%s\n",
(int)m,(unsigned long)udata,keystr(e),oname);
nerrs++;
}
}
}
/* Walk vector and mark corresponding hash entry*/
if(nclistlength(l) == 0 || lm->map->active == 0)
goto done; /* cannot verify */
for(i=0;i < nclistlength(l); i++) {
int match;
const char** xp = (const char**)nclistget(l,i);
/* Walk map looking for *xp */
for(match=0,m=0;m < lm->map->active; m++) {
NC_hentry* e = &lm->map->table[m];
if((e->flags & 1) == 0) continue;
if(strcmp(keystr(e),*xp)==0) {
if((e->flags & 128) == 128) {
fprintf(stderr,"%ld: %s already in map at %ld\n",(unsigned long)i,keystr(e),(unsigned long)m);
nerrs++;
}
match = 1;
e->flags += 128;
}
}
if(!match) {
fprintf(stderr,"mismatch: %d: %s in vector, not in map\n",(int)i,*xp);
nerrs++;
}
}
/* Verify that every element in map in in vector */
for(m=0;m < lm->map->active; m++) {
NC_hentry* e = &lm->map->table[m];
if((e->flags & 1) == 0) continue;
if((e->flags & 128) == 128) continue;
/* We have a hash entry not in the vector */
fprintf(stderr,"mismatch: %d: %s->%lu in hash, not in vector\n",(int)m,keystr(e),(unsigned long)e->data);
nerrs++;
}
/* clear the 'touched' flag */
for(m=0;m < lm->map->active; m++) {
NC_hentry* e = &lm->map->table[m];
e->flags &= ~128;
}
done:
#endif /*NCNOHASH*/
fflush(stderr);
return (nerrs > 0 ? 0: 1);
}
static const char*
sortname(NC_SORT sort)
{
switch(sort) {
case NCNAT: return "NCNAT";
case NCVAR: return "NCVAR";
case NCDIM: return "NCDIM";
case NCATT: return "NCATT";
case NCTYP: return "NCTYP";
case NCGRP: return "NCGRP";
default: break;
}
return "unknown";
}
void
printindexlist(NClist* lm)
{
int i;
if(lm == NULL) {
fprintf(stderr,"<empty>\n");
return;
}
for(i=0;i<nclistlength(lm);i++) {
NC_OBJ* o = (NC_OBJ*)nclistget(lm,i);
if(o == NULL)
fprintf(stderr,"[%ld] <null>\n",(unsigned long)i);
else
fprintf(stderr,"[%ld] sort=%s name=|%s| id=%lu hashkey=%lu\n",
(unsigned long)i,sortname(o->sort),o->name,(unsigned long)o->id,(unsigned long)o->hashkey);
}
}
#ifndef NCNOHASH
void
printindexmap(NCindex* lm)
{
if(lm == NULL) {
fprintf(stderr,"<empty>\n");
return;
}
printhashmap(lm->map);
}
#endif
void
printindex(NCindex* lm)
{
if(lm == NULL) {
fprintf(stderr,"<empty>\n");
return;
}
printindexlist(lm->list);
#ifndef NCNOHASH
printindexmap(lm);
#endif
}