netcdf-c/libdispatch/dhttp.c
Dennis Heimbigner bf2746b8ea Provide byte-range reading of remote datasets
re: issue https://github.com/Unidata/netcdf-c/issues/1251

Assume that you have the URL to a remote dataset
which is a normal netcdf-3 or netcdf-4 file.

This PR allows the netcdf-c to read that dataset's
contents as a netcdf file using HTTP byte ranges
if the remote server supports byte-range access.

Originally, this PR was set up to access Amazon S3 objects,
but it can also access other remote datasets such as those
provided by a Thredds server via the HTTPServer access protocol.
It may also work for other kinds of servers.

Note that this is not intended as a true production
capability because, as is known, this kind of access to
can be quite slow. In addition, the byte-range IO drivers
do not currently do any sort of optimization or caching.

An additional goal here is to gain some experience with
the Amazon S3 REST protocol.

This architecture and its use documented in
the file docs/byterange.dox.

There are currently two test cases:

1. nc_test/tst_s3raw.c - this does a simple open, check format, close cycle
   for a remote netcdf-3 file and a remote netcdf-4 file.
2. nc_test/test_s3raw.sh - this uses ncdump to investigate some remote
   datasets.

This PR also incorporates significantly changed model inference code
(see the superceded PR https://github.com/Unidata/netcdf-c/pull/1259).

1. It centralizes the code that infers the dispatcher.
2. It adds support for byte-range URLs

Other changes:

1. NC_HDF5_finalize was not being properly called by nc_finalize().
2. Fix minor bug in ncgen3.l
3. fix memory leak in nc4info.c
4. add code to walk the .daprc triples and to replace protocol=
   fragment tag with a more general mode= tag.

Final Note:
Th inference code is still way too complicated. We need to move
to the validfile() model used by netcdf Java, where each
dispatcher is asked if it can process the file. This decentralizes
the inference code. This will be done after all the major new
dispatchers (PIO, Zarr, etc) have been implemented.
2019-01-01 18:27:36 -07:00

351 lines
8.2 KiB
C

/**
* @file
*
* Read a range of data from a remote dataset.
*
* Copyright 2018 University Corporation for Atmospheric
* Research/Unidata. See COPYRIGHT file for more info.
*/
#include "config.h"
#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#ifdef HAVE_UNISTD_H
#include <unistd.h>
#endif
#define CURL_DISABLE_TYPECHECK 1
#include <curl/curl.h>
#include "netcdf.h"
#include "nclog.h"
#include "ncbytes.h"
#include "nclist.h"
#include "nchttp.h"
#undef TRACE
#define CURLERR(e) (e)
/* Mnemonics */
#define GETCMD 0
#define HEADCMD 1
/* Forward */
static int setupconn(CURL* curl, const char* objecturl, NCbytes* buf);
static int execute(CURL* curl, int headcmd, long* httpcodep);
static int headerson(CURL* curl, NClist* list);
static void headersoff(CURL* curl);
#ifdef TRACE
static void
dbgflush() {
fflush(stderr);
fflush(stdout);
}
static void
Trace(const char* fcn)
{
fprintf(stdout,"xxx: %s\n",fcn);
dbgflush();
}
#else
#define dbgflush()
#define Trace(fcn)
#endif /*TRACE*/
/**************************************************/
/**
@param objecturl url we propose to access
@param curlp curl handle stored here if non-NULL
@param filelenp store length of the file here if non-NULL
*/
int
nc_http_open(const char* objecturl, void** curlp, long long* filelenp)
{
int stat = NC_NOERR;
CURL* curl = NULL;
int i;
NClist* list = NULL;
Trace("open");
/* initialize curl*/
curl = curl_easy_init();
if (curl == NULL) stat = NC_ECURL;
if(curlp && curl) *curlp = (void*)curl;
if(filelenp) {
*filelenp = -1;
/* Attempt to get the file length using HEAD */
list = nclistnew();
if((stat = setupconn(curl,objecturl,NULL))) goto done;
if((stat = headerson(curl,list))) goto done;
if((stat = execute(curl,HEADCMD,NULL))) goto done;
headersoff(curl);
for(i=0;i<nclistlength(list);i+=2) {
char* s = nclistget(list,i);
if(strcasecmp(s,"content-length")==0) {
s = nclistget(list,i+1);
sscanf(s,"%lld",filelenp);
break;
}
/* Also check for the Accept-Ranges header */
if(strcasecmp(s,"accept-ranges")==0) {
s = nclistget(list,i+1);
if(strcasecmp(s,"bytes")!=0) /* oops! */
{stat = NC_EACCESS; goto done;}
}
}
}
done:
nclistfreeall(list);
dbgflush();
return stat;
}
int
nc_http_close(void* curl0)
{
int stat = NC_NOERR;
CURL* curl = curl0;
Trace("close");
if(curl != NULL)
(void)curl_easy_cleanup(curl);
dbgflush();
return stat;
}
/**
Assume URL etc has already been set.
@param curl curl handle
@param start starting offset
@param count number of bytes to read
@param buf store read data here -- caller must allocate and free
*/
int
nc_http_read(CURL* curl, const char* objecturl, fileoffset_t start, fileoffset_t count, NCbytes* buf)
{
int stat = NC_NOERR;
char range[64];
long httpcode = 200;
CURLcode cstat = CURLE_OK;
Trace("read");
if(count == 0)
goto done; /* do not attempt to read */
if((stat = setupconn(curl,objecturl,buf)))
goto fail;
/* Set to read byte range */
snprintf(range,sizeof(range),"%ld-%ld",(long)start,(long)((start+count)-1));
cstat = CURLERR(curl_easy_setopt(curl, CURLOPT_RANGE, range));
if(cstat != CURLE_OK)
{stat = NC_ECURL; goto done;}
if((stat = execute(curl,GETCMD,&httpcode)))
goto done;
done:
dbgflush();
return stat;
fail:
stat = NC_ECURL;
goto done;
}
static size_t
WriteMemoryCallback(void *ptr, size_t size, size_t nmemb, void *data)
{
NCbytes* buf = data;
size_t realsize = size * nmemb;
Trace("WriteMemoryCallback");
if(realsize == 0)
nclog(NCLOGWARN,"WriteMemoryCallback: zero sized chunk");
ncbytesappendn(buf, ptr, realsize);
return realsize;
}
static void
trim(char* s)
{
size_t l = strlen(s);
char* p = s;
char* q = s + l;
if(l == 0) return;
q--; /* point to last char of string */
/* Walk backward to first non-whitespace */
for(;q > p;q--) {
if(*q > ' ') break; /* found last non-whitespace */
}
/* invariant: p == q || *q > ' ' */
if(p == q) /* string is all whitespace */
{*p = '\0';}
else {/* *q is last non-whitespace */
q++; /* point to actual whitespace */
*q = '\0';
}
/* Ok, skip past leading whitespace */
for(p=s;*p;p++) {if(*p > ' ') break;}
/* invariant: *p == '\0' || *p > ' ' */
if(*p == 0) return; /* no leading whitespace */
/* Ok, overwrite any leading whitespace */
for(q=s;*p;) {*q++ = *p++;}
*q = '\0';
return;
}
static size_t
HeaderCallback(char *buffer, size_t size, size_t nitems, void *data)
{
NClist* list = data;
size_t realsize = size * nitems;
char* name = NULL;
char* value = NULL;
char* p = NULL;
size_t i;
int havecolon;
Trace("HeaderCallback");
if(realsize == 0)
nclog(NCLOGWARN,"HeaderCallback: zero sized chunk");
i = 0;
/* Look for colon separator */
for(p=buffer;(i < realsize) && (*p != ':');p++,i++);
havecolon = (i < realsize);
if(i == 0)
nclog(NCLOGWARN,"HeaderCallback: malformed header: %s",buffer);
name = malloc(i+1);
memcpy(name,buffer,i);
name[i] = '\0';
value = NULL;
if(havecolon) {
size_t vlen = (realsize - i);
value = malloc(vlen+1);
p++; /* skip colon */
memcpy(value,p,vlen);
value[vlen] = '\0';
trim(value);
}
nclistpush(list,name);
name = NULL;
if(value == NULL) value = strdup("");
nclistpush(list,value);
value = NULL;
return realsize;
}
static int
setupconn(CURL* curl, const char* objecturl, NCbytes* buf)
{
int stat = NC_NOERR;
CURLcode cstat = CURLE_OK;
if(objecturl != NULL) {
/* Set the URL */
cstat = CURLERR(curl_easy_setopt(curl, CURLOPT_URL, (void*)objecturl));
if (cstat != CURLE_OK) goto fail;
}
/* Set options */
cstat = CURLERR(curl_easy_setopt(curl, CURLOPT_TIMEOUT, 100)); /* 30sec timeout*/
if (cstat != CURLE_OK) goto fail;
cstat = CURLERR(curl_easy_setopt(curl, CURLOPT_CONNECTTIMEOUT, 100));
if (cstat != CURLE_OK) goto fail;
cstat = CURLERR(curl_easy_setopt(curl, CURLOPT_NOPROGRESS, 1));
if (cstat != CURLE_OK) goto fail;
if(buf != NULL) {
/* send all data to this function */
cstat = CURLERR(curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, WriteMemoryCallback));
if (cstat != CURLE_OK) goto fail;
/* Set argument for WriteMemoryCallback */
cstat = CURLERR(curl_easy_setopt(curl, CURLOPT_WRITEDATA, (void*)buf));
if (cstat != CURLE_OK) goto fail;
} else {/* turn off data capture */
(void)CURLERR(curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, NULL));
(void)CURLERR(curl_easy_setopt(curl, CURLOPT_WRITEDATA, NULL));
}
/* Turn off header capture */
headersoff(curl);
done:
return stat;
fail:
stat = NC_ECURL;
goto done;
}
static int
execute(CURL* curl, int headcmd, long* httpcodep)
{
int stat = NC_NOERR;
CURLcode cstat = CURLE_OK;
long httpcode = 0;
if(headcmd) {
cstat = CURLERR(curl_easy_setopt(curl, CURLOPT_NOBODY, 1L));
if(cstat != CURLE_OK) goto fail;
}
cstat = CURLERR(curl_easy_perform(curl));
if(cstat != CURLE_OK) goto fail;
cstat = CURLERR(curl_easy_getinfo(curl,CURLINFO_RESPONSE_CODE,&httpcode));
if(cstat != CURLE_OK) httpcode = 0;
if(httpcodep) *httpcodep = httpcode;
if(headcmd) {
cstat = CURLERR(curl_easy_setopt(curl, CURLOPT_NOBODY, 0L));
if(cstat != CURLE_OK) goto fail;
}
done:
return stat;
fail:
stat = NC_ECURL;
goto done;
}
static int
headerson(CURL* curl, NClist* list)
{
int stat = NC_NOERR;
CURLcode cstat = CURLE_OK;
cstat = CURLERR(curl_easy_setopt(curl, CURLOPT_HEADERFUNCTION, HeaderCallback));
if(cstat != CURLE_OK) goto fail;
cstat = CURLERR(curl_easy_setopt(curl, CURLOPT_HEADERDATA, (void*)list));
if (cstat != CURLE_OK) goto fail;
done:
return stat;
fail:
stat = NC_ECURL;
goto done;
}
static void
headersoff(CURL* curl)
{
(void)CURLERR(curl_easy_setopt(curl, CURLOPT_HEADERFUNCTION, NULL));
(void)CURLERR(curl_easy_setopt(curl, CURLOPT_HEADERDATA, NULL));
}
#ifdef IGNORE
static void
reset(CURL* curl)
{
(void)curl_easy_reset(curl);
}
#endif