| nested structures in dlm [message #52210] |
Fri, 12 January 2007 04:55  |
lbusoni
Messages: 7
Registered: January 2007
|
Junior Member |
|
|
HI Guru's of DLMs,
I need to convert a rather complex C++ structure to something easy to
manage in IDL.
The data I have to convert are a list of dictionaries of (time,force)
pairs:
For instance to object V000 belong 2 array ("time and "force") of n0
elements,
to object V001 belong 2 array ("time and "force") of n1 elements,
to object V00M belong 2 array ("time and "force") of nm elements.
I know the value of M and of the varius n0, n1, ..., nm only at
run-time, i.e. I don't know in advance how many object I will get from
the query, nor how many elements each one of these objects will
contain.
Hope the description is clear enough.
I tested with a small function that assign to its argument a structure
FOO of structures V000 to V004.
Each substructure contains a pair of array of different length:
v000.time and v000.force are dblarr[3], v001.time and v001.force are
dblarr[4] and so on.
Here is the code:
static IDL_MEMINT times_dims[] = { 1, 1 };
static IDL_MEMINT force_dims[] = { 1, 1 };
static IDL_STRUCT_TAG_DEF substruct_tags[] = {
{"TIME", times_dims, (void *) IDL_TYP_DOUBLE},
{"FORCE", force_dims, (void *) IDL_TYP_DOUBLE},
{0}
};
IDL_VPTR IDLTestS(int Argc, IDL_VPTR Argv[])
{
IDL_VPTR variabile = Argv[0];
void *s;
void *struct_s;
IDL_MEMINT n_ele=1;
// Dummy data
int size=100000;
double *s_data = (double*)malloc(size*sizeof(double));
for (int i=0; i<size; i++){
s_data[i]=i;
}
// I need to create the IDL_STRUCT_TAG_DEF [] at run time
// because I don't know a priori the number of objects
int n_of_objects=5;
IDL_STRUCT_TAG_DEF *struct_tags = (IDL_STRUCT_TAG_DEF*)
malloc(sizeof(IDL_STRUCT_TAG_DEF) * (n_of_objects+1) );
IDL_STRUCT_TAG_DEF *tag;
for (int i=0; i<n_of_objects; i++){
tag = &struct_tags[i];
tag->name=(char*)malloc(5);
snprintf(tag->name,5,"V%03d",i);
tag->dims=(IDL_MEMINT*) malloc(2*sizeof(IDL_MEMINT));
tag->dims[0]=1;
tag->dims[1]=1;
tag->type=NULL;
}
// terminating the array of IDL_STRUCT_TAG_DEF
tag = &struct_tags[n_of_objects];
tag->name=0;
// create substructs
for (int i=0; i<n_of_objects; i++){
char nome[5];
snprintf(nome,5,"V%03d",i);
times_dims[1] = 3+i;
force_dims[1] = 3+i;
s = IDL_MakeStruct(nome, substruct_tags);
struct_tags[i].type =s ;
}
// create main struct
struct_s = IDL_MakeStruct("FOO", struct_tags);
// see if IDL_STRUCT_TAG_DEF [] is correct
printf("printing struct_tags\n");
{
int itag=0;
IDL_STRUCT_TAG_DEF *tag = &struct_tags[itag];
while ( ((char*)tag)[0] != 0) {
printf("%s - (%ld %ld) - %p\n",tag->name, tag->dims[0],
tag->dims[1], tag->type);
tag = &struct_tags[++itag];
}
}
// attach data to the created structure
IDL_VPTR vv = IDL_ImportArray( 1, &n_ele, IDL_TYP_STRUCT, (UCHAR
*)s_data, idl_free_cb, struct_s);
IDL_VarCopy(vv, variabile);
return IDL_GettmpLong(1);
}
The main point is that, instead of having the IDL_STRUCT_TAG_DEF array
defined at compile time as usual, I need to create this structure at
run time.
That's the output of the test:
IDL> print, tests(a)
% Loaded DLM: TESTS.
printing struct_tags
V000 - (1 1) - 0x823dec4
V001 - (1 1) - 0x823dfbc
V002 - (1 1) - 0x823e0b4
V003 - (1 1) - 0x823e1ac
V004 - (1 1) - 0x823e2a4
1
IDL> help ,a , /str
** Structure FOO, 7 tags, length=400, data length=400:
TIME DOUBLE Array[3]
FORCE DOUBLE Array[3]
V001 STRUCT -> V001 Array[1]
V002 STRUCT -> V002 Array[1]
TIME DOUBLE Array[6]
FORCE DOUBLE Array[6]
V004 STRUCT -> V004 Array[1]
IDL> print, a
{ 0.0000000 1.0000000 2.0000000
3.0000000 4.0000000 5.0000000
{
6.0000000 7.0000000 8.0000000 9.0000000
10.000000 11.000000 12.000000 13.000000
}{
14.000000 15.000000 16.000000 17.000000
18.000000
19.000000 20.000000 21.000000 22.000000
23.000000
}
24.000000 25.000000 26.000000 27.000000
28.000000 29.000000
30.000000 31.000000 32.000000 33.000000
34.000000 35.000000
{
36.000000 37.000000 38.000000 39.000000
40.000000 41.000000 42.000000
43.000000 44.000000 45.000000 46.000000
47.000000 48.000000 49.000000
}}
IDL> help, a.v001, /str
** Structure V001, 2 tags, length=64, data length=64:
TIME DOUBLE Array[4]
FORCE DOUBLE Array[4]
IDL>
As you can see, the variable "a" now contains a structure named "FOO"
containing the correct data, but instead of having a.v000, a.v001, ...,
a.v004, V000 and V003 (in this example) have not been created
correctly.
Of course there's nothing special in V000 and v0003 and indeed this
behaviour changes from time to time, (sometimes all the substructures
are OK).
It seems that me and IDL_MakeStruct got confused :)
Any idea of what's happening? My code is completely crazy?
Thanks
Lorenzo
|
|
|
|
| Re: nested structures [message #84441 is a reply to message #52210] |
Tue, 28 May 2013 00:28  |
hannah_ue
Messages: 14
Registered: November 2012
|
Junior Member |
|
|
Thank you Phillip, that helped. I finally figured out (I hope) how to reference to the array of structures in the array of structures and how to replicate those independently. I think I'm getting on now.
Am Montag, 27. Mai 2013 17:03:01 UTC+2 schrieb Phillip Bitzer:
> You can certainly do what you're after. In fact, I do this sort of thing when building arrays of radar data, which may have different lengths, sizes, etc.
>
>
>
> First, some basic pointer stuff:
>
>
>
> Consider:
>
> IDL> s1 = {tag1:0L, tag2:PTR_NEW(/ALLOCATE)}
>
>
>
> Then,
>
> IDL> help, s1
>
> ** Structure <314b91d8>, 2 tags, length=8, data length=8, refs=1:
>
> TAG1 LONG 0
>
> TAG2 POINTER <PtrHeapVar14>
>
>
>
> So, we see tag2 is a pointer. Fine, let's assign the pointer to a (new) structure:
>
> IDL> *s1.tag2 = {ntag1:0L, nTag2:0L}
>
>
>
> Okey doke. So, s1.tag2 is the pointer, and when we dereference this:
>
> IDL> help, *s1.tag2
>
> ** Structure <1dc84338>, 2 tags, length=8, data length=8, refs=1:
>
> NTAG1 LONG 0
>
> NTAG2 LONG 0
>
>
>
> we see our (new) structure.
>
>
>
> What about getting to one of these tags? Notice this doesn't work:
>
> IDL> help, *s1.tag2.ntag2
>
> % Expression must be a structure in this context: <No name>.
>
> % Execution halted at: $MAIN$
>
>
>
> But this does:
>
> IDL> help, (*s1.tag2).ntag2
>
> <Expression> LONG = 0
>
>
>
> Remember, *s1.tag2 is the pointer, and that's what what we want to dereference. That's why the parentheses are where they are.
>
>
>
> Arrays of structures with pointers can be a little more tricky, because you'll be throwing brackets in there too. Just keep in mind where the pointer is.
>
>
>
> Further, you'll want to take a look a this for the initialization:
>
>
>
> http://www.idlcoyote.com/code_tips/structptrinit.html
|
|
|
|
| Re: nested structures [message #84454 is a reply to message #52210] |
Mon, 27 May 2013 08:03 |
Phillip Bitzer
Messages: 223
Registered: June 2006
|
Senior Member |
|
|
You can certainly do what you're after. In fact, I do this sort of thing when building arrays of radar data, which may have different lengths, sizes, etc.
First, some basic pointer stuff:
Consider:
IDL> s1 = {tag1:0L, tag2:PTR_NEW(/ALLOCATE)}
Then,
IDL> help, s1
** Structure <314b91d8>, 2 tags, length=8, data length=8, refs=1:
TAG1 LONG 0
TAG2 POINTER <PtrHeapVar14>
So, we see tag2 is a pointer. Fine, let's assign the pointer to a (new) structure:
IDL> *s1.tag2 = {ntag1:0L, nTag2:0L}
Okey doke. So, s1.tag2 is the pointer, and when we dereference this:
IDL> help, *s1.tag2
** Structure <1dc84338>, 2 tags, length=8, data length=8, refs=1:
NTAG1 LONG 0
NTAG2 LONG 0
we see our (new) structure.
What about getting to one of these tags? Notice this doesn't work:
IDL> help, *s1.tag2.ntag2
% Expression must be a structure in this context: <No name>.
% Execution halted at: $MAIN$
But this does:
IDL> help, (*s1.tag2).ntag2
<Expression> LONG = 0
Remember, *s1.tag2 is the pointer, and that's what what we want to dereference. That's why the parentheses are where they are.
Arrays of structures with pointers can be a little more tricky, because you'll be throwing brackets in there too. Just keep in mind where the pointer is.
Further, you'll want to take a look a this for the initialization:
http://www.idlcoyote.com/code_tips/structptrinit.html
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
