Convert data in nested structure to array

Hi everybody,

I have a nested structure like this one:

S.a1.b1.x=1
S.a1.b1.y=2
S.a1.b1.u=3
S.a1.b1.v=4

S.a1.b2.x=5
S.a1.b2.y=6
S.a1.b2.u=7
S.a1.b2.v=8

S.a2.b1.x=9
S.a2.b1.y=10
S.a2.b1.u=11
S.a2.b1.v=12

S.a2.b2.x=13
S.a2.b2.y=14
S.a2.b2.u=15
S.a2.b2.v=16

Now I would like to plot all the data in the structure using quiver. To do so I need to extract all the x, y, u and v in the structure with something like:

X=S.(:).(:).x
Y=S.(:).(:).y
U=S.(:).(:).u
V=S.(:).(:).v

quiver(X,Y,U,V)

Which is the best (readable, fast, best-practice) way to do so? I'm starting to think that maybe it has been wrong to save these data into a nested struct. Is there a better way?

In order to answer this latter question I need to point out that the user must be able to specify which "a" (a1, a2, ...) sub structure to extract. That's the reason why I decided to use a structure in first instance: to be able to use commands like

S.(choice).b1
S.(choice).b2

ecc

Thank you very much,

Luca Amerio

Answer 1

[Guillaume]

The most straightforward I can think of getting at your data is using structfun repeatedly on each subfield

points = cell2mat(structfun(@(fa) {cell2mat(structfun(@(fb) {structfun(@(fxyzu) fxyzu, fb)'}, fa))}, S);
XYUV = num2cell(points, 1);
quiver(XYUV{:})

Answer 2

[Tim Baur]

@Guillaume

Would you know a solution that could extract the desired values/rangens dynamically from a structured array, too?

So depending on the number of levels and the selected indeces, e.g.:

X = S(1:3).a(:).b(1).c(4:5).d >> 4x nested structfun

X = S(1:3).a(:) >> 1x nested structfun

X = S(1:3).a(:).b(1).c(4:5).d.e.f.g.h(:) >> 8x nested structfun

etc.

That would be super useful in case the depth of the structure is not always the same