Is there a way to extract specific fields from a structure?

I have a 1x1 structure with a large number of fields of which I'd like to extract some subset. I don't want to have to use

[myst.field1 myst.field2 myst.field3]...

Is there a way to shorten the code so I only have to type myst once and just extract the fields necessary? The structure name is nested within another structure so the code can get lengthy.

Answer 1

[Stephen23]

Solution

>> S.a = 'Hello';
>> S.b = 'World';
>> C = {'a','b'};
>> D = cellfun(@(f)getfield(S,f),C,'UniformOutput',false);
>> horzcat(D{:})
ans = HelloWorld

Take a look at fieldnames.

Improvement

Note that if the fieldnames really are numbered like that then a much better solution would be to consider using a non-scalar array. A non-scalar structure is a much better way to store data, compared to numbered fieldnames. Using a non-scalar structure would avoid having to perform slow and complicated tricks like this, instead you could use the neat and fast methods shown in the link:

>> S(1).f = 'Hello';
>> S(2).f = 'World';
>> [S.f]
ans = HelloWorld

See how choosing the best data storage makes a huge difference to the quality of code!

[TastyPastry]

Oh yeah they're not numbered fieldnames. The issue I'm facing is that the fieldname data is user inputted and comes in random order so I'm trying to avoid indices.

[Stephen23]

@TastyPastry: fair enough. Then my solution does exactly what your question asks for: it concatenates the fields together.

Answer 2

[Thorsten]

help getfield

[TastyPastry]

See I did look at getfield(), but it seems like getfield() only works on single fields. I can see that it's able to get the same field for multiple structures in an array and retrieve multiple pieces of data for that field using indexing, but I can't see how it works for multiple fields.

Okay so I tried

getfield(myst,'field1','field2','field3'...);

but that gives me an error Attempt to reference field of non-structure array.

Answer 3

[Itsik Adiv]

You can use rmfield instead:

fieldsList = fieldnames(myst); myst = rmfields(myst, fieldnames{4:end});

Of course you can replace indices 4:end with any valid indexing logic you wish