avoid inserting input as a string with ' '

I have what shoud be a short and simple question in Matlab: let's say I have a function with some inputs: either a filename i.e.:

function DataSet = extractData(filename,.......)
    [~,~,DataSet]=xlsread(filename)
end

or like a value I want to switch in order to let my function use different solutions, i.e.:

function result = operation(valuesSwitched,.......)
    switch valuesSwitched
        case one
            result=1;
        case two
            result=2
            ...
    end
end

and IN BOTH CASES I WISH TO AVOID calling the inputs as strings. Is there a possibility to do this, or in both cases inputs should be always represented as strings?

thanks

Gabriele

[matquest]

What do you mean by "avoid calling the inputs as strings"? In your first example, at some point you would have to use a string because you need to specify what file you want to use (unless you want to use a file pointer rather than a string filename). E.g.,

>>fname = 'filelocation/filename.xslx';
>>res = extractData(fname,[vars]);

In the second case, you could use an integer or float (matlab seems to be loosely-typed in that regard) for example.

Or do you want to prevent the user from calling the functions with strings as inputs? E.g., return an error if they try to call the function with a string.

[gabriele fadanelli]

Ok, thank you very much that was exactly wgat I wanted to know. So in case i want to call my input variable (valuesSwitched) in the second case one or two, I should ALWAYS use a string like 'one' or 'two'? No way to escape this? Thank you very much

[dpb]

"... you could use an integer or float (matlab seems to be loosely-typed in that regard)"

MATLAB default numeric variable is the double; everything is a double unless explicitly cast as something else. Integers of up to roughly 15 decimal digits are exactly representable as a double; there is no difference internally; the difference is all simply in the command window display doesn't display trailing fractional zeros if the value is an integral value although it is still stored as double.

>> n1=1; n2=1.0;      % store integer, floating integral value
>> whos n*            % see what they are internally
  Name      Size            Bytes  Class     Attributes

  n1        1x1                 8  double              
  n2        1x1                 8  double              

>> 

Both are doubles...this is the general rule in MATLAB--its weak typing is obtained by assuming everything is a double unless you declare it different--and the only way to declare it different is to cast it explicitly.

So, in fact, as opposed to being weakly or untyped, MATLAB is actually strongly typed -- it's just hidden from the user by the syntax adopted.

[Steven Lord]

and the only way to declare it different is to cast it explicitly.

This is a bit of a tangent, but do you consider the "0x" and "0b" prefixes and the u{8, 16, 32, 64} and s{8, 16, 32, 64} suffixes introduced in release R2019b to be explicit casts?

>> x = 0xff
x =
  uint8
   255

>> y = 0xbadbeef
y =
  uint32
   195935983
   
>> z = 0b10011001
z =
  uint8
   153
   
>> w = 0b1001s32
w =
  int32
   9

[dpb]

Wasn't aware of them yet, Steven...thanks for pointing out the feature.

I'll have to mull it over some before making a rash judgement statement. :)

LATER:

OK, I think I would say it is the same as and ergo, yes, an explicit cast.

There's still no statement-like feature such as

INTEGER [KIND=scalar_int] variable_name

for Fortran to specify the type of integer for the name first; it's all implied that unless the special rule of the type prefix or suffix is appended comes to play that any variable will be double with no work necessary.

To me, while I applaud its introduction, the above are "syntactic sugar" that sweetens the pot and eases the amount of typing otherwise required to reach the same goal so in my mind the idea presented above still holds.

It is somewhat of a sidebar overall, granted, but the poster to whom I responded seemed think there really were two variable types there in whether was or was not an integral value.

Answer 1

[gabriele fadanelli]

very thankful. Now everything is clear to me

[dpb]

If the above answered the Q?, please Accept an Answer so folks can know...

Answer 2

[dpb]

"So in case i want to call my input variable (valuesSwitched) in the second case one or two, I should ALWAYS use a string like 'one' or 'two'? No way to escape this?"

It all depends on the way you write the switch construct case statements. If you write

function result = operation(valuesSwitched,.......)
    switch valuesSwitched
        case one
            result=1;
        case two
            result=2
            ...
    end
end

then the two case comparisons are

valuesSwitched==one
valuesSwitched==two

but one and two are undefined variables in the function as shown. Instead you undoubtedly meant to write the functionality expressed as

function result = operation(valuesSwitched,.......)
    switch valuesSwitched
        case 1
            result=1;
        case 2
            result=2
            ...
    end
end

to compare against a numeric value of 1 or 2. If you really did want to use strings comparison instead, you'd have to write

function result = operation(valuesSwitched,.......)
    switch valuesSwitched
        case 'one'
            result=1;
        case 'two'
            result=2
            ...
    end
end

I don't understand at all the first lament of "IN BOTH CASES I WISH TO AVOID calling the inputs as strings." As pointed out above, the two functions are something entirely different -- in the first you need a file name at some point; may as well pass it to begin with unless there is some (small) set of possible files to select from and it's a choice of one of those you're after as a shortcut. If that's the case, if it's user input, use a uilistbox or uigetfile to just let the user select from the limited choices you'll give him.

[dpb]

ADDENDUM:

BTW,

function result = operation(valuesSwitched)
   switch valuesSwitched
     case {'one',1}
       result=1;
     case {'two',2}
       result=2
   end
end

would allow for either form of the argument to be passed...by make the case expression a cell array, the match is for any one of the cell content to satisfy the condition statement.