get every item in mp_obj_list_t

[tejalbarnwal]

Hey, I am actually trying to store a value from micro python list into a temporary variable. How can I do it?
For example, let say I have a list pointer variable named input_list which was declared by mp_obj_list_t *input_list;
Now I wish to get the elements

for (size_t i = 0; i < input_list->len; i++) {
       float temp = (float)(input_list->items[i]);
}

But when i try to compiled the library, i get the following error:

error: pointer value used where a floating-point was expected
                 float temp = (float)(arg->input_list->items[size]);
                 ^~~~~

What is the correct way of storing value in temp variable, assuming that list is only made of integers or floating values?

[jimmo]

In MicroPython, every object is an mp_obj_t. This is an opaque type that encodes strings, integers, floats, or pointers-to-objects.

So you need to know what type you expect. In this case, float, so use mp_float_t temp = mp_obj_float_get(input_list->items[i])

Note the mp_float_t is important, because it depends on whether MicroPython is built for single or double precision float as to what the built-in type will be.

[tejalbarnwal]

Hey, thankyou so much for the quick response!

[tejalbarnwal]

Hey @jimmo , I encountered something while i was debugging my code.

The function that I defined in a user defined module is something along the following lines:

STATIC mp_obj_t py_tf_regression(uint n_args, const mp_obj_t *args, mp_map_t *kw_args)
{
    // did something with args[0] which works fine
   
    // reading the remaining three elements from the args

    // args[1] passed is an integer
    printf("printing arg 1: \t");
    printf("%d \n", (int)mp_obj_float_get(args[1]));

    // args[2] passed is an integer
    printf("printing arg 2: \t");
    printf("%d \n", (int) args[2]);

    //args[3] passed is list of integers
    mp_obj_list_t *arg_list = args[3];
    printf("read into arg list from arg 3 \n");
    for (size_t i = 0; i < arg_list->len; i++) {
        printf( "%d \t", (int) mp_obj_float_get(arg_list->items[i]));
    }
    printf("\n printed arg 3 \n");

}

Micropython code for calling the function:

a = 1
b = 1
input_list = [-8, -5, -8,  2]
output = tf.regression(net, a, b, input_list)

The output that I get:

printing arg 1:         2147483647 
printing arg 2:         3
read into arg list from arg 2 
2147483647      2147483647      2147483647
 printed arg 2

Doubts:

1. Why does argument 1 becomes a garbage value? How can I read it correctly?
2. I found that if I initialize b as 3, it printed as 5. I ran the function with other values of b as well, noticing the printed value of is always 2*b+1. What do you think what could be causing it?
3. Further, the list that I passed also contains garbage values. How can I correct this?

[stinos]

An integer doesn't have the same representation as a floating point value, so using mp_float_get on an mp_obj_t which is actually an integer isn't going to work, you should check what type the argument is using mp_obj_is_xxx, then use mp_obj_get_int (or similar), see obj.h.

[v923z]

You could also use

micropython/py/obj.c

Line 326 in b41aaaa

bool mp_obj_get_int_maybe(mp_const_obj_t arg, mp_int_t *value) {

, which makes sure that you read an integer, if the object is indeed an integer.

[tejalbarnwal]

It would be fantastic if this could be added in the documentation under internals here.
Its a really, really great documentation with a lot of examples.
Thankyou @v923z