Question

How to improve a data structure for storing functions in arithmetic parser converting from infix to postfix notation?

At this moment I am using an array of char arrays:

char *funct[] = { "sin", "cos", "tan"... }
char text[] = "tan";

This impementation is a little bit confused and leads to the following comparisions, if we test char to be a function

if ( strcmp ( funct[0], text) == 0 ) || ( strcmp ( funct[1], "text ) == 0 ) || ( strcmp ( func[2], text) == 0 ))
{
  ... do something
}

( or to the for cycle version).

If there are a lot of functions (and a lot of comparisions), the index referencing leads to errors and it is not clear. There is also a necessity to change the index when we remove/add a new function….

How to improve such a structure so as it is easy to read, easy to maintain and easy to scale up?

I was thinking about enum

typedef enum
{
  Fsin=0,
  Fcos,
  Ftan
} TFunctions;

which results to

if ( strcmp ( funct[Fsin], text) == 0 ) || ( strcmp ( funct[Fcos], "text ) == 0 ) || ( strcmp ( func[Ftan], text) == 0 ))
{
...

but there may be a better solution…

Answer 1

You can use std::map.

enum functions
{
    sin,
    cos,
    tan
};

std::map<std::string, unsigned char> func_map;
func_map["sin"] = sin;
func_map["cos"] = cos;
func_map["tan"] = tan;

// then:
std::string text = "cos";

std::map<char*, unsigned char>::iterator it;
it = func_map.find(text);

if(it != func_map.end())
{
    // ELEMENT FOUND
    unsigned char func_id = it->second;
}
else
{
    // NOT FOUND
}