CODEWITHSUNDEEP
ocaml
JKL
JKL

Reputation: 33

OCaml - difference between read_int and Scanf

I would like to know what is the difference between : 

let b = read_int();;

and 

let scan_int () = Scanf.scanf " %d" (fun x -> x)

?

Moreover why Scanf take a function as an entry ?

Upvotes: 3

Views: 4515

Answers (3)

octachron
octachron

Reputation: 18902

In brief, the Scanf module provides a way to read multiple inputs from an abstract source of characters. For instance, the following function reads two string inputs separated by an = character from a string:

let f string = Scanf.sscanf string "%s@=%s" (fun x y -> x, y )
assert (f "x=y" = "x","y")

The function argument gives a generic way to combine together the multiple inputs read by the scanf function into a single ouput.

It is therefore a generalization of the few read functions provided in Pervasives that can read only one input by line and only from a Pervasives in_channel.

Upvotes: 3

PatJ
PatJ

Reputation: 6144

read_int () reads a whole line, changes that whole line into an integer. The whole line must contain only an integer.

Scanf.scanf " %d" (fun x -> x) also reads a whole line, ditches the white spaces at the beginning of the line then reads an integer written in decimal notation. If there are characters after that integer, they are left in the buffer.

Here is a table:

| line in stdin | read_int | scanf " %d"     |
|---------------|----------|-----------------|
| "1234"        | 1234     | 1234            |
| " 1234"       | ERROR    | 1234            |
| "1234 foo"    | ERROR    | 1234 (foo left) |
| "0xff"        | 255      | 0 (xff left)    |

What's left in your buffer will usually hinder your next scanf. I would advise you not to use scanf on an unknown input.

About the identity function being needed, that is because scanf can have very complex format string, which will create complex data. How you want that data returned cannot be guessed. Tuples could be used, but OCaml doesn't support arbitrary flattened product in that specific case. Giving what we call a continuation does the job nicely and efficiently.

Upvotes: 12

Virgile
Virgile

Reputation: 10148

The main difference is that, as mentioned in the manual, read_int reads a whole line and expects that the entire line (except the final \n of course) consists in the integer, whereas with your definition of scan_int, you can have some non-digits characters after the integer.

The function argument of Scanf.scanf lets you perform an arbitrary transformation over the input(s).

Upvotes: 3

Related Questions