Type detection

This draft task needs a purpose, a description and some way to tell whether examples satisfy or do not satisfy it.

Type detection is a draft programming task. It is not yet considered ready to be promoted as a complete task, for reasons that should be found in its talk page.

Task

Show a function/procedure that processes a block of text by printing it.

The function takes one parameter (ideally) that describes the text.

Demonstrate by calling the function twice, each time passing in a different type.

This can be done with pattern matching, multi-methods, dynamic type detection, structure(s) with a tag, etc.

The objective is write a [e.g. library] function that processes text from multiple sources (such as a string/char *, socket, file, etc).

If not practical, show how the caller would coerce a type that can be passed to the library function.

ATS

Using garbage collection and datatype

<lang ats>#include "share/atspre_staload.hats"

datatype source_t = | source_t_string of string | source_t_FILEref of FILEref

extern fun print_text (source : source_t) : void

implement print_text (source) =

 case+ source of
 | source_t_string s => print! (s)
 | source_t_FILEref f =>
   let
     var c : int = fileref_getc (f)
   in
     while (0 <= c)
       begin
         fileref_putc (stdout_ref, c);
         c := fileref_getc (f)
       end
   end

implement main0 () =

 let
   val f = fileref_open_exn ("type_detection-postiats.dats",
                             file_mode_r)
 in
   print_text (source_t_string "This\nis a\ntext.\n");
   print_text (source_t_FILEref f)
 end</lang>

Output:

$ patscc -DATS_MEMALLOC_GCBDW type_detection-postiats.dats -lgc && ./a.out

This
is a
text.
#include "share/atspre_staload.hats"

datatype source_t =
| source_t_string of string
| source_t_FILEref of FILEref

extern fun
print_text (source : source_t) : void

implement
print_text (source) =
  case+ source of
  | source_t_string s => print! (s)
  | source_t_FILEref f =>
    let
      var c : int = fileref_getc (f)
    in
      while (0 <= c)
        begin
          fileref_putc (stdout_ref, c);
          c := fileref_getc (f)
        end
    end

implement
main0 () =
  let
    val f = fileref_open_exn ("type_detection-postiats.dats",
                              file_mode_r)
  in
    print_text (source_t_string "This\nis a\ntext.\n");
    print_text (source_t_FILEref f)
  end

Using a linear dataviewtype

In this implementation, print_text consumes its argument; the argument’s value cannot be used again. However, the object wrapped in the source_vt might be reusable.

<lang ats>#include "share/atspre_staload.hats"

dataviewtype source_vt = | source_vt_string of string | source_vt_FILEref of FILEref

extern fun print_text (source : source_vt) : void

implement print_text (source) =

 case+ source of
 | ~ source_vt_string s => print! (s)
 | ~ source_vt_FILEref f =>
   let
     var c : int = fileref_getc (f)
   in
     while (0 <= c)
       begin
         fileref_putc (stdout_ref, c);
         c := fileref_getc (f)
       end
   end

implement main0 () =

 let
   val f_opt = fileref_open_opt ("type_detection-postiats-2.dats",
                                 file_mode_r)
 in
   print_text (source_vt_string "This\nis a\ntext.\n");
   case+ f_opt of
   | ~ Some_vt f => print_text (source_vt_FILEref f)
   | ~ None_vt () => fprintln! (stderr_ref, "Failed to open the file.")
 end</lang>

Output:

$ patscc -DATS_MEMALLOC_LIBC type_detection-postiats-2.dats && ./a.out

This
is a
text.
#include "share/atspre_staload.hats"

dataviewtype source_vt =
| source_vt_string of string
| source_vt_FILEref of FILEref

extern fun
print_text (source : source_vt) : void

implement
print_text (source) =
  case+ source of
  | ~ source_vt_string s => print! (s)
  | ~ source_vt_FILEref f =>
    let
      var c : int = fileref_getc (f)
    in
      while (0 <= c)
        begin
          fileref_putc (stdout_ref, c);
          c := fileref_getc (f)
        end
    end

implement
main0 () =
  let
    val f_opt = fileref_open_opt ("type_detection-postiats-2.dats",
                                  file_mode_r)
  in
    print_text (source_vt_string "This\nis a\ntext.\n");
    case+ f_opt of
    | ~ Some_vt f => print_text (source_vt_FILEref f)
    | ~ None_vt () => fprintln! (stderr_ref, "Failed to open the file.")
  end

AWK

<lang AWK>

1. syntax: TAWK -f TYPE_DETECTION.AWK
2. uses Thompson Automation's TAWK 5.0c

BEGIN {

   arr[0] = 0
   print(typeof(arr))
   print(typeof(0.))
   print(typeof(0))
   print(typeof(/0/))
   print(typeof("0"))
   print(typeof(x))
   print(typeof(addressof("x")))
   print(typeof(fopen("x","r")))
   exit(0)

} </lang>

Output:

array
float
int
regular_expression
string
uninitialized
address
fileid

C

The closest C comes to meeting this task, short of building it into the compiler or accessing memory segments via pointers, which is not guaranteed to be portable, is the ctype.h header file. It is part of the C Standard Library and provides 11 methods for detecting the type of a character, out of which the following 7 called in the wrapper function below can be called to be unique. The function accepts a string, but it actually checks the first character. An if ladder is used instead of if-else so that all function calls which return a non-zero value for the character are satisfied and the information is printed. <lang C>

1. include<stdio.h>
2. include<ctype.h>

void typeDetector(char* str){ if(isalnum(str[0])!=0) printf("\n%c is alphanumeric",str[0]); if(isalpha(str[0])!=0) printf("\n%c is alphabetic",str[0]); if(iscntrl(str[0])!=0) printf("\n%c is a control character",str[0]); if(isdigit(str[0])!=0) printf("\n%c is a digit",str[0]); if(isprint(str[0])!=0) printf("\n%c is printable",str[0]); if(ispunct(str[0])!=0) printf("\n%c is a punctuation character",str[0]); if(isxdigit(str[0])!=0) printf("\n%c is a hexadecimal digit",str[0]); }

int main(int argC, char* argV[]) { int i;

if(argC==1) printf("Usage : %s <followed by ASCII characters>"); else{ for(i=1;i<argC;i++) typeDetector(argV[i]); } return 0; } </lang> Output, shown for multiple inputs, as well as single ones:

C:\rosettaCode>typeDetector.exe s 3 $ f ! as 3

s is alphanumeric
s is alphabetic
s is printable
3 is alphanumeric
3 is a digit
3 is printable
3 is a hexadecimal digit
$ is printable
$ is a punctuation character
f is alphanumeric
f is alphabetic
f is printable
f is a hexadecimal digit
! is printable
! is a punctuation character
a is alphanumeric
a is alphabetic
a is printable
a is a hexadecimal digit
3 is alphanumeric
3 is a digit
3 is printable
3 is a hexadecimal digit

C:\rosettaCode>typeDetector.exe a

a is alphanumeric
a is alphabetic
a is printable
a is a hexadecimal digit
C:\rosettaCode>typeDetector.exe $

$ is printable
$ is a punctuation character
C:\rosettaCode>typeDetector.exe 3

3 is alphanumeric
3 is a digit
3 is printable
3 is a hexadecimal digit

Dispatch by tagged union

<lang c>#include <stdio.h>

1. include <stdbool.h>
2. include <assert.h>

typedef enum {

 STRING,
 INPUT_FILE

} tag_t;

typedef struct {

 tag_t tag;
 union {
   char *string;
   FILE *input_file;
 } value;

} source_t;

void print_text (source_t source) {

 switch (source.tag)
   {
   case STRING:
     fputs (source.value.string, stdout);
     break;
   case INPUT_FILE:
     {
       int c;
       c = getc (source.value.input_file);
       while (c != EOF)
         {
           putc (c, stdout);
           c = getc (source.value.input_file);
         }
     }
     break;
   default:
     assert (false);
   }

}

int main () {

 source_t source;

 source.tag = STRING;
 source.value.string = "This\nis a\ntext.\n";
 print_text (source);

 source.tag = INPUT_FILE;
 source.value.input_file = fopen ("type_detection-c.c", "r");
 print_text (source);

}</lang>

Output:

$ cc type_detection-c.c && ./a.out

This
is a
text.
#include <stdio.h>
#include <stdbool.h>
#include <assert.h>

typedef enum {
  STRING,
  INPUT_FILE
} tag_t;

typedef struct {
  tag_t tag;
  union {
    char *string;
    FILE *input_file;
  } value;
} source_t;

void
print_text (source_t source)
{
  switch (source.tag)
    {
    case STRING:
      fputs (source.value.string, stdout);
      break;
    case INPUT_FILE:
      {
        int c;
        c = getc (source.value.input_file);
        while (c != EOF)
          {
            putc (c, stdout);
            c = getc (source.value.input_file);
          }
      }
      break;
    default:
      assert (false);
    }
}

int
main ()
{
  source_t source;

  source.tag = STRING;
  source.value.string = "This\nis a\ntext.\n";
  print_text (source);

  source.tag = INPUT_FILE;
  source.value.input_file = fopen ("type_detection-c.c", "r");
  print_text (source);
}

C#

<lang csharp>using System;

namespace TypeDetection {

   class C { }
   struct S { }
   enum E {
       NONE,
   }

   class Program {
       static void ShowType<T>(T t) {
           Console.WriteLine("The type of '{0}' is {1}", t, t.GetType());
       }

       static void Main() {
           ShowType(5);
           ShowType(7.5);
           ShowType('d');
           ShowType(true);
           ShowType("Rosetta");
           ShowType(new C());
           ShowType(new S());
           ShowType(E.NONE);
           ShowType(new int[] { 1, 2, 3 });
       }
   }

}</lang>

Output:

The type of '5' is System.Int32
The type of '7.5' is System.Double
The type of 'd' is System.Char
The type of 'True' is System.Boolean
The type of 'Rosetta' is System.String
The type of 'TypeDetection.C' is TypeDetection.C
The type of 'TypeDetection.S' is TypeDetection.S
The type of 'NONE' is TypeDetection.E
The type of 'System.Int32[]' is System.Int32[]

C++

Translation of: D

<lang cpp>#include <iostream>

template <typename T> auto typeString(const T&) {

   return typeid(T).name();

}

class C {}; struct S {};

int main() {

   std::cout << typeString(1) << '\n';
   std::cout << typeString(1L) << '\n';
   std::cout << typeString(1.0f) << '\n';
   std::cout << typeString(1.0) << '\n';
   std::cout << typeString('c') << '\n';
   std::cout << typeString("string") << '\n';
   std::cout << typeString(C{}) << '\n';
   std::cout << typeString(S{}) << '\n';
   std::cout << typeString(nullptr) << '\n';

}</lang>

Output:

int
long
float
double
char
char [7]
class C
struct S
std::nullptr_t

Crystal

<lang Ruby>def print_type(x)

 puts "Compile-time type of #{x} is #{typeof(x)}"
 puts "    Actual runtime type is #{x.class}" if x.class != typeof(x)

end

print_type 123 print_type 123.45 print_type rand < 0.5 ? "1" : 0 print_type rand < 1.5 print_type nil print_type 'c' print_type "str" print_type [1,2] print_type({ 2, "two" }) print_type({a: 1, b: 2}) print_type ->(x : Int32){ x+2 > 0 } </lang>

Output:

Compile-time type of 123 is Int32
Compile-time type of 123.45 is Float64
Compile-time type of 0 is (Int32 | String)
    Actual runtime type is Int32
Compile-time type of true is Bool
Compile-time type of  is Nil
Compile-time type of c is Char
Compile-time type of str is String
Compile-time type of [1, 2] is Array(Int32)
Compile-time type of {2, "two"} is Tuple(Int32, String)
Compile-time type of {a: 1, b: 2} is NamedTuple(a: Int32, b: Int32)
Compile-time type of #<Proc(Int32, Bool):0x5645695ab9f0> is Proc(Int32, Bool)

D

<lang D>import std.stdio;

auto typeString(T)(T _) {

   return T.stringof;

}

class C {} struct S {}

void main() {

   writeln(typeString(1));
   writeln(typeString(1L));
   writeln(typeString(1.0f));
   writeln(typeString(1.0));
   writeln(typeString('c'));
   writeln(typeString("string"));
   writeln(typeString(new C()));
   writeln(typeString(S()));
   writeln(typeString(null));

}</lang>

Output:

int
long
float
double
char
string
C
S
typeof(null)

F#

<lang fsharp> printfn "%A" (3.14.GetType()) let inline fN g=g.GetType()|>string printfn "%s" (fN "Nigel") printfn "%s" (fN 23) </lang>

Output:

System.Double
System.String
System.Int32

Factor

Using dynamic dispatch: <lang>USING: arrays formatting io kernel math prettyprint sequences strings ; IN: rosetta-code.type-detection

GENERIC: myprint ( object -- )

M: object myprint drop "I don't know how to print this." print ; M: string myprint "I'm a string: \"%s\"\n" printf ; M: fixnum myprint "I'm a fixnum: " write . ; M: array myprint "I'm an array: { " write

   [ pprint bl ] each "}" print ;
   

"Hello world." myprint { 1 2 3 4 5 } myprint 123 myprint 3.1415 myprint</lang>

Output:

I'm a string: "Hello world."
I'm an array: { 1 2 3 4 5 }
I'm a fixnum: 123
I don't know how to print this.

Fortran

Fortran 2008 and later can do this with general types, using the unlimited polymorphic class class(*).

Polymorphism involving types derived from a base type can be done in Fortran 2003 and later.

There is also, in modern Fortran, an overload mechanism that can be used to give procedures that take different types the same name.

Legacy versions of Fortran have ways to store data of one type in a variable declared as some other type, and this mechanism could be employed.

Unlimited procedure polymorphism

Below I use ‘class(*)’. The ‘print_text’ subroutine is used to print a string, an array of strings, and an input file.

<lang fortran>program input_type_detection_demo

 implicit none

 type text_block_t
    character(len = 10000), allocatable :: lines(:)
 end type text_block_t

 type(text_block_t) :: text_block
 integer :: i

 call print_text ('Print me.')

 allocate (text_block%lines(1:10))
 do i = 1, 10
    write (text_block%lines(i), '("i = ", I0)') i
 end do
 call print_text (text_block)

 open (100, file = 'type_detection-fortran.f90', action = 'read')
 call print_text (100)
 close (100)

contains

 subroutine print_text (source)
   class(*), intent(in) :: source

   select type (source)

   type is (character(len = *))
      ! Print a single character string.
      write (*, '(A)') source

   class is (text_block_t)
      ! Print an array of lines.
      block
        integer :: i
        do i = lbound (source%lines, 1), ubound (source%lines, 1)
           write (*, '(A)') trim (source%lines(i))
        end do
      end block

   type is (integer)
      ! Print a file.
      block
        character(len = 10000) :: line_buffer
        integer :: stat
        read (source, '(A)', iostat = stat) line_buffer
        do while (stat == 0)
           write (*, '(A)') trim (line_buffer)
           read (source, '(A)', iostat = stat) line_buffer
        end do
      end block

   class default
      ! There is no handler for the type.
      error stop

   end select
 end subroutine print_text

end program input_type_detection_demo</lang>

Output:

$ gfortran type_detection-fortran.f90 && ./a.out

Print me.
i = 1
i = 2
i = 3
i = 4
i = 5
i = 6
i = 7
i = 8
i = 9
i = 10
program input_type_detection_demo
  implicit none

  type text_block_t
     character(len = 10000), allocatable :: lines(:)
  end type text_block_t

  type(text_block_t) :: text_block
  integer :: i

  call print_text ('Print me.')

  allocate (text_block%lines(1:10))
  do i = 1, 10
     write (text_block%lines(i), '("i = ", I0)') i
  end do
  call print_text (text_block)

  open (100, file = 'type_detection-fortran.f90', action = 'read')
  call print_text (100)
  close (100)

contains

  subroutine print_text (source)
    class(*), intent(in) :: source

    select type (source)

    type is (character(len = *))
       ! Print a single character string.
       write (*, '(A)') source

    class is (text_block_t)
       ! Print an array of lines.
       block
         integer :: i
         do i = lbound (source%lines, 1), ubound (source%lines, 1)
            write (*, '(A)') trim (source%lines(i))
         end do
       end block

    type is (integer)
       ! Print a file.
       block
         character(len = 10000) :: line_buffer
         integer :: stat
         read (source, '(A)', iostat = stat) line_buffer
         do while (stat == 0)
            write (*, '(A)') trim (line_buffer)
            read (source, '(A)', iostat = stat) line_buffer
         end do
       end block

    class default
       ! There is no handler for the type.
       error stop

    end select
  end subroutine print_text

end program input_type_detection_demo

Generic procedures (overloading)

In the following, the name print_text refers to two distinct subroutines. Which of the two is called depends on the type of the argument in the call; the decision is made at compile time.

(To use this mechanism in a library, one would want to use modules. Nevertheless, the program below illustrates the principle more simply.)

<lang fortran>program type_detection_demo

 implicit none

 type text_block_t
    character(len = 10000), allocatable :: lines(:)
 end type text_block_t

 interface print_text
    procedure print_text_text_block_t
    procedure print_text_file_unit
 end interface print_text

 type(text_block_t) :: text_block
 integer :: i

 allocate (text_block%lines(1:10))
 do i = 1, 10
    write (text_block%lines(i), '("i = ", I0)') i
 end do
 call print_text (text_block)

 open (100, file = 'type_detection-fortran-2.f90', action = 'read')
 call print_text (100)
 close (100)

contains

 subroutine print_text_text_block_t (source)
   class(text_block_t), intent(in) :: source

   integer :: i

   do i = lbound (source%lines, 1), ubound (source%lines, 1)
      write (*, '(A)') trim (source%lines(i))
   end do
 end subroutine print_text_text_block_t

 subroutine print_text_file_unit (source)
   integer, intent(in) :: source

   character(len = 10000) :: line_buffer
   integer :: stat

   read (source, '(A)', iostat = stat) line_buffer
   do while (stat == 0)
      write (*, '(A)') trim (line_buffer)
      read (source, '(A)', iostat = stat) line_buffer
   end do
 end subroutine print_text_file_unit

end program type_detection_demo</lang>

Output:

$ gfortran type_detection-fortran-2.f90 && ./a.out

i = 1
i = 2
i = 3
i = 4
i = 5
i = 6
i = 7
i = 8
i = 9
i = 10
program type_detection_demo

  implicit none

  type text_block_t
     character(len = 10000), allocatable :: lines(:)
  end type text_block_t

  interface print_text
     procedure print_text_text_block_t
     procedure print_text_file_unit
  end interface print_text

  type(text_block_t) :: text_block
  integer :: i

  allocate (text_block%lines(1:10))
  do i = 1, 10
     write (text_block%lines(i), '("i = ", I0)') i
  end do
  call print_text (text_block)

  open (100, file = 'type_detection-fortran-2.f90', action = 'read')
  call print_text (100)
  close (100)

contains

  subroutine print_text_text_block_t (source)
    class(text_block_t), intent(in) :: source

    integer :: i

    do i = lbound (source%lines, 1), ubound (source%lines, 1)
       write (*, '(A)') trim (source%lines(i))
    end do
  end subroutine print_text_text_block_t

  subroutine print_text_file_unit (source)
    integer, intent(in) :: source

    character(len = 10000) :: line_buffer
    integer :: stat

    read (source, '(A)', iostat = stat) line_buffer
    do while (stat == 0)
       write (*, '(A)') trim (line_buffer)
       read (source, '(A)', iostat = stat) line_buffer
    end do
  end subroutine print_text_file_unit

end program type_detection_demo

Go

Note that Go doesn't really have a character type. A single quoted character (such as 'd') is by default a rune (or 32 bit integer) literal representing its Unicode code-point. <lang go>package main

import "fmt"

type any = interface{}

func showType(a any) {

   switch a.(type) {
   case rune:
       fmt.Printf("The type of '%c' is %T\n", a, a)
   default:
       fmt.Printf("The type of '%v' is %T\n", a, a)
   }

}

func main() {

   values := []any{5, 7.5, 2 + 3i, 'd', true, "Rosetta"}
   for _, value := range values {
       showType(value)
   }

}</lang>

Output:

The type of '5' is int
The type of '7.5' is float64
The type of '(2+3i)' is complex128
The type of 'd' is int32
The type of 'true' is bool
The type of 'Rosetta' is string

Goaldi

Translation of: Icon

<lang goaldi>procedure main() {

 print_text("This\nis a\ntext.\n")
 print_text(file("type_detection-goaldi.gd"))

}

procedure print_text(source) {

 case type(source) of {
   string : writes(source)
   file : while write(read(source))
 }

}</lang>

Output:

$ goaldi type_detection-goaldi.gd

This
is a
text.
procedure main() {
  print_text("This\nis a\ntext.\n")
  print_text(file("type_detection-goaldi.gd"))
}

procedure print_text(source) {
  case type(source) of {
    string : writes(source)
    file : while write(read(source))
  }
}

Icon

This should work with the Unicon compiler, but I currently have only the Arizona Icon compiler to test it with.

See also ObjectIcon and Goaldi.

<lang icon>procedure main()

 print_text("This\nis\na text.\n")
 print_text(open("type_detection-icon.icn"))

end

procedure print_text(source)

 case type(source) of {
   "string" : writes(source)
   "file" : while write(read(source))
 }

end</lang>

Output:

$ icont -u -s type_detection-icon.icn && ./type_detection-icon

This
is
a text.
procedure main()
  print_text("This\nis\na text.\n")
  print_text(open("type_detection-icon.icn"))
end

procedure print_text(source)
  case type(source) of {
    "string" : writes(source)
    "file" : while write(read(source))
  }
end

J

Presumably this satisfies the task requirements...

<lang J> echo 'one' one

  echo 1

1</lang>

Java

Translation of: Kotlin

<lang Java>public class TypeDetection {

   private static void showType(Object a) {
       if (a instanceof Integer) {
           System.out.printf("'%s' is an integer\n", a);
       } else if (a instanceof Double) {
           System.out.printf("'%s' is a double\n", a);
       } else if (a instanceof Character) {
           System.out.printf("'%s' is a character\n", a);
       } else {
           System.out.printf("'%s' is some other type\n", a);
       }
   }

   public static void main(String[] args) {
       showType(5);
       showType(7.5);
       showType('d');
       showType(true);
   }

}</lang>

Output:

'5' is an integer
'7.5' is a double
'd' is a character
'true' is some other type

JavaScript

[1]

console.log(typeof('foo')); // Returns string
console.log(typeof(12345)); // Returns number

jq

In jq, the function that returns the JSON type of a JSON entity is type/0. It returns "object", "array", "boolean", "string", or "number".

Given arbitrary UTF-8 input, it could be used like so:

       try type catch "invalid JSON"

Given some text of unknown type, a "typeof" function could be written to determine whether the string could be interpreted as a JSON document, and if so, what JSON type it would have, as follows:

   def typeof:
     try (fromjson | type) catch "string" ;

Here is an illustrative transcript from an interactive session showing input and output on alternate lines:

$ jq -R 'try (fromjson | type) catch "string"'
abc
"string"
{"a":1,"b":2}
"object"
[1,"a"]
"array"

Julia

In Julia, the function that returns the type of an object is the typeof function, and the function isa tests whether an object is of that type. <lang Julia>

julia> a = 1 1

julia> typeof(a) Int32

julia> b = 1.0 1.0

julia> typeof(b) Float64

julia> 1.0 isa Number true

julia> 1.0 isa Int false

julia> 1 isa Int true

julia> typeof("hello") String

julia> typeof(typeof("hello")) DataType

julia> typeof(Set([1,3,4])) Set{Int64}

julia> 1 isa String false

julia> "1" isa Number false

julia> "1" isa String true

julia> isa(1.0,Float32) false

julia> isa(1.0,Float64) true

</lang>

OASYS Assembler

<lang oasys_oaa>

The following method checks if a global variable or property is an

object type. Does not work with locals and arguments.

[&OBJ#,^]

 ,^<,^<<    ; Remember old value
 ,^<*>      ; Create new object
 ,^<<DES    ; Destroy the object
 ,^<<EX     ; Check if variable has been cleared
 />1RF      ; It is clear
 :>0RF      ; It is not clear

</lang>

Kotlin

<lang scala>// version 1.0.6 fun showType(a: Any) = when (a) {

       is Int    -> println("'$a' is an integer")
       is Double -> println("'$a' is a double")
       is Char   -> println("'$a' is a character")
       else      -> println("'$a' is some other type")
   }

fun main(args: Array<String>) {

   showType(5)
   showType(7.5)
   showType('d')
   showType(true)

}</lang>

Output:

'5' is an integer
'7.5' is a double
'd' is a character
'true' is some other type

Lua

<lang lua>function writeValue(v)

   local t = type(v)
   if t == "number" then
       io.write(v)
   elseif t == "string" then
       io.write("`" .. v .. "`")
   elseif t == "table" then
       local c = 0
       io.write("{")
       for k,v in pairs(v) do
           if c > 0 then
               io.write(", ")
           end
           writeValue(k)
           io.write(" => ")
           writeValue(v)
           c = c + 1
       end
       io.write("}")
   elseif t == "function" then
       io.write("`" .. tostring(v) .. "`")
   else
       io.write("Unhandled type: " .. t)
   end

end

function printType(v)

   io.write("The value ")
   writeValue(v)
   print(" is of type " .. type(v))

end

function main()

   printType(42)
   printType(3.14)
   printType("hello world")
   printType({1, 2, 3, 4, 5})
   printType(main)

end

main()</lang>

Output:

The value 42 is of type number
The value 3.14 is of type number
The value `hello world` is of type string
The value {1 => 1, 2 => 2, 3 => 3, 4 => 4, 5 => 5} is of type table
The value `function: 00C6C4A8` is of type function

Nim

As Nim is a statically typed language, there is no way to detect the type at runtime. But it is possible to write a procedure which writes text in ways depending on the type of a parameter.

Using a generic procedure

All is done at compile time. The generated code depends on the type of the parameter. We accept here any type, but only provide code to display text from a string and from a file and we emit an error for the other types. <lang Nim>proc writeText[T](source: T) =

 when T is string:
   echo source
 elif T is File:
   echo source.readAll()
 else:
   echo "Unable to write text for type “", T, "”."

writeText("Hello world!") writeText(stdin) writeText(3) # Emit an error.</lang>

Using variant objects

We define a type “Source” which contains a discriminator (tag) which is used to define branches. The code to display the text contains a test on the tag. The only types allowed are those for which a tag value exists. Here, we defined two possible values: “kString” and “kFile”. The “type detection” is done at runtime, but these are not actual types but tag values. <lang Nim>type Kind = enum kString, kFile

type Source = object

 case kind: Kind
 of kString: str: string
 of kFile: file: File

proc writeText(source: Source) =

 case source.kind
 of kString: echo source.str
 of kFile: echo source.file.readAll()

let s1 = Source(kind: kString, str: "Hello world!") let s2 = Source(kind: kFile, file: stdin)

s1.writeText() s2.writeText()</lang>

ObjectIcon

Translation of: Icon

<lang objecticon>import io

procedure main()

 print_text("This\nis\na text.\n")
 print_text(open("type_detection-oi.icn"))

end

procedure print_text(source)

 case type(source) of {
   "string" : writes(source)
   "object" : while write(source.read())
 }

end</lang>

Output:

$ oit -s type_detection-oi.icn && ./type_detection-oi

This
is
a text.
import io

procedure main()
  print_text("This\nis\na text.\n")
  print_text(open("type_detection-oi.icn"))
end

procedure print_text(source)
  case type(source) of {
    "string" : writes(source)
    "object" : while write(source.read())
  }
end

Pascal

Dispatch by variant record

Pascal has a plethora of dialects, and so I have tried to be as close to Algorithms + Data Structures = Programs (Wirth) style as I could figure out how, when using the Free Pascal Compiler.

<lang pascal>program typedetectiondemo (input, output); type

  sourcetype = record case kind : (builtintext, filetext) of
                 builtintext : (i : integer);
                 filetext    : (f : file of char);
               end;

var

  source : sourcetype;
  input  : file of char;
  c      : char;

procedure printtext (source : sourcetype); begin

  case source.kind of
    builtintext : case source.i of
                   1 : writeln ('This is text 1.');
                   2 : writeln ('This is text 2.')
                  end;
    filetext    : while not eof (source.f) do
                  begin
                     read (source.f, c);
                     write (c)
                  end
  end

end;

begin

  assign (input, 'type_detection-pascal.pas');
  reset (input);
  with source do
     begin
        kind := builtintext;
        i := 1;
        printtext (source);
        i := 2;
        printtext (source);
        kind := filetext;
        f := input;
        printtext (source)
     end

end.</lang>

Output:

$ fpc -Miso type_detection-pascal.pas && ./type_detection-pascal

This is text 1.
This is text 2.
program typedetectiondemo (input, output);
type
   sourcetype = record case kind : (builtintext, filetext) of
                  builtintext : (i : integer);
                  filetext    : (f : file of char);
                end;

var
   source : sourcetype;
   input  : file of char;
   c      : char;

procedure printtext (source : sourcetype);
begin
   case source.kind of
     builtintext : case source.i of
                    1 : writeln ('This is text 1.');
                    2 : writeln ('This is text 2.')
                   end;
     filetext    : while not eof (source.f) do
                   begin
                      read (source.f, c);
                      write (c)
                   end
   end
end;

begin
   assign (input, 'type_detection-pascal.pas');
   reset (input);
   with source do
      begin
         kind := builtintext;
         i := 1;
         printtext (source);
         i := 2;
         printtext (source);
         kind := filetext;
         f := input;
         printtext (source)
      end
end.

Perl

The function ref takes a reference to a variable, via '\', and returns the type. Some of the more common are shown here. In the cases where the value in question is already a reference ($regex and $subref) the '\' is not used. <lang perl>$scalar = 1; @array = (1, 2); %hash = ('a' => 1); $regex = qr/foo.*bar/; $reference = \%hash; sub greet { print "Hello world!" }; $subref = \&greet;

$fmt = "%-11s is type: %s\n"; printf $fmt, '$scalar', ref(\$scalar); printf $fmt, '@array', ref(\@array); printf $fmt, '%hash', ref(\%hash); printf $fmt, '$regex', ref( $regex); printf $fmt, '$reference', ref(\$reference); printf $fmt, '$subref', ref( $subref);</lang>

Output:

$scalar     is type:  SCALAR
@array      is type:  ARRAY
%hash       is type:  HASH
$regex      is type:  Regexp
$reference  is type:  REF
$subref     is type:  CODE

Phix

Library: Phix/basics

Phix builtin type tests are: integer(), atom(), string(), sequence(), and object(). The latter returns true unless arg is unassigned, also notice that showtype never even attempts to set t to "object", since it is guaranteed to be one of the other four.

procedure showtype(object o)
    string t = iff(atom(o)?iff(integer(o)?"integer":"atom")
                          :iff(string(o)?"string":"sequence"))
    ?{t,o}
end procedure
 
showtype(5)
showtype(7.5)
showtype("string")
showtype({5,7.5,"string"})

Output:

{"integer",5}
{"atom",7.5}
{"string","string"}
{"sequence",{5,7.5,"string"}}

PicoLisp

PicoLisp have only three base data types.

: (num? 123)
-> 123
: (num? (1 2 3))
-> NIL
: (sym? 'a)
-> T
: (sym? 123)
-> NIL
: (lst? NIL)
-> T
: (lst? (1 . 2))
-> T
: (lst? (1 2 3))
-> T

PHP

[2]

echo gettype('foo'); // Returns string
echo gettype(12345); // Returns integer

Specific tester functions

• is_array()

PowerShell

In PowerShell everything is an object and all objects have the GetType() method: <lang PowerShell> [string]$str = "123" $str.GetType() </lang>

Output:

IsPublic IsSerial Name                                     BaseType     
-------- -------- ----                                     --------                                                                                                                  
True     True     String                                   System.Object

<lang PowerShell> [int]$int = $str -as [int] $int.GetType() </lang>

Output:

IsPublic IsSerial Name                                     BaseType        
-------- -------- ----                                     --------                                                                                                                  
True     True     Int32                                    System.ValueType

Python

Built-in function type()

>>> type('foo')
<class 'str'>
>>> type(12345)
<class 'int'>

Testing types

>>> type('foo') is str
True
>>> type(123.0) is not int
True
>>> type([]) is list
True
>>> type({}) is dict
True

Specific tester functions

• isinstance()

Racket

Hopefully you can see how to extend the code to add all sorts of other types. If I did this, I’d swamp the task page. A good list of types supported/provided by Racket can be found in the Typed Racket reference: http://docs.racket-lang.org/ts-reference/type-ref.html

<lang racket>#lang racket

(require racket/undefined)

(define fooer<%> (interface ())) (define foo% (class* object% (fooer<%>)

              (super-new)))

(struct my-tree (l v r))

-----------------------------------------------------------------------------

(define (n.t f)

 (list f (regexp-replace #rx"\\?" (symbol->string (object-name f)) "")))

listed in the order (as close as) shown in

http://docs.racket-lang.org/guide/datatypes.html (section numbers next to

some entries)

(define type-tests.names

 `(,@(map n.t
          (list boolean? immutable? ; 3.1
                ))
   ;; the famous scheme numerical tower
   ,@(map n.t ; 3.2
          (list number? complex? real? rational? integer? exact-integer?
                exact-nonnegative-integer? exact-positive-integer?
                inexact-real? fixnum? flonum? double-flonum? single-flonum?
                zero? positive? negative? odd? even? exact? inexact?))
   ,@(map n.t
          (list char? ; 3.3 --- there are also char-alphabetic? etc -- but they're not
                      ;         types as such
                string? ; 3.4
                byte? bytes? ; 3.5
                symbol? ; 3.6
                keyword? ; 3.7
                pair? null? list? ; 3.8
                vector? ; 3.9
                hash? hash-equal? hash-eqv? hash-eq? hash-weak? ; 3.10
                box? ; 3.11
                void? ; 3.12
                ))
   ,(list (λ (v) (eq? v undefined)) "undefined") ; 3. 12
   ;; now we move to http://docs.racket-lang.org/reference/data.html
   ;; for section numbering
   ,@(map n.t
          (list
           regexp? pregexp? byte-regexp? byte-pregexp? ; 4.7
           stream? sequence? ; 4.14
           dict? ; 4.15
           set-equal? set-eqv? set-eq? set? set-mutable? set-weak? ; 4.16
           continuation? procedure? ; 4.17            
           ))
   ;; class/interface testing
   ,(list (λ (v) (is-a? v object%)) "object%")
   ,(list (λ (v) (is-a? v foo%)) "foo%")
   ,(list (λ (v) (is-a? v fooer<%>)) "fooer<%>")

   ;; more types from reference (sections are top-level, mostly)
   ,@(map n.t
          (list
           syntax? ; 3.
           my-tree? ; 5.
           exn? exn:fail? exn:fail:filesystem? ; 10.2
           promise? ; 10.3
           ))

   ;; there's all sorts of other types to test!
   ))

(define (->type-names v)

 (let ((rv (for/list ((t.n (in-list type-tests.names))
                      #:when (with-handlers
                                 ((exn? (λ (x) #f)))
                               ((car t.n) v))) (cadr t.n))))
   (if (null? rv) (list "UNKNOWN") rv)))

(module+ test

 (require xml/xml)

 (define test-values
   (list 3.+4.i 3+4i (- pi) pi 0. 0 -0.5 0.5 -1/3 1/3
         -12345678909876543210123456789 12345678909876543210123456788 -132 133
         #\t #\null
         "" "monkeys" "\u03BB"
         -1 255 256
         #"" #"nibble"
         'hello '||
         '#:woo
         '() '(1 . 2) '(3) '(5 6)

         #() #(1) #("foo" 2 'bar)

         (make-hash)
         (make-hasheq)
         (make-hasheqv)
         (hash)
         (hasheq)
         (hasheqv)
         (make-weak-hash)
         (make-weak-hasheq)
         (make-weak-hasheqv)     
         (make-immutable-hash)
         (make-immutable-hasheq)
         (make-immutable-hasheqv)

         (box "x")
         (void)
         undefined
         #rx".*" #px"3?" #rx#"t.m" #px#".i."
         (in-vector #(1 2 3)) (stream 1 2 3)
         #hash((a . "apple")) #("apple" "binana") '("apple" "binana")
          '((a . "apple") (b . "binana"))
          (set 1 2 3) (seteq 1 2 3) (seteqv 1 2 3)
          (mutable-set 1 2 3) (mutable-seteq 1 2 3) (mutable-seteqv 1 2 3)
          (weak-set 1 2 3) (weak-seteq 1 2 3) (weak-seteqv 1 2 3)

          + (λ (x) #t) (call/cc (λ (k) k))

          (new object%) (new foo%)

          #'(xxy zzy)
          (my-tree (my-tree #f 1 #f) 2 #f)
          (with-handlers ((exn? values)) (error 'aargh!))
          (with-handlers ((exn? values))
            (file->string "/tmp/there-is-no-way-this-file-exists---surely?"))
          (delay 3)
         ))

 ;; tempted to print a cross-reference table, but that would be too wide for RC (maybe)
 (write-xexpr #:insert-newlines? #f
  `(table (thead (tr (th "Value [~s]") (th "->type-name")))
          "\n"
          (tbody ,@(map (λ (v) `(tr "\n" (td ,(~s v))
                                    (td ,(string-join (->type-names v) ", "))))
                        test-values)))))

</lang>

Output:

The following table is generated:

<thead></thead> <tbody></tbody>

Value [~s]	->type-name
3.0+4.0i	number, complex, inexact
3+4i	number, complex, exact
-3.141592653589793	number, complex, real, rational, inexact-real, flonum, double-flonum, negative, inexact
3.141592653589793	number, complex, real, rational, inexact-real, flonum, double-flonum, positive, inexact
0.0	number, complex, real, rational, integer, inexact-real, flonum, double-flonum, zero, even, inexact
0	number, complex, real, rational, integer, exact-integer, exact-nonnegative-integer, fixnum, zero, even, exact, byte, sequence
-0.5	number, complex, real, rational, inexact-real, flonum, double-flonum, negative, inexact
0.5	number, complex, real, rational, inexact-real, flonum, double-flonum, positive, inexact
-1/3	number, complex, real, rational, negative, exact
1/3	number, complex, real, rational, positive, exact
-12345678909876543210123456789	number, complex, real, rational, integer, exact-integer, negative, odd, exact
12345678909876543210123456788	number, complex, real, rational, integer, exact-integer, exact-nonnegative-integer, exact-positive-integer, positive, even, exact, sequence
-132	number, complex, real, rational, integer, exact-integer, fixnum, negative, even, exact
133	number, complex, real, rational, integer, exact-integer, exact-nonnegative-integer, exact-positive-integer, fixnum, positive, odd, exact, byte, sequence
#\t	char
#\nul	char
""	immutable, string, sequence
"monkeys"	immutable, string, sequence
"λ"	immutable, string, sequence
-1	number, complex, real, rational, integer, exact-integer, fixnum, negative, odd, exact
255	number, complex, real, rational, integer, exact-integer, exact-nonnegative-integer, exact-positive-integer, fixnum, positive, odd, exact, byte, sequence
256	number, complex, real, rational, integer, exact-integer, exact-nonnegative-integer, exact-positive-integer, fixnum, positive, even, exact, sequence
#""	immutable, bytes, sequence
#"nibble"	immutable, bytes, sequence
hello	symbol
||	symbol
#:woo	keyword
()	null, list, stream, sequence, dict
(1 . 2)	pair
(3)	pair, list, stream, sequence
(5 6)	pair, list, stream, sequence
#()	immutable, vector, sequence, dict
#(1)	immutable, vector, sequence, dict
#("foo" 2 (quote bar))	immutable, vector, sequence, dict
#hash()	hash, hash-equal, sequence, dict
#hasheq()	hash, hash-eq, sequence, dict
#hasheqv()	hash, hash-eqv, sequence, dict
#hash()	immutable, hash, hash-equal, sequence, dict
#hasheq()	immutable, hash, hash-eq, sequence, dict
#hasheqv()	immutable, hash, hash-eqv, sequence, dict
#<hash>	hash, hash-equal, hash-weak, sequence, dict
#<hash>	hash, hash-eq, hash-weak, sequence, dict
#<hash>	hash, hash-eqv, hash-weak, sequence, dict
#hash()	immutable, hash, hash-equal, sequence, dict
#hasheq()	immutable, hash, hash-eq, sequence, dict
#hasheqv()	immutable, hash, hash-eqv, sequence, dict
#&"x"	box
#<void>	void
#<undefined>	undefined
#rx".*"	regexp
#px"3?"	regexp, pregexp
#rx#"t.m"	byte-regexp
#px#".i."	byte-regexp, byte-pregexp
#<sequence>	sequence
#<stream>	stream, sequence
#hash((a . "apple"))	immutable, hash, hash-equal, sequence, dict
#("apple" "binana")	immutable, vector, sequence, dict
("apple" "binana")	pair, list, stream, sequence
((a . "apple") (b . "binana"))	pair, list, stream, sequence, dict
#<set: 1 3 2>	stream, sequence, set-equal, set
#<seteq: 1 2 3>	stream, sequence, set-eq, set
#<seteqv: 1 2 3>	stream, sequence, set-eqv, set
#<mutable-set: 1 2 3>	sequence, set-equal, set-mutable
#<mutable-seteq: 1 2 3>	sequence, set-eq, set-mutable
#<mutable-seteqv: 1 2 3>	sequence, set-eqv, set-mutable
#<weak-set: 1 3 2>	sequence, set-equal, set-weak
#<weak-seteq: 1 2 3>	sequence, set-eq, set-weak
#<weak-seteqv: 1 2 3>	sequence, set-eqv, set-weak
#<procedure:+>	procedure
#<procedure:...pe-detection.rkt:113:13>	procedure
#<continuation>	continuation, procedure
#(struct:object)	object%
#(struct:object:foo% ...)	object%, foo%, fooer<%>
#<syntax:D:\Users\tim\Dropbox\hacking\rosettacode\type-detection.rkt:117:13 (xxy zzy)>	syntax
#<my-tree>	my-tree
#(struct:exn:fail "error: aargh!" #<continuation-mark-set>)	exn, exn:fail
#(struct:exn:fail:filesystem "file-size: file not found\n path: D:/tmp/there-is-no-way-this-file-exists---surely?" #<continuation-mark-set>)	exn, exn:fail, exn:fail:filesystem
#<promise:...e/type-detection.rkt:122:11>	promise

Raku

(formerly Perl 6)

Works with: Rakudo version 2020.08.1

Raku is a dynamic language that has gradual, duck typing. It provides introspection methods through its comprehensive MOP (Meta Object Protocol) making it easy to do type detection, subroutine signatures and multi-dispatch. Raku types have two general flavors: content types and container types. Different container types have varying restrictions on what sort of content they can contain and in return provide specialized methods to operate on those contents. Content types give the compiler hints on how to best handle the information, what storage requirements it may have, what operators will work with it, etc.

This is really a very broad and kind of hand-wavey overview of Raku types. For much more in-depth coverage see Raku Types

<lang perl6>sub type ($t) { say $t.raku, "\tis type: ", $t.WHAT }

1. some content types

.&type for 1, 2.0, 3e0, 4i, π, Inf, NaN, 'String';

1. some primitive container types

.&type for $, [ ], @, { }, %, (5 .. 7), (8 ... 10), /0/, {;}, sub {}, ( );

1. undefined things

.&type for Any, Nil;

1. user defined types

class my-type { };

my my-type $object;

$object.&type;</lang>

Output:

1	is type: (Int)
2.0	is type: (Rat)
3e0	is type: (Num)
<0+4i>	is type: (Complex)
3.14159265358979e0	is type: (Num)
Inf	is type: (Num)
NaN	is type: (Num)
"String"	is type: (Str)
Any	is type: (Any)
$[]	is type: (Array)
$[]	is type: (Array)
{}	is type: (Hash)
{}	is type: (Hash)
5..7	is type: (Range)
(8, 9, 10).Seq	is type: (Seq)
/0/	is type: (Regex)
-> ;; $_? is raw { #`(Block|61385680) ... }	is type: (Block)
sub () { #`(Sub|62948936) ... }	is type: (Sub)
$()	is type: (List)
Any	is type: (Any)
Nil	is type: Nil
my-type	is type: (my-type)

REXX

These are some of the tests that can be performed on REXX variables (values) to determine which   type   they are.

Although everything   (as far as variables are concerning)   in the REXX language is a character string,   character
strings can be classified by having certain characteristics,   or in other words, types.
Characteristics of these   types   can overlap. <lang rexx>/*REXX program displays what "type" a variable is (based on the variable's value). */ signal on noValue /*trap for undefined REXX variables. */ y= 1938  ; call showType y /*╔═══════════════════════════════════╗*/ y= 77.1  ; call showType y /*║ All REXX variables are stored as ║*/ y=  ; call showType y /*║ character strings, even numbers. ║*/ y= ' '  ; call showType y /*║ If a variable string is numeric, ║*/ y= 'abc'  ; call showType y /*║ all comparisons (IF statements) ║*/ y= 'ABC'  ; call showType y /*║ that are made with numbers are ║*/ y= 'aBc'  ; call showType y /*║ compared numerically. If not ║*/ y= '1515'x  ; call showType y /*║ numeric, the string is compared ║*/ y= '10 11'x  ; call showType y /*║ char by char after leading and ║*/ y= '00 0001'b ; call showType y /*║ trailing blanks are removed, and ║*/ y= '1'b  ; call showType y /*║ shorter strings are padded with ║*/ y= ' + 1938 ' ; call showType y /*║ blanks to match the longer string.║*/ y= ' - 1.2e4' ; call showType y /*╚═══════════════════════════════════╝*/ y= '1'  ; call showType y /* */

                         call showType yyy      /*note:  the variable YYY is undefined.*/

exit /*stick a fork in it, we're all done. */ /*──────────────────────────────────────────────────────────────────────────────────────*/ noValue: say ' REXX variable ' condition("D") ' is undefined.'; exit /*──────────────────────────────────────────────────────────────────────────────────────*/ showType: procedure; parse arg x 1 xu; upper xu /*get true value & an uppercase version*/

          @= '      value is';              say  @  x
                                            say  @  'of length'       length(x)
          if  x ==                   then say  @  "null."
          if  x\==  &  x=          then say  @  "all blank."
          if  datatype(x, 'N')         then say  @  "numeric (decimal)."
                                       else say  @  "a character string (not numeric)."
          if  datatype(x, 'W')         then say  @  "an integer (a whole number)."
          if  datatype(x, 'N') &,
             \datatype(x, 'W')         then say  @  "not an integer."
          if  datatype(x, 'N') &,
              pos('E', xu)\==0         then say  @  "a number in exponential format."
          if  datatype(x, 'A')         then say  @  "an alphanumeric string."
          if  datatype(x, 'U')         then say  @  "all uppercase (Latin) letters."
          if  datatype(x, 'L')         then say  @  "all lowercase (Latin) letters."
          if \datatype(x, 'L') &,
             \datatype(x, 'U') &,
              datatype(x, 'M')         then say  @  "of mixed case (Latin) letters."
          if  datatype(x, 'B')         then say  @  "binary."
          if  datatype(x, 'X')         then say  @  "hexadecimal."
          if  datatype(x, 'S')         then say  @  "a REXX symbol."
          say copies('▒',  50)                  /*a fence that is used as a separator. */
          return</lang>

output   when using the internal default data:

      value is 1938
      value is of length 4
      value is numeric (decimal).
      value is an integer (a whole number).
      value is an alphanumeric string.
      value is hexadecimal.
      value is a REXX symbol.
▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
      value is 77.1
      value is of length 4
      value is numeric (decimal).
      value is not an integer.
      value is a REXX symbol.
▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
      value is
      value is of length 0
      value is null.
      value is a character string (not numeric).
      value is binary.
      value is hexadecimal.
▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
      value is
      value is of length 3
      value is all blank.
      value is a character string (not numeric).
▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
      value is abc
      value is of length 3
      value is a character string (not numeric).
      value is an alphanumeric string.
      value is all lowercase (Latin) letters.
      value is hexadecimal.
      value is a REXX symbol.
▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
      value is ABC
      value is of length 3
      value is a character string (not numeric).
      value is an alphanumeric string.
      value is all uppercase (Latin) letters.
      value is hexadecimal.
      value is a REXX symbol.
▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
      value is aBc
      value is of length 3
      value is a character string (not numeric).
      value is an alphanumeric string.
      value is of mixed case (Latin) letters.
      value is hexadecimal.
      value is a REXX symbol.
▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
      value is §§
      value is of length 2
      value is a character string (not numeric).
▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
      value is ►◄
      value is of length 2
      value is a character string (not numeric).
▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
      value is ☺
      value is of length 1
      value is a character string (not numeric).
▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
      value is ☺
      value is of length 1
      value is a character string (not numeric).
▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
      value is  + 1938
      value is of length 8
      value is numeric (decimal).
      value is an integer (a whole number).
▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
      value is  - 1.2e4
      value is of length 8
      value is numeric (decimal).
      value is an integer (a whole number).
      value is a number in exponential format.
▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
      value is 1
      value is of length 1
      value is numeric (decimal).
      value is an integer (a whole number).
      value is an alphanumeric string.
      value is binary.
      value is hexadecimal.
      value is a REXX symbol.
▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
      REXX variable  YYY  is undefined.

Ring

<lang ring>

1. Project : Type detection

see "5 -> " + type(5) + nl see "7.5 -> " + type(7.5) + nl see "d -> " + type('d') + nl </lang> Output:

5 -> NUMBER
7.5 -> NUMBER
d -> STRING

Rust

While Rust compiles to machine code, it does provide multiple mechanisms for requesting that type information be preserved so that it can be queried at runtime.

enum

Rust has first-class support for tagged unions via the enum keyword.

The compiler will require that all possible values are handled, even if it's as simple as a fallthrough _ => unreachable!() which causes the program to die in a memory-safe way.

<lang rust>enum ExpectedTypes {

   Int(i64),
   UInt(u64),
   Real(f64),
   Text(String),
   Uncertain,

}

// Avoid having to prefix each variant name with ExpectedTypes:: use ExpectedTypes::*;

fn main() {

   let enum_test = &[Int(-5), UInt(10), Real(-15.5), Text("Twenty".to_owned()),
                    Uncertain];

   for entry in enum_test {
       match entry {
           Int(x) => println!("Got an integer: {}", x),
           UInt(x) => println!("Got an unsigned integer: {}", x),
           Real(x) => println!("Got a floating-point number: {}", x),
           Text(x) => println!("Got a string of text: {}", x),
           Uncertain => println!("Value is uncertain"),
       }
   }

}</lang>

Output:

Got an integer: -5
Got an unsigned integer: 10
Got a floating-point number: -15.5
Got a string of text: Twenty
Value is uncertain

(Rust also supports untagged unions, but they're only intended to be used for calling or exposing interfaces which use the C ABI.)

Traits

Rust traits are analogous to interfaces in other languages and support both monomorphic dispatch via generics (the compiler will generate a copy of the function for each combination of input arguments used) and polymorphic dispatch via type erasure (known as "Trait objects").

As Rust allows you to implement your own traits on other people's types, this can be used to implement type detection.

<lang rust>use std::error::Error; use std::fs::File; use std::io::{self,prelude::*}; use std::net::TcpStream;

// Declare an Identify trait and implement it for a bunch of types pub trait Identify {

   fn id(&self) -> &'static str;

}

// Declare a macro to compact away the boilerplate macro_rules! declare_id {

   ( $struct:ty, $id:ident ) => (
       impl Identify for $struct {
           fn id(&self) -> &'static str {
               return stringify!($id);
           }
       }
   )

}

// Use the macro to `impl` the Identify trait for a bunch of types declare_id!(io::Empty, empty); declare_id!(File, file_handle); declare_id!(TcpStream, tcp_stream); declare_id!(u8, int8); declare_id!(&str, string);

// This uses monomorphic dispatch via generics. // A copy of the function will be generated for each input type encountered pub fn calc_size<R: Read + Identify>(readable: R) {

   let id = readable.id();
   let mut size = 0;

   for _byte in readable.bytes() {
       size += 1;
   }
   println!(" {}: {} bytes", id, size);

}

// This uses polymorphic dispatch via type erasure pub fn identify(thing: &dyn Identify) {

   println!(" Got {}", thing.id());

}

fn main() -> Result<(), Box<dyn Error>> {

   println!("Monomorphic Generic Interface:");
   calc_size(File::open("/bin/sh")?);
   calc_size(io::empty());
   calc_size(TcpStream::connect("127.0.0.1:37")?);

   println!("\nPolymorphic Interface:");
   for x in &[&15u8 as &dyn Identify, &"Hello" as &dyn Identify] {
       identify(*x);
   }

   Ok(())

}</lang>

Output:

Monomorphic Generic Interface:
 file_handle: 154072 bytes
 empty: 0 bytes
 tcp_stream: 4 bytes

Polymorphic Interface:
 Got int8
 Got string

The Any trait

Finally, while it's more limited than it appears, Rust does have some small degree of support for type introspection.

<lang rust>use std::any::Any;

pub fn is_string(thing: &dyn Any) {

   if thing.is::<&str>() {
       println!("It's a string slice!");
   } else {
       println!("Dunno");
   }

}

fn main() {

   is_string(&"Hello, World!");
   is_string(&5u16);

}</lang>

Output:

It's a string slice!
Dunno

Scala

 <lang scala>object TypeDetection extends App {
 def showType(a: Any) = a match {
   case a: Int => println(s"'$a' is an integer")
   case a: Double => println(s"'$a' is a double")
   case a: Char => println(s"'$a' is a character")
   case _ => println(s"'$a' is some other type")
 }

 showType(5)
 showType(7.5)
 showType('d')
 showType(true)

 println(s"\nSuccessfully completed without errors. [total ${scala.compat.Platform.currentTime - executionStart} ms]")

}</lang>

Scheme

<lang scheme>(define (print-text source)

 (cond ((string? source)
        ;; The source is a string.
        (display source))

       ((and (list? source)
             (or (null? source) (string? (car source))))
        ;; The source is a list of strings.
        (for-each display source))

       ((input-port? source) 
        ;; The source is a file or similar.
        (do ((s (read-line source) (read-line source)))
            ((eof-object? s))
          (display s)
          (newline)))))

(print-text "Print me.\n")

(print-text '("Print\n" "a list\n" "of strings.\n"))

(call-with-input-file "type_detection-scheme.scm" print-text)</lang>

Output:

Using CHICKEN 5 as an R7RS Scheme compiler.

$ csc -R r7rs type_detection-scheme.scm && ./type_detection-scheme

Print me.
Print
a list
of strings.
(define (print-text source)
  (cond ((string? source)
         ;; The source is a string.
         (display source))

        ((and (list? source)
              (or (null? source) (string? (car source))))
         ;; The source is a list of strings.
         (for-each display source))

        ((input-port? source) 
         ;; The source is a file or similar.
         (do ((s (read-line source) (read-line source)))
             ((eof-object? s))
           (display s)
           (newline)))))

(print-text "Print me.\n")

(print-text '("Print\n" "a list\n" "of strings.\n"))

(call-with-input-file "type_detection-scheme.scm" print-text)

Smalltalk

Just send class to any object and ask the class for its name... <lang smalltalk>typeOf := [:anyObject |

   e'arg is of type {anyObject class name} and prints itself as: "{anyObject printString}"' printCR

]. typeOf value:1234. typeOf value:1234.456. typeOf value:(1/2). typeOf value:(3+4i). typeOf value:25 factorial. typeOf value:true. typeOf value:nil. typeOf value:'hello world'. typeOf value: (1 to:100). typeOf value: (TopView new). typeOf value: OperatingSystem. "a reference to the concrete OS" typeOf value: Timestamp now. typeOf value: #() class. typeOf value: #() class class. typeOf value: #() class new. typeOf value: [ 1 print ]. typeOf value: thisContext. "that is the current continuation (stack frame)" typeOf value: #+. typeOf value: (SmallInteger >> #+). [ 1 / 0 ] on:Error do:[:ex | typeOf value: ex ]. "catch the exception" [ 1 fooBar ] on:Error do:[:ex | typeOf value: ex ].</lang>

Output:

arg is of type SmallInteger and prints itself as: "1234"
arg is of type Float and prints itself as: "1234.456"
arg is of type Fraction and prints itself as: "(1/2)"
arg is of type Complex and prints itself as: "(3+4i)"
arg is of type LargeInteger and prints itself as: "15511210043330985984000000"
arg is of type True and prints itself as: "true"
arg is of type UndefinedObject and prints itself as: "nil"
arg is of type String and prints itself as: "hello world"
arg is of type Interval and prints itself as: "1 to:100"
arg is of type TopView and prints itself as: "a TopView"
arg is of type OSXOperatingSystem class and prints itself as: "OSXOperatingSystem"
arg is of type Timestamp and prints itself as: "2020-12-25 23:51:33.275"
arg is of type ImmutableArray class and prints itself as: "ImmutableArray"
arg is of type Metaclass and prints itself as: "ImmutableArray class"
arg is of type ImmutableArray and prints itself as: "#()"
arg is of type Block and prints itself as: "[] in UnboundMethod>>doIt"
arg is of type Context and prints itself as: "UndefinedObject(**DIRECTED**) >> doIt [20]"
arg is of type Symbol and prints itself as: "+"
arg is of type Method and prints itself as: "a Method(SmallInteger >> +)"
arg is of type ZeroDivide and prints itself as: "division by zero"
arg is of type MessageNotUnderstood and prints itself as: "SmallInteger does not understand: "fooBar""

VBA

VBA has a built-in function TypeName (VarType returns a number), which can also recognize arrays. <lang vb>Public Sub main()

   Dim c(1) As Currency
   Dim d(1) As Double
   Dim dt(1) As Date
   Dim a(1) As Integer
   Dim l(1) As Long
   Dim s(1) As Single
   Dim e As Variant
   Dim o As Object
   Set o = New Application
   Debug.Print TypeName(o)
   Debug.Print TypeName(1 = 1)
   Debug.Print TypeName(CByte(1))
   Set o = New Collection
   Debug.Print TypeName(o)
   Debug.Print TypeName(1@)
   Debug.Print TypeName(c)
   Debug.Print TypeName(CDate(1))
   Debug.Print TypeName(dt)
   Debug.Print TypeName(CDec(1))
   Debug.Print TypeName(1#)
   Debug.Print TypeName(d)
   Debug.Print TypeName(e)
   Debug.Print TypeName(CVErr(1))
   Debug.Print TypeName(1)
   Debug.Print TypeName(a)
   Debug.Print TypeName(1&)
   Debug.Print TypeName(l)
   Set o = Nothing
   Debug.Print TypeName(o)
   Debug.Print TypeName([A1])
   Debug.Print TypeName(1!)
   Debug.Print TypeName(s)
   Debug.Print TypeName(CStr(1))
   Debug.Print TypeName(Worksheets(1))

End Sub</lang>

Output:

Application
Boolean
Byte
Collection
Currency
Currency()
Date
Date()
Decimal
Double
Double()
Empty
Error
Integer
Integer()
Long
Long()
Nothing
Range
Single
Single()
String
Worksheet

Visual Basic .NET

Works with: Visual Basic .NET version 9.0+

<lang vbnet>Module TypeDetection

   Sub Main()
       printTypeOf(5)
       printTypeOf("VB.Net")
       printTypeOf(7.2)
       printTypeOf(True)
   End Sub

   Private Sub printTypeOf(obj As Object)
       Console.WriteLine(obj.GetType.ToString)
   End Sub

End Module </lang>

Output:

System.Int32
System.String
System.Double
System.Boolean

Vlang

<lang vlang>fn show_type<T>(a T) {

   println('The type of $a is ${typeof(a).name}')

}

fn main() {

   show_type(-556461841)
   show_type('Rosetta')
   show_type(7.4)
   show_type(`s`)
   show_type([0x32,0x22])

}</lang>

Output:

The type of -556461841 is int
The type of Rosetta is string
The type of 7.4 is f64
The type of s is rune
The type of [50, 34] is []int

Wren

Library: Wren-fmt

<lang ecmascript>import "./fmt" for Fmt

var showType = Fn.new { |obj|

   Fmt.print("$10n has type $q", obj, obj.type)

}

var a = [4, 3.728, [1, 2], { 1: "first" }, true, null, 1..6, "Rosetta"] a.each { |e| showType.call(e) }</lang>

Output:

         4 has type "Num"
     3.728 has type "Num"
    [1, 2] has type "List"
{1: first} has type "Map"
      true has type "Bool"
      null has type "Null"
      1..6 has type "Range"
   Rosetta has type "String"

Z80 Assembly

<lang z80>PrintChar equ &BB5A ;Amstrad CPC BIOS Call

       org &8000

ld de,TestString call PrintDispatch call NewLine

ld de,TestHex call PrintDispatch call NewLine

ld de,TestBCD call PrintDispatch call NewLine

ReturnToBasic: ret

SUBROUTINES

align 8 TypeTable: word PrintString_Text ;string word PrintHexByte ;a single hexadecimal 8-bit value word PrintHexByte ;a single binary-coded-decimal 8 bit value

PrintDispatch: ;input: DE - pointer to "string" of bytes.

ld a,(de) ;read the type header. This is an index into TypeTable inc de ;point DE to actual data. ld h,>TypeTable add a ;double the index, since these are words. ld L,a ;now HL is offset to the desired routine.

;We need to dereference HL and get the actual subroutine address. ; You'd think the JP (HL) would do that, but it does not! ; Despite the syntax, JP (HL) doesn't actually dereference HL, ; it just copies the value stored in HL directly to the program counter. ; There are a couple ways to do this. Self-modifying code uses fewer registers but is a ; bit slow. We'll still use that as I think it's the easiest to follow.

ld a,(hl) ld (Go+1),a ;store low byte in the low byte of the jump's operand. inc hl ;inc to high byte.

ld a,(hl) ld (Go+2),a ;store high byte in the high byte of the jump's operand Go: jp &ABCD ;the &ABCD gets overwritten with the address of the desired function.

PrintString_Text: ld a,(de) ;get next char or a ;is it the terminator? ret z ;if so, exit call PrintChar ;prints accumulator as an ASCII character inc de ;next entry in string jp PrintString_Text ;go back to start

PrintHexByte: ld a,(de) jp ShowHex ;it returns for us.

ShowHex: push af and %11110000 rrca rrca rrca rrca call PrintHexChar pop af and %00001111 PrintHexChar: or a ;Clear Carry Flag daa add a,&F0 adc a,&40 ;this converts hexadecimal and BCD to ASCII - somehow! jp PrintChar ;it returns for us.

NewLine: push af ld a,13 call PrintChar ld a,10 call PrintChar pop af ret

TestString: byte 0,"Hello World",0 TestHex: byte 1,&46 TestBCD: byte 2,&99</lang>

Output:

Hello World
46
99

zkl

<lang zkl>fcn processText(data_or_fileName){ // unknown

  if (data_or_fileName.isType(String)) // == .isType("")
     data_or_fileName=File(data_or_fileName,"rb").read(); //-->Data
  text:=data_or_fileName.text;  //-->String
  doTheActualTextProcessing(text);

} fcn doTheActualTextProcessing(text){ println(text) }</lang> If an int is passed in, (123).text --> "123", other objects might throw an exception.

How to use: <lang zkl>processText("foo.txt"); processText(Data(Void,"This is some text")); // fake up a class that holds a string: cs:=class{ var text }; cs.text="this is more text"; processText(cs);</lang>

Output:

this is foo.txt

This is some text
this is more text