I want to expand the interpreter to understand numbers - both as tokens directly in the code (literals) and as values in variables. So that LET A = 123 no longer sets the variable A to the string "123" but to the integer 123. And so that LET B = A no longer sets the variable B to the string "A" but to the value of the variable A, in this case the integer 123.

This is probably going to be a bit of work - first I have to let the tokenizer understand numbers, so that LET A = 123 becomes LET <identifier ‘A’> <equal> <number 123>.

Number token

That means that the nextToken function has to be extended so any digit begins s number token, and not an identifier token

flowchart LR
  A(read character)
  A --> B{ }
  B -->|whitespace| C([begin **whitespace** token])
  B -->|digit 0-9| N([begin **number** token])
  B -->|#quot;| D([begin **string** token])
  B -->|=| E([**equal sign** token])
  B -- letter --> I([begin **identifier** token])

By the way, I’m going to use the builtin functions in ctype.h to check for the different types - no longer use if (c == ' ' || c == '\\t') but simply if (isspace(c)) - and to check for digits, if (isdigit(c)).

So if the character is a digit, we continue to tokenize_number, that will continue as long as there are more digits. Almost the exact same as tokenize_whitespace:

---
title: Number
---
flowchart LR
  a(next character)
  a-->b{ }
  b-->|digit 0-9|f
  b-->|anything else|g([end token])
  f(continue token)-.->a

And of course I need to add number as a token type:

typedef enum
{
  WHITESPACE,
  NUMBER,
  STRING,
  IDENTIFIER,
  EQUALS,
  END
} TOKEN_TYPE;

static const char *TOKEN_NAMES[] = {
    "WHITESPACE",
    "NUMBER",
    "STRING",
    "IDENTIFIER",
    "EQUALS",
    "END"};

And that should be it - there really is nothing to it - so lets test it!

Great, an error - so it works … It works that we get a number - now we just need to be able to use numbers. Let’s take a look at LET first!

kw_let

Earlier I allowed the variable value to be set to the value of either a string or an identifier - simply changing that to string or number should fix that!

if(token->type == STRING || token->type == NUMBER)
{
  value = token->value;
}

Test it by creating a variable with the value of a number, and printing it:

Cool, that already works - and the variable can be printed because a is an identifier, so PRINT doesn’t know that it is a number.

Now it should also be possible to handle <identifiers> correctly, so I can write let b = a, and thus set the value of b to 1234. Just add a special if-statement: