// Expression Compiler
// Author: Yevano

#define T_PLUS     0
#define T_DASH     1
#define T_ASTERISK 2
#define T_FSLASH   3
#define T_LPAREN   4
#define T_RPAREN   5
#define T_NUMBER   6

//////// TOKENIZER ////////

// Returns pointer to an array of tokens (ebx);
// Parameters:
//     eax: pointer to code string
tokenize:
    tokens: alloc 3000;  // Allocate 4k for tokens (tokens are 3 bytes each (type, strindex1, strindex2), so max of 2k tokens)

    // Preserve registers
    push R0;             // R0 points to characters in the code string
    push R1;             // R1 points to current token
    push R2;             // Misc

    mov R0, eax;         // Point to the code string

    t_loop:
        // Check which kind of token to add
            cmp #R0, '+';
            jne check_dash;
            mov #R1, T_PLUS;
            add #R1, 3;                // Ready for next token
        check_dash:
            cmp #R0, '-';
            jne check_asterisk;
            mov #R1, T_DASH;
            add #R1, 3;                // Ready for next token
        check_asterisk:
            cmp #R0, '*';
            jne check_fslash;
            mov #R1, T_ASTERISK;
            add #R1, 3;                // Ready for next token
        check_fslash:
            cmp #R0, '/';
            jne check_lparen;
            mov #R1, T_FSLASH;
            add #R1, 3;                // Ready for next token
        check_lparen:
            cmp #R0, '(';
            jne check_rparen;
            mov #R1, T_LPAREN;
            add #R1, 3;                // Ready for next token
        check_rparen:
            cmp #R0, ')';
            jne check_number;
            mov #R1, T_RPAREN;
            add #R1, 3;                // Ready for next token
        check_number:
            mov R2, R0;                // R2 gets incremented to last index
            push eax;                  // Preserve eax before calling char_is_digit
            mov eax, #R2;              // Set the char parameter to the current char
            call char_is_digit;        // Call char_is_digit :)
            pop eax;                   // Restore eax
            cmp ebx, 0;                // Jump to done if not a digit
            je check_number_loop_done; // Finish the loop if no more digits
            inc R2;                    // Increment R2
            jmp check_number;          // Loop again

            check_number_loop_done:
            cmp R2, R0;                // This means R2 wasn't incremented, so no digits.
            je error;                  // Error, we can't tokenize dis shit
            mov #R1, T_NUMBER;         // Set token type
            inc R1;
            mov #R1, R0;               // Start index
            inc R1;
            mov #R1, R2;               // End index
            inc R1;                    // Ready for next token
        check_number_done:

        cmp #R0, 0;       // Continue tokenizing until we get a null character
        jne t_loop;

    pop R2;              // Restore used registers
    pop R1;
    pop R0;
    mov ebx, tokens;     // Return the tokens in ebx
    ret;

    error:
    pop R2;              // Restore used registers
    pop R1;
    pop R0;
    mov ebx, 0;          // Return null

// Returns 1 if current given char is a digit (ebx);
// Paramaters:
//     eax: char
char_is_digit:
    cmp eax, '0';
    jl char_is_digit_false;
    cmp eax, '9';
    jg char_is_digit_false;
    mov ebx, 1;
    ret;
    char_is_digit_false: mov ebx, 0;
    ret;

//////// END TOKENIZER ////////