#include <vector>
#include <ctype.h>
#include <string>

typedef std::vector<std::string> string_list;
typedef std::vector<long long > int_list;
typedef std::vector<long double> float_list;

std::string substr(const char* value, size_t length){
    std::string v;
    v.resize(length);
    memcpy(&v[0], value, length * sizeof(char));
    return v;
}

long long string_to_int(const char* value, size_t length){
    return atoll(substr(value, length).c_str());
}
long double string_to_float(const char* value, size_t length){
    return atof(substr(value, length).c_str());
}


void int_list_add(int_list& list, long long value){
    list.push_back(value);
}
void string_list_add(string_list& list, const char* value, size_t length){
    list.push_back(substr(value, length));
}
void float_list_add(float_list& list, long double value){
    list.push_back(value);
}
size_t int_list_last(int_list& list){
    return list.size();
}
size_t string_list_last(string_list& list){
    return list.size();
}
size_t float_list_last(float_list& list){
    return list.size();
}



typedef struct{
    string_list identifiers;
    string_list constants_string;
    int_list constants_int;
    float_list constants_float;
    size_t id;
} *state, state_value;

state tok_state_create(){
    state ret = new state_value;
    ret->id = 0;
    return ret;
}
void tok_state_destroy(state t_state){
    delete t_state;
}
const char* tok_state_read_identifier(state t_state, size_t id){
    return t_state->identifiers[id - 1].c_str();
}
const char* tok_state_read_string(state t_state, size_t id){
    return t_state->constants_string[id - 1].c_str();
}
long long tok_state_read_int(state t_state, size_t id){
    return t_state->constants_int[id - 1];
}
long double tok_state_read_float(state t_state, size_t id){
    return t_state->constants_float[id - 1];
}



const char* punct_tokens[] = { "Not A Token (Dummy)",
".", ",", "<", "<<", ">", ">>",
";", "+", "-", "/", "*", "!", "%", "^",
"&", "(", ")", "=", "==", "[", "]", "{",
"}", "?", ":", "|", "||", "&&", "~", 0
};

const char* key_tokens[] = { "Not A Token (Dummy)",
"if", "while", "do", "then", "end", 0
};

typedef enum{
    TOK_TYPE_INTEGER = 500,
    TOK_TYPE_FLOAT,
    TOK_TYPE_STRING,
    TOK_TYPE_IDENTIFIER,
    TOK_TYPE_NONE
} tok_type;

const char* get_token_from_id(size_t id){
    if (id < 100){
        return punct_tokens[id];
    }
    if (id < 200){
        return key_tokens[id - 100];
    }
    if (id >= 500){
        switch (id){
        case TOK_TYPE_INTEGER:      return "Integer Constant";
        case TOK_TYPE_FLOAT:        return "Float Constant  ";
        case TOK_TYPE_STRING:       return "String Constant ";
        case TOK_TYPE_IDENTIFIER:   return "Identifier      ";
        case TOK_TYPE_NONE:         return "Unknown         ";
        default:
            break;
        }
    }
    return "Not A Token (Dummy)";
}

int is_identifier_char(char c){
    if (isalpha(c) || c == '_'){
        return 1;
    }
    return 0;
}

size_t read_punct_token(const char* input, size_t size){
    size_t max_len = 0;
    size_t token_id = 0;
    for (size_t i = 1; punct_tokens[i] != 0; ++i){
        size_t len = strlen(punct_tokens[i]);
        if (len > max_len && len <= size && strncmp(punct_tokens[i], input, len) == 0){
            max_len = len;
            if (i == 1 && size > 1 && isdigit(input[1])){
                return 0; //Special case for floats
            }
            token_id = i;
        }
    }
    return token_id;
}

size_t read_key_token(const char* input, size_t size){
    size_t max_len = 0;
    size_t token_id = 0;
    for (size_t i = 1; key_tokens[i] != 0; ++i){
        size_t len = strlen(key_tokens[i]);
        if (len > max_len && len <= size && strncmp(key_tokens[i], input, len) == 0){
            max_len = len;
            token_id = i + 100;
        }
    }
    return token_id;
}


size_t is_punct_token_char(char c){
    for (size_t i = 1; punct_tokens[i] != 0; ++i){
        if (punct_tokens[i][0] == c){
            return 1;
        }
    }
    return 0;
}


void add_token(state t_state, tok_type type, const char* string, size_t length){
    switch (type){
    case TOK_TYPE_INTEGER:
        int_list_add(t_state->constants_int, string_to_int(string, length));
        t_state->id = int_list_last(t_state->constants_int);
        break;
    case TOK_TYPE_FLOAT:
        float_list_add(t_state->constants_float, string_to_float(string, length));
        t_state->id = float_list_last(t_state->constants_float);
        break;
    case TOK_TYPE_STRING:
        string_list_add(t_state->constants_string, string, length);
        t_state->id = string_list_last(t_state->constants_string);
        break;
    case TOK_TYPE_IDENTIFIER:
        string_list_add(t_state->identifiers, string, length);
        t_state->id = string_list_last(t_state->identifiers);
        break;
    default:
        //Do some error here
        break;
    }
}

size_t get_token(state t_state, char** input, size_t *size){
    if (t_state->id != 0){
        size_t id = t_state->id;
        t_state->id = 0;
        return id;
    }
    char* base = *input;
    size_t padding = 0;
    size_t length = 0;
    tok_type type = TOK_TYPE_NONE;
    while (*size > 0){
        if (isspace(*base)){
            base++;
            (*size)--;
        }
        else{
            break;
        }
    }

    size_t tok = read_punct_token(base, *size);
    if (tok){
        size_t len = +strlen(get_token_from_id(tok));
        *input = base + len;
        *size -= len;
        return tok;
    }
    tok = read_key_token(base, *size);
    if (tok){
        size_t len = +strlen(get_token_from_id(tok));
        *input = base + len;
        *size -= len;
        return tok;
    }

    while (*size - length > 0){
        if (length == 0 && type == TOK_TYPE_NONE){
            if (is_identifier_char(*base)){
                type = TOK_TYPE_IDENTIFIER;
                length++;
            }
            else if (*base == '"'){
                type = TOK_TYPE_STRING;
                padding = 1;
                base++;
                (*size)--;
            }
            else if (*base == '.' && *size > 1 && isdigit(base[1])){
                type = TOK_TYPE_FLOAT;
            }
            else if (isdigit(*base)){
                type = TOK_TYPE_INTEGER;
            }
            else if (is_punct_token_char(*base)){
                tok = read_punct_token(base, *size);
                if (tok){
                    size_t len = strlen(punct_tokens[tok]);
                    *input += len;
                    *size -= len;
                    return tok;
                }
                else{
                    //do error
                }
            }
        }
        else{
            if (!isspace(base[length]) || type == TOK_TYPE_STRING){
                switch (type){
                case TOK_TYPE_INTEGER:
                    if (isdigit(base[length])){
                        length++;
                        continue;
                    }
                    else if (base[length] == '.' || tolower(base[length]) == 'e'){
                        type = TOK_TYPE_FLOAT;
                        length++;
                        continue;
                    }
                    break;
                case TOK_TYPE_FLOAT:
                    if (isdigit(base[length]) || base[length] == '.' || base[length] == 'e'){
                        length++;
                        continue;
                    }
                    break;
                case TOK_TYPE_STRING:
                    if (base[length] != '"'){
                        length++;
                        continue;
                    }
                    break;
                case TOK_TYPE_IDENTIFIER:
                    if (is_identifier_char(base[length])){
                        length++;
                        continue;
                    }
                    break;
                default:
                    break;
                }
            }
            //We only get here if this is a space or any of the switch cases didn't continue.
            add_token(t_state, type, base, length);
            *input = base + length + padding;
            *size -= length + padding;
            return type;
        }
    }
    *input = base + length + padding;
    *size -= length + padding;
    return 0;
}

int main(){
    const char* input = "if(1+1==4)then print"hi!";end";
    state s = tok_state_create();
    size_t size = strlen(input);
    size_t token;
    size_t token_prev = 0;
    printf("TokentMeaningnn");

    while ((token = get_token(s, (char**)&input, &size)) != 0){
        if (token_prev < 500){
            if (token < 500){
                printf("%dt%sn", token, get_token_from_id(token));
            }
            else{
                printf("%dt%s #", token, get_token_from_id(token));
            }
        }
        else{
            printf("%dt", token);
            switch (token_prev){
            case TOK_TYPE_IDENTIFIER: printf("%sn", tok_state_read_identifier(s, token)); break;
            case TOK_TYPE_STRING: printf("%sn", tok_state_read_string(s, token)); break;
            case TOK_TYPE_INTEGER: printf("%dn", tok_state_read_int(s, token)); break;
            case TOK_TYPE_FLOAT: printf("%fn", tok_state_read_float(s, token)); break;

            }
        }
        token_prev = token;
    }

    tok_state_destroy(s);
}

Token   Meaning

101     if
16      (
500     Integer Constant #1     1
8       +
500     Integer Constant #2     1
19      ==
500     Integer Constant #3     4
17      )
104     then
503     Identifier       #1     print
502     String Constant  #1     hi!
7       ;
105     end