检查输入字符串是否为数字，是否使用C ++，如果是，则将其转换为int（正则表达式？）

Question

I feel like this is a fairly basic question, but I've searched the internet for over an hour and I haven't found the answer.

I'm writing a text interface that takes an input as a string. If the input string is a number, I want to convert the string to an integer and push it onto a stack that I created.

The code of the text interface is as follows:

#include <iostream>
#include "textInterface.h"
#include "Stack.h"
#include <string>
#include <regex>

using namespace std;

void Interface(){
    Stack stack = Stack();

    string input;

    cout << "Please enter a number or operation";
    cin >> input;


     if (input == "."){
         cout << stack.pop();
     } //this pops the stack

     if (input == "+"){
         int a = stack.pop();
         int b = stack.pop();
         int c = a + b;
         stack.push(c);
     } //pops the first two things off the stack, adds them, and pushes the result

     if (input == "-"){
        int a = stack.pop();
        int b = stack.pop();
        int c = a - b;
        stack.push(c);
     } //pops the first two things off the stack, subtracts them, and pushes the result

     if (input == "*"){
        int a = stack.pop();
        int b = stack.pop();
        int c = a * b;
        stack.push(c);
     } //pops the first two things off the stack, multiplies them, and pushes the result

     if (input == ".s"){
         cout << stack.count();
     } //returns the size of the stack

     if (regex_match(input, "[0-9]")){
         int num;
         stringstream convert(input);
         convert >> num;
         stack.push(num);
     } //This is the part with the error!!! 

}

Like I said, I'd like to check if input is a number, and if so, convert the string to an int and push it onto the stack. I've worked with regular expressions before, but it's been a while and it was in Python (I'm new to C++). I know my regex_match is not formulated correctly, does anyone have any advice on how to make it correct, or suggestions for resources to read?

Answer 1

Don't check if it looks like an integer and then convert it to an integer. Rather, convert it to an integer and see if that worked. Use std::stoi (or stol or stoll , depending on how big you expect the number to be.) (in <string> ): See here .

If the string cannot be converted to an integer of the indicated size, an exception will be thrown, so you have to execute that function inside of a try . (That style should be familiar if you're a pythonista.)

Also, if there was a number but it didn't occupy the entire string (ie trailing garbage), the size_t pointed to be the second argument will be set to the index of the first unused character, so if you want to check that the entire string is a number, you should also check to make sure that the returned index is the size of the input string.

If you're uncomfortable with exceptions, even those thrown by the standard library, it's possible to just use underlying standard c functions, strtol and friends. Those have a similar interface, but use an arbitrary value return combined with setting errno to try to communicate failure. Personally I think the exception interface is less fiddly, but, as always, tastes vary.

Answer 2

I upvoted your question because I had the same problem and found no obvious solution. I developped a, somewhat flawed, solution but good enough for common cases. Still it is definitely something that lacks answers when looking from google. So for the sake of humanity knowledge I'm going to post my solution here.

bool StringCtr::IsNumeric(bool onlyIntegersYieldTrue /*= false*/, int intRadix /*= 0*/) const
{
    _ASSERT(intRadix == 0 || intRadix >= 2 && intRadix <= 36);

    bool const acceptsFloats = !onlyIntegersYieldTrue;
    bool ok = false;
    if (!Empty())
    {
        char* pEnd = 0;
        long li1;
        li1 = strtol(Cstr(), &pEnd, intRadix);
        ok = ! (li1 == 0L && pEnd == Cstr());
        if (!ok && acceptsFloats)
        {
            double d1;
            d1 = strtod(Cstr(), &pEnd);
            ok = ! (d1 == 0.0 && pEnd == Cstr());
        }
        if (ok && !acceptsFloats)
        {
            ok = ok && !Contains(".");
            ok = ok && !Contains("e");
        }
    }           
    return ok;
}

so this is a method of my string class, the other methods called from there are self explanatory enough I believe. This is trying to use standard C tokenizer (strtol) which is able to accept a lot of formats. That is actually something to be real careful about, because for example number starting with 0 will be interpreted as octal base, when using 0 (auto) radix. otherwize it is convenient because it will recognize 0x.. hexa notation pattern, and distinguish between integers and floating points. it should even understand exponent notation eg "-1.05e12" for example.

Anyways, it depends on the parsing power you really need, what kind of notation convention you think your users will excpect the program to recognize ? And in the end, do not listen to people saying "it is easy, just check foreach (char c) { ok = c >= '0' && c <= '9'; } because this is bullshit and obviously not sufficient. What about negative numbers, floating points, and all that I mentioned before ?

cheers !