[C++] How do I judge the type of input is right or not?

chuanyung · 07-20-2004, 05:20 AM

Dear all,

I use C++'s cin to get data from screen.
If I want 1 string and 1 floating number,
but the input is not like above.
When this case appears, I must show a error.
How can I do ?
-------------------
string A;
float B;

cin >> A >> B;

/* then ??? */
--------------------
Thanks for your help.

Best Regards,
Chuanyung.

spacer · 07-20-2004, 09:22 AM

Quote:

Originally posted by chuanyung
Dear all,

I use C++'s cin to get data from screen.
If I want 1 string and 1 floating number,
but the input is not like above.
When this case appears, I must show a error.
How can I do ?
-------------------
string A;
float B;

cin >> A >> B;

/* then ??? */
--------------------
Thanks for your help.

Best Regards,
Chuanyung.

Anything cam be thought of as a string, so u cant actually check for it .
As reg float or not, u can typecast it to int n then check whether the both r same or not.

i.e int C = (int)B;
if (C==B)
then not float;
else float;

Hope it helps u

Hko · 07-20-2004, 09:31 AM

You may want to use strtod() for conversion. Read the "RETURN VALUE" section of its man -page on how to detect if a floating point number was correctly read.

Tesl · 07-20-2004, 10:37 AM

Could you accept both the inputs, and then in the function that uses them surround it with a try{ catch{ block; and catch the input error that way?

There might be better ways of doing it, but iv always found exception handling really helpful for things like that

dakensta · 07-20-2004, 11:10 AM

If you provide input that cannot be formatted into required type, for example you send "Hello" to cin >> a_float; then you will find that cin.fail() returns true.

You can also put your input into a 'while' statement which will implicitly cast the returned stream to void* and thus a bool test.

For reading a string, you might want to think about std::getline, rather than cin >> a_string; as this lets you use whitespace in the string and will remove the terminating '\n', if you want to.

N.B. This does depend on the type of input you expect ... from the screen I would expect a prompt before each data input and to hit 'return' after typing the desired data, from a file (or piped file) I would expect different, maybe whitespace delimited columns which would require a slightly different strategy, (like reading the whole lot into a string as suggested above) so post back if you have a particular input format and get stuck. I would tend to use istringstream's to convert strings into other datatypes but feel free to use whatever you like.

Example:

Code:

#include <iostream>
#include <string>
#include <limits>

using namespace std;

int main()
{

  float f;
  string s1;
  cout <<"Enter string\n";
  
  // Assume this function will work, pretty rare not to ...
  getline(cin, s1)

  cout << "Enter float\n";

  // Until we get good input, keep trying ...
  while (!(cin >> f))
    {
      // Error message
      cout << "Float input failed\nEnter float\n";
      // clear the failbit set by bad input
      cin.clear();
      // We probably still have bad input lying in the stream
      // so we need to dispose of it
      cin.ignore(numeric_limits<streamsize>::max(), '\n');
    }

  cout << "Data Entered " << s1 << " and " << f << endl;

}

chuanyung · 07-20-2004, 08:31 PM

Quote:

Originally posted by spacer
Anything cam be thought of as a string, so u cant actually check for it .
As reg float or not, u can typecast it to int n then check whether the both r same or not.

i.e int C = (int)B;
if (C==B)
then not float;
else float;

Hope it helps u

This has a bug when input is such as 10.0 .
10.0 will be not a float number.

chuanyung · 07-20-2004, 08:37 PM

Quote:

Originally posted by dakensta
If you provide input that cannot be formatted into required type, for example you send "Hello" to cin >> a_float; then you will find that cin.fail() returns true.

You can also put your input into a 'while' statement which will implicitly cast the returned stream to void* and thus a bool test.

For reading a string, you might want to think about std::getline, rather than cin >> a_string; as this lets you use whitespace in the string and will remove the terminating '\n', if you want to.

N.B. This does depend on the type of input you expect ... from the screen I would expect a prompt before each data input and to hit 'return' after typing the desired data, from a file (or piped file) I would expect different, maybe whitespace delimited columns which would require a slightly different strategy, (like reading the whole lot into a string as suggested above) so post back if you have a particular input format and get stuck. I would tend to use istringstream's to convert strings into other datatypes but feel free to use whatever you like.

Example:

Code:

#include <iostream> #include <string> #include <limits> using namespace std; int main() { float f; string s1; cout <<"Enter string\n"; // Assume this function will work, pretty rare not to ... getline(cin, s1) cout << "Enter float\n"; // Until we get good input, keep trying ... while (!(cin >> f)) { // Error message cout << "Float input failed\nEnter float\n"; // clear the failbit set by bad input cin.clear(); // We probably still have bad input lying in the stream // so we need to dispose of it cin.ignore(numeric_limits<streamsize>::max(), '\n'); } cout << "Data Entered " << s1 << " and " << f << endl; }

This works almost well.
But
(a) if I want a realy float, input is a integer.
(b) if I want a realy integer,input is a float.
For these 2 cases, cin.fail() is always return false.
So it seems not to check perfectly.

dakensta · 07-21-2004, 04:11 AM

I am not sure I quite follow but this is what I think you are saying is that you want to determine a float by the presence of a decimal point.

So, for the statement (using float f):
cin >> f;

input:
a) 12.34
b) 12

a) is acceptable to you because it has a decimal point, b) is not because there is no decimal point.
To c++ both are acceptable and you have a value of 12.34 or 12.0.

For the statement (using int i):
cin >> i

input:
a) 12.34
b) 12

a) is unacceptable because there is a decimal point, b) is acceptable because there is no decimal point.

To c++ BOTH are acceptable and i will take on the value 12. This is because, in both cases you have provided something that is convertable to an integer.
What you need to be aware of is that in a) you still have ".34\n" still waiting in the buffer.

The quick way to check this is write:
if (cin.peek() == '.') { //call error handler }

But, if you are going to require the presence of a decimal point, then you need to read in a string, validate the string and decide if you are going to accept it or not.

From a c++ point of view, it is not malformed input unless the supplied input CANNOT be converted into the desired type.
An int 12 from 12.34 is c++ 'best guess'.
A float 12 from input 12 is still a float despite no decimal point being supplied.

HTH