What causes character counts from cin inaccurate and really large?

Question

I'm doing a character count via char c = cin.get(). Example of my output would be: a: 53 b: 32 c: 29 etc..

For one of the alphabets, I'm getting an insanely large 9-digit number. And then when I use another input file that has more characters, the numbers are 6-digits and too large to be accurate. Any thoughts of this phenotype?

Again, sorry it's crunch time with the end of the semester. I appreciate any help out there.

int main (int argc, char *argv[])
{

int count [26] = { };   
char alpha [26] = { };

char c;

c = cin.get();     
while(!cin.eof())
{
  if (isalpha(c))
  {
  c = tolower(c);      
  }    

  count [ c - 'a']++;
  alpha [ c - 'a'] = c;

  c = cin.get();
  }     

  for ( int i = 0; i<26; i++ )
     {
     cout << alpha[i] << ":" << count[i] << endl;
     } 

  } //end main

Here's the output: (edited)

a:224
b:50
c:70
d:20
e:167772180
f:10
g:40
h:66
i:28

Here's the input: (edited)

aaaaaaaaAAAAAaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaAAAaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaBBBB

Made these changes, but it's still "hanging":

c = cin.get();     
while(c=cin.get())
{
  c = tolower(c);
  if (isalpha(c))
      continue;      


  count [ c - 'a']++;
  alpha [ c - 'a'] = c;

  }     

  for ( int i = 0; i<26; i++ )
     {
     cout << alpha[i] << ":" << count[i] << endl;
     } 

  } //end main

You shouldn't assume ASCII is in use, or use `while (!eof())`. Use `while (cin.get(c))` instead. — chris, Nov 30 '12 at 07:55
@KarthikT All 26 alphabets, some capitalized. But, each alphabet either consistently capitalized or in lower case. — harman2012, Nov 30 '12 at 08:00
I would suggest `int count [26] = {0};` and initialize your alphabet array somewhr else. — Karthik T, Nov 30 '12 at 08:01
indeed while we are going into suggestions.. `std::map charCounts;` — Karthik T, Nov 30 '12 at 08:02
@chris Could you explain why while(!eof()) shouldn't be used? I see it being used in class... (sometimes) — harman2012, Nov 30 '12 at 08:04
@chris Changing to while(cin.get()) makes my compiled code "hang"... — harman2012, Nov 30 '12 at 08:07
@harman2012, Work your loop around it. As for why the other is wrong, [see here](http://stackoverflow.com/questions/5605125/why-is-iostreameof-inside-a-loop-condition-considered-wrong). — chris, Nov 30 '12 at 08:09
@KarthikT Okay I just added the input/output text. Sorry it takes up so much space! And I just realized that not all letters are consistently lower or upper case. — harman2012, Nov 30 '12 at 08:14

Henrik · Accepted Answer · 2012-11-30T08:37:26.000

4

If the input file contains anyting else than letters, anything can happen because you are accessing the array wtih out of bounds indices. Maybe you meant:

while(!cin.eof())
{
    if (isalpha(c))
    {
        c = tolower(c);      

        count [ c - 'a']++;
        alpha [ c - 'a'] = c;
     }

    c = cin.get();
}

E.g. when your data contains a line feed character (10) you are accessing alpha with index 10 - 97 = -87. And this probably writes 10 into the most significant byte of count[4].

edited Nov 30 '12 at 08:37

answered Nov 30 '12 at 08:04

Henrik

22,652
6
38
90

Did you mean `(isalpha(c))`? – harman2012 Nov 30 '12 at 08:33
@harman2012 - just realized that `continue` would also skip the `cin.get()` at the end of the loop. Edited my answer. – Henrik Nov 30 '12 at 08:38
Yes, the new line makes sense! Although looking at the file, a new line is returned at "s". I will scrutinize this new discovery. So the smaller file works and that makes sense to put the counts under isalpha(). Thanks so much! – harman2012 Nov 30 '12 at 08:54

What causes character counts from cin inaccurate and really large?

1 Answers1

Linked