# how to extract numbers from string

I have a string

OWNER_5477854

And i would like to simply return the sum of just the numbers. If my string does not contain any numbers, i would like any 2 digit number returned.
LVL 1
###### Who is Participating?

x

Commented:
``````#include <stdio.h>
#include <string>
#include <iostream>
using namespace std;

int sumOfDigits( string text ) {
int sum = 0;

for( size_t i=0; i<text.size(); ++i ) { // loop over each char
const char character = text[i];      // save the char
if( isdigit( character ) ) {             // if the char is a digit, then we can
char digitBuf[2] = {character, '\0'}; // convert it to a c-style string, so that
sum += atoi( digitBuf );              // we can convert it to an int.
}                                        // Could have just done a (character - '0')
}                                           // to convert to int if standard ASCII code used
return sum;
}

int main() {
string text = "OWNER_5477854";
char sumString[20]; // string that holds the sum
sprintf( sumString, "%02d", sumOfDigits("OWNER_5477854") );
cout << "Sum = " << sumString << endl;
sprintf( sumString, "%02d", sumOfDigits("OWNER_NO_DIGITS") );
cout << "Sum No Digits Present:  " << sumString << endl;
}
``````
Output is:
``````Sum = 40
Sum No Digits Present: 00
``````
0

Commented:
Does 00 count as "any 2 digit number"?
0

Author Commented:
Sure  doesn't matter what.
0

Author Commented:
Wow, that is a fantastic solution! Very well documented, thank you very much. More importantly than working code, i understand the way you presented it.
0

Commented:
Points already assigned - this is just for interest.

Just cos I am old school and this was an intersting problem - here is a solution that is about 100x faster than the one posted - actually there are two solutions depending on the question

If
a) from OW_49_NER_5477854 you want the answer to be 53
OR
b) from OW_49_NER_5477854 you want the answer to be 13 (i.e. break after the number is broken
Option 1
``````int SoD(const char * text)
{
const char *s = text;
int sum = 0;
while(*s) {
char d = *(s++) - 48; // Get digit value
sum += (d >= 0 && d <=9) ? d : 0; // add to sum only if between 0-9
}

return sum;
}
``````
Option 2
``````int SoD2(const char * text)
{
const char *s = text;
int sum = 0;
bool flag = false; // Flag to tell us if we have found a number yet
while(*s) {
char d = *(s++) - 48;
if (d >=0 && d<=9){
flag = true; // found one
}
else if (flag) break; // this char is not a number and we already found a number so break
// If we are in a number string then add to sum (can probably drop the condition
if(flag){
sum += (d >= 0 && d <=9) ? d : 0;
}
}

return sum;
}
``````
0

Commented:
>>that is about 100x faster than the one posted
Is there a typo here? If not, try measuring and let me know what you actually discover. Why don't you post the measuring driver that you used - I kind of think your performance factors are off.
Also, which one posted are you referring to, since I posted two algorithms?
0

Commented:
Ok here is what I did.

I put the accepted solution into a function and the posted code above.

Then for each of the functions I ran a loop of 100,000 iterations surround by

long start = GetTickCount();
... loop ...
long end = GetTickCount();

I then dumped the difference between the start and end times for each of the loops.

For the accepted solution the ranges in time were between 1500ms and 2100ms

For the code posted above the ranges in time were between 15ms and 20ms which indicates an approximate factor of 100 in terms of increase in speed.

Of course I may be totally off here because I knocked this together in a couple of minutes but it makes sense - the accepted solution is doing a lot of unnecessary work to achieve the same result. Also, in the grand scheme of things it makes absolutely no difference because you have to call the function 100,000 times before it makes a significant difference. I am just an old school programmer who looks for the optimal solution - which is why I posted it for interest.

Full source here

``````// ee1.cpp : Defines the entry point for the console application.
//

#include "stdafx.h"
#include "windows.h"
#include <stdio.h>
#include <iostream>
using namespace std;

int SoD(const char * text)
{
const char *s = text;
int sum = 0;
while(*s) {
char d = *(s++) - 48;
sum += (d >= 0 && d <=9) ? d : 0;
}

return sum;
}

int SoD2(const char * text)
{
const char *s = text;
int sum = 0;
bool flag = false;
while(*s) {
char d = *(s++) - 48;
if (d >=0 && d<=9){
flag = true;
}
else if (flag) break;

if(flag){
sum += (d >= 0 && d <=9) ? d : 0;
}
}

return sum;
}

int sumOfDigits( string text ) {
int sum = 0;

for( size_t i=0; i<text.size(); ++i ) { // loop over each char
const char character = text[i];      // save the char
if( isdigit( character ) ) {             // if the char is a digit, then we can
char digitBuf[2] = {character, '\0'}; // convert it to a c-style string, so that
sum += atoi( digitBuf );              // we can convert it to an int.
}                                        // Could have just done a (character - '0')
}                                           // to convert to int if standard ASCII code used
return sum;
}

void test1()
{
const char * input  = "OWNER_5477854";
char sumString[20];

int sum;
long end, start = GetTickCount();
for(int i = 0; i < 100000; i++) sum = SoD(input);
end = GetTickCount();
sprintf( sumString, "%02d (%ld)", sum, (end - start));
cout << "Sum = " << sumString << endl ;
}

void test2()
{
string text = "OWNER_5477854";
int sum;
char sumString[20]; // string that holds the sum
long end, start = GetTickCount();
for(int i = 0; i < 100000; i++) sum = sumOfDigits("OWNER_5477854");
end = GetTickCount();
sprintf( sumString, "%02d (%ld)", sum, (end - start) );
cout << "Sum = " << sumString << endl;
sprintf( sumString, "%02d", sumOfDigits("OWNER_NO_DIGITS") );
cout << "Sum No Digits Present:  " << sumString << endl;
}

int _tmain(int argc, _TCHAR* argv[])
{
test1();
test2();

int r ;
cin >> r;
return 0;
}
``````
0

Commented:
What results did you get for my other algorithm? See lines 14-15
0

Commented:
This is the output from the code above - what other algorithm are you referring to not sure what the 14-15 is pointing at?

For 100,000 iterations

Sum = 40 (16)
sumOfDigits = 40 (1515)

For 1,000,000 iterations

Sum = 40 (156)
Sum = 40 (13812)

If you want to take this offline mail me - address is in my profile.
0

Commented:
// we can convert it to an int.
// Could have just done a (character - '0')
This is the other algorithm that I suggested to simplify things. I showed general usage of library functions for educational purposes, but then added this extra option. And I do not believe the numeric representation has to be ASCII, as I mentioned.
0

Commented:
Confused as to where this is going - I posted a solution (out of interest - after points were assigned) making the point that the solution was potentially 100x faster than the code in the accepted solution. You questioned where I got that result from - I posted code for that. What are we trying to achieve here?
0

Commented:
I was wondering whether you got a 100x faster measurement for the alternative algorithm I proposed, where we replace my code where we "Could have just done a (character - '0')" to get the int. You explained that you only tested with the first algorithm presented. Thanks for including your test program. Concluded.
0
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.