Skip lines in std::istream

Question

I'm using std::getline() to read lines from an std::istream-derived class, how can I move forward a few lines?

Do I have to just read and discard them?

Solution 1

Edit: You can also use std::istream::ignore, see https://stackoverflow.com/a/25012566/492336

---

Do I have to use getline the number of lines I want to skip?

No, but it's probably going to be the clearest solution to those reading your code. If the number of lines you're skipping is large, you can improve performance by reading large blocks and counting newlines in each block, stopping and repositioning the file to the last newline's location. But unless you are having performance problems, I'd just put getline in a loop for the number of lines you want to skip.

Solution 2

No, you don't have to use getline

The more efficient way is ignoring strings with std::istream::ignore

for (int currLineNumber = 0; currLineNumber < startLineNumber; ++currLineNumber){
    if (addressesFile.ignore(numeric_limits<streamsize>::max(), addressesFile.widen('\n'))){ 
        //just skipping the line
    } else 
        return HandleReadingLineError(addressesFile, currLineNumber);
}

HandleReadingLineError is not standart but hand-made, of course. The first parameter is maximum number of characters to extract. If this is exactly numeric_limits::max(), there is no limit: Link at cplusplus.com: std::istream::ignore

If you are going to skip a lot of lines you definitely should use it instead of getline: when i needed to skip 100000 lines in my file it took about a second in opposite to 22 seconds with getline.

Solution 3

Yes use std::getline unless you know the location of the newlines.

If for some strange reason you happen to know the location of where the newlines appear then you can use ifstream::seekg first.

You can read in other ways such as ifstream::read but std::getline is probably the easiest and most clear solution.

Comments

[GManNickG]

Counting newlines is sorta what a looped getline would do, right?

[Billy ONeal]

@GMan - Save the Unicorns: Yes, but you can use larger blocks and shift forward a larger distance if you know n is large (i.e. counting more than one newline in a buffer block)

[ufk]

what if I'll load all the file to memory and parse it with regular expressions ? the file is less then 1MB.

[Brian R. Bondy]

@Billy: Actually your tardiness gave me 10 extra rep since I was already at max cap until 10 min ago.

[Billy ONeal]

LOL! I have yet to bump my head into said cap.