Question

I have a program in C++ that needs to return a line that a specific word appears in. For instance, if my file looks like this:

the cow jumped over
the moon with the
green cheese in his mouth

and I need to print the line that has “with”. All the program gets is the offset from the beginning of the file (in this case 24, since “with” is 24 characters from the beginning of the file).

How do I print the whole line “the moon with the”, with just the offset?

Thanks a lot!

Answer 1

A good solution is reading the file from the beginning until the desired position (answer by @Chet Simpson). If you want optimization (e.g. very large file, position somewhere in the middle, typical lines rather short), you can read the file backwards. However, this only works with files opened in binary mode (any file on unix-like platforms; open the file with ios_base::binary parameter on Windows).

The algorithm goes as follows:

• Go back a few bytes in file
• Read the few bytes
• If there is an end-of-line there, the rest is easy
• Otherwise, repeat

Code (tested on Windows):

std::string GetSurroundingLine(std::istream& f, std::istream::pos_type start_pos)
{
    std::istream::pos_type prev_pos = start_pos;
    std::istream::pos_type pos;
    char buffer[40]; // typical line length, so typical iteration count is 1
    std::istream::pos_type size = sizeof(buffer);

    // Look for the beginning of the line that includes the given position
    while (true)
    {
        // Move back 40 bytes from prev_pos
        if (prev_pos < size)
            pos = 0;
        else
            pos = prev_pos - size;
        f.seekg(pos);

        // Read 40 bytes
        f.read(buffer, prev_pos - pos);
        if (!f)
            throw;

        // Look for a newline byte, which terminates previous line
        int eol_pos;
        for (eol_pos = sizeof(buffer) - 1; eol_pos >= 0; --eol_pos)
            if (buffer[eol_pos] == '\n')
                break;

        // If found newline or got to beginning of file - done looking
        if (eol_pos >= 0 || pos == (std::istream::pos_type)0)
        {
            pos += eol_pos + 1;
            break;
        }
    }

    // Position the read pointer
    f.seekg(pos);

    // Read the line
    std::string s;
    std::getline(f, s, '\n');

    return s;
}

Edit: On Windows-like platforms, where end-of-line is marked by \r\n, since you have to use binary mode, the output string will contain the extra character \r (unless there is no end-of-line at end-of-file), which you can throw away.