为什么vector.size（）在一行中读得太少？

Question

when running the following code, the amount of lines will read on less then there actually is (if the input file is main itself, or otherwise) why is this and how can i change that fact (besides for just adding 1)?

#include <fstream>
#include <iostream>
#include <string>
#include <vector>

using namespace std;

int main()
{
    // open text file for input
    string file_name;

    cout << "please enter file name: ";
    cin  >> file_name;

    // associate the input file stream with a text file
    ifstream infile(file_name.c_str());

    // error checking for a valid filename
    if ( !infile ) {
        cerr << "Unable to open file "
             << file_name << " -- quitting!\n";
        return( -1 );
        }
        else cout << "\n";

    // some data structures to perform the function
    vector<string> lines_of_text;
    string textline;

    // read in text file, line by line
    while (getline( infile, textline, '\n' ))   {
        // add the new element to the vector
        lines_of_text.push_back( textline );

        // print the 'back' vector element - see the STL documentation
        cout << "line read: " << lines_of_text.back() << "\n";
    }
cout<<lines_of_text.size();
    return 0;
}

Answer 1

The code you have is sound. Here's a small test case that might help:

void read_lines(std::istream& input) {
  using namespace std;
  vector<string> lines;
  for (string line; getline(input, line);) {
    lines.push_back(line);
    cout << "read: " << lines.back() << '\n';
  }
  cout << "size: " << lines.size() << '\n';
}

int main() {
  {
    std::istringstream ss ("abc\n\n");
    read_lines(ss);
  }
  std::cout << "---\n";
  {
    std::istringstream ss ("abc\n123\n");
    read_lines(ss);
  }
  std::cout << "---\n";
  {
    std::istringstream ss ("abc\n123");  // last line missing newline
    read_lines(ss);
  }
  return 0;
}

Output:

read: abc
read: 
size: 2
---
read: abc
read: 123
size: 2
---
read: abc
read: 123
size: 2

Answer 2

I think I have tracked down the source of your problem. In Code::Blocks, a completely empty file will report that there is 1 line in it (the current one) in the gizmo on the status bar at the bottom of the IDE. This means that were you actually to enter a line of text, it would be line 1. In other words, Code::Blocks will normally over-report the number of actual lines in a file. You should never depend on CB, or any other IDE, to find out info on files - that's not what they are for.

Answer 3

Well, if the last line of your file is just '\\n', you don't push it into the vector. If you want it to be there, change the loop to:

while (getline( infile, textline, '\n' ).gcount() > 0) 
{
    if (infile.fail()) break; //An error occurred, break or do something else

    // add the new element to the vector
    lines_of_text.push_back( textline );

    // print the 'back' vector element - see the STL documentation
    cout << "line read: " << lines_of_text.back() << "\n";
}

Use the gcount() member to check how many characters were read in the last read - this will return 1 if it only read a delimiter character.

Answer 4

Ok so here is an explanation that you will hopefully understand. Your code should work fine if the file we're talking about doesn't end with newline. But what if it does? Let's say it looks like this:

"line 1"
"line 2"
""

Or as a sequence of characters:

line 1\nline 2\n

This file has THREE lines -- the last one being empty but it's there. After calling getline twice, you've read all the characters from the file. The third call to getline will say oops, end of file, sorry no more characters so you'll see only two lines of text.