How to copy a file into a vector? (C++)

[Kyral]

Hey all.

For a program I'm writing, I need to copy a file character for character into a vector of type char. That includes spaces and endlines. Thing is I kinda don't know how to do it. I was thinking that I could do


vector.push_back ( file );

But I'm not sure. Thanks for your help!

PS. If anyone could explain the system() function to me it would be great

[LordHunter317]

The most trivial way would be to do something like:
#include <algorithm>
#include <fstream>
#include <iterator>
#include <vector>

int
main()
{
typedef std::istream_iterator<char> istream_iterator;
std::ifstream file("example.txt");
std::vector<char> input;

file >> std::noskipws;
std::copy(istream_iterator(file), istream_iterator(),
std::back_inserter(input));
}Really, the file >> std::noskipws; is the critical operation here: it turns off whitespace skipping on formatted input streams. Using copy is just convient for slurping a whole file.

[bihe]

read a file and process every single char:


#include <iostream>
#include <fstream>
using namespace std;

int main()
{
int blank_count = 0;
int char_count = 0;
int sentence_count = 0;
char ch;

ifstream iFile("file.txt");

if (! iFile)
{
cout << "Error opening input file" << endl;
return -1;
}

while (iFile.get(ch))
{
switch (ch) {
case ' ':
blank_count++;
break;
case '\n':
break;
case '.':
sentence_count++;
break;
default:
char_count++;
break;
}
}

cout << "There are " << blank_count << " blanks" << endl;
cout << "There are " << char_count << " characters" << endl;
cout << "There are " << sentence_count << " sentences" << endl;

return 0;
}


google ... (http://www.google.com/search?q=c%2B%2B+read+file&start=0&ie=utf-8&oe=utf-8)

[LordHunter317]

Your posted code has a problem: that check won't capture all whitespace. One should use isspace() instead.

I should point out that even though my example uses char, unless you're going to completely ignore lines in whatever you're going to do, reading into a vector<std::string> might be more wortwhile, as then doing line based manipulations is eaiser.

Though, if you're going to do that, slurping into a strstream may be the best solution. Then you can treat your in-memory copy just like any other I/O stream.

[thumper]

The most trivial way would be to do something like:
#include <algorithm>
#include <fstream>
#include <iterator>
#include <vector>

int
main()
{
typedef std::istream_iterator<char> istream_iterator;
std::ifstream file("example.txt");
std::vector<char> input;

file >> std::noskipws;
std::copy(istream_iterator(file), istream_iterator(),
std::back_inserter(input));
}Really, the file >> std::noskipws; is the critical operation here: it turns off whitespace skipping on formatted input streams. Using copy is just convient for slurping a whole file.An alternative is to use the streaming of the file buffer directly using the rdbuf. This way goes through a temporary string, and constructs the vector from that:

#include <vector>
#include <fstream>
#include <sstream>
#include <iostream>

int main(int argc, char* argv[])
{
if (argc < 2) {
std::cout << "usage: " << argv[0] << " <filename>\n";
return 2;
}

std::ifstream fin(argv[1]);
if (fin) {
std::stringstream ss;
// this copies the entire contents of the file into the string stream
ss << fin.rdbuf();
// get the string out of the string stream
std::string contents = ss.str();
std::cout << contents;
// construct the vector from the string.
std::vector<char> buff(contents.begin(), contents.end());
}
else {
std::cout << "Couldn't open " << argv[1] << "\n";
return 1;
}
return 0;
}

[Kyral]

Wow I didn't expect this much response....

[LordHunter317]

An alternative is to use the streaming of the file buffer directly using the rdbuf.I mentioned that, though didn't post code ;), and called it a strstream for some reason :(

However, doing that for the sole purpose of putting the data in a vector is pointless, and requires 2X the RAM to complete the copy, to boot.

However, for about any other task where you're not doing line-based I/O, this is probably a superior solution. :)

Though for all practical textual I/O purposes I do, I end up using boost::tokenizer or something wrapped around it, which makes dealing with textual data intuitive and generally easy, for a lot of formats.

[jerome bettis]

i think the answer to most questions like this is "look at the boost libraries". and it's usually a very good answer.

[LordHunter317]

Well, I can think of a lot of I/O cases where boost::tokenizer would be useless, even for formatted I/O. XML is the most obvious and biggest one: a traditional tokenizer (e.g., boost::tokenizer) is rather inadequate here.

That being said, boost::tokenizer is a templated tokenizer, so you could write your template that parsed XML in a more sane fashion.

However, dealing with XML in a stream like fashion almost never make sense, ignoring a few cases.

But yes, boost:: is wonderful and anyone beginning or experienced with C++ should always check there first before writing any code.

[jerome bettis]

But yes, boost:: is wonderful and anyone beginning or experienced with C++ should always check there first before writing any code.

right that's really all i was trying to say. whether or not it will be a help to him here i have no idea. in fact i have no idea why you would actually need to do this but whatever. i would be interested in hearing why he's doing this

btw you just seem to know everything about everything programming all the time, and that deserves a lot of respect.

[thumper]

But yes, boost:: is wonderful and anyone beginning or experienced with C++ should always check there first before writing any code.C'mon, any experienced C++ person should already know what it has to offer and shouldn't need to check every time 8-)

[LordHunter317]

in fact i have no idea why you would actually need to do this but whatever.I can think of a few reasons. Mostly dealing with statistical distribution of letters/words in a file. Doing a lot of searching is easier in a vector than a string.


i would be interested in hearing why he's doing thisAye. Me as well.


btw you just seem to know everything about everything programming all the time, and that deserves a lot of respect.Not hardly ;)

[Kyral]

right that's really all i was trying to say. whether or not it will be a help to him here i have no idea. in fact i have no idea why you would actually need to do this but whatever. i would be interested in hearing why he's doing this

btw you just seem to know everything about everything programming all the time, and that deserves a lot of respect.


I can show you why. If you would just follow the link to the project page...

http://www.clarkson.edu/projects/cosi/fa2005/students/petermcv/

Tis the second one :P

Comments welcome, flames will be redirected to /dev/null :P

[thumper]

That doesn't enlighten any. What does the testing procedure have to do with copying a file into a vector?

Why do it in C++?

[Kyral]

1) I only know C++ right now

2) The program reads a file into the vector to act as a buffer, then outputs it into script to be echoed in the shell script

[LordHunter317]

It's silly to read it in to a vector just as a buffer, and especially silly to do it as a vector<char>. Use a stringstream, as thumper showed how. That gives you I/O streams semantics with in-memory storage.

But if you're just going to write it out to a script, I don't see the point.

Anyway, what you want to do can easily be accomplish by rdiff-backup, a simple command to pull the package list, and some glue.