Some bytes are missing after reading an std::ifstream to a vector of lines

Question

For copying what I read from input file to vector, I used std::copy() as recommended in Reading an std::ifstream to a vector of lines.

The problem occurs if I use:

std::copy(std::istream_iterator<unsigned char>(inputfile),
          std::istream_iterator<unsigned char>(),
          std::back_inserter(myVector));

The 16th byte of my file is missing in the myVector variable.

But if I use the following code:

inputfile.read((char*)&myVector[0], sizeof(int)*getfilesize(nameOfFile));

Then the byte is not missing anymore.

I am trying to parse WAV files and I lost too much time on this, I hope I will learn something new out of this. Can you please tell me what is wrong with the first version of the code above?

The first version is using a formatted input function, thus skipping e.g. what it interprets as whitespaces. — dyp, Jul 21 '13 at 19:32
Yes , I didn't changed anything except above. I used ifstream like; std::ifstream inputfile(nameOfFile.c_str(),std::ifstream::in|std::ifstream::binary); — Kadir Erdem Demir, Jul 21 '13 at 19:40
Note also how the example you referenced deals with a file of "formatted" numbers, i.e. an integer represented as a sequence of (probably) ASCII characters, whereas the WAV file format contains a lot of binary representations. — dyp, Jul 21 '13 at 19:47
Note IMO this `sizeof(int)*getfilesize(nameOfFile)` as the second argument of `read` is error-prone. Better use something like `myVector.size() * sizeof(myVector[0])`. — dyp, Jul 21 '13 at 19:49

RiaD · Accepted Answer · 2013-07-21T20:03:33.310

7

istream_iterator uses operator >> to read elements, but operator >> skip whitespaces.

You may try using noskipws

inputfile >> noskipws;

§ 24.6.1 p1. (my emphasis)

The class template istream_iterator is an input iterator (24.2.3) that reads (using operator>>) successive elements from the input stream for which it was constructed....

edited Jul 21 '13 at 20:03

answered Jul 21 '13 at 19:43

RiaD

46,822
11
79
123

... iff the skipws flag is set, which it is per default. – dyp Jul 21 '13 at 19:44
1

A better solution if you want to read binary bytes. might be to use `std::istreambuf_iterator`. – James Kanze Jul 21 '13 at 20:53

score 5 · Answer 2 · answered Jul 21 '13 at 19:49

5

Like RiaD said, istream_iterator performs formatted input via operator >>. The solution is to use unformatted reading on the underlying buffer. To do this, use istreambuf_iterator:

std::copy(std::istreambuf_iterator<char>(inputfile),
          std::istreambuf_iterator<char>(),
          std::back_inserter(myVector));

answered Jul 21 '13 at 19:49

Konrad Rudolph

530,221
131
937
1,214

1

Shouldn't it be `char` for the `std::ifstream`? – dyp Jul 21 '13 at 19:51
@DyP Right, I *always* make that mistake. – Konrad Rudolph Jul 21 '13 at 19:52

score 1 · Answer 3 · answered Jul 21 '13 at 19:43

First of all wav file is a binary data, so you should treat it as such, you should open your file in binary mode:

ifstream ifs;
ifs.open ("test.wav", ifstream::in | ifstream::binary);

Then, you have to use ordinary read function that works as you stated.

ifstream documentation

Some bytes are missing after reading an std::ifstream to a vector of lines

3 Answers3