Downloading file via TcpClient from HTTP server - http

Question

I`m learning HTTP, and decide to use TcpClient for downloading file from internet.
private void requireBtn_Click(object sender, EventArgs e) {
var client = new TcpClient(addressBox.Text, 80);
var networkStream = client.GetStream();
var sWriter = new StreamWriter(networkStream, Encoding.Default);
foreach (string querLine in queryBox.Lines) {
sWriter.WriteLine(querLine);
}
sWriter.WriteLine();
sWriter.Flush();
string responceText = "";
var sReader = new StreamReader(networkStream);
string respLine;
while ((respLine = sReader.ReadLine()).Length > 0) {
responceText += respLine + "\r\n";
}
responceBox.Text = responceText;
Regex reContentLength = new Regex(#"(?<=Content-Length:\s)\d+", RegexOptions.IgnoreCase);
Int32 contentLength = Int32.Parse(reContentLength.Match(responceText).Value);
this.Text = contentLength.ToString();
if (networkStream.CanRead) {
var fileStream = new FileStream(#"C:\img.png", FileMode.Create);
byte[] buffer = new byte[1024];
int numberOfBytesRead = 0;
do {
numberOfBytesRead = networkStream.Read(buffer, 0, buffer.Length);
fileStream.Write(buffer, 0, numberOfBytesRead);
}
while (networkStream.DataAvailable);
fileStream.Flush();
fileStream.Close();
MessageBox.Show("Done!");
}
else
MessageBox.Show("Fail!");
client.Close();
}
But downloaded file is corrupted. I can't understand why...
Source file

Answer 1

private void requireBtn_Click(object sender, EventArgs e) {
string host = "atomAltera.SkyHost.ge";
string query = "GET /mems.txt HTTP/1.1\r\n" +
"Host: atomAltera.SkyHost.ge\r\n" +
"User-Agent: NuclightWeb\r\n" +
"Connection: close\r\n"+
"\r\n";
var client = new Socket(AddressFamily.InterNetwork, SocketType.Stream, ProtocolType.IP);
client.Connect(host, 80);
var networkStream = new NetworkStream(client);
var bytes = Encoding.Default.GetBytes(query);
networkStream.Write(bytes, 0, bytes.Length);
var bReader = new BinaryReader(networkStream, Encoding.Default);
string responce = "";
string line;
char c;
do {
line = "";
c = '\u0000';
while (true) {
c = bReader.ReadChar();
if (c == '\r')
break;
line += c;
}
c = bReader.ReadChar();
responce += line + "\r\n";
} while (line.Length > 0);
responceBox.Text = responce;
Regex reContentLength = new Regex(#"(?<=Content-Length:\s)\d+", RegexOptions.IgnoreCase);
Int32 contentLength = Int32.Parse(reContentLength.Match(responce).Value);
this.Text = contentLength.ToString();
var fileStream = new FileStream(#"C:\m.txt", FileMode.Create);
byte[] buffer = new byte[4 * 1024];
int n = 0;
int read = 0;
while (n < contentLength) {
if (networkStream.DataAvailable) {
read = networkStream.Read(buffer, 0, buffer.Length);
n += read;
fileStream.Write(buffer, 0, read);
}
}
fileStream.Flush();
fileStream.Close();
client.Close();
}
It Works! May be StreamReader reads superfluous bytes?

Answer 2

I suggest using contentLength instead of networkStream.DataAvailable to detect when the data transfer is complete.
DataAvailable simply means that there is data received by the OS stack, however depending on the network routing and fragmentation it might not have arrived by the time you're checking the condition, so you stop receiving that data too early.
To prove or disprove this, you can do a binary file comparison between what your program has downloaded versus what a normal browser would download. If the former is truncated, then we've found the problem.

Answer 3

You need to use the value from the Content-Length header (which can be larger than a 32-bit integer, btw). Keep reading until that many bytes have been received.
However, it is possible for the server to not use a Content-Length header at all. It may use a Transfer-Encoding header that includes the chunked coding, or it may not use any header at all. HTTP defines several different ways to determine the length of the data, and you should support as many as you can in order to account for all kinds of servers. See RFC 2616 Sections 4.3 and 4.4 for the particular rules you need to follow.