Transcript ppt - CS

Networking with Java 1:
The Client Side
Introduction to Networking
Protocols
Hi
Hi
TCP connection
request
Client
Got the
time?
TCP connection
Server
reply
GET http://www.cs.huji.ac.il/~dbi
2:00
<file>
time
DBI 2008
HUJI-CS
3
Networking as Layers
DATA
Application (HTTP, FTP)
HEADER
DATA
HEADER
HEADER
DATA
HEADER
HEADER
DATA
Transport (TCP,UDP)
Network (IP)
Link (LINK)
DBI 2008
HEADER
HUJI-CS
4
TCP (Transmission-Control Protocol)
• Enables symmetric byte-stream transmission
between two endpoints (applications)
• Reliable communication channel
• TCP perform these tasks:
– connection establishment by handshake (relatively
slow)
– division to numbered packets (transferred by IP)
– error correction of packets (checksum)
– acknowledgement and retransmission of packets
– connection termination by handshake
DBI 2008
HUJI-CS
5
UDP (User Datagram Protocol)
• Enables direct datagram (packet) transmission
from one endpoint to another
• No reliability (except for data correction)
– sender does not wait for acknowledgements
– arrival order is not guaranteed
– arrival is not guaranteed
• Used when speed is essential, even in cost of
reliability
– e.g., streaming media, games, Internet telephony, etc.
DBI 2008
HUJI-CS
6
Ports
• A computer may have several applications that
communicate with applications on remote
computers through the same physical connection
to the network
• When receiving a packet, how can the computer
tell which application is the destination?
• Solution: each channel endpoint is assigned a
unique port that is known to both the computer
and the other endpoint
DBI 2008
HUJI-CS
7
Ports (cont)
• Thus, an endpoint application on the
Internet is identified by
– A host name → 32 bits IP-address
– A 16 bits port
• Why don’t we specify the port in a Web
browser?
DBI 2008
HUJI-CS
8
Known Ports
• Some known ports are
Client
Application
– 20, 21: FTP
– 22: SSH
mail client
– 23: TELNET
– 25: SMTP
– 110: POP3
– 80: HTTP
21
23
25
web browser
110 80 119
– 119: NNTP
DBI 2008
HUJI-CS
9
Sockets
• A socket is a construct that represents one endpoint of a two-way communication channel
between two programs running on the network
• Using sockets, the OS provides processes a filelike access to the channel
– i.e., sockets are allocated a file descriptor, and
processes can access (read/write) the socket by
specifying that descriptor
• A specific socket is identified by the machine's IP
and a port within that machine
DBI 2008
HUJI-CS
10
Sockets (cont)
• A socket stores the IP and port number of the
other end-point computer of the channel
• When writing to a socket, the written bytes are
sent to the other computer and port (e.g., over
TCP/IP)
– That is, remote IP and port are attached to the packets
• When OS receives packets on the network, it
uses their destination port to decide which socket
should get the received bytes
DBI 2008
HUJI-CS
11
Java Sockets
Low-Level Networking
Java Sockets
• Java wraps OS sockets (over TCP) by the objects
of class java.net.Socket
• new Socket(String remoteHost, int remotePort) creates
a TCP socket and connects it to the remote host
on the remote port (hand shake)
• Write and read using streams:
– InputStream getInputStream()
– OutputStream getOutputStream()
DBI 2008
HUJI-CS
13
A Socket Example
import java.net.*;
import java.io.*;
public class SimpleSocket {
public static void main(String[] args) throws IOException {
... next slide ...
}
}
DBI 2008
HUJI-CS
14
Socket socket = new Socket("www.cs.huji.ac.il", 80);
InputStream istream = socket.getInputStream();
OutputStream ostream = socket.getOutputStream();
String request =
"GET /~dbi/admin.html HTTP/1.1\r\n" +
"Host: www.cs.huji.ac.il\r\n" +
"Connection: close\r\n\r\n";
ostream.write(request.getBytes());
Needed for
forwarding for
example
byte[] response = new byte[4096]; int bytesRead = -1;
while ((bytesRead = istream.read(response)) >= 0) {
System.out.write(response, 0, bytesRead);
}
socket.close();
DBI 2008
HUJI-CS
15
Timeout
• You can set timeout values to blocking
method read() of Socket
• Use the method
socket.setSoTimeout(milliseconds)
• If timeout is reached before the method
returns, java.net.SocketTimeoutException is
thrown
Read more about Socket Class
DBI 2008
HUJI-CS
16
Java Sockets and HTTP
HTTP Message Structure
• A HTTP message has the following
structure:
Request/Status-Line \r\n
Header1: value1 \r\n
Header2: value2 \r\n
...
HeaderN: valueN \r\n
\r\n
Message-Body
DBI 2008
HUJI-CS
18
Reading HTTP Messages
• Several ways to interpret the bytes of the
body
– Binary: images, compressed files, class files, ...
– Text: ASCII, Latin-1, UTF-8, ...
• Commonly, applications parse the headers of the
message, and process the body according to the
information supplied by the headers
– E.g., Content-Type, Content-Encoding, TransferEncoding
DBI 2008
HUJI-CS
19
An Example
DBI 2008
HUJI-CS
20
Parsing the Headers
• So how are the headers themselves
represented?
• Headers of a HTTP message must be in
US-ASCII format (1 byte per character)
DBI 2008
HUJI-CS
21
Example: Extracting the Headers
Socket socket = new Socket(argv[0], 80);
InputStream istream = socket.getInputStream();
OutputStream ostream = socket.getOutputStream();
String request = "GET / HTTP/1.1\r\n" +
"Host: " + argv[0] + "\r\n" +
"Connection: close\r\n\r\n";
ostream.write(request.getBytes());
StringBuffer headers = new StringBuffer(); int byteRead = 0;
while ( !endOfHeaders(headers) &&
(byteRead = istream.read()) >= 0) {
headers.append((char) byteRead); }
System.out.print(headers);
socket.close();
DBI 2008
HUJI-CS
22
Example: Extracting the Headers (cont)
public static boolean endOfHeaders(StringBuffer headers) {
int lastIndex = headers.length() - 1;
if (lastIndex < 3 || headers.charAt(lastIndex) != '\n')
return false;
return (headers.substring(lastIndex - 3, lastIndex + 1)
.equals("\r\n\r\n"));
}
• Why did we (inefficiently) read one byte at a time?
• Is there any way to avoid this inefficiency?
DBI 2008
HUJI-CS
23
Persistent Connections
• According to HTTP/1.1, a server does not have to
close the connection after fulfilling your request
• One connection (socket) can be used for several
requests and responses send more requests
– even while earlier responses are being transferred
(pipelining)
– saves “slow start” time
• How can the client know when one response
ends and a new one begins?
• To avoid persistency, require explicitly by the
header Connection: close
DBI 2008
HUJI-CS
24
Parsing URLs
Working with URLs
• URL (Uniform/Universal Resource Locator): a
reference (address) to a resource on the Internet
http://www.cs.huji.ac.il:80/~dbi/main.html#notes
Protocol
Host
Name
Port
Number
File
Name
Reference
Query
http://www.google.com/search?hl=en&q=dbi+huji&btnG=Search
DBI 2008
HUJI-CS
26
The Class URL
• The class URL is used for parsing URLs
• Constructing URLs:
– URL w3c1 = new URL("http://www.w3.org/TR/");
– URL w3c2 = new URL("http","www.w3.org",80,"TR/");
– URL w3c3 = new URL(w3c2, "xhtml1/");
• If the string is not an absolute URL, then it is
considered relative to the URL
DBI 2008
HUJI-CS
27
Parsing URLs
• The following methods of URL can be used
for parsing URLs
getProtocol(), getHost(), getPort(),
getPath(), getFile(), getQuery(), getRef()
Read more about URL Class
DBI 2008
HUJI-CS
28
URLEncoder
• Contains a utility method encode for converting a
string into an encoded format (used in URLs, e.g.
for searches)
• To convert a string, each char is examined in turn:
– Space is converted into a plus sign +
– a-z, A-Z, 0-9, ., -, * and _ remain the same.
– The bytes of all special characters are replaced by
hexadecimal numbers, preceded with %
• To decode an encoded string, use decode() of the
class URLDecoder
Read more about URLEncoder Class
DBI 2008
HUJI-CS
29
Class URLConnection
High-Level Networking
The class URLConnection
• To establish the actual resource, we can use the
object URLConnection obtained by
url.openConnection()
• If the protocol of the URL is HTTP, the returned
object is of class HttpURLConnection
• This class encapsulates all socket management
and HTTP directions required to obtain the
resource
Read more about URLConnection Class
DBI 2008
HUJI-CS
31
public class ContentExtractor {
public static void main(String[] argv) throws Exception {
URL url = new URL(argv[0]);
System.out.println("Host: " + url.getHost());
System.out.println("Protocol: " + url.getProtocol());
System.out.println("----");
URLConnection con = url.openConnection();
InputStream stream = con.getInputStream();
byte[] data = new byte[4096]; int bytesRead = 0;
while((bytesRead=stream.read(data))>=0) {
System.out.write(data,0,bytesRead);
}}}
DBI 2008
HUJI-CS
32
About URLConnection
• The life cycle of a URLConnection object has two
parts:
– Before actual connection establishment
• Connection configuration
– After actual connection establishment
• Content retrieval
• Passage from the first phase to the second is
implicit
– A result of calling some committing methods, like
getDate()
DBI 2008
HUJI-CS
33
About HttpURLConnection
• The HttpURLConnection class encapsulates all
HTTP transaction over sockets, e.g.,
– Content decoding
– Redirection
– Proxy indirection
• You can control requests by its methods
– setRequestMethod, setFollowRedirects,
setRequestProperty, ...
Read more about HttpURLConnection Class
DBI 2008
HUJI-CS
34