Servlets

1-Nov-13

Servers

A server is a computer that responds to requests from a client

Typical requests: provide a web page, upload or download a file, send email

A server is also the software that responds to these requests; a client could be the browser or other software making these requests Typically, your little computer is the client, and someone elses big computer is the server

However, any computer can be a server It is not unusual to have server software and client software running on the same computer

Apache

Apache is a very popular server

66% of the web sites on the Internet use Apache Full-featured and extensible Efficient Robust Secure (at least, more secure than other servers) Up to date with current standards Open source Free

Apache is:

Why use anything else?

3

Ports

A port is a connection between a server and a client

Ports are identified by positive integers A port is a software notion, not a hardware notion, so there may be very many of them Typical port numbers:

A service is associated with a specific port

21FTP, File Transfer Protocol 22SSH, Secure Shell 25SMTP, Simple Mail Transfer Protocol 53DNS, Domain Name Service 80HTTP, Hypertext Transfer Protocol 8080HTTP (used for testing HTTP) 7648, 7649CU-SeeMe 27960Quake III

These are the ports of most interest to us

Ports II

My UPenn Web page is:

[Link]

But it is also:
[Link]

The http: at the beginning signifies a particular protocol (communication language), the Hypertext Transfer Protocol The :80 specifies a port By default, the Web server listens to port 80

The Web server could listen to any port it chose This could lead to problems if the port was in use by some other server For testing servlets, we typically have the server listen to port 8080 If I had sent it to some other port, say, 99, my request would either go unheard, or would (probably) not be understood
5

In the second URL above, I explicitly sent my request to port 80

CGI Scripts

CGI stands for Common Gateway Interface

Client sends a request to server Server starts a CGI script client

server
script

Script computes a result for server and quits

Server returns response to client Another client sends a request

client

Server starts the CGI script again

Etc.

Servlets

A servlet is like an applet, but on the server side

Client sends a request to server Server starts a servlet client

server
servlet

Servlet computes a result for server and does not quit

Server returns response to client Another client sends a request

client

Server calls the servlet again

Etc.

Servlets vs. CGI scripts

Advantages:

Running a servlet doesnt require creating a separate process each time A servlet stays in memory, so it doesnt have to be reloaded each time There is only one instance handling multiple requests, not a separate instance for every request Untrusted servlets can be run in a sandbox Servlets must be in Java (CGI scripts can be in any language)
8

Disadvantage:

Tomcat

Tomcat is the Servlet Engine than handles servlet requests for Apache

Tomcat is a helper application for Apache Its best to think of Tomcat as a servlet container Apache can be installed without Tomcat Tomcat can be installed without Apache

Apache can handle many types of web services

Its easier to install Tomcat standalone than as part of Apache

By itself, Tomcat can handle web pages, servlets, and JSP

Apache and Tomcat are open source (and therefore free)

Servlets

A servlet is any class that implements the [Link] interface

In practice, most servlets extend the [Link] class Some servlets extend [Link] instead

Servlets, like applets, usually lack a main method, but must implement or override certain other methods

10

What does a servlet do?

1. Read any data sent by the user 2. Look up any information about the request that is embedded in the HTTP request 3. Generate the results 4. Format the results inside a document 5. Set the appropriate HTTP parameters 6. Send the document back to the client

From: Core Servlets and JavaServerPages, by Marty Hall

11

Important servlet methods, I

When a servlet is first started up, its init(ServletConfig config) method is called

init should perform any necessary initializations init is called only once, and does not need to be thread-safe

Every servlet request results in a call to service(ServletRequest request, ServletResponse response)

service calls another method depending on the type of service requested Usually you would override the called methods of interest, not service itself service handles multiple simultaneous requests, so it and the methods it calls must be thread safe
destroy is called only once, but must be thread safe (because other threads may still be running)
12

When the servlet is shut down, destroy() is called

HTTP requests

When a request is submitted from a Web page, it is almost always a GET or a POST request The HTTP <form> tag has an attribute action, whose value can be "get" or "post" The "get" action results in the form information being put after a ? in the URL

Example: [Link] The & separates the various parameters Only a limited amount of information can be sent this way The information is in the body of the HTTP request
13

"post" can send large amounts of information

Important servlet methods, II

The service method dispatches (sends to another method) the following kinds of requests: DELETE, GET, HEAD, OPTIONS, POST, PUT, and TRACE

A GET request is dispatched to doGet(HttpServletRequest request, HttpServletResponse response) A POST request is dispatched to doPost(HttpServletRequest request, HttpServletResponse response) These are the two methods you will usually override

Regardless of whether the client sends a GET or a POST request, the information it sends is in the HttpServletRequest parameter

doGet and doPost typically do the same thing, so usually you do the real work in one, and have the other just call it public void doGet(HttpServletRequest request, HttpServletResponse response) { doPost(request, response); }

14

Less important requests

HEAD

Asks for only the header part of what a GET would return Asks to send the received request message back (for debugging) Says to put the body at the requested URL Says to delete the resource or file at the requested URL Asks for a list of the HTTP methods to which the thing at the requested URL can respond Says to connect for purposes of tunneling
15

TRACE

PUT

DELETE

OPTIONS

CONNECT

Each of these (except CONNECT) an associated doXXX method

A Hello World servlet

(from the Tomcat installation documentation)

public class HelloServlet extends HttpServlet { public void doGet(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException { [Link]("text/html"); PrintWriter out = [Link](); String docType = "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0 " + "Transitional//EN\">\n"; [Link](docType + "<HTML>\n" + "<HEAD><TITLE>Hello</TITLE></HEAD>\n" + "<BODY BGCOLOR=\"#FDF5E6\">\n" + "<H1>Hello World</H1>\n" + "</BODY></HTML>"); } Dont worry, well take this a little at a time! }
16

The superclass

public class HelloServlet extends HttpServlet {

Every class must extend GenericServlet or a subclass of GenericServlet

GenericServlet is protocol independent, so you could write a servlet to process any protocol In practice, you almost always want to respond to an HTTP request, so you extend HttpServlet

A subclass of HttpServlet must override at least one method, usually one doGet, doPost, doPut, doDelete, init and destroy, or getServletInfo
17

The doGet method

public void doGet(HttpServletRequest request, HttpServletResponse response) throws ServletException, IOException { This method, doGet, services a GET request The method uses request to get the information that was sent to it The method does not return a value; instead, it uses response to get an I/O stream, and outputs its response Since the method does I/O, it can throw an IOException Any other type of exception must be encapsulated as a ServletException The doPost method works exactly the same way

18

Parameters to doGet and doPost

Input is from the HttpServletRequest parameter

Our first example doesnt get any input, so well discuss this a bit later

Output is via the HttpServletResponse object, which we have named response

I/O in Java is very flexible but also quite complex, so the response object acts as an assistant

19

doGet and doPost

Actual differences between doGet and doPost:

You can send only a small amount of data with doGet Data sent via doGet is visible in the URL Pages returned by doPost cannot easily be bookmarked
doGet is a request for information; it does not (or should not)

Intended difference between doGet and doPost:

change anything on the server

Jargon term: doGet should be idempotent

doPost provides information (such as placing an order for

merchandise) that the server is expected to remember

20

Using the HttpServletResponse

The second parameter to doGet (or doPost) is HttpServletResponse response Everything sent via the Web has a MIME type The first thing we must do with response is set the MIME type of our reply: [Link]("text/html");

This tells the client to interpret the page as HTML

Because we will be outputting character data, we need a PrintWriter, handily provided for us by the getWriter method of response: PrintWriter out = [Link](); Now were ready to create the actual page to be returned

21

Using the PrintWriter

From here on, its just a matter of using our PrintWriter, named out, to produce the Web page First we create a header string:
String docType = "<!DOCTYPE HTML PUBLIC \"-//W3C//DTD HTML 4.0 " + "Transitional//EN\">\n"; This line is technically required by the HTML spec Browsers mostly dont care, but HTML validators do care

Then use the println method of out one or more times

[Link](docType + "<HTML>\n" + "<HEAD> ... </BODY></HTML>");

And were done!

22

Input to a servlet

A GET request supplies parameters in the form

URL ? name=value & name=value & name=value (Illegal spaces added to make it more legible) Actual spaces in the parameter values are encoded by + signs Other special characters are encoded in hex; for example, an ampersand is represented by %26

Parameter names can occur more than once, with different values A POST request supplies parameters in the same syntax, only it is in the body section of the request and is therefore harder for the user to see

23

Getting the parameters

Input parameters are retrieved via messages to the HttpServletRequest object request

Most of the interesting methods are inherited from the superinterface ServletRequest
Returns an Enumeration of the parameter names

public Enumeration getParameterNames()

An Enumeration is like an Iterator, only older

If no parameters, returns an empty Enumeration Returns the value of the parameter name as a String If the parameter doesnt exist, returns null If name has multiple values, only the first is returned Returns an array of values of the parameter name If the parameter doesnt exist, returns null
24

public String getParameter(String name)

public String[] getParameterValues(String name)

Enumeration review

An Enumeration is almost the same as Iterator

Its an older class, and the names are longer

Iterator boolean hasNext() Object next() void remove()

Enumeration boolean hasMoreElements() Object nextElement() --

Example use:

Enumeration e = [Link](); while ([Link]()) { [Link]([Link]()); }

25

Example of input parameters

public void doGet(HttpServletRequest request, HttpServletResponse response) { ... stuff omitted ... [Link]("<H1>Hello"); String names[] = [Link]("name"); if (names != null) for (int i = 0; i < [Link]; i++) [Link](" " + names[i]); [Link]("!"); }
26

Java review: Data from Strings

All parameter values are retrieved as Strings Frequently these Strings represent numbers, and you want the numeric value

int n = new Integer(param).intValue(); double d = new Double(param).doubleValue(); byte b = new Byte(param).byteValue();

Similarly for short, float, and long These can all throw a NumberFormatException, which is a subclass of RuntimeException

boolean p = new Boolean(param).booleanValue();

char c = [Link](0);
27

But:

Whats left?

Weve covered enough so far to write simple servlets, but not enough to write useful servlets

We still need to be able to:

Configure the server to recognize the servlet! Use other configuration information Authenticate users Keep track of users during a session Retain information across different sessions Make sure our servlets are thread safe Communicate between servlets

But remember: The most difficult program in any language is Hello World!
28

The End

29