Author image Ralf S. Engelschall


LWP - Library for WWW access in Perl


  use LWP;
  print "This is libwww-perl-$LWP::VERSION\n";


Libwww-perl is a collection of Perl modules which provides a simple and consistent programming interface (API) to the World-Wide Web. The main focus of the library is to provide classes and functions that allow you to write WWW clients, thus libwww-perl said to be a WWW client library. The library also contain modules that are of more general use.

The main architecture of the library is object oriented. The user agent, requests sent and responses received from the WWW server are all represented by objects. This makes a simple and powerful interface to these services. The interface should be easy to extend and customize for your needs.

The main features of the library are:

  • Contains various reusable components (modules) that can be used separately or together.

  • Provides an object oriented model of HTTP-style communication. Within this framework we currently support access to http, gopher, ftp, news, file, and mailto resources.

  • The library be used through the full object oriented interface or through a very simple procedural interface.

  • Support the basic and digest authorization schemes.

  • Transparent redirect handling.

  • Supports access through proxy servers.

  • URL handling (both absolute and relative URLs are supported).

  • A parser for robots.txt files and a framework for constructing robots.

  • An experimental HTML parser and formatters (for PostScript and plain text).

  • The library can cooperate with Tk. A simple Tk-based GUI browser called 'tkweb' is distributed with the Tk extension for perl.

  • An implementation of the HTTP content negotiation algorithm that can be used both in protocol modules and in server scripts (like CGI scripts).

  • A simple command line client application called lwp-request.


The libwww-perl library is based on HTTP style communication. This section try to describe what that means.

Let us start with this quote from the HTTP specification document <URL:>:

  • The HTTP protocol is based on a request/response paradigm. A client establishes a connection with a server and sends a request to the server in the form of a request method, URI, and protocol version, followed by a MIME-like message containing request modifiers, client information, and possible body content. The server responds with a status line, including the message's protocol version and a success or error code, followed by a MIME-like message containing server information, entity meta-information, and possible body content.

What this means to libwww-perl is that communication always take place through these steps: First a request object is created and configured. This object is then passed to a server and we get a response object in return that we can examine. A request is always independent of any previous requests, i.e. the service is stateless. The same simple model is used for any kind of service we want to access.

For example, if we want to fetch a document from a remote file server, then we send it a request that contains a name for that document and the response will contain the document itself. If we access a search engine, then the content of the request will contain the query parameters and the response will contain the query result. If we want to send a mail message to somebody then we send a request object which contains our message to the mail server and the response object will contain an acknowledgment that tells us that the message has been accepted and will be forwarded to the recipient(s).

It is as simple as that!

The Request Object

The request object has the class name HTTP::Request in libwww-perl. The fact that the class name use HTTP:: as a name prefix only implies that we use the HTTP model of communication. It does not limit the kind of services we can try to pass this request to. For instance, we will send HTTP::Requests both to ftp and gopher servers, as well as to the local file system.

The main attributes of the request objects are:

  • The method is a short string that tells what kind of request this is. The most used methods are GET, PUT, POST and HEAD.

  • The url is a string denoting the protocol, server and the name of the "document" we want to access. The url might also encode various other parameters.

  • The headers contain additional information about the request and can also used to describe the content. The headers is a set of keyword/value pairs.

  • The content is an arbitrary amount of data.

The Response Object

The request object has the class name HTTP::Response in libwww-perl. The main attributes of objects of this class are:

  • The code is a numerical value that encode the overall outcome of the request.

  • The message is a short (human readable) string that corresponds to the code.

  • The headers contain additional information about the response and they also describe the content.

  • The content is an arbitrary amount of data.

Since we don't want to handle all possible code values directly in our programs, the libwww-perl response object have methods that can be used to query what kind of response this is. The most commonly used response classification methods are:


The request was was successfully received, understood or accepted.


The request failed. The server or the resource might not be available, access to the resource might be denied or other things might have failed for some reason.

The User Agent

Let us assume that we have created a request object. What do we actually do with it in order to receive a response?

The answer is that you pass it on to a user agent object and this object will take care of all the things that need to be done (low-level communication and error handling). The user agent will give you back a response object. The user agent represents your application on the network and it provides you with an interface that can accept requests and will return responses.

You should think about the user agent as an interface layer between your application code and the network. Through this interface you are able to access the various servers on the network.

The libwww-perl class name for the user agent is LWP::UserAgent. Every libwww-perl application that wants to communicate should create at least one object of this kind. The main method provided by this object is request(). This method takes an HTTP::Request object as argument and will (eventually) return a HTTP::Response object.

The user agent has many other attributes that lets you configure how it will interact with the network and with your application code.

  • The timeout specify how much time we give remote servers in creating responses before the library disconnect and creates an internal timeout response.

  • The agent specify the name that your application should use when it presents itself on the network.

  • The from attribute can be set to the e-mail address of the person responsible for running the application. If this is set, then the address will be sent to the servers with every request.

  • The use_alarm specify if it is OK for the user agent to use the alarm(2) system to implement timeouts.

  • The use_eval specify if the agent should raise an exception (die in perl) if an error condition occur.

  • The parse_head specify whether we should initialize response headers from the <head> section of HTML documents.

  • The proxy and no_proxy specify if and when communication should go through a proxy server. <URL:>

  • The credentials provide a way to set up user names and passwords that is needed to access certain services.

Many applications would want even more control over how they interact with the network and they get this by specializing the LWP::UserAgent by sub-classing. The library provide a specialization called LWP::RobotUA that is used by robot applications.

An Example

This example shows how the user agent, a request and a response are represented in actual perl code:

  # Create a user agent object
  use LWP::UserAgent;
  $ua = new LWP::UserAgent;
  $ua->agent("AgentName/0.1 " . $ua->agent);

  # Create a request
  my $req = new HTTP::Request POST => '';

  # Pass request to the user agent and get a response back
  my $res = $ua->request($req);

  # Check the outcome of the response
  if ($res->is_success) {
      print $res->content;
  } else {
      print "Bad luck this time\n";

The $ua is created once when the application starts up. New request objects are normally created for each request sent.


This section goes through the various protocol schemes and describe the HTTP style methods that are supported and the headers that might have any effect.

For all requests, a "User-Agent" header is added and initialized from the $ua->agent value before the request is handed to the network layer. In the same way, a "From" header is initialized from the $ua->from value.

For all responses, the library will add a header called "Client-Date". This header will encode the time when the response was received by your application. This format and semantics of the header is just like the server created "Date" header.

HTTP Requests

HTTP request are really just handed off to an HTTP server and it will decide what happens. Few servers implement methods beside the usual "GET", "HEAD", "POST" and "PUT" but CGI-scripts can really implement any method they like.

If the server is not available then the library will generate an internal error response.

The library automatically adds a "Host" and a "Content-Length" header to the HTTP request before it is sent over the network.

For GET request you might want to add the "If-Modified-Since" header to make the request conditional.

For POST request you should add the "Content-Type" header. When you try to emulate HTML <FORM> handling you should usually let the value of the "Content-Type" header be "application/x-www-form-urlencoded". See lwpcook for examples of this.

The libwww-perl HTTP implementation currently support the HTTP/1.0 protocol. HTTP/0.9 servers are also handled correctly.

The library allows you to access proxy server through HTTP. This means that you can set up the library to forward all types of request through the HTTP protocol module. See LWP::UserAgent for documentation of this.

FTP Requests

The library currently support GET, HEAD and PUT requests. GET will retrieve a file or a directory listing from an FTP server. PUT will store a file on a ftp server.

You can specify a ftp account for servers that want this in addition user name and password. This is specified by passing an "Account" header in the request.

User name/password can be specified using basic authorization or be encoded in the URL. Bad logins return an UNAUTHORIZED response with "WWW-Authenticate: Basic" and can be treated as basic authorization for HTTP.

The library support ftp ASCII transfer mode by specifying the "type=a" parameter in the URL.

Directory listings are by default returned unprocessed (as returned from the ftp server) with the content media type reported to be "text/ftp-dir-listing". The File::Listing module provide functionality for parsing of these directory listing.

The ftp module is also able to convert directory listings to HTML and this can be requested via the standard HTTP content negotiation mechanisms (add an "Accept: text/html" header in the request if you want this).

The normal file retrievals, the "Content-Type" is guessed based on the file name suffix. See LWP::MediaTypes.

The "If-Modified-Since" header is not honored yet.


  $req = HTTP::Request->new(GET => '');
  $req->header(Accept => "text/html, */*;q=0.1");

News Requests

Access to the USENET News system is implemented through the NNTP protocol. The name of the news server is obtained from the NNTP_SERVER environment variable and defaults to "news". It is not possible to specify the hostname of the NNTP server in the news:-URLs.

The library support GET and HEAD to retrieve news articles through the NNTP protocol. You can also post articles to newsgroups by using (surprise!) the POST method.

GET on newsgroups is not implemented yet.


  $req = HTTP::Request->new(GET => '');

  $req = HTTP::Request->new(POST => 'news:comp.lang.perl.test');
  $req->header(Subject => 'This is a test',
               From    => '');
  This is the content of the message that we are sending to
  the world.

Gopher Request

The library supports the GET and HEAD method for gopher request. All request header values are ignored. HEAD cheats and will return a response without even talking to server.

Gopher menus are always converted to HTML.

The response "Content-Type" is generated from the document type encoded (as the first letter) in the request URL path itself.


  $req = HTTP::Request->new('GET', 'gopher://');

File Request

The library supports GET and HEAD methods for file requests. The "If-Modified-Since" header is supported. All other headers are ignored. The host component of the file URL must be empty or set to "localhost". Any other host value will be treated as an error.

Directories are always converted to an HTML document. For normal files, the "Content-Type" and "Content-Encoding" in the response are guessed based on the file suffix.


  $req = HTTP::Request->new(GET => 'file:/etc/passwd');

Mailto Request

You can send (aka "POST") mail messages using the library. All headers specified for the request are passed on to the mail system. The "To" header is initialized from the mail address in the URL.


  $req = HTTP::Request->new(POST => '');
  $req->header("Subject", "subscribe");
  $req->content("Please subscribe me to the libwww-perl mailing list!\n");


This table should give you a quick overview of the classes provided by the library. Indentation shows class inheritance.

 LWP::MemberMixin   -- Access to member variables of Perl5 classes
   LWP::UserAgent   -- WWW user agent class
     LWP::RobotUA   -- When developing a robot applications
   LWP::Protocol          -- Interface to various protocol schemes
     LWP::Protocol::http  -- http:// access
     LWP::Protocol::file  -- file:// access
     LWP::Protocol::ftp   -- ftp:// access

 LWP::Socket        -- Socket creation and IO

 HTTP::Headers      -- MIME/RFC822 style header (used by HTTP::Message)
 HTTP::Message      -- HTTP style message
   HTTP::Request    -- HTTP request
   HTTP::Response   -- HTTP response
 HTTP::Daemon       -- A HTTP server class

 URI::URL           -- Uniform Resource Locators

 WWW::RobotRules    -- Parse robots.txt files
   WWW::RobotRules::AnyDBM_File -- Persistent RobotRules

 HTML::Parser       -- Parse HTML documents
   HTML::TreeBuilder-- Build a HTML syntax tree
   HTML::HeadParser -- Parse the <HEAD> section of a HTML document
   HTML::LinkExtor  -- Extract links from a HTML document
 HTML::Element      -- Building block for the HTML::TreeBuilder
 HTML::Formatter    -- Convert HTML syntax trees to readable formats
   HTML::FormatText -- Output is plain text
   HTML::FormatPS   -- Output is PostScript

The following modules provide various functions and definitions.

 LWP                -- This file.  Library version number and documentation.
 LWP::MediaTypes    -- MIME types configuration (text/html etc.)
 LWP::Debug         -- Debug logging module
 LWP::Simple        -- Simplified procedural interface for common functions
 HTTP::Status       -- HTTP status code (200 OK etc)
 HTTP::Date         -- Date parsing module for HTTP date formats
 HTTP::Negotiate    -- HTTP content negotiation calculation
 HTML::Entities     -- Expand or unexpand entities in HTML text
 File::Listing      -- Parse directory listings

HTTP use the Base64 encoding at some places. The QuotedPrint module is just included to make the MIME:: collection more complete.

 MIME::Base64       -- Base64 encoding/decoding routines
 MIME::QuotedPrint  -- Quoted Printable encoding/decoding routines

The following modules does not have much to do with the World Wide Web, but are included just because I am lazy and did not bother to make separate distributions for them. Regard them as bonus, provided free for your pleasure.

 Font::AFM          -- Parse Adobe Font Metric files
 File::CounterFile  -- Persistent counter class


All modules contain detailed information on the interfaces they provide. The lwpcook is the libwww-perl cookbook that contain examples of typical usage of the library. You might want to take a look at how the scripts lwp-request, lwp-rget and lwp-mirror are implemented.


The library can not handle multiple simultaneous requests yet. The HTML:: modules are still experimental. Also, check out what's left in the TODO file.


This package owes a lot in motivation, design, and code, to the libwww-perl library for Perl 4, maintained by Roy Fielding <>.

That package used work from Alberto Accomazzi, James Casey, Brooks Cutter, Martijn Koster, Oscar Nierstrasz, Mel Melchner, Gertjan van Oosten, Jared Rhine, Jack Shirazi, Gene Spafford, Marc VanHeyningen, Steven E. Brenner, Marion Hakanson, Waldemar Kebsch, Tony Sanders, and Larry Wall; see the libwww-perl-0.40 library for details.

The primary architect for this Perl 5 library is Martijn Koster and Gisle Aas, with lots of help from Graham Barr, Tim Bunce, Andreas Koenig, Jared Rhine, and Jack Shirazi.


  Copyright 1995-1997, Gisle Aas
  Copyright 1995, Martijn Koster

This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself.


The latest version of this library is likely to be available from:

The best place to discuss this code is on the <> mailing list.