hc-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From httpclient-...@jakarta.apache.org
Subject [Jakarta-HttpClient Wiki] New: UseCases/AccessURLs
Date Sat, 12 Feb 2005 23:20:49 GMT
   Date: 2005-02-12T15:20:48
   Editor: RolandWeber
   Wiki: Jakarta-HttpClient Wiki
   Page: UseCases/AccessURLs
   URL: http://wiki.apache.org/jakarta-httpclient/UseCases/AccessURLs

   no comment

New Page:

= Use Case: Access some URLs =

== Description ==

I have some URLs, and want to access those documents in my application.


== Background ==

This is the most obvious use of an HTTP client component.
It is important for the ease of use of the API.
[[BR]]
This use case is about a list of URLs rather than a single URL to
highlight the difference between initializations and per-access
activities.


== Variations ==

 1. I want to save the documents to the file system, too.


== Related / Out of Scope ==

 * access from multiple threads
 * access with session cookie
 * access with authentication
 * access through proxy


= Discussion =


There is a trade-off between ease of use and flexibility. The standard
Java library provides the required functionality for this use case in
the {{{java.net}}} package through the {{{URL}}} and {{{HttpURLConnection}}} classes.
These are easy to use, but exhibit major drawbacks once you require
flexibility.

The HttpClient project was originally started because {{{HttpURLConnection}}}
is not good enough. HttpClient is not and will hardly ever be as
easy to use. That said, it still shouldn't be unnecessary hard to use either.

If accessing documents is your only concern you may be
better off using the Java built-in stuff. But as soon as you have to deal with
timeouts, error handling, different cookie specs, connection management,
multiple user sessions and all kinds of stuff, HttpClient starts
to shine.



= Solution =

== Standard Solution ==

Additional preparations are required for HTTPS.

 * create an {{{HttpClient}}} object
 * for each URL:
  1. create a {{{GetMethod}}} for the URL
  1. ask the {{{HttpClient}}} to execute the {{{GetMethod}}} to obtain an {{{HttpResponse}}}
  1. get the {{{HttpIncomingEntity}}} type you want from the {{{HttpResponse}}}
  1. once you're done with the entity, release the connection

== Bare Bones Solution ==

The bare bones solution will not work for HTTPS.

 * create an {{{NIOHttpClientConnection}}} object
 * for each URL:
  1. create a {{{GetMethod}}} for the URL
  1. send the {{{GetMethod}}} over the {{{HttpConnection}}} to obtain an {{{HttpResponse}}}
  1. get the {{{HttpIncomingEntity}}} you want from the {{{HttpResponse}}}
  1. reset the connection (? - implementation detail, to be decided later)

== Variation: Save to File System ==

in addition to one of the solutions above,...

 * for each URL:
  1. the {{{HttpIncomingEntity}}} you want is a {{{StreamIcEntity}}}
  1. get the {{{InputStream}}} from the {{{StreamIcEntity}}}
  1. open a {{{FileOutputStream}}}
  1. read chunks of data from the input stream into a buffer and write them to the output
stream. 4K or 16K are common choices for the buffer size.
  1. close the input and output streams

---------------------------------------------------------------------
To unsubscribe, e-mail: httpclient-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: httpclient-dev-help@jakarta.apache.org


Mime
View raw message