Hacking

    ______  _     _  ______   _______  ______   _______  _  _  _
   / _____)(_)   (_)(_____ \ (_______)(_____ \ (_______)(_)(_)(_)
  ( (____   _     _  _____) ) _____    _____) ) _______  _  _  _
   \____ \ | |   | ||  __  / |  ___)  |  __  / |  ___  || || || |
   _____) )| |___| || |  \ \ | |      | |  \ \ | |   | || || || |
  (______/  \_____/ |_|   |_||_|      |_|   |_||_|   |_| \_____/

  Surfraw - Shell Users' Revolutionary Front Rage Against the Web

			                 Oh Baybe
			                I need some
		                   Deep Linking
			                 Let us go
		                Surfin' in the raw!

		                    H A C K I N G

Or... How to Eat The CHEESE

First, find a front end ('an elvi, types of elvis -- makes evil nice') in the elvi/ directory that does something similar to what you want. In this example we shall use the google elvis.
Copy the google elvis to the name of your new elvis. In this example we shall create an elvis for the Peoples' Front Against Forum Stealing Dogs http://www.pfafsd.org/ mailing-list search service
At the top of the file, you will see something like:

elvis: google -- Search the web using Google (http://www.google.com)

Keep the "elvis:" take and change the rest of the line. This description is used at Surfraw installation time to generate a list of Surfraw clients

Immediately after we see:

    . surfraw || exit 1

Which pulls in the Surfraw functions. Keep it.

Next we (optionally) define w3_config_hook(). It is called by w3_config and is used to define default configuration variables. It is called after the global configuration file is interpreted, but before ~/.surfraw.

You will notice the use of:

    def    -- define configuration variable unless already defined
              (i.e by the environment)
    defyn  -- define boolean configuration variable unless already defined

Which are pre-defined surfraw functions, but you can of course use your own arbitrary broken shell code instead.

Any non-trivial program will then define w3_usage_hook(). $w3_argv0 is set to contain $0, otherwise usage is self explanatory. You must call w3_global_usage at the end of this function and follow the surfraw usage style least you be marked a quiche eater.
Next define w3_parse_option_hook(). This is called at w3_config() time, and takes two arguments. The first is the current option being examined, the second is only present if the option was of the form -foo=bar. It's the bar.

w3_parse_option_hook() is called before Surfraw global options are parsed and so you can selectively over-ride them.

Return value: 0 if the option was recognised, 1 otherwise.

After this, w3_config() and w3_parse_args "[email protected]" are called. w3_config() scans the configuration files and calls w3_config_hook(). w3_parse_args "[email protected]" processes options, calls w3_parse_option_hook() as needed and builds $w3_args -- the list non-option arguments.

This ends the regular initialisation code.

7.5) Note that surfraw now supports quoting arguments, eg with:

sr google foo "bar baz" bam

it adds the quotes back.

Not all sites support arguments in double quotes. If you want to disable it for your elvis, add SURFRAW_quote_ifs=0 in between the calls to w3_config and w3_parse_args.

What happens next depends on the type of evil we are dealing with. In the Peoples' Front Against Forum Stealing Dogs case, we have previously performed a search on the http://www.pfafsd.org search service, and noted (with the = key in lynx, links) the URL returned. We then analyse this URL to work out which CGI variable is being used to convey search the search string, number of results et al.

Q: What about those proslytizers of iniquity that worship idols like method=POST?

A: Most sites will take GETs in the place of POSTs, even if they preach method=POST in forms. For those that don't? Walking with the devil should not be a an exercise in relaxed discourse. The problem is getting the browser to automatically submit the data using POST as a method. One partial solution is to have surfraw generate a pre-filled method=POST form as a local file and have the browser read it. Exploration of other ideas is encouraged.

w3_url_of_arg() is then used to escape the command line search string argument into something that can be seen in a URL without embarrassment.

i.e escaped_args=w3_url_of_arg $w3_args

In our last exhilarating moments we then build the final URL and pass it onto the currently selected browser with w3_browse_url, proud in the knowledge that the Shell Users' Revolutionary Front Against the WWW has brought the holy word of the command line to yet another heathen land.

-Julian Assange [email protected]

elvis: google -- Search the web using Google (http://www.google.com)