From: James Britt <jamesUNDERBARb@...>
Date: 2004-08-17T05:40:43+09:00
Subject: Re: Ruby as a first language

Gawnsoft wrote:
>>No, seriously, I've written a number of simple scripts to fetch web 
>>pages.  Usually they just construct a URL based on some command-line 
>>input, then make a system call to launch a browser.  I search Google 
>>this way, as well as look for stuff on buy.com, check TV listings, read 
>>the news, grep/navigate my local wiki, a few other things.
> 
> 
> Could I see them please?
> 


Here's a simple case: google

# file g.rb
$: << File.dirname( __FILE__ )
require 'browser'

$url = 'http://www.google.com/search?hl=en&ie=UTF-8&oe=UTF-8&q='
if ARGV.size < 1
   puts "You need to enter at least one search word."
else
   $url << ARGV.join("+")
   $stderr.puts "Calling #{$url}"
   Browser.new( 'FB' ).get( $url )
end
# end of file

Next is the code for browser.rb.  It's meant to run on Windows.
I've never bothered to try it on Linux; executable paths would
have to change, as well as the calling convention (e.g., 'call
someApp.exe'), but maybe not much else.  (Actually, I think the big 
issue for me is that when I launch Mozilla or Firefox , and there is 
already an instance running, it stupidly insists I select another 
profile, or create a new one, or I can't use the new instance.  That 
doesn't happen on Windows.)

I believe I use a thread to launch the browser so that the script ends
and I get my command prompt back; I run my browser scripts from the
command line,  like so:

C:\> g some search words

Windows handles the file extension stuff automagically, as I've added 
the script directory to my PATH variable.

But, without using a thread, the script would just wait until I closed 
the browser, which is annoying.

Anyway, once I put together the Browser class I needn't think about it
when I write other scripts for pulling up web sites.  (And 'less 
thinking' is one of the goals of coding. :) )

# browser.rb

class Browser
   DEFAULT = 'FB'
   CB = 'CB'
   IE = 'IE'
   MOZ = 'MOZ'
   OP = 'OP'
   FB = 'FB'

   @@exe = Hash.new( 'explorer')
   @@exe['IE'] = 'explorer'
   @@exe['CB'] = '"E:/Program Files/Crazy Browser/Crazy Browser.exe"'
   @@exe['OP'] = 'e:/Program Files/Opera7/opera.exe'
   @@exe['MOZ'] = 'e:/Program Files/mozilla.org/Mozilla/mozilla.exe'
   @@exe['FB'] = 'E:/MozillaFirebird/MozillaFirebird.exe '

   def initialize( browser_name = Browser::DEFAULT  )
     @broswser = browser_name
     @b_exe = select_exe( browser_name )
     $stderr.puts( "Using broswer #@broswser ")
   end

   def get( url )
     $stderr.puts( "get is using #@b_exe" )
     Thread.new( url ){ |t|  `call  "#{@b_exe}" "#{url}"`  }
   end

   def select_exe( bname )
     @@exe[ bname ]
   end

   def self.get( url,
     browser_name = ( @broswser ? @broswser : Browser::DEFAULT ) )
     exe = Browser.new(browser_name).select_exe( browser_name )
     Thread.new( url ){ |t| `start  "#{exe}" "#{url}"`  }
   end
end


# End of browser.rb

Lookng at the code now, there's cruft and goofiness, but it works so I
haven't bothered to clean stuff up.  But for a newcomer, it might be 
worthwhile to try writing a class that allows the user to 1) optionally 
select a particular browser to use (but have a default browser defined), 
  2) pass in the URL to load, 3) ensures that the script can run and 
exit without having to wait until the browser closes.

Here's another script; this one calls Yahoo maps, like this:

c:\> map 11123 Main Street, Scottsdale, AZ 85251

# map.rb
$: << File.dirname( __FILE__ )
require 'browser'
$base_url = 'http://maps.yahoo.com/py/maps.py?Pyt=Tmap'

def argstr_to_qstring( str )
   parts = str.split( ',', 2)
   street = parts[0].strip.gsub( / /, '+')
   state = parts[1].strip.gsub( / /, '+')
   "&addr=#{street}&csz=#{state}"
end

def args_to_mapurl( args )
   a  = args.join(' ')
    "#{$base_url}#{argstr_to_qstring( a )}&Get%A0Map=Get+Map"
end

if ARGV.size < 1
   puts "You need to enter an address."
else
   url = args_to_mapurl( ARGV )
   Browser.get( url , Browser::MOZ )
end

# end of map.rb

This one is a bit gnarlier.  But, once someone has written a general way 
to launch a Web browser, a next step might be to explore manipulating 
the input to construct more complex URLs.  The goal is to make it easy 
to type at the command line, and let the code do the busy work.

James