read data from East Side Traders site

eastsidrocks.rb

#!/usr/bin/ruby -rubygems
require 'open-uri'
base='http://www.visiteastside.co.uk/search_eastside/index.php?name=&&&offset=0#searchresults'

poidb = {}

# This Ruby script will extract data from the pages of the Visit East Side site.
#
# Notes:
#  Please play nice and only use the data for purposes that further the goals of 
#  Bristol East Side Traders. http://www.bristoleastsidetraders.co.uk/
#  "BEST (Bristol East Side Traders) is a not-for-profit company that aims to 
#   develop a thriving local economy in Bristol's inner city that benefits 
#   residents - as entrepreneurs, employees and customers."
#
# I haven't yet asked permission or forgiveness to do anything with this data.
# If you make something, please bear that in mind, and raise it with them before announcing.
#
# It pulls out the postcode field from address. For this see 
# http://data.gov.uk/ideas/link-post-code-government-services
# http://www.epsiplus.net/news/news/ordnance_survey_postcodes_as_linked_data  etc.
#
# Dan Brickley <[email protected]>

# Usage:
# TellyClub:hackbristol danbri$ ./eastsiderocks.rb 
#1	BS5 9LT	Clarks Pies	 	114b Church Road, Redfield, Bristol BS5 9LT	 
#2	BS5 9LH	Claudette Reid Financial Consultancy		31 Verrier Road, Redfield, Bristol BS5 9LH	 
#3	BS5 9LJ	Clear Financial Solutions		112a Church Road, Redfield, Bristol BS5 9LJ	 
#5	BS5 6NQ	Coach House Pub	 	380 Stapleton Road, Easton, Bristol BS5 6NQ	12pm - late
#6	BS6 5NU	Co-Creative		119a Ashley Road, Montpelier, Bristol BS6 5NU
# ...etc. 

#
# Tech notes: you need to 
# get a PHPSESSID from eg. Chrome browser, settings / under the hood / content settings / cookies
# I've included a session ID of mine in the script; not sure how long it'll work for.


myCookie = "PHPSESSID=166a38e8088a5ea0e39796b10aabcae1"
base = 'http://www.visiteastside.co.uk/search_eastside/details.php?trader_id=0&offset='
hp = open('http://www.visiteastside.co.uk/search_eastside/index.php#searchresults')

0.step(700,1) do |i|
  
  sleep 1 # play nice
  
  u = String.new(base)
  u.gsub!(/trader_id=0/, "trader_id=#{i}")
  # puts "Fetching #{u}"
  # :( Sorry, you need cookies enabled to use the search. Need to copy one in by hand for now:
  
  open(u,  "Cookie" => myCookie ) do |f|
     page = f.read
     next if page =~ /<h1>Organisation Not Found<\/h1>/
     myPoi = {}
     poidb[i] = myPoi

	# <h1>Search Eastside</h1><h2>Clarks Pies</h2>
     if page =~ /<h1>Search Eastside<\/h1><h2>([^<]+)<\/h2>/
       myPoi['name']=$1
     end

	# <p><span class="label">Address:</span> 114b Church Road, Redfield, Bristol BS5 9LT</p>
     if page =~ /Address:<\/span>\s*([^<]+)<\/p>/
#       puts "ADDRESS! #{$1}"
       myPoi['addr']=$1

       myPoi['addr'] =~ /\s*(BS\d+\s+\d+\S+)/
#       puts "POSTCODE: #{$1}"
       myPoi['postcode'] = $1
     end


	# <p><span class="label">Telephone:</span> 0117 955 6702</p>
     if page =~ /Telephone:<\/span>\s*([^<]+)<\/p>/
#       puts "TEL! #{$1}"
       myPoi['tel']=$1
     end

	# <p><span class="label">Mobile:</span> </p>
     if page =~ /Mobile:<\/span>\s*([^<]+)<\/p>/
#       puts "MOB! #{$1}"
       myPoi['mob']=$1
     end

	# <p><span class="label">Fax:</span> </p>
     if page =~ /Fax:<\/span>\s*([^<]+)<\/p>/
#       puts "FAX! #{$1}"
       myPoi['fax']=$1
     end

	# <p><span class="label">Website:</span> </p>
     if page =~ /Website:<\/span>\s*([^<]+)<\/p>/
#       puts "WEB! #{$1}"
       myPoi['web']=$1
     end

	# <p><span class="label">Opening Times:</span> </p>
     if page =~ /Times:<\/span>\s*([^<]+)<\/p>/
#       puts "TIMES! #{$1}"
       myPoi['times']=$1
     end
   
     puts "#{i}\t#{myPoi['postcode']}\t#{myPoi['name']}\t#{myPoi['web']}\t#{myPoi['addr']}\t#{myPoi['times']}"

  end
end

#poidb.each do |p| 
#  puts "#{p.dump}"
#end

# more re cookie handling - http://stackoverflow.com/questions/1360808/rubys-open-uri-and-cookies


# cookie
# content: 166a38e8088a5ea0e39796b10aabcae1
# domain: www.visiteastside.co.uk
# path /