Scraping CA NOI's

scraper.rb

require 'rubygems'
require 'mechanize'
require 'byebug'

a = Mechanize.new { |agent|
  agent.user_agent_alias = 'Mac Safari'
  key = 'JSESSIONID'
  value = '3a70858e14aca4f6aac9ec59d95d13d86e0b4e755b40b67bb56f96f27fbf05b2.e3qSc3iTc34Oe3iNaO0'

  cookie = Mechanize::Cookie.new(key, value)
  cookie.domain = "smarts.waterboards.ca.gov"
  cookie.path = "/smarts"
  agent.cookie_jar.add(cookie)
}


postVars = {
  'form1:searchButton' => 'Search',
  'form1:selectOneMenu1' => 6021,
  'form1:regMeasId' => nil,
  'form1:processedFromDate' => nil,
  'form1:processedToDate' => nil,
  'form1:wdid' => nil,
  'form1:ownerName' => nil,
  'form1:selectOneMenu2' => 0,
  'form1:ownerStreet' => nil,
  'form1:selectOneMenu4' => 0,
  'form1:devName' => nil,
  'form1:facilityName' => nil,
  'form1:sicCode' => nil,
  'form1:facilityStreet' => nil,
  'form1:selectOneMenu3' => 4,
  'form1:facilityCity' => nil,
  'com.sun.faces.VIEW' => '_id62823:_id62824',
  'form1' => 'form1'
}


url = 'https://smarts.waterboards.ca.gov/smarts/faces/PublicDataAccess/PublicNoiSearch.jsp';


page = a.post(url, postVars)
hiddenForm = page.form_with(:id => 'form1')

links = page.links.select { |link| link.text == 'NOI Form' }
links.each { |link|

  uglyJs = link.attributes.attributes['onclick'].value
  submitData = {
    'com.sun.faces.VIEW' => '_id62823:_id62827',
    'form1' => 'form1',
    'regMeasId' => 453212
    'form1:_idcl' => 'form1:publicSearchNOItable:19:_id27'
  }

}

byebug

#

puts 'hella'