hostelify 0.3.7

data/lib/hostelify.rb

@@ -0,0 +1,10 @@
+require 'rubygems'
+require 'mechanize'
+require 'hpricot'
+require 'open-uri'
+require 'date'
+require 'htmlentities'
+require 'iconv'
+require 'hostelify/hostelify'
+require 'hostelify/hostelworld'
+require 'hostelify/hostelbookers'

data/pkg/hostelify-0.3.6/Manifest

@@ -0,0 +1,30 @@
+Manifest
+README.rdoc
+Rakefile
+hostelify.gemspec
+lib/hostelify.rb
+lib/hostelify/gomio.rb
+lib/hostelify/hostelbookers.rb
+lib/hostelify/hostelify.rb
+lib/hostelify/hostelworld.rb
+pkg/hostelify-0.3.6.gem
+pkg/hostelify-0.3.6.tar.gz
+pkg/hostelify-0.3.6/Manifest
+pkg/hostelify-0.3.6/README.rdoc
+pkg/hostelify-0.3.6/Rakefile
+pkg/hostelify-0.3.6/hostelify.gemspec
+pkg/hostelify-0.3.6/lib/hostelify.rb
+pkg/hostelify-0.3.6/lib/hostelify/gomio.rb
+pkg/hostelify-0.3.6/lib/hostelify/hostelbookers.rb
+pkg/hostelify-0.3.6/lib/hostelify/hostelify.rb
+pkg/hostelify-0.3.6/lib/hostelify/hostelworld.rb
+pkg/hostelify-0.3.6/spec/_helper.rb
+pkg/hostelify-0.3.6/spec/hb_find_by_hostel.spec
+pkg/hostelify-0.3.6/spec/hb_find_hostels.spec
+pkg/hostelify-0.3.6/spec/hw_find_by_hostel.spec
+pkg/hostelify-0.3.6/spec/hw_find_hostels.spec
+spec/_helper.rb
+spec/hb_find_by_hostel.spec
+spec/hb_find_hostels.spec
+spec/hw_find_by_hostel.spec
+spec/hw_find_hostels.spec

data/pkg/hostelify-0.3.6/README.rdoc

@@ -0,0 +1 @@
+Nothin yet... still...

data/pkg/hostelify-0.3.6/Rakefile

@@ -0,0 +1,14 @@
+require 'rubygems'
+require 'rake'
+require 'echoe'
+ 
+Echoe.new('hostelify', '0.3.7') do |p|
+  p.description = "Simple Hostel Webscrapper."
+  p.url = "http://github.com/holden/hostelify"
+  p.author = "Holden Thomas"
+  p.email = "holden.thomas@gmail.com"
+  p.ignore_pattern = ["tmp/*", "script/*"]
+  p.development_dependencies = []
+end
+ 
+Dir["#{File.dirname(__FILE__)}/tasks/*.rake"].sort.each { |ext| load ext }

data/pkg/hostelify-0.3.6/hostelify.gemspec

@@ -0,0 +1,31 @@
+# -*- encoding: utf-8 -*-
+
+Gem::Specification.new do |s|
+  s.name = %q{hostelify}
+  s.version = "0.3.6"
+
+  s.required_rubygems_version = Gem::Requirement.new(">= 1.2") if s.respond_to? :required_rubygems_version=
+  s.authors = ["Holden Thomas"]
+  s.date = %q{2009-09-23}
+  s.description = %q{Simple Hostel Webscrapper.}
+  s.email = %q{holden.thomas@gmail.com}
+  s.extra_rdoc_files = ["README.rdoc", "lib/hostelify.rb", "lib/hostelify/gomio.rb", "lib/hostelify/hostelbookers.rb", "lib/hostelify/hostelify.rb", "lib/hostelify/hostelworld.rb"]
+  s.files = ["Manifest", "README.rdoc", "Rakefile", "hostelify.gemspec", "lib/hostelify.rb", "lib/hostelify/gomio.rb", "lib/hostelify/hostelbookers.rb", "lib/hostelify/hostelify.rb", "lib/hostelify/hostelworld.rb", "spec/_helper.rb", "spec/hb_find_by_hostel.spec", "spec/hb_find_hostels.spec", "spec/hw_find_by_hostel.spec", "spec/hw_find_hostels.spec"]
+  s.has_rdoc = true
+  s.homepage = %q{http://github.com/holden/hostelify}
+  s.rdoc_options = ["--line-numbers", "--inline-source", "--title", "Hostelify", "--main", "README.rdoc"]
+  s.require_paths = ["lib"]
+  s.rubyforge_project = %q{hostelify}
+  s.rubygems_version = %q{1.3.1}
+  s.summary = %q{Simple Hostel Webscrapper.}
+
+  if s.respond_to? :specification_version then
+    current_version = Gem::Specification::CURRENT_SPECIFICATION_VERSION
+    s.specification_version = 2
+
+    if Gem::Version.new(Gem::RubyGemsVersion) >= Gem::Version.new('1.2.0') then
+    else
+    end
+  else
+  end
+end

data/pkg/hostelify-0.3.6/lib/hostelify/gomio.rb

@@ -0,0 +1,102 @@
+require 'rubygems'
+require 'mechanize'
+require 'hpricot'
+require 'open-uri'
+require 'date'
+require 'htmlentities'
+
+Hpricot.buffer_size = 262144
+
+class Gomio
+    
+  #constants
+  GOMIO_SINGULAR = "http://www.gomio.com/reservation/chooseBed.aspx?HostelId="
+  GOMIO_PLURAL_HOSTELS_URL = "http://www.gomio.com/hostels/europe/" #europe/poland/krakow/mama%27s%20hostel/overview.htm
+    
+  def self.find_hostel_by_id(options)
+    #opts = { :directions => false, :images => false, :all => false }.merge options
+    city = options[:location].split(',').first.gsub(' ','')
+    country = options[:location].split(',').last.gsub(' ','')
+    id = options[:id]
+    url = GOMIO_PLURAL_HOSTELS_URL + "#{country}/#{city}/#{id}/overview.htm"
+    
+    #setSearch(url,"2009-09-20",2,7)
+    data = Hpricot(open(url))
+    
+    data = data.search('div[@id="main"]')
+    name = data.at("h3/span").inner_text.lstrip.rstrip
+    address = data.at("span.br_address").inner_text.lstrip.rstrip
+    desc = data.at("div.br_detail").inner_text.lstrip.rstrip
+    available1 = data.at("td.HasNoAvail_Outer")
+    available2 = data.at("td.HasNoAvail")
+    puts "#{available1}, #{available2}"
+  end
+  
+  def self.find_hostels_by_location(options) #location
+    city = options[:location].split(',').first.gsub(' ','')
+    country = options[:location].split(',').last.gsub(' ','')
+    
+    url = GOMIO_PLURAL_HOSTELS_URL + "#{country}/#{city}/search.htm"
+    
+    #data = Hpricot(open(url))
+    data = setSearch(url,"2009-09-09",2,5)
+    
+    data = data.search("div.SearchResultMembers")
+    
+    (data/"div.SearchResultsHostel").each do |row|
+      name = row.at("h3").inner_text.lstrip.rstrip
+      desc = row.at("p").inner_text.lstrip.rstrip
+      url = row.at("h3/a")['href']
+      gomio_id = url.match(/(#{city}).([\d\D]*)(\/overview.htm)/)[2]
+      
+      available1 = row/("td.HasNoAvail_Outer/text()").to_a.join(',').split(',')
+      available2 = row/("td.HasNoAvail_Outer/text()").to_a.join(',').split(',')
+      available = available1 + available2
+
+      @results = { :gomio_id => gomio_id, :name => name, :unavailable => available }
+      puts @results
+    end
+    return @results
+  end
+  
+  def self.setSearch(url,date,no_ppl,no_days)
+
+      date = Date.strptime(date)
+      month = date.strftime("%m").to_i
+      day = date.strftime("%d").to_i
+      if Time.now.strftime("%y") == date.strftime("%y") then year = 0 else year = 1 end
+
+      agent = WWW::Mechanize.new
+      page = agent.get(url)
+
+      #the form name
+      form = page.forms.first # => WWW::Mechanize::Form
+      #page = agent.submit(form)
+
+      #ctl00_searchbox_sb_ddlMonth
+      #ctl00_searchbox_sb_ddlDay
+      #ctl00_searchbox_sb_ddlYear
+
+      #ctl00_searchbox_sb_ddlNights
+      #ctl00_searchbox_sb_ddlBeds
+
+      form.field_with(:name => 'ctl00$searchbox$sb$ddlMonth').options[month-1].select
+      form.field_with(:name => 'ctl00$searchbox$sb$ddlDay').options[day-1].select
+      form.field_with(:name => 'ctl00$searchbox$sb$ddlYear').options[year].select
+      form.field_with(:name => 'ctl00$searchbox$sb$ddlNights').options[no_days.to_i-1].select
+      form.field_with(:name => 'ctl00$searchbox$sb$ddlBeds').options[no_ppl.to_i-1].select
+      #form.field_with(:id => 'Currency').options[4].select #US Currency
+
+      page = agent.submit(form)
+      data = page.search('//div[@id="main"]')
+      #puts data
+      return data
+    end
+    
+  #url = GOMIO_PLURAL_HOSTELS_URL + "poland/krakow/search.htm"
+  
+  #Gomio.setSearch(url,"2009-09-20",2,7)
+  Gomio.find_hostels_by_location(:location => "krakow,poland")
+  #Gomio.find_hostel_by_id(:id => "mama's%20hostel", :location => "krakow,poland")
+  
+end

data/pkg/hostelify-0.3.6/lib/hostelify/hostelbookers.rb

@@ -0,0 +1,180 @@
+class Hostelbookers
+  
+  #constants
+  HB_SINGULAR_DETAIL_URL = "http://www.hostelbookers.com/hostels/" #poland/krakow/
+  HB_PLURAL_HOSTELS_URL = "http://www.hostelbookers.com/hostels/" #poland/wroclaw/
+  
+  #options
+  @default_options = { :date => date=(Date.today+4).to_s, :no_days => "7", :live => true }
+  
+  def self.find_hostels_by_location(options) #location
+    city = options[:location].split(',').first.rstrip.lstrip.gsub(' ','-').squeeze("-")
+    country = options[:location].split(',').last.rstrip.lstrip.gsub(' ','-').squeeze("-")
+
+    url = HB_PLURAL_HOSTELS_URL + "#{country}/#{city}"
+    
+    if options[:date]
+      options = @default_options.merge(options)
+      date = Date.strptime(options[:date])
+      data = setSearch(url,options[:date],options[:no_days])
+    else
+      Retryable.try 3 do
+        data = Hpricot(open(url))
+      end
+    end
+    
+    data = data.search("//div[@id='propertyResultsList']")
+    #@results = []
+    @results = HostelifyCollection.new
+      #coder = HTMLEntities.new
+    (data/"tr.propertyRow").each do |row|
+      name = row.at("a.propertyTitle").inner_text
+      url = row.at("a.propertyTitle")['href']
+      desc = row.at("p.shortDescription").inner_text
+      rating = row.at("td.rating/text()")
+      rating = rating.to_s.to_i unless rating.nil?
+      dorm = row.at("td.shared/text()")
+      single = row.at("td.private/text()")
+      hb_id = url.match(/[\d]{2,5}.$/).to_s.to_i
+      
+      #@results << Hostelify.new(:hostel_id => hb_id, :name => name, :description => desc, :rating => rating, :dorm => dorm, :single => single)
+      @results << Hostelify.new(:hostel_id => hb_id, :name => name, :description => desc, :rating => rating, :dorm => dorm, :single => single)
+    end    
+    return @results
+  end
+  
+  def self.find_hostel_by_id(options)
+    city = options[:location].split(',').first.gsub(' ','')
+    country = options[:location].split(',').last.gsub(' ','')
+    id = options[:id]
+    url = HB_SINGULAR_DETAIL_URL + "#{country}/#{city}/#{id}"
+    
+    hostel = Hostelify.new
+
+    if options[:date]
+      options = @default_options.merge(options)
+      data = setSearch_id(url,options[:date],options[:no_days])
+    else
+      Retryable.try 3 do
+        data = Hpricot(open(url))
+      end
+    end
+    
+    hostel.hostel_id = id
+    hostel.name = data.at("h1").inner_text
+    hostel.address = data.at("p.address").inner_text
+    hostel.description = data.at('div[@id="overviewPane"]').inner_text
+    facilities_td = data.at("table.tableFacilities")
+    
+    facilities = []
+    (facilities_td/"td").each do |row|
+       facilities << row.inner_text
+    end
+    hostel.facilities = facilities
+    extras = []
+    extras_td = data.at("table.tableFeatures")
+    (extras_td/"td.name").each do |row|
+      extras << "Free " + row.inner_text.to_s
+    end
+    facilities = facilities + extras
+    
+    ratings = []
+    ratings_td = data.at('div[@id="overviewIndRtng"]/table')
+    
+    (ratings_td/"tr").each do |row|
+      ratings << row.at("td").inner_text.to_s.to_f
+    end
+    
+    hostel.ratings = ratings
+    images = []
+    image = data.at('div[@id="propMedia"]/table')
+    (image/"td").each do |row|
+      img = row.at("img")['onclick']
+      if img =~ /(http).*(jpg|gif|png|jpeg)/
+        images << img.match(/(http).*(jpg|gif|png|jpeg)/)[0]
+      else
+        #add youtube?
+      end
+    end
+    hostel.images = images
+    
+    if options[:all]
+      data = Hpricot(open(url + "/map"))
+      data.search("h2").remove #get rid of header
+      hostel.directions = data.at('div[@id="directions"]').inner_text
+      hostel.geo = data.to_s.scan(/-{0,1}\d{1,3}\.\d{7}/).uniq!
+    end
+    
+    if options[:date]
+      date = Date.strptime(options[:date])
+      @availables = []
+      available = data.at("div.tableAvailability/table")
+      if available
+        (available/"tr").each do |row|
+          name = row.at("td.roomType/label/text()")
+          people = row.at("td.people/select")
+          people = people.at("option:last-child").inner_text unless people.nil?
+          price = row.at("td.price")
+          price = price.inner_text.to_s.match(/[\d.]{1,5}/)[0] unless price.nil?
+          (0..(options[:no_days].to_i-1)).each do |x|
+            #@availables << { :name => name, :spots => people, :price => price, :bookdate => (date+x).to_s } unless price.nil?
+            @availables << HostelifyAvailable.new(name,price,people,(date+x).to_s) unless price.nil?
+          end
+        end
+      end
+      hostel.availability = @availables
+    end
+    
+    return hostel
+  end
+  
+  def self.setSearch(url,date,no_days)
+      date = Date.strptime(date).strftime("%d/%m/%Y")
+      agent = WWW::Mechanize.new
+      page = agent.get(url)
+      form = page.form_with(:name => 'searchForm') # => WWW::Mechanize::Form
+      form.field_with(:name => 'intnights').options[no_days.to_i-1].select
+      form.dtearrival = date #d/m/y
+      
+      Retryable.try 3 do
+        page = agent.submit(form)
+      end
+      
+      #to dollars!
+      form = page.forms[0]
+      form.field_with(:name => 'strSelectedCurrencyCode').options[5].select
+      
+      Retryable.try 3 do
+        page = agent.submit(form)
+      end
+      
+      data = page.search('//div[@id="content"]')
+
+      return data
+    end
+    
+    def self.setSearch_id(url,date,no_days)
+        date = Date.strptime(date).strftime("%d/%m/%Y")
+        agent = WWW::Mechanize.new
+        page = agent.get(url)
+        form = page.form_with(:name => 'frmCheckAvailBook') # => WWW::Mechanize::Form
+        form.field_with(:name => 'intNights').options[no_days.to_i-1].select
+        form.dteArrival = date #d/m/y
+        
+        Retryable.try 3 do
+          page = agent.submit(form)
+        end
+        #change currency to dollars
+        form = page.forms[1]
+        #puts form.name
+        form.field_with(:name => 'strSelectedCurrencyCode').options[5].select
+        
+        Retryable.try 3 do
+          page = agent.submit(form)
+        end
+        data = page.search('//div[@id="content"]')
+
+        return data
+      end
+  
+end

data/pkg/hostelify-0.3.6/lib/hostelify/hostelify.rb

@@ -0,0 +1,61 @@
+class Hostelify
+  attr_accessor :hostel_id, :name, :address, :description, :facilities, :ratings, :directions, :geo, :images, :video, :availability
+  attr_accessor :rating, :dorm, :single, :unavailable
+
+  def initialize(options = {})
+      options.each {
+        |k,v|
+        self.send( "#{k.to_s}=".intern, v)
+      }
+  end
+
+end
+
+class HostelifyCollection < Array
+  # This collection does everything an Array does, plus
+  # you can add utility methods like names.
+
+  def ids
+    collect do |i|
+      i.hostel_id
+    end
+  end
+  
+  def names
+    collect do |i|
+      i.name
+    end
+  end
+  
+  def descs
+    collect do |i|
+      i.description
+    end
+  end
+  
+end
+
+class HostelifyAvailable
+  attr_accessor :name, :price, :spots, :bookdate
+  
+  def initialize(name, price, spots, bookdate)
+    @name = name
+    @price = price
+    @spots = spots
+    @bookdate = bookdate
+  end
+  
+end
+
+module Retryable
+  extend self
+
+  def try times = 1, options = {}, &block
+    val = yield
+  rescue options[:on] || Exception
+    retry if (times -= 1) > 0
+  else
+    val
+  end
+end
+