schedule-scraper 0.1.0 → 0.2.0

data/README.md

@@ -8,7 +8,8 @@ Supported schedule sites:
 
 Supported output formats:
 
-* CSV
+* CSV (plain dump)
+* Google Calendar formatted CSV
 
 ## Installation
 
@@ -26,14 +27,28 @@ Or install it yourself as:
 
 ## Usage
 
+### Pointstreak Example
+
+Locate the printable version of the scheulde:
+
+1. Visit the leagues home page
+2. Click on the team in question
+3. Click the SCHEDULE link under TEAM MENU
+4. Click PRINT THIS PAGE
+
 Request a schedule:
 
-    schedule = ScheduleScrape.fetch(:point_streak, :season => 123, :team => 456)
+    url = "http://www.pointstreak.com/players/print/players-team-schedule.html?teamid=385368&seasonid=9162"
+    schedule = ScheduleScrape.fetch(:point_streak, url)
 
 Export the schedule to CSV:
 
     schedule.to_csv
 
+or
+
+    schedule.to_gcal
+
 ## TODO
 
 1. Add more export options: iCal, Google Calendar (csv)

data/lib/schedule-scraper/event.rb

@@ -15,6 +15,20 @@ module ScheduleScraper
         self.send(field)
       end
     end
+
+    def to_gcal
+      [
+        title,
+        start_date,
+        start_time,
+        end_date,
+        "",
+        all_day?,
+        description,
+        "",
+        private?
+      ]
+    end
   end
 end
 

data/lib/schedule-scraper/pointstreak/event.rb

@@ -10,6 +10,38 @@ module ScheduleScraper
       element 'td:nth(3)' => :date, :with => cleaner
       element 'td:nth(4)' => :time, :with => cleaner
       element 'td:nth(5)' => :rink, :with => cleaner
+
+      def title
+        "#{home_team} vs. #{away_team}"
+      end
+
+      def start_date
+        Date.parse(date).strftime("%m/%d/%y")
+      end
+
+      def end_date
+        start_date
+      end
+
+      def start_time
+        time
+      end
+
+      # def end_time
+      #   will default to one hour?
+      # end
+
+      def all_day?
+        false
+      end
+
+      def description
+        title
+      end
+
+      def private?
+        true
+      end
     end
   end
 end

data/lib/schedule-scraper/pointstreak/schedule.rb

@@ -3,14 +3,12 @@ module ScheduleScraper
     class Schedule < Nibbler
       include ScheduleScraper::Schedule
 
-      POINT_STREAK_URL = "http://www.pointstreak.com/players/print/players-team-schedule.html"
-
       element 'table table:last' => :list do
         elements 'tr:not(.fields)' => :event_list, :with => Event
       end
 
-      def self.fetch(options)
-        parse html(options[:season], options[:team])
+      def self.fetch(url)
+        parse open(url)
       end
 
       def events
@@ -19,14 +17,6 @@ module ScheduleScraper
 
       private
 
-      def self.html(season, team)
-        open(source_url(season, team))
-      end
-
-      def self.source_url(season, team)
-        "#{POINT_STREAK_URL}?teamid=#{team}&seasonid=#{season}"
-      end
-
       def event_class
         ScheduleScraper::Pointstreak::Event
       end

data/lib/schedule-scraper/schedule.rb

@@ -15,5 +15,26 @@ module ScheduleScraper
         end
       end
     end
+
+    def to_gcal
+      headers = [
+        "Subject",
+        "Start Date",
+        "Start Time",
+        "End Date",
+        "End Time",
+        "All Day Event",
+        "Description",
+        "Location",
+        "Private"
+      ]
+
+      CSV.generate do |csv|
+        csv << headers
+        events.each do |event|
+          csv << event.to_gcal
+        end
+      end
+    end
   end
 end

data/lib/schedule-scraper/version.rb

@@ -1,3 +1,3 @@
 module ScheduleScraper
-  VERSION = "0.1.0"
+  VERSION = "0.2.0"
 end

data/lib/schedule-scraper.rb

@@ -1,6 +1,8 @@
 require 'nibbler'
 require 'open-uri'
 require 'csv'
+require 'uri'
+require 'nokogiri'
 require "schedule-scraper/version"
 require "schedule-scraper/event"
 require "schedule-scraper/schedule"
@@ -8,10 +10,11 @@ require "schedule-scraper/pointstreak/event"
 require "schedule-scraper/pointstreak/schedule"
 
 module ScheduleScraper
-  def self.fetch(type, options = {})
+  def self.fetch(type, url)
     raise UnsupportedSchedule unless supported_schedules.include?(type.to_sym)
+    raise InvalidURL unless valid_url?(url)
 
-    type_class(type).fetch(options)
+    type_class(type).fetch(url)
   end
 
   def self.type_class(type)
@@ -26,5 +29,13 @@ module ScheduleScraper
     ]
   end
 
+  def self.valid_url?(url)
+    uri = URI.parse(url)
+    uri.kind_of?(URI::HTTP)
+  rescue URI::InvalidURIError
+    false
+  end
+
   class UnsupportedSchedule < StandardError; end
+  class InvalidURL < StandardError; end
 end

data/schedule-scraper.gemspec

@@ -16,6 +16,7 @@ Gem::Specification.new do |gem|
   gem.version       = ScheduleScraper::VERSION
 
   gem.add_dependency 'nibbler',       '~> 1.3.0'
+  gem.add_dependency 'nokogiri',      '~> 1.5.4'
 
   gem.add_development_dependency 'minitest',            '~> 3.1.0'
   gem.add_development_dependency 'rake',                '~> 0.9.2'

data/spec/schedule-scraper/pointstreak/event_spec.rb

@@ -3,6 +3,15 @@ require File.expand_path(File.join(File.dirname(__FILE__), '../../spec_helper'))
 describe ScheduleScraper::Pointstreak::Event do
   let(:options) { POINTSTREAK_OPTIONS }
   let(:fields) { [:home_team, :away_team, :date, :time, :rink] }
+  let(:expected_values) do
+    {
+      :home_team => "BLADES 6",
+      :away_team => "SUMMIT 8",
+      :date => "Sun, Jun 03",
+      :time => "7:45 pm",
+      :rink => "final"
+    }
+  end
 
   before do
     VCR.use_cassette('summit_summer_2012') do
@@ -20,11 +29,60 @@ describe ScheduleScraper::Pointstreak::Event do
 
   it "uses elements to define fields for csv" do
     klass = ScheduleScraper::Pointstreak::Event
-    klass.send(:export_fields).must_equal fields
+    klass.send(:export_fields).must_equal expected_values.keys
   end
 
   it "returns a list of fields when you ask for csv" do
-    expected = ["BLADES 6", "SUMMIT 8", "Sun, Jun 03", "7:45 pm", "final"]
-    subject.to_csv.must_equal expected
+    subject.to_csv.must_equal expected_values.values
+  end
+
+  describe "output helper methods" do
+    it "defines a title" do
+      expected = "#{expected_values[:home_team]} vs. #{expected_values[:away_team]}"
+      subject.title.must_equal expected
+    end
+
+    it "defines a start date" do
+      expected = Date.parse(expected_values[:date]).strftime("%m/%d/%y")
+      subject.start_date.must_equal expected
+    end
+
+    it "defines an end date" do
+      subject.end_date.must_equal subject.start_date
+    end
+
+    it "defines a start time" do
+      subject.start_time.must_equal expected_values[:time]
+    end
+
+    it "defines a description" do
+      subject.description.must_equal subject.title
+    end
+
+    it "defines all day event" do
+      subject.all_day?.must_equal false
+    end
+
+    it "defines all private" do
+      subject.private?.must_equal true
+    end
+  end
+
+  describe "#to_gcal" do
+    it "provides an array ready to export to csv" do
+      expected = [
+        subject.title,
+        subject.start_date,
+        subject.start_time,
+        subject.end_date,
+        "",
+        false,
+        subject.description,
+        "",
+        true
+      ]
+
+      subject.to_gcal.must_equal expected
+    end
   end
 end

data/spec/schedule-scraper/pointstreak/schedule_spec.rb

@@ -4,23 +4,6 @@ describe ScheduleScraper::Pointstreak::Schedule do
   subject() { ScheduleScraper::Pointstreak::Schedule }
   let(:options) { POINTSTREAK_OPTIONS }
 
-  it "knows the root pointstreak url" do
-    subject::POINT_STREAK_URL.must_match /pointstreak/
-  end
-
-  it "builds a valid url" do
-    url = subject.send(:source_url, "123", "456")
-    expected = "#{subject::POINT_STREAK_URL}?teamid=456&seasonid=123"
-
-    url.must_equal expected
-  end
-
-  it "fetches html from pointstreak" do
-    VCR.use_cassette('summit_summer_2012') do
-      subject.html(options[:season], options[:team])
-    end # wont_raise
-  end
-
   describe "schedule instance" do
     subject() do
       VCR.use_cassette('summit_summer_2012') do
@@ -44,5 +27,10 @@ describe ScheduleScraper::Pointstreak::Schedule do
     it "generates a csv file" do
       subject.to_csv.must_be_instance_of String
     end
+
+    it "generates a google calendar formatted csv" do
+      p subject.to_gcal
+      subject.to_gcal.must_be_instance_of String
+    end
   end
 end

data/spec/schedule-scraper/schedule_spec.rb

@@ -1,11 +1,9 @@
 # require File.expand_path(File.join(File.dirname(__FILE__), '../spec_helper'))
 
-# describe ScheduleScraper::Schedule do
-#   subject() { ScheduleScraper::Schedule.new({}) }
+# class ScraperTest
+#   include ScheduleScraper::Schedule
+# end
 
-#   describe "#to_csv" do
-#     it "returns a schedule in CSV format" do
-#       subject.to_csv.must_be_instance_of String
-#     end
-#   end
+# describe ScheduleScraper::Schedule do
+#   subject() { ScraperTest.new(:xyz, :url => "http://www.xyz.com") }
 # end

data/spec/schedule-scraper_spec.rb

@@ -11,10 +11,16 @@ describe ScheduleScraper do
       end
 
       -> {
-        ScheduleScraper.fetch(:xyz)
+        ScheduleScraper.fetch(:xyz, options)
       }.must_raise ScheduleScraper::UnsupportedSchedule
     end
 
+    it "validates the url" do
+      -> {
+        ScheduleScraper.fetch(:pointstreak, "abc")
+      }.must_raise ScheduleScraper::InvalidURL
+    end
+
     it "returns a schedule" do
       VCR.use_cassette('summit_summer_2012') do
         schedule = ScheduleScraper.fetch(:pointstreak, options)

data/spec/spec_helper.rb

@@ -5,7 +5,8 @@ require 'vcr'
 
 require File.expand_path(File.join(File.dirname(__FILE__), '../lib/schedule-scraper'))
 
-POINTSTREAK_OPTIONS = { :season => "9162", :team => "385368" }
+# POINTSTREAK_OPTIONS = { :season => "9162", :team => "385368" }
+POINTSTREAK_OPTIONS = "http://www.pointstreak.com/players/print/players-team-schedule.html?teamid=385368&seasonid=9162"
 
 VCR.configure do |c|
   c.cassette_library_dir = 'spec/vcr_cassettes'

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: schedule-scraper
 version: !ruby/object:Gem::Version
-  version: 0.1.0
+  version: 0.2.0
   prerelease: 
 platform: ruby
 authors:
@@ -9,11 +9,11 @@ authors:
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2012-06-18 00:00:00.000000000 Z
+date: 2012-06-19 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: nibbler
-  requirement: &70142865159900 !ruby/object:Gem::Requirement
+  requirement: &70168497621260 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -21,10 +21,21 @@ dependencies:
         version: 1.3.0
   type: :runtime
   prerelease: false
-  version_requirements: *70142865159900
+  version_requirements: *70168497621260
+- !ruby/object:Gem::Dependency
+  name: nokogiri
+  requirement: &70168497620760 !ruby/object:Gem::Requirement
+    none: false
+    requirements:
+    - - ~>
+      - !ruby/object:Gem::Version
+        version: 1.5.4
+  type: :runtime
+  prerelease: false
+  version_requirements: *70168497620760
 - !ruby/object:Gem::Dependency
   name: minitest
-  requirement: &70142865156300 !ruby/object:Gem::Requirement
+  requirement: &70168497620300 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -32,10 +43,10 @@ dependencies:
         version: 3.1.0
   type: :development
   prerelease: false
-  version_requirements: *70142865156300
+  version_requirements: *70168497620300
 - !ruby/object:Gem::Dependency
   name: rake
-  requirement: &70142865155280 !ruby/object:Gem::Requirement
+  requirement: &70168497619820 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -43,10 +54,10 @@ dependencies:
         version: 0.9.2
   type: :development
   prerelease: false
-  version_requirements: *70142865155280
+  version_requirements: *70168497619820
 - !ruby/object:Gem::Dependency
   name: turn
-  requirement: &70142865153960 !ruby/object:Gem::Requirement
+  requirement: &70168497619320 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -54,10 +65,10 @@ dependencies:
         version: 0.9.5
   type: :development
   prerelease: false
-  version_requirements: *70142865153960
+  version_requirements: *70168497619320
 - !ruby/object:Gem::Dependency
   name: vcr
-  requirement: &70142865171400 !ruby/object:Gem::Requirement
+  requirement: &70168497635140 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -65,10 +76,10 @@ dependencies:
         version: 2.2.0
   type: :development
   prerelease: false
-  version_requirements: *70142865171400
+  version_requirements: *70168497635140
 - !ruby/object:Gem::Dependency
   name: fakeweb
-  requirement: &70142865170560 !ruby/object:Gem::Requirement
+  requirement: &70168497634600 !ruby/object:Gem::Requirement
     none: false
     requirements:
     - - ~>
@@ -76,7 +87,7 @@ dependencies:
         version: 1.3.0
   type: :development
   prerelease: false
-  version_requirements: *70142865170560
+  version_requirements: *70168497634600
 description: Scrapes online schedules and provides portable versions
 email:
 - john@threedogconsulting.com