publicstorage 0.1.2 → 0.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: fc55c7ceabfcca39423c75b3315ded193267c2d02da648056b472c4f6e5ca64a
-  data.tar.gz: ed7aaea3035fe7aaa876febfa809a90cf9559080e61aeca1fdffb8ed338bba61
+  metadata.gz: 34a6d2f42dac72b5aac92461444a57e40f3dd42ace468fc24f61ac62c81d09e1
+  data.tar.gz: bbb5c737bcf5236e4236fe6d0b336497557f1e300938924b1eb9727c204a9ff1
 SHA512:
-  metadata.gz: b4d628a25a0af8d0754c98868d3a7cd18c2a8b7d6d9dc6f66fbd1a1a83454aea72c108806f3e25a4311d5aa80862cb4a7798cd5645e08484b4f44179139f1cd3
-  data.tar.gz: 46684da48469510751828b452b5f00381caf94b460da89383675429ed8974878355e03cb5e2fdc879db59b5a8e591a83a557b77f131558a4ad91fafb84dc05d9
+  metadata.gz: b1bed2e588707393e89bf67a416bc8f3622dda29e3c8b4754cfd015f224555530a628cce7cbb68e5c4787c745905493ae8c6e6efe14a8c6d84fae852f213474b
+  data.tar.gz: e46336139bbf2f3054cff093e650b88c5e84b7bb0beaa876b5f833d192c641fa3e48449ca27a3af1215a2fdd9dafd284c937348951248e5704ca62668b43f11d

data/README.md

@@ -12,6 +12,17 @@
 gem install publicstorage
 ```
 
+## Configuration
+
+```ruby
+require 'publicstorage'
+
+PublicStorage.configure do |config|
+  config.user_agent = '../..' # ENV['PUBLICSTORAGE_USER_AGENT']
+  config.timeout = 30 # ENV['PUBLICSTORAGE_TIMEOUT']
+end
+```
+
 ## Usage
 
 ```ruby
@@ -20,24 +31,20 @@ require 'publicstorage'
 sitemap = PublicStorage::Facility.sitemap
 sitemap.links.each do |link|
   url = link.loc
+  facility = ExtraSpace::Facility.fetch(url:)
 
-  facility = PublicStorage::Facility.fetch(url:)
-
-  puts "Street: #{facility.address.street}"
-  puts "City: #{facility.address.city}"
-  puts "State: #{facility.address.state}"
-  puts "ZIP: #{facility.address.zip}"
-  puts "Latitude: #{facility.geocode.latitude}"
-  puts "Longitude: #{facility.geocode.longitude}"
-  puts
+  puts facility.text
 
   facility.prices.each do |price|
-    puts "ID: #{price.id}"
-    puts "Width: #{price.dimensions.width}"
-    puts "Depth: #{price.dimensions.depth}"
-    puts "SQFT: #{price.dimensions.sqft}"
-    puts "Rates: $#{price.rates.street} (street) / $#{price.rates.web} (web)"
-    puts
+    puts price.text
   end
+
+  puts
 end
 ```
+
+## CLI
+
+```bash
+publicstorage crawl
+```

data/exe/publicstorage

@@ -0,0 +1,7 @@
+#!/usr/bin/env ruby
+# frozen_string_literal: true
+
+require 'publicstorage'
+
+cli = PublicStorage::CLI.new
+cli.parse

data/lib/publicstorage/address.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
 module PublicStorage
-  # Represents an address associated with a facility.
+  # The address (street + city + state + zip) of a facility.
   class Address
     # @attribute [rw] street
     #   @return [String]
@@ -41,6 +41,11 @@ module PublicStorage
       "#<#{self.class.name} #{props.join(' ')}>"
     end
 
+    # @return [String]
+    def text
+      "#{street}, #{city}, #{state} #{zip}"
+    end
+
     # @param data [Hash]
     #
     # @return [Address]

data/lib/publicstorage/cli.rb

@@ -0,0 +1,58 @@
+# frozen_string_literal: true
+
+require 'optparse'
+
+module PublicStorage
+  # Used when interacting with the library from the command line interface (CLI).
+  #
+  # Usage:
+  #
+  #   cli = PublicStorage::CLI.new
+  #   cli.parse
+  class CLI
+    module Code
+      OK = 0
+      ERROR = 1
+    end
+
+    # @param argv [Array<String>]
+    def parse(argv = ARGV)
+      parser.parse!(argv)
+      command = argv.shift
+
+      case command
+      when 'crawl' then crawl
+      else
+        warn("unsupported command=#{command.inspect}")
+        exit(Code::ERROR)
+      end
+    end
+
+    private
+
+    def crawl
+      PublicStorage::Facility.crawl
+      exit(Code::OK)
+    end
+
+    def help(options)
+      puts(options)
+      exit(Code::OK)
+    end
+
+    def version
+      puts(VERSION)
+      exit(Code::OK)
+    end
+
+    # @return [OptionParser]
+    def parser
+      OptionParser.new do |options|
+        options.banner = 'usage: PublicStorage [options] <command> [<args>]'
+
+        options.on('-h', '--help', 'help') { help(options) }
+        options.on('-v', '--version', 'version') { version }
+      end
+    end
+  end
+end

data/lib/publicstorage/config.rb

@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+
+module PublicStorage
+  # The core configuration.
+  class Config
+    # @attribute [rw] user_agent
+    #   @return [String]
+    attr_accessor :user_agent
+
+    # @attribute [rw] timeout
+    #   @return [Integer]
+    attr_accessor :timeout
+
+    def initialize
+      @user_agent = ENV.fetch('PUBLIC_STORAGE_USER_AGENT', "publicstorage.rb/#{VERSION}")
+      @timeout = Integer(ENV.fetch('PUBLICSTORAGE_TIMEOUT', 60))
+    end
+  end
+end

data/lib/publicstorage/crawler.rb

@@ -3,6 +3,8 @@
 module PublicStorage
   # Used to fetch and parse either HTML or XML via a URL.
   class Crawler
+    HOST = 'https://www.publicstorage.com'
+
     # Raised for unexpected HTTP responses.
     class FetchError < StandardError
       # @param url [String]
@@ -26,11 +28,23 @@ module PublicStorage
       new.xml(url:)
     end
 
+    # @return [HTTP::Client]
+    def connection
+      @connection ||= begin
+        config = PublicStorage.config
+
+        connection = HTTP.persistent(HOST)
+        connection = connection.headers('User-Agent' => config.user_agent) if config.user_agent
+        connection = connection.timeout(config.timeout) if config.timeout
+        connection
+      end
+    end
+
     # @param url [String]
     # @return [HTTP::Response]
     def fetch(url:)
-      response = HTTP.get(url)
-      raise FetchError(url:, response: response.flush) unless response.status.ok?
+      response = connection.get(url)
+      raise FetchError.new(url:, response: response.flush) unless response.status.ok?
 
       response
     end

data/lib/publicstorage/dimensions.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
 module PublicStorage
-  # The dimensions associated with a price.
+  # The dimensions (width + depth + sqft) of a price.
   class Dimensions
     SELECTOR = '.unit-size'
 
@@ -36,6 +36,11 @@ module PublicStorage
       "#<#{self.class.name} #{props.join(' ')}>"
     end
 
+    # @return [String] e.g. "10' × 10' (100 sqft)"
+    def text
+      "#{format('%g', @width)}' × #{format('%g', @depth)}' (#{@sqft} sqft)"
+    end
+
     # @param element [Nokogiri::XML::Element]
     #
     # @return [Dimensions]

data/lib/publicstorage/facility.rb

@@ -6,6 +6,17 @@ module PublicStorage
     SITEMAP_URL = 'https://www.publicstorage.com/sitemap_0-product.xml'
 
     PRICE_SELECTOR = '.units-results-section .unit-list-group .unit-list-item'
+    LD_SELECTOR = 'script[type="application/ld+json"]'
+
+    ID_REGEX = /(?<id>\d+)\.html/
+
+    # @attribute [rw] id
+    #   @return [Integer]
+    attr_accessor :id
+
+    # @attribute [rw] name
+    #   @return [String]
+    attr_accessor :name
 
     # @attribute [rw] address
     #   @return [Address]
@@ -19,25 +30,6 @@ module PublicStorage
     #   @return [Array<Price>]
     attr_accessor :prices
 
-    # @param address [Address]
-    # @param geocode [Geocode]
-    # @param prices [Array<Price>]
-    def initialize(address:, geocode:, prices:)
-      @address = address
-      @geocode = geocode
-      @prices = prices
-    end
-
-    # @return [String]
-    def inspect
-      props = [
-        "address=#{@address.inspect}",
-        "geocode=#{@geocode.inspect}",
-        "prices=#{@prices.inspect}"
-      ]
-      "#<#{self.class.name} #{props.join(' ')}>"
-    end
-
     # @return [Sitemap]
     def self.sitemap
       Sitemap.fetch(url: SITEMAP_URL)
@@ -55,14 +47,65 @@ module PublicStorage
     #
     # @return [Facility]
     def self.parse(document:)
-      data = JSON.parse(document.at_css('script[type="application/ld+json"]').text)
-      item = data.find { |entry| entry['@type'] == 'SelfStorage' }
-      address = Address.parse(data: item['address'])
-      geocode = Geocode.parse(data: item['geo'])
+      data = parse_ld(document:)
+      id = Integer(data['url'].match(ID_REGEX)[:id])
+      name = data['name']
+      address = Address.parse(data: data['address'])
+      geocode = Geocode.parse(data: data['geo'])
 
       prices = document.css(PRICE_SELECTOR).map { |element| Price.parse(element:) }
 
-      new(address:, geocode:, prices:)
+      new(id:, name:, address:, geocode:, prices:)
+    end
+
+    # @param document [NokoGiri::XML::Document]
+    #
+    # @return [Hash]
+    def self.parse_ld(document:)
+      JSON.parse(document.at_css(LD_SELECTOR).text).find { |entry| entry['@type'] == 'SelfStorage' }
+    end
+
+    def self.crawl
+      sitemap.links.each do |link|
+        url = link.loc
+
+        facility = fetch(url:)
+        puts facility.text
+
+        facility.prices.each do |price|
+          puts price.text
+        end
+
+        puts
+      end
+    end
+
+    # @param id [String]
+    # @param name [String]
+    # @param address [Address]
+    # @param geocode [Geocode]
+    # @param prices [Array<Price>]
+    def initialize(id:, name:, address:, geocode:, prices:)
+      @id = id
+      @name = name
+      @address = address
+      @geocode = geocode
+      @prices = prices
+    end
+
+    # @return [String]
+    def inspect
+      props = [
+        "address=#{@address.inspect}",
+        "geocode=#{@geocode.inspect}",
+        "prices=#{@prices.inspect}"
+      ]
+      "#<#{self.class.name} #{props.join(' ')}>"
+    end
+
+    # @return [String]
+    def text
+      "#{@id} | #{@name} | #{@address.text} | #{@geocode.text}"
     end
   end
 end

data/lib/publicstorage/geocode.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
 module PublicStorage
-  # A geocode associated with a facility.
+  # The geocode (latitude + longitude) of a facility.
   class Geocode
     # @attribute [rw] latitude
     #   @return [Float]
@@ -27,6 +27,11 @@ module PublicStorage
       "#<#{self.class.name} #{props.join(' ')}>"
     end
 
+    # @return [String]
+    def text
+      "#{@latitude},#{@longitude}"
+    end
+
     # @param data [Hash]
     #
     # @return [Geocode]

data/lib/publicstorage/price.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
 module PublicStorage
-  # A price associated with a unit.
+  # The price (id + dimensions + rate) for a facility
   class Price
     # @attribute [rw] id
     #   @return [String]
@@ -34,6 +34,11 @@ module PublicStorage
       "#<#{self.class.name} #{props.join(' ')}>"
     end
 
+    # @return [String] e.g. "123 | 5' × 5' (25 sqft) | $100 (street) / $90 (web)"
+    def text
+      "#{@id} | #{@dimensions.text} | #{@rates.text}"
+    end
+
     # @param element [Nokogiri::XML::Element]
     #
     # @return [Price]

data/lib/publicstorage/rates.rb

@@ -1,7 +1,7 @@
 # frozen_string_literal: true
 
 module PublicStorage
-  # The rates associated with a price
+  # The rates (street + web) for a facility
   class Rates
     STREET_SELECTOR = '.unit-prices .unit-pricing .unit-strike-through-price'
     WEB_SELECTOR = '.unit-prices .unit-pricing .unit-price'
@@ -30,6 +30,11 @@ module PublicStorage
       "#<#{self.class.name} #{props.join(' ')}>"
     end
 
+    # @return [String] e.g. "$80 (street) | $60 (web)"
+    def text
+      "$#{@street} (street) | $#{@web} (web)"
+    end
+
     # @param element [Nokogiri::XML::Element]
     #
     # @return [Rates]

data/lib/publicstorage/sitemap.rb

@@ -1,7 +1,9 @@
 # frozen_string_literal: true
 
 module PublicStorage
-  # A sitemap.
+  # A sitemap on publicstorage.com.
+  #
+  # e.g. https://www.publicstorage.com/sitemap_0-product.xml
   class Sitemap
     # @attribute [rw] links
     #   @return [Array<Link>]

data/lib/publicstorage/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module PublicStorage
-  VERSION = '0.1.2'
+  VERSION = '0.3.0'
 end

data/lib/publicstorage.rb

@@ -6,8 +6,21 @@ require 'zeitwerk'
 
 loader = Zeitwerk::Loader.for_gem
 loader.inflector.inflect 'publicstorage' => 'PublicStorage'
+loader.inflector.inflect 'cli' => 'CLI'
 loader.setup
 
+# An interface for PublicStorage.
 module PublicStorage
   class Error < StandardError; end
+
+  # @return [Config]
+  def self.config
+    @config ||= Config.new
+  end
+
+  # @yield [config]
+  # @yieldparam config [Config]
+  def self.configure
+    yield config
+  end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: publicstorage
 version: !ruby/object:Gem::Version
-  version: 0.1.2
+  version: 0.3.0
 platform: ruby
 authors:
 - Kevin Sylvestre
@@ -69,7 +69,8 @@ dependencies:
 description: Uses HTTP.rb to scrape publicstorage.com.
 email:
 - kevin@ksylvest.com
-executables: []
+executables:
+- publicstorage
 extensions: []
 extra_rdoc_files: []
 files:
@@ -77,8 +78,11 @@ files:
 - README.md
 - bin/console
 - bin/setup
+- exe/publicstorage
 - lib/publicstorage.rb
 - lib/publicstorage/address.rb
+- lib/publicstorage/cli.rb
+- lib/publicstorage/config.rb
 - lib/publicstorage/crawler.rb
 - lib/publicstorage/dimensions.rb
 - lib/publicstorage/facility.rb