proto 0.0.4 → 0.0.5

data/lib/proto/scraper.rb

@@ -3,7 +3,7 @@ module Proto
     attr_accessor :url, :doc, :url_collection
 
     def initialize(url)
-      @url = url.chomp '/'
+      @url = url.chomp '/' #remove trailing slash
       @doc = Nokogiri::HTML(open(url))
     end
 
@@ -15,45 +15,38 @@ module Proto
 
     def fetch(name='Type', args)
       if url_collection
-        attributes = visit_urls_and_fetch(args)
-        protos     = create_return_objects(name, attributes)
-        return protos        
+        attributes = scrape_multiple_pages(args)
       else
-        attributes = scrape_attribute_data(args)
-        protos     = create_return_objects(name, attributes)
-        return protos
+        attributes = scrape_single_page(args)
       end
+      protos = create_return_objects(name, attributes)
+      return protos
     end
     alias_method :fetch_and_create!, :fetch
 
   private
 
-    def visit_urls_and_fetch(attributes)
-      hash_array = []
-      final_array = url_collection.map do |url|
-        page  = Nokogiri::HTML(open(url))
-        attrs_hash = gather_data(page, attributes)
-        hash_array << attrs_hash
+    def scrape_multiple_pages(attributes)
+      url_collection.each_with_object([]).map do |url, hash_array|
+         gather_data(url, attributes)
       end
-      return hash_array
     end
     
-    def gather_data(page, attributes)
-      job_hash = attributes.each_with_object({}) do |(key, selector), attrs|
+    def gather_data(url, attributes)
+      page = Nokogiri::HTML(open(url))
+      attributes.each_with_object({}) do |(key, selector), attrs|
         attrs[key] = page.css(selector).text.strip
       end
     end
 
-    def scrape_attribute_data(document=self.doc, attributes)
-      length_of_scrape = document.css(attributes.first[1]).count
+    def scrape_single_page(attributes)
+      length_of_scrape = doc.css(attributes.first[1]).count
       
-      final_array = length_of_scrape.times.map do |index|
-        attributes.inject(Hash.new) do |hash, (attr_name, selector)|
-          hash.merge(attr_name => document.css(selector)[index].text.strip) if document.css(selector)[index]
+      length_of_scrape.times.map do |index|
+        attributes.inject({}) do |hash, (attr_name, selector)|
+          hash.merge(attr_name => doc.css(selector)[index].text.strip) if doc.css(selector)[index]
         end
-      end
-
-      final_array.compact
+      end.compact
     end
 
     def create_return_objects(name, attributes)
@@ -62,4 +55,4 @@ module Proto
       attributes.map { |hash| Proto.const_get(name).new(hash) }
     end
   end
-end
+end

data/lib/proto/version.rb

@@ -1,3 +1,3 @@
 module Proto
-  VERSION = "0.0.4"
+  VERSION = "0.0.5"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: proto
 version: !ruby/object:Gem::Version
-  version: 0.0.4
+  version: 0.0.5
   prerelease: 
 platform: ruby
 authors: