oald_parser 0.2.1 → 0.2.2

data/lib/oald_parser/facade.rb

@@ -16,7 +16,7 @@ module OaldParser
     def self.create_facade
       downloader = PageDownloader.new('http://www.oxfordadvancedlearnersdictionary.com/dictionary')
       parser = PageParser.new
-      formatter = Formatter.new(items: 5)
+      formatter = Formatter.new
       extractor = WordExtractor.new
       Facade.new(downloader, parser, formatter, extractor)
     end

data/lib/oald_parser/formatter.rb

@@ -2,36 +2,29 @@ require 'nokogiri'
 
 module OaldParser
   class Formatter
-    def initialize(options)
-      @options = options
-    end
-
     def format(page)
-      if !page.blocks.empty?
-        format_blocks(page.blocks)
-      else
-        format_items(page.items)  
-      end
+      format_blocks(page.blocks)
     end
 
     private
-    def format_blocks(blocks, limit = 1000)
+    def format_blocks(blocks)
       blocks.collect do |block|
         res = ''
-        res += block.text.upcase
-        res += "\n"
-        res += '-' * 20
-        res += "\n"
+        unless block.text.empty?
+          res += block.text.upcase
+          res += "\n"
+          res += '-' * 20
+          res += "\n"
+        end
         res += format_items(block.items)
         res
       end.join("\n\n")
     end
 
-    def format_items(items, limit = 1000)
+    def format_items(items)
       items.collect do |item|
-        res = ''
-        res += item.text
-        if !item.examples.empty?
+        res = item.text
+        unless item.examples.empty?
           res += "\n"
           res += format_examples(item.examples)
         end
@@ -40,7 +33,7 @@ module OaldParser
     end
 
     def format_examples(examples)
-      examples.collect {|e| "+ #{e}"}.join("\n")
+      examples.collect{|e| "+ #{e}"}.join("\n")
     end
   end
 end

data/lib/oald_parser/page_parser.rb

@@ -2,9 +2,13 @@ require 'nokogiri'
 
 module OaldParser
   class Page
-    attr_reader :blocks, :items
-    def initialize(blocks, items)
-      @blocks, @items = blocks, items
+    attr_reader :blocks
+    def initialize(blocks)
+      @blocks = blocks
+    end
+
+    def self.empty
+      Page.new []
     end
   end
 
@@ -26,26 +30,28 @@ module OaldParser
     def parse(page)
       parsed = Nokogiri::HTML(page)
       if blocks_on_page? parsed
-        Page.new(parse_block(parsed), [])
-      elsif def_on_page? parsed
-        Page.new([], parse_def(parsed))      
+        Page.new(parse_block(parsed))
       else
-        Page.new([], parse_items(parsed))
+        parse_page_from_items(parsed)
       end
     end
 
     private
-    def blocks_on_page?(page)
-      page.css('div.sd-g').first
+    def parse_page_from_items(parsed)
+      items = parse_items(parsed)
+      if items.empty?
+        Page.empty
+      else
+        Page.new([Block.new("", items)])
+      end      
     end
 
-    def def_on_page?(page)
-      page.css('div.h-g').first
+    def blocks_on_page?(page)
+      page.css('div.sd-g').first
     end
 
     def parse_block(page)
-      block_nodes = page.css('div.sd-g')
-      block_nodes.collect do |block|
+      page.css('div.sd-g').collect do |block|
         block_text  = all_except(block, 'n-g')
         items = parse_items(block)
         Block.new(block_text, items)
@@ -53,19 +59,16 @@ module OaldParser
     end
 
     def parse_items(block)
-      item_nodes = block.css('span.n-g')
-      item_nodes.collect do |item|
+      items = block.css('span.n-g').collect do |item|
         item_text = all_except(item, 'x-g')
         Item.new(item_text, parse_examples(item))
-      end      
-    end
+      end
 
-    def parse_def(page)
-      item_nodes = page.css('div.h-g')
-      item_nodes.collect do |item|
-        item_text = item.css('div.def_block').first.text.strip
-        Item.new(item_text, parse_examples(item))
-      end        
+      if block.css('div.def_block').first
+        item_text = block.css('div.def_block').first.text.strip
+        items << Item.new(item_text, parse_examples(block))
+      end
+      items
     end
 
     def parse_examples(item)

data/lib/oald_parser.rb

@@ -8,9 +8,13 @@ require_relative 'oald_parser/page_parser'
 #include OaldParser
 #
 #downloader = PageDownloader.new("http://www.oxfordadvancedlearnersdictionary.com/dictionary")
-#page = downloader.download("prevent")
+#page = downloader.download("a")
+##puts page
+#
 #parser = PageParser.new
 #parsed = parser.parse(page)
+#puts parsed.inspect
+#
 #formatter = Formatter.new(items: 15)
 #puts formatter.format(parsed)
 

metadata

@@ -5,8 +5,8 @@ version: !ruby/object:Gem::Version
   segments: 
   - 0
   - 2
-  - 1
-  version: 0.2.1
+  - 2
+  version: 0.2.2
 platform: ruby
 authors: 
 - Victor Savkin