added support for Comment and Post entities. This is more complicated…

… than I thought

Gemfile

@@ -1,4 +1,5 @@
 source "http://rubygems.org"
 
 gem "rest-client", "~> 1.6.7"
-gem "nokogiri", "~> 1.5.5"
+gem "nokogiri", "~> 1.5.5"
+gem "chronic", "~> 0.7.0"

hn2json.gemspec

@@ -15,6 +15,7 @@ Gem::Specification.new do |s|
 
   s.add_dependency    "rest-client",  "~> 1.6.7"
   s.add_dependency    "nokogiri",  "~> 1.5.5"
+  s.add_dependency    "chronic",  "~> 0.7.0"
 
   #s.files            += Dir.glob("bin/**/*")
   #s.executables       = %w( hn2json )

lib/hn2json.rb

@@ -1,5 +1,6 @@
 require 'rest-client'
 require 'nokogiri'
+require 'chronic'
 
 module HN2JSON
   autoload :Request,    'hn2json/request'

lib/hn2json/entity.rb

@@ -3,7 +3,7 @@ module HN2JSON
   class Entity
 
     attr_accessor :type, :id, :parent, :url, :title, :comments, :votes
-    attr_accessor :full_text, :posted_by, :date_posted, :voting_on
+    attr_accessor :fulltext, :posted_by, :date_posted, :voting_on
 
     def initialize id
       @id = id
@@ -12,7 +12,7 @@ def initialize id
       @parent = nil
       @url = nil
       @title = nil
-      @full_text = nil
+      @fulltext = nil
       @posted_by = nil
       @date_posted = nil
       @voting_on = nil
@@ -21,6 +21,8 @@ def initialize id
 
       get_page
       determine_type
+
+      get_attrs
     end
 
     def get_page
@@ -32,6 +34,14 @@ def determine_type
       @type = @parser.determine_type
     end
 
+    def get_attrs
+      eval("@parser.get_attrs_#{@type.to_s} self")
+    end
+
+    def add_attrs
+      yield self
+    end
+
   end
 
 end

lib/hn2json/parser.rb

@@ -29,6 +29,103 @@ def determine_type
 
     end
 
+    def get_attrs_comment entity
+
+      parent_url = @doc.css('.comhead a')[2]['href']
+      parent_regex = /id\=(.*)/
+      parent = parent_regex.match(parent_url)[1]
+
+      fulltext_nolinks = @doc.css('.comment')[0].to_s
+      fulltext_nolinks = fulltext_nolinks.gsub(/\<a\shref\=['"](.*)['"].*rel\=.*\>.*\<\/a\>/, '\1')
+      fulltext = fulltext_nolinks.gsub(/<\/?[^>]*>/, '')
+
+      comhead = @doc.css('.comhead')[0]
+
+      date_regex = /.*\s(.*\s.*\sago)/
+      ago = date_regex.match(comhead.content)[1]
+      date_posted = Chronic.parse(ago).to_s
+
+      posted_by = comhead.css('a')[0].content
+
+      comments = []
+
+      full_comments = @doc.css('td > img[width="0"]').xpath("..").xpath("..").css('.default')
+
+      full_comments.each do |comment|
+        comment_id = comment.css('span a')[1]['href'].gsub("item?id=", '')
+        comments.push comment_id
+      end
+
+      entity.add_attrs do |e|
+        e.parent = parent
+        e.fulltext = fulltext
+        e.date_posted = date_posted
+        e.comments = comments
+        e.posted_by = posted_by
+      end
+    end
+
+
+    def get_attrs_post entity
+
+      subtext = @doc.css('.subtext')[0]
+
+      date_regex = /.*\s(.*\s.*\sago)/
+      ago = date_regex.match(subtext.content)[1]
+      date_posted = Chronic.parse(ago).to_s
+
+      posted_by = subtext.css('a')[0].content
+
+      votes = subtext.css('span')[0].content.to_i
+
+      comments = []
+
+      full_comments = @doc.css('td > img[width="0"]').xpath("..").xpath("..").css('.default')
+
+      full_comments.each do |comment|
+        comment_id = comment.css('span a')[1]['href'].gsub("item?id=", '')
+        comments.push comment_id
+      end
+
+      head = @doc.css('.title a')[0]
+
+      title = head.content
+
+      url = head['href']
+
+      entity.add_attrs do |e|
+        e.url = url
+        e.title = title
+        e.date_posted = date_posted
+        e.comments = comments
+        e.votes = votes
+        e.posted_by = posted_by
+      end
+    end
+
+#    def get_attrs_poll entity
+#      entity.add_attrs do |e|
+#        e.title =
+#        e.fulltext =
+#        e.date_posted =
+#        e.posted_by =
+#        e.votes =
+#        e.comments =
+#        voting_on =
+#      end
+#    end
+#
+#    def get_attrs_discussion entity
+#      entity.add_attrs do |e|
+#        e.title =
+#        e.fulltext =
+#        e.date_posted =
+#        e.posted_by =
+#        e.comments =
+#        e.votes =
+#      end
+#    end
+
   end
 
 end

lib/hn2json/version.rb

@@ -1,3 +1,3 @@
 module HN2JSON
-  VERSION = '0.0.0'
+  VERSION = '0.0.1'
 end