feed_yamlizer 0.0.9 → 0.1.0

data/lib/feed_yamlizer/html_listener.rb

@@ -28,11 +28,9 @@ class FeedYamlizer
 
       digits = @links.size.to_s.size 
 
+      x = Nokogiri::HTML.parse(x).inner_text
       # wrap the text
-      x = format(x)
-
-      # delete extra blank lines
-      x = x.split(/\n\n+/).join("\n\n")
+      x = FeedYamlizer.format(x)
 
       # format the blockquotes
       line_buffer = []
@@ -44,7 +42,7 @@ class FeedYamlizer
         elsif line == '[/blockquote]'
           inblock = false
           block = blockquote_buffer.join("\n")
-          line_buffer << format(block, '-c')
+          line_buffer << ( FeedYamlizer.format(block, '-c') )
           blockquote_buffer = []
         else
           if inblock 
@@ -55,8 +53,9 @@ class FeedYamlizer
         end
       end
       x = line_buffer.join("\n")
+      
 
-      x + "\n\n" + @links.map {|x| 
+      res = x + "\n\n" + @links.map {|x| 
         gutter = x[:index].to_s.rjust(digits)
         if x[:content] && x[:content].strip.length > 0
           %Q|#{gutter}. "#{x[:content].gsub(/[\r\n]+/, ' ').strip}"\n#{' ' * (digits + 2)}#{x[:href]}|
@@ -64,6 +63,7 @@ class FeedYamlizer
           "#{gutter}. #{x[:href]}"
         end
       }.join("\n")
+      res 
     end
 
     def strip_empty_tags(line)
@@ -91,7 +91,7 @@ class FeedYamlizer
       when 'ul', 'ol', 'dl'
         @content << "<#{name}>\n"
       when 'li', 'dt', 'dd'
-        @content[-1] << "\n  * "
+        @content += ["[blockquote]", "", "* "]
       when 'strong', 'em'
         @content[-1] << "<#{name}>"
       when *BLOCK_TAGS
@@ -112,11 +112,11 @@ class FeedYamlizer
       when *HEADER_TAGS
         @content[-1] << "</#{UNIFORM_HEADER_TAG}>" 
       when 'blockquote'
-        @content << '[/blockquote]'
+        @content += ["","[/blockquote]"]
       when 'ul', 'ol', 'dl'
         @content[-1] << "</#{name}>"
       when 'li', 'dt', 'dd'
-        @content[-1] << "\n"
+        @content += ["", "[/blockquote]"]
       when 'strong', 'em'
         @content[-1] << "</#{name}>"
       when *BLOCK_TAGS
@@ -149,13 +149,6 @@ class FeedYamlizer
       @nested_tags.join('/')
     end
 
-    def format(x, flags='')
-      IO.popen("fmt #{flags}", "r+") do |pipe| 
-        pipe.puts x
-        pipe.close_write
-        pipe.read
-      end
-    end
 
   end
 end

data/lib/feed_yamlizer/version.rb

@@ -1,3 +1,3 @@
 class FeedYamlizer
-  VERSION = "0.0.9"
+  VERSION = "0.1.0"
 end

data/lib/feed_yamlizer.rb

@@ -10,7 +10,6 @@ require 'feed_yamlizer/feed_parser'
 require 'feed_yamlizer/html_listener'
 require 'feed_yamlizer/html_cleaner'
 require 'nokogiri'
-require 'feed_yamlizer/textifier'
 require 'fileutils'
 require 'yaml'
 require 'htmlentities'
@@ -18,6 +17,13 @@ require 'string_ext'
 
 class FeedYamlizer 
   include FileUtils::Verbose
+  def self.format(x, flags='')
+    IO.popen("fmt #{flags}", "r+") do |pipe| 
+      pipe.puts x
+      pipe.close_write
+      pipe.read
+    end
+  end
 
   def initialize(feed)
     @feed = feed
@@ -64,14 +70,15 @@ class FeedYamlizer
     @result[:items][-1][:content] = {:html => content}
     # TODO check if HTML or plain text!
     simplified = HtmlCleaner.new(content).output
-    textified = Textifier.new(simplified).output 
     #@result[:items][-1][:content][:simplified] = simplified
-    textified = textified.gsub(FeedYamlizer::NEWLINE_PLACEHOLDER, "\n").
+    textified = simplified.gsub(FeedYamlizer::NEWLINE_PLACEHOLDER, "\n").
         gsub(SPACE_PLACEHOLDER, " ").
         gsub(TAB_PLACEHOLDER, "  ").
         gsub(/^\s+$/, "").
-        # eliminate extra blank lines
-        gsub(/\n{3,}/, "\n\n")
+        # eliminate extra blank lines 
+        #gsub(/\n{3,}(?!\s)/, "awdkljalwkdjalwkjd lawkdj GOLD klajw d\n\n")
+        #gsub(/\n{3,}(?!\s)/m, "\n\n").
+        gsub(/\n *\n *\n *$/ , "\n\n")
     # next two lines are dev lines
     #puts textified
     #exit

metadata

@@ -4,9 +4,9 @@ version: !ruby/object:Gem::Version
   prerelease: false
   segments: 
   - 0
+  - 1
   - 0
-  - 9
-  version: 0.0.9
+  version: 0.1.0
 platform: ruby
 authors: 
 - Daniel Choi
@@ -64,7 +64,6 @@ files:
 - lib/feed_yamlizer/feed_parser.rb
 - lib/feed_yamlizer/html_cleaner.rb
 - lib/feed_yamlizer/html_listener.rb
-- lib/feed_yamlizer/textifier.rb
 - lib/feed_yamlizer/version.rb
 has_rdoc: true
 homepage: https://github.com/danchoi/feed_yamlizer

data/lib/feed_yamlizer/textifier.rb

@@ -1,16 +0,0 @@
-# just takes simplified HTML and converts it to plain text
-class FeedYamlizer
-  class Textifier
-    def initialize(html)
-      @doc = Nokogiri::HTML.parse(html)
-    end
-
-    # TODO beef this up with real effects
-
-    def output
-      @doc.inner_text
-    end
-  end
-end
-
-