julien51-feedzirra 0.0.13

data/lib/feedzirra/parser/rss_entry.rb

@@ -0,0 +1,40 @@
+module Feedzirra
+  
+  module Parser
+    # == Summary
+    # Parser for dealing with RDF feed entries.
+    #
+    # == Attributes
+    # * title
+    # * url
+    # * author
+    # * content
+    # * summary
+    # * published
+    # * categories
+    class RSSEntry
+      include SAXMachine
+      include FeedEntryUtilities
+      element :title
+      element :link, :as => :url
+
+      element :"dc:creator", :as => :author
+      element :"content:encoded", :as => :content
+      element :description, :as => :summary
+
+      element :pubDate, :as => :published
+      element :"dc:date", :as => :published
+      element :"dc:Date", :as => :published
+      element :"dcterms:created", :as => :published
+
+
+      element :"dcterms:modified", :as => :updated
+      element :issued, :as => :published
+      elements :category, :as => :categories
+
+      element :guid, :as => :id
+    end
+
+  end
+  
+end

data/lib/feedzirra/push_parser.rb

@@ -0,0 +1,56 @@
+module Feedzirra
+  ##
+  # Contrary to Feedzirra::Feed, Feedzirra::PushParser doesn't expect
+  # the whole document to be given in one String, but allows
+  # subsequent parsing of chunks.
+  class PushParser
+    ##
+    # How many bytes to buffer before starting to parse, helps
+    # Feedzirra's content detection.
+    BUF_MIN_THRESHOLD = 1000
+
+    ##
+    # Just resets instance variables
+    def initialize
+      @buf = ''
+      @parser = nil
+    end
+
+    ##
+    # Either buffer up til BUF_MIN_THRESHOLD or, if reached, actually
+    # parse a chunk
+    def push(chunk)
+      if @parser
+        @parser.parse(chunk)
+      else
+        @buf += chunk
+        if @buf.size > BUF_MIN_THRESHOLD
+          start_parsing
+        end
+      end
+    end
+
+    ##
+    # Really start parsing, if BUF_MIN_THRESHOLD wasn't reached yet,
+    # finalize, and return the actual parser/document
+    def finish
+      # TODO: if we haven't started yet we won't even need a
+      # PushParser
+      start_parsing unless @parser
+
+      @parser.parse_finish
+      @parser
+    end
+
+    private
+
+    def start_parsing
+      unless klass = Feed::determine_feed_parser_for_xml(@buf)
+        raise NoParserAvailable.new("No valid parser for XML.")
+      end
+      @parser = klass.new
+      @parser.parse(@buf)
+      @buf = nil
+    end
+  end
+end

data/lib/feedzirra.rb

@@ -0,0 +1,37 @@
+$LOAD_PATH.unshift(File.dirname(__FILE__)) unless $LOAD_PATH.include?(File.dirname(__FILE__))
+
+gem 'activesupport'
+
+require 'zlib'
+require 'curb'
+require 'sax-machine'
+require 'dryopteris'
+require 'uri'
+require 'active_support/basic_object'
+require 'active_support/core_ext/object'
+require 'active_support/core_ext/time'
+
+require 'core_ext/date'
+require 'core_ext/string'
+
+require 'feedzirra/feed_utilities'
+require 'feedzirra/feed_entry_utilities'
+require 'feedzirra/feed'
+require 'feedzirra/push_parser'
+
+require 'feedzirra/push_parser'
+
+require 'feedzirra/parser/rss_entry'
+require 'feedzirra/parser/itunes_rss_owner'
+require 'feedzirra/parser/itunes_rss_item'
+require 'feedzirra/parser/atom_entry'
+require 'feedzirra/parser/atom_feed_burner_entry'
+
+require 'feedzirra/parser/rss'
+require 'feedzirra/parser/itunes_rss'
+require 'feedzirra/parser/atom'
+require 'feedzirra/parser/atom_feed_burner'
+
+module Feedzirra
+  VERSION = "0.0.12"
+end

data/spec/feedzirra/feed_entry_utilities_spec.rb

@@ -0,0 +1,52 @@
+require File.dirname(__FILE__) + '/../spec_helper'
+
+describe Feedzirra::FeedUtilities do
+  before(:each) do
+    @klass = Class.new do
+      include Feedzirra::FeedEntryUtilities
+    end
+  end
+  
+  describe "handling dates" do
+    it "should parse an ISO 8601 formatted datetime into Time" do
+      time = @klass.new.parse_datetime("2008-02-20T8:05:00-010:00")
+      time.class.should == Time
+      time.to_s.should == "Wed Feb 20 18:05:00 UTC 2008"
+    end
+  end
+  
+  describe "sanitizing" do
+    before(:each) do
+      @feed = Feedzirra::Feed.parse(sample_atom_feed)
+      @entry = @feed.entries.first
+    end
+    
+    it "should provide a sanitized title" do
+      new_title = "<script>" + @entry.title
+      @entry.title = new_title
+      @entry.title.sanitize.should == Dryopteris.sanitize(new_title)
+    end
+    
+    it "should sanitize content in place" do
+      new_content = "<script>" + @entry.content
+      @entry.content = new_content.dup
+      @entry.content.sanitize!.should == Dryopteris.sanitize(new_content)
+      @entry.content.should == Dryopteris.sanitize(new_content)
+    end
+    
+    it "should sanitize things in place" do
+      @entry.title   += "<script>"
+      @entry.author  += "<script>"
+      @entry.content += "<script>"
+
+      cleaned_title   = Dryopteris.sanitize(@entry.title)
+      cleaned_author  = Dryopteris.sanitize(@entry.author)
+      cleaned_content = Dryopteris.sanitize(@entry.content)
+      
+      @entry.sanitize!
+      @entry.title.should   == cleaned_title
+      @entry.author.should  == cleaned_author
+      @entry.content.should == cleaned_content
+    end
+  end
+end