Parsing feeds into structs

lib/opds.rb

@@ -1,4 +1,10 @@
 require 'opds/opds'
 require 'opds/support/logging'
 require 'opds/support/browser'
+require 'opds/support/linkset'
 require 'opds/parser'
+
+require 'opds/feed'
+require 'opds/entry'
+require 'opds/acquisition_feed'
+require 'opds/navigation_feed'

lib/opds/acquisition_feed.rb

@@ -0,0 +1,4 @@
+module OPDS
+	class AcquisitionFeed < Feed
+	end
+end

lib/opds/entry.rb

@@ -0,0 +1,7 @@
+module OPDS
+	class Entry
+		def self.from_nokogiri(content)
+			self.new
+		end
+	end
+end

lib/opds/feed.rb

@@ -0,0 +1,125 @@
+module OPDS
+	class Feed
+		include Logging
+		attr_reader :raw_doc
+		attr_reader :entries
+
+
+		def initialize(browser=nil)
+			@browser=browser
+			@browser||=OPDS::Support::Browser.new
+		end
+
+		# access root catalog
+		def root
+			return @root unless root?
+			self
+		end
+
+		# root catalog predicate
+		def root?
+		end
+
+		def self.parse_url(url,browser=nil,parser_opts={})
+			@browser=browser
+			@browser||=OPDS::Support::Browser.new
+			@browser.go_to(url)
+			if @browser.ok?
+				return self.parse_raw(@browser.body,parser_opts)
+			else
+				return false
+			end
+		end
+
+		def self.parse_raw(txt,opts={})
+			parser=OPDSParser.new(opts)
+			pfeed=parser.parse(txt)
+			type=parser.sniffed_type
+			return pfeed
+		end
+
+		def self.from_nokogiri(content)
+			z=self.new
+			z.instance_variable_set('@raw_doc',content)
+			z.serialize!
+			z
+		end
+
+		#read xml entries into entry struct
+		def serialize!
+			@entries=raw_doc.xpath('/xmlns:feed/xmlns:entry',raw_doc.root.namespaces).map do |el|
+				OPDS::Entry.from_nokogiri(el)
+			end
+		end
+
+		def title
+			raw_doc.at('/xmlns:feed/xmlns:title',raw_doc.root.namespaces).text
+		end
+
+		def icon
+			raw_doc.at('/xmlns:feed/xmlns:icon',raw_doc.root.namespaces).text
+		end
+
+		def links
+			if !@links || @links.size ==0
+				@links=OPDS::Support::LinkSet.new
+				raw_doc.xpath('/xmlns:feed/xmlns:link',raw_doc.root.namespaces).each do |n|
+					text=nil
+					text=n.attributes['title'].value unless n.attributes['title'].nil?
+					link=n.attributes['href'].value
+					unless n.attributes['rel'].nil?
+						n.attributes['rel'].value.split.each do |rel|
+							@links.push(rel,link,text)
+						end
+					else
+						@links.push(nil,link,text)
+					end
+				end
+
+			end
+			@links
+		end
+
+		def id
+			raw_doc.at('/xmlns:feed/xmlns:id',raw_doc.root.namespaces).text
+		end
+
+		def author
+			{
+			:name => raw_doc.at('/xmlns:feed/xmlns:author/xmlns:name',raw_doc.root.namespaces).text,
+			:uri => raw_doc.at('/xmlns:feed/xmlns:author/xmlns:uri',raw_doc.root.namespaces).text,
+			:email => raw_doc.at('/xmlns:feed/xmlns:author/xmlns:email',raw_doc.root.namespaces).text
+			}
+		end
+
+
+		def next_page_url
+			links.link_url(:rel => 'next')
+		end
+
+		def prev_page_url
+			links.link_url(:rel => 'prev')
+		end
+
+		def paginated?
+			!next_page_url.nil?||!prev_page_url.nil?
+		end
+
+		def first_page?
+			!prev_page_url if paginated?
+		end
+
+		def last_page?
+			!next_page_url if paginated?
+		end
+
+		def next_page
+			Feed.parse_url(next_page_url,@browser)
+		end
+
+		def prev_page
+			Feed.parse_url(prev_page_url,@browser)
+		end
+
+	end
+end

lib/opds/navigation_feed.rb

@@ -0,0 +1,4 @@
+module OPDS
+	class NavigationFeed  < Feed
+	end
+end

lib/opds/parser.rb

@@ -12,8 +12,13 @@ def initialize(opts={})
 		def parse(content)
 			@ret=Nokogiri::XML(content)
 			@sniffed_type=sniff(@ret)
-			@ret
+			case @sniffed_type
+			when :acquisition then return OPDS::AcquisitionFeed.from_nokogiri(@ret)
+			when :navigation then return OPDS::NavigationFeed.from_nokogiri(@ret)
+			when :entry then return OPDS::Entry.from_nokogiri(@ret)
+			end
 		end
+
 		protected 
 		def sniff(doc)
 			return :entry if doc.root.name=='entry'

lib/opds/support/linkset.rb

@@ -0,0 +1,90 @@
+module OPDS
+	module Support
+		class LinkSet 
+			include Enumerable
+			def initialize
+				@rel_store=Hash.new
+				@txt_store=Hash.new
+				@lnk_store=Hash.new
+				@store=[]
+			end
+
+			def []=(k,v)
+				@store.push [k]+v
+				i=@store.size-1
+				@rel_store[k]=[] unless @rel_store[k]
+				@rel_store[k].push i
+				@txt_store[v.last]=[] unless @txt_store[v.last]
+				@txt_store[v.last].push i
+				@lnk_store[v.first]=[] unless @lnk_store[v.first]
+				@lnk_store[v.first].push i
+
+			end
+
+			def [](k)
+				remap(@rel_store[k])
+			end
+
+			def each(&block)
+				@store.each(&block)
+			end
+
+			def push(rel,link,text=nil)
+				self[rel]=[link,text]
+			end
+
+			def link_url(k)
+				ty,v=k.first
+				t=remap(collection(ty)[v])
+				t.first[1] unless t.nil?
+			end
+
+			def link_rel(k)
+				ty,v=k.first
+				t=remap(collection(ty)[v])
+				t.first[0] unless t.nil?
+			end
+
+			def link_text(k)
+				ty,v=k.first
+				t=remap(collection(ty)[v])
+				t.first[2] unless t.nil?
+			end
+
+			def size
+				@store.size
+			end
+
+			def by(type)
+				Hash[collection(type).map{|k,v| [k,remap(v)]}]
+			end
+
+			def links
+				@lnk_store.keys
+			end
+
+			def rels
+				@rel_store.keys
+			end
+
+			def texts
+				@txt_store.keys
+			end
+
+			protected 
+			def collection(type)
+				case type.to_s
+				when 'link' then @lnk_store
+				when 'rel' then @rel_store
+				when 'txt' then @txt_store
+				end
+			end
+
+			def remap(tab)
+				return nil if tab.nil? || tab.size==0
+				tab.map{|i| @store[i]}
+			end
+
+		end
+	end
+end

spec/linkset_spec.rb

@@ -0,0 +1,34 @@
+require File.expand_path(File.dirname(__FILE__) + '/spec_helper')
+
+describe OPDS::Support::LinkSet do 
+	before(:each) do 
+		subject.push('root','http://feedbooks.com','Racine')
+		subject.push('subsection','http://feedbooks.com/publicdomain','Domaine pub')
+		subject.push('subsection','http://feedbooks.com/original','Original')
+		subject.push('subsection','http://feedbooks.com/feed','feeds')
+		subject.push('http://opds-spec.org/shelf','http://feedbooks.com/shelf','shelf')
+		subject.push('related','http://feedbooks.com/shelf',nil)
+	end
+
+	it do
+		subject.size.should be(6)
+	end
+
+	it do
+		subject.map(&:first).size.should be(6)
+	end
+
+	it "should find 3 subsection" do
+		subject['subsection'].size.should be(3)
+		subject.by(:rel)['subsection'].size.should be(3)
+	end
+
+	it "should give root url" do
+		subject.link_url(:rel => 'root').should == ('http://feedbooks.com')
+		subject.by(:rel)['root'].first[1].should == ('http://feedbooks.com')
+	end
+
+	it "get all text values" do
+		subject.texts.size.should be(6)
+	end
+end

spec/opdsparser_spec.rb

@@ -11,9 +11,58 @@ def sample(type)
 			lambda { subject.parse(sample(feed_type)) }.should_not raise_error
 		end
 
-		it "should should sniff  entry" do
+		it "should sniff entry" do
 			subject.parse(sample(feed_type))
 			subject.sniffed_type.should be(feed_type)
 		end
+
+		it "should return an instance of the correct class " do
+			subject.parse(sample(feed_type)).class.should be({:entry => OPDS::Entry, :navigation => OPDS::NavigationFeed,:acquisition => OPDS::AcquisitionFeed}[feed_type])
+		end
+	end
+
+	[ :acquisition, :navigation].each do |feed_type|
+		it "should have a feed title" do 
+			subject.parse(sample(feed_type)).title.size.should_not be(0)
+		end
+
+		it "should have a feed icon" do 
+			subject.parse(sample(feed_type)).icon.size.should_not be(0)
+		end
+
+		it "should have feed links" do 
+			subject.parse(sample(feed_type)).links.size.should_not be(0)
+		end
+
+		it "should have feed id" do 
+			subject.parse(sample(feed_type)).id.size.should_not be(0)
+		end
+		it "should have a feed author" do 
+			auth=subject.parse(sample(feed_type)).author
+			auth[:name].should == ('Feedbooks')
+			auth[:uri].should == ('http://www.feedbooks.com')
+			auth[:email].should == ('support@feedbooks.com')
+		end
+
+		it "should have entries" do 
+			subject.parse(sample(feed_type)).entries.size.should_not be(0)
+		end
+
+
 	end
+
+		it do
+			subject.parse(sample(:acquisition)).should be_paginated()
+		end
+
+		it do
+			subject.parse(sample(:acquisition)).should be_first_page()
+		end
+
+		it do
+			feed=nil
+			lambda { feed=subject.parse(sample(:acquisition)).next_page  }.should_not raise_error
+
+			feed.class.should be(OPDS::AcquisitionFeed)
+		end
 end