legislation-uk 0.0.2 → 0.0.3

data/CHANGELOG

@@ -1,2 +1,3 @@
+v0.0.3 now returns opsi uri and legislation uri for a section
 v0.0.2 now returns opsi url, statutelaw url and legislation url
 v0.0.1 initial alpha release

data/README

@@ -28,18 +28,38 @@ Can be used from command line if you run irb:
 
  legislation = Legislation::UK.find('Channel Tunnel Rail Link Act 1996')
 
- legislation.title           #=> "Channel Tunnel Rail Link Act 1996"
+ legislation.title
+ #=> "Channel Tunnel Rail Link Act 1996"
 
- legislation.legislation_url #=> "http://www.legislation.gov.uk/ukpga/1996/61"
+ legislation.legislation_uri
+ #=> "http://www.legislation.gov.uk/ukpga/1996/61"
 
- legislation.opsi_url        #=> "http://www.opsi.gov.uk/acts/acts1996/ukpga_19960061_en_1"
+ legislation.opsi_uri
+ #=> "http://www.opsi.gov.uk/acts/acts1996/ukpga_19960061_en_1"
 
- legislation.statutelaw_url  #=> "http://www.statutelaw.gov.uk/documents/1996/61/ukpga/c61"
+ legislation.statutelaw_uri
+ #=> "http://www.statutelaw.gov.uk/documents/1996/61/ukpga/c61"
 
- legislation.parts.size      #=> 3
+ legislation.parts.size
+ #=> 3
 
- legislation.parts.collect(&:number) #=> ["Part I", "Part II", "Part III"]
+ legislation.parts.collect(&:number)
+ #=> ["Part I", "Part II", "Part III"]
 
- legislation.parts.collect(&:title)  #=> ["The Channel Tunnel Rail Link",
-                                     #    "The A2 and M2 Improvement Works",
-                                     #    "Miscellaneous and General"]
+ legislation.parts.collect(&:title)
+ #=> ["The Channel Tunnel Rail Link", "The A2 and M2 Improvement Works",
+ #    "Miscellaneous and General"]
+
+ section = legislation.sections.first
+
+ section.title
+ #=> "Construction and maintenance of scheduled works"
+
+ section.number
+ #=> "1"
+
+ section.legislation_uri
+ #=> "http://www.legislation.gov.uk/ukpga/1996/61/section/1"
+
+ section.opsi_uri
+ #=> "http://www.opsi.gov.uk/acts/acts1996/ukpga_19960061_en_2#pt1-pb1-l1g1"

data/lib/legislation_uk.rb

@@ -4,31 +4,74 @@ require 'open-uri'
 
 module LegislationUK
 
-  module Title
+  module TitleHelper
     def title
       contents_title.is_a?(String) ? contents_title.strip : contents_title.title.strip
     end
   end
 
-  module ItemNumber
+  module ItemNumberHelper
     def number
       contents_number
     end
   end
 
+  module LegislationUriHelper
+    def legislation_uri
+      document_uri
+    end
+  end
+
+  module Helper
+    def return_values many
+      if respond_to?(many)
+        send(many) ? send(many) : []
+      else
+        one = many.to_s.singularize.to_sym
+        if respond_to?(one)
+          send(one) ? [send(one)] : []
+        else
+          []
+        end
+      end
+    end
+  end
+
   class Legislation
     include Morph
+    include LegislationUriHelper
 
     def self.open_uri uri
       open(uri).read
     end
 
-    def legislation_url
-      document_uri
+    def title
+      metadata.title
+    end
+
+    def populate
+      parts.each {|x| x.legislation= self}
+      sections.each {|x| x.legislation= self}
+    end
+
+    def sections
+      if parts
+        parts.collect(&:sections).flatten
+      else
+        []
+      end
+    end
+
+    def parts
+      if respond_to?(:contents) && contents
+        contents.parts
+      else
+        []
+      end
     end
 
-    def statutelaw_url
-      if legislation_url[%r|http://www.legislation.gov.uk/(.+)/(\d\d\d\d)/(\d+)|]
+    def statutelaw_uri
+      if legislation_uri[%r|http://www.legislation.gov.uk/(.+)/(\d\d\d\d)/(\d+)|]
         type = $1
         year = $2
         chapter = $3
@@ -38,16 +81,8 @@ module LegislationUK
       end
     end
 
-    def title
-      metadata.title
-    end
-
-    def parts
-      contents ? contents.contents_parts : []
-    end
-
-    def opsi_url
-      unless @opsi_url
+    def opsi_uri
+      unless @opsi_uri
         search_url = "http://search.opsi.gov.uk/search?q=#{URI.escape(title)}&output=xml_no_dtd&client=opsisearch_semaphore&site=opsi_collection"
         begin
           doc = Hpricot.XML Legislation.open_uri(search_url)
@@ -60,39 +95,101 @@ module LegislationUK
             end
           end
 
-          @opsi_url = url
+          @opsi_uri = url
         rescue Exception => e
           puts 'error retrieving: ' + search_url
           puts e.class.name
           puts e.to_s
         end
       end
-      @opsi_url
+      @opsi_uri
+    end
+
+    def opsi_uri_for_section section_number
+      if opsi_uri && !@opsi_sections
+        section_number = section_number.to_s
+        doc = Hpricot Legislation.open_uri(opsi_uri)
+
+        (doc/'span[@class="LegDS LegContentsNo"]').each do |span|
+          number_of_section = span.inner_text.chomp('.')
+          if span.at('a')
+            path = span.at('a')['href']
+            base = opsi_uri[/^(.+\/)[^\/]+$/,1]
+            section_title = span.next_sibling.inner_text
+
+            @opsi_sections ||= {}
+            @opsi_sections[number_of_section] = { :title => section_title, :opsi_uri => "#{base}#{path}"}
+          else
+            puts "cannot find opsi url for section #{number_of_section} of #{name}"
+          end
+        end
+        @opsi_sections[section_number][:opsi_uri]
+      elsif @opsi_sections
+        if @opsi_sections[section_number]
+          @opsi_sections[section_number][:opsi_uri]
+        else
+          puts "no opsi url for #{section_number}\n" + @opsi_sections.inspect + "\n\nno opsi url for #{section_number}\n"
+          nil
+        end
+      else
+        nil
+      end
+    end
+  end
+
+  class Contents
+    include Morph
+    include Helper
+
+    def parts
+      return_values :contents_parts
     end
   end
 
   class ContentsPart
     include Morph
-    include Title
-    include ItemNumber
+    include Helper
+    include TitleHelper
+    include ItemNumberHelper
+    include LegislationUriHelper
+
+    attr_accessor :legislation
+
     def blocks
-      contents_pblocks ? contents_pblocks : []
+      return_values :contents_pblocks
+    end
+
+    def sections
+      if blocks.empty?
+        return_values :contents_items
+      else
+        blocks.collect(&:sections).flatten
+      end
     end
   end
 
   class ContentsPblock
     include Morph
-    include Title
+    include Helper
+    include TitleHelper
+    include LegislationUriHelper
 
     def sections
-      contents_items
+      return_values :contents_items
     end
   end
 
   class ContentsItem
     include Morph
-    include Title
-    include ItemNumber
+    include TitleHelper
+    include ItemNumberHelper
+    include LegislationUriHelper
+
+    attr_accessor :legislation
+
+    def opsi_uri
+      @legislation.opsi_uri_for_section(number)
+    end
   end
 end
 
@@ -100,7 +197,7 @@ end
 # See README for usage documentation.
 module Legislation
   module UK
-    VERSION = "0.0.2"
+    VERSION = "0.0.3"
 
     def self.open_uri uri
       open(uri).read
@@ -111,7 +208,9 @@ module Legislation
       xml.gsub!('dc:type','dc:the_type')
       hash = Hash.from_xml(xml)
       namespace = LegislationUK
-      Morph.from_hash(hash, namespace)
+      legislation = Morph.from_hash(hash, namespace)
+      legislation.populate
+      legislation
     end
 
     def self.find title, number=nil

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: legislation-uk
 version: !ruby/object:Gem::Version 
-  version: 0.0.2
+  version: 0.0.3
 platform: ruby
 authors: 
 - Rob McKinnon
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2009-09-29 00:00:00 +01:00
+date: 2009-10-02 00:00:00 +01:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency