Hacked in support for redirecting from Wikidata identifiers

dbpedialite.rb

@@ -2,6 +2,7 @@
 
 require 'wikipedia_thing'
 require 'wikipedia_category'
+require 'wikidata_api'
 require 'formats'
 
 
@@ -79,6 +80,18 @@ def redirect_from_title(title)
     end
   end
 
+  def redirect_from_wikidata(id)
+    begin
+      sitelink = WikidataApi.get_sitelink(id)
+      redirect_from_title sitelink['title']
+    rescue MediaWiki::NotFound => e
+      not_found e.to_s
+    rescue MediaWiki::Exception => e
+      error 500, "Wikidata API excpetion: #{e}"
+    end
+    redirect_from_title(title)
+  end
+
   helpers do
     include Sinatra::ContentFor
     include Sinatra::UrlForHelper
@@ -176,6 +189,10 @@ def format_iso8061(datetime)
     redirect_from_title(title)
   end
 
+  get %r{^/wikidata/Q(\d+)$} do |id|
+    redirect_from_wikidata(id)
+  end
+
   get %r{^/things/(\d+)\.?([a-z0-9]*)$} do |pageid,format|
     begin
       @thing = WikipediaThing.load(pageid)
@@ -222,6 +239,8 @@ def format_iso8061(datetime)
       redirect_from_title($2)
     elsif params[:url] =~ %r{^http://dbpedia.org/(page|resource|data)/(.+)$}
       redirect_from_title($2)
+    elsif params[:url] =~ %r{^http://(www\.)?wikidata.org/wiki/Q(\d+)$}
+      redirect_from_wikidata($2)
     elsif params[:url] =~ %r{^http://www.freebase.com/(view|inspect|edit/topic)(/.+)$}
       begin
         data = FreebaseApi.lookup_by_id($2)

lib/mediawiki_api.rb

@@ -0,0 +1,111 @@
+require 'net/http'
+require 'uri'
+
+class MediaWikiApi
+
+  class Exception < Exception
+  end
+
+  class NotFound < MediaWikiApi::Exception
+  end
+
+  USER_AGENT = 'DbpediaLite/1'
+  HTTP_TIMEOUT = 5
+  NBSP = Nokogiri::HTML("&nbsp;").text
+  UNSAFE_REGEXP = Regexp.new('[^-_\.!~*\'()a-zA-Z0-9;/:@&=$,]', false, 'N').freeze
+  DBPEDIA_UNSAFE_REGEXP = Regexp.new('[^a-zA-Z0-9\.\-*/:_,&]', false, 'N').freeze
+
+  def self.escape_query(str)
+    URI::escape(str, UNSAFE_REGEXP)
+  end
+
+  def self.escape_title(title)
+    URI::escape(title.gsub(' ','_'), ' ?#%"+=')
+  end
+
+  def self.title_to_dbpedia_key(title)
+    # From http://dbpedia.org/URIencoding
+    URI::escape(title.gsub(' ', '_').squeeze('_'), DBPEDIA_UNSAFE_REGEXP)
+  end
+
+  def self.clean_displaytitle(hash)
+    if hash['displaytitle']
+      hash['displaytitle'] = Nokogiri::HTML(hash['displaytitle']).text
+    end
+  end
+
+  def self.get(action, args={})
+    items = []
+    args.merge!(:action => action, :format => 'json')
+
+    keys = args.keys.sort {|a,b| a.to_s <=> b.to_s}
+    keys.each do |key|
+     items << escape_query(key.to_s)+'='+escape_query(args[key].to_s)
+    end
+
+    uri = self.api_uri
+    uri.query = items.join('&')
+    res = Net::HTTP.start(uri.host, uri.port) do |http|
+      http.read_timeout = HTTP_TIMEOUT
+      http.open_timeout = HTTP_TIMEOUT
+      http.get(uri.request_uri, {'User-Agent' => USER_AGENT})
+    end
+
+    # Throw exception if unsuccessful
+    res.value
+
+    # Parse the response if it is JSON
+    if res.content_type == 'application/json'
+      data = JSON.parse(res.body)
+    else
+      raise MediaWikiApi::Exception.new(
+        "Response from MediaWiki API was not of type application/json."
+      )
+    end
+
+    # Check for errors in the response
+    if data.nil?
+      raise MediaWikiApi::Exception.new('Empty response')
+    elsif data.has_key?('error')
+      if data['error']['code'] == 'nosuchpageid'
+        raise MediaWikiApi::NotFound.new(
+          data['error']['info']
+        )
+      else
+        raise MediaWikiApi::Exception.new(
+          data['error']['info']
+        )
+      end
+    end
+
+    return data
+  end
+
+  def self.page_info(args)
+    data = self.get('query', {
+      :prop => 'info',
+      :inprop => 'displaytitle',
+      :redirects => 1
+    }.merge(args))
+
+    if data['query'].nil? or data['query']['pages'].empty?
+      raise WikipediaApi::Exception.new('Empty response')
+    else
+      info = data['query']['pages'].values.first
+      if info.has_key?('missing')
+        raise WikipediaApi::PageNotFound.new
+      else
+        clean_displaytitle(info)
+        return info
+      end
+    end
+  end
+
+  def self.search(query, args={})
+    data = self.get('query', {:list => 'search', :srprop => 'snippet|titlesnippet', :srsearch => query}.merge(args))
+
+    data['query']['search']
+  end
+
+
+end

lib/wikidata_api.rb

@@ -0,0 +1,28 @@
+require 'mediawiki_api'
+
+class WikidataApi < MediaWikiApi
+
+  def self.api_uri
+    URI.parse('http://wikidata.org/w/api.php')
+  end
+
+  def self.get_sitelink(id, site='enwiki')
+    data = self.get('wbgetitems', {
+      :ids => id,
+      :props => 'sitelinks',
+      :languages => 'en',
+      :sites => site
+    })
+
+    if data['items'].nil?
+      raise MediaWiki::Exception.new('Empty response')
+    elsif data['items'][id].nil?
+      raise MediaWiki::NotFound.new('Wikidata identifier does not exist')
+    elsif data['items'][id]['sitelinks'][site].nil?
+      raise MediaWiki::NotFound.new('Sitelink does not exist for Wikidata identifier')
+    else
+      return data['items'][id]['sitelinks'][site]
+    end
+  end
+
+end