umlaut 3.1.0 → 3.1.1

data/app/controllers/resolve_controller.rb

@@ -5,6 +5,12 @@
 class ResolveController < UmlautController
   before_filter :init_processing
 
+  # We need to NOT require a CSRF token on post to #index,
+  # to allow POSTed OpenURLs
+  protect_from_forgery :except => :index
+  # POST'ed OpenURLs are a mess, redirect them to GETs
+  before_filter :post_to_get, :only => :index
+
   # Init processing will look at this list, and for actions mentioned,
   # will not create a @user_request if an existing one can't be found.
   # Used for actions meant only to deal with existing requests.
@@ -129,14 +135,22 @@ class ResolveController < UmlautController
 
   protected
 
+  def post_to_get
+    if request.method == "POST"
+      redirect_to url_for(params)
+    end
+  end
+
   # Retrives or sets up the relevant Umlaut Request, and returns it.
-  def init_processing
+  def init_processing    
     # intentionally trigger creation of session if it didn't already exist
     # because we need to track session ID for caching. Can't find any
     # way to force session creation without setting a value in session,
     # so we do this weird one.
     session[nil] = nil
 
+
+
     # We have to clean the params of bad char encoding bytes, or it causes
     # no end of problems later. We can't just refuse to process, sources
     # do send us bad bytes, I'm afraid.

data/app/controllers/search_methods/sfx4.rb

@@ -40,7 +40,7 @@ module SearchMethods
     end
 
     # Needs to return ContextObjects
-    def find_by_title                  
+    def find_by_title
       connection = sfx4_db_connection
       query_match_clause = case search_type_param
         when "contains"
@@ -51,8 +51,12 @@ module SearchMethods
           end
           # Then make each term required, but stemmed. Seems to match SFX4, 
           # and more importantly give us decent results. 
+          #
+          # For reasons we can't entirely tell, the wildcard "*" on terms of less
+          # than 2 causes false negatives. Otherwise we use it to be consistent
+          # with SFX. This reverse-engineering is full of pitfalls.
           query = terms.collect do |term|
-            "+" + connection.quote_string(term) + "*"
+            "+" + connection.quote_string(term) + (term.length > 2 ? "*" : "")
           end.join(" ")
           "MATCH (TS.TITLE_SEARCH) AGAINST ('#{query}' IN BOOLEAN MODE)"
         when "begins"

data/app/controllers/search_methods/sfx4_solr/searcher.rb

@@ -28,24 +28,52 @@ module SearchMethods
       end
 
       def _search_by_title(query, search_type, page=1)
+        # 20 per page
+        # TODO: Make this configurable
+        per_page = 20
+        # Upcase the query since SFX normalizes titles
+        # by doing the same.
+        query = query.upcase
+        # Remove "the", "a", and "an" from the query
+        # so that a search for "the economist" and a 
+        # search for "economist" return the same results.
+        # Seems to be remotely similar to how SFX works
+        # though I'm not sure if that's the best example
+        # to follow.
+        [/^THE /, /^AN? /].each do |article|
+          query.gsub! article, ""
+        end
+        # Sunspot DSL per search type
         search = case search_type
           when "contains"
             az_title_klass.search {
               keywords query, :fields => [:title]
+              # Order by relevance, since we can
+              order_by(:score, :desc)
+              # Order by title after relevance
               order_by(:title_sort, :asc)
-              paginate(:page => page, :per_page => 20)
+              # Pagination
+              paginate(:page => page, :per_page => per_page)
             }
           when "begins"
             az_title_klass.search {
               with(:title_exact).starting_with(query)
+              # Order by relevance, since we can
+              order_by(:score, :desc)
+              # Order by title after relevance
               order_by(:title_sort, :asc)
-              paginate(:page => page, :per_page => 20)
+              # Pagination
+              paginate(:page => page, :per_page => per_page)
             }
           else # exact
             az_title_klass.search {
               with(:title_exact, query)
+              # Order by relevance, since we can
+              order_by(:score, :desc)
+              # Order by title after relevance
               order_by(:title_sort, :asc)
-              paginate(:page => page, :per_page => 20)
+              # Pagination
+              paginate(:page => page, :per_page => per_page)
             }
           end
       end

data/app/service_adaptors/bx.rb

@@ -30,11 +30,9 @@ class Bx < Service
     format = "rss"
     bx_url = "#{@base_url}?res_dat=format%3D#{format}%26source%3D#{@source}%26token%3D#{@token}%26maxRecords%3D#{@max_records}%26threshold%3D#{@threshold}%26baseUrl%3D#{@openurl_base}&#{request.to_context_object.kev}"
     response = open(bx_url)
-    Rails.logger.debug("bX URL #{bx_url.inspect}")
     response_xml_str = response.read
-    Rails.logger.debug("bX Response #{response_xml_str.inspect}")
     response_xml = Nokogiri::XML(response_xml_str)
-    response_xml.search("//item") do |item|
+    response_xml.search("//item").each do |item|
       title = item.at("title").inner_text
       author = item.at("author").inner_text
       display_text = (author.nil?)? "#{title}" : "#{author}; #{title}"