lib/MESH/tree.rb in mesh-medical-subject-headings-2.0.3 vs lib/MESH/tree.rb in mesh-medical-subject-headings-2.0.4

@@ -9,10 +9,11 @@
 
       @headings = []
       @by_unique_id = {}
       @by_tree_number = {}
       @by_original_heading = {}
+      @by_entry = {}
       @locales = [@@default_locale]
 
       filename = File.expand_path('../../../data/mesh_data_2014/d2014.bin.gz', __FILE__)
       gzipped_file = File.open(filename)
       file = Zlib::GzipReader.new(gzipped_file)
@@ -28,12 +29,18 @@
               current_heading.entries.sort!
               @headings << current_heading
               @by_unique_id[current_heading.unique_id] = current_heading
               @by_original_heading[current_heading.original_heading] = current_heading
               current_heading.tree_numbers.each do |tree_number|
+                raise if @by_tree_number[tree_number]
                 @by_tree_number[tree_number] = current_heading
               end
+              match_headings = current_heading.entries.map { |e| entry_match_key(e) }.uniq
+              match_headings.each do |entry|
+                raise "#{@by_entry[entry]} vs #{current_heading} on #{entry}\n\n#{@by_entry[entry].entries}\n\n#{current_heading.entries}" if @by_entry[entry]
+                @by_entry[entry] = current_heading
+              end
             end
             current_heading = MESH::Heading.new(self)
             current_heading.default_locale = @@default_locale
 
           when matches = line.match(/^UI = (.*)/)
@@ -50,18 +57,18 @@
             current_heading.descriptor_class = @@descriptor_classes[matches[1].to_i]
 
           when matches = line.match(/^MH = (.*)/)
             mh = matches[1]
             current_heading.set_original_heading(mh)
-            current_heading.entries << mh
+            current_heading.entries << mh unless current_heading.entries.include? mh
             librarian_parts = mh.match(/(.*), (.*)/)
             nln = librarian_parts.nil? ? mh : "#{librarian_parts[2]} #{librarian_parts[1]}"
             current_heading.set_natural_language_name(nln)
 
           when matches = line.match(/^(?:PRINT )?ENTRY = ([^|]+)/)
             entry = matches[1].chomp
-            current_heading.entries << entry
+            current_heading.entries << entry unless current_heading.entries.include? entry
 
         end
 
       end
 
@@ -79,10 +86,14 @@
         end
       end
 
     end
 
+    def entry_match_key(e)
+      e.strip.upcase
+    end
+
     def load_translation(locale)
       return if @locales.include? locale
       filename = File.expand_path("../../../data/mesh_data_2014/d2014.#{locale}.bin.gz", __FILE__)
       gzipped_file = File.open(filename)
       file = Zlib::GzipReader.new(gzipped_file)
@@ -134,10 +145,16 @@
 
       end
       @locales << locale
     end
 
+    def linkify_summaries &block
+      @headings.each do |h|
+        h.linkify_summary &block
+      end
+    end
+
     # NO LONGER COVERED BY TESTS
     # def translate(locale, tr)
     #   return if @locales.include? locale
     #   @headings.each_with_index do |h, i|
     #     h.set_original_heading(tr.translate(h.original_heading), locale)
@@ -158,9 +175,13 @@
       return @by_tree_number[tree_number]
     end
 
     def find_by_original_heading(heading)
       return @by_original_heading[heading]
+    end
+
+    def find_by_entry(entry)
+      return @by_entry[entry_match_key(entry)]
     end
 
     def where(conditions)
       matches = []
       @headings.each do |heading|
\ No newline at end of file