lib/MESH/tree.rb in mesh-medical-subject-headings-2.2.1 vs lib/MESH/tree.rb in mesh-medical-subject-headings-2.3.0

@@ -9,10 +9,11 @@
       @headings = []
       @by_unique_id = {}
       @by_tree_number = {}
       @by_original_heading = {}
       @by_entry = {}
+      @by_entry_word = Hash.new { |h, k| h[k] = Set.new }
       @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)
@@ -40,21 +41,30 @@
 
     def add_heading_to_hashes(mh)
       @headings << mh
       @by_unique_id[mh.unique_id] = mh
       @by_original_heading[mh.original_heading] = mh
+      add_heading_by_entry_word(mh, mh.original_heading)
       mh.tree_numbers.each do |tree_number|
         raise if @by_tree_number[tree_number]
         @by_tree_number[tree_number] = mh
       end
       match_headings = mh.entries.map { |e| entry_match_key(e) }.uniq
       match_headings.each do |entry|
         raise if @by_entry[entry]
         @by_entry[entry] = mh
+        add_heading_by_entry_word(mh, entry)
       end
     end
 
+    def add_heading_by_entry_word(mh, entry)
+      entry.split.each do |word|
+        word.downcase!
+        @by_entry_word[word] << mh
+      end
+    end
+
     def entry_match_key(e)
       e.strip.upcase
     end
 
     def load_translation(locale)
@@ -78,11 +88,14 @@
               entries.uniq!
               if heading = find(unique_id)
                 heading.set_original_heading(original_heading, locale) unless original_heading.nil?
                 heading.set_natural_language_name(natural_language_name, locale) unless natural_language_name.nil?
                 heading.set_summary(summary, locale) unless summary.nil?
-                entries.each { |entry| heading.entries(locale) << entry }
+                entries.each do |entry|
+                  heading.entries(locale) << entry
+                  add_heading_by_entry_word(heading, entry)
+                end
               end
 
               entries = []
               original_heading = nil
               summary = nil
@@ -185,10 +198,14 @@
 
     def find_by_entry(entry)
       return @by_entry[entry_match_key(entry)]
     end
 
+    def find_by_entry_word(word)
+      return @by_entry_word[word]
+    end
+
     def where(conditions)
       matches = []
       @headings.each do |heading|
         matches << heading if heading.matches(conditions)
       end
@@ -202,33 +219,38 @@
     end
 
     def match_in_text(text)
       return [] if text.nil?
       downcased = text.downcase
+      candidate_headings = Set.new
+      downcased.split(/\W+/).uniq.each do |word|
+        candidate_headings.merge(find_by_entry_word(word))
+      end
       matches = []
-      @headings.each do |heading|
+      candidate_headings.each do |heading|
         next unless heading.useful
         @locales.each do |locale|
           heading.entries(locale).each do |entry|
             if downcased.include? entry.downcase #This is a looser check than the regex but much, much faster
               if /^[A-Z0-9]+$/ =~ entry
                 regex = /(^|\W)#{Regexp.quote(entry)}(\W|$)/
               else
                 regex = /(^|\W)#{Regexp.quote(entry)}(\W|$)/i
               end
               text.to_enum(:scan, regex).map do |m,|
-                matches << {heading: heading, matched: entry, index: $`.size}
+                match = Regexp.last_match
+                matches << {heading: heading, matched: entry, index: match.offset(0)}
               end
             end
           end
         end
       end
       confirmed_matches = []
       matches.combination(2) do |l, r|
-        if (r[:index] >= l[:index]) && (r[:index] + r[:matched].length <= l[:index] + l[:matched].length)
+        if (r[:index][0] >= l[:index][0]) && (r[:index][1] <= l[:index][1])
           #r is within l
           r[:delete] = true
-        elsif (l[:index] >= r[:index]) && (l[:index] + l[:matched].length <= r[:index] + r[:matched].length)
+        elsif (l[:index][0] >= r[:index][0]) && (l[:index][1] <= r[:index][1])
           #l is within r
           l[:delete] = true
         end
       end
       matches.delete_if { |match| match[:delete] }
\ No newline at end of file