mesh-medical-subject-headings 2.2.0 → 2.2.1

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 9e0d8d67721e16e0e016d35056b1ce147d318bd3
-  data.tar.gz: 31ec3da2e71025e3bff77a82fe5178531aff3eeb
+  metadata.gz: cd1eb2f1f911abc7f2421de83fb5e7d6c83dc3cb
+  data.tar.gz: 70b0081953e3cdfb68335d8cd33d4841b6424205
 SHA512:
-  metadata.gz: 3ea3dd0a355ca538a471afb7846aa5ab35ab206de3440719217722c6f70ec97488fc15a30610e929be39dad7e9eeb8c560b06e277567bdacc6f2b8715c68eadd
-  data.tar.gz: db140b61b574f7baa5bd26bba5e5b7d9570851f0992e208da8ce61549233bea5c98016c10ad66ae90ac7b45604a9485df4f2859be87bef56e4fea18c97f2a9e2
+  metadata.gz: 8b4d63674735db2a1675056d65446fadec07be1b3f81d0d157008c5073f8c1733b165a2c785f220e9578c8e3e53173657c621abe4ba97da9821fcbc79ede1280
+  data.tar.gz: cd889c0a8a95d462d4df42b26980ce3506d9e2e52a4ad973aac0f45dd0e1a63af3c5a241e812cdf134e63d577074a69766991f937d005e812863f8890cba3709

data/CHANGELOG.md

@@ -1,3 +1,6 @@
+#2.2.1 / 2014-07-18
+* [FEATURE] Headings now have forward references
+
 #2.2.0 / 2014-07-10
 * [FEATURE] Headings now have structured_entries which include semantic and lexical information
 

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    mesh-medical-subject-headings (2.2.0)
+    mesh-medical-subject-headings (2.2.1)
 
 GEM
   remote: https://rubygems.org/

data/lib/MESH/heading.rb

@@ -1,8 +1,10 @@
 module MESH
   class Heading
 
+    @@descriptor_classes = [:make_array_start_at_1, :topical_descriptor, :publication_type, :check_tag, :geographic_descriptor]
+
     include Comparable
-    attr_accessor :unique_id, :tree_numbers, :roots, :parents, :children, :useful, :descriptor_class, :default_locale, :semantic_types, :wikipedia_links, :structured_entries
+    attr_accessor :unique_id, :tree_numbers, :roots, :parents, :children, :useful, :descriptor_class, :default_locale, :semantic_types, :wikipedia_links, :structured_entries, :forward_references
     attr_reader :linkified_summary
 
     def <=> other
@@ -100,25 +102,94 @@ module MESH
       @summary[locale] = summary
     end
 
+    def connect_to_parents
+      if !@connected_to_parents
+        @tree_numbers.each do |tree_number|
+          #D03.438.221.173
+          parts = tree_number.split('.')
+          if parts.size > 1
+            parts.pop
+            parent_tree_number = parts.join '.'
+            parent = @tree.find_by_tree_number(parent_tree_number)
+            @parents << parent unless parent.nil? || @parents.include?(parent)
+            parent.children << self unless parent.nil? || parent.children.include?(self)
+          end
+        end
+        @connected_to_parents = true
+      end
+    end
+
+    def connect_to_forward_references
+      if !@connected_to_forward_references
+        @forward_references = @forward_reference_terms.map do |term|
+          @tree.find_by_original_heading(term)
+        end
+        @connected_to_forward_references = true
+      end
+    end
+
     private
 
-    def initialize(tree)
+    def initialize(tree, default_locale, lines)
       @tree = tree
+      @default_locale = default_locale
       @useful = true
       @tree_numbers = []
       @semantic_types = []
       @roots = []
       @parents = []
       @children = []
-      @entries = {}
+      @forward_references = []
+      @forward_reference_terms = []
+      @entries = {@default_locale => []}
       @structured_entries = []
       @original_heading = {}
       @natural_language_name = {}
       @summary = {}
       @wikipedia_links = []
-    end
 
+      lines.each do |line|
+        case
+
+          when matches = line.match(/^UI = (.*)/)
+            @unique_id = matches[1]
+
+          when matches = line.match(/^MN = (.*)/)
+            @tree_numbers << matches[1]
+            @roots << matches[1][0] unless @roots.include?(matches[1][0])
 
+          when matches = line.match(/^MS = (.*)/)
+            set_summary(matches[1])
+
+          when matches = line.match(/^DC = (.*)/)
+            @descriptor_class = @@descriptor_classes[matches[1].to_i]
+
+          when matches = line.match(/^ST = (.*)/)
+            @semantic_types << MESH::SemanticTypes[matches[1]]
+
+          when matches = line.match(/^MH = (.*)/)
+            mh = matches[1]
+            set_original_heading(mh)
+            @entries[@default_locale] << mh unless @entries.include? mh
+            librarian_parts = mh.match(/(.*), (.*)/)
+            nln = librarian_parts.nil? ? mh : "#{librarian_parts[2]} #{librarian_parts[1]}"
+            set_natural_language_name(nln)
+
+          when matches = line.match(/^(?:PRINT )?ENTRY = (.*)/)
+            entry = matches[1]
+            term = entry.match(/([^|]+)/)
+            @entries[@default_locale] << term[1] unless @entries.include? term[1]
+            @structured_entries << MESH::Entry.new(self, entry)
+
+          when matches = line.match(/^FX = (.*)/)
+            @forward_reference_terms << matches[1]
+
+        end
+
+      end
+      @entries[@default_locale].sort!
+
+    end
   end
 end
 

data/lib/MESH/tree.rb

@@ -2,7 +2,6 @@ module MESH
 
   class Tree
 
-    @@descriptor_classes = [:make_array_start_at_1, :topical_descriptor, :publication_type, :check_tag, :geographic_descriptor]
     @@default_locale = :en_us
 
     def initialize
@@ -18,85 +17,42 @@ module MESH
       gzipped_file = File.open(filename)
       file = Zlib::GzipReader.new(gzipped_file)
 
-      current_heading = MESH::Heading.new(self)
-      current_heading.default_locale = @@default_locale
+      lines = []
       file.each_line do |line|
-
         case
-
           when line.match(/^\*NEWRECORD$/)
-            unless current_heading.unique_id.nil?
-              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
+            unless lines.empty?
+              mh = MESH::Heading.new(self, @@default_locale, lines)
+              add_heading_to_hashes(mh)
+              lines = [line]
             end
-            current_heading = MESH::Heading.new(self)
-            current_heading.default_locale = @@default_locale
-
-          when matches = line.match(/^UI = (.*)/)
-            current_heading.unique_id = matches[1]
-
-          when matches = line.match(/^MN = (.*)/)
-            current_heading.tree_numbers << matches[1]
-            current_heading.roots << matches[1][0] unless current_heading.roots.include?(matches[1][0])
-
-          when matches = line.match(/^MS = (.*)/)
-            current_heading.set_summary(matches[1])
-
-          when matches = line.match(/^DC = (.*)/)
-            current_heading.descriptor_class = @@descriptor_classes[matches[1].to_i]
-
-          when matches = line.match(/^ST = (.*)/)
-            current_heading.semantic_types << MESH::SemanticTypes[matches[1]]
-
-          when matches = line.match(/^MH = (.*)/)
-            mh = matches[1]
-            current_heading.set_original_heading(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 unless current_heading.entries.include? entry
-          #
-          when matches = line.match(/^(?:PRINT )?ENTRY = (.*)/)
-            entry = matches[1]
-            term = entry.match(/([^|]+)/)
-            current_heading.entries << term[1] unless current_heading.entries.include? term[1]
-            current_heading.structured_entries << MESH::Entry.new(current_heading, entry)
-
+          else
+            lines << line
         end
-
       end
 
-      @by_unique_id.each do |id, heading|
-        heading.tree_numbers.each do |tree_number|
-          #D03.438.221.173
-          parts = tree_number.split('.')
-          if parts.size > 1
-            parts.pop
-            parent_tree_number = parts.join '.'
-            parent = @by_tree_number[parent_tree_number]
-            heading.parents << parent unless parent.nil? || heading.parents.include?(parent)
-            parent.children << heading unless parent.nil? || parent.children.include?(heading)
-          end
-        end
+      @headings.each do |heading|
+        heading.connect_to_parents
+        heading.connect_to_forward_references
       end
 
     end
 
+    def add_heading_to_hashes(mh)
+      @headings << mh
+      @by_unique_id[mh.unique_id] = mh
+      @by_original_heading[mh.original_heading] = mh
+      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
+      end
+    end
+
     def entry_match_key(e)
       e.strip.upcase
     end

data/lib/MESH/version.rb

@@ -1,3 +1,3 @@
 module Mesh
-  VERSION = "2.2.0"
+  VERSION = '2.2.1'
 end

data/test/heading_test.rb

@@ -336,6 +336,24 @@ module MESH
       end
     end
 
+    def test_has_one_forward_reference
+      mh = @mesh_tree.find_by_original_heading('Abdominal Muscles')
+      fx = @mesh_tree.find_by_original_heading('Abdominal Wall')
+      assert_equal [fx], mh.forward_references
+    end
+
+    def test_has_several_forward_references
+      mh = @mesh_tree.find_by_original_heading('Acquired Immunodeficiency Syndrome')
+      fx1 = @mesh_tree.find_by_original_heading('AIDS Arteritis, Central Nervous System')
+      fx2 = @mesh_tree.find_by_original_heading('AIDS Dementia Complex')
+      fx3 = @mesh_tree.find_by_original_heading('AIDS Serodiagnosis')
+      fx4 = @mesh_tree.find_by_original_heading('HIV Seropositivity')
+      fx5 = @mesh_tree.find_by_original_heading('HIV Seroprevalence')
+      fx6 = @mesh_tree.find_by_original_heading('Lymphoma, AIDS-Related')
+
+      assert_equal [fx1, fx2, fx3, fx4, fx5, fx6], mh.forward_references
+    end
+
     def test_have_the_correct_siblings
       skip
     end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: mesh-medical-subject-headings
 version: !ruby/object:Gem::Version
-  version: 2.2.0
+  version: 2.2.1
 platform: ruby
 authors:
 - Rob Styles
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2014-07-11 00:00:00.000000000 Z
+date: 2014-07-18 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler