mesh-medical-subject-headings 2.0.3 → 2.0.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: b2319bc27d6bc4f2564f73648af6fc13cbcb3d8c
-  data.tar.gz: ad0787c9a3229e15d0b591504c2882bd8756379b
+  metadata.gz: 1779cff991e94c43a9adf2429b4ec0c487c2d0b4
+  data.tar.gz: 4ef7e94c5f20c098648bcb1a1a8874e1c4905b8c
 SHA512:
-  metadata.gz: e0ff231b13d0223d9f8a33972117e97a3ee0a5459ea40587e11b11ab6c69b4233fb713601e6e64c3c14e5c90a09ccd4ae33fc39f87a87f776624a6922fcf09fe
-  data.tar.gz: a86aa7a5f3db453cc37d5528900fcf1f006793b8eb53b44fa54cce0057887dfc816e4b62b801396e0dd98e54f97db69ac3fe89bfcf90ab2dd6ad60667dcedfb7
+  metadata.gz: 0cf3b2a10bec165bf2652cd0d8ece4ab04900c8743c052ed9174e05526ccbb35cce6e68ac66e9703060e726aaa3c3e7aac77e360e80f57b5f3b235648f4efaaa
+  data.tar.gz: d8a2662c49f5235c8a0aa0165fe1188b2ab6a9a347f66b23dfd99f1cee40d7b1b91a2f6e3b48dc002d73e566db7ab9fc9fa7033956dcc78f2e6f316c187e87d5

data/CHANGELOG.md

@@ -1,3 +1,6 @@
+#2.0.4 / 2014-06-25
+* [FEATURE] tree can now linkify_summaries for all headings. Performance optimisations to make this viable.
+
 #2.0.3 / 2014-06-25
 * [FEATURE] You can now link cross-references in heading.summary by passing a block to heading.linkify_summary
 

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    mesh-medical-subject-headings (2.0.3)
+    mesh-medical-subject-headings (2.0.4)
 
 GEM
   remote: https://rubygems.org/

data/lib/MESH/heading.rb

@@ -3,6 +3,7 @@ module MESH
 
     include Comparable
     attr_accessor :unique_id, :tree_numbers, :roots, :parents, :children, :useful, :descriptor_class, :default_locale
+    attr_reader :linkified_summary
 
     def <=> other
       self.unique_id <=> other.unique_id
@@ -21,13 +22,16 @@ module MESH
     end
 
     def linkify_summary
-      linkified_summary = summary.dup
-      linkified_summary.scan(/[A-Z]+[A-Z,\s]+[A-Z]+/).each do |text|
-        heading = @tree.where(entries: /^#{text.strip}$/i).first
-        replacement_text = yield text, heading
-        linkified_summary.sub!(text, replacement_text)
+      return if summary.nil?
+      @linkified_summary = summary.dup
+      @linkified_summary.scan(/[A-Z]+[A-Z,\s]+[A-Z]+/).each do |text|
+        heading = @tree.find_by_entry(text)
+        if heading
+          replacement_text = yield text, heading
+          @linkified_summary.sub!(text, replacement_text)
+        end
       end
-      linkified_summary
+      @linkified_summary
     end
 
     def entries(locale = default_locale)

data/lib/MESH/tree.rb

@@ -11,6 +11,7 @@ module MESH
       @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__)
@@ -30,8 +31,14 @@ module MESH
               @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
@@ -52,14 +59,14 @@ module MESH
           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
 
@@ -81,6 +88,10 @@ module MESH
 
     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__)
@@ -136,6 +147,12 @@ module MESH
       @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
@@ -162,6 +179,10 @@ module MESH
       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|

data/lib/MESH/version.rb

@@ -1,3 +1,3 @@
 module Mesh
-  VERSION = "2.0.3"
+  VERSION = "2.0.4"
 end

data/test/mesh_core_test.rb

@@ -40,6 +40,39 @@ module MESH
       assert_nil mh
     end
 
+    def test_find_by_entry
+
+      expected_entries = [
+        'Adult Reye Syndrome',
+        'Adult Reye\'s Syndrome',
+        'Fatty Liver with Encephalopathy',
+        'Reye Johnson Syndrome',
+        'Reye Like Syndrome',
+        'Reye Syndrome',
+        'Reye Syndrome, Adult',
+        'Reye\'s Like Syndrome',
+        'Reye\'s Syndrome',
+        'Reye\'s Syndrome, Adult',
+        'Reye\'s-Like Syndrome',
+        'Reye-Johnson Syndrome',
+        'Reye-Like Syndrome'
+      ]
+
+      entries_to_test = expected_entries.flat_map do |e|
+        [e, e.upcase, e.downcase, " #{e.downcase} ", "\n\n\t #{e.downcase}\t "]
+      end
+
+      entries_to_test.each do |entry|
+        refute_nil mh = @mesh_tree.find_by_entry(entry), "Failed to find heading by entry '#{entry}'"
+        assert_equal 'D012202', mh.unique_id, "Found wrong heading by entry '#{entry}'"
+      end
+
+    end
+
+    def test_find_by_entry_doesnt_match
+      assert_nil @mesh_tree.find_by_entry('foo')
+    end
+
     def test_have_the_correct_unique_id
       mh = @mesh_tree.find('D000001')
       assert_equal 'D000001', mh.unique_id
@@ -121,10 +154,15 @@ module MESH
         'ADENOCARCINOMA' => false
       }
 
+      # start = Time.now
       linkified_summary = mh.linkify_summary do |text, heading|
         found[text] = true unless heading.nil?
         "<foo>#{text.downcase}</foo>"
       end
+      # finish = Time.now
+      # puts start
+      # puts finish
+      # puts finish - start
 
       assert_equal 5, found.length
       assert found['ESOPHAGUS']
@@ -137,6 +175,15 @@ module MESH
 
     end
 
+    def test_linkifies_all_summaries
+      mesh = MESH::Tree.new
+      mesh.linkify_summaries do |text, heading|
+        "<bar>#{text.downcase}</bar>"
+      end
+      mh = mesh.find('D001471')
+      assert_equal 'A condition with damage to the lining of the lower <bar>esophagus</bar> resulting from chronic acid reflux (<bar>esophagitis, reflux</bar>). Through the process of metaplasia, the squamous cells are replaced by a columnar epithelium with cells resembling those of the <bar>intestine</bar> or the salmon-pink mucosa of the <bar>stomach</bar>. Barrett\'s columnar epithelium is a marker for severe reflux and precursor to <bar>adenocarcinoma</bar> of the esophagus.', mh.linkified_summary
+    end
+
     def test_to_s
       mh = @mesh_tree.find('D001471')
       assert_equal 'D001471, Barrett Esophagus, [C06.198.102,C06.405.117.102]', mh.to_s

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: mesh-medical-subject-headings
 version: !ruby/object:Gem::Version
-  version: 2.0.3
+  version: 2.0.4
 platform: ruby
 authors:
 - Rob Styles