simple_bioc 0.0.11 → 0.0.12

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 3a4b12838e8618ee607ff56a7170cda15e0f242c
-  data.tar.gz: 275bff96a672d49519d81b47fb97f00d1eda0ad4
+  metadata.gz: 85e59da540b4e77a5802b72315e9a9a8e7c60a71
+  data.tar.gz: 5d0db6bfd1c2fd8fd3ea713de9f60aeff0c7b365
 SHA512:
-  metadata.gz: c6a49eada64585f46811c22d6e76f3791ab24b4f8ef961aac2c7ff6bf6e19d764c54f35f519d6fad2eddbeff324bc9819fdecb318579cc55fc022b7ee73b702b
-  data.tar.gz: a835ff1a99236baf0ebcfca85069c1821b137cb9412feb873a277579cb55d1d052699e0782ccb9fddfdc7ed1c4caedc3b8be44d26d4066262c1d5492cc72e886
+  metadata.gz: db1b62532952cfc88237d359dc70d1c5aa0abbc6152455b2245f0229b4e8138985fc716adba6f9648e358de50342e5de984f48f52ea59a59560b3ebc46efd75a
+  data.tar.gz: 398312f95e42cd2f36371509a4a85a1c35bd670f60cb30b54556aa0fbb75810764578bb1ff726d5a140966480d6e64510ee28f93a047b146d48543362e4b3395

data/lib/simple_bioc/bioc_merger.rb

@@ -49,7 +49,7 @@ module BioCMerger
           copy_text(s_d, s_s)
           copy_relations(doc_d, s_d, s_s, id_map)
           copy_annotations(doc_d, s_d, s_s, id_map)
-          adjust_annotation_offset(s_d)
+          s_d.adjust_annotation_offsets
         end
       elsif p_d.sentences.size == 0
         p_d.text = p_s.sentences.map{|s| s.text}.join(" ") if blank?(p_d.text)
@@ -77,21 +77,11 @@ module BioCMerger
         end
       end
       copy_annotations(doc_d, p_d, p_s, id_map)
-      adjust_annotation_offset(p_d)
+      p_d.adjust_annotation_offsets
     end
     puts warnings
   end
 
-  def adjust_annotation_offset(obj)
-    return if obj.nil?
-    obj.annotations.each do |a|
-      positions = find_all_locations(obj, a.text)
-      a.locations.each do |l|
-        l.offset = choose_offset_candidate(l.offset, positions)
-      end
-    end
-  end
-
   def adjust_relation_refids(doc, id_map)
     adjust_relation_refid(doc, id_map)
     doc.passages.each do |p|
@@ -236,29 +226,4 @@ module BioCMerger
       end
     end
   end
-
-  def find_all_locations(obj, text)
-    positions = []
-    return positions if obj.nil? || obj.text.nil?
-    pos = obj.text.index(text)
-    until pos.nil? 
-      positions << (pos + obj.offset)
-      pos = obj.text.index(text, pos + 1)
-    end
-    return positions
-  end
-
-  def choose_offset_candidate(offset, positions)
-    min_diff = 99999
-    ret = offset
-    offset = offset.to_i
-    positions.each do |p|
-      diff = (offset - p).abs
-      if diff < min_diff
-        offset = p 
-        min_diff = diff
-      end
-    end
-    return ret
-  end
 end

data/lib/simple_bioc/bioc_reader.rb

@@ -79,6 +79,7 @@ module BioCReader
     read_recursive(xml, passage, "sentence")
     read_recursive(xml, passage, "annotation")
     read_recursive(xml, passage, "relation")
+    passage.adjust_annotation_offsets
     true
   end
 
@@ -88,6 +89,7 @@ module BioCReader
     read_infon(xml, sentence)
     read_recursive(xml, sentence, "annotation")
     read_recursive(xml, sentence, "relation")
+    sentence.adjust_annotation_offsets
     true
   end
 

data/lib/simple_bioc/location_adjuster.rb

@@ -0,0 +1,39 @@
+module LocationAdjuster
+  def adjust_annotation_offsets
+    obj = self
+    return if obj.nil?
+    obj.annotations.each do |a|
+      positions = find_all_locations(obj, a.text)
+      a.locations.each do |l|
+        l.offset = choose_offset_candidate(l.offset, positions)
+      end
+    end
+  end
+  
+  module_function
+
+  def find_all_locations(obj, text)
+    positions = []
+    return positions if obj.nil? || obj.text.nil?
+    pos = obj.text.index(text)
+    until pos.nil? 
+      positions << (pos + obj.offset)
+      pos = obj.text.index(text, pos + 1)
+    end
+    return positions
+  end
+
+  def choose_offset_candidate(offset, positions)
+    min_diff = 99999
+    offset = offset.to_i
+    ret = offset
+    positions.each do |p|
+      diff = (offset - p).abs
+      if diff < min_diff
+        ret = p 
+        min_diff = diff
+      end
+    end
+    return ret
+  end
+end

data/lib/simple_bioc/passage.rb

@@ -1,5 +1,7 @@
 module SimpleBioC
   class Passage
+    include LocationAdjuster
+
     attr_accessor :offset, :text, :infons, :sentences, :annotations, :relations
     attr_reader :document
 

data/lib/simple_bioc/sentence.rb

@@ -1,5 +1,7 @@
 module SimpleBioC
   class Sentence
+    include LocationAdjuster
+
     attr_accessor :offset, :text, :infons, :annotations, :relations
     attr_reader :passage
 

data/lib/simple_bioc/version.rb

@@ -1,3 +1,3 @@
 module SimpleBioC
-  VERSION = "0.0.11"
+  VERSION = "0.0.12"
 end

data/spec/file_check_spec.rb

@@ -32,11 +32,19 @@ describe "File Check" do
 
     SimpleBioC.merge(col1, col2)
     output = SimpleBioC.to_xml(col1)
-    File.write("./xml/merge/output.xml", output)
+    File.write("./xml/merge/output_10330397.xml", output)
     puts "merge2"
-    col5 = SimpleBioC.from_xml("./xml/merge/output.xml")
+    col5 = SimpleBioC.from_xml("./xml/merge/output_10330397.xml")
+  end 
+
+  it "should fix location problem" do
+    col1 = SimpleBioC.from_xml("./xml/merge/10366597_error.xml")
+    output = SimpleBioC.to_xml(col1)
+    File.write("./xml/merge/output_10366597.xml", output)
+    col5 = SimpleBioC.from_xml("./xml/merge/output_10366597.xml")
   end 
 
+
   it "should merge documents successfully with different order" do
     col4 = SimpleBioC.from_xml("./xml/merge/9864355.xml")
     col3 = SimpleBioC.from_xml("./xml/merge/9864355_1.xml")