open_nlp 0.0.4-java → 0.0.5-java

data/lib/open_nlp/chunker.rb

@@ -30,8 +30,15 @@ module OpenNlp
 
       data.inject([]) do |acc, val|
         chunk = val[2]
-        acc << [{val[0] => val[1]}] if chunk[0] == 'B'
-        acc.last << {val[0] => val[1]} if chunk[0] == 'I'
+        acc << [{val[0] => val[1]}] if chunk[0] == 'B' # add token to chunk if it is a start of chunk
+
+        if chunk[0] == 'I'
+          if acc.last
+            acc.last << {val[0] => val[1]} # add token to chunk if it is a continuation of chunk
+          else
+            acc << [{val[0] => val[1]}] # add token to new chunk if no chunks exists
+          end
+        end
 
         acc
       end

data/lib/open_nlp/parser.rb

@@ -35,7 +35,6 @@ module OpenNlp
                          )
       end
 
-      #code_tree @j_instance.parse(parse_obj)
       Parser::Parse.new(@j_instance.parse(parse_obj))
     end
 
@@ -49,53 +48,5 @@ module OpenNlp
 
       offset
     end
-
-    #def build_tree(parse_obj)
-    #  span = parse_obj.getSpan
-    #  start = span.getStart
-    #  text = parse_obj.getText
-    #  type = parse_obj.getType
-    #
-    #  res = {}
-    #  res[:type] = type unless type == Java::opennlp.tools.parser.AbstractBottomUpParser::TOK_NODE
-    #
-    #  children = parse_obj.getChildren.inject([]) do |acc,c|
-    #    s = c.span
-    #
-    #    h = {}
-    #
-    #    if start < s.getStart
-    #      token = text[start..s.getStart-1]
-    #      h[:token] = token unless token.strip.empty?
-    #    end
-    #
-    #    subtree = build_tree(c)
-    #    h[:children] = subtree unless subtree.empty?
-    #
-    #    start = s.getEnd
-    #
-    #    acc << h
-    #    acc
-    #  end
-    #
-    #  res[:token] = text[start..span.getEnd-1] if start < span.getEnd
-    #
-    #  res[:children] = children unless children.empty?
-    #
-    #  res
-    #end
-
-    def code_tree(parse_obj)
-      kids = parse_obj.getChildren
-
-      kids.inject([]) do |acc,kid|
-        data = {type: kid.getType, parent_type: parse_obj.getType, token: kid.toString}
-        subtree = code_tree(kid)
-        data[:children] = subtree unless subtree.empty?
-        acc << data
-
-        acc
-      end
-    end
   end
 end

data/lib/open_nlp/version.rb

@@ -1,3 +1,3 @@
 module OpenNlp
-  VERSION = '0.0.4'
+  VERSION = '0.0.5'
 end

metadata

@@ -2,14 +2,14 @@
 name: open_nlp
 version: !ruby/object:Gem::Version
   prerelease:
-  version: 0.0.4
+  version: 0.0.5
 platform: java
 authors:
 - Hck
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2012-09-26 00:00:00.000000000 Z
+date: 2012-09-28 00:00:00.000000000 Z
 dependencies: []
 description: JRuby tools wrapper for Apache OpenNLP
 email: