anystyle-parser 0.0.5 → 0.0.6

data/HISTORY.md

@@ -1,5 +1,6 @@
-0.0.5 / 2011-09-06
+0.0.6 / 2011-09-06
 ==================
+* Added location normalizer
 * Improved punctuation feature elicitation
 * Improved name tokenizing
 * Bugfixes

data/lib/anystyle/parser/features.rb

@@ -136,6 +136,14 @@ module Anystyle
 				:'internal-braces'
 			when /^[\(\[\{<].*[>\}\]\)]$/
 				:braces
+			when /^[\(\[\{<]/
+				:'opening-braces'
+			when /[>\}\]\)][\.]$/
+				:'terminal-closing-braces'
+			when /[>\}\]\)][,;:-]$/
+				:'internal-closing-braces'
+			when /^[>\}\]\)]$/
+				:'closing-braces'
 			when /[,;:-]$/
 				:internal
 			when /[!\?\."']$/

data/lib/anystyle/parser/normalizer.rb

@@ -311,6 +311,24 @@ module Anystyle
 				warn e.message
 				hash
 			end
+			
+			def normalize_location(hash)
+				location, *dangling = hash[:location]
+				unmatched(:pages, hash, dangling) unless dangling.empty?
+
+				location.gsub!(/^\W+|\W+$/, '')
+
+				if !hash.has_key?(:publisher) && location =~ /:/
+					location, publisher = location.split(/\s*:\s*/)
+					hash[:publisher] = publisher
+				end
+				
+				hash[:location] = location
+				hash
+			rescue => e
+				warn e.message
+				hash
+			end
 
 			private
 			

data/lib/anystyle/parser/parser.rb

@@ -169,7 +169,7 @@ module Anystyle
 					hash[:type] = :book
 				when keys.include?(:institution)
 					hash[:type] = :techreport
-				when keys.include?(:school)
+				when keys.include?(:school) || text =~ /master('s)?\s+thesis/i
 					hash[:type] = :mastersthesis
 				when text =~ /unpublished/i
 					hash[:type] = :unpublished