anystyle-parser 0.5.2 → 0.5.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: c5471f4465579c6aec53ffee37f8e337075934fd
-  data.tar.gz: e32a68bbbe093abd3c44df90df5524325afc9f7e
+  metadata.gz: 71ded6c875b62e8abc138caa27b2fa9c4d423ca0
+  data.tar.gz: 011cb817c6dfde3ed323b34fc7196bc3e43fa47c
 SHA512:
-  metadata.gz: cf520db7b3e31bfe1bd02b03b3d8a951a6288e6d10055ceeccf0b515d3f45e1a83abce911cb698839169c75cf9cac8d2e4169ea54d5db900b25b134b7caffd39
-  data.tar.gz: ebc4173949674553f396dbba6631fae165493f47f748288edad5ccf4fdee33b4e0a2fe7b6a54c5fc0c95e79f8bbbf991a03f62dff202fb6ec949a1098ec1ab24
+  metadata.gz: 8e7787ab5548ef848e7045c8c340630ccda3158884471bf40c3a408543f7367eeb0f6b8cecb005a57e08b578100c87ccf96a583a130c82a8c9b1e308e9894f3c
+  data.tar.gz: 368d50c41245e8456e5937a3b2cd81e9b6f2b2b2ac655d026e9c4716c5c8c031a363b73d49674b6d7d3ad6733cf945db84cdb8a4b37e64c2d5f681c64d224ab4

data/HISTORY.md

@@ -1,3 +1,8 @@
+0.5.3 / 2014-03-14
+==================
+* Added source, director, producer and section labels
+* Updated model
+
 0.5.2 / 2014-03-13
 ==================
 * Add XML output

data/lib/anystyle/parser/normalizer.rb

@@ -109,7 +109,7 @@ module Anystyle
       def normalize_translator(hash)
         translators = hash[:translator]
 
-        editors.gsub!(/^\W+|\W+$/, '')
+        translators.gsub!(/^\W+|\W+$/, '')
         translators.gsub!(/[^[:alpha:]]*trans(lated)?[^[:alpha:]]*/i, '')
         translators.gsub!(/\bby\b/i, '')
 
@@ -117,7 +117,31 @@ module Anystyle
         hash
       end
 
+      def normalize_director(hash)
+        directors = hash[:director]
+
+        directors.gsub!(/^\W+|\W+$/, '')
+        directors.gsub!(/[^[:alpha:]]*direct(or|ed)?[^[:alpha:]]*/i, '')
+        directors.gsub!(/\bby\b/i, '')
+
+        hash[:director] = normalize_names(directors)
+        hash
+      end
+
+      def normalize_producer(hash)
+        producers = hash[:producer]
+
+        producers.gsub!(/^\W+|\W+$/, '')
+        producers.gsub!(/[^[:alpha:]]*produc(er|ed)?[^[:alpha:]]*/i, '')
+        producers.gsub!(/\bby\b/i, '')
+
+        hash[:director] = normalize_names(producers)
+        hash
+      end
+
       def normalize_names(names)
+        names.gsub!(/\s*(\.\.\.|…)\s*/, '')
+
         Namae.parse!(names).map { |name|
           unless name.given.nil? || name.family.nil?
             name.given.gsub!(/\b([[:upper:]])(\s|$)/, '\1.\2')
@@ -135,11 +159,11 @@ module Anystyle
       Namae.options[:prefer_comma_as_separator] = true
 
       def normalize_title(hash)
-        title, container = hash[:title]
+        title, source = hash[:title]
 
-        unless container.nil?
-          hash[:container] = container
-          normalize(:container, hash)
+        unless source.nil?
+          hash[:source] = source
+          normalize(:source, hash)
         end
 
         extract_edition(title, hash)
@@ -202,18 +226,18 @@ module Anystyle
         hash
       end
 
-      def normalize_container(hash)
-        container, *dangling = hash[:container]
-        unmatched(:container, hash, dangling) unless dangling.empty?
+      def normalize_source(hash)
+        source, *dangling = hash[:source]
+        unmatched(:source, hash, dangling) unless dangling.empty?
 
-        case container
+        case source
         when /dissertation abstracts/i
-          container.gsub!(/\s*section \w: ([[:alnum:]\s]+).*$/i, '')
+          source.gsub!(/\s*section \w: ([[:alnum:]\s]+).*$/i, '')
           hash[:category] = $1 unless $1.nil?
           hash[:type] = :phdthesis
         end
 
-        hash[:container] = container
+        hash[:source] = source
         hash
       end
 
@@ -227,6 +251,11 @@ module Anystyle
 
         if date =~ /(\d{4})/
           hash[:year] = $1.to_i
+
+          if hash.key?(:month) && date =~ /(\d{1,2})\b/
+            hash[:day] = $1.to_i
+          end
+
           hash.delete(:date)
         end
 
@@ -260,6 +289,24 @@ module Anystyle
         hash
       end
 
+      def normalize_publisher(hash)
+        normalize :publisher, hash
+
+        case hash[:publisher]
+        when /^producers?$/i
+          hash[:publisher] = hash[:producer]
+
+        when /^authors?$/i
+          hash[:publisher] = hash[:author]
+
+        when /^editor?$/i
+          hash[:publisher] = hash[:editor]
+
+        end
+
+        hash
+      end
+
       def normalize_pages(hash)
         pages, *dangling = hash[:pages]
         unmatched(:pages, hash, dangling) unless dangling.empty?
@@ -316,6 +363,14 @@ module Anystyle
         hash
       end
 
+      def normalize_medium(hash)
+        medium, *dangling = hash[:medium]
+        unmatched(:medium, hash, dangling) unless dangling.empty?
+
+        hash[:medium] = medium.split(/\W+/).reject(&:empty?).join('-')
+        hash
+      end
+
       private
 
       def unmatched(label, hash, tokens)

data/lib/anystyle/parser/parser.rb

@@ -180,11 +180,13 @@ module Anystyle
         text = hash.values.flatten.join
 
         case
+        when keys.include?(:medium)
+          hash[:type] = hash[:medium]
         when keys.include?(:journal)
           hash[:type] = :article
         when text =~ /proceedings/i
           hash[:type] = :inproceedings
-        when keys.include?(:booktitle), keys.include?(:container)
+        when keys.include?(:booktitle), keys.include?(:source)
           hash[:type] = :incollection
         when keys.include?(:publisher)
           hash[:type] = :book