Pass prev context to normalizers

inukshuk · May 29, 2018 · e268f50 · e268f50
1 parent 411c60c
commit e268f50
Show file tree

Hide file tree

Showing 18 changed files with 24 additions and 42 deletions.
diff --git a/lib/anystyle/normalizer.rb b/lib/anystyle/normalizer.rb
@@ -18,7 +18,7 @@ def name
       self.class.name
     end
 
-    def normalize(item)
+    def normalize(item, **opts)
       raise NotImplementedError
     end
 

diff --git a/lib/anystyle/normalizer/brackets.rb b/lib/anystyle/normalizer/brackets.rb
@@ -3,7 +3,7 @@ class Normalizer
     class Brackets < Normalizer
       @keys = [:'citation-number', :note]
 
-      def normalize(item)
+      def normalize(item, **opts)
         each_value(item) do |_, value|
           value.gsub!(/^[\(\[\{]|[\]\)\}]$/, '')
         end

diff --git a/lib/anystyle/normalizer/container.rb b/lib/anystyle/normalizer/container.rb
@@ -3,7 +3,7 @@ class Normalizer
     class Container < Normalizer
       @keys = [:'container-title']
 
-      def normalize(item)
+      def normalize(item, **opts)
         map_values(item) do |_, value|
           value.gsub(/^[Ii]n:?\s+/, '')
         end

diff --git a/lib/anystyle/normalizer/date.rb b/lib/anystyle/normalizer/date.rb
@@ -3,15 +3,15 @@ class Normalizer
     class Date < Normalizer
       @keys = [:date]
 
-      # TODO return seasons/ intervals in ISO/EDTF format
-      def normalize(item)
+      def normalize(item, **opts)
         map_values(item) do |_, value|
           case
           when unknown?(value)
             'XXXX'
           when interval?(value)
             value
           # TODO AD/BC
+          # TODO Seasons
           when iso?(value)
             value
           else

diff --git a/lib/anystyle/normalizer/edition.rb b/lib/anystyle/normalizer/edition.rb
@@ -3,7 +3,7 @@ class Normalizer
     class Edition < Normalizer
       @keys = [:edition]
 
-      def normalize(item)
+      def normalize(item, **opts)
         map_values(item) do |_, value|
           value
             .gsub(/rev\./, 'revised')

diff --git a/lib/anystyle/normalizer/journal.rb b/lib/anystyle/normalizer/journal.rb
@@ -1,7 +1,7 @@
 module AnyStyle
   class Normalizer
     class Journal < Normalizer
-      def normalize(item)
+      def normalize(item, **opts)
         if item.key?(:journal)
           item[:type] = 'article-journal'
           item[:journal].each { |journal| append item, :'container-title', journal }

diff --git a/lib/anystyle/normalizer/locale.rb b/lib/anystyle/normalizer/locale.rb
@@ -7,14 +7,15 @@ def initialize
         @ld = LanguageDetector.new if defined?(LanguageDetector)
       end
 
-      def normalize(item)
+      def normalize(item, **opts)
         return item if @ld.nil? || item.key?(:language)
 
         sample = item.values_at(
           :title,
           :'container-title',
 #          :'collection-title',
           :location,
+          :journal,
           :publisher
 #          :note
         ).flatten.compact.join(' ')

diff --git a/lib/anystyle/normalizer/location.rb b/lib/anystyle/normalizer/location.rb
@@ -3,7 +3,7 @@ class Normalizer
     class Location < Normalizer
       @keys = [:location]
 
-      def normalize(item)
+      def normalize(item, **opts)
         map_values(item) do |_, value|
           location = strip value
 

diff --git a/lib/anystyle/normalizer/locator.rb b/lib/anystyle/normalizer/locator.rb
@@ -5,7 +5,7 @@ class Normalizer
     class Locator < Normalizer
       @keys = [:isbn, :url]
 
-      def normalize(item)
+      def normalize(item, **opts)
         map_values(item) do |key, value|
           case key
           when :isbn

diff --git a/lib/anystyle/normalizer/names.rb b/lib/anystyle/normalizer/names.rb
@@ -20,7 +20,7 @@ def initialize(**options)
         })
       end
 
-      def normalize(item)
+      def normalize(item, **opts)
         map_values(item) do |_, value|
           begin
             parse(strip(value))

diff --git a/lib/anystyle/normalizer/page.rb b/lib/anystyle/normalizer/page.rb
@@ -3,7 +3,7 @@ class Normalizer
     class Page < Normalizer
       @keys = [:pages]
 
-      def normalize(item)
+      def normalize(item, **opts)
         map_values(item) do |_, value|
           pages = case value
             when /(\d+)(?:\.(\d+))?(?:\((\d{4})\))?:(\d.*)/

diff --git a/lib/anystyle/normalizer/publisher.rb b/lib/anystyle/normalizer/publisher.rb
@@ -3,7 +3,7 @@ class Normalizer
     class Publisher < Normalizer
       @keys = [:publisher]
 
-      def normalize(item)
+      def normalize(item, **opts)
         replace_author(item) if item.key?(:author)
         item
       end

diff --git a/lib/anystyle/normalizer/pubmed.rb b/lib/anystyle/normalizer/pubmed.rb
@@ -3,7 +3,7 @@ class Normalizer
     class PubMed < Normalizer
       @keys = [:note]
 
-      def normalize(item)
+      def normalize(item, **opts)
         each_value(item) do |_, value|
           if (value =~ /PMID:?\s*(\d+)/)
             append :pmid, $1

diff --git a/lib/anystyle/normalizer/punctuation.rb b/lib/anystyle/normalizer/punctuation.rb
@@ -11,7 +11,7 @@ class Punctuation < Normalizer
         :location
       ]
 
-      def normalize(item)
+      def normalize(item, **opts)
         each_value(item) do |_, value|
           value.gsub!(/[\)\]\.,:;\p{Pd}\p{Z}\p{C}]+$/, '')
           value.gsub!(/^[\(\[]/, '')

diff --git a/lib/anystyle/normalizer/quotes.rb b/lib/anystyle/normalizer/quotes.rb
@@ -4,7 +4,7 @@ class Quotes < Normalizer
       QUOTES = /^[«‹»›„‚“‟‘‛”’"❛❜❟❝❞⹂〝〞〟\[]|[«‹»›„‚“‟‘‛”’"❛❜❟❝❞⹂〝〞〟\]]$/
       @keys = [:title, :'citation-number', :medium]
 
-      def normalize(item)
+      def normalize(item, **opts)
         each_value(item) do |_, value|
           value.gsub! QUOTES, ''
         end

diff --git a/lib/anystyle/normalizer/type.rb b/lib/anystyle/normalizer/type.rb
@@ -1,7 +1,7 @@
 module AnyStyle
   class Normalizer
     class Type < Normalizer
-      def normalize(item)
+      def normalize(item, **opts)
         item[:type] = classify item unless item.key?(:type)
         item
       end

diff --git a/lib/anystyle/normalizer/volume.rb b/lib/anystyle/normalizer/volume.rb
@@ -3,28 +3,7 @@ class Normalizer
     class Volume < Normalizer
       @keys = [:volume, :pages, :date]
 
-      def normalize(item)
-        # TODO
-        #if !hash.has_key?(:pages) && volume =~ /\D*(\d+):(\d+(?:[—–-]+)\d+)/
-        #  hash[:volume], hash[:pages] = $1.to_i, $2
-        #  hash = normalize_pages(hash)
-        #else
-        #  case volume
-        #  when /\D*(\d+)\D+(\d+[\s\/&—–-]+\d+)/
-        #    hash[:volume], hash[:number] = $1.to_i, $2
-        #  when /(\d+)?\D+no\.\s*(\d+\D+\d+)/
-        #    hash[:volume] = $1.to_i unless $1.nil?
-        #    hash[:number] = $2
-        #  when /(\d+)?\D+no\.\s*(\d+)/
-        #    hash[:volume] = $1.to_i unless $1.nil?
-        #    hash[:number] = $2.to_i
-        #  when /\D*(\d+)\D+(\d+)/
-        #    hash[:volume], hash[:number] = $1.to_i, $2.to_i
-        #  when /(\d+)/
-        #    hash[:volume] = $1.to_i
-        #  end
-        #end
-
+      def normalize(item, **opts)
         map_values(item, [:volume]) do |_, volume|
           case volume
           when /(\p{Lu}?\d+)\s?\(([^)]+)\)/

diff --git a/lib/anystyle/parser.rb b/lib/anystyle/parser.rb
@@ -54,10 +54,10 @@ def learn(input)
       train(input, truncate: false)
     end
 
-    def normalize(hash)
+    def normalize(hash, **opts)
       normalizers.each do |n|
         begin
-          hash = n.normalize(hash) unless n.skip?
+          hash = n.normalize(hash, **opts) unless n.skip?
         rescue => e
           warn "Error in #{n.name} normalizer: #{e.message}"
         end
@@ -156,7 +156,9 @@ def expand(dataset)
     end
 
     def format_hash(dataset, symbolize_keys: true)
-      dataset.map { |seq| normalize(seq.to_h(symbolize_keys: symbolize_keys)) }
+      dataset.inject([]) { |out, seq|
+        out << normalize(seq.to_h(symbolize_keys: symbolize_keys), prev: out)
+      }
     end
 
     def flatten_values(hash, skip: [])