opener-polarity-tagger 3.2.4 → 3.4.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 062fd54795d572d62bb63481dbce896c918f9b9e80a55df36b5e19f2a0be97ad
-  data.tar.gz: 6a4ff6d3ef761a22d044b7301064bc7eb0e2f75df8f806d6359aa5e5a01c42f4
+  metadata.gz: e7d99cc90032fe712bac6fa6952d1dc482d7ea5a86d9db9544f17e52d75d68f6
+  data.tar.gz: 8c8a5126f8a63a55bea7062c06d078bc87b0ff9286201956524755fb163e1172
 SHA512:
-  metadata.gz: f40d2703af5e936a3f2c747c2fc5938606ec4a8d8d65bfc8b7bd5aeaa2836de9090912c91d7f699b978d324be921128d743bfb69367539947e3202dca50abcb7
-  data.tar.gz: 884bbe39961248e5158f0b3b4c8c84ebc46bdce1fc7292a7b27eb4737fac7497a273f3a9fa91706aa283f2a814bf88b92439df3bd2dcfb8cfe0991461061b06d
+  metadata.gz: 185160b82b01008ed03ce4a26e0d93ca18b1aab91e20c57acfe7425fa103c557606e0151e161ad7d36dc7380b2f26c9daae9e9649e4b1f7b65ce29c98f270ccc
+  data.tar.gz: 73f6b42eba61db2004fff08f54a1636767e8e77e93ec817717530accfa9fcd3001097e630c043fea0d3afd462c42b755a03f274c4d74e427c1753cf70fb30648

data/lib/opener/{polarity_tagger/kaf → kaf}/term.rb

@@ -18,6 +18,10 @@ module Opener
         @node.attr :lemma
       end
 
+      def text
+        @node.attr :text
+      end
+
       def pos
         @node.attr :pos
       end

data/lib/opener/polarity_tagger/internal.rb

@@ -1,6 +1,6 @@
 require_relative 'lexicons_cache'
 require_relative 'lexicon_map'
-require_relative 'kaf/document'
+require_relative '../kaf/document'
 
 module Opener
   class PolarityTagger
@@ -16,44 +16,46 @@ module Opener
         @ignore_pos = ignore_pos
       end
 
-      def clear_cache lang: nil, environment:
-      end
-
       def run input, params = {}
-        @kaf = KAF::Document.from_xml input
+        kaf = KAF::Document.from_xml input
 
         @cache_keys = params[:cache_keys] ||= {}
-        @cache_keys.merge! lang: @kaf.language
-        @map = @kaf.map = CACHE[**@cache_keys].lexicons
+        @cache_keys.merge! lang: kaf.language
+        @map = kaf.map = CACHE[**@cache_keys].lexicons
 
-        @kaf.terms.each do |t|
+        raise Opener::Core::UnsupportedLanguageError, kaf.language if @map.blank?
+
+        kaf.terms.each do |t|
           lemma = t.lemma&.downcase
+          text  = t.text.to_s.downcase
           pos   = if @ignore_pos then nil else t.pos end
           attrs = Hashie::Mash.new
 
-          lexicon, polarity_pos = @map.by_polarity lemma, pos
+          # text matching have priority as sometimes
+          # the lemma provided by Stanza is a different word
+          lexicon, polarity_pos = @map.by_polarity text, pos
+          lexicon, polarity_pos = @map.by_polarity lemma, pos if lexicon.polarity == 'unknown'
 
-          if lexicon.polarity != 'unknown'
-            attrs.polarity = lexicon.polarity
-          end
-          if l = @map.by_negator(lemma)
+          if l = @map.by_negator(text) || @map.by_negator(lemma)
             lexicon, polarity_pos = l, nil
             attrs.sentiment_modifier = 'shifter'
-          end
-          if l = @map.by_intensifier(lemma)
+          elsif l = @map.by_intensifier(text) || @map.by_intensifier(lemma)
             lexicon, polarity_pos = l, nil
             attrs.sentiment_modifier = 'intensifier'
+          elsif lexicon.polarity != 'unknown'
+            attrs.polarity = lexicon.polarity
           end
 
           if attrs.size > 0
-            attrs.resource = lexicon.resource if lexicon.resource
+            attrs['lexicon-id'] = lexicon.id.to_s  if lexicon.id
+            attrs.resource      = lexicon.resource if lexicon.resource
             t.setPolarity attrs, polarity_pos
           end
         end
 
-        @kaf.add_linguistic_processor DESC, "#{LAST_EDITED}_#{VERSION}", 'terms'
+        kaf.add_linguistic_processor DESC, "#{LAST_EDITED}_#{VERSION}", 'terms'
 
-        @kaf.to_xml
+        kaf.to_xml
       end
 
     end

data/lib/opener/polarity_tagger/lexicon_map.rb

@@ -16,11 +16,11 @@ module Opener
         adv:         'A',
         noun:        'N',
         propernoun:  'N',
-        other:       'O',
         prep:        'P',
         verb:        'V',
+        other:       DEFAULT_POS,
         nil =>       DEFAULT_POS,
-        multi_word_expression: 'O',
+        multi_word_expression: DEFAULT_POS,
       }
 
       def initialize lang:, lexicons:
@@ -46,13 +46,12 @@ module Opener
       end
 
       def by_polarity lemma, short_pos
-        return [@with_polarity[lemma+short_pos] || UNKNOWN, short_pos] if short_pos
+        l = @with_polarity[lemma+short_pos] if short_pos
+        return [l, short_pos] if l
 
         POS_ORDER.chars.each do |short_pos|
-          if l = @with_polarity[lemma+short_pos]
-            puts "Found polarify #{l.polarity} for #{lemma} with PoS #{short_pos}"
-            return [l, short_pos]
-          end
+          l = @with_polarity[lemma+short_pos]
+          return [l, short_pos] if l
         end
 
         [UNKNOWN, 'unknown']
@@ -60,6 +59,14 @@ module Opener
 
       protected
 
+      def map_one_polarity l
+        poses = if l.poses.present? then l.poses else [l.pos] end
+        poses.each do |pos|
+          short_pos = POS_SHORT_MAP[pos&.to_sym] || DEFAULT_POS
+          @with_polarity[l.lemma+short_pos] = l
+        end
+      end
+
       def map lexicons
         return if blank?
 
@@ -71,8 +78,10 @@ module Opener
           when 'intensifier'     then @intensifiers[l.lemma] = l
           else
             if l.polarity
-              short_pos = POS_SHORT_MAP[l.pos&.to_sym] || DEFAULT_POS
-              @with_polarity[l.lemma+short_pos] = l
+              map_one_polarity l
+              l.variants&.each do |v|
+                map_one_polarity v
+              end
             end
           end
         end

data/lib/opener/polarity_tagger/version.rb

@@ -1,7 +1,7 @@
 module Opener
   class PolarityTagger
 
-    VERSION = '3.2.4'
+    VERSION = '3.4.0'
 
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: opener-polarity-tagger
 version: !ruby/object:Gem::Version
-  version: 3.2.4
+  version: 3.4.0
 platform: ruby
 authors:
 - development@olery.com
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2020-11-02 00:00:00.000000000 Z
+date: 2021-01-25 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: opener-daemons
@@ -181,10 +181,10 @@ dependencies:
 description: Polarity tagger for various languages.
 email:
 executables:
-- polarity-tagger-server
-- console
 - polarity-tagger
 - polarity-tagger-daemon
+- polarity-tagger-server
+- console
 extensions:
 - ext/hack/Rakefile
 extra_rdoc_files: []
@@ -212,12 +212,12 @@ files:
 - core/poltagger-basic-multi.py
 - exec/polarity-tagger.rb
 - ext/hack/Rakefile
+- lib/opener/kaf/document.rb
+- lib/opener/kaf/term.rb
 - lib/opener/polarity_tagger.rb
 - lib/opener/polarity_tagger/cli.rb
 - lib/opener/polarity_tagger/external.rb
 - lib/opener/polarity_tagger/internal.rb
-- lib/opener/polarity_tagger/kaf/document.rb
-- lib/opener/polarity_tagger/kaf/term.rb
 - lib/opener/polarity_tagger/lexicon_map.rb
 - lib/opener/polarity_tagger/lexicons_cache.rb
 - lib/opener/polarity_tagger/public/markdown.css