opener-polarity-tagger 3.2.3 → 3.3.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: a62617dcfcc50787c56ac1243ecdacdd6c85d39621662e3c585a944453153469
-  data.tar.gz: 818c317c7b7ce7ddc4668bcef407ceb41e9a96ed781c0dcf47e6bedfd016cfaf
+  metadata.gz: dcce63ff8ce9916a2387c5d0f48209b203eacbc3834eedcf3b5b4990084ac1ad
+  data.tar.gz: 495e79bfdec921d459ff78e50bdae28a2c6a0a8bc2aed150de27f49a7f99c423
 SHA512:
-  metadata.gz: babc5afc6e4ab7b7845386e2493c88aa975429397aa695eac4b8705f3aa92417ce240b367844b7cdeec1f3f446a792e5d501a38f7e303c1587196bb4020ebf14
-  data.tar.gz: 221d76e8fe87f56861e7a48ac5dd1ce3bae9dd85397ce0d165f67c13cfaf9241fc7cd31cd5f630b0fdb15b5d373ff3b307e73f7e993e6b576faf2ddcbaf2938b
+  metadata.gz: e51288c630084ce9be780b022ae23012337622b6df30de91962f5c73fd2ef32054435247bf850924a40ca723c3fe1a71fc9f688455b1ea5141acf80959fbe37a
+  data.tar.gz: e8f00bd36a27aab5242d91e3f350d4da94d2661ef5565590e1793e02eb65075cf7dd789682b2cd0e06cf4b84bb21a5d5671d51f7e9d94ec1748cd1c8012f33ac

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
@@ -16,33 +16,34 @@ 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
@@ -51,9 +52,9 @@ module Opener
           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']

data/lib/opener/polarity_tagger/lexicons_cache.rb

@@ -46,7 +46,7 @@ module Opener
 
       def load_from_url lang:, cache:, **params
         url  = "#{@url}&language_code=#{lang}&#{params.to_query}"
-        url += "&if_updated_since=#{cache.from.iso8601}" if cache
+        url += "&if_updated_since=#{cache.from.utc.iso8601}" if cache
         puts "#{lang}: loading lexicons from url #{url}"
 
         lexicons = JSON.parse http.get(url).body

data/lib/opener/polarity_tagger/version.rb

@@ -1,7 +1,7 @@
 module Opener
   class PolarityTagger
 
-    VERSION = '3.2.3'
+    VERSION = '3.3.0'
 
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: opener-polarity-tagger
 version: !ruby/object:Gem::Version
-  version: 3.2.3
+  version: 3.3.0
 platform: ruby
 authors:
 - development@olery.com
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2020-11-02 00:00:00.000000000 Z
+date: 2020-12-10 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