opener-property-tagger 3.3.4 → 3.3.5

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 436af27476deb0372f325ed255151bf5128608405c740289931ecb630b111ab4
-  data.tar.gz: 5ff2d6bf101e606a786f0b4a9da8021a231ea07fac29fb51c452a1aefa737954
+  metadata.gz: 4498897f273195326764dae62875d3901919f69014df57d0919d88b0755ee777
+  data.tar.gz: e0c3c1cfcc32cf260024c2c4b5fee4afb7e6a2b4e46ba690499e8b151e959e52
 SHA512:
-  metadata.gz: 4918a0be797e2776329863a1f9f43609b9b30390e6165e73c4e089f533ce769a077e82fa928091285dbf866039dad589b7a650b7c6ffd9cc403c355862a9bd5b
-  data.tar.gz: bcac64dc7d1f229b264976070c05f64e5ff42c54ba2f809f905beb4d7597acea6c71923c271c4bc5dd77c979aa78ffdcb74757c5d27f6819bb6e62186d41fed8
+  metadata.gz: 0f0557b5cb1604b6edec19c6d5c9b2d3cc892792d7562209d619cfe0f1f09c04fa0a340f3197e8f5963d0dbefbfb890d18e9051f82705eaa346f20682b724eb1
+  data.tar.gz: 828fb63990b968d4511548b5ed6b05a44574cc9b0ed933745ddb7053cc642400ddf1ff7966f50a55b6dc84223a1fbbb8882c9cb28850dd3a8e98e550e9453183

data/lib/opener/property_tagger/file_aspects_cache.rb

@@ -36,7 +36,7 @@ module Opener
         mapping = Hash.new { |hash, key| hash[key] = [] }
 
         File.foreach(path) do |line|
-          lemma, pos, aspect = line.chomp.split("\t")
+          lemma, _pos, aspect = line.chomp.split("\t")
 
           mapping[lemma.to_sym] << aspect
         end

data/lib/opener/property_tagger/processor.rb

@@ -57,11 +57,11 @@ module Opener
 
         add_linguistic_processor
 
-        return pretty ? pretty_print(document) : document.to_xml
+        pretty ? pretty_print(document) : document.to_xml
       end
 
       def language
-        return @language ||= document.at_xpath('KAF').attr('xml:lang')
+        @language ||= document.at_xpath('KAF').attr('xml:lang')
       end
 
       def terms
@@ -69,11 +69,11 @@ module Opener
           @terms = {}
 
           document.xpath('KAF/terms/term').each do |term|
-            @terms[term.attr('tid').to_sym] = term.attr('lemma')
+            @terms[term.attr('tid').to_sym] = { lemma: term.attr('lemma'), text: term.attr('text')}
           end
         end
 
-        return @terms
+        @terms
       end
 
       ##
@@ -81,37 +81,39 @@ module Opener
       # @return [Hash]
       #
       def extract_aspects
-        term_ids  = terms.keys
-        lemmas    = terms.values
+        term_ids     = terms.keys
+        lemmas       = terms.values
+        uniq_aspects = Hash.new { |hash, key| hash[key] = [] }
 
-        current_token = 0
-        # Use of n-grams to determine if a unigram (1 lemma) or bigram (2
-        # lemmas) belong to a property.
-        max_ngram = 2
+        [:lemma, :text].each do |k|
+          current_token = 0
+          # Use of n-grams to determine if a unigram (1 lemma) or bigram (2
+          # lemmas) belong to a property.
+          max_ngram = 2
 
-        uniq_aspects = Hash.new { |hash, key| hash[key] = [] }
 
-        while current_token < terms.count
-          (0..max_ngram).each do |tam_ngram|
-            if current_token + tam_ngram <= terms.count
-              ngram = lemmas[current_token..current_token+tam_ngram].join(" ").downcase
+          while current_token < terms.count
+            (0..max_ngram).each do |tam_ngram|
+              if current_token + tam_ngram <= terms.count
+                ngram = lemmas[current_token..current_token+tam_ngram].map{|a| a[k] }.join(" ").downcase
 
-              if aspects[ngram.to_sym]
-                properties = aspects[ngram.to_sym]
-                ids        = term_ids[current_token..current_token+tam_ngram]
+                if aspects[ngram.to_sym]
+                  properties = aspects[ngram.to_sym]
+                  ids        = term_ids[current_token..current_token+tam_ngram]
 
-                properties.uniq.each do |property|
-                  next if !property or property.strip.empty?
+                  properties.uniq.each do |property|
+                    next if !property or property.strip.empty?
 
-                  uniq_aspects[property.to_sym] << [ids,ngram]
+                    uniq_aspects[property.to_sym] << [ids,ngram] unless uniq_aspects[property.to_sym].include? [ids,ngram]
+                  end
                 end
               end
             end
+            current_token += 1
           end
-          current_token += 1
         end
 
-        return Hash[uniq_aspects.sort]
+        Hash[uniq_aspects.sort]
       end
 
       ##
@@ -190,7 +192,7 @@ module Opener
         formatter.compact = true
         formatter.write(doc, out)
 
-        return out.strip
+        out.strip
       end
 
       protected
@@ -206,7 +208,7 @@ module Opener
 
         parent_node.add_child node
 
-        return node
+        node
       end
 
       ##
@@ -214,7 +216,7 @@ module Opener
       # @return [Boolean]
       #
       def is_kaf?
-        return !!document.at_xpath('KAF')
+        !!document.at_xpath('KAF')
       end
 
       ##

data/lib/opener/property_tagger/version.rb

@@ -1,7 +1,7 @@
 module Opener
   class PropertyTagger
 
-    VERSION = '3.3.4'
+    VERSION = '3.3.5'
 
   end
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: opener-property-tagger
 version: !ruby/object:Gem::Version
-  version: 3.3.4
+  version: 3.3.5
 platform: ruby
 authors:
 - development@olery.com
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2020-11-09 00:00:00.000000000 Z
+date: 2020-11-10 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: opener-daemons
@@ -167,9 +167,9 @@ dependencies:
 description: Property tagger for hotels in Dutch and English.
 email:
 executables:
-- property-tagger
 - property-tagger-daemon
 - property-tagger-server
+- property-tagger
 extensions: []
 extra_rdoc_files: []
 files:
@@ -213,7 +213,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project:
-rubygems_version: 2.7.8
+rubygems_version: 2.7.6
 signing_key:
 specification_version: 4
 summary: Property tagger for hotels in Dutch and English.