opener-chained-daemon 3.1.5 → 3.3.3

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1a3d261875bffe0427ec49e37eb23a13dc8ddddb6a0677681f7a21a550961da5
-  data.tar.gz: eaf306aa0f178899dde39b6995efa08985cb00020636a3bcaa81061b2d5bc1c2
+  metadata.gz: 5bdd841299458126e66df2dff1857ea3c2d2386efab9f092def36357935e2f9d
+  data.tar.gz: 140ff88d7ddd90a90bc21106bc9a0d3370802969bca20820ee04b45398e975c3
 SHA512:
-  metadata.gz: 98a9a18b0130a61f41697c38deb88fe48769be4c5263a05aad4bd91b472db3d47b3e1cd36e69a6f6c6a5860a7413a4322cf8853837e7155d0e0adf0087c492cf
-  data.tar.gz: 2783a9005454540407af1a2fd07f748db06ee02716720abc9695f532448e367322fe09e6c28df59e8b690435aa1958c7ae10cc773a23418a1ff6993480984fc2
+  metadata.gz: 7026baaf0c0541725cb3459575bec129354bada59be1aa30174425051d26ff973f4a0359985cb58b1ad70035ba6c6c9db4409b29631cf9b1d8fbe16152a59268
+  data.tar.gz: d947f1bc75e048b56d18b5406ed455a365c406fc1b01bf6e4b6ee0ae4584f2513e45589f93d5e1486c47fe34e1305d4a0105612ec7b17dc5d75e67c0658e0d0a

data/lib/opener/chained_daemon.rb

@@ -8,14 +8,14 @@ require 'rexml/formatters/pretty'
 
 require 'opener/daemons'
 
+require_relative 'sym_mash'
+
 require_relative 'chained_daemon/languages_cache'
 require 'opener/language_identifier'
-require 'opener/tokenizer'
-require 'opener/pos_tagger' if RUBY_ENGINE == 'jruby'
 require 'opener/polarity_tagger'
 require 'opener/property_tagger'
 require 'opener/opinion_detector_basic'
-require 'opener/stanza/tokenizer_pos'
+require 'opener/stanza/processor'
 
 require_relative 'chained_daemon/chained_daemon'
 require_relative 'chained_daemon/cli'

data/lib/opener/chained_daemon/chained_daemon.rb

@@ -8,20 +8,21 @@ module Opener
       @options   = DEFAULT_OPTIONS.merge options
       @queue_map = {
         'opener-language-identifier':    Opener::LanguageIdentifier.new,
-        'stanza-tokenizer-pos':          Stanza::TokenizerPos.new, # replace this tokenizer-pos with both below with you dont have a stanza server
-        #'opener-tokenizer':              Opener::Tokenizer.new,
-        #'opener-pos-tagger':             Opener::POSTagger.new,
+        'stanza-processor':              Stanza::Processor.new,
         'opener-property-tagger':        Opener::PropertyTagger.new,
         'opener-polarity-tagger':        Opener::PolarityTagger.new,
         'opener-opinion-detector-basic': Opener::OpinionDetectorBasic.new,
       }
     end
 
-    def run input, params = {}
-      params ||= {}
-      params.deep_symbolize_keys!
-      params[:translate_languages] ||= []
-      params[:cache_keys] = params[:cache_keys]&.sort&.to_h || {}
+    def run input, _params = {}
+      params = SymMash.new _params
+      params.translate_languages ||= []
+      params.cache_keys = SymMash.new params.cache_keys&.to_h&.sort&.to_h || {}
+      if params.filter_vertical and params.property_type.present?
+        params.cache_keys.property_type = params.property_type
+      end
+      params.cache_keys.environment ||= 'production'
 
       lang     = nil
       output   = nil
@@ -34,7 +35,7 @@ module Opener
       rescue Core::UnsupportedLanguageError
         xml  = Nokogiri.parse input
         lang = xml.root.attr('xml:lang')
-        raise unless lang.in? params[:translate_languages]
+        raise unless lang.in? params.translate_languages
 
         input = translate xml, params
         retry
@@ -47,7 +48,7 @@ module Opener
         output = xml.to_s
       end
 
-      output = pretty_print output if params[:cache_keys][:environment] == 'staging'
+      output = pretty_print output if params.cache_keys.environment == 'staging'
       output
 
     rescue Core::UnsupportedLanguageError
@@ -84,7 +85,7 @@ module Opener
     protected
 
     def translate_service params
-      params[:translate_service]&.to_sym || :google
+      params.translate_service&.to_sym || :google
     end
 
     def google_translator

data/lib/opener/chained_daemon/cli.rb

@@ -47,16 +47,19 @@ Example:
           run do |opts, args|
             daemon = ChainedDaemon.new args: args
             input  = STDIN.tty? ? nil : STDIN.read
+            params = if ENV['PARAMS'] then JSON.parse ENV['PARAMS'] else {} end
 
-            params = if ENV['PARAMS']
-                       JSON.parse ENV['PARAMS']
-                     else
-                       {}
-                     end
             # Set environment as staging from console for testing purposes
-            params[:cache_keys] = { environment: 'staging', merged: true }
+            env = ENV['LEXICONS_ENV'] || 'staging'
+            pt  = ENV['LEXICONS_PROPERTY_TYPE']
+            params[:cache_keys] = {
+              environment:   env,
+              property_type: pt,
+              merged:        (true if env == 'staging'),
+            }
 
-            puts daemon.run input, params || {}
+            output = daemon.run input, params
+            puts output
           end
         end
       end

data/lib/opener/chained_daemon/languages_cache.rb

@@ -19,15 +19,15 @@ module Opener
           break @cache if @last_updated and @last_updated > UPDATE_INTERVAL.ago
           cache_update
         end
-        @cache
       end
 
       def cache_update
         puts "loading supported languages from url #{@url}" if ENV['DEBUG']
 
-        languages     = JSON.parse http.get(@url).body
-        @cache        = languages['data'].map { |l| l['code'] }
+        languages     = SymMash.new JSON.parse http.get(@url).body
         @last_updated = Time.now
+        @cache        = languages.data.each.with_object({}){ |l,h| h[l.code] = l }
+        @cache
       end
 
       def http

data/lib/opener/chained_daemon/version.rb

@@ -1,7 +1,7 @@
 module Opener
   class ChainedDaemon
 
-    VERSION = '3.1.5'
+    VERSION = '3.3.3'
 
   end
 end

data/lib/opener/kaf/document.rb

@@ -48,13 +48,14 @@ module Opener
 
       def add_word_form params
         text = @document.at('text') || @document.root.add_child('<text/>').first
-        wf   = text.add_child("<wf>#{params[:text]}</wf>")
+        wf   = text.add_child("<wf>#{params.text}</wf>")
         attrs = {
-          wid:    "w#{params[:wid]}",
-          sent:   params[:sid],
-          para:   params[:para],
-          offset: params[:offset],
-          length: params[:length],
+          wid:    "w#{params.wid}",
+          sent:   params.sid,
+          para:   params.para,
+          offset: params.offset,
+          length: params.length,
+          head:   params.head,
         }
         wf.attr attrs
       end
@@ -63,15 +64,16 @@ module Opener
         text  = @document.at('terms') || @document.root.add_child('<terms/>').first
         term  = text.add_child("<term/>")
         attrs = {
-          tid:        "t#{params[:tid]}",
-          type:       params[:type],
-          lemma:      params[:lemma],
-          text:       params[:text],
-          pos:        params[:pos],
-          morphofeat: params[:morphofeat],
+          tid:        "t#{params.tid}",
+          type:       params.type,
+          lemma:      params.lemma,
+          text:       params.text,
+          pos:        params.pos,
+          morphofeat: params.morphofeat,
+          head:       params.head,
         }
         term.attr attrs
-        term.first.add_child("<span><target id='w#{params[:wid]}' /></span>")
+        term.first.add_child("<span><target id='w#{params.wid}'/></span>")
       end
 
       def to_xml

data/lib/opener/kaf/text.rb

@@ -1,6 +1,7 @@
 module Opener
   module KAF
     class WordForm
+
       def initialize(document, xml_node)
         @document = document
         @xml_node = xml_node
@@ -25,6 +26,7 @@ module Opener
       def paragraph
         return @paragraph ||= @xml_node.attr('para').to_i
       end
+
     end
   end
 end

data/lib/opener/stanza/{tokenizer_pos.rb → processor.rb}

@@ -1,6 +1,6 @@
 module Opener
   module Stanza
-    class TokenizerPos
+    class Processor
 
       DESC            = 'Tokenizer / POS by Stanza'
       VERSION         = '1.0'
@@ -8,8 +8,10 @@ module Opener
       BASE_URL        = ENV['STANZA_SERVER']
       LANGUAGES_CACHE = Opener::ChainedDaemon::LanguagesCache.new
 
-      RTL_LANGUAGES   = [ "ar", "ara", "arc", "ae", "ave", "egy", "he", "heb", "nqo", "pal", "phn", "sam",
-                          "syc", "syr", "fa", "per", "fas", "ku", "kur", "ur", "urd" ]
+      RTL_LANGUAGES   = %w[
+        ar ara arc ae ave egy he heb nqo pal phn sam
+        syc syr fa per fas ku kur ur urd
+      ]
 
       POS             = {
         'DET'   => 'D',
@@ -37,66 +39,65 @@ module Opener
         raise 'missing Stanza server' if ENV['STANZA_SERVER'].blank?
 
         kaf      = KAF::Document.from_xml input
-
-        prod     = params[:cache_keys][:environment] != 'staging'
-        if prod and !LANGUAGES_CACHE.get.include?(kaf.language)
+        lang     = LANGUAGES_CACHE.get[kaf.language]
+        env      = params.cache_keys.environment
+        unless lang&.environments&.include? env or (env == 'staging' and lang&.environments&.include? 'production')
           raise Core::UnsupportedLanguageError.new kaf.language
         end
 
-        input    = kaf.raw
-        input    = input.gsub(/\,[^\ ]/, ', ')
-        response = Faraday.post BASE_URL, {lang: kaf.language, input: input}.to_query
+        input     = kaf.raw
+        input     = input.gsub(/\,[^\ ]/, ', ')
+        response  = Faraday.post BASE_URL, {lang: kaf.language, input: input}.to_query
         raise Core::UnsupportedLanguageError, kaf.language if response.status == 406
         raise response.body if response.status >= 400
-        tokens   = JSON.parse response.body
+        sentences = JSON.parse response.body
+        sentences.each{ |s| s.map!{ |t| Hashie::Mash.new t } }
 
         w_index = 0
 
         miscs = {}
-        tokens.each_with_index do |t, i|
-          miscs[i] ||= {}
-          t.each do |word|
-            word['id'].is_a?(Array) && word['id'].each do |id|
-              puts id
-              miscs[i][id] = word['misc']
-            end
+        sentences.each.with_index do |s, i|
+          miscs[i] = {}
+          s.each do |word|
+            word.id.is_a?(Array) && word.id.each{ |id| miscs[i][id] = word.misc }
           end
         end
 
-        tokens.map{ |t| t.reverse! } if RTL_LANGUAGES.include? kaf.language
-        tokens.each_with_index do |sentence, s_index|
-          sentence.each_with_index do |word|
+        sentences.map{ |s| s.reverse! } if RTL_LANGUAGES.include? kaf.language
+        sentences.each.with_index do |s, s_index|
+          s.each do |word|
             w_index += 1
             # save misc for later usase in a MWT case
-            next if word['id'].is_a? Array
+            next if word.id.is_a? Array
 
-            misc = word['misc'] || miscs[s_index][word['id']]
+            misc = word.misc || miscs[s_index][word.id]
 
-            Rollbar.scoped({ input: input, params: params, tokens: tokens, word: word }) do
+            Rollbar.scoped({ input: input, params: params, sentences: sentences, word: word }) do
               raise 'Missing misc'
             end if misc.nil?
 
             offset = misc.match(/start_char=(\d+)|/)[1].to_i
             length = misc.match(/end_char=(\d+)/)[1].to_i - offset
 
-            u_pos  = word['upos']
+            u_pos  = word.upos
             pos    = POS[u_pos]
             raise "Didn't find a map for #{u_pos}" if pos.nil?
             type   = if POS_OPEN.include? pos then 'open' else 'close' end
 
-            params = {
+            params = Hashie::Mash.new(
               wid:        w_index,
               sid:        s_index + 1,
               tid:        w_index,
               para:       1,
               offset:     offset,
               length:     length,
-              text:       word['text'],
-              lemma:      word['lemma'],
+              text:       word.text,
+              lemma:      word.lemma,
               morphofeat: u_pos,
               pos:        pos,
               type:       type,
-            }
+              head:       word.head,
+            )
 
             kaf.add_word_form params
             kaf.add_term params

data/lib/opener/sym_mash.rb

@@ -0,0 +1,14 @@
+class SymMash < ::Hashie::Mash
+
+  disable_warnings
+
+  include Hashie::Extensions::Mash::SymbolizeKeys
+
+  def inspect
+    map do |k,v|
+      v = "{#{v.inspect}}" if v.is_a? SymMash
+      "#{k}=#{v}"
+    end.join ' '
+  end
+
+end

data/opener-chained-daemon.gemspec

@@ -35,8 +35,8 @@ Gem::Specification.new do |spec|
   spec.add_dependency 'opener-language-identifier', '>= 4.4.0'
   spec.add_dependency 'opener-tokenizer',  '>= 2.2.0'
   spec.add_dependency 'opener-pos-tagger', '>= 3.2.0'
-  spec.add_dependency 'opener-property-tagger', '>= 3.3.6'
-  spec.add_dependency 'opener-polarity-tagger', '>= 3.3.0'
+  spec.add_dependency 'opener-property-tagger', '>= 3.4.0'
+  spec.add_dependency 'opener-polarity-tagger', '>= 3.4.0'
   spec.add_dependency 'opener-opinion-detector-basic', '>= 3.2.3'
 
   spec.add_development_dependency 'bundler', '~> 1.3'

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: opener-chained-daemon
 version: !ruby/object:Gem::Version
-  version: 3.1.5
+  version: 3.3.3
 platform: ruby
 authors:
 - development@olery.com
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2020-12-11 00:00:00.000000000 Z
+date: 2021-02-23 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -184,28 +184,28 @@ dependencies:
     requirements:
     - - ">="
       - !ruby/object:Gem::Version
-        version: 3.3.6
+        version: 3.4.0
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - ">="
       - !ruby/object:Gem::Version
-        version: 3.3.6
+        version: 3.4.0
 - !ruby/object:Gem::Dependency
   name: opener-polarity-tagger
   requirement: !ruby/object:Gem::Requirement
     requirements:
     - - ">="
       - !ruby/object:Gem::Version
-        version: 3.3.0
+        version: 3.4.0
   type: :runtime
   prerelease: false
   version_requirements: !ruby/object:Gem::Requirement
     requirements:
     - - ">="
       - !ruby/object:Gem::Version
-        version: 3.3.0
+        version: 3.4.0
 - !ruby/object:Gem::Dependency
   name: opener-opinion-detector-basic
   requirement: !ruby/object:Gem::Requirement
@@ -307,10 +307,10 @@ dependencies:
 description: OpeNER daemon for processing multiple queues at once
 email:
 executables:
-- chained-daemon
 - chained-daemon-daemon
 - chained-daemon-csv
 - console
+- chained-daemon
 extensions: []
 extra_rdoc_files: []
 files:
@@ -330,7 +330,8 @@ files:
 - lib/opener/kaf/document.rb
 - lib/opener/kaf/term.rb
 - lib/opener/kaf/text.rb
-- lib/opener/stanza/tokenizer_pos.rb
+- lib/opener/stanza/processor.rb
+- lib/opener/sym_mash.rb
 - opener-chained-daemon.gemspec
 homepage:
 licenses: