opener-chained-daemon 3.0.7 → 3.1.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1f37e21fff5f35b6736a17e4d4efcad5d496c1c2ec63e74c7b8907db45210770
-  data.tar.gz: 6e3b80ba665a950407b1c4bcba9e0f6074db41db86b2da233b21c1db638c792f
+  metadata.gz: 8f4ae996f2569fd72fbb224a89aa6f1bd6335419a514db0714858e9c60e51511
+  data.tar.gz: ce58d5715abd85083f70e2b6bb7789581f7007a3be4583050af0c401c61a224e
 SHA512:
-  metadata.gz: 5738bfe0a4dbf5ff66010a9aa54ef7de4e951e7728d753000b99effbb0f4c17884791a8eb6c58fcfcafd51f0ba94520148f6d3699dd592c74145766e2c06917f
-  data.tar.gz: 7100f63ac6fd859b98f28eac22dd73e60419f21fc40e3b59c34c6550915182ca0598eb74f2ec9eddaa870fd4b91f94b31a694f70fc62ed20307b1415c0e82804
+  metadata.gz: 9cfb8fc04839f942f05d182bbac523764502a8f204679bf81515272839763572d72707ec9156f662f1538149a7f20b79741e3ad941b7da98bdb1ab0a7dcfd82b
+  data.tar.gz: 6d03594f42b21240fc2210747884cd0de4296c469b581561b9c3845a86cb2696265052cae19002c7e1308cc28ab118927c2978ff67e88e6ea7e57dba2eceefaf

data/lib/opener/chained_daemon.rb

@@ -3,6 +3,9 @@ require 'oga'
 require 'hashie'
 require 'google/cloud/translate'
 
+require 'rexml/document'
+require 'rexml/formatters/pretty'
+
 require 'opener/daemons'
 
 require_relative 'chained_daemon/languages_cache'

data/lib/opener/chained_daemon/chained_daemon.rb

@@ -47,6 +47,7 @@ module Opener
         output = xml.to_s
       end
 
+      output = pretty_print output if params[:cache_keys][:environment] == 'staging'
       output
 
     rescue Core::UnsupportedLanguageError
@@ -54,6 +55,17 @@ module Opener
       output
     end
 
+    def pretty_print(output)
+      doc = REXML::Document.new output
+      doc.context[:attribute_quote] = :quote
+      out = ""
+      formatter = REXML::Formatters::Pretty.new
+      formatter.compact = true
+      formatter.write(doc, out)
+
+      out.strip
+    end
+
     def translate xml, params
       raw = xml.at :raw
       case translate_service params

data/lib/opener/chained_daemon/cli.rb

@@ -47,7 +47,14 @@ Example:
           run do |opts, args|
             daemon = ChainedDaemon.new args: args
             input  = STDIN.tty? ? nil : STDIN.read
-            params = JSON.parse ENV['PARAMS'] if ENV['PARAMS']
+
+            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 }
 
             puts daemon.run input, params || {}
           end

data/lib/opener/chained_daemon/version.rb

@@ -1,7 +1,7 @@
 module Opener
   class ChainedDaemon
 
-    VERSION = '3.0.7'
+    VERSION = '3.1.4'
 
   end
 end

data/lib/opener/stanza/tokenizer_pos.rb

@@ -8,6 +8,9 @@ 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" ]
+
       POS             = {
         'DET'   => 'D',
         'ADJ'   => 'G',
@@ -40,16 +43,31 @@ module Opener
           raise Core::UnsupportedLanguageError.new kaf.language
         end
 
-        response = Faraday.post BASE_URL, {lang: kaf.language, input: kaf.raw}.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
 
         w_index = 0
+
+        tokens.map{ |t| t.reverse! } if RTL_LANGUAGES.include? kaf.language
         tokens.each_with_index do |sentence, s_index|
+          miscs = {}
           sentence.each_with_index do |word|
             w_index += 1
-            misc   = word['misc']
+            # save misc for later usase in a MWT case
+            if word['id'].is_a? Array
+              word['id'].each { |id| miscs[id] = word['misc'] }
+              next
+            end
+            misc = word['misc'] || miscs[word['id']]
+
+            Rollbar.scoped({ input: input, params: params, tokens: tokens, 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
 

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: opener-chained-daemon
 version: !ruby/object:Gem::Version
-  version: 3.0.7
+  version: 3.1.4
 platform: ruby
 authors:
 - development@olery.com
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2020-11-11 00:00:00.000000000 Z
+date: 2020-11-23 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport