immosquare-translate 0.1.5 → 0.1.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 9daf467e21100501ea1833692e80742d341950fb05f1a451b25134a83e7ba7e1
-  data.tar.gz: 7fe077c183fb8beee8d917abad7d014eafd811ef300574cfc776427efad51f0e
+  metadata.gz: fdfdc479af37a288d9d1b16fad5aa331509facfaf906982f2eeec13c4f88bf89
+  data.tar.gz: d4025b70638d454d4363abf40dd29f0a6d86bae51406cf60f0b5dd4b3e77646f
 SHA512:
-  metadata.gz: 5758dc40a24ab4721d4ec72184d181567e96949b03aeb71e8bbdd14ffc107f0def5197719bf069ab65486fb65876223029d67f5f82de12a5ee96391c5dd61b97
-  data.tar.gz: 8ade76c458bb26cec34a1c369703b6e1141a10fc99fbb56774e423ea3ccb95658c82863c7c86cb9169d3fd82e2434cafeb200390f2b7d283a391de44c574e868
+  metadata.gz: e5da8f3f28db2e39bc251d9a9f466cafdadd0f90a2a122b7e62f05ca66b413e6afda8685c634ab78aa3a5dadc15de513596bef9163516a40d97623a87d454662
+  data.tar.gz: b9d956d253702bb30027f90a8052184be56ce86fc246e728211847f038bbc5ceea885dbc6345f0955d1d59fb7ab102948fa81429c5d71129050ddf6ae8ce77de

data/lib/immosquare-translate/translator.rb

@@ -27,23 +27,26 @@ module ImmosquareTranslate
           }
 
           prompt_system = "As a sophisticated translation AI, your role is to translate sentences from a specified source language to multiple target languages.\n" \
-                          "It is imperative that you return the translations in a single, pure JSON string format. Use ISO 639-1 language codes for specifying languages.\n" \
-                          "If string is html, you should return the translated html.\n" \
-                          "Ensure that the output does not include markdown (```json) or any other formatting characters. Adhere to the JSON structure meticulously."
+                          "Rules to respect:\n" \
+                          "- Use ISO 639-1 language codes for specifying languages." \
+                          "- Correct any spelling or grammatical errors in the source text before translating.\n" \
+                          "- If the source language is also a target language, include the corrected version of the sentence for that language as well, if not dont include it.\n" \
+                          "- If string to translate is html, you should return the translated html.\n" \
+                          "- If string to translate contains 3 underscores in row ___, keep them, don't translate the them, don't remove them, don't change them for dashes (they correspond to the newline)\n" \
+                          "- If string to translate contains 3 dashes in row ---, keep them, don't translate the them, don't remove them, don't change them for underscores (they correspond to the tabulation)\n" \
+                          "- Ensure that the output does not include markdown (```json) or any other formatting characters. Adhere to the JSON structure meticulously.\n" \
+                          "- Ensure the output strictly follows this JSON format, without any extraneous characters or formatting.\n" \
+                          "- Format the translation output as a JSON string adhering to the following structure: {\"datas\":[{\"locale_iso\": \"Translated Text\"}]}, where 'locale_iso' is replaced with the appropriate ISO 639-1 code for each translation."
 
 
-
-          prompt = "Translate the #{text.size} #{text.size == 1 ? "sentence" : "sentences"} provided below from the source language (ISO 639-1 code: #{from}) to the target languages specified: #{to_iso.map {|iso, language| "#{language} (ISO 639-1 code: #{iso})" }.join(", ")}. " \
-                   "Before translating, correct any spelling or grammatical errors in each sentence. " \
-                   "If the source language is also a target language, include the corrected version of the sentence for that language as well. " \
-                   "Format the translation output as a JSON string adhering to the following structure: " \
-                   "{\"datas\":[{\"locale_iso\": \"Translated Text\"}]}, where 'locale_iso' is replaced with the appropriate ISO 639-1 code for each translation. " \
-                   "Ensure the output strictly follows this JSON format, without any extraneous characters or formatting. " \
-                   "Sentences to translate:"
+          prompt = "Translate the #{text.size} following #{text.size == 1 ? "sentence" : "sentences"} from the source language (ISO 639-1 code: #{from}) to the target languages specified: #{to_iso.map {|iso, language| "#{language} (ISO 639-1 code: #{iso})" }.join(", ")}. "
 
 
+          ##============================================================##
+          ## we replace the \n by ___ to avoid the API to remove them
+          ##============================================================##
           text.each_with_index do |sentence, index|
-            prompt += "\n#{index + 1}: #{sentence}"
+            prompt += "\n#{index + 1}: #{sentence.gsub("\n", "___").gsub("\t", "---")}"
           end
 
 
@@ -88,7 +91,9 @@ module ImmosquareTranslate
           content = JSON.parse(choice["message"]["content"])
           datas   = content["datas"]
           datas.map do |hash|
-            hash.select {|key, _| to.include?(key) }
+            hash
+              .select {|key, _| to.include?(key) }
+              .transform_values {|value| value.gsub("___", "\n").gsub("---", "\t") }
           end.reject(&:empty?)
         rescue StandardError => e
           puts(e.message)

data/lib/immosquare-translate/version.rb

@@ -1,3 +1,3 @@
 module ImmosquareTranslate
-  VERSION = "0.1.5".freeze
+  VERSION = "0.1.7".freeze
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: immosquare-translate
 version: !ruby/object:Gem::Version
-  version: 0.1.5
+  version: 0.1.7
 platform: ruby
 authors:
 - IMMO SQUARE