immosquare-yaml 0.1.25 → 0.1.26
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +4 -4
- data/lib/immosquare-yaml/version.rb +1 -1
- data/lib/immosquare-yaml.rb +1 -1
- data/lib/tasks/immosquare-yaml.rake +0 -26
- metadata +3 -32
- data/lib/immosquare-yaml/translate.rb +0 -365
checksums.yaml
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
---
|
|
2
2
|
SHA256:
|
|
3
|
-
metadata.gz:
|
|
4
|
-
data.tar.gz:
|
|
3
|
+
metadata.gz: 896c1bf0647c93ce8c4d3c0a94da529d1f63f8e7e91541c6cac3d88a7edc6f1f
|
|
4
|
+
data.tar.gz: 48a6746f1840b50eb8be76985da5f8e7e6a2812ad532cc45b0ffebb07c6f2275
|
|
5
5
|
SHA512:
|
|
6
|
-
metadata.gz:
|
|
7
|
-
data.tar.gz:
|
|
6
|
+
metadata.gz: 5d53aeb26beb740b12f71715e9320fe230e817f0926e8cabf2cdbb404cf5ec9d8b28e9d5d4ea6b62b090bd5a2bd87a983fb1520e87c8edb77b511342a9ad537e
|
|
7
|
+
data.tar.gz: 89ba2a15d0f64e377dd4d1484e6afcf6f5b821631c16cf3b90d8af124095f863e947d8e10aea9373d27d2edf90e2cacf3f3051359fbbf5c17c96352086374282
|
data/lib/immosquare-yaml.rb
CHANGED
|
@@ -1,9 +1,9 @@
|
|
|
1
1
|
require "English"
|
|
2
2
|
require "psych"
|
|
3
|
+
require "date"
|
|
3
4
|
require "immosquare-extensions"
|
|
4
5
|
require_relative "immosquare-yaml/configuration"
|
|
5
6
|
require_relative "immosquare-yaml/shared_methods"
|
|
6
|
-
require_relative "immosquare-yaml/translate"
|
|
7
7
|
require_relative "immosquare-yaml/railtie" if defined?(Rails)
|
|
8
8
|
|
|
9
9
|
##===========================================================================##
|
|
@@ -1,30 +1,4 @@
|
|
|
1
1
|
namespace :immosquare_yaml do
|
|
2
|
-
##============================================================##
|
|
3
|
-
## Function to translate translation files in rails app
|
|
4
|
-
## rake immosquare_yaml:translate SOURCE_LOCALE=fr
|
|
5
|
-
##============================================================##
|
|
6
|
-
desc "Translate translation files in rails app"
|
|
7
|
-
task :translate => :environment do
|
|
8
|
-
begin
|
|
9
|
-
source_locale = ENV.fetch("SOURCE_LOCALE", nil) || "fr"
|
|
10
|
-
reset_translations = ENV.fetch("RESET_TRANSLATIONS", nil) || false
|
|
11
|
-
reset_translations = reset_translations == "true"
|
|
12
|
-
|
|
13
|
-
raise("Please provide a valid locale") if !I18n.available_locales.map(&:to_s).include?(source_locale)
|
|
14
|
-
raise("Please provide a valid boolean for reset_translations") if ![true, false].include?(reset_translations)
|
|
15
|
-
|
|
16
|
-
locales = I18n.available_locales.map(&:to_s).reject {|l| l == source_locale }
|
|
17
|
-
puts("Translating from #{source_locale} to #{locales.join(", ")} with reset_translations=#{reset_translations}")
|
|
18
|
-
Dir.glob("#{Rails.root}/config/locales/**/*#{source_locale}.yml").each do |file|
|
|
19
|
-
locales.each do |locale|
|
|
20
|
-
ImmosquareYaml::Translate.translate(file, locale, :reset_translations => reset_translations)
|
|
21
|
-
end
|
|
22
|
-
end
|
|
23
|
-
rescue StandardError => e
|
|
24
|
-
puts(e.message)
|
|
25
|
-
end
|
|
26
|
-
end
|
|
27
|
-
|
|
28
2
|
##============================================================##
|
|
29
3
|
## Function to clean translation files in rails app
|
|
30
4
|
##============================================================##
|
metadata
CHANGED
|
@@ -1,7 +1,7 @@
|
|
|
1
1
|
--- !ruby/object:Gem::Specification
|
|
2
2
|
name: immosquare-yaml
|
|
3
3
|
version: !ruby/object:Gem::Version
|
|
4
|
-
version: 0.1.
|
|
4
|
+
version: 0.1.26
|
|
5
5
|
platform: ruby
|
|
6
6
|
authors:
|
|
7
7
|
- IMMO SQUARE
|
|
@@ -10,20 +10,6 @@ bindir: bin
|
|
|
10
10
|
cert_chain: []
|
|
11
11
|
date: 2024-03-15 00:00:00.000000000 Z
|
|
12
12
|
dependencies:
|
|
13
|
-
- !ruby/object:Gem::Dependency
|
|
14
|
-
name: httparty
|
|
15
|
-
requirement: !ruby/object:Gem::Requirement
|
|
16
|
-
requirements:
|
|
17
|
-
- - "~>"
|
|
18
|
-
- !ruby/object:Gem::Version
|
|
19
|
-
version: '0'
|
|
20
|
-
type: :runtime
|
|
21
|
-
prerelease: false
|
|
22
|
-
version_requirements: !ruby/object:Gem::Requirement
|
|
23
|
-
requirements:
|
|
24
|
-
- - "~>"
|
|
25
|
-
- !ruby/object:Gem::Version
|
|
26
|
-
version: '0'
|
|
27
13
|
- !ruby/object:Gem::Dependency
|
|
28
14
|
name: immosquare-extensions
|
|
29
15
|
requirement: !ruby/object:Gem::Requirement
|
|
@@ -33,7 +19,7 @@ dependencies:
|
|
|
33
19
|
version: '0'
|
|
34
20
|
- - ">="
|
|
35
21
|
- !ruby/object:Gem::Version
|
|
36
|
-
version: 0.1.
|
|
22
|
+
version: 0.1.18
|
|
37
23
|
type: :runtime
|
|
38
24
|
prerelease: false
|
|
39
25
|
version_requirements: !ruby/object:Gem::Requirement
|
|
@@ -43,21 +29,7 @@ dependencies:
|
|
|
43
29
|
version: '0'
|
|
44
30
|
- - ">="
|
|
45
31
|
- !ruby/object:Gem::Version
|
|
46
|
-
version: 0.1.
|
|
47
|
-
- !ruby/object:Gem::Dependency
|
|
48
|
-
name: iso-639
|
|
49
|
-
requirement: !ruby/object:Gem::Requirement
|
|
50
|
-
requirements:
|
|
51
|
-
- - "~>"
|
|
52
|
-
- !ruby/object:Gem::Version
|
|
53
|
-
version: '0'
|
|
54
|
-
type: :runtime
|
|
55
|
-
prerelease: false
|
|
56
|
-
version_requirements: !ruby/object:Gem::Requirement
|
|
57
|
-
requirements:
|
|
58
|
-
- - "~>"
|
|
59
|
-
- !ruby/object:Gem::Version
|
|
60
|
-
version: '0'
|
|
32
|
+
version: 0.1.18
|
|
61
33
|
description: IMMOSQUARE-YAML is a specialized Ruby gem tailored primarily for parsing
|
|
62
34
|
and dumping YML translation files, addressing challenges faced with other parsers
|
|
63
35
|
like interpreting translation keys as booleans, multi-line strings, and more.
|
|
@@ -71,7 +43,6 @@ files:
|
|
|
71
43
|
- lib/immosquare-yaml/configuration.rb
|
|
72
44
|
- lib/immosquare-yaml/railtie.rb
|
|
73
45
|
- lib/immosquare-yaml/shared_methods.rb
|
|
74
|
-
- lib/immosquare-yaml/translate.rb
|
|
75
46
|
- lib/immosquare-yaml/version.rb
|
|
76
47
|
- lib/tasks/immosquare-yaml.rake
|
|
77
48
|
homepage: https://github.com/IMMOSQUARE/immosquare-yaml
|
|
@@ -1,365 +0,0 @@
|
|
|
1
|
-
require "iso-639"
|
|
2
|
-
require "httparty"
|
|
3
|
-
|
|
4
|
-
|
|
5
|
-
module ImmosquareYaml
|
|
6
|
-
module Translate
|
|
7
|
-
extend SharedMethods
|
|
8
|
-
|
|
9
|
-
class << self
|
|
10
|
-
|
|
11
|
-
def translate(file_path, locale_to, options = {})
|
|
12
|
-
begin
|
|
13
|
-
##=============================================================##
|
|
14
|
-
## options
|
|
15
|
-
##=============================================================##
|
|
16
|
-
options = {
|
|
17
|
-
:reset_translations => false
|
|
18
|
-
}.merge(options)
|
|
19
|
-
options[:reset_translations] = false if ![true, false].include?(options[:reset_translations])
|
|
20
|
-
|
|
21
|
-
|
|
22
|
-
##=============================================================##
|
|
23
|
-
## Load config keys from config_dev.yml
|
|
24
|
-
##=============================================================##
|
|
25
|
-
raise("Error: openai_api_key not found in config_dev.yml") if ImmosquareYaml.configuration.openai_api_key.nil?
|
|
26
|
-
raise("Error: File #{file_path} not found") if !File.exist?(file_path)
|
|
27
|
-
raise("Error: locale is not a locale") if !locale_to.is_a?(String) || locale_to.size != 2
|
|
28
|
-
|
|
29
|
-
##============================================================##
|
|
30
|
-
## We clean the file before translation
|
|
31
|
-
##============================================================##
|
|
32
|
-
ImmosquareYaml.clean(file_path)
|
|
33
|
-
|
|
34
|
-
##============================================================##
|
|
35
|
-
## We parse the clean input file
|
|
36
|
-
##============================================================##
|
|
37
|
-
hash_from = ImmosquareYaml.parse(file_path)
|
|
38
|
-
raise("#{file_path} is not a correct yml translation file") if !hash_from.is_a?(Hash) && hash_from.keys.size > 1
|
|
39
|
-
|
|
40
|
-
##============================================================##
|
|
41
|
-
## Check if the locale is present in the file
|
|
42
|
-
##============================================================##
|
|
43
|
-
locale_from = hash_from.keys.first.to_s
|
|
44
|
-
raise("Error: The destination file (#{locale_to}) is the same as the source file (#{locale_from}).") if locale_from == locale_to
|
|
45
|
-
raise("Error: Expected the source file (#{file_path}) to end with '#{locale_from}.yml' but it didn't.") if !file_path.end_with?("#{locale_from}.yml")
|
|
46
|
-
|
|
47
|
-
|
|
48
|
-
##============================================================##
|
|
49
|
-
## Prepare the output file
|
|
50
|
-
##============================================================##
|
|
51
|
-
file_basename = File.basename(file_path)
|
|
52
|
-
file_dirname = File.dirname(file_path)
|
|
53
|
-
translated_file_path = "#{file_dirname}/#{file_basename.gsub("#{locale_from}.yml", "#{locale_to}.yml")}"
|
|
54
|
-
|
|
55
|
-
##============================================================##
|
|
56
|
-
## We create a hash with all keys from the source file
|
|
57
|
-
##============================================================##
|
|
58
|
-
hash_to = {locale_to => hash_from.delete(locale_from)}
|
|
59
|
-
|
|
60
|
-
##============================================================##
|
|
61
|
-
## We create a array with all keys from the source file
|
|
62
|
-
##============================================================##
|
|
63
|
-
array_to = translatable_array(hash_to)
|
|
64
|
-
array_to = array_to.map {|k, v| [k, v, nil] }
|
|
65
|
-
|
|
66
|
-
##============================================================##
|
|
67
|
-
## If we already have a translation file for the language
|
|
68
|
-
## we get the values in it and put it in our
|
|
69
|
-
## file... You have to do well with !nil?
|
|
70
|
-
## to retrieve the values "" and " "...
|
|
71
|
-
##============================================================##
|
|
72
|
-
if File.exist?(translated_file_path) && options[:reset_translations] == false
|
|
73
|
-
temp_hash = ImmosquareYaml.parse(translated_file_path)
|
|
74
|
-
raise("#{translated_file_path} is not a correct yml translation file") if !temp_hash.is_a?(Hash) && temp_hash.keys.size > 1
|
|
75
|
-
|
|
76
|
-
##============================================================##
|
|
77
|
-
## t can be nil if the key is not present in the source file
|
|
78
|
-
##============================================================##
|
|
79
|
-
translatable_array(temp_hash).each do |key, value|
|
|
80
|
-
t = array_to.find {|k, _v| k == key }
|
|
81
|
-
t[2] = value if !t.nil? && !value.nil?
|
|
82
|
-
end
|
|
83
|
-
end
|
|
84
|
-
|
|
85
|
-
##============================================================##
|
|
86
|
-
## Here we have to do all the translation logic...
|
|
87
|
-
## For the moment we use the OPENAI API, but we can imagine
|
|
88
|
-
## using other translation APIs in the future.
|
|
89
|
-
##============================================================##
|
|
90
|
-
translated_array = translate_with_open_ai(array_to, locale_from, locale_to)
|
|
91
|
-
|
|
92
|
-
##============================================================##
|
|
93
|
-
## Then we have to reformat the output yml file
|
|
94
|
-
##============================================================##
|
|
95
|
-
final_array = translated_array.map do |k, _from, to|
|
|
96
|
-
parsed_to = !to.nil? && to.start_with?("[") && to.end_with?("]") ? JSON.parse(to) : to
|
|
97
|
-
[k, parsed_to]
|
|
98
|
-
end
|
|
99
|
-
final_hash = translatable_hash(final_array)
|
|
100
|
-
|
|
101
|
-
|
|
102
|
-
##============================================================##
|
|
103
|
-
## We write the output file and clean it
|
|
104
|
-
##============================================================##
|
|
105
|
-
File.write(translated_file_path, ImmosquareYaml.dump(final_hash))
|
|
106
|
-
ImmosquareYaml.clean(translated_file_path)
|
|
107
|
-
rescue StandardError => e
|
|
108
|
-
puts(e.message)
|
|
109
|
-
puts(e.backtrace)
|
|
110
|
-
false
|
|
111
|
-
end
|
|
112
|
-
end
|
|
113
|
-
|
|
114
|
-
|
|
115
|
-
private
|
|
116
|
-
|
|
117
|
-
##============================================================##
|
|
118
|
-
## To translatable hash to array
|
|
119
|
-
## opitons are :
|
|
120
|
-
## :format => "string" or "array"
|
|
121
|
-
## :keys_only => true or false
|
|
122
|
-
## {:fr=>{"demo1"=>"demo1", "demo2"=>{"demo2-1"=>"demo2-1"}}}
|
|
123
|
-
## format = "string" and keys_only = false => [["fr.demo1", "demo1"], ["fr.demo2.demo2-1", "demo2-1"]]
|
|
124
|
-
## format = "string" and keys_only = true => ["fr.demo1", "fr.demo2.demo2-1"]
|
|
125
|
-
## format = "array" and keys_only = false => [[["fr", "demo1"], "demo1"], [["fr", "demo2", "demo2-1"], "demo2-1"]]
|
|
126
|
-
## format = "array" and keys_only = true => [["fr", "demo1"], ["fr", "demo2", "demo2-1"]]
|
|
127
|
-
## ============================================================
|
|
128
|
-
def translatable_array(hash, key = nil, result = [], **options)
|
|
129
|
-
options = {
|
|
130
|
-
:format => "string",
|
|
131
|
-
:keys_only => false
|
|
132
|
-
}.merge(options)
|
|
133
|
-
options[:keys_only] = false if ![true, false].include?(options[:keys_only])
|
|
134
|
-
options[:format] = "string" if !["string", "array"].include?(options[:format])
|
|
135
|
-
|
|
136
|
-
|
|
137
|
-
if hash.is_a?(Hash)
|
|
138
|
-
hash.each_key do |k|
|
|
139
|
-
translatable_array(hash[k], "#{key}#{":" if !key.nil?}#{k}", result, **options)
|
|
140
|
-
end
|
|
141
|
-
else
|
|
142
|
-
r2 = options[:format] == "string" ? key.split(":").join(".") : key.split(":")
|
|
143
|
-
result << (options[:keys_only] ? r2 : [r2, hash.is_a?(Array) ? hash.to_json : hash])
|
|
144
|
-
end
|
|
145
|
-
result
|
|
146
|
-
end
|
|
147
|
-
|
|
148
|
-
##============================================================##
|
|
149
|
-
## We can do the inverse of the previous function
|
|
150
|
-
##============================================================##
|
|
151
|
-
def translatable_hash(array)
|
|
152
|
-
data_hash = array.to_h
|
|
153
|
-
final = {}
|
|
154
|
-
data_hash.each do |key, value|
|
|
155
|
-
key_parts = key.split(".")
|
|
156
|
-
leaf = key_parts.pop
|
|
157
|
-
parent = key_parts.inject(final) {|h, k| h[k] ||= {} }
|
|
158
|
-
parent[leaf] = value
|
|
159
|
-
end
|
|
160
|
-
final
|
|
161
|
-
end
|
|
162
|
-
|
|
163
|
-
##============================================================##
|
|
164
|
-
## Translate with OpenAI
|
|
165
|
-
##
|
|
166
|
-
## [
|
|
167
|
-
## ["en.mlsconnect.contact_us", "Nous contacter", "Contact us"],
|
|
168
|
-
## ["en.mlsconnect.description", "Description", nil],
|
|
169
|
-
## ...
|
|
170
|
-
## ]
|
|
171
|
-
##============================================================##
|
|
172
|
-
def translate_with_open_ai(array, from, to)
|
|
173
|
-
##============================================================##
|
|
174
|
-
## https://platform.openai.com/docs/models/
|
|
175
|
-
## https://openai.com/pricing
|
|
176
|
-
##============================================================##
|
|
177
|
-
model_name = ImmosquareYaml.configuration.openai_model
|
|
178
|
-
models = [
|
|
179
|
-
{:name => "gpt-3.5-turbo-0125", :window_tokens => 16_385, :output_tokens => 4096, :input_price_for_1m => 0.50, :output_price_for_1m => 1.50, :group_size => 75},
|
|
180
|
-
{:name => "gpt-4-0125-preview", :window_tokens => 128_000, :output_tokens => 4096, :input_price_for_1m => 10.00, :output_price_for_1m => 30.00, :group_size => 75}
|
|
181
|
-
]
|
|
182
|
-
model = models.find {|m| m[:name] == model_name }
|
|
183
|
-
model = models.find {|m| m[:name] == "gpt-4-0125-preview" } if model.nil?
|
|
184
|
-
|
|
185
|
-
##============================================================##
|
|
186
|
-
## Manage blank values
|
|
187
|
-
##============================================================##
|
|
188
|
-
blank_values = [NOTHING, SPACE, "\"\"", "\"#{SPACE}\""]
|
|
189
|
-
cant_be_translated = "CANNOT-BE-TRANSLATED"
|
|
190
|
-
array = array.map do |key, from, to|
|
|
191
|
-
[key, from, blank_values.include?(from) ? from : to]
|
|
192
|
-
end
|
|
193
|
-
|
|
194
|
-
|
|
195
|
-
##============================================================##
|
|
196
|
-
## we want to send as little data as possible to openAI because
|
|
197
|
-
## we pay for the volume of data sent. So we're going to send. We put
|
|
198
|
-
## a number rather than a string for the translations to be made.
|
|
199
|
-
## --------
|
|
200
|
-
## Remove the translations that have already been made
|
|
201
|
-
##============================================================##
|
|
202
|
-
data_open_ai = array.clone
|
|
203
|
-
data_open_ai = data_open_ai.map.with_index {|(_k, from, to), index| [index, from, to] }
|
|
204
|
-
data_open_ai = data_open_ai.select {|_index, from, to| !from.nil? && to.nil? }
|
|
205
|
-
|
|
206
|
-
##============================================================##
|
|
207
|
-
## Remove quotes surrounding the value if they are present.
|
|
208
|
-
## and remove to to avoid error in translation
|
|
209
|
-
##============================================================##
|
|
210
|
-
data_open_ai = data_open_ai.map do |index, from, _to|
|
|
211
|
-
from = from.to_s
|
|
212
|
-
from = from[1..-2] while (from.start_with?(DOUBLE_QUOTE) && from.end_with?(DOUBLE_QUOTE)) || (from.start_with?(SIMPLE_QUOTE) && from.end_with?(SIMPLE_QUOTE))
|
|
213
|
-
[index, from]
|
|
214
|
-
end
|
|
215
|
-
|
|
216
|
-
return array if data_open_ai.empty?
|
|
217
|
-
|
|
218
|
-
|
|
219
|
-
##============================================================##
|
|
220
|
-
## Call OpenAI API
|
|
221
|
-
##============================================================##
|
|
222
|
-
index = 0
|
|
223
|
-
group_size = model[:group_size]
|
|
224
|
-
from_iso = ISO_639.find_by_code(from).english_name.split(";").first
|
|
225
|
-
to_iso = ISO_639.find_by_code(to).english_name.split(";").first
|
|
226
|
-
ai_resuslts = []
|
|
227
|
-
prompt_system = "You are a translation tool from #{from_iso} to #{to_iso}\n" \
|
|
228
|
-
"The input is an array of pairs, where each pair contains an index and a string to translate, formatted as [index, string_to_translate]\n" \
|
|
229
|
-
"Your task is to create an output ARRAY where each element is a pair consisting of the index and the translated string, formatted as [index, 'string_translated']\n" \
|
|
230
|
-
"If a string_to_translate starts with [ and ends with ], it is considered a special string that should be treated as a JSON object. Otherwise, it's a normal string.\n" \
|
|
231
|
-
"\nRules to respect for JSON objects:\n" \
|
|
232
|
-
"- You need to translate ONLY the values of the JSON object, not the keys. Do not change anything in the format, just translate the values.\n" \
|
|
233
|
-
"- Respect all following rules for normal strings to translate the values\n" \
|
|
234
|
-
"\nRules to respect for normal strings:\n" \
|
|
235
|
-
"- Do not escape apostrophes in translated strings; leave them as they are.\n" \
|
|
236
|
-
"- Special characters, except apostrophes, that need to be escaped in translated strings should be escaped using a single backslash (\\), not double (\\\\).\n" \
|
|
237
|
-
"- If a string cannot be translated use the string '#{cant_be_translated}' translated as the translation value witouth quote (simple or double) quote, just the string\n" \
|
|
238
|
-
"- If you dont know the correct translatation use the #{cant_be_translated} strategy of the preceding point\n" \
|
|
239
|
-
"- Use only double quotes (\") to enclose translated strings and avoid using single quotes (').\n" \
|
|
240
|
-
"- Your output must ONLY be an array with the same number of pairs as the input, without any additional text or explanation. DO NOT COMMENT!\n" \
|
|
241
|
-
"- You need to check that the globle array is correctly closed at the end of the response. (the response must therefore end with ]] to to be consistent)"
|
|
242
|
-
prompt_init = "Please proceed with translating the following array:"
|
|
243
|
-
headers = {
|
|
244
|
-
"Content-Type" => "application/json",
|
|
245
|
-
"Authorization" => "Bearer #{ImmosquareYaml.configuration.openai_api_key}"
|
|
246
|
-
}
|
|
247
|
-
|
|
248
|
-
|
|
249
|
-
##============================================================##
|
|
250
|
-
## Estimate the number of window_tokens
|
|
251
|
-
## https://platform.openai.com/tokenizer
|
|
252
|
-
## English: 75 words => 100 tokens
|
|
253
|
-
## French : 55 words => 100 tokens
|
|
254
|
-
## -----------------
|
|
255
|
-
## For each array value we add 5 tokens for the array format.
|
|
256
|
-
## [1, "my_word"],
|
|
257
|
-
## [ => first token
|
|
258
|
-
## 2 => second token
|
|
259
|
-
## , => third token
|
|
260
|
-
## " => fourth token
|
|
261
|
-
## ]" => fifth token
|
|
262
|
-
## -----------------
|
|
263
|
-
# data_open_ai.inspect.size => to get the total number of characters in the array
|
|
264
|
-
## with the array structure [""],
|
|
265
|
-
##============================================================##
|
|
266
|
-
estimation_for_100_tokens = from == "fr" ? 55 : 75
|
|
267
|
-
prompt_tokens_estimation = (((prompt_system.split.size + prompt_init.split.size + data_open_ai.map {|_index, from| from.split.size }.sum) / estimation_for_100_tokens * 100.0) + (data_open_ai.size * 5)).round
|
|
268
|
-
split_array = (prompt_tokens_estimation / model[:window_tokens].to_f).ceil
|
|
269
|
-
slice_size = (data_open_ai.size / split_array.to_f).round
|
|
270
|
-
data_open_ai_sliced = data_open_ai.each_slice(slice_size).to_a
|
|
271
|
-
|
|
272
|
-
|
|
273
|
-
##============================================================##
|
|
274
|
-
## Now each slice of the array should no be more than window_tokens
|
|
275
|
-
## of the model.... We can now translate each slice.
|
|
276
|
-
## ---------------------------------
|
|
277
|
-
## Normally we could send the whole slice at once and tell the api to continue if its response is not tarnished...
|
|
278
|
-
## But it should manage if a word is cut etc...
|
|
279
|
-
## For the moment we cut it into small group for which we are sure not to exceed the limit
|
|
280
|
-
##============================================================##
|
|
281
|
-
puts("fields to translate from #{from_iso} (#{from}) to #{to_iso} (#{to}) : #{data_open_ai.size}#{" by group of #{group_size}" if data_open_ai.size > group_size}")
|
|
282
|
-
while index < data_open_ai.size
|
|
283
|
-
data_group = data_open_ai[index, group_size]
|
|
284
|
-
|
|
285
|
-
|
|
286
|
-
begin
|
|
287
|
-
puts("call OPENAI Api (with model #{model[:name]}) #{" for #{data_group.size} fields (#{index}-#{index + data_group.size})" if data_open_ai.size > group_size}")
|
|
288
|
-
prompt = "#{prompt_init}:\n\n#{data_group.inspect}\n\n"
|
|
289
|
-
body = {
|
|
290
|
-
:model => model[:name],
|
|
291
|
-
:messages => [
|
|
292
|
-
{:role => "system", :content => prompt_system},
|
|
293
|
-
{:role => "user", :content => prompt}
|
|
294
|
-
],
|
|
295
|
-
:temperature => 0.0
|
|
296
|
-
}
|
|
297
|
-
t0 = Time.now
|
|
298
|
-
call = HTTParty.post("https://api.openai.com/v1/chat/completions", :body => body.to_json, :headers => headers, :timeout => 500)
|
|
299
|
-
|
|
300
|
-
puts("responded in #{(Time.now - t0).round(2)} seconds")
|
|
301
|
-
raise(call["error"]["message"]) if call.code != 200
|
|
302
|
-
|
|
303
|
-
|
|
304
|
-
##============================================================##
|
|
305
|
-
## We check that the result is complete
|
|
306
|
-
##============================================================##
|
|
307
|
-
response = JSON.parse(call.body)
|
|
308
|
-
choice = response["choices"][0]
|
|
309
|
-
raise("Result is not complete") if choice["finish_reason"] != "stop"
|
|
310
|
-
|
|
311
|
-
|
|
312
|
-
##============================================================##
|
|
313
|
-
## We calculate the estimate price of the call
|
|
314
|
-
##============================================================##
|
|
315
|
-
input_price = response["usage"]["prompt_tokens"] * (model[:input_price_for_1m] / 1_000_000)
|
|
316
|
-
output_price = response["usage"]["completion_tokens"] * (model[:output_price_for_1m] / 1_000_000)
|
|
317
|
-
price = input_price + output_price
|
|
318
|
-
puts("Estimate price => #{input_price.round(3)} + #{output_price.round(3)} = #{price.round(3)} USD")
|
|
319
|
-
|
|
320
|
-
##============================================================##
|
|
321
|
-
## We check that the result is an array
|
|
322
|
-
##============================================================##
|
|
323
|
-
content = eval(choice["message"]["content"])
|
|
324
|
-
raise("Is not an array") if !content.is_a?(Array)
|
|
325
|
-
|
|
326
|
-
##============================================================##
|
|
327
|
-
## We save the result
|
|
328
|
-
##============================================================##
|
|
329
|
-
content.each do |index, translation|
|
|
330
|
-
ai_resuslts << [index, translation == cant_be_translated ? nil : translation]
|
|
331
|
-
end
|
|
332
|
-
rescue StandardError => e
|
|
333
|
-
puts("error OPEN AI API => #{e.message}")
|
|
334
|
-
puts(e.message)
|
|
335
|
-
puts(e.backtrace)
|
|
336
|
-
end
|
|
337
|
-
index += group_size
|
|
338
|
-
end
|
|
339
|
-
|
|
340
|
-
|
|
341
|
-
##============================================================##
|
|
342
|
-
## We put the translations in the original array
|
|
343
|
-
##============================================================##
|
|
344
|
-
ai_resuslts.each do |index, translation|
|
|
345
|
-
begin
|
|
346
|
-
array[index.to_i][2] = translation
|
|
347
|
-
rescue StandardError => e
|
|
348
|
-
puts(e.message)
|
|
349
|
-
end
|
|
350
|
-
end
|
|
351
|
-
|
|
352
|
-
##============================================================##
|
|
353
|
-
## We return the modified array
|
|
354
|
-
##============================================================##
|
|
355
|
-
array.map.with_index do |(k, from, to), index|
|
|
356
|
-
from = from.to_s
|
|
357
|
-
to = "#{DOUBLE_QUOTE}#{to}#{DOUBLE_QUOTE}" if ai_resuslts.find {|i, _t| i == index } && ((from.start_with?(DOUBLE_QUOTE) && from.end_with?(DOUBLE_QUOTE)) || (from.start_with?(SIMPLE_QUOTE) && from.end_with?(SIMPLE_QUOTE)))
|
|
358
|
-
[k, from, to]
|
|
359
|
-
end
|
|
360
|
-
end
|
|
361
|
-
|
|
362
|
-
|
|
363
|
-
end
|
|
364
|
-
end
|
|
365
|
-
end
|