jekyll-chatgpt-translate 0.0.16 → 0.0.18

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f3a1fa14a50caf08c780825e889b16c100e4afcd4042db089e49b33d01eece4a
-  data.tar.gz: 87cf1124880695e6efad136115b7dfd9e33a629e03f154adc5f7325f7f6e43e9
+  metadata.gz: 1f593e5927f89dc9fe25dcbe28cc61ed938b3fdc42aee742d9ec31ebc86417fe
+  data.tar.gz: c5486a9adaf6e838bd016ea01175169786dbdd1807601cf922b0acc065c0a337
 SHA512:
-  metadata.gz: fb6b987ce7934d9294ba69f73d4d23bb3c74d6394cc5537ed956ee0b54b280f0ae79e5d12c9974ef75402f0586422700727d02304e44c4d7f5c74090c7b7c03c
-  data.tar.gz: 597a0397823b9d91a65ce1b8269f3117f37152b2ab9c8eafd501a9c4c0aaeec9ecd70022460616c2ae8c212605fb2fd1df347ab07b9accc5323e5931d7cfd037
+  metadata.gz: 794664182ebba232506ab97b026d02a7060ee369fb16756308d43362df24aed862703836fb0af142c07240fd1eeeeeb9305e43804ed5b34300ce2dff1c30a03c
+  data.tar.gz: f1db90b466e6ea441519ee35351a2bf62b9182ed772ab020b5cd4692fc8932a22115fb994089b1cb70724c92a4344865810312b7a58fe1fe666ba0e5c308aaeb

data/README.md

@@ -7,7 +7,7 @@ If you have a [Jekyll](https://jekyllrb.com/) static site, this plugin may help
 translate its pages to another language, through [ChatGPT](https://chat.openai.com/). See how it 
 works for [my blog](https://github.com/yegor256/ru.yegor256.com), 
 for example [this page](https://ru.yegor256.com/2023-08-13-dictators.html) is translated to 
-[English]().
+[English](https://ru.yegor256.com/english/2023-08-13-dictators.html).
 
 Install it first:
 
@@ -18,13 +18,16 @@ gem install jekyll-chatgpt-translate
 Then, add this to `_config.yml`:
 
 ```yaml
+plugins:
+  - ... your other plugins here ...
+  - jekyll-chatgpt-translate
 chatgpt-translate:
   model: gpt-3.5-turbo
   source: en
   layout: translated
   targets: 
     - 
-      language: cn
+      language: zh
       permalink: :year-:month-:day-:slug-chinese.html
       layout: chinese-translated
     - 
@@ -32,7 +35,7 @@ chatgpt-translate:
       permalink: :year-:month-:day-:title-french.html
 ```
 
-Here, the source language is English (`en`), the target one is Chinese (`cn`),
+Here, the source language is English (`en`), the target one is Chinese (`zh`),
 the layout is `_layout/translated.html` (you must have this file).
 
 OpenAI API KEY must be set in `OPENAI_API_KEY` environment variable, otherwise
@@ -40,10 +43,15 @@ the plugin will not do any translation and won't generate translated pages.
 You can get your key [here](https://help.openai.com/en/articles/4936850-where-do-i-find-my-secret-api-key).
 
 Inside the original page you can use `{{ page.translated-XX-url }}` in order to render the URL
-of the translated page, where `XX` is the ISO-839-1 code of the target language..
+of the translated page, where `XX` is the [ISO-639-1](https://en.wikipedia.org/wiki/List_of_ISO_639-1_codes) 
+code of the target language.
 Inside the translated page you can use `{{ page.translated-original-url }}` in order
-to get the URL of the page that was translated. You can also use `{{ page.chatgpt-model }}`
+to get the URL of the page that was translated. 
+
+You can also use `{{ page.chatgpt-model }}`
 inside both the original page and the translated one, to refer to the model of ChatGPT.
+The presence of this attribute in the `{{ page }}` means that the
+page was translated or the translated HTML was downloaded and placed into the `_site` directory.
 
 ## Options
 
@@ -56,6 +64,11 @@ Full list of options available to specify in `_config.yml`:
 
   * `source` (optional) — is the ISO-839-1 code of the source language.
 
+  * `no_download` (optional) — if this attribute is present, the plugin won't try
+    to find HTML versions of translated pages in the Internet and won't try to 
+    download them and place into the `_site` directory. Thus, your entire site
+    will have to be re-translated on every build (might be very ineffective if the site is big!)
+
   * `layout` (optional) — is name of the file in `_layouts` directory, without the extension. 
 This layout will be specified for the pages generated by this plugin.
 

data/features/cli.feature

@@ -13,7 +13,7 @@ Feature: Simple site building
       layout: translated
       targets:
         -
-          language: cn
+          language: zh
           permalink: :year-:month-:day-:slug-chinese.html
           layout: chinese-translated
         -
@@ -22,7 +22,7 @@ Feature: Simple site building
     """
     And I have a "_layouts/default.html" file with content:
     """
-    The Chinese: {{ page.translated-cn-url }}
+    The Chinese: {{ page.translated-zh-url }}
     The French: {{ page.translated-fr-url }}
     {{ content }}
     """
@@ -40,8 +40,9 @@ Feature: Simple site building
     Hello, world!
     """
     Then I build Jekyll site
-    And File "_chatgpt-translated/cn/2023-01-01-hello-cn.md" exists
-    And File "_chatgpt-translated/cn/2023-01-01-hello-cn.md" contains "/2023-01-01-hello-chinese.html"
+    And File "_chatgpt-translated/zh/2023-01-01-hello-zh.md" exists
+    And File "_chatgpt-translated/zh/2023-01-01-hello-zh.md" contains "/2023-01-01-hello-chinese.html"
+    And File "_chatgpt-translated/zh/2023-01-01-hello-zh.md" contains "translated-language: \"zh\""
     And File "_site/2023/01/01/hello.html" exists
     And File "_site/2023/01/01/hello.html" contains "The Chinese: /2023-01-01-hello-chinese.html"
     And File "_site/2023-01-01-hello-chinese.html" exists

data/features/gem_package.feature

@@ -3,7 +3,7 @@ Feature: Gem Package
   package the Gem into .gem file
 
   Scenario: Gem can be packaged
-    When it is Unix
+    When It is Unix
     Given I have a "execs.rb" file with content:
     """
     #!/usr/bin/env ruby

data/features/step_definitions/steps.rb

@@ -40,39 +40,35 @@ Given(/^I have a "([^"]*)" file with content:$/) do |file, text|
   File.write(file, text.gsub('\\xFF', 0xFF.chr))
 end
 
-When(/^I build Jekyll site$/) do
+When('I build Jekyll site') do
   @stdout = `jekyll build`
   @exitstatus = $CHILD_STATUS.exitstatus
 end
 
-Then(/^Stdout contains "([^"]*)"$/) do |txt|
-  raise "STDOUT doesn't contain '#{txt}':\n#{@stdout}" unless @stdout.include?(txt)
+Then('Stdout contains {string}') do |string|
+  raise "STDOUT doesn't contain '#{string}':\n#{@stdout}" unless @stdout.include?(string)
 end
 
-Then(/^File "([^"]*)" exists$/) do |name|
-  raise "The file \"#{name}\" is absent:\n#{`tree -s`}" unless File.exist?(name)
+Then('File {string} exists') do |string|
+  raise "The file \"#{string}\" is absent:\n#{`tree -s`}" unless File.exist?(string)
 end
 
-Then(/^File "([^"]*)" contains "([^"]*)"$/) do |name, text|
-  raise "The file \"#{name}\" is absent" unless File.exist?(name)
-  content = File.read(name)
-  raise "The file \"#{name}\" doesn't contain \"#{text}\":\n#{content}" unless content.include?(text)
+Then('File {string} contains {string}') do |string, string2|
+  raise "The file \"#{string}\" is absent" unless File.exist?(string)
+  content = File.read(string)
+  raise "The file \"#{string}\" doesn't contain \"#{string2}\":\n#{content}" unless content.include?(string2)
 end
 
-Then(/^Stdout is empty$/) do
-  raise "STDOUT is not empty:\n#{@stdout}" unless @stdout == ''
-end
-
-Then(/^Exit code is zero$/) do
+Then('Exit code is zero') do
   raise "Non-zero exit #{@exitstatus}:\n#{@stdout}" unless @exitstatus.zero?
 end
 
-Then(/^Exit code is not zero$/) do
+Then('Exit code is not zero') do
   raise 'Zero exit code' if @exitstatus.zero?
 end
 
-When(/^I run bash with "([^"]*)"$/) do |text|
-  @stdout = `#{text}`
+When('I run bash with {string}') do |string|
+  @stdout = `#{string}`
   @exitstatus = $CHILD_STATUS.exitstatus
 end
 
@@ -81,14 +77,14 @@ When(/^I run bash with:$/) do |text|
   @exitstatus = $CHILD_STATUS.exitstatus
 end
 
-When(/^I copy this gem into temp dir$/) do
+When('I copy this gem into temp dir') do
   FileUtils.copy_entry(@cwd, File.join(@dir, 'jekyll-chatgpt-translate'))
 end
 
-Given(/^It is Unix$/) do
+Given('It is Unix') do
   pending if Gem.win_platform?
 end
 
-Given(/^It is Windows$/) do
+Given('It is Windows') do
   pending unless Gem.win_platform?
 end

data/jekyll-chatgpt-translate.gemspec

@@ -28,7 +28,7 @@ Gem::Specification.new do |s|
   s.required_rubygems_version = Gem::Requirement.new('>= 0') if s.respond_to? :required_rubygems_version=
   s.required_ruby_version = '>= 2.6'
   s.name = 'jekyll-chatgpt-translate'
-  s.version = '0.0.16'
+  s.version = '0.0.18'
   s.license = 'MIT'
   s.summary = 'Translate Jekyll Pages Through ChatGPT'
   s.description = [

data/lib/jekyll-chatgpt-translate/generator.rb

@@ -54,7 +54,8 @@ class GptTranslate::Generator < Jekyll::Generator
     version = config['version'] || GptTranslate::VERSION
     threshold = config['threshold'] || 1024
     start = Time.now
-    total = 0
+    translated = 0
+    copied = 0
     model = config['model'] || 'gpt-3.5-turbo'
     marker = "Translated by ChatGPT #{model}/#{version}"
     site.posts.docs.shuffle.each do |doc|
@@ -63,48 +64,49 @@ class GptTranslate::Generator < Jekyll::Generator
         link = GptTranslate::Permalink.new(doc, target['permalink']).to_path
         lang = target['language']
         raise 'Language must be defined for each target' if target.nil?
-        if total >= threshold
-          Jekyll.logger.info("Already generated #{total} pages, that's enough for today")
-          break
-        end
         path = File.join(home, lang, doc.basename.gsub(/\.md$/, "-#{lang}.md"))
         FileUtils.mkdir_p(File.dirname(path))
         File.write(path, '') # in order to surpress warnings in Page ctor
         dest = Jekyll::Page.new(site, site.source, File.dirname(path), File.basename(path)).destination(site.dest)
+        if config['no_download'].nil? && GptTranslate::Ping.new(site, link).found?(dest, version)
+          copied += 1
+        elsif translated >= threshold
+          next
+        else
+          gpt = GptTranslate::ChatGPT.new(
+            key,
+            model,
+            config['source'] || 'en',
+            lang
+          )
+          foreign = gpt.translate(plain)
+          File.write(
+            path,
+            [
+              '---',
+              "layout: #{target['layout'] || layout}",
+              "title: #{doc['title'].to_json}",
+              "description: #{doc['description'].to_json}",
+              "permalink: #{link.to_json}",
+              "translated-original-url: #{doc.url.to_json}",
+              "translated-language: #{lang.to_json}",
+              "chatgpt-model: #{model.to_json}",
+              '---',
+              '',
+              foreign,
+              '',
+              "#{marker} on #{Time.now.strftime('%d/%m/%Y %H:%M')}\n{: .jekyll-chatgpt-translate}"
+            ].join("\n")
+          )
+          site.pages << Jekyll::Page.new(site, site.source, File.dirname(path), File.basename(path))
+          translated += 1
+          Jekyll.logger.info("Translated via ChatGPT: #{path} (#{File.size(path)} bytes)")
+        end
         doc.data["translated-#{lang}-url"] = link
         doc.data['chatgpt-model'] = model
-        next if GptTranslate::Ping.new(site, link).found?(dest, version)
-        gpt = GptTranslate::ChatGPT.new(
-          key,
-          model,
-          config['source'] || 'en',
-          lang
-        )
-        translated = gpt.translate(plain)
-        File.write(
-          path,
-          [
-            '---',
-            "layout: #{target['layout'] || layout}",
-            "title: #{doc['title'].to_json}",
-            "description: #{doc['description'].to_json}",
-            "permalink: #{link.to_json}",
-            "translated-original-url: #{doc.url.to_json}",
-            "chatgpt-model: #{model.to_json}",
-            '---',
-            '',
-            translated,
-            '',
-            "#{marker} on #{Time.now.strftime('%d/%m/%Y %H:%M')}\n{: .jekyll-chatgpt-translate}"
-          ].join("\n")
-        )
-        site.pages << Jekyll::Page.new(site, site.source, File.dirname(path), File.basename(path))
-        total += 1
-        Jekyll.logger.info("Translated via ChatGPT: #{path}")
       end
-      break if total >= threshold
     end
-    Jekyll.logger.info("#{total} pages translated in #{(Time.now - start).round(2)}s")
+    Jekyll.logger.info("#{translated} pages translated and #{copied} pages copied in #{(Time.now - start).round(2)}s")
   end
 
   private

data/lib/jekyll-chatgpt-translate/plain.rb

@@ -74,6 +74,8 @@ class GptTranslate::Plain
     def codespan(content)
       if content.start_with?("\n")
         "```#{content}```"
+      elsif content.end_with?("\n")
+        "```\n#{content.split("\n", 2)[1]}```"
       else
         content
       end

data/lib/jekyll-chatgpt-translate/version.rb

@@ -23,5 +23,5 @@
 # SOFTWARE.
 
 module GptTranslate
-  VERSION = '0.0.16'
+  VERSION = '0.0.18'
 end

data/test/test_generator.rb

@@ -33,19 +33,16 @@ require_relative '../lib/jekyll-chatgpt-translate/generator'
 # License:: MIT
 class GptTranslate::GeneratorTest < Minitest::Test
   class FakeSite
-    attr_reader :config
+    attr_reader :config, :pages
 
-    def initialize(config, doc)
+    def initialize(config, docs)
       @config = config
-      @doc = doc
+      @docs = docs
+      @pages = []
     end
 
     def posts
-      FakePosts.new(@doc)
-    end
-
-    def pages
-      []
+      FakePosts.new(@docs)
     end
 
     def permalink_style
@@ -65,7 +62,7 @@ class GptTranslate::GeneratorTest < Minitest::Test
     end
 
     def dest
-      ''
+      File.dirname(@docs[0])
     end
 
     def in_theme_dir(base, _foo = nil, _bar = nil)
@@ -78,28 +75,23 @@ class GptTranslate::GeneratorTest < Minitest::Test
   end
 
   class FakeDocument
+    attr_reader :data
+
     def initialize(path)
       @path = path
+      @data = { 'date' => Time.now, 'title' => 'Hello!' }
     end
 
     def content
       'Hello, world!'
     end
 
-    def data
-      {}
+    def []=(key, value)
+      @data[key] = value
     end
 
-    def []=(key, value); end
-
     def [](key)
-      if key == 'date'
-        Time.now
-      elsif key == 'title'
-        'Hello!'
-      else
-        ''
-      end
+      @data[key] || ''
     end
 
     def relative_path
@@ -118,12 +110,12 @@ class GptTranslate::GeneratorTest < Minitest::Test
   class FakePosts
     attr_reader :config
 
-    def initialize(doc)
-      @doc = doc
+    def initialize(docs)
+      @docs = docs
     end
 
     def docs
-      [FakeDocument.new(@doc)]
+      @docs.map { |d| FakeDocument.new(d) }
     end
   end
 
@@ -137,18 +129,48 @@ class GptTranslate::GeneratorTest < Minitest::Test
           'chatgpt-translate' => {
             'targets' => [
               {
-                'language' => 'cn',
+                'language' => 'zh',
                 'layout' => 'chinese',
                 'permalink' => ':slug.html'
               }
             ]
           }
         },
-        FakeDocument.new({})
+        [post]
+      )
+      gen = GptTranslate::Generator.new
+      stub_request(:get, 'https://www.yegor256.com/.html').to_return(body: '')
+      gen.generate(site)
+      assert_equal(1, site.pages.count)
+    end
+  end
+
+  def test_threshold_stops
+    Dir.mktmpdir do |home|
+      post = File.join(home, '2023-01-01-hello.md')
+      File.write(post, "---\ntitle: Hello\n---\n\nHello, world!")
+      site = FakeSite.new(
+        {
+          'chatgpt-translate' => {
+            'threshold' => 1,
+            'targets' => [
+              {
+                'language' => 'zh',
+                'permalink' => ':slug.html'
+              },
+              {
+                'language' => 'fr',
+                'permalink' => ':year/:slug.html'
+              }
+            ]
+          }
+        },
+        [post, post]
       )
       gen = GptTranslate::Generator.new
       stub_request(:get, 'https://www.yegor256.com/.html').to_return(body: '')
       gen.generate(site)
+      assert_equal(1, site.pages.count)
     end
   end
 end

data/test/test_plain.rb

@@ -96,6 +96,10 @@ class GptTranslate::PlainTest < Minitest::Test
       "```\nHello\n```",
       GptTranslate::Plain.new("```\nHello\n```").to_s
     )
+    assert_equal(
+      "```\nprint('hi!')\n```",
+      GptTranslate::Plain.new("```java\nprint('hi!')\n```").to_s
+    )
   end
 
   def test_liquid_tags

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: jekyll-chatgpt-translate
 version: !ruby/object:Gem::Version
-  version: 0.0.16
+  version: 0.0.18
 platform: ruby
 authors:
 - Yegor Bugayenko