jekyll-chatgpt-translate 0.0.12 → 0.0.14

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: ed6617e3b8bf46db264f484e1286ea1f91295f4fb47512cb983a6fc59c7922df
-  data.tar.gz: a0f6c744cd4bf2b2558d7b5568f1ca961a3f854e5264b9d64bb8b24249f03e3a
+  metadata.gz: 4b8a34c60ab2378b8dc26ab4b189ff3fa168cbb864c80db92a854d68b58def59
+  data.tar.gz: 3654fe32f201e8a155cd5f154129c24350333b6f2459cd86bb511c0acc22c0c1
 SHA512:
-  metadata.gz: 2d58f41b76840e524e9f1d72c2a4816b33d9e57ad6b207f54baa226abf68ae6cca848cb8cce80cf1f4502a3ad9b52871db7f7708463bf4101fdbf8c21e5f99bb
-  data.tar.gz: f85b41ad75176c9989a8a3715fd84485b3a2fcbe03e6cca879b7a201ac4b5745dda0aa16695f26cbaad1ec901b84c6837f983f7a04cee94aa26752d13ef41593
+  metadata.gz: ea02f9faba8ffedaa3e25bd428bde6aaba7a7a484bda745cc270a9f97f1a40211b776cff7d5a8c5d297854f107ba311b10dd5885bb6eff8de353f4a8aa9778d6
+  data.tar.gz: 6f3cc93df6fff8290f3a6d45bc980a9c5b273154a98ab7c18ddff1571c55dd8c2796e9ef965c3a0df72ff175c42ed54b72f4b4ecb5d2d8bb21864ae3b4aa0c7c

data/.gitignore

@@ -2,4 +2,5 @@ Gemfile.lock
 *.gem
 .bundle/
 .DS_Store
-coverage/
+coverage/
+_chatgpt-translated/

data/features/cli.feature

@@ -40,11 +40,12 @@ Feature: Simple site building
     Hello, world!
     """
     Then I build Jekyll site
-    Then File "_chatgpt-translated/cn/2023-01-01-hello-cn.md" exists
-    Then File "_site/2023/01/01/hello.html" exists
-    Then File "_site/2023/01/01/hello.html" contains "The Chinese: /2023-01-01-hello-chinese.html"
-    Then File "_site/2023-01-01-hello-chinese.html" exists
-    Then File "_site/2023-01-01-hello-chinese.html" contains "The original: /2023/01/01/hello.html"
-    Then File "_site/2023/hello-french.html" exists
+    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 "_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
+    And File "_site/2023-01-01-hello-chinese.html" contains "The original: /2023/01/01/hello.html"
+    And File "_site/2023/hello-french.html" exists
     And Exit code is zero
 

data/features/step_definitions/steps.rb

@@ -50,10 +50,11 @@ Then(/^Stdout contains "([^"]*)"$/) do |txt|
 end
 
 Then(/^File "([^"]*)" exists$/) do |name|
-  raise "The file \"#{name}\" is absent:\n#{`tree`}" unless File.exist?(name)
+  raise "The file \"#{name}\" is absent:\n#{`tree -s`}" unless File.exist?(name)
 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)
 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 = '0.0.12'
+  s.version = '0.0.14'
   s.license = 'MIT'
   s.summary = 'Translate Jekyll Pages Through ChatGPT'
   s.description = [

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

@@ -92,17 +92,12 @@ through #{@model} in #{(Time.now - start).round(2)}s")
   end
 
   def prompt
-    if (@source == 'ru') && (@target == 'en')
-      'Пожалуйста, переведи этот параграф на английский язык'
-    elsif (@source == 'en') && (@target == 'ru')
-      'Please, translate this paragraph to Russian'
-    else
-      [
-        'Please, translate this paragraph from',
-        ISO_639.find_by_code(@source),
-        'to',
-        ISO_639.find_by_code(@target)
-      ].join(' ')
-    end
+    [
+      'Please, translate this paragraph from',
+      ISO_639.find_by_code(@source),
+      'to',
+      ISO_639.find_by_code(@target),
+      ', don\'t change proper nouns'
+    ].join(' ')
   end
 end

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

@@ -23,6 +23,7 @@
 # SOFTWARE.
 
 require 'jekyll'
+require 'json'
 require_relative 'chatgpt'
 require_relative 'permalink'
 require_relative 'ping'
@@ -60,13 +61,19 @@ class GptTranslate::Generator < Jekyll::Generator
       plain = GptTranslate::Plain.new(doc.content).to_s
       config['targets'].each do |target|
         link = GptTranslate::Permalink.new(doc, target['permalink']).to_path
-        next if GptTranslate::Ping.new(site, link).found?(doc.path, version)
         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$/, "-#{lang}.md"))
+        FileUtils.mkdir_p(File.dirname(path))
+        File.write(path, '') # in order to surpress warnings in Page ctor
+        dest = Jekyll::Page.new(site, site.source, File.dirname(path), File.basename(path)).destination(site.dest)
+        doc.data["translated-#{lang}-url"] = link
+        doc.data['chatgpt-model'] = model
+        next if GptTranslate::Ping.new(site, link).found?(dest, version)
         gpt = GptTranslate::ChatGPT.new(
           key,
           model,
@@ -74,17 +81,16 @@ class GptTranslate::Generator < Jekyll::Generator
           lang
         )
         translated = gpt.translate(plain)
-        path = File.join(home, lang, doc.basename.gsub(/\.md$/, "-#{lang}.md"))
-        FileUtils.mkdir_p(File.dirname(path))
         File.write(
           path,
           [
             '---',
             "layout: #{target['layout'] || layout}",
-            "title: #{doc.data['title']}",
-            "permalink: #{link}",
-            "translated-original-url: #{doc.url}",
-            "chatgpt-model: #{model}",
+            "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,
@@ -92,8 +98,6 @@ class GptTranslate::Generator < Jekyll::Generator
             "#{marker}\n{: .jekyll-chatgpt-translate}"
           ].join("\n")
         )
-        doc.data["translated-#{lang}-url"] = link
-        doc.data['chatgpt-model'] = model
         site.pages << Jekyll::Page.new(site, site.source, File.dirname(path), File.basename(path))
         total += 1
         Jekyll.logger.info("Translated via ChatGPT: #{path}")

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

@@ -51,21 +51,26 @@ class GptTranslate::Ping
     home = @site.config['url']
     return false if home.nil?
     uri = Iri.new(home).path(@path)
-    before = Net::HTTP.get_response(URI(uri.to_s))
-    if before.is_a?(Net::HTTPSuccess)
-      html = before.body
-      if html.include?(marker)
-        Jekyll.logger.info("No need to translate, the page exists at \
-#{uri} (#{html.split.count} words), saved to #{file}")
-        FileUtils.mkdir_p(File.dirname(file))
-        File.write(file, html)
-        return true
+    begin
+      before = Net::HTTP.get_response(URI(uri.to_s))
+      if before.is_a?(Net::HTTPSuccess)
+        html = before.body
+        if html.include?(marker)
+          Jekyll.logger.info("No need to translate, the page exists at \
+\"#{uri}\" (#{html.split.count} words), saved to \"#{file}\"")
+          FileUtils.mkdir_p(File.dirname(file))
+          File.write(file, html)
+          return true
+        end
+        Jekyll.logger.info("Re-translation required for \"#{uri}\"")
+      else
+        Jekyll.logger.info("The page is absent, will translate \"#{uri}\"")
       end
-      Jekyll.logger.info("Re-translation required for #{uri}")
-    else
-      Jekyll.logger.info("The page is absent, will translate: #{uri}")
+      Jekyll.logger.debug("GET \"#{uri}\": #{before.code}")
+    rescue StandardError => e
+      Jekyll.logger.debug("Failed to ping \"#{uri}\": #{e.message}")
+      Jekyll.logger.info("The page is absent: \"#{uri}\"")
     end
-    Jekyll.logger.debug("GET #{uri}: #{before.code}")
     false
   end
 end

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

@@ -38,28 +38,31 @@ class GptTranslate::Plain
   end
 
   def to_s
-    @markdown.split(/\n{2,}/).compact.map do |par|
-      par.gsub!("\n", ' ')
-      par.gsub!(/\s{2,}/, ' ')
+    # To turn compact lists into proper lists
+    @markdown.gsub(/([^\n])\n(\s*\*)/, "\\1\n\n\\2").split(/\n{2,}/).compact.map do |par|
+      # par.gsub!("\n", ' ')
+      par.gsub!("\t", ' ')
+      par.gsub!(/ {2,}/, ' ')
       # Liquid tags are removed, but this implementation is primitive
       # Seehttps://stackoverflow.com/questions/
       par.gsub!(/{{[^}]+}}/, '')
-      par.gsub!(/{%[^%]+%}/, '')
+      par.gsub!(/{%.+?%}/, '')
+      par.gsub!(/<!--.+?-->/m, '')
       par.strip!
       Redcarpet::Markdown.new(Strip).render(par)
     end.join("\n\n").gsub(/\n{2,}/, "\n\n").strip
   end
 
-  # To ignore/remove Liquid tags.
-  class NullDrop < Liquid::Drop
-    def method_missing(*)
-      nil
-    end
+  # # To ignore/remove Liquid tags.
+  # class NullDrop < Liquid::Drop
+  #   def method_missing(*)
+  #     nil
+  #   end
 
-    def respond_to_missing?(*)
-      true
-    end
-  end
+  #   def respond_to_missing?(*)
+  #     true
+  #   end
+  # end
 
   # Markdown to pain text.
   class Strip < Redcarpet::Render::Base

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

@@ -23,5 +23,5 @@
 # SOFTWARE.
 
 module GptTranslate
-  VERSION = '0.0.12'
+  VERSION = '0.0.14'
 end

data/test/test_generator.rb

@@ -23,6 +23,8 @@
 # SOFTWARE.
 
 require 'minitest/autorun'
+require 'tmpdir'
+require 'webmock/minitest'
 require_relative '../lib/jekyll-chatgpt-translate/generator'
 
 # Generator test.
@@ -30,7 +32,123 @@ require_relative '../lib/jekyll-chatgpt-translate/generator'
 # Copyright:: Copyright (c) 2023 Yegor Bugayenko
 # License:: MIT
 class GptTranslate::GeneratorTest < Minitest::Test
+  class FakeSite
+    attr_reader :config
+
+    def initialize(config, doc)
+      @config = config
+      @doc = doc
+    end
+
+    def posts
+      FakePosts.new(@doc)
+    end
+
+    def pages
+      []
+    end
+
+    def permalink_style
+      ''
+    end
+
+    def frontmatter_defaults
+      Jekyll::FrontmatterDefaults.new(self)
+    end
+
+    def converters
+      [Jekyll::Converters::Markdown.new({ 'markdown_ext' => 'md' })]
+    end
+
+    def source
+      ''
+    end
+
+    def dest
+      ''
+    end
+
+    def in_theme_dir(base, _foo = nil, _bar = nil)
+      base
+    end
+
+    def in_dest_dir(*paths)
+      paths[0].dup
+    end
+  end
+
+  class FakeDocument
+    def initialize(path)
+      @path = path
+    end
+
+    def content
+      'Hello, world!'
+    end
+
+    def data
+      {}
+    end
+
+    def []=(key, value); end
+
+    def [](key)
+      if key == 'date'
+        Time.now
+      elsif key == 'title'
+        'Hello!'
+      else
+        ''
+      end
+    end
+
+    def relative_path
+      @path
+    end
+
+    def url
+      '2023-01-01-hello.html'
+    end
+
+    def basename
+      '2023-01-01-hello.md'
+    end
+  end
+
+  class FakePosts
+    attr_reader :config
+
+    def initialize(doc)
+      @doc = doc
+    end
+
+    def docs
+      [FakeDocument.new(@doc)]
+    end
+  end
+
   def test_simple_scenario
-    # assert(stdout.include?('DEBUG'))
+    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(
+        {
+          'url' => 'https://www.yegor256.com/',
+          'chatgpt-translate' => {
+            'targets' => [
+              {
+                'language' => 'cn',
+                'layout' => 'chinese',
+                'permalink' => ':slug.html'
+              }
+            ]
+          }
+        },
+        FakeDocument.new({})
+      )
+      gen = GptTranslate::Generator.new
+      stub_request(:get, 'https://www.yegor256.com/.html').to_return(body: '')
+      gen.generate(site)
+    end
   end
 end

data/test/test_ping.rb

@@ -59,6 +59,15 @@ class GptTranslate::PingTest < Minitest::Test
     end
   end
 
+  def test_wrong_address
+    WebMock.allow_net_connect!
+    site = FakeSite.new({ 'url' => 'https://localhost:1/' })
+    ping = GptTranslate::Ping.new(site, '/boom.html')
+    Tempfile.open do |f|
+      assert(!ping.found?(f, ''))
+    end
+  end
+
   def test_relative_path
     assert_raises do
       GptTranslate::Ping.new({}, '404.html')

data/test/test_plain.rb

@@ -47,6 +47,20 @@ class GptTranslate::PlainTest < Minitest::Test
     )
   end
 
+  def test_ordered_list
+    assert_equal(
+      "first\n\nsecond\n\nthird",
+      GptTranslate::Plain.new("1. first\n\n2. second\n\n3. third").to_s
+    )
+  end
+
+  def test_compact_list
+    assert_equal(
+      "first\n\nsecond\n\nthird",
+      GptTranslate::Plain.new("* first\n* second\n* third").to_s
+    )
+  end
+
   def test_links
     assert_equal(
       'Hello, dude!',
@@ -89,5 +103,20 @@ class GptTranslate::PlainTest < Minitest::Test
       'Hello, !',
       GptTranslate::Plain.new('Hello, {% Java %}!').to_s
     )
+    assert_equal(
+      'Hello, !',
+      GptTranslate::Plain.new('Hello, {% plantuml "width=50%" %}!').to_s
+    )
+  end
+
+  def test_html_comments
+    assert_equal(
+      'Hello, !',
+      GptTranslate::Plain.new('Hello, <!-- Java -->!').to_s
+    )
+    assert_equal(
+      'Hello, !',
+      GptTranslate::Plain.new("Hello, <!-- \nJava\n -->!").to_s
+    )
   end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: jekyll-chatgpt-translate
 version: !ruby/object:Gem::Version
-  version: 0.0.12
+  version: 0.0.14
 platform: ruby
 authors:
 - Yegor Bugayenko