jekyll-chatgpt-translate 0.0.18 → 0.0.20

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 1f593e5927f89dc9fe25dcbe28cc61ed938b3fdc42aee742d9ec31ebc86417fe
-  data.tar.gz: c5486a9adaf6e838bd016ea01175169786dbdd1807601cf922b0acc065c0a337
+  metadata.gz: 76da34820682d38fd53ca0c5c02bf8d9e23f5665af2a6f0553f11a62d6fce452
+  data.tar.gz: 413def3d7da352c1a6c2d7dbd67722329573732d53bfbae96a1bf86c39c9c16e
 SHA512:
-  metadata.gz: 794664182ebba232506ab97b026d02a7060ee369fb16756308d43362df24aed862703836fb0af142c07240fd1eeeeeb9305e43804ed5b34300ce2dff1c30a03c
-  data.tar.gz: f1db90b466e6ea441519ee35351a2bf62b9182ed772ab020b5cd4692fc8932a22115fb994089b1cb70724c92a4344865810312b7a58fe1fe666ba0e5c308aaeb
+  metadata.gz: 715bdfed0bf33249c0091a66a0f338d89469aaa43db6fb6f80bf79a2e633d68ed0087997ce7b08b259bc24a47196061d3114d2c3f66ab8f1c7f76cc9758e0a2c
+  data.tar.gz: 6a0eeb27d20053284535b0a053ab3e0832eb6e20477bb3451ac095ce5ab48e967b7e5e607aa777269475ce846b784414382e1367ceb88daf5e24ad458440c745

data/.rubocop.yml

@@ -33,3 +33,5 @@ Layout/ElseAlignment:
   Enabled: false
 Layout/EndAlignment:
   Enabled: false
+Metrics/ClassLength:
+  Max: 200

data/README.md

@@ -1,3 +1,4 @@
+[![Gem Version](https://badge.fury.io/rb/glogin.svg)](http://badge.fury.io/rb/glogin)
 <img src="logo.png" style="width:256px;"/>
 
 [![rake](https://github.com/yegor256/jekyll-chatgpt-translate/actions/workflows/rake.yml/badge.svg)](https://github.com/yegor256/jekyll-chatgpt-translate/actions/workflows/rake.yml)
@@ -69,6 +70,11 @@ Full list of options available to specify in `_config.yml`:
     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!)
 
+  * `min_chars` (optional) — minimum number of chars that must be present in
+    a paragraph in order for it to be feasible to go to ChatGPT. The robot
+    doesn't translate short paragraphs pretty enough. It's better to keep this
+    number big enough, to avoid silly translations. The default is 128.
+
   * `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/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 = '0.0.18'
+  s.version = '0.0.20'
   s.license = 'MIT'
   s.summary = 'Translate Jekyll Pages Through ChatGPT'
   s.description = [

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

@@ -47,13 +47,17 @@ class GptTranslate::ChatGPT
     @target = target
   end
 
-  def translate(text)
-    text.split("\n\n").compact.map do |par|
-      if par.length <= 32
-        Jekyll.logger.debug("Not translating this, b/c too short: \"#{par}\"")
+  def translate(markdown, min: 32)
+    markdown.split(/\n{2,}/).compact.map do |par|
+      par.strip!
+      if par.length < min
+        Jekyll.logger.debug("Not translating this, b/c too short: #{par.inspect}")
         par
-      elsif par !~ /^[[:alpha:]]/
-        Jekyll.logger.debug("Not translating this, b/c it's not a plain text: \"#{par}\"")
+      elsif par.start_with?('```') || par.end_with?('```')
+        Jekyll.logger.debug("Not translating this code block: #{par.inspect}")
+        par
+      elsif par =~ /^[^\p{Alnum}\*'"]/
+        Jekyll.logger.debug("Not translating this, b/c it's not a plain text: #{par.inspect}")
         par
       elsif @key.empty?
         par
@@ -79,15 +83,19 @@ class GptTranslate::ChatGPT
         }
       )
       answer = response.dig('choices', 0, 'message', 'content')
-      Jekyll.logger.debug("ChatGPT prompt: \"#{prompt}\", ChatGPT answer: \"#{answer}\"")
+      Jekyll.logger.debug("ChatGPT prompt: #{prompt.inspect}, ChatGPT answer: #{answer.inspect}")
     rescue StandardError => e
       attempt += 1
-      retry if attempt < 4
+      if attempt < 4
+        Jekyll.logger.error("ChatGPT failed to answer to #{prompt.inspect} \
+(attempt no.#{attempt}): #{e.message.inspect}")
+        retry
+      end
       raise e
     end
     Jekyll.logger.info("Translated #{par.split.count} #{@source.upcase} words \
 to #{answer.split.count} #{@target.upcase} words \
-through #{@model} in #{(Time.now - start).round(2)}s")
+through #{@model} in #{(Time.now - start).round(2)}s: #{"#{par[0..24]}...".inspect}")
     answer
   end
 end

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

@@ -53,6 +53,7 @@ class GptTranslate::Generator < Jekyll::Generator
     layout = config['layout'] || 'translated'
     version = config['version'] || GptTranslate::VERSION
     threshold = config['threshold'] || 1024
+    min_chars = config['min_chars'] || 128
     start = Time.now
     translated = 0
     copied = 0
@@ -79,7 +80,7 @@ class GptTranslate::Generator < Jekyll::Generator
             config['source'] || 'en',
             lang
           )
-          foreign = gpt.translate(plain)
+          foreign = gpt.translate(plain, min: min_chars)
           File.write(
             path,
             [
@@ -95,7 +96,7 @@ class GptTranslate::Generator < Jekyll::Generator
               '',
               foreign,
               '',
-              "#{marker} on #{Time.now.strftime('%d/%m/%Y %H:%M')}\n{: .jekyll-chatgpt-translate}"
+              "#{marker} on #{Time.now.strftime('%Y-%m-%d at %H:%M')}\n{: .jekyll-chatgpt-translate}"
             ].join("\n")
           )
           site.pages << Jekyll::Page.new(site, site.source, File.dirname(path), File.basename(path))
@@ -121,10 +122,10 @@ class GptTranslate::Generator < Jekyll::Generator
       k
     elsif File.exist?(file)
       k = File.read(file).strip
-      Jekyll.logger.info("The OpenAI API key taken from the file: \"#{file}\" (#{k.length} chars)")
+      Jekyll.logger.info("The OpenAI API key taken from the file: #{file.inspect} (#{k.length} chars)")
       k
     else
-      Jekyll.logger.info("The file with the OpenAI API key is not found: \"#{file}\"")
+      Jekyll.logger.info("The file with the OpenAI API key is not found: #{file.inspect}")
       nil
     end
     if key.nil? && Jekyll.env == 'development'

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

@@ -50,26 +50,26 @@ class GptTranslate::Ping
   def found?(file, marker)
     home = @site.config['url']
     return false if home.nil?
-    uri = Iri.new(home).path(@path)
+    uri = Iri.new(home).path(@path).to_s
     begin
-      before = Net::HTTP.get_response(URI(uri.to_s))
+      before = Net::HTTP.get_response(URI(uri))
       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}\"")
+#{uri.inspect} (#{html.split.count} words), saved to #{file.inspect}")
           FileUtils.mkdir_p(File.dirname(file))
           File.write(file, html)
           return true
         end
-        Jekyll.logger.info("Re-translation required for \"#{uri}\"")
+        Jekyll.logger.info("Re-translation required for #{uri.inspect}")
       else
-        Jekyll.logger.info("The page is absent, will translate \"#{uri}\"")
+        Jekyll.logger.info("The page is absent, will translate #{uri.inspect}")
       end
-      Jekyll.logger.debug("GET \"#{uri}\": #{before.code}")
+      Jekyll.logger.debug("GET #{uri.inspect}: #{before.code}")
     rescue StandardError => e
-      Jekyll.logger.debug("Failed to ping \"#{uri}\": #{e.message}")
-      Jekyll.logger.info("The page is absent: \"#{uri}\"")
+      Jekyll.logger.debug("Failed to ping #{uri.inspect}: #{e.message}")
+      Jekyll.logger.info("The page is absent (#{e.class.name}): #{uri.inspect}")
     end
     false
   end

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

@@ -37,31 +37,28 @@ class GptTranslate::Plain
     @markdown = markdown
   end
 
+  # Liquid tags are removed, but this implementation is primitive
+  # Seehttps://stackoverflow.com/questions/
   def to_s
-    # To turn compact lists into proper lists
-    @markdown.gsub(/([^\n])\n(\s*\*)/, "\\1\n\n\\2").split(/\n{2,}/).compact.map do |par|
-      par.strip!
-      # Liquid tags are removed, but this implementation is primitive
-      # Seehttps://stackoverflow.com/questions/
-      par.gsub!(/{{[^}]+}}/, '')
-      par.gsub!(/{%.+?%}/, '')
-      par.gsub!(/<!--.+?-->/m, '')
-      par = Redcarpet::Markdown.new(Strip).render(par)
-      par.gsub!("\t", '  ')
-      par.gsub!(/\n+/, ' ') unless par.start_with?('```')
-      par.gsub!(/ {2,}/, ' ') unless par.start_with?('```')
-      par.strip
-    end.join("\n\n").gsub(/\n{2,}/, "\n\n").strip
+    Redcarpet::Markdown.new(Strip).render(
+      @markdown
+        .gsub(/([^\n])\n(\s*\* )/, "\\1\n\n\\2")
+        .gsub(/<!--.+?-->/m, '')
+        .gsub(/{{[^}]+}}/, '')
+        .gsub(/{%.+?%}/, '')
+    ).strip
   end
 
   # Markdown to pain text.
+  # Motivated by https://github.com/vmg/redcarpet/blob/master/lib/redcarpet/render_strip.rb
   class Strip < Redcarpet::Render::Base
     %i[
-      block_code block_quote
+      block_quote
       block_html
-      autolink double_emphasis
-      emphasis underline
-      triple_emphasis strikethrough
+      autolink
+      underline
+      triple_emphasis
+      strikethrough
       superscript highlight quote
       footnotes footnote_def footnote_ref
       entity normal_text
@@ -71,18 +68,38 @@ class GptTranslate::Plain
       end
     end
 
+    def double_emphasis(txt)
+      "**#{txt}**"
+    end
+
+    def block_code(code, _lang)
+      code
+    end
+
+    def emphasis(txt)
+      "*#{txt}*"
+    end
+
+    def header(text, level)
+      "#{'#' * level} #{text}\n\n"
+    end
+
     def codespan(content)
       if content.start_with?("\n")
         "```#{content}```"
       elsif content.end_with?("\n")
         "```\n#{content.split("\n", 2)[1]}```"
       else
-        content
+        "`#{content}`"
       end
     end
 
-    def raw_html(content)
-      content
+    def image(link, title, alt)
+      "![#{alt}](#{link} \"#{title}\")"
+    end
+
+    def raw_html(html)
+      html
     end
 
     def list(content, _type)
@@ -90,15 +107,19 @@ class GptTranslate::Plain
     end
 
     def list_item(content, _type)
-      content
+      "#{content.strip}\n\n"
     end
 
     def paragraph(text)
-      text
+      unless text.start_with?('```')
+        text.gsub!(/\n+/, ' ')
+        text.gsub!(/\s{2,}/, ' ')
+      end
+      "#{text}\n\n"
     end
 
-    def link(_link, _title, content)
-      content
+    def link(link, _title, content)
+      "[#{content}](#{link})"
     end
   end
 end

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

@@ -44,15 +44,15 @@ class GptTranslate::Prompt
 
   def to_s
     from = ISO_639.find_by_code(@source)
-    raise "Unknown source language ISO-639 code: \"#{@source}\"" if from.nil?
+    raise "Unknown source language ISO-639 code: #{@source.inspect}" if from.nil?
     to = ISO_639.find_by_code(@target)
-    raise "Unknown source language ISO-639 code: \"#{@target}\"" if to.nil?
+    raise "Unknown source language ISO-639 code: #{@target.inspect}" if to.nil?
     head = [
-      'Please, translate the following paragraph from ',
+      'Please, translate the following Markdown paragraph from ',
       from[3],
       ' to ',
       to[3],
-      ', don\'t change proper nouns'
+      ', don\'t translate technical terms and proper nouns'
     ].join
     "#{head}:\n\n#{@par}"
   end

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

@@ -23,5 +23,5 @@
 # SOFTWARE.
 
 module GptTranslate
-  VERSION = '0.0.18'
+  VERSION = '0.0.20'
 end

data/test/test__helper.rb

@@ -24,3 +24,6 @@ $stdout.sync = true
 
 require 'simplecov'
 SimpleCov.start
+
+require 'jekyll'
+Jekyll.logger.adjust_verbosity(verbose: true)

data/test/test_chatgpt.rb

@@ -24,6 +24,7 @@
 
 require 'minitest/autorun'
 require 'webmock/minitest'
+require_relative 'test__helper'
 require_relative '../lib/jekyll-chatgpt-translate/chatgpt'
 
 # ChatGPT test.
@@ -32,20 +33,45 @@ require_relative '../lib/jekyll-chatgpt-translate/chatgpt'
 # License:: MIT
 class GptTranslate::ChatGPTTest < Minitest::Test
   def test_short_text
-    chat = GptTranslate::ChatGPT.new('fake-key', 'gpt-3.5-turbo', 'en', 'ru')
+    chat = GptTranslate::ChatGPT.new('fake-key', 'foo', 'xx', 'xx')
     assert_equal('Hello, world!', chat.translate('Hello, world!'))
   end
 
   def test_dry_mode
-    chat = GptTranslate::ChatGPT.new('', 'gpt-3.5-turbo', 'en', 'ru')
-    assert_equal(38, chat.translate('This text should not be sent to OpenAI').length)
+    chat = GptTranslate::ChatGPT.new('', 'foo', 'xx', 'xx')
+    assert_equal(38, chat.translate('This text should not be sent to OpenAI', min: 100).length)
+  end
+
+  def test_no_translation
+    chat = GptTranslate::ChatGPT.new('', 'foo', 'xx', 'xx')
+    chat.translate(
+      "
+      How are you, my friend?
+
+      Read this Java code:
+
+      ```
+      System.out.println(\"Hello, dude!\");
+      System.out.println(\"Good bye!\");
+      System.out.println(\"Done!\");
+      ```
+
+      This is it.
+      ",
+      min: 40
+    )
   end
 
   def test_markup
-    chat = GptTranslate::ChatGPT.new('fake-key', 'gpt-3.5-turbo', 'en', 'ru')
+    chat = GptTranslate::ChatGPT.new('fake-key', 'gpt-3.5-turbo', 'xx', 'xx')
     assert_equal('<img src="a"/>', chat.translate('<img src="a"/>'))
   end
 
+  def test_code_block
+    chat = GptTranslate::ChatGPT.new('fake-key', '', 'xx', 'xx')
+    chat.translate("```\ntest\n```", min: 0)
+  end
+
   def test_through_webmock
     stub_request(:any, 'https://api.openai.com/v1/chat/completions')
       .to_return(body: '{"choices":[{"message":{"content": "boom!"}}]}')

data/test/test_generator.rb

@@ -25,6 +25,7 @@
 require 'minitest/autorun'
 require 'tmpdir'
 require 'webmock/minitest'
+require_relative 'test__helper'
 require_relative '../lib/jekyll-chatgpt-translate/generator'
 
 # Generator test.

data/test/test_permalink.rb

@@ -23,6 +23,7 @@
 # SOFTWARE.
 
 require 'minitest/autorun'
+require_relative 'test__helper'
 require_relative '../lib/jekyll-chatgpt-translate/permalink'
 
 # Permalink test.

data/test/test_ping.rb

@@ -26,6 +26,7 @@ require 'minitest/autorun'
 require 'webmock/minitest'
 require 'jekyll'
 require 'tempfile'
+require_relative 'test__helper'
 require_relative '../lib/jekyll-chatgpt-translate/ping'
 
 # Ping test.

data/test/test_plain.rb

@@ -23,6 +23,7 @@
 # SOFTWARE.
 
 require 'minitest/autorun'
+require_relative 'test__helper'
 require_relative '../lib/jekyll-chatgpt-translate/plain'
 
 # Plain test.
@@ -31,8 +32,10 @@ require_relative '../lib/jekyll-chatgpt-translate/plain'
 # License:: MIT
 class GptTranslate::PlainTest < Minitest::Test
   def test_simple_map
-    assert_equal('Hello, world!', GptTranslate::Plain.new('Hello, **world**!').to_s)
-    assert_equal('Hello, Jeff!', GptTranslate::Plain.new('Hello, _Jeff_!').to_s)
+    assert_equal('Hello, **world**!', GptTranslate::Plain.new("Hello,\n**world**!").to_s)
+    assert_equal('Hello, [world](/x.html)!', GptTranslate::Plain.new("Hello,\n[world](/x.html)!").to_s)
+    assert_equal('Hello, *Jeff*!', GptTranslate::Plain.new('Hello, _Jeff_!').to_s)
+    # assert_equal('Hello, Walter!', GptTranslate::Plain.new('Hello, ~Walter~!').to_s)
     assert_equal("Hi\n\nBye", GptTranslate::Plain.new("  Hi\n\nBye\n\n\n").to_s)
     assert_equal('Hi, dude!', GptTranslate::Plain.new("  Hi,\ndude!\n").to_s)
   end
@@ -64,34 +67,27 @@ class GptTranslate::PlainTest < Minitest::Test
 
   def test_links
     assert_equal(
-      'Hello, dude!',
-      GptTranslate::Plain.new('Hello, [dude](https://www.google.com)!').to_s
+      'Hello, [dude](/a.html)!',
+      GptTranslate::Plain.new('Hello, [dude](/a.html)!').to_s
     )
   end
 
   def test_code
     assert_equal(
-      'Hello, Java!',
+      'Hello, `Java`!',
       GptTranslate::Plain.new('Hello, `Java`!').to_s
     )
   end
 
   def test_code_block
     assert_equal(
-      "Hello:\n\n```\nJava\n```",
-      GptTranslate::Plain.new("Hello:\n\n```\nJava\n```\n").to_s
+      "```\na\na\na\na\na\na\na\n\n```",
+      GptTranslate::Plain.new("```\na\na\na\na\na\na\na\n\n```").to_s
     )
-  end
-
-  def test_html
     assert_equal(
-      'This is picture: <img src="a"/>!',
-      GptTranslate::Plain.new('This is picture: <img src="a"/>!').to_s
+      "Hello:\n\n```\nJava\n```",
+      GptTranslate::Plain.new("Hello:\n\n```\nJava\n```\n").to_s
     )
-    assert_equal('<img src="a"/>', GptTranslate::Plain.new('<img src="a"/>').to_s)
-  end
-
-  def test_big_code
     assert_equal(
       "```\nHello\n```",
       GptTranslate::Plain.new("```\nHello\n```").to_s
@@ -102,6 +98,24 @@ class GptTranslate::PlainTest < Minitest::Test
     )
   end
 
+  def test_titles
+    assert_equal('# Hello', GptTranslate::Plain.new('# Hello').to_s)
+    assert_equal('## Hello', GptTranslate::Plain.new('## Hello').to_s)
+    assert_equal('### Hello', GptTranslate::Plain.new('### Hello').to_s)
+  end
+
+  def test_image
+    assert_equal('![alt](a.png "hello")', GptTranslate::Plain.new('![alt](a.png "hello")').to_s)
+  end
+
+  def test_html
+    assert_equal(
+      'This is picture: <img src="a"/>!',
+      GptTranslate::Plain.new('This is picture: <img src="a"/>!').to_s
+    )
+    assert_equal('<img src="a"/>', GptTranslate::Plain.new('<img src="a"/>').to_s)
+  end
+
   def test_liquid_tags
     assert_equal(
       'Hello, !',
@@ -131,4 +145,64 @@ class GptTranslate::PlainTest < Minitest::Test
       GptTranslate::Plain.new("Hello, <!-- \nJava\n -->!").to_s
     )
   end
+
+  def test_big_text
+    expected = "Hi, dear **friend**!
+
+In this *lovely* letter I will explain how objects work in C++:
+
+Declare a class
+
+Make an instance of it
+
+Delete the instance
+
+## More details
+
+Something like this:
+
+```
+class Foo {};
+Foo f = Foo();
+```
+
+And then use `new` and `delete` like this:
+
+```
+Foo* f = new Foo();
+delete f;
+```
+
+Should work!"
+    input = "
+Hi, dear **friend**!
+
+In this _lovely_ letter I will
+explain how objects
+work in C++:
+
+  * \tDeclare a class
+  * \tMake an instance of it
+  * \tDelete the instance
+
+## More details
+
+Something like this:
+
+```
+class Foo {};
+Foo f = Foo();
+```
+
+And then use `new` and `delete` like this:
+
+```cpp
+Foo* f = new Foo();
+delete f;
+```
+
+Should work!
+"
+    assert_equal(expected, GptTranslate::Plain.new(input).to_s)
+  end
 end

data/test/test_prompt.rb

@@ -23,6 +23,7 @@
 # SOFTWARE.
 
 require 'minitest/autorun'
+require_relative 'test__helper'
 require_relative '../lib/jekyll-chatgpt-translate/prompt'
 
 # Prompt test.
@@ -31,7 +32,11 @@ require_relative '../lib/jekyll-chatgpt-translate/prompt'
 # License:: MIT
 class GptTranslate::PromptTest < Minitest::Test
   def par(body, source, target)
-    "Please, translate the following paragraph from #{source} to #{target}, don't change proper nouns:\n\n#{body}"
+    [
+      'Please, translate the following Markdown paragraph',
+      " from #{source} to #{target},",
+      " don't translate technical terms and proper nouns:\n\n#{body}"
+    ].join
   end
 
   def test_english_to_russian

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: jekyll-chatgpt-translate
 version: !ruby/object:Gem::Version
-  version: 0.0.18
+  version: 0.0.20
 platform: ruby
 authors:
 - Yegor Bugayenko
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2023-08-22 00:00:00.000000000 Z
+date: 2023-08-23 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: iri