jekyll-chatgpt-translate 0.0.17 → 0.0.19

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f48ffb6334ada2deb878ede0622065326b23ed765946070aff20c9ce5888715d
-  data.tar.gz: df98890a24fb215b1ba1efc2794a6e4bf1cb0b9949efb94648b267c343052562
+  metadata.gz: d7cc9cfc295b4eeb123b972094f4fa1c648a88f3df384a78b489b9b0bafbd0c1
+  data.tar.gz: 06cd12eabf0fbf6e77a64f52384609bcdc33edb8c6324aa4555de4f7fb275c99
 SHA512:
-  metadata.gz: 158df2fdeb7459cace117be39d3e94f9830c852034719ae049777b863bbbc61dc869c362534cc651f53e21af1fa37203865ec7b295254768274921a406cf858f
-  data.tar.gz: f8e34b0bd3432d598b9e7b2f5582ede75231f0c8c9e3dd71db348a5711ddf56106f7685adbf28c02a84f8ef84c7589f4a4909c650afb08d9a110751a317b8bbd
+  metadata.gz: d0b55a94c132688afe28b9edbc4115570e6fd449b001bc606bf33c9d30712f2c1639fa8e0334652c0e8aa36b79e356f40845162e810e5b980f8b2dbcd7641746
+  data.tar.gz: e058ea19c7ffb630ab48bac0b433c8fde5f5b76d6087833cb2f7a70c04eaaca32f7273cacd881caf7b3b9355b5cb6a15f539060ff14bd3184baf5a6214b79914

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)
@@ -50,6 +51,8 @@ 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
 
@@ -62,6 +65,16 @@ 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!)
+
+  * `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/features/step_definitions/steps.rb

@@ -40,7 +40,7 @@ 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
@@ -59,20 +59,16 @@ Then('File {string} contains {string}') do |string, string2|
   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
 
@@ -81,7 +77,7 @@ 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
 

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 = '0.0.17'
+  s.version = '0.0.19'
   s.license = 'MIT'
   s.summary = 'Translate Jekyll Pages Through ChatGPT'
   s.description = [

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

@@ -47,12 +47,12 @@ class GptTranslate::ChatGPT
     @target = target
   end
 
-  def translate(text)
+  def translate(text, min: 32)
     text.split("\n\n").compact.map do |par|
-      if par.length <= 32
+      if par.length < min
         Jekyll.logger.debug("Not translating this, b/c too short: \"#{par}\"")
         par
-      elsif par !~ /^[[:alpha:]]/
+      elsif par !~ /^[[[:alpha:]]'"]/
         Jekyll.logger.debug("Not translating this, b/c it's not a plain text: \"#{par}\"")
         par
       elsif @key.empty?

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
@@ -68,41 +69,42 @@ class GptTranslate::Generator < Jekyll::Generator
         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
-        if GptTranslate::Ping.new(site, link).found?(dest, version)
+        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, min: min_chars)
+          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('%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))
+          translated += 1
+          Jekyll.logger.info("Translated via ChatGPT: #{path} (#{File.size(path)} bytes)")
         end
-        next if translated >= threshold
-        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)")
+        doc.data["translated-#{lang}-url"] = link
+        doc.data['chatgpt-model'] = model
       end
     end
     Jekyll.logger.info("#{translated} pages translated and #{copied} pages copied in #{(Time.now - start).round(2)}s")

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

@@ -39,7 +39,7 @@ class GptTranslate::Plain
 
   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|
+    @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/
@@ -55,6 +55,7 @@ class GptTranslate::Plain
   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
@@ -71,6 +72,10 @@ class GptTranslate::Plain
       end
     end
 
+    def header(text, level)
+      "#{'#' * level} #{text}"
+    end
+
     def codespan(content)
       if content.start_with?("\n")
         "```#{content}```"
@@ -81,6 +86,10 @@ class GptTranslate::Plain
       end
     end
 
+    def image(link, title, alt)
+      "![#{alt}](#{link} \"#{title}\")"
+    end
+
     def raw_html(content)
       content
     end

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

@@ -23,5 +23,5 @@
 # SOFTWARE.
 
 module GptTranslate
-  VERSION = '0.0.17'
+  VERSION = '0.0.19'
 end

data/test/test_generator.rb

@@ -75,28 +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

data/test/test_plain.rb

@@ -31,8 +31,10 @@ 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, world!', GptTranslate::Plain.new("Hello,\n**world**!").to_s)
+    assert_equal('Hello, world!', 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
@@ -83,6 +85,16 @@ 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"/>!',

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: jekyll-chatgpt-translate
 version: !ruby/object:Gem::Version
-  version: 0.0.17
+  version: 0.0.19
 platform: ruby
 authors:
 - Yegor Bugayenko