nhkore 0.3.17 → 0.3.19

data/lib/nhkore/sifter.rb

@@ -3,17 +3,15 @@
 
 #--
 # This file is part of NHKore.
-# Copyright (c) 2020-2021 Jonathan Bradley Whited
+# Copyright (c) 2020 Bradley Whited
 #
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 require 'nhkore/article'
 require 'nhkore/fileable'
 require 'nhkore/util'
 
-
 module NHKore
   class Sifter
     include Fileable
@@ -87,7 +85,7 @@ module NHKore
         datetime = article.datetime
 
         return true if datetime.nil? ||
-          datetime < datetime_filter[:from] || datetime > datetime_filter[:to]
+                       datetime < datetime_filter[:from] || datetime > datetime_filter[:to]
       end
 
       if !title_filter.nil?
@@ -109,7 +107,7 @@ module NHKore
       return false
     end
 
-    def filter_by_datetime(datetime_filter=nil,from: nil,to: nil)
+    def filter_by_datetime(datetime_filter = nil,from: nil,to: nil)
       if !datetime_filter.nil?
         if datetime_filter.respond_to?(:[])
           # If out-of-bounds, just nil.
@@ -234,10 +232,10 @@ module NHKore
       HTML
 
       # If have too few or too many '<col>', invalid HTML.
-      @output << %Q(<col style="width:6em;">\n) unless @ignores[:freq]
-      @output << %Q(<col style="width:17em;">\n) unless @ignores[:word]
-      @output << %Q(<col style="width:17em;">\n) unless @ignores[:kana]
-      @output << %Q(<col style="width:5em;">\n) unless @ignores[:eng]
+      @output << %(<col style="width:6em;">\n) unless @ignores[:freq]
+      @output << %(<col style="width:17em;">\n) unless @ignores[:word]
+      @output << %(<col style="width:17em;">\n) unless @ignores[:kana]
+      @output << %(<col style="width:5em;">\n) unless @ignores[:eng]
       @output << "<col>\n" unless @ignores[:defn] # No width for defn, fills rest of page
 
       @output << '<tr>'

data/lib/nhkore/splitter.rb

@@ -3,15 +3,13 @@
 
 #--
 # This file is part of NHKore.
-# Copyright (c) 2020-2021 Jonathan Bradley Whited
+# Copyright (c) 2020 Bradley Whited
 #
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 require 'nhkore/util'
 
-
 module NHKore
   class Splitter
     def begin_split(str)

data/lib/nhkore/util.rb

@@ -3,18 +3,16 @@
 
 #--
 # This file is part of NHKore.
-# Copyright (c) 2020-2021 Jonathan Bradley Whited
+# Copyright (c) 2020 Bradley Whited
 #
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 require 'cgi'
 require 'set'
 require 'time'
 require 'uri'
 
-
 module NHKore
   module Util
     CORE_DIR = 'core'
@@ -64,7 +62,8 @@ module NHKore
 
       stylers = Array(stylers)
 
-      return Psychgus.dump(obj,
+      return Psychgus.dump(
+        obj,
         deref_aliases: true, # Dereference aliases for load_yaml()
         header: true, # %YAML [version]
         line_width: 10_000, # Try not to wrap; ichiman!
@@ -117,10 +116,11 @@ module NHKore
     def self.load_yaml(data,file: nil,**kargs)
       require 'psychgus'
 
-      return Psych.safe_load(data,
+      return Psych.safe_load(
+        data,
         aliases: false,
         filename: file,
-        #freeze: true, # Not in this current version of Psych
+        # freeze: true, # Not in this current version of Psych
         permitted_classes: [Symbol],
         symbolize_names: true,
         **kargs,
@@ -180,8 +180,8 @@ module NHKore
     # String's normal strip() method doesn't work with special Unicode/HTML white space.
     def self.strip_web_str(str)
       # After testing with Benchmark, this is slower than one regex.
-      #str = str.gsub(/\A[[:space:]]+/,'')
-      #str = str.gsub(/[[:space:]]+\z/,'')
+      # str = str.gsub(/\A[[:space:]]+/,'')
+      # str = str.gsub(/[[:space:]]+\z/,'')
 
       str = str.gsub(STRIP_WEB_STR_REGEX,'')
 

data/lib/nhkore/variator.rb

@@ -3,12 +3,11 @@
 
 #--
 # This file is part of NHKore.
-# Copyright (c) 2020-2021 Jonathan Bradley Whited
+# Copyright (c) 2020 Bradley Whited
 #
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 module NHKore
   class Variator
     def begin_variate(str)
@@ -24,7 +23,7 @@ module NHKore
   end
 
   class BasicVariator < Variator
-    def end_variate(str)
+    def end_variate(_str)
       return [] # No variations; don't return nil
     end
   end
@@ -49,7 +48,7 @@ module NHKore
     def end_variate(str)
       guess = @deinflector.deinflect(str)
 
-      return [] if guess.length < 1
+      return [] if guess.empty?
       return [] if (guess = guess[0])[:weight] < 0.5
 
       return [guess[:word]]

data/lib/nhkore/version.rb

@@ -3,12 +3,11 @@
 
 #--
 # This file is part of NHKore.
-# Copyright (c) 2020-2021 Jonathan Bradley Whited
+# Copyright (c) 2020 Bradley Whited
 #
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 module NHKore
-  VERSION = '0.3.17'
+  VERSION = '0.3.19'
 end

data/lib/nhkore/word.rb

@@ -3,18 +3,16 @@
 
 #--
 # This file is part of NHKore.
-# Copyright (c) 2020-2021 Jonathan Bradley Whited
+# Copyright (c) 2020 Bradley Whited
 #
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 require 'nokogiri'
 
 require 'nhkore/error'
 require 'nhkore/util'
 
-
 module NHKore
   class Word
     attr_accessor :defn
@@ -24,7 +22,7 @@ module NHKore
     attr_reader :kanji
     attr_reader :key
 
-    def initialize(defn: nil,eng: nil,freq: 1,kana: nil,kanji: nil,unknown: nil,word: nil,**kargs)
+    def initialize(defn: nil,eng: nil,freq: 1,kana: nil,kanji: nil,unknown: nil,word: nil,**_kargs)
       super()
 
       if !word.nil?
@@ -110,14 +108,14 @@ module NHKore
       # First, try <rb> tags.
       kanjis = tag.css('rb')
       # Second, try text nodes.
-      kanjis = tag.search('./text()') if kanjis.length < 1
+      kanjis = tag.search('./text()') if kanjis.empty?
       # Third, try non-<rt> tags, in case of being surrounded by <span>, <b>, etc.
-      kanjis = tag.search("./*[not(name()='rt')]") if kanjis.length < 1
+      kanjis = tag.search("./*[not(name()='rt')]") if kanjis.empty?
 
       kanas = tag.css('rt')
 
-      raise ScrapeError,"no kanji at URL[#{url}] in tag[#{tag}]" if kanjis.length < 1
-      raise ScrapeError,"no kana at URL[#{url}] in tag[#{tag}]" if kanas.length < 1
+      raise ScrapeError,"no kanji at URL[#{url}] in tag[#{tag}]" if kanjis.empty?
+      raise ScrapeError,"no kana at URL[#{url}] in tag[#{tag}]" if kanas.empty?
 
       if kanjis.length != kanas.length
         raise ScrapeError,"number of kanji & kana mismatch at URL[#{url}] in tag[#{tag}]"
@@ -130,7 +128,7 @@ module NHKore
         kana = kanas[i].text
 
         # Uncomment for debugging; really need a logger.
-        #puts "Word[#{i}]: #{kanji} => #{kana}"
+        # puts "Word[#{i}]: #{kanji} => #{kana}"
 
         if !missingno.nil?
           # Check kana first, since this is the typical scenario.
@@ -162,7 +160,7 @@ module NHKore
 
     # Do not clean and/or strip spaces, as the raw text is important for
     # Defn and ArticleScraper.
-    def self.scrape_text_node(tag,url: nil)
+    def self.scrape_text_node(tag,url: nil) # rubocop:disable Lint/UnusedMethodArgument
       text = tag.text
 
       # No error; empty text is fine (not strictly kanji/kana only).

data/lib/nhkore.rb

@@ -3,12 +3,11 @@
 
 #--
 # This file is part of NHKore.
-# Copyright (c) 2020-2021 Jonathan Bradley Whited
+# Copyright (c) 2020 Bradley Whited
 #
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 TESTING = ($PROGRAM_NAME == __FILE__)
 
 if TESTING
@@ -16,18 +15,11 @@ if TESTING
   require 'bundler/setup'
 end
 
-require 'nhkore/app'
 require 'nhkore/lib'
-
-require 'nhkore/cli/fx_cmd'
-require 'nhkore/cli/get_cmd'
-require 'nhkore/cli/news_cmd'
-require 'nhkore/cli/search_cmd'
-require 'nhkore/cli/sift_cmd'
-
+require 'nhkore/app'
 
 module NHKore
-  def self.run(args=ARGV)
+  def self.run(args = ARGV)
     app = App.new(args)
 
     begin

data/nhkore.gemspec

@@ -1,29 +1,30 @@
 # encoding: UTF-8
 # frozen_string_literal: true
 
-
 require_relative 'lib/nhkore/version'
 
 Gem::Specification.new do |spec|
   spec.name        = 'nhkore'
   spec.version     = NHKore::VERSION
-  spec.authors     = ['Jonathan Bradley Whited']
+  spec.authors     = ['Bradley Whited']
   spec.email       = ['code@esotericpig.com']
   spec.licenses    = ['LGPL-3.0-or-later']
   spec.homepage    = 'https://github.com/esotericpig/nhkore'
   spec.summary     = 'NHK News Web (Easy) word frequency (core) scraper for Japanese language learners.'
-  spec.description =
-    'Scrapes NHK News Web (Easy) for the word frequency (core list) for Japanese language learners.' \
-    ' Includes a CLI app and a scraper library.'
+  spec.description = <<~DESC
+    Scrapes NHK News Web (Easy) for the word frequency (core list) for Japanese language learners.
+    Includes a CLI app and a scraper library.
+  DESC
 
   spec.metadata = {
-    'homepage_uri'      => 'https://github.com/esotericpig/nhkore',
-    'source_code_uri'   => 'https://github.com/esotericpig/nhkore',
-    'bug_tracker_uri'   => 'https://github.com/esotericpig/nhkore/issues',
-    'changelog_uri'     => 'https://github.com/esotericpig/nhkore/blob/master/CHANGELOG.md',
-    #'documentation_uri' => '',
-    #'wiki_uri'          => '',
-    #'mailing_list_uri'  => '',
+    'rubygems_mfa_required' => 'true',
+    'homepage_uri'          => 'https://github.com/esotericpig/nhkore',
+    'source_code_uri'       => 'https://github.com/esotericpig/nhkore',
+    'bug_tracker_uri'       => 'https://github.com/esotericpig/nhkore/issues',
+    'changelog_uri'         => 'https://github.com/esotericpig/nhkore/blob/master/CHANGELOG.md',
+    # 'documentation_uri'     => '',
+    # 'wiki_uri'              => '',
+    # 'mailing_list_uri'      => '',
   }
 
   spec.required_ruby_version = '>= 3.1'
@@ -35,7 +36,7 @@ Gem::Specification.new do |spec|
   spec.bindir        = 'bin'
   spec.executables   = [spec.name]
 
-  spec.extra_rdoc_files = %w[ CHANGELOG.md LICENSE.txt README.md ]
+  spec.extra_rdoc_files = %w[LICENSE.txt CHANGELOG.md README.md]
   spec.rdoc_options = [
     '--hyperlink-all','--show-hash',
     '--title',"NHKore v#{NHKore::VERSION} Doc",
@@ -43,40 +44,36 @@ Gem::Specification.new do |spec|
   ]
 
   spec.files = [
-    Dir.glob(File.join("{#{spec.require_paths.join(',')}}",'**','*.{erb,rb}')),
-    Dir.glob(File.join(spec.bindir,'*')),
-    Dir.glob(File.join('{samples,test,yard}','**','*.{erb,rb}')),
-    %W[ Gemfile Gemfile.lock #{spec.name}.gemspec Rakefile .yardopts ],
+    Dir.glob("{#{spec.require_paths.join(',')}}/**/*.{erb,rb}"),
+    Dir.glob("#{spec.bindir}/{#{spec.executables.join(',')}}"),
+    Dir.glob('{samples,spec,test,yard}/**/*.{erb,rb}'),
+    %W[Gemfile Gemfile.lock #{spec.name}.gemspec Rakefile .yardopts],
     spec.extra_rdoc_files,
   ].flatten
 
-  run_dep = spec.method(:add_runtime_dependency)
-  run_dep[ 'attr_bool'           ,'~> 0.2'  ] # attr_accessor?/attr_reader?.
-  run_dep[ 'bimyou_segmenter'    ,'~> 1.2'  ] # Splitting Japanese sentences into words.
-  run_dep[ 'cri'                 ,'~> 2.15' ] # CLI commands/options.
-  run_dep[ 'down'                ,'~> 5.4'  ] # Downloading files (GetCmd).
-  run_dep[ 'highline'            ,'~> 3.1'  ] # CLI input/output.
-  run_dep[ 'http-cookie'         ,'~> 1.0'  ] # Parsing/Setting cookies [(Bing)Scraper].
-  run_dep[ 'japanese_deinflector','~> 0.0'  ] # Unconjugating Japanese words (dictionary form).
-  run_dep[ 'nokogiri'            ,'~> 1.16' ] # Scraping/Hacking.
-  run_dep[ 'psychgus'            ,'~> 1.3'  ] # Styling Psych YAML.
-  run_dep[ 'public_suffix'       ,'~> 6.0'  ] # Parsing URL domain names.
-  run_dep[ 'rainbow'             ,'~> 3.1'  ] # CLI color output.
-  run_dep[ 'rss'                 ,'~> 0.3'  ] # Scraping [(Bing)Scraper].
-  run_dep[ 'rubyzip'             ,'~> 2.3'  ] # Extracting Zip files (GetCmd).
-  run_dep[ 'tiny_segmenter'      ,'~> 0.0'  ] # Splitting Japanese sentences into words.
-  run_dep[ 'tty-progressbar'     ,'~> 0.18' ] # CLI progress bars.
-  run_dep[ 'tty-spinner'         ,'~> 0.9'  ] # CLI spinning progress.
-
-  dev_dep = spec.method(:add_development_dependency)
-  dev_dep[ 'bundler'   ,'~> 2.5'  ]
-  dev_dep[ 'minitest'  ,'~> 5.25' ]
-  dev_dep[ 'rake'      ,'~> 13.2' ]
-  dev_dep[ 'raketeer'  ,'~> 0.2'  ] # Extra Rake tasks.
-  dev_dep[ 'rdoc'      ,'~> 6.7'  ] # YARDoc RDoc (*.rb).
-  dev_dep[ 'redcarpet' ,'~> 3.6'  ] # YARDoc Markdown (*.md).
-  dev_dep[ 'yard'      ,'~> 0.9'  ] # Doc.
-  dev_dep[ 'yard_ghurt','~> 1.2'  ] # Extra YARDoc Rake tasks.
+  # Japanese.
+  spec.add_dependency 'bimyou_segmenter'     ,'~> 1.2'  # Splits Japanese sentences into words.
+  spec.add_dependency 'japanese_deinflector' ,'~> 0.0'  # Un-conjugates Japanese words (dictionary form).
+  spec.add_dependency 'tiny_segmenter'       ,'~> 0.0'  # Splits Japanese sentences into words.
+  # Network/Scraping.
+  spec.add_dependency 'down'                 ,'~> 5.4'  # Downloads files (GetCmd).
+  spec.add_dependency 'http-cookie'          ,'~> 1.0'  # Parses/Sets cookies [(Bing)Scraper].
+  spec.add_dependency 'nokogiri'             ,'~> 1'    # Parses HTML.
+  spec.add_dependency 'public_suffix'        ,'~> 6.0'  # Parses URL domain names.
+  spec.add_dependency 'ronin-web-user_agents','~> 0.1'  # Generates a random User-Agent.
+  spec.add_dependency 'rss'                  ,'~> 0.3'  # Parses RSS feeds [(Bing)Scraper].
+  # Data/Files.
+  spec.add_dependency 'csv'                  ,'~> 3.3'  # Outputs CSV.
+  spec.add_dependency 'psychgus'             ,'~> 1.3'  # Styles Psych YAML.
+  spec.add_dependency 'rubyzip'              ,'~> 2.4'  # Extracts Zip files (GetCmd).
+  # CLI.
+  spec.add_dependency 'cri'                  ,'~> 2.15' # CLI commands/options.
+  spec.add_dependency 'highline'             ,'~> 3.1'  # CLI IO.
+  spec.add_dependency 'rainbow'              ,'~> 3.1'  # CLI color output.
+  spec.add_dependency 'tty-progressbar'      ,'~> 0.18' # CLI progress bars.
+  spec.add_dependency 'tty-spinner'          ,'~> 0.9'  # CLI spinning progress.
+  # Utils.
+  spec.add_dependency 'attr_bool'            ,'~> 0.2'  # attr_accessor?/attr_reader?.
 
   spec.post_install_message = <<~MSG
     +=============================================================================+
@@ -90,7 +87,4 @@ Gem::Specification.new do |spec|
     | Changelog: #{spec.metadata['changelog_uri']}
     +=============================================================================+
   MSG
-
-  # Uncomment to see max line length:
-  #puts spec.post_install_message.split("\n").map(&:length).max
 end

data/samples/looper.rb

@@ -4,12 +4,11 @@
 
 #--
 # This file is part of NHKore.
-# Copyright (c) 2020-2021 Jonathan Bradley Whited
+# Copyright (c) 2020 Bradley Whited
 #
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 ###
 # If you run this script, be aware that it uses the +-F+ force option
 # (which overwrites files without prompting).

data/test/nhkore/test_helper.rb

@@ -3,18 +3,11 @@
 
 #--
 # This file is part of NHKore.
-# Copyright (c) 2020-2021 Jonathan Bradley Whited
+# Copyright (c) 2020 Bradley Whited
 #
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 require 'minitest/autorun'
 
 require 'nhkore'
-
-
-module NHKore
-  class TestHelper < Minitest::Test
-  end
-end

data/test/nhkore_test.rb

@@ -3,21 +3,17 @@
 
 #--
 # This file is part of NHKore.
-# Copyright (c) 2020-2021 Jonathan Bradley Whited
+# Copyright (c) 2020 Bradley Whited
 #
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 require 'nhkore/test_helper'
 
+describe(NHKore) do
+  subject { NHKore }
 
-module NHKore
-  class NHKoreTest < TestHelper
-    def setup
-    end
-
-    def test_something
-    end
+  it 'has version' do
+    expect(subject::VERSION).wont_be_nil
   end
 end