nhkore 0.3.16 → 0.3.18

data/lib/nhkore/entry.rb

@@ -8,11 +8,9 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 require 'nhkore/defn'
 require 'nhkore/util'
 
-
 module NHKore
   class Entry
     HYOUKI_SEP = '・'

data/lib/nhkore/error.rb

@@ -8,7 +8,6 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 module NHKore
   class Error < ::StandardError; end
 

data/lib/nhkore/fileable.rb

@@ -8,7 +8,6 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 module NHKore
   module Fileable
     def self.included(mod)

data/lib/nhkore/lib.rb

@@ -8,7 +8,6 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 require 'nhkore/article'
 require 'nhkore/article_scraper'
 require 'nhkore/cleaner'
@@ -33,7 +32,6 @@ require 'nhkore/variator'
 require 'nhkore/version'
 require 'nhkore/word'
 
-
 module NHKore
   ###
   # Include this file to only require the files needed to use this

data/lib/nhkore/missingno.rb

@@ -8,10 +8,8 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 require 'nhkore/util'
 
-
 module NHKore
   class Missingno
     attr_reader :kanas

data/lib/nhkore/news.rb

@@ -8,13 +8,11 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 require 'nhkore/article'
 require 'nhkore/error'
 require 'nhkore/fileable'
 require 'nhkore/util'
 
-
 module NHKore
   class News
     include Fileable
@@ -60,14 +58,14 @@ module NHKore
       coder[:articles] = @articles
     end
 
-    def self.load_data(data,article_class: Article,file: nil,news_class: News,overwrite: false,**kargs)
+    def self.load_data(data,article_class: Article,file: nil,news_class: News,overwrite: false,**_kargs)
       data = Util.load_yaml(data,file: file)
 
       articles = data[:articles]
 
       news = news_class.new
 
-      articles&.each() do |key,hash|
+      articles&.each do |key,hash|
         key = key.to_s # Change from a symbol
         news.add_article(article_class.load_data(key,hash),key: key,overwrite: overwrite)
       end
@@ -99,7 +97,6 @@ module NHKore
       @articles.each_value do |a|
         if a.sha256 == sha256
           article = a
-
           break
         end
       end
@@ -131,11 +128,11 @@ module NHKore
       return News.load_data(data,article_class: Article,news_class: FutsuuNews,**kargs)
     end
 
-    def self.load_file(file=DEFAULT_FILE,**kargs)
+    def self.load_file(file = DEFAULT_FILE,**kargs)
       return News.load_file(file,article_class: Article,news_class: FutsuuNews,**kargs)
     end
 
-    def save_file(file=DEFAULT_FILE,**kargs)
+    def save_file(file = DEFAULT_FILE,**kargs)
       super
     end
   end
@@ -148,11 +145,11 @@ module NHKore
       return News.load_data(data,article_class: Article,news_class: YasashiiNews,**kargs)
     end
 
-    def self.load_file(file=DEFAULT_FILE,**kargs)
+    def self.load_file(file = DEFAULT_FILE,**kargs)
       return News.load_file(file,article_class: Article,news_class: YasashiiNews,**kargs)
     end
 
-    def save_file(file=DEFAULT_FILE,**kargs)
+    def save_file(file = DEFAULT_FILE,**kargs)
       super
     end
   end

data/lib/nhkore/polisher.rb

@@ -8,10 +8,8 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 require 'nhkore/word'
 
-
 module NHKore
   class Polisher
     def begin_polish(str)

data/lib/nhkore/scraper.rb

@@ -8,7 +8,6 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 require 'attr_bool'
 require 'nokogiri'
 require 'open-uri'
@@ -17,7 +16,6 @@ require 'nhkore/error'
 require 'nhkore/user_agents'
 require 'nhkore/util'
 
-
 module NHKore
   class Scraper
     extend AttrBool::Ext
@@ -48,7 +46,7 @@ module NHKore
     #                   (time-consuming) operation since it opens the URL again, but necessary for some URLs.
     # @param redirect_rule [nil,:lenient,:strict]
     def initialize(url,eat_cookie: false,header: nil,is_file: false,max_redirects: 3,max_retries: 3,
-        redirect_rule: :strict,str_or_io: nil,**kargs)
+                   redirect_rule: :strict,str_or_io: nil,**kargs)
       super()
 
       if !header.nil? && !is_file
@@ -106,7 +104,7 @@ module NHKore
       return URI.join(@url,relative_url)
     end
 
-    def open(url,str_or_io=nil,is_file: @is_file)
+    def open(url,str_or_io = nil,is_file: @is_file)
       @is_file = is_file
       @str_or_io = str_or_io
       @url = url
@@ -155,16 +153,20 @@ module NHKore
         case @redirect_rule
         when :lenient,:strict
           if redirect_uri.scheme != top_uri.scheme
-            raise redirect.exception("redirect scheme[#{redirect_uri.scheme}] does not match original " \
-              "scheme[#{top_uri.scheme}] at redirect URL[#{redirect_uri}]: #{redirect}")
+            raise redirect.exception(
+              "redirect scheme[#{redirect_uri.scheme}] does not match original " \
+              "scheme[#{top_uri.scheme}] at redirect URL[#{redirect_uri}]: #{redirect}"
+            )
           end
 
           if @redirect_rule == :strict
             redirect_domain = Util.domain(redirect_uri.host)
 
             if redirect_domain != top_domain
-              raise redirect.exception("redirect domain[#{redirect_domain}] does not match original " \
-                "domain[#{top_domain}] at redirect URL[#{redirect_uri}]: #{redirect}")
+              raise redirect.exception(
+                "redirect domain[#{redirect_domain}] does not match original " \
+                "domain[#{top_domain}] at redirect URL[#{redirect_uri}]: #{redirect}"
+              )
             end
           end
         end

data/lib/nhkore/search_link.rb

@@ -8,14 +8,12 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 require 'attr_bool'
 require 'time'
 
 require 'nhkore/fileable'
 require 'nhkore/util'
 
-
 module NHKore
   class SearchLink
     extend AttrBool::Ext
@@ -49,7 +47,7 @@ module NHKore
       coder[:sha256] = @sha256
     end
 
-    def self.load_data(key,hash)
+    def self.load_data(_key,hash)
       slink = SearchLink.new(
         hash[:url],
         scraped: hash[:scraped],
@@ -74,11 +72,11 @@ module NHKore
     end
 
     def datetime=(value)
-      if value.is_a?(Time)
-        @datetime = value
-      else
-        @datetime = Util.empty_web_str?(value) ? nil : Time.iso8601(value)
-      end
+      @datetime = if value.is_a?(Time)
+                    value
+                  else
+                    Util.empty_web_str?(value) ? nil : Time.iso8601(value)
+                  end
     end
 
     def futsuurl=(value)
@@ -143,8 +141,8 @@ module NHKore
       return self
     end
 
-    def each(&block)
-      return @links.each(&block)
+    def each(&)
+      return @links.each(&)
     end
 
     def encode_with(coder)
@@ -153,14 +151,14 @@ module NHKore
       coder[:links] = @links
     end
 
-    def self.load_data(data,file: nil,**kargs)
+    def self.load_data(data,file: nil,**_kargs)
       data = Util.load_yaml(data,file: file)
 
       links = data[:links]
 
       slinks = SearchLinks.new
 
-      links&.each() do |key,hash|
+      links&.each do |key,hash|
         key = key.to_s unless key.nil?
         slinks.links[key] = SearchLink.load_data(key,hash)
       end

data/lib/nhkore/search_scraper.rb

@@ -8,7 +8,6 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 require 'net/http'
 require 'uri'
 
@@ -17,7 +16,6 @@ require 'nhkore/scraper'
 require 'nhkore/search_link'
 require 'nhkore/util'
 
-
 module NHKore
   class SearchScraper < Scraper
     DEFAULT_RESULT_COUNT = 100
@@ -60,8 +58,10 @@ module NHKore
 
     # Example: https://www3.nhk.or.jp/news/easy/k10014150691000/k10014150691000.html
     def fetch_valid_link?(link)
-      uri = begin
-        URI(link)
+      uri = nil
+
+      begin
+        uri = URI(link)
       rescue StandardError
         return false # Bad URL.
       end
@@ -111,19 +111,21 @@ module NHKore
       super(url,**kargs)
     end
 
-    def self.build_url(site,count: DEFAULT_RESULT_COUNT,**kargs)
+    # rubocop:disable Lint/UnusedMethodArgument
+    def self.build_url(site,count: DEFAULT_RESULT_COUNT,**_kargs)
       url = ''.dup
 
       url << 'https://www.bing.com/search?'
       url << URI.encode_www_form(
         q: "site:#{site}",
-        count: count
+        # count: count # FIXME: `count` is no longer allowed on Bing.
       )
 
       return url
     end
+    # rubocop:enable Lint/UnusedMethodArgument
 
-    def scrape(slinks,page=NextPage.new())
+    def scrape(slinks,page = NextPage.new())
       next_page,link_count = scrape_html(slinks,page)
 
       if link_count <= 0
@@ -133,7 +135,7 @@ module NHKore
       return next_page
     end
 
-    def scrape_html(slinks,page,next_page=NextPage.new())
+    def scrape_html(slinks,page,next_page = NextPage.new())
       doc = html_doc
       link_count = 0
 
@@ -161,7 +163,7 @@ module NHKore
       return [next_page,link_count]
     end
 
-    def scrape_rss(slinks,page,next_page=NextPage.new())
+    def scrape_rss(slinks,page,next_page = NextPage.new())
       link_count = 0
 
       if !@is_file

data/lib/nhkore/sifter.rb

@@ -8,12 +8,10 @@
 # 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

@@ -8,10 +8,8 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 require 'nhkore/util'
 
-
 module NHKore
   class Splitter
     def begin_split(str)

data/lib/nhkore/user_agents.rb

@@ -8,7 +8,6 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 if $PROGRAM_NAME == __FILE__
   require 'bundler/inline'
 

data/lib/nhkore/util.rb

@@ -8,13 +8,11 @@
 # 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

@@ -8,7 +8,6 @@
 # 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

@@ -8,7 +8,6 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 module NHKore
-  VERSION = '0.3.16'
+  VERSION = '0.3.18'
 end

data/lib/nhkore/word.rb

@@ -8,13 +8,11 @@
 # 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

@@ -8,7 +8,6 @@
 # SPDX-License-Identifier: LGPL-3.0-or-later
 #++
 
-
 TESTING = ($PROGRAM_NAME == __FILE__)
 
 if TESTING
@@ -25,9 +24,8 @@ require 'nhkore/cli/news_cmd'
 require 'nhkore/cli/search_cmd'
 require 'nhkore/cli/sift_cmd'
 
-
 module NHKore
-  def self.run(args=ARGV)
+  def self.run(args = ARGV)
     app = App.new(args)
 
     begin