mangdown 0.10.2 → 0.11.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 0fcefd9f91de261d15fdb0c165ec5be1852c9507
-  data.tar.gz: 23b77f0de9dfd3b57bea6d7d23a2d63ed51165b6
+  metadata.gz: b33eb9d7a4f9548de2e7317602f8d2bfff7e2a29
+  data.tar.gz: d7bc3c791d10bfac54b24e212e382cddaab04923
 SHA512:
-  metadata.gz: fffa3aca22c07425b59007d0d50500a01f28d8292763494ee2518e07348d98f1f8a921e54914db654b297be3c38ce99dae0c7c7cd2d2d87e7b6d3e3aead7acff
-  data.tar.gz: 69c6e148cfefb9dd5906860ece93b5d8ead402b4ee71814cabcb33bbb991b6a81af747e4b3d74d2977561a467b3de5fee45c25181ec340891e02b23b1360c615
+  metadata.gz: 54deffd02793da865629ad2e655036fe014b1d9c532521b4c16ae75bebd2baecc8959fa2c7429f639dc5613df076ab895095fbf6bcfb6cde01bdd9bdc7a8834b
+  data.tar.gz: a1e6274cde3cab1e2b2c5c51fa0e33f3b880a7f2f811c4a758ce1e45651523aa2cf6acabd94555340db033605f7cedbf2d7b891bfc14f011155557de10e0f34b

data/lib/mangdown/adapter/mangafox.rb

@@ -0,0 +1,42 @@
+module Mangdown
+  class Mangafox < Adapter::Base
+    Mangdown::ADAPTERS << self
+
+    def initialize(uri, doc)
+      super
+			@manga_list_css        = 'div.manga_list li a'
+			@chapter_list_css      = 'a.tips'
+			@root                  = 'http://mangafox.me'
+      @manga_list_uri        = "#{@root}/manga"
+			@manga_link_prefix     = ''
+			@reverse_chapters      = true
+      @manga_uri_regex       = 
+        /#{@root}\/manga\/[^\/]+?\//i
+      @chapter_uri_regex     = 
+        /#{@manga_uri_regex}(v\d+\/)?(c\d+\/)(1\.html)/i
+      @page_uri_regex        = /.+\.(png|jpg|jpeg)$/i
+		end
+
+    def build_page_uri(uri, manga, chapter, page_num)
+      uri.sub(/\d+\.html/, "#{page_num}.html")
+    end
+
+    def num_pages
+      doc.css('select')[1].css('option').length - 1
+    end
+
+    def page_image_src
+      page_image[:src]
+    end
+
+    def page_image_name
+      doc.css('select')[1].css('option[selected]').first().text
+        .rjust(3,'0')
+    end
+
+    private
+    def page_image
+      doc.css('img')[0]
+    end
+  end
+end

data/lib/mangdown/adapter/mangahere.rb

@@ -0,0 +1,46 @@
+module Mangdown
+  class Mangahere < Adapter::Base
+    Mangdown::ADAPTERS << self
+
+    def initialize(uri, doc)
+      super
+			@root                  = 'http://www.mangahere.co'
+			@manga_list_css        = 'a.manga_info'
+			@chapter_list_css      = '.detail_list ul a'
+      @manga_list_uri        = "#{@root}/mangalist/"
+			@manga_link_prefix     = ''
+			@reverse_chapters      = true
+      @manga_uri_regex       = 
+        /#{@root}\/manga\/([^\/]+\/)?/i
+      @chapter_uri_regex     = 
+        /#{@manga_uri_regex}(v\d+\/)?(c\d+\/)(1\.html)?/i
+      @page_uri_regex        = /.+\.(png|jpg|jpeg)$/i
+		end
+
+    def build_page_uri(uri, manga, chapter, page_num)
+      if page_num == 1
+        uri
+      else
+        uri.sub(/\/[^\/]*$/, "/#{page_num}.html")
+      end
+    end
+
+    def num_pages
+      doc.css('.go_page span.right select')[0].css('option').length
+    end
+
+    def page_image_src
+      page_image[:src].sub(/\?.*$/, '')
+    end
+
+    def page_image_name
+      doc.css('.go_page span.right select')[0].css('option[selected]')
+        .text.rjust(3, '0')
+    end
+
+    private
+    def page_image
+      doc.css('.read_img img')[0]
+    end
+  end
+end

data/lib/mangdown/adapter/mangapanda.rb

@@ -0,0 +1,13 @@
+require_relative 'mangareader'
+
+module Mangdown
+  class Mangapanda < Mangareader
+    Mangdown::ADAPTERS << self
+
+    def initialize(uri, doc)
+      @root = 'http://www.mangapanda.com'
+      super
+    end
+  end
+end
+

data/lib/mangdown/adapter/mangareader.rb

@@ -0,0 +1,45 @@
+module Mangdown
+  class Mangareader < Adapter::Base
+    Mangdown::ADAPTERS << self
+
+    def initialize(uri, doc)
+      super
+			@root                  ||= 'http://www.mangareader.net'
+			@manga_list_css        = 'ul.series_alpha li a'
+			@chapter_list_css      = 'div#chapterlist td a'
+      @manga_list_uri        = "#{@root}/alphabetical"
+			@manga_link_prefix     = @root 
+			@reverse_chapters      = false
+      @manga_uri_regex       = 
+        /#{@root}(\/\d+)?(\/[^\/]+)(\.html)?/i
+      @chapter_uri_regex     = 
+        /#{@root}(\/[^\/]+){1,2}\/(\d+|chapter-\d+\.html)/i
+      @page_uri_regex        = /.+\.(png|jpg|jpeg)$/i
+    end
+
+    def build_page_uri(uri, manga, chapter, page_num)
+      "#{root}/#{manga.gsub(' ', '-')}/#{chapter}/#{page_num}"
+    end
+
+    def num_pages
+      doc.css('select')[1].css('option').length
+    end
+
+    def page_image_src
+      page_image[:src]
+    end
+
+    def page_image_name
+      page_image[:alt].sub(/([^\d]*)(\d+)(\.\w+)?$/) { 
+        "#{Regexp.last_match[1]}" +
+        "#{Regexp.last_match[2].to_s.rjust(3, '0')}"
+      }
+    end
+
+    private
+    def page_image
+      doc.css('img')[0]
+    end
+  end
+end
+

data/lib/mangdown/adapter/wiemanga.rb

@@ -0,0 +1,44 @@
+module Mangdown
+  class Wiemanga < Adapter::Base
+    Mangdown::ADAPTERS << self
+
+    def initialize(uri, doc)
+      super
+			@root                  = 'http://www.wiemanga.com'
+      @manga_list_uri        = 
+        "#{@root}/search/?name_sel=contain&author_sel=contain" +
+        "&completed_series=either"
+			@manga_list_css        = 'a.resultbookname'
+			@chapter_list_css      = 
+        '.chapterlist tr:not(:first-child) .col1 a'
+			@manga_link_prefix     = '' 
+			@reverse_chapters      = true
+      @manga_uri_regex       = /#{@root}\/manga\/([^\/]+)(\.html)?/i
+      @chapter_uri_regex     = 
+        /#{@root}\/chapter\/([^\/]+)\/(\d+)(\/|\.html)?/i
+      @page_uri_regex        = /.+\.(png|jpg|jpeg)$/i
+		end
+
+    def build_page_uri(uri, manga, chapter, page_num)
+      "#{uri}-#{page_num}.html"
+    end
+
+    def num_pages
+      doc.css('select#page')[0].css('option').length
+    end
+
+    def page_image_src
+      page_image[:src]
+    end
+
+    def page_image_name
+      doc.css('select#page')[0].css('option[selected]')[0]
+        .text.rjust(3, '0')
+    end
+
+    private
+    def page_image
+      doc.css('img#comicpic')[0]
+    end
+  end
+end

data/lib/mangdown/adapter.rb

@@ -0,0 +1,85 @@
+module Mangdown
+  module Adapter
+    class Base
+      
+      attr_reader :root
+      def initialize(uri, doc)
+        @uri, @doc = uri, doc
+        #@root                  = '' 
+        #@manga_list_css        = ''
+        #@chapter_list_css      = ''
+        #@manga_list_uri        = '' 
+        #@manga_link_prefix     = '' 
+        #@reverse_chapters      = true || false
+        #@manga_uri_regex       = /.*/i 
+        #@chapter_uri_regex     = /.*/i
+        #@page_uri_regex        = /.*/i
+      end
+
+      def type
+        self.class.to_s.split('::').last.downcase.to_sym
+      end
+
+      # Must return true/false if uri represents a manga for adapter
+      def is_manga?(uri = @uri)
+        uri.slice(@manga_uri_regex) == uri
+      end
+
+      # Must return true/false if uri represents a chapter for adapter
+      def is_chapter?(uri = @uri)
+        uri.slice(@chapter_uri_regex) == uri
+      end
+
+      # Must return true/false if uri represents a page for adapter
+      def is_page?(uri = @uri)
+        uri.slice(@page_uri_regex) == uri
+      end
+
+      # Must return a uri for a page given the arguments
+      def build_page_uri(uri, manga, chapter, page_num)
+      end
+
+      # Must return the number of pages for a chapter 
+      def num_pages
+      end
+
+      # Must return the src for the page image
+      def page_image_src
+      end
+
+      # Must return the name of the page image
+      def page_image_name
+      end
+
+      # If no block given, must return an array arrays 
+      # [manga_uri, manga_name, adapter_type]
+      # If block given, then the block may alter this array
+      # Only valid mangas should be returned (using is_manga?(uri))
+      def manga_list
+        doc.css(@manga_list_css).map { |a| 
+          manga = ["#{@manga_link_prefix}#{a[:href]}",a.text.strip,type]
+          next(nil) unless is_manga?(manga.first)
+          block_given? ? yield(manga) : manga
+        }.compact
+      end
+
+      # If no block given, must return an array arrays 
+      # [chapter_uri, chapter_name, adapter_type]
+      # If block given, then the block may alter this array
+      # Only valid chapters should be returned (using is_chapter?(uri))
+      def manga_chapters
+        chapters = doc.css(@chapter_list_css).map { |a|
+          chapter = [(root + a[:href].sub(root, '')),a.text.strip,type]
+          next(nil) unless is_chapter?(chapter.first)
+          block_given? ? yield(chapter) : chapter 
+        }.compact
+        @reverse_chapters ? chapters.reverse : chapters
+      end
+
+      private
+      def doc
+        @doc ||= Tools.get_doc(@uri)
+      end
+    end
+  end
+end

data/lib/mangdown/chapter.rb

@@ -39,110 +39,65 @@ module Mangdown
 
 		# download all pages in a chapter
 		def download_to(dir)
-      dir = File.expand_path(dir + '/' + @name)
+      dir   = Tools.relative_or_absolute_path(dir, @name)
+      pages = map {|page| page.to_page}
+      failed    = []
+      succeeded = []
+
       Dir.mkdir(dir) unless Dir.exists?(dir)
-      
-      # Limit active threads to 20
-      each_slice(20) do |slice| 
-        threads = []
-        slice.each do |page|
-          threads << Thread.new(page) do |this_page| 
-            this_page.to_page.download_to(dir)
+      Tools.hydra_streaming(pages) do |stage, page, data=nil|
+        case stage
+        when :failed
+          failed << page
+        when :succeeded
+          succeeded << page
+        when :before
+          path = page.file_path(dir)
+          !File.exist?(path)
+        when :body
+          unless failed.include?(page)
+            path = page.file_path(dir)
+            File.open(path, 'ab') { |file| file.write(data) } 
           end
-        end
-        threads.each {|thread| thread.join}
-      end
-		end
-
-		private
-			# get page objects for all pages in a chapter
-			def get_pages
-        threads = []
-
-        num_pages = get_num_pages(Tools.get_doc(uri))
-        1.upto(num_pages) do |num|
-          threads << Thread.new(num) do |this_num|
-            Tools.no_time_out do
-              tries = 0
-              until doc = get_page_doc(this_num) || tries > 2
-                tries += 1
-              end
-              return unless doc
-              @pages << get_page(doc)
-            end
+        when :complete
+          unless failed.include?(page)
+            path = page.file_path(dir)
+            FileUtils.mv(path, "#{path}.#{Tools.file_type(path)}")
           end
         end
-
-        threads.each {|thread| thread.join}
-			end
-
-      # get the number of pages in a chapter
-			def get_num_pages(doc)
-				# the select is a dropdown menu of chapter pages
-        doc.css('select')[1].css('option').length
-			end	
-
-  end
-
-	# mangareader chapter object
-	class MRChapter < Chapter
-		private
-      # get the doc for a given page number
-      def get_page_doc(num)
-        root     = @properties.root
-        manga    = @manga.gsub(' ', '-')
-        uri_str  = "#{root}/#{manga}/#{@chapter}/#{num}"
-        page_uri = Mangdown::Uri.new(uri_str).downcase
-
-        Tools.get_doc(page_uri)
-      rescue SocketError => error 
-        STDERR.puts( "#{error.message} | #{name} | #{num}" )
       end
-		
-			# get the page uri and name
-			def get_page(doc)
-				image = doc.css('img')[0]
+      FileUtils.rm_rf(dir) if succeeded.empty?
 
-				MDHash.new(
-					uri: image['src'], 
-					name: (image['alt'] + ".jpg"),
-          site: @properties.type,
-        )
-      rescue NoMethodError => error
-        puts 'doc was ' + doc.class
-			end
-	end
+      !succeeded.empty?
+		end
 
-	# mangafox chapter object
-	class MFChapter < Chapter
 		private
-      # get the doc for a given page number
-      def get_page_doc(num)
-        Tools.get_doc(
-          Mangdown::Uri.new(
-            @properties.root                                + 
-            "/manga/#{@manga.gsub(' ', '_')}/c#{@chapter}/" + 
-            "#{num}.html"
-          ).downcase
-        )
-      rescue SocketError => error 
-        STDERR.puts( "#{error.message} | #{name} | #{num}" )
+    # get page objects for all pages in a chapter
+    def get_pages
+      pages = (1..@properties.num_pages).map {|num| get_page_hash(num)}
+
+      Tools.hydra(pages) do |page, body|
+        @pages << get_page(page.uri, Nokogiri::HTML(body))
       end
+    end
 
-			# get the page name and uri
-			def get_page(doc)
-				image = doc.css('img')[0]
+    # get the doc for a given page number
+    def get_page_hash(num)
+      uri_str = @properties.build_page_uri(uri, @manga, @chapter, num)
 
-				MDHash.new(
-					uri: image[:src], 
-					name: image[:src].sub(/.+\//, ''),
-          site: @properties.type,
-        )
-			end
+      MDHash.new(
+        uri: Mangdown::Uri.new(uri_str).downcase, name: num
+      )
+    end
 
-			# get the number of pages
-			def get_num_pages(doc)
-        super - 1
-			end
-	end
+    # get the page name and uri
+    def get_page(uri, doc)
+      properties = Properties.new(uri, nil, doc)
+      MDHash.new(
+        uri:  properties.page_image_src, 
+        name: properties.page_image_name,
+        site: properties.type
+      )
+    end
+  end
 end

data/lib/mangdown/commands.rb

@@ -7,6 +7,14 @@ module M
   HELP_FILE_PATH = File.expand_path(
     '../../doc/help.txt', File.dirname(__FILE__)
   )
+  MANGA_PAGES = (1..9).map { |p| 
+      "http://www.wiemanga.com/search/?name_sel=contain" +
+      "&author_sel=contain&completed_series=either&page=#{p}.html"
+    } +
+    [
+      'http://www.mangareader.net/alphabetical',
+      'http://mangafox.me/manga/'
+    ]
 
   # return a list of hash with :uri and :name of mangas found in list
   def find(search)
@@ -53,12 +61,12 @@ module M
     data = data_from_file
     return MangaList.from_data(data) if data.is_a? Array 
 
-    MangaList.new(
-      'http://www.mangareader.net/alphabetical',
-      'http://mangafox.me/manga/'
-    ).tap { |list| File.open(path,'w+') {|f| f.write(list.to_yaml)} }
+    MangaList.new(*MANGA_PAGES).tap { |list| 
+      File.open(path,'w+') {|f| f.write(list.to_yaml)} 
+    }
   rescue Object => error
     puts "#{path} is corrupt: #{error.message}"
+    raise
   end
 
   # check if the search key contains letters or numbers

data/lib/mangdown/manga.rb

@@ -15,10 +15,9 @@ module Mangdown
 			@name = name
 			@uri  = Mangdown::Uri.new(uri)
       @chapters = []
-      @properties = Properties.new(@uri)
+      @properties = Properties.new(uri)
 
       get_chapters
-      @chapters.select! { |chapter| @properties.is_chapter?(chapter) }
     end
 
     def inspect
@@ -41,8 +40,12 @@ module Mangdown
       bar = progress_bar(start, stop)
       reset(start, stop)
       loop do
-        self.next.to_chapter.download_to(dir)
-        bar.increment!
+        chapter = self.next.to_chapter
+        if chapter.download_to(dir)
+          bar.increment!
+        else
+          STDERR.puts("error: #{chapter.name} was not downloaded") 
+        end
       end
     end
 
@@ -70,19 +73,9 @@ module Mangdown
 
     # get push MDHashes of manga chapters to @chapters 
     def get_chapters
-      doc        = Tools.get_doc(@uri)
-      root       = @properties.root
-
-      #get the link with chapter name and uri
-      doc.css(@properties.manga_css_klass).each do |chapter|
-        @chapters << MDHash.new(
-          uri: (root + chapter[:href].sub(root, '')), 
-          name: chapter.text,
-          site: @properties.type,
-        ) 
+      @chapters += @properties.manga_chapters do |uri, name, site|
+         MDHash.new(uri: uri, name: name, site: site)
       end
-
-      @chapters.reverse! if @properties.reverse 
     end
 
     def chapter_indeces(start, stop)
@@ -90,14 +83,16 @@ module Mangdown
     end
 
     def setup_download_dir!(dir)
-      dir += "/#{name}"
-      Dir.mkdir(dir) unless Dir.exist?(dir)
-      dir
+      "#{dir}/#{name}".tap {|dir| Dir.mkdir(dir) unless Dir.exist?(dir)}
     end
 
     def validate_indeces!(start, stop)
       i_start, i_stop = chapter_indeces(start, stop)
-      if i_stop < i_stop
+      if i_start.nil? || i_stop.nil?
+        last  = chapters.length - 1
+        error = "This manga has chapters in the range (0..#{last})"
+        raise ArgumentError, error
+      elsif i_stop < i_stop
         error = 'Last index must be greater than first index'
         raise ArgumentError, error
       end 

data/lib/mangdown/manga_list.rb

@@ -22,21 +22,11 @@ module Mangdown
       @mangas.map(&:to_hash).to_yaml
     end
 
+    private
 		# get a list of mangas from the uri
     def get_mangas(uri)
-			properties = Properties.new(uri)
-      if properties.empty?
-        raise ArgumentError, 
-          "Bad URI: No Properties Specified for URI <#{uri}>"
-      end
-
-      doc = Tools.get_doc(uri)
-      # This should be put in a tool
-			doc.css(properties.manga_list_css_klass).each do |a|
-				@mangas << MDHash.new( 
-					uri: "#{properties.manga_link_prefix}#{a[:href]}", 
-					name: a.text
-			  )
+			@mangas += Properties.new(uri).manga_list do |uri, name|
+				MDHash.new(uri: uri, name: name) 
       end
     end
   end

data/lib/mangdown/mangdown.rb

@@ -1,4 +1,6 @@
 module Mangdown 
+  ADAPTERS = []
+
   module Equality
     def eql?(other)
       (self.name == other.name) && (self.uri == other.uri)
@@ -8,5 +10,10 @@ module Mangdown
       # puts "You may want to use :eql?"
       super
     end
+
+    # space ship operator for sorting
+    def <=>(other)
+      self.name <=> other.name
+    end
   end
 end

data/lib/mangdown/mangdown_hash.rb

@@ -2,10 +2,8 @@ module Mangdown
   class MDHash
     include Equality
 
-    attr_reader :properties
-
 		def initialize(options = {})
-      @properties = Properties.new(options[:site] || options[:uri])
+      @properties = Properties.new(options[:uri], options[:site])
 
       @hash = {}
       [:uri, :name].each {|key| @hash[key] = options.fetch(key)}
@@ -15,7 +13,7 @@ module Mangdown
 
 		# explicit conversion to manga 
     def to_manga
-      if @properties.is_manga?(self)
+      if @properties.is_manga?
         Manga.new(name, uri)
       else
         raise NoMethodError, 'This is not a known manga type'
@@ -24,8 +22,8 @@ module Mangdown
 
 		# explicit conversion to chapter 
 		def to_chapter
-      if @properties.is_chapter?(self)
-        @properties.chapter_klass.new(name, uri)
+      if @properties.is_chapter?
+        Chapter.new(name, uri)
       else
         raise NoMethodError, 'This is not a known chapter type'
       end
@@ -33,7 +31,7 @@ module Mangdown
 
 		# explicit conversion to page 
  	  def to_page 
-      if @properties.is_page?(self)
+      if @properties.is_page?
         Page.new(name, uri)
       else
         raise NoMethodError, 'This is not a known page type'
@@ -50,6 +48,16 @@ module Mangdown
       @hash[:uri]
     end
 
+    # name writer
+    def name=(other)
+      @hash[:name] = other
+    end
+
+    # uri writer
+    def uri=(other)
+      @hash[:uri] = other
+    end
+
     def [](key)
       @hash[key]
     end