iwadon-text-hatena 0.12.20080627.0

data/lib/text/hatena/auto_link/hatena_rss.rb

@@ -0,0 +1,26 @@
+require "text/hatena/auto_link/scheme"
+
+module Text
+  class Hatena
+    class AutoLink
+      class HatenaRSS < Scheme
+        @@pattern = /\[?(r:id:([A-Za-z][a-zA-Z0-9_\-]{2,14}))\]?/i
+
+        def patterns
+          [@@pattern]
+        end
+
+        def init
+          super
+          @domain = 'r.hatena.ne.jp'
+        end
+
+        def parse(text, opt = {})
+          return if @@pattern !~ text
+          return sprintf('<a href="http://%s/%s/"%s>%s</a>',
+                         @domain, $2, @a_target_string, $1)
+        end
+      end
+    end
+  end
+end

data/lib/text/hatena/auto_link/hatena_search.rb

@@ -0,0 +1,44 @@
+require "text/hatena/auto_link/scheme"
+
+module Text
+  class Hatena
+    class AutoLink
+      class HatenaSearch < Scheme
+        @@pattern = /\[search:(?:(keyword|question|asin|web):)?([^\]]+?)\]/i
+
+        def patterns
+          [@@pattern]
+        end
+
+        def init
+          super
+          @domain = 'search.hatena.ne.jp'
+        end
+
+        def parse(text, opt)
+          return if @@pattern !~ text
+          type, word = $1, $2
+          enword = html_encode(word)
+          case type.to_s.downcase
+          when 'question'
+            return sprintf('<a href="http://%s/questsearch?word=%s&ie=utf8"%s>search:%s:%s</a>',
+                           @domain, enword, @a_target_string,
+                           type, word)
+          when 'asin'
+            return sprintf('<a href="http://%s/asinsearch?word=%s&ie=utf8"%s>search:%s:%s</a>',
+                           @domain, enword, @a_target_string,
+                           type, word)
+          when 'web'
+            return sprintf('<a href="http://%s/websearch?word=%s&ie=utf8"%s>search:%s:%s</a>',
+                           @domain, enword, @a_target_string,
+                           type, word)
+          else
+            return sprintf('<a href="http://%s/keyword?word=%s&ie=utf8"%s>search:%s%s</a>',
+                           @domain, enword, @a_target_string,
+                           type ? "#{type}:" : '', word)
+          end
+        end
+      end
+    end
+  end
+end

data/lib/text/hatena/auto_link/http.rb

@@ -0,0 +1,102 @@
+require "open-uri"
+require "text/hatena/auto_link/scheme"
+
+module Text
+  class Hatena
+    class AutoLink
+      class HTTP < Scheme
+        @@pattern_simple = /\[?(https?:\/\/[A-Za-z0-9~\/._\?\&=\-%#\+:\;,\@\'\(\)\!]+)\]?/i
+        @@pattern_useful = /\[(https?:\/\/[A-Za-z0-9~\/._\?\&=\-%#\+:\;,\@\'\(\)\!]+?):(title(?:=([^\]]*))?|barcode|detail|image(?::([hw]\d+))?)\]/i
+
+        def patterns
+          [@@pattern_useful, @@pattern_simple]
+        end
+
+        def parse(text, opt = {})
+          case text
+          when @@pattern_useful
+            return _parse_useful(text, opt)
+          when @@pattern_simple
+            return _parse_simple(text)
+          end
+        end
+
+        def _parse_simple(url)
+          return nil if url.nil? or url.empty?
+          url.sub!(/^\[/, '')
+          url.sub!(/\]$/, '')
+          sprintf('<a href="%s"%s>%s</a>', url, @a_target_string, url)
+        end
+
+        def _parse_useful(text, opt)
+          return unless @@pattern_useful =~ text
+          url, type, title, size = $1, $2, $3, $4
+          case type
+          when /^title/i
+            title ||= _get_page_title(url)
+            sprintf('<a href="%s"%s>%s</a>', url, @a_target_string, title)
+          when /^detail/i
+            title ||= _get_page_title(url)
+            html = sprintf('<div class="hatena-http-detail"><p class="hatena-http-detail-url"><a href="%s"%s>%s</a></p><p class="hatena-http-detail-title">%s</p></div>', url, @a_target_string, url, title)
+            html = "</p>#{html}<p>" if opt[:in_paragraph]
+            html
+          when /^image/i
+            if /\.(jpe?g|gif|png|bmp)$/i =~ url
+              size_string = ""
+              if /^h(\d+)$/i =~ size
+                size_string = sprintf(' height="%s"', $1)
+              elsif /^w(\d+)$/i =~ size
+                size_string = sprintf(' width="%s"', $1)
+              end
+              sprintf('<a href="%s"%s><img src="%s" alt="%s" class="hatena-http-image"%s></a>',
+                      url,
+                      @a_target_string,
+                      url,
+                      url,
+                      size_string
+                      )
+            else
+              sprintf('<a href="%s"%s>%s</a>',
+                      url,
+                      @a_target_string,
+                      url
+                      )
+            end
+          when /^barcode/i
+            str = html_encode(url)
+            sprintf('<a href="%s"%s><img src="http://d.hatena.ne.jp/barcode?str=%s" class="barcode" alt="%s"></a>',
+                    url,
+                    @a_target_string,
+                    str,
+                    url)
+          end
+        end
+
+        private
+
+        def _get_page_title(url)
+          begin
+            open(url) do |f|
+              content = f.read(131072) # 2^17
+              return "#{url} (notitle)" unless /<title.*?>(.*?)<\/title>/i =~ content
+              title = $1
+              if h = @option[:title_handler]
+                if /charset="?(.+?)"?$/i =~ f.content_type
+                  cset = $1.downcase
+                elsif /<meta[^>]+charset="?([\w\d\s\-]+)"?/i =~ content
+                  cset = $1.downcase
+                end
+                title = h.call(title, cset)
+              end
+              return title
+            end
+          rescue Timeout::Error
+            return "#{url} (timeout)"
+          rescue Exception => e
+            return "#{url} (#{e.message})"
+          end
+        end
+      end
+    end
+  end
+end

data/lib/text/hatena/auto_link/mailto.rb

@@ -0,0 +1,24 @@
+require "text/hatena/auto_link/scheme"
+
+module Text
+  class Hatena
+    class AutoLink
+      class Mailto < Scheme
+        @@pattern = /\[?mailto:([a-zA-Z0-9_][a-zA-Z0-9_\.\-]+\@[a-zA-Z0-9_]+[a-zA-Z0-9_\.\-]*[a-zA-Z0-9_])\]?/i
+
+        def patterns
+          [@@pattern]
+        end
+
+        def parse(text, opt)
+          return unless @@pattern =~ text
+          addr = $1
+          return sprintf('<a href="mailto:%s">mailto:%s</a>',
+                         addr,
+                         addr
+                         )
+        end
+      end
+    end
+  end
+end

data/lib/text/hatena/auto_link/rakuten.rb

@@ -0,0 +1,24 @@
+require "kconv"
+require "text/hatena/auto_link/scheme"
+
+module Text
+  class Hatena
+    class AutoLink
+      class Rakuten < Scheme
+        @@pattern = /\[rakuten:([^\]]+?)\]/i
+
+        def patterns
+          [@@pattern]
+        end
+
+        def parse(text, opt)
+          return if @@pattern !~ text
+          word = $1
+          return sprintf('<a href="http://pt.afl.rakuten.co.jp/c/002e8f0a.89099887/?sv=2&v=3&p=0&sitem=%s"%s>rakuten:%s</a>',
+                         html_encode(word.toeuc),
+                         @a_target_string, word)
+        end
+      end
+    end
+  end
+end

data/lib/text/hatena/auto_link/scheme.rb

@@ -0,0 +1,40 @@
+module Text
+  class Hatena
+    class AutoLink
+      class Scheme
+        attr_accessor :patterns
+
+        def initialize(args = {})
+          @option = args
+          init
+        end
+
+        def init
+          @a_target = @option[:a_target]
+          @a_target_string = @a_target ?
+            %Q! target="#{escape_attr(@a_target)}"! :
+            ""
+        end
+
+        def parse(text)
+          text
+        end
+
+        def pattern
+          /#{patterns.join("|")}/
+        end
+
+        def escape_attr(str)
+          str.gsub(/\"/, "&quote;")
+        end
+
+        def html_encode(text)
+          return nil if text.nil? or text.empty?
+          text.gsub(/(\W)/n) do
+            sprintf("%%%02x", $1[0])
+          end
+        end
+      end
+    end
+  end
+end

data/lib/text/hatena/auto_link/tex.rb

@@ -0,0 +1,26 @@
+require "text/hatena/auto_link/scheme"
+
+module Text
+  class Hatena
+    class AutoLink
+      class Tex < Scheme
+        @@pattern = /\[tex:(.*?[^\\\\])\]/i
+
+        def patterns
+          [@@pattern]
+        end
+
+        def parse(text, opt = {})
+          return if @@pattern !~ text
+          alt = escape_attr($1)
+          tex = $1
+          tex.gsub!(/\\([\[\]])/, '\1')
+          tex.gsub!(/\s/, '~')
+          tex.gsub!(/"/, '&quot;')
+          return sprintf('<img src="http://d.hatena.ne.jp/cgi-bin/mimetex.cgi?%s" class="tex" alt="%s">',
+                         tex, alt)
+        end
+      end
+    end
+  end
+end

data/lib/text/hatena/auto_link/unbracket.rb

@@ -0,0 +1,23 @@
+require "text/hatena/auto_link/scheme"
+
+module Text
+  class Hatena
+    class AutoLink
+      class Unbracket < Scheme
+        @@pattern = /\[\](.+?)\[\]/i
+        
+        def patterns
+          [@@pattern]
+        end
+
+        def parse(text, opt = {})
+          if @@pattern =~ text
+            $1
+          else
+            nil
+          end
+        end
+      end
+    end
+  end
+end

data/lib/text/hatena/blockquote_node.rb

@@ -0,0 +1,43 @@
+require "text/hatena/section_node"
+
+module Text
+  class Hatena
+    class BlockquoteNode < SectionNode
+      def init
+        @pattern = /^>((?!<).*)>$/
+        @endpattern = /^<<$/
+        @childnode = %w(h4 h5 blockquote dl list pre superpre table tagline tag)
+        # @startstring = "<blockquote>"
+        @endstring = "</blockquote>"
+      end
+
+      def parse
+        c = @context
+        return unless @pattern =~ c.nextline
+        url = $1
+        c.shiftline
+        t = "\t" * @ilevel
+        _set_child_node_refs
+        startstring = "<blockquote"
+        if url and not url.empty?
+          html = c.autolink.parse('[' << url << ']')
+          if /<a href="([^"]+?)">([^<]+)<\/a>/ =~ html
+            startstring << " title=\"#{$2}\" cite=\"#{$1}\""
+          end
+        end
+        startstring << ">"
+        c.htmllines(t + startstring)
+        while c.hasnext
+          l = c.nextline
+          if @endpattern =~ l
+            c.shiftline
+            break
+          end
+          break unless node = _findnode(l)
+          node.parse
+        end
+        c.htmllines(t + @endstring)
+      end
+    end
+  end
+end

data/lib/text/hatena/body_node.rb

@@ -0,0 +1,18 @@
+require "text/hatena/footnote_node"
+require "text/hatena/node"
+require "text/hatena/section_node"
+
+module Text
+  class Hatena
+    class BodyNode < Node
+      def parse
+        c = @context
+        while c.hasnext
+          node = SectionNode.new({ :context => c,
+                                   :ilevel => @ilevel })
+          node.parse
+        end
+      end
+    end
+  end
+end

data/lib/text/hatena/br_node.rb

@@ -0,0 +1,19 @@
+require "text/hatena/node"
+
+module Text
+  class Hatena
+    class BrNode < Node
+      def parse
+        c = @context
+        l = c.shiftline
+        return unless l.empty?
+        t = "\t" * @ilevel
+        if c.lasthtmlline == "#{t}<br>" or c.lasthtmlline == t
+          c.htmllines("#{t}<br>")
+        else
+          c.htmllines(t)
+        end
+      end
+    end
+  end
+end

data/lib/text/hatena/cdata_node.rb

@@ -0,0 +1,22 @@
+require "text/hatena/node"
+
+module Text
+  class Hatena
+    class CDataNode < Node
+      def parse
+        c = @context
+        l = c.shiftline
+        t = "\t" * @ilevel
+        #if c.lasthtmlline == "#{t}<br>" or c.lasthtmlline == t
+        #  c.htmllines("#{t}<br>")
+        #else
+        #  c.htmllines(t)
+        #end
+        text = Text.new({:context => @context})
+        text.parse(l)
+        l = text.html
+        c.htmllines("#{t}#{l}")
+      end
+    end
+  end
+end

data/lib/text/hatena/context.rb

@@ -0,0 +1,113 @@
+require "text/hatena/auto_link"
+
+module Text
+  class Hatena
+    class Context
+      def initialize(args = {})
+        @text = args[:text]
+        @baseuri = args[:baseuri]
+        @permalink = args[:permalink]
+        @invalidnode = args[:invalidnode]
+        @sectionanchor = args[:sectionanchor]
+        @autolink_option = args[:autolink_option]
+        @texthandler = args[:texthandler]
+        @htmllines = []
+        @html = ""
+        @footnotes = []
+        @sectioncount = 0
+        @syntaxrefs = []
+        @noparagraph = false
+        init
+      end
+
+      def init
+        @text.gsub!(/\r/, "")
+        @lines = @text.split(/\n/)
+        @index = -1
+      end
+
+      def hasnext
+        not @lines[@index + 1].nil?
+      end
+
+      def nextline
+        @lines[@index + 1]
+      end
+
+      def shiftline
+        @lines[@index += 1]
+      end
+
+      def currentline
+        @lines[@index]
+      end
+
+      def html
+        @htmllines.join("\n")
+      end
+
+      def htmllines(arg = nil)
+        @htmllines.push(arg) unless arg.nil?
+        @htmllines
+      end
+
+      def lasthtmlline
+        @htmllines[-1]
+      end
+
+      def footnotes(arg = nil)
+        @footnotes.push(arg) unless arg.nil?
+        @footnotes
+      end
+
+      def syntaxrefs(arg = nil)
+        @syntaxrefs.push(arg) unless arg.nil?
+        @syntaxrefs
+      end
+
+      def syntaxpattern(arg = nil)
+        @syntaxpattern.push(arg) unless arg.nil?
+        @syntaxpattern
+      end
+
+      def noparagraph(*args)
+        @noparagraph = args[0] unless args.empty?
+        @noparagraph
+      end
+
+      def autolink(*args)
+        @autolink = args[0] unless args.empty?
+        @autolink ||= AutoLink.new(@autolink_option)
+        @autolink
+      end
+
+      def sectioncount
+        @sectioncount
+      end
+
+      def incrementsection
+        @sectioncount += 1
+      end
+
+      def baseuri
+        @baseuri
+      end
+
+      def permalink
+        @permalink
+      end
+
+      def invalidnode
+        @invalidnode
+      end
+
+      def sectionanchor
+        @sectionanchor
+      end
+
+      def texthandler
+        @texthandler
+      end
+    end
+  end
+end