pink_shirt 0.0.1

data/.gitignore

@@ -0,0 +1,4 @@
+*.gem
+.bundle
+Gemfile.lock
+pkg/*

data/Gemfile

@@ -0,0 +1,4 @@
+source "http://rubygems.org"
+
+# Specify your gem's dependencies in pink_shirt.gemspec
+gemspec

data/Rakefile

@@ -0,0 +1 @@
+require "bundler/gem_tasks"

data/Readme.txt

@@ -0,0 +1,26 @@
+I had a problem: some html needed to be textile
+
+None of the available options could get back to exactly where you started.
+They weren't 'reversable'
+
+I learned a lot by reading other's who went before.
+I've looked at: html2texile https://github.com/thickpaddy/html2textile
+I've looked at: clothred https://github.com/CynicalRyan/ClothRed
+I've looked at: a deeply buried module in humpyard https://github.com/humpyard/humpyard
+
+This library still has some problems with whitespace management, but all the tags convert.
+
+It's based around https://github.com/textile/textile-spec
+and built to produce mirror images of html produced by
+RedCloth https://github.com/jgarber/redcloth
+
+Usage
+============================
+text =  "This is *my* text."
+html = RedCloth.new(text).to_html
+same = PinkShirt.new(html).to_textile
+same #=> "This is *my* text."
+
+
+
+

data/lib/pink_shirt/attributes.rb

@@ -0,0 +1,169 @@
+class PinkShirt
+  # Attributes
+  # -------------------------
+  # the sax parser passes an array of attributes along with each tag
+  # attrs = ['href', 'http://www.example.com', 'style', 'background-color: snake;']
+  #
+  # Textile displays attributes in a certain way
+  # 
+  # USAGE
+  # ---------------------------
+  # Attributes.new(attrs).write
+  #
+  #
+  # colspan=2                 => \2
+  # rowspan=3                 => /3
+  # style='padding-left:1em'  => (
+  # style='padding-right:2em' => ))
+  # style='text-align:right'  => >
+  # style='text-align:left'   => <
+  # style='text-align:center' => =
+  # style='text-align:justify'=> <>
+  # class = 'panthers'        => (panthers)
+  # id = 'banner'             => (#banner)
+  # class='this' id='that'    => (this#that)
+  # lang='fr'                 => [fr]
+  # style='color:red'         => {color:red}
+  
+  
+  class Attributes
+
+    def initialize(attrs)
+      @attrs = attrs
+      @attrs_hash  = Hash[attrs]
+      @styles_hash = parse_styles
+      @nudges  = steal_nudges
+      @padding = steal_padding
+    end
+
+    def attrs
+      @attrs_hash
+    end
+
+    def styles
+      @styles_hash ||= {}
+    end
+
+
+    def write
+      add = []
+      add << colspan
+      add << rowspan
+      add << nudges
+      add << padding
+      add << klass_and_id
+      add << style
+      add << lang
+      out = add.join
+      return nil if out == ""
+      out
+    end
+
+    def parse_styles
+      return nil unless attrs['style']
+      rules_list = attrs['style'].split(";").map{|rule|
+        rule.split(":")
+      }
+      Hash[rules_list]
+    end
+
+    def styles_to_s
+      @styles_hash.map{|k, v| "#{k}:#{v}"}.join(";")
+    end
+
+    def steal_nudges
+
+      return nil unless attrs['style']
+      nudges = []
+      style = attrs['style']
+
+      text_align = case styles.delete('text-align')
+      when 'left'
+        "<"
+      when 'right'
+        ">"
+      when 'center'
+        "="
+      when 'justify'
+        "<>"
+      end
+
+      nudges << text_align
+
+      nudges
+    end
+
+    def steal_padding
+
+      return nil unless attrs['style']
+      left = case styles.delete('padding-left')
+      when '1em'; "(" ;
+      when '2em'; "((";
+      end
+
+      #Alternately, but i'd rather not encourage it
+      # left_count = styles.delete('padding-left').to_i
+      # left = Array.new(left_count).map{"("}.join
+
+      right = case styles.delete('padding-right')
+      when '1em'; ")" ;
+      when '2em'; "))";
+      end
+
+      padding = "#{left}#{right}"
+
+    end
+
+    def colspan
+      return nil unless attrs['colspan']
+      width = attrs['colspan']
+      colspan = ""
+      colspan << "\\" #literal backslash
+      colspan << "#{width}"
+      colspan if width
+    end
+
+    def rowspan
+      return nil unless attrs['rowspan']
+      height = attrs['rowspan']
+      colspan = '/' + "#{height}" if height
+    end
+
+    def nudges
+      @nudges
+    end
+
+    def padding
+      @padding
+    end
+
+
+    def klass_and_id
+      klass = attrs['class']
+      id    = attrs['id']
+      return nil unless klass || id
+      output  = ""
+      output += "("
+      output += klass if klass
+      output += "##{id}" if id
+      output += ")"
+    end
+
+    def style
+      return nil unless styles.length > 0
+
+      output  = ""
+      output += "{"
+      output += styles_to_s
+      output += "}"
+    end
+
+    def lang
+      return nil unless attrs.include?('lang')
+      output  = ""
+      output += "["
+      output += attrs['lang']
+      output += "]"
+    end
+  end
+end

data/lib/pink_shirt/entities.rb

@@ -0,0 +1,32 @@
+class PinkShirt
+  # Gets rid of smart quotes, dashes and some pesky unicode
+  # Usage
+  # ---------------------
+  # Entities.sanitize(input)
+  #
+  class Entities
+    REPLACEMENTS = {
+      8217 => "'", #single quote
+      8216 => "'", #left single quote
+      8220 => '"', #right double quote
+      8221 => '"', #left double quote
+      8211 => "-", #endash
+      8212 => "--", #emdash
+      8230 => "...", #ellipsis
+      215  => "x", #times
+      8242 => "'", #inch
+      174  => "(r)", #registered trademark
+      8482 => "(tm)", # trademark
+      169  => "(c)", #copyright
+    }
+
+
+    def self.sanitize(string)
+      string_chars =  string.unpack("U*")
+      string_chars.map! {|x| REPLACEMENTS[x] ? REPLACEMENTS[x].unpack('U*') : x}
+      string_chars.flatten
+      string =  string_chars.flatten.pack('U*')
+      string
+    end
+  end
+end

data/lib/pink_shirt/flags.rb

@@ -0,0 +1,5 @@
+# Used for moving flags around during processing.
+# Clearly underused
+class PinkShirt::Flags
+  attr_accessor :pre
+end

data/lib/pink_shirt/output.rb

@@ -0,0 +1,64 @@
+class PinkShirt
+  # 
+  # Nokogiri::Sax builds output as a long stream.
+  # Output collects all the writes as an array, and then joins them when required
+  # you can also lock it for writing.
+  # 
+  # Usage
+  # ------------------------
+  # stream = Output.new
+  # stream << 'goods... '
+  # stream.inspect #=> ['goods... ']
+  # stream.lock('suspicious tag')
+  # stream << 'bad stuff'
+  # stream.inspect #=> ['goods... ']
+  # stream.unlock
+  # stream << 'good again'
+  # stream.inspect #=> ['goods... ', 'good again']
+  #
+  class Output
+    def initialize
+      @contents = []
+      @silenced = false
+      @locks = []
+      @ouput = ""
+    end
+
+    def lock(key)
+      @locks << key
+      @silenced = true
+    end
+
+    def unlock
+      @locks.pop
+      @silenced = false if @locks.empty?
+    end
+
+    def << (more)
+      @contents << more unless @silenced
+    end
+
+    def join
+      @output = @contents.join
+    end
+
+    def sanitize
+      @output = PinkShirt::Entities.sanitize(@output)
+    end
+
+    def trim
+      @output = @output.chomp.chomp.chomp
+    end
+
+    def to_s
+      join
+      sanitize
+      trim
+      @output
+    end
+
+    def inspect
+      @contents
+    end
+  end
+end

data/lib/pink_shirt/sax/acronym.rb

@@ -0,0 +1,15 @@
+class PinkShirt
+  class SAX::Acronym < SAX::Base
+    TAGS = %(acronym)
+    def initialize(*args)
+      @current_acronym = []
+      super
+    end
+    def start_acronym attrs
+      @current_acronym << attrs['title']
+    end
+    def end_acronym
+      @output << "(#{@current_acronym.pop})"
+    end
+  end
+end

data/lib/pink_shirt/sax/base.rb

@@ -0,0 +1,20 @@
+class PinkShirt
+  class SAX::Base
+    TAGS = []
+    def initialize(input, flags)
+      @output = input
+      @flags  = flags
+    end
+
+    def add_attributes(attrs)
+      PinkShirt::Attributes.new(attrs).write
+    end
+
+    def to_s
+      @output
+    end
+
+    def method_missing(*args)
+    end
+  end
+end

data/lib/pink_shirt/sax/basic.rb

@@ -0,0 +1,114 @@
+class PinkShirt
+  class SAX::Basic < SAX::Base
+    TAGS = %w(strong b em i cite del ins sub sup span code)
+    attr_accessor :nospan
+    def start_strong attrs
+      @output << "*"
+      @output << add_attributes(attrs) if add_attributes(attrs)
+    end
+
+    alias :start_b :start_strong
+
+    def end_strong
+      @output << "*"
+
+    end
+
+    def start_b attrs
+      @output << "**"
+      @output << add_attributes(attrs) if add_attributes(attrs)
+    end
+
+    def end_b
+      @output << "**"
+    end
+
+    def start_cite attrs
+      @output << "??"
+      @output << add_attributes(attrs) if add_attributes(attrs)
+    end
+
+    def end_cite
+      @output << "??"
+    end
+
+    def start_em attrs
+      @output << "_"
+      @output << add_attributes(attrs) if add_attributes(attrs)
+    end
+
+    def end_em
+      @output << "_"
+    end
+
+    def start_i attrs
+      @output << "__"
+      @output << add_attributes(attrs) if add_attributes(attrs)
+    end
+
+    def end_i
+      @output << "__"
+    end
+
+    def start_del attrs
+      @output << "-"
+      @output << add_attributes(attrs) if add_attributes(attrs)
+    end
+
+    def end_del
+      @output << "-"
+    end
+
+    def start_ins attrs
+      @output << "+"
+      @output << add_attributes(attrs) if add_attributes(attrs)
+    end
+
+    def end_ins
+      @output << "+"
+    end
+
+    def start_sub attrs
+      @output << "~"
+      @output << add_attributes(attrs) if add_attributes(attrs)
+    end
+
+    def end_sub
+      @output << "~"
+    end
+
+    def start_sup attrs
+      @output << "^"
+      @output << add_attributes(attrs) if add_attributes(attrs)
+    end
+
+    def end_sup
+      @output << "^"
+    end
+
+    # redcloth loves to put caps in 'span-caps', it's lame
+    def start_span attrs
+      if attrs['class'] == 'caps'
+        @nospan = true
+      else
+        @output << "%"
+        @output << add_attributes(attrs) if add_attributes(attrs)
+      end
+    end
+
+    def end_span
+      @output << "%" unless @nospan
+      @nospan = false
+    end
+
+    def start_code attrs
+      @output << "@"
+      @output << add_attributes(attrs) if add_attributes(attrs)
+    end
+
+    def end_code
+      @output << "@"
+    end
+
+  end
+end

data/lib/pink_shirt/sax/block_level.rb

@@ -0,0 +1,69 @@
+class PinkShirt
+  class SAX::BlockLevel < SAX::Base
+    TAGS = %w(p br h1 h2 h3 h4 h5 h6 div blockquote)
+    def start_p attrs
+      @output << "p#{add_attributes(attrs)}. " if add_attributes(attrs)
+    end
+
+    def end_p
+      @output << "\n\n"
+    end
+
+    def start_div attrs
+      @output << "div#{add_attributes(attrs)}. "
+    end
+
+    def end_div
+      @output << "\n"
+    end
+
+    def start_blockquote attrs
+      if attrs['cite']
+        @output << "bq.:#{attrs['cite']} "
+      else
+        @output << "bq#{add_attributes(attrs)}. "
+      end
+    end
+
+    def start_br attrs
+    end
+
+    def end_br
+      @output << "\n"
+    end
+
+    def start_h1 attrs
+      @output << "h1#{add_attributes(attrs)}. "
+    end
+
+    def start_h2 attrs
+      @output << "h2#{add_attributes(attrs)}. "
+    end
+
+    def start_h3 attrs
+      @output << "h3#{add_attributes(attrs)}. "
+    end
+
+    def start_h4 attrs
+      @output << "h4#{add_attributes(attrs)}. "
+    end
+
+    def start_h5 attrs
+      @output << "h5#{add_attributes(attrs)}. "
+    end
+
+    def start_h6 attrs
+      @output << "h6#{add_attributes(attrs)}. "
+    end
+
+    def end_h
+      @output << "\n\n"
+    end
+    alias :end_h1 :end_h
+    alias :end_h2 :end_h
+    alias :end_h3 :end_h
+    alias :end_h4 :end_h
+    alias :end_h5 :end_h
+    alias :end_h6 :end_h
+  end
+end

data/lib/pink_shirt/sax/boiler_plate.rb

@@ -0,0 +1,11 @@
+class PinkShirt
+  class SAX::BoilerPlate < SAX::Base
+    TAGS = %w(html body head)
+    def start_html(attrs); end
+    def end_html; end
+    def start_body(attrs); end
+    def end_body; end
+    def start_head(attrs); end
+    def end_head; end
+  end
+end

data/lib/pink_shirt/sax/images.rb

@@ -0,0 +1,17 @@
+class PinkShirt
+  class SAX::Images < SAX::Base
+    TAGS = %(img)
+    def start_img attrs
+      title   = attrs['alt'] || attrs['title']
+      title   = nil if title == "" || title == " "
+      image   = "!"
+      image  += attrs['src']
+      image  += "(#{title})" if title
+      image  += "!"
+      @output << image
+    end
+    def end_img
+
+    end
+  end
+end

data/lib/pink_shirt/sax/links.rb

@@ -0,0 +1,19 @@
+class PinkShirt
+  class SAX::Links < SAX::Base
+    TAGS = %(a)
+    def start_a attrs
+      @link_info = attrs
+      link  = ""
+      link += '"'
+      link += "(#{attrs['class']}). " if attrs['class']
+      @output << link
+    end
+    def end_a
+      link  = ""
+      link += " (#{@link_info['title']})" if @link_info['title']
+      link += '"'
+      link += ":#{@link_info['href']}"
+      @output << link
+    end
+  end
+end

data/lib/pink_shirt/sax/lists.rb

@@ -0,0 +1,67 @@
+class PinkShirt
+  class SAX::Lists < SAX::Base
+    TAGS = %w(ul ol li dl dt dd)
+    def initialize(*args)
+      @last_depth = 0
+      @nesting = []
+      super
+    end
+    def start_ul(attrs)
+      @nesting.push "ul"
+      @in_ul = true
+      @output << "\n" unless @nesting.length == 1
+    end
+
+    def end_ul
+      @nesting.pop
+      @in_ul = false
+    end
+
+    def start_ol(attrs)
+      @nesting.push('ol')
+      @in_ol = true
+      @output << "\n" unless @nesting.length == 1
+    end
+
+    def end_ol
+      @nesting.pop
+      @in_ol = false
+    end
+
+    def start_li(attrs)
+      current_nest = @nesting.last
+      if current_nest == "ol"
+        chr = "#"
+      else
+        chr = "*"
+      end
+      @output << Array.new(@nesting.length).map{chr}.join + " "
+    end
+
+    def end_li
+      @output << "\n" unless @last_depth > @nesting.length
+      @last_depth = @nesting.length
+    end
+
+    def start_dl attrs
+    end
+
+    def end_dl
+    end
+
+    def start_dt attrs
+      @output << "- "
+    end
+
+    def end_dt
+    end
+
+    def start_dd attrs
+      @output << " := "
+    end
+
+    def end_dd
+      @output << "\n"
+    end
+  end
+end

data/lib/pink_shirt/sax/preformatted.rb

@@ -0,0 +1,13 @@
+class PinkShirt
+  
+  class SAX::Preformatted < SAX::Base
+    TAGS = %(pre)
+    def start_pre attrs
+      @flags.pre = true
+      @output << "pre#{add_attributes(attrs)}. "
+    end
+    def end_pre
+      @flags.pre = false
+    end
+  end
+end

data/lib/pink_shirt/sax/script.rb

@@ -0,0 +1,12 @@
+class PinkShirt
+  # One does not textilize the contents of script tags, it is not done.
+  class SAX::Script < SAX::Base
+    TAGS = %(script)
+    def start_script(attrs)
+      @output.lock('script')
+    end
+    def end_script
+      @output.unlock
+    end
+  end
+end