Custom html parser. This was used to convert html files which were exported from an older Adobe InDesign file into markdown. The html was inconsistent and varied but this handled 90 percent of the files. The remaining cases were edited by hand in a form with a markdown enabled text area.

catalog_entry.rb

# renders markdown with footnotes matching Pandoc's syntax,
# but renders html with custom formatting in the class methods link_to_text_replace, link_to_fn_replace
class CatalogEntry

  LINK_TO_FN = "[^%d] "
  LINK_TO_TEXT = "[^%d]: %s"
  CITATION = "%{last}, %{first}%{other}. \"%{title}.\" In _Bits & Pieces Put Together to Present a Semblance of a Whole: Walker Art Center Collections_, edited by Joan Rothfuss and Elizabeth Carpenter. Minneapolis, MN: Walker Art Center, 2005."

  FILE_TEMPLATE = "B&Pcat-%{last},%{first}_2005"
  FILE_NAME_REGEX = /B&Pcat-(.+),(.+)_2005.*/
  P = "\n \n"

  attr_accessor :source, :source_file_name, :artist_name, :sidebar, :author_string, :title

  def initialize opts={}
    return unless opts[:path]
    self.sidebar = false
    format = opts[:format] || File.extname(opts[:path])
    @source_file_name = File.basename(opts[:path]).chomp(format)
    if opts[:path] and format == '.html'
      @source = Nokogiri::HTML(open(opts[:path]))
    end
  end


  def body
    process_body + "

<div class=\"footnotes\">
  <ol>
    " + process_notes + "
  </ol>
</div>

"
  end

  def author
    self.author_string ||= begin
                             text = @source.css('span.author-s-name').first.try :text
                             if text.nil?
                               text = @source.css('span.sidebar-normal').last.try :text
                               if AuthorMap.keys.include? text
                                 AuthorMap[text]
                               elsif AuthorMap.values.include? text
                                 text
                               end
                             else
                               process_author(text)
                             end
                           end
  end

  def process_author text
    authors = text.split(/ with | and /)
    authors.collect{ |a| AuthorMap[a.try(:strip)] || a || '' }.join(', ')
  end

  def notes
    process_notes
  end

  # may or may not exist
  def title
    if sidebar?
      @source.css('span.sidebar-normal').first.try :text
    else
      artist_name
    end
  end

  def sidebar?
    @source.css('span.sidebar-normal').present? || /_ADD/.match(@source_file_name).present?
  end

  def citation
    return unless author.present?
    first, other = author.split(',')
    a = first.split(' ')
    last = a.pop
    first_middle = a.join(' ').chomp('.') # remove . in middle abbrev to avoid ..
    CITATION % {last: last, first: first_middle, title: title, other: (", #{other}" if other)}
  end


  def artist_name
    @artist_name ||= process_artist_name
  end

  def valid?
    @source and @source.validate
  end


  private


  def process_notes
    @source.css('p').collect do |el|
      next unless el.css('.notes').present?
      next if el.text == "Notes"
      number, text = *extract_footnote(process(el))
      format_footnote(number,text) if number
    end.compact.join(P)
  end


  def process_artist_name
    matchdata, last, first = *FILE_NAME_REGEX.match(source_file_name)
    [first,last].join(' ')
  end

  PROCESSER = Hash.new(->(key){ ->(el){ el.text }}).merge({
    'body-copy'           => ->(el){ el.text },
    'notes'               => ->(el){ el.text },
    'sidebar-normal'      => ->(el){ el.text },
    'body-copy-italic'    => ->(el){ el.text.sub(/\A(\s*)/){ $1.to_s+'_'}.sub(/(\s*)\Z/){ '_'+$1.to_s} }, # retain space within the element
    'notes-italic'        => ->(el){ '<em>'+el.text+'</em>' }, #unfortunate hack html within html
    'body-copy-footnotes' => ->(el){ LINK_TO_FN % el.text },
    'author-s-name'       => ->(el){ raise "BODY END" },
  })

  def process_body
    raise 'Invalid Content' unless valid?
    story = @source.css('.story p.body-copy')
    if sidebar?
      self.sidebar = true
      self.title = story.shift
    end
    story.collect do |el|
      unless el.css('.author-s-name').present? || el.css('.notes').present?
        process(el)
      end
    end.compact.join(P)
  end

  def process el
    el.children.collect{|e| PROCESSER[e[:class]].call(e) }.join('')
  end

  def extract_footnote note_text
    matchdata, number, text =  */(\d+)\. (.*)/.match(note_text)
    [number,text] unless number.nil?
  end

  def format_footnote number, text
    LINK_TO_TEXT % [number, text]
  end

  # id = ref
  def self.link_to_fn_replace html, prefix
    html.gsub(/\[\^(\d+)\](\s+)/) do
      id   = 'fnref' + prefix + "-" + $1.to_s
      href = '#fn'   + prefix + "-" + $1.to_s
      "<sup><a href=\"#{href}\" id=\"#{id}\" class=\"footnoteRef\" rel=\"footnote\">#{$1}</a></sup>#{$2}"
    end
  end

  # id = fn
  def self.link_to_text_replace html, prefix
    html.split("\n").collect do |line|
      line.sub(/(\s)*\[\^(\d+)\]\: (.*)/) do
        href = '#fnref' + prefix + '-' + $2.to_s
        id   = 'fn'     + prefix + '-' + $2.to_s
        "<li id=\"#{id}\"><p>#{$3} <a href=\"#{href}\" class=\"jump\" rev=\"footnote\">&#8617;</a></p></li>"
      end
    end.join("\n")
  end

  AuthorMap = {
    'A.B.'     => 'Andrew Blauvelt',
    'D.B.'     => 'Doug Benidt',
    'D.C.'     => 'Doryun Chong',
    'D.F.'     => 'Douglas Fogle',
    'D.K.'     => 'Diana Kim',
    'D.O.'     => 'Dean Otto',
    'D.S.'     => 'Daniel Smith',
    'E.C.'     => 'Elizabeth Carpenter',
    'J.C.P.'   => 'Jennifer Case Phelps',
    'J.R.'     => 'Joan Rothfuss',
    'J.V.'     => 'Jill Vetter',
    'L.D.'     => 'Lynn Dierks',
    'M.A.'     => 'Max Andrews',
    'O.I.'     => 'Olukemi Ilesanmi',
    'P.B.'     => 'Philip Bither',
    'P.V.'     => 'Philippe Vergne',
    'R.F.'     => 'Richard Flood',
    'R.Furtak' => 'Rosemary Furtak',
    'S.E.'     => 'Siri Engberg',
    'S.M.'     => 'Sheryl Mously',
  }

end