stylesheets/evergreen_docbook_files/docbook-xsl-1.75.2/epub/bin/lib/docbook.rb

   1 require 'fileutils'
   2 require 'rexml/parsers/pullparser'
   3
   4 module DocBook
   5
   6   class Epub
   7     CHECKER = "epubcheck"
   8     STYLESHEET = File.expand_path(File.join(File.dirname(__FILE__), '..', '..', "docbook.xsl"))
   9     CALLOUT_PATH = File.join('images', 'callouts')
  10     CALLOUT_FULL_PATH = File.expand_path(File.join(File.dirname(__FILE__), '..', '..', '..', CALLOUT_PATH))
  11     CALLOUT_LIMIT = 15
  12     CALLOUT_EXT = ".png"
  13     XSLT_PROCESSOR = "xsltproc"
  14     OUTPUT_DIR = ".epubtmp#{Time.now.to_f.to_s}"
  15     MIMETYPE = "application/epub+zip"
  16     META_DIR = "META-INF"
  17     OEBPS_DIR = "OEBPS"
  18     ZIPPER = "zip"
  19
  20     attr_reader :output_dir
  21
  22     def initialize(docbook_file, output_dir=OUTPUT_DIR, css_file=nil, customization_layer=nil, embedded_fonts=[])
  23       @docbook_file = docbook_file
  24       @output_dir = output_dir
  25       @meta_dir  = File.join(@output_dir, META_DIR)
  26       @oebps_dir = File.join(@output_dir, OEBPS_DIR)
  27       @css_file = css_file ? File.expand_path(css_file) : css_file
  28       @embedded_fonts = embedded_fonts
  29       raise NotImplementedError if @embedded_fonts.length > 1
  30       @to_delete = []
  31
  32       if customization_layer
  33         @stylesheet = File.expand_path(customization_layer)
  34       else
  35         @stylesheet = STYLESHEET
  36       end
  37
  38       unless File.exist?(@docbook_file)
  39         raise ArgumentError.new("File #{@docbook_file} does not exist")
  40       end
  41     end
  42
  43     def render_to_file(output_file, verbose=false)
  44       render_to_epub(output_file, verbose)
  45       bundle_epub(output_file, verbose)
  46       cleanup_files(@to_delete)
  47     end
  48
  49     def self.invalid?(file)
  50       # Obnoxiously, we can't just check for a non-zero output...
  51       cmd = "#{CHECKER} #{file}"
  52       output = `#{cmd} 2>&1`
  53
  54       if $?.to_i == 0
  55         return false
  56       else
  57         STDERR.puts output if $DEBUG
  58         return output
  59       end
  60     end
  61
  62     private
  63     def render_to_epub(output_file, verbose)
  64       @collapsed_docbook_file = collapse_docbook()
  65
  66       chunk_quietly =   "--stringparam chunk.quietly " + (verbose ? '0' : '1')
  67       callout_path =    "--stringparam callout.graphics.path #{CALLOUT_PATH}/"
  68       callout_limit =   "--stringparam callout.graphics.number.limit #{CALLOUT_LIMIT}"
  69       callout_ext =     "--stringparam callout.graphics.extension #{CALLOUT_EXT}"
  70       html_stylesheet = "--stringparam html.stylesheet #{File.basename(@css_file)}" if @css_file
  71       base =            "--stringparam base.dir #{OEBPS_DIR}/"
  72       unless @embedded_fonts.empty?
  73         font =            "--stringparam epub.embedded.font \"#{File.basename(@embedded_fonts.first)}\""
  74       end
  75       meta =            "--stringparam epub.metainf.dir #{META_DIR}/"
  76       oebps =           "--stringparam epub.oebps.dir #{OEBPS_DIR}/"
  77       options = [chunk_quietly,
  78                  callout_path,
  79                  callout_limit,
  80                  callout_ext,
  81                  base,
  82                  font,
  83                  meta,
  84                  oebps,
  85                  html_stylesheet,
  86                 ].join(" ")
  87       # Double-quote stylesheet & file to help Windows cmd.exe
  88       db2epub_cmd = "cd #{@output_dir} && #{XSLT_PROCESSOR} #{options} \"#{@stylesheet}\" \"#{@collapsed_docbook_file}\""
  89       STDERR.puts db2epub_cmd if $DEBUG
  90       success = system(db2epub_cmd)
  91       raise "Could not render as .epub to #{output_file} (#{db2epub_cmd})" unless success
  92       @to_delete << Dir["#{@meta_dir}/*"]
  93       @to_delete << Dir["#{@oebps_dir}/*"]
  94     end
  95
  96     def bundle_epub(output_file, verbose)
  97
  98       quiet = verbose ? "" : "-q"
  99       mimetype_filename = write_mimetype()
 100       meta   = File.basename(@meta_dir)
 101       oebps  = File.basename(@oebps_dir)
 102       images = copy_images()
 103       csses  = copy_csses()
 104       fonts  = copy_fonts()
 105       callouts = copy_callouts()
 106       # zip -X -r ../book.epub mimetype META-INF OEBPS
 107       # Double-quote stylesheet & file to help Windows cmd.exe
 108       zip_cmd = "cd \"#{@output_dir}\" &&  #{ZIPPER} #{quiet} -X -r  \"#{File.expand_path(output_file)}\" \"#{mimetype_filename}\" \"#{meta}\" \"#{oebps}\""
 109       puts zip_cmd if $DEBUG
 110       success = system(zip_cmd)
 111       raise "Could not bundle into .epub file to #{output_file}" unless success
 112     end
 113
 114     # Input must be collapsed because REXML couldn't find figures in files that
 115     # were XIncluded or added by ENTITY
 116     #   http://sourceforge.net/tracker/?func=detail&aid=2750442&group_id=21935&atid=373747
 117     def collapse_docbook
 118       collapsed_file = File.join(File.expand_path(File.dirname(@docbook_file)),
 119                                  '.collapsed.' + File.basename(@docbook_file))
 120       entity_collapse_command = "xmllint --loaddtd --noent -o '#{collapsed_file}' '#{@docbook_file}'"
 121       entity_success = system(entity_collapse_command)
 122       raise "Could not collapse named entites in #{@docbook_file}" unless entity_success
 123
 124       xinclude_collapse_command = "xmllint --xinclude -o '#{collapsed_file}' '#{collapsed_file}'"
 125       xinclude_success = system(xinclude_collapse_command)
 126       raise "Could not collapse XIncludes in #{@docbook_file}" unless xinclude_success
 127
 128       @to_delete << collapsed_file
 129       return collapsed_file
 130     end
 131
 132     def copy_callouts
 133       new_callout_images = []
 134       if has_callouts?
 135         calloutglob = "#{CALLOUT_FULL_PATH}/*#{CALLOUT_EXT}"
 136         Dir.glob(calloutglob).each {|img|
 137           img_new_filename = File.join(@oebps_dir, CALLOUT_PATH, File.basename(img))
 138
 139           # TODO: What to rescue for these two?
 140           FileUtils.mkdir_p(File.dirname(img_new_filename))
 141           FileUtils.cp(img, img_new_filename)
 142           @to_delete << img_new_filename
 143           new_callout_images << img
 144         }
 145       end
 146       return new_callout_images
 147     end
 148
 149     def copy_fonts
 150       new_fonts = []
 151       @embedded_fonts.each {|font_file|
 152         font_new_filename = File.join(@oebps_dir, File.basename(font_file))
 153         FileUtils.cp(font_file, font_new_filename)
 154         new_fonts << font_file
 155       }
 156       return new_fonts
 157     end
 158
 159     def copy_csses
 160       if @css_file
 161         css_new_filename = File.join(@oebps_dir, File.basename(@css_file))
 162         FileUtils.cp(@css_file, css_new_filename)
 163       end
 164     end
 165
 166     def copy_images
 167       image_references = get_image_refs()
 168       new_images = []
 169       image_references.each {|img|
 170         # TODO: It'd be cooler if we had a filetype lookup rather than just
 171         # extension
 172         if img =~ /\.(svg|png|gif|jpe?g|xml)/i
 173           img_new_filename = File.join(@oebps_dir, img)
 174           img_full = File.join(File.expand_path(File.dirname(@docbook_file)), img)
 175
 176           # TODO: What to rescue for these two?
 177           FileUtils.mkdir_p(File.dirname(img_new_filename))
 178           puts(img_full + ": " + img_new_filename) if $DEBUG
 179           FileUtils.cp(img_full, img_new_filename)
 180           @to_delete << img_new_filename
 181           new_images << img_full
 182         end
 183       }
 184       return new_images
 185     end
 186
 187     def write_mimetype
 188       mimetype_filename = File.join(@output_dir, "mimetype")
 189       File.open(mimetype_filename, "w") {|f| f.print MIMETYPE}
 190       @to_delete << mimetype_filename
 191       return File.basename(mimetype_filename)
 192     end
 193
 194     def cleanup_files(file_list)
 195       file_list.flatten.each {|f|
 196         # Yikes
 197         FileUtils.rm_r(f, :force => true )
 198       }
 199     end
 200
 201     # Returns an Array of all of the (image) @filerefs in a document
 202     def get_image_refs
 203       parser = REXML::Parsers::PullParser.new(File.new(@collapsed_docbook_file))
 204       image_refs = []
 205       while parser.has_next?
 206         el = parser.pull
 207         if el.start_element? and (el[0] == "imagedata" or el[0] == "graphic")
 208           image_refs << el[1]['fileref']
 209         end
 210       end
 211       return image_refs
 212     end
 213
 214     # Returns true if the document has code callouts
 215     def has_callouts?
 216       parser = REXML::Parsers::PullParser.new(File.new(@collapsed_docbook_file))
 217       while parser.has_next?
 218         el = parser.pull
 219         if el.start_element? and (el[0] == "calloutlist" or el[0] == "co")
 220           return true
 221         end
 222       end
 223       return false
 224     end
 225   end
 226 end