doc/xslt/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       @to_delete = []
  30
  31       if customization_layer
  32         @stylesheet = File.expand_path(customization_layer)
  33       else
  34         @stylesheet = STYLESHEET
  35       end
  36
  37       unless File.exist?(@docbook_file)
  38         raise ArgumentError.new("File #{@docbook_file} does not exist")
  39       end
  40     end
  41
  42     def render_to_file(output_file, verbose=false)
  43       render_to_epub(output_file, verbose)
  44       bundle_epub(output_file, verbose)
  45       cleanup_files(@to_delete)
  46     end
  47
  48     def self.invalid?(file)
  49       # Obnoxiously, we can't just check for a non-zero output...
  50       cmd = %Q(#{CHECKER} "#{file}")
  51       output = `#{cmd} 2>&1`
  52
  53       if $?.to_i == 0
  54         return false
  55       else
  56         STDERR.puts output if $DEBUG
  57         return output
  58       end
  59     end
  60
  61     private
  62     def render_to_epub(output_file, verbose)
  63       @collapsed_docbook_file = collapse_docbook()
  64
  65       chunk_quietly =   "--stringparam chunk.quietly " + (verbose ? '0' : '1')
  66       callout_path =    "--stringparam callout.graphics.path #{CALLOUT_PATH}/"
  67       callout_limit =   "--stringparam callout.graphics.number.limit #{CALLOUT_LIMIT}"
  68       callout_ext =     "--stringparam callout.graphics.extension #{CALLOUT_EXT}"
  69       html_stylesheet = "--stringparam html.stylesheet #{File.basename(@css_file)}" if @css_file
  70       base =            "--stringparam base.dir #{OEBPS_DIR}/"
  71       unless @embedded_fonts.empty?
  72         embedded_fonts = @embedded_fonts.map {|f| File.basename(f)}.join(',')
  73         font =            "--stringparam epub.embedded.fonts \"#{embedded_fonts}\""
  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 = %Q(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 = %Q(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       # Double-quote stylesheet & file to help Windows cmd.exe
 119       collapsed_file = File.join(File.expand_path(File.dirname(@docbook_file)),
 120                                  '.collapsed.' + File.basename(@docbook_file))
 121       entity_collapse_command = %Q(xmllint --loaddtd --noent -o "#{collapsed_file}" "#{@docbook_file}")
 122       entity_success = system(entity_collapse_command)
 123       raise "Could not collapse named entites in #{@docbook_file}" unless entity_success
 124
 125       xinclude_collapse_command = %Q(xmllint --xinclude -o "#{collapsed_file}" "#{collapsed_file}")
 126       xinclude_success = system(xinclude_collapse_command)
 127       raise "Could not collapse XIncludes in #{@docbook_file}" unless xinclude_success
 128
 129       @to_delete << collapsed_file
 130       return collapsed_file
 131     end
 132
 133     def copy_callouts
 134       new_callout_images = []
 135       if has_callouts?
 136         calloutglob = "#{CALLOUT_FULL_PATH}/*#{CALLOUT_EXT}"
 137         Dir.glob(calloutglob).each {|img|
 138           img_new_filename = File.join(@oebps_dir, CALLOUT_PATH, File.basename(img))
 139
 140           # TODO: What to rescue for these two?
 141           FileUtils.mkdir_p(File.dirname(img_new_filename))
 142           FileUtils.cp(img, img_new_filename)
 143           @to_delete << img_new_filename
 144           new_callout_images << img
 145         }
 146       end
 147       return new_callout_images
 148     end
 149
 150     def copy_fonts
 151       new_fonts = []
 152       @embedded_fonts.each {|font_file|
 153         font_new_filename = File.join(@oebps_dir, File.basename(font_file))
 154         FileUtils.cp(font_file, font_new_filename)
 155         new_fonts << font_file
 156       }
 157       return new_fonts
 158     end
 159
 160     def copy_csses
 161       if @css_file
 162         css_new_filename = File.join(@oebps_dir, File.basename(@css_file))
 163         FileUtils.cp(@css_file, css_new_filename)
 164       end
 165     end
 166
 167     def copy_images
 168       image_references = get_image_refs()
 169       new_images = []
 170       image_references.each {|img|
 171         # TODO: It'd be cooler if we had a filetype lookup rather than just
 172         # extension
 173         if img =~ /\.(svg|png|gif|jpe?g|xml)/i
 174           img_new_filename = File.join(@oebps_dir, img)
 175           img_full = File.join(File.expand_path(File.dirname(@docbook_file)), img)
 176
 177           # TODO: What to rescue for these two?
 178           FileUtils.mkdir_p(File.dirname(img_new_filename))
 179           puts(img_full + ": " + img_new_filename) if $DEBUG
 180           FileUtils.cp(img_full, img_new_filename)
 181           @to_delete << img_new_filename
 182           new_images << img_full
 183         end
 184       }
 185       return new_images
 186     end
 187
 188     def write_mimetype
 189       mimetype_filename = File.join(@output_dir, "mimetype")
 190       File.open(mimetype_filename, "w") {|f| f.print MIMETYPE}
 191       @to_delete << mimetype_filename
 192       return File.basename(mimetype_filename)
 193     end
 194
 195     def cleanup_files(file_list)
 196       file_list.flatten.each {|f|
 197         # Yikes
 198         FileUtils.rm_r(f, :force => true )
 199       }
 200     end
 201
 202     # Returns an Array of all of the (image) @filerefs in a document
 203     def get_image_refs
 204       parser = REXML::Parsers::PullParser.new(File.new(@collapsed_docbook_file))
 205       image_refs = []
 206       while parser.has_next?
 207         el = parser.pull
 208         if el.start_element? and (el[0] == "imagedata" or el[0] == "graphic")
 209           image_refs << el[1]['fileref']
 210         end
 211       end
 212       return image_refs.uniq
 213     end
 214
 215     # Returns true if the document has code callouts
 216     def has_callouts?
 217       parser = REXML::Parsers::PullParser.new(File.new(@collapsed_docbook_file))
 218       while parser.has_next?
 219         el = parser.pull
 220         if el.start_element? and (el[0] == "calloutlist" or el[0] == "co")
 221           return true
 222         end
 223       end
 224       return false
 225     end
 226   end
 227 end