fixup-docbook.rb

   1 #!/usr/bin/ruby
   2 # coding: utf-8
   3
   4 require 'nokogiri'
   5
   6 LogLevel=1
   7 raise ArgumentError, 'Source/destination files not specified' if ARGV.size != 2
   8 srcfile = ARGV[0]
   9 dstfile = ARGV[1]
  10
  11 f=File.open(srcfile)
  12 xml = Nokogiri::XML(f)
  13
  14 def log(level,what)
  15   indicators = %w(! • # -)
  16   if level >= LogLevel
  17     print indicators[level]
  18   else
  19     print "\n%s %s\n" % ['*' * (level+1), what]
  20   end
  21 end
  22
  23 def partreplace(xml, partid, tag)
  24   xml.css('part[id=' + partid + ']').each do |part|
  25     part.name = tag
  26   end
  27 end
  28
  29
  30 log 0, 'replace article* with book*'
  31 xml.css('articleinfo').each do |node|
  32   node.name = 'bookinfo'
  33   node.first_element_child.before(<<'XML')
  34 <publisher>
  35     <publishername>Gunnar Wolf</publishername>
  36     <address><city>Mexico City</city></address>
  37 </publisher>
  38 XML
  39   node.first_element_child.before(<<'XML')
  40 <copyright>
  41     <year>2017</year>
  42     <holder>Creative Commons</holder>
  43 </copyright>
  44 XML
  45 end
  46 xml.css('article').each do |node|
  47   node.name = 'book'
  48   node['lang'] = 'en'
  49 end
  50
  51 log 0, 'change parts to colophon, dedication and chapter'
  52 partreplace(xml, 'colophon', 'colophon')
  53 partreplace(xml, 'dedication', 'dedication')
  54 partreplace(xml, 'foreword', 'preface')
  55 partreplace(xml, 'introduction', 'preface')
  56 partreplace(xml, 'bibliography', 'chapter')
  57 partreplace(xml, 'acknowledgments', 'chapter')
  58
  59 log 0, 'place part introduction into <partintro>'
  60 s = xml.xpath("//part/title[text()='The Case Studies']")[0]
  61 if s
  62   s.after('<partintro>')
  63   p = xml.css('part partintro')[0]
  64   s.parent.xpath("//part/para").each do |node|
  65     node.parent = p
  66   end
  67 end
  68
  69
  70 log 0, 'remove empty notes/web links sections'
  71 [
  72   'Notes',
  73   'Web links',
  74   'Web link',
  75 ].each do |title|
  76   xml.xpath("//title[text()='%s']" % title).each do |node|
  77     p = node.parent
  78     node.remove
  79     if p.content =~ /^\s*$/
  80       p.remove
  81     else
  82       raise RuntimeError, 'Non-empty «%s» found' % title
  83     end
  84   end
  85 end
  86
  87 log 0, 'remove title from dedication'
  88 xml.css('dedication title')[0].content = ""
  89
  90 log 0, 'move legal notice to bookinfo'
  91 xml.css('book bookinfo')[0].first_element_child.before('<legalnotice>')
  92 ln = xml.css('book bookinfo legalnotice')[0]
  93 xml.css('para').each do |para|
  94   if para.content =~ /This book is published under a/
  95     log 0, 'found legal'
  96     para.parent = ln
  97     break
  98   end
  99 end
 100
 101 log 0, 'replace colophon page with one for this edition'
 102  xml.xpath('//colophon/para').remove
 103 s = xml.xpath('//colophon')[0]
 104 s.first_element_child.after(<<'XML')
 105 <para>Made with Creative Commons</para>
 106
 107 <para>by Paul Stacey & Sarah Hinchliff Pearson</para>
 108
 109 <para>© 2017 by the Creative Commons Foundation.</para>
 110
 111 <para>Published under a Creative Commons Attribution-ShareAlike
 112 license (CC BY-SA), version 4.0.</para>
 113
 114 <para>ISBN: YET-TO-BE-DECIDED (PDF), YET-TO-BE-DECIDED (ePub),
 115 YET-TO-BE-DECIDED (Paperback) </para>
 116
 117 <para>Illustrations by Bryan Mathers, <ulink url="https://bryanmathers.com/"/></para>
 118
 119 <para>Publisher: Gunnar Wolf.</para>
 120
 121 <!-- space for information about translators -->
 122 <para>&nbsp;</para>
 123
 124 <para>Downloadable e-book available at
 125 <ulink url="https://madewith.cc/"/></para>
 126
 127 <para>This book is published under a CC BY-SA license, which means that you
 128 can copy, redistribute, remix, transform, and build upon the content for
 129 any purpose, even commercially, as long as you give appropriate credit,
 130 provide a link to the license, and indicate if changes were made. If you
 131 remix, transform, or build upon the material, you must distribute your
 132 contributions under the same license as the original. License details:
 133 <ulink url="http://creativecommons.org/licenses/by-sa/4.0/"/></para>
 134
 135 <para>Made With Creative Commons is published with the kind support of
 136 Creative Commons and backers of our crowdfunding-campaign on the
 137 Kickstarter.com platform.</para>
 138
 139 XML
 140
 141 log 0, 'remove title from colophon'
 142 xml.css('colophon title')[0].content = ""
 143
 144 log 0, 'change CC logo images to informalfigure'
 145 xml.css('figure mediaobject imageobject imagedata[width="40.0%"]').each do |id|
 146   f = id.parent.parent.parent
 147   f.name = 'informalfigure'
 148 end
 149
 150 log 0, 'assigning IDs to formal figures'
 151 seq = 1
 152 xml.css('figure').each do |fig|
 153   fig['id'] = 'fig-%d' % seq
 154   seq = seq + 1
 155 end
 156
 157 log 0, 'Writing processed file'
 158 # Unable to figure out API way to replace DOCTYPE
 159 data = xml.to_xml().gsub!(/DOCTYPE article/, 'DOCTYPE book')
 160 File.open(dstfile, 'w') {|f| f.write(data)}