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.last_element_child.after(<<'XML')
  34 <copyright>
  35     <year>2017</year>
  36     <holder>Creative Commons</holder>
  37 </copyright>
  38 XML
  39   node.last_element_child.after(<<'XML')
  40 <publisher>
  41     <publishername>Gunnar Wolf</publishername>
  42     <address><city>Mexico City</city></address>
  43 </publisher>
  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 if ! xml.css('dedication title').empty?
  89   xml.css('dedication title')[0].content = ""
  90 end
  91
  92 log 0, 'move legal notice to bookinfo'
  93 xml.css('book bookinfo')[0].last_element_child.after('<legalnotice>')
  94 ln = xml.css('book bookinfo legalnotice')[0]
  95 xml.css('para').each do |para|
  96   if para.content =~ /This book is published under a/
  97     log 0, 'found legal'
  98     para.parent = ln
  99     break
 100   end
 101 end
 102
 103 log 0, 'replace colophon page with one for this edition'
 104  xml.xpath('//colophon/para').remove
 105 s = xml.xpath('//colophon')[0]
 106 s.first_element_child.after(<<'XML')
 107 <para>Made with Creative Commons</para>
 108
 109 <para>by Paul Stacey & Sarah Hinchliff Pearson</para>
 110
 111 <para>© 2017 by the Creative Commons Foundation.</para>
 112
 113 <para>Published under a Creative Commons Attribution-ShareAlike
 114 license (CC BY-SA), version 4.0.</para>
 115
 116 <para>ISBN: YET-TO-BE-DECIDED (PDF), YET-TO-BE-DECIDED (ePub),
 117 YET-TO-BE-DECIDED (Paperback) </para>
 118
 119 <para>Illustrations by Bryan Mathers,
 120 <ulink url="https://bryanmmathers.com/"/>.</para>
 121
 122 <para>Publisher: Gunnar Wolf.</para>
 123
 124 <para>
 125 <!--space for information about translators-->
 126 &nbsp;
 127 </para>
 128
 129 <para>Downloadable e-book available at
 130 <ulink url="https://madewith.cc/"/>.</para>
 131
 132 <para>This book is published under a CC BY-SA license, which means that you
 133 can copy, redistribute, remix, transform, and build upon the content for
 134 any purpose, even commercially, as long as you give appropriate credit,
 135 provide a link to the license, and indicate if changes were made. If you
 136 remix, transform, or build upon the material, you must distribute your
 137 contributions under the same license as the original. License details:
 138 <ulink url="http://creativecommons.org/licenses/by-sa/4.0/"/></para>
 139
 140 <para>Made With Creative Commons is published with the kind support of
 141 Creative Commons and backers of our crowdfunding-campaign on the
 142 Kickstarter.com platform.</para>
 143
 144 <para>This edition of the book is maintained on
 145 <ulink url="https://gitlab.com/gunnarwolf/madewithcc-es/"/>, and the
 146 translations are maintained on
 147 <ulink url="https://hosted.weblate.org/projects/madewithcc/"/>.  If
 148 you find any error in the book, please let us know via Gitlab or Weblate.</para>
 149
 150 <para>
 151 Classifications:
 152 </para>
 153
 154 <para>
 155 (Dewey) 346.048, 347.78
 156 </para>
 157
 158 <para>
 159 (UDK) ?
 160 </para>
 161
 162 <para>
 163 (US Library of Congress) Z286 O63 S73 2017
 164 </para>
 165
 166 <para>
 167 (Melvil) 025.523
 168 </para>
 169
 170 <para>
 171 (ACM CRCS) ?
 172 </para>
 173
 174 XML
 175
 176 log 0, 'remove title from colophon'
 177 xml.css('colophon title')[0].content = ""
 178
 179 log 0, 'change CC logo images to informalfigure'
 180 xml.css('figure mediaobject imageobject imagedata[width="40.0%"]').each do |id|
 181   f = id.parent.parent.parent
 182   f.name = 'informalfigure'
 183 end
 184
 185 log 0, 'assigning IDs to formal figures'
 186 seq = 1
 187 xml.css('figure').each do |fig|
 188   fig['id'] = 'fig-%d' % seq
 189   seq = seq + 1
 190 end
 191
 192 # Disabled as dblatex do not understand chapter/chapterinfo/author,
 193 # see <URL: https://bugs.debian.org/891183 >.
 194 if false
 195 log 0, 'migrate chapter author into <chapterinfo> where relevant'
 196 xml.css('chapter para').each do |para|
 197   if para.content =~ /^\s*((Paul|Sarah Hinchliff) (Stacey|Pearson))\s*$/
 198     log 1, 'migrated %s %s' % [$2, $3]
 199     para.parent.css('title')[0].before(<<'XML' % [$2, $3])
 200       <chapterinfo>
 201         <author>
 202           <firstname>%s</firstname><surname>%s</surname>
 203         </author>
 204       </chapterinfo>
 205 XML
 206     para.remove
 207   end
 208 end
 209 end
 210
 211 log 0, 'Writing processed file'
 212 # Unable to figure out API way to replace DOCTYPE
 213 data = xml.to_xml().gsub!(/DOCTYPE article/, 'DOCTYPE book')
 214 File.open(dstfile, 'w') {|f| f.write(data)}