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