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').select {|p| p.attributes['id'].value == partid}.each do |part|
  25     part.name = tag
  26   end
  27 end
  28
  29
  30 log 0, 'replace article* with book*'
  31 xml.children[0].name = 'book'
  32 xml.children[1].name = 'book'
  33 xml.css('info').each do |node|
  34   node.last_element_child.after(<<'XML')
  35 <copyright>
  36     <year>2017</year>
  37     <holder>Creative Commons</holder>
  38 </copyright>
  39 XML
  40   node.last_element_child.after(<<'XML')
  41 <publisher>
  42     <publishername>Gunnar Wolf</publishername>
  43     <address><city>Mexico City</city></address>
  44 </publisher>
  45 XML
  46 end
  47 xml.css('article').each do |node|
  48   node.name = 'book'
  49   node['lang'] = 'en'
  50 end
  51
  52 log 0, 'change parts to colophon, dedication and chapter'
  53 partreplace(xml, 'colophon', 'colophon')
  54 partreplace(xml, 'dedication', 'dedication')
  55 partreplace(xml, 'foreword', 'preface')
  56 partreplace(xml, 'introduction', 'preface')
  57 partreplace(xml, 'bibliography', 'chapter')
  58 partreplace(xml, 'acknowledgments', 'chapter')
  59
  60 log 0, 'place part introduction into <partintro>'
  61 s = xml.search('part title').select {|tit| tit.content =~ /The Case Studies/}[0]
  62
  63 if s
  64   s.after('<partintro>')
  65   p = xml.css('part partintro')[0]
  66   s.parent.children.select{|c| c.name == 'para'}.each do |node|
  67     node.parent = p
  68   end
  69 else
  70   log 0, 'Adding <partintro> failed!'
  71 end
  72
  73
  74 log 0, 'remove empty notes/web links sections'
  75 [
  76   'Notes',
  77   'Web links',
  78   'Web link',
  79 ].each do |title|
  80   xml.search('title').select  {|t| t.content == title}.each do |node|
  81     p = node.parent
  82     node.remove
  83     if p.content =~ /^\s*$/
  84       p.remove
  85     else
  86       raise RuntimeError, 'Non-empty «%s» found' % title
  87     end
  88   end
  89 end
  90
  91 log 0, 'remove title from dedication'
  92 xml.css('dedication title')[0].content = ""
  93
  94 log 0, 'move legal notice to bookinfo'
  95 xml.css('book info')[0].last_element_child.after('<legalnotice>')
  96 ln = xml.css('book info legalnotice')[0]
  97 xml.css('para').each do |para|
  98   if para.content =~ /This book is published under a/
  99     log 0, 'found legal'
 100     para.parent = ln
 101     break
 102   end
 103 end
 104
 105 log 0, 'replace colophon page with one for this edition'
 106  xml.search('colophon para').map {|p| p.remove}
 107 s = xml.search('colophon')[0]
 108 s.first_element_child.after(<<'XML')
 109 <para>Made with Creative Commons</para>
 110
 111 <para>by Paul Stacey & Sarah Hinchliff Pearson</para>
 112
 113 <para>© 2017 by the Creative Commons Foundation.</para>
 114
 115 <para>Published under a Creative Commons Attribution-ShareAlike
 116 license (CC BY-SA), version 4.0.</para>
 117
 118 <para>ISBN: YET-TO-BE-DECIDED (PDF), YET-TO-BE-DECIDED (ePub),
 119 YET-TO-BE-DECIDED (Paperback) </para>
 120
 121 <para>Illustrations by Bryan Mathers,
 122 <ulink url="https://bryanmmathers.com/"/>.</para>
 123
 124 <para>Publisher: Gunnar Wolf.</para>
 125
 126 <para>
 127 <!--space for information about translators-->
 128 &nbsp;
 129 </para>
 130
 131 <para>Downloadable e-book available at
 132 <ulink url="https://madewith.cc/"/>.</para>
 133
 134 <para>This book is published under a CC BY-SA license, which means that you
 135 can copy, redistribute, remix, transform, and build upon the content for
 136 any purpose, even commercially, as long as you give appropriate credit,
 137 provide a link to the license, and indicate if changes were made. If you
 138 remix, transform, or build upon the material, you must distribute your
 139 contributions under the same license as the original. License details:
 140 <ulink url="http://creativecommons.org/licenses/by-sa/4.0/"/></para>
 141
 142 <para>Made With Creative Commons is published with the kind support of
 143 Creative Commons and backers of our crowdfunding-campaign on the
 144 Kickstarter.com platform.</para>
 145
 146 <para>This edition of the book is maintained on
 147 <ulink url="https://gitlab.com/gunnarwolf/madewithcc-es/"/>, and the
 148 translations are maintained on
 149 <ulink url="https://hosted.weblate.org/projects/madewithcc/"/>.  If
 150 you find any error in the book, please let us know via gitlab.</para>
 151
 152 <para>
 153 Classifications:
 154 </para>
 155
 156 <para>
 157 (Dewey) 346.048, 347.78
 158 </para>
 159
 160 <para>
 161 (UDK) ?
 162 </para>
 163
 164 <para>
 165 (US Library of Congress) Z286 O63 S73 2017
 166 </para>
 167
 168 <para>
 169 (Melvil) 025.523
 170 </para>
 171
 172 <para>
 173 (ACM CRCS) ?
 174 </para>
 175
 176 XML
 177
 178 log 0, 'remove title from colophon'
 179 xml.css('colophon title')[0].content = ""
 180
 181 log 0, 'change CC logo images to informalfigure'
 182 xml.css('figure mediaobject imageobject imagedata[width="40.0%"]').each do |id|
 183   f = id.parent.parent.parent
 184   f.name = 'informalfigure'
 185 end
 186
 187 log 0, 'assigning IDs to formal figures'
 188 seq = 1
 189 xml.css('figure').each do |fig|
 190   fig['id'] = 'fig-%d' % seq
 191   seq = seq + 1
 192 end
 193
 194 # Disabled as dblatex do not understand chapter/chapterinfo/author,
 195 # see <URL: https://bugs.debian.org/891183 >.
 196 if false
 197 log 0, 'migrate chapter author into <chapterinfo> where relevant'
 198 xml.css('chapter para').each do |para|
 199   if para.content =~ /^\s*((Paul|Sarah Hinchliff) (Stacey|Pearson))\s*$/
 200     log 1, 'migrated %s %s' % [$2, $3]
 201     para.parent.css('title')[0].before(<<'XML' % [$2, $3])
 202       <chapterinfo>
 203         <author>
 204           <firstname>%s</firstname><surname>%s</surname>
 205         </author>
 206       </chapterinfo>
 207 XML
 208     para.remove
 209   end
 210 end
 211 end
 212
 213 log 0, 'Writing processed file (%s)' % dstfile
 214 File.open(dstfile, 'w') {|f| f.write(xml.to_xml())}