maint/htagsfix

   1 #!/usr/bin/env ruby
   2
   3 #
   4 # Fixes htags' output so that files have fixed names. E.g., S/542.html -> D/whatever.c.h.
   5 #
   6 # It addresses the following problem:
   7 #
   8 #    "Can htags create permanent addresses?"
   9 #    http://lists.gnu.org/archive/html/help-global/2015-11/msg00002.html
  10 #
  11
  12 require 'fileutils'
  13 require 'pp'
  14
  15 Encoding.default_external = "BINARY"
  16
  17 $base = '.'  # Where the HTML is stored.
  18 $verbose = false
  19 $debug = false
  20 $create_subdirs = false  # Experimental feature. Doesn't really work.
  21
  22 # Bonus: highlight the target line.
  23 $css = "
  24 a:target {
  25   position: absolute;
  26   left: 0;
  27   height: 1.35em;
  28   width: 100%;
  29   background: yellow;
  30   opacity: .5;
  31   z-index: -1000;
  32 }
  33 "
  34
  35 #
  36 # Computes pretty names for files:
  37 #
  38 #    '.../S/456.html' => 'S/modules/tty.c.html'
  39 #
  40 def prettyname(full_path, prefix, kind, nid, ext)
  41   gist = nid
  42   start = IO.read(full_path, 1024)
  43   if start =~ /<title>(.*?)<\/title>/ then
  44     gist = $1
  45     gist.gsub!(%r{/+$}, '')  # remove trailing '/'s.
  46     if not $create_subdirs then
  47       gist.gsub!('/', '--')
  48     end
  49     gist.gsub!(%r{[^a-zA-Z0-9_.,/-]}, '-')  # clean special characters.
  50   end
  51   return kind + '/' + gist + ext
  52 end
  53
  54 class Counter
  55   class << self
  56     def init(total)
  57       @total = total
  58     end
  59     def reset()
  60       @count = 0
  61     end
  62     def report()
  63       print "[#{@count}/#{@total}]\r"
  64       @count += 1
  65     end
  66     def finish()
  67       puts
  68     end
  69   end
  70 end
  71
  72 #
  73 # Returns a table explaining how to rename the files. E.g.:
  74 #
  75 # {
  76 #    'S/456.html' => 'S/modules/tty.c.html'
  77 #    ...
  78 # }
  79 #
  80 def build_cnv_tbl()
  81   Counter.reset()
  82   cnv = {}
  83   Dir[$base + '/{[A-Z],files}/*.html'].each do |full_path|
  84   #p full_path
  85     if full_path =~ /(.*)\/((\w+)\/(\d+)(\.\w+))$/ then
  86       prefix = $1  # "/home/joe/.../HTML"
  87       path = $2    # "S/323.html"
  88       kind = $3    # "S"
  89       nid = $4     # "323"
  90       ext = $5     # ".html"
  91       pretty = prettyname(full_path, prefix, kind, nid, ext)
  92       cnv[path] = pretty
  93     end
  94     Counter.report()
  95   end
  96   Counter.finish()
  97   return cnv
  98 end
  99
 100 #
 101 # Make all the substiutions inside the files:
 102 #
 103 #   <a href='../S/456.html'>  -->  <a href='../S/modules/tty.c.html'>
 104 #   ...
 105 #
 106 def mksub(cnv)
 107   Counter.reset()
 108   Dir[$base + '/**/*.html'].each do |path|
 109     kind = path[$base.length .. -1][/\w+/]
 110     text = IO.read(path)
 111     fixed = text.gsub(/(<a href='(?:..\/)?)([^'#]+)/) do |a|
 112       prefix = $1
 113       target = $2
 114       if cnv[target] then
 115         target = cnv[target]
 116       elsif cnv[kind + '/' + target] then
 117         # This is a relative link of the form: href='456.html'
 118         if cnv[kind + '/' + target].match /\/(.*)/  # converts "s/whatever.html" to "whatever.html"
 119           target = $1
 120         end
 121       end
 122       prefix + target
 123     end
 124
 125     # Fix a problem in Opera, where empty lines are squashed.
 126     fixed.gsub!(%r{(<a id='L\d+' name='L\d+'></a>)\n}, "\\1 \n")
 127
 128     if text != fixed then
 129       IO.write(path, fixed)
 130       puts(path + " modified") if $debug
 131     end
 132     Counter.report()
 133   end
 134   Counter.finish()
 135 end
 136
 137 #
 138 # Rename the files.
 139 #
 140 def rename(cnv)
 141   Counter.reset()
 142   cnv.each do |a, b|
 143     src = $base + '/' + a
 144     trg = $base + '/' + b
 145     if $create_subdirs then
 146       FileUtils.mkdir_p(File.dirname(trg))
 147     end
 148     File.rename(src, trg)
 149     Counter.report()
 150   end
 151   Counter.finish()
 152 end
 153
 154 def better_css(css_path)
 155   text = IO.read(css_path)
 156   IO.write(css_path, text + $css)
 157 end
 158
 159 def main()
 160   if not File.exists?($base + '/help.html') then
 161     $base += '/HTML'
 162     if not File.exists?($base + '/help.html') then
 163       puts "Error: This doesn't look like htags' output directory."
 164       exit(1)
 165     end
 166   end
 167   Counter.init( Dir[$base + '/**/*.html'].length )
 168   puts "Figuring out pretty names..."
 169   cnv = build_cnv_tbl()
 170   pp cnv if $verbose
 171   puts "Updating links in files..."
 172   mksub(cnv)
 173   puts "Renaming files..."
 174   rename(cnv)
 175   puts "Enhancing the css..."
 176   better_css($base + '/style.css')
 177   puts "Done."
 178 end
 179
 180 main()