mc/maint/htagsfix

#!/usr/bin/env ruby

#
# Fixes htags' output so that files have fixed names. E.g., S/542.html -> D/whatever.c.h.
#
# It addresses the following problem:
#
#    "Can htags create permanent addresses?"
#    http://lists.gnu.org/archive/html/help-global/2015-11/msg00002.html
#

require 'fileutils'
require 'pp'

Encoding.default_external = "BINARY"

$base = '.'  # Where the HTML is stored.
$verbose = false
$debug = false
$create_subdirs = false  # Experimental feature. Doesn't really work.

# Bonus: highlight the target line.
$css = "
a:target {
  position: absolute;
  left: 0;
  height: 1.35em;
  width: 100%;
  background: yellow;
  opacity: .5;
  z-index: -1000;
}
"

#
# Computes pretty names for files:
#
#    '.../S/456.html' => 'S/modules/tty.c.html'
#
def prettyname(full_path, prefix, kind, nid, ext)
  gist = nid
  start = IO.read(full_path, 1024)
  if start =~ /<title>(.*?)<\/title>/ then
    gist = $1
    gist.gsub!(%r{/+$}, '')  # remove trailing '/'s.
    if not $create_subdirs then
      gist.gsub!('/', '--')
    end
    gist.gsub!(%r{[^a-zA-Z0-9_.,/-]}, '-')  # clean special characters.
  end
  return kind + '/' + gist + ext
end

class Counter
  class << self
    def init(total)
      @total = total
    end
    def reset()
      @count = 0
    end
    def report()
      print "[#{@count}/#{@total}]\r"
      @count += 1
    end
    def finish()
      puts
    end
  end
end

#
# Returns a table explaining how to rename the files. E.g.:
#
# {
#    'S/456.html' => 'S/modules/tty.c.html'
#    ...
# }
#
def build_cnv_tbl()
  Counter.reset()
  cnv = {}
  Dir[$base + '/{[A-Z],files}/*.html'].each do |full_path|
  #p full_path
    if full_path =~ /(.*)\/((\w+)\/(\d+)(\.\w+))$/ then
      prefix = $1  # "/home/joe/.../HTML"
      path = $2    # "S/323.html"
      kind = $3    # "S"
      nid = $4     # "323"
      ext = $5     # ".html"
      pretty = prettyname(full_path, prefix, kind, nid, ext)
      cnv[path] = pretty
    end
    Counter.report()
  end
  Counter.finish()
  return cnv
end

#
# Make all the substiutions inside the files:
#
#   <a href='../S/456.html'>  -->  <a href='../S/modules/tty.c.html'>
#   ...
#
def mksub(cnv)
  Counter.reset()
  Dir[$base + '/**/*.html'].each do |path|
    kind = path[$base.length .. -1][/\w+/]
    text = IO.read(path)
    fixed = text.gsub(/(<a href='(?:..\/)?)([^'#]+)/) do |a|
      prefix = $1
      target = $2
      if cnv[target] then
        target = cnv[target]
      elsif cnv[kind + '/' + target] then
        # This is a relative link of the form: href='456.html'
        if cnv[kind + '/' + target].match /\/(.*)/  # converts "s/whatever.html" to "whatever.html"
          target = $1
        end
      end
      prefix + target
    end

    # Fix a problem in Opera, where empty lines are squashed.
    fixed.gsub!(%r{(<a id='L\d+' name='L\d+'></a>)\n}, "\\1 \n")

    if text != fixed then
      IO.write(path, fixed)
      puts(path + " modified") if $debug
    end
    Counter.report()
  end
  Counter.finish()
end

#
# Rename the files.
#
def rename(cnv)
  Counter.reset()
  cnv.each do |a, b|
    src = $base + '/' + a
    trg = $base + '/' + b
    if $create_subdirs then
      FileUtils.mkdir_p(File.dirname(trg))
    end
    File.rename(src, trg)
    Counter.report()
  end
  Counter.finish()
end

def better_css(css_path)
  text = IO.read(css_path)
  IO.write(css_path, text + $css)
end

def main()
  if not File.exists?($base + '/help.html') then
    $base += '/HTML'
    if not File.exists?($base + '/help.html') then
      puts "Error: This doesn't look like htags' output directory."
      exit(1)
    end
  end
  Counter.init( Dir[$base + '/**/*.html'].length )
  puts "Figuring out pretty names..."
  cnv = build_cnv_tbl()
  pp cnv if $verbose
  puts "Updating links in files..."
  mksub(cnv)
  puts "Renaming files..."
  rename(cnv)
  puts "Enhancing the css..."
  better_css($base + '/style.css')  
  puts "Done."
end

main()