Get available chars

get-available-chars.rb

#!/usr/bin/env ruby
require 'optparse'

# Get available characters
#
# Written by Phoenix Enero.
#
# It takes a font input in OTF/TTF format and returns a text file with the
# cmap.

## Lets start!
stopwatch_start = Time.now



# Option handling

options = {:extension => '.unc', :pattern => '*.woff'}
OptionParser.new do |o|
	o.banner = "Usage: get-available-chars.rb [options]"

	o.on("-f", "--files [FILES]", "The file path or glob pattern") do |pattern|
		options[:pattern] = pattern
	end

	o.on("-x", "--extension [EXTENSION]", "The extension for the output file(s).") do |ext|
		options[:extension] = ext || '.unc'

		# Ensure extension begins with dot `.`
		options[:extension].sub!(/\A\.?(?=.)/, ".")
	end
end.parse!

files = Dir.glob( options[:pattern] )





# Make a .ttx file containing only the CMAP
def make_cmap_ttx( name, source )
	# puts "ttx -f -o #{name} -t cmap #{source}"
	system *%W(ttx -f -o #{name} -t cmap #{source})

	# Check if it worked
	if ! File.file?( name )
		raise "Failed to create ttx cmap `#{name}`"
		exit
	end
end


# Return all hexadecimal values starting with `0x` without including `0x`
def grab_hex( str )
	return str.scan( /(?<=0[xX])([0-9a-fA-F]+)/ )
end





# Now we loop
files.each do |font|

	font_name = File.basename( font )
	font_base_name = File.basename( font, ".*" )

	# Make a temporary .ttx file containing only the cmap
	temp_name = "__#{Time.now.to_i.to_s}.cmap.ttx"
	make_cmap_ttx( temp_name, font )

	# Grab the temp file's contents
	temp = File.read( temp_name )

	# Get the Unicode values from the XML file
	# Assumptions: UTF strings begin with '0x'
	characters = grab_hex( temp )

	# delete temp file
	File.delete( temp_name )

	# Make an output file with the same name and dir as the font file
	output = File.open( File.join(File.dirname(font), File.basename(font, '.*')) + options[:extension], "w" );

	# Output the unicode values line-by-line
	output << characters.join( "\n" )
	output.close
end

# And we're done!
stopwatch_finish = Time.now
stopwatch_delta = stopwatch_finish - stopwatch_start

puts "Finished in #{stopwatch_delta}s"