research/hhe/dump-strings

#!/usr/bin/env ruby

require "json"
require "common"

# HHE seems to stop when it has found this many strings:

NUM_STRINGS=785

set_config(ARGV[1])

def read_block(file)
        result = ""

        4.times do
                break if file.eof?
                c = file.readchar
                result += sprintf("%c", c)
        end

        end_index = result.index(0)
        if end_index
                result = result[0, end_index]
        end

        result
end

def bad_string?(str)
	str.each_byte do |b|
		if b >= 0x80
			return true
		end
	end

	return false
end

File.open(ARGV[0]) do |file|
        current_str = ""

	file.seek(STRINGS_OFFSET)

	offset = 0
	start_offset = nil
	num_strings = 0

        while num_strings < NUM_STRINGS
                block = read_block(file)

		if start_offset == nil
			start_offset = offset
		end

		current_str += block

		# End of string?

		if block.length < 4

			# Extended-ASCII characters cannot be output
			# to JSON.

			if bad_string?(current_str)
				current_str = "XXXXX BAD STRING"
			end

			printf "    { %6i, %s },\n", start_offset,
			                             current_str.to_json
			current_str = ""
			start_offset = nil
			num_strings += 1
		end

		offset += 4
        end
end