research/hhe/gen-string-tables

#!/usr/bin/env ruby

require "common.rb"

# Generate string -> offset lookup tables for each executable.

def load_string_lookups(exe_files)
	string_lookups = []

	CONFIGS.each_with_index do |config, i|
		strings = find_strings(exe_files[i], config)
		string_lookups.push(strings)
	end

	string_lookups
end

# Load list of "good" strings.

def load_good_strings(filename)
	File.readlines(filename).map do |string|
		JSON::load("["+ string.chomp + "]")[0]
	end
end

# Look up a string's offset, removing it from the specified lookup table.
# This is done because a string can appear multiple times at different offsets.

def lookup_string(lookup_table, string)
	lookup_table.each_with_index do |entry, i|
		if entry[1] == string
			lookup_table.delete_at(i)
			return entry[0]
		end
	end

	nil
end

# Find offsets for the specified string.

def get_string_offsets(string, string_lookups)
	string_lookups.map do |lookup|
		offset = lookup_string(lookup, string)

		if offset != nil
			offset
		else
			0
		end
	end
end

def format_offsets(offsets)
	formatted_offsets = offsets.map do |offset|
		sprintf("%5i", offset)
	end

	"{ " + formatted_offsets.join(", ") + " }"
end

# Print strings left over in the specified lookup table
# that were not in the good strings list.

def print_unsupported_strings(name, strings)

	unsupported_offsets = strings.map do |entry|
		entry[0]
	end

	puts "static int #{name}[] = {"

	numbers = 0

	for offset in unsupported_offsets.sort + [ -1 ]
		if numbers == 0
			print "    "
		else
			print " "
		end

		printf "%5i,", offset
		numbers += 1

		if numbers >= 8
			print "\n"
			numbers = 0
		end
	end

	if numbers > 0
		print "\n"
	end

	puts "};"
	puts
end

string_lookups = load_string_lookups(ARGV)
good_strings = load_good_strings("good-strings.txt")

# Output list of supported strings with their offset for each version

puts "static hhe_string_t strings[] = {"

for string in good_strings

	offsets = get_string_offsets(string, string_lookups)

	print "    { " + format_offsets(offsets) + ", "
#	print "\n      "
	print string.to_json + " },\n"
end

puts "};"
puts

# Output strings that appear in HHE but are not supported (ie. in good_strings)

CONFIGS.each_with_index do |config, i|
	name = "unsupported_strings_#{config::SUFFIX}"
	print_unsupported_strings(name, string_lookups[i])
end