Hide/find messages hidden in text indentation.

hidespace.py

#!/usr/bin/env python
"""Hide/find messages hidden in text indentation."""
import argparse
from collections import deque
import os
from os.path import exists


# Command line parsing stuff.
PARSER = argparse.ArgumentParser(description="Hide/find messages hidden in text indentation.")
PARSER.add_argument("in_file", metavar="IN_FILE", help="The file to hide/find messages in")
PARSER.add_argument("message", metavar="MESSAGE", nargs="?", help="The message to hide in the file")
PARSER.add_argument("-o", metavar="OUT_FILE", dest="out_file",
                    help="An output path for modified files, defaults to appending"
                         " \"-hidden\" to IN_FILE")
PARSER.add_argument("-t", "--tab-width", metavar="N", default=4, type=int,
                    help="The number of spaces per indentation level in file")
ARGS = PARSER.parse_args()

BITS = (128, 64, 32, 16, 8, 4, 2, 1)


def tabify_line(line):
    """Convert space indentation to tabs."""
    new_line = ""
    index = 0
    spaces = 0
    tabs = 0
    while line[index] in (" ", "\t"):
        if line[index] == " ":
            spaces += 1
            if spaces >= ARGS.tab_width:
                new_line += "\t"
                tabs += 1
                spaces = 0
        else:
            spaces = 0  # If there were spaces ahead of this tab,
                        #  we want to strip them.
            new_line += line[index]
            tabs += 1
        index += 1
    if spaces:
        new_line += " " * spaces
    new_line += line[index:]
    return new_line, tabs


def prep_file(in_file):
    """Convert a file to tabs."""
    prepped_file = in_file + "-tmp"
    total_tabs = 0
    with open(in_file) as in_fd, open(prepped_file, "w") as out_fd:
        for line in in_fd.readlines():
            new_line, tabs = tabify_line(line)
            out_fd.write(new_line)
            total_tabs += tabs
    return prepped_file, total_tabs


def hide_message_in_file(in_file, out_file, message):
    """Hide a message in a file's indentation."""
    prepped_file, tabs = prep_file(in_file)
    if tabs < len(message) * 8:
        os.remove(in_file + "-tmp")
        raise ValueError("message too long for amount of indentation in file")
    bit_queue = deque()
    for character in message:
        ordinal = ord(character)
        if ordinal > 255:
            raise ValueError("character cannot be hidden: " + character)
        for bit in BITS:
            bit_queue.append(int(bool(ordinal & bit)))
    with open(prepped_file) as in_fd, open(out_file, "w") as out_fd:
        for line in in_fd.readlines():
            new_line = ""
            index = 0
            while bit_queue and line[index] in (" ", "\t"):
                if line[index] == "\t":
                    bit = bit_queue.popleft()
                    new_line += (" " * (bit + 1)) + "\t"
                else:
                    new_line += " "
                index += 1
            new_line += line[index:]
            out_fd.write(new_line)
    os.remove(prepped_file)


def show_message_in_file(in_file):
    """Find a message hidden in a file's indentation."""
    found_bits = []
    with open(in_file) as in_fd:
        for line in in_fd.readlines():
            index = 0
            spaces = 0
            while line[index] in (" ", "\t"):
                if line[index] == " ":
                    spaces += 1
                else:
                    if spaces and spaces < 3:
                        found_bits.append(spaces - 1)
                        spaces = 0
                index += 1
    if len(found_bits) < 8:
        print("No message found.")
        return
    message = ""
    while found_bits:
        if len(found_bits) < 8:
            print("WARNING: number of bits found not multiple of 8")
            break
        byte = found_bits[:8]
        found_bits = found_bits[8:]
        ordinal = 0
        for index, bit in enumerate(byte):
            if bit:
                ordinal += BITS[index]
        message += chr(ordinal)
    print("Found message:", message)


def start():
    """The main entrypoint for this script."""
    if not exists(ARGS.in_file):
        raise FileNotFoundError(ARGS.in_file)
    if ARGS.message:
        out_file = ARGS.out_file or ARGS.in_file + "-hidden"
        hide_message_in_file(ARGS.in_file, out_file, ARGS.message)
    else:
        show_message_in_file(ARGS.in_file)


if __name__ == "__main__":
    start()