I have a binary file that is filled with FF values. I filled its start with many \000. Then, I padded its start with 10 \000, in order to get some kind of offset, and then I wrote a shorter string, also terminated with \000

I used this printf:

printf \000\000\000\000\000\000\000\000\000\000MAC_ADDRESS=12:34:56:78:90,PCB_MAIN_ID=m/SF-1V/MAIN/0.0,PCB_PIGGY1_ID=n/SF-1V/PS/0.0,CSL_HW_VARIANT=D\000' > eeprom

This is how it looks like when I show the hexdump of the file

enter image description here

Now what I want to know is how can I read the string. I can use MY_STR=${eeprom:OFFSET} (eeprom is the file name) which will give me the string, but will also give me the rest of the file which I don't want. How can I stop it when it first encounters \000?

  • Can't use MY_STR=${eeprom:OFFSET:LENGTH} because string's length is unknown
  • Another thing - how can I fill it up again with FF?
  • Using sh (busybox)


Trying to do some minor example of this... I have one file input with this values (after xxd -c 1 input):

0000000: 68  h
0000001: 65  e
0000002: 6c  l
0000003: 6c  l
0000004: 6f  o
0000005: 2c  ,
0000006: 20
0000007: 00  .
0000008: 69  i
0000009: 74  t
000000a: 27  '
000000b: 73  s
000000c: 20
000000d: 6d  m
000000e: 65  e
000000f: 2c  ,
0000010: 00  .

and I have this script s.sh:

for c in $(xxd -p input); do
    if [ "${c}" != 00 ]; then

echo $BUF

and I expected it to echo "hello", however, nothing is printed

  • busybox's sh doesn't sound like something that would work well with NUL-separated data. Zsh or even Bash would better. Or some actual programming language, like Perl. Are there any other tools than busybox you can use? – ilkkachu May 16 '18 at 13:51
  • strings will extract null terminated strings from binary files – ajeh May 16 '18 at 14:14

Solution 1: Direct Variable Assignment

If all you're worried about are the null bytes then you should just be able to directly read the data from the file into a variable using whichever standard method you prefer, i.e. you should be able to just ignore the null bytes and read the data from the file. Here's an example using the cat command and command substitution:

$ data="$(cat eeprom)"
$ echo "${data}"

This worked for me inside of a BusyBox Docker container.

Solution 2: Using xxd and a for loop

If you want more control than you can use xxd to convert the bytes to hexadecimal strings and iterate over these strings. Then, while iterating over these string, you can apply whatever logic you'd like, e.g. you could explicitly skip over the initial null values and print the rest of the data until you reach some break condition.

Here's a script that specifies a "white-list" of valid characters (ASCII 32 through 127), treats any subsequence of other characters as a separator, and extracts all valid substrings:

# get_hex_substrings.sh

# Get the path to the data-file as a command-line argument

# Keep track of state using environment variables

# NOTE: The '-p' flag is for "plain" output (no additional formatting)
# and the '-c 1' option specifies that the representation of each byte
# will be printed on a separate line
for h in $(xxd -p -c 1 "${datafile}"); do

    # Convert the hex character to standard decimal

    # Case where we're still inside the initial padding block
    if [ "${inside_padding_block}" == "true" ]; then
        if [ "${d}" -ge 32 ] && [ "${d}" -le 127 ]; then
            printf '\x'"${h}";

    # Case where we're passed the initial padding, but inside another
    # block of non-printable characters
    elif [ "${inside_bad_block}" == "true" ]; then
        if [ "${d}" -ge 32 ] && [ "${d}" -le 127 ]; then
            printf '\x'"${h}";

    # Case where we're inside of a substring that we want to extract
        if [ "${d}" -ge 32 ] && [ "${d}" -le 127 ]; then
            printf '\x'"${h}";

if [ "${inside_bad_block}" == "false" ]; then

Now we can test this out by creating an example file which has both \x00 and \xff subsequences separating substrings:

printf '\x00\x00\x00string1\xff\xff\xffstring2\x00\x00\x00string3\x00\x00\x00' > data.hex

And here's the output we get when running the script:

$ sh get_hex_substrings.sh data.hex

Solution 3: Using the tr and cut commands

You could also try using the tr and cut commands to deal with the null bytes. Here's an example of extracting the first null-terminated string from a list of null-terminated strings by squeezing/collapsing adjacent null-characters and converting them to newlines:

$ printf '\000\000\000string1\000\000\000string2\000\000\000string3\000\000\000' > file.dat
$ tr -s '\000' '\n' < file.dat | cut -d$'\n' -f2
  • How would it handle the FF bytes? and what if there will be other garbage values after the null byte that ends the string? – CIsForCookies May 16 '18 at 14:35
  • 1
    @CIsForCookies The first method (dat="$(cat eeprom)" probably won't work for that case, but the loop obviously will - you can include whatever logic you want in there. Maybe update your question to include a more complete input example along with the desired output? – igal May 16 '18 at 15:15
  • Can you please explain the xxd parameters? can't understand why -c is needed (the others I'm not sure about) – CIsForCookies May 16 '18 at 16:08
  • 1
    @CIsForCookies The -c parameter sets the number of columns used when outputting. Using -c 1 should mean that each byte is printed on its own line. I don't think it's necessary, strictly speaking - just a personal choice. Anyway, I'll update my post. – igal May 16 '18 at 16:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.