Extract cargo from a secure shipment

Content as provided by sship has to be extracted and decrypted in order for the recipient to gain access the actual content. The prerequisite is that the recipient holds the private key associated with the public key used for encryption. This key pair must have been generated using the RSA crypto system and should use 2048 bits key size, or more. The recipient may use tools within sship to establish the associated content or any other tool that handle the cryptographic algorithms applied by sship.

Conventions

The following conventions and assumptions are used:

• cargo of shipment is found within a compressed file named [filename]__[datestamp]_[timestamp].tar.gz where [filename] is the name of the file that was encrypted prior to shipment, [datestamp] is the date of shipment formatted as YYYYMMDD and [timestamp] is the time of shipment formatted as HHMMSS
• for the below examples the filename data.csv__20200101_081500.tar.gz will be used
• the above file exists in the current working directory
• the private key is represented by the file ~/.ssh/id_rsa

Extract and decrypt content in R

Install the latest version of the sship packages from the R command prompt:

remotes::install_github("Rapporteket/sship")

and load it into the R session:

library("sship")

Use the dec()-function to extract and decrypt the content:

dec(tarfile = "data.csv__20200101_081500.tar.gz",
    keyfile = "~/.ssh/id_rsa",
    target_dir = ".")

From the example above the file data.csv will be written to the current working directory. Please use

?dec

at the R prompt for more information.

Extract and decrypt content outside R

Extraction and decryption of content that was encrypted and packed by sship (in R) is perfectly possible also outside R. In the above R-function a few things is taken care of “under the hood” and to aid a better understanding more information should be added to the above list of “conventions”:

• when encrypting content sship uses the AES block cipher in CBC mode applying a key of 256 random bits (32 bytes long)
• when encryption content sship uses an initialization vector of 128 random bits (16 bytes long)
• the shipment, i.e. the compressed tar archive file, consists of three files that will be named (according to the current example):
  • data.csv.enc: the encrypted data of the shipment
  • key.enc: the encrypted key
  • iv: the initialization vector used for encryption and decryption
• just for the record then, names of encrypted files ends with .enc
• the filename of the original content (data) is represented by the first part of the archive filename (shipment) up until the double underscore (__), e.g. given the archive filename data.csv__20200101_081500.tar.gz the decrypted content will be established in the file data.csv

Below is an example of how data.csv__20200101_081500.tar.gz can be processed from a Bash shell (unix) command prompt. First, extract the archive:

tar -xvzf data.csv__20200101_081500.tar.gz

Then, decrypt the key by using the private key:

openssl rsautl -decrypt -inkey ~/.ssh/id_rsa -in key.enc -out key

Finally, decrypt the content to the file data.csv:

openssl aes-256-cbc -d -in data.csv.enc -out data.csv \
-K $(hexdump -e '32/1 "%02x"' key) -iv $(hexdump -e '16/1 "%02x"' iv)

In the above example both the key and iv files are binary. When using OpenSSL for decryption these have to be provided as strings of hexadecimal digits as can be seen by the use of inline hexdump conversion in the last command. Methods for conversion from binary to hex may vary between systems. It might also be worth noting that here AES uses a key directly and not a (hash of some) passphrase that often will be the case elsewhere.