make-self-extracting

A library for the creation of self extracting shell scripts supporting Node 12 and newer.

Github: https://github.com/danishcake/make-self-extracting
NPM: https://www.npmjs.com/package/make-self-extracting

The shell scripts are designed to run under bash.

Why?

It's easier for your users if you distribute a single file, and it's called install.sh.

Installation

Run

npm install make-self-extracting --save

Usage

The script runs in two stages:

1. A 'pre-extraction' section. This runs before the embedded files have been extracted to a temporary directory, and is a good place to display a banner, gather input etc.
2. A 'post-extraction' section'. This runs after the embedded files have been extracted. The working directory will have been changed to a temporary directory containing the extracted files for the duration of this section

await makeSelfExtractingScript(
  {
    preExtraction: 'echo This section runs in `pwd`',
    postExtraction: 'cat header.txt\necho This section runs in `pwd`\nls -l'
  },
  [
    {
      filename: 'header.txt',
      content: Buffer.from('This text comes from an embedded file\n', 'utf-8')
    },
    {
      filename: 'example_zip.mjs',
      content: fs.createReadStream('examples/zip_format.mjs')
    },
    {
      filename: 'example_docker.mjs',
      content: fs.createReadStream('examples/docker.mjs')
    }
  ],
  fs.createWriteStream('simple.sh')
);

This generates output that looks like this:

#!/bin/bash
# Self extracting self script created with 'make-self-extracting
# https://www.npmjs.com/package/make-self-extracting
# https://github.com/danishcake/make-self-extracting
echo This section runs in `pwd`
readonly TMPDIR=`mktemp -d`
readonly PAYLOAD_START=16
tail -n+$PAYLOAD_START $0 | tar -xz -C $TMPDIR
pushd $TMPDIR > /dev/null
cat header.txt
echo This section runs in `pwd`
ls -l
popd > /dev/null
rm -rf $TMPDIR
exit 0
# ... Compressed data

You can control how the payload files are stored using the SelfExtractingScriptOptions argument.

type SelfExtractingScriptOptions = {
  // Script to execute pre-extraction. This does not need a shebang
  preExtraction?: string;
  // Script content to execute post-extraction. This will be executed in a temporary directory
  // containing the extracted payload
  postExtraction?: string;
  // If the 'Generated using' header should be omitted. Defaults to false
  omitLibraryHeader?: boolean;
  // Archive format. Zip has lower memory consumption during generation, but is less likely be available
  // Zip uses more disk space during extraction, as it must extract the zip to a temporary directory first
  // Defaults to 'tar'
  archiveFormat?: 'zip' | 'tar';
  // If the archive should be compressed. If the input files are already compressed it's usually better
  // to just store the files instead
  // Defaults to 5
  compressionLevel?: 0 | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9;
}

Additional examples are in the examples folder. They expect to be run from the project root.

You can add files for Buffers, Readables and paths to local files. The file mode defaults to 644 but can be set on a per file basis is required.

// Embedded shell script must be executable
{
  filename: 'setup.sh',
  path: 'setup.sh',
  mode: 0o755
}

How it works

The embedded files are stored in an archive and appended to the shell script. The script then uses tail on itself to extract the archive to a temporary directory when run.

Remarks

The code herein will probably work fine under older versions of Node. The limiting factor is the jest unit test library.

Refences

linuxjournal