Convert Markdown Code Blocks to Shell Scripts

Keeping documentation updated is tedious and error prone. Particularly documentation that contain actual commands or e.g. configuration data. General documentation can be high-level and thus remain correct even when changes are being implemented, but actual commands or configuration data cannot - it must be correct as documented.

This tool allows for extraction of code blocks from Markdown documentation into shell-scripts that can be exercised in automated tests and thus provide an indication of documentation correctness.

Getting Started

Imagine the following example with three code blocks. One that sets environment variables:

export FOO=hello
export BAR=world

One that prints an output based on the variables:

echo "$FOO, $BAR"

And finally the expected output:

hello, world

We can automate testing of this documentation using the markdown2bash tool. First extract code blocks:

export IMAGE_VERSION=0.0.2
cat README.md | docker run --rm -i ghcr.io/michaelvl/markdown2bash:$IMAGE_VERSION > readme.sh

This will generate a shell script that e.g. includes the first code block both as a function and a variable:

read -r -d '' _v_getting_started_0001 <<'EOF_5577006791947779410'
export FOO=hello
export BAR=world
EOF_5577006791947779410

_f_getting_started_0001() {
  export FOO=hello
  export BAR=world
}

Note how the function/variable names are constructed from the lower-case version of the Headers (getting_started) in the markdown, followed by an incremented id (0001) to handle multiple code blocks per heading and function names start with a _f_ prefix and variables with _v_.

From these function and variable definitions, we can construct tests that use the actual code from markdown documentation. See test/readme-example-test.sh for an example that test the code blocks above. See also the GitHub action e2e-tests.yaml for how markdown2bash is used to test the code blocks above.