bash2asciicast.sh - Convert a bash script into an asciinema recording

bash2asciicast.sh

#!/usr/bin/env bash

# Convert a bash script into an asciinema recording

set -o errexit -o nounset -o pipefail

INPUT_FILE=${1:-}
if [[ -z "${INPUT_FILE}" ]]; then
  echo "Syntax: $0 <path/to/bash-scipt.sh> <path/to/out.cast>"
  exit 1
fi

OUTPUT_FILE=${2:-}
if [[ -z "${OUTPUT_FILE}" ]]; then
  echo "Syntax: $0 <path/to/bash-scipt.sh> <path/to/out.cast>"
  exit 1
fi

CHAR_STEP=0.05
LINE_STEP=0.5

IFS='' read -r -d '' BG_SCRIPT << EOF || true
set -o errexit -o nounset -o pipefail

sleep 1

while IFS="" read -r line; do

  if [[ "\${line}" == "#!"* ]]; then
    # skip shebang
    continue
  fi

  # loop utf-8 characters and print them individually
  while IFS="" read -r char; do
    echo -n "\${char}"
    sleep ${CHAR_STEP}
  done <<< "\$(echo "\${line}" | sed -e 's/\(.\)/\1\n/g')"

  echo
  sleep ${LINE_STEP}

done < "/dev/stdin"

echo
echo "exit"

EOF

cat "${INPUT_FILE}" | bash -c "${BG_SCRIPT}" | asciinema rec --stdin "${OUTPUT_FILE}"

Note: There's no backpressure from asciinema to wait fo commands to complete, so if you have long commands, you may want to add a longer sleep to replace the sleep ${LINE_STEP}.

For example:

  if [[ "\${line}" == "kpt pkg get "* ]]; then
    sleep 10.0
  elif [[ "\${line}" == "kpt cfg set "* ]]; then
    sleep 2.0
  elif [[ "\${line}" == "kubectl "* ]]; then
    sleep 10.0
  else
    sleep ${LINE_SLEEP}
  fi

This will of course create long pauses in the recording, but you can work around that by adding -i 2 to the asciinema command, which will shorten idle pauses to a max of 2 seconds in the output recording.

The only other solution I could come up with would be to write the asciicast manually, as in https://gist.github.com/karlkfi/3e286f507c2d0ae7f8a1994e4e137a7e