Bash script to merge all mp4 videos in current directory (recursively 2 levels). It also updates the chapter marks to retain the folder/filename of source dir

merge-mp4.sh

#!/bin/bash
#http://redsymbol.net/articles/unofficial-bash-strict-mode/
set -euo pipefail
IFS=$'\n\t'

## Script to merge all mp4 videos in current directory (recursively 2 levels)
## And update chapter marks to retain the folder/filename

## Script for merging videos

temp_dir=$(mktemp -d)
function finish {
  rc=$?
  if [[ $rc != 0 ]]; then
  	echo
  	echo "FAILED!"
  fi
  echo -n "Cleaning up "
  rm -rf "${temp_dir}"
  echo "..........[ DONE ]"
  exit $rc
}
trap finish EXIT

current_dir=$(pwd)
bname=$(basename ${current_dir})
final_mp4=${bname}.mp4
input_list=${temp_dir}/${bname}-input_list.txt
file_list=${temp_dir}/${bname}-file_list
meta_file=${temp_dir}/${bname}-metadata.txt

#Hopefully this will work for either BSD or GNU sed
extended_match="-r"
echo "" | sed ${extended_match} 's|foo|bar|' 2>/dev/null || extended_match="-E"

if [ -e "${final_mp4}" ]; then
    echo "${final_mp4} already exists, please remove it."
    exit 1
fi

echo -n "Generating file lists "
find -s . -maxdepth 2 -type f -iname '*.mp4' |
    sed -e 's|^./||' |
    tee "${file_list}" |
    awk "{printf \"file '${current_dir}/%s'\n\", \$0}" > "${input_list}"
echo "..........[ DONE ]"  

## chapter marks
#Do this first so we fail early
#TODO: Test (‘=’, ‘;’, ‘#’, ‘\’) are escaped
ts=0
echo -n "Generating chapter marks "
ffmpeg -i "$(head -1 "${file_list}")" -f ffmetadata "${meta_file}" -v quiet
cat "${file_list}" | while read file
do
    ds=$(ffprobe -v quiet -of csv=p=0 -show_entries format=duration "${file}")
    # echo "$ds"
    escaped_title=$(echo ${file} | sed ${extended_match} -e 's|([=;#\])|\\\1|g' -e 's|.[Mm][Pp]4$||' )
    echo "[CHAPTER]" >> "${meta_file}"
    echo "TIMEBASE=1/1" >> "${meta_file}"
    echo "START=${ts}" >> "${meta_file}"
    ts=$(awk "BEGIN {print ${ts}+${ds}; exit}")
    echo "END=${ts}" >> "${meta_file}"
    echo "TITLE=${escaped_title}" >> "${meta_file}"
done
echo "..........[ DONE ]"

echo -n "Merging the files "
ffmpeg -f concat -safe 0 -i "${input_list}" -i "${meta_file}" -map_metadata 1 -codec copy "${final_mp4}" -v quiet
echo "..........[ DONE ]"

echo "Job Completed."

Great script, but sadly it sorts the chapters strictly alphabetical and not via the natural sort order. I've already tried to modify it myself, but I wasn't succesful. Could you help me out there?

Okay, I figured it out for anyone who has the same problem: The -s parameter in the "find" command in line 42 causes find to traverse the file hierarchies in lexicographical order , meaning 10 comes after 1, instead of 2 (which is a problem if you, say, have files that have the chapter number in their filename). If you want the chapters to be in their natural sort order (1, 2, 10), remove the -s parameter and add "sort -V |" at the end of the line. (Be aware though, there shouldn't be any quotes in the filenames using this method, otherwise some chapters were missing as evident in a small fill size.)
Thanks again for your work!

As a note, filenames with colons do not work correctly. ffmpeg interprets them within the commandline. This would seem to be a fairly common situation due to using timestamps as filenames.

Also, GNU find doesn't support -s. A quick edit to substitute it for sort in the pipeline works.