m4b-tool/to-m4b.sh

#!/usr/bin/env bash
set -x
set -eufo pipefail

# CONSTANTS
declare -r script_dir="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"

# VARS
LEADING_ZEROES=2
SRC="${script_dir}/src"
OUT="${script_dir}/out"

# FUNCTIONS
parse-vars() {
  # Series structure: ${SRC}/<author>/<series>/Book <part> - <year> - <book> {<narrator>}/
  # Standalone structure: ${SRC}/<author>/<year> - <book> {<narrator>}/
  # Return variables as key=value lines; <part> zero-padded to LEADING_ZEROES as series_index
  local dir="${1%/}"
  local leaf="${dir##*/}"
  local parent="${dir%/*}"
  local grandparent="${parent%/*}"

  local author=""
  local series=""
  local title=""
  local year=""
  local narrator=""
  local part=""
  local series_index=""
  local rest=""

  # Helper: trim leading/trailing whitespace
  trim() { local s="$1"; s="${s#${s%%[![:space:]]*}}"; s="${s%${s##*[![:space:]]}}"; printf '%s' "$s"; }

  # Detect pattern and extract fields
  if [[ "$leaf" =~ ^[Bb]ook[[:space:]]+([0-9]+)[[:space:]]*-[[:space:]]*([0-9]{4})[[:space:]]*-[[:space:]]*(.+)$ ]]; then
    part="${BASH_REMATCH[1]}"
    year="${BASH_REMATCH[2]}"
    rest="${BASH_REMATCH[3]}"
    if [[ "$rest" =~ ^(.+)[[:space:]]*\{([^}]*)\}[[:space:]]*$ ]]; then
      title="${BASH_REMATCH[1]}"
      narrator="${BASH_REMATCH[2]}"
    else
      title="$rest"
    fi
    series="${parent##*/}"
    author="${grandparent##*/}"
  elif [[ "$leaf" =~ ^([0-9]{4})[[:space:]]*-[[:space:]]*(.+)$ ]]; then
    year="${BASH_REMATCH[1]}"
    rest="${BASH_REMATCH[2]}"
    if [[ "$rest" =~ ^(.+)[[:space:]]*\{([^}]*)\}[[:space:]]*$ ]]; then
      title="${BASH_REMATCH[1]}"
      narrator="${BASH_REMATCH[2]}"
    else
      title="$rest"
    fi
    author="${parent##*/}"
  else
    # Fallback best-effort parsing
    rest="$leaf"
    if [[ "$rest" =~ \{([^}]*)\}[[:space:]]*$ ]]; then
      narrator="${BASH_REMATCH[1]}"
      rest="${rest%\{${narrator}\}*}"
    fi
    if [[ "$rest" =~ ^([0-9]{4})[[:space:]]*-[[:space:]]*(.+)$ ]]; then
      year="${BASH_REMATCH[1]}"
      title="${BASH_REMATCH[2]}"
    else
      title="$rest"
    fi
    author="${parent##*/}"
  fi

  # Normalize whitespace
  title="$(trim "$title")"
  author="$(trim "$author")"
  series="$(trim "$series")"
  narrator="$(trim "$narrator")"

  # Zero-pad part for series_index if available
  if [[ -n "$part" ]]; then
    part="${part//[^0-9]/}"
    series_index=$(printf "%0*d" "$LEADING_ZEROES" "$part")
  fi

  printf 'author=%q\n' "$author"
  printf 'series=%q\n' "$series"
  printf 'series_index=%q\n' "$series_index"
  printf 'title=%q\n' "$title"
  printf 'year=%q\n' "$year"
  printf 'narrator=%q\n' "$narrator"
}

get-book-directories() {
  # Discover book directories under ${SRC} (or ${script_dir}/src if unset)
  # Matches:
  # - ${SRC}/<author>/<series>/Book <part> - <year> - <title> {<narrator>}/
  # - ${SRC}/<author>/<year> - <title> {<narrator>}/
  # Prints matched directories, one per line
  local src_root="${1}"
  local -a results=()
  local dir leaf

  # Iterate candidate depths using fast globbing (avoids slow full-recursive find)
  for dir in "${src_root}"/*/*/ "${src_root}"/*/*/*/; do
    [[ -d "$dir" ]] || continue
    leaf="${dir%/}"; leaf="${leaf##*/}"

    # Match standalone: "YYYY - Title {Narrator}" or similar
    if [[ "$leaf" =~ ^[0-9]{4}[[:space:]]*-[[:space:]]*.+$ ]]; then
      :
    # Match series: "Book N - YYYY - Title {Narrator}" (Book/book, flexible spaces)
    elif [[ "$leaf" =~ ^[Bb]ook[[:space:]]+[0-9]+[[:space:]]*-[[:space:]]*[0-9]{4}[[:space:]]*-[[:space:]]*.+$ ]]; then
      :
    else
      continue
    fi

    # Quick check: directory contains at least one likely audio file
    if compgen -G "$dir"*.{mp3,m4a,m4b,aac,flac,wav} >/dev/null 2>&1; then
      results+=("${dir%/}")
    fi
  done

  printf '%s\n' "${results[@]}"
}

m4b-merge() {
  local output_file="${1}"
  local source_dir="${2}"
  local author="${3}"
  local narrator="${4}"
  local title="${5}"
  local year="${6}"
  local series="${7}"
  local series_index=${8}

  mkdir -p "$(dirname "${output_file}")"
  nix run github:sandreas/m4b-tool#m4b-tool-libfdk -- \
    merge \
      -v \
      --jobs=6 \
      --audio-samplerate=44100 \
      --audio-quality=100 \
      --writer="${author}" \
      --artist="${narrator}" \
      --title="${title}" \
      --year="${year}" \
      --album="${series}" \
      --album-sort="${series_index}" \
      --output-file="${output_file}/" \
      -- "${source_dir}"
}

main() {
  local src_root="${SRC:-${script_dir}/src}"
  local out_root="${OUT:-${script_dir}/out}"
  local dir author narrator title year series series_index output_file

  mapfile -t dirs < <(get-book-directories "${src_root}")

  for dir in "${dirs[@]}"; do
    eval "$(parse-vars "$dir")"
    if [[ -z "$title" ]]; then
      echo "Skipping '$dir': could not parse title" >&2
      continue
    fi
    if [[ -z "$author" ]]; then
      echo "Skipping '$dir': could not parse author" >&2
      continue
    fi
    # Construct output path: ${OUT}/<author>/<series or standalone>/<series_index - >title {narrator}.m4b
    if [[ -n "$series" ]]; then
      output_file="${out_root}/${author}/${series}/${series_index:+${series_index} - }${title}${narrator:+ {${narrator}}}.m4b"
    else
      output_file="${out_root}/${author}/${year:+${year} - }${title}${narrator:+ {${narrator}}}.m4b"
    fi

    echo "Processing '$dir' -> '$output_file'"
    # m4b-merge "$output_file" "$dir" "$author" "$narrator" "$title" "$year" "$series" "$series_index"
  done
}

main "$@"