library_works/encode_all

#!/bin/bash
# encode_all encodes the whole library into specified format
#
ulimit -n "$(ulimit -Hn)"
. env_parallel.bash

function ts() {
  date '+%Y-%m-%d_%H:%M'
}

# overwrite_all=true
overwrite_all=false
debug_only=false
# debug_only=true

src_library="${HOME}/flac"
output_dir="${HOME}/opus"
logdir="${HOME}/encoding_logs"
[[ -d "$logdir" ]] || mkdir "$logdir"
start_ts="$(ts)"
log="${logdir}/${start_ts}_encoding_log"
fixes="${logdir}/${start_ts}_fixes.sh"
temp_dir="$(mktemp -d)"

printf '#!/bin/bash\n# FIXES generated by library_works/encode_all\n\n' > "$fixes"

function encode() {
  file="$1"
  # file is full original path + filename
  src_dir="$(dirname "$file")"
  file_name="$(basename "$file")"
  output_dir="$(sed "s_${HOME}/flac_${HOME}/opus_" <<< "$src_dir")"
  final_opus="$(sed \
                -e 's/\.flac$/.opus/' \
                -e "s_${HOME}/flac_${HOME}/opus_" <<< "$file")"

  sum="$(sha256sum <<< "$final_opus" | cut -d\  -f1)"
  printf '%s\n' "$final_opus" > "$temp_dir/$sum"
  if [[ -f "$final_opus" && "$overwrite_all" == 'false' ]]; then
    printf '%s exists, not reencoding\n' >> "$log"
    return
  fi

  unset tags
  unset dest_dir

  declare -A tags
  while read -r line; do
    tag_type="$(cut -d= -f1 <<< "$line")"
    tag_type="${tag_type^^}"
    tag_val="$(cut -d= -f2- <<< "$line")"
    # TODO: suggest fixes for lower case tags
    tags["$tag_type"]="$tag_val"
  done < <(metaflac --list --block-type='VORBIS_COMMENT' "$file" | \
           grep 'comment\[' | \
           cut -d: -f2- | \
           sed 's/^ //')
  album="$(basename "$output_dir")"

  album_dir_date="${album%% *}"
  album_dir_date_valid='true'
  if [[ ! "$album_dir_date" =~ ^[-0-9]+$ ]]; then
    printf '%s invalid album date extracted from album directory name: %s\n' "$file" "$album_dir_date" >> "$log"
    album_dir_date_valid='false'
  fi

  # FLAC tags to read in order: DATE RELEASEDATE YEAR
  album_flac_date_valid='false'
  for date_tag in DATE RELEASEDATE YEAR; do
    album_flac_date="${tags[$date_tag]}"
    if [[ "$album_flac_date" =~ ^[-0-9]+$ ]]; then
      album_flac_date_valid='true'
      break
    fi
  done

  if [[ "$album_flac_date_valid" == 'false' ]]; then
    [[ -n "$album_flac_date" ]] && printf '%s invalid album date extracted from flac data: %s\n' "$file" "$album_flac_date" >> "$log"
    [[ ! -n "$album_flac_date" ]] && printf '%s no album date extracted from flac data\n' "$file" >> "$log"
  fi

  album_date='unknown'
  if [[ "$album_dir_date_valid" && ! "$album_flac_date_valid" ]]; then
    printf '%s invalid flac tag date\n' >> "$log"
    album_date="$album_dir_date"
    if [[ "$file" != *\'* ]]; then
      printf "metaflac --set-tag=DATE='%s' '%s'\n" "$album_date" "$file" >> "$fixes"
    elif [[ "$file" != *\"* && "$file" == *\'* ]]; then
      printf 'metaflac --set-tag=DATE="%s" "%s"\n' "$album_date" "$file" >> "$fixes"
    elif [[ "$file" == *\"* && "$file" == *\'* ]]; then
      printf "# metaflac --set-tag=DATE='%s' '%s'\n" "$album_date" "$file" >> "$fixes"
      printf '# cant quote\n' >> "$fixes"
    fi
    [[ -n "$album_flac_date" ]] && printf "# old tag: '%s'\n" "$album_flac_date" >> "$fixes"
    [[ -z "$album_flac_date" ]] && printf "# no old tag\n" >> "$fixes"
  elif [[ ! "$album_dir_date_valid" && "$album_flac_date_valid" ]]; then
    printf '%s invalid album dir date\n' "$file" >> "$log"
    album_date="$album_flac_date"
  elif [[ "$album_dir_date_valid" && "$album_flac_date_valid" ]]; then
    album_dir_date_length="$(wc -c <<< "$album_dir_date")"
    album_flac_date_length="$(wc -c <<< "$album_flac_date")"
    if [[ "$album_dir_date_length" -gt "$album_flac_date_length" ]]; then
      printf '%s has a longer date set in album dir name\n' "$file" >> "$log"
      album_date="$album_dir_date"
      if [[ "$file" != *\'* ]]; then
        printf "metaflac --set-tag=DATE='%s' '%s'\n" "$album_date" "$file" >> "$fixes"
      elif [[ "$file" != *\"* && "$file" == *\'* ]]; then
        printf 'metaflac --set-tag=DATE="%s" "%s"\n' "$album_date" "$file" >> "$fixes"
      elif [[ "$file" == *\"* && "$file" == *\'* ]]; then
        printf "# metaflac --set-tag=DATE='%s' '%s'\n" "$album_date" "$file" >> "$fixes"
        printf '# cant quote\n' >> "$fixes"
      fi
      printf "# old tag: '%s'\n" "$album_flac_date" >> "$fixes"
    else
      printf '%s has a longer date set in flac tag\n' "$file" >> "$log"
      album_date="$album_flac_date"
    fi
  fi

  if [[ "$file_name" =~ ^D[0-9]+ ]]; then
    disk="$(cut -d\  -f1 <<< "$file_name")"
    tags['DISK']="${disk#D}"
  fi
  declare -a tag_ensure
  tag_ensure=( 'TRACKNUMBER' 'TITLE' 'ALBUM' 'ARTIST' )
  for tag_type in "${tag_ensure[@]}"; do
    if [[ -z "${tags["$tag_type"]}" ]]; then
      printf '%s is missing tag: %s\n' "$file" "$tag_type" >> "$log"
      tags["$tag_type"]="missing $tag_type"
    fi
  done

  [[ -d "$output_dir" ]] || mkdir -p "$output_dir"
  tag_flags=(
      '--title' "${tags['TITLE']}"
      '--artist' "${tags['ARTIST']}"
      '--album' "${tags['ALBUM']}"
      '--tracknumber' "${tags['TRACKNUMBER']}"
      '--date' "${album_date}")
  [[ -n "${tags[DISK]}" ]] && tag_flags+=('--comment' "DISK=${tags['DISK']}")
  if [[ "$debug_only" == 'false' ]]; then
    err="$(opusenc --quiet --discard-comments --discard-pictures "${tag_flags[@]}"  "$file" "$final_opus" 2>&1)"
  else
    echo opusenc --quiet --discard-comments --discard-pictures "${tag_flags[@]}"  "$file" "$final_opus"
  fi
  if [[ -n "$err" ]]; then
    printf '%s\n' "$err" >> "$log"
  fi
}

echo starting processing
find "$src_library" -type f -name '*.flac' -print0 | env_parallel -P 0 -0 --bar encode {}
echo done processing

echo removing obsolete opus files
diff -d --suppress-common-lines <(cat "$temp_dir"/* | sort -V) <(find "$output_dir" -type f | sort -V) | grep -E '^>' | sed 's/^> //' | while read old_opus; do
  rm -f "$old_opus"
done
rm -rf "$temp_dir"

echo deleting any empty directories
find "$output_dir" -empty -type d -delete