m4b2opus

#!/bin/bash
#depends opustags, ffmpeg, indexopus, opus.book.4, mplayer, mediaduration
#need to address existing m4a files in the convert directory prior to the m4b to m4a split.

#trap pause DEBUG
shopt -s nullglob


## definitions #################################################################################
[[ "$TERM" = screen ]] && TERM=screen.xterm-256color
[[ "$TERM" = dummy ]] && TERM=xterm-256color
export TERM

threads=4                #this should be assigned when ~/.config/indexopus.conf is sourced
scriptpath="$(realpath "$0")"
scriptname="${scriptpath##*\/}"
ulon="$(tput smul)"
uloff="$(tput rmul)"
bold="$(tput bold)"
tput0="$(tput sgr0)"
red="$(tput setaf 9)"
white="$(tput setaf 15)"
relipsis="$red..."
stamp="$(date +%s)"

## load functions ##############################################################################
## printline()
## pause()
## checkdur() ## reviewed and refactored Tue Jan 14 08:15:13 AM EST 2025

. "$HOME/bin/gits/indexopus/indexopus.lib"

## load defaults ###############################################################################
. "$HOME/.config/indexopus.conf"

fnrmr(){
  local red relipsis="${red}..."
  [[ "$1" = "-r" ]] && red="${tput setaf 7}"

  fn_rmfile(){
    printf '%s\n%s is a file:\n%s' "$relipsis" "$bold" "$1" "$white"
    ls -ltr "$1"
    printf '%s\nrmr: remove this file? (y/n) %s' "$red" "$tput0"
    old_stty_cfg=$(stty -g) # https://stackoverflow.com/questions/226703/how-do-i-prompt-for-yes-no-cancel-input-in-a-linux-shell-script
    stty raw -echo
    answer=$(head -c 1)
    stty "$old_stty_cfg"  # Careful playing with stty

    if [ "$answer" != "${answer#[Yy]}" ]
      then
      rm "$1"
      printf "\n\n"
      if [ ! -e "$1" ]
        then
        printf '\n%sDone! File \"%s\" removed.\n\n%s' "$relipsis" "$1" "$tput0"
      else
        printf "\n\n%s%s%s not removed, check permissions:\n\n%s" "$relipsis" "$bold" "$1" "$tput0"
        ls -ltr "$1"
      fi
    else
      printf '\n\n%sFiles remain:\n%s \n%s\n' "$relipsis" "$(ls -ltr "$1")" "$tput0"
    fi

#	There isn't really anything wrong with this code, but the above method removes the CR
#	There're other ways to do this https://stackoverflow.com/questions/226703/how-do-i-prompt-for-yes-no-cancel-input-in-a-linux-shell-script

   }
  ##--> fn_rmfile() <--###############################################################################


for path in "$@"
  do
    if [[ -d "$path" ]]; then
	printf '%s%s\n' "$red" "$bold"
	if rmdir "$path" 2>/dev/null; then
	   printf 'Removing %s... \n\n$ ls %s/ \n\n' "$path" "$path"
	else
	   printf '\n%s' "$tput0"
	   find "$path"
 	   printf '%s' "$bold"
	   rm -rI "$path"
	   printf '%s' "$tput0"
	fi

	if [ ! -e "$path" ]; then
      	   printf '%s\nDone!\n' "$relipsis"
      	else
      	   printf '\n\n%s%sRemoval Failed!%s\n\n%s is still present check ownership and permissions:\n' "$relipsis" "$bold" "$tput0" "$path"
      	   ls -lad --color=always "$path"
	fi
  elif [[ -f "$path" ]]; then
    fn_rmfile "$path"

  else
    printf '\n\n No such directory or file found!\n\n'

  fi

done
}
##--> fnrmr() <--###################################################################################

editscript(){
  local script path; script="${scriptname##*/}"; path="${scriptname%/*}"; swp="$path/.$script.swp"
     if [[ ! -e "$swp" ]]; then printf "\n\n%s\n\n" "$swp"; (/usr/bin/nano "$scriptname"); exit
     else printf "\n%s is already being edited.\n%s exists; try fg or look in another window.\n" "$scriptname" "$swp"; exit;
  fi; }
##--> editscript() <--##############################################################################

filecheck(){
   clifilename="$1"  #if a filename is passed via the cli.  this doesn't work anymore now with -t??
   [[ -d "$clifilename" ]] && clidir="${clifilename%/*}" && cd "$clidir" && unset clifilename
   [[ -f "$clifilename" ]] && clidir="${clifilename%/*}" && cd "$clidir" && clifilename="${clifilename##*/}"

  if [[ -n "$clifilename" ]] && [[ "$clifilename" != *\ --\ Audiobook.m4b ]]
    then
    printf 'The file specified: %s\nis not in the correct format of\n* -- Audiobook.m4b, checking directory for an m4b file...' "$clifilename"
  elif [[ "$clifilename" == *\ --\ Audiobook.m4b ]] && [[ -f "$1" ]]
    then
    m4bin="$1"
  else
    files=(*.m4[ba])
    if (( "${#files[@]}" == 1 )); then

      # to handle the case where there's an unindexed m4a or a misnamed m4b,
      # for simplicity sake, just rename to m4b
      # probably want to change it back at the end, but it is going to get deleted anyway
      [[ "$files" = *m4a ]] && startm4a="${files}" && mv "$files" "${files%.m4a}.mb4"

      if [[ "${files[@]}" == *\ --\ Audiobook.m4b ]]; then
        m4bin="$files"
      elif [[ "${files[@]}" == *m4b ]]; then
        m4bin="${files/.m4b/ -- Audiobook.m4b}"
        mvoutput="$(mv "$files" "$m4bin" 2>&1)"
        mvec="$?"
        if (( mvec > 0 )); then
          if [[ "$mvoutput" = "mv: cannot move "*\ to\ *": Permission denied" ]]; then
            own && mv "$files" "$m4bin" || { printf 'Unable to gain write permission for %s. exit 1\n\n' "${files[@]}"; exit 1; }
          else
            printf 'Unknown error renaming %s. exit 1\n\n' "${files[@]}"
            exit 1
          fi
        fi
      fi
    else
      echo "No single m4b/m4a input file found or specified, please check and retry..."
      ls
      exit 1
    fi
  fi; }
##--> filecheck() <--#################################################################################

calctime(){
  duration=$(find -type f -name "$1" -print0 |
			xargs -0 mplayer -vo dummy -ao dummy -identify 2>/dev/null |
			perl -nle '/ID_LENGTH=([0-9\.]+)/ && ($t +=$1) && printf "%02d:%02d:%02d\n",$t/3600,$t/60%60,$t%60' |
			tail -n 1)
  hours=$(find -type f -name "$1" -print0 |
			xargs -0 mplayer -vo dummy -ao dummy -identify 2>/dev/null |
			perl -nle '/ID_LENGTH=([0-9\.]+)/ && ($t +=$1) && printf "%02d\n",$t/3600' |
			tail -n 1)
  minutes=$(find -type f -name "$1" -print0 |
			xargs -0 mplayer -vo dummy -ao dummy -identify 2>/dev/null |
			perl -nle '/ID_LENGTH=([0-9\.]+)/ && ($t +=$1) && printf "%02d\n",$t/60%60' |
			tail -n 1)
  seconds=$(find -type f -name "*$1" -print0 |
			xargs -0 mplayer -vo dummy -ao dummy -identify 2>/dev/null |
			perl -nle '/ID_LENGTH=([0-9\.]+)/ && ($t +=$1) && printf "%02d\n",$t%60' |
			tail -n 1)

  [[ "$hours" =~ "0*(.*)" ]] && hours=${BASH_REMATCH[1]}
  [[ "$minutes" =~ "0*(.*)" ]] && minutes=${BASH_REMATCH[1]}
  [[ "$seconds" =~ "0*(.*)" ]] && seconds=${BASH_REMATCH[1]}

#BASH_REMATCH is an array variable in Bash that contains the results of the most recent pattern match operation performed using the =~ operator with the [[ ... ]] conditional expression.
#
#When you use =~ to match a pattern against a string, Bash populates the BASH_REMATCH array with the text that matches the pattern. The first element (BASH_REMATCH[0]) contains the portion of the string that matches the entire regular expression, while subsequent elements (BASH_REMATCH[1], BASH_REMATCH[2], etc.) contain the portions of the string that match the corresponding capture groups in the regular expression.
#
#Here's a breakdown of how BASH_REMATCH is populated:
#
#BASH_REMATCH[0]: Contains the portion of the string that matches the entire regular expression.
#BASH_REMATCH[1]: Contains the portion of the string that matches the first capture group ( ... ).
#BASH_REMATCH[2]: Contains the portion of the string that matches the second capture group ( ... ), if present.
#and so on...
#In the script provided earlier, we used BASH_REMATCH[1] to extract the content inside double quotes. If the pattern \"(.*)\" matches successfully, the content inside the double quotes is captured and stored in BASH_REMATCH[1].

# printf "Duration %s\n" "$duration"

 hours="${hours#0}"
 hourmin=$(( hours*60 ))
 ##echo "hourmin $hourmin"
 minutes="${minutes#0}"
 secmin=$((hourmin+minutes))
 secmin=$((secmin*60))
 seconds="${seconds#0}"
 ##echo "secmin  $secmin"
 ##echo "seconds $seconds"
 totalsec=$(( "$seconds" + "$secmin" ))
 ##echo "totalsec $totalsec"
 divsec=$(python3 -c "print($totalsec/4)")
 ##echo "divsec $divsec"
}
##--> calctime{} <--#############################################################################

#m4bsplit(){
## Original code from: http://crunchbang.org/forums/viewtopic.php?id=38748#p414992
## m4bronto
#
##     Chapter #0:0: start 0.000000, end 1290.013333
##       first   _     _     start    _     end
#local bpath outputdur START first start end partnum n
#outputdir="$2"
#START=1
#bpath="$(basename "$PWD")"
##while [ "$#" -gt 0 ]; do
#
#ffmpeg -i "$1" 2> ".tmp$stamp"
#
#n="$START"
#
#while read -r first _ _ start _ end
# do
#  if [[ "$first" = Chapter ]]
#   then
#      read  # discard line with Metadata:
#      read _ _ chapter
#      [[ "$chapter" = Chapter\ ?\:\ * ]] && chapter="${chapter/Chapter /Chapter 0}"
#
## at some point i'd like to make a menu for what to do here, but for now this works.
##      [[ "$chapter" =~ \ \/\  ]] && chapter="${chapter// \/ /: }"
#      [[ "$chapter" =~ \/ ]] && chapter="${chapter//\//\\}"
#
##      read -rp "$title" </dev/tty
#
#      [[ "$n" -lt 10 ]] && partnum="0$n" || partnum="$n"
#      printf '%sCreating %s -- Part %s: %s.m4a%s\n' "$relipsis" "$title" "$partnum" "$chapter" "$tput0"
#      readarray -t ffoutput < <(ffmpeg -n -nostdin -hide_banner -stats -v error -vsync 2 -ss "${start%?}" -to "$end" -i "$1" -vn -codec copy "$outputdir/$title -- Part $partnum: $chapter.m4a" 2>&1)
#      ((n++))
#  fi
#done <".tmp$stamp"
#
#rm ".tmp$stamp"
#
##shift
##done
#}

m4bsplit() {
  local bpath outputdir START first start end partnum n chapter max_digits
  outputdir="$2"
  START=1
  bpath="$(basename "$PWD")"
  ffmpeg -i "$1" 2> ".tmp$stamp"

  chapters=$(grep -cE 'Chapter #[[:digit:]]:[[:digit:]]{1,3}: start' ".tmp$stamp")

  (( chapters < 100 )) && max_digits=2 || max_digits=3

  n="$START"
  while read -r first _ _ start _ end; do
    if [[ "$first" == "Chapter" ]]; then
      read  # discard the Metadata line
      read _ _ chapter

      chapter="${chapter//\//\\}"  # Escape slashes in chapter titles
      partnum=$(printf "%0${max_digits}d" "$n")

      local outfile="$outputdir/$title -- Part $partnum: $chapter.m4a"
      printf '%sCreating %s\n' "$relipsis" "$outfile"

      ffmpeg -n \
             -nostdin \
             -hide_banner \
             -stats \
             -v error \
             -ss "${start%?}" \
             -to "$end" \
             -i "$1" \
             -vn \
             -codec copy \
             "$outfile"

      ((n++))
    fi
  done < ".tmp$stamp"

  rm ".tmp$stamp"
}

##--> m4bsplit() <--##################################################################################

rmyn(){
 local quiet
 [[ "$1" = '-q' ]] && quiet=true && shift
  while true
  read -rp "$1" rmyn
   do
     case "$rmyn" in
       [yY] ) #"${@:3}"; break;;
         if [[ $quiet ]]; then
            "${@:3}" >/dev/null
          else
            "${@:3}"
          fi
          break;;
       [nN] ) printf '%s\n' "$2"; break;;
          * ) printf "Invalid response...\n";;
     esac
   done
}
##--> rmyn() <--######################################################################################


noindexsplit(){
  local i n partno m4bin="$1" convertdir="$2"
  printf '%s%sNo index found in %s!\n...%s%sThe item will be spiit into %s parts for efficient ffmpeg conversion.\n...Reindex later with chapter information pulled from the web.\n\n' "$relipsis" "$bold" "$m4bin" "$tput0" "$red" "$partno"
    partno=1
    calctime "$m4bin"  #why?  i think this is the only call to calctime?  there's no need for it here...
                       #oh contrare! this is what provides $divsec and the script fails without that line!
    for ((partno=1; partno < $(( threads + 1)); partno++ )); do
       printf '%s%s: %s' "$relipsis" "$partno" "$tput0"
       ffmpeg -n -nostdin -hide_banner -stats -v error -ss $(python3 -c "print(($partno-1)*$divsec)") -to $(python3 -c "print($partno*$divsec)") -i "$m4bin" -vn -acodec copy "$convertdir/${m4bin%% -- *} -- Part 0$partno.m4a"
       printf '%s' "$tput0"
    done
    cd "$convertdir"
    #for i in First Second Third Fourth
    for ((i=1; i<=parts; i++)); do
      printf -v i '%02d\n' "$i"
      ((n++))
      for f in *\ --\ Part\ 0"$n".*; do
        mv "$f" "${f/./: Part $i.}"
      done
    done
}

##--> Main Code <--################################################################################
##--> m4b2opus <--####################################################################################

echo

while (( $# > 0 )); do
  [[ "$1" = @(edit|e|nano) ]] && editscript
  [[ "$1" = "-y" ]] && shift && rmmatch=true
  [[ "$1" = "--stats" ]] && shift && stats=true
  [[ "$1" = @(--screen|-s) ]] && shift && screened=true
  [[ "$1" = @(-ys|-yes) ]] && shift && rmmatch=true && screened=true
  [[ -f "$1" ]] && clifile="$1" && shift  #these don't actually do anything
  [[ -d "$1" ]] && clidir="$1" && shift   #at the moment but they could...there has to be some catchall condition for things not in this list or it will loop endlessly...
done

if "${screened:=false}"; then
  printline "$bold Calling m4b2opus in GNU screen $tput0"
  printf \\n\\n
  allm4s=(*.m4[ab])
  screenname="m4b2opus-${allm4s[0]}"
  echo "m4b2opus ${rmmatch:+$rmmatch}"
  screen -dmS "${screenname:0:16}" m4b2opus "${rmmatch:+-y}" #"${rmmatch:--y }"  #should this be a -y here?  probably.
  screen -ls                                                                     #could probably add some logic to not delete shit
  printf \\n\\n                                                                  #leaving that for later, but screen implies -y as
  exit                                                                           #it doesn't really work otherwise?
fi

printline "${bold}  Welcome to ${white}$script  ${tput0}"

allm4s=(*.m4[ab])

printf '\n%s/*.m4[ab]: \n' "$PWD"
printf ' %s%s%s\n' "$red" "${allm4s[@]}" "$white"

filecheck "$1" && [[ ! "$?" = 0 ]] && exit  # this function returns "$m4bin"

[[ "$stats" = true ]] && mediainfo "$m4bin"

m4bindur="$(checkdur m4b)"                  # this function returns the duration of m4bs

printf '\n%sStarting .m4b file: %s%s%s' "$bold" "$red" "$m4bin" "$tput0"
printf ' %s%s%s\n' "$boldred" "$m4bin" "$tput0"
printf '%sStarting duration :%s' "$bold" "$tput0"
printf ' %s%s%s\n' "$boldred" "$m4bindur" "$tput0"
title="${m4bin% -- *}"
swd="$PWD"
convertdir="$CACHE/m4b2opus-$title"

[[ "$convertdir" = *\'* ]] && convertdir="${convertdir//\'}"  #check $convertdir for \'s and remove them

if [[ ! -d "$convertdir" ]]; then
  mkdir -p "$convertdir" ||
  { printf '%s[ERROR]%s Could not create "%s"; please investigate.\n(exit 1)\n%s' "$boldred" "$white" "$convertdir" "$tput0"; exit 1; }
fi

### CODE REVIEW STOPPED Tue Jan 14 08:30:49 AM EST 2025

if [[ ! $(mediainfo "$m4bin"|grep Menu) =~ ^Menu ]]; then  #process unindexed m4b files
  noindexsplit "$m4bin" "$convertdir"
else
  m4bsplit "$m4bin" "$convertdir"
  cd "$convertdir"
  (( $(ls *m4a|wc -l) > 99 )) && for i in *\ --\ Part\ ??\:\ *.m4a; do mv "$i" "${i/Part /Part 0}"; done
  regexexp='.*\ --\ Part\ [[:digit:]]{2}:\ Chapter\ [[:digit:]][\ |\.]'
  for m4afile in *m4a; do
    [[ "$m4afile" =~ .*\ --\ Part\ [[:digit:]]{2}:\ Chapter\ [[:digit:]][\.\ ] ]] &&
      mv "$m4afile" "${m4afile/Chapter /Chapter 0}"
  done
fi

cd "$convertdir"
echo

[[ "$rmmatch" != true ]] && pause "${bold}${red}Control-Z to suspend and adjust chapter names in $convertdir? $tput0"

#opus.book.4##################################################################################
echo
printline "${bold}  Calling external dependency:${red} opus.book.4  ${tput0}"

opus.book.4 --stats -d "${m4bindur#*: }"  #this is a dependency; needs to be made a function?
if (( "$?" > 0 ));  then
  printf '%s\n\nopus.book.4 failed; exit 1\n\n%s' "$bold" "$tput0"
  printline "  Leaving $scriptname: Error!  "
  exit 1
fi

printline "${bold}  Returning from external dependency: ${red} opus.book.4  ${tput0}"
#opus.book.4##################################################################################

clear -x
#need to renable rmyn below if these are uncommented
#[[ ! -d "$convertdir/m4a" ]] && mkdir "$convertdir/m4a"
#mv *.m4a ./m4a
opuspartdur="$(checkdur opus)"
m4bindur="${m4bindur#*: }"

printf '%sDuration of the original m4b file: %s%s%s\n' "$bold" "$red" "$m4bindur" "$tput0"
printf '%sDuration of the  .opus part files: %s%s%s\n' "$bold" "$red" "$opuspartdur" "$tput0"

##indexopus##################################################################################

echo
printline "${bold}  Calling external dependency:${red} indexopus  ${tput0}"

if [[ "$rmmatch" = true ]]
  then
    indexopus -y -t "$title" -m "$convertdir" -s "$stamp"	-d "$opuspartdur"
  else
    indexopus -t "$title" -m "$convertdir" -s "$stamp" -d "$opuspartdur"
fi
   ec4io="$?" 				# error code 4 indexopus
   if (( ec4io == 2 )); then
     iorm=true
   elif (( ec4io == 1 )); then
     printf 'How the hell did you manage to get here?!'
     if ! [[ -f "$convertdir/$title -- Audiobook.opus" ]]; then
       pause "\$convertdir/\$title -- Audiobook.opus: $convertdir/$title -- Audiobook.opus"
       exit
     fi
   elif (( ec4io == 130 )); then
     printf '%sindexopus was aborted by user, exit (130). Investigate...\n%s\n%s\n\n' "$bold" "$tput0" "$(printline "  ${bold}Leaving ${scriptname}: ${red}indexopus aborted, exit (1)${tput0}  ")"
     exit
   fi


printline "${bold}  Returning from external dependency:${red} indexopus  ${tput0}"
##indexopus##################################################################################


tmp="/tmp/indexopus-$stamp"  #this probably doesn't exist...
titleopus="$title -- Audiobook.opus"

[[ ! -f "$titleopus" ]] &&
   printf '%s%sError: Where the fuck is %s?! Exiting (1)\n%s' "$bold" "$red" "$titleopus" "$tput0" &&
     printline "$scriptname" &&
     exit 1

mv "$titleopus" "$swd"
opusparts=( *opus )
cd "$swd"

[[ ! "$rmmatch"=true ]] && pause "Need to check to see if $tmp is being created... when indexopus isn't run with -y"

checkdur # probably do something different here since you have m4bindur and have to calc opusdur...
         # yeah, this is a terrible idea.  do this by heand instead of relying on something no
         # indended for this

#echo "Duration(s):"

#echo "${red}${bold}${tput0}$(m4bindur) the $m4bin"  #this threw an error saying m4bindur could not be found: not a function?

#echo "${red}${bold}${tput0}$m4bindur $m4bin"     #removed the () from above.
#echo "${red}${bold}${tput0}$(checkdur opus) $titleopus"

opusdursec="$(sex2sec "$(checkdur opus)")"
m4bindursec="$(sex2sec $m4bindur)"
if [[ "$rmmatch" = true ]] && (( opusdursec > 0 )); then
#  verifydur=$(bc <<< "scale=4; ($m4bindursec - $opusdursec) / $opusdursec " )
  verifydur=$(( (1000 * (m4bindursec - opusdursec)) / opusdursec ))
  verifydur="${verifydur/#-}"
#  if [[ $(bc <<< "scale=4; $verifydur < 0.001") ]]; then
  if (( verifydur < 1 )); then  #this is 0.1% difference
    rm -r *m4b "$convertdir"
    [[ "$m4bindursec" != "$opusdursec" ]] &&
       printf '\n%sDurations don'\''t %sexactly%s match but fuckin'\'' close '\''nuff!%s\n' "$relipsis" "$(tput sitm)" "$(tput ritm)" "$tput0" ||
       printf '%sDurations match!%s\n' "$relipsis" "$tput0"
    printf '\n%sDone.\n\nPlease see %s/%s%s\n\n' "$relipsis" "$PWD" "$ioutput" "$tput0"
    exit
  elif [[ "$rmmatch" = true ]] && [[ "$(checkdur m4b)" != "$(checkdur opus)" ]]; then
    printf '%s%sThe m4b and opus audiobook durations do not match!\n' "$bold" "$red"
    printf '%s%sm4b2opus will not automatically delete starting and temporary files!\n' "$white" "$tput0"
    printf '\n\n'
    printf '%s-y/-f specified, but the durations do not match!\n%s' "$bold" "$tput0"
  fi
fi

  printf '\n%sCleaning up remaining files:%s\n\n' "$red" "$tput0"

#  ls "$tmp" "$convertdir" 2>/dev/null

  printf '\n\n'
# the 2>/dev/null does not work for rmyn -- rewrite!!
#  rmyn -q "${red}${bold}Remove the conversion ./m4a directory? ${tput0}" "${red}...Leaving ./m4a.${tput0}" rm -r "$convertdir/m4a"
  [[ ! "$iorm" ]] && [[ "$opusparts" ]] && rmyn -q "${red}${bold}Remove $ititle -- Part *.opus files? ${tput0}" "${red}...Leaving $ititle -- Part *.opus files...${tput0}" rm "$convertdir/$title"\ --\ Part\ ??:*.opus
  rmyn "${red}${bold}Remove the starting $m4bin? " "${red}Leaving starting $m4bin${tput0}" rm "$m4bin"
  for i in "$convertdir" .tmp* "$tmp"; do [[ -a "$i" ]] && fnrmr "$i"; done
  printf '\n\n...Done.\n\nPlease see %s/%s\n\n' "$PWD" "$ioutput"

printline "${bold}  Exiting (0) ${white}m4b2opus  ${tput0}"
exit 0