shrink

#!/bin/bash
source /home/fabian/d/programs/bash_scripts/sane
# This script forces a video below a certain size, with the best quality possible for that size. It also converts to MP4, because why not.
# Exit value is 3 on FFMPEG error, 2 if a so far unsupported file type was provided (currently only certain videos work), 1 if the file couldn't be made small enough, otherwise 0 indicates success and then the last line of output is the path/file that worked. To get the filename in a script without letting "tail" overwrite the exit value, use one of these:
# temp=$(resize.sh); if(($?==0)); then echo $temp | tail -n 1; fi
# if temp=$(resize.sh); then echo $temp | tail -n 1; fi
# Effective working directory will be the one with the input file in it.

### INITIALISATION AND VALIDATION
## interactive or command line parameters
if [[ -r "$1" && ( "$2" == "d" || "$2" == "t" || "$2" == "g" || "$2" == "b" || "$2" =~ ^[0-9]+$ )]]; then
 # input file path/name
 in="$1"
 # maximum file size allowed, examples:
 if [[ "$2" == "d" ]]; then
 # Discord: 25MiB=25*1024² Bytes
  limit=26214400
 elif [[ "$2" == "t" ]]; then
 # Telegram: 2000 MiB=2000*1024² Bytes
  limit=2097152000
 elif [[ "$2" == "g" ]]; then
 # GermanZero Sharepoint: 2 SI GB=2*1000³ Bytes
  limit=2000000000
 elif [[ "$2" == "b" ]]; then
 # 32bit systems: <2GiB→2*1024³-1 Bytes
  limit=2147483647
 else
  limit=$2
 fi
else
 echo -n "Input file: "
 while true; do
  read in
  if [[ -r "$1" ]]; then break; fi
  echo "Invalid or unreadable, try again or abort with Ctrl+Esc."
 done
 echo -n "File size limit in bytes: "
 while true; do
  read limit
  if [[ "$2" =~ ^[0-9]+$ ]]; then break; fi
  echo "Not an integer, try again or abort with Ctrl+Esc."
 done
fi
# If this is set to "true", the files that resulted from all the attempts except for the best one are deleted. If set to "false", they are kept. This does not guarantee that any specific file will exist (but it is predictable, for example if CRF 18 is the best, then the files with CRF 51, 25, 12, 18, 15, 16 and 17 will exist).
if [[ "$3" == "false" ]]; then
 delete_intermediate_results=false
else
 # default is true
 delete_intermediate_results=true
fi

# start values, crf=51 is worst possible
good=0
bad=51
curr=51

# filename without extension
name="${in%.*}"
# file extension
ext="${in#*.}"

# reason: CRF 0..51 range apparently only apples to h264
if ! ffprobe "$in" 2>&1 | grep -q h264; then
 echo "Not a supported file! Currently only h264 (usually .mp4) video is supported. ffprobe output:"
 ffprobe "$in"
 exit 2
fi

output(){
 echo "This file succeeded ($(stat --printf="%s" "$out")<$limit):"
 echo "$out"
 exit
}

in_size=$(stat --printf="%s" "$in")
if((in_size<limit)); then
 echo "Input is already below limit!"
 out="$in"
 output
fi

### MAIN PART

echo "Input file size: $in_size"

while true; do
 out="$name""_$curr.$ext"
 echo -n "$(date +"%H:%M:%S") testing crf=$curr, filename: $name""_$curr.$ext"
#############################################################
# This is where the script will spend 99% of its time!      #
 ffmpeg -nostdin -i "$in" -crf $curr -c:s copy -v 16 "$out" #
#############################################################
 if(($?!=0)); then
  echo "\nFFMPEG error!"
  exit 3
 fi
 size=$(stat --printf="%s" "$out")
 echo -n ", resulting size: $size"
 if((size>limit)); then
  if((curr==51)); then
   echo ">$limit, even with CRF=51"
   if [[ $delete_intermediate_results == true ]]; then rm "$out"; fi
   exit 1
  fi
  echo ">$limit"
   # If the current result is above the limit, it gets deleted. Since some valid result should already exist with CRF=51 or whatever else happened later (except if the script fails completely), there will always be a file left.
  if [[ $delete_intermediate_results == true ]]; then rm "$name""_$curr.$ext"; fi
  good=$curr
 else
  echo "≤$limit"
   # If the current result is below the limit, then the previous worse/smaller result is no longer needed.
  if [[ $delete_intermediate_results == true && $curr != 51 ]]; then rm "$name""_$bad.$ext"; fi
  bad=$curr
 fi

 # Stop searching if narrowed down to exact number, pick best value that is "worse" than the limit, which luckily already is the current value of $bad.
 # CRF 0 will never be picked. Normally it shouldn't make the file smaller anyway and it leads to issues in some programs.
 if((bad-good<2)); then
  out="$name""_$bad.$ext"
  output
 fi
 curr=$(((good+bad)/2))
done