From e995a596621e5405b55ace787a5efb354283c1f9 Mon Sep 17 00:00:00 2001
From: Dan Snyder <dan.snyder.3rd@gmail.com>
Date: Sun, 16 Feb 2025 16:59:20 -0500
Subject: [PATCH] Initial version of the script

---
 reencode.sh | 190 ++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 190 insertions(+)
 create mode 100644 reencode.sh

diff --git a/reencode.sh b/reencode.sh
new file mode 100644
index 0000000..ea6803c
--- /dev/null
+++ b/reencode.sh
@@ -0,0 +1,190 @@
+#!/bin/sh
+
+#
+# Section: Argument parsing
+#
+
+options="sim run full"
+
+if [ $# -ne 2 ]; then
+  echo "reencode <$( echo "$options" | sed 's/ /|/g' )> <directory>"
+  exit 1
+fi
+
+mode=""
+
+for opt in $options; do 
+	if [ "$opt" = "$1" ]; then 
+		mode="$opt"
+	fi
+done
+
+if [ "$mode" = "" ]; then
+	echo "'$1' is not a valid option"
+	exit 1
+fi
+
+path="$2"
+cd "$path"
+
+if [ $? -ne 0 ]; then
+	echo "Path does not exist: $path"
+	exit 1
+fi 
+
+echo "Searching for MKVs in $( pwd )"
+
+#
+# Section: Functions
+# 
+
+getCodec() {
+	ffprobe -v error -select_streams v:0 -show_entries stream=codec_name -of compact=p=0:nk=1 "$1"
+}
+
+fileSize() {
+	du -sb "$1" | awk '{print $1}'
+}
+
+getBitrate() {
+	ffprobe -v quiet -select_streams v:0 -show_entries format=bit_rate -of compact=p=0:nk=1 "$1"
+}
+
+formatSize() {
+	echo "$1" | numfmt --to=iec-i | awk '{print $1"B"}'
+}
+
+formatRate() {
+	echo "$1" | numfmt --to=si | awk '{print $1"bps"}'
+} 
+
+encodeFile() {
+	process=$( echo $1 | awk -v FPAT="([^' ]+)|'([^']+)'" '{print $1}' )
+	filesize=$( echo $1 | awk -v FPAT="([^' ]+)|'([^']+)'" '{print $2}' )
+	bitrate=$( echo $1 | awk -v FPAT="([^' ]+)|'([^']+)'" '{print $3}' )
+	codec=$( echo $1 | awk -v FPAT="([^' ]+)|'([^']+)'" '{print $4}' )
+	infile=$( echo $1 | awk -v FPAT="([^' ]+)|'([^']+)'" '{print $5}' | sed "s/^'\|'\$//g" )
+	outfile=$( echo $1 | awk -v FPAT="([^' ]+)|'([^']+)'" '{print $6}' | sed "s/^'\|'\$//g" )
+
+	if [ "$process" = "Ignore" ]; then
+		return 1
+	fi
+
+	tmpfile="$infile.bak"
+
+	mv "$infile" "$tmpfile"
+
+	if [ $? -ne 0 ]; then
+		echo "Failed to make backup file: $infile"
+		return 2
+	fi
+
+	ffmpeg -nostdin -hide_banner -i "$tmpfile" -c:v hevc_nvenc -c:a copy -c:s copy -map 0 -vtag hvc1 "$outfile"
+
+	if [ $? -ne 0 ]; then
+		echo "FFMPEG re-encode failed: $infile"
+		echo "Restoring from backup"
+		mv -f "$tmpfile" "$infile"
+		return 1
+	fi
+
+	echo "Output file: $outfile"
+
+	rm -f "$tmpfile"
+	oldsize=$( echo "$filesize" | formatSize )
+	oldbitr=$( echo "$bitrate" | formatRate )
+	newsize=$( fileSize "$outfile" | formatSize )
+	newbitr=$( getBitrate "$outfile" | formatRate )
+	echo "Converted $infile:" >> summary.txt
+	echo "\t$oldsize ($oldbitr) => $newsize ($newbitr)\n" >> summary.txt
+	return 0
+}
+
+getFileInfo() {
+	# Args:
+	# 	$1 - video file name
+	#
+	# Returns a single line:
+	# 	filesize bitrate codec
+	echo "$( fileSize "$1" ) $( getBitrate "$1" ) $( getCodec "$1" )"
+}
+
+selectFile() {
+	# Arg: 
+	# 	$1 - filename
+	# 	$2 - filesize
+	# 	$3 - bitrate
+	# 	$4 - codec
+	# 	
+	# Note: args line up with return values from getFileInfo
+	#
+	# Echo "Ignore" or "Process" with an optional reason
+	file=$1
+	size=$2
+	bitrate=$3
+	codec=$4
+
+	#if [ "$codec" = "hevc" ]; then
+	#	echo "Ignore Already HEVC"
+	#	return
+	#fi
+
+	if [ $bitrate -lt 4000000 ]; then
+		echo "Ignore bitrate too low"
+	else
+		echo "Process"
+	fi
+}
+
+simFile() {
+	# Arg:
+	# 	$1 - video file name
+	#
+	# Prints out a line to simulate.txt formatted like so:
+	# 	process filesize bitrate codec infile outfile
+	fileinfo=$( getFileInfo "$1" )
+	selected=$( selectFile "$1" $fileinfo )
+	outputfile="${1%.*}.mkv"
+
+	reason=${selected#* }
+	selected=$( echo "$selected" | awk '{print $1}' )
+
+	if [ "$selected" = "Ignore" ]; then
+		echo "$selected $fileinfo '$1' '$reason'" >> simulation.txt
+		echo "Ignored $1"
+		echo "\t => $reason"
+	else
+		echo "$selected $fileinfo '$1' '$outputfile' '$reason'" >> simulation.txt
+		echo "Selected $1"
+	fi
+}
+
+#
+# Section: Main code - do mode-specific stuff
+#
+
+if [ "$mode" != "run" ]; then
+	# if mode is not run (sim or full) then we're doing a simulation
+	rm -f simulation.txt
+
+	find -type f \( -name "*.mkv" -o -name "*.mp4" \) | while read file; do
+		simFile "$file"
+	done
+fi
+
+if [ "$mode" != "sim" ]; then 
+	# if mode is not sim (run or full) then we're actually executing re-encodings
+	rm -f output.mkv
+
+	if [ ! -f "simulation.txt" ]; then
+		echo "No simulation file found. Cannot proceed."
+		exit 1
+	fi
+
+	cat simulation.txt | while read line; do
+		encodeFile "$line"
+		if [ $? -ne 0 ]; then
+			exit 2
+		fi
+	done
+fi