pdftomp4

#!/bin/bash

start=1  # start page (from)
end=1000  # end page (to)

win_height=400 # the height of sliding window (view port)
stride=32 # the number of pixels to slide

cut_left=100 # left margin to crop
cut_top=100   # top margin to crop

cols=0  # number of paper columns 


direction="ltr"
default_delay=0.16
start_point=0
remove="none"
del_all="no"
negate=""
trim="none"
scale_h=100
cuts=""
dups="none"
linger=1
linger_frame=20
top_trim="auto"
skip=0
frame_rate=24
crf=17
split=500
from=0
last_page=0
base_folder="date"
update_rate=0
mode="1x"
support_lines="yes"

while [ $# -gt 0 ]; do

if [[ $1 == *"--"* ]]; then
param="${1/--/}"
declare $param="$2"
# echo $1 $2 // Optional to see the parameter:value result
fi

shift
done

if [ ! -f "$input" ]; then
echo "$input doesn't exist"
echo "Please provide the input file using --input option"
exit 1
fi
if [ -z ${reset+x} ]; then 
echo "reset is unset"; 
else 
remove="all"
echo "reset is set to '$reset'"; 
fi


if [ -z ${delay+x} ]; then 
 delay=$default_delay  
fi


start_page=$start
end_page=$end

echo "input = $input, remove = $remove, start=$start_page end = $end_page"

#echo "./conv.sh " $@ >> "$input".sh


cut_pages=()
cut_locs=()
for elem in $cuts; do
	arr=(${elem//:/ })
	p=${arr[0]}
	loc=${arr[1]}
	cut_pages=( "${cut_pages[@]}" "$p" ) 
	cut_locs=( "${cut_locs[@]}" "$loc" )
done


function get_index() {
	value="$1"
	shift
	local index=0
	for i in "$@"; do
	if [[ "$i" = "${value}" ]]; then
	echo "${index}";
	return
	fi
	((index++))
	done
	return 1
}

fullpath=$input
filename="${fullpath##*/}"                      # Strip longest match of */ from start
dir="${fullpath:0:${#fullpath} - ${#filename}}" # Substring from 0 thru pos of filename
base="${filename%.[^.]*}"                       # Strip shortest match of . plus at least one non-dot char from end
ext="${filename:${#base} + 1}"                  # Substring from len of base thru end

pages_dir="$dir""$base"_pages

echo "directory is $pages_dir"

if [[ $remove = "pages" ]] || [[ $remove = "all" ]]; then
	echo "Removing $pages_dir"
	rm -r "$pages_dir"
fi


echo "delay:$delay win_hegith:$win_height stride:$stride"
dt=$(date '+%d-%m %H');

if [[ ! -d "$pages_dir" ]]; then
	echo "Generating pages.. "
	mkdir "$pages_dir"
	if [[ $end -ne 1000 ]]; then
		pdftoppm "$fullpath" -f $start -l $end -png "$pages_dir"/page 
	else
		pdftoppm "$fullpath" -png "$pages_dir"/page
	fi
	if [[ $trim != "none" ]]; then
		for img in "$pages_dir"/*.png; do    
			convert "$img"  "$img"
		done      
	fi

	
	image_files=( "$pages_dir"/*.png )
	
	IFS=$'\n' image_files=($(sort -V <<<"${image_files[*]}"))
    unset IFS
	
	all_images_num=${#image_files[@]}
	current="${image_files[0]}"
	page_width=$(identify -format "%w" $current)
	page_height=$(identify -format "%h" $current)
	pad_x=$((44-(page_width%4)))

	echo "page_width:$page_width pad_x:$pad_x "

	page_width=$((page_width + pad_x))
	middle=$((page_width /2))

	trim=0
	if [[ $cols -eq 0 ]]; then   
		convert "$current" -crop +0+600 +repage -alpha off -colorspace gray -negate -scale x1\! -threshold 2% txt: > "$pages_dir"/col_detect.txt
		filled_seen=0
		x_begin=0
		x_end=0
	  counter=0
		while IFS= read -r line; do
			line_arr=($line)
			pos=${line_arr[0]}
			color=${line_arr[3]}
			if [[ $color = "white" ]] || [[ $color = "gray(255)"  ]] ; then # filled space
				pos_arr=(${pos//,/ })
			  xpos=${pos_arr[0]}
				if [[ $x_begin -gt 0 ]]; then
					x_end=$xpos
					echo "x_end: $x_end"
					break;
		  else
			if [[ $trim = 0 ]]; then
					  trim=$((xpos - 20))                      
					  echo "Trim 1: $trim"
			fi
				fi
				filled_seen=1
		  counter=0
			else # empty space
				if [[ $filled_seen -eq 1 ]] && [[ $x_begin -eq 0 ]]; then
					pos_arr=(${pos//,/ })
					xpos=${pos_arr[0]}
			counter=$((counter + 1))
			if [[ $counter -gt 10 ]]; then
					if [[ $xpos -gt $((page_width / 3)) ]] && [[ $xpos -lt $((2*(page_width / 3))) ]]; then
						x_begin=$xpos
						echo "x_begin: $x_begin"
					else
				if [[ $xpos -lt $((page_width / 3)) ]]; then        
						 trim=0
						   echo "Trim 2: $trim, xpos:$xpos"
				fi
					fi
			fi
				fi
			fi 
			
		done < "$pages_dir"/col_detect.txt
		
		if [[ $x_end -gt 0 ]]; then
			middle=$(((x_begin + x_end) / 2))
			middle=$((middle + (pad_x/2)))
			cols=2
			echo "Two columns detected ... middle: $middle"
		fi      
	fi  

	if [[ $cols != 2 ]]; then
	  scale_h=200
	else
		# if [[ $negate = "" ]]; then
			# negate="all"
		# fi
		convert "$current" -alpha off -colorspace gray -negate -scale 1x\! -threshold 2% txt: > "$pages_dir"/vert.txt  
		filled_seen=0
		counter=0
		v_cut=0
		while IFS= read -r line; do
			line_arr=($line)
			pos=${line_arr[0]}
			color=${line_arr[3]}
			pos_arr=(${pos//,/ })
			ypos=${pos_arr[1]}
			ypos=${ypos%?}
			if [[ $color = "white" ]] || [[ $color = "gray(255)"  ]] ; then # filled space
				filled_seen=1
				counter=0
			else # empty space
				if [[ $filled_seen -eq 1 ]]; then #empty space after filled space
					counter=$((counter + 1))
					if [[ $counter -gt 30 ]]; then
						if [[ $ypos -lt $((page_height / 3)) ]]; then        
							v_cut=$((ypos - 10))
						fi
					fi
				fi
			fi 		
		done < "$pages_dir"/vert.txt	
		if [[ $v_cut -gt 0 ]]; then
			echo "V_cut is $v_cut "
			p=1
			loc=$v_cut
			cut_pages=( "${cut_pages[@]}" "$p" ) 
			cut_locs=( "${cut_locs[@]}" "$loc" )
		fi
	fi 

	ff="%02d"
	if [[ $all_images_num -gt 99 ]]; then
	ff="%03d"
	fi

	do_negate=""
  bgcolor="white"
	if [[ $negate = "all" ]]; then
    echo "Nagation is for all pages"
		do_negate="-negate"
    bgcolor="black"
	fi 
 
	do_trim=""
	if [[ $top_trim = "auto" ]]; then
		do_trim="-trim +repage -bordercolor $bgcolor -border 10x20"
	fi



	shave=""
	win_half=$((winh / 2))
	if [[ $trim != "none" ]]; then
		shave="$trim"x0
		shave="-shave $shave +repage"		
	  echo "Shaving at $trim"
	fi
	
	for ((ii=0; ii<${#image_files[@]}; ii++)); do
		img=${image_files[$ii]}
	    iii=$((ii+1))
		pnum=$(printf $ff $iii)
		convert "$img" $shave $do_negate -adaptive-resize "$page_width"x +repage "$img" 
		
		if [[ ii -eq 0 ]]; then
		   # convert "$img" -pointsize 24 -fill white \
		   # -undercolor '#0004' -gravity "North" \
		   # -annotate +10+20  "$dt | $all_images_num | $stride" \
		   # "$img"
		    echo "first page"
			convert                 \
			 -background '#0004'    \
			 -gravity Center         \
			 -fill white            \
			 -size 500x100     \
			  caption:"$dt | $all_images_num | $stride"      \
			  "${img}"              \
			 +swap                  \
			 -gravity North         \
			 -composite             \
			  "${img}"		
		fi
		if [[ $support_lines = "yes" ]]; then
			convert "$img" \
			\( -size 256x250 xc:"#0502" -size 256x20 xc:none -append -write mpr:tile +delete \) \
			\( -clone 0 -tile mpr:tile -draw "color 0,0 reset" \) \
			-compose over -composite \
			"$img"
		fi
		
		
		
		num=${pnum#0}
		if [[ $cols -eq 2 ]]; then
		cut=0
		if page_index=$(get_index $num "${cut_pages[@]}"); then
			echo "Cutting page $pnum ..."
			cut="${cut_locs[$page_index]}"
		fi
		if [[ $cut -gt 0 ]]; then          
			convert "$img" -crop x"$cut"+0+0 +repage $do_trim -adaptive-resize "$page_width"x "$pages_dir"/page-"$pnum"-cut-0.png           
		fi
	  safe_margin=0
		m1=$((middle + $safe_margin))
		m2=$((middle - $safe_margin))
		col1="$m1"x"0"+0+"$cut"
		col2="$m1"x"0"+"$m2"+"$cut"
		
		if [[ $direction = "rtl" ]]; then
			col1="$middle"x"0"+"$middle"+"$cut"
			col2="$middle"x"0"+0+"$cut"
		fi
		convert "$img" -crop $col1 +repage $do_trim -background black -adaptive-resize "$page_width"x  "$pages_dir"/page-"$pnum"-cut-1.png           
		convert "$img" -crop $col2 +repage $do_trim -background black -adaptive-resize "$page_width"x  "$pages_dir"/page-"$pnum"-cut-2.png           
		rm "$img"
		else
			convert "$img" $do_trim +repage -adaptive-resize "$page_width"x "$img" 		
		fi
		if [[ $dups != "none" ]]; then	
			if [[ $dups = "all" ]]; then
				cp "$img" "$pages_dir"/page-"$pnum"-2.png
				mv "$img" "$pages_dir"/page-"$pnum"-1.png	
			else 
				if [[ " ${dups[@]} " =~ " ${num} " ]]; then
					echo "Duplicating page $pnum ..."
					cp "$img" "$pages_dir"/page-"$pnum"-2.png
					mv "$img" "$pages_dir"/page-"$pnum"-1.png	
				fi
			fi
		fi
	done
fi

image_files=( "$pages_dir"/*.png )
all_images_num=${#image_files[@]}

if [[ $end_page -eq 1000 ]]; then
	end_page=$all_images_num
	if [[ $cols = 2 ]]; then
		end_page=$((all_images_num/2))
	fi
fi

factor=$((win_height/stride))
factor=5
first_delay=$(echo $delay+5 | bc)
beg=$((0 + win_height)) 

frames_dir="$dir""$base"_frames_"$stride"

if [[ $remove = "frames" ]] || [[ $remove = "all" ]]; then
	echo "Removing $frames_dir"
	rm -r "$frames_dir"
fi
shopt -s nullglob

folders=("$dir""$base"_frames_*)

echo "$dir""$base"_frames_*
exist=1
if [ ${#folders[*]} -ge 1  ]; then
	frames_dir=${folders[0]}
    echo "Frames ($frames_dir) aleardy exist, it adds new frames to it"
	frames=($(ls -a "$frames_dir"/*.png | head -2))
	echo "frames : ${frames[@]}"
	first_frame=$(basename ${frames[0]})
	first_frame="${first_frame%.*}"
	sec_frame=$(basename ${frames[1]})
	sec_frame="${sec_frame%.*}"
	
	echo "first_frame: $first_frame, sec_frame:$sec_frame"
	
	arr=(${first_frame//_/ })
	num1=${arr[3]}
	num1=$((10#$num1))

	arr=(${sec_frame//_/ })
	num2=${arr[3]}
	num2=$((10#$num2))
	
	dist=$((num2 - num1))
	start_point=$((dist /2)) 
	stride=$dist
	delay=$((dist / 100)) | bc -l
	
	rm "$frames_dir"/*.txt
	mv "$frames_dir" "$dir""$base"_frames_"$start_point"
	frames_dir="$dir""$base"_frames_"$start_point"
	echo "stride: $stride, start_point: $start_point, delay:$delay"
else
    exist=0
	mkdir "$frames_dir"	
fi



dt=$(date '+%d-%m');
echo "$dt"

page_num=-1

echo "Generating images for $start_page to $end_page"
first=-1
image_files=( "$pages_dir"/*.png )

echo "${#image_files[@]}  images in $pages_dir"
page_height=0
col_num=1
frame=-1
mdelay=$delay
last_frame=""
for ((ii=0; ii<${#image_files[@]}; ii++)); do

	page=${image_files[$ii]}
	filename="${page##*/}"
	name=$(echo "$filename" | cut -f 1 -d '.')
	arr=(${name//-/ })
	num=${arr[1]}
	new_page=0

	num=$((10#$num))
	if [[ $num != ${page_num#0} ]]; then
		new_page=1
		frame=-1
	fi

	page_num=$num
	echo "page number : $page_num, $start_page, $end_page"
	if [[ $num -lt $start_page ]]; then
		continue
	fi
	if [[ $num -gt $end_page ]]; then
		break
	fi
	echo "Processing page $page_num of $end_page"

	prev_page_height=$page_height

	page_height=$(identify -format "%h" $page)
	page_width=$(identify -format "%w" $page)
	#read page_width page_height < <(identify -format "%w %h" $page)
		
	if [[ $first -lt 0 ]] && [[ $ii -eq 0 ]]; then
		cp "${image_files[$ii]}" "$base"_temp.png
		current="$base"_temp.png
		first=1	
		limit=$page_height
		top=$start_point
		mdelay=$first_delay
	else
		# echo "combining..."
		t1="${image_files[$ii-1]}"
		#t1="$last_frame"
		t2="${image_files[$ii]}"    
		if [[ $negate = "even" ]]; then
			if [ $((ii%2)) -eq 0 ]; then
				convert -negate "$t2" t2.png
				t1="${image_files[$ii-1]}"
				t2="t2.png"
			else 
				convert -negate "$t1" t1.png
				t1="t1.png"
				t2="${image_files[$ii]}"
			fi
		fi
		convert "$t1" "$t2" -append "$base"_temp.png
		current="$base"_temp.png
		limit=$((prev_page_height + page_height))
		top=$((limit-page_height-win_height+stride+start_point))
		#top=$((stride+start_point))
		# echo "top :$top"
	fi 
	#echo "current image: $current top:$top"
	bottom=$((top+win_height))
	left=0
	win_width=$page_width

	pnum=$(printf "%03d" $num) 
	iii=$(printf "%03d" $ii)

  # echo "$name"
  list_file="$frames_dir"/list-"$start_point"-"$pnum"-"$iii".txt
		
	if [[ -f $list_file ]]; then
		continue
	fi
  # echo "limit: $limit"
	while [  $bottom -lt $limit ]; do
		frame=$((frame+1))
		i=$((frame+1000))
		ttt=$(printf "%04d" $top)
		fname="page_"$pnum"_"$iii"_"$ttt".png"
								
		# if [ -f "$frames_dir"/$fname ]; then
			# top=$((top+stride))
			# bottom=$((top+win_height))
			# continue
		# fi

		crop="$win_width"x"$win_height"+"$left"+"$top"				

		scaledH=$((win_height+scale_h))
		# -adaptive-resize x"$scaledH"\!
		convert $current -crop "$crop" +repage -quality 99 -adaptive-resize x"$scaledH"\! +repage -pointsize 24 -fill white -undercolor '#56f' -gravity "NorthEast" -annotate +10+20  "$page_num" "$frames_dir"/$fname
		
		# last_frame="$frames_dir"/$fname
		top=$((top+stride))
		bottom=$((top+win_height))
		#echo "$bottom of $limit"
	done #while
	echo "done" >> $list_file
done

rm "$base"_temp.png
echo "$stride $start_point $delay" >> "$base"_"$dt".txt


./frames2mp4 --frames "$frames_dir" --start_page $start_page --end_page $end_page --bg