#!/bin/sh

set -e

currDir=$pwd
framespath=${1:-}
lang=${2:-deu}

[ -z "$framespath" ] && echo "no framespath provided" && exit 1

parentDIR=$(dirname "$framespath")
textDIR="$parentDIR"/text
grayDIR="$parentDIR"/gray
mkdir -p ${textDIR} ${grayDIR}

# split video on frames (every 90 seconds)
# ffmpeg -i "$videopath" -r 0.011 frames/ffmpeg_%0d.jpg
# yes | ffmpeg -i "$videopath" -ss 00:00:05 -vframes 1 frames/ffmpeg_2.jpg

# lets say we dont know extention
# ext=".jpg"


# convert each frame to grayscale then crop to text if any
for img in "$framespath"/*
do
    iname=$(basename "$img")
    sub_img_name=$"${grayDIR}/sub_${iname}"
    sub_out="$textDIR/$(echo "$iname" | sed 's/.\(png\|gif\|jpg\|jpeg\|bmp\)//')"
    magick convert "$img" -chop 220x0 -gravity East -chop 220x0 \
        -gravity South -chop 0x50 \
        -colorspace Gray -resize 600x "$sub_img_name"
    # dpi=$(magick identify -format '%x' $sub_img_name)
    # echo "$sub_img_name - dpi: $dpi"
    # dpi=300

    printf "\r%s" "$sub_img_name"

    # call tesseract to get the text
    # tesseract -l $lang --dpi $dpi $sub_img_name $sub_out
    tesseract -l "$lang" "$sub_img_name" "$sub_out"
done

resultfile=$(basename "$framespath")
find "$textDIR" -type f | sort -n | xargs cat > "${parentDIR}/${resultfile}.txt"
cd $currDir

./remove_dups.py "$textDIR" "${parentDIR}/${resultfile}_clean"

# rm -rf text
# rm -rf frames