├── .gitignore
├── README.md
└── summarize-movie-dialog.sh


/.gitignore:
--------------------------------------------------------------------------------
1 | /in
2 | /tmp
3 | /out
4 | /.idea


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
 1 | # movie-dialog-summarizer
 2 | Cuts movie dialog summary video from video file and subtitles.
 3 | 
 4 | # Examples
 5 | - [The Shawshank Redemption](https://youtu.be/FZdDk7A4t1A)
 6 | - [The Gotfather Part I](https://youtu.be/6pY6qu0AZ2Y)
 7 | - [The Dark Knight](https://youtu.be/Px-f24xC0q0)
 8 | - [Schindler's List](https://youtu.be/3JdCfARm4IQ)
 9 | - [12 Angry Men](https://youtu.be/dbfdSBm9jmU)
10 | - [The Fellowship Of The Ring](https://youtu.be/75M5UUtTUgA)
11 | 
12 | # Technologies
13 | - [LexRank](https://en.wikipedia.org/wiki/Automatic_summarization#TextRank_and_LexRank)
14 | - [Sumy](https://github.com/miso-belica/sumy)
15 | - [ffmpeg](https://github.com/FFmpeg/FFmpeg)
16 | 
17 | # Usage
18 |     ./summarize-movie-dialog.sh [video file] [subtitles ".srt" file]
19 | 


--------------------------------------------------------------------------------
/summarize-movie-dialog.sh:
--------------------------------------------------------------------------------
  1 | getStamps () {
  2 | 	grep -F -B1 "${match}" "tmp/$name-subtitles.txt" |
  3 |                 grep '^[0-9][0-9]:[0-9][0-9]' | tr ',' '.'
  4 | }
  5 | getStart () {
  6 | 	getStamps | awk '{print $1}'; 
  7 | }
  8 | getEnd () {
  9 | 	getStamps | awk '{print $3}'; 
 10 | }
 11 | addSeconds() {
 12 | 	echo "$(date --date "@$(($(date --date "2015-01-01 ${1}" +%s)+${2}))" +%H:%M:%S)";
 13 | } && (addSeconds "01:03:22.777" "1" |grep -q '01:03:23') || exit 1;
 14 | matchStart () {
 15 | 	end=$lenght;
 16 |         while ! [ ${end} = ${position} ]; do
 17 |                 match="${line:position:end}";
 18 |                 if grep -qF "${match}" "tmp/$name-subtitles.txt"; then
 19 |                         addSeconds "$(getStart)" "-1";
 20 |                         return;
 21 |                 fi;
 22 |                 end=$((end-1));
 23 |         done;
 24 | 	echo "Failed to match start of line: $line"; exit 1;
 25 | }
 26 | matchEnd () {
 27 | 	matchStart > /dev/null;
 28 | 	if [ ${end} = ${lenght} ]; then 
 29 | 		addSeconds "$(getEnd)" "3";
 30 | 		return;
 31 | 	fi
 32 | 	position=$((end+1));
 33 | 	end=${lenght};
 34 |         while ! [ ${end} = ${position} ]; do
 35 |                 match="${line:position:end}";
 36 |                 if grep -qF "${match}" "tmp/$name-subtitles.txt"; then
 37 | 			addSeconds "$(getEnd)" "3";
 38 |                         return;
 39 |                 fi;
 40 |                 position=$((position+1));
 41 |         done;
 42 | 	echo "Failed to match end of line: $line"; exit 1;
 43 | }
 44 | match () {
 45 | 	cat tmp/$name-dialogs.utf.summary.txt |
 46 | 	while read line; do
 47 | 		lenght="${#line}";
 48 | 		position=0;
 49 | 		start="$(matchStart)";
 50 | 		finish="$(matchEnd)";
 51 | 		echo "${start} ${finish} ${line}" >> tmp/$name-extracts.txt;
 52 | 	done;
 53 | }
 54 | extract () {
 55 | 	lineId=0;
 56 | 	cat tmp/$name-extracts.txt |sort >tmp/$name-extracts-sorted.txt;
 57 | 	cat tmp/$name-extracts-sorted.txt |
 58 | 	while read line; do
 59 | 		start="$(echo "$line"|awk '{print $1}')";
 60 | 		finish="$(echo "$line"|awk '{print $2}')";
 61 | 		echo "$start $finish";
 62 | 		echo "file '${PWD}/tmp/$name-summary${lineId}.mp4'" >> tmp/$name-concate.txt;
 63 | 		cat /dev/null | 
 64 | 			ffmpeg -nostats -loglevel panic -i "$movie" -ss "${start}" -to "${finish}" -map_metadata -1 $OPTIONS tmp/$name-summary${lineId}.mp4;
 65 | 		lineId=$((lineId+1));
 66 | 	done;
 67 | 	cat /dev/null |ffmpeg -nostats -loglevel panic -f concat -i tmp/$name-concate.txt -c copy out/$name-summary.mp4
 68 | }
 69 | sanitizeSubtitles () {
 70 | 	cat "$subtitles" |
 71 |                 iconv -c -t UTF-8 |
 72 | 		# Tripple dot overuse causes trouble.
 73 |                 #sed 's/\.\+/\./g' | 
 74 |                 sed 's/([^)]*)//g' |
 75 |                 sed 's/<[^>]*>//g' |
 76 |                 tr '-' '_' | tr -d '\r' | grep -v '^[0-9]*$' |
 77 |                 while read line; do
 78 |                         if ! echo "$line" | grep -q '^[0-9]'; then
 79 |                                 echo -n "$line ";
 80 |                         else
 81 |                                 echo; 
 82 |                                 echo "$line";
 83 |                         fi;
 84 |                 done |
 85 | 		grep -v '^[ ]*$' > tmp/$name-subtitles.txt;
 86 | }
 87 | summarize () {
 88 | 	cat tmp/$name-subtitles.txt |grep -v '^[0-9]' > tmp/$name-dialogs.utf.txt;
 89 | 	~/.local/bin/sumy lex-rank --length=15 --file tmp/$name-dialogs.utf.txt > tmp/$name-dialogs.utf.summary.txt;
 90 | }
 91 | main () {
 92 | 	movie="$1";
 93 | 	name="$(basename "$1")";
 94 | 	subtitles="$2";
 95 | 	if [ "$#" = 3 ] && [ "$3" = "-fast" ]; then
 96 | 		OPTIONS="-c copy";	
 97 | 	else
 98 | 		OPTIONS="-strict -2";
 99 | 	fi
100 | 	rm tmp/$name-* || true;
101 | 	rm out/$name-summary.mp4 || true;
102 | 	sanitizeSubtitles;
103 | 	summarize;
104 | 	match;
105 | 	extract;
106 | }
107 | set -ue
108 | main "$@";
109 | 


--------------------------------------------------------------------------------