Compress MKV and MP4 to MP4 with options

gistfile1.txt · 5.8 KiB · Text Raw

#!/usr/bin/env bash

##########################################################################
#
# This bash script was written by ChatGPT, but not this comment section.
# Everything works after testing, but not 2 functions:
# - Adding metadata to compressed MP4 file.
# - Checking if the file has already been compressed by this script.
#
##########################################################################

set -u
export LC_NUMERIC=C

SCRIPT_NAME="$(basename "$0")"
UNIQUE_PREFIX="r-"

########################################
# Encoding presets overview:
#
# none  : balanced quality, AAC 96k
# vlog  : spoken voice / crowd, copy audio
# tiny  : max compression, AAC 64k
#
# Resolutions: 720p | 1080p
########################################

show_help() {
  cat <<EOF
Usage:
  $SCRIPT_NAME <file>
  $SCRIPT_NAME [none|vlog|tiny] [720p|1080p] <file>
  $SCRIPT_NAME batch

Batch:
  Compresses all MP4 + MKV files using:
  mode=none, resolution=1080p

Defaults:
  mode=none
  resolution=720p

EOF
  exit 0
}

[[ "${1:-}" == "--help" ]] && show_help

MODE="none"
RES="720p"

# --- Batch mode ---
if [[ "${1:-}" == "batch" ]]; then
  MODE="none"
  RES="1080p"
  shift || true

shopt -s nullglob
  FILES=( *.mp4 *.MP4 *.mkv )
  shopt -u nullglob

if [[ ${#FILES[@]} -eq 0 ]]; then
    echo "No MP4 or MKV files found."
    exit 0
  fi

for f in "${FILES[@]}"; do
    "$0" "$MODE" "$RES" "$f"
  done
  exit 0
fi

# --- Parse arguments ---
if [[ $# -eq 1 ]]; then
  INPUT="$1"
elif [[ $# -eq 2 ]]; then
  MODE="$1"
  INPUT="$2"
elif [[ $# -eq 3 ]]; then
  MODE="$1"
  RES="$2"
  INPUT="$3"
else
  show_help
fi

[[ ! -f "$INPUT" ]] && { echo "File not found: $INPUT"; exit 1; }

BASENAME="${INPUT%.*}"
OUTPUT="${BASENAME}-r.mp4"
SUMMARY="${BASENAME}.json"

########################################
# Resolution
########################################
case "$RES" in
  720p) SCALE="scale=-2:720" ;;
  1080p) SCALE="scale=-2:1080" ;;
  *) echo "Invalid resolution"; exit 1 ;;
esac

########################################
# Mode settings
########################################
CQ=23
AUDIO_OPTS=("-c:a" "aac" "-b:a" "96k")

case "$MODE" in
  vlog)
    AUDIO_OPTS=("-c:a" "copy")
    CQ=23
    ;;
  tiny)
    AUDIO_OPTS=("-c:a" "aac" "-b:a" "64k")
    CQ=28
    ;;
  none) ;;
  *)
    echo "Invalid mode"
    exit 1
    ;;
esac

########################################
# Metadata detection (robust)
########################################
EXISTING_ID="$(ffprobe -v error \
  -show_entries format_tags=compressed_by \
  -of default=nk=1:nw=1 "$INPUT" 2>/dev/null || true)"

if [[ "$EXISTING_ID" =~ ^${UNIQUE_PREFIX}[A-Za-z0-9]+$ ]]; then
  echo "Skipping already-compressed file: $INPUT"
  exit 0
fi

########################################
# Generate unique ID
########################################
COMPRESS_ID="${UNIQUE_PREFIX}$(tr -dc A-Za-z0-9 </dev/urandom | head -c 12)"

########################################
# Encode
########################################
echo
echo "Compressing: $INPUT > $OUTPUT"

DURATION=$(ffprobe -v error -show_entries format=duration \
  -of default=nk=1:nw=1 "$INPUT")

START_TIME=$(date +%s)

ffmpeg -y -hide_banner -loglevel error \
  -i "$INPUT" \
  -map 0:v:0 -map 0:a? \
  -vf "$SCALE" \
  -c:v hevc_nvenc \
  -preset p6 -tune hq \
  -rc vbr -multipass fullres \
  -profile:v main -pix_fmt yuv420p \
  -spatial_aq 1 -temporal_aq 1 -aq-strength 8 \
  -cq "$CQ" \
  "${AUDIO_OPTS[@]}" \
  -movflags +faststart \
  -metadata compressed_by="$COMPRESS_ID" \
  -progress pipe:1 \
  "$OUTPUT" 2>/dev/null | while IFS='=' read -r key value; do

if [[ "$key" == "out_time_ms" ]]; then
      CURRENT_SEC=$(awk "BEGIN { print $value / 1000000 }")

PERCENT=$(awk -v c="$CURRENT_SEC" -v d="$DURATION" \
        'BEGIN { if (d>0) printf "%.1f", (c/d)*100; else print 0 }')

NOW=$(date +%s)
      ELAPSED_SEC=$(( NOW - START_TIME ))

ETA_SEC=$(awk -v e="$ELAPSED_SEC" -v p="$PERCENT" '
        BEGIN {
          if (p>0) printf "%.0f", e*(100-p)/p;
          else print 0
        }')

printf "\rProgress: %5.1f%% | Elapsed: %02d:%02d | ETA: %02d:%02d" \
        "$PERCENT" \
        $((ELAPSED_SEC/60)) $((ELAPSED_SEC%60)) \
        $((ETA_SEC/60)) $((ETA_SEC%60))
    fi

done

########################################
# Size check
########################################
ORIG_SIZE=$(stat -c%s "$INPUT")
NEW_SIZE=$(stat -c%s "$OUTPUT")

if (( NEW_SIZE >= ORIG_SIZE )); then
  echo
  echo "Result larger than original - skipping"
  rm -f "$OUTPUT"
  exit 0
fi

########################################
# JSON summary
########################################
HUMAN_ORIG=$(numfmt --to=iec --suffix=B "$ORIG_SIZE")
HUMAN_NEW=$(numfmt --to=iec --suffix=B "$NEW_SIZE")
OCCURRED=$(date +"%Y-%m-%d %H:%M:%S")

########################################
# Safe reduction percent calculation
########################################
if [[ -z "$ORIG_SIZE" ]] || (( ORIG_SIZE == 0 )); then
    REDUCTION_PERCENT=0
else
    REDUCTION_PERCENT=$(awk "BEGIN { pct=(1-$NEW_SIZE/$ORIG_SIZE)*100; if(pct<0) pct=0; printf \"%.2f\", pct }")
fi

cat > "$SUMMARY" <<EOF
{
  "input": "$(basename "$INPUT")",
  "output": "$(basename "$OUTPUT")",

"original_bytes": $ORIG_SIZE,
  "compressed_bytes": $NEW_SIZE,

"original": "$HUMAN_ORIG",
  "compressed": "$HUMAN_NEW",

"reduction_percent": $REDUCTION_PERCENT,
  "occurred": "$OCCURRED",

"compression_id": "$COMPRESS_ID",

"settings": {
    "codec": "hevc_nvenc",
    "resolution": "$RES",
    "mode": "$MODE",
    "cq": $CQ,
    "scale_filter": "$SCALE",
    "audio": "${AUDIO_OPTS[*]}",
    "preset": "p6",
    "rate_control": "vbr",
    "multipass": "fullres",
    "pixel_format": "yuv420p",
    "aq": {
      "spatial": 1,
      "temporal": 1,
      "strength": 8
    }
  }
}
EOF

REDUCTION_HR=$(awk "BEGIN { printf \"%.2f\", (1 - $NEW_SIZE / $ORIG_SIZE) * 100 }")

echo
echo "Done. Final reduction: $REDUCTION_HR% ($HUMAN_ORIG > $HUMAN_NEW)"
echo

1	#!/usr/bin/env bash
2
3	##########################################################################
4	#
5	# This bash script was written by ChatGPT, but not this comment section.
6	# Everything works after testing, but not 2 functions:
7	# - Adding metadata to compressed MP4 file.
8	# - Checking if the file has already been compressed by this script.
9	#
10	##########################################################################
11
12	set -u
13	export LC_NUMERIC=C
14
15	SCRIPT_NAME="$(basename "$0")"
16	UNIQUE_PREFIX="r-"
17
18	########################################
19	# Encoding presets overview:
20	#
21	# none : balanced quality, AAC 96k
22	# vlog : spoken voice / crowd, copy audio
23	# tiny : max compression, AAC 64k
24	#
25	# Resolutions: 720p \| 1080p
26	########################################
27
28	show_help() {
29	cat <<EOF
30	Usage:
31	$SCRIPT_NAME <file>
32	$SCRIPT_NAME [none\|vlog\|tiny] [720p\|1080p] <file>
33	$SCRIPT_NAME batch
34
35	Batch:
36	Compresses all MP4 + MKV files using:
37	mode=none, resolution=1080p
38
39	Defaults:
40	mode=none
41	resolution=720p
42
43	EOF
44	exit 0
45	}
46
47	[[ "${1:-}" == "--help" ]] && show_help
48
49	MODE="none"
50	RES="720p"
51
52	# --- Batch mode ---
53	if [[ "${1:-}" == "batch" ]]; then
54	MODE="none"
55	RES="1080p"
56	shift \|\| true
57
58	shopt -s nullglob
59	FILES=( .mp4 .MP4 *.mkv )
60	shopt -u nullglob
61
62	if [[ ${#FILES[@]} -eq 0 ]]; then
63	echo "No MP4 or MKV files found."
64	exit 0
65	fi
66
67	for f in "${FILES[@]}"; do
68	"$0" "$MODE" "$RES" "$f"
69	done
70	exit 0
71	fi
72
73	# --- Parse arguments ---
74	if [[ $# -eq 1 ]]; then
75	INPUT="$1"
76	elif [[ $# -eq 2 ]]; then
77	MODE="$1"
78	INPUT="$2"
79	elif [[ $# -eq 3 ]]; then
80	MODE="$1"
81	RES="$2"
82	INPUT="$3"
83	else
84	show_help
85	fi
86
87	[[ ! -f "$INPUT" ]] && { echo "File not found: $INPUT"; exit 1; }
88
89	BASENAME="${INPUT%.*}"
90	OUTPUT="${BASENAME}-r.mp4"
91	SUMMARY="${BASENAME}.json"
92
93	########################################
94	# Resolution
95	########################################
96	case "$RES" in
97	720p) SCALE="scale=-2:720" ;;
98	1080p) SCALE="scale=-2:1080" ;;
99	*) echo "Invalid resolution"; exit 1 ;;
100	esac
101
102	########################################
103	# Mode settings
104	########################################
105	CQ=23
106	AUDIO_OPTS=("-c:a" "aac" "-b:a" "96k")
107
108	case "$MODE" in
109	vlog)
110	AUDIO_OPTS=("-c:a" "copy")
111	CQ=23
112	;;
113	tiny)
114	AUDIO_OPTS=("-c:a" "aac" "-b:a" "64k")
115	CQ=28
116	;;
117	none) ;;
118	*)
119	echo "Invalid mode"
120	exit 1
121	;;
122	esac
123
124	########################################
125	# Metadata detection (robust)
126	########################################
127	EXISTING_ID="$(ffprobe -v error \
128	-show_entries format_tags=compressed_by \
129	-of default=nk=1:nw=1 "$INPUT" 2>/dev/null \|\| true)"
130
131	if [[ "$EXISTING_ID" =~ ^${UNIQUE_PREFIX}[A-Za-z0-9]+$ ]]; then
132	echo "Skipping already-compressed file: $INPUT"
133	exit 0
134	fi
135
136	########################################
137	# Generate unique ID
138	########################################
139	COMPRESS_ID="${UNIQUE_PREFIX}$(tr -dc A-Za-z0-9 </dev/urandom \| head -c 12)"
140
141	########################################
142	# Encode
143	########################################
144	echo
145	echo "Compressing: $INPUT > $OUTPUT"
146
147	DURATION=$(ffprobe -v error -show_entries format=duration \
148	-of default=nk=1:nw=1 "$INPUT")
149
150	START_TIME=$(date +%s)
151
152	ffmpeg -y -hide_banner -loglevel error \
153	-i "$INPUT" \
154	-map 0:v:0 -map 0:a? \
155	-vf "$SCALE" \
156	-c:v hevc_nvenc \
157	-preset p6 -tune hq \
158	-rc vbr -multipass fullres \
159	-profile:v main -pix_fmt yuv420p \
160	-spatial_aq 1 -temporal_aq 1 -aq-strength 8 \
161	-cq "$CQ" \
162	"${AUDIO_OPTS[@]}" \
163	-movflags +faststart \
164	-metadata compressed_by="$COMPRESS_ID" \
165	-progress pipe:1 \
166	"$OUTPUT" 2>/dev/null \| while IFS='=' read -r key value; do
167
168	if [[ "$key" == "out_time_ms" ]]; then
169	CURRENT_SEC=$(awk "BEGIN { print $value / 1000000 }")
170
171	PERCENT=$(awk -v c="$CURRENT_SEC" -v d="$DURATION" \
172	'BEGIN { if (d>0) printf "%.1f", (c/d)*100; else print 0 }')
173
174	NOW=$(date +%s)
175	ELAPSED_SEC=$(( NOW - START_TIME ))
176
177	ETA_SEC=$(awk -v e="$ELAPSED_SEC" -v p="$PERCENT" '
178	BEGIN {
179	if (p>0) printf "%.0f", e*(100-p)/p;
180	else print 0
181	}')
182
183	printf "\rProgress: %5.1f%% \| Elapsed: %02d:%02d \| ETA: %02d:%02d" \
184	"$PERCENT" \
185	$((ELAPSED_SEC/60)) $((ELAPSED_SEC%60)) \
186	$((ETA_SEC/60)) $((ETA_SEC%60))
187	fi
188
189	done
190
191	########################################
192	# Size check
193	########################################
194	ORIG_SIZE=$(stat -c%s "$INPUT")
195	NEW_SIZE=$(stat -c%s "$OUTPUT")
196
197	if (( NEW_SIZE >= ORIG_SIZE )); then
198	echo
199	echo "Result larger than original - skipping"
200	rm -f "$OUTPUT"
201	exit 0
202	fi
203
204	########################################
205	# JSON summary
206	########################################
207	HUMAN_ORIG=$(numfmt --to=iec --suffix=B "$ORIG_SIZE")
208	HUMAN_NEW=$(numfmt --to=iec --suffix=B "$NEW_SIZE")
209	OCCURRED=$(date +"%Y-%m-%d %H:%M:%S")
210
211	########################################
212	# Safe reduction percent calculation
213	########################################
214	if [[ -z "$ORIG_SIZE" ]] \|\| (( ORIG_SIZE == 0 )); then
215	REDUCTION_PERCENT=0
216	else
217	REDUCTION_PERCENT=$(awk "BEGIN { pct=(1-$NEW_SIZE/$ORIG_SIZE)*100; if(pct<0) pct=0; printf \"%.2f\", pct }")
218	fi
219
220	cat > "$SUMMARY" <<EOF
221	{
222	"input": "$(basename "$INPUT")",
223	"output": "$(basename "$OUTPUT")",
224
225	"original_bytes": $ORIG_SIZE,
226	"compressed_bytes": $NEW_SIZE,
227
228	"original": "$HUMAN_ORIG",
229	"compressed": "$HUMAN_NEW",
230
231	"reduction_percent": $REDUCTION_PERCENT,
232	"occurred": "$OCCURRED",
233
234	"compression_id": "$COMPRESS_ID",
235
236	"settings": {
237	"codec": "hevc_nvenc",
238	"resolution": "$RES",
239	"mode": "$MODE",
240	"cq": $CQ,
241	"scale_filter": "$SCALE",
242	"audio": "${AUDIO_OPTS[*]}",
243	"preset": "p6",
244	"rate_control": "vbr",
245	"multipass": "fullres",
246	"pixel_format": "yuv420p",
247	"aq": {
248	"spatial": 1,
249	"temporal": 1,
250	"strength": 8
251	}
252	}
253	}
254	EOF
255
256	REDUCTION_HR=$(awk "BEGIN { printf \"%.2f\", (1 - $NEW_SIZE / $ORIG_SIZE) * 100 }")
257
258	echo
259	echo "Done. Final reduction: $REDUCTION_HR% ($HUMAN_ORIG > $HUMAN_NEW)"
260	echo