-
Notifications
You must be signed in to change notification settings - Fork 36
/
split_and_stitch.yaml
196 lines (175 loc) · 6.12 KB
/
split_and_stitch.yaml
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
name: split and stitch demo
description: |
this job splits an input video into 10s chunks, transcodes the various chunks in parallel,
and finally stitches the final result into a single file.
This demo assumes a locally running instace of Minio (AWS S3-like service) which will be used to store the
video chunks and the final video output
You can get a running instance of minio using the following command:
docker run --name=minio -d -p 9000:9000 -p 9001:9001 minio/minio server /data --console-address ":9001"
You'll like have to change the endpointURL below to the IP address of your minio server
The default credentials for a Minio Server are minioadmin/minioadmin
inputs:
accessKeyID: minioadmin
endpointURL: http://my-minio-server:9000
secretKeyID: minioadmin
source: s3://master/master.mov
tasks:
- name: presign the s3 source
var: signedURL
run: aws --endpoint-url $ENDPOINT_URL s3 presign $SOURCE --expires-in 86400 > $TORK_OUTPUT
image: amazon/aws-cli:2.13.10
env:
AWS_ACCESS_KEY_ID: "{{inputs.accessKeyID}}"
AWS_SECRET_ACCESS_KEY: "{{inputs.secretKeyID}}"
ENDPOINT_URL: "{{inputs.endpointURL}}"
SOURCE: "{{ inputs.source }}"
- name: get the video metadata
var: ffprobe
run: |
ffprobe \
-v quiet \
-print_format json \
-show_error \
-show_format \
-show_streams \
$SOURCE > $TORK_OUTPUT
image: jrottenberg/ffmpeg:3.4-alpine
env:
SOURCE: "{{ tasks.signedURL }}"
- name: extract the duration of the video
var: duration
run: |
DURATION=$(echo -n $FFPROBE | jq -r '.format.duration')
echo -n $DURATION >> $TORK_OUTPUT
image: badouralix/curl-jq
env:
FFPROBE: "{{ tasks.ffprobe }}"
- name: extract the framerate of the video
var: framerate
run: |
FRAMERATE=$(echo -n $FFPROBE | jq -r '.streams[] | select (.codec_type=="video") | .r_frame_rate')
echo -n $FRAMERATE >> $TORK_OUTPUT
image: badouralix/curl-jq
env:
FFPROBE: "{{ tasks.ffprobe }}"
- name: clean and parse framerate
var: framerate
run: |
python script.py $FRAMERATE > $TORK_OUTPUT
image: python:3-slim
env:
FRAMERATE: "{{ tasks.framerate }}"
files:
script.py: |
import re
import sys
cfrate = re.sub(r"[^0-9/\\.]", "", sys.argv[1])
pfrate = cfrate.split("/")
frate = float(pfrate[0])/float(pfrate[1])
print(frate,end="")
- name: calculate chunks times
var: chunks
run: |
python script.py $DURATION $FRAMERATE > $TORK_OUTPUT
image: python:3-slim
env:
DURATION: "{{ tasks.duration }}"
FRAMERATE: "{{ tasks.framerate }}"
files:
script.py: |
import math
import json
import sys
duration = float(sys.argv[1])
frate = float(sys.argv[2])
frate_ceil = math.ceil(frate)
time_unit = frate_ceil/frate
chunk_size = 30*time_unit
chunks = []
start = 0
length = 0
while(start<duration):
if(duration-start<chunk_size):
length=duration-start
else:
length=chunk_size
if duration-(start+length) < 5:
length=duration-start
chunks.append({"start":start,"length":length})
start = start+length;
print(json.dumps(chunks))
- name: generate a random bucket name to store the results
var: bucketName
run: echo -n "video-$(shuf -i 1-10000 -n1)" > $TORK_OUTPUT
image: ubuntu:mantic
- name: create a temporary bucket to house the chunks
run: |
aws \
--endpoint-url $ENDPOINT_URL s3api create-bucket --bucket $BUCKET_NAME
image: amazon/aws-cli:2.13.10
env:
AWS_ACCESS_KEY_ID: "{{inputs.accessKeyID}}"
AWS_SECRET_ACCESS_KEY: "{{inputs.secretKeyID}}"
BUCKET_NAME: "{{tasks.bucketName}}"
ENDPOINT_URL: "{{inputs.endpointURL}}"
- name: transcode the chunks in parallel
each:
list: "{{ fromJSON(tasks.chunks) }}"
task:
name: encode the chunk
var: chunk{{ item.index }}
run: |
ffmpeg -ss ${START} -i $SOURCE -to $LENGTH /tmp/chunk.mp4
image: jrottenberg/ffmpeg:3.4-alpine
env:
LENGTH: "{{ item.value.length }}"
SOURCE: "{{ tasks.signedURL }}"
START: "{{ item.value.start }}"
post:
- name: upload the chunk to minio
run: aws --endpoint-url $ENDPOINT_URL s3 cp /tmp/chunk.mp4 s3://$BUCKET_NAME/chunks/chunk_$NUMBER.mp4
image: amazon/aws-cli:2.13.10
env:
AWS_ACCESS_KEY_ID: "{{inputs.accessKeyID}}"
AWS_SECRET_ACCESS_KEY: "{{inputs.secretKeyID}}"
BUCKET_NAME: "{{tasks.bucketName}}"
ENDPOINT_URL: "{{inputs.endpointURL}}"
NUMBER: "{{ item.index }}"
mounts:
- type: volume
target: /tmp
retry:
limit: 2
- name: stitch the chunks into a single video
run: |
for filename in /tmp/chunks/*.mp4; do
echo "file $filename" >> /tmp/chunks.txt
done
ffmpeg -f concat -safe 0 -i /tmp/chunks.txt -c copy /tmp/output.mp4
image: jrottenberg/ffmpeg:3.4-alpine
env:
BUCKET_NAME: "{{tasks.bucketName}}"
pre:
- name: download the chunks
run: aws --endpoint-url $ENDPOINT_URL s3 sync s3://$BUCKET_NAME/chunks /tmp/chunks
image: amazon/aws-cli:2.13.10
env:
AWS_ACCESS_KEY_ID: "{{inputs.accessKeyID}}"
AWS_SECRET_ACCESS_KEY: "{{inputs.secretKeyID}}"
BUCKET_NAME: "{{tasks.bucketName}}"
ENDPOINT_URL: "{{inputs.endpointURL}}"
post:
- name: upload the final video to minio
run: aws --endpoint-url $ENDPOINT_URL s3 cp /tmp/output.mp4 s3://$BUCKET_NAME/output.mp4
image: amazon/aws-cli:2.13.10
env:
AWS_ACCESS_KEY_ID: "{{inputs.accessKeyID}}"
AWS_SECRET_ACCESS_KEY: "{{inputs.secretKeyID}}"
BUCKET_NAME: "{{tasks.bucketName}}"
ENDPOINT_URL: "{{inputs.endpointURL}}"
mounts:
- type: volume
target: /tmp
retry:
limit: 2
timeout: 120s