1
0
mirror of https://github.com/mxpv/podsync.git synced 2024-05-11 05:55:04 +00:00

Drop ytdl

This commit is contained in:
Maksym Pavlenko
2019-03-24 16:58:45 -07:00
parent 8cbd235976
commit 21ff51f4f9
7 changed files with 1 additions and 172 deletions

View File

@ -1,4 +1,4 @@
SUBDIRS := cmd/api cmd/nginx cmd/ytdl
SUBDIRS := cmd/api cmd/nginx
GOLANGCI := ./bin/golangci-lint
.PHONY: push

View File

@ -1,12 +0,0 @@
FROM python:3.6.0-alpine
WORKDIR /app
COPY . .
ENV REDIS_CONNECTION_STRING localhost
RUN apk add --update --no-cache gcc g++ make && \
pip install --no-cache-dir --requirement /app/requirements.txt
EXPOSE 5002
ENTRYPOINT ["python", "/app/ytdl.py"]

View File

@ -1,4 +0,0 @@
.PHONY: push
push:
docker build -t mxpv/ytdl .
docker push mxpv/ytdl

View File

@ -1,4 +0,0 @@
requests==2.19.1
youtube_dl==2019.03.09
sanic==18.12
redis==2.10.6

View File

@ -1,27 +0,0 @@
import ytdl
import unittest
class TestYtdl(unittest.TestCase):
def test_resolve(self):
self.assertIsNotNone(
ytdl._resolve('https://youtube.com/watch?v=ygIUF678y40',
{'format': 'video', 'quality': 'low', 'provider': 'youtube'}))
self.assertIsNotNone(
ytdl._resolve('https://youtube.com/watch?v=WyaEiO4hyik',
{'format': 'audio', 'quality': 'high', 'provider': 'youtube'}))
self.assertIsNotNone(
ytdl._resolve('https://youtube.com/watch?v=5mjUF2j9dgA',
{'format': 'video', 'quality': 'low', 'provider': 'youtube'})
)
self.assertIsNotNone(
ytdl._resolve('https://www.youtube.com/watch?v=2nH7xAMqD2g',
{'format': 'video', 'quality': 'high', 'provider': 'youtube'})
)
def test_vimeo(self):
self.assertIsNotNone(
ytdl._resolve('https://vimeo.com/237715420', {'format': 'video', 'quality': 'low', 'provider': 'vimeo'}))
self.assertIsNotNone(
ytdl._resolve('https://vimeo.com/275211960', {'format': 'video', 'quality': 'high', 'provider': 'vimeo'})
)

View File

@ -1,116 +0,0 @@
import os
import requests
import youtube_dl
from sanic import Sanic
from sanic.exceptions import InvalidUsage
from sanic.response import text, redirect
from youtube_dl.utils import DownloadError
METADATA_URL = os.getenv('METADATA_URL', 'http://app:5001/api/metadata/{feed_id}')
print('Using metadata URL template: ' + METADATA_URL)
app = Sanic()
opts = {
'quiet': True,
'no_warnings': True,
'forceurl': True,
'simulate': True,
'skip_download': True,
'call_home': False,
'nocheckcertificate': True
}
url_formats = {
'youtube': 'https://youtube.com/watch?v={}',
'vimeo': 'https://vimeo.com/{}',
}
@app.route('/download/<feed_id>/<video_id>', methods=['GET', 'HEAD'])
async def download(req, feed_id, video_id):
if not feed_id:
raise InvalidUsage('Invalid feed id')
# Remote extension and check if video id is ok
video_id = os.path.splitext(video_id)[0]
if not video_id:
raise InvalidUsage('Invalid video id')
# Pull metadata from API server
metadata_url = METADATA_URL.format(feed_id=feed_id, video_id=video_id)
r = requests.get(url=metadata_url)
json = r.json()
# Build URL
provider = json['provider']
tpl = url_formats[provider]
if not tpl:
raise InvalidUsage('Invalid feed')
url = tpl.format(video_id)
try:
redirect_url = _resolve(url, json)
return redirect(redirect_url)
except DownloadError as e:
msg = str(e)
return text(msg, status=511)
def _resolve(url, metadata):
if not url:
raise InvalidUsage('Invalid URL')
try:
provider = metadata['provider']
with youtube_dl.YoutubeDL(opts) as ytdl:
info = ytdl.extract_info(url, download=False)
if provider == 'youtube':
return _yt_choose_url(info, metadata)
elif provider == 'vimeo':
return _vimeo_choose_url(info, metadata)
else:
raise ValueError('undefined provider')
except DownloadError:
raise
except Exception as e:
print(e)
raise
def _yt_choose_url(info, metadata):
is_video = metadata['format'] == 'video'
# Filter formats by file extension
ext = 'mp4' if is_video else 'm4a'
fmt_list = [x for x in info['formats'] if x['ext'] == ext and 'acodec' in x and x['acodec'] != 'none']
if not len(fmt_list):
return info['url']
# Sort list by field (width for videos, file size for audio)
sort_field = 'width' if is_video else 'filesize'
# Sometime 'filesize' field can be None
if not all(x[sort_field] is not None for x in fmt_list):
sort_field = 'format_id'
ordered = sorted(fmt_list, key=lambda x: x[sort_field], reverse=True)
# Choose an item depending on quality, better at the beginning
is_high_quality = metadata['quality'] == 'high'
item = ordered[0] if is_high_quality else ordered[-1]
return item['url']
def _vimeo_choose_url(info, metadata):
# Query formats with 'extension' = mp4 and 'format_id' = http-1080p/http-720p/../http-360p
fmt_list = [x for x in info['formats'] if x['ext'] == 'mp4' and x['format_id'].startswith('http-')]
ordered = sorted(fmt_list, key=lambda x: x['width'], reverse=True)
is_high_quality = metadata['quality'] == 'high'
item = ordered[0] if is_high_quality else ordered[-1]
return item['url']
if __name__ == '__main__':
app.run(host='0.0.0.0', port=5002, workers=32)

View File

@ -29,14 +29,6 @@ services:
awslogs-group: "Prod"
awslogs-stream: "Backend"
awslogs-multiline-pattern: "{"
ytdl:
image: mxpv/ytdl
container_name: ytdl
restart: always
ports:
- 5002
environment:
- METADATA_URL=http://api:5001/api/metadata/{feed_id}
redis:
image: redis:5.0.3
container_name: redis