mirror/youtube-dl
1
0
Fork 0
mirror of https://codeberg.org/polarisfm/youtube-dl.git synced 2024-12-29 11:05:13 +00:00
Code Releases Activity

[tele13] improve extraction

Browse source 
- improve jwplayer setup regex
- sort formats
- remove duplicate formats
- update youtube test
This commit is contained in:
remitamine 2015-12-20 15:48:19 +01:00
parent 436416afe2
commit 6882c0870e
1 changed files with 32 additions and 32 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

64
youtube_dl/extractor/tele13.py
View file

@ -1,10 +1,13 @@
# coding: utf-8 # coding: utf-8
from __future__ import unicode_literals from __future__ import unicode_literals
import re
from .common import InfoExtractor from .common import InfoExtractor
from ..utils import js_to_json from .youtube import YoutubeIE
from ..utils import (
js_to_json,
qualities,
determine_ext,
)
class Tele13IE(InfoExtractor): class Tele13IE(InfoExtractor):
@ -25,12 +28,12 @@ class Tele13IE(InfoExtractor):
}, },
{ {
'url': 'http://www.t13.cl/videos/mundo/tendencias/video-captan-misteriosa-bola-fuego-cielos-bangkok', 'url': 'http://www.t13.cl/videos/mundo/tendencias/video-captan-misteriosa-bola-fuego-cielos-bangkok',
'md5': '65d1ae54812c96f4b345dd21d3bb1adc', 'md5': '867adf6a3b3fef932c68a71d70b70946',
'info_dict': { 'info_dict': {
'id': 'rOoKv2OMpOw', 'id': 'rOoKv2OMpOw',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Shooting star seen on 7-Sep-2015', 'title': 'Shooting star seen on 7-Sep-2015',
'description': 'md5:a1cd2e74f6ee6851552c9cf5851d6b06', 'description': 'md5:7292ff2a34b2f673da77da222ae77e1e',
'uploader': 'Porjai Jaturongkhakun', 'uploader': 'Porjai Jaturongkhakun',
'upload_date': '20150906', 'upload_date': '20150906',
'uploader_id': 'UCnLY_3ezwNcDSC_Wc6suZxw', 'uploader_id': 'UCnLY_3ezwNcDSC_Wc6suZxw',
@ -41,41 +44,38 @@ class Tele13IE(InfoExtractor):
def _real_extract(self, url): def _real_extract(self, url):
display_id = self._match_id(url) display_id = self._match_id(url)
webpage = self._download_webpage(url, display_id) webpage = self._download_webpage(url, display_id)
setup_js = self._parse_json( setup_js = self._search_regex(r"(?s)jwplayer\('player-vivo'\).setup\((\{.*?\})\)", webpage, 'setup code')
js_to_json( sources = self._parse_json(self._search_regex(r'sources\s*:\s*(\[[^\]]+\])', setup_js, 'sources'), display_id, js_to_json)
self._search_regex(
r"jwplayer\('player-vivo'\).setup\((\{.*?\})\)",
webpage,
'setup code',
flags=re.DOTALL
).replace('\n//', '')
),
display_id
)
title = setup_js['title']
thumbnail = setup_js.get('image') or setup_js['playlist'][0].get('image')
description = self._html_search_meta(
'description', webpage, 'description')
preference = qualities(['Móvil', 'SD', 'HD'])
formats = [] formats = []
for f in setup_js['playlist'][0]['sources']: urls = []
for f in sources:
format_url = f['file'] format_url = f['file']
if format_url != '': if format_url and format_url not in urls:
if '.m3u8' in format_url: ext = determine_ext(format_url)
formats.extend(self._extract_m3u8_formats(format_url, display_id)) if ext == 'm3u8':
m3u8_formats = self._extract_m3u8_formats(format_url, display_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False)
if m3u8_formats:
formats.extend(m3u8_formats)
elif YoutubeIE.suitable(format_url):
return self.url_result(format_url, 'Youtube')
else: else:
if 'youtube.com' in format_url: formats.append({
return self.url_result(format_url, 'Youtube') 'url': format_url,
else: 'format_id': f.get('label'),
formats.append({'url': format_url, 'format_id': f.get('label')}) 'preference': preference(f.get('label')),
'ext': ext,
})
urls.append(format_url)
self._sort_formats(formats)
return { return {
'id': display_id, 'id': display_id,
'title': title, 'title': self._search_regex(r'title\s*:\s*"([^"]+)"', setup_js, 'title'),
'description': description, 'description': self._html_search_meta('description', webpage, 'description'),
'thumbnail': thumbnail, 'thumbnail': self._search_regex(r'image\s*:\s*"([^"]+)"', setup_js, 'thumbnail', default=None),
'formats': formats, 'formats': formats,
} }