Merge branch 'hugohaa-alura'

pull/8/head
Tom-Oliver Heidel 4 years ago
commit 88f365113d

@ -0,0 +1,180 @@
# coding: utf-8
from __future__ import unicode_literals
import re
from .common import InfoExtractor
from ..compat import (
compat_urlparse,
)
from ..utils import (
urlencode_postdata,
urljoin,
int_or_none,
clean_html,
ExtractorError
)
class AluraIE(InfoExtractor):
_VALID_URL = r'https?://(?:cursos\.)?alura\.com\.br/course/(?P<course_name>[^/]+)/task/(?P<id>\d+)'
_LOGIN_URL = 'https://cursos.alura.com.br/loginForm?urlAfterLogin=/loginForm'
_VIDEO_URL = 'https://cursos.alura.com.br/course/%s/task/%s/video'
_NETRC_MACHINE = 'alura'
_TESTS = [{
'url': 'https://cursos.alura.com.br/course/clojure-mutabilidade-com-atoms-e-refs/task/60095',
'info_dict': {
'id': '60095',
'ext': 'mp4',
'title': 'Referências, ref-set e alter'
},
'skip': 'Requires alura account credentials'},
{
# URL without video
'url': 'https://cursos.alura.com.br/course/clojure-mutabilidade-com-atoms-e-refs/task/60098',
'only_matching': True},
{
'url': 'https://cursos.alura.com.br/course/fundamentos-market-digital/task/55219',
'only_matching': True}
]
def _real_extract(self, url):
video_id = self._match_id(url)
course = self._search_regex(self._VALID_URL, url, 'post url', group='course_name')
video_url = self._VIDEO_URL % (course, video_id)
video_dict = self._download_json(video_url, video_id, 'Searching for videos')
if video_dict:
webpage = self._download_webpage(url, video_id)
video_title = clean_html(self._search_regex(
r'<span[^>]+class=(["\'])task-body-header-title-text\1[^>]*>(?P<title>[^<]+)',
webpage, 'title', group='title'))
formats = []
for video_obj in video_dict:
video_url_m3u8 = video_obj.get('link')
video_format = self._extract_m3u8_formats(
video_url_m3u8, None, 'mp4', entry_protocol='m3u8_native',
m3u8_id='hls', fatal=False)
for f in video_format:
m = re.search(r'^[\w \W]*-(?P<res>\w*).mp4[\W \w]*', f['url'])
if m:
if not f.get('height'):
f['height'] = int('720' if m.group('res') == 'hd' else '480')
formats.extend(video_format)
self._sort_formats(formats, field_preference=('height', 'width', 'tbr', 'format_id'))
return {
'id': video_id,
'title': video_title,
"formats": formats
}
def _real_initialize(self):
self._login()
def _login(self):
username, password = self._get_login_info()
if username is None:
return
pass
login_page = self._download_webpage(
self._LOGIN_URL, None, 'Downloading login popup')
def is_logged(webpage):
return any(re.search(p, webpage) for p in (
r'href=[\"|\']?/signout[\"|\']',
r'>Logout<'))
# already logged in
if is_logged(login_page):
return
login_form = self._hidden_inputs(login_page)
login_form.update({
'username': username,
'password': password,
})
post_url = self._search_regex(
r'<form[^>]+class=["|\']signin-form["|\'] action=["|\'](?P<url>.+?)["|\']', login_page,
'post url', default=self._LOGIN_URL, group='url')
if not post_url.startswith('http'):
post_url = compat_urlparse.urljoin(self._LOGIN_URL, post_url)
response = self._download_webpage(
post_url, None, 'Logging in',
data=urlencode_postdata(login_form),
headers={'Content-Type': 'application/x-www-form-urlencoded'})
if not is_logged(response):
error = self._html_search_regex(
r'(?s)<p[^>]+class="alert-message[^"]*">(.+?)</p>',
response, 'error message', default=None)
if error:
raise ExtractorError('Unable to login: %s' % error, expected=True)
raise ExtractorError('Unable to log in')
class AluraCourseIE(AluraIE):
_VALID_URL = r'https?://(?:cursos\.)?alura\.com\.br/course/(?P<id>[^/]+)'
_LOGIN_URL = 'https://cursos.alura.com.br/loginForm?urlAfterLogin=/loginForm'
_NETRC_MACHINE = 'aluracourse'
_TESTS = [{
'url': 'https://cursos.alura.com.br/course/clojure-mutabilidade-com-atoms-e-refs',
'only_matching': True,
}]
@classmethod
def suitable(cls, url):
return False if AluraIE.suitable(url) else super(AluraCourseIE, cls).suitable(url)
def _real_extract(self, url):
course_path = self._match_id(url)
webpage = self._download_webpage(url, course_path)
course_title = self._search_regex(
r'<h1.*?>(.*?)<strong>(?P<course_title>.*?)</strong></h[0-9]>', webpage,
'course title', default=course_path, group='course_title')
entries = []
if webpage:
for path in re.findall(r'<a\b(?=[^>]* class="[^"]*(?<=[" ])courseSectionList-section[" ])(?=[^>]* href="([^"]*))', webpage):
page_url = urljoin(url, path)
section_path = self._download_webpage(page_url, course_path)
for path_video in re.findall(r'<a\b(?=[^>]* class="[^"]*(?<=[" ])task-menu-nav-item-link-VIDEO[" ])(?=[^>]* href="([^"]*))', section_path):
chapter = clean_html(
self._search_regex(
r'<h3[^>]+class=(["\'])task-menu-section-title-text\1[^>]*>(?P<chapter>[^<]+)',
section_path,
'chapter',
group='chapter'))
chapter_number = int_or_none(
self._search_regex(
r'<span[^>]+class=(["\'])task-menu-section-title-number[^>]*>(.*?)<strong>(?P<chapter_number>[^<]+)</strong>',
section_path,
'chapter number',
group='chapter_number'))
video_url = urljoin(url, path_video)
entry = {
'_type': 'url_transparent',
'id': self._match_id(video_url),
'url': video_url,
'id_key': self.ie_key(),
'chapter': chapter,
'chapter_number': chapter_number
}
entries.append(entry)
return self.playlist_result(entries, course_path, course_title)

@ -36,6 +36,10 @@ from .afreecatv import AfreecaTVIE
from .airmozilla import AirMozillaIE from .airmozilla import AirMozillaIE
from .aljazeera import AlJazeeraIE from .aljazeera import AlJazeeraIE
from .alphaporno import AlphaPornoIE from .alphaporno import AlphaPornoIE
from .alura import (
AluraIE,
AluraCourseIE
)
from .amcnetworks import AMCNetworksIE from .amcnetworks import AMCNetworksIE
from .americastestkitchen import AmericasTestKitchenIE from .americastestkitchen import AmericasTestKitchenIE
from .animeondemand import AnimeOnDemandIE from .animeondemand import AnimeOnDemandIE

Loading…
Cancel
Save