def _get_streams(self):
player_js = self.session.http.get(
self.url,
schema=validate.Schema(
validate.transform(self._re_player.search),
validate.any(
None,
validate.Schema(
validate.get(1),
validate.transform(
lambda url: update_scheme("https:", url))))))
if not player_js:
return
log.debug(f"Found player js {player_js}")
data = self.session.http.get(
player_js,
schema=validate.Schema(
validate.transform(self._re_json.match), validate.get(1),
validate.parse_json(), validate.get("mediaResource"),
validate.get("dflt"), {
validate.optional("audioURL"): validate.url(),
validate.optional("videoURL"): validate.url()
}))
if data.get("videoURL"):
yield from HLSStream.parse_variant_playlist(
self.session, update_scheme("https:",
data.get("videoURL"))).items()
if data.get("audioURL"):
yield "audio", HTTPStream(
self.session, update_scheme("https:", data.get("audioURL")))
def test_update_scheme():
assert update_scheme("https://other.com/bar", "//example.com/foo"
) == "https://example.com/foo", "should become https"
assert update_scheme(
"http://other.com/bar",
"//example.com/foo") == "http://example.com/foo", "should become http"
assert update_scheme("https://other.com/bar", "http://example.com/foo"
) == "http://example.com/foo", "should remain http"
assert update_scheme(
"https://other.com/bar",
"example.com/foo") == "https://example.com/foo", "should become https"
def test_update_scheme(self):
self.assertEqual(
"https://example.com/foo", # becomes https
update_scheme("https://other.com/bar", "//example.com/foo"))
self.assertEqual(
"http://example.com/foo", # becomes http
update_scheme("http://other.com/bar", "//example.com/foo"))
self.assertEqual(
"http://example.com/foo", # remains unchanged
update_scheme("https://other.com/bar", "http://example.com/foo"))
self.assertEqual(
"https://example.com/foo", # becomes https
update_scheme("https://other.com/bar", "example.com/foo"))
def _get_vod(self, root):
schema_vod = validate.Schema(
validate.xml_xpath_string(
".//script[@type='application/ld+json'][contains(text(),'VideoObject')][1]/text()"
), str,
validate.transform(
lambda jsonlike: re.sub(r"[\r\n]+", "", jsonlike)),
validate.parse_json(),
validate.any(
validate.all(
{"@graph": [dict]}, validate.get("@graph"),
validate.filter(lambda obj: obj["@type"] == "VideoObject"),
validate.get(0)), dict), {"contentUrl": validate.url()},
validate.get("contentUrl"),
validate.transform(
lambda content_url: update_scheme("https://", content_url)))
try:
vod = schema_vod.validate(root)
except PluginError:
return
if urlparse(vod).path.endswith(".m3u8"):
return HLSStream.parse_variant_playlist(self.session, vod)
return {"vod": HTTPStream(self.session, vod)}
def _get_live_streams(self):
video_id = self.session.http.get(
self.url,
schema=validate.Schema(
validate.parse_html(),
validate.xml_xpath_string(
".//div[@data-google-src]/@data-video-id")))
if video_id:
return self.session.streams(
f"https://www.youtube.com/watch?v={video_id}")
info_url = self.session.http.get(
self.API_URL.format(subdomain=self.match.group("subdomain")),
schema=validate.Schema(
validate.parse_json(), {"url": validate.url()},
validate.get("url"),
validate.transform(
lambda url: update_scheme("https://", url))))
hls_url = self.session.http.get(info_url,
schema=validate.Schema(
validate.parse_json(), {
"status": "ok",
"protocol": "hls",
"primary": validate.url()
}, validate.get("primary")))
return HLSStream.parse_variant_playlist(self.session, hls_url)
def _get_streams(self):
postdata = {
"clientID": self.match.group(1),
"showEncoder": True,
"showMediaAssets": True,
"showStreams": True,
"includePrivate": False,
"advancedDetails": True,
"VAST": True,
"eventID": self.match.group(2)
}
headers = {
"Content-Type": "application/json",
"wsc-api-key": self.WSC_API_KEY,
"Authorization": "embedder"
}
res = self.session.http.post(self.API_URL,
data=json.dumps(postdata),
headers=headers)
api_response = self.session.http.json(res,
schema=self.api_response_schema)
if api_response is None:
return
hls_url = api_response["data"]["streamingURIs"]["main"]
return HLSStream.parse_variant_playlist(
self.session, update_scheme("https://", hls_url))
def _get_streams(self):
data = self.match.groupdict()
url = update_scheme("https://", data.get("url"), force=False)
params = parse_params(data.get("params"))
log.debug(f"URL={url}; params={params}")
return DASHStream.parse_manifest(self.session, url, **params)
def _get_streams_api(self, video_id):
log.debug("Found video ID: {0}".format(video_id))
tld = self.match.group("tld")
try:
data = self.session.http.get(
self._api.get(tld, "lt"),
params=dict(video_id=video_id),
schema=validate.Schema(
validate.parse_json(),
{
"success": True,
"data": {
"versions": {
validate.text: validate.all(
[{
"type": validate.text,
"src": validate.text,
}],
validate.filter(lambda item: item["type"] == "application/x-mpegurl")
)
}
}
},
validate.get(("data", "versions"))
)
)
except PluginError:
log.error("Failed to get streams from API")
return
for stream in itertools.chain(*data.values()):
src = update_scheme("https://", stream["src"], force=False)
for s in HLSStream.parse_variant_playlist(self.session, src).items():
yield s
class TV999(Plugin):
iframe_re = re.compile(r'<iframe.*src="([^"]+)"')
hls_re = re.compile(r'src="([^"]+)"\s+type="application/x-mpegURL"')
iframe_schema = validate.Schema(
validate.transform(iframe_re.search),
validate.any(None, validate.all(
validate.get(1),
validate.url(),
)),
)
hls_schema = validate.Schema(
validate.transform(hls_re.search),
validate.any(
None,
validate.all(
validate.get(1),
validate.transform(lambda x: update_scheme('http:', x)),
validate.url(),
)),
)
def _get_streams(self):
iframe_url = self.session.http.get(self.url, schema=self.iframe_schema)
if not iframe_url:
log.error('Failed to find IFRAME URL')
return
hls_url = self.session.http.get(iframe_url, schema=self.hls_schema)
if hls_url:
return {'live': HLSStream(self.session, hls_url)}
def _get_streams(self):
"""
Find the streams for web.tv
:return:
"""
headers = {}
res = self.session.http.get(self.url, headers=headers)
headers["Referer"] = self.url
sources = self._sources_re.findall(res.text)
if len(sources):
sdata = parse_json(sources[0], schema=self._sources_schema)
for source in sdata:
log.debug(f"Found stream of type: {source['type']}")
if source["type"] == "application/vnd.apple.mpegurl":
url = update_scheme("https://", source["src"], force=False)
try:
# try to parse the stream as a variant playlist
variant = HLSStream.parse_variant_playlist(
self.session, url, headers=headers)
if variant:
yield from variant.items()
else:
# and if that fails, try it as a plain HLS stream
yield 'live', HLSStream(self.session,
url,
headers=headers)
except OSError:
log.warning(
"Could not open the stream, perhaps the channel is offline"
)
def _get_live_streams(self, match):
"""
Get the live stream in a particular language
:param match:
:return:
"""
live_url = self._live_api_url.format(match.get("subdomain"))
live_res = self.session.http.json(self.session.http.get(live_url),
schema=self._live_schema)
api_url = update_scheme("{0}:///".format(match.get("scheme")),
live_res["url"])
api_res = self.session.http.json(self.session.http.get(api_url),
schema=self._stream_api_schema)
try:
parse_hls = bool(strtobool(self.get_option('parse_hls')))
except AttributeError:
parse_hls = True
if parse_hls:
return HLSStream.parse_variant_playlist(self.session,
api_res["primary"])
else:
return dict(stream=HTTPStream(self.session, api_res["primary"]))
def _get_streams(self):
data = self.match.groupdict()
url = update_scheme("https://", data.get("url"), force=False)
params = parse_params(data.get("params"))
log.debug("URL={0}; params={1}".format(url, params))
return {"live": HTTPStream(self.session, url, **params)}
def _get_streams_delfi(self, src):
try:
data = self.session.http.get(src, schema=validate.Schema(
validate.parse_html(),
validate.xml_xpath_string(".//script[contains(text(),'embedJs.setAttribute(')][1]/text()"),
validate.any(None, validate.all(
validate.text,
validate.transform(re.compile(r"embedJs\.setAttribute\('src',\s*'(.+?)'").search),
validate.any(None, validate.all(
validate.get(1),
validate.transform(lambda url: parse_qsd(urlparse(url).fragment)),
{"stream": validate.text},
validate.get("stream"),
validate.parse_json(),
{"versions": [{
"hls": validate.text
}]},
validate.get("versions")
))
))
))
except PluginError:
log.error("Failed to get streams from iframe")
return
for stream in data:
src = update_scheme("https://", stream["hls"], force=False)
for s in HLSStream.parse_variant_playlist(self.session, src).items():
yield s
def _get_streams(self):
res = self.session.http.get(self.url)
# Look for Youtube embedded video first
for iframe in itertags(res.text, "iframe"):
if urlparse(iframe.attributes.get("src")).netloc.endswith(
"youtube.com"):
log.debug("Handing off to YouTube plugin")
return self.session.streams(iframe.attributes.get("src"))
# Next check for HLS URL with token
mobile_url_m = self.mobile_url_re.search(res.text)
mobile_url = mobile_url_m and update_scheme(
"https://", mobile_url_m.group("url"), force=False)
if mobile_url:
log.debug("Found mobile stream: {0}".format(mobile_url_m.group(0)))
token = mobile_url_m and mobile_url_m.group("token")
if not token and "kralmuzik" in self.url:
log.debug("Getting Kral Muzik HLS stream token from API")
token = self.session.http.get(self.kral_token_url).text
elif not token:
# if no token is in the url, try to find it else where in the page
log.debug("Searching for HLS stream token in URL")
token_m = self.token_re.search(res.text)
token = token_m and token_m.group("token")
return HLSStream.parse_variant_playlist(
self.session,
mobile_url + token,
headers={"Referer": self.url})
def _get_streams(self):
data = self.session.http.get(self.url, schema=self.config_schema)
for info in data["files"].values():
stream_url = update_scheme("https://", info["url"])
# pick the smaller of the two dimensions, for landscape v. portrait videos
res = min(info["width"], info["height"])
yield "{0}p".format(res), HTTPStream(self.session, stream_url)
def test_update_scheme(self):
self.assertEqual(
"https://example.com/foo", # becomes https
update_scheme("https://other.com/bar", "//example.com/foo")
)
self.assertEqual(
"http://example.com/foo", # becomes http
update_scheme("http://other.com/bar", "//example.com/foo")
)
self.assertEqual(
"http://example.com/foo", # remains unchanged
update_scheme("https://other.com/bar", "http://example.com/foo")
)
self.assertEqual(
"https://example.com/foo", # becomes https
update_scheme("https://other.com/bar", "example.com/foo")
)
def merge_path_list(self, static, user):
for _path_url in user:
if not _path_url.startswith(('http', '//')):
_path_url = update_scheme('http://', _path_url, force=False)
_parsed_path_url = urlparse(_path_url)
if _parsed_path_url.netloc and _parsed_path_url.path:
static += [(_parsed_path_url.netloc, _parsed_path_url.path)]
return static
def _get_streams(self):
self.session.http.headers.update({
'Referer': 'http://www.abweb.com/BIS-TV-Online/bistvo-tele-universal.aspx'
})
login_username = self.get_option('username')
login_password = self.get_option('password')
if self.options.get('purge_credentials'):
self.clear_cookies()
self._authed = False
log.info('All credentials were successfully removed.')
if self._authed:
log.info('Attempting to authenticate using cached cookies')
elif not self._authed and not (login_username and login_password):
log.error('A login for ABweb is required, use --abweb-username USERNAME --abweb-password PASSWORD')
return
elif not self._authed and not self._login(login_username, login_password):
return
log.debug('get iframe_url')
res = self.session.http.get(self.url)
for iframe in itertags(res.text, 'iframe'):
iframe_url = iframe.attributes.get('src')
if iframe_url.startswith('/'):
iframe_url = url_concat('https://www.abweb.com', iframe_url)
else:
iframe_url = update_scheme('https://', iframe_url)
log.debug(f'iframe_url={iframe_url}')
break
else:
raise PluginError('No iframe_url found.')
self.session.http.headers.update({'Referer': iframe_url})
res = self.session.http.get(iframe_url)
m = self._hls_re.search(res.text)
if not m:
raise PluginError('No hls_url found.')
hls_url = update_scheme('https://', m.group('url'))
streams = HLSStream.parse_variant_playlist(self.session, hls_url)
if streams:
yield from streams.items()
else:
yield 'live', HLSStream(self.session, hls_url)
def resolve_url(self, url: str, follow_redirect: bool = True) -> Plugin:
"""Attempts to find a plugin that can use this URL.
The default protocol (http) will be prefixed to the URL if
not specified.
Raises :exc:`NoPluginError` on failure.
:param url: a URL to match against loaded plugins
:param follow_redirect: follow redirects
"""
url = update_scheme("https://", url, force=False)
matcher: Matcher
candidate: Optional[Type[Plugin]] = None
priority = NO_PRIORITY
for name, plugin in self.plugins.items():
if plugin.matchers:
for matcher in plugin.matchers:
if matcher.priority > priority and matcher.pattern.match(
url) is not None:
candidate = plugin
priority = matcher.priority
# TODO: remove deprecated plugin resolver
elif hasattr(plugin, "can_handle_url") and callable(
plugin.can_handle_url) and plugin.can_handle_url(url):
prio = plugin.priority(url) if hasattr(
plugin, "priority") and callable(
plugin.priority) else NORMAL_PRIORITY
if prio > priority:
log.info(
f"Resolved plugin {name} with deprecated can_handle_url API"
)
candidate = plugin
priority = prio
if candidate:
return candidate(url)
if follow_redirect:
# Attempt to handle a redirect URL
try:
res = self.http.head(url,
allow_redirects=True,
acceptable_status=[501])
# Fall back to GET request if server doesn't handle HEAD.
if res.status_code == 501:
res = self.http.get(url, stream=True)
if res.url != url:
return self.resolve_url(res.url,
follow_redirect=follow_redirect)
except PluginError:
pass
raise NoPluginError
class Mico(Plugin):
author = None
category = None
title = None
url_re = re.compile(r'https?://(?:www\.)?micous\.com/live/\d+')
json_data_re = re.compile(r'win._profile\s*=\s*({.*})')
_json_data_schema = validate.Schema(
validate.transform(json_data_re.search),
validate.any(None, validate.all(
validate.get(1),
validate.transform(parse_json),
validate.any(None, validate.all({
'mico_id': int,
'nickname': validate.text,
'h5_url': validate.all(
validate.transform(lambda x: update_scheme('http:', x)),
validate.url(),
),
'is_live': bool,
})),
)),
)
@classmethod
def can_handle_url(cls, url):
return cls.url_re.match(url) is not None
def get_author(self):
if self.author is not None:
return self.author
def get_category(self):
if self.category is not None:
return self.category
def get_title(self):
if self.title is not None:
return self.title
def _get_streams(self):
json_data = self.session.http.get(self.url, schema=self._json_data_schema)
if not json_data:
log.error('Failed to get JSON data')
return
if not json_data['is_live']:
log.info('This stream is no longer online')
return
self.author = json_data['mico_id']
self.category = 'Live'
self.title = json_data['nickname']
return HLSStream.parse_variant_playlist(self.session, json_data['h5_url'])
def _get_http_streams(self, info):
name = QUALITY_MAP.get(info["_quality"], "vod")
urls = info["_stream"]
if not isinstance(info["_stream"], list):
urls = [urls]
for url in urls:
stream = HTTPStream(self.session, update_scheme("https://", url))
yield name, stream
def _get_streams(self):
data = self.match.groupdict()
url = update_scheme("https://", data.get("url"), force=False)
params = parse_params(data.get("params"))
log.debug("URL={0}; params={1}".format(url, params))
streams = HLSStream.parse_variant_playlist(self.session, url, **params)
return streams if streams else {
"live": HLSStream(self.session, url, **params)
}
def _get_streams(self):
iframe_url = self._get_xpath_string(self.url, ".//iframe[@src]/@src")
if not iframe_url:
return
hls_url = self._get_xpath_string(
iframe_url, ".//source[contains(@src,'m3u8')]/@src")
if not hls_url:
return
return {
"live": HLSStream(self.session, update_scheme("http://", hls_url))
}
def _get_streams(self):
res = self.session.http.get(self.url)
m = self.embed_url_re.search(res.text)
platform_url = m and m.group("url")
if platform_url:
url = update_scheme("https://", platform_url)
# hand off to ThePlatform plugin
p = ThePlatform(url)
p.bind(self.session, "plugin.nbc")
return p.streams()
def _get_streams(self):
params = self.session.http.get(
self.url,
schema=validate.Schema(
validate.transform(self._re_player_manager.search),
validate.any(
None,
validate.all(
validate.get("json"), validate.parse_json(), {
"contentId": validate.any(str, int),
validate.optional("streamId"): str,
validate.optional("idec"): str,
validate.optional("token"): str
}))))
if not params:
log.error("Could not find player manager data")
return
params.update({
"video": (unquote(params.pop("token")) if params.get("token")
is not None else params.pop("streamId")),
"noflash":
"yes",
"embedded":
"0",
})
url_parsed = urlparse(self.url)
skip_vods = url_parsed.netloc.endswith(
"m4sport.hu") and url_parsed.path.startswith("/elo")
self.session.http.headers.update({"Referer": self.url})
playlists = self.session.http.get(
self.PLAYER_URL,
params=params,
schema=validate.Schema(
validate.transform(self._re_player_json.search),
validate.any(
None,
validate.all(
validate.get("json"), validate.parse_json(),
{"playlist": [{
"file": validate.url(),
"type": str
}]}, validate.get("playlist"),
validate.filter(lambda p: p["type"] == "hls"),
validate.filter(
lambda p: not skip_vods or "vod" not in p["file"]),
validate.map(
lambda p: update_scheme("https://", p["file"]))))))
for url in playlists or []:
yield from HLSStream.parse_variant_playlist(self.session,
url).items()
def _get_streams(self):
m = self._live_url_re.search(self.page.text)
playlist_url = m and update_scheme("https://", m.group(1), force=False)
player_url = self.url
live_channel = None
p = urlparse(player_url)
channelnumber = 0
if p.netloc.endswith("tvc.com.ec"):
live_channel = "Canal5"
elif p.netloc.endswith("rts.com.ec"):
live_channel = "Guayaquil"
elif p.netloc.endswith("atv.pe"):
if p.path.endswith(("ATVMas", "ATVMas/")):
live_channel = "ATVMas"
channelnumber = 1
else:
live_channel = "ATV"
token = self._get_token(channelnumber)
log.debug("token {0}".format(token))
if playlist_url:
log.debug("Found playlist URL in the page")
else:
if live_channel:
log.debug("Live channel: {0}".format(live_channel))
player_url = self._channel_urls[live_channel] + quote(token)
page = self.session.http.get(player_url,
raise_for_status=False)
if "block access from your country." in page.text:
raise PluginError("Content is geo-locked")
m = self._playlist_re.search(page.text)
playlist_url = m and update_scheme(
"https://", m.group(1), force=False)
else:
log.error("Could not find the live channel")
if playlist_url:
stream_url = "{0}?{1}".format(playlist_url,
urlencode({"iut": token}))
return HLSStream.parse_variant_playlist(
self.session, stream_url, headers={"referer": player_url})
def _get_streams(self):
if "cdn.bg" in urlparse(self.url).netloc:
iframe_url = self.url
h = self.session.get_option("http-headers")
if h and h.get("Referer"):
_referer = h.get("Referer")
else:
log.error(
"Missing Referer for iframe URL, use --http-header \"Referer=URL\" "
)
return
else:
_referer = self.url
res = self.session.http.get(self.url)
m = self._re_frame.search(res.text)
if m:
iframe_url = m.group(1)
else:
for iframe in itertags(res.text, "iframe"):
iframe_url = iframe.attributes.get("src")
if iframe_url and "cdn.bg" in iframe_url:
iframe_url = update_scheme("https://",
html_unescape(iframe_url),
force=False)
break
else:
return
log.debug(f"Found iframe: {iframe_url}")
res = self.session.http.get(iframe_url, headers={"Referer": _referer})
stream_url = self.stream_schema.validate(res.text)
if "geoblock" in stream_url:
log.error("Geo-restricted content")
return
return HLSStream.parse_variant_playlist(
self.session,
update_scheme(iframe_url, stream_url),
headers={"Referer": "https://i.cdn.bg/"},
)
def _get_live_streams(self, match):
"""
Get the live stream in a particular language
:param match:
:return:
"""
live_url = self._live_api_url.format(match.get("subdomain"))
live_res = self.session.http.json(self.session.http.get(live_url), schema=self._live_schema)
api_url = update_scheme("{0}:///".format(match.get("scheme")), live_res["url"])
api_res = self.session.http.json(self.session.http.get(api_url), schema=self._stream_api_schema)
return HLSStream.parse_variant_playlist(self.session, api_res["primary"])
def _get_streams(self):
data = self.session.http.get(
self.url,
schema=validate.Schema(
validate.transform(self._re_json_base.search),
validate.any(
None,
validate.all(
validate.get("json"), validate.parse_json(), {
"cam": {
validate.text: {
"live_type": validate.text,
"html5_streamingdomain": validate.text,
"html5_streampath": validate.text,
"group": validate.text,
"location": validate.text,
"title": validate.text,
"liveon": validate.text,
"defaulttab": validate.text,
}
}
}, validate.get("cam")))))
if not data:
return
cam_name = parse_qsd(self.url).get("cam") or next(
iter(data.keys()), None)
cam_data = data.get(cam_name)
if not cam_data:
return
# exclude everything other than live video streams
if cam_data["live_type"] != "flashvideo" or cam_data[
"liveon"] != "true" or cam_data["defaulttab"] != "live":
return
log.debug("Found cam {0}".format(cam_name))
hls_domain = cam_data["html5_streamingdomain"]
hls_playpath = cam_data["html5_streampath"]
self.author = cam_data["group"]
self.category = cam_data["location"]
self.title = cam_data["title"]
if hls_playpath:
hls_url = update_scheme("https://",
"{0}{1}".format(hls_domain, hls_playpath))
for s in HLSStream.parse_variant_playlist(self.session,
hls_url).items():
yield s
def __init__(self, url):
super(Generic, self).__init__(url)
self.url = update_scheme('http://', self.match.group('url'), force=False)
self.html_text = ''
# START - cache every used url and set a referer
if hasattr(GenericCache, 'cache_url_list'):
GenericCache.cache_url_list += [self.url]
# set the last url as a referer
self.referer = GenericCache.cache_url_list[-2]
else:
GenericCache.cache_url_list = [self.url]
self.referer = self.url
self.session.http.headers.update({'Referer': self.referer})
# END
# START - how often _get_streams already run
self._run = len(GenericCache.cache_url_list)
def _get_hls(self, root):
schema_live = validate.Schema(
validate.xml_xpath_string(
".//*[contains(@data-broadcast,'m3u8')]/@data-broadcast"), str,
validate.parse_json(),
validate.any(validate.all({"files": list}, validate.get("files")),
list),
[{
"url": validate.url(path=validate.endswith(".m3u8"))
}], validate.get((0, "url")),
validate.transform(
lambda content_url: update_scheme("https://", content_url)))
try:
live = schema_live.validate(root)
except PluginError:
return
return HLSStream.parse_variant_playlist(self.session, live)