[yt-dlp.git] / youtube_dl / downloader / hls.py

from __future__ import unicode_literals

import re
import binascii
try:
    from Crypto.Cipher import AES
    can_decrypt_frag = True
except ImportError:
    can_decrypt_frag = False

from .fragment import FragmentFD
from .external import FFmpegFD

from ..compat import (
    compat_urllib_error,
    compat_urlparse,
    compat_struct_pack,
)
from ..utils import (
    parse_m3u8_attributes,
    update_url_query,
)


class HlsFD(FragmentFD):
    """ A limited implementation that does not require ffmpeg """

    FD_NAME = 'hlsnative'

    @staticmethod
    def can_download(manifest, info_dict):
        UNSUPPORTED_FEATURES = (
            r'#EXT-X-KEY:METHOD=(?!NONE|AES-128)',  # encrypted streams [1]
            # r'#EXT-X-BYTERANGE',  # playlists composed of byte ranges of media files [2]

            # Live streams heuristic does not always work (e.g. geo restricted to Germany
            # http://hls-geo.daserste.de/i/videoportal/Film/c_620000/622873/format,716451,716457,716450,716458,716459,.mp4.csmil/index_4_av.m3u8?null=0)
            # r'#EXT-X-MEDIA-SEQUENCE:(?!0$)',  # live streams [3]

            # This heuristic also is not correct since segments may not be appended as well.
            # Twitch vods of finished streams have EXT-X-PLAYLIST-TYPE:EVENT despite
            # no segments will definitely be appended to the end of the playlist.
            # r'#EXT-X-PLAYLIST-TYPE:EVENT',  # media segments may be appended to the end of
            #                                 # event media playlists [4]

            # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.4
            # 2. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
            # 3. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.2
            # 4. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.5
        )
        check_results = [not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES]
        is_aes128_enc = '#EXT-X-KEY:METHOD=AES-128' in manifest
        check_results.append(can_decrypt_frag or not is_aes128_enc)
        check_results.append(not (is_aes128_enc and r'#EXT-X-BYTERANGE' in manifest))
        check_results.append(not info_dict.get('is_live'))
        return all(check_results)

    def real_download(self, filename, info_dict):
        man_url = info_dict['url']
        self.to_screen('[%s] Downloading m3u8 manifest' % self.FD_NAME)

        urlh = self.ydl.urlopen(self._prepare_url(info_dict, man_url))
        man_url = urlh.geturl()
        s = urlh.read().decode('utf-8', 'ignore')

        if not self.can_download(s, info_dict):
            if info_dict.get('extra_param_to_segment_url'):
                self.report_error('pycrypto not found. Please install it.')
                return False
            self.report_warning(
                'hlsnative has detected features it does not support, '
                'extraction will be delegated to ffmpeg')
            fd = FFmpegFD(self.ydl, self.params)
            for ph in self._progress_hooks:
                fd.add_progress_hook(ph)
            return fd.real_download(filename, info_dict)

        def anvato_ad(s):
            return s.startswith('#ANVATO-SEGMENT-INFO') and 'type=ad' in s

        media_frags = 0
        ad_frags = 0
        ad_frag_next = False
        for line in s.splitlines():
            line = line.strip()
            if not line:
                continue
            if line.startswith('#'):
                if anvato_ad(line):
                    ad_frags += 1
                    ad_frag_next = True
                continue
            if ad_frag_next:
                ad_frag_next = False
                continue
            media_frags += 1

        ctx = {
            'filename': filename,
            'total_frags': media_frags,
            'ad_frags': ad_frags,
        }

        self._prepare_and_start_frag_download(ctx)

        fragment_retries = self.params.get('fragment_retries', 0)
        skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
        test = self.params.get('test', False)

        extra_query = None
        extra_param_to_segment_url = info_dict.get('extra_param_to_segment_url')
        if extra_param_to_segment_url:
            extra_query = compat_urlparse.parse_qs(extra_param_to_segment_url)
        i = 0
        media_sequence = 0
        decrypt_info = {'METHOD': 'NONE'}
        byte_range = {}
        frag_index = 0
        ad_frag_next = False
        for line in s.splitlines():
            line = line.strip()
            if line:
                if not line.startswith('#'):
                    if ad_frag_next:
                        ad_frag_next = False
                        continue
                    frag_index += 1
                    if frag_index <= ctx['fragment_index']:
                        continue
                    frag_url = (
                        line
                        if re.match(r'^https?://', line)
                        else compat_urlparse.urljoin(man_url, line))
                    if extra_query:
                        frag_url = update_url_query(frag_url, extra_query)
                    count = 0
                    headers = info_dict.get('http_headers', {})
                    if byte_range:
                        headers['Range'] = 'bytes=%d-%d' % (byte_range['start'], byte_range['end'])
                    while count <= fragment_retries:
                        try:
                            success, frag_content = self._download_fragment(
                                ctx, frag_url, info_dict, headers)
                            if not success:
                                return False
                            break
                        except compat_urllib_error.HTTPError as err:
                            # Unavailable (possibly temporary) fragments may be served.
                            # First we try to retry then either skip or abort.
                            # See https://github.com/rg3/youtube-dl/issues/10165,
                            # https://github.com/rg3/youtube-dl/issues/10448).
                            count += 1
                            if count <= fragment_retries:
                                self.report_retry_fragment(err, frag_index, count, fragment_retries)
                    if count > fragment_retries:
                        if skip_unavailable_fragments:
                            i += 1
                            media_sequence += 1
                            self.report_skip_fragment(frag_index)
                            continue
                        self.report_error(
                            'giving up after %s fragment retries' % fragment_retries)
                        return False
                    if decrypt_info['METHOD'] == 'AES-128':
                        iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', media_sequence)
                        decrypt_info['KEY'] = decrypt_info.get('KEY') or self.ydl.urlopen(decrypt_info['URI']).read()
                        frag_content = AES.new(
                            decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(frag_content)
                    self._append_fragment(ctx, frag_content)
                    # We only download the first fragment during the test
                    if test:
                        break
                    i += 1
                    media_sequence += 1
                elif line.startswith('#EXT-X-KEY'):
                    decrypt_url = decrypt_info.get('URI')
                    decrypt_info = parse_m3u8_attributes(line[11:])
                    if decrypt_info['METHOD'] == 'AES-128':
                        if 'IV' in decrypt_info:
                            decrypt_info['IV'] = binascii.unhexlify(decrypt_info['IV'][2:].zfill(32))
                        if not re.match(r'^https?://', decrypt_info['URI']):
                            decrypt_info['URI'] = compat_urlparse.urljoin(
                                man_url, decrypt_info['URI'])
                        if extra_query:
                            decrypt_info['URI'] = update_url_query(decrypt_info['URI'], extra_query)
                        if decrypt_url != decrypt_info['URI']:
                            decrypt_info['KEY'] = None
                elif line.startswith('#EXT-X-MEDIA-SEQUENCE'):
                    media_sequence = int(line[22:])
                elif line.startswith('#EXT-X-BYTERANGE'):
                    splitted_byte_range = line[17:].split('@')
                    sub_range_start = int(splitted_byte_range[1]) if len(splitted_byte_range) == 2 else byte_range['end']
                    byte_range = {
                        'start': sub_range_start,
                        'end': sub_range_start + int(splitted_byte_range[0]),
                    }
                elif anvato_ad(line):
                    ad_frag_next = True

        self._finish_frag_download(ctx)

        return True
Commit	Line	Data
f0b5d6af PH	1	from __future__ import unicode_literals
f0b5d6af PH	2
f0b5d6af	3	import re
e154c651	4	import binascii
	5	try:
	6	from Crypto.Cipher import AES
	7	can_decrypt_frag = True
	8	except ImportError:
	9	can_decrypt_frag = False
3bc2ddcc	10
f9a5affa	11	from .fragment import FragmentFD
0d66bd0e	12	from .external import FFmpegFD
f9a5affa	13
e154c651	14	from ..compat import (
25afc2a7	15	compat_urllib_error,
e154c651	16	compat_urlparse,
	17	compat_struct_pack,
	18	)
1cc79574	19	from ..utils import (
e154c651	20	parse_m3u8_attributes,
aaf44a2f	21	update_url_query,
3bc2ddcc JMF	22	)
	23
	24
12b84ac8	25	class HlsFD(FragmentFD):
12b84ac8	26	""" A limited implementation that does not require ffmpeg """
f0b5d6af	27
f9a5affa S	28	FD_NAME = 'hlsnative'
f9a5affa S	29
0d66bd0e	30	@staticmethod
6f126d90	31	def can_download(manifest, info_dict):
0d66bd0e	32	UNSUPPORTED_FEATURES = (
e154c651	33	r'#EXT-X-KEY:METHOD=(?!NONE\|AES-128)', # encrypted streams [1]
f5974637	34	# r'#EXT-X-BYTERANGE', # playlists composed of byte ranges of media files [2]
1e236d7e	35
c15c47d1 S	36	# Live streams heuristic does not always work (e.g. geo restricted to Germany
c15c47d1 S	37	# http://hls-geo.daserste.de/i/videoportal/Film/c_620000/622873/format,716451,716457,716450,716458,716459,.mp4.csmil/index_4_av.m3u8?null=0)
2937590e	38	# r'#EXT-X-MEDIA-SEQUENCE:(?!0$)', # live streams [3]
1e236d7e S	39
1e236d7e S	40	# This heuristic also is not correct since segments may not be appended as well.
633b444f S	41	# Twitch vods of finished streams have EXT-X-PLAYLIST-TYPE:EVENT despite
633b444f S	42	# no segments will definitely be appended to the end of the playlist.
1e236d7e	43	# r'#EXT-X-PLAYLIST-TYPE:EVENT', # media segments may be appended to the end of
51c4d85c	44	# # event media playlists [4]
1e236d7e	45
0d66bd0e S	46	# 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.4
	47	# 2. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.2.2
	48	# 3. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.2
6104cc29	49	# 4. https://tools.ietf.org/html/draft-pantos-http-live-streaming-17#section-4.3.3.5
0d66bd0e	50	)
e154c651	51	check_results = [not re.search(feature, manifest) for feature in UNSUPPORTED_FEATURES]
f5974637 RA	52	is_aes128_enc = '#EXT-X-KEY:METHOD=AES-128' in manifest
	53	check_results.append(can_decrypt_frag or not is_aes128_enc)
	54	check_results.append(not (is_aes128_enc and r'#EXT-X-BYTERANGE' in manifest))
2bfaf89b	55	check_results.append(not info_dict.get('is_live'))
e154c651	56	return all(check_results)
0d66bd0e	57
f0b5d6af	58	def real_download(self, filename, info_dict):
f9a5affa S	59	man_url = info_dict['url']
f9a5affa S	60	self.to_screen('[%s] Downloading m3u8 manifest' % self.FD_NAME)
69035555	61
c5a49ff0 S	62	urlh = self.ydl.urlopen(self._prepare_url(info_dict, man_url))
	63	man_url = urlh.geturl()
	64	s = urlh.read().decode('utf-8', 'ignore')
0d66bd0e	65
6f126d90	66	if not self.can_download(s, info_dict):
bfa1073e RA	67	if info_dict.get('extra_param_to_segment_url'):
	68	self.report_error('pycrypto not found. Please install it.')
	69	return False
2bfaf89b RA	70	self.report_warning(
	71	'hlsnative has detected features it does not support, '
	72	'extraction will be delegated to ffmpeg')
	73	fd = FFmpegFD(self.ydl, self.params)
	74	for ph in self._progress_hooks:
	75	fd.add_progress_hook(ph)
	76	return fd.real_download(filename, info_dict)
0d66bd0e	77
74c42d9e S	78	def anvato_ad(s):
	79	return s.startswith('#ANVATO-SEGMENT-INFO') and 'type=ad' in s
	80
	81	media_frags = 0
	82	ad_frags = 0
	83	ad_frag_next = False
f0b5d6af PH	84	for line in s.splitlines():
f0b5d6af PH	85	line = line.strip()
74c42d9e S	86	if not line:
	87	continue
	88	if line.startswith('#'):
	89	if anvato_ad(line):
	90	ad_frags += 1
a9ee4f6e	91	ad_frag_next = True
74c42d9e S	92	continue
	93	if ad_frag_next:
	94	ad_frag_next = False
	95	continue
	96	media_frags += 1
f0b5d6af	97
f9a5affa	98	ctx = {
f0b5d6af	99	'filename': filename,
74c42d9e S	100	'total_frags': media_frags,
74c42d9e S	101	'ad_frags': ad_frags,
f9a5affa S	102	}
	103
	104	self._prepare_and_start_frag_download(ctx)
	105
25afc2a7 S	106	fragment_retries = self.params.get('fragment_retries', 0)
	107	skip_unavailable_fragments = self.params.get('skip_unavailable_fragments', True)
	108	test = self.params.get('test', False)
	109
b8079a40	110	extra_query = None
aaf44a2f	111	extra_param_to_segment_url = info_dict.get('extra_param_to_segment_url')
b8079a40 RA	112	if extra_param_to_segment_url:
b8079a40 RA	113	extra_query = compat_urlparse.parse_qs(extra_param_to_segment_url)
e154c651	114	i = 0
	115	media_sequence = 0
	116	decrypt_info = {'METHOD': 'NONE'}
f5974637	117	byte_range = {}
75a24854	118	frag_index = 0
74c42d9e	119	ad_frag_next = False
e154c651	120	for line in s.splitlines():
	121	line = line.strip()
	122	if line:
	123	if not line.startswith('#'):
74c42d9e S	124	if ad_frag_next:
	125	ad_frag_next = False
	126	continue
75a24854	127	frag_index += 1
3e0304fe	128	if frag_index <= ctx['fragment_index']:
75a24854	129	continue
e154c651	130	frag_url = (
	131	line
	132	if re.match(r'^https?://', line)
	133	else compat_urlparse.urljoin(man_url, line))
b8079a40 RA	134	if extra_query:
b8079a40 RA	135	frag_url = update_url_query(frag_url, extra_query)
25afc2a7	136	count = 0
f5974637 RA	137	headers = info_dict.get('http_headers', {})
	138	if byte_range:
	139	headers['Range'] = 'bytes=%d-%d' % (byte_range['start'], byte_range['end'])
25afc2a7 S	140	while count <= fragment_retries:
25afc2a7 S	141	try:
75a24854 RA	142	success, frag_content = self._download_fragment(
75a24854 RA	143	ctx, frag_url, info_dict, headers)
25afc2a7 S	144	if not success:
25afc2a7 S	145	return False
25afc2a7	146	break
2e99cd30	147	except compat_urllib_error.HTTPError as err:
25afc2a7 S	148	# Unavailable (possibly temporary) fragments may be served.
	149	# First we try to retry then either skip or abort.
	150	# See https://github.com/rg3/youtube-dl/issues/10165,
	151	# https://github.com/rg3/youtube-dl/issues/10448).
	152	count += 1
	153	if count <= fragment_retries:
75a24854	154	self.report_retry_fragment(err, frag_index, count, fragment_retries)
25afc2a7 S	155	if count > fragment_retries:
	156	if skip_unavailable_fragments:
	157	i += 1
	158	media_sequence += 1
75a24854	159	self.report_skip_fragment(frag_index)
25afc2a7 S	160	continue
	161	self.report_error(
	162	'giving up after %s fragment retries' % fragment_retries)
e154c651	163	return False
e154c651	164	if decrypt_info['METHOD'] == 'AES-128':
8369a4fe	165	iv = decrypt_info.get('IV') or compat_struct_pack('>8xq', media_sequence)
75a24854	166	decrypt_info['KEY'] = decrypt_info.get('KEY') or self.ydl.urlopen(decrypt_info['URI']).read()
8369a4fe S	167	frag_content = AES.new(
8369a4fe S	168	decrypt_info['KEY'], AES.MODE_CBC, iv).decrypt(frag_content)
75a24854	169	self._append_fragment(ctx, frag_content)
e154c651	170	# We only download the first fragment during the test
25afc2a7	171	if test:
e154c651	172	break
	173	i += 1
	174	media_sequence += 1
	175	elif line.startswith('#EXT-X-KEY'):
75a24854	176	decrypt_url = decrypt_info.get('URI')
e154c651	177	decrypt_info = parse_m3u8_attributes(line[11:])
	178	if decrypt_info['METHOD'] == 'AES-128':
	179	if 'IV' in decrypt_info:
07ea9c9b	180	decrypt_info['IV'] = binascii.unhexlify(decrypt_info['IV'][2:].zfill(32))
e154c651	181	if not re.match(r'^https?://', decrypt_info['URI']):
8369a4fe S	182	decrypt_info['URI'] = compat_urlparse.urljoin(
8369a4fe S	183	man_url, decrypt_info['URI'])
b8079a40 RA	184	if extra_query:
b8079a40 RA	185	decrypt_info['URI'] = update_url_query(decrypt_info['URI'], extra_query)
75a24854 RA	186	if decrypt_url != decrypt_info['URI']:
75a24854 RA	187	decrypt_info['KEY'] = None
e154c651	188	elif line.startswith('#EXT-X-MEDIA-SEQUENCE'):
e154c651	189	media_sequence = int(line[22:])
f5974637 RA	190	elif line.startswith('#EXT-X-BYTERANGE'):
	191	splitted_byte_range = line[17:].split('@')
	192	sub_range_start = int(splitted_byte_range[1]) if len(splitted_byte_range) == 2 else byte_range['end']
	193	byte_range = {
	194	'start': sub_range_start,
	195	'end': sub_range_start + int(splitted_byte_range[0]),
	196	}
74c42d9e S	197	elif anvato_ad(line):
74c42d9e S	198	ad_frag_next = True
f9a5affa S	199
	200	self._finish_frag_download(ctx)
	201
f0b5d6af	202	return True