[downloader/hls] Allow extractors to provide AES key (#6158)

and related cleanup

Authored by: bashonly, Grub4K

Co-authored-by: Simon Sawicki <contact@grub4k.xyz>

yt_dlp/YoutubeDL.py

@@ -554,7 +554,7 @@ class YoutubeDL:
         'vbr', 'fps', 'vcodec', 'container', 'filesize', 'filesize_approx', 'rows', 'columns',
         'player_url', 'protocol', 'fragment_base_url', 'fragments', 'is_from_start',
         'preference', 'language', 'language_preference', 'quality', 'source_preference',
-        'http_headers', 'stretched_ratio', 'no_resume', 'has_drm', 'downloader_options',
+        'http_headers', 'stretched_ratio', 'no_resume', 'has_drm', 'extra_param_to_segment_url', 'hls_aes', 'downloader_options',
         'page_url', 'app', 'play_path', 'tc_url', 'flash_version', 'rtmp_live', 'rtmp_conn', 'rtmp_protocol', 'rtmp_real_time'
     }
     _format_selection_exts = {

yt_dlp/downloader/external.py

@@ -104,6 +104,7 @@ class ExternalFD(FragmentFD):
         return all((
             not info_dict.get('to_stdout') or Features.TO_STDOUT in cls.SUPPORTED_FEATURES,
             '+' not in info_dict['protocol'] or Features.MULTIPLE_FORMATS in cls.SUPPORTED_FEATURES,
+            not traverse_obj(info_dict, ('hls_aes', ...), 'extra_param_to_segment_url'),
             all(proto in cls.SUPPORTED_PROTOCOLS for proto in info_dict['protocol'].split('+')),
         ))
 

yt_dlp/downloader/fragment.py

@@ -360,7 +360,8 @@ class FragmentFD(FileDownloader):
             if not decrypt_info or decrypt_info['METHOD'] != 'AES-128':
                 return frag_content
             iv = decrypt_info.get('IV') or struct.pack('>8xq', fragment['media_sequence'])
-            decrypt_info['KEY'] = decrypt_info.get('KEY') or _get_key(info_dict.get('_decryption_key_url') or decrypt_info['URI'])
+            decrypt_info['KEY'] = (decrypt_info.get('KEY')
+                                   or _get_key(traverse_obj(info_dict, ('hls_aes', 'uri')) or decrypt_info['URI']))
             # Don't decrypt the content in tests since the data is explicitly truncated and it's not to a valid block
             # size (see https://github.com/ytdl-org/youtube-dl/pull/27660). Tests only care that the correct data downloaded,
             # not what it decrypts to.

yt_dlp/downloader/hls.py

@@ -8,7 +8,14 @@ from .external import FFmpegFD
 from .fragment import FragmentFD
 from .. import webvtt
 from ..dependencies import Cryptodome
-from ..utils import bug_reports_message, parse_m3u8_attributes, update_url_query
+from ..utils import (
+    bug_reports_message,
+    parse_m3u8_attributes,
+    remove_start,
+    traverse_obj,
+    update_url_query,
+    urljoin,
+)
 
 
 class HlsFD(FragmentFD):
@@ -150,6 +157,13 @@ class HlsFD(FragmentFD):
         i = 0
         media_sequence = 0
         decrypt_info = {'METHOD': 'NONE'}
+        external_aes_key = traverse_obj(info_dict, ('hls_aes', 'key'))
+        if external_aes_key:
+            external_aes_key = binascii.unhexlify(remove_start(external_aes_key, '0x'))
+            assert len(external_aes_key) in (16, 24, 32), 'Invalid length for HLS AES-128 key'
+        external_aes_iv = traverse_obj(info_dict, ('hls_aes', 'iv'))
+        if external_aes_iv:
+            external_aes_iv = binascii.unhexlify(remove_start(external_aes_iv, '0x').zfill(32))
         byte_range = {}
         discontinuity_count = 0
         frag_index = 0
@@ -165,10 +179,7 @@ class HlsFD(FragmentFD):
                     frag_index += 1
                     if frag_index <= ctx['fragment_index']:
                         continue
-                    frag_url = (
+                    frag_url = urljoin(man_url, line)
-                        line
-                        if re.match(r'^https?://', line)
-                        else urllib.parse.urljoin(man_url, line))
                     if extra_query:
                         frag_url = update_url_query(frag_url, extra_query)
 
@@ -190,10 +201,7 @@ class HlsFD(FragmentFD):
                         return False
                     frag_index += 1
                     map_info = parse_m3u8_attributes(line[11:])
-                    frag_url = (
+                    frag_url = urljoin(man_url, map_info.get('URI'))
-                        map_info.get('URI')
-                        if re.match(r'^https?://', map_info.get('URI'))
-                        else urllib.parse.urljoin(man_url, map_info.get('URI')))
                     if extra_query:
                         frag_url = update_url_query(frag_url, extra_query)
 
@@ -218,15 +226,18 @@ class HlsFD(FragmentFD):
                     decrypt_url = decrypt_info.get('URI')
                     decrypt_info = parse_m3u8_attributes(line[11:])
                     if decrypt_info['METHOD'] == 'AES-128':
-                        if 'IV' in decrypt_info:
+                        if external_aes_iv:
+                            decrypt_info['IV'] = external_aes_iv
+                        elif 'IV' in decrypt_info:
                             decrypt_info['IV'] = binascii.unhexlify(decrypt_info['IV'][2:].zfill(32))
-                        if not re.match(r'^https?://', decrypt_info['URI']):
+                        if external_aes_key:
-                            decrypt_info['URI'] = urllib.parse.urljoin(
+                            decrypt_info['KEY'] = external_aes_key
-                                man_url, decrypt_info['URI'])
+                        else:
-                        if extra_query:
+                            decrypt_info['URI'] = urljoin(man_url, decrypt_info['URI'])
-                            decrypt_info['URI'] = update_url_query(decrypt_info['URI'], extra_query)
+                            if extra_query:
-                        if decrypt_url != decrypt_info['URI']:
+                                decrypt_info['URI'] = update_url_query(decrypt_info['URI'], extra_query)
-                            decrypt_info['KEY'] = None
+                            if decrypt_url != decrypt_info['URI']:
+                                decrypt_info['KEY'] = None
 
                 elif line.startswith('#EXT-X-MEDIA-SEQUENCE'):
                     media_sequence = int(line[22:])

yt_dlp/extractor/common.py

@@ -81,8 +81,8 @@ from ..utils import (
     update_Request,
     update_url_query,
     url_basename,
-    urlhandle_detect_ext,
     url_or_none,
+    urlhandle_detect_ext,
     urljoin,
     variadic,
     xpath_element,
@@ -220,6 +220,17 @@ class InfoExtractor:
                     * no_resume  The server does not support resuming the
                                  (HTTP or RTMP) download. Boolean.
                     * has_drm    The format has DRM and cannot be downloaded. Boolean
+                    * extra_param_to_segment_url  A query string to append to each
+                                 fragment's URL, or to update each existing query string
+                                 with. Only applied by the native HLS/DASH downloaders.
+                    * hls_aes    A dictionary of HLS AES-128 decryption information
+                                 used by the native HLS downloader to override the
+                                 values in the media playlist when an '#EXT-X-KEY' tag
+                                 is present in the playlist:
+                                 * uri  The URI from which the key will be downloaded
+                                 * key  The key (as hex) used to decrypt fragments.
+                                        If `key` is given, any key URI will be ignored
+                                 * iv   The IV (as hex) used to decrypt fragments
                     * downloader_options  A dictionary of downloader options
                                  (For internal use only)
                                  * http_chunk_size Chunk size for HTTP downloads

yt_dlp/extractor/vzaar.py

@@ -87,7 +87,7 @@ class VzaarIE(InfoExtractor):
                 m3u8_id='hls', fatal=False)
             if hls_aes:
                 for f in m3u8_formats:
-                    f['_decryption_key_url'] = url_templ % ('goose', '') + qs
+                    f['hls_aes'] = {'uri': url_templ % ('goose', '') + qs}
             formats.extend(m3u8_formats)
 
         return {