Merge 6abf89a9c7 into f2a4983df7

Closes #11526
Authored by: jshumphrey

supportedsites.md

@@ -1468,6 +1468,8 @@
  - **ThisVid**
  - **ThisVidMember**
  - **ThisVidPlaylist**
+ - **Threads**
+ - **ThreadsIOS**: Threads' iOS `barcelona://` URL
  - **ThreeSpeak**
  - **ThreeSpeakUser**
  - **TikTok**

yt_dlp/extractor/_extractors.py

@@ -2090,6 +2090,10 @@ from .thisvid import (
     ThisVidMemberIE,
     ThisVidPlaylistIE,
 )
+from .threads import (
+    ThreadsIE,
+    ThreadsIOSIE,
+)
 from .threeqsdn import ThreeQSDNIE
 from .threespeak import (
     ThreeSpeakIE,

yt_dlp/extractor/archiveorg.py

@@ -205,6 +205,26 @@ class ArchiveOrgIE(InfoExtractor):
                 },
             },
         ],
+    }, {
+        # The reviewbody is None for one of the reviews; just need to extract data without crashing
+        'url': 'https://archive.org/details/gd95-04-02.sbd.11622.sbeok.shnf/gd95-04-02d1t04.shn',
+        'info_dict': {
+            'id': 'gd95-04-02.sbd.11622.sbeok.shnf/gd95-04-02d1t04.shn',
+            'ext': 'mp3',
+            'title': 'Stuck Inside of Mobile with the Memphis Blues Again',
+            'creators': ['Grateful Dead'],
+            'duration': 338.31,
+            'track': 'Stuck Inside of Mobile with the Memphis Blues Again',
+            'description': 'md5:764348a470b986f1217ffd38d6ac7b72',
+            'display_id': 'gd95-04-02d1t04.shn',
+            'location': 'Pyramid Arena',
+            'uploader': 'jon@archive.org',
+            'album': '1995-04-02 - Pyramid Arena',
+            'upload_date': '20040519',
+            'track_number': 4,
+            'release_date': '19950402',
+            'timestamp': 1084927901,
+        },
     }]
 
     @staticmethod
@@ -335,7 +355,7 @@ class ArchiveOrgIE(InfoExtractor):
                 info['comments'].append({
                     'id': review.get('review_id'),
                     'author': review.get('reviewer'),
-                    'text': str_or_none(review.get('reviewtitle'), '') + '\n\n' + review.get('reviewbody'),
+                    'text': join_nonempty('reviewtitle', 'reviewbody', from_dict=review, delim='\n\n'),
                     'timestamp': unified_timestamp(review.get('createdate')),
                     'parent': 'root'})
 

yt_dlp/extractor/facebook.py

@@ -563,13 +563,13 @@ class FacebookIE(InfoExtractor):
                 return extract_video_data(try_get(
                     js_data, lambda x: x['jsmods']['instances'], list) or [])
 
-        def extract_dash_manifest(video, formats):
+        def extract_dash_manifest(vid_data, formats, mpd_url=None):
             dash_manifest = traverse_obj(
-                video, 'dash_manifest', 'playlist', 'dash_manifest_xml_string', expected_type=str)
+                vid_data, 'dash_manifest', 'playlist', 'dash_manifest_xml_string', 'manifest_xml', expected_type=str)
             if dash_manifest:
                 formats.extend(self._parse_mpd_formats(
                     compat_etree_fromstring(urllib.parse.unquote_plus(dash_manifest)),
-                    mpd_url=url_or_none(video.get('dash_manifest_url'))))
+                    mpd_url=url_or_none(video.get('dash_manifest_url')) or mpd_url))
 
         def process_formats(info):
             # Downloads with browser's User-Agent are rate limited. Working around
@@ -619,9 +619,12 @@ class FacebookIE(InfoExtractor):
                         video = video['creation_story']
                         video['owner'] = traverse_obj(video, ('short_form_video_context', 'video_owner'))
                         video.update(reel_info)
-                    fmt_data = traverse_obj(video, ('videoDeliveryLegacyFields', {dict})) or video
+
                     formats = []
                     q = qualities(['sd', 'hd'])
+
+                    # Legacy formats extraction
+                    fmt_data = traverse_obj(video, ('videoDeliveryLegacyFields', {dict})) or video
                     for key, format_id in (('playable_url', 'sd'), ('playable_url_quality_hd', 'hd'),
                                            ('playable_url_dash', ''), ('browser_native_hd_url', 'hd'),
                                            ('browser_native_sd_url', 'sd')):
@@ -629,7 +632,7 @@ class FacebookIE(InfoExtractor):
                         if not playable_url:
                             continue
                         if determine_ext(playable_url) == 'mpd':
-                            formats.extend(self._extract_mpd_formats(playable_url, video_id))
+                            formats.extend(self._extract_mpd_formats(playable_url, video_id, fatal=False))
                         else:
                             formats.append({
                                 'format_id': format_id,
@@ -638,6 +641,28 @@ class FacebookIE(InfoExtractor):
                                 'url': playable_url,
                             })
                     extract_dash_manifest(fmt_data, formats)
+
+                    # New videoDeliveryResponse formats extraction
+                    fmt_data = traverse_obj(video, ('videoDeliveryResponseFragment', 'videoDeliveryResponseResult'))
+                    mpd_urls = traverse_obj(fmt_data, ('dash_manifest_urls', ..., 'manifest_url', {url_or_none}))
+                    dash_manifests = traverse_obj(fmt_data, ('dash_manifests', lambda _, v: v['manifest_xml']))
+                    for idx, dash_manifest in enumerate(dash_manifests):
+                        extract_dash_manifest(dash_manifest, formats, mpd_url=traverse_obj(mpd_urls, idx))
+                    if not dash_manifests:
+                        # Only extract from MPD URLs if the manifests are not already provided
+                        for mpd_url in mpd_urls:
+                            formats.extend(self._extract_mpd_formats(mpd_url, video_id, fatal=False))
+                    for prog_fmt in traverse_obj(fmt_data, ('progressive_urls', lambda _, v: v['progressive_url'])):
+                        format_id = traverse_obj(prog_fmt, ('metadata', 'quality', {str.lower}))
+                        formats.append({
+                            'format_id': format_id,
+                            # sd, hd formats w/o resolution info should be deprioritized below DASH
+                            'quality': q(format_id) - 3,
+                            'url': prog_fmt['progressive_url'],
+                        })
+                    for m3u8_url in traverse_obj(fmt_data, ('hls_playlist_urls', ..., 'hls_playlist_url', {url_or_none})):
+                        formats.extend(self._extract_m3u8_formats(m3u8_url, video_id, 'mp4', fatal=False, m3u8_id='hls'))
+
                     if not formats:
                         # Do not append false positive entry w/o any formats
                         return

yt_dlp/extractor/threads.py

@@ -0,0 +1,158 @@
+from .common import InfoExtractor
+from ..utils import (
+    remove_end,
+    strftime_or_none,
+    strip_or_none,
+)
+from ..utils.traversal import traverse_obj
+
+
+class ThreadsIE(InfoExtractor):
+    _VALID_URL = r'https?://(?:www\.)?threads\.net/(?P<uploader>[^/]+)/post/(?P<id>[^/?#&]+)/?(?P<embed>embed.*?)?'
+
+    _TESTS = [{
+        'url': 'https://www.threads.net/@tntsportsbr/post/C6cqebdCfBi',
+        'info_dict': {
+            'id': 'C6cqebdCfBi',
+            'ext': 'mp4',
+            'title': 'md5:062673d04195aa2d99b8d7a11798cb9d',
+            'description': 'md5:fe0c73f9a892fb92efcc67cc075561b0',
+            'uploader': 'TNT Sports Brasil',
+            'uploader_id': 'tntsportsbr',
+            'uploader_url': 'https://www.threads.net/@tntsportsbr',
+            'channel': 'tntsportsbr',
+            'channel_url': 'https://www.threads.net/@tntsportsbr',
+            'timestamp': 1714613811,
+            'upload_date': '20240502',
+            'like_count': int,
+            'channel_is_verified': bool,
+            'thumbnail': r're:^https?://.*\.jpg',
+        },
+    }, {
+        'url': 'https://www.threads.net/@felipebecari/post/C6cM_yNPHCF',
+        'info_dict': {
+            'id': 'C6cM_yNPHCF',
+            'ext': 'mp4',
+            'title': '@felipebecari • Sobre o futuro dos dois últimos resgatados: tem muita notícia boa! 🐶❤️',
+            'description': 'Sobre o futuro dos dois últimos resgatados: tem muita notícia boa! 🐶❤️',
+            'uploader': 'Felipe Becari',
+            'uploader_id': 'felipebecari',
+            'uploader_url': 'https://www.threads.net/@felipebecari',
+            'channel': 'felipebecari',
+            'channel_url': 'https://www.threads.net/@felipebecari',
+            'timestamp': 1714598318,
+            'upload_date': '20240501',
+            'like_count': int,
+            'channel_is_verified': bool,
+            'thumbnail': r're:^https?://.*\.jpg',
+        },
+    }]
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+        webpage = self._download_webpage(url, video_id)
+        metadata = {}
+
+        # Try getting videos from json
+        json_data = self._search_regex(
+            rf'<script[^>]+>(.*"code":"{video_id}".*)</script>',
+            webpage, 'main json', fatal=True)
+
+        result = self._search_json(
+            r'"result":', json_data,
+            'result data', video_id, fatal=True)
+
+        edges = traverse_obj(result, ('data', 'data', 'edges'))
+
+        for node in edges:
+            items = traverse_obj(node, ('node', 'thread_items'))
+
+            for item in items:
+                post = item.get('post')
+
+                if post and post.get('code') == video_id:
+                    formats = []
+                    thumbnails = []
+
+                    # Videos
+                    if post.get('carousel_media') is not None:  # Handle multiple videos posts
+                        media_list = post.get('carousel_media')
+                    else:
+                        media_list = [post]
+
+                    for media in media_list:
+                        videos = media.get('video_versions')
+
+                        if videos:
+                            for video in videos:
+                                formats.append({
+                                    'format_id': '{}-{}'.format(media.get('pk'), video['type']),  # id-type
+                                    'url': video['url'],
+                                    'width': media.get('original_width'),
+                                    'height': media.get('original_height'),
+                                })
+
+                    # Thumbnails
+                    thumbs = traverse_obj(post, ('image_versions2', 'candidates'))
+
+                    for thumb in thumbs:
+                        thumbnails.append({
+                            'url': thumb['url'],
+                            'width': thumb['width'],
+                            'height': thumb['height'],
+                        })
+
+                    # Metadata
+                    metadata.setdefault('uploader_id', traverse_obj(post, ('user', 'username')))
+                    metadata.setdefault('channel_is_verified', traverse_obj(post, ('user', 'is_verified')))
+                    metadata.setdefault('uploader_url', 'https://www.threads.net/@{}'.format(traverse_obj(post, ('user', 'username'))))
+                    metadata.setdefault('timestamp', post.get('taken_at'))
+                    metadata.setdefault('like_count', post.get('like_count'))
+
+        # Try getting metadata
+        metadata['id'] = video_id
+        metadata['title'] = strip_or_none(remove_end(self._html_extract_title(webpage), '• Threads'))
+        metadata['description'] = self._og_search_description(webpage)
+
+        metadata['channel'] = metadata.get('uploader_id')
+        metadata['channel_url'] = metadata.get('uploader_url')
+        metadata['uploader'] = self._search_regex(r'(.*?) \(', self._og_search_title(webpage), 'uploader', metadata.get('uploader_id'))
+        metadata['upload_date'] = strftime_or_none(metadata.get('timestamp'))
+
+        return {
+            **metadata,
+            'formats': formats,
+            'thumbnails': thumbnails,
+        }
+
+
+class ThreadsIOSIE(InfoExtractor):
+    IE_DESC = 'IOS barcelona:// URL'
+    _VALID_URL = r'barcelona://media\?shortcode=(?P<id>[^/?#&]+)'
+    _TESTS = [{
+        'url': 'barcelona://media?shortcode=C6fDehepo5D',
+        'info_dict': {
+            'id': 'C6fDehepo5D',
+            'ext': 'mp4',
+            'title': 'md5:dc92f960981b8b3a33eba9681e9fdfc6',
+            'description': 'md5:0c36a7e67e1517459bc0334dba932164',
+            'uploader': 'Sa\u0303o Paulo Futebol Clube',
+            'uploader_id': 'saopaulofc',
+            'uploader_url': 'https://www.threads.net/@saopaulofc',
+            'channel': 'saopaulofc',
+            'channel_url': 'https://www.threads.net/@saopaulofc',
+            'timestamp': 1714694014,
+            'upload_date': '20240502',
+            'like_count': int,
+            'channel_is_verified': bool,
+            'thumbnail': r're:^https?://.*\.jpg',
+        },
+        'add_ie': ['Threads'],
+    }]
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+
+        # Threads doesn't care about the user url, it redirects to the right one
+        # So we use ** instead so that we don't need to find it
+        return self.url_result(f'http://www.threads.net/**/post/{video_id}', ThreadsIE, video_id)