From a0a1bc3d8d8e3bb9a48a06e835815a0460e90e77 Mon Sep 17 00:00:00 2001
From: bashonly <88596187+bashonly@users.noreply.github.com>
Date: Tue, 23 Jul 2024 17:00:57 -0500
Subject: [PATCH] [ie/vimeo] Fix chapters extraction (#10544)
Closes #5308
Authored by: bashonly
---
yt_dlp/extractor/vimeo.py | 51 ++++++++++++++++++++++++++++++---------
1 file changed, 39 insertions(+), 12 deletions(-)
diff --git a/yt_dlp/extractor/vimeo.py b/yt_dlp/extractor/vimeo.py
index 18eb08444..d10689cd8 100644
--- a/yt_dlp/extractor/vimeo.py
+++ b/yt_dlp/extractor/vimeo.py
@@ -212,16 +212,6 @@ class VimeoBaseInfoExtractor(InfoExtractor):
owner = video_data.get('owner') or {}
video_uploader_url = owner.get('url')
- duration = int_or_none(video_data.get('duration'))
- chapter_data = try_get(config, lambda x: x['embed']['chapters']) or []
- chapters = [{
- 'title': current_chapter.get('title'),
- 'start_time': current_chapter.get('timecode'),
- 'end_time': next_chapter.get('timecode'),
- } for current_chapter, next_chapter in zip(chapter_data, chapter_data[1:] + [{'timecode': duration}])]
- if chapters and chapters[0]['start_time']: # Chapters may not start from 0
- chapters[:0] = [{'title': '<Untitled>', 'start_time': 0, 'end_time': chapters[0]['start_time']}]
-
return {
'id': str_or_none(video_data.get('id')) or video_id,
'title': video_title,
@@ -229,8 +219,12 @@ class VimeoBaseInfoExtractor(InfoExtractor):
'uploader_id': video_uploader_url.split('/')[-1] if video_uploader_url else None,
'uploader_url': video_uploader_url,
'thumbnails': thumbnails,
- 'duration': duration,
- 'chapters': chapters or None,
+ 'duration': int_or_none(video_data.get('duration')),
+ 'chapters': sorted(traverse_obj(config, (
+ 'embed', 'chapters', lambda _, v: int(v['timecode']) is not None, {
+ 'title': ('title', {str}),
+ 'start_time': ('timecode', {int_or_none}),
+ })), key=lambda c: c['start_time']) or None,
'formats': formats,
'subtitles': subtitles,
'live_status': live_status,
@@ -708,6 +702,39 @@ class VimeoIE(VimeoBaseInfoExtractor):
'skip_download': True,
},
},
+ {
+ # chapters must be sorted, see: https://github.com/yt-dlp/yt-dlp/issues/5308
+ 'url': 'https://player.vimeo.com/video/756714419',
+ 'info_dict': {
+ 'id': '756714419',
+ 'ext': 'mp4',
+ 'title': 'Dr Arielle Schwartz - Therapeutic yoga for optimum sleep',
+ 'uploader': 'Alex Howard',
+ 'uploader_id': 'user54729178',
+ 'uploader_url': 'https://vimeo.com/user54729178',
+ 'thumbnail': r're:https://i\.vimeocdn\.com/video/1520099929-[\da-f]+-d_1280',
+ 'duration': 2636,
+ 'chapters': [
+ {'start_time': 0, 'end_time': 10, 'title': '<Untitled Chapter 1>'},
+ {'start_time': 10, 'end_time': 106, 'title': 'Welcoming Dr Arielle Schwartz'},
+ {'start_time': 106, 'end_time': 305, 'title': 'What is therapeutic yoga?'},
+ {'start_time': 305, 'end_time': 594, 'title': 'Vagal toning practices'},
+ {'start_time': 594, 'end_time': 888, 'title': 'Trauma and difficulty letting go'},
+ {'start_time': 888, 'end_time': 1059, 'title': "Dr Schwartz' insomnia experience"},
+ {'start_time': 1059, 'end_time': 1471, 'title': 'A strategy for helping sleep issues'},
+ {'start_time': 1471, 'end_time': 1667, 'title': 'Yoga nidra'},
+ {'start_time': 1667, 'end_time': 2121, 'title': 'Wisdom in stillness'},
+ {'start_time': 2121, 'end_time': 2386, 'title': 'What helps us be more able to let go?'},
+ {'start_time': 2386, 'end_time': 2510, 'title': 'Practical tips to help ourselves'},
+ {'start_time': 2510, 'end_time': 2636, 'title': 'Where to find out more'},
+ ],
+ },
+ 'params': {
+ 'http_headers': {'Referer': 'https://sleepsuperconference.com'},
+ 'skip_download': 'm3u8',
+ },
+ 'expected_warnings': ['Failed to parse XML: not well-formed'],
+ },
{
# user playlist alias -> https://vimeo.com/258705797
'url': 'https://vimeo.com/user26785108/newspiritualguide',