From 367429e23879ae127b82e7e8cacd62b878033e75 Mon Sep 17 00:00:00 2001
From: Elyse <26639800+elyse0@users.noreply.github.com>
Date: Thu, 9 Mar 2023 12:08:20 -0600
Subject: [PATCH] [common] Extract start and end keys for Dash fragments

---
 yt_dlp/extractor/common.py | 21 ++++++++++++++++++---
 1 file changed, 18 insertions(+), 3 deletions(-)

diff --git a/yt_dlp/extractor/common.py b/yt_dlp/extractor/common.py
index 2091df7fa..815538248 100644
--- a/yt_dlp/extractor/common.py
+++ b/yt_dlp/extractor/common.py
@@ -2566,7 +2566,7 @@ def extract_common(source):
                             r = int(s.get('r', 0))
                             ms_info['total_number'] += 1 + r
                             ms_info['s'].append({
-                                't': int(s.get('t', 0)),
+                                't': int_or_none(s.get('t')),
                                 # @d is mandatory (see [1, 5.3.9.6.2, Table 17, page 60])
                                 'd': int(s.attrib['d']),
                                 'r': r,
@@ -2608,9 +2608,16 @@ def extract_Initialization(source):
             return ms_info
 
         mpd_duration = parse_duration(mpd_doc.get('mediaPresentationDuration'))
+        availability_start_time = unified_timestamp(
+            mpd_doc.get('availabilityStartTime'), with_milliseconds=True) or 0
         formats, subtitles = [], {}
         stream_numbers = collections.defaultdict(int)
         for period in mpd_doc.findall(_add_ns('Period')):
+            # segmentIngestTime is completely out of spec, but YT Livestream do this
+            segment_ingest_time = period.get('{http://youtube.com/yt/2012/10/10}segmentIngestTime')
+            if segment_ingest_time:
+                availability_start_time = unified_timestamp(segment_ingest_time, with_milliseconds=True)
+
             period_duration = parse_duration(period.get('duration')) or mpd_duration
             period_ms_info = extract_multisegment_info(period, {
                 'start_number': 1,
@@ -2784,13 +2791,17 @@ def add_segment_url():
                                     'Bandwidth': bandwidth,
                                     'Number': segment_number,
                                 }
+                                duration = float_or_none(segment_d, representation_ms_info['timescale'])
+                                start = float_or_none(segment_time, representation_ms_info['timescale'])
                                 representation_ms_info['fragments'].append({
                                     media_location_key: segment_url,
-                                    'duration': float_or_none(segment_d, representation_ms_info['timescale']),
+                                    'duration': duration,
+                                    'start': availability_start_time + start,
+                                    'end': availability_start_time + start + duration,
                                 })
 
                             for num, s in enumerate(representation_ms_info['s']):
-                                segment_time = s.get('t') or segment_time
+                                segment_time = s['t'] if s.get('t') is not None else segment_time
                                 segment_d = s['d']
                                 add_segment_url()
                                 segment_number += 1
@@ -2806,6 +2817,7 @@ def add_segment_url():
                         fragments = []
                         segment_index = 0
                         timescale = representation_ms_info['timescale']
+                        start = 0
                         for s in representation_ms_info['s']:
                             duration = float_or_none(s['d'], timescale)
                             for r in range(s.get('r', 0) + 1):
@@ -2813,8 +2825,11 @@ def add_segment_url():
                                 fragments.append({
                                     location_key(segment_uri): segment_uri,
                                     'duration': duration,
+                                    'start': availability_start_time + start,
+                                    'end': availability_start_time + start + duration,
                                 })
                                 segment_index += 1
+                                start += duration
                         representation_ms_info['fragments'] = fragments
                     elif 'segment_urls' in representation_ms_info:
                         # Segment URLs with no SegmentTimeline