Regardless of which chapter I choose, the audio starts playing from the beginning every time

[winstonma]

I cloned the repo 7b37cae version and built the flatpak image.

Then I opened the .opus audiobook using cozy. It could read the chapters from the audiobook and display in cozy. However when I click a specific chapter it always play the beginning of the audiobook (maybe the first chapter). In this version all chapter name and chapter time is correct. The only unexpected thing here is cozy would play the beginning of the audiobook when I click other chapters.

The flatpak version using c02e73e with PR #969 works flawlessly so my primary suspect is commit ff983a0. BTW I saved the working flatpak version is here.

Pls don't forget to Scan Library everytime when cozy is started

[winstonma]

I printed out the media_file.chapters for my audiobook:

This is the .M4B files that Cozy can go to the correct chapter:

 Chapter(name='End Credits',
         position=16681144000000,
         length=37.76800000000003,
         number=22)]

This is a Chapter inside .Opus file, which the player always start at the beginning of the audiofile:

 Chapter(name='A Note on the Recording',
         position=43.899,
         length=89.584,
         number=3),

I am not sure if the position of the Opus chapter is too small that Cozy would always seek the first chapter

EDIT: According to chapter.py the position parameter should be in seconds. So the M4B is version is not correct. However the M4B could go to the correct chapter.

[winstonma]

I added this few lines and opus chapter works for me.

diff --git a/cozy/media/tag_reader.py b/cozy/media/tag_reader.py
index 010e5176..141d43b2 100644
--- a/cozy/media/tag_reader.py
+++ b/cozy/media/tag_reader.py
@@ -231,7 +231,7 @@ class TagReader:
             if tag.lower().endswith("name"):
                 chapter_dict[chapter_num].name = value
             elif len(tag) == 10:
-                chapter_dict[chapter_num].position = self._vorbis_timestamp_to_secs(value)
+                chapter_dict[chapter_num].position = self._vorbis_timestamp_to_secs(value) * Gst.SECOND
 
         if not chapter_dict:
             return self._get_single_file_chapter()
@@ -242,13 +242,11 @@ class TagReader:
                 return self._get_single_file_chapter()
 
             if prev_chapter:
-                prev_chapter.length = chapter.position - prev_chapter.position
+                prev_chapter.length = (chapter.position - prev_chapter.position) / Gst.SECOND
 
             chapter_list.append(chapter)
             prev_chapter = chapter
 
-        prev_chapter.length = self._get_length_in_seconds() - prev_chapter.position
-
         return chapter_list

But I guess the position parameter is still not in seconds. Also in tag_reader.py the length calculation for each chapter could be refactored across _get_mp4_chapters, _get_mp3_chapters and _get_ogg_chapters.