Determine merge container better (See desc) (#1482)

* Determine the container early. Closes #4069
* Use codecs instead of just file extensions
* Obey `--prefer-free-formats`
* Allow fallbacks in `--merge-output`

Authored by: pukkandan, selfisekai

1 files changed, 40 insertions, 0 deletions

diff --git a/yt_dlp/utils.py b/yt_dlp/utils.py
index c3ccb3a78..d405ed3e3 100644
--- a/yt_dlp/utils.py
+++ b/yt_dlp/utils.py
@@ -3456,6 +3456,46 @@ def parse_codecs(codecs_str):
     return {}
 
 
+def get_compatible_ext(*, vcodecs, acodecs, vexts, aexts, preferences=None):
+    assert len(vcodecs) == len(vexts) and len(acodecs) == len(aexts)
+
+    allow_mkv = not preferences or 'mkv' in preferences
+
+    if allow_mkv and max(len(acodecs), len(vcodecs)) > 1:
+        return 'mkv'  # TODO: any other format allows this?
+
+    # TODO: All codecs supported by parse_codecs isn't handled here
+    COMPATIBLE_CODECS = {
+        'mp4': {
+            'av1', 'hevc', 'avc1', 'mp4a',  # fourcc (m3u8, mpd)
+            'h264', 'aacl',  # Set in ISM
+        },
+        'webm': {
+            'av1', 'vp9', 'vp8', 'opus', 'vrbs',
+            'vp9x', 'vp8x',  # in the webm spec
+        },
+    }
+
+    sanitize_codec = functools.partial(try_get, getter=lambda x: x.split('.')[0].replace('0', ''))
+    vcodec, acodec = sanitize_codec(vcodecs[0]), sanitize_codec(acodecs[0])
+
+    for ext in preferences or COMPATIBLE_CODECS.keys():
+        codec_set = COMPATIBLE_CODECS.get(ext, set())
+        if ext == 'mkv' or codec_set.issuperset((vcodec, acodec)):
+            return ext
+
+    COMPATIBLE_EXTS = (
+        {'mp3', 'mp4', 'm4a', 'm4p', 'm4b', 'm4r', 'm4v', 'ismv', 'isma', 'mov'},
+        {'webm'},
+    )
+    for ext in preferences or vexts:
+        current_exts = {ext, *vexts, *aexts}
+        if ext == 'mkv' or current_exts == {ext} or any(
+                ext_sets.issuperset(current_exts) for ext_sets in COMPATIBLE_EXTS):
+            return ext
+    return 'mkv' if allow_mkv else preferences[-1]
+
+
 def urlhandle_detect_ext(url_handle):
     getheader = url_handle.headers.get