mirror of
https://github.com/unshackle-dl/unshackle.git
synced 2026-11-11 19:45:06 +00:00
imSp4rky
dc188b7276
Some packagers renumber AdaptationSet and Representation ids in every Period, so audio took video segments after the first Period. A match never crosses codec, size or language.
166 lines
7.4 KiB
Python
166 lines
7.4 KiB
Python
"""DASH.resolve_representation must continue a track into a later Period without crossing codec, size or language."""
|
|
|
|
from __future__ import annotations
|
|
|
|
from typing import Any
|
|
|
|
import pytest
|
|
|
|
from tests.core.test_manifest_corpus import CORPUS, INDEX, KNOWN_FAILURES, parse
|
|
from unshackle.core.manifests import DASH
|
|
from unshackle.core.utils.xml import load_xml
|
|
|
|
RENUMBERED = b"""<MPD type="static" mediaPresentationDuration="PT20S">
|
|
<Period id="0" duration="PT10S">
|
|
<AdaptationSet id="0" contentType="text" lang="en">
|
|
<Representation id="0" bandwidth="1000" codecs="wvtt" mimeType="application/mp4"/>
|
|
</AdaptationSet>
|
|
<AdaptationSet id="1" contentType="video">
|
|
<Representation id="1" bandwidth="2000000" codecs="avc1" width="720" height="540"/>
|
|
<Representation id="2" bandwidth="500000" codecs="avc1" width="640" height="480"/>
|
|
</AdaptationSet>
|
|
<AdaptationSet id="2" contentType="audio" lang="en">
|
|
<Representation id="3" bandwidth="128000" codecs="mp4a.40.2" audioSamplingRate="48000"/>
|
|
</AdaptationSet>
|
|
</Period>
|
|
<Period id="1" duration="PT10S">
|
|
<AdaptationSet id="0" contentType="audio" lang="en">
|
|
<Representation id="0" bandwidth="129000" codecs="mp4a.40.2" audioSamplingRate="48000"/>
|
|
</AdaptationSet>
|
|
<AdaptationSet id="1" contentType="video">
|
|
<Representation id="1" bandwidth="480000" codecs="avc1" width="640" height="480"/>
|
|
<Representation id="2" bandwidth="1900000" codecs="avc1" width="720" height="540"/>
|
|
<Representation id="3" bandwidth="3000000" codecs="avc1" width="720" height="540"/>
|
|
</AdaptationSet>
|
|
</Period>
|
|
</MPD>"""
|
|
|
|
|
|
def video(rep_id: str, bandwidth: str, codecs: str = "avc1.640028") -> str:
|
|
return f'<Representation id="{rep_id}" bandwidth="{bandwidth}" codecs="{codecs}" width="1920" height="1080" mimeType="video/mp4"/>'
|
|
|
|
|
|
def audio(rep_id: str) -> str:
|
|
return f'<Representation id="{rep_id}" bandwidth="128000" codecs="mp4a.40.2" audioSamplingRate="48000" mimeType="audio/mp4"/>'
|
|
|
|
|
|
def two_periods(first: str, second: str) -> bytes:
|
|
return f'<MPD><Period id="0">{first}</Period><Period id="1">{second}</Period></MPD>'.encode()
|
|
|
|
|
|
def resolve(mpd: bytes, set_index: int, rep_id: str) -> tuple[str | None, str | None, str | None] | None:
|
|
p0, p1 = load_xml(mpd).findall("Period")
|
|
found = DASH.resolve_representation(p1, rep_id, p0.findall("AdaptationSet")[set_index])
|
|
if found is None:
|
|
return None
|
|
adaptation_set, rep = found
|
|
return adaptation_set.get("lang"), rep.get("id"), DASH.get_attr("codecs", adaptation_set, rep)
|
|
|
|
|
|
def test_renumbered_audio_continues_into_audio_not_a_video_rep_with_its_id():
|
|
assert resolve(RENUMBERED, 2, "3") == ("en", "0", "mp4a.40.2")
|
|
|
|
|
|
def test_renumbered_video_continues_into_the_same_size():
|
|
assert resolve(RENUMBERED, 1, "1") == (None, "2", "avc1")
|
|
assert resolve(RENUMBERED, 1, "2") == (None, "1", "avc1")
|
|
|
|
|
|
def test_track_missing_from_the_period_is_not_matched_to_another_type():
|
|
assert resolve(RENUMBERED, 0, "0") is None
|
|
|
|
|
|
def test_untyped_sets_that_swap_order_keep_audio_on_audio():
|
|
mpd = two_periods(
|
|
f"<AdaptationSet>{video('v', '5000000')}</AdaptationSet><AdaptationSet>{audio('a')}</AdaptationSet>",
|
|
f"<AdaptationSet>{audio('a')}</AdaptationSet><AdaptationSet>{video('v', '5000000')}</AdaptationSet>",
|
|
)
|
|
assert resolve(mpd, 1, "a") == (None, "a", "mp4a.40.2")
|
|
|
|
|
|
def test_codec_missing_from_the_period_is_skipped_not_swapped():
|
|
mpd = two_periods(
|
|
f'<AdaptationSet contentType="video">{video("avc", "5000000")}</AdaptationSet>'
|
|
f'<AdaptationSet contentType="video">{video("hevc", "3000000", "hvc1.2.4.L150")}</AdaptationSet>',
|
|
f'<AdaptationSet contentType="video">{video("avc", "5000000")}</AdaptationSet>',
|
|
)
|
|
assert resolve(mpd, 1, "hevc") is None
|
|
|
|
|
|
def test_renumbered_sets_that_swap_order_keep_the_codec():
|
|
mpd = two_periods(
|
|
f'<AdaptationSet id="0" contentType="video">{video("0", "8000000", "hvc1.2.4.L150")}</AdaptationSet>'
|
|
f'<AdaptationSet id="1" contentType="video">{video("1", "8000000", "dvh1.08.06")}</AdaptationSet>',
|
|
f'<AdaptationSet id="0" contentType="video">{video("0", "8000000", "dvh1.08.06")}</AdaptationSet>'
|
|
f'<AdaptationSet id="1" contentType="video">{video("1", "8000000", "hvc1.2.4.L150")}</AdaptationSet>',
|
|
)
|
|
assert resolve(mpd, 1, "1") == (None, "0", "dvh1.08.06")
|
|
|
|
|
|
@pytest.mark.parametrize("content", ["audio", "text"])
|
|
def test_language_missing_from_the_period_is_skipped_not_swapped(content: str):
|
|
def rep(rep_id: str) -> str:
|
|
return (
|
|
audio(rep_id) if content == "audio" else f'<Representation id="{rep_id}" bandwidth="1000" codecs="wvtt"/>'
|
|
)
|
|
|
|
mpd = two_periods(
|
|
f'<AdaptationSet id="1" contentType="{content}" lang="en">{rep("x")}</AdaptationSet>',
|
|
f'<AdaptationSet id="2" contentType="{content}" lang="de">{rep("y")}</AdaptationSet>',
|
|
)
|
|
assert resolve(mpd, 0, "x") is None
|
|
|
|
|
|
def test_regional_tag_continues_into_the_same_language():
|
|
mpd = two_periods(
|
|
f'<AdaptationSet id="1" contentType="audio" lang="en">{audio("a")}</AdaptationSet>',
|
|
f'<AdaptationSet id="5" contentType="audio" lang="en-US">{audio("b")}</AdaptationSet>',
|
|
)
|
|
assert resolve(mpd, 0, "a") == ("en-US", "b", "mp4a.40.2")
|
|
|
|
|
|
def test_same_identity_sets_that_swap_order_follow_the_rep_id():
|
|
dialog = f'<AdaptationSet contentType="audio" lang="en">{audio("dialog")}</AdaptationSet>'
|
|
described = f'<AdaptationSet contentType="audio" lang="en">{audio("ad")}</AdaptationSet>'
|
|
assert resolve(two_periods(dialog + described, described + dialog), 0, "dialog") == ("en", "dialog", "mp4a.40.2")
|
|
|
|
|
|
def test_stable_ladder_follows_bandwidth_rank_when_every_rung_drifts():
|
|
mpd = two_periods(
|
|
f'<AdaptationSet id="1" contentType="video">{video("v1", "5000000")}{video("v2", "6000000")}</AdaptationSet>',
|
|
f'<AdaptationSet id="1" contentType="video">{video("v1", "3000000")}{video("v2", "4500000")}</AdaptationSet>',
|
|
)
|
|
assert resolve(mpd, 0, "v1") == (None, "v1", "avc1.640028")
|
|
assert resolve(mpd, 0, "v2") == (None, "v2", "avc1.640028")
|
|
|
|
|
|
def test_non_integer_bandwidth_does_not_raise():
|
|
mpd = two_periods(
|
|
f'<AdaptationSet id="1" contentType="video">{video("v1", "5000000")}</AdaptationSet>',
|
|
f'<AdaptationSet id="1" contentType="video">{video("v1", "5e6")}</AdaptationSet>',
|
|
)
|
|
assert resolve(mpd, 0, "v1") == (None, "v1", "avc1.640028")
|
|
|
|
|
|
def multi_period(entry: dict[str, Any]) -> bool:
|
|
periods = load_xml((CORPUS / entry["file"]).read_bytes()).findall("Period")
|
|
return sum(DASH.is_content_period(p, []) for p in periods) > 1
|
|
|
|
|
|
@pytest.mark.parametrize(
|
|
"entry",
|
|
[e for e in INDEX if e["kind"] == "dash" and e["file"] not in KNOWN_FAILURES and multi_period(e)],
|
|
ids=lambda e: e["file"],
|
|
)
|
|
def test_corpus_matches_never_cross_codec_or_size(entry: dict[str, Any]):
|
|
shape = ("codecs", "width", "height")
|
|
for track in parse(entry):
|
|
data = track.data["dash"]
|
|
want = tuple(DASH.get_attr(a, data["adaptation_set"], data["representation"]) for a in shape)
|
|
for period in data["manifest"].findall("Period"):
|
|
if period is data["period"] or not DASH.is_content_period(period, []):
|
|
continue
|
|
found = DASH.resolve_representation(period, data["representation_id"], data["adaptation_set"])
|
|
if found is not None:
|
|
assert tuple(DASH.get_attr(a, *found) for a in shape) == want
|