Merge remote-tracking branch 'origin/development' into development

libs/fese/__init__.py

@@ -14,7 +14,7 @@ from babelfish import Language
 from babelfish.exceptions import LanguageError
 import pysubs2
 
-__version__ = "0.1.3"
+__version__ = "0.1.4"
 
 logger = logging.getLogger(__name__)
 
@@ -127,6 +127,11 @@ class FFprobeSubtitleStream:
         self.time_base = stream.get("time_base")
         self.tags = stream.get("tags", {})
         self.start_time = float(stream.get("start_time", 0))
+        # TODO: separate tags
+        self.number_of_frames = int(self.tags.get("NUMBER_OF_FRAMES", 0))
+        self.number_of_frames_eng = int(
+            self.tags.get("NUMBER_OF_FRAMES-eng", self.number_of_frames)
+        )
 
         self.duration, self.duration_ts = 0, 0
 

libs/subliminal_patch/core.py

@@ -105,10 +105,12 @@ class SZProviderPool(ProviderPool):
         # Check if the pool was initialized enough hours ago
         self._check_lifetime()
 
+        providers = set(providers or [])
+
         # Check if any new provider has been added
-        updated = set(providers) != self.providers or ban_list != self.ban_list
+        updated = providers != self.providers or ban_list != self.ban_list
-        removed_providers = list(sorted(self.providers - set(providers)))
+        removed_providers = list(sorted(self.providers - providers))
-        new_providers = list(sorted(set(providers) - self.providers))
+        new_providers = list(sorted(providers - self.providers))
 
         # Terminate and delete removed providers from instance
         for removed in removed_providers:
@@ -128,8 +130,6 @@ class SZProviderPool(ProviderPool):
             self.providers.difference_update(removed_providers)
             self.providers.update(list(providers))
 
-        self.blacklist = blacklist
-
         # Restart providers with new configs
         for key, val in provider_configs.items():
             # Don't restart providers that are not enabled
@@ -154,6 +154,9 @@ class SZProviderPool(ProviderPool):
 
         self.provider_configs = provider_configs
 
+        self.blacklist = blacklist or []
+        self.ban_list = ban_list or {'must_contain': [], 'must_not_contain': []}
+
         return updated
 
     def _check_lifetime(self):

libs/subliminal_patch/providers/embeddedsubtitles.py

@@ -40,7 +40,7 @@ class EmbeddedSubtitle(Subtitle):
         self.container: FFprobeVideoContainer = container
         self.forced = stream.disposition.forced
         self.page_link = self.container.path
-        self.release_info = os.path.basename(self.page_link)
+        self.release_info = _get_pretty_release_name(stream, container)
         self.media_type = media_type
 
         self._matches: set = matches
@@ -117,6 +117,8 @@ class EmbeddedSubtitlesProvider(Provider):
             self._blacklist.add(path)
             streams = []
 
+        streams = _discard_possible_incomplete_subtitles(list(streams))
+
         if not streams:
             logger.debug("No subtitles found for container: %s", video)
 
@@ -260,6 +262,39 @@ def _check_hi_fallback(streams, languages):
             logger.debug("HI fallback not needed: %s", compatible_streams)
 
 
+def _discard_possible_incomplete_subtitles(streams):
+    """Check number_of_frames attributes from subtitle streams in order to find
+    supposedly incomplete subtitles"""
+    try:
+        max_frames = max(stream.number_of_frames for stream in streams)
+    except ValueError:
+        return []
+
+    # Blatantly assume there's nothing to discard as some ffprobe streams don't
+    # have number_of_frames tags
+    if not max_frames:
+        return streams
+
+    logger.debug("Checking possible incomplete subtitles (max frames: %d)", max_frames)
+
+    valid_streams = []
+
+    for stream in streams:
+        # 500 < 1200
+        if stream.number_of_frames < max_frames // 2:
+            logger.debug(
+                "Possible bad subtitle found: %s (%s frames - %s frames)",
+                stream,
+                stream.number_of_frames,
+                max_frames,
+            )
+            continue
+
+        valid_streams.append(stream)
+
+    return valid_streams
+
+
 def _is_fuse_rclone_mount(path: str):
     # Experimental!
 
@@ -272,3 +307,8 @@ def _is_fuse_rclone_mount(path: str):
     # https://forum.rclone.org/t/fuse-inode-number-aufs/215/5
     # https://pkg.go.dev/bazil.org/fuse/fs?utm_source=godoc#GenerateDynamicInode
     return len(str(os.stat(path).st_ino)) > 18
+
+
+def _get_pretty_release_name(stream, container):
+    bname = os.path.basename(container.path)
+    return f"{os.path.splitext(bname)[0]}.{stream.suffix}"

tests/subliminal_patch/test_embeddedsubtitles.py

@@ -9,10 +9,11 @@ import subliminal_patch
 from subliminal_patch.core import Episode
 from subliminal_patch.core import Movie
 from subliminal_patch.exceptions import MustGetBlacklisted
-from subliminal_patch.providers.embeddedsubtitles import \
+from subliminal_patch.providers.embeddedsubtitles import _MemoizedFFprobeVideoContainer
-    _MemoizedFFprobeVideoContainer
+from subliminal_patch.providers.embeddedsubtitles import EmbeddedSubtitlesProvider
-from subliminal_patch.providers.embeddedsubtitles import \
+from subliminal_patch.providers.embeddedsubtitles import (
-    EmbeddedSubtitlesProvider
+    _discard_possible_incomplete_subtitles,
+)
 from subzero.language import Language
 
 
@@ -249,3 +250,30 @@ def test_memoized(video_single_language, mocker):
             ]
             is not None
         )
+
+
+@pytest.mark.parametrize(
+    "number_of_frames,expected_len",
+    [((34, 811), 1), ((0, 0), 2), ((811, 34), 1), ((900, 1000), 2), ((0, 900), 1)],
+)
+def test_discard_possible_incomplete_subtitles(number_of_frames, expected_len):
+    subtitle_1 = FFprobeSubtitleStream(
+        {
+            "index": 1,
+            "codec_name": "subrip",
+            "codec_long_name": "SubRip subtitle",
+            "disposition": {},
+            "tags": {"language": "eng", "NUMBER_OF_FRAMES": number_of_frames[0]},
+        }
+    )
+    subtitle_2 = FFprobeSubtitleStream(
+        {
+            "index": 2,
+            "codec_name": "subrip",
+            "codec_long_name": "SubRip subtitle",
+            "disposition": {},
+            "tags": {"language": "eng", "NUMBER_OF_FRAMES": number_of_frames[1]},
+        }
+    )
+    new_list = _discard_possible_incomplete_subtitles([subtitle_1, subtitle_2])
+    assert len(new_list) == expected_len