Add Embedded Subtitles provider

README.md

@@ -46,6 +46,7 @@ If you need something that is not already part of Bazarr, feel free to create a
 * Assrt
 * BetaSeries
 * BSplayer
+* Embedded Subtitles
 * GreekSubtitles
 * Hosszupuska
 * LegendasDivx

bazarr/config.py

@@ -190,6 +190,9 @@ def get(self, section, option, raw=False, vars=None):
         'approved_only': 'False',
         'multithreading': 'True'
     },
+    'embeddedsubtitles': {
+        'include_ass': 'True',
+    },
     'subsync': {
         'use_subsync': 'False',
         'use_subsync_threshold': 'False',

bazarr/get_providers.py

@@ -13,6 +13,7 @@
 from get_args import args
 from config import settings, get_array_from
 from event_handler import event_stream
+from utils import get_binary
 from subliminal_patch.exceptions import TooManyRequests, APIThrottled, ParseResponseError, IPAddressBlocked
 from subliminal.providers.opensubtitles import DownloadLimitReached
 from subliminal.exceptions import DownloadLimitExceeded, ServiceUnavailable
@@ -198,6 +199,12 @@ def get_providers_auth():
             'email': settings.ktuvit.email,
             'hashed_password': settings.ktuvit.hashed_password,
         },
+        'embeddedsubtitles': {
+            'include_ass': settings.embeddedsubtitles.getboolean('include_ass'),
+            'cache_dir': os.path.join(args.config_dir, "cache"),
+            'ffprobe_path': get_binary("ffprobe"),
+            'ffmpeg_path': get_binary("ffmpeg"),
+        }
     }
 
 

frontend/src/Settings/Providers/list.ts

@@ -46,6 +46,17 @@ export const ProviderList: Readonly<ProviderInfo[]> = [
     key: "bsplayer",
     name: "BSplayer",
   },
+  {
+    key: "embeddedsubtitles",
+    name: "Embedded Subtitles",
+    description: "Embedded Subtitles from your Media Files",
+    defaultKey: {
+      include_ass: true,
+    },
+    keyNameOverride: {
+      include_ass: "Convert embedded ASS to SRT",
+    },
+  },
   {
     key: "greeksubs",
     name: "GreekSubs",

libs/subliminal_patch/providers/embeddedsubtitles.py

@@ -0,0 +1,162 @@
+# -*- coding: utf-8 -*-
+
+import logging
+import os
+import shutil
+import tempfile
+
+from babelfish import language_converters
+import fese
+from fese import check_integrity
+from fese import FFprobeSubtitleStream
+from fese import FFprobeVideoContainer
+from fese import to_srt
+from subliminal.subtitle import fix_line_ending
+from subliminal_patch.core import Episode
+from subliminal_patch.core import Movie
+from subliminal_patch.providers import Provider
+from subliminal_patch.subtitle import Subtitle
+from subzero.language import Language
+
+logger = logging.getLogger(__name__)
+
+# Replace Babelfish's Language with Subzero's Language
+fese.Language = Language
+
+
+class EmbeddedSubtitle(Subtitle):
+    provider_name = "embeddedsubtitles"
+    hash_verifiable = False
+
+    def __init__(self, stream, container, matches):
+        super().__init__(stream.language, stream.disposition.hearing_impaired)
+        self.stream: FFprobeSubtitleStream = stream
+        self.container: FFprobeVideoContainer = container
+        self._matches: set = matches
+        self.page_link = self.container.path
+        self.release_info = os.path.basename(self.page_link)
+
+    def get_matches(self, video):
+        if self.hearing_impaired:
+            self._matches.add("hearing_impaired")
+
+        self._matches.add("hash")
+        return self._matches
+
+    @property
+    def id(self):
+        return f"{self.container.path}_{self.stream.index}"
+
+
+class EmbeddedSubtitlesProvider(Provider):
+    provider_name = "embeddedsubtitles"
+
+    languages = {Language("por", "BR"), Language("spa", "MX")} | {
+        Language.fromalpha2(l) for l in language_converters["alpha2"].codes
+    }
+    languages.update(set(Language.rebuild(lang, hi=True) for lang in languages))
+
+    # TODO: add forced support
+    # languages.update(set(Language.rebuild(lang, forced=True) for lang in languages))
+
+    video_types = (Episode, Movie)
+    subtitle_class = EmbeddedSubtitle
+
+    def __init__(
+        self, include_ass=True, cache_dir=None, ffprobe_path=None, ffmpeg_path=None
+    ):
+        self._include_ass = include_ass
+        self._cache_dir = os.path.join(
+            cache_dir or tempfile.gettempdir(), self.__class__.__name__.lower()
+        )
+        self._cached_paths = {}
+
+        fese.FFPROBE_PATH = ffprobe_path or fese.FFPROBE_PATH
+        fese.FFMPEG_PATH = ffmpeg_path or fese.FFMPEG_PATH
+
+        if logger.getEffectiveLevel() == logging.DEBUG:
+            fese.FF_LOG_LEVEL = "warning"
+        else:
+            # Default is True
+            fese.FFMPEG_STATS = False
+
+    def initialize(self):
+        os.makedirs(self._cache_dir, exist_ok=True)
+
+    def terminate(self):
+        # Remove leftovers
+        shutil.rmtree(self._cache_dir, ignore_errors=True)
+
+    def query(self, path: str, languages):
+        video = FFprobeVideoContainer(path)
+
+        try:
+            streams = video.get_subtitles()
+        except fese.InvalidSource as error:
+            logger.error("Error trying to get subtitles for %s: %s", video, error)
+            streams = []
+
+        if not streams:
+            logger.debug("No subtitles found for container: %s", video)
+
+        subtitles = []
+
+        for stream in streams:
+            # Only subrip and ass are currently supported
+            if stream.codec_name not in ("subrip", "ass"):
+                logger.debug("Ignoring codec: %s", stream)
+                continue
+
+            if not self._include_ass and stream.codec_name == "ass":
+                logger.debug("Ignoring ASS subtitle: %s", stream)
+                continue
+
+            if stream.language not in languages:
+                continue
+
+            disposition = stream.disposition
+            if disposition.generic or disposition.hearing_impaired:
+                logger.debug("Appending subtitle: %s", stream)
+                subtitles.append(EmbeddedSubtitle(stream, video, {"hash"}))
+            else:
+                logger.debug("Ignoring unwanted subtitle: %s", stream)
+
+        return subtitles
+
+    def list_subtitles(self, video, languages):
+        return self.query(video.name, languages)
+
+    def download_subtitle(self, subtitle):
+        path = self._get_subtitle_path(subtitle)
+        with open(path, "rb") as sub:
+            content = sub.read()
+            subtitle.content = fix_line_ending(content)
+
+    def _get_subtitle_path(self, subtitle: EmbeddedSubtitle):
+        container = subtitle.container
+
+        # Check if the container is not already in the instance
+        if container.path not in self._cached_paths:
+            # Extract all subittle streams to avoid reading the entire
+            # container over and over
+            streams = filter(_check_allowed_extensions, container.get_subtitles())
+            extracted = container.extract_subtitles(list(streams), self._cache_dir)
+            # Add the extracted paths to the containter path key
+            self._cached_paths[container.path] = extracted
+
+        cached_path = self._cached_paths[container.path]
+        # Get the subtitle file by index
+        subtitle_path = cached_path[subtitle.stream.index]
+
+        check_integrity(subtitle.stream, subtitle_path)
+
+        # Convert to SRT if the subtitle is ASS
+        new_subtitle_path = to_srt(subtitle_path, remove_source=True)
+        if new_subtitle_path != subtitle_path:
+            cached_path[subtitle.stream.index] = new_subtitle_path
+
+        return new_subtitle_path
+
+
+def _check_allowed_extensions(subtitle: FFprobeSubtitleStream):
+    return subtitle.extension in ("ass", "srt")

tests/subliminal_patch/test_embeddedsubtitles.py

@@ -0,0 +1,115 @@
+# -*- coding: utf-8 -*-
+import os
+
+import fese
+import pytest
+from subliminal_patch.core import Episode, Movie
+from subliminal_patch.providers.embeddedsubtitles import EmbeddedSubtitlesProvider
+from subzero.language import Language
+
+_DATA = os.path.join(os.path.abspath(os.path.dirname(__file__)), "data")
+
+
+fese.Language = Language
+
+
+@pytest.fixture
+def video_single_language():
+    # Has only ASS streams in english
+    return Episode(
+        os.path.join(_DATA, "file_1.mkv"),
+        "Serial Experiments Lain",
+        1,
+        1,
+        source="Web",
+    )
+
+
+@pytest.fixture
+def video_multiple_languages():
+    # Has SubRip streams in multiple languages
+    return Movie(
+        os.path.join(_DATA, "file_2.mkv"),
+        "I'm No Longer Here",
+        year=2019,
+        source="Web",
+    )
+
+
+@pytest.fixture
+def video_inexistent(tmpdir):
+    return Movie(
+        os.path.join(tmpdir, "inexistent_video.mkv"),
+        "Dummy",
+        year=2021,
+        source="Web",
+    )
+
+
+def test_inexistent_video(video_inexistent):
+    with EmbeddedSubtitlesProvider() as provider:
+        subtitles = provider.list_subtitles(video_inexistent, {})
+        assert len(subtitles) == 0
+
+
+def test_list_subtitles_single_language(video_single_language):
+    with EmbeddedSubtitlesProvider() as provider:
+        subs = provider.list_subtitles(
+            video_single_language, {Language.fromalpha2("en")}
+        )
+
+        for sub in subs:
+            assert sub.language == Language.fromalpha2("en")
+
+
+def test_list_subtitles_multiple_languages(video_multiple_languages):
+    with EmbeddedSubtitlesProvider() as provider:
+        languages = {Language.fromalpha2(code) for code in ("en", "it", "fr", "es")} | {
+            Language("por", "BR")
+        }
+
+        subs = provider.list_subtitles(video_multiple_languages, languages)
+        for expected in languages:
+            assert any(sub.language == expected for sub in subs)
+
+
+def test_list_subtitles_wo_ass(video_single_language):
+    with EmbeddedSubtitlesProvider(include_ass=False) as provider:
+        subs = provider.list_subtitles(
+            video_single_language, {Language.fromalpha2("en")}
+        )
+        assert not subs
+
+
+def test_download_subtitle_multiple(video_multiple_languages):
+    with EmbeddedSubtitlesProvider() as provider:
+        languages = {Language.fromalpha2(code) for code in ("en", "it", "fr")} | {
+            Language("por", "BR")
+        }
+
+        subs = provider.list_subtitles(video_multiple_languages, languages)
+        for sub in subs:
+            provider.download_subtitle(sub)
+            assert sub.content is not None
+
+
+def test_download_subtitle_single(video_single_language):
+    with EmbeddedSubtitlesProvider() as provider:
+        subtitle = provider.list_subtitles(
+            video_single_language, {Language.fromalpha2("en")}
+        )[0]
+        provider.download_subtitle(subtitle)
+        assert subtitle.content is not None
+
+
+def test_download_invalid_subtitle(video_single_language):
+    with EmbeddedSubtitlesProvider() as provider:
+        subtitle = provider.list_subtitles(
+            video_single_language, {Language.fromalpha2("en")}
+        )[0]
+
+        provider._cached_paths[subtitle.container.path] = {
+            subtitle.stream.index: "dummy.srt"
+        }
+        with pytest.raises(fese.InvalidFile):
+            provider.download_subtitle(subtitle)