summaryrefslogtreecommitdiffhomepage
diff options
context:
space:
mode:
authorvitiko98 <[email protected]>2022-09-02 23:35:56 -0400
committervitiko98 <[email protected]>2022-09-02 23:37:17 -0400
commit6e521143e1de1f720dd9cf908aa7c907696187f0 (patch)
tree5709e69b4590daeb274a4255bf679244d4db9ddc
parent0bc05232ed69be8dd727ce70e24e68e9d7362367 (diff)
downloadbazarr-6e521143e1de1f720dd9cf908aa7c907696187f0.tar.gz
bazarr-6e521143e1de1f720dd9cf908aa7c907696187f0.zip
Embedded Subtitles provider: improve ASS subtitles filtering
-rw-r--r--libs/subliminal_patch/providers/embeddedsubtitles.py39
-rw-r--r--tests/subliminal_patch/test_embeddedsubtitles.py15
2 files changed, 53 insertions, 1 deletions
diff --git a/libs/subliminal_patch/providers/embeddedsubtitles.py b/libs/subliminal_patch/providers/embeddedsubtitles.py
index e253dd310..38089070d 100644
--- a/libs/subliminal_patch/providers/embeddedsubtitles.py
+++ b/libs/subliminal_patch/providers/embeddedsubtitles.py
@@ -3,6 +3,7 @@
import functools
import logging
import os
+import re
import shutil
import tempfile
@@ -182,8 +183,19 @@ class EmbeddedSubtitlesProvider(Provider):
"series" if isinstance(video, Episode) else "movie",
)
- def download_subtitle(self, subtitle):
+ def download_subtitle(self, subtitle: EmbeddedSubtitle):
path = self._get_subtitle_path(subtitle)
+
+ modifiers = _type_modifiers.get(subtitle.stream.codec_name)
+ logger.debug(
+ "Found modifiers for %s type: %s", subtitle.stream.codec_name, modifiers
+ )
+
+ if modifiers is not None:
+ for mod in modifiers:
+ logger.debug("Running %s modifier for %s", mod, path)
+ mod(path, path)
+
with open(path, "rb") as sub:
content = sub.read()
subtitle.content = fix_line_ending(content)
@@ -303,3 +315,28 @@ def _discard_possible_incomplete_subtitles(streams):
def _get_pretty_release_name(stream, container):
bname = os.path.basename(container.path)
return f"{os.path.splitext(bname)[0]}.{stream.suffix}"
+
+
+# TODO: improve this
+_SIGNS_LINE_RE = re.compile(r",([\w|_]{,15}(sign|fx|karaoke))", flags=re.IGNORECASE)
+
+
+def _clean_ass_subtitles(path, output_path):
+ """An attempt to ignore extraneous lines from ASS anime subtitles. Experimental."""
+
+ clean_lines = []
+
+ with open(path, "r") as f:
+ lines = f.readlines()
+ for line in lines:
+ if _SIGNS_LINE_RE.search(line) is None:
+ clean_lines.append(line)
+
+ logger.debug("Cleaned lines: %d", abs(len(lines) - len(clean_lines)))
+
+ with open(output_path, "w") as f:
+ f.writelines(clean_lines)
+ logger.debug("Lines written to output path: %s", output_path)
+
+
+_type_modifiers = {"ass": {_clean_ass_subtitles}}
diff --git a/tests/subliminal_patch/test_embeddedsubtitles.py b/tests/subliminal_patch/test_embeddedsubtitles.py
index d858581f8..ab0dc1c41 100644
--- a/tests/subliminal_patch/test_embeddedsubtitles.py
+++ b/tests/subliminal_patch/test_embeddedsubtitles.py
@@ -10,6 +10,7 @@ from subliminal_patch.core import Episode
from subliminal_patch.core import Movie
from subliminal_patch.providers.embeddedsubtitles import (
_discard_possible_incomplete_subtitles,
+ _clean_ass_subtitles,
)
from subliminal_patch.providers.embeddedsubtitles import _get_pretty_release_name
from subliminal_patch.providers.embeddedsubtitles import _MemoizedFFprobeVideoContainer
@@ -283,6 +284,20 @@ def test_get_pretty_release_name():
assert _get_pretty_release_name(stream, container) == "foo.en.forced.srt"
+def test_clean_ass_subtitles(data, tmp_path):
+ path = os.path.join(data, "subs.ass")
+
+ with open(path, "r") as f:
+ og_lines_len = len(f.readlines())
+
+ output_path = os.path.join(tmp_path, "subs.ass")
+
+ _clean_ass_subtitles(path, output_path)
+
+ with open(output_path, "r") as f:
+ assert og_lines_len > len(f.readlines())
+
+
def test_download_subtitle_multiple(video_multiple_languages):
with EmbeddedSubtitlesProvider() as provider:
languages = {Language.fromalpha2(code) for code in ("en", "it", "fr")} | {