You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
bazarr/libs/fese/__init__.py

439 lines
13 KiB

# -*- coding: utf-8 -*-
# License: GPL
from __future__ import annotations
import json
import logging
import os
import re
import subprocess
from typing import List, Optional
from babelfish import Language
from babelfish.exceptions import LanguageError
import pysubs2
__version__ = "0.1.3"
logger = logging.getLogger(__name__)
# Paths to executables
FFPROBE_PATH = os.environ.get("FFPROBE_PATH", "ffprobe")
FFMPEG_PATH = os.environ.get("FFMPEG_PATH", "ffmpeg")
FFMPEG_STATS = True
FF_LOG_LEVEL = "quiet"
class FeseError(Exception):
pass
class ExtractionError(FeseError):
pass
class InvalidFile(FeseError):
pass
class InvalidStream(FeseError):
pass
class InvalidSource(FeseError):
pass
class ConversionError(FeseError):
pass
class LanguageNotFound(FeseError):
pass
# Extensions
SRT = "srt"
ASS = "ass"
class FFprobeSubtitleDisposition:
def __init__(self, data: dict):
self.default = False
self.generic = False
self.dub = False
self.original = False
self.comment = False
self.lyrics = False
self.karaoke = False
self.forced = False
self.hearing_impaired = False
self.visual_impaired = False
self.clean_effects = False
self.attached_pic = False
self.timed_thumbnails = False
self._content_type = None
for key, val in data.items():
if hasattr(self, key):
setattr(self, key, bool(val))
def update_from_tags(self, tags):
tag_title = tags.get("title")
if tag_title is None:
logger.debug("Title not found. Marking as generic")
self.generic = True
return None
l_tag_title = tag_title.lower()
for key, val in _content_types.items():
if val.search(l_tag_title) is not None:
logger.debug("Found %s: %s", key, l_tag_title)
self._content_type = key
setattr(self, key, True)
return None
logger.debug("Generic disposition title found: %s", l_tag_title)
self.generic = True
return None
@property
def suffix(self):
if self._content_type is not None:
return f"-{self._content_type}"
return ""
def __str__(self):
return self.suffix.lstrip("-").upper() or "GENERIC"
class FFprobeSubtitleStream:
"""Base class for FFprobe (FFmpeg) extractable subtitle streams."""
def __init__(self, stream: dict):
"""
:raises: LanguageNotFound
"""
self.index = int(stream.get("index", 0))
self.codec_name = stream.get("codec_name", "Unknown")
self.extension = _subtitle_extensions.get(self.codec_name, self.codec_name)
self.r_frame_rate = stream.get("r_frame_rate")
self.avg_frame_rate = stream.get("avg_frame_rate")
self.time_base = stream.get("time_base")
self.tags = stream.get("tags", {})
self.start_time = float(stream.get("start_time", 0))
self.duration, self.duration_ts = 0, 0
# some subtitles streams miss the duration_ts field and only have tags->DURATION field
# fixme: we still don't know if "DURATION" is a common tag/key
if "DURATION" in self.tags:
try:
h, m, s = [
ts.replace(",", ".").strip()
for ts in self.tags["DURATION"].split(":")
]
self.duration = float(s) + float(m) * 60 + float(h) * 60 * 60
self.duration_ts = int(self.duration * 1000)
except ValueError as error:
logger.warning("Couldn't get duration field: %s. Using 0", error)
else:
try:
self.duration = float(stream.get("duration", "0").replace(",", "."))
self.duration_ts = int(stream.get("duration_ts", self.duration * 1000))
# some subtitles streams miss a duration completely and has "N/A" as value
except ValueError as error:
logger.warning("Couldn't get duration field: %s. Using 0", error)
self.start_pts = int(stream.get("start_pts", 0))
self.disposition = FFprobeSubtitleDisposition(stream.get("disposition", {}))
if self.tags:
self.disposition.update_from_tags(self.tags)
self.language: Language = self._language()
@property
def suffix(self):
lang = self.language.alpha2
if self.language.country is not None:
lang = f"{lang}-{self.language.country}"
return f"{lang}{self.disposition.suffix}.{self.extension}"
def _language(self) -> Language:
og_lang = self.tags.get("language")
last_exc = None
if og_lang is not None:
if og_lang in _extra_languages:
extra = _extra_languages[og_lang]
title = self.tags.get("title", "n/a").lower()
if any(possible in title for possible in extra["matches"]):
logger.debug("Found extra language %s", extra["language_args"])
return Language(*extra["language_args"])
try:
lang = Language.fromalpha3b(og_lang)
# Test for suffix
assert lang.alpha2
return lang
except LanguageError as error:
last_exc = error
logger.debug("Error with '%s' language: %s", og_lang, error)
raise LanguageNotFound(
f"Couldn't detect language for stream: {self.tags}"
) from last_exc
def __repr__(self) -> str:
return f"<{self.codec_name.upper()}: {self.language}@{self.disposition}>"
class FFprobeVideoContainer:
def __init__(self, path: str):
self.path = path
@property
def extension(self):
return os.path.splitext(self.path)[-1].lstrip(".")
def get_subtitles(self, timeout: int = 600) -> List[FFprobeSubtitleStream]:
"""Factory function to create subtitle instances from FFprobe.
:param timeout: subprocess timeout in seconds (default: 600)
:raises: InvalidSource"""
ff_command = [
FFPROBE_PATH,
"-v",
FF_LOG_LEVEL,
"-print_format",
"json",
"-show_format",
"-show_streams",
self.path,
]
try:
result = subprocess.run(
ff_command, stdout=subprocess.PIPE, check=True, timeout=timeout
)
streams = json.loads(result.stdout)["streams"]
except _ffprobe_exceptions as error:
raise InvalidSource(
f"{error} trying to get information from {self.path}"
) from error # We want to see the traceback
subs = []
for stream in streams:
if stream.get("codec_type", "n/a") != "subtitle":
continue
try:
subs.append(FFprobeSubtitleStream(stream))
except LanguageNotFound:
pass
if not subs:
logger.debug("Source doesn't have any subtitle valid streams")
return []
logger.debug("Found subtitle streams: %s", subs)
return subs
def extract_subtitles(
self,
subtitles: List[FFprobeSubtitleStream],
custom_dir=None,
overwrite=True,
timeout=600,
):
"""Extracts a list of subtitles. Returns a dictionary of the extracted
filenames by index.
:param subtitles: a list of FFprobeSubtitle instances
:param custom_dir: a custom directory to save the subtitles. Defaults to
same directory as the media file
:param overwrite: overwrite files with the same name (default: True)
:param timeout: subprocess timeout in seconds (default: 600)
:raises: ExtractionError, OSError
"""
extract_command = [FFMPEG_PATH, "-v", FF_LOG_LEVEL]
if FFMPEG_STATS:
extract_command.append("-stats")
extract_command.extend(["-y", "-i", self.path])
if custom_dir is not None:
# May raise OSError
os.makedirs(custom_dir, exist_ok=True)
items = {}
collected_paths = set()
for subtitle in subtitles:
sub_path = f"{os.path.splitext(self.path)[0]}.{subtitle.suffix}"
if custom_dir is not None:
sub_path = os.path.join(custom_dir, os.path.basename(sub_path))
if sub_path in collected_paths:
sub_path = (
f"{sub_path.rstrip(f'.{subtitle.suffix}')}"
f"-{len(collected_paths)}.{subtitle.suffix}"
)
if not overwrite and os.path.isfile(sub_path):
logger.debug("Ignoring path (OVERWRITE TRUE): %s", sub_path)
continue
extract_command.extend(
["-map", f"0:{subtitle.index}", "-c", "copy", sub_path]
)
logger.debug("Appending subtitle path: %s", sub_path)
collected_paths.add(sub_path)
items[subtitle.index] = sub_path
if not items:
logger.debug("No subtitles to extract")
return {}
logger.debug("Extracting subtitle with command %s", " ".join(extract_command))
try:
subprocess.run(extract_command, timeout=timeout, check=True)
except (subprocess.SubprocessError, FileNotFoundError) as error:
raise ExtractionError(f"Error calling ffmpeg: {error}") from error
for path in items.values():
if not os.path.isfile(path):
logger.debug("%s was not extracted", path)
return items
def __repr__(self) -> str:
return f"<FFprobeVideoContainer {self.extension}: {self.path}>"
def check_integrity(
subtitle: FFprobeSubtitleStream, path: str, sec_offset_threshold=900
):
"""A relative check for the integriy of a file. This can be used to find a failed
ffmpeg extraction where the final file might not be complete or might be corrupted.
Currently, only ASS and Subrip are supported.
:param subtitle: FFprobeSubtitle instance
:param path: the path of the subtitle file (ass or srt)
:param sec_offset_threshold: the maximum seconds offset to determine if the file is complete
:raises: InvalidFile
"""
if subtitle.extension not in (ASS, SRT):
raise InvalidFile(f"Extension not supported: {subtitle.extension}")
try:
sub = pysubs2.load(path)
except (pysubs2.Pysubs2Error, UnicodeError, OSError, FileNotFoundError) as error:
raise InvalidFile(error) from error
else:
# ignore the duration check if the stream has no duration listed at all
if subtitle.duration_ts:
off = abs(int(sub[-1].end) - subtitle.duration_ts)
if off > abs(sec_offset_threshold) * 1000:
raise InvalidFile(
f"The last subtitle timestamp ({sub[-1].end/1000} sec) is {off/1000} sec ahead"
f" from the subtitle stream total duration ({subtitle.duration} sec)"
)
logger.debug("Integrity check passed (%d sec offset)", off / 1000)
else:
logger.warning(
"Ignoring duration check, subtitle stream has bad duration values: %s",
subtitle,
)
def to_srt(
source: str, output: Optional[str] = None, remove_source: bool = False
) -> str:
"""Convert a subtitle to SubRip. Currently, only ASS is supported. SubRip
files will be silently ignored.
raises: ConversionError, OSError"""
if source.endswith(".srt"):
return source
split_path = os.path.splitext(source)
if split_path[-1] not in (".ass"):
raise ConversionError(
f"No converter found for extension: {split_path[-1]}"
) from None
output = output or f"{split_path[0]}.srt"
try:
parsed = pysubs2.load(source)
parsed.save(output)
except (pysubs2.Pysubs2Error, UnicodeError) as error:
raise ConversionError(f"Exception converting {output}: {error}") from error
logger.debug("Converted: %s", output)
if remove_source and source != output:
try:
os.remove(source)
except OSError as error:
logger.debug("Can't remove source: %s (%s)", source, error)
return output
_subtitle_extensions = {
"subrip": "srt",
"ass": "ass",
"hdmv_pgs_subtitle": "sup",
"pgs": "sup",
}
_content_types = {
"hearing_impaired": re.compile(r"sdh|hearing impaired"),
"forced": re.compile(r"forced"),
"comment": re.compile(r"comment"),
"visual_impaired": re.compile(r"signs|visual impair"),
"karaoke": re.compile(r"karaoke|songs"),
}
_ffprobe_exceptions = (
subprocess.SubprocessError,
json.JSONDecodeError,
FileNotFoundError,
KeyError,
)
_extra_languages = {
"spa": {
"matches": (
"es-la",
"spa-la",
"spl",
"mx",
"latin",
"mexic",
"argent",
"latam",
),
"language_args": ("spa", "MX"),
},
"por": {
"matches": ("pt-br", "pob", "pb", "brazilian", "brasil", "brazil"),
"language_args": ("por", "BR"),
},
}