import io
import re
-from .compat import compat_Match, compat_Pattern
from .utils import int_or_none, timetuple_from_msec
self._pos = 0
def match(self, r):
- if isinstance(r, compat_Pattern):
+ if isinstance(r, re.Pattern):
return r.match(self._data, self._pos)
if isinstance(r, str):
if self._data.startswith(r, self._pos):
def advance(self, by):
if by is None:
amt = 0
- elif isinstance(by, compat_Match):
+ elif isinstance(by, re.Match):
amt = len(by.group(0))
elif isinstance(by, str):
amt = len(by)
class ParseError(Exception):
def __init__(self, parser):
super().__init__("Parse error at position %u (near %r)" % (
- parser._pos, parser._data[parser._pos:parser._pos + 20]
+ parser._pos, parser._data[parser._pos:parser._pos + 100]
))
([0-9]{3})?
''')
_REGEX_EOF = re.compile(r'\Z')
-_REGEX_NL = re.compile(r'(?:\r\n|[\r\n])')
+_REGEX_NL = re.compile(r'(?:\r\n|[\r\n]|$)')
_REGEX_BLANK = re.compile(r'(?:\r\n|[\r\n])+')
+_REGEX_OPTIONAL_WHITESPACE = re.compile(r'[ \t]*')
def _parse_ts(ts):
Convert a parsed WebVTT timestamp (a re.Match obtained from _REGEX_TS)
into an MPEG PES timestamp: a tick counter at 90 kHz resolution.
"""
-
- h, min, s, ms = ts.groups()
- return 90 * (
- int(h or 0) * 3600000 + # noqa: W504,E221,E222
- int(min) * 60000 + # noqa: W504,E221,E222
- int(s) * 1000 + # noqa: W504,E221,E222
- int(ms) # noqa: W504,E221,E222
- )
+ return 90 * sum(
+ int(part or 0) * mult for part, mult in zip(ts.groups(), (3600_000, 60_000, 1000, 1)))
def _format_ts(ts):
A WebVTT block that may only appear in the header part of the file,
i.e. before any cue blocks.
"""
-
pass
_REGEX_TSMAP_MPEGTS = re.compile(r'MPEGTS:([0-9]+)')
_REGEX_TSMAP_SEP = re.compile(r'[ \t]*,[ \t]*')
+ # This was removed from the spec in the 2017 revision;
+ # the last spec draft to describe this syntax element is
+ # <https://www.w3.org/TR/2015/WD-webvtt1-20151208/#webvtt-metadata-header>.
+ # Nevertheless, YouTube keeps serving those
+ _REGEX_META = re.compile(r'(?:(?!-->)[^\r\n])+:(?:(?!-->)[^\r\n])+(?:\r\n|[\r\n])')
+
@classmethod
def __parse_tsmap(cls, parser):
parser = parser.child()
raise ParseError(parser)
extra = m.group(1)
- local, mpegts = None, None
- if parser.consume(cls._REGEX_TSMAP):
- local, mpegts = cls.__parse_tsmap(parser)
- if not parser.consume(_REGEX_NL):
+ local, mpegts, meta = None, None, ''
+ while not parser.consume(_REGEX_NL):
+ if parser.consume(cls._REGEX_TSMAP):
+ local, mpegts = cls.__parse_tsmap(parser)
+ continue
+ m = parser.consume(cls._REGEX_META)
+ if m:
+ meta += m.group(0)
+ continue
raise ParseError(parser)
parser.commit()
- return cls(extra=extra, mpegts=mpegts, local=local)
+ return cls(extra=extra, mpegts=mpegts, local=local, meta=meta)
def write_into(self, stream):
stream.write('WEBVTT')
stream.write(',MPEGTS:')
stream.write(str(self.mpegts if self.mpegts is not None else 0))
stream.write('\n')
+ if self.meta:
+ stream.write(self.meta)
stream.write('\n')
if not m1:
return None
m2 = parser.consume(cls._REGEX_SETTINGS)
+ parser.consume(_REGEX_OPTIONAL_WHITESPACE)
if not parser.consume(_REGEX_NL):
return None
a bytes object containing the raw contents of a WebVTT file.
"""
- parser = _MatchParser(frag_content.decode('utf-8'))
+ parser = _MatchParser(frag_content.decode())
yield Magic.parse(parser)