def _html_to_json(self, view_html, lecture_id):
data = parse_json(
search_regex(
r'videojs-setup-data=(["\'])(?P<data>{.+?})\1',
view_html,
"setup data",
default="{}",
group="data",
),
lecture_id,
transform_source=unescapeHTML,
fatal=False,
)
text_tracks = parse_json(
search_regex(
r'text-tracks=(["\'])(?P<data>\[.+?\])\1',
view_html,
"text tracks",
default="{}",
group="data",
),
lecture_id,
transform_source=lambda s: js_to_json(unescapeHTML(s)),
fatal=False,
)
return data, text_tracks
def convert(self, filename=None, keep_vtt=False):
if filename:
seq = 1
fname = filename.replace(".vtt", ".srt")
content = self._vttcontents(fname=filename)
if content and isinstance(content, list):
timecode_loc = self._locate_timecode(content)
if not timecode_loc.get("status"):
return {
"status": "False",
"msg": "subtitle file seems to have malfunction skipping conversion ..",
}
for line in content[timecode_loc.get("location") :]:
flag = self._is_timecode(timecode=line)
if flag:
timecode = self._generate_timecode(seq, unescapeHTML(line))
self._write_srtcontent(fname, timecode)
seq += 1
if not flag:
match = re.match("^([0-9]{1,3})$", line, flags=re.U)
if not match:
data = "{content}\r\n".format(content=line)
self._write_srtcontent(fname, data)
else:
return content
if not keep_vtt:
try:
os.unlink(filename)
except Exception: # pylint: disable=W
pass
return {"status": "True", "msg": "successfully generated subtitle in srt ..."}