def process_subtitle(self, subtitle: Subtitle) -> Sequence[Subtitle]:
"""Try to split a subtitle taking into account the post-processing
options.
:param subtitle:
:type subtitle: Subtitle
:rtype: Sequence[Subtitle]
"""
subtitle.start = datetime.timedelta(
seconds=subtitle.start.seconds,
microseconds=subtitle.start.microseconds + (self.milli * 1000),
)
subtitle.content = normalize_request_str(subtitle.content, False)
split_sub = _split_dialogue(subtitle)
if len(split_sub) == 2:
if self.postproc.remove_first:
logger.debug("Removing first quote: %s", split_sub)
return [split_sub[1]]
if self.postproc.remove_second:
logger.debug("Removing second quote %s", split_sub)
return [split_sub[0]]
return split_sub
def leftGap(self):
""""""
inlist = self.inlist
mingap = self.mingap
Left = self.Left
gaps = 0
overlap = 0
new_s = []
for FSUB in zip(inlist, inlist[1:]):
end_1 = mTime(FSUB[0].end)
start_1 = mTime(FSUB[1].start)
if start_1 < end_1: overlap += 1
gap = start_1 - end_1
if gap < mingap:
gaps += 1
new_end = DT.timedelta(milliseconds=(start_1 - Left))
new_s.append(
Subtitle(FSUB[0].index, FSUB[0].start, new_end,
FSUB[0].content))
else:
new_s.append(FSUB[0])
new_s.append(inlist[len(inlist) - 1])
return gaps, overlap, new_s
def make_subs(filename,sub_data):
filename = filename.replace('.mov','.srt')
print filename, sub_data
subtitles = []
for item in sorted(sub_data,key=int):
print item, sub_data[item]
seconds = int(item)
index = 1
start = timedelta(0,seconds,0)
end = timedelta(0,seconds+10,0)
content = sub_data[item]
s = Subtitle(index, start, end, content, proprietary='')
subtitles.append(s)
# print srt.compose(subtitles)
print len(subtitles)
file = open(path + filename, "wb")
file.write(srt.compose(subtitles).encode('utf-8'))
file.close()
print "output to:",path+filename
def transformBack(notez: Iterator[Message], is_lyrics: bool,
k_time: float) -> List[Subtitle]:
out = []
def read(ty, blanks=["set_tempo"]):
note = next(notez)
if note.type != ty:
if note.type == "end_of_track": raise StopIteration("EOT")
while note.type in blanks:
note = next(notez) #< jump off!
if note.type != ty:
raise ValueError(f"unexpected msg {note}, expecting {ty}")
return note
timeof = lambda n: timedelta(seconds=n / k_time)
t_acc = timedelta()
lyric = None
index = 1 #< for subtitles
while True:
try:
if is_lyrics: lyric = read("lyrics")
on = read("note_on")
t_on = timeof(lyric.time if is_lyrics else on.time)
t_off = timeof(read("note_off").time)
except StopIteration:
break #v pitch back
out.append(
Subtitle(index, t_acc + t_on, t_acc + t_on + t_off,
lyric.text if is_lyrics else str(on.note)))
t_acc += (t_on + t_off)
index += 1
return out
def replaceCurrent(self):
""""""
if self.Replaced:
text = self.text_3.GetValue()
sub = self.Replaced[0]
changed = Subtitle(sub.index, sub.start, sub.end, text)
self.text_2.AppendText(self.composeSub(changed))
self.Replaced.clear()
self.new_subs.append(changed)
self.default_subs = self.GetText()
def process_subtitle(self, subtitle: Subtitle) -> Sequence[Subtitle]:
logger.debug("Milliseconds value: %s", self.milli)
subtitle.start = datetime.timedelta(
seconds=subtitle.start.seconds,
microseconds=subtitle.start.microseconds + (self.milli * 1000),
)
logger.debug("New start: %s", subtitle.start)
subtitle.content = _normalize_request_str(subtitle.content, False)
split_sub = _split_dialogue(subtitle)
logger.debug("Result: %s", [item.content for item in split_sub])
if len(split_sub) == 2:
if self._args.get("remove_first"):
logger.debug("Removing first quote: %s", split_sub)
return [split_sub[1]]
if self._args.get("remove_second"):
logger.debug("Removing second quote %s", split_sub)
return [split_sub[0]]
return split_sub
def _guess_timestamps(
og_quote: Subtitle, quotes: Sequence[str]
) -> Tuple[Subtitle, Subtitle]:
"""Guess new timestamps in order to split dialogue.
:param og_quote:
:type og_quote: Subtitle
:param quotes:
:type quotes: Sequence[str]
"""
start_sec = og_quote.start.seconds
end_sec = og_quote.end.seconds
start_micro = og_quote.start.microseconds
end_micro = og_quote.end.microseconds
extra_secs = (start_micro * 0.000001) + (end_micro * 0.000001)
total_secs = end_sec - start_sec + extra_secs
new_time = list(_gen_quote_time(quotes, total_secs))
first_new = Subtitle(
index=og_quote.index,
start=datetime.timedelta(seconds=start_sec + 1, microseconds=0),
end=og_quote.end,
content=quotes[0],
)
index = first_new.index + 0
content = quotes[1]
start = datetime.timedelta(
seconds=new_time[0][0] + start_sec, microseconds=new_time[1][1]
)
end = datetime.timedelta(seconds=new_time[0][0] + start_sec + 1, microseconds=0)
second_new = Subtitle(index, start, end, content)
logger.debug("Result: %s %s", first_new, second_new)
return first_new, second_new
def create_subtitle(self):
subs = []
out_filename = self.search_term + '.srt'
for i, (text, time) in enumerate(
zip(self.content, list(self.clips_asset.values()))):
ti = timedelta(seconds=time['clipStart'])
te = timedelta(seconds=time['clipEnd'] - 1, microseconds=416000)
subs.append(
Subtitle(index=i + 1, start=ti, end=te, content=text['text']))
try:
with open(out_filename, 'w') as out_file:
out_file.write(compose(subs))
except Exception:
print(out_filename + ' not created.')
def merge_lines(inPar, inNepar):
re_pattern = re.compile(r'<[^<]*>')
new_j = []
for (
first,
second,
) in zip_longest(inNepar, inPar, fillvalue=dsub):
gap = mTime(second.start) - mTime(first.end)
trajanje = mTime(second.end) - mTime(first.start)
tekst1 = re.sub(re_pattern, '', first.content)
tekst2 = re.sub(re_pattern, '', second.content)
text_len = len(tekst1) + len(tekst2)
if gap <= _gap and trajanje <= max_time and text_len <= max_char:
if (first.content == second.content
and first.start == second.start
and first.end == second.end):
new_j.append(
Subtitle(first.index, first.start, second.end,
first.content))
else:
new_j.append(
Subtitle(first.index, first.start, second.end,
f"{first.content} {second.content}"))
else:
new_j.append(
Subtitle(first.index, first.start, first.end,
first.content))
new_j.append(
Subtitle(second.index, second.start, second.end,
second.content))
parni = new_j[1::2]
neparni = new_j[0::2]
return new_j, parni, neparni
def convert_to_srt(json_sub: str):
print("converting to srt file...")
srt_list = []
for i in range(len(json_sub)):
cur = Subtitle(
index=i + 1,
start=timedelta(seconds=json_sub[i]['start']),
end=timedelta(seconds=json_sub[i]['start'] +
json_sub[i]['duration'] if i +
1 == len(json_sub) else min(
json_sub[i + 1]['start'], json_sub[i]['start'] +
json_sub[i]['duration'])),
content=json_sub[i]['text'])
srt_list.append(cur)
return compose(srt_list)
...
def rightGap(self, in_list):
""""""
inlist = in_list
mingap = self.mingap
Right = self.Right
new_f = []
new_f.append(inlist[0])
for FSUB in zip(inlist, inlist[1:]):
end_1 = mTime(FSUB[0].end)
start_1 = mTime(FSUB[1].start)
gap = start_1 - end_1
if gap < mingap:
new_start = DT.timedelta(milliseconds=(start_1 + Right))
new_f.append(
Subtitle(FSUB[1].index, new_start, FSUB[1].end,
FSUB[1].content))
else:
new_f.append(FSUB[1])
return new_f
def getValues(self, iterator):
""""""
c = 0
wdict = self.wdict
r1 = re.compile(r"\b(" + "|".join(map(re.escape, wdict.keys())) +
r")\b")
try:
sub = next(iterator)
c += 1
except StopIteration:
logger.debug("Iterator was empty")
finally:
p = "=" * 20
self.text_3.SetValue(f"{p}\nEnd of subtitles reached\n{p}")
try:
t1 = list(set(r1.findall(sub.content)))
newd = {}
self.text_1.Clear()
for i in range(len(t1)):
self.text_1.AppendText(f"{t1[i]} ")
v = wdict[t1[i]]
newd[t1[i]] = v
for k, v in newd.items():
ctext = re.compile(r'\b' + k + r'\b')
sub.content = ctext.sub(v, sub.content)
self.text_3.SetValue(sub.content)
self.text_3.SetFocus()
for v in newd.values():
self.textStyle(self.text_3, sub.content, "RED", "", v)
if t1:
changed = Subtitle(sub.index, sub.start, sub.end, sub.content)
self.Replaced.append(changed)
for v in newd.values():
self.ReplacedAll.append(v)
self.new_d = newd
else:
self.text_2.AppendText(self.composeSub(sub))
return c
except Exception as e:
logger.debug(f"Error: {e}")
def finish(self):
td = lambda s: timedelta(seconds=s)
return compose([
Subtitle(i + 1, td(p[0]), td(p[1]), str(p[2]))
for (i, p) in enumerate(self.items)
])
Using mainly spoons, 3 00:31:41,933 --> 00:31:43,435 we dig a tunnel under the city and release it into the wild. ''') subtitles = list(subtitle_generator) # print subtitles[0].start # datetime.timedelta(0, 1897, 894000) # print subtitles[1].content # 'Using mainly spoons,' # pprint(subtitles) subtitles = [] # s = Subtitle(index=1,content="text",start=datetime(0,1,0),end=datetime(0,2,0),proprietary='') index = 1 start = timedelta(0, 1, 0) end = timedelta(0, 2, 0) content = 'text' s = Subtitle(index, start, end, content, proprietary='') print s print srt.compose([s, s, s, s, s])
def intoSrt(srts, sep=None):
def newContent(line):
words = [srt.content for srt in line]
return cfgOrDefault(sep, sepDeft, words).join(words)
time = lambda it: it.start
return [Subtitle(i+1, min(line, key=time).start, max(line, key=time).end, newContent(line)) for (i, line) in enumerate(srts)]
def fromLrc(text, min_len): td = lambda t: timedelta(seconds=t) return [Subtitle(i+1, td(t), td(t+min_len), s) for i, (t, s) in enumerate(readLrc(text))]
srt_str = ''
with open('image_classes.yml', 'r') as f:
class_to_name = yaml.load(f)
start_time = 0
n_valid = 1
for i in range(len(gradm)):
sec = i * frame_time
cl_ = class_frames[i]
test = last_cat is None
if cl_ != last_cat:
last_cat_end = sec
srt_block = Subtitle(n_valid, pd.to_timedelta(start_time,
unit='s'),
pd.to_timedelta(last_cat_end, unit='s'),
str([class_to_name[x]
for x in last_cat])).to_srt()
srt_str += srt_block
start_time = sec
n_valid += 1
last_cat = cl_
with open('test_upper.srt', 'w') as f:
f.write(srt_str)
lyric_dfs[0]['end'] = pd.to_timedelta('00:' + lyric_dfs[0]['end'])
with open('test_lower.srt', 'w') as f:
for i, row in lyric_dfs[0].iterrows():
if pd.isnull(row['end']):
def myMerger(subs_in, max_time, max_char, _gap):
subs = subs_in
dsub = Subtitle(
subs[-1].index + 1,
DT.timedelta(milliseconds=(mTime(subs[-1].start) + 6000)),
DT.timedelta(milliseconds=(mTime(subs[-1].end) + 11000)),
'Darkstar test appliance',
)
parni = subs[1::2]
neparni = subs[0::2]
def merge_lines(inPar, inNepar):
re_pattern = re.compile(r'<[^<]*>')
new_j = []
for (
first,
second,
) in zip_longest(inNepar, inPar, fillvalue=dsub):
gap = mTime(second.start) - mTime(first.end)
trajanje = mTime(second.end) - mTime(first.start)
tekst1 = re.sub(re_pattern, '', first.content)
tekst2 = re.sub(re_pattern, '', second.content)
text_len = len(tekst1) + len(tekst2)
if gap <= _gap and trajanje <= max_time and text_len <= max_char:
if (first.content == second.content
and first.start == second.start
and first.end == second.end):
new_j.append(
Subtitle(first.index, first.start, second.end,
first.content))
else:
new_j.append(
Subtitle(first.index, first.start, second.end,
f"{first.content} {second.content}"))
else:
new_j.append(
Subtitle(first.index, first.start, first.end,
first.content))
new_j.append(
Subtitle(second.index, second.start, second.end,
second.content))
parni = new_j[1::2]
neparni = new_j[0::2]
return new_j, parni, neparni
out_f, par1, nep1 = merge_lines(parni, neparni)
out_f, par2, nep2 = merge_lines(par1, nep1)
out_f, par3, nep3 = merge_lines(par2, nep2)
out_f, par4, nep4 = merge_lines(par3, nep3)
for i in out_f[-4:]:
if i.content == "Darkstar test appliance": out_f.remove(i)
WORK_TEXT.truncate(0)
WORK_TEXT.write(srt.compose(out_f))
WORK_TEXT.seek(0)
