def test_keep_unknown_html_tags():
# see issue #26
text = dedent("""\
1
00:00:10,500 --> 00:00:13,000
<i>Elephant's <sub>Little</sub> Dream</i>
2
00:00:15,000 --> 00:00:18,000
<font color="cyan">At the left we can see...</font>
""")
ref_default = SSAFile()
ref_default.append(SSAEvent(start=make_time(s=10.5), end=make_time(s=13), text="{\\i1}Elephant's Little Dream{\\i0}"))
ref_default.append(SSAEvent(start=make_time(s=15), end=make_time(s=18), text="At the left we can see..."))
ref_keep = SSAFile()
ref_keep.append(SSAEvent(start=make_time(s=10.5), end=make_time(s=13), text="{\\i1}Elephant's <sub>Little</sub> Dream{\\i0}"))
ref_keep.append(SSAEvent(start=make_time(s=15), end=make_time(s=18), text="<font color=\"cyan\">At the left we can see...</font>"))
subs_default = SSAFile.from_string(text)
subs_keep = SSAFile.from_string(text, keep_unknown_html_tags=True)
assert subs_default.equals(ref_default)
assert subs_keep.equals(ref_keep)
assert subs_keep.to_string("srt") == ref_keep.to_string("srt")
def test_repr_comment():
ev = SSAEvent(start=make_time(m=1, s=30),
end=make_time(m=1, s=35),
text="Hello\\Nworld!")
ev.is_comment = True
ref = r"<SSAEvent type=Comment start=0:01:30 end=0:01:35 text='Hello\\Nworld!'>"
assert repr(ev) == ref
def test_simple_parsing():
test_input1 = "[123][456] Line 1"
subs1 = SSAFile.from_string(test_input1)
assert len(subs1) == 1
assert subs1[0] == SSAEvent(start=make_time(ms=12300),
end=make_time(ms=45600),
text="Line 1")
test_input2 = "[123][456] / Line 1| Line 2/2"
subs2 = SSAFile.from_string(test_input2)
assert len(subs2) == 1
assert subs2[0] == SSAEvent(start=make_time(ms=12300),
end=make_time(ms=45600),
text="{\i1}Line 1{\i0}\\NLine2/2")
test_input3 = dedent("""
[123][456] Line 1
[321][456] / Line 2| Line 3
(123)(456)This line should not be parsed
This line should also not be parsed
[789][1234] /Line 4""")
subs3 = SSAFile.from_string(test_input3)
assert len(subs3) == 3
assert subs3[0] == SSAEvent(start=make_time(ms=12300),
end=make_time(ms=45600),
text="Line 1")
assert subs3[1] == SSAEvent(start=make_time(ms=32100),
end=make_time(ms=45600),
text="{\i1}Line 2{\i0}\\NLine 3")
assert subs3[2] == SSAEvent(start=make_time(ms=78900),
end=make_time(ms=123400),
text="{\i1}Line 4{\i0}")
def create_ass_file(subtitle_blocks: List[SubtitleBlock], ass_file,
styles: Dict[str, StyleConfig]):
subs = SSAFile()
colors = [Color(255, 255, 255), Color(100, 100, 255), Color(255, 100, 100)]
for k, name in enumerate(subtitle_blocks[0].names):
my_style = subs.styles["Default"].copy()
my_style.primarycolor = colors[k]
my_style.fontsize = styles[name].fontsize
my_style.shadow = 0
subs.styles[name] = my_style
for sb in subtitle_blocks:
start, end = None, None
for name, text in sb.name_texts:
if len(text) > 0:
text = text.replace("_", " ")
if start is None:
start = sb.start
end = sb.end
sub_line = SSAEvent(
start=start,
end=end,
text=text,
)
sub_line.style = name
subs.append(sub_line)
else:
print(f"WARNING: got empty block! {name} ")
subs.save(ass_file)
def test_write_drawing():
# test for 7bde9a6c3a250cf0880a8a9fe31d1b6a69ff21a0
subs = SSAFile()
e1 = SSAEvent()
e1.start = 0
e1.end = 60000
e1.text = r"{\p1}m 0 0 l 100 0 100 100 0 100{\p0}test"
e2 = SSAEvent()
e2.start = 60000
e2.end = 120000
e2.text = "Subtitle number\\Ntwo."
subs.append(e1)
subs.append(e2)
ref = dedent("""\
1
00:01:00,000 --> 00:02:00,000
Subtitle number
two.
""")
text = subs.to_string("srt")
assert text.strip() == ref.strip()
def test_empty_subtitles():
# regression test for issue #11
text = dedent("""
392
00:29:27,46 --> 00:29:29,83
I'm Liza Minnelli..
393
00:00:00,00 --> 00:00:00,00
394
00:00:00,00 --> 00:00:00,00
""")
ref = SSAFile()
ref.append(
SSAEvent(start=make_time(m=29, s=27, ms=460),
end=make_time(m=29, s=29, ms=830),
text="I'm Liza Minnelli.."))
ref.append(SSAEvent(start=0, end=0, text=""))
ref.append(SSAEvent(start=0, end=0, text=""))
subs = SSAFile.from_string(text)
assert subs.equals(ref)
def test_read_position_styling():
"""position is ignored, italic is converted, color is ignored"""
text = dedent("""\
1
00:00:10,500 --> 00:00:13,000 X1:63 X2:223 Y1:43 Y2:58
<i>Elephant's Dream</i>
2
00:00:15,000 --> 00:00:18,000 X1:53 X2:303 Y1:438 Y2:453
<font color="cyan">At the left we can see...</font>
""")
ref = SSAFile()
ref.append(
SSAEvent(start=make_time(s=10.5),
end=make_time(s=13),
text="{\\i1}Elephant's Dream{\\i0}"))
ref.append(
SSAEvent(start=make_time(s=15),
end=make_time(s=18),
text="At the left we can see..."))
subs = SSAFile.from_string(text)
assert subs.equals(ref)
def test_read_complex():
# regression test for #30
text = dedent("""\
WEBVTT
X-TIMESTAMP-MAP=LOCAL:00:00:00.000,MPEGTS:0
00:50.099 --> 00:53.299 line:85% align:middle
Cuidem do seu grupo.
Cuidem de suas fileiras.
01:54.255 --> 01:55.455 line:85% align:middle
Parem!
01:58.155 --> 01:59.555 line:85% align:middle
E, parem!
""")
ref = SSAFile()
ref.append(
SSAEvent(start=make_time(s=50, ms=99),
end=make_time(s=53, ms=299),
text=r"Cuidem do seu grupo.\NCuidem de suas fileiras."))
ref.append(
SSAEvent(start=make_time(m=1, s=54, ms=255),
end=make_time(m=1, s=55, ms=455),
text="Parem!"))
ref.append(
SSAEvent(start=make_time(m=1, s=58, ms=155),
end=make_time(m=1, s=59, ms=555),
text="E, parem!"))
subs = SSAFile.from_string(text)
assert subs.equals(ref)
def test_repr_simple():
subs = SSAFile()
subs.append(SSAEvent(start=make_time(m=5), end=make_time(m=6)))
subs.append(SSAEvent(start=make_time(m=125), end=make_time(m=126)))
subs.append(SSAEvent(start=make_time(m=15), end=make_time(m=16)))
subs.styles["style1"] = SSAStyle()
subs.styles["style2"] = SSAStyle()
ref = "<SSAFile with 3 events and 3 styles, last timestamp 2:06:00>"
assert repr(subs) == ref
def build_ref():
subs = SSAFile()
subs.info["My Custom Info"] = "Some: Test, String."
subs.styles["left"] = SSAStyle(alignment=7, bold=True)
subs.styles["topleft"] = SSAStyle(alignment=4)
subs.append(SSAEvent(start=0, end=make_time(m=1), text="An, example, subtitle."))
subs.append(SSAEvent(start=0, end=make_time(m=1), type="Comment", text="You can't see this one."))
subs.append(SSAEvent(start=make_time(m=1), end=make_time(m=2), text="Subtitle number\\Ntwo."))
return subs
def save_to_subtitles(results, f):
subs = create_ssa_file()
for result in results:
event = SSAEvent(start=make_time(s=result['start']),
end=make_time(s=result['end']),
text=f(result))
if 'highlight' in result and result['highlight']:
event.style = 'red'
subs.append(event)
logger.info(f'Processed {len(results)} results to subtitle events')
return subs
def test_plaintext():
e = SSAEvent(text=r"First\NSecond\NThird\hline{with hidden text}")
assert e.plaintext == "First\nSecond\nThird line"
e.plaintext = "My\n Text "
assert e.text == r"My\N Text "
# SubStation has no way to escape braces, thus this wart
text = "My text{with braces}"
e.plaintext = text
assert e.plaintext != text
def test_plaintext():
e = SSAEvent(text=r"First\NSecond\NThird\hline{with hidden text}")
assert e.plaintext == "First\nSecond\nThird line"
e.plaintext = "My\n Text "
assert e.text == r"My\N Text "
# SubStation has no way to escape braces, thus this wart
text = "My text{with braces}"
e.plaintext = text
assert e.plaintext != text
def test_simple_read():
text = dedent("""\
00:00:00:ten--chars
00:01:00:ten--chars-ten-chars
""")
#calculate endtime from starttime + 500 miliseconds + 67 miliseconds per each character (15 chars per second)
ref = SSAFile()
ref.append(SSAEvent(start=0, end=make_time(ms=1840), text="ten--chars"))
ref.append(SSAEvent(start=make_time(m=1), end=make_time(ms=62510), text="ten--chars-ten-chars"))
subs = SSAFile.from_string(text)
assert subs.equals(ref)
def test_writer_skips_comment_lines():
subs = SSAFile()
subs.append(SSAEvent(start=0, end=10, text="Hello!"))
subs.append(SSAEvent(start=0, end=10, text="World!"))
subs[0].is_comment = True
f = dedent("""\
{0}{0}1000
{0}{10}World!
""")
assert subs.to_string("microdvd", fps=1000) == f
def test_write_drawing():
subs = SSAFile()
subs.append(
SSAEvent(start=0,
end=10,
text=r"{\p1}m 0 0 l 100 0 100 100 0 100{\p0}test"))
subs.append(SSAEvent(start=10, end=20, text="Let me tell you."))
f = dedent("""\
{0}{0}1000
{10}{20}Let me tell you.
""")
assert subs.to_string("microdvd", fps=1000) == f
def test_duration():
e = SSAEvent(start=0, end=10)
assert e.duration == 10
e.duration = 20
assert e.start == 0 and e.end == 20
e.duration = 5
assert e.start == 0 and e.end == 5
e.duration = 0
assert e.start == 0 and e.end == 0
with assert_raises(ValueError):
e.duration = -20
def test_write_read():
subs = SSAFile()
e1 = SSAEvent(text="Hello, world!")
e2 = SSAEvent(text="The other subtitle.\\NWith two lines.",
style="custom style")
s1 = SSAStyle(italic=True, primarycolor=Color(r=255, g=0, b=0, a=0))
subs.append(e1)
subs.append(e2)
subs.styles["custom style"] = s1
json_text = subs.to_string("json")
subs2 = SSAFile.from_string(json_text, "json")
assert subs2.equals(subs)
def build(self, **extra):
from pysubs2 import SSAFile, SSAEvent, Color # type: ignore[import]
millis = lambda td: td / timedelta(milliseconds=1)
subs = (
SSAEvent(
start=millis(t - self.start),
end =millis(t - self.start + timedelta(seconds=length)),
text=text.replace('\n', r'\N'), # \N necessary for SSA files
) for t, text, length in self.l
)
sf = SSAFile()
style = sf.styles['Default'].copy()
style.fontsize = 16 # default is 20, bit too much??
style.outlinecolor = Color(0, 0, 0, 50) # semitransparent
style.shadow = 0.0
style.outline = 0.1
style.borderstyle = 3 # no idea why 3, but it makes the background apperar in conjunction with outline
for k, v in extra.items():
setattr(style, k, v)
sf.styles['Default'] = style
for s in subs:
sf.append(s)
return sf.to_string('ass')
def test_read_malformed():
"""no line number, no empty line, leading whitespace, bad timestamp format"""
text = dedent("""\
00:00:00.000 ->00:01:00.000
An example subtitle.
0:01:00,00 --> 0:02:00,00
Subtitle number
two.
""")
ref = SSAFile()
ref.append(SSAEvent(start=0, end=make_time(m=1), text="An example subtitle."))
ref.append(SSAEvent(start=make_time(m=1), end=make_time(m=2), text="Subtitle number\\Ntwo."))
subs = SSAFile.from_string(text)
assert subs.equals(ref)
def test_transform_framerate():
subs = SSAFile()
subs.append(SSAEvent(start=0, end=10))
subs.append(SSAEvent(start=1000, end=1010))
with assert_raises(ValueError):
subs.transform_framerate(1, 0)
with assert_raises(ValueError):
subs.transform_framerate(1, -1)
with assert_raises(ValueError):
subs.transform_framerate(0, 1)
with assert_raises(ValueError):
subs.transform_framerate(-1, 1)
subs.transform_framerate(10, 20)
assert subs[0] == SSAEvent(start=0, end=5)
assert subs[1] == SSAEvent(start=500, end=505)
def save_to_subtitles(results, formatter):
"""
Save to subtitle file
:param results: Dictionary containing info and start/end times
:param formatter: Apply text formating to the subtitle
:return: New subtitle file
"""
subs = SSAFile()
for result in results:
event = SSAEvent(start=make_time(s=result['start']),
end=make_time(s=result['end']),
text=formatter(result))
if 'highlight' in result and result['highlight']:
event.style = 'red'
subs.append(event)
logger.info(f'Processed {len(results)} results to subtitle events')
return subs
def save_to_srt(results, file_name):
"""Converts the output of the model to the standard subtitle format .srt"""
subs = SSAFile()
for result in results:
event = SSAEvent(start=make_time(s=result['start']),
end=make_time(s=result['end']),
text=result['word'])
subs.append(event)
subs.save(file_name)
def test_simple_read():
text = dedent("""\
1
00:00:00,000 --> 00:01:00,000
An example subtitle.
2
00:01:00,000 --> 00:02:00,000
Subtitle number
two.
""")
ref = SSAFile()
ref.append(SSAEvent(start=0, end=make_time(m=1), text="An example subtitle."))
ref.append(SSAEvent(start=make_time(m=1), end=make_time(m=2), text="Subtitle number\\Ntwo."))
subs = SSAFile.from_string(text)
assert subs.equals(ref)
def test_framerate_inference():
fps = 1000.0
has_fps = dedent("""\
{0}{0}1000.0
{10}{20}Hello!
""")
no_fps = dedent("""\
{10}{20}Hello!
""")
ignored_fps = dedent("""\
{0}{0}23.976
{10}{20}Hello!
""")
illegal_fps = dedent("""\
{0}{0}-23.976
{10}{20}Hello!
""")
subs1 = SSAFile.from_string(has_fps)
assert subs1.fps == fps
assert len(subs1) == 1
assert subs1[0] == SSAEvent(start=10, end=20, text="Hello!")
subs2 = SSAFile.from_string(no_fps, fps=fps)
assert subs2.fps == fps
assert len(subs2) == 1
assert subs2[0] == SSAEvent(start=10, end=20, text="Hello!")
# fps argument takes preference over what the file says, first line is kept
subs3 = SSAFile.from_string(ignored_fps, fps=fps)
assert subs3.fps == fps
assert len(subs3) == 2
assert subs3[0] == SSAEvent(start=0, end=0, text="23.976")
assert subs3[1] == SSAEvent(start=10, end=20, text="Hello!")
with assert_raises(UnknownFPSError):
SSAFile.from_string(no_fps)
with assert_raises(ValueError):
SSAFile.from_string(illegal_fps)
def combine_subs(first_subs,
second_subs,
third_subs,
fourth_subs,
one_only=False):
# Only ass files keep styling information properly
combined_subs = create_ssa_file()
for sub in second_subs:
combined_subs.append(
SSAEvent(start=sub.start,
end=sub.end,
text=sub.text,
style='bottom'))
for sub in first_subs:
combined_subs.append(
SSAEvent(start=sub.start,
end=sub.end,
text=f'({sub.text})',
style='top'))
for sub in third_subs:
if sub.style == 'red':
combined_subs.append(
SSAEvent(start=sub.start,
end=sub.end,
text=f'[{sub.text}]',
style='left_red'))
else:
combined_subs.append(
SSAEvent(start=sub.start,
end=sub.end,
text=f'[{sub.text}]',
style='left'))
for sub in fourth_subs:
combined_subs.append(
SSAEvent(start=sub.start,
end=sub.end,
text=f'[{sub.text}]',
style='right'))
combined_subs.sort()
if one_only:
combined_subs = filter_subs(combined_subs)
logger.info(f'Combined subtitles into {len(combined_subs)} subtitles')
return combined_subs
def test_insertion_of_wrong_type():
subs = SSAFile()
subs.append(SSAEvent())
with assert_raises(TypeError):
subs.append(42)
with assert_raises(TypeError):
subs.insert(42)
with assert_raises(TypeError):
subs[0] = 42
def test_writer_strips_tags():
subs = SSAFile()
subs.append(SSAEvent(start=0, end=10, text="Let me tell you{a secret}."))
f = dedent("""\
{0}{0}1000
{0}{10}Let me tell you.
""")
assert subs.to_string("microdvd", fps=1000) == f
def test_writer_uses_original_fps():
subs = SSAFile()
subs.append(SSAEvent(start=0, end=10, text="Hello!"))
subs.fps = 1000
f = dedent("""\
{0}{0}1000
{0}{10}Hello!
""")
assert subs.to_string("microdvd") == f
def to_subtitles(response):
subtitles = SSAFile()
subtitles.events = [
SSAEvent(
text=result.alternatives[0].transcript,
start=duration_to_ms(result.alternatives[0].words[0].start_time),
end=duration_to_ms(result.alternatives[0].words[-1].end_time),
) for result in response.results
]
return subtitles
def test_extra_whitespace_parsing():
f = dedent("""\
{ 0 } { 0 } 1000.0
{ 10 }{ 20} Hello!
""")
subs = SSAFile.from_string(f)
assert subs[0] == SSAEvent(start=10, end=20, text="Hello!")
def test_writer_handles_whitespace():
subs = SSAFile()
subs.append(
SSAEvent(start=0, end=10,
text=r"Hello,\hworld!\NSo many\N\nNewlines."))
f = dedent("""\
{0}{0}1000
{0}{10}Hello, world!|So many||Newlines.
""")
assert subs.to_string("microdvd", fps=1000) == f
def test_simple_write():
subs = SSAFile()
e1 = SSAEvent()
e1.start = 0
e1.end = 60000
e1.text = "An example subtitle."
e2 = SSAEvent()
e2.start = 60000
e2.end = 120000
e2.text = "Subtitle number\\Ntwo."
e3 = SSAEvent()
e3.start = 60000
e3.end = 120000
e3.text = "Invisible subtitle."
e3.is_comment = True
subs.append(e1)
subs.append(e2)
subs.append(e3)
ref = dedent("""\
1
00:00:00,000 --> 00:01:00,000
An example subtitle.
2
00:01:00,000 --> 00:02:00,000
Subtitle number
two.
""")
text = subs.to_string("srt")
assert text.strip() == ref.strip()
def test_shift():
e = SSAEvent(start=0, end=10)
with assert_raises(ValueError):
e.shift(frames=5)
with assert_raises(ValueError):
e.shift(fps=23.976)
with assert_raises(ValueError):
e.shift(frames=5, fps=-1)
e2 = e.copy(); e2.shift(ms=5)
assert e2 == SSAEvent(start=5, end=15)
e2 = e.copy(); e2.shift(ms=-5)
assert e2 == SSAEvent(start=-5, end=5)
e2 = e.copy(); e2.shift(frames=1, fps=100.0)
assert e2 == SSAEvent(start=10, end=20)
e2 = e.copy(); e2.shift(frames=-1, fps=100.0)
assert e2 == SSAEvent(start=-10, end=0)
e2 = e.copy(); e2.shift(h=1, m=-60, s=2, ms=-2000)
assert e2 == e
def test_repr_comment():
ev = SSAEvent(start=make_time(m=1, s=30), end=make_time(m=1, s=35), text="Hello\\Nworld!")
ev.is_comment = True
ref = "<SSAEvent type=Comment start=0:01:30 end=0:01:35 text='Hello\\Nworld!'>"
assert repr(ev) == ref
