def test_len(self):
assert len(Line("- <i>It's working.</i>")) == len("- It's working.")
assert len(Line("{\an8}appears at top")) == len("appears at top")
assert len(
Line(
'<i>Previously on <font color="#ffff00">"TV Show"</font>...</i>'
)) == len('Previously on "TV Show"...')
def test_fix_styles(self, processor: StyleProcessor):
assert processor.fix_styles(Line("<i></i>")) == ""
assert processor.fix_styles(Line("<i> </i>")) == " "
assert processor.fix_styles(Line("</i> <i>")) == " "
assert processor.fix_styles(Line("<i></i><i></i>")) == ""
assert (processor.fix_styles(
Line("<i></i> <i> </i> <i>sentence</i>")) == " <i>sentence</i>")
def test_clean_parentheses(self, fake_processor: SDHProcessor):
assert (fake_processor.clean_parentheses(
Line("that's for you. [sighs]")) == "that's for you. ")
assert (fake_processor.clean_parentheses(
Line("on my part, I mean, [laughs] utter idiocy.")) ==
"on my part, I mean, utter idiocy.")
assert (fake_processor.clean_parentheses(
Line(
"telling a joke [laughs], I mean, [continues laughing] you should've seen him."
)) == "telling a joke , I mean, you should've seen him.")
def test_strip(self):
assert Line("").strip() == ""
assert Line(" ").strip() == ""
assert Line(" ").strip() == ""
assert Line("test").strip() == "test"
assert Line(" test").strip() == "test"
assert Line(" test ").strip() == "test"
assert Line("test <i>").strip() == "test<i>"
assert Line("<i> test").strip() == "<i>test"
assert Line(" <i>test").strip() == "<i>test"
assert Line("<i>test</i>").strip() == "<i>test</i>"
def process_section(cls, section: Section) -> Section:
if not len(section) > 1:
return section
chunks = cls.split_dialog_chunks(section.lines)
section.lines = []
for chunk in chunks:
if not len(chunk) > 1:
section.lines.append(chunk[0])
elif cls.is_short(Line.merge(chunk)):
section.lines.append(Line.merge(chunk))
else:
section.lines += chunk
return section
def clean_hi(cls, line: Line) -> Line:
"""Clean hearing impaired."""
line = line.sub(
r"^([-‐\s<i>]+)?((\b[-A-Za-z.']+\s?#?\d?){1,2}(?!\.)([\[(][\w\s]*[\])])?:(?!\w)|[\[]+.*[\]:]+)(<\/?i>)?([\s])*",
r"\1\5",
)
line = cls.clean_parentheses(line)
return line
def test_contains_hi(self, fake_processor: SDHProcessor):
assert fake_processor.contains_hi(Line("that's for you. [sighs]"))
assert fake_processor.contains_hi(Line("‐TEACHER: blabla..."))
assert fake_processor.contains_hi(Line("[Laura] sentence"))
assert fake_processor.contains_hi(
Line("<i>[Laura]</i> <i>sentence</i>"))
assert fake_processor.contains_hi(Line("- CHRISTOPHER:<i> Hello?</i>"))
assert not fake_processor.contains_hi(
Line("9:17 a.m., to be specific,"))
assert not fake_processor.contains_hi(Line("between 4:00 and 6:00."))
assert not fake_processor.contains_hi(
Line("I got some time between 4:00 and 6:00."))
def test_in_blacklist(self, fake_processor: BlacklistProcessor):
assert fake_processor.in_blacklist(
Line("Advertise your product or brand here"))
assert fake_processor.in_blacklist(
Line("contact www.OpenSubtitles.org today"))
assert fake_processor.in_blacklist(
Line('<font color="#ffff00">Provided by username</font>'))
assert fake_processor.in_blacklist(Line("[http://example.com]"))
assert fake_processor.in_blacklist(Line("http://foo.network"))
assert fake_processor.in_blacklist(
Line("Visit https://another-example.com"))
assert fake_processor.in_blacklist(Line("find more under subs.link"))
assert fake_processor.in_blacklist(Line("twitter.com/username"))
def test_clean_dashes(self, processor: DialogProcessor):
assert processor.clean_dashes(Line("-dialog.")) == "- dialog."
assert processor.clean_dashes(Line("- dialog.")) == "- dialog."
assert processor.clean_dashes(Line("i-in")) == "i-in"
assert processor.clean_dashes(Line("<i>-dialog.</i>")) == "<i>- dialog.</i>"
assert processor.clean_dashes(Line("<i>-</i>dialog.")) == "<i>- </i>dialog."
assert processor.clean_dashes(Line("-...dialog.")) == "- ...dialog."
def test_is_parentheses(self, fake_processor: SDHProcessor):
assert fake_processor.is_parentheses(Line("(distant shouting))"))
assert fake_processor.is_parentheses(Line("[ distant shouting ]"))
assert fake_processor.is_parentheses(Line("-[camera shutter]"))
assert fake_processor.is_parentheses(
Line("(distant shouting,")) is False
assert fake_processor.is_parentheses(
Line("weapons clashing))")) is False
assert fake_processor.is_parentheses(
Line("[laughing nervously]:")) is False
def test_fix_space_punctuation(self, processor: ErrorProcessor):
assert (processor.fix_space_punctuation(
Line("First sentence . Second sentence , blabla.")) ==
"First sentence. Second sentence, blabla.")
assert (processor.fix_space_punctuation(
Line("First sentence... Second sentence.")) ==
"First sentence... Second sentence.")
assert processor.fix_space_punctuation(Line("Whoa ...")) == "Whoa..."
assert processor.fix_space_punctuation(Line("Whoa...")) == "Whoa..."
assert (processor.fix_space_punctuation(
Line("Yeah. ..maybe.")) == "Yeah...maybe.")
assert (processor.fix_space_punctuation(
Line("Begin... ...end.")) == "Begin... ...end.")
assert processor.fix_space_punctuation(
Line("- ...dialog")) == "- ...dialog"
def test_clean_hi(self, fake_processor: SDHProcessor):
assert (fake_processor.clean_hi(
Line("that's for you. [sighs]")) == "that's for you. ")
assert fake_processor.clean_hi(
Line("‐TEACHER: blabla...")) == "‐blabla..."
assert fake_processor.clean_hi(Line("[Laura] sentence")) == "sentence"
assert (fake_processor.clean_hi(
Line("<i>[Laura]</i> <i>sentence</i>")) == "<i></i><i>sentence</i>"
)
assert (fake_processor.clean_hi(
Line("- CHRISTOPHER:<i> Hello?</i>")) == "- <i>Hello?</i>")
assert (fake_processor.clean_hi(
Line("9:17 a.m., to be specific,")) == "9:17 a.m., to be specific,"
)
def test_strip_styles(self):
assert Line(
"- <i>It's working.</i>").strip_styles() == "- It's working."
assert Line("{\an8}appears at top").strip_styles() == "appears at top"
def test_is_dialog(self):
assert Line("- <i>This is a dialog.</i>").is_dialog()
assert Line("-this is also a dialog").is_dialog()
assert not Line("not a dialog").is_dialog()
def fix_spaces(line: Line) -> Line:
"""Add missing spaces between sentences."""
return line.sub(r"\b([.?!]+)([A-Z][a-z])", r"\1 \2")
def trim_whitespace(line: Line) -> Line:
"""Trim multiple spaces between words, also if there are style tags in between"""
return line.sub(r"\s+(<\/?i>)*\s*", r" \1").strip()
def test_split_dialog_chunks(self, processor: LineLengthProcessor):
assert processor.split_dialog_chunks([Line("hello"),
Line("there")]) == [[
Line("hello"),
Line("there"),
]]
assert processor.split_dialog_chunks([Line("- hello"),
Line("there")]) == [[
Line("- hello"),
Line("there"),
]]
assert processor.split_dialog_chunks([Line("hello"),
Line("- there")]) == [
[Line("hello")],
[Line("- there")],
]
assert processor.split_dialog_chunks(
[Line("hello"), Line("- there"),
Line("man")]) == [
[Line("hello")],
[Line("- there"), Line("man")],
]
assert processor.split_dialog_chunks([Line("- hi"),
Line("- bye")]) == [
[Line("- hi")],
[Line("- bye")],
]
assert processor.split_dialog_chunks(
[Line("- hi"),
Line("bob"),
Line("- bye"),
Line("bob")]) == [[Line("- hi"), Line("bob")],
[Line("- bye"), Line("bob")]]
assert processor.split_dialog_chunks(
[Line("hi"), Line("-bob"),
Line("- bye"),
Line("bob")]) == [[Line("hi")], [Line("-bob")],
[Line("- bye"), Line("bob")]]
assert processor.split_dialog_chunks([
Line("-I'm gonna call the police,"),
Line("this can't keep happening.")
]) == [[
Line("-I'm gonna call the police,"),
Line("this can't keep happening.")
]]
def fix_styles(line: Line) -> Line:
"""Remove leftover style tags"""
return line.sub(r"<\/?i>(\s*)<\/?i>", r"\1")
def join(self) -> Line:
return Line(" ".join(self.lines))
def test_is_simple_hi(self, fake_processor: SDHProcessor):
assert fake_processor.is_simple_hi(Line("♪"))
assert fake_processor.is_simple_hi(Line("- ♪ ♪"))
def fix_space_punctuation(line: Line) -> Line:
line = line.sub(r"(?<!^-)(?<!\.{3})\s+([.,!?]+)",
r"\1") # remove space before punctuation
line = line.sub(r"([.,!?]+)\s{2,}(?!$)",
r"\1 ") # fix multiple spaces after punctuation
return line
def fix_hyphen(line: Line) -> Line:
return line.sub(r"'’", "'")
def test_is_hi(self, fake_processor: SDHProcessor):
assert fake_processor.is_hi(Line("[camera shutter]"))
assert fake_processor.is_hi(Line("-[camera shutter]"))
assert fake_processor.is_hi(Line("- [camera shutter]"))
assert fake_processor.is_hi(Line("(distant shouting))"))
assert fake_processor.is_hi(Line("[ distant shouting ]"))
assert fake_processor.is_hi(Line("♪"))
assert fake_processor.is_hi(Line("- ♪ ♪"))
assert fake_processor.is_hi(Line("(distant shouting,")) is False
assert fake_processor.is_hi(Line("weapons clashing))")) is False
assert fake_processor.is_hi(Line("-[journalists] Christine!")) is False
assert fake_processor.is_hi(Line("- TAMIKA: Yeah.")) is False
assert fake_processor.is_hi(Line("♪ (SOFT PIANO MUSIC PLAYS)) ♪"))
def clean_dashes(line: Line) -> Line:
return line.sub(r"^(<\/?i>)*([-‐]+)(\s+)?", r"\1- ")
def clean_parentheses(line: Line) -> Line:
"""Clean parentheses ()[]."""
return line.sub(r"[(\[*].*?[)\]*:]+", "")
def fix_quote(line: Line) -> Line:
return line.sub(r""", '"')
def is_parenthesis_not_matching(line: Line) -> bool:
return bool(
re.search(r"[()\[\]]", line)
and (line.count("(") != line.count(")")
or line.count("[") != line.count("]")))
def test_is_music(self, fake_processor: SDHProcessor):
assert fake_processor.is_music(Line("♪ ominous music ♪"))
assert fake_processor.is_music(Line("- ♪ mysterious music ♪"))
assert fake_processor.is_music(Line("♪ somber music ♪"))
assert fake_processor.is_music(Line("♪ foreboding music ♪"))
assert fake_processor.is_music(Line("♪ chilling music ♪"))
assert fake_processor.is_music(Line("♪ solemn music ♪"))
assert fake_processor.is_music(Line(" ♪ dramatic music ♪"))
assert fake_processor.is_music(Line("♪ poignant music ♪"))
assert fake_processor.is_music(Line("♪ emotional music ♪"))
assert fake_processor.is_music(Line("♪ uneasy music ♪"))
assert fake_processor.is_music(Line("♪ harrowing music ♪"))
assert fake_processor.is_music(Line("♪ sinister music ♪"))
assert fake_processor.is_music(Line("♪ upbeat music plays ♪"))
assert fake_processor.is_music(Line("♪ gentle music ♪"))
assert fake_processor.is_music(Line("♪ light orchestral music ♪"))
assert fake_processor.is_music(Line("♪ upbeat folk music playing ♪"))
assert fake_processor.is_music(Line("♪ dramatic music playing ♪"))
assert fake_processor.is_music(Line("♪ ominous music playing ♪"))
assert fake_processor.is_music(Line("♪ dramatic music playing ♪"))
assert fake_processor.is_music(Line("♪ upbeat music playing ♪"))
assert fake_processor.is_music(Line("♪ dramatic music playing ♪"))
assert fake_processor.is_music(Line("♪ dramatic music playing ♪"))
assert fake_processor.is_music(Line("♪ smooth music playing ♪"))
assert fake_processor.is_music(Line("♪ dance music playing ♪"))
assert fake_processor.is_music(Line("♪ dramatic music playing ♪"))
assert fake_processor.is_music(Line("♪ dramatic music playing ♪"))
assert fake_processor.is_music(Line("♪ dramatic music playing ♪"))
assert fake_processor.is_music(Line("♪ dramatic music playing ♪"))
assert fake_processor.is_music(
Line("♪ swelling orchestral music plays ♪"))
assert fake_processor.is_music(Line("♪ dramatic music playing ♪"))
assert fake_processor.is_music(
Line("♪ upbeat song playing over speakers ♪"))
assert fake_processor.is_music(
Line("♪ upbeat song playing over headphones ♪"))
assert fake_processor.is_music(Line("♪ soft, dramatic music ♪"))
assert fake_processor.is_music(Line("♪ soft, stirring music ♪"))
assert fake_processor.is_music(Line("♪ music intensifies ♪"))
assert fake_processor.is_music(Line("♪ music swells ♪"))
assert fake_processor.is_music(Line("♪ dark music swells ♪"))
assert fake_processor.is_music(Line("♪ dramatic musical sting ♪"))
assert fake_processor.is_music(Line("♪ uneasy musical crescendo ♪"))
assert fake_processor.is_music(
Line("♪ up‐tempo percussive music playing ♪"))
assert fake_processor.is_music(Line("<i>quiet, eerie music </i>"))
assert fake_processor.is_music(Line("<i>droning music fades in </i>"))
assert fake_processor.is_music(Line("<i>women vocalizing eerily </i>"))
assert not fake_processor.is_music(
Line("this is not a music subtitle"))
def fix_ampersand(line: Line) -> Line:
return line.sub(r"&", "&")
def fix_music(line: Line) -> Line:
return line.sub(r"^#\s", "♪ ")
