def read(self, content, lang='en-US'):
caption_set = CaptionSet()
lines = content.splitlines()
start_line = 0
captions = []
while start_line < len(lines):
if not lines[start_line].isdigit():
break
caption = Caption()
end_line = self._find_text_line(start_line, lines)
timing = lines[start_line + 1].split('-->')
caption.start = self._srttomicro(timing[0].strip(' \r\n'))
caption.end = self._srttomicro(timing[1].strip(' \r\n'))
for line in lines[start_line + 2:end_line - 1]:
# skip extra blank lines
if not caption.nodes or line != '':
caption.nodes.append(CaptionNode.create_text(line))
caption.nodes.append(CaptionNode.create_break())
# remove last line break from end of caption list
caption.nodes.pop()
captions.append(caption)
start_line = end_line
caption_set.set_captions(lang, captions)
return caption_set
def _translate_break(self, caption):
# if break appears at start of caption, skip break
if self.first_element == True:
return
# if the last caption was a break, skip this break
elif caption.nodes[-1].type == CaptionNode.BREAK:
return
# close any open italics
elif self.open_italic == True:
caption.nodes.append(CaptionNode.create_style(False, {'italics': True}))
self.open_italic = False
# add line break
caption.nodes.append(CaptionNode.create_break())
def _combine_matching_captions(self, caption_set):
for lang in caption_set.get_languages():
captions = caption_set.get_captions(lang)
new_caps = captions[:1]
for caption in captions[1:]:
if caption.start == new_caps[-1].start and caption.end == new_caps[-1].end:
new_caps[-1].nodes.append(CaptionNode.create_break())
new_caps[-1].nodes.extend(caption.nodes)
else:
new_caps.append(caption)
caption_set.set_captions(lang, new_caps)
return caption_set
def _translate_tag(self, tag):
# convert text
if isinstance(tag, NavigableString):
if tag.strip() != '':
node = CaptionNode.create_text(tag.strip())
self.nodes.append(node)
# convert line breaks
elif tag.name == 'br':
self.nodes.append(CaptionNode.create_break())
# convert italics
elif tag.name == 'span':
# convert span
self._translate_span(tag)
else:
# recursively call function for any children elements
for a in tag.contents:
self._translate_tag(a)
def _translate_tag(self, tag):
# convert text
if isinstance(tag, NavigableString):
self.line.append(CaptionNode.create_text(tag.strip()))
# convert line breaks
elif tag.name == 'br':
self.line.append(CaptionNode.create_break())
# convert italics
elif tag.name == 'i':
self.line.append(CaptionNode.create_style(True, {'italics': True}))
# recursively call function for any children elements
for a in tag.contents:
self._translate_tag(a)
self.line.append(
CaptionNode.create_style(False, {'italics': True}))
elif tag.name == 'span':
self._translate_span(tag)
else:
# recursively call function for any children elements
for a in tag.contents:
self._translate_tag(a)
