def parse(self, data):
if self.skip_transcript(data):
return
date = data.get('date')
top_section = self.get_or_create(
Section,
instance=self.instance,
source_url=data['url'],
heading=self.top_section_heading(data),
parent=self.get_parent_section(data),
)
for speech in self.parse_transcript(data):
if not speech:
continue
if speech.section:
if speech.section.object:
section = speech.section.object
else:
heading = self.prettify(speech.section.heading)
section = Section(
instance=self.instance,
heading=heading,
parent=top_section,
)
if self.commit:
section.save()
speech.section.object = section
else:
section = top_section
if speech.speaker:
speaker = self.prettify(speech.speaker)
speaker = self.get_or_create(Speaker,
instance=self.instance,
name=speaker)
else:
speaker = None
if not speech.type:
speech.type = ('speech' if speaker or speech.speaker_display
else 'narrative')
text = '</p>\n<p>'.join([' '.join(s) for s in speech.text])
text = '<p>%s</p>' % text
speech_date = speech.date or date
speech = Speech(
instance=self.instance,
section=section,
text=text,
speaker=speaker,
speaker_display=speech.speaker_display,
type=speech.type,
start_date=speech_date,
start_time=speech.time,
)
if self.commit:
speech.save()
def parse(self, data):
url, date, heading, speaker, text = data
text, speaker = parse_speech(text, speaker)
speaker = self.get_or_create(
Speaker,
instance=self.instance,
name=speaker,
)
speech = Speech(
instance=self.instance,
text=text,
speaker=speaker,
start_date=date,
heading=heading,
source_url=url,
type='speech',
)
if self.commit:
speech.save()
