def save_info(self):
config = as_dict(self.config)
self.parser_info = ParserInfo(name=self.name, config=json.dumps(config))
self.parser_info.save()
return self.parser_info
class RSSParser:
name = 'RSS'
class Config:
resource = 'test.xml'
def __init__(self, **config):
self.config = self.Config()
self.config.__dict__.update(config)
url = urlparse(self.config.resource)
if url.scheme == 'file':
self.config.resource = url.path
elif url.scheme:
raise NotImplementedError('Only local documents are supported')
def get_document(self):
with open(self.config.resource) as f:
return etree.parse(f)
def iter_events(self, doc):
events = doc.findall('events/event')
for e in events:
event = ParsedEvent(e)
event.parse()
yield event
def iter_places(self, doc):
places = doc.findall('places/place')
for e in places:
place = ParsedPlace(e)
place.parse()
yield place
def iter_sessions(self, doc):
sessions = doc.findall('schedule/session')
for e in sessions:
session = Session(e)
session.parse()
yield session
def save_info(self):
config = as_dict(self.config)
self.parser_info = ParserInfo(name=self.name, config=json.dumps(config))
self.parser_info.save()
return self.parser_info
def parse_all(self):
'''good if all parsers would implement this
'''
doc = self.get_document()
return {
'events': [e.parsed_data for e in self.iter_events(doc)],
'places': [p.parsed_data for p in self.iter_places(doc)],
'sessions': [s.parsed_data for s in self.iter_sessions(doc)],
}
def save_all(self):
'''main method
'''
doc = self.get_document()
info = self.save_info()
for session in self.iter_sessions(doc):
session.save(info)
for event in self.iter_events(doc):
event.save(info)
for place in self.iter_places(doc):
place.save(info)
return {
'parser_info': info.id
}