def import_feed(self, feed):
from molly.apps.feeds.models import Item, vCard
xml = etree.parse(urllib2.urlopen(feed.rss_url))
feed.last_modified = datetime.now()
feed.save(update_last_modified=True)
items = set()
for talk in xml.findall('talk'):
item, created = Item.objects.get_or_create(
feed=feed, guid=talk.find('id').text)
item.last_modified = self.parse_date(talk.find('updated_at').text)
item.title = talk.find('title').text.strip()
item.description = sanitise_html(
talk.find('abstract').text.strip())
item.link = talk.find('url').text
item.dt_start = self.parse_date(talk.find('start_time').text)
item.dt_end = self.parse_date(talk.find('end_time').text)
location, created = vCard.objects.get_or_create(
name=talk.find('venue').text.strip())
location.save()
item.venue = location
item.save()
items.add(item)
for item in Item.objects.filter(feed=feed):
if item not in items:
item.delete()
return items
def import_feed(self, feed):
from molly.apps.feeds.models import Item, vCard
xml = etree.parse(urllib2.urlopen(feed.rss_url))
feed.last_modified = datetime.now()
feed.save(update_last_modified=True)
items = set()
for talk in xml.findall('talk'):
item, created = Item.objects.get_or_create(feed=feed, guid=talk.find('id').text)
item.last_modified = self.parse_date(talk.find('updated_at').text)
item.title = talk.find('title').text.strip()
item.description = sanitise_html(talk.find('abstract').text.strip())
item.link = talk.find('url').text
item.dt_start = self.parse_date(talk.find('start_time').text)
item.dt_end = self.parse_date(talk.find('end_time').text)
location, created = vCard.objects.get_or_create(name=talk.find('venue').text.strip())
location.save()
item.venue = location
item.save()
items.add(item)
for item in Item.objects.filter(feed=feed):
if item not in items:
item.delete()
return items
def import_feed(self, feed):
from molly.apps.feeds.models import Item, vCard
calendar = Calendar.from_string(urllib2.urlopen(feed.rss_url).read())
items = set()
for component in calendar.walk():
if component.name == 'VEVENT':
item, created = Item.objects.get_or_create(feed=feed,
guid=str(component.get('UID')))
# Do not create the event if one the property is not correct,
# first tries to parse DT as datetime then as date, if it still
# fails, then ignore
try:
try:
item.dt_start = vDatetime.from_ical(str(
component.get('DTSTART')))
except ValueError, ve:
item.dt_start = vDate.from_ical(str(
component.get('DTSTART')))
if component.get('DTEND'):
try:
item.dt_end = vDatetime.from_ical(str(
component.get('DTEND')))
except ValueError, ve:
item.dt_end = vDate.from_ical(str(
component.get('DTEND')))
item.title = vText.from_ical(str(
component.get('SUMMARY')).strip())
if component.get('URL'):
item.link = str(component.get('URL'))
if component.get('DESCRIPTION'):
item.description = sanitise_html(vText.from_ical(str(
component.get('DESCRIPTION'))))
if str(component.get('LOCATION')) != '':
location, created = vCard.objects.get_or_create(
name=vText.from_ical(str(
component.get('LOCATION')).strip()))
# in the future, we could imagine to (try to) geocode
# the location to get a point field...
location.save()
item.venue = location
try:
item.last_modified = vDatetime.from_ical(str(
component.get('LAST-MODIFIED')))
except Exception, e:
item.last_modified = datetime.now()
item.save()
items.add(item)
def import_feed(self, feed):
from molly.apps.feeds.models import Item
feed_data = feedparser.parse(feed.rss_url)
try:
feed.last_modified = \
struct_to_datetime(feed_data.feed.updated_parsed)
except:
feed.last_modified = \
parse_date(feed_data.headers.get('last-modified',
datetime.now().strftime("%a, %d %b %Y %H:%M:%S +0000")))
feed.save()
items = set()
for x_item in feed_data.entries:
if hasattr(x_item, 'id'):
guid = x_item.id
else:
# Some stupid feeds don't have any GUIDs, fall back to the URL
# and hope it's unique
guid = x_item.link
try:
last_modified = datetime(*x_item.date_parsed[:7])
except:
last_modified = None
for i in items:
if i.guid == guid:
item = i
break
else:
try:
item = Item.objects.get(guid=guid, feed=feed)
except Item.DoesNotExist:
item = Item(guid=guid,
last_modified=datetime(1900, 1, 1),
feed=feed)
if True or item.last_modified < last_modified:
item.title = x_item.title
item.description = sanitise_html(x_item.get('description', ''))
item.link = x_item.link
item.last_modified = last_modified
item.save()
items.add(item)
for item in Item.objects.filter(feed=feed):
if item not in items:
item.delete()
return items
def import_feed(self, feed):
from molly.apps.feeds.models import Item
feed_data = feedparser.parse(feed.rss_url)
try:
feed.last_modified = struct_to_datetime(feed_data.feed.updated_parsed)
except:
feed.last_modified = parse_date(
feed_data.headers.get("last-modified", datetime.now().strftime("%a, %d %b %Y %H:%M:%S +0000"))
)
feed.save()
items = set()
for x_item in feed_data.entries:
if hasattr(x_item, "id"):
guid = x_item.id
else:
# Some stupid feeds don't have any GUIDs, fall back to the URL
# and hope it's unique
guid = x_item.link
try:
last_modified = datetime(*x_item.date_parsed[:7])
except:
last_modified = None
for i in items:
if i.guid == guid:
item = i
break
else:
try:
item = Item.objects.get(guid=guid, feed=feed)
except Item.DoesNotExist:
item = Item(guid=guid, last_modified=datetime(1900, 1, 1), feed=feed)
if True or item.last_modified < last_modified:
item.title = x_item.title
item.description = sanitise_html(x_item.get("description", ""))
item.link = x_item.link
item.last_modified = last_modified
item.save()
items.add(item)
for item in Item.objects.filter(feed=feed):
if item not in items:
item.delete()
return items
def import_feed(self, feed):
from molly.apps.feeds.models import Item
feed_data = feedparser.parse(feed.rss_url)
try:
feed.last_modified = struct_to_datetime(feed_data.feed.updated_parsed)
except:
feed.last_modified = parse_date(feed_data.headers.get('last-modified', datetime.now().strftime("%a, %d %b %Y %H:%M:%S +0000")))
feed.save()
items = set()
for x_item in feed_data.entries:
guid, last_modified = x_item.id, datetime(*x_item.date_parsed[:7])
for i in items:
if i.guid == guid:
item = i
break
else:
try:
item = Item.objects.get(guid=guid, feed=feed)
except Item.DoesNotExist:
item = Item(guid=guid, last_modified=datetime(1900,1,1), feed=feed)
if True or item.last_modified < last_modified:
item.title = x_item.title
item.description = sanitise_html(x_item.get('description', ''))
item.link = x_item.link
item.last_modified = last_modified
item.save()
items.add(item)
for item in Item.objects.filter(feed=feed):
if item not in items:
item.delete()
return items
