def get_creator(self, tag='dc:creator'):
creator = ''
if len(self.item.getElementsByTagName(tag)) > 0:
creator = util.getNodeText(self.item.getElementsByTagName(tag)[0].childNodes)
if creator == None or creator.strip() == '':
creator = util.getCData(self.item.getElementsByTagName(tag)[0])
if creator == None or creator.strip() == '':
if len(self.item.getElementsByTagName('author')) > 0 and len(self.item.getElementsByTagName('author')[0].getElementsByTagName('name')) > 0:
creator = util.getNodeText(self.item.getElementsByTagName('author')[0].getElementsByTagName('name')[0].childNodes)
return creator.strip()
def set_locations_from_api(self):
params = {"method": "geo.getmetros"}
xmlevents = self.call(params)
root = xmlevents.documentElement
metros = root.getElementsByTagName("metros")[0]
for m in metros.getElementsByTagName("metro"):
city_name = util.getNodeText(m.getElementsByTagName("name")[0].childNodes)
country_name = util.getNodeText(m.getElementsByTagName("country")[0].childNodes)
try:
dbcountry, created = Country.objects.get_or_create(name=country_name, short_name=slugify(country_name))
City.objects.get_or_create(name=city_name, slug=slugify(city_name), country=dbcountry)
except:
print "FAILED ON", city_name, country_name
def get_creator(self, tag='dc:creator'):
creator = ''
if len(self.item.getElementsByTagName(tag)) > 0:
creator = util.getNodeText(
self.item.getElementsByTagName(tag)[0].childNodes)
if creator == None or creator.strip() == '':
creator = util.getCData(self.item.getElementsByTagName(tag)[0])
if creator == None or creator.strip() == '':
if len(self.item.getElementsByTagName('author')) > 0 and len(
self.item.getElementsByTagName('author')
[0].getElementsByTagName('name')) > 0:
creator = util.getNodeText(
self.item.getElementsByTagName('author')
[0].getElementsByTagName('name')[0].childNodes)
return creator.strip()
def get_title(self):
title = util.getNodeText(
self.item.getElementsByTagName('title')[0].childNodes)
if title.find('CDATA') > -1 or title.strip() == '':
title = util.getCData(self.item.getElementsByTagName('title')
[0]) # village voice, style peterson
return title.strip()
def get_category(self):
if self.source.category == '':
if (len(self.item.getElementsByTagName('category')) > 0):
return util.getNodeText(self.item.getElementsByTagName('category')[0].childNodes)
else:
return self.source.category
return ''
def get_description(self, tag='description'):
desc = ''
if len(self.item.getElementsByTagName(tag)) > 0:
desc = util.getNodeText(self.item.getElementsByTagName(tag)[0].childNodes)
if desc == None or desc.strip() == '' or desc.find('CDATA') > -1:
desc = util.getCData(self.item.getElementsByTagName(tag)[0])
if desc == None or desc.strip() == '':
if len(self.item.getElementsByTagName('content')) > 0:
desc = util.getNodeText(self.item.getElementsByTagName('content')[0].childNodes)
if desc == None or desc.strip() == '' or desc.find('CDATA') > -1:
desc = util.getCData(self.item.getElementsByTagName('content')[0])
if desc == None or desc.strip() == '':
if len(self.item.getElementsByTagName('summary')) > 0:
desc = util.getNodeText(self.item.getElementsByTagName('summary')[0].childNodes)
return desc.strip()
def get_category(self):
if self.source.category == '':
if (len(self.item.getElementsByTagName('category')) > 0):
return util.getNodeText(
self.item.getElementsByTagName('category')[0].childNodes)
else:
return self.source.category
return ''
def get_date(self):
if (len(self.item.getElementsByTagName('dc:date')) > 0):
dateStr = util.getNodeText(self.item.getElementsByTagName('dc:date')[0].childNodes)
if dateStr =='':
return None
#2011-01-01T14:00:18-05:00
date = datetime.datetime.strptime(dateStr[0:dateStr.rfind('-')], datetime_formats.GOTHAMIST_TIME_FORMAT)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('updated')) > 0):
dateStr = util.getNodeText(self.item.getElementsByTagName('updated')[0].childNodes)
if dateStr =='':
return None
date = datetime.datetime.strptime(dateStr, datetime_formats.BK_VEGAN_FORMAT)
date = date - datetime.timedelta(hours=5)
return date
return None
def set_locations_from_api(self):
params = {"method": "geo.getmetros"}
xmlevents = self.call(params)
root = xmlevents.documentElement
metros = root.getElementsByTagName('metros')[0]
for m in metros.getElementsByTagName('metro'):
city_name = util.getNodeText(
m.getElementsByTagName('name')[0].childNodes)
country_name = util.getNodeText(
m.getElementsByTagName('country')[0].childNodes)
try:
dbcountry, created = Country.objects.get_or_create(
name=country_name, short_name=slugify(country_name))
City.objects.get_or_create(name=city_name,
slug=slugify(city_name),
country=dbcountry)
except:
print 'FAILED ON', city_name, country_name
def get_date(self):
if (len(self.item.getElementsByTagName('updated')) > 0):
dateStr = util.getNodeText(
self.item.getElementsByTagName('updated')[0].childNodes)
if dateStr == '':
return None
date = datetime.datetime.strptime(dateStr,
datetime_formats.BK_VEGAN_FORMAT)
date = date - datetime.timedelta(hours=5)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('published')) > 0):
dateStr = util.getNodeText(self.item.getElementsByTagName('published')[0].childNodes)
if dateStr =='':
return None
#2010-12-24T08:27:44-05:00
dateStr = dateStr[0:dateStr.rfind('-')].strip()
# 2010-12-24T08:27:44
date = datetime.datetime.strptime(dateStr, datetime_formats.GOTHAMIST_TIME_FORMAT)
return date
return None
def get_category(self):
if (len(self.item.getElementsByTagName('category')) > 0):
cat = util.getNodeText(self.item.getElementsByTagName('category')[0].childNodes)
if cat == 'Film':
return 'Movies'
if cat == 'Art':
return 'Art'
if cat == 'Theater':
return 'Art'
return cat
return 'News'
def get_date(self):
if (len(self.item.getElementsByTagName('published')) > 0):
dateStr = util.getNodeText(self.item.getElementsByTagName('published')[0].childNodes)
if dateStr =='':
return None
#2011-01-16T14:03:20.364-05:00
dateStr = dateStr[0:dateStr.find('.')]
#2011-01-16T14:03:20
date = datetime.datetime.strptime(dateStr, datetime_formats.GOTHAMIST_TIME_FORMAT)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('updated')) > 0):
dateStr = util.getNodeText(self.item.getElementsByTagName('updated')[0].childNodes)
if dateStr =='':
return None
#2009-08-13T17:33:20.000Z
dateStr = dateStr[0:dateStr.find('.')]
#2009-08-13T17:33:20
date = datetime.datetime.strptime(dateStr, datetime_formats.GOTHAMIST_TIME_FORMAT)
date = date - datetime.timedelta(hours=5)
return date
return None
def get_description(self, tag='description'):
desc = ''
if len(self.item.getElementsByTagName(tag)) > 0:
desc = util.getNodeText(
self.item.getElementsByTagName(tag)[0].childNodes)
if desc == None or desc.strip() == '' or desc.find('CDATA') > -1:
desc = util.getCData(self.item.getElementsByTagName(tag)[0])
if desc == None or desc.strip() == '':
if len(self.item.getElementsByTagName('content')) > 0:
desc = util.getNodeText(
self.item.getElementsByTagName('content')[0].childNodes)
if desc == None or desc.strip(
) == '' or desc.find('CDATA') > -1:
desc = util.getCData(
self.item.getElementsByTagName('content')[0])
if desc == None or desc.strip() == '':
if len(self.item.getElementsByTagName('summary')) > 0:
desc = util.getNodeText(
self.item.getElementsByTagName('summary')[0].childNodes)
return desc.strip()
def get_category(self):
if (len(self.item.getElementsByTagName('category')) > 0):
cat = util.getNodeText(
self.item.getElementsByTagName('category')[0].childNodes)
if cat == 'Film':
return 'Movies'
if cat == 'Art':
return 'Art'
if cat == 'Theater':
return 'Art'
return cat
return 'News'
def get_date(self):
if (len(self.item.getElementsByTagName('pubDate')) > 0):
dateStr = util.getNodeText(self.item.getElementsByTagName('pubDate')[0].childNodes)
if dateStr =='':
return None
#<pubDate>Mon, 03 Jan 2011 14:19:06 +0000</pubDate>
dateStr = dateStr[dateStr.find(',')+1:].strip()
dateStr = dateStr[0:dateStr.rfind('+')].strip()
date = datetime.datetime.strptime(dateStr, datetime_formats.VILLAGE_VOICE_TIME_FORMAT)
date = date - datetime.timedelta(hours=5)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('pubDate')) > 0):
dateStr = util.getNodeText(self.item.getElementsByTagName('pubDate')[0].childNodes)
if dateStr =='':
return None
#<pubDate>Fri, 24 Dec 2010 21:00:08 -0800</pubDate>
dateStr = dateStr[dateStr.find(',')+1:].strip()
dateStr = dateStr[0:dateStr.rfind('-')].strip()
date = datetime.datetime.strptime(dateStr, datetime_formats.L_MAG_TIME_FORMAT)
date = date + datetime.timedelta(hours=3)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('dc:date')) > 0):
dateStr = util.getNodeText(
self.item.getElementsByTagName('dc:date')[0].childNodes)
if dateStr == '':
return None
#2011-01-01T14:00:18-05:00
date = datetime.datetime.strptime(
dateStr[0:dateStr.rfind('-')],
datetime_formats.GOTHAMIST_TIME_FORMAT)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('pubDate')) > 0):
dateStr = util.getNodeText(self.item.getElementsByTagName('pubDate')[0].childNodes)
if dateStr =='':
return None
#<pubDate>Fri, 31 Dec 2010 15:56:00 -0500</pubDate>
dateStr = dateStr[dateStr.find(',')+1:].strip()
# 31 Dec 2010 15:56:00 -0500
dateStr = dateStr[0:dateStr.rfind('-')].strip()
# 31 Dec 2010 15:56:00
date = datetime.datetime.strptime(dateStr, datetime_formats.L_MAG_TIME_FORMAT)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('published')) > 0):
dateStr = util.getNodeText(
self.item.getElementsByTagName('published')[0].childNodes)
if dateStr == '':
return None
#2011-01-16T14:03:20.364-05:00
dateStr = dateStr[0:dateStr.find('.')]
#2011-01-16T14:03:20
date = datetime.datetime.strptime(
dateStr, datetime_formats.GOTHAMIST_TIME_FORMAT)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('published')) > 0):
dateStr = util.getNodeText(
self.item.getElementsByTagName('published')[0].childNodes)
if dateStr == '':
return None
#2010-12-24T08:27:44-05:00
dateStr = dateStr[0:dateStr.rfind('-')].strip()
# 2010-12-24T08:27:44
date = datetime.datetime.strptime(
dateStr, datetime_formats.GOTHAMIST_TIME_FORMAT)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('pubDate')) > 0):
dateStr = util.getNodeText(self.item.getElementsByTagName('pubDate')[0].childNodes)
if dateStr =='':
return None
#<pubDate>Sun, 02 Jan 2011 23:35:40 GMT</pubDate>
dateStr = dateStr[dateStr.find(',')+1:].strip()
# 31 Dec 2010 15:56:00 GMT
dateStr = dateStr.replace('GMT', '').strip()
# 31 Dec 2010 15:56:00
date = datetime.datetime.strptime(dateStr, datetime_formats.L_MAG_TIME_FORMAT)
date = date - datetime.timedelta(hours=5)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('pubDate')) > 0):
dateStr = util.getNodeText(
self.item.getElementsByTagName('pubDate')[0].childNodes)
if dateStr == '':
return None
#<pubDate>Fri, 24 Dec 2010 21:00:08 -0800</pubDate>
dateStr = dateStr[dateStr.find(',') + 1:].strip()
dateStr = dateStr[0:dateStr.rfind('-')].strip()
date = datetime.datetime.strptime(
dateStr, datetime_formats.L_MAG_TIME_FORMAT)
date = date + datetime.timedelta(hours=3)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('pubDate')) > 0):
dateStr = util.getNodeText(
self.item.getElementsByTagName('pubDate')[0].childNodes)
if dateStr == '':
return None
#<pubDate>Mon, 03 Jan 2011 14:19:06 +0000</pubDate>
dateStr = dateStr[dateStr.find(',') + 1:].strip()
dateStr = dateStr[0:dateStr.rfind('+')].strip()
date = datetime.datetime.strptime(
dateStr, datetime_formats.VILLAGE_VOICE_TIME_FORMAT)
date = date - datetime.timedelta(hours=5)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('updated')) > 0):
dateStr = util.getNodeText(
self.item.getElementsByTagName('updated')[0].childNodes)
if dateStr == '':
return None
#2009-08-13T17:33:20.000Z
dateStr = dateStr[0:dateStr.find('.')]
#2009-08-13T17:33:20
date = datetime.datetime.strptime(
dateStr, datetime_formats.GOTHAMIST_TIME_FORMAT)
date = date - datetime.timedelta(hours=5)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('pubDate')) > 0):
dateStr = util.getNodeText(self.item.getElementsByTagName('pubDate')[0].childNodes)
if dateStr =='':
return None
# <pubDate>Fri, 24 Sep 2010 10:44:43 -0400</pubDate>
dateStr = dateStr[dateStr.find(',')+1:].strip()
# 31 Dec 2010 15:56:00 -0400
dateStr = dateStr[0:dateStr.rfind('-')].strip()
# 31 Dec 2010 15:56:00
date = datetime.datetime.strptime(dateStr, datetime_formats.VILLAGE_VOICE_TIME_FORMAT)
date = date - datetime.timedelta(hours=1)
return date
return None
def get_url(self):
url = ''
if len(self.item.getElementsByTagName('link')) == 1:
url = util.getNodeText(self.item.getElementsByTagName('link')[0].childNodes)
if url == None or url.strip() == '':
url = self.item.getElementsByTagName('link')[0].getAttribute('href') #bk vegan, flaming pab
else:
for l in self.item.getElementsByTagName('link'): #style peterson
if l.getAttribute('rel') == 'alternate':
url = l.getAttribute('href')
break
if url == None:
url = ''
return url.strip()
def get_date(self):
if (len(self.item.getElementsByTagName('pubDate')) > 0):
dateStr = util.getNodeText(
self.item.getElementsByTagName('pubDate')[0].childNodes)
if dateStr == '':
return None
#<pubDate>Fri, 31 Dec 2010 15:56:00 -0500</pubDate>
dateStr = dateStr[dateStr.find(',') + 1:].strip()
# 31 Dec 2010 15:56:00 -0500
dateStr = dateStr[0:dateStr.rfind('-')].strip()
# 31 Dec 2010 15:56:00
date = datetime.datetime.strptime(
dateStr, datetime_formats.L_MAG_TIME_FORMAT)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('pubDate')) > 0):
dateStr = util.getNodeText(
self.item.getElementsByTagName('pubDate')[0].childNodes)
if dateStr == '':
return None
#<pubDate>Sun, 02 Jan 2011 23:35:40 GMT</pubDate>
dateStr = dateStr[dateStr.find(',') + 1:].strip()
# 31 Dec 2010 15:56:00 GMT
dateStr = dateStr.replace('GMT', '').strip()
# 31 Dec 2010 15:56:00
date = datetime.datetime.strptime(
dateStr, datetime_formats.L_MAG_TIME_FORMAT)
date = date - datetime.timedelta(hours=5)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('pubDate')) > 0):
dateStr = util.getNodeText(
self.item.getElementsByTagName('pubDate')[0].childNodes)
if dateStr == '':
return None
# <pubDate>Fri, 24 Sep 2010 10:44:43 -0400</pubDate>
dateStr = dateStr[dateStr.find(',') + 1:].strip()
# 31 Dec 2010 15:56:00 -0400
dateStr = dateStr[0:dateStr.rfind('-')].strip()
# 31 Dec 2010 15:56:00
date = datetime.datetime.strptime(
dateStr, datetime_formats.VILLAGE_VOICE_TIME_FORMAT)
date = date - datetime.timedelta(hours=1)
return date
return None
def get_url(self):
url = ''
if len(self.item.getElementsByTagName('link')) == 1:
url = util.getNodeText(
self.item.getElementsByTagName('link')[0].childNodes)
if url == None or url.strip() == '':
url = self.item.getElementsByTagName('link')[0].getAttribute(
'href') #bk vegan, flaming pab
else:
for l in self.item.getElementsByTagName('link'): #style peterson
if l.getAttribute('rel') == 'alternate':
url = l.getAttribute('href')
break
if url == None:
url = ''
return url.strip()
def get_date(self):
if (len(self.item.getElementsByTagName('pubDate')) > 0):
link = self.source.url;
dateStr = util.getNodeText(self.item.getElementsByTagName('pubDate')[0].childNodes)
if dateStr =='':
return None
if link.find('promoEvents') > -0:
#<pubDate>2010-10-14 15:39:09</pubDate>
date = datetime.datetime.strptime(dateStr, datetime_formats.VILLAGE_VOICE_PROMO_TIME_FORMAT)
return date
else:
#<pubDate>Fri, 31 Dec 2010 15:56:00 -0500</pubDate>
dateStr = dateStr[dateStr.find(',')+1:].strip()
# 31 Dec 2010 15:56:00 -0500
dateStr = dateStr[0:dateStr.rfind('-')].strip()
# 31 Dec 2010 15:56:00
date = datetime.datetime.strptime(dateStr, datetime_formats.VILLAGE_VOICE_TIME_FORMAT)
return date
return None
def get_date(self):
if (len(self.item.getElementsByTagName('pubDate')) > 0):
link = self.source.url
dateStr = util.getNodeText(
self.item.getElementsByTagName('pubDate')[0].childNodes)
if dateStr == '':
return None
if link.find('promoEvents') > -0:
#<pubDate>2010-10-14 15:39:09</pubDate>
date = datetime.datetime.strptime(
dateStr, datetime_formats.VILLAGE_VOICE_PROMO_TIME_FORMAT)
return date
else:
#<pubDate>Fri, 31 Dec 2010 15:56:00 -0500</pubDate>
dateStr = dateStr[dateStr.find(',') + 1:].strip()
# 31 Dec 2010 15:56:00 -0500
dateStr = dateStr[0:dateStr.rfind('-')].strip()
# 31 Dec 2010 15:56:00
date = datetime.datetime.strptime(
dateStr, datetime_formats.VILLAGE_VOICE_TIME_FORMAT)
return date
return None
def save_event_to_db(self, eventNode, loc):
artists = []
headliners = []
# load up artists
artistsNodes = eventNode.getElementsByTagName("artists")[0]
for a in artistsNodes.getElementsByTagName("artist"):
try:
artist = util.getNodeText(a.childNodes)
a = Artist.objects.get_or_create(name=artist)[0]
artists.append(a)
except:
print exception_util.get_exception_str()
for h in artistsNodes.getElementsByTagName("headliner"):
try:
headliner = util.getNodeText(h.childNodes)
h = Artist.objects.get_or_create(name=headliner)[0]
headliners.append(h)
except:
print exception_util.get_exception_str()
# venue stuff
vNode = eventNode.getElementsByTagName("venue")[0]
vId = util.getNodeText(vNode.getElementsByTagName("id")[0].childNodes)
vName = util.getNodeText(vNode.getElementsByTagName("name")[0].childNodes)
vExternalUrl = util.getNodeText(vNode.getElementsByTagName("url")[0].childNodes)
vUrl = util.getNodeText(vNode.getElementsByTagName("website")[0].childNodes)
vPhone = util.getNodeText(vNode.getElementsByTagName("phonenumber")[0].childNodes)
imgNodes = vNode.getElementsByTagName("image")
vImgSm = ""
vImgMed = ""
vImgLg = ""
vImgXLg = ""
vImgMega = ""
for iNode in imgNodes:
if iNode.hasChildNodes():
if iNode.getAttribute("size") == "small":
vImgSm = util.getNodeText(iNode.childNodes)
elif iNode.getAttribute("size") == "medium":
vImgMed = util.getNodeText(iNode.childNodes)
elif iNode.getAttribute("size") == "large":
vImgLg = util.getNodeText(iNode.childNodes)
elif iNode.getAttribute("size") == "extralarge":
vImgXLg = util.getNodeText(iNode.childNodes)
elif iNode.getAttribute("size") == "mega":
vImgMega = util.getNodeText(iNode.childNodes)
locNode = vNode.getElementsByTagName("location")[0]
# vCity = util.getNodeText(locNode.getElementsByTagName('city')[0].childNodes)
vCountry = util.getNodeText(locNode.getElementsByTagName("country")[0].childNodes)
vStreet = util.getNodeText(locNode.getElementsByTagName("street")[0].childNodes)
vZip = util.getNodeText(locNode.getElementsByTagName("postalcode")[0].childNodes)
geoNode = locNode.getElementsByTagName("geo:point")[0]
vLat = util.getNodeText(geoNode.getElementsByTagName("geo:lat")[0].childNodes)
vLong = util.getNodeText(geoNode.getElementsByTagName("geo:long")[0].childNodes)
city = City.objects.get(name=loc)
v = None
try:
v = Venue.objects.get_or_create(
external_id=vId, name=vName, city=city, country=vCountry, source=self.source
)[0]
v.street = vStreet
v.zip = vZip
v.lat = vLat
v.long = vLong
v.external_url = vExternalUrl
v.url = vUrl
v.phone_number = vPhone
v.image_size_sm = vImgSm
v.image_size_med = vImgMed
v.image_size_lg = vImgLg
v.image_size_xl = vImgXLg
v.image_size_mega = vImgMega
v.save()
except:
print "VENUE PROB", exception_util.get_exception_str()
# Ticket stuff
tickets = []
ticketsNode = eventNode.getElementsByTagName("tickets")
if len(ticketsNode) != 0:
ticketsNode = ticketsNode[0]
for tNode in ticketsNode.getElementsByTagName("ticket"):
supplier = tNode.getAttribute("supplier")
tixUrl = util.getNodeText(tNode.childNodes)
ticket = Ticket.objects.get_or_create(name=supplier, url=tixUrl)[0]
tickets.append(ticket)
# Tags stuff
tags = []
tagsNode = eventNode.getElementsByTagName("tags")
if len(tagsNode) != 0:
tagsNode = tagsNode[0]
for tNode in tagsNode.getElementsByTagName("tag"):
tagTxt = util.getNodeText(tNode.childNodes)
tag = Tag.objects.get_or_create(name=tagTxt)[0]
tags.append(tag)
# Event stuff
lastfm_id = util.getNodeText(eventNode.getElementsByTagName("id")[0].childNodes)
lastfm_id = "lfm" + lastfm_id
title = util.getNodeText(eventNode.getElementsByTagName("title")[0].childNodes)
startDate = util.getNodeText(eventNode.getElementsByTagName("startDate")[0].childNodes)
start_date = datetime.datetime.fromtimestamp(
time.mktime(time.strptime(startDate, datetime_formats.LAST_FM_FORMAT))
)
startDate = start_date.strftime(datetime_formats.MYSQL_FORMAT)
description = util.getNodeText(eventNode.getElementsByTagName("description")[0].childNodes) # TODO get text???
external_url = ""
url = ""
for urlNode in eventNode.getElementsByTagName("url"):
if urlNode.parentNode == eventNode:
external_url = util.getNodeText(urlNode.childNodes)
for urlNode in eventNode.getElementsByTagName("website"):
if urlNode.parentNode == eventNode:
url = util.getNodeText(urlNode.childNodes)
cancelled = util.getNodeText(eventNode.getElementsByTagName("cancelled")[0].childNodes)
if cancelled == "":
cancelled = 0
else:
cancelled = int(cancelled)
imgSm = ""
imgMed = ""
imgLg = ""
imgXLg = ""
imgNodes = eventNode.getElementsByTagName("image")
for iNode in imgNodes:
if iNode.parentNode == eventNode:
if iNode.getAttribute("size") == "small":
imgSm = util.getNodeText(iNode.childNodes)
elif iNode.getAttribute("size") == "medium":
imgMed = util.getNodeText(iNode.childNodes)
elif iNode.getAttribute("size") == "large":
imgLg = util.getNodeText(iNode.childNodes)
elif iNode.getAttribute("size") == "extralarge":
imgXLg = util.getNodeText(iNode.childNodes)
if v != None:
event, created = Event.objects.get_or_create(external_id=lastfm_id)
event.name = title
event.start_time = start_date
# default all last.fm events to end at 6
event.end_time = start_date + datetime.timedelta(hours=6)
event.description = description
event.image_size_sm = imgSm
event.image_size_med = imgMed
event.image_size_lg = imgLg
event.image_size_xl = imgXLg
event.cancelled = cancelled
event.venue = v
event.category = self.source.category
event.source = self.source
event.external_url = external_url
event.url = url
for a in artists:
event.artists.add(a)
for h in headliners:
event.headliners.add(h)
for t in tags:
event.tags.add(t)
for t in tickets:
event.tickets.add(t)
try:
event.save()
except:
print "EVENT ERR", exception_util.get_exception_str()
def save_event_to_db(self, eventNode, loc):
artists = []
headliners = []
# load up artists
artistsNodes = eventNode.getElementsByTagName('artists')[0]
for a in artistsNodes.getElementsByTagName('artist'):
try:
artist = util.getNodeText(a.childNodes)
a = Artist.objects.get_or_create(name=artist)[0]
artists.append(a)
except:
print exception_util.get_exception_str()
for h in artistsNodes.getElementsByTagName('headliner'):
try:
headliner = util.getNodeText(h.childNodes)
h = Artist.objects.get_or_create(name=headliner)[0]
headliners.append(h)
except:
print exception_util.get_exception_str()
# venue stuff
vNode = eventNode.getElementsByTagName('venue')[0]
vId = util.getNodeText(vNode.getElementsByTagName('id')[0].childNodes)
vName = util.getNodeText(
vNode.getElementsByTagName('name')[0].childNodes)
vExternalUrl = util.getNodeText(
vNode.getElementsByTagName('url')[0].childNodes)
vUrl = util.getNodeText(
vNode.getElementsByTagName('website')[0].childNodes)
vPhone = util.getNodeText(
vNode.getElementsByTagName('phonenumber')[0].childNodes)
imgNodes = vNode.getElementsByTagName('image')
vImgSm = ""
vImgMed = ""
vImgLg = ""
vImgXLg = ""
vImgMega = ""
for iNode in imgNodes:
if iNode.hasChildNodes():
if iNode.getAttribute('size') == "small":
vImgSm = util.getNodeText(iNode.childNodes)
elif iNode.getAttribute('size') == "medium":
vImgMed = util.getNodeText(iNode.childNodes)
elif iNode.getAttribute('size') == "large":
vImgLg = util.getNodeText(iNode.childNodes)
elif iNode.getAttribute('size') == "extralarge":
vImgXLg = util.getNodeText(iNode.childNodes)
elif iNode.getAttribute('size') == "mega":
vImgMega = util.getNodeText(iNode.childNodes)
locNode = vNode.getElementsByTagName('location')[0]
#vCity = util.getNodeText(locNode.getElementsByTagName('city')[0].childNodes)
vCountry = util.getNodeText(
locNode.getElementsByTagName('country')[0].childNodes)
vStreet = util.getNodeText(
locNode.getElementsByTagName('street')[0].childNodes)
vZip = util.getNodeText(
locNode.getElementsByTagName('postalcode')[0].childNodes)
geoNode = locNode.getElementsByTagName('geo:point')[0]
vLat = util.getNodeText(
geoNode.getElementsByTagName('geo:lat')[0].childNodes)
vLong = util.getNodeText(
geoNode.getElementsByTagName('geo:long')[0].childNodes)
city = City.objects.get(name=loc)
v = None
try:
v = Venue.objects.get_or_create(external_id=vId,
name=vName,
city=city,
country=vCountry,
source=self.source)[0]
v.street = vStreet
v.zip = vZip
v.lat = vLat
v.long = vLong
v.external_url = vExternalUrl
v.url = vUrl
v.phone_number = vPhone
v.image_size_sm = vImgSm
v.image_size_med = vImgMed
v.image_size_lg = vImgLg
v.image_size_xl = vImgXLg
v.image_size_mega = vImgMega
v.save()
except:
print "VENUE PROB", exception_util.get_exception_str()
# Ticket stuff
tickets = []
ticketsNode = eventNode.getElementsByTagName('tickets')
if len(ticketsNode) != 0:
ticketsNode = ticketsNode[0]
for tNode in ticketsNode.getElementsByTagName('ticket'):
supplier = tNode.getAttribute('supplier')
tixUrl = util.getNodeText(tNode.childNodes)
ticket = Ticket.objects.get_or_create(name=supplier,
url=tixUrl)[0]
tickets.append(ticket)
# Tags stuff
tags = []
tagsNode = eventNode.getElementsByTagName('tags')
if len(tagsNode) != 0:
tagsNode = tagsNode[0]
for tNode in tagsNode.getElementsByTagName('tag'):
tagTxt = util.getNodeText(tNode.childNodes)
tag = Tag.objects.get_or_create(name=tagTxt)[0]
tags.append(tag)
# Event stuff
lastfm_id = util.getNodeText(
eventNode.getElementsByTagName('id')[0].childNodes)
lastfm_id = 'lfm' + lastfm_id
title = util.getNodeText(
eventNode.getElementsByTagName('title')[0].childNodes)
startDate = util.getNodeText(
eventNode.getElementsByTagName('startDate')[0].childNodes)
start_date = datetime.datetime.fromtimestamp(
time.mktime(
time.strptime(startDate, datetime_formats.LAST_FM_FORMAT)))
startDate = start_date.strftime(datetime_formats.MYSQL_FORMAT)
description = util.getNodeText(
eventNode.getElementsByTagName('description')
[0].childNodes) #TODO get text???
external_url = ""
url = ""
for urlNode in eventNode.getElementsByTagName('url'):
if urlNode.parentNode == eventNode:
external_url = util.getNodeText(urlNode.childNodes)
for urlNode in eventNode.getElementsByTagName('website'):
if urlNode.parentNode == eventNode:
url = util.getNodeText(urlNode.childNodes)
cancelled = util.getNodeText(
eventNode.getElementsByTagName('cancelled')[0].childNodes)
if cancelled == '':
cancelled = 0
else:
cancelled = int(cancelled)
imgSm = ""
imgMed = ""
imgLg = ""
imgXLg = ""
imgNodes = eventNode.getElementsByTagName('image')
for iNode in imgNodes:
if iNode.parentNode == eventNode:
if iNode.getAttribute('size') == "small":
imgSm = util.getNodeText(iNode.childNodes)
elif iNode.getAttribute('size') == "medium":
imgMed = util.getNodeText(iNode.childNodes)
elif iNode.getAttribute('size') == "large":
imgLg = util.getNodeText(iNode.childNodes)
elif iNode.getAttribute('size') == "extralarge":
imgXLg = util.getNodeText(iNode.childNodes)
if v != None:
event, created = Event.objects.get_or_create(external_id=lastfm_id)
event.name = title
event.start_time = start_date
# default all last.fm events to end at 6
event.end_time = start_date + datetime.timedelta(hours=6)
event.description = description
event.image_size_sm = imgSm
event.image_size_med = imgMed
event.image_size_lg = imgLg
event.image_size_xl = imgXLg
event.cancelled = cancelled
event.venue = v
event.category = self.source.category
event.source = self.source
event.external_url = external_url
event.url = url
for a in artists:
event.artists.add(a)
for h in headliners:
event.headliners.add(h)
for t in tags:
event.tags.add(t)
for t in tickets:
event.tickets.add(t)
try:
event.save()
except:
print "EVENT ERR", exception_util.get_exception_str()
def get_category(self):
if (len(self.item.getElementsByTagName('dc:subject')) > 0):
return util.getNodeText(self.item.getElementsByTagName('dc:subject')[0].childNodes)
return 'News'
def get_category(self):
if (len(self.item.getElementsByTagName('dc:subject')) > 0):
return util.getNodeText(
self.item.getElementsByTagName('dc:subject')[0].childNodes)
return 'News'
def get_title(self):
title = util.getNodeText(self.item.getElementsByTagName('title')[0].childNodes)
if title.find('CDATA') > -1 or title.strip() == '':
title = util.getCData(self.item.getElementsByTagName('title')[0]) # village voice, style peterson
return title.strip()
