def job_redirect(request, slug, source, job_id):
if request.method == 'GET' and request.GET.has_key('redirect'):
try:
elastic = ES(settings.SEARCH_HOSTS)
data = elastic.get(source, 'job', job_id)
elastic.connection.close()
return HttpResponseRedirect(data['_source']['details_url'])
except NotFoundException:
raise Http404
return direct_to_template(request, 'pages/redirect.html')
def job_redirect(request, slug, source, job_id):
if request.method == "GET" and request.GET.has_key("redirect"):
try:
elastic = ES(settings.SEARCH_HOSTS)
data = elastic.get(source, "job", job_id)
elastic.connection.close()
return HttpResponseRedirect(data["_source"]["details_url"])
except NotFoundException:
raise Http404
return direct_to_template(request, "pages/redirect.html")
class KVStore(KVStoreBase):
def __init__(self, *args, **kwargs):
super(KVStore, self).__init__(*args, **kwargs)
self.connection = ES(settings.THUMBNAIL_ELASTIC_SEARCH_SERVERS)
def _get_raw(self, key):
try:
#import pdb; pdb.set_trace()
value = self.connection.get(settings.THUMBNAIL_ELASTIC_SEARCH_INDEX,
settings.THUMBNAIL_ELASTIC_SEARCH_DOCUMENT_TYPE,
key)
return value['_source']['value']
except:
return None
def _set_raw(self, key, value):
ret = self.connection.index({"value": value},
settings.THUMBNAIL_ELASTIC_SEARCH_INDEX,
settings.THUMBNAIL_ELASTIC_SEARCH_DOCUMENT_TYPE,
key)
return ret['ok']
def _delete_raw(self, *keys):
rets = []
for key in keys:
try:
ret = self.connection.delete(settings.THUMBNAIL_ELASTIC_SEARCH_INDEX,
settings.THUMBNAIL_ELASTIC_SEARCH_DOCUMENT_TYPE,
key)
rets.append(ret['ok'])
except:
rets.append(False)
return rets
def _find_keys_raw(self, prefix):
search = Search(query=PrefixQuery("_id", prefix), size=1000, start=0, fields=[])
results = self.connection.search(search,
indexes=[settings.THUMBNAIL_ELASTIC_SEARCH_INDEX,],
doc_types=[settings.THUMBNAIL_ELASTIC_SEARCH_DOCUMENT_TYPE,])
return [hit['_id'] for hit in results['hits']['hits']]
class ProcessSpiderData(Task):
def run(self, spider_name):
cities = []
backup_source = []
backup_created_date = None
self.elastic = ES(settings.SEARCH_HOSTS, timeout=22.0, bulk_size=1500)
java = JavaInterface()
self.extractor = java.ArticleSentencesExtractor.INSTANCE
self.logger = ProcessSpiderData.get_logger()
spider = Data.objects.get(name=spider_name)
source = spider.source
if spider and len(source):
backup_created_date = spider.created_date
index_new = '%s_%d' % (spider.name, int(time.time()))
# create new index (not connected to alias)
self.elastic.create_index(index_new)
self.elastic.put_mapping('job', {'job':{'properties':mapping}}, index_new)
for item in source:
item = self._process_content(item)
item = self._get_location(item)
if item.has_key('city'):
cities.append(item['city'])
self._create_index(index_new, item)
backup_source.append(item)
# save new index (in bulk)
self.elastic.force_bulk()
# create alias
indices_old = self.elastic.get_alias(spider.name)
self.elastic.set_alias(spider.name, [index_new])
# delete all indices
for index in indices_old:
self.elastic.delete_index_if_exists(index)
# optimize
self.elastic.optimize(index_new, refresh=True)
# save backup (currently processed data)
if len(backup_source) and backup_created_date:
self._process_cities(set(cities), spider_name)
cache.clear()
obj = DataBackup.objects.get_or_create(
name=spider_name,
created_date=backup_created_date
)
obj[0].source = binascii.hexlify(bz2.compress(
JSONEncoder().encode(backup_source)
))
obj[0].save()
# force java & ES garbage collection
self.elastic.connection.close()
del self.extractor
del java
return True
def _process_content(self, item):
if len(item['content']):
item['content'] = self.extractor.getText(jpype.JString(item['content']))
return item
def _get_location(self, item):
if not item.has_key('city'):
return item
try:
geo = geocoders.GeoNames()
places = geo.geocode(item['city'].encode('utf-8'), exactly_one=False)
if places:
place, (lat, lon) = places[0] if isinstance(places, list) else places
if place: item['pin'] = {
'location': { 'lat': lat, 'lon': lon }
}
except: pass
return item
def _create_index(self, index, item):
id = item['id']
del item['id']
try:
self.elastic.get(index, 'job', id)
except ElasticSearchException:
self.elastic.index(
dumps(item, cls=DjangoJSONEncoder),
index, 'job', id, bulk=True
)
def _process_cities(self, cities, spider_name):
cities_current = City.objects.filter(indices__contains='"%s"' % spider_name)
# save lists of saved cities
cities_old_single = [ city.name for city in cities_current if city.indices and spider_name in city.indices and len(city.indices) == 1 ]
cities_old_multi = [ city.name for city in cities_current if city.indices and spider_name in city.indices and len(city.indices) > 1 ]
for city in cities:
city = unicode(city.strip().lower())
city = normalize_spaces.sub(' ', city)
city = remove_braces.sub('', city)
city_clean = [remove_none_chars.sub('', word) for word in city.split(' ')]
city_clean = ' '.join(filter(None, city_clean))
city, created = City.objects.get_or_create(name = city_clean[:255])
if created:
city.indices = [spider_name]
else:
city.indices.append(spider_name)
city.indices = list(set(city.indices))
city.save()
if city.name in cities_old_single: cities_old_single.remove(city.name)
if city.name in cities_old_multi: cities_old_multi.remove(city.name)
# remove unlinked citie
City.objects.filter(name__in=cities_old_single).delete()
for item in City.objects.filter(name__in=cities_old_multi):
if spider_name in item.indices:
item.indices.remove(spider_name)
item.save()
