class QueueWatcher(threading.Thread):
queue_name = None
types = None
logger = None
batch_obj = None
batch_queue = None
def __init__(self, queue_name, types, batch_obj, error_queue):
threading.Thread.__init__(self)
self.queue_name = queue_name
self.types = types
self.batch_obj = batch_obj
self.batch_queue = BatchQueue(queue_name, batch_obj)
self.error_queue = error_queue
def run(self):
logger.info("Starting watcher on queue: %s" % self.queue_name)
while True:
try:
while True:
self.batch_queue.dequeue()
logger.info("Dequeuing %s batch." % self.queue_name)
for key in self.types:
if self.batch_obj.type == key:
self.types[key].import_batch(self.batch_obj)
break
except django.db.utils.IntegrityError:
logger.warning("Tried to create a dup in queue: %s. Ignoring." % self.queue_name)
except:
logger.debug("Problem in %s queue:" % self.queue_name, exc_info=sys.exc_info())
self.error_queue.put(sys.exc_info())
raise
def __init__(self, queue_name, types, batch_obj, error_queue):
threading.Thread.__init__(self)
self.queue_name = queue_name
self.types = types
self.batch_obj = batch_obj
self.batch_queue = BatchQueue(queue_name, batch_obj)
self.error_queue = error_queue
def register_schema_map(self, sourceid):
cur = self.conn.cursor()
mappings = self.get_field_mappings_by_source(sourceid,
only_unmapped=True)
cmd = "SELECT id, name FROM global_attributes"
cur.execute(cmd)
global_attributes = self.dictfetchall(cur)
batch_obj = expertsrc_pb2.QuestionBatch()
batch_obj.type = expertsrc_pb2.QuestionBatch.SCHEMAMAP
# TODO: make sure to grab this from auth service using provided cookie
batch_obj.asker_name = 'data-tamer'
cmd = """SELECT local_id, value
FROM local_sources ls, local_source_meta lsm
WHERE ls.id = %s and lsm.source_id = ls.id and
lsm.meta_name = 'expertsrc_domain'"""
cur.execute(cmd, (sourceid, ))
source_name, domain_name = cur.fetchone()
logger.info('source_name -> %s' % source_name)
logger.info('domain_name -> %s' % domain_name)
batch_obj.source_name = source_name
batch = BatchQueue('question', batch_obj)
for fid in mappings.keys():
question = batch.getbatchobj().question.add()
question.domain_name = domain_name
question.local_field_id = fid
question.local_field_name = mappings[fid]['name']
choices = mappings[fid]['matches']
ids = list()
choice_count = 10
for c in choices:
if choice_count > 0:
choice = question.choice.add()
choice.global_attribute_id = c['id']
choice.global_attribute_name = c['name']
choice.confidence_score = c['score']
ids.append(c['id'])
choice_count -= 1
# uncomment this if you want to add all global attributes as
# potential choices.
id_set = set(ids)
for a in global_attributes:
if a['id'] not in id_set:
choice = question.choice.add()
choice.global_attribute_id = a['id']
choice.global_attribute_name = a['name']
batch.enqueue()
self.conn.commit()
def register_schema_map(self, sourceid):
cur = self.conn.cursor()
mappings = self.get_field_mappings_by_source(sourceid, only_unmapped=True)
cmd = "SELECT id, name FROM global_attributes"
cur.execute(cmd)
global_attributes = self.dictfetchall(cur)
batch_obj = expertsrc_pb2.QuestionBatch()
batch_obj.type = expertsrc_pb2.QuestionBatch.SCHEMAMAP
# TODO: make sure to grab this from auth service using provided cookie
batch_obj.asker_name = 'data-tamer'
cmd = """SELECT local_id, value
FROM local_sources ls, local_source_meta lsm
WHERE ls.id = %s and lsm.source_id = ls.id and
lsm.meta_name = 'expertsrc_domain'"""
cur.execute(cmd, (sourceid,))
source_name, domain_name = cur.fetchone()
logger.info('source_name -> %s' % source_name)
logger.info('domain_name -> %s' % domain_name)
batch_obj.source_name = source_name
batch = BatchQueue('question', batch_obj)
for fid in mappings.keys():
question = batch.getbatchobj().question.add()
question.domain_name = domain_name
question.local_field_id = fid
question.local_field_name = mappings[fid]['name']
choices = mappings[fid]['matches']
ids = list()
choice_count = 10
for c in choices:
if choice_count > 0:
choice = question.choice.add()
choice.global_attribute_id = c['id']
choice.global_attribute_name = c['name']
choice.confidence_score = c['score']
ids.append(c['id'])
choice_count -= 1
# uncomment this if you want to add all global attributes as
# potential choices.
id_set = set(ids)
for a in global_attributes:
if a['id'] not in id_set:
choice = question.choice.add()
choice.global_attribute_id = a['id']
choice.global_attribute_name = a['name']
batch.enqueue()
self.conn.commit()
def create_mappings(self,
pairs,
anti=False,
answerer_id=0,
answerer_auth=0.5):
if len(pairs) == 0:
return
cur = self.conn.cursor()
params = []
batch_obj = expertsrc_pb2.AnswerBatch()
batch_obj.type = expertsrc_pb2.AnswerBatch.SCHEMAMAP
batch = BatchQueue('answer', batch_obj)
cmd = 'INSERT INTO attribute_mappings (' \
' local_id, global_id, confidence, authority, who_created, ' \
' when_created, why_created) ' \
'VALUES '
if anti:
cmd = cmd.replace('mappings', 'antimappings')
for local_id, global_id, conf in pairs:
cmd = cmd + '(%s, %s, %s, %s, %s, NOW(), %s), '
params.append(local_id)
params.append(global_id)
params.append(int(conf) / 100.0)
params.append(answerer_auth)
params.append(answerer_id)
params.append("Expertsrc")
# register answers with expertsrc
answer = batch.getbatchobj().answer.add()
answer.answerer_id = answerer_id
answer.confidence = int(conf) / 100.0
answer.authority = answerer_auth
answer.global_attribute_id = int(global_id)
answer.local_field_id = int(local_id)
answer.is_match = not anti
batch.enqueue()
cmd = cmd[0:-2] + ';'
cur.execute(cmd, params)
self.conn.commit()
return str([cmd, params])
def create_mappings(self, pairs, anti=False, answerer_id=0, answerer_auth=0.5):
if len(pairs) == 0:
return
cur = self.conn.cursor()
params = []
batch_obj = expertsrc_pb2.AnswerBatch()
batch_obj.type = expertsrc_pb2.AnswerBatch.SCHEMAMAP
batch = BatchQueue('answer', batch_obj)
cmd = 'INSERT INTO attribute_mappings (' \
' local_id, global_id, confidence, authority, who_created, ' \
' when_created, why_created) ' \
'VALUES '
if anti:
cmd = cmd.replace('mappings', 'antimappings')
for local_id, global_id, conf in pairs:
cmd = cmd + '(%s, %s, %s, %s, %s, NOW(), %s), '
params.append(local_id)
params.append(global_id)
params.append(int(conf) / 100.0)
params.append(answerer_auth)
params.append(answerer_id)
params.append("Expertsrc")
# register answers with expertsrc
answer = batch.getbatchobj().answer.add()
answer.answerer_id = answerer_id
answer.confidence = int(conf) / 100.0
answer.authority = answerer_auth
answer.global_attribute_id = int(global_id)
answer.local_field_id = int(local_id)
answer.is_match = not anti
batch.enqueue()
cmd = cmd[0:-2] + ';'
cur.execute(cmd, params)
self.conn.commit()
return str([cmd, params])
