def producer_queue(queue, ranges):
for response_id, cleaned_content in read_session.query(Response.id, Response.cleaned_content).filter(
and_(Response.schemas is not None, Response.schemas != '{}')).limit(ranges[0]).offset(ranges[1]).all():
stderr = validate_in_memory(cleaned_content)
validated_data = {
"response_id": response_id,
"valid": 'Error at' not in stderr,
"validated_on": datetime.now().isoformat()
}
if stderr:
validated_data.update({
"errors": [s.strip() for s in stderr.split('\n\n') if s]
})
queue.put(validated_data)
queue.put('STOP')
# set up the connection
with open('local_rds.conf', 'r') as f:
conf = json.loads(f.read())
reader = ResponseReader(conf)
loader = Loader(conf)
# get the set, validate, store outputs
# but need to paginate because of ram issues
for i in xrange(0, 668110, 25):
print 'QUERYING {0}:{1}'.format(i, 25)
for response in reader.read('', limit=25, offset=i):
print response.source_url
xml = response.cleaned_content
stderr = validate_in_memory(xml)
data = {
"response_id": response.id,
"valid": 'Error at' not in stderr,
"validated_on": datetime.now()
}
if stderr:
data.update({"errors": [s.strip() for s in stderr.split('\n\n')]})
print '\t{0}'.format(stderr[:100])
try:
v = Validation()
v.create(data)
loader.load(v)
except Exception as ex:
# set up the connection
with open('local_rds.conf', 'r') as f:
conf = json.loads(f.read())
reader = ResponseReader(conf)
loader = Loader(conf)
# get the set, validate, store outputs
# but need to paginate because of ram issues
for i in xrange(0, 668110, 25):
print 'QUERYING {0}:{1}'.format(i, 25)
for response in reader.read('', limit=25, offset=i):
print response.source_url
xml = response.cleaned_content
stderr = validate_in_memory(xml)
data = {
"response_id": response.id,
"valid": 'Error at' not in stderr,
"validated_on": datetime.now()
}
if stderr:
data.update({
"errors": [s.strip() for s in stderr.split('\n\n')]
})
print '\t{0}'.format(stderr[:100])
try:
v = Validation()
v.create(data)