def consumer_queue(proc_id, queue):
loader = Loader(conf)
while True:
try:
consumer_data = queue.get(proc_id, 1)
if consumer_data == 'STOP':
queue.put('STOP')
break
for data in consumer_data:
v = Validation()
v.create(data)
loader.load(v)
except Empty:
pass
loader.close()
def main():
op = OptionParser()
op.add_option('--connection', '-c')
op.add_option('--start', '-s')
op.add_option('--end', '-e')
op.add_option('--interval', '-i')
options, arguments = op.parse_args()
if not options.connection:
op.error('No RDS Connection provided')
try:
LIMIT = int(options.interval)
s = int(options.start)
e = int(options.end)
except Exception as ex:
op.error('Invalid pagination integer: {0}'.format(ex))
with open('big_rds.conf', 'r') as f:
conf = js.loads(f.read())
# our connection
engine = sqla.create_engine(conf.get('connection'))
Session = sessionmaker()
Session.configure(bind=engine)
session = Session()
cmd = "StdInParse -v=always -n -s -f < %s"
for OFFSET in xrange(s, e, LIMIT):
appends = []
for response in session.query(Response).filter(
and_(Response.schemas is not None, Response.schemas != '{}')
).limit(LIMIT).offset(OFFSET).all():
if response.validations:
continue
response_id = response.id
cleaned_content = response.cleaned_content
handle, name = tempfile.mkstemp(suffix='.xml')
write(handle, cleaned_content)
close(handle)
# stderr = validate_in_memory(cleaned_content)
tc = TimedCmd(cmd % name)
try:
status, output, error = tc.run(120)
except:
print 'failed validation: ', response_id
continue
finally:
unlink(name)
validated_data = {
"response_id": response_id,
"valid": 'Error at' not in error,
"validated_on": datetime.now().isoformat()
}
if error:
validated_data.update({
"errors": [v.strip() for v in error.split('\n\n') if v]
})
v = Validation()
v.create(validated_data)
appends.append(v)
try:
session.add_all(appends)
session.commit()
except Exception as ex:
print ex
print [a['response_id'] for a in appends]
print
session.rollback()
continue
def main():
op = OptionParser()
op.add_option("--connection", "-c")
op.add_option("--files", "-f") # comma-delim list, one response id/line
options, arguments = op.parse_args()
if not options.connection:
op.error("No RDS Connection provided")
if not options.files:
op.error("No file list")
with open("big_rds.conf", "r") as f:
conf = js.loads(f.read())
# our connection
engine = sqla.create_engine(conf.get("connection"))
Session = sessionmaker()
Session.configure(bind=engine)
session = Session()
cmd = "StdInParse -v=always -n -s -f < %s"
for f in options.files.split(","):
with open(f, "r") as g:
data = [int(a.strip()) for a in g.readlines() if a]
for d in data:
response = session.query(Response).filter(Response.id == d).first()
if response.validations:
continue
response_id = response.id
cleaned_content = response.cleaned_content
handle, name = tempfile.mkstemp(suffix=".xml")
write(handle, cleaned_content)
close(handle)
# stderr = validate_in_memory(cleaned_content)
tc = TimedCmd(cmd % name)
try:
status, output, error = tc.run(60)
except:
print "failed validation: ", response_id
error = "Error at validation CLI: timeout error"
# continue
finally:
unlink(name)
validated_data = {
"response_id": response_id,
"valid": "Error at" not in error,
"validated_on": datetime.now().isoformat(),
}
if error:
validated_data.update({"errors": [v.strip() for v in error.split("\n\n") if v]})
v = Validation()
v.create(validated_data)
try:
session.add(v)
session.commit()
except Exception as ex:
print ex
print d
print
session.rollback()
continue
session.close()
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:
print ex
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:
print ex