def run(queue_name=None,
queue_name_output=None,
write_delay=None,
flush_queue=None,
batch_size=None,
format_any=None,
trigger_none_msg=None, trigger_topic=None,
delete_on_error=False, dont_pass_through=False,
simulate_error=False, error_msg=None,
log_network_error=False
,**_):
if write_delay>900 or write_delay<1:
raise Exception("Invalid value for 'delay_seconds': must be between 1 and 900")
# SETUP PRIVATE QUEUE
#####################
def setup_queue(qn):
""" closure for setting up private queue """
conn = boto.connect_sqs()
q=conn.create_queue(qn)
if format_any:
q.set_message_class(RawMessage)
else:
q.set_message_class(JSONMessage)
return q
logging.info("Creating Input Queue... (automatic retry)")
try:
qi=retry(partial(setup_queue, queue_name))
logging.info("Input Queue successfully created")
except Exception,e:
raise Exception("Creating queue '%s': %s" % (queue_name, str(e)))
def run(queue_name=None, flush_queue=None,
batch_size=None, polling_interval=None,
format_any=None, propagate_error=None,
retry_always=None, wait_trigger=None,
trigger_none_msg=None, trigger_topic=None,
delete_on_error=False, dont_pass_through=False,
simulate_error=False, error_msg=None
,**_ ):
## we need a minimum of second between polls
polling_interval=max(1, polling_interval)
# SETUP PRIVATE QUEUE
#####################
def setup_private_queue():
""" closure for setting up private queue """
conn = boto.connect_sqs()
q=conn.create_queue(queue_name)
if format_any:
q.set_message_class(RawMessage)
else:
q.set_message_class(JSONMessage)
return q
try:
q=retry(setup_private_queue, always=retry_always)
logging.info("Queue successfully created")
except Exception,e:
raise Exception("Creating queue '%s': %s" % (queue_name, str(e)))
def run(
queue_name=None,
flush_queue=None,
format_any=None,
retry_always=None,
topics=None,
error_msg=None,
simulate_error=None,
**_
):
queue_name = queue_name.strip()
# SETUP PRIVATE QUEUE
def setup_private_queue():
conn = boto.connect_sqs()
q = conn.create_queue(queue_name)
if not format_any:
q.set_message_class(JSONMessage)
else:
q.set_message_class(RawMessage)
return q
try:
q = retry(setup_private_queue, always=retry_always)
except Exception, e:
raise Exception("Creating queue '%s': %s" % (queue_name, str(e)))
def run(bucket_name=None,
path_source=None,
path_move=None,
delete_source=False,
polling_interval=60,
extd=None,
extf=None
,**_):
if not delete_source and path_move is None:
raise Exception("Options 'delete source' or 'move path' is required")
if delete_source and path_move is not None:
raise Exception("Options 'delete source' and 'move path' are mutually exclusive")
#if args.enable_debug:
# logger=logging.getLogger()
# logger.setLevel(logging.DEBUG)
bucket_name=bucket_name.strip()
path_source=path_source.strip()
code, p_src=resolve_path(path_source)
if not code.startswith("ok"):
raise Exception("Invalid source path: %s" % path_source)
mkdir_p(p_src)
if path_move is not None:
code, path_move=resolve_path(path_move)
if not code.startswith("ok"):
raise Exception("Invalid move path: %s" % path_move)
code,_=mkdir_p(path_move)
if not code.startswith("ok"):
raise Exception("Can't create move path: %s" % path_move)
try:
conn = boto.connect_s3()
except:
## not much we can do
## but actually no remote calls are made
## at this point so it should be highly improbable
raise Exception("Can't 'connect' to S3")
###################### BUCKET
logging.info("Getting/creating bucket (unlimited retries with backoff)")
def _get_create_bucket():
return conn.create_bucket(bucket_name)
bucket=retry(_get_create_bucket)
logging.info("Got bucket: %s" % bucket_name)
#############################
logging.debug("Starting loop...")
ppid=os.getppid()
logging.info("Process pid: %s" % os.getpid())
logging.info("Parent pid: %s" % ppid)
while True:
if os.getppid()!=ppid:
logging.warning("Parent terminated... exiting")
break
#################################################
logging.debug("Start processing...")
code, dirs=get_root_dirs(p_src)
if not code.startswith("ok"):
raise Warning("Source path disappeared: %s" % p_src)
dirs=filter_dirs(extd, dirs)
for _dir in dirs:
process_dir(bucket, _dir, delete_source, extf, path_move)
#####################################################
logging.debug("...sleeping for %s seconds" % polling_interval)
sleep(polling_interval)
if format_any:
q.set_message_class(RawMessage)
else:
q.set_message_class(JSONMessage)
return q
logging.info("Creating Input Queue... (automatic retry)")
try:
qi=retry(partial(setup_queue, queue_name))
logging.info("Input Queue successfully created")
except Exception,e:
raise Exception("Creating queue '%s': %s" % (queue_name, str(e)))
logging.info("Creating Output Queue... (automatic retry)")
try:
qo=retry(partial(setup_queue, queue_name_output))
logging.info("Output Queue successfully created")
except Exception,e:
raise Exception("Creating queue '%s': %s" % (queue_name_output, str(e)))
# FLUSH QUEUE
#############
if flush_queue:
try:
qi.clear()
logging.info("Input Queue flushed")
except: pass
# MAIN LOOP
###########
def run(
enable_simulate=False,
bucket_name=None,
path_source=None,
path_moveto=None,
path_check=None,
num_files=5,
enable_delete=False,
propagate_error=False,
prefix=None,
polling_interval=None,
only_ext=None,
filename_input_full=False,
filename_input_regex=None,
key_output_format=None,
enable_progress_report=False,
write_done=False,
**_
):
if key_output_format is not None:
if filename_input_regex is None:
raise Exception("-ifnr and -okf options work in tandem")
if filename_input_regex is not None:
if key_output_format is None:
raise Exception("Input filename regex specified but no output S3 key format specified")
logging.info("Compiling input filename regex...")
try:
ireg = re.compile(filename_input_regex.strip("'"))
ofmt = key_output_format.strip("'")
except:
raise Exception("Can't compile input filename regex pattern")
else:
ireg = None
ofmt = None
# if args.enable_debug:
# logger=logging.getLogger()
# logger.setLevel(logging.DEBUG)
bucket_name = bucket_name.strip()
path_source = path_source.strip()
try:
prefix = prefix.strip()
except:
prefix = None
try:
path_moveto = path_moveto.strip()
except:
path_moveto = None
if path_check is not None:
code, path_check = resolve_path(path_check)
if not code.startswith("ok"):
logging.warning("path_check '%s' might be in error..." % path_check)
### VALIDATE PARAMETERS
if not enable_delete and path_moveto is None:
raise Exception("either -d or -m must be used")
if enable_delete and path_moveto is not None:
raise Exception("-d can't be used with -m")
code, p_src = resolve_path(path_source)
if not code.startswith("ok"):
raise Exception("Invalid source path: %s" % path_source)
if path_moveto is not None:
code, p_dst = resolve_path(path_moveto)
if not code.startswith("ok"):
raise Exception("Invalid moveto path: %s" % path_moveto)
else:
p_dst = None
### wait for 'source' path to be available
logging.info("Waiting for source path to be accessible... CTRL-c to stop")
while True:
if os.path.isdir(p_src):
break
sleep(1)
logging.info("* Source path accessible")
if path_moveto is not None:
logging.info("Creating 'moveto' directory if required")
code, _ = mkdir_p(p_dst)
if not code.startswith("ok"):
raise Exception("Can't create 'moveto' directory: %s" % p_dst)
logging.info("* Created moveto directory")
if not enable_simulate:
try:
conn = boto.connect_s3()
except:
## not much we can do
## but actually no remote calls are made
## at this point so it should be highly improbable
raise Exception("Can't 'connect' to S3")
if not enable_simulate:
###################### BUCKET
logging.info("Getting/creating bucket (unlimited retries with backoff)")
def _get_create_bucket():
return conn.create_bucket(bucket_name)
bucket = retry(_get_create_bucket)
logging.info("Got bucket")
#############################
if enable_simulate:
logging.info("Begin simulation...")
else:
logging.debug("Starting loop...")
ppid = os.getppid()
logging.info("Process pid: %s" % os.getpid())
logging.info("Parent pid: %s" % ppid)
while True:
if os.getppid() != ppid:
logging.warning("Parent terminated... exiting")
break
#################################################
_code, path_exists = safe_path_exists(path_check)
if path_check is None or path_exists:
try:
logging.debug("Start processing...")
count = 0
gen = gen_walk(p_src, max_files=num_files, only_ext=only_ext)
for src_filename in gen:
if enable_progress_report:
logging.info("Processing file: %s" % src_filename)
if write_done:
if is_done_file(src_filename):
continue
try:
s3key_name = gen_s3_key(ireg, ofmt, p_src, src_filename, prefix, filename_input_full)
except Exception, e:
raise Exception(
"Error generating S3 key... check your command line parameters... use the 'simulate' facility: %s"
% e
)
if enable_simulate:
simulate(src_filename, s3key_name, enable_delete, p_dst)
else:
k = S3Key(bucket)
k.key = s3key_name
was_uploaded = process_file(
enable_progress_report,
bucket_name,
prefix,
k,
src_filename,
p_dst,
enable_delete,
propagate_error,
write_done,
)
if was_uploaded:
count = count + 1
except Exception, e:
logging.error("Error processing files...(%s)" % str(e))
else:
logging.info()
if count > 0:
logging.info("Progress> uploaded %s files" % count)
#####################################################
logging.debug("...sleeping for %s seconds" % polling_interval)
sleep(polling_interval)
def run_aws(node_id, proc, polling_interval, queue_name, topic_name, dst_path, delete_queue):
if topic_name is None:
raise Exception("Need a topic_name")
auto_queue = False
if queue_name is None:
auto_queue = True
queue_name = gen_queue_name()
def setup_private_queue():
conn = boto.connect_sqs()
q = conn.create_queue(queue_name)
q.set_message_class(RawMessage)
return (conn, q)
# SETUP PRIVATE QUEUE
logging.info("Creating queue '%s'" % queue_name)
sqs_conn, q = retry(setup_private_queue, logmsg="Having trouble creating queue...")
topic_arn = build_topic_arn(sqs_conn, topic_name)
logging.info("topic_arn: %s" % topic_arn)
### create topic
def create_topic():
"""
{'CreateTopicResponse':
{'ResponseMetadata':
{'RequestId': '5e2c6700-4dd0-11e1-b421-41716ce69b95'},
'CreateTopicResult': {'TopicArn': 'arn:aws:sns:us-east-1:674707187858:election'}}}
"""
snsconn = boto.connect_sns()
snsconn.create_topic(topic_name)
retry(create_topic, logmsg="Having trouble creating topic...")
# SUBSCRIBE TO TOPIC
def sub_topic():
snsconn = boto.connect_sns()
snsconn.subscribe_sqs_queue(topic_arn, q)
return snsconn
snsconn = retry(sub_topic, logmsg="Having trouble subscribing queue to topic...")
logging.info("Subscribed to topic '%s'" % topic_name)
current_state = "NL"
MSGS = {"NL": "Leadership lost", "L": "Leadership gained", "ML": "Leadership momentum"}
poll_count = 0
def cleanup():
rm(dst_path)
if auto_queue or delete_queue:
logging.info("... deleting queue")
try:
sqs_conn.delete_queue(q)
except:
pass
ppid = os.getppid()
logging.info("Process pid: %s" % os.getpid())
logging.info("Parent pid: %s" % ppid)
logging.info("Starting loop...")
while True:
if os.getppid() != ppid:
logging.warning("Parent terminated... exiting")
cleanup()
break
try:
## do a bit of garbage collection :)
global sigtermReceived
if sigtermReceived:
cleanup()
raise SignalTerminate()
#########################################
try:
### BATCH PROCESS - required!!!
while True:
rawmsg = q.read()
if rawmsg is not None:
jsonmsg = json.loads(rawmsg.get_body())
q.delete_message(rawmsg)
## SNS encapsulates the original message...
nodeid = str(jsonmsg["Message"])
transition, current_state = proc.send((poll_count, nodeid))
jstdout({"state": current_state})
if transition:
logging.info(MSGS[current_state])
if current_state == "L":
code, _ = touch(dst_path)
logging.info("Created '%s': %s" % (dst_path, code))
else:
code, _ = rm(dst_path)
logging.info("Deleted '%s': %s" % (dst_path, code))
else:
break
except SQSDecodeError:
logging.warning("Message decoding error")
except Exception, e:
logging.error(str(e))
continue
msg = str(node_id)
logging.debug("Publishing our 'node id': %s" % node_id)
try:
snsconn.publish(topic_arn, msg)
except:
try:
snsconn.publish(topic_arn, msg)
except:
logging.warning("Can't publish to topic '%s'" % topic_name)
logging.debug("... sleeping for %s seconds" % polling_interval)
sleep(polling_interval)
poll_count = poll_count + 1
except KeyboardInterrupt:
cleanup()
raise
