def match_new_io(base_io, new_io, match_map=None, copy=True):
"""
Generate notifications for new I/O data being added to base I/O data.
Args:
base_io: The existing (base) I/O data being added to, and possibly
referred to by the arriving (new) I/O data. Will be used
to complete the data being notified about. Can already
contain the new I/O data, it will be considered "new"
regardless.
new_io: The arriving (new) I/O data being added to the existing
(base) data. Can refer to the existing I/O data. The new
I/O data can already be added to the base I/O data,
anything in the new data will be considered "new"
regardless.
match_map: The map of subscription match functions: a dictionary with
OO data object list names and a list of tuples, each
containing the name of the subscription and a match
function.
Each function must accept an object from the corresponding
object list in OO data, and return an iterable producing
kcidb.monitor.output.NotificationMessage objects, or None,
which is equivalent to an empty iterable.
The default is a dictionary of matching functions from all
kcidb.subscriptions.* modules, where each is called
"match_<OBJ_NAME>", where "<OBJ_NAME>" is an object list
name without the "s" ending.
copy: True, if the data should be copied before
referencing/modifying. False, if the data could be
referenced and modified in-place.
Optional, default is True.
Returns:
The list of notifications: kcidb.monitor.output.Notification objects.
"""
assert LIGHT_ASSERTS or io.schema.is_valid(base_io)
assert LIGHT_ASSERTS or io.schema.is_valid(new_io)
# Merge the new data into the base (*copy* new data as we'll need it)
merged_io = io.merge(base_io, [new_io],
copy_target=copy,
copy_sources=True)
# Convert both to OO representation
merged_oo = oo.from_io(merged_io, copy=False)
new_oo = oo.from_io(new_io, copy=copy)
# Remove all objects with missing parents from the merged data
rooted_oo = oo.remove_orphans(merged_oo)
# Delist everything except loaded or modified objects, but keep references
masked_oo = oo.apply_mask(rooted_oo, new_oo)
# Generate notifications
return match_oo(masked_oo, match_map)
def notify_main():
"""Execute the kcidb-notify command-line tool"""
sys.excepthook = misc.log_and_print_excepthook
description = 'kcidb-notify - Generate notifications for new I/O data'
parser = misc.ArgumentParser(description=description)
parser.add_argument('base',
metavar='BASE_FILE',
nargs='?',
help='Path to a JSON file with base I/O data')
args = parser.parse_args()
base = io.new()
if args.base is not None:
try:
with open(args.base, "r") as json_file:
base_reports = [
io.schema.validate(data)
for data in misc.json_load_stream_fd(json_file.fileno())
]
base = io.merge(base,
base_reports,
copy_target=False,
copy_sources=False)
except (jq.JSONParseError,
jsonschema.exceptions.ValidationError) as err:
raise Exception("Failed reading base file") from err
try:
for new in misc.json_load_stream_fd(sys.stdin.fileno()):
new = io.schema.validate(new)
for notification in subscriptions.match_new_io(base, new):
sys.stdout.write(notification.render().as_string(
policy=email.policy.SMTPUTF8))
sys.stdout.write("\x00")
sys.stdout.flush()
base = io.merge(base, [new], copy_target=False, copy_sources=False)
except (jq.JSONParseError, jsonschema.exceptions.ValidationError) as err:
raise Exception("Failed reading new I/O data") from err
def merge_main():
"""Execute the kcidb-merge command-line tool"""
sys.excepthook = misc.log_and_print_excepthook
description = 'kcidb-merge - Upgrade and merge I/O data sets'
parser = misc.OutputArgumentParser(description=description)
args = parser.parse_args()
sources = [
io.schema.validate(data)
for data in misc.json_load_stream_fd(sys.stdin.fileno())
]
merged_data = io.merge(io.new(),
sources,
copy_target=False,
copy_sources=False)
misc.json_dump(merged_data, sys.stdout, indent=args.indent, seq=args.seq)
def kcidb_load_queue(event, context):
"""
Load multiple KCIDB data messages from the LOAD_QUEUE_SUBSCRIBER queue
into the database, if it stayed unmodified for at least
DATASET_LOAD_PERIOD.
"""
# Do nothing, if updated recently
now = datetime.datetime.now(datetime.timezone.utc)
last_modified = DB_CLIENT.get_last_modified()
LOGGER.debug("Now: %s, Last modified: %s", now, last_modified)
if last_modified and now - last_modified < DATASET_LOAD_PERIOD:
LOGGER.info("Database too fresh, exiting")
return
# Pull messages
msgs = kcidb_load_queue_msgs(LOAD_QUEUE_SUBSCRIBER, LOAD_QUEUE_MSG_MAX,
LOAD_QUEUE_OBJ_MAX, LOAD_QUEUE_TIMEOUT_SEC)
if msgs:
LOGGER.info("Pulled %u messages", len(msgs))
else:
LOGGER.info("Pulled nothing, exiting")
return
# Create merged data referencing the pulled pieces
LOGGER.debug("Merging %u messages...", len(msgs))
data = kcidb_io.merge(kcidb_io.new(), (msg[1] for msg in msgs),
copy_target=False,
copy_sources=False)
LOGGER.info("Merged %u messages", len(msgs))
# Load the merged data into the database
obj_num = kcidb_io.get_obj_num(data)
LOGGER.debug("Loading %u objects...", obj_num)
DB_CLIENT.load(data)
LOGGER.info("Loaded %u objects", obj_num)
# Acknowledge all the loaded data
for msg in msgs:
LOAD_QUEUE_SUBSCRIBER.ack(msg[0])
LOGGER.debug("ACK'ed %u messages", len(msgs))
# Forward the loaded data to the "loaded" topic
for msg in msgs:
LOADED_QUEUE_PUBLISHER.publish(msg[1])
LOGGER.debug("Forwarded %u messages", len(msgs))
