def __init__(self, path, storage=None):
from datetime import datetime
self._descriptor_path = os.path.join(path, "descriptor.json")
with open(self._descriptor_path, "r") as f:
self.descriptor_json = f.read()
self._descriptor = json.loads(self.descriptor_json)
self._powerlog_path = os.path.join(path, "power.log")
with open(self._powerlog_path, "r") as f:
self._log = f.read()
api_key_str = self._descriptor["event"]["headers"]["X-Api-Key"]
self._api_key = APIKey.objects.get_or_create(api_key=api_key_str,
defaults={
"full_name":
"Test Client",
"email":
"*****@*****.**",
"website":
"https://example.org",
})[0]
auth_token_str = self._descriptor["event"]["headers"][
"Authorization"].split()[1]
self._auth_token = AuthToken.objects.get_or_create(
key=auth_token_str, creation_apikey=self._api_key)[0]
self._api_key.tokens.add(self._auth_token)
timestamp_str = self._descriptor["upload_metadata"]["match_start"][
0:16]
self._timestamp = datetime.strptime(timestamp_str, "%Y-%m-%dT%H:%M")
self._shortid = self._descriptor["shortid"]
if storage:
key = _generate_upload_key(self._timestamp, self._shortid)
os.makedirs(os.path.dirname(storage.path(key)), exist_ok=True)
with storage.open(key, mode="w") as log_file:
log_file.write(self._log)
self._reason = None
self._delete_was_called = False
def __init__(self, path, storage=None):
self._descriptor_path = os.path.join(path, "descriptor.json")
self._descriptor = json.load(open(self._descriptor_path))
self._powerlog_path = os.path.join(path, "power.log")
self._log = open(self._powerlog_path).read()
api_key_str = self._descriptor["gateway_headers"]["X-Api-Key"]
self._api_key = APIKey.objects.get_or_create(api_key=api_key_str, defaults={
"full_name": "Test Client",
"email": "*****@*****.**",
"website": "https://example.org",
})[0]
auth_token_str = self._descriptor["gateway_headers"]["Authorization"].split()[1]
self._auth_token = AuthToken.objects.get_or_create(
key=auth_token_str,
creation_apikey=self._api_key
)[0]
self._api_key.tokens.add(self._auth_token)
timestamp_str = self._descriptor["upload_metadata"]["match_start"][0:16]
self._timestamp = datetime.strptime(timestamp_str, "%Y-%m-%dT%H:%M")
self._shortid = self._descriptor["shortid"]
if storage:
key = _generate_upload_key(self._timestamp, self._shortid)
os.makedirs(os.path.dirname(storage.path(key)), exist_ok=True)
with storage.open(key, mode="w") as log_file:
log_file.write(self._log)
self._upload_event_log_bucket = None
self._upload_event_log_key = None
self._reason = None
self._delete_was_called = False
def process_raw_upload(raw_upload,
reprocess=False,
log_group_name="",
log_stream_name=""):
"""
Generic processing logic for raw log files.
"""
from ..games.serializers import UploadEventSerializer
logger = logging.getLogger("hsreplaynet.lambdas.process_raw_upload")
obj, created = UploadEvent.objects.get_or_create(
shortid=raw_upload.shortid,
defaults={"status": UploadEventStatus.PENDING})
logger.debug("UploadEvent Created: %r", created)
if not created and not reprocess:
# This can occur two ways:
# 1) The client sends the PUT request twice
# 2) Re-enabling processing queues an upload to the stream and the S3 event fires
logger.info("Invocation is an instance of double_put. Exiting Early.")
influx_metric("raw_log_double_put", {
"count": 1,
"shortid": raw_upload.shortid,
"key": raw_upload.log_key
})
return
obj.log_group_name = log_group_name
obj.log_stream_name = log_stream_name
descriptor = raw_upload.descriptor
new_log_key = _generate_upload_key(raw_upload.timestamp,
raw_upload.shortid)
new_bucket = settings.AWS_STORAGE_BUCKET_NAME
# Move power.log to the other bucket if it's needed
raw_upload.prepare_upload_event_log_location(new_bucket, new_log_key)
upload_metadata = descriptor["upload_metadata"]
event = descriptor["event"]
headers = {k.lower(): v for k, v in event.get("headers", {}).items()}
user_agent = headers.get("user-agent", "")
logger.debug("User Agent: %r", user_agent)
obj.file = new_log_key
obj.descriptor_data = raw_upload.descriptor_json
obj.upload_ip = event.get("requestContext",
{}).get("identity", {}).get("sourceIp", "")
obj.canary = "canary" in upload_metadata and upload_metadata["canary"]
obj.user_agent = user_agent[:100]
obj.status = UploadEventStatus.VALIDATING
try:
if not obj.user_agent:
raise ValidationError("Missing User-Agent header")
header = headers.get("authorization", "")
token = auth_token_from_header(header)
if not token:
msg = "Malformed or Invalid Authorization Header: %r" % (header)
logger.error(msg)
raise ValidationError(msg)
obj.token = token
api_key = headers.get("x-api-key", "")
if not api_key:
raise ValidationError(
"Missing X-Api-Key header. Please contact us for an API key.")
obj.api_key = LegacyAPIKey.objects.get(api_key=api_key)
except (ValidationError, LegacyAPIKey.DoesNotExist) as e:
logger.error("Exception: %r", e)
obj.status = UploadEventStatus.VALIDATION_ERROR
obj.error = e
obj.save()
logger.info("All state successfully saved to UploadEvent with id: %r",
obj.id)
# If we get here, now everything is in the DB.
# Clear out the raw upload so it doesn't clog up the pipeline.
raw_upload.delete()
logger.info("Deleting objects from S3 succeeded.")
logger.info(
"Validation Error will be raised and we will not proceed to processing"
)
raise
else:
if "test_data" in upload_metadata or obj.token.test_data:
logger.debug("Upload Event Is TEST DATA")
if obj.token.test_data:
# When token.test_data = True, then all UploadEvents are test_data = True
obj.test_data = True
# Only old clients released during beta do not include a user agent
is_unsupported_client = obj.user_agent.startswith(
settings.UPLOAD_USER_AGENT_BLACKLIST)
if is_unsupported_client:
logger.info(
"No UA provided. Marking as unsupported (client too old).")
influx_metric(
"upload_from_unsupported_client", {
"count": 1,
"shortid": raw_upload.shortid,
"api_key": obj.api_key.full_name
})
obj.status = UploadEventStatus.UNSUPPORTED_CLIENT
obj.save()
logger.debug("Saved: UploadEvent.id = %r", obj.id)
# If we get here, now everything is in the DB.
raw_upload.delete()
logger.debug("Deleting objects from S3 succeeded")
if is_unsupported_client:
# Wait until after we have deleted the raw_upload to exit
# But do not start processing if it's an unsupported client
logger.info("Exiting Without Processing - Unsupported Client")
return
serializer = UploadEventSerializer(obj, data=upload_metadata)
if serializer.is_valid():
logger.debug("UploadEvent passed serializer validation")
obj.status = UploadEventStatus.PROCESSING
serializer.save()
logger.debug("Starting GameReplay processing for UploadEvent")
obj.process()
else:
obj.error = serializer.errors
logger.info("UploadEvent failed validation with errors: %r", obj.error)
obj.status = UploadEventStatus.VALIDATION_ERROR
obj.save()
logger.debug("Done")
def process_raw_upload(raw_upload):
"""A method for processing a raw upload in S3.
This will usually be invoked by process_s3_create_handler, however
it can also be invoked when a raw upload is queued for reprocessing via SNS.
"""
logger = logging.getLogger("hsreplaynet.lambdas.process_raw_upload")
logger.info("Starting processing for RawUpload: %s", str(raw_upload))
descriptor = raw_upload.descriptor
new_key = _generate_upload_key(raw_upload.timestamp, raw_upload.shortid)
new_bucket = settings.AWS_STORAGE_BUCKET_NAME
# First we copy the log to the proper location
copy_source = "%s/%s" % (raw_upload.bucket, raw_upload.log_key)
logger.info("*** COPY RAW LOG TO NEW LOCATION ***")
logger.info("SOURCE: %s" % copy_source)
logger.info("DESTINATION: %s/%s" % (new_bucket, new_key))
aws.S3.copy_object(
Bucket=new_bucket,
Key=new_key,
CopySource=copy_source,
)
# Then we build the request and send it to DRF
# If "file" is a string, DRF will interpret as a S3 Key
upload_metadata = descriptor["upload_metadata"]
upload_metadata["shortid"] = descriptor["shortid"]
upload_metadata["file"] = new_key
upload_metadata["type"] = int(UploadEventType.POWER_LOG)
gateway_headers = descriptor["gateway_headers"]
headers = {
"HTTP_X_FORWARDED_FOR": descriptor["source_ip"],
"HTTP_AUTHORIZATION": gateway_headers["Authorization"],
"HTTP_X_API_KEY": gateway_headers["X-Api-Key"],
"format": "json",
}
path = descriptor["event"]["path"]
request = emulate_api_request(path, upload_metadata, headers)
try:
result = create_upload_event_from_request(request)
except Exception as e:
logger.info("Create Upload Event Failed!!")
# If DRF fails: delete the copy of the log to not leave orphans around.
aws.S3.delete_object(Bucket=new_bucket, Key=new_key)
# Now move the failed upload into the failed location for easier inspection.
raw_upload.make_failed(str(e))
logger.info("RawUpload has been marked failed: %s", str(raw_upload))
raise
else:
logger.info("Create Upload Event Success - RawUpload will be deleted.")
# If DRF returns success, then we delete the raw_upload
raw_upload.delete()
logger.info("Processing RawUpload Complete.")
return result
def process_raw_upload(raw_upload, reprocess=False, log_group_name="", log_stream_name=""):
"""
Generic processing logic for raw log files.
"""
logger = logging.getLogger("hsreplaynet.lambdas.process_raw_upload")
obj, created = UploadEvent.objects.get_or_create(
shortid=raw_upload.shortid,
defaults={"status": UploadEventStatus.PENDING}
)
logger.debug("UploadEvent Created: %r", created)
if not created and not reprocess:
# This can occur two ways:
# 1) The client sends the PUT request twice
# 2) Re-enabling processing queues an upload to the stream and the S3 event fires
logger.info("Invocation is an instance of double_put. Exiting Early.")
influx_metric("raw_log_double_put", {
"count": 1,
"shortid": raw_upload.shortid,
"key": raw_upload.log_key
})
return
obj.log_group_name = log_group_name
obj.log_stream_name = log_stream_name
descriptor = raw_upload.descriptor
new_log_key = _generate_upload_key(raw_upload.timestamp, raw_upload.shortid)
new_bucket = settings.AWS_STORAGE_BUCKET_NAME
# Move power.log/descriptor.json to the other bucket if it's needed
raw_upload.prepare_upload_event_log_location(new_bucket, new_log_key)
upload_metadata = descriptor["upload_metadata"]
gateway_headers = descriptor["gateway_headers"]
if "User-Agent" in gateway_headers:
logger.debug("User Agent: %s", gateway_headers["User-Agent"])
else:
logger.debug("User Agent: UNKNOWN")
obj.file = new_log_key
obj.descriptor_data = json.dumps(descriptor)
obj.upload_ip = descriptor["source_ip"]
obj.canary = "canary" in upload_metadata and upload_metadata["canary"]
obj.user_agent = gateway_headers.get("User-Agent", "")[:100]
obj.status = UploadEventStatus.VALIDATING
try:
header = gateway_headers.get("Authorization", "")
token = AuthToken.get_token_from_header(header)
if not token:
msg = "Malformed or Invalid Authorization Header: %r" % (header)
logger.error(msg)
raise ValidationError(msg)
obj.token = token
api_key = gateway_headers.get("X-Api-Key", "")
if not api_key:
raise ValidationError("Missing X-Api-Key header. Please contact us for an API key.")
obj.api_key = APIKey.objects.get(api_key=api_key)
except (ValidationError, APIKey.DoesNotExist) as e:
logger.error("Exception: %r", e)
obj.status = UploadEventStatus.VALIDATION_ERROR
obj.error = e
obj.save()
logger.info("All state successfully saved to UploadEvent with id: %r", obj.id)
# If we get here, now everything is in the DB.
# Clear out the raw upload so it doesn't clog up the pipeline.
raw_upload.delete()
logger.info("Deleting objects from S3 succeeded.")
logger.info("Validation Error will be raised and we will not proceed to processing")
raise
else:
if "test_data" in upload_metadata or obj.token.test_data:
logger.debug("Upload Event Is TEST DATA")
if obj.token.test_data:
# When token.test_data = True, then all UploadEvents are test_data = True
obj.test_data = True
# Only old clients released during beta do not include a user agent
is_unsupported_client = not obj.user_agent
if is_unsupported_client:
logger.info("No UA provided. Marking as unsupported (client too old).")
influx_metric("upload_from_unsupported_client", {
"count": 1,
"shortid": raw_upload.shortid,
"api_key": obj.api_key.full_name
})
obj.status = UploadEventStatus.UNSUPPORTED_CLIENT
obj.save()
logger.debug("Saved: UploadEvent.id = %r", obj.id)
# If we get here, now everything is in the DB.
raw_upload.delete()
logger.debug("Deleting objects from S3 succeeded")
if is_unsupported_client:
# Wait until after we have deleted the raw_upload to exit
# But do not start processing if it's an unsupported client
logger.info("Exiting Without Processing - Unsupported Client")
return
serializer = UploadEventSerializer(obj, data=upload_metadata)
if serializer.is_valid():
logger.debug("UploadEvent passed serializer validation")
obj.status = UploadEventStatus.PROCESSING
serializer.save()
logger.debug("Starting GameReplay processing for UploadEvent")
obj.process()
else:
obj.error = serializer.errors
logger.info("UploadEvent failed validation with errors: %r", obj.error)
obj.status = UploadEventStatus.VALIDATION_ERROR
obj.save()
logger.debug("Done")
def process_raw_upload(raw_upload,
reprocess=False,
log_group_name="",
log_stream_name=""):
"""
Generic processing logic for raw log files.
"""
logger = logging.getLogger("hsreplaynet.lambdas.process_raw_upload")
obj, created = UploadEvent.objects.get_or_create(
shortid=raw_upload.shortid,
defaults={"status": UploadEventStatus.PENDING})
logger.info("The created flag for this upload event is: %r", created)
if not created and not reprocess:
# This can occur two ways:
# 1) The client sends the PUT request twice
# 2) Re-enabling processing queues an upload to the stream and the S3 event fires
logger.info("Invocation is an instance of double_put. Exiting Early.")
influx_metric("raw_log_double_put", {
"count": 1,
"shortid": raw_upload.shortid,
"key": raw_upload.log_key
})
return
obj.log_group_name = log_group_name
obj.log_stream_name = log_stream_name
descriptor = raw_upload.descriptor
new_log_key = _generate_upload_key(raw_upload.timestamp,
raw_upload.shortid)
new_descriptor_key = _generate_upload_key(raw_upload.timestamp,
raw_upload.shortid,
"descriptor.json")
new_bucket = settings.AWS_STORAGE_BUCKET_NAME
# Move power.log/descriptor.json to the other bucket if it's needed
raw_upload.prepare_upload_event_log_location(new_bucket, new_log_key,
new_descriptor_key)
upload_metadata = descriptor["upload_metadata"]
gateway_headers = descriptor["gateway_headers"]
if "User-Agent" in gateway_headers:
logger.info("The uploading user agent is: %s",
gateway_headers["User-Agent"])
else:
logger.info("A User-Agent header was not provided.")
obj.file = new_log_key
obj.descriptor = new_descriptor_key
obj.upload_ip = descriptor["source_ip"]
obj.canary = "canary" in upload_metadata and upload_metadata["canary"]
obj.user_agent = gateway_headers.get("User-Agent", "")[:100]
obj.status = UploadEventStatus.VALIDATING
try:
header = gateway_headers["Authorization"]
token = AuthToken.get_token_from_header(header)
if not token:
msg = "Malformed or Invalid Authorization Header: %r" % (header)
logger.error(msg)
raise Exception(msg)
obj.token = token
obj.api_key = APIKey.objects.get(api_key=gateway_headers["X-Api-Key"])
except Exception as e:
logger.error("Exception: %r", e)
obj.status = UploadEventStatus.VALIDATION_ERROR
obj.error = e
obj.save()
logger.info("All state successfully saved to UploadEvent with id: %r",
obj.id)
# If we get here, now everything is in the DB.
# Clear out the raw upload so it doesn't clog up the pipeline.
raw_upload.delete()
logger.info("Deleting objects from S3 succeeded.")
logger.info(
"Validation Error will be raised and we will not proceed to processing"
)
raise
else:
if "test_data" in upload_metadata or obj.token.test_data:
logger.info("Upload Event Is TEST DATA")
if obj.token.test_data:
# When token.test_data = True, then all UploadEvents are test_data = True
obj.test_data = True
# Only old clients released during beta do not include a user agent
is_unsupported_client = not obj.user_agent
if is_unsupported_client:
logger.info(
"No UA provided. Marking as unsupported (client too old).")
influx_metric(
"upload_from_unsupported_client", {
"count": 1,
"shortid": raw_upload.shortid,
"api_key": obj.api_key.full_name
})
obj.status = UploadEventStatus.UNSUPPORTED_CLIENT
obj.save()
logger.info("All state successfully saved to UploadEvent with id: %r",
obj.id)
# If we get here, now everything is in the DB.
raw_upload.delete()
logger.info("Deleting objects from S3 succeeded")
if is_unsupported_client:
# Wait until after we have deleted the raw_upload to exit
# But do not start processing if it's an unsupported client
logger.info("Exiting Without Processing - Unsupported Client")
return
serializer = UploadEventSerializer(obj, data=upload_metadata)
if serializer.is_valid():
logger.info("UploadEvent passed serializer validation")
obj.status = UploadEventStatus.PROCESSING
serializer.save()
logger.info("Starting GameReplay processing for UploadEvent")
obj.process()
else:
obj.error = serializer.errors
logger.info("UploadEvent failed validation with errors: %r", obj.error)
obj.status = UploadEventStatus.VALIDATION_ERROR
obj.save()
logger.info("RawUpload event processing is complete")
