def process_insert(
self,
event: Mapping[str, Any],
metadata: Optional[KafkaMessageMetadata] = None
) -> Optional[Mapping[str, Any]]:
if not self._should_process(event):
return None
processed = {"deleted": 0}
extract_project_id(processed, event)
self._extract_event_id(processed, event)
processed["retention_days"] = enforce_retention(
event,
datetime.strptime(event["datetime"],
settings.PAYLOAD_DATETIME_FORMAT),
)
self.extract_required(processed, event)
data = event.get("data", {})
# HACK: https://sentry.io/sentry/snuba/issues/802102397/
if not data:
logger.error("No data for event: %s", event, exc_info=True)
return None
self.extract_common(processed, event, metadata)
self.extract_custom(processed, event, metadata)
sdk = data.get("sdk", None) or {}
self.extract_sdk(processed, sdk)
tags = _as_dict_safe(data.get("tags", None))
self.extract_promoted_tags(processed, tags)
self.extract_tags_custom(processed, event, tags, metadata)
contexts = data.get("contexts", None) or {}
self.extract_promoted_contexts(processed, contexts, tags)
self.extract_contexts_custom(processed, event, contexts, metadata)
processed["contexts.key"], processed[
"contexts.value"] = extract_extra_contexts(contexts)
processed["tags.key"], processed["tags.value"] = extract_extra_tags(
tags)
processed["_tags_flattened"] = flatten_nested_field(
processed["tags.key"], processed["tags.value"])
exception = (data.get("exception",
data.get("sentry.interfaces.Exception", None))
or {})
stacks = exception.get("values", None) or []
self.extract_stacktraces(processed, stacks)
if metadata is not None:
processed["offset"] = metadata.offset
processed["partition"] = metadata.partition
return processed
def __fill_common(self, span: MutableMapping[str, Any],
data: Mapping[str, Any]) -> None:
"""
Fills in the fields that have the same structure between transactions and spans
but that come from a different dictionary.
"""
span["start_ts"], span["start_ns"] = self.__extract_timestamp(
data["start_timestamp"], )
span["finish_ts"], span["finish_ns"] = self.__extract_timestamp(
data["timestamp"], )
duration_secs = (span["finish_ts"] - span["start_ts"]).total_seconds()
# duration is in milliseconds
span["duration_ms"] = max(int(duration_secs * 1000), 0)
tags = _as_dict_safe(data.get("tags", None))
span["tags.key"], span["tags.value"] = extract_extra_tags(tags)
def test_extract_tags(self):
orig_tags = {
"sentry:user": "******",
"level": "the_level",
"logger": "the_logger",
"server_name": "the_servername",
"transaction": "the_transaction",
"environment": "the_enviroment",
"sentry:release": "the_release",
"sentry:dist": "the_dist",
"site": "the_site",
"url": "the_url",
"extra_tag": "extra_value",
"null_tag": None,
}
tags = orig_tags.copy()
output = {}
enforce_table_writer(self.dataset).get_stream_loader().get_processor(
).extract_promoted_tags(output, tags)
assert output == {
"sentry:dist": "the_dist",
"environment": u"the_enviroment",
"level": u"the_level",
"logger": u"the_logger",
"sentry:release": "the_release",
"server_name": u"the_servername",
"site": u"the_site",
"transaction": u"the_transaction",
"url": u"the_url",
"sentry:user": u"the_user",
}
assert tags == orig_tags
extra_output = {}
extra_output["tags.key"], extra_output[
"tags.value"] = extract_extra_tags(tags)
valid_items = [(k, v) for k, v in sorted(orig_tags.items()) if v]
assert extra_output == {
"tags.key": [k for k, v in valid_items],
"tags.value": [v for k, v in valid_items],
}
def _process_tags(
self,
processed: MutableMapping[str, Any],
event_dict: EventDict,
) -> None:
tags: Mapping[str, Any] = _as_dict_safe(event_dict["data"].get(
"tags", None))
processed["tags.key"], processed["tags.value"] = extract_extra_tags(
tags)
promoted_tags = {
col: tags[col]
for col in self.PROMOTED_TAGS if col in tags
}
processed["release"] = promoted_tags.get(
"sentry:release",
event_dict.get("release"),
)
processed["environment"] = promoted_tags.get("environment")
processed["user"] = promoted_tags.get("sentry:user", "")
processed["dist"] = _unicodify(
promoted_tags.get("sentry:dist", event_dict["data"].get("dist")), )
def process_message(self, message, metadata) -> Optional[ProcessedMessage]:
processed = {"deleted": 0}
if not (isinstance(message, (list, tuple)) and len(message) >= 2):
return None
version = message[0]
if version not in (0, 1, 2):
return None
type_, event = message[1:3]
if type_ != "insert":
return None
data = event["data"]
event_type = data.get("type")
if event_type != "transaction":
return None
extract_base(processed, event)
processed["retention_days"] = enforce_retention(
event,
datetime.fromtimestamp(data["timestamp"]),
)
if not data.get("contexts", {}).get("trace"):
return None
transaction_ctx = data["contexts"]["trace"]
trace_id = transaction_ctx["trace_id"]
try:
processed["event_id"] = str(uuid.UUID(processed["event_id"]))
processed["trace_id"] = str(uuid.UUID(trace_id))
processed["span_id"] = int(transaction_ctx["span_id"], 16)
processed["transaction_op"] = _unicodify(
transaction_ctx.get("op") or "")
processed["transaction_name"] = _unicodify(
data.get("transaction") or "")
processed[
"start_ts"], processed["start_ms"] = self.__extract_timestamp(
data["start_timestamp"], )
status = transaction_ctx.get("status", None)
if status:
int_status = SPAN_STATUS_NAME_TO_CODE.get(
status, UNKNOWN_SPAN_STATUS)
else:
int_status = UNKNOWN_SPAN_STATUS
processed["transaction_status"] = int_status
if data["timestamp"] - data["start_timestamp"] < 0:
# Seems we have some negative durations in the DB
metrics.increment("negative_duration")
except Exception:
# all these fields are required but we saw some events go through here
# in the past. For now bail.
return
processed["finish_ts"], processed[
"finish_ms"] = self.__extract_timestamp(data["timestamp"], )
duration_secs = (processed["finish_ts"] -
processed["start_ts"]).total_seconds()
processed["duration"] = max(int(duration_secs * 1000), 0)
processed["platform"] = _unicodify(event["platform"])
tags = _as_dict_safe(data.get("tags", None))
processed["tags.key"], processed["tags.value"] = extract_extra_tags(
tags)
processed["_tags_flattened"] = flatten_nested_field(
processed["tags.key"], processed["tags.value"])
promoted_tags = {
col: tags[col]
for col in self.PROMOTED_TAGS if col in tags
}
processed["release"] = promoted_tags.get(
"sentry:release",
event.get("release"),
)
processed["environment"] = promoted_tags.get("environment")
contexts = _as_dict_safe(data.get("contexts", None))
user_dict = data.get("user", data.get("sentry.interfaces.User",
None)) or {}
geo = user_dict.get("geo", None) or {}
if "geo" not in contexts and isinstance(geo, dict):
contexts["geo"] = geo
measurements = data.get("measurements")
if measurements is not None:
try:
(
processed["measurements.key"],
processed["measurements.value"],
) = extract_nested(measurements,
lambda value: float(value["value"]))
except Exception:
# Not failing the event in this case just yet, because we are still
# developing this feature.
logger.error(
"Invalid measurements field.",
extra={"measurements": measurements},
exc_info=True,
)
request = data.get("request", data.get("sentry.interfaces.Http",
None)) or {}
http_data: MutableMapping[str, Any] = {}
extract_http(http_data, request)
processed["http_method"] = http_data["http_method"]
processed["http_referer"] = http_data["http_referer"]
processed["contexts.key"], processed[
"contexts.value"] = extract_extra_contexts(contexts)
processed["_contexts_flattened"] = flatten_nested_field(
processed["contexts.key"], processed["contexts.value"])
processed["dist"] = _unicodify(
promoted_tags.get("sentry:dist", data.get("dist")), )
user_data = {}
extract_user(user_data, user_dict)
processed["user"] = promoted_tags.get("sentry:user", "")
processed["user_name"] = user_data["username"]
processed["user_id"] = user_data["user_id"]
processed["user_email"] = user_data["email"]
ip_address = _ensure_valid_ip(user_data["ip_address"])
if ip_address:
if ip_address.version == 4:
processed["ip_address_v4"] = str(ip_address)
elif ip_address.version == 6:
processed["ip_address_v6"] = str(ip_address)
processed["partition"] = metadata.partition
processed["offset"] = metadata.offset
sdk = data.get("sdk", None) or {}
processed["sdk_name"] = _unicodify(sdk.get("name") or "")
processed["sdk_version"] = _unicodify(sdk.get("version") or "")
if processed["sdk_name"] == "":
metrics.increment("missing_sdk_name")
if processed["sdk_version"] == "":
metrics.increment("missing_sdk_version")
return InsertBatch([processed])
def process_message(
self, message: Tuple[int, str, Any],
metadata: KafkaMessageMetadata) -> Optional[ProcessedMessage]:
processed: MutableMapping[str, Any] = {"deleted": 0}
if not (isinstance(message, (list, tuple)) and len(message) >= 2):
return None
version = message[0]
if version not in (0, 1, 2):
return None
type_, event = message[1:3]
if type_ != "insert":
return None
data = event["data"]
event_type = data.get("type")
if event_type != "transaction":
return None
extract_base(processed, event)
try:
# We are purposely using a naive datetime here to work with the
# rest of the codebase. We can be confident that clients are only
# sending UTC dates.
processed["retention_days"] = enforce_retention(
event,
datetime.utcfromtimestamp(data["timestamp"]),
)
except EventTooOld:
return None
if not data.get("contexts", {}).get("trace"):
return None
transaction_ctx = data["contexts"]["trace"]
trace_id = transaction_ctx["trace_id"]
processed["event_id"] = str(uuid.UUID(processed["event_id"]))
processed["trace_id"] = str(uuid.UUID(trace_id))
processed["span_id"] = int(transaction_ctx["span_id"], 16)
processed["transaction_op"] = _unicodify(
transaction_ctx.get("op") or "")
processed["transaction_name"] = _unicodify(
data.get("transaction") or "")
processed["start_ts"], processed[
"start_ms"] = self.__extract_timestamp(data["start_timestamp"], )
status = transaction_ctx.get("status", None)
if status:
int_status = SPAN_STATUS_NAME_TO_CODE.get(status,
UNKNOWN_SPAN_STATUS)
else:
int_status = UNKNOWN_SPAN_STATUS
processed["transaction_status"] = int_status
if data["timestamp"] - data["start_timestamp"] < 0:
# Seems we have some negative durations in the DB
metrics.increment("negative_duration")
processed["finish_ts"], processed[
"finish_ms"] = self.__extract_timestamp(data["timestamp"], )
duration_secs = (processed["finish_ts"] -
processed["start_ts"]).total_seconds()
processed["duration"] = max(int(duration_secs * 1000), 0)
processed["platform"] = _unicodify(event["platform"])
tags: Mapping[str, Any] = _as_dict_safe(data.get("tags", None))
processed["tags.key"], processed["tags.value"] = extract_extra_tags(
tags)
promoted_tags = {
col: tags[col]
for col in self.PROMOTED_TAGS if col in tags
}
processed["release"] = promoted_tags.get(
"sentry:release",
event.get("release"),
)
processed["environment"] = promoted_tags.get("environment")
contexts: MutableMapping[str, Any] = _as_dict_safe(
data.get("contexts", None))
user_dict = data.get("user", data.get("sentry.interfaces.User",
None)) or {}
geo = user_dict.get("geo", None) or {}
if "geo" not in contexts and isinstance(geo, dict):
contexts["geo"] = geo
measurements = data.get("measurements")
if measurements is not None:
try:
(
processed["measurements.key"],
processed["measurements.value"],
) = extract_nested(
measurements,
lambda value: float(value["value"])
if (value is not None and isinstance(
value.get("value"), numbers.Number)) else None,
)
except Exception:
# Not failing the event in this case just yet, because we are still
# developing this feature.
logger.error(
"Invalid measurements field.",
extra={"measurements": measurements},
exc_info=True,
)
breakdowns = data.get("breakdowns")
if breakdowns is not None:
span_op_breakdowns = breakdowns.get("span_ops")
if span_op_breakdowns is not None:
try:
(
processed["span_op_breakdowns.key"],
processed["span_op_breakdowns.value"],
) = extract_nested(
span_op_breakdowns,
lambda value: float(value["value"])
if (value is not None and isinstance(
value.get("value"), numbers.Number)) else None,
)
except Exception:
# Not failing the event in this case just yet, because we are still
# developing this feature.
logger.error(
"Invalid breakdowns.span_ops field.",
extra={"span_op_breakdowns": span_op_breakdowns},
exc_info=True,
)
request = data.get("request", data.get("sentry.interfaces.Http",
None)) or {}
http_data: MutableMapping[str, Any] = {}
extract_http(http_data, request)
processed["http_method"] = http_data["http_method"]
processed["http_referer"] = http_data["http_referer"]
skipped_contexts = settings.TRANSACT_SKIP_CONTEXT_STORE.get(
processed["project_id"], set())
for context in skipped_contexts:
if context in contexts:
del contexts[context]
processed["contexts.key"], processed[
"contexts.value"] = extract_extra_contexts(contexts)
processed["dist"] = _unicodify(
promoted_tags.get("sentry:dist", data.get("dist")), )
user_data: MutableMapping[str, Any] = {}
extract_user(user_data, user_dict)
processed["user"] = promoted_tags.get("sentry:user", "")
processed["user_name"] = user_data["username"]
processed["user_id"] = user_data["user_id"]
processed["user_email"] = user_data["email"]
ip_address = _ensure_valid_ip(user_data["ip_address"])
if ip_address:
if ip_address.version == 4:
processed["ip_address_v4"] = str(ip_address)
elif ip_address.version == 6:
processed["ip_address_v6"] = str(ip_address)
processed["partition"] = metadata.partition
processed["offset"] = metadata.offset
sdk = data.get("sdk", None) or {}
processed["sdk_name"] = _unicodify(sdk.get("name") or "")
processed["sdk_version"] = _unicodify(sdk.get("version") or "")
if processed["sdk_name"] == "":
metrics.increment("missing_sdk_name")
if processed["sdk_version"] == "":
metrics.increment("missing_sdk_version")
return InsertBatch([processed], None)
def process_message(self, message, metadata) -> Optional[ProcessedMessage]:
if not (isinstance(message, (list, tuple)) and len(message) >= 2):
return None
version = message[0]
if version not in (0, 1, 2):
return None
type_, event = message[1:3]
if type_ != "insert":
return None
data = event["data"]
event_type = data.get("type")
if event_type != "transaction":
return None
ret: List[MutableMapping[str, Any]] = []
# Add the transaction span
transaction_ctx = data["contexts"].get("trace")
if not transaction_ctx:
metrics.increment("missing_trace_ctx")
return None
# Add the transaction root span
processed = self.__init_span(event)
processed["span_id"] = self.__safe_extract_int(
"transaction:span_id", transaction_ctx["span_id"], False)
if processed["span_id"] is None:
return None
processed["transaction_name"] = _unicodify(
data.get("transaction") or "")
processed["parent_span_id"] = self.__safe_extract_int(
"transaction:parent_span_id",
transaction_ctx.get("parent_span_id"), True)
processed["description"] = _unicodify(data.get("transaction") or "")
processed["op"] = _unicodify(transaction_ctx.get("op") or "")
status = transaction_ctx.get("status", None)
self.__fill_status(processed, status)
self.__fill_common(processed, event["data"])
ret.append(processed)
spans = data.get("spans", [])
for span in spans:
processed = self.__init_span(event)
processed["span_id"] = self.__safe_extract_int(
"span:span_id", span["span_id"], False)
if processed["span_id"] is None:
return None
processed["parent_span_id"] = self.__safe_extract_int(
"span:parent_span_id", span.get("parent_span_id"), True)
processed["description"] = span.get("description", "") or ""
processed["op"] = span["op"]
tags = _as_dict_safe(span.get("tags", None))
processed["tags.key"], processed[
"tags.value"] = extract_extra_tags(tags)
status = span.get("status", None)
self.__fill_status(processed, status)
self.__fill_common(processed, span)
ret.append(processed)
if ret:
return InsertBatch(ret)
else:
return None
