def extract_custom(
self,
output: MutableMapping[str, Any],
event: InsertEvent,
metadata: KafkaMessageMetadata,
) -> None:
data = event.get("data", {})
output["message"] = _unicodify(event["message"])
# USER REQUEST GEO
user = data.get("user", data.get("sentry.interfaces.User", None)) or {}
extract_user(output, user)
geo = user.get("geo", None) or {}
self.extract_geo(output, geo)
request = data.get("request", data.get("sentry.interfaces.Http",
None)) or {}
http_data: MutableMapping[str, Any] = {}
extract_http(http_data, request)
output["http_method"] = http_data["http_method"]
output["http_referer"] = http_data["http_referer"]
output["primary_hash"] = _hashify(event["primary_hash"])
output["hierarchical_hashes"] = list(
_hashify(x) for x in data.get("hierarchical_hashes") or ())
output["culprit"] = _unicodify(data.get("culprit", None))
output["type"] = _unicodify(data.get("type", None))
output["title"] = _unicodify(data.get("title", None))
def extract_custom(
self,
output: MutableMapping[str, Any],
event: InsertEvent,
metadata: KafkaMessageMetadata,
) -> None:
data = event.get("data", {})
user_dict = data.get("user", data.get("sentry.interfaces.User",
None)) or {}
user_data: MutableMapping[str, Any] = {}
extract_user(user_data, user_dict)
output["user_name"] = user_data["username"]
output["user_id"] = user_data["user_id"]
output["user_email"] = user_data["email"]
ip_address = _ensure_valid_ip(user_data["ip_address"])
if ip_address:
if ip_address.version == 4:
output["ip_address_v4"] = str(ip_address)
elif ip_address.version == 6:
output["ip_address_v6"] = str(ip_address)
contexts: MutableMapping[str, Any] = _as_dict_safe(
data.get("contexts", None))
geo = user_dict.get("geo", {})
if "geo" not in contexts and isinstance(geo, dict):
contexts["geo"] = geo
request = data.get("request", data.get("sentry.interfaces.Http",
None)) or {}
http_data: MutableMapping[str, Any] = {}
extract_http(http_data, request)
output["http_method"] = http_data["http_method"]
output["http_referer"] = http_data["http_referer"]
# _as_dict_safe may not return a reference to the entry in the data
# dictionary in some cases.
data["contexts"] = contexts
output["message"] = _unicodify(event["message"])
output["primary_hash"] = str(uuid.UUID(_hashify(
event["primary_hash"])))
output["hierarchical_hashes"] = list(
str(uuid.UUID(_hashify(x)))
for x in data.get("hierarchical_hashes") or ())
output["culprit"] = _unicodify(data.get("culprit", ""))
output["type"] = _unicodify(data.get("type", ""))
output["title"] = _unicodify(data.get("title", ""))
def test_extract_user(self):
user = {
"id": "user_id",
"email": "user_email",
"username": "******",
"ip_address": "127.0.0.2",
}
output = {}
extract_user(output, user)
assert output == {
"email": u"user_email",
"ip_address": u"127.0.0.2",
"user_id": u"user_id",
"username": u"user_username",
}
def extract_custom(
self,
output: MutableMapping[str, Any],
event: Mapping[str, Any],
metadata: Optional[KafkaMessageMetadata] = None,
) -> None:
data = event.get("data", {})
user_dict = data.get("user", data.get("sentry.interfaces.User",
None)) or {}
user_data: MutableMapping[str, Any] = {}
extract_user(user_data, user_dict)
output["user_name"] = user_data["username"]
output["user_id"] = user_data["user_id"]
output["user_email"] = user_data["email"]
ip_address = _ensure_valid_ip(user_data["ip_address"])
if ip_address:
if ip_address.version == 4:
output["ip_address_v4"] = str(ip_address)
elif ip_address.version == 6:
output["ip_address_v6"] = str(ip_address)
contexts = _as_dict_safe(data.get("contexts", None))
geo = user_dict.get("geo", {})
if "geo" not in contexts and isinstance(geo, dict):
contexts["geo"] = geo
request = data.get("request", data.get("sentry.interfaces.Http",
None)) or {}
if "request" not in contexts and isinstance(request, dict):
http = {}
http["http_method"] = _unicodify(request.get("method", None))
http_headers = _as_dict_safe(request.get("headers", None))
http["http_referer"] = _unicodify(http_headers.get(
"Referer", None))
contexts["request"] = http
# _as_dict_safe may not return a reference to the entry in the data
# dictionary in some cases.
data["contexts"] = contexts
output["message"] = _unicodify(event["message"])
output["org_id"] = event["organization_id"]
def extract_custom(
self,
output: MutableMapping[str, Any],
event: Mapping[str, Any],
metadata: Optional[KafkaMessageMetadata] = None,
) -> None:
data = event.get("data", {})
# The following concerns the change to message/search_message
# There are 2 Scenarios:
# Pre-rename:
# - Payload contains:
# "message": "a long search message"
# - Does NOT contain a `search_message` property
# - "message" value saved in `message` column
# - `search_message` column nonexistent or Null
# Post-rename:
# - Payload contains:
# "search_message": "a long search message"
# - Optionally the payload's "data" dict (event body) contains:
# "message": "short message"
# - "search_message" value stored in `search_message` column
# - "message" value stored in `message` column
#
output["search_message"] = _unicodify(event.get(
"search_message", None))
if output["search_message"] is None:
# Pre-rename scenario, we expect to find "message" at the top level
output["message"] = _unicodify(event["message"])
else:
# Post-rename scenario, we check in case we have the optional
# "message" in the event body.
output["message"] = _unicodify(data.get("message", None))
# USER REQUEST GEO
user = data.get("user", data.get("sentry.interfaces.User", None)) or {}
extract_user(output, user)
geo = user.get("geo", None) or {}
self.extract_geo(output, geo)
http = data.get("request", data.get("sentry.interfaces.Http",
None)) or {}
self.extract_http(output, http)
def _process_user(
self,
processed: MutableMapping[str, Any],
event_dict: EventDict,
) -> None:
user_dict = (event_dict["data"].get(
"user", event_dict["data"].get("sentry.interfaces.User", None))
or {})
user_data: MutableMapping[str, Any] = {}
extract_user(user_data, user_dict)
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)
def extract_custom(
self,
output: MutableMapping[str, Any],
event: InsertEvent,
metadata: KafkaMessageMetadata,
) -> None:
data = event.get("data", {})
output["message"] = _unicodify(event["message"])
# USER REQUEST GEO
user = data.get("user", data.get("sentry.interfaces.User", None)) or {}
extract_user(output, user)
geo = user.get("geo", None) or {}
self.extract_geo(output, geo)
request = data.get("request", data.get("sentry.interfaces.Http",
None)) or {}
http_data: MutableMapping[str, Any] = {}
extract_http(http_data, request)
output["http_method"] = http_data["http_method"]
output["http_referer"] = http_data["http_referer"]
def _process_contexts_and_user(
self,
processed: MutableMapping[str, Any],
event_dict: EventDict,
) -> None:
contexts: MutableMapping[str,
Any] = _as_dict_safe(event_dict["data"].get(
"contexts", None))
user_dict = (event_dict["data"].get(
"user", event_dict["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
skipped_contexts = settings.TRANSACT_SKIP_CONTEXT_STORE.get(
processed["project_id"], set())
for context in skipped_contexts:
if context in contexts:
del contexts[context]
sanitized_contexts = self._sanitize_contexts(processed, event_dict)
processed["contexts.key"], processed[
"contexts.value"] = extract_extra_contexts(sanitized_contexts)
user_data: MutableMapping[str, Any] = {}
extract_user(user_data, user_dict)
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)
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)
