예제 #1
0
 def extract_contexts_custom(
     self,
     output: MutableMapping[str, Any],
     event: Mapping[str, Any],
     contexts: Mapping[str, Any],
     metadata: Optional[KafkaMessageMetadata] = None,
 ) -> None:
     key, value = extract_extra_contexts(contexts)
     output["_contexts_flattened"] = flatten_nested_field(key, value)
예제 #2
0
    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
예제 #3
0
    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)
예제 #4
0
    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])
예제 #5
0
    def test_extract_contexts(self):
        contexts = {
            "app": {
                "device_app_hash": "the_app_device_uuid"
            },
            "os": {
                "name": "the_os_name",
                "version": "the_os_version",
                "rooted": True,
                "build": "the_os_build",
                "kernel_version": "the_os_kernel_version",
            },
            "runtime": {
                "name": "the_runtime_name",
                "version": "the_runtime_version"
            },
            "browser": {
                "name": "the_browser_name",
                "version": "the_browser_version"
            },
            "device": {
                "model": "the_device_model",
                "family": "the_device_family",
                "name": "the_device_name",
                "brand": "the_device_brand",
                "locale": "the_device_locale",
                "uuid": "the_device_uuid",
                "model_id": "the_device_model_id",
                "arch": "the_device_arch",
                "battery_level": 30,
                "orientation": "the_device_orientation",
                "simulator": False,
                "online": True,
                "charging": True,
            },
            "extra": {
                "type": "extra",  # unnecessary
                "null": None,
                "int": 0,
                "float": 1.3,
                "list": [1, 2, 3],
                "dict": {
                    "key": "value"
                },
                "str": "string",
                "\ud83c": "invalid utf-8 surrogate",
            },
        }
        orig_tags = {
            "app.device": "the_app_device_uuid",
            "os": "the_os_name the_os_version",
            "os.name": "the_os_name",
            "os.rooted": True,
            "runtime": "the_runtime_name the_runtime_version",
            "runtime.name": "the_runtime_name",
            "browser": "the_browser_name the_browser_version",
            "browser.name": "the_browser_name",
            "device": "the_device_model",
            "device.family": "the_device_family",
            "extra_tag": "extra_value",
        }
        tags = orig_tags.copy()
        output = {}

        enforce_table_writer(self.dataset).get_stream_loader().get_processor(
        ).extract_promoted_contexts(output, contexts, tags)

        assert output == {
            "app_device": u"the_app_device_uuid",
            "browser": u"the_browser_name the_browser_version",
            "browser_name": u"the_browser_name",
            "device": u"the_device_model",
            "device_arch": u"the_device_arch",
            "device_battery_level": 30.0,
            "device_brand": u"the_device_brand",
            "device_charging": True,
            "device_family": u"the_device_family",
            "device_locale": u"the_device_locale",
            "device_model_id": u"the_device_model_id",
            "device_name": u"the_device_name",
            "device_online": True,
            "device_orientation": u"the_device_orientation",
            "device_simulator": False,
            "device_uuid": u"the_device_uuid",
            "os": u"the_os_name the_os_version",
            "os_build": u"the_os_build",
            "os_kernel_version": u"the_os_kernel_version",
            "os_name": u"the_os_name",
            "os_rooted": True,
            "runtime": u"the_runtime_name the_runtime_version",
            "runtime_name": u"the_runtime_name",
        }
        assert contexts == {
            "app": {},
            "browser": {},
            "device": {},
            "extra": {
                "dict": {
                    "key": "value"
                },
                "\ud83c": "invalid utf-8 surrogate",
                "float": 1.3,
                "int": 0,
                "list": [1, 2, 3],
                "null": None,
                "type": "extra",
                "str": "string",
            },
            "os": {},
            "runtime": {},
        }
        assert tags == orig_tags

        extra_output = {}
        (
            extra_output["contexts.key"],
            extra_output["contexts.value"],
        ) = extract_extra_contexts(contexts)

        assert extra_output == {
            "contexts.key":
            ["extra.int", "extra.float", "extra.str", "extra.\\ud83c"],
            "contexts.value":
            [u"0", u"1.3", u"string", u"invalid utf-8 surrogate"],
        }
예제 #6
0
    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)