feat(use-case): Add visibility modifiers to use cases and use internal MRI parser (#67774)

Author: Riccardo Busetti

src/sentry/search/events/builder/metrics.py

@@ -52,7 +52,7 @@
     WhereType,
 )
 from sentry.sentry_metrics import indexer
-from sentry.sentry_metrics.use_case_id_registry import UseCaseID, extract_use_case_id
+from sentry.sentry_metrics.use_case_id_registry import UseCaseID
 from sentry.snuba.dataset import Dataset
 from sentry.snuba.discover import create_result_key
 from sentry.snuba.metrics.extraction import (
@@ -64,6 +64,7 @@
     should_use_on_demand_metrics_for_querying,
 )
 from sentry.snuba.metrics.fields import histogram as metrics_histogram
+from sentry.snuba.metrics.naming_layer.mri import extract_use_case_id
 from sentry.snuba.metrics.query import (
     MetricField,
     MetricGroupByField,

src/sentry/sentry_metrics/aggregation_option_registry.py

@@ -1,7 +1,8 @@
 from enum import Enum
 
 from sentry import options
-from sentry.sentry_metrics.use_case_id_registry import UseCaseID, extract_use_case_id
+from sentry.sentry_metrics.use_case_id_registry import UseCaseID
+from sentry.snuba.metrics.naming_layer.mri import extract_use_case_id
 
 
 class AggregationOption(Enum):

src/sentry/sentry_metrics/consumers/indexer/batch.py

@@ -27,7 +27,8 @@
 from sentry.sentry_metrics.consumers.indexer.parsed_message import ParsedMessage
 from sentry.sentry_metrics.consumers.indexer.routing_producer import RoutingPayload
 from sentry.sentry_metrics.indexer.base import Metadata
-from sentry.sentry_metrics.use_case_id_registry import UseCaseID, extract_use_case_id
+from sentry.sentry_metrics.use_case_id_registry import UseCaseID
+from sentry.snuba.metrics.naming_layer.mri import extract_use_case_id
 from sentry.utils import json, metrics
 
 logger = logging.getLogger(__name__)

src/sentry/sentry_metrics/use_case_id_registry.py

@@ -1,14 +1,18 @@
 from __future__ import annotations
 
-import re
 from collections.abc import Mapping
 from enum import Enum
 
-from sentry_kafka_schemas.codecs import ValidationError
-
 from sentry.sentry_metrics.configuration import UseCaseKey
 
-MRI_RE_PATTERN = re.compile("^([c|s|d|g|e]):([a-zA-Z0-9_]+)/.*$")
+
+class UseCaseIDAPIAccess(Enum):
+    """
+    Represents the access levels of a UseCaseID for sentry's APIs.
+    """
+
+    PUBLIC = 0
+    PRIVATE = 1
 
 
 class UseCaseID(Enum):
@@ -22,6 +26,17 @@ class UseCaseID(Enum):
     METRIC_STATS = "metric_stats"
 
 
+USE_CASE_ID_API_ACCESSES: Mapping[UseCaseID, UseCaseIDAPIAccess] = {
+    UseCaseID.SPANS: UseCaseIDAPIAccess.PUBLIC,
+    UseCaseID.TRANSACTIONS: UseCaseIDAPIAccess.PUBLIC,
+    UseCaseID.SESSIONS: UseCaseIDAPIAccess.PUBLIC,
+    UseCaseID.ESCALATING_ISSUES: UseCaseIDAPIAccess.PRIVATE,
+    UseCaseID.CUSTOM: UseCaseIDAPIAccess.PUBLIC,
+    UseCaseID.PROFILES: UseCaseIDAPIAccess.PRIVATE,
+    UseCaseID.BUNDLE_ANALYSIS: UseCaseIDAPIAccess.PRIVATE,
+    UseCaseID.METRIC_STATS: UseCaseIDAPIAccess.PRIVATE,
+}
+
 # UseCaseKey will be renamed to MetricPathKey
 METRIC_PATH_MAPPING: Mapping[UseCaseID, UseCaseKey] = {
     UseCaseID.SPANS: UseCaseKey.PERFORMANCE,
@@ -58,16 +73,11 @@ class UseCaseID(Enum):
 }
 
 
-def get_use_case_key(use_case_id: UseCaseID) -> UseCaseKey | None:
-    return METRIC_PATH_MAPPING.get(use_case_id)
-
-
-def extract_use_case_id(mri: str) -> UseCaseID:
+def get_use_case_id_api_access(use_case_id: UseCaseID) -> UseCaseIDAPIAccess:
     """
-    Returns the use case ID given the MRI, returns None if MRI is invalid.
+    Returns the api access visibility of a use case and defaults to private in case no api access is provided.
+
+    The rationale for defaulting to private visibility is that we do not want to leak by mistake any internal metrics
+    that users should not have access to.
     """
-    if matched := MRI_RE_PATTERN.match(mri):
-        use_case_str = matched.group(2)
-        if use_case_str in {id.value for id in UseCaseID}:
-            return UseCaseID(use_case_str)
-    raise ValidationError(f"Invalid mri: {mri}")
+    return USE_CASE_ID_API_ACCESSES.get(use_case_id, UseCaseIDAPIAccess.PRIVATE)

src/sentry/snuba/metrics/naming_layer/mri.py

@@ -39,6 +39,8 @@
 from enum import Enum
 from typing import cast
 
+from sentry_kafka_schemas.codecs import ValidationError
+
 from sentry.exceptions import InvalidParams
 from sentry.sentry_metrics.use_case_id_registry import UseCaseID
 from sentry.snuba.dataset import EntityKey
@@ -358,3 +360,17 @@ def get_available_operations(parsed_mri: ParsedMRI) -> Sequence[str]:
     else:
         entity_key = get_entity_key_from_entity_type(parsed_mri.entity, True).value
         return AVAILABLE_GENERIC_OPERATIONS[entity_key]
+
+
+def extract_use_case_id(mri: str) -> UseCaseID:
+    """
+    Returns the use case ID given the MRI, throws an error if MRI is invalid or the use case doesn't exist.
+    """
+    parsed_mri = parse_mri(mri)
+    if parsed_mri is not None:
+        if parsed_mri.namespace in {id.value for id in UseCaseID}:
+            return UseCaseID(parsed_mri.namespace)
+
+        raise ValidationError(f"The use case of the MRI {parsed_mri.namespace} does not exist")
+
+    raise ValidationError(f"The MRI {mri} is not valid")

tests/sentry/sentry_metrics/test_gen_metrics_multiprocess_steps.py

@@ -419,7 +419,7 @@ def test_process_messages_default_card_rollout(set_sentry_option: Callable[...,
     ),
     (
         {
-            "name": "c:transactions/alert@none" * 21,
+            "name": "c:transactions/alert@" + ("none" * 50),
             "tags": {
                 "environment": "production",
                 "session.status": "errored",

tests/sentry/sentry_metrics/test_rh_metrics_multiprocess_steps.py

@@ -363,7 +363,7 @@ def test_process_messages() -> None:
     ),
     (
         {
-            "name": SessionMRI.RAW_ERROR.value * 21,
+            "name": "s:sessions/error@" + ("none" * 50),
             "tags": {
                 "environment": "production",
                 "session.status": "errored",