create grouping component taxonomy

Author: lobsterkatie

src/sentry/grouping/component.py

@@ -1,7 +1,7 @@
 from __future__ import annotations
 
 from collections.abc import Generator, Iterator, Sequence
-from typing import Any, TypeVar
+from typing import Any, Generic, TypeVar
 
 from sentry.grouping.utils import hash_from_values
 
@@ -174,3 +174,141 @@ def __repr__(self) -> str:
 
 
 GroupingComponent = TypeVar("GroupingComponent", bound=BaseGroupingComponent)
+ValueType = TypeVar("ValueType", bound=(str | int))
+
+
+class ValueGroupingComponent(BaseGroupingComponent, Generic[ValueType]):
+    """
+    A component whose `values` is a list of a single actual value rather than a lit of other
+    grouping components.
+    """
+
+    values: list[ValueType]
+
+
+# Top-level components
+
+
+class MessageGroupingComponent(ValueGroupingComponent[str]):
+    id: str = "message"
+
+
+class ExceptionGroupingComponent(BaseGroupingComponent):
+    id: str = "exception"
+    values: list[
+        ErrorTypeGroupingComponent
+        | ErrorValueGroupingComponent
+        | NSErrorGroupingComponent
+        | StacktraceGroupingComponent
+    ]
+
+
+class ChainedExceptionGroupingComponent(BaseGroupingComponent):
+    id: str = "chained-exception"
+    values: list[ExceptionGroupingComponent]
+
+
+class StacktraceGroupingComponent(BaseGroupingComponent):
+    id: str = "stacktrace"
+    values: list[FrameGroupingComponent]
+
+
+class ThreadsGroupingComponent(BaseGroupingComponent):
+    id: str = "threads"
+    values: list[StacktraceGroupingComponent]
+
+
+class CSPGroupingComponent(BaseGroupingComponent):
+    id: str = "csp"
+    values: list[SaltGroupingComponent | ViolationGroupingComponent | URIGroupingComponent]
+
+
+class ExpectCTGroupingComponent(BaseGroupingComponent):
+    id: str = "expect-ct"
+    values: list[HostnameGroupingComponent | SaltGroupingComponent]
+
+
+class ExpectStapleGroupingComponent(BaseGroupingComponent):
+    id: str = "expect-staple"
+    values: list[HostnameGroupingComponent | SaltGroupingComponent]
+
+
+class HPKPGroupingComponent(BaseGroupingComponent):
+    id: str = "hpkp"
+    values: list[HostnameGroupingComponent | SaltGroupingComponent]
+
+
+class TemplateGroupingComponent(BaseGroupingComponent):
+    id: str = "template"
+    values: list[ContextLineGroupingComponent | FilenameGroupingComponent]
+
+
+# Error-related inner components
+
+
+class FrameGroupingComponent(BaseGroupingComponent):
+    id: str = "frame"
+    values: list[
+        ContextLineGroupingComponent
+        | FilenameGroupingComponent
+        | FunctionGroupingComponent
+        | LineNumberGroupingComponent  # only in legacy config
+        | ModuleGroupingComponent
+        | SymbolGroupingComponent  # only in legacy config
+    ]
+
+
+class ContextLineGroupingComponent(ValueGroupingComponent[str]):
+    id: str = "context-line"
+
+
+class ErrorTypeGroupingComponent(ValueGroupingComponent[str]):
+    id: str = "type"
+
+
+class ErrorValueGroupingComponent(ValueGroupingComponent[str]):
+    id: str = "value"
+
+
+class FilenameGroupingComponent(ValueGroupingComponent[str]):
+    id: str = "filename"
+
+
+class FunctionGroupingComponent(ValueGroupingComponent[str]):
+    id: str = "function"
+
+
+class LineNumberGroupingComponent(ValueGroupingComponent[str]):
+    id: str = "lineno"
+
+
+class ModuleGroupingComponent(ValueGroupingComponent[str]):
+    id: str = "module"
+
+
+class NSErrorGroupingComponent(ValueGroupingComponent[str | int]):
+    id: str = "ns-error"
+
+
+class SymbolGroupingComponent(ValueGroupingComponent[str]):
+    id: str = "symbol"
+
+
+# Security-related inner components
+
+
+class HostnameGroupingComponent(ValueGroupingComponent[str]):
+    id: str = "hostname"
+
+
+class SaltGroupingComponent(ValueGroupingComponent[str]):
+    id: str = "salt"
+    hint: str = "a static salt"
+
+
+class ViolationGroupingComponent(ValueGroupingComponent[str]):
+    id: str = "violation"
+
+
+class URIGroupingComponent(ValueGroupingComponent[str]):
+    id: str = "uri"

src/sentry/grouping/enhancer/__init__.py

@@ -15,7 +15,7 @@
 from sentry_ophio.enhancers import Enhancements as RustEnhancements
 
 from sentry import projectoptions
-from sentry.grouping.component import BaseGroupingComponent
+from sentry.grouping.component import StacktraceGroupingComponent
 from sentry.stacktraces.functions import set_in_app
 from sentry.utils.safe import get_path, set_path
 
@@ -179,8 +179,7 @@ def assemble_stacktrace_component(self, components, frames, platform, exception_
         for py_component, rust_component in zip(components, rust_components):
             py_component.update(contributes=rust_component.contributes, hint=rust_component.hint)
 
-        component = BaseGroupingComponent(
-            id="stacktrace",
+        component = StacktraceGroupingComponent(
             values=components,
             hint=rust_results.hint,
             contributes=rust_results.contributes,

src/sentry/grouping/strategies/legacy.py

@@ -3,7 +3,22 @@
 from typing import Any
 
 from sentry.eventstore.models import Event
-from sentry.grouping.component import BaseGroupingComponent
+from sentry.grouping.component import (
+    ChainedExceptionGroupingComponent,
+    ContextLineGroupingComponent,
+    ErrorTypeGroupingComponent,
+    ErrorValueGroupingComponent,
+    ExceptionGroupingComponent,
+    FilenameGroupingComponent,
+    FrameGroupingComponent,
+    FunctionGroupingComponent,
+    LineNumberGroupingComponent,
+    ModuleGroupingComponent,
+    SaltGroupingComponent,
+    StacktraceGroupingComponent,
+    SymbolGroupingComponent,
+    ThreadsGroupingComponent,
+)
 from sentry.grouping.strategies.base import (
     GroupingContext,
     ReturnedVariants,
@@ -164,17 +179,15 @@ def single_exception_legacy(
     interface: SingleException, event: Event, context: GroupingContext, **meta: Any
 ) -> ReturnedVariants:
 
-    type_component = BaseGroupingComponent(
-        id="type",
+    type_component = ErrorTypeGroupingComponent(
         values=[interface.type] if interface.type else [],
         contributes=False,
     )
-    value_component = BaseGroupingComponent(
-        id="value",
+    value_component = ErrorValueGroupingComponent(
         values=[interface.value] if interface.value else [],
         contributes=False,
     )
-    stacktrace_component = BaseGroupingComponent(id="stacktrace")
+    stacktrace_component = StacktraceGroupingComponent()
 
     if interface.stacktrace is not None:
         stacktrace_component = context.get_single_grouping_component(
@@ -195,8 +208,8 @@ def single_exception_legacy(
             value_component.update(contributes=True)
 
     return {
-        context["variant"]: BaseGroupingComponent(
-            id="exception", values=[stacktrace_component, type_component, value_component]
+        context["variant"]: ExceptionGroupingComponent(
+            values=[stacktrace_component, type_component, value_component]
         )
     }
 
@@ -231,7 +244,7 @@ def chained_exception_legacy(
             if stacktrace_component is None or not stacktrace_component.contributes:
                 value.update(contributes=False, hint="exception has no stacktrace")
 
-    return {context["variant"]: BaseGroupingComponent(id="chained-exception", values=values)}
+    return {context["variant"]: ChainedExceptionGroupingComponent(values=values)}
 
 
 @chained_exception_legacy.variant_processor
@@ -262,7 +275,7 @@ def frame_legacy(
     # Safari throws [native code] frames in for calls like ``forEach``
     # whereas Chrome ignores these. Let's remove it from the hashing algo
     # so that they're more likely to group together
-    filename_component = BaseGroupingComponent(id="filename")
+    filename_component = FilenameGroupingComponent()
     if interface.filename == "<anonymous>":
         filename_component.update(
             contributes=False, values=[interface.filename], hint="anonymous filename discarded"
@@ -293,13 +306,13 @@ def frame_legacy(
     # if we have a module we use that for grouping.  This will always
     # take precedence over the filename, even if the module is
     # considered unhashable.
-    module_component = BaseGroupingComponent(id="module")
+    module_component = ModuleGroupingComponent()
     if interface.module:
         if is_unhashable_module_legacy(interface, platform):
             module_component.update(
                 values=[
-                    BaseGroupingComponent(
-                        id="salt", values=["<module>"], hint="normalized generated module name"
+                    SaltGroupingComponent(
+                        values=["<module>"], hint="normalized generated module name"
                     )
                 ],
                 hint="ignored module",
@@ -313,7 +326,7 @@ def frame_legacy(
             )
 
     # Context line when available is the primary contributor
-    context_line_component = BaseGroupingComponent(id="context-line")
+    context_line_component = ContextLineGroupingComponent()
     if interface.context_line is not None:
         if len(interface.context_line) > 120:
             context_line_component.update(hint="discarded because line too long")
@@ -322,9 +335,9 @@ def frame_legacy(
         else:
             context_line_component.update(values=[interface.context_line])
 
-    symbol_component = BaseGroupingComponent(id="symbol")
-    function_component = BaseGroupingComponent(id="function")
-    lineno_component = BaseGroupingComponent(id="lineno")
+    symbol_component = SymbolGroupingComponent()
+    function_component = FunctionGroupingComponent()
+    lineno_component = LineNumberGroupingComponent()
 
     # The context line grouping information is the most reliable one.
     # If we did not manage to find some information there, we want to
@@ -347,8 +360,8 @@ def frame_legacy(
             if is_unhashable_function_legacy(func):
                 function_component.update(
                     values=[
-                        BaseGroupingComponent(
-                            id="salt", values=["<function>"], hint="normalized lambda function name"
+                        SaltGroupingComponent(
+                            values=["<function>"], hint="normalized lambda function name"
                         )
                     ]
                 )
@@ -380,8 +393,7 @@ def frame_legacy(
             )
 
     return {
-        context["variant"]: BaseGroupingComponent(
-            id="frame",
+        context["variant"]: FrameGroupingComponent(
             values=[
                 module_component,
                 filename_component,
@@ -461,8 +473,7 @@ def threads_legacy(
     thread_count = len(interface.values)
     if thread_count != 1:
         return {
-            context["variant"]: BaseGroupingComponent(
-                id="threads",
+            context["variant"]: ThreadsGroupingComponent(
                 contributes=False,
                 hint="ignored because contains %d threads" % thread_count,
             )
@@ -471,14 +482,13 @@ def threads_legacy(
     stacktrace = interface.values[0].get("stacktrace")
     if not stacktrace:
         return {
-            context["variant"]: BaseGroupingComponent(
-                id="threads", contributes=False, hint="thread has no stacktrace"
+            context["variant"]: ThreadsGroupingComponent(
+                contributes=False, hint="thread has no stacktrace"
             )
         }
 
     return {
-        context["variant"]: BaseGroupingComponent(
-            id="threads",
+        context["variant"]: ThreadsGroupingComponent(
             values=[context.get_single_grouping_component(stacktrace, event=event, **meta)],
         )
     }

src/sentry/grouping/strategies/message.py

@@ -4,7 +4,7 @@
 from sentry import analytics
 from sentry.eventstore.models import Event
 from sentry.features.rollout import in_rollout_group
-from sentry.grouping.component import BaseGroupingComponent
+from sentry.grouping.component import MessageGroupingComponent
 from sentry.grouping.parameterization import Parameterizer, UniqueIdExperiment
 from sentry.grouping.strategies.base import (
     GroupingContext,
@@ -108,17 +108,10 @@ def message_v1(
         raw = interface.message or interface.formatted or ""
         normalized = normalize_message_for_grouping(raw, event)
         hint = "stripped event-specific values" if raw != normalized else None
-        return {
-            context["variant"]: BaseGroupingComponent(
-                id="message",
-                values=[normalized],
-                hint=hint,
-            )
-        }
+        return {context["variant"]: MessageGroupingComponent(values=[normalized], hint=hint)}
     else:
         return {
-            context["variant"]: BaseGroupingComponent(
-                id="message",
+            context["variant"]: MessageGroupingComponent(
                 values=[interface.message or interface.formatted or ""],
             )
         }