ref(proguard): Deobfuscate view hierarchies as part of symbolication (#74196)

Depends on getsentry/symbolicator#1496.

This moves view hierarchy deobfuscation out of the plugin/preprocessor
and into symbolication.

Author: loewenheim

src/sentry/lang/java/plugin.py

@@ -4,7 +4,7 @@
 from typing import Any
 
 from sentry.lang.java.processing import deobfuscate_exception_value
-from sentry.lang.java.utils import deobfuscate_view_hierarchy, has_proguard_file
+from sentry.lang.java.utils import has_proguard_file
 from sentry.plugins.base.v2 import EventPreprocessor, Plugin2
 
 
@@ -19,6 +19,6 @@ def get_stacktrace_processors(self, data, stacktrace_infos, platforms, **kwargs)
 
     def get_event_preprocessors(self, data: Mapping[str, Any]) -> Sequence[EventPreprocessor]:
         if has_proguard_file(data):
-            return [deobfuscate_exception_value, deobfuscate_view_hierarchy]
+            return [deobfuscate_exception_value]
         else:
             return []

src/sentry/lang/java/processing.py

@@ -3,6 +3,10 @@
 from collections.abc import Mapping
 from typing import Any
 
+import orjson
+
+from sentry.attachments import CachedAttachment, attachment_cache
+from sentry.ingest.consumer.processors import CACHE_TIMEOUT
 from sentry.lang.java.utils import get_jvm_images, get_proguard_images
 from sentry.lang.native.error import SymbolicationFailed, write_error
 from sentry.lang.native.symbolicator import Symbolicator
@@ -11,6 +15,7 @@
 from sentry.models.release import Release
 from sentry.stacktraces.processing import find_stacktraces_in_data
 from sentry.utils import metrics
+from sentry.utils.cache import cache_key_for_event
 from sentry.utils.safe import get_path
 
 logger = logging.getLogger(__name__)
@@ -139,6 +144,76 @@ def _get_release_package(project: Project, release_name: str | None) -> str | No
     return release.package if release else None
 
 
+def _get_window_class_names(attachments: list[CachedAttachment]) -> list[str]:
+    """Returns the class names of all windows in all view hierarchies
+    contained in `attachments`."""
+
+    class_names = []
+    windows_to_deobfuscate = []
+
+    for attachment in attachments:
+        if attachment.type == "event.view_hierarchy":
+            view_hierarchy = orjson.loads(attachment_cache.get_data(attachment))
+            windows_to_deobfuscate.extend(view_hierarchy.get("windows"))
+
+    while windows_to_deobfuscate:
+        window = windows_to_deobfuscate.pop()
+        if window.get("type") is not None:
+            class_names.append(window["type"])
+        if children := window.get("children"):
+            windows_to_deobfuscate.extend(children)
+
+    return class_names
+
+
+def _deobfuscate_view_hierarchy(view_hierarchy: Any, class_names: dict[str, str]) -> None:
+    """Deobfuscates a view hierarchy in-place.
+
+    The `class_names` dict is used to resolve obfuscated to deobfuscated names. If
+    an obfuscated class name isn't present in `class_names`, it is left unchanged."""
+
+    windows_to_deobfuscate = [*view_hierarchy.get("windows")]
+
+    while windows_to_deobfuscate:
+        window = windows_to_deobfuscate.pop()
+        if (
+            window.get("type") is not None
+            and (mapped_type := class_names.get(window["type"])) is not None
+        ):
+            window["type"] = mapped_type
+        if children := window.get("children"):
+            windows_to_deobfuscate.extend(children)
+
+
+def _deobfuscate_view_hierarchies(
+    attachments: list[CachedAttachment], class_names: dict[str, str]
+) -> list[CachedAttachment]:
+    """Deobfuscates all view hierarchies contained in `attachments`, returning a new list of attachments.
+
+    Non-view-hierarchy attachments are unchanged.
+    """
+    new_attachments = []
+    for attachment in attachments:
+        if attachment.type == "event.view_hierarchy":
+            view_hierarchy = orjson.loads(attachment_cache.get_data(attachment))
+            _deobfuscate_view_hierarchy(view_hierarchy, class_names)
+            # Reupload to cache as a unchunked data
+            new_attachments.append(
+                CachedAttachment(
+                    type=attachment.type,
+                    id=attachment.id,
+                    name=attachment.name,
+                    content_type=attachment.content_type,
+                    data=orjson.dumps(view_hierarchy),
+                    chunks=None,
+                )
+            )
+        else:
+            new_attachments.append(attachment)
+
+    return new_attachments
+
+
 def map_symbolicator_process_jvm_errors(
     errors: list[dict[str, Any]] | None,
 ) -> list[dict[str, Any]]:
@@ -195,10 +270,17 @@ def process_jvm_stacktraces(symbolicator: Symbolicator, data: Any) -> Any:
     ]
 
     processable_exceptions = _get_exceptions_for_symbolication(data)
+    cache_key = cache_key_for_event(data)
+    attachments = [*attachment_cache.get(cache_key)]
+    window_class_names = _get_window_class_names(attachments)
 
     metrics.incr("proguard.symbolicator.events")
 
-    if not any(stacktrace["frames"] for stacktrace in stacktraces) and not processable_exceptions:
+    if (
+        not any(stacktrace["frames"] for stacktrace in stacktraces)
+        and not processable_exceptions
+        and not window_class_names
+    ):
         metrics.incr("proguard.symbolicator.events.skipped")
         return
 
@@ -211,6 +293,7 @@ def process_jvm_stacktraces(symbolicator: Symbolicator, data: Any) -> Any:
         stacktraces=stacktraces,
         modules=modules,
         release_package=release_package,
+        classes=window_class_names,
     )
 
     if not _handle_response_status(data, response):
@@ -248,4 +331,8 @@ def process_jvm_stacktraces(symbolicator: Symbolicator, data: Any) -> Any:
         raw_exc["module"] = exc["module"]
         raw_exc["type"] = exc["type"]
 
+    classes = response.get("classes")
+    new_attachments = _deobfuscate_view_hierarchies(attachments, classes)
+    attachment_cache.set(cache_key, attachments=new_attachments, timeout=CACHE_TIMEOUT)
+
     return data

src/sentry/lang/java/utils.py

@@ -11,7 +11,6 @@
 from sentry.lang.java.proguard import open_proguard_mapper
 from sentry.models.debugfile import ProjectDebugFile
 from sentry.models.project import Project
-from sentry.stacktraces.processing import StacktraceInfo
 from sentry.utils.cache import cache_key_for_event
 from sentry.utils.safe import get_path
 
@@ -70,31 +69,6 @@ def get_proguard_mapper(uuid: str, project: Project):
     return mapper
 
 
-def _deobfuscate_view_hierarchy(event_data: dict[str, Any], project: Project, view_hierarchy):
-    """
-    Deobfuscates a view hierarchy in-place.
-
-    If we're unable to fetch a ProGuard uuid or unable to init the mapper,
-    then the view hierarchy remains unmodified.
-    """
-    proguard_uuids = get_proguard_images(event_data)
-    if len(proguard_uuids) == 0:
-        return
-
-    with sentry_sdk.start_span(op="proguard.deobfuscate_view_hierarchy_data"):
-        for proguard_uuid in proguard_uuids:
-            mapper = get_proguard_mapper(proguard_uuid, project)
-            if mapper is None:
-                return
-
-            windows_to_deobfuscate = [*view_hierarchy.get("windows")]
-            while windows_to_deobfuscate:
-                window = windows_to_deobfuscate.pop()
-                window["type"] = mapper.remap_class(window.get("type")) or window.get("type")
-                if children := window.get("children"):
-                    windows_to_deobfuscate.extend(children)
-
-
 @sentry_sdk.trace
 def deobfuscation_template(data, map_type, deobfuscation_fn):
     """
@@ -133,13 +107,8 @@ def deobfuscation_template(data, map_type, deobfuscation_fn):
     attachment_cache.set(cache_key, attachments=new_attachments, timeout=CACHE_TIMEOUT)
 
 
-def deobfuscate_view_hierarchy(data):
-    deobfuscation_template(data, "proguard", _deobfuscate_view_hierarchy)
-
-
-def is_jvm_event(data: Any, stacktraces: list[StacktraceInfo]) -> bool:
-    """Returns whether `data` is a JVM event, based on its platform, images, and
-    the supplied stacktraces."""
+def is_jvm_event(data: Any) -> bool:
+    """Returns whether `data` is a JVM event, based on its images."""
 
     # check if there are any JVM or Proguard images
     images = get_path(
@@ -149,14 +118,4 @@ def is_jvm_event(data: Any, stacktraces: list[StacktraceInfo]) -> bool:
         filter=lambda x: is_valid_jvm_image(x) or is_valid_proguard_image(x),
         default=(),
     )
-    if not images:
-        return False
-
-    if data.get("platform") == "java":
-        return True
-
-    for stacktrace in stacktraces:
-        if any(x == "java" for x in stacktrace.platforms):
-            return True
-
-    return False
+    return bool(images)

src/sentry/lang/native/symbolicator.py

@@ -241,6 +241,7 @@ def process_jvm(
         stacktraces,
         modules,
         release_package,
+        classes,
         apply_source_context=True,
     ):
         """
@@ -262,6 +263,7 @@ def process_jvm(
             "exceptions": exceptions,
             "stacktraces": stacktraces,
             "modules": modules,
+            "classes": classes,
             "options": {"apply_source_context": apply_source_context},
         }
 

src/sentry/profiles/task.py

@@ -489,6 +489,7 @@ def symbolicate(
             modules=modules,
             release_package=profile.get("transaction_metadata", {}).get("app.identifier"),
             apply_source_context=False,
+            classes=[],
         )
     return symbolicator.process_payload(
         stacktraces=stacktraces, modules=modules, apply_source_context=False

src/sentry/tasks/symbolication.py

@@ -109,7 +109,7 @@ def get_symbolication_platforms(
 
     platforms = []
 
-    if is_jvm_event(data, stacktraces):
+    if is_jvm_event(data):
         platforms.append(SymbolicatorPlatform.jvm)
     if is_js_event(data, stacktraces):
         platforms.append(SymbolicatorPlatform.js)

tests/relay_integration/lang/java/test_plugin.py

@@ -1556,7 +1556,6 @@ def test_invalid_exception(self):
 
     def test_is_jvm_event(self):
         from sentry.lang.java.utils import is_jvm_event
-        from sentry.stacktraces.processing import find_stacktraces_in_data
 
         event = {
             "user": {"ip_address": "31.172.207.97"},
@@ -1586,8 +1585,7 @@ def test_is_jvm_event(self):
             },
             "timestamp": iso_format(before_now(seconds=1)),
         }
-        stacktraces = find_stacktraces_in_data(event)
-        assert is_jvm_event(event, stacktraces)
+        assert is_jvm_event(event)
 
         event = {
             "user": {"ip_address": "31.172.207.97"},
@@ -1616,9 +1614,8 @@ def test_is_jvm_event(self):
             },
             "timestamp": iso_format(before_now(seconds=1)),
         }
-        stacktraces = find_stacktraces_in_data(event)
         # has no platform
-        assert not is_jvm_event(event, stacktraces)
+        assert is_jvm_event(event)
 
         event = {
             "user": {"ip_address": "31.172.207.97"},
@@ -1648,6 +1645,5 @@ def test_is_jvm_event(self):
             },
             "timestamp": iso_format(before_now(seconds=1)),
         }
-        stacktraces = find_stacktraces_in_data(event)
         # has no modules
-        assert not is_jvm_event(event, stacktraces)
+        assert not is_jvm_event(event)