Remove File-based EventAttachment code (#93804)

Already 17 months ago in #62318, I switched the `EventAttachment` code
over from using the `file_id` and django `File`-models to storing things
in filestore (the go service) directly via the `blob_path`.

As attachments have a 30 day TTL, it is high time to remove the
backwards compatibility code.

Author: Swatinem

src/sentry/api/serializers/models/eventattachment.py

@@ -7,18 +7,10 @@
 
 @register(EventAttachment)
 class EventAttachmentSerializer(Serializer):
-    def get_attrs(self, item_list, user, **kwargs):
-        files = {
-            f.id: f
-            for f in File.objects.filter(id__in=[ea.file_id for ea in item_list if ea.file_id])
-        }
-        return {ea: {"file": files[ea.file_id]} for ea in item_list if ea.file_id}
-
     def serialize(self, obj, attrs, user, **kwargs):
-        file = attrs.get("file")
-        content_type = obj.content_type or get_mimetype(file)
-        size = obj.size if obj.size is not None else file.size
-        sha1 = obj.sha1 or file.checksum
+        content_type = obj.content_type
+        size = obj.size or 0
+        sha1 = obj.sha1
         headers = {"Content-Type": content_type}
 
         return {

src/sentry/event_manager.py

@@ -2401,7 +2401,6 @@ def save_attachment(
         size=file.size,
         sha1=file.sha1,
         # storage:
-        file_id=file.file_id,
         blob_path=file.blob_path,
     )
 

src/sentry/models/eventattachment.py

@@ -8,7 +8,6 @@
 
 import zstandard
 from django.core.cache import cache
-from django.core.exceptions import ObjectDoesNotExist
 from django.db import models
 from django.utils import timezone
 
@@ -42,7 +41,6 @@ class PutfileResult:
     content_type: str
     size: int
     sha1: str
-    file_id: int | None = None
     blob_path: str | None = None
 
 
@@ -61,8 +59,7 @@ class EventAttachment(Model):
     """Attachment Metadata and Storage
 
     The actual attachment data can be saved in different backing stores:
-    - Using the :class:`File` model using the `file_id` field.
-      This stores attachments chunked and deduplicated.
+    - When the attachment is empty (0-size), `blob_path is None`.
     - When the `blob_path` field has a `:` prefix:
       It is saved inline in `blob_path` following the `:` prefix.
       This happens for "small" and ASCII-only (see `can_store_inline`) attachments.
@@ -88,8 +85,7 @@ class EventAttachment(Model):
 
     date_added = models.DateTimeField(default=timezone.now, db_index=True)
 
-    # the backing blob, either going through the `File` model,
-    # or directly to a backing blob store
+    # The `file_id` will be removed in a multi-part migration soon.
     file_id = BoundedBigIntegerField(null=True, db_index=True)
     blob_path = models.TextField(null=True)
 
@@ -114,51 +110,29 @@ def delete(self, *args: Any, **kwargs: Any) -> tuple[int, dict[str, int]]:
 
         if self.blob_path:
             if self.blob_path.startswith(":"):
-                return rv
+                pass  # nothing to do for inline-stored attachments
             elif self.blob_path.startswith("eventattachments/v1/"):
                 storage = get_storage()
+                storage.delete(self.blob_path)
             else:
                 raise NotImplementedError()
 
-            storage.delete(self.blob_path)
-            return rv
-
-        try:
-            from sentry.models.files.file import File
-
-            file = File.objects.get(id=self.file_id)
-        except ObjectDoesNotExist:
-            # It's possible that the File itself was deleted
-            # before we were deleted when the object is in memory
-            # This seems to be a case that happens during deletion
-            # code.
-            pass
-        else:
-            file.delete()
-
         return rv
 
     def getfile(self) -> IO[bytes]:
-        if self.size == 0:
+        if not self.blob_path:
             return BytesIO(b"")
 
-        if self.blob_path:
-            if self.blob_path.startswith(":"):
-                return BytesIO(self.blob_path[1:].encode())
-
-            elif self.blob_path.startswith("eventattachments/v1/"):
-                storage = get_storage()
-                compressed_blob = storage.open(self.blob_path)
-                dctx = zstandard.ZstdDecompressor()
-                return dctx.stream_reader(compressed_blob, read_across_frames=True)
+        if self.blob_path.startswith(":"):
+            return BytesIO(self.blob_path[1:].encode())
 
-            else:
-                raise NotImplementedError()
-
-        from sentry.models.files.file import File
+        elif self.blob_path.startswith("eventattachments/v1/"):
+            storage = get_storage()
+            compressed_blob = storage.open(self.blob_path)
+            dctx = zstandard.ZstdDecompressor()
+            return dctx.stream_reader(compressed_blob, read_across_frames=True)
 
-        file = File.objects.get(id=self.file_id)
-        return file.getfile()
+        raise NotImplementedError()
 
     @classmethod
     def putfile(cls, project_id: int, attachment: CachedAttachment) -> PutfileResult:
@@ -171,7 +145,6 @@ def putfile(cls, project_id: int, attachment: CachedAttachment) -> PutfileResult
             return PutfileResult(content_type=content_type, size=0, sha1=sha1().hexdigest())
 
         blob = BytesIO(data)
-
         size, checksum = get_size_and_checksum(blob)
 
         if can_store_inline(data):

src/sentry/testutils/factories.py

@@ -87,7 +87,6 @@
 )
 from sentry.models.debugfile import ProjectDebugFile
 from sentry.models.environment import Environment
-from sentry.models.eventattachment import EventAttachment
 from sentry.models.files.control_file import ControlFile
 from sentry.models.files.file import File
 from sentry.models.group import Group
@@ -1106,25 +1105,6 @@ def create_file_from_path(path, name=None, **kwargs):
             file.putfile(f)
         return file
 
-    @staticmethod
-    @assume_test_silo_mode(SiloMode.REGION)
-    def create_event_attachment(event, file=None, **kwargs):
-        if file is None:
-            file = Factories.create_file(
-                name="log.txt",
-                size=32,
-                headers={"Content-Type": "text/plain"},
-                checksum="dc1e3f3e411979d336c3057cce64294f3420f93a",
-            )
-
-        return EventAttachment.objects.create(
-            project_id=event.project_id,
-            event_id=event.event_id,
-            file_id=file.id,
-            type=file.type,
-            **kwargs,
-        )
-
     @staticmethod
     @assume_test_silo_mode(SiloMode.REGION)
     def create_dif_file(

src/sentry/testutils/fixtures.py

@@ -317,11 +317,6 @@ def create_file(self, **kwargs):
     def create_file_from_path(self, *args, **kwargs):
         return Factories.create_file_from_path(*args, **kwargs)
 
-    def create_event_attachment(self, event=None, *args, **kwargs):
-        if event is None:
-            event = self.event
-        return Factories.create_event_attachment(event, *args, **kwargs)
-
     def create_dif_file(self, project: Project | None = None, *args, **kwargs):
         if project is None:
             project = self.project

src/sentry/utils/mockdata/core.py

@@ -647,12 +647,9 @@ def generate_events(
             project_id=project.id,
             event_id=event1.event_id,
             name="example-logfile.txt",
-            file_id=File.objects.get_or_create(
-                name="example-logfile.txt",
-                type="text/plain",
-                checksum="abcde" * 8,
-                size=13043,
-            )[0].id,
+            type="text/plain",
+            sha1="abcde" * 8,
+            size=13043,
         )
 
         event2 = create_sample_event(

tests/sentry/api/endpoints/test_event_attachment_details.py

@@ -35,10 +35,7 @@ def create_attachment(self, content: bytes | None = None, group_id: int | None =
         attachment = CachedAttachment(
             name="hello.png", content_type="image/png; foo=bar", data=data
         )
-        file = EventAttachment.putfile(
-            self.project.id,
-            attachment,
-        )
+        file = EventAttachment.putfile(self.project.id, attachment)
 
         self.attachment = EventAttachment.objects.create(
             event_id=self.event.event_id,
@@ -50,7 +47,6 @@ def create_attachment(self, content: bytes | None = None, group_id: int | None =
             size=file.size,
             sha1=file.sha1,
             # storage:
-            file_id=file.file_id,
             blob_path=file.blob_path,
         )
 

tests/sentry/api/endpoints/test_event_attachments.py

@@ -1,7 +1,4 @@
-from io import BytesIO
-
 from sentry.models.eventattachment import EventAttachment
-from sentry.models.files.file import File
 from sentry.testutils.cases import APITestCase
 from sentry.testutils.helpers.datetime import before_now
 from sentry.testutils.skips import requires_snuba
@@ -21,14 +18,15 @@ def test_simple(self):
             data={"fingerprint": ["group1"], "timestamp": min_ago}, project_id=self.project.id
         )
 
-        file1 = File.objects.create(name="hello.png", type="event.attachment")
-        file1.putfile(BytesIO(b"File contents here"))
         attachment1 = EventAttachment.objects.create(
             project_id=event1.project_id,
             event_id=event1.event_id,
             type="event.attachment",
-            name=file1.name,
-            file_id=file1.id,
+            name="hello.png",
+            content_type="image/png",
+            size=18,
+            sha1="d3f299af02d6abbe92dd8368bab781824a9702ed",
+            blob_path=":File contents here",
         )
 
         attachment2 = EventAttachment.objects.create(
@@ -39,7 +37,6 @@ def test_simple(self):
             content_type="image/png",
             size=1234,
             sha1="1234",
-            # NOTE: we are not actually attaching the `file_id` here
         )
 
         path = f"/api/0/projects/{event1.project.organization.slug}/{event1.project.slug}/events/{event1.event_id}/attachments/"
@@ -82,26 +79,21 @@ def test_is_screenshot(self):
             data={"fingerprint": ["group1"], "timestamp": min_ago}, project_id=self.project.id
         )
 
-        file = File.objects.create(name="screenshot.png", type="image/png")
         EventAttachment.objects.create(
             event_id=event1.event_id,
             project_id=event1.project_id,
-            file_id=file.id,
-            name=file.name,
+            name="screenshot.png",
+            content_type="image/png",
         )
-        file = File.objects.create(name="crash_screenshot.png")
         EventAttachment.objects.create(
             event_id=event1.event_id,
             project_id=event1.project_id,
-            file_id=file.id,
-            name=file.name,
+            name="crash_screenshot.png",
         )
-        file = File.objects.create(name="foo.png")
         EventAttachment.objects.create(
             event_id=event1.event_id,
             project_id=event1.project_id,
-            file_id=file.id,
-            name=file.name,
+            name="foo.png",
         )
 
         path = f"/api/0/projects/{event1.project.organization.slug}/{event1.project.slug}/events/{event1.event_id}/attachments/"

tests/sentry/api/endpoints/test_group_attachments.py

@@ -1,8 +1,6 @@
-from io import BytesIO
 from urllib.parse import urlencode
 
 from sentry.models.eventattachment import EventAttachment
-from sentry.models.files.file import File
 from sentry.testutils.cases import APITestCase
 from sentry.testutils.helpers.datetime import before_now
 from sentry.testutils.skips import requires_snuba
@@ -15,16 +13,13 @@ def create_attachment(self, type=None, event_id=None, file_name="hello.png", gro
         if type is None:
             type = "event.attachment"
 
-        self.file = File.objects.create(name=file_name, type=type)
-        self.file.putfile(BytesIO(b"File contents here"))
-
         self.attachment = EventAttachment.objects.create(
             event_id=event_id or self.event.event_id,
             project_id=self.event.project_id,
             group_id=group_id or self.group.id,
-            file_id=self.file.id,
-            type=self.file.type,
+            type=type,
             name=file_name,
+            blob_path=":File contents here",
         )
 
         return self.attachment

tests/sentry/deletions/test_group.py

@@ -15,7 +15,6 @@
 from sentry.issues.grouptype import FeedbackGroup, GroupCategory
 from sentry.issues.issue_occurrence import IssueOccurrence
 from sentry.models.eventattachment import EventAttachment
-from sentry.models.files.file import File
 from sentry.models.group import Group
 from sentry.models.groupassignee import GroupAssignee
 from sentry.models.grouphash import GroupHash
@@ -57,13 +56,11 @@ def setUp(self) -> None:
         UserReport.objects.create(
             event_id=self.event.event_id, project_id=self.event.project_id, name="With event id"
         )
-        file = File.objects.create(name="hello.png", type="image/png")
         EventAttachment.objects.create(
             event_id=self.event.event_id,
             project_id=self.event.project_id,
-            file_id=file.id,
-            type=file.type,
             name="hello.png",
+            content_type="image/png",
         )
         GroupAssignee.objects.create(group=group, project=self.project, user_id=self.user.id)
         GroupHash.objects.create(project=self.project, group=group, hash=uuid4().hex)