From 265a7a5f6f450f97794a6b25731a634ae7448c41 Mon Sep 17 00:00:00 2001
From: Dan Fuller <dfuller@sentry.io>
Date: Thu, 22 Jun 2023 10:00:58 -0700
Subject: [PATCH] feat(search-issues): Start writing message column in search
 issues (#4387)

This adds the `message` column to search issues and starts writing to it. `message` is already being
sent, but we missed including it a column initially.

I will let this write for a week or two before switching over to use it. We don't need the full 90
days of data for this to be effective.
---
 .../configuration/issues/entities/search_issues.yaml   |  1 +
 .../configuration/issues/storages/search_issues.yaml   |  1 +
 snuba/datasets/processors/search_issues_processor.py   |  2 ++
 snuba/snuba_migrations/search_issues/__init__.py       |  0
 tests/datasets/test_search_issues_processor.py         | 10 ++++++++++
 tests/test_search_issues_api.py                        |  2 ++
 6 files changed, 16 insertions(+)
 create mode 100644 snuba/snuba_migrations/search_issues/__init__.py

diff --git a/snuba/datasets/configuration/issues/entities/search_issues.yaml b/snuba/datasets/configuration/issues/entities/search_issues.yaml
index 508030901f..3a81f8cfc6 100644
--- a/snuba/datasets/configuration/issues/entities/search_issues.yaml
+++ b/snuba/datasets/configuration/issues/entities/search_issues.yaml
@@ -9,6 +9,7 @@ schema:
     { name: group_id, type: UInt, args: { size: 64 } },
     { name: search_title, type: String },
     { name: resource_id, type: String },
+    { name: message, type: String },
     { name: subtitle, type: String },
     { name: culprit, type: String },
     { name: level, type: String, args: { schema_modifiers: [ low_cardinality ] } },
diff --git a/snuba/datasets/configuration/issues/storages/search_issues.yaml b/snuba/datasets/configuration/issues/storages/search_issues.yaml
index 87f79ce9fc..e9e30c71d1 100644
--- a/snuba/datasets/configuration/issues/storages/search_issues.yaml
+++ b/snuba/datasets/configuration/issues/storages/search_issues.yaml
@@ -23,6 +23,7 @@ schema:
 
 
       { name: resource_id, type: String, args: { schema_modifiers: [ nullable ] } },
+      { name: message, type: String },
       { name: subtitle, type: String, args: { schema_modifiers: [ nullable ] } },
       { name: culprit, type: String, args: { schema_modifiers: [ nullable ] } },
       { name: level, type: String, args: { schema_modifiers: [ nullable ] } },
diff --git a/snuba/datasets/processors/search_issues_processor.py b/snuba/datasets/processors/search_issues_processor.py
index 1de9291d33..c0c65b0f8d 100644
--- a/snuba/datasets/processors/search_issues_processor.py
+++ b/snuba/datasets/processors/search_issues_processor.py
@@ -92,6 +92,7 @@ class SearchIssueEvent(TypedDict, total=False):
     group_id: int
     platform: str
     primary_hash: str
+    message: str
     datetime: str
 
     data: IssueEventData
@@ -276,6 +277,7 @@ def process_insert_v1(
             "receive_timestamp": receive_timestamp,
             "client_timestamp": client_timestamp,
             "platform": event["platform"],
+            "message": _unicodify(event["message"]),
         }
 
         # optional fields
diff --git a/snuba/snuba_migrations/search_issues/__init__.py b/snuba/snuba_migrations/search_issues/__init__.py
new file mode 100644
index 0000000000..e69de29bb2
diff --git a/tests/datasets/test_search_issues_processor.py b/tests/datasets/test_search_issues_processor.py
index bd9b635b16..e2a36ce4db 100644
--- a/tests/datasets/test_search_issues_processor.py
+++ b/tests/datasets/test_search_issues_processor.py
@@ -36,6 +36,7 @@ def message_base() -> SearchIssueEvent:
         "primary_hash": str(uuid.uuid4()),
         "datetime": datetime.utcnow().isoformat() + "Z",
         "platform": "other",
+        "message": "something",
         "data": {
             "received": datetime.now().timestamp(),
         },
@@ -70,6 +71,7 @@ class TestSearchIssuesMessageProcessor:
         "platform",
         "tags.key",
         "tags.value",
+        "message",
     }
 
     def process_message(
@@ -439,6 +441,14 @@ def test_extract_replay_id(self, message_base):
             with pytest.raises(ValueError):
                 self.process_message(message_base)
 
+    def test_extract_message(self, message_base):
+        message = "a message"
+        message_base["message"] = message
+        processed = self.process_message(message_base)
+        self.assert_required_columns(processed)
+        insert_row = processed.rows[0]
+        assert insert_row["message"] == message
+
     def test_ensure_uuid(self):
         with pytest.raises(ValueError):
             ensure_uuid("not_a_uuid")
diff --git a/tests/test_search_issues_api.py b/tests/test_search_issues_api.py
index 425e7b2dc0..29704990c9 100644
--- a/tests/test_search_issues_api.py
+++ b/tests/test_search_issues_api.py
@@ -29,6 +29,7 @@ def base_insert_event(
             "primary_hash": str(uuid.uuid4()),
             "datetime": datetime.utcnow().isoformat() + "Z",
             "platform": "other",
+            "message": "message",
             "data": {
                 "received": now.timestamp(),
             },
@@ -94,6 +95,7 @@ def test_simple_search_query(self) -> None:
             primary_hash=str(uuid.uuid4().hex),
             datetime=datetime.utcnow().isoformat() + "Z",
             platform="other",
+            message="message",
             data={"received": now.timestamp()},
             occurrence_data=dict(
                 id=str(uuid.uuid4().hex),