fix(pr-metrics): Send the canonical Seer RepoDefinition in the judge forward

The judge forward sent a flat repo dict with a combined "owner/repo" name, which
won't validate into Seer's RepoDefinition — it requires owner and name split, a
bare provider slug, and the base/org/privacy fields. Reuse code review's repo
definition builder so the forward emits the exact shape Seer already parses, with
no normalization on the Seer side.

Promote build_repo_definition to public and make its event_payload optional: the
forward runs in a task with no webhook payload, so it leaves is_private unknown.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Author: vaind

src/sentry/pr_metrics/judge.py

@@ -32,6 +32,7 @@
 from sentry.net.http import connection_from_url
 from sentry.pr_metrics.attribution import record_attribution_signal
 from sentry.pr_metrics.emit import active_attributions, emit_pr_metrics_row, resolved_group_ids
+from sentry.seer.code_review.utils import build_repo_definition
 from sentry.seer.signed_seer_api import SeerViewerContext, make_signed_seer_api_request
 from sentry.utils import metrics
 
@@ -81,14 +82,10 @@ def _build_judge_request(pull_request: PullRequest, repository: Repository) -> d
         "organization_id": pull_request.organization_id,
         "repository_id": pull_request.repository_id,
         "pull_request_id": pull_request.id,
-        "repo": {
-            "provider": repository.provider,
-            "external_id": repository.external_id,
-            "name": repository.name,
-            "integration_id": (
-                str(repository.integration_id) if repository.integration_id is not None else None
-            ),
-        },
+        # The shared Seer RepoDefinition shape (split owner/name, bare provider
+        # slug) so Seer parses it directly; head_commit_sha is the PR tip Seer
+        # resolves the repo at, with the merge/head SHAs also sent below.
+        "repo": build_repo_definition(repository, head_commit_sha),
         "pr_number": pull_request.key,
         "close_action": close_action,
         "head_commit_sha": head_commit_sha,

src/sentry/seer/code_review/utils.py

@@ -267,7 +267,7 @@ def _common_codegen_request_payload(
     event_payload: Mapping[str, Any],
 ) -> dict[str, Any]:
     data: dict[str, Any] = {
-        "repo": _build_repo_definition(repo, target_commit_sha, event_payload),
+        "repo": build_repo_definition(repo, target_commit_sha, event_payload),
         "bug_prediction_specific_information": {
             "organization_id": organization.id,
             "organization_slug": organization.slug,
@@ -356,17 +356,22 @@ def transform_pull_request_to_codegen_request(
     return payload
 
 
-def _build_repo_definition(
+def build_repo_definition(
     repo: Repository,
     target_commit_sha: str,
-    event_payload: Mapping[str, Any],
+    event_payload: Mapping[str, Any] | None = None,
 ) -> dict[str, Any]:
     """
-    Build the repository definition for code review requests.
+    Build the Seer RepoDefinition for a repository.
 
     GitHub and GitLab expose repo identity and visibility differently, so each
     provider has its own builder. Anything unrecognized falls back to GitHub.
+
+    ``event_payload`` supplies only the repo's privacy flag; a payload-less caller
+    (e.g. the PR metrics judge forward, which runs in a task off the webhook) may
+    omit it, leaving ``is_private`` unknown.
     """
+    event_payload = event_payload or {}
     # repo.provider uses the "integrations:<slug>" format; Seer expects the bare slug
     provider = repo.provider.removeprefix("integrations:") if repo.provider else "github"
 

tests/sentry/pr_metrics/test_judge.py

@@ -326,10 +326,15 @@ def test_forwards_terminal_facts_and_repo_identity(self, mock_request: Any) -> N
         assert body["close_action"] == "merged"
         assert body["head_commit_sha"] == HEAD_SHA
         assert body["merge_commit_sha"] == MERGE_SHA
+        # The shared Seer RepoDefinition shape: split owner/name and bare provider slug.
         assert body["repo"] == {
-            "provider": "integrations:github",
+            "provider": "github",
+            "owner": "getsentry",
+            "name": "sentry",
             "external_id": "10270250",
-            "name": "getsentry/sentry",
+            "base_commit_sha": HEAD_SHA,
+            "organization_id": self.organization.id,
+            "is_private": None,
             "integration_id": "99",
         }
         assert body["additions"] == 12

tests/sentry/seer/code_review/test_utils.py

@@ -866,34 +866,34 @@ def _make_repo(self, provider: str = "integrations:github") -> MagicMock:
         return repo
 
     def test_provider_is_github_for_cloud_integration(self) -> None:
-        from sentry.seer.code_review.utils import _build_repo_definition
+        from sentry.seer.code_review.utils import build_repo_definition
 
-        result = _build_repo_definition(
+        result = build_repo_definition(
             repo=self._make_repo("integrations:github"),
             target_commit_sha="abc123",
             event_payload={"repository": {"private": False}},
         )
         assert result["provider"] == "github"
 
     def test_provider_is_github_enterprise_for_ghe_integration(self) -> None:
-        from sentry.seer.code_review.utils import _build_repo_definition
+        from sentry.seer.code_review.utils import build_repo_definition
 
-        result = _build_repo_definition(
+        result = build_repo_definition(
             repo=self._make_repo("integrations:github_enterprise"),
             target_commit_sha="abc123",
             event_payload={"repository": {"private": False}},
         )
         assert result["provider"] == "github_enterprise"
 
     def test_gitlab_uses_config_path_not_display_name(self) -> None:
-        from sentry.seer.code_review.utils import _build_repo_definition
+        from sentry.seer.code_review.utils import build_repo_definition
 
         repo = self._make_repo("integrations:gitlab")
         # Repository.name is the display "name_with_namespace"; it must not be used.
         repo.name = "Cool Group / Sentry"
         repo.config = {"path": "cool-group/sentry"}
 
-        result = _build_repo_definition(
+        result = build_repo_definition(
             repo=repo,
             target_commit_sha="abc123",
             event_payload={"project": {"visibility_level": 0}},
@@ -903,14 +903,14 @@ def test_gitlab_uses_config_path_not_display_name(self) -> None:
         assert result["name"] == "sentry"
 
     def test_gitlab_missing_config_path_raises_rather_than_using_display_name(self) -> None:
-        from sentry.seer.code_review.utils import _build_repo_definition
+        from sentry.seer.code_review.utils import build_repo_definition
 
         repo = self._make_repo("integrations:gitlab")
         repo.name = "Cool Group / Sentry"
         repo.config = {}
 
         with pytest.raises(ValueError):
-            _build_repo_definition(
+            build_repo_definition(
                 repo=repo,
                 target_commit_sha="abc123",
                 event_payload={"project": {}},