Deploy PR search API contributor surface fallback

src/slop_farmer/app/pr_search_api.py

@@ -152,8 +152,15 @@ def create_app(settings: PrSearchApiSettings | None = None) -> FastAPI:
         settings = request.app.state.settings
         repo_slug = _repo_slug(settings, owner, repo)
         status = get_pr_search_status(settings.index_path, repo=repo_slug)
-        snapshot_dir = _status_snapshot_dir(status)
-        return {**status, "surfaces": get_snapshot_surfaces(snapshot_dir)}
+        issue_snapshot_dir = _surface_snapshot_dir(settings, repo_slug, surface="issues")
+        contributor_snapshot_dir = _surface_snapshot_dir(
+            settings, repo_slug, surface="contributors"
+        )
+        surface_payload = {
+            "issues": get_snapshot_surfaces(issue_snapshot_dir)["issues"],
+            "contributors": get_snapshot_surfaces(contributor_snapshot_dir)["contributors"],
+        }
+        return {**status, "surfaces": surface_payload}
 
     @app.get("/v1/repos/{owner}/{repo}/pulls/{number}/similar")
     async def pr_similar(
@@ -240,7 +247,10 @@ def create_app(settings: PrSearchApiSettings | None = None) -> FastAPI:
     ) -> dict[str, Any]:
         settings = request.app.state.settings
         repo_slug = _repo_slug(settings, owner, repo)
-        return get_issue_cluster_status(_active_snapshot_dir(settings, repo_slug), variant=variant)
+        return get_issue_cluster_status(
+            _surface_snapshot_dir(settings, repo_slug, surface="issues"),
+            variant=variant,
+        )
 
     @app.get("/v1/repos/{owner}/{repo}/issues/clusters")
     async def issue_clusters(
@@ -253,7 +263,7 @@ def create_app(settings: PrSearchApiSettings | None = None) -> FastAPI:
         settings = request.app.state.settings
         repo_slug = _repo_slug(settings, owner, repo)
         return list_issue_clusters(
-            _active_snapshot_dir(settings, repo_slug),
+            _surface_snapshot_dir(settings, repo_slug, surface="issues"),
             limit=_limit(
                 limit,
                 default=settings.issue_list_limit_default,
@@ -273,7 +283,7 @@ def create_app(settings: PrSearchApiSettings | None = None) -> FastAPI:
         settings = request.app.state.settings
         repo_slug = _repo_slug(settings, owner, repo)
         return get_issue_cluster(
-            _active_snapshot_dir(settings, repo_slug),
+            _surface_snapshot_dir(settings, repo_slug, surface="issues"),
             cluster_id=cluster_id,
             variant=variant,
         )
@@ -289,7 +299,7 @@ def create_app(settings: PrSearchApiSettings | None = None) -> FastAPI:
         settings = request.app.state.settings
         repo_slug = _repo_slug(settings, owner, repo)
         return get_issue_clusters_for_pr(
-            _active_snapshot_dir(settings, repo_slug),
+            _surface_snapshot_dir(settings, repo_slug, surface="issues"),
             pr_number=number,
             variant=variant,
         )
@@ -306,7 +316,7 @@ def create_app(settings: PrSearchApiSettings | None = None) -> FastAPI:
         settings = request.app.state.settings
         repo_slug = _repo_slug(settings, owner, repo)
         return check_issue_cluster_membership(
-            _active_snapshot_dir(settings, repo_slug),
+            _surface_snapshot_dir(settings, repo_slug, surface="issues"),
             pr_number=number,
             cluster_id=cluster_id,
             variant=variant,
@@ -323,7 +333,7 @@ def create_app(settings: PrSearchApiSettings | None = None) -> FastAPI:
         settings = request.app.state.settings
         repo_slug = _repo_slug(settings, owner, repo)
         return list_issue_duplicate_prs(
-            _active_snapshot_dir(settings, repo_slug),
+            _surface_snapshot_dir(settings, repo_slug, surface="issues"),
             limit=_limit(
                 limit,
                 default=settings.issue_list_limit_default,
@@ -341,7 +351,10 @@ def create_app(settings: PrSearchApiSettings | None = None) -> FastAPI:
     ) -> dict[str, Any]:
         settings = request.app.state.settings
         repo_slug = _repo_slug(settings, owner, repo)
-        return get_issue_best(_active_snapshot_dir(settings, repo_slug), variant=variant)
+        return get_issue_best(
+            _surface_snapshot_dir(settings, repo_slug, surface="issues"),
+            variant=variant,
+        )
 
     @app.get("/v1/repos/{owner}/{repo}/contributors/status")
     async def contributor_status(
@@ -351,7 +364,7 @@ def create_app(settings: PrSearchApiSettings | None = None) -> FastAPI:
     ) -> dict[str, Any]:
         settings = request.app.state.settings
         repo_slug = _repo_slug(settings, owner, repo)
-        return get_contributor_status(_active_snapshot_dir(settings, repo_slug))
+        return get_contributor_status(_surface_snapshot_dir(settings, repo_slug, surface="contributors"))
 
     @app.get("/v1/repos/{owner}/{repo}/contributors")
     async def contributors(
@@ -363,7 +376,7 @@ def create_app(settings: PrSearchApiSettings | None = None) -> FastAPI:
         settings = request.app.state.settings
         repo_slug = _repo_slug(settings, owner, repo)
         return list_contributors(
-            _active_snapshot_dir(settings, repo_slug),
+            _surface_snapshot_dir(settings, repo_slug, surface="contributors"),
             limit=_limit(
                 limit,
                 default=settings.contributor_list_limit_default,
@@ -380,7 +393,10 @@ def create_app(settings: PrSearchApiSettings | None = None) -> FastAPI:
     ) -> dict[str, Any]:
         settings = request.app.state.settings
         repo_slug = _repo_slug(settings, owner, repo)
-        return get_contributor(_active_snapshot_dir(settings, repo_slug), author_login=login)
+        return get_contributor(
+            _surface_snapshot_dir(settings, repo_slug, surface="contributors"),
+            author_login=login,
+        )
 
     @app.get("/v1/repos/{owner}/{repo}/contributors/{login}/risk")
     async def contributor_risk(
@@ -391,7 +407,10 @@ def create_app(settings: PrSearchApiSettings | None = None) -> FastAPI:
     ) -> dict[str, Any]:
         settings = request.app.state.settings
         repo_slug = _repo_slug(settings, owner, repo)
-        return get_contributor_risk(_active_snapshot_dir(settings, repo_slug), author_login=login)
+        return get_contributor_risk(
+            _surface_snapshot_dir(settings, repo_slug, surface="contributors"),
+            author_login=login,
+        )
 
     return app
 
@@ -474,6 +493,23 @@ def _active_snapshot_dir(settings: PrSearchApiSettings, repo_slug: str) -> Path:
     return _status_snapshot_dir(get_pr_search_status(settings.index_path, repo=repo_slug))
 
 
+def _surface_snapshot_dir(
+    settings: PrSearchApiSettings,
+    repo_slug: str,
+    *,
+    surface: Literal["issues", "contributors"],
+) -> Path:
+    active_snapshot_dir = _active_snapshot_dir(settings, repo_slug)
+    if _surface_available(active_snapshot_dir, surface=surface):
+        return active_snapshot_dir
+    materialized_snapshot_dir = _materialized_snapshot_dir(settings)
+    if materialized_snapshot_dir is not None and _surface_available(
+        materialized_snapshot_dir, surface=surface
+    ):
+        return materialized_snapshot_dir
+    return active_snapshot_dir
+
+
 def _status_snapshot_dir(status: dict[str, Any]) -> Path:
     snapshot_dir = status.get("snapshot_dir")
     if not snapshot_dir:
@@ -481,6 +517,24 @@ def _status_snapshot_dir(status: dict[str, Any]) -> Path:
     return Path(str(snapshot_dir))
 
 
+def _materialized_snapshot_dir(settings: PrSearchApiSettings) -> Path | None:
+    if settings.hf_repo_id is None:
+        return None
+    return settings.hf_materialize_dir or default_hf_materialize_dir(
+        settings.output_dir,
+        settings.hf_repo_id,
+        settings.hf_revision,
+    )
+
+
+def _surface_available(snapshot_dir: Path, *, surface: Literal["issues", "contributors"]) -> bool:
+    if not snapshot_dir.exists():
+        return False
+    if surface == "issues":
+        return any(snapshot_dir.glob("analysis-report*.json"))
+    return (snapshot_dir / "new-contributors-report.json").exists()
+
+
 def _limit(value: int | None, *, default: int, maximum: int) -> int:
     limit = default if value is None else value
     if limit < 1: