Complete publish preview Phase 2A with retention, refresh and cache consistency

backend/app/services/publish_preview_cache.py

@@ -5,7 +5,11 @@ from pathlib import Path
 from uuid import uuid4
 
 from app.core.config import settings
-from app.repositories.scheme_artifacts import create_scheme_artifact, list_scheme_artifacts
+from app.repositories.scheme_artifacts import (
+    create_scheme_artifact,
+    delete_scheme_artifacts_by_artifact_ids,
+    list_scheme_artifacts,
+)
 
 
 def _preview_storage_dir() -> Path:
@@ -14,6 +18,64 @@ def _preview_storage_dir() -> Path:
     return path
 
 
+def _cleanup_preview_file(storage_path: str) -> None:
+    path = Path(storage_path)
+    try:
+        if path.exists() and path.is_file():
+            path.unlink()
+    except FileNotFoundError:
+        pass
+
+    parent = path.parent
+    preview_root = Path(settings.storage_preview_dir)
+
+    try:
+        if parent != preview_root and parent.exists():
+            parent.rmdir()
+    except OSError:
+        pass
+
+
+async def cleanup_publish_preview_artifacts(
+    *,
+    scheme_version_id: str,
+    baseline_scheme_version_id: str | None,
+) -> dict:
+    retention = max(1, settings.publish_preview_retention_per_variant)
+    variant = baseline_scheme_version_id or "default"
+
+    rows = await list_scheme_artifacts(
+        scheme_version_id=scheme_version_id,
+        artifact_type="publish_preview",
+        artifact_variant=variant,
+    )
+    if len(rows) <= retention:
+        return {
+            "retention": retention,
+            "deleted_count": 0,
+            "deleted_artifact_ids": [],
+        }
+
+    rows_sorted = sorted(
+        rows,
+        key=lambda row: (row.created_at, row.id),
+        reverse=True,
+    )
+    to_delete = rows_sorted[retention:]
+
+    deleted_artifact_ids = [row.artifact_id for row in to_delete]
+    for row in to_delete:
+        _cleanup_preview_file(row.storage_path)
+
+    deleted_count = await delete_scheme_artifacts_by_artifact_ids(deleted_artifact_ids)
+
+    return {
+        "retention": retention,
+        "deleted_count": deleted_count,
+        "deleted_artifact_ids": deleted_artifact_ids,
+    }
+
+
 async def save_publish_preview_artifact(
     *,
     scheme_id: str,
@@ -37,7 +99,16 @@ async def save_publish_preview_artifact(
             "summary": payload.get("summary"),
         },
     )
-    return artifact
+
+    cleanup = await cleanup_publish_preview_artifacts(
+        scheme_version_id=scheme_version_id,
+        baseline_scheme_version_id=baseline_scheme_version_id,
+    )
+
+    return {
+        "artifact": artifact,
+        "cleanup": cleanup,
+    }
 
 
 async def get_latest_publish_preview_artifact(
@@ -45,15 +116,13 @@ async def get_latest_publish_preview_artifact(
     scheme_version_id: str,
     baseline_scheme_version_id: str | None,
 ):
-    rows = await list_scheme_artifacts(scheme_version_id=scheme_version_id)
-    variant = baseline_scheme_version_id or "default"
-
-    matching = [
-        row for row in rows
-        if row.artifact_type == "publish_preview" and row.artifact_variant == variant
-    ]
-    if not matching:
+    rows = await list_scheme_artifacts(
+        scheme_version_id=scheme_version_id,
+        artifact_type="publish_preview",
+        artifact_variant=baseline_scheme_version_id or "default",
+    )
+    if not rows:
         return None
 
-    matching.sort(key=lambda row: (row.created_at, row.id), reverse=True)
-    return matching[0]
+    rows.sort(key=lambda row: (row.created_at, row.id), reverse=True)
+    return rows[0]