From d1d03d5c6aa35b0d9d66d4a1944e0a6cf49618ea Mon Sep 17 00:00:00 2001
From: Friedrich Lindenberg <friedrich@pudo.org>
Date: Sun, 8 Oct 2023 23:04:24 +0200
Subject: [PATCH] add an option to immediately cancel the indexer if a specific
 index already exists

---
 yente/search/indexer.py | 16 ++++++++++++++--
 yente/settings.py       |  1 +
 2 files changed, 15 insertions(+), 2 deletions(-)

diff --git a/yente/search/indexer.py b/yente/search/indexer.py
index 936f4de6..081e09f8 100644
--- a/yente/search/indexer.py
+++ b/yente/search/indexer.py
@@ -92,7 +92,13 @@ async def index_entities(es: AsyncElasticsearch, dataset: Dataset, force: bool)
     )
     dataset_prefix = f"{settings.ENTITY_INDEX}-{dataset.name}-"
     next_index = f"{dataset_prefix}{version}"
-    exists = await es.indices.exists_alias(name=settings.ENTITY_INDEX, index=next_index)
+    if settings.INDEX_EXISTS_ABORT:
+        exists = await es.indices.exists(index=next_index)
+    else:
+        exists = await es.indices.exists_alias(
+            name=settings.ENTITY_INDEX,
+            index=next_index,
+        )
     if exists.body and not force:
         log.info("Index is up to date.", index=next_index)
         return False
@@ -134,7 +140,13 @@ async def index_entities(es: AsyncElasticsearch, dataset: Dataset, force: bool)
             errors=errors,
             entities_url=dataset.entities_url,
         )
-        await es.indices.delete(index=next_index)
+        if settings.INDEX_EXISTS_ABORT:
+            is_linked = await es.indices.exists_alias(
+                name=settings.ENTITY_INDEX,
+                index=next_index,
+            )
+            if not is_linked.body:
+                await es.indices.delete(index=next_index)
         return False
 
     await es.indices.refresh(index=next_index)
diff --git a/yente/settings.py b/yente/settings.py
index 73f36849..c58387f7 100644
--- a/yente/settings.py
+++ b/yente/settings.py
@@ -150,6 +150,7 @@ def env_str(name: str, default: Optional[str] = None) -> Optional[str]:
 ES_SHARDS = int(env_str("YENTE_ELASTICSEARCH_SHARDS") or "1")
 ENTITY_INDEX = f"{ES_INDEX}-entities"
 INDEX_VERSION = env_str("YENTE_INDEX_VERSION", "008")
+INDEX_EXISTS_ABORT = as_bool(env_str("YENTE_INDEX_EXISTS_ABORT") or "false")
 
 # Log output can be formatted as JSON:
 LOG_JSON = as_bool(env_str("YENTE_LOG_JSON", "false"))