From 1ce8b3f40773e3eff5e21e9944c9e3a2f0897384 Mon Sep 17 00:00:00 2001
From: Daniel K <dk1844@gmail.com>
Date: Thu, 11 Apr 2024 09:41:09 +0200
Subject: [PATCH] trailing paths cleanup as bulk operations - not limited to
 created to ECS mapping script created one

---
 scripts/mongo/ecs_trailing_slash_cleanup.js | 13 ++++++-------
 1 file changed, 6 insertions(+), 7 deletions(-)

diff --git a/scripts/mongo/ecs_trailing_slash_cleanup.js b/scripts/mongo/ecs_trailing_slash_cleanup.js
index 44fde7212..737efe72c 100644
--- a/scripts/mongo/ecs_trailing_slash_cleanup.js
+++ b/scripts/mongo/ecs_trailing_slash_cleanup.js
@@ -13,16 +13,15 @@
  * limitations under the License.
  */
 
-// Removes trailing slashes from ECS-mapped paths
+// Removes trailing slashes from 's3a://.../' paths (ECS-mapped paths and derived further from there)
 
-function stripTrailingSlashOps(collectionName, fieldToStrip, requiredFieldToExist) {
+function stripTrailingSlashOps(collectionName, fieldToStrip) {
   print(`PrepOps: Stripping trailing / from field ${fieldToStrip} collection in ${collectionName}`);
   var count = 0;
   var ops = db[collectionName].find(
     {
       "$and": [
-        {[requiredFieldToExist]: {$exists: true}},
-        {[fieldToStrip]: {$regex: "/$"}}
+        {[fieldToStrip]: {$regex: "s3a://.*/$"}}
       ]
     }
   ).map(function (doc) {
@@ -45,12 +44,12 @@ function stripTrailingSlashOps(collectionName, fieldToStrip, requiredFieldToExis
   return ops;
 }
 
-var ops_d1 = stripTrailingSlashOps("dataset_v1", "hdfsPath", "bakHdfsPath");
+var ops_d1 = stripTrailingSlashOps("dataset_v1", "hdfsPath");
 db.getCollection('dataset_v1').bulkWrite(ops_d1);
 
-var ops_d2 = stripTrailingSlashOps("dataset_v1", "hdfsPublishPath", "bakHdfsPublishPath");
+var ops_d2 = stripTrailingSlashOps("dataset_v1", "hdfsPublishPath");
 db.getCollection('dataset_v1').bulkWrite(ops_d2);
 
-var ops_mt1 = stripTrailingSlashOps("mapping_table_v1", "hdfsPath", "bakHdfsPath");
+var ops_mt1 = stripTrailingSlashOps("mapping_table_v1", "hdfsPath");
 db.getCollection('mapping_table_v1').bulkWrite(ops_mt1);