From de4263376a79eeb427ed0ae46efe3298de15e9cd Mon Sep 17 00:00:00 2001
From: Cee Chen <549407+cee-chen@users.noreply.github.com>
Date: Fri, 26 Apr 2024 09:44:15 -0700
Subject: [PATCH] [buildkite] Delete all but most recent 50 PR staging subdirs
 (#7707)

---
 .../pipelines/pipeline_prune_staging_docs.sh  | 21 +++++++++++--------
 1 file changed, 12 insertions(+), 9 deletions(-)

diff --git a/.buildkite/scripts/pipelines/pipeline_prune_staging_docs.sh b/.buildkite/scripts/pipelines/pipeline_prune_staging_docs.sh
index 2fa808b33024..b66f377b93c3 100644
--- a/.buildkite/scripts/pipelines/pipeline_prune_staging_docs.sh
+++ b/.buildkite/scripts/pipelines/pipeline_prune_staging_docs.sh
@@ -25,14 +25,17 @@ BUCKET=${GPROJECT}-${EUI_DOCS_PROJECT}
 # https://cloud.google.com/storage/docs/gsutil/commands/ls
 ls_options=(
   -d # only list directories
-  -l # Print additional details about the subdir
+  -l # include additional details about the subdir, notably the date as a second field
 )
-echo "Listing all PR staging links"
-gsutil ls "${ls_options[@]}" "gs://${BUCKET}/pr_*" # | sort -k 2 # sort by the 2nd field returned by -l which is the 'created by' timestamp
-
-# https://cloud.google.com/storage/docs/gsutil/commands/rm
-rm_options=(
-  -r # recursive, delete everything inside subdir
-  -m # enables multi-threading for large numbers of objects
+echo "Getting all but the most recent 50 PR staging links..."
+list=$(gsutil ls "${ls_options[@]}" "gs://${BUCKET}/pr_*" \
+  | sort -k 2 `# sort by the 2nd field returned by -l which is a timestamp` \
+  | head -n -50 `# remove the last 50 items, so basically keep the latest 50 staging docs` \
 )
-# gsutil rm "${rm_options[@]}" "gs://${BUCKET}/pr_TODO"
+while IFS= read -r line || [[ -n $line ]]; do
+  url="$(echo -e "${line}" | tr -d '[:space:]')" # trim the leading whitespaces
+  url=${url%/} # trim the trailing slash
+  echo "Deleting $url"
+  # https://cloud.google.com/storage/docs/gsutil/commands/rm
+  gsutil -m rm -r "$url"
+done < <(printf '%s' "$list")