From 5f48ef87bc5bf361f05b37853bd4f1ddc0153b55 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Guido=20Mart=C3=ADnez?= <mtzguido@gmail.com>
Date: Wed, 23 Oct 2024 21:02:26 -0700
Subject: [PATCH 1/2] actions: adding a workflow to check for stale hints

---
 .github/workflows/stale_hints.yml | 19 +++++++++++++++++++
 .scripts/remove_stale_hints.sh    | 13 +++++++++++--
 2 files changed, 30 insertions(+), 2 deletions(-)
 create mode 100644 .github/workflows/stale_hints.yml

diff --git a/.github/workflows/stale_hints.yml b/.github/workflows/stale_hints.yml
new file mode 100644
index 00000000000..f7388d7001a
--- /dev/null
+++ b/.github/workflows/stale_hints.yml
@@ -0,0 +1,19 @@
+name: Check for stale hints
+on:
+  pull_request:
+  workflow_dispatch:
+jobs:
+  check_stale_hints:
+    runs-on: ubuntu-latest
+    defaults:
+      run:
+        shell: bash
+    steps:
+      - uses: actions/checkout@v4
+      - run: |
+          L=$(.scripts/remove_stale_hints.sh list)
+          if [ "$L" != "" ]; then
+            echo "There are stale hints:"
+            echo "$L"
+            false
+          fi
diff --git a/.scripts/remove_stale_hints.sh b/.scripts/remove_stale_hints.sh
index e8e8f20570c..db029f17465 100755
--- a/.scripts/remove_stale_hints.sh
+++ b/.scripts/remove_stale_hints.sh
@@ -2,6 +2,11 @@
 
 set -eu
 
+list=false
+if [ $# -gt 0 ] && [ $1 == "list" ]; then
+	list=true
+fi
+
 declare -A files # Store all basenames in repo
 declare -A hints # Store all paths of hints in repo
 
@@ -27,7 +32,11 @@ for h0 in "${!hints[@]}"; do
 	# Given a/b/c/Foo.Bar.fst.hints, if there is no Foo.Bar.fst
 	# anywhere, then delete the hint file.
 	if ! [ -v "files[$h]" ]; then
-		echo "DELETING HINT $h0"
-		rm -f "${h0}"
+		if $list; then
+			echo "$h0"
+		else
+			echo "DELETING HINT $h0"
+			rm -f "${h0}"
+		fi
 	fi
 done

From 23f2b1fb2d882dee83836aca4bc6fff50ed28a99 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Guido=20Mart=C3=ADnez?= <mtzguido@gmail.com>
Date: Wed, 23 Oct 2024 21:31:05 -0700
Subject: [PATCH 2/2] scripts: remove_stale_hints: make it faster

---
 .scripts/remove_stale_hints.sh | 19 +++++++++++--------
 1 file changed, 11 insertions(+), 8 deletions(-)

diff --git a/.scripts/remove_stale_hints.sh b/.scripts/remove_stale_hints.sh
index db029f17465..720f5b34382 100755
--- a/.scripts/remove_stale_hints.sh
+++ b/.scripts/remove_stale_hints.sh
@@ -7,8 +7,8 @@ if [ $# -gt 0 ] && [ $1 == "list" ]; then
 	list=true
 fi
 
-declare -A files # Store all basenames in repo
-declare -A hints # Store all paths of hints in repo
+declare -A files # Store all basenames for non-hint files in repo (value = 1, this is just a set)
+declare -A hints # Store a map from hint paths into basenames (turns out basename computation was a bottleneck)
 
 trap 'RC=$?; rm -f .filelist; exit $RC' EXIT INT
 
@@ -16,18 +16,21 @@ trap 'RC=$?; rm -f .filelist; exit $RC' EXIT INT
 # the array. Using a pipe here would be better, but a pipe creates a
 # subshell, so changes to the variables won't have any effect in the
 # parent.
-find . -name '.git' -prune -o \( -type f \) > .filelist
+find . -name '.git' -prune -false -o \( -type f -name '*.hints' \) > .filelist
 while read f0; do
 	f="$(basename "${f0}")"
-	files[$f]=1;
-	if [[ "$f0" == *.hints ]]; then
-		hints[$f0]=1
-	fi
+	hints[$f0]=$f
+done < .filelist
+
+find . -name '.git' -prune -false -o \( -type f -not -name '*.hints' \) -printf '%f\n' > .filelist
+while read f; do
+	files[$f]=1
 done < .filelist
+
 rm -f .filelist
 
 for h0 in "${!hints[@]}"; do
-	h="$(basename "${h0}")"
+	h=${hints[$h0]}
 	h="${h%.hints}"
 	# Given a/b/c/Foo.Bar.fst.hints, if there is no Foo.Bar.fst
 	# anywhere, then delete the hint file.