From b4397ac91fa17972c4820da7d3acf80ecd6f93ba Mon Sep 17 00:00:00 2001
From: =?utf8?q?Paul=20H=C3=A4nsch?= <paul@plutz.net>
Date: Wed, 28 Feb 2024 15:57:10 +0100
Subject: [PATCH] bugfix: accidentally wiped search index when pruning

---
 searchindex.sh | 25 ++++++++++++++++++++-----
 1 file changed, 20 insertions(+), 5 deletions(-)

diff --git a/searchindex.sh b/searchindex.sh
index 162fad1..28bd76a 100755
--- a/searchindex.sh
+++ b/searchindex.sh
@@ -65,18 +65,25 @@ help() {
 	    script is called, if it can be determined.
 
 	--data SITE_DIR
-	    Point to the location of your site installation. I.e. the directory
-	    containing your "pages/" and "index/" dir. Defaults to the
-	    environment variable "\$_DATA" or the working directory.
+	    Point to the location of your site installation. I.e. the
+	    directory containing your "pages/" and "index/" dir. Defaults to
+	    the environment variable "\$_DATA" or the working directory.
 
 	--force
 	    Add pages to index even if they seem to be indexed already.
 
-	--loction /PAGE
+	--location /PAGE
 	    Index only the given page and its children. The path is given
 	    relative to the web root, i.e. without the DATA and "page/"
 	    directory.
 
+	--purge
+	    During pruning, remove empty database files. This will look
+	    tidyer in the index/ directory, but the operation does not
+	    guarantee atomicity and can disrupt index operations happening at
+	    the same time. Use it carefully and maybe don't use it in
+	    automatic runs via cron.
+
 	-v
 	    Be more verbose.
 	EOF
@@ -90,6 +97,7 @@ while [ $# -gt 0 ]; do case $1 in
   --location) location="${2}"; shift 2;;
   --verbose|-v) verb=true; shift 1;;
   --force) force=true; shift 1;;
+  --purge) purge=true; shift 1;;
   --help) help 0 2>&1;;
   prune|index)
     [ ! "$cmd" ] && cmd="$1" || help 1
@@ -112,6 +120,8 @@ fi
 . "$_EXEC/cgilite/storage.sh"
 
 prune() {
+  local v l d date location freq num total
+
   for word in "$_DATA/index"/*; do
     [ "$word" = "$_DATA/index/*" ] && continue
   
@@ -120,7 +130,7 @@ prune() {
     mv -- "$word" "${word}.$$"
   
     while read -r date location freq num total; do
-      l="$_DATA/pages$(UNSTRING "$location")#index.flag"
+      l="$_DATA/pages$(UNSTRING "$location")/#index.flag"
       d="$(stat -c %Y "$l")" 2>&-
   
       if [ "$date" -ge "$d" ] 2>&-; then
@@ -131,10 +141,15 @@ prune() {
       fi
     done <"${word}.$$" >>"${word}"
     rm -- "${word}.$$"
+    if [ "${purge}" -a -f "${word}" -a ! -s "${word}" ]; then
+      rm -- "${word}"
+      [ "$verb" ] && printf '\rRemoving empty search key "%s"\n' "${word}" >&2
+    fi
   done
 }
 
 index() {
+  local PATH_INFO _DATE SEARCH_INDEX location v
   export PATH_INFO="" _DATE="$(date +%s)" SEARCH_INDEX=true
 
   if [ "$location" ]; then
-- 
2.39.2