From b45089e38287dd6da2cab48069cae40805f98da3 Mon Sep 17 00:00:00 2001
From: =?utf8?q?Paul=20H=C3=A4nsch?= <paul@plutz.net>
Date: Fri, 22 Sep 2023 10:07:31 +0200
Subject: [PATCH] search handler

---
 handlers/40_search.sh | 40 ++++++++++++++++++++++++++++++++++++++++
 1 file changed, 40 insertions(+)
 create mode 100644 handlers/40_search.sh

diff --git a/handlers/40_search.sh b/handlers/40_search.sh
new file mode 100644
index 0000000..3db8987
--- /dev/null
+++ b/handlers/40_search.sh
@@ -0,0 +1,40 @@
+#!/bin/sh
+
+[ "${PATH_INFO%\[search\]}" = "$PATH_INFO" ] && return 1
+
+. "$_EXEC/cgilite/storage.sh"
+
+I="$_DATA/index"
+words="$(GET q |tr '] \t\n\r!\"#'\''()*+,./:;<=>?\\^_`{|}~[-' ' ')"
+
+for w in ${words}; do
+  [ ! -f "$I/$w" ] && continue
+
+  while read date doc freq num total; do
+    P="$_DATA/pages$(UNSTRING "$doc")"
+    d="$(stat -c %Y -- "$P/#page.md")"
+    [ "$d" -gt "$date" ] && continue
+
+    printf '%s	%f\n' "$doc" "$freq"
+  done <"$I/$w"
+done \
+| awk '
+      { cnt[$1]++; weight[$1] = weight[$1] ? weight[$1] * $2 : $2; }
+  END { m = 0; for (d in cnt) m = ( m < cnt[d] ) ? cnt[d] : m;
+        for (d in cnt) if ( cnt[d] == m ) printf "%f	%s\n", weight[d], d;
+      }
+' \
+| sort -nr \
+| while read freq doc; do
+  page="$(UNSTRING "$doc")"
+  acl_read "$page" || continue
+  printf '<li><a href="%s">%s</a></li>' "$(URL "$page")" "$(HTML "$page")"
+done \
+| theme_page - <<-EOF
+	<article>
+	  <h1>$(_ "Search results:")</h1>
+	  <ol class="searchresults">
+	    $(cat)
+	  </ol>
+	</article>
+	EOF
-- 
2.39.2