From c693d4e78ae27e4f29bb1e5d93f778de1cb3b97a Mon Sep 17 00:00:00 2001
From: Bill Thiede <git@xinu.tv>
Date: Sun, 19 Jan 2025 17:20:18 -0800
Subject: [PATCH] server: strip html from search index of summaries

---
 .../20250117225159_strip-html-index.down.sql          |  5 +++++
 .../migrations/20250117225159_strip-html-index.up.sql | 11 +++++++++++
 2 files changed, 16 insertions(+)
 create mode 100644 server/migrations/20250117225159_strip-html-index.down.sql
 create mode 100644 server/migrations/20250117225159_strip-html-index.up.sql

diff --git a/server/migrations/20250117225159_strip-html-index.down.sql b/server/migrations/20250117225159_strip-html-index.down.sql
new file mode 100644
index 0000000..4b01805
--- /dev/null
+++ b/server/migrations/20250117225159_strip-html-index.down.sql
@@ -0,0 +1,5 @@
+-- Add down migration script here
+DROP INDEX post_summary_idx;
+CREATE INDEX post_summary_idx ON post USING gin (
+    to_tsvector('english', summary)
+);
diff --git a/server/migrations/20250117225159_strip-html-index.up.sql b/server/migrations/20250117225159_strip-html-index.up.sql
new file mode 100644
index 0000000..4e99e01
--- /dev/null
+++ b/server/migrations/20250117225159_strip-html-index.up.sql
@@ -0,0 +1,11 @@
+-- Something like this around summary in the idx w/ tsvector
+DROP INDEX post_summary_idx;
+CREATE INDEX post_summary_idx ON post USING gin (to_tsvector(
+    'english',
+    regexp_replace(
+        regexp_replace(summary, '<[^>]+>', ' ', 'g'),
+        '\s+',
+        ' ',
+        'g'
+    )
+));