From 852495fdb08298168c90fbe540b3b79d69c6793e Mon Sep 17 00:00:00 2001
From: Helmut Merz <helmutm@cy55.de>
Date: Fri, 23 Jun 2023 11:31:37 +0200
Subject: [PATCH] provide SQL scripts for populating search index tables from
 index.ejsl

---
 sql/load.sql   | 17 +++++++++++++++++
 sql/tables.sql | 19 +++++++++++++++++++
 2 files changed, 36 insertions(+)
 create mode 100644 sql/load.sql
 create mode 100644 sql/tables.sql

diff --git a/sql/load.sql b/sql/load.sql
new file mode 100644
index 0000000..925385f
--- /dev/null
+++ b/sql/load.sql
@@ -0,0 +1,17 @@
+-- load.sql
+
+delete from hugo_input;
+
+\copy hugo_input (data) from 'public/index.ejsl'
+
+insert into hugo_text (site, url, title, content)
+  select data ->> 'site', data ->> 'url', data ->> 'title', data ->> 'content'
+  from hugo_input;
+-- TODO: fill more columns
+-- TODO: on conflict update ...
+
+-- sample query:
+-- select to_tsquery('german', 'prolog') as q \gset
+-- select url, title, 
+--     ts_headline(content, :q, 'MaxFragments=3, MaxWords=6, MinWords=3')  
+--   from hugo_text where :q @@ content_tsv;
diff --git a/sql/tables.sql b/sql/tables.sql
new file mode 100644
index 0000000..8567d99
--- /dev/null
+++ b/sql/tables.sql
@@ -0,0 +1,19 @@
+-- tables.sql
+
+create table hugo_input (data jsonb);
+
+create table hugo_text (
+	site text,
+	url text,
+	title text,
+	content text,
+	title_tsv tsvector generated always as 
+		(to_tsvector('german', coalesce(title, ''))) stored,
+	content_tsv tsvector generated always as 
+		(to_tsvector('german', coalesce(content, ''))) stored,
+	primary key (site, url)
+);
+
+create index title_tsv_idx on hugo_text using gin (title_tsv);
+create index content_tsv_idx on hugo_text using gin (content_tsv);
+