From 99c5bcdb011d3d630afbff0fa98fa19023443205 Mon Sep 17 00:00:00 2001
From: jvoisin <julien.voisin@dustri.org>
Date: Sat, 21 Jun 2025 01:04:37 +0200
Subject: [PATCH] perf(storage): truncate strings on go's side instead of
 pgsql's

There is no need to send the whole title and content to have them truncated on
postgresql's side when we can do this client-side. This should save some
memory on the database's side, as well as some bandwidth
if it's located on another server. And it makes the SQL queries a tad more
readable as well.
---
 internal/storage/entry.go | 25 +++++++++++++++++--------
 1 file changed, 17 insertions(+), 8 deletions(-)

diff --git a/internal/storage/entry.go b/internal/storage/entry.go
index d2063318..75248643 100644
--- a/internal/storage/entry.go
+++ b/internal/storage/entry.go
@@ -18,6 +18,8 @@ import (
 	"github.com/lib/pq"
 )
 
+const truncationLen = 500000
+
 // CountAllEntries returns the number of entries for each status in the database.
 func (s *Storage) CountAllEntries() map[string]int64 {
 	rows, err := s.db.Query(`SELECT status, count(*) FROM entries GROUP BY status`)
@@ -78,12 +80,12 @@ func (s *Storage) UpdateEntryTitleAndContent(entry *model.Entry) error {
 			title=$1,
 			content=$2,
 			reading_time=$3,
-			document_vectors = setweight(to_tsvector(left(coalesce($1, ''), 500000)), 'A') || setweight(to_tsvector(left(coalesce($2, ''), 500000)), 'B')
+			document_vectors = setweight(to_tsvector($1), 'A') || setweight(to_tsvector($2), 'B')
 		WHERE
 			id=$4 AND user_id=$5
 	`
 
-	if _, err := s.db.Exec(query, entry.Title, entry.Content, entry.ReadingTime, entry.ID, entry.UserID); err != nil {
+	if _, err := s.db.Exec(query, truncateString(entry.Title), truncateString(entry.Content), entry.ReadingTime, entry.ID, entry.UserID); err != nil {
 		return fmt.Errorf(`store: unable to update entry #%d: %v`, entry.ID, err)
 	}
 
@@ -122,7 +124,7 @@ func (s *Storage) createEntry(tx *sql.Tx, entry *model.Entry) error {
 				$9,
 				$10,
 				now(),
-				setweight(to_tsvector(left(coalesce($1, ''), 500000)), 'A') || setweight(to_tsvector(left(coalesce($6, ''), 500000)), 'B'),
+				setweight(to_tsvector($1), 'A') || setweight(to_tsvector($6), 'B'),
 				$11
 			)
 		RETURNING
@@ -130,12 +132,12 @@ func (s *Storage) createEntry(tx *sql.Tx, entry *model.Entry) error {
 	`
 	err := tx.QueryRow(
 		query,
-		entry.Title,
+		truncateString(entry.Title),
 		entry.Hash,
 		entry.URL,
 		entry.CommentsURL,
 		entry.Date,
-		entry.Content,
+		truncateString(entry.Content),
 		entry.Author,
 		entry.UserID,
 		entry.FeedID,
@@ -178,7 +180,7 @@ func (s *Storage) updateEntry(tx *sql.Tx, entry *model.Entry) error {
 			content=$4,
 			author=$5,
 			reading_time=$6,
-			document_vectors = setweight(to_tsvector(left(coalesce($1, ''), 500000)), 'A') || setweight(to_tsvector(left(coalesce($4, ''), 500000)), 'B'),
+			document_vectors = setweight(to_tsvector($1), 'A') || setweight(to_tsvector($4), 'B'),
 			tags=$10
 		WHERE
 			user_id=$7 AND feed_id=$8 AND hash=$9
@@ -187,10 +189,10 @@ func (s *Storage) updateEntry(tx *sql.Tx, entry *model.Entry) error {
 	`
 	err := tx.QueryRow(
 		query,
-		entry.Title,
+		truncateString(entry.Title),
 		entry.URL,
 		entry.CommentsURL,
-		entry.Content,
+		truncateString(entry.Content),
 		entry.Author,
 		entry.ReadingTime,
 		entry.UserID,
@@ -642,3 +644,10 @@ func removeEmpty(l []string) []string {
 	}
 	return finalSlice
 }
+
+func truncateString(s string) string {
+	if len(s) > truncationLen {
+		return s[:truncationLen]
+	}
+	return s
+}