Просмотр исходного кода

Set arbitrary maximum size for tsvector column

- The max size for tsvector is 1 MiB
- We index only the first million of characters,
  it should be enough for most feed entries.
Frédéric Guillot 7 лет назад
Родитель
Сommit
a614f98063
3 измененных файлов с 6 добавлено и 6 удалено
  1. 2 2
      database/sql.go
  2. 1 1
      database/sql/schema_version_20.sql
  3. 3 3
      storage/entry.go

+ 2 - 2
database/sql.go

@@ -142,7 +142,7 @@ alter table users add column extra hstore;
 create index users_extra_idx on users using gin(extra);
 `,
 	"schema_version_20": `alter table entries add column document_vectors tsvector;
-update entries set document_vectors = to_tsvector(title || ' ' || coalesce(content, ''));
+update entries set document_vectors = to_tsvector(substring(title || ' ' || coalesce(content, '') for 1000000));
 create index document_vectors_idx on entries using gin(document_vectors);`,
 	"schema_version_21": `alter table feeds add column user_agent text default '';`,
 	"schema_version_3": `create table tokens (
@@ -192,7 +192,7 @@ var SqlMapChecksums = map[string]string{
 	"schema_version_18": "c0ec24847612c7f2dc326cf735baffba79391a56aedd73292371a39f38724a71",
 	"schema_version_19": "a83f77b41cc213d282805a5b518f15abbf96331599119f0ef4aca4be037add7b",
 	"schema_version_2":  "e8e9ff32478df04fcddad10a34cba2e8bb1e67e7977b5bd6cdc4c31ec94282b4",
-	"schema_version_20": "6c4e9b2c5bccdc3243c239c390fb1caa5e15624e669b2c07e14c126f6d2e2cd6",
+	"schema_version_20": "5d414c0cfc0da2863c641079afa58b7ff42dccb0f0e01c822ad435c3e3aa9201",
 	"schema_version_21": "77da01ee38918ff4fe33985fbb20ed3276a717a7584c2ca9ebcf4d4ab6cb6910",
 	"schema_version_3":  "a54745dbc1c51c000f74d4e5068f1e2f43e83309f023415b1749a47d5c1e0f12",
 	"schema_version_4":  "216ea3a7d3e1704e40c797b5dc47456517c27dbb6ca98bf88812f4f63d74b5d9",

+ 1 - 1
database/sql/schema_version_20.sql

@@ -1,3 +1,3 @@
 alter table entries add column document_vectors tsvector;
-update entries set document_vectors = to_tsvector(title || ' ' || coalesce(content, ''));
+update entries set document_vectors = to_tsvector(substring(title || ' ' || coalesce(content, '') for 1000000));
 create index document_vectors_idx on entries using gin(document_vectors);

+ 3 - 3
storage/entry.go

@@ -50,7 +50,7 @@ func (s *Storage) UpdateEntryContent(entry *model.Entry) error {
 
 	query := `
 		UPDATE entries
-		SET document_vectors = to_tsvector(title || ' ' || coalesce(content, ''))
+		SET document_vectors = to_tsvector(substring(title || ' ' || coalesce(content, '') for 1000000))
 		WHERE id=$1 AND user_id=$2
 	`
 	_, err = tx.Exec(query, entry.ID, entry.UserID)
@@ -68,7 +68,7 @@ func (s *Storage) createEntry(entry *model.Entry) error {
 		INSERT INTO entries
 		(title, hash, url, comments_url, published_at, content, author, user_id, feed_id, document_vectors)
 		VALUES
-		($1, $2, $3, $4, $5, $6, $7, $8, $9, to_tsvector($1 || ' ' || coalesce($6, '')))
+		($1, $2, $3, $4, $5, $6, $7, $8, $9, to_tsvector(substring($1 || ' ' || coalesce($6, '') for 1000000)))
 		RETURNING id, status
 	`
 	err := s.db.QueryRow(
@@ -107,7 +107,7 @@ func (s *Storage) updateEntry(entry *model.Entry) error {
 	query := `
 		UPDATE entries SET
 		title=$1, url=$2, comments_url=$3, content=$4, author=$5,
-		document_vectors=to_tsvector($1 || ' ' || coalesce($4, ''))
+		document_vectors=to_tsvector(substring($1 || ' ' || coalesce($4, '') for 1000000))
 		WHERE user_id=$6 AND feed_id=$7 AND hash=$8
 		RETURNING id
 	`