Greetings,
After updating PostgreSQL from version 17.7 to 17.8, we encountered a problem when extracting a substring from a UTF8 encoded string:
ERROR: invalid byte sequence for encoding "UTF8": 0xe2
Server:
Linux i-db-sandbox1.4vrs.com 6.1.0-43-cloud-amd64 #1 SMP PREEMPT_DYNAMIC Debian 6.1.162-1 (2026-02-08) x86_64 GNU/Linux
$ cat /etc/apt/sources.list.d/pgdg.list
deb http://apt.postgresql.org/pub/repos/apt/ bookworm-pgdg main
deb-src http://apt.postgresql.org/pub/repos/apt/ bookworm-pgdg main
deb https://apt-archive.postgresql.org/pub/repos/apt bookworm-pgdg-archive main
PostgreSQL 17.8 (Debian 17.8-1.pgdg12+1) on x86_64-pc-linux-gnu, compiled by gcc (Debian 12.2.0-14+deb12u1) 12.2.0, 64-bit
Steps to reproduce (psql):
db_fev:vladimir@i-db-sandbox1 => create table test123(id integer, m text);
CREATE TABLE
db_fev:vladimir@i-db-sandbox1 => insert into test123 (id,m) values (1, repeat('a', 1027)||E'\xe2\x80\x8d'||repeat('a', 1027));
INSERT 0 1
db_fev:vladimir@i-db-sandbox1 => select length(SUBSTRING(m from 1 for 256)) from test123;
ERROR: invalid byte sequence for encoding "UTF8": 0xe2
db_fev:vladimir@i-db-sandbox1 => select length(SUBSTRING(substring(m from 1 for length(m)) from 1 for 256)) from test123;
length
--------
256
(1 row)
Database db_feb:
Name | Encoding | Locale Provider | LC_COLLATE | LC_CTYPE | Locale | ICU Rules |
------------+-----------+-----------------+------------+----------+--------+-----------+
db_feb | UTF8 | libc | C | C | [NULL] | [NULL] |
The problem does not appear on PostgreSQL 17.7. Also, the problem does not occur if the string is fully loaded into memory:
db_feb:vladimir@i-db-sandbox1 =# select length(SUBSTRING(substring(m from 1 for length(m)) from 1 for 256)) from test123;
length
--------
256
(1 row)
The bugreport has also been sent to bugs@postgrespro.ru
--
Best Regards,
Vladimir Valikaev
Streamline - Property Management Software
