From 5cfbb59a11d099fa9b8703502fd8aac936a02c4d Mon Sep 17 00:00:00 2001 From: Roman Zharkov Date: Fri, 21 Oct 2022 19:56:19 +0300 Subject: [PATCH] Fix citext_utf8 test's "Turkish I" with ICU collation provider MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit With the ICU collation provider the Turkish unicode symbol "İ" (U+0130 "LATIN CAPITAL LETTER I WITH DOT ABOVE") has two lowercase variants: - "i", i.e. "U+0069 LATIN SMALL LETTER I", in "tr" and "az" locales. - "i̇", i.e. "U+0069 LATIN SMALL LETTER I" followed by "U+0307 COMBINING DOT ABOVE" in all other locales I've tried (including "en-US", "de", "ru", etc). So, add both variants to the test. --- contrib/citext/expected/citext_utf8.out | 6 ++++-- contrib/citext/sql/citext_utf8.sql | 6 ++++-- 2 files changed, 8 insertions(+), 4 deletions(-) diff --git a/contrib/citext/expected/citext_utf8.out b/contrib/citext/expected/citext_utf8.out index 666b07ccec4..62f0794028f 100644 --- a/contrib/citext/expected/citext_utf8.out +++ b/contrib/citext/expected/citext_utf8.out @@ -48,10 +48,12 @@ SELECT 'Ä'::citext <> 'Ä'::citext AS t; t (1 row) --- Test the Turkish dotted I. The lowercase is a single byte while the +-- Test the Turkish dotted I. The lowercase might be a single byte while the -- uppercase is multibyte. This is why the comparison code can't be optimized -- to compare string lengths. -SELECT 'i'::citext = 'İ'::citext AS t; +-- Note that lower('İ') is 'i' (U+0069) in tr and az locales, +-- but 'i̇' (U+0069 U+0307) in C and most (all?) other locales. +SELECT 'İ'::citext in ('i'::citext, 'i̇'::citext) AS t; t --- t diff --git a/contrib/citext/sql/citext_utf8.sql b/contrib/citext/sql/citext_utf8.sql index d068000b423..942daa9ce50 100644 --- a/contrib/citext/sql/citext_utf8.sql +++ b/contrib/citext/sql/citext_utf8.sql @@ -24,10 +24,12 @@ SELECT 'À'::citext <> 'B'::citext AS t; SELECT 'Ä'::text <> 'Ä'::text AS t; SELECT 'Ä'::citext <> 'Ä'::citext AS t; --- Test the Turkish dotted I. The lowercase is a single byte while the +-- Test the Turkish dotted I. The lowercase might be a single byte while the -- uppercase is multibyte. This is why the comparison code can't be optimized -- to compare string lengths. -SELECT 'i'::citext = 'İ'::citext AS t; +-- Note that lower('İ') is 'i' (U+0069) in tr and az locales, +-- but 'i̇' (U+0069 U+0307) in C and most (all?) other locales. +SELECT 'İ'::citext in ('i'::citext, 'i̇'::citext) AS t; -- Regression. SELECT 'láska'::citext <> 'laská'::citext AS t; -- 2.38.1