Skip to content

Apache DataFusion SQL Query Engine

Add more regexp_replace test coverage (#21485)

Andrew Lamb6cf94c73d613
diff --git a/datafusion/sqllogictest/test_files/regexp/regexp_replace.slt b/datafusion/sqllogictest/test_files/regexp/regexp_replace.slt
index 99e0b10430..e27ff1e9c1 100644
--- a/datafusion/sqllogictest/test_files/regexp/regexp_replace.slt
+++ b/datafusion/sqllogictest/test_files/regexp/regexp_replace.slt
@@ -128,43 +128,6 @@ from (values ('a'), ('b')) as tbl(col);
NULL NULL NULL
NULL NULL NULL
-# Extract domain from URL using anchored pattern with trailing .*
-# This tests that the full URL suffix is replaced, not just the matched prefix
-query T
-SELECT regexp_replace(url, '^https?://(?:www\.)?([^/]+)/.*$', '\1') FROM (VALUES
- ('https://www.example.com/path/to/page?q=1'),
- ('http://test.org/foo/bar'),
- ('https://example.com/'),
- ('not-a-url')
-) AS t(url);
-----
-example.com
-test.org
-example.com
-not-a-url
-
-# More than one capture group should disable the short-regex fast path.
-# This still uses replacement \1, but captures_len() will be > 2, so the
-# implementation must fall back to the normal regexp_replace path.
-query T
-SELECT regexp_replace(url, '^https?://((www\.)?([^/]+))/.*$', '\1') FROM (VALUES
- ('https://www.example.com/path/to/page?q=1'),
- ('http://test.org/foo/bar'),
- ('not-a-url')
-) AS t(url);
-----
-www.example.com
-test.org
-not-a-url
-
-# If the overall pattern matches but capture group 1 does not participate,
-# regexp_replace(..., '\1') should substitute the empty string, not keep
-# the original input.
-query B
-SELECT regexp_replace('bzzz', '^(a)?b.*$', '\1') = '';
-----
-true
-
# Stripping trailing .*$ must not change match semantics for inputs with
# newlines when the original pattern does not use the 's' flag.
query B
@@ -183,3 +146,111 @@ SELECT regexp_replace(
) = concat('x', chr(10), 'rest');
----
true
+
+
+# Fixture for testing optimizations in regexp_replace
+statement ok
+CREATE TABLE regexp_replace_optimized_cases (
+ value string,
+ regexp string,
+ replacement string,
+ expected string
+);
+
+# Extract domain from URL using anchored pattern with trailing .*
+# This tests that the full URL suffix is replaced, not just the matched prefix.
+statement ok
+INSERT INTO regexp_replace_optimized_cases VALUES
+ ('https://www.example.com/path/to/page?q=1', '^https?://(?:www\.)?([^/]+)/.*$', '\1', 'example.com'),
+ ('http://test.org/foo/bar', '^https?://(?:www\.)?([^/]+)/.*$', '\1', 'test.org'),
+ ('https://example.com/', '^https?://(?:www\.)?([^/]+)/.*$', '\1', 'example.com'),
+ ('not-a-url', '^https?://(?:www\.)?([^/]+)/.*$', '\1', 'not-a-url');
+
+# More than one capture group should disable the short-regex fast path.
+# This still uses replacement \1, but captures_len() will be > 2, so the
+# implementation must fall back to the normal regexp_replace path.
+statement ok
+INSERT INTO regexp_replace_optimized_cases VALUES
+ ('https://www.example.com/path/to/page?q=1', '^https?://((www\.)?([^/]+))/.*$', '\1', 'www.example.com'),
+ ('http://test.org/foo/bar', '^https?://((www\.)?([^/]+))/.*$', '\1', 'test.org'),
+ ('not-a-url', '^https?://((www\.)?([^/]+))/.*$', '\1', 'not-a-url');
+
+# If the overall pattern matches but capture group 1 does not participate,
+# regexp_replace(..., '\1') should substitute the empty string, not keep
+# the original input.
+statement ok
+INSERT INTO regexp_replace_optimized_cases VALUES
+ ('bzzz', '^(a)?b.*$', '\1', '');
+
+
+query TB
+SELECT value, regexp_replace(value, regexp, replacement) = expected
+FROM regexp_replace_optimized_cases
+ORDER BY regexp, value, replacement, expected;
+----
+bzzz true
+http://test.org/foo/bar true
+https://www.example.com/path/to/page?q=1 true
+not-a-url true
+http://test.org/foo/bar true
+https://example.com/ true
+https://www.example.com/path/to/page?q=1 true
+not-a-url true
+
+query TB
+SELECT value, regexp_replace(
+ arrow_cast(value, 'LargeUtf8'),
+ arrow_cast(regexp, 'LargeUtf8'),
+ arrow_cast(replacement, 'LargeUtf8')
+ ) = arrow_cast(expected, 'LargeUtf8')
+FROM regexp_replace_optimized_cases
+ORDER BY regexp, value, replacement, expected;
+----
+bzzz true
+http://test.org/foo/bar true
+https://www.example.com/path/to/page?q=1 true
+not-a-url true
+http://test.org/foo/bar true
+https://example.com/ true
+https://www.example.com/path/to/page?q=1 true
+not-a-url true
+
+query TB
+SELECT value, regexp_replace(
+ arrow_cast(value, 'Utf8View'),
+ arrow_cast(regexp, 'Utf8View'),
+ arrow_cast(replacement, 'Utf8View')
+ ) = arrow_cast(expected, 'Utf8View')
+FROM regexp_replace_optimized_cases
+ORDER BY regexp, value, replacement, expected;
+----
+bzzz true
+http://test.org/foo/bar true
+https://www.example.com/path/to/page?q=1 true
+not-a-url true
+http://test.org/foo/bar true
+https://example.com/ true
+https://www.example.com/path/to/page?q=1 true
+not-a-url true
+
+query TB
+SELECT value, regexp_replace(
+ arrow_cast(value, 'Dictionary(Int32, Utf8)'),
+ arrow_cast(regexp, 'Dictionary(Int32, Utf8)'),
+ arrow_cast(replacement, 'Dictionary(Int32, Utf8)')
+ ) = arrow_cast(expected, 'Dictionary(Int32, Utf8)')
+FROM regexp_replace_optimized_cases
+ORDER BY regexp, value, replacement, expected;
+----
+bzzz true
+http://test.org/foo/bar true
+https://www.example.com/path/to/page?q=1 true
+not-a-url true
+http://test.org/foo/bar true
+https://example.com/ true
+https://www.example.com/path/to/page?q=1 true
+not-a-url true
+
+# cleanup
+statement ok
+DROP TABLE regexp_replace_optimized_cases;