# name: test/sql/function/string/regex_search.test
# description: regex search test
# group: [string]

statement ok
PRAGMA enable_verification

statement ok
CREATE TABLE t0 as FROM VALUES('asdf') t(c0);

# null
query I
SELECT regexp_matches(c0, NULL) from t0
----
NULL

# constant strings
query T
SELECT regexp_matches(c0, '.*sd.*') from t0;
----
1

query T
SELECT regexp_matches(c0, '.*yu.*') from t0;
----
0

query T
SELECT regexp_matches(c0, '') from t0;
----
1

# partial matches okay
query T
SELECT regexp_matches(c0, 'sd') from t0;
----
1

query T
SELECT regexp_full_match(c0, 'sd') from t0;
----
0

query T
SELECT regexp_full_match(c0, '.sd.') from t0;
----
1

query T
SELECT regexp_matches(c0, '^sdf$') from t0;
----
0

# empty strings
query T
SELECT regexp_matches('', '.*yu.*')
----
0

query T
SELECT regexp_matches('', '.*')
----
1

# NULLs
query T
SELECT regexp_matches(c0, CAST(NULL AS STRING)) from t0;
----
NULL

query T
SELECT regexp_matches(CAST(NULL AS STRING), '.*sd.*')
----
NULL

query T
SELECT regexp_matches(CAST(NULL AS STRING), CAST(NULL AS STRING))
----
NULL

query T
SELECT regexp_matches('foobarbequebaz', '(bar)(beque)')
----
1

# postgres says throw error on invalid regex
statement error
SELECT regexp_matches('', '\X')
----
<REGEX>:.*Invalid Input Error: invalid escape sequence.*

statement ok
CREATE TABLE regex(s STRING, p STRING)

statement ok
INSERT INTO regex VALUES ('asdf', 'sd'), ('asdf', '^sd'), (NULL, '^sd'), ('asdf', NULL)

query T
SELECT regexp_matches(s, '.*') FROM regex
----
1
1
NULL
1

query T
SELECT regexp_matches(s, p) FROM regex
----
1
0
NULL
NULL

# test regex_matches with options
# case sensitivity
query T
SELECT regexp_matches(c0, '.*SD.*', 'i') from t0;
----
1

query T
SELECT regexp_matches(c0, '.*SD.*', 'c') from t0;
----
0

# literal match
query T
SELECT regexp_matches('as^/$df', '^/$', 'l')
----
1

query T
SELECT regexp_matches('as^/$df', '^/$')
----
0

# dot matches newline
query T
SELECT regexp_matches('hello
world', '.*', 's')
----
1

query T
SELECT regexp_full_match('hello
world', '.*', 'n')
----
0

# whitespace is ignored
query T
SELECT regexp_matches(c0, '.*SD.*', ' i 	') from t0;
----
1

# NULL in options is an error
statement error
SELECT regexp_matches(c0, '.*SD.*', NULL) from t0;
----
<REGEX>:.*Invalid Input Error.*must not be NULL.*

# this also works with tables
statement ok
CREATE TABLE test(v VARCHAR);

statement ok
INSERT INTO test VALUES ('hello'), ('HELLO');

query T
SELECT regexp_matches(v, 'h.*', 'i') FROM test ORDER BY v
----
1
1

query T
SELECT regexp_matches(v, 'h.*', 'c') FROM test ORDER BY v
----
0
1

statement error
SELECT regexp_matches(v, 'h.*', v) FROM test ORDER BY v
----
<REGEX>:.*Invalid Input Error.*must be a constant.*

# throw on invalid options
statement error
SELECT regexp_matches(c0, '.*SD.*', 'q') from t0;
----
<REGEX>:.*Invalid Input Error.*Unrecognized.*

# can only use "g" with regexp replace
statement error
SELECT regexp_matches(c0, '.*SD.*', 'g') from t0;
----
<REGEX>:.*Invalid Input Error.*only valid for regexp_replace.*

# error in non-constant regex
statement ok
INSERT INTO regex VALUES ('asdf', '(')

statement error
SELECT regexp_matches(s, p) FROM regex
----
<REGEX>:.*Invalid Input Error.*missing.*