Avoid determining regexp subexpression matches, when possible.

Identifying the precise match locations for parenthesized subexpressions
is a fairly expensive task given the way our regexp engine works, both
at regexp compile time (where we must create an optimized NFA for each
parenthesized subexpression) and at runtime (where determining exact
match locations requires laborious search).

Up to now we've made little attempt to optimize this situation.  This
patch identifies cases where we know at compile time that we won't
need to know subexpression match locations, and teaches the regexp
compiler to not bother creating per-subexpression regexps for
parenthesis pairs that are not referenced by backrefs elsewhere in
the regexp.  (To preserve semantics, we obviously still have to
pin down the match locations of backref references.)  Users could
have obtained the same results before this by being careful to
write "non capturing" parentheses wherever possible, but few people
bother with that.

Discussion: https://postgr.es/m/2219936.1628115334@sss.pgh.pa.us

1 files changed, 11 insertions, 2 deletions

diff --git a/src/test/modules/test_regex/sql/test_regex.sql b/src/test/modules/test_regex/sql/test_regex.sql
index 9224fdfdd3a..3419564203a 100644
--- a/src/test/modules/test_regex/sql/test_regex.sql
+++ b/src/test/modules/test_regex/sql/test_regex.sql
@@ -63,8 +63,8 @@ select * from test_regex('ab', 'ab', 'b');
 
 -- expectMatch	4.1  -		(a)e		ae	ae	a
 select * from test_regex('(a)e', 'ae', '-');
--- expectMatch	4.2  o		(a)e		ae
-select * from test_regex('(a)e', 'ae', 'o');
+-- expectMatch	4.2  oPR	(.)\1e		abeaae	aae	{}
+select * from test_regex('(.)\1e', 'abeaae', 'oPR');
 -- expectMatch	4.3  b		{\(a\)b}	ab	ab	a
 select * from test_regex('\(a\)b', 'ab', 'b');
 -- expectMatch	4.4  -		a((b)c)		abc	abc	bc	b
@@ -775,6 +775,11 @@ select * from test_regex('(^\w+).*\1', 'abc abc abc', 'LRP');
 select * from test_regex('(^\w+\M).*\1', 'abc abcd abd', 'LRP');
 select * from test_regex('(\w+(?= )).*\1', 'abc abcd abd', 'HLRP');
 
+-- exercise oversize-regmatch_t-array paths in regexec()
+-- (that case is not reachable via test_regex, sadly)
+select substring('fffoooooooooooooooooooooooooooooooo', '^(.)\1(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)(.)');
+select regexp_split_to_array('abcxxxdefyyyghi', '((.))(\1\2)');
+
 -- doing 15 "octal escapes vs back references"
 
 -- # initial zero is always octal
@@ -1011,6 +1016,10 @@ select * from test_regex('(a*)*', 'bc', 'N');
 select * from test_regex(' TO (([a-z0-9._]+|"([^"]+|"")+")+)', 'asd TO foo', 'M');
 -- expectMatch	21.36 RPQ	((.))(\2){0}	xy	x	x	x	{}
 select * from test_regex('((.))(\2){0}', 'xy', 'RPQ');
+-- expectMatch	21.37 RP	((.))(\2)	xyy	yy	y	y	y
+select * from test_regex('((.))(\2)', 'xyy', 'RP');
+-- expectMatch	21.38 oRP	((.))(\2)	xyy	yy	{}	{}	{}
+select * from test_regex('((.))(\2)', 'xyy', 'oRP');
 
 -- doing 22 "multicharacter collating elements"
 -- # again ugh