Skip to content

<regex>: Avoid generating unnecessary single-branch _N_if nodes #5539

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
StephanTLavavej merged 1 commit into from
May 28, 2025
Merged

<regex>: Avoid generating unnecessary single-branch _N_if nodes #5539

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
34 changes: 24 additions & 10 deletions stl/inc/regex
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4125,6 +4125,9 @@ _BidIt _Matcher2<_BidIt, _Elem, _RxTraits, _It, _Alloc>::_Skip(_BidIt _First_arg
// GH-5452: If this node has two or more branches,
// examining all these branches has quadratic worst-case complexity.
// Thus, we only continue if this node has a single branch only.
// TRANSITION, ABI: After GH-5539, the parser no longer generates single-branch _N_if nodes.
// But we have to retain this special handling to avoid performance regression
// when an old parser gets mixed with a new matcher.
_Node_if* _Node = static_cast<_Node_if*>(_Nx);

if (_Node->_Child) {
Expand Down Expand Up @@ -4903,15 +4906,17 @@ void _Parser2<_FwdIt, _Elem, _RxTraits>::_Disjunction() { // check for valid dis
_Nfa._End_group(_Pos3);
}

_Node_base* _Pos2 = _Nfa._Begin_if(_Pos1);
while (_Mchar == _Meta_bar) { // append terms as long as we keep finding | characters
_Next();
if (!_Alternative()) { // zero-length trailing alternative
_Node_base* _Pos3 = _Nfa._Begin_group();
_Nfa._End_group(_Pos3);
}
if (_Mchar == _Meta_bar) {
_Node_base* _Pos2 = _Nfa._Begin_if(_Pos1);
do { // append terms as long as we keep finding | characters
_Next();
if (!_Alternative()) { // zero-length trailing alternative
_Node_base* _Pos3 = _Nfa._Begin_group();
_Nfa._End_group(_Pos3);
}

_Nfa._Else_if(_Pos1, _Pos2);
_Nfa._Else_if(_Pos1, _Pos2);
} while (_Mchar == _Meta_bar);
}
}

Expand Down Expand Up @@ -4979,17 +4984,26 @@ void _Parser2<_FwdIt, _Elem, _RxTraits>::_Calculate_loop_simplicity(
}
}
break;

case _N_group:
case _N_capture:
// TRANSITION, requires more research to decide on the subset of loops that we can make simple:
// - Simple mode can square the running time when matching a regex to an input string in the current matcher
// - The optimal subset of simple loops for a non-recursive rewrite of the matcher aren't clear yet
Copy link
Member

@StephanTLavavej StephanTLavavej May 28, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This should say "The optimal subset ... isn't clear yet" but it isn't worth resetting testing.

if (_Outer_rep) {
_Outer_rep->_Simple_loop = 0;
}
break;

case _N_none:
case _N_nop:
case _N_bol:
case _N_eol:
case _N_wbound:
case _N_dot:
case _N_str:
case _N_group:
case _N_end_group:
case _N_end_assert:
case _N_capture:
case _N_end_capture:
case _N_back:
case _N_endif:
Expand Down