boostorg/regex
1
0
Fork 1
mirror of https://github.com/boostorg/regex.git synced 2025-07-18 14:52:08 +02:00
Code Issues Packages Projects Releases Wiki Activity

Fix bug that effects recursive expressions combined with repeats.

Browse Source 
[SVN r58722]
This commit is contained in:
John Maddock
2010-01-05 18:04:08 +00:00
parent f7e16093de
commit 0e1e9804da
6 changed files with 63 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

35
include/boost/regex/v4/basic_regex_creator.hpp
View File

@ -811,8 +811,43 @@ void basic_regex_creator<charT, traits>::fixup_recursions(re_syntax_base* state)
{
if((p->type == syntax_element_startmark) && (static_cast<re_brace*>(p)->index == id))
{
//
// We've found the target of the recursion, set the jump target:
//
static_cast<re_jump*>(state)->alt.p = p;
ok = true;
//
// Now scan the target for nested repeats:
//
p = p->next.p;
int next_rep_id = 0;
while(p)
{
switch(p->type)
{
case syntax_element_rep:
case syntax_element_dot_rep:
case syntax_element_char_rep:
case syntax_element_short_set_rep:
case syntax_element_long_set_rep:
next_rep_id = static_cast<re_repeat*>(p)->state_id;
break;
case syntax_element_endmark:
if(static_cast<const re_brace*>(p)->index == id)
next_rep_id = -1;
break;
default:
break;
}
if(next_rep_id)
break;
p = p->next.p;
}
if(next_rep_id > 0)
{
static_cast<re_recurse*>(state)->state_id = next_rep_id - 1;
}
break;
}
p = p->next.p;

4
include/boost/regex/v4/basic_regex_parser.hpp
View File

@ -1939,7 +1939,9 @@ bool basic_regex_parser<charT, traits>::parse_perl_extension()
}
insert_recursion:
pb->index = markid = 0;
static_cast<re_jump*>(this->append_state(syntax_element_recurse, sizeof(re_jump)))->alt.i = v;
re_recurse* pr = static_cast<re_recurse*>(this->append_state(syntax_element_recurse, sizeof(re_recurse)));
pr->alt.i = v;
pr->state_id = 0;
static_cast<re_case*>(
this->append_state(syntax_element_toggle_case, sizeof(re_case))
)->icase = this->flags() & regbase::icase;

7
include/boost/regex/v4/perl_matcher.hpp
View File

@ -277,11 +277,16 @@ public:
else
{
repeater_count* p = next;
while(p->state_id != state_id)
while(p && (p->state_id != state_id))
p = p->next;
if(p)
{
count = p->count;
start_pos = p->start_pos;
}
else
count = 0;
}
}
~repeater_count()
{

5
include/boost/regex/v4/perl_matcher_non_recursive.hpp
View File

@ -904,10 +904,15 @@ bool perl_matcher<BidiIterator, Allocator, traits>::match_recursion()
}
recursion_stack[recursion_stack_position].preturn_address = pstate->next.p;
recursion_stack[recursion_stack_position].results = *m_presult;
if(static_cast<const re_recurse*>(pstate)->state_id > 0)
{
push_repeater_count(static_cast<const re_recurse*>(pstate)->state_id, &next_count);
}
pstate = static_cast<const re_jump*>(pstate)->alt.p;
recursion_stack[recursion_stack_position].id = static_cast<const re_brace*>(pstate)->index;
++recursion_stack_position;
//BOOST_ASSERT(recursion_stack[recursion_stack_position-1].id);
return true;
}

8
include/boost/regex/v4/states.hpp
View File

@ -248,6 +248,14 @@ struct re_repeat : public re_alt
bool greedy; // True if this is a greedy repeat
};
/*** struct re_recurse ************************************************
Recurse to a particular subexpression.
**********************************************************************/
struct re_recurse : public re_jump
{
int state_id; // identifier of first nested repeat within the recursion.
};
/*** enum re_jump_size_type *******************************************
Provides compiled size of re_jump structure (allowing for trailing alignment).
We provide this so we know how manybytes to insert when constructing the machine

4
test/regress/test_perl_ex.cpp
View File

@ -892,5 +892,9 @@ void test_recursion()
TEST_REGEX_SEARCH("\\b(?&byte)(\\.(?&byte)){3}(?(DEFINE)(?<byte>2[0-4]\\d|25[0-5]|1\\d\\d|[1-9]?\\d))", perl|mod_x, "10.0.0.0", match_default, make_array(0, 8, 6, 8, -1, -1, -2, -2));
TEST_REGEX_SEARCH("\\b(?&byte)(\\.(?&byte)){3}(?(DEFINE)(?<byte>2[0-4]\\d|25[0-5]|1\\d\\d|[1-9]?\\d))", perl|mod_x, "10.6", match_default, make_array(-2, -2));
TEST_REGEX_SEARCH("\\b(?&byte)(\\.(?&byte)){3}(?(DEFINE)(?<byte>2[0-4]\\d|25[0-5]|1\\d\\d|[1-9]?\\d))", perl|mod_x, "455.3.4.5", match_default, make_array(-2, -2));
// Bugs:
TEST_REGEX_SEARCH("namespace\\s+(\\w+)\\s+(\\{(?:[^{}]*(?:(?2)[^{}]*)*)?\\})", perl, "namespace one { namespace two { int foo(); } }", match_default, make_array(0, 46, 10, 13, 14, 46, -2, -2));
TEST_REGEX_SEARCH("namespace\\s+(\\w+)\\s+(\\{(?:[^{}]*(?:(?2)[^{}]*)*)?\\})", perl, "namespace one { namespace two { int foo(){} } { {{{ } } } } {}}", match_default, make_array(0, 64, 10, 13, 14, 64, -2, -2));
}