diff --git a/build/Jamfile.v2 b/build/Jamfile.v2
index a8359dda..312632f8 100644
--- a/build/Jamfile.v2
+++ b/build/Jamfile.v2
@@ -102,11 +102,13 @@ SOURCES =
lib boost_regex : ../src/$(SOURCES) icu_options
:
shared:BOOST_REGEX_DYN_LINK=1
- gcc-mw:static
- gcc-mingw:static
+ #gcc-mw:static
+ #gcc-mingw:static
gcc-cygwin:static
;
boost-install boost_regex ;
+
+
diff --git a/example/Jamfile.v2 b/example/Jamfile.v2
index 82098387..dd958c7d 100644
--- a/example/Jamfile.v2
+++ b/example/Jamfile.v2
@@ -14,8 +14,8 @@ project
gcc:all
gcc:-Wextra
U_USING_ICU_NAMESPACE=0
- gcc-mw:static
- gcc-mingw:static
+ #gcc-mw:static
+ #gcc-mingw:static
gcc-cygwin:static
;
@@ -78,3 +78,4 @@ test-suite regex-examples :
+
diff --git a/include/boost/regex/v4/basic_regex.hpp b/include/boost/regex/v4/basic_regex.hpp
index f807fce9..04c7bb37 100644
--- a/include/boost/regex/v4/basic_regex.hpp
+++ b/include/boost/regex/v4/basic_regex.hpp
@@ -53,7 +53,7 @@ void bubble_down_one(I first, I last)
if(first != last)
{
I next = last - 1;
- while((next != first) && !(*(next-1) < *next))
+ while((next != first) && (*next < *(next-1)))
{
(next-1)->swap(*next);
--next;
@@ -61,70 +61,59 @@ void bubble_down_one(I first, I last)
}
}
-//
-// Class named_subexpressions
-// Contains information about named subexpressions within the regex.
-//
-template
-class named_subexpressions_base
-{
-public:
- virtual int get_id(const charT* i, const charT* j)const = 0;
- virtual int get_id(std::size_t h)const = 0;
-#ifdef __GNUC__
- // warning supression:
- virtual ~named_subexpressions_base(){}
-#endif
-};
-
template
-inline std::size_t hash_value_from_capture_name(Iterator i, Iterator j)
+inline int hash_value_from_capture_name(Iterator i, Iterator j)
{
std::size_t r = boost::hash_range(i, j);
r %= ((std::numeric_limits::max)() - 10001);
r += 10000;
- return r;
+ return static_cast(r);
}
-template
-class named_subexpressions : public named_subexpressions_base
+class named_subexpressions
{
+public:
struct name
{
+ template
name(const charT* i, const charT* j, int idx)
- : /*n(i, j), */ index(idx)
+ : index(idx)
{
hash = hash_value_from_capture_name(i, j);
}
- name(std::size_t h, int idx)
+ name(int h, int idx)
: index(idx), hash(h)
{
}
- //std::vector n;
int index;
- std::size_t hash;
+ int hash;
bool operator < (const name& other)const
{
- return hash < other.hash; //std::lexicographical_compare(n.begin(), n.end(), other.n.begin(), other.n.end());
+ return hash < other.hash;
}
bool operator == (const name& other)const
{
- return hash == other.hash; //n == other.n;
+ return hash == other.hash;
}
void swap(name& other)
{
- //n.swap(other.n);
std::swap(index, other.index);
std::swap(hash, other.hash);
}
};
-public:
+
+ typedef std::vector::const_iterator const_iterator;
+ typedef std::pair range_type;
+
named_subexpressions(){}
+
+ template
void set_name(const charT* i, const charT* j, int index)
{
m_sub_names.push_back(name(i, j, index));
bubble_down_one(m_sub_names.begin(), m_sub_names.end());
}
+ template
int get_id(const charT* i, const charT* j)const
{
name t(i, j, 0);
@@ -135,72 +124,37 @@ public:
}
return -1;
}
- int get_id(std::size_t h)const
+ template
+ range_type equal_range(const charT* i, const charT* j)const
+ {
+ name t(i, j, 0);
+ return std::equal_range(m_sub_names.begin(), m_sub_names.end(), t);
+ }
+ int get_id(int h)const
{
name t(h, 0);
- typename std::vector::const_iterator pos = std::lower_bound(m_sub_names.begin(), m_sub_names.end(), t);
+ std::vector::const_iterator pos = std::lower_bound(m_sub_names.begin(), m_sub_names.end(), t);
if((pos != m_sub_names.end()) && (*pos == t))
{
return pos->index;
}
return -1;
}
+ range_type equal_range(int h)const
+ {
+ name t(h, 0);
+ return std::equal_range(m_sub_names.begin(), m_sub_names.end(), t);
+ }
private:
std::vector m_sub_names;
};
-template
-class named_subexpressions_converter : public named_subexpressions_base
-{
- boost::shared_ptr > m_converter;
-public:
- named_subexpressions_converter(boost::shared_ptr > s)
- : m_converter(s) {}
- int get_id(const charT* i, const charT* j)const
- {
- if(i == j)
- return -1;
- std::vector v;
- while(i != j)
- {
- v.push_back(*i);
- ++i;
- }
- return m_converter->get_id(&v[0], &v[0] + v.size());
- }
- int get_id(std::size_t h)const
- {
- return m_converter->get_id(h);
- }
-};
-
-template
-inline boost::shared_ptr > convert_to_named_subs_imp(
- boost::shared_ptr > s,
- boost::integral_constant const&)
-{
- return s;
-}
-template
-inline boost::shared_ptr > convert_to_named_subs_imp(
- boost::shared_ptr > s,
- boost::integral_constant const&)
-{
- return boost::shared_ptr >(new named_subexpressions_converter(s));
-}
-template
-inline boost::shared_ptr > convert_to_named_subs(
- boost::shared_ptr > s)
-{
- typedef typename boost::is_same::type tag_type;
- return convert_to_named_subs_imp(s, tag_type());
-}
//
// class regex_data:
// represents the data we wish to expose to the matching algorithms.
//
template
-struct regex_data : public named_subexpressions
+struct regex_data : public named_subexpressions
{
typedef regex_constants::syntax_option_type flag_type;
typedef std::size_t size_type;
@@ -672,7 +626,7 @@ public:
BOOST_ASSERT(0 != m_pimpl.get());
return m_pimpl->get_data();
}
- boost::shared_ptr > get_named_subs()const
+ boost::shared_ptr get_named_subs()const
{
return m_pimpl;
}
diff --git a/include/boost/regex/v4/basic_regex_creator.hpp b/include/boost/regex/v4/basic_regex_creator.hpp
index f3bc0061..efa9f7dd 100644
--- a/include/boost/regex/v4/basic_regex_creator.hpp
+++ b/include/boost/regex/v4/basic_regex_creator.hpp
@@ -806,7 +806,13 @@ void basic_regex_creator::fixup_recursions(re_syntax_base* state)
re_syntax_base* p = base;
std::ptrdiff_t idx = static_cast(state)->alt.i;
if(idx > 10000)
- idx = m_pdata->get_id(idx);
+ {
+ //
+ // There may be more than one capture group with this hash, just do what Perl
+ // does and recurse to the leftmost:
+ //
+ idx = m_pdata->get_id(static_cast(idx));
+ }
while(p)
{
if((p->type == syntax_element_startmark) && (static_cast(p)->index == idx))
diff --git a/include/boost/regex/v4/basic_regex_parser.hpp b/include/boost/regex/v4/basic_regex_parser.hpp
index 6779383e..4dacfc6f 100644
--- a/include/boost/regex/v4/basic_regex_parser.hpp
+++ b/include/boost/regex/v4/basic_regex_parser.hpp
@@ -820,7 +820,11 @@ escape_type_class_jump:
return false;
}
// maybe have \g{ddd}
- if(this->m_traits.syntax_type(*m_position) == regex_constants::syntax_open_brace)
+ regex_constants::syntax_type syn = this->m_traits.syntax_type(*m_position);
+ regex_constants::syntax_type syn_end = 0;
+ if((syn == regex_constants::syntax_open_brace)
+ || (syn == regex_constants::escape_type_left_word)
+ || (syn == regex_constants::escape_type_end_buffer))
{
if(++m_position == m_end)
{
@@ -828,6 +832,18 @@ escape_type_class_jump:
return false;
}
have_brace = true;
+ switch(syn)
+ {
+ case regex_constants::syntax_open_brace:
+ syn_end = regex_constants::syntax_close_brace;
+ break;
+ case regex_constants::escape_type_left_word:
+ syn_end = regex_constants::escape_type_right_word;
+ break;
+ default:
+ syn_end = regex_constants::escape_type_end_buffer;
+ break;
+ }
}
negative = (*m_position == static_cast('-'));
if((negative) && (++m_position == m_end))
@@ -837,18 +853,20 @@ escape_type_class_jump:
}
const charT* pc = m_position;
int i = this->m_traits.toi(pc, m_end, 10);
- if(i < 0)
+ if((i < 0) && syn_end)
{
- // Check for a named capture:
+ // Check for a named capture, get the leftmost one if there is more than one:
const charT* base = m_position;
- while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_brace))
+ while((m_position != m_end) && (this->m_traits.syntax_type(*m_position) != syn_end))
+ {
++m_position;
- i = this->m_pdata->get_id(base, m_position);
+ }
+ i = hash_value_from_capture_name(base, m_position);
pc = m_position;
}
if(negative)
i = 1 + m_mark_count - i;
- if((i > 0) && (this->m_backrefs & (1u << (i-1))))
+ if(((i > 0) && (this->m_backrefs & (1u << (i-1)))) || ((i > 10000) && (this->m_pdata->get_id(i) > 0) && (this->m_backrefs & (1u << (this->m_pdata->get_id(i)-1)))))
{
m_position = pc;
re_brace* pb = static_cast(this->append_state(syntax_element_backref, sizeof(re_brace)));
@@ -863,7 +881,7 @@ escape_type_class_jump:
m_position = pc;
if(have_brace)
{
- if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != regex_constants::syntax_close_brace))
+ if((m_position == m_end) || (this->m_traits.syntax_type(*m_position) != syn_end))
{
fail(regex_constants::error_escape, m_position - m_base, incomplete_message);
return false;
diff --git a/include/boost/regex/v4/match_results.hpp b/include/boost/regex/v4/match_results.hpp
index 8f11bbbe..ca9898f4 100644
--- a/include/boost/regex/v4/match_results.hpp
+++ b/include/boost/regex/v4/match_results.hpp
@@ -38,7 +38,6 @@ namespace boost{
namespace re_detail{
-template
class named_subexpressions;
}
@@ -69,7 +68,7 @@ public:
typedef typename re_detail::regex_iterator_traits<
BidiIterator>::value_type char_type;
typedef std::basic_string string_type;
- typedef re_detail::named_subexpressions_base named_sub_type;
+ typedef re_detail::named_subexpressions named_sub_type;
// construct/copy/destroy:
explicit match_results(const Allocator& a = Allocator())
@@ -225,10 +224,15 @@ public:
//
const_reference named_subexpression(const char_type* i, const char_type* j) const
{
+ //
+ // Scan for the leftmost *matched* subexpression with the specified named:
+ //
if(m_is_singular)
raise_logic_error();
- int index = m_named_subs->get_id(i, j);
- return index > 0 ? (*this)[index] : m_null;
+ re_detail::named_subexpressions::range_type r = m_named_subs->equal_range(i, j);
+ while((r.first != r.second) && ((*this)[r.first->index].matched == false))
+ ++r.first;
+ return r.first != r.second ? (*this)[r.first->index] : m_null;
}
template
const_reference named_subexpression(const charT* i, const charT* j) const
@@ -243,10 +247,20 @@ public:
}
int named_subexpression_index(const char_type* i, const char_type* j) const
{
+ //
+ // Scan for the leftmost *matched* subexpression with the specified named.
+ // If none found then return the leftmost expression with that name,
+ // otherwise an invalid index:
+ //
if(m_is_singular)
raise_logic_error();
- int index = m_named_subs->get_id(i, j);
- return index > 0 ? index : -20;
+ re_detail::named_subexpressions::range_type s, r;
+ s = r = m_named_subs->equal_range(i, j);
+ while((r.first != r.second) && ((*this)[r.first->index].matched == false))
+ ++r.first;
+ if(r.first == r.second)
+ r = s;
+ return r.first != r.second ? r.first->index : -20;
}
template
int named_subexpression_index(const charT* i, const charT* j) const
diff --git a/include/boost/regex/v4/perl_matcher_common.hpp b/include/boost/regex/v4/perl_matcher_common.hpp
index fe224210..b8c4e963 100644
--- a/include/boost/regex/v4/perl_matcher_common.hpp
+++ b/include/boost/regex/v4/perl_matcher_common.hpp
@@ -200,7 +200,7 @@ bool perl_matcher::match_imp()
m_match_flags |= regex_constants::match_all;
m_presult->set_size((m_match_flags & match_nosubs) ? 1 : re.mark_count(), search_base, last);
m_presult->set_base(base);
- m_presult->set_named_subs(re_detail::convert_to_named_subs::char_type>(this->re.get_named_subs()));
+ m_presult->set_named_subs(this->re.get_named_subs());
if(m_match_flags & match_posix)
m_result = *m_presult;
verify_options(re.flags(), m_match_flags);
@@ -262,7 +262,7 @@ bool perl_matcher::find_imp()
pstate = re.get_first_state();
m_presult->set_size((m_match_flags & match_nosubs) ? 1 : re.mark_count(), base, last);
m_presult->set_base(base);
- m_presult->set_named_subs(re_detail::convert_to_named_subs::char_type>(this->re.get_named_subs()));
+ m_presult->set_named_subs(this->re.get_named_subs());
m_match_flags |= regex_constants::match_init;
}
else
@@ -588,8 +588,23 @@ bool perl_matcher::match_backref()
// in the match, this is in line with ECMAScript, but not Perl
// or PCRE.
//
- BidiIterator i = (*m_presult)[static_cast(pstate)->index].first;
- BidiIterator j = (*m_presult)[static_cast(pstate)->index].second;
+ int index = static_cast(pstate)->index;
+ if(index >= 10000)
+ {
+ named_subexpressions::range_type r = re.get_data().equal_range(index);
+ BOOST_ASSERT(r.first != r.second);
+ do
+ {
+ index = r.first->index;
+ ++r.first;
+ }while((r.first != r.second) && ((*m_presult)[index].matched != true));
+ }
+
+ if((m_match_flags & match_perl) && !(*m_presult)[index].matched)
+ return false;
+
+ BidiIterator i = (*m_presult)[index].first;
+ BidiIterator j = (*m_presult)[index].second;
while(i != j)
{
if((position == last) || (traits_inst.translate(*position, icase) != traits_inst.translate(*i, icase)))
@@ -713,7 +728,7 @@ inline bool perl_matcher::match_assert_backref(
{
// return true if marked sub-expression N has been matched:
int index = static_cast(pstate)->index;
- bool result;
+ bool result = false;
if(index == 9999)
{
// Magic value for a (DEFINE) block:
@@ -721,11 +736,25 @@ inline bool perl_matcher::match_assert_backref(
}
else if(index > 0)
{
+ // Have we matched subexpression "index"?
// Check if index is a hash value:
if(index >= 10000)
- index = re.get_data().get_id(index);
- // Have we matched subexpression "index"?
- result = (*m_presult)[index].matched;
+ {
+ named_subexpressions::range_type r = re.get_data().equal_range(index);
+ while(r.first != r.second)
+ {
+ if((*m_presult)[r.first->index].matched)
+ {
+ result = true;
+ break;
+ }
+ ++r.first;
+ }
+ }
+ else
+ {
+ result = (*m_presult)[index].matched;
+ }
pstate = pstate->next.p;
}
else
@@ -734,8 +763,20 @@ inline bool perl_matcher::match_assert_backref(
// If index == 0 then check for any recursion at all, otherwise for recursion to -index-1.
int idx = -index-1;
if(idx >= 10000)
- idx = re.get_data().get_id(idx);
- result = !recursion_stack.empty() && ((recursion_stack.back().idx == idx) || (index == 0));
+ {
+ named_subexpressions::range_type r = re.get_data().equal_range(idx);
+ int stack_index = recursion_stack.empty() ? -1 : recursion_stack.back().idx;
+ while(r.first != r.second)
+ {
+ result |= (stack_index == r.first->index);
+ if(result)break;
+ ++r.first;
+ }
+ }
+ else
+ {
+ result = !recursion_stack.empty() && ((recursion_stack.back().idx == idx) || (index == 0));
+ }
pstate = pstate->next.p;
}
return result;
diff --git a/include/boost/regex/v4/sub_match.hpp b/include/boost/regex/v4/sub_match.hpp
index 1c79e39a..34a86840 100644
--- a/include/boost/regex/v4/sub_match.hpp
+++ b/include/boost/regex/v4/sub_match.hpp
@@ -56,7 +56,7 @@ struct sub_match : public std::pair
template
operator std::basic_string ()const
{
- return std::basic_string(this->first, this->second);
+ return matched ? std::basic_string(this->first, this->second) : std::basic_string();
}
#else
operator std::basic_string ()const
@@ -66,19 +66,22 @@ struct sub_match : public std::pair
#endif
difference_type BOOST_REGEX_CALL length()const
{
- difference_type n = ::boost::re_detail::distance((BidiIterator)this->first, (BidiIterator)this->second);
+ difference_type n = matched ? ::boost::re_detail::distance((BidiIterator)this->first, (BidiIterator)this->second) : 0;
return n;
}
std::basic_string str()const
{
std::basic_string result;
- std::size_t len = ::boost::re_detail::distance((BidiIterator)this->first, (BidiIterator)this->second);
- result.reserve(len);
- BidiIterator i = this->first;
- while(i != this->second)
+ if(matched)
{
- result.append(1, *i);
- ++i;
+ std::size_t len = ::boost::re_detail::distance((BidiIterator)this->first, (BidiIterator)this->second);
+ result.reserve(len);
+ BidiIterator i = this->first;
+ while(i != this->second)
+ {
+ result.append(1, *i);
+ ++i;
+ }
}
return result;
}
diff --git a/src/regex_traits_defaults.cpp b/src/regex_traits_defaults.cpp
index 31b79184..5f06149c 100644
--- a/src/regex_traits_defaults.cpp
+++ b/src/regex_traits_defaults.cpp
@@ -100,7 +100,7 @@ BOOST_REGEX_DECL const char* BOOST_REGEX_CALL get_default_syntax(regex_constants
"p",
"P",
"N",
- "g",
+ "gk",
"K",
"R",
};
diff --git a/test/Jamfile.v2 b/test/Jamfile.v2
index 18af0d7d..ce198b56 100644
--- a/test/Jamfile.v2
+++ b/test/Jamfile.v2
@@ -17,8 +17,8 @@ project
gcc:-Wextra
gcc:-Wshadow
U_USING_ICU_NAMESPACE=0
- gcc-mw:static
- gcc-mingw:static
+ #gcc-mw:static
+ #gcc-mingw:static
gcc-cygwin:static
;
@@ -177,3 +177,5 @@ test-suite regex
;
build-project ../example ;
+
+
diff --git a/test/regress/test_perl_ex.cpp b/test/regress/test_perl_ex.cpp
index 05abf22d..f96be261 100644
--- a/test/regress/test_perl_ex.cpp
+++ b/test/regress/test_perl_ex.cpp
@@ -907,5 +907,16 @@ void test_recursion()
// 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));
+
+ // Recursion to a named sub with a name that is used multiple times:
+ TEST_REGEX_SEARCH("(?:(?a+)|(?b+))\\.(?&A)", perl, "aaaa.aa", match_default, make_array(0, 7, 0, 4, -1, -1, -2, -2));
+ TEST_REGEX_SEARCH("(?:(?a+)|(?b+))\\.(?&A)", perl, "bbbb.aa", match_default, make_array(0, 7, -1, -1, 0, 4, -2, -2));
+ TEST_REGEX_SEARCH("(?:(?a+)|(?b+))\\.(?&A)", perl, "bbbb.bb", match_default, make_array(-2, -2));
+ // Back reference to a named sub with a name that is used multiple times:
+ TEST_REGEX_SEARCH("(?:(?a+)|(?b+))\\.\\k", perl, "aaaa.aaaa", match_default, make_array(0, 9, 0, 4, -1, -1, -2, -2));
+ TEST_REGEX_SEARCH("(?:(?a+)|(?b+))\\.\\k", perl, "bbbb.aaaa", match_default, make_array(-2, -2));
+ TEST_REGEX_SEARCH("(?:(?a+)|(?b+))\\.\\k", perl, "aaaa.bbbb", match_default, make_array(-2, -2));
+ TEST_REGEX_SEARCH("(?:(?a+)|(?b+))\\.\\k", perl, "bbbb.bbbb", match_default, make_array(0, 9, -1, -1, 0, 4, -2, -2));
+ TEST_REGEX_SEARCH("(?:(?a+)|(?b+)|c+)\\.\\k", perl, "cccc.cccc", match_default, make_array(-2, -2));
}
diff --git a/test/regress/test_replace.cpp b/test/regress/test_replace.cpp
index 06c137bf..2d8724eb 100644
--- a/test/regress/test_replace.cpp
+++ b/test/regress/test_replace.cpp
@@ -184,5 +184,11 @@ void test_replace()
TEST_REGEX_REPLACE("(?a+)(?b+)", perl, " ...aabb,,", match_default|format_no_copy, "$2", "bb");
TEST_REGEX_REPLACE("(?a+)(?b+)", perl, " ...aabb,,", match_default|format_no_copy, "d$+{one}c", "daac");
TEST_REGEX_REPLACE("(?a+)(?b+)", perl, " ...aabb,,", match_default|format_no_copy, "c$+{two}d", "cbbd");
+
+ TEST_REGEX_REPLACE("(?a)(?b)(c)", perl, " ...abc,,", match_default, "$1.$2.$3.$+{one}", " ...a.b.c.a,,");
+ TEST_REGEX_REPLACE("(?:(?a)|(?b))", perl, " ...a,,", match_default, "$1.$2.$+{one}", " ...a..a,,");
+ TEST_REGEX_REPLACE("(?:(?a)|(?b))", perl, " ...b,,", match_default, "$1.$2.$+{one}", " ....b.b,,");
+ TEST_REGEX_REPLACE("(?:(?a)(?b))", perl, " ...ab,,", match_default, "$1.$2.$+{one}", " ...a.b.a,,");
+
}
diff --git a/test/regress/test_tricky_cases.cpp b/test/regress/test_tricky_cases.cpp
index e025dc43..033c3dea 100644
--- a/test/regress/test_tricky_cases.cpp
+++ b/test/regress/test_tricky_cases.cpp
@@ -49,7 +49,7 @@ void test_tricky_cases()
TEST_REGEX_SEARCH("(a)d|(b)c", perl, "abc", match_default, make_array(1, 3, -1, -1, 1, 2, -2, -2));
TEST_REGEX_SEARCH("_+((www)|(ftp)|(mailto)):_*", perl, "_wwwnocolon _mailto:", match_default, make_array(12, 20, 13, 19, -1, -1, -1, -1, 13, 19, -2, -2));
// subtleties of matching
- TEST_REGEX_SEARCH("a(b)?c\\1d", perl, "acd", match_default, make_array(0, 3, -1, -1, -2, -2));
+ TEST_REGEX_SEARCH("a(b)?c\\1d", perl, "acd", match_default, make_array(-2, -2));
TEST_REGEX_SEARCH("a(b?c)+d", perl, "accd", match_default, make_array(0, 4, 2, 3, -2, -2));
TEST_REGEX_SEARCH("(wee|week)(knights|night)", perl, "weeknights", match_default, make_array(0, 10, 0, 3, 3, 10, -2, -2));
TEST_REGEX_SEARCH(".*", perl, "abc", match_default, make_array(0, 3, -2, 3, 3, -2, -2));