boostorg/regex
1
0
Fork 1
mirror of https://github.com/boostorg/regex.git synced 2025-07-22 16:47:17 +02:00
Code Issues Packages Projects Releases Wiki Activity

Added possessive modifiers ++ *+ ?+ {}+.

Browse Source 
Added support for \v and \h as character classes as per Perl-5.10. 

[SVN r52558]
This commit is contained in:
John Maddock
2009-04-23 09:51:31 +00:00
parent ccf465daac
commit 7b10b5dac5
96 changed files with 521 additions and 286 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
src/c_regex_traits.cpp
View File

@ -122,7 +122,9 @@ enum
char_class_graph=char_class_alnum|char_class_punct,
char_class_blank=1<<9,
char_class_word=1<<10,
char_class_unicode=1<<11
char_class_unicode=1<<11,
char_class_horizontal=1<<12,
char_class_vertical=1<<13
};
c_regex_traits<char>::char_class_type BOOST_REGEX_CALL c_regex_traits<char>::lookup_classname(const char* p1, const char* p2)
@ -137,6 +139,7 @@ c_regex_traits<char>::char_class_type BOOST_REGEX_CALL c_regex_traits<char>::loo
char_class_digit,
char_class_digit,
char_class_graph,
char_class_horizontal,
char_class_lower,
char_class_lower,
char_class_print,
@ -146,6 +149,7 @@ c_regex_traits<char>::char_class_type BOOST_REGEX_CALL c_regex_traits<char>::loo
char_class_upper,
char_class_unicode,
char_class_upper,
char_class_vertical,
char_class_alnum | char_class_word,
char_class_alnum | char_class_word,
char_class_xdigit,
@ -176,7 +180,9 @@ bool BOOST_REGEX_CALL c_regex_traits<char>::isctype(char c, char_class_type mask
|| ((mask & char_class_punct) && (std::ispunct)(static_cast<unsigned char>(c)))
|| ((mask & char_class_xdigit) && (std::isxdigit)(static_cast<unsigned char>(c)))
|| ((mask & char_class_blank) && (std::isspace)(static_cast<unsigned char>(c)) && !::boost::re_detail::is_separator(c))
|| ((mask & char_class_word) && (c == '_'));
|| ((mask & char_class_word) && (c == '_'))
|| ((mask & char_class_vertical) && (::boost::re_detail::is_separator(c) || (c == '\v')))
|| ((mask & char_class_horizontal) && (std::isspace)(static_cast<unsigned char>(c)) && !::boost::re_detail::is_separator(c) && (c != '\v'));
}
c_regex_traits<char>::string_type BOOST_REGEX_CALL c_regex_traits<char>::lookup_collatename(const char* p1, const char* p2)

8
src/icu.cpp
View File

@ -101,6 +101,8 @@ const icu_regex_traits::char_class_type icu_regex_traits::mask_underscore = icu_
const icu_regex_traits::char_class_type icu_regex_traits::mask_unicode = icu_regex_traits::char_class_type(1) << offset_unicode;
const icu_regex_traits::char_class_type icu_regex_traits::mask_any = icu_regex_traits::char_class_type(1) << offset_any;
const icu_regex_traits::char_class_type icu_regex_traits::mask_ascii = icu_regex_traits::char_class_type(1) << offset_ascii;
const icu_regex_traits::char_class_type icu_regex_traits::mask_horizontal = icu_regex_traits::char_class_type(1) << offset_horizontal;
const icu_regex_traits::char_class_type icu_regex_traits::mask_vertical = icu_regex_traits::char_class_type(1) << offset_vertical;
icu_regex_traits::char_class_type icu_regex_traits::lookup_icu_mask(const ::UChar32* p1, const ::UChar32* p2)
{
@ -370,6 +372,7 @@ icu_regex_traits::char_class_type icu_regex_traits::lookup_classname(const char_
U_GC_ND_MASK,
U_GC_ND_MASK,
(0x3FFFFFFFu) & ~(U_GC_CC_MASK | U_GC_CF_MASK | U_GC_CS_MASK | U_GC_CN_MASK | U_GC_Z_MASK),
mask_horizontal,
U_GC_LL_MASK,
U_GC_LL_MASK,
~(U_GC_C_MASK),
@ -379,6 +382,7 @@ icu_regex_traits::char_class_type icu_regex_traits::lookup_classname(const char_
U_GC_LU_MASK,
mask_unicode,
U_GC_LU_MASK,
mask_vertical,
char_class_type(U_GC_L_MASK | U_GC_ND_MASK | U_GC_MN_MASK) | mask_underscore,
char_class_type(U_GC_L_MASK | U_GC_ND_MASK | U_GC_MN_MASK) | mask_underscore,
char_class_type(U_GC_ND_MASK) | mask_xdigit,
@ -487,6 +491,10 @@ bool icu_regex_traits::isctype(char_type c, char_class_type f) const
return true;
if(((f & mask_ascii) != 0) && (c <= 0x7F))
return true;
if(((f & mask_vertical) != 0) && (::boost::re_detail::is_separator(c) || (c == static_cast<char_type>('\v')) || (m == U_GC_ZL_MASK) || (m == U_GC_ZP_MASK)))
return true;
if(((f & mask_horizontal) != 0) && !::boost::re_detail::is_separator(c) && u_isspace(c) && (c != static_cast<char_type>('\v')))
return true;
return false;
}

10
src/wc_regex_traits.cpp
View File

@ -161,7 +161,9 @@ enum
char_class_graph=char_class_alnum|char_class_punct,
char_class_blank=1<<9,
char_class_word=1<<10,
char_class_unicode=1<<11
char_class_unicode=1<<11,
char_class_horizontal=1<<12,
char_class_vertical=1<<13
};
c_regex_traits<wchar_t>::char_class_type BOOST_REGEX_CALL c_regex_traits<wchar_t>::lookup_classname(const wchar_t* p1, const wchar_t* p2)
@ -176,6 +178,7 @@ c_regex_traits<wchar_t>::char_class_type BOOST_REGEX_CALL c_regex_traits<wchar_t
char_class_digit,
char_class_digit,
char_class_graph,
char_class_horizontal,
char_class_lower,
char_class_lower,
char_class_print,
@ -185,6 +188,7 @@ c_regex_traits<wchar_t>::char_class_type BOOST_REGEX_CALL c_regex_traits<wchar_t
char_class_upper,
char_class_unicode,
char_class_upper,
char_class_vertical,
char_class_alnum | char_class_word,
char_class_alnum | char_class_word,
char_class_xdigit,
@ -216,7 +220,9 @@ bool BOOST_REGEX_CALL c_regex_traits<wchar_t>::isctype(wchar_t c, char_class_typ
|| ((mask & char_class_xdigit) && (std::iswxdigit)(c))
|| ((mask & char_class_blank) && (std::iswspace)(c) && !::boost::re_detail::is_separator(c))
|| ((mask & char_class_word) && (c == '_'))
|| ((mask & char_class_unicode) && (c & ~static_cast<wchar_t>(0xff)));
|| ((mask & char_class_unicode) && (c & ~static_cast<wchar_t>(0xff)))
|| ((mask & char_class_vertical) && (::boost::re_detail::is_separator(c) || (c == L'\v')))
|| ((mask & char_class_horizontal) && (std::iswspace)(c) && !::boost::re_detail::is_separator(c) && (c != L'\v'));
}
c_regex_traits<wchar_t>::string_type BOOST_REGEX_CALL c_regex_traits<wchar_t>::lookup_collatename(const wchar_t* p1, const wchar_t* p2)