3 * Copyright (c) 1998-2002
6 * Use, modification and distribution are subject to the
7 * Boost Software License, Version 1.0. (See accompanying file
8 * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
13 * LOCATION: see http://www.boost.org for most recent version.
15 * VERSION see <boost/version.hpp>
16 * DESCRIPTION: Declares internal state machine structures.
19 #ifndef BOOST_REGEX_V4_STATES_HPP
20 #define BOOST_REGEX_V4_STATES_HPP
24 #pragma warning(disable: 4103)
26 #ifdef BOOST_HAS_ABI_HEADERS
27 # include BOOST_ABI_PREFIX
36 /*** mask_type *******************************************************
37 Whenever we have a choice of two alternatives, we use an array of bytes
38 to indicate which of the two alternatives it is possible to take for any
39 given input character. If mask_take is set, then we can take the next
40 state, and if mask_skip is set then we can take the alternative.
41 ***********************************************************************/
47 mask_any = mask_skip | mask_take,
51 /*** helpers **********************************************************
52 These helpers let us use function overload resolution to detect whether
53 we have narrow or wide character strings:
54 ***********************************************************************/
55 struct _narrow_type{};
57 template <class charT> struct is_byte;
58 template<> struct is_byte<char> { typedef _narrow_type width_type; };
59 template<> struct is_byte<unsigned char>{ typedef _narrow_type width_type; };
60 template<> struct is_byte<signed char> { typedef _narrow_type width_type; };
61 template <class charT> struct is_byte { typedef _wide_type width_type; };
63 /*** enum syntax_element_type ******************************************
64 Every record in the state machine falls into one of the following types:
65 ***********************************************************************/
66 enum syntax_element_type
68 // start of a marked sub-expression, or perl-style (?...) extension
69 syntax_element_startmark = 0,
70 // end of a marked sub-expression, or perl-style (?...) extension
71 syntax_element_endmark = syntax_element_startmark + 1,
72 // any sequence of literal characters
73 syntax_element_literal = syntax_element_endmark + 1,
74 // start of line assertion: ^
75 syntax_element_start_line = syntax_element_literal + 1,
76 // end of line assertion $
77 syntax_element_end_line = syntax_element_start_line + 1,
78 // match any character: .
79 syntax_element_wild = syntax_element_end_line + 1,
80 // end of expression: we have a match when we get here
81 syntax_element_match = syntax_element_wild + 1,
82 // perl style word boundary: \b
83 syntax_element_word_boundary = syntax_element_match + 1,
84 // perl style within word boundary: \B
85 syntax_element_within_word = syntax_element_word_boundary + 1,
86 // start of word assertion: \<
87 syntax_element_word_start = syntax_element_within_word + 1,
88 // end of word assertion: \>
89 syntax_element_word_end = syntax_element_word_start + 1,
90 // start of buffer assertion: \`
91 syntax_element_buffer_start = syntax_element_word_end + 1,
92 // end of buffer assertion: \'
93 syntax_element_buffer_end = syntax_element_buffer_start + 1,
94 // backreference to previously matched sub-expression
95 syntax_element_backref = syntax_element_buffer_end + 1,
96 // either a wide character set [..] or one with multicharacter collating elements:
97 syntax_element_long_set = syntax_element_backref + 1,
98 // narrow character set: [...]
99 syntax_element_set = syntax_element_long_set + 1,
100 // jump to a new state in the machine:
101 syntax_element_jump = syntax_element_set + 1,
102 // choose between two production states:
103 syntax_element_alt = syntax_element_jump + 1,
105 syntax_element_rep = syntax_element_alt + 1,
106 // match a combining character sequence
107 syntax_element_combining = syntax_element_rep + 1,
108 // perl style soft buffer end: \z
109 syntax_element_soft_buffer_end = syntax_element_combining + 1,
110 // perl style continuation: \G
111 syntax_element_restart_continue = syntax_element_soft_buffer_end + 1,
112 // single character repeats:
113 syntax_element_dot_rep = syntax_element_restart_continue + 1,
114 syntax_element_char_rep = syntax_element_dot_rep + 1,
115 syntax_element_short_set_rep = syntax_element_char_rep + 1,
116 syntax_element_long_set_rep = syntax_element_short_set_rep + 1,
117 // a backstep for lookbehind repeats:
118 syntax_element_backstep = syntax_element_long_set_rep + 1,
119 // an assertion that a mark was matched:
120 syntax_element_assert_backref = syntax_element_backstep + 1,
121 syntax_element_toggle_case = syntax_element_assert_backref + 1
124 #ifdef BOOST_REGEX_DEBUG
125 // dwa 09/26/00 - This is needed to suppress warnings about an ambiguous conversion
126 std::ostream& operator<<(std::ostream&, syntax_element_type);
129 struct re_syntax_base;
131 /*** union offset_type ************************************************
132 Points to another state in the machine. During machine construction
133 we use integral offsets, but these are converted to pointers before
134 execution of the machine.
135 ***********************************************************************/
142 /*** struct re_syntax_base ********************************************
143 Base class for all states in the machine.
144 ***********************************************************************/
145 struct re_syntax_base
147 syntax_element_type type; // what kind of state this is
148 offset_type next; // next state in the machine
151 /*** struct re_brace **************************************************
152 A marked parenthesis.
153 ***********************************************************************/
154 struct re_brace : public re_syntax_base
156 // The index to match, can be zero (don't mark the sub-expression)
157 // or negative (for perl style (?...) extentions):
161 /*** struct re_dot **************************************************
163 ***********************************************************************/
167 force_not_newline = 0,
170 test_not_newline = 2,
173 struct re_dot : public re_syntax_base
178 /*** struct re_literal ************************************************
179 A string of literals, following this structure will be an
180 array of characters: charT[length]
181 ***********************************************************************/
182 struct re_literal : public re_syntax_base
187 /*** struct re_case ************************************************
188 Indicates whether we are moving to a case insensive block or not
189 ***********************************************************************/
190 struct re_case : public re_syntax_base
195 /*** struct re_set_long ***********************************************
196 A wide character set of characters, following this structure will be
197 an array of type charT:
198 First csingles null-terminated strings
199 Then 2 * cranges NULL terminated strings
200 Then cequivalents NULL terminated strings
201 ***********************************************************************/
202 template <class mask_type>
203 struct re_set_long : public re_syntax_base
205 unsigned int csingles, cranges, cequivalents;
212 /*** struct re_set ****************************************************
213 A set of narrow-characters, matches any of _map which is none-zero
214 ***********************************************************************/
215 struct re_set : public re_syntax_base
217 unsigned char _map[1 << CHAR_BIT];
220 /*** struct re_jump ***************************************************
221 Jump to a new location in the machine (not next).
222 ***********************************************************************/
223 struct re_jump : public re_syntax_base
225 offset_type alt; // location to jump to
228 /*** struct re_alt ***************************************************
229 Jump to a new location in the machine (possibly next).
230 ***********************************************************************/
231 struct re_alt : public re_jump
233 unsigned char _map[1 << CHAR_BIT]; // which characters can take the jump
234 unsigned int can_be_null; // true if we match a NULL string
237 /*** struct re_repeat *************************************************
238 Repeat a section of the machine
239 ***********************************************************************/
240 struct re_repeat : public re_alt
242 std::size_t min, max; // min and max allowable repeats
243 int state_id; // Unique identifier for this repeat
244 bool leading; // True if this repeat is at the start of the machine (lets us optimize some searches)
245 bool greedy; // True if this is a greedy repeat
248 /*** enum re_jump_size_type *******************************************
249 Provides compiled size of re_jump structure (allowing for trailing alignment).
250 We provide this so we know how manybytes to insert when constructing the machine
251 (The value of padding_mask is defined in regex_raw_buffer.hpp).
252 ***********************************************************************/
253 enum re_jump_size_type
255 re_jump_size = (sizeof(re_jump) + padding_mask) & ~(padding_mask),
256 re_repeater_size = (sizeof(re_repeat) + padding_mask) & ~(padding_mask),
257 re_alt_size = (sizeof(re_alt) + padding_mask) & ~(padding_mask)
260 /*** proc re_is_set_member *********************************************
261 Forward declaration: we'll need this one later...
262 ***********************************************************************/
264 template<class charT, class traits>
267 template <class iterator, class charT, class traits_type, class char_classT>
268 iterator BOOST_REGEX_CALL re_is_set_member(iterator next,
270 const re_set_long<char_classT>* set_,
271 const regex_data<charT, traits_type>& e, bool icase);
273 } // namespace re_detail
278 #pragma warning(push)
279 #pragma warning(disable: 4103)
281 #ifdef BOOST_HAS_ABI_HEADERS
282 # include BOOST_ABI_SUFFIX