Chris@16
|
1 /*
|
Chris@16
|
2 *
|
Chris@16
|
3 * Copyright (c) 1998-2002
|
Chris@16
|
4 * John Maddock
|
Chris@16
|
5 *
|
Chris@16
|
6 * Use, modification and distribution are subject to the
|
Chris@16
|
7 * Boost Software License, Version 1.0. (See accompanying file
|
Chris@16
|
8 * LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_0.txt)
|
Chris@16
|
9 *
|
Chris@16
|
10 */
|
Chris@16
|
11
|
Chris@16
|
12 /*
|
Chris@16
|
13 * LOCATION: see http://www.boost.org for most recent version.
|
Chris@16
|
14 * FILE cregex.cpp
|
Chris@16
|
15 * VERSION see <boost/version.hpp>
|
Chris@16
|
16 * DESCRIPTION: Declares POSIX API functions
|
Chris@16
|
17 * + boost::RegEx high level wrapper.
|
Chris@16
|
18 */
|
Chris@16
|
19
|
Chris@16
|
20 #ifndef BOOST_RE_CREGEX_HPP_INCLUDED
|
Chris@16
|
21 #define BOOST_RE_CREGEX_HPP_INCLUDED
|
Chris@16
|
22
|
Chris@16
|
23 #ifndef BOOST_REGEX_CONFIG_HPP
|
Chris@16
|
24 #include <boost/regex/config.hpp>
|
Chris@16
|
25 #endif
|
Chris@16
|
26 #include <boost/regex/v4/match_flags.hpp>
|
Chris@16
|
27 #include <boost/regex/v4/error_type.hpp>
|
Chris@16
|
28
|
Chris@16
|
29 #ifdef __cplusplus
|
Chris@16
|
30 #include <cstddef>
|
Chris@16
|
31 #else
|
Chris@16
|
32 #include <stddef.h>
|
Chris@16
|
33 #endif
|
Chris@16
|
34
|
Chris@16
|
35 #ifdef BOOST_MSVC
|
Chris@16
|
36 #pragma warning(push)
|
Chris@16
|
37 #pragma warning(disable: 4103)
|
Chris@16
|
38 #endif
|
Chris@16
|
39 #ifdef BOOST_HAS_ABI_HEADERS
|
Chris@16
|
40 # include BOOST_ABI_PREFIX
|
Chris@16
|
41 #endif
|
Chris@16
|
42 #ifdef BOOST_MSVC
|
Chris@16
|
43 #pragma warning(pop)
|
Chris@16
|
44 #endif
|
Chris@16
|
45
|
Chris@16
|
46 /* include these defs only for POSIX compatablity */
|
Chris@16
|
47 #ifdef __cplusplus
|
Chris@16
|
48 namespace boost{
|
Chris@16
|
49 extern "C" {
|
Chris@16
|
50 #endif
|
Chris@16
|
51
|
Chris@16
|
52 #if defined(__cplusplus) && !defined(BOOST_NO_STDC_NAMESPACE)
|
Chris@16
|
53 typedef std::ptrdiff_t regoff_t;
|
Chris@16
|
54 typedef std::size_t regsize_t;
|
Chris@16
|
55 #else
|
Chris@16
|
56 typedef ptrdiff_t regoff_t;
|
Chris@16
|
57 typedef size_t regsize_t;
|
Chris@16
|
58 #endif
|
Chris@16
|
59
|
Chris@16
|
60 typedef struct
|
Chris@16
|
61 {
|
Chris@16
|
62 unsigned int re_magic;
|
Chris@16
|
63 #ifdef __cplusplus
|
Chris@16
|
64 std::size_t re_nsub; /* number of parenthesized subexpressions */
|
Chris@16
|
65 #else
|
Chris@16
|
66 size_t re_nsub;
|
Chris@16
|
67 #endif
|
Chris@16
|
68 const char* re_endp; /* end pointer for REG_PEND */
|
Chris@16
|
69 void* guts; /* none of your business :-) */
|
Chris@16
|
70 match_flag_type eflags; /* none of your business :-) */
|
Chris@16
|
71 } regex_tA;
|
Chris@16
|
72
|
Chris@16
|
73 #ifndef BOOST_NO_WREGEX
|
Chris@16
|
74 typedef struct
|
Chris@16
|
75 {
|
Chris@16
|
76 unsigned int re_magic;
|
Chris@16
|
77 #ifdef __cplusplus
|
Chris@16
|
78 std::size_t re_nsub; /* number of parenthesized subexpressions */
|
Chris@16
|
79 #else
|
Chris@16
|
80 size_t re_nsub;
|
Chris@16
|
81 #endif
|
Chris@16
|
82 const wchar_t* re_endp; /* end pointer for REG_PEND */
|
Chris@16
|
83 void* guts; /* none of your business :-) */
|
Chris@16
|
84 match_flag_type eflags; /* none of your business :-) */
|
Chris@16
|
85 } regex_tW;
|
Chris@16
|
86 #endif
|
Chris@16
|
87
|
Chris@16
|
88 typedef struct
|
Chris@16
|
89 {
|
Chris@16
|
90 regoff_t rm_so; /* start of match */
|
Chris@16
|
91 regoff_t rm_eo; /* end of match */
|
Chris@16
|
92 } regmatch_t;
|
Chris@16
|
93
|
Chris@16
|
94 /* regcomp() flags */
|
Chris@16
|
95 typedef enum{
|
Chris@16
|
96 REG_BASIC = 0000,
|
Chris@16
|
97 REG_EXTENDED = 0001,
|
Chris@16
|
98 REG_ICASE = 0002,
|
Chris@16
|
99 REG_NOSUB = 0004,
|
Chris@16
|
100 REG_NEWLINE = 0010,
|
Chris@16
|
101 REG_NOSPEC = 0020,
|
Chris@16
|
102 REG_PEND = 0040,
|
Chris@16
|
103 REG_DUMP = 0200,
|
Chris@16
|
104 REG_NOCOLLATE = 0400,
|
Chris@16
|
105 REG_ESCAPE_IN_LISTS = 01000,
|
Chris@16
|
106 REG_NEWLINE_ALT = 02000,
|
Chris@16
|
107 REG_PERLEX = 04000,
|
Chris@16
|
108
|
Chris@16
|
109 REG_PERL = REG_EXTENDED | REG_NOCOLLATE | REG_ESCAPE_IN_LISTS | REG_PERLEX,
|
Chris@16
|
110 REG_AWK = REG_EXTENDED | REG_ESCAPE_IN_LISTS,
|
Chris@16
|
111 REG_GREP = REG_BASIC | REG_NEWLINE_ALT,
|
Chris@16
|
112 REG_EGREP = REG_EXTENDED | REG_NEWLINE_ALT,
|
Chris@16
|
113
|
Chris@16
|
114 REG_ASSERT = 15,
|
Chris@16
|
115 REG_INVARG = 16,
|
Chris@16
|
116 REG_ATOI = 255, /* convert name to number (!) */
|
Chris@16
|
117 REG_ITOA = 0400 /* convert number to name (!) */
|
Chris@16
|
118 } reg_comp_flags;
|
Chris@16
|
119
|
Chris@16
|
120 /* regexec() flags */
|
Chris@16
|
121 typedef enum{
|
Chris@16
|
122 REG_NOTBOL = 00001,
|
Chris@16
|
123 REG_NOTEOL = 00002,
|
Chris@16
|
124 REG_STARTEND = 00004
|
Chris@16
|
125 } reg_exec_flags;
|
Chris@16
|
126
|
Chris@16
|
127 /*
|
Chris@16
|
128 * POSIX error codes:
|
Chris@16
|
129 */
|
Chris@16
|
130 typedef unsigned reg_error_t;
|
Chris@16
|
131 typedef reg_error_t reg_errcode_t; /* backwards compatibility */
|
Chris@16
|
132
|
Chris@16
|
133 static const reg_error_t REG_NOERROR = 0; /* Success. */
|
Chris@16
|
134 static const reg_error_t REG_NOMATCH = 1; /* Didn't find a match (for regexec). */
|
Chris@16
|
135
|
Chris@16
|
136 /* POSIX regcomp return error codes. (In the order listed in the
|
Chris@16
|
137 standard.) */
|
Chris@16
|
138 static const reg_error_t REG_BADPAT = 2; /* Invalid pattern. */
|
Chris@16
|
139 static const reg_error_t REG_ECOLLATE = 3; /* Undefined collating element. */
|
Chris@16
|
140 static const reg_error_t REG_ECTYPE = 4; /* Invalid character class name. */
|
Chris@16
|
141 static const reg_error_t REG_EESCAPE = 5; /* Trailing backslash. */
|
Chris@16
|
142 static const reg_error_t REG_ESUBREG = 6; /* Invalid back reference. */
|
Chris@16
|
143 static const reg_error_t REG_EBRACK = 7; /* Unmatched left bracket. */
|
Chris@16
|
144 static const reg_error_t REG_EPAREN = 8; /* Parenthesis imbalance. */
|
Chris@16
|
145 static const reg_error_t REG_EBRACE = 9; /* Unmatched \{. */
|
Chris@16
|
146 static const reg_error_t REG_BADBR = 10; /* Invalid contents of \{\}. */
|
Chris@16
|
147 static const reg_error_t REG_ERANGE = 11; /* Invalid range end. */
|
Chris@16
|
148 static const reg_error_t REG_ESPACE = 12; /* Ran out of memory. */
|
Chris@16
|
149 static const reg_error_t REG_BADRPT = 13; /* No preceding re for repetition op. */
|
Chris@16
|
150 static const reg_error_t REG_EEND = 14; /* unexpected end of expression */
|
Chris@16
|
151 static const reg_error_t REG_ESIZE = 15; /* expression too big */
|
Chris@16
|
152 static const reg_error_t REG_ERPAREN = 8; /* = REG_EPAREN : unmatched right parenthesis */
|
Chris@16
|
153 static const reg_error_t REG_EMPTY = 17; /* empty expression */
|
Chris@16
|
154 static const reg_error_t REG_E_MEMORY = 15; /* = REG_ESIZE : out of memory */
|
Chris@16
|
155 static const reg_error_t REG_ECOMPLEXITY = 18; /* complexity too high */
|
Chris@16
|
156 static const reg_error_t REG_ESTACK = 19; /* out of stack space */
|
Chris@16
|
157 static const reg_error_t REG_E_PERL = 20; /* Perl (?...) error */
|
Chris@16
|
158 static const reg_error_t REG_E_UNKNOWN = 21; /* unknown error */
|
Chris@16
|
159 static const reg_error_t REG_ENOSYS = 21; /* = REG_E_UNKNOWN : Reserved. */
|
Chris@16
|
160
|
Chris@16
|
161 BOOST_REGEX_DECL int BOOST_REGEX_CCALL regcompA(regex_tA*, const char*, int);
|
Chris@16
|
162 BOOST_REGEX_DECL regsize_t BOOST_REGEX_CCALL regerrorA(int, const regex_tA*, char*, regsize_t);
|
Chris@16
|
163 BOOST_REGEX_DECL int BOOST_REGEX_CCALL regexecA(const regex_tA*, const char*, regsize_t, regmatch_t*, int);
|
Chris@16
|
164 BOOST_REGEX_DECL void BOOST_REGEX_CCALL regfreeA(regex_tA*);
|
Chris@16
|
165
|
Chris@16
|
166 #ifndef BOOST_NO_WREGEX
|
Chris@16
|
167 BOOST_REGEX_DECL int BOOST_REGEX_CCALL regcompW(regex_tW*, const wchar_t*, int);
|
Chris@16
|
168 BOOST_REGEX_DECL regsize_t BOOST_REGEX_CCALL regerrorW(int, const regex_tW*, wchar_t*, regsize_t);
|
Chris@16
|
169 BOOST_REGEX_DECL int BOOST_REGEX_CCALL regexecW(const regex_tW*, const wchar_t*, regsize_t, regmatch_t*, int);
|
Chris@16
|
170 BOOST_REGEX_DECL void BOOST_REGEX_CCALL regfreeW(regex_tW*);
|
Chris@16
|
171 #endif
|
Chris@16
|
172
|
Chris@16
|
173 #ifdef UNICODE
|
Chris@16
|
174 #define regcomp regcompW
|
Chris@16
|
175 #define regerror regerrorW
|
Chris@16
|
176 #define regexec regexecW
|
Chris@16
|
177 #define regfree regfreeW
|
Chris@16
|
178 #define regex_t regex_tW
|
Chris@16
|
179 #else
|
Chris@16
|
180 #define regcomp regcompA
|
Chris@16
|
181 #define regerror regerrorA
|
Chris@16
|
182 #define regexec regexecA
|
Chris@16
|
183 #define regfree regfreeA
|
Chris@16
|
184 #define regex_t regex_tA
|
Chris@16
|
185 #endif
|
Chris@16
|
186
|
Chris@16
|
187 #ifdef BOOST_MSVC
|
Chris@16
|
188 #pragma warning(push)
|
Chris@16
|
189 #pragma warning(disable: 4103)
|
Chris@16
|
190 #endif
|
Chris@16
|
191 #ifdef BOOST_HAS_ABI_HEADERS
|
Chris@16
|
192 # include BOOST_ABI_SUFFIX
|
Chris@16
|
193 #endif
|
Chris@16
|
194 #ifdef BOOST_MSVC
|
Chris@16
|
195 #pragma warning(pop)
|
Chris@16
|
196 #endif
|
Chris@16
|
197
|
Chris@16
|
198 #ifdef __cplusplus
|
Chris@16
|
199 } /* extern "C" */
|
Chris@16
|
200 } /* namespace */
|
Chris@16
|
201 #endif
|
Chris@16
|
202
|
Chris@16
|
203 #if defined(__cplusplus)
|
Chris@16
|
204 /*
|
Chris@16
|
205 * C++ high level wrapper goes here:
|
Chris@16
|
206 */
|
Chris@16
|
207 #include <string>
|
Chris@16
|
208 #include <vector>
|
Chris@16
|
209 namespace boost{
|
Chris@16
|
210
|
Chris@16
|
211 #ifdef BOOST_MSVC
|
Chris@16
|
212 #pragma warning(push)
|
Chris@16
|
213 #pragma warning(disable: 4103)
|
Chris@16
|
214 #endif
|
Chris@16
|
215 #ifdef BOOST_HAS_ABI_HEADERS
|
Chris@16
|
216 # include BOOST_ABI_PREFIX
|
Chris@16
|
217 #endif
|
Chris@16
|
218 #ifdef BOOST_MSVC
|
Chris@16
|
219 #pragma warning(pop)
|
Chris@16
|
220 #endif
|
Chris@16
|
221
|
Chris@16
|
222 class RegEx;
|
Chris@16
|
223
|
Chris@16
|
224 namespace re_detail{
|
Chris@16
|
225
|
Chris@16
|
226 class RegExData;
|
Chris@16
|
227 struct pred1;
|
Chris@16
|
228 struct pred2;
|
Chris@16
|
229 struct pred3;
|
Chris@16
|
230 struct pred4;
|
Chris@16
|
231
|
Chris@16
|
232 } /* namespace re_detail */
|
Chris@16
|
233
|
Chris@16
|
234 #if (defined(BOOST_MSVC) || defined(__BORLANDC__)) && !defined(BOOST_DISABLE_WIN32)
|
Chris@16
|
235 typedef bool (__cdecl *GrepCallback)(const RegEx& expression);
|
Chris@16
|
236 typedef bool (__cdecl *GrepFileCallback)(const char* file, const RegEx& expression);
|
Chris@16
|
237 typedef bool (__cdecl *FindFilesCallback)(const char* file);
|
Chris@16
|
238 #else
|
Chris@16
|
239 typedef bool (*GrepCallback)(const RegEx& expression);
|
Chris@16
|
240 typedef bool (*GrepFileCallback)(const char* file, const RegEx& expression);
|
Chris@16
|
241 typedef bool (*FindFilesCallback)(const char* file);
|
Chris@16
|
242 #endif
|
Chris@16
|
243
|
Chris@16
|
244 class BOOST_REGEX_DECL RegEx
|
Chris@16
|
245 {
|
Chris@16
|
246 private:
|
Chris@16
|
247 re_detail::RegExData* pdata;
|
Chris@16
|
248 public:
|
Chris@16
|
249 RegEx();
|
Chris@16
|
250 RegEx(const RegEx& o);
|
Chris@16
|
251 ~RegEx();
|
Chris@16
|
252 explicit RegEx(const char* c, bool icase = false);
|
Chris@16
|
253 explicit RegEx(const std::string& s, bool icase = false);
|
Chris@16
|
254 RegEx& operator=(const RegEx& o);
|
Chris@16
|
255 RegEx& operator=(const char* p);
|
Chris@16
|
256 RegEx& operator=(const std::string& s){ return this->operator=(s.c_str()); }
|
Chris@16
|
257 unsigned int SetExpression(const char* p, bool icase = false);
|
Chris@16
|
258 unsigned int SetExpression(const std::string& s, bool icase = false){ return SetExpression(s.c_str(), icase); }
|
Chris@16
|
259 std::string Expression()const;
|
Chris@16
|
260 unsigned int error_code()const;
|
Chris@16
|
261 /*
|
Chris@16
|
262 * now matching operators:
|
Chris@16
|
263 */
|
Chris@16
|
264 bool Match(const char* p, match_flag_type flags = match_default);
|
Chris@16
|
265 bool Match(const std::string& s, match_flag_type flags = match_default) { return Match(s.c_str(), flags); }
|
Chris@16
|
266 bool Search(const char* p, match_flag_type flags = match_default);
|
Chris@16
|
267 bool Search(const std::string& s, match_flag_type flags = match_default) { return Search(s.c_str(), flags); }
|
Chris@16
|
268 unsigned int Grep(GrepCallback cb, const char* p, match_flag_type flags = match_default);
|
Chris@16
|
269 unsigned int Grep(GrepCallback cb, const std::string& s, match_flag_type flags = match_default) { return Grep(cb, s.c_str(), flags); }
|
Chris@16
|
270 unsigned int Grep(std::vector<std::string>& v, const char* p, match_flag_type flags = match_default);
|
Chris@16
|
271 unsigned int Grep(std::vector<std::string>& v, const std::string& s, match_flag_type flags = match_default) { return Grep(v, s.c_str(), flags); }
|
Chris@16
|
272 unsigned int Grep(std::vector<std::size_t>& v, const char* p, match_flag_type flags = match_default);
|
Chris@16
|
273 unsigned int Grep(std::vector<std::size_t>& v, const std::string& s, match_flag_type flags = match_default) { return Grep(v, s.c_str(), flags); }
|
Chris@16
|
274 #ifndef BOOST_REGEX_NO_FILEITER
|
Chris@16
|
275 unsigned int GrepFiles(GrepFileCallback cb, const char* files, bool recurse = false, match_flag_type flags = match_default);
|
Chris@16
|
276 unsigned int GrepFiles(GrepFileCallback cb, const std::string& files, bool recurse = false, match_flag_type flags = match_default) { return GrepFiles(cb, files.c_str(), recurse, flags); }
|
Chris@16
|
277 unsigned int FindFiles(FindFilesCallback cb, const char* files, bool recurse = false, match_flag_type flags = match_default);
|
Chris@16
|
278 unsigned int FindFiles(FindFilesCallback cb, const std::string& files, bool recurse = false, match_flag_type flags = match_default) { return FindFiles(cb, files.c_str(), recurse, flags); }
|
Chris@16
|
279 #endif
|
Chris@16
|
280
|
Chris@16
|
281 std::string Merge(const std::string& in, const std::string& fmt,
|
Chris@16
|
282 bool copy = true, match_flag_type flags = match_default);
|
Chris@16
|
283 std::string Merge(const char* in, const char* fmt,
|
Chris@16
|
284 bool copy = true, match_flag_type flags = match_default);
|
Chris@16
|
285
|
Chris@16
|
286 std::size_t Split(std::vector<std::string>& v, std::string& s, match_flag_type flags = match_default, unsigned max_count = ~0);
|
Chris@16
|
287 /*
|
Chris@16
|
288 * now operators for returning what matched in more detail:
|
Chris@16
|
289 */
|
Chris@16
|
290 std::size_t Position(int i = 0)const;
|
Chris@16
|
291 std::size_t Length(int i = 0)const;
|
Chris@16
|
292 bool Matched(int i = 0)const;
|
Chris@16
|
293 std::size_t Marks()const;
|
Chris@16
|
294 std::string What(int i = 0)const;
|
Chris@16
|
295 std::string operator[](int i)const { return What(i); }
|
Chris@16
|
296
|
Chris@16
|
297 static const std::size_t npos;
|
Chris@16
|
298
|
Chris@16
|
299 friend struct re_detail::pred1;
|
Chris@16
|
300 friend struct re_detail::pred2;
|
Chris@16
|
301 friend struct re_detail::pred3;
|
Chris@16
|
302 friend struct re_detail::pred4;
|
Chris@16
|
303 };
|
Chris@16
|
304
|
Chris@16
|
305 #ifdef BOOST_MSVC
|
Chris@16
|
306 #pragma warning(push)
|
Chris@16
|
307 #pragma warning(disable: 4103)
|
Chris@16
|
308 #endif
|
Chris@16
|
309 #ifdef BOOST_HAS_ABI_HEADERS
|
Chris@16
|
310 # include BOOST_ABI_SUFFIX
|
Chris@16
|
311 #endif
|
Chris@16
|
312 #ifdef BOOST_MSVC
|
Chris@16
|
313 #pragma warning(pop)
|
Chris@16
|
314 #endif
|
Chris@16
|
315
|
Chris@16
|
316 } /* namespace boost */
|
Chris@16
|
317
|
Chris@16
|
318 #endif /* __cplusplus */
|
Chris@16
|
319
|
Chris@16
|
320 #endif /* include guard */
|
Chris@16
|
321
|
Chris@16
|
322
|
Chris@16
|
323
|
Chris@16
|
324
|
Chris@16
|
325
|
Chris@16
|
326
|
Chris@16
|
327
|
Chris@16
|
328
|
Chris@16
|
329
|
Chris@16
|
330
|