Commit | Line | Data |
---|---|---|
0e23f538 TJ |
1 | /* |
2 | The software in this package is distributed under the GNU General | |
3 | Public License version 2 (with a special exception described below). | |
4 | ||
5 | A copy of GNU General Public License (GPL) is included in this distribution, | |
6 | in the file COPYING.GPL. | |
7 | ||
8 | As a special exception, if other files instantiate templates or use macros | |
9 | or inline functions from this file, or you compile this file and link it | |
10 | with other works to produce a work based on this file, this file | |
11 | does not by itself cause the resulting work to be covered | |
12 | by the GNU General Public License. | |
13 | ||
14 | However the source code for this file must still be made available | |
15 | in accordance with section (3) of the GNU General Public License. | |
16 | ||
17 | This exception does not invalidate any other reasons why a work based | |
18 | on this file might be covered by the GNU General Public License. | |
19 | */ | |
6a93d84a TJ |
20 | /** @file |
21 | * @brief collection of string tools (/ functions). | |
22 | * | |
23 | * contains a collection of miscellaneous functions for dealing with strings. | |
24 | * | |
25 | * some functions (like trim, lower case, upper case ...) are available in two versions: | |
26 | * - a modifying one (suffix "Mod") which modifies the given string argument in place. | |
27 | * - a non modifying one which take a constant string reference and returns a new string. | |
28 | * . | |
29 | * | |
30 | * | |
31 | * (c) Copyright 2007-2008 by Intra2net AG | |
6a93d84a | 32 | */ |
e93545dd GE |
33 | |
34 | #ifndef __STRINGFUNC_HXX | |
35 | #define __STRINGFUNC_HXX | |
36 | ||
5ad80381 | 37 | #include <stdio.h> |
6a93d84a | 38 | #include <list> |
5ad80381 | 39 | #include <numeric> |
0aaf13b5 | 40 | #include <set> |
376ec4fa | 41 | #include <vector> |
e93545dd | 42 | #include <string> |
6a93d84a TJ |
43 | #include <sstream> |
44 | #include <stdexcept> | |
b04b4f7e | 45 | #include <sys/types.h> |
6a93d84a | 46 | |
6ab3bc95 RP |
47 | namespace I2n |
48 | { | |
6a93d84a TJ |
49 | |
50 | /* | |
51 | ** some useful constants: | |
52 | */ | |
53 | ||
6ab3bc95 RP |
54 | extern const std::string Whitespaces; |
55 | extern const std::string LineEndings; | |
6a93d84a TJ |
56 | |
57 | ||
58 | /* | |
59 | ** predicates: | |
60 | */ | |
61 | ||
62 | ||
6ab3bc95 | 63 | bool has_prefix(const std::string& str, const std::string& prefix); |
6a93d84a | 64 | |
6ab3bc95 | 65 | bool has_suffix(const std::string& str, const std::string& suffix); |
6a93d84a TJ |
66 | |
67 | ||
68 | /* | |
6ab3bc95 | 69 | ** tool functions(modifying): |
6a93d84a TJ |
70 | */ |
71 | ||
6ab3bc95 | 72 | std::string trim_mod(std::string& str, const std::string& charlist = Whitespaces); |
6a93d84a | 73 | |
6ab3bc95 | 74 | std::string chomp_mod(std::string& str, const std::string& what= LineEndings ); |
6a93d84a | 75 | |
6ab3bc95 | 76 | std::string to_lower_mod(std::string& str); |
6a93d84a | 77 | |
6ab3bc95 | 78 | std::string to_upper_mod(std::string& str); |
6a93d84a TJ |
79 | |
80 | ||
81 | /* | |
82 | ** tool functions (not modifying): | |
83 | */ | |
84 | ||
6ab3bc95 | 85 | std::string trim(const std::string& str, const std::string& charlist = Whitespaces); |
6a93d84a | 86 | |
6ab3bc95 | 87 | std::string chomp(const std::string& str, const std::string& what= LineEndings ); |
6a93d84a | 88 | |
6ab3bc95 | 89 | std::string to_lower(const std::string& str); |
6a93d84a | 90 | |
6ab3bc95 | 91 | std::string to_upper(const std::string& str); |
6a93d84a TJ |
92 | |
93 | ||
6ab3bc95 | 94 | std::string remove_suffix(const std::string& str, const std::string& suffix); |
6a93d84a | 95 | |
6ab3bc95 | 96 | std::string remove_prefix(const std::string& str, const std::string& prefix); |
6a93d84a TJ |
97 | |
98 | ||
99 | ||
100 | /* | |
101 | ** split and join: | |
102 | */ | |
103 | ||
104 | ||
6ab3bc95 RP |
105 | bool pair_split( |
106 | const std::string& str, | |
107 | std::string& key, | |
108 | std::string& value, | |
109 | char delimiter = '='); | |
6a93d84a TJ |
110 | |
111 | ||
6ab3bc95 RP |
112 | void split_string( |
113 | const std::string& str, | |
114 | std::list< std::string >& result, | |
115 | const std::string& delimiter= "\n", | |
116 | bool omit_empty= false, | |
117 | const std::string& trim_list= std::string() | |
6a93d84a TJ |
118 | ); |
119 | ||
338da253 CH |
120 | void split_string( |
121 | const std::string& str, | |
122 | std::vector< std::string >& result, | |
123 | const std::string& delimiter= "\n", | |
124 | bool omit_empty= false, | |
125 | const std::string& trim_list= std::string() | |
126 | ); | |
127 | ||
6ab3bc95 RP |
128 | std::list< std::string > split_string( |
129 | const std::string& str, | |
130 | const std::string& delimiter = "\n", | |
131 | bool omit_empty= false, | |
132 | const std::string& trim_list= std::string() | |
6a93d84a TJ |
133 | ); |
134 | ||
5ad80381 PG |
135 | struct concatenator { |
136 | std::string delim; | |
6a93d84a | 137 | |
5ad80381 | 138 | concatenator (const std::string &delim) : delim (delim) { } |
6a93d84a | 139 | |
5ad80381 PG |
140 | inline std::string operator() (const std::string &acc, const std::string &elt) const |
141 | { return acc + delim + elt; } | |
142 | }; | |
376ec4fa | 143 | |
5ad80381 PG |
144 | template<typename Iter> |
145 | std::string | |
146 | join_string ( | |
147 | Iter first, | |
148 | Iter last, | |
149 | const std::string &delimiter = "\n" | |
150 | ) | |
151 | { | |
152 | if (first == last) { return ""; } | |
153 | ||
154 | const std::string &init = *first++; | |
155 | if (first == last) { return init; } | |
156 | ||
157 | return std::accumulate (first, last, init, concatenator (delimiter)); | |
158 | } | |
159 | ||
160 | /** | |
161 | * @brief joins a container of strings into a single string. | |
162 | * | |
163 | * This funtion is (basically) the reverse operation of @a split_string. | |
164 | * | |
165 | * @param parts the container of strings. | |
166 | * @param delimiter the delimiter to insert between the strings. | |
167 | * @return the joined string. | |
168 | */ | |
169 | template<typename Cont> | |
170 | inline std::string join_string( | |
171 | const Cont& parts, | |
0aaf13b5 | 172 | const std::string& delimiter = "\n" |
5ad80381 PG |
173 | ) |
174 | { return join_string (parts.begin (), parts.end (), delimiter); } | |
0aaf13b5 PG |
175 | |
176 | std::string join_string( | |
4f7a7b9f PG |
177 | const char *const parts [], |
178 | const std::string& delimiter = "\n" | |
179 | ); | |
180 | ||
6a93d84a | 181 | /* |
f3b61bd6 PG |
182 | * provide instantiation wrappers for ABI compatibility |
183 | */ | |
184 | std::string join_string (const std::list<std::string> &l, | |
185 | const std::string &d = "\n"); | |
186 | ||
187 | std::string join_string (const std::vector<std::string> &l, | |
188 | const std::string &d = "\n"); | |
189 | ||
190 | /* | |
6a93d84a TJ |
191 | ** conversions: |
192 | */ | |
193 | ||
194 | ||
6ab3bc95 | 195 | std::string convert_binary_to_hex(const std::string&str, bool upper_case_digits= false); |
6a93d84a | 196 | |
6ab3bc95 | 197 | std::string convert_hex_to_binary(const std::string& str) throw(std::runtime_error); |
6a93d84a TJ |
198 | |
199 | ||
6a93d84a TJ |
200 | /* |
201 | ** "type conversions": | |
202 | */ | |
203 | ||
204 | ||
205 | /** | |
206 | * convert a datatype @a T to a string via string stream. | |
207 | * | |
1fe2e899 CH |
208 | * This will not report trouble in conversion; for example: |
209 | * string_to<int>("christian") | |
210 | * will return 0 and not throw an error. | |
211 | * Use boost::lexical_cast<T>(string) to get error-checked results. | |
212 | * | |
6a93d84a TJ |
213 | * @param s the string which should be converted to @a T. |
214 | * @return the value of type T. | |
215 | */ | |
216 | template< | |
6ab3bc95 | 217 | class T |
6a93d84a | 218 | > |
6ab3bc95 | 219 | T string_to(const std::string& s) |
6a93d84a | 220 | { |
6ab3bc95 RP |
221 | std::istringstream istr(s); |
222 | T result; | |
223 | istr >> result; | |
224 | return result; | |
225 | } // eo string_to(const std::string&) | |
6a93d84a TJ |
226 | |
227 | ||
228 | /** | |
229 | * convert a datatype @a T to a string via string stream. | |
230 | * | |
231 | * @param s the string which should be converted to @a T. | |
232 | * @param result the resulting value of type @a T. | |
233 | * @return @a true iff the internal string stream was EOF after the conversion. | |
08f2d184 GE |
234 | * |
235 | * @attention: does not return if the conversion was successful. So check for empty strings before. | |
6a93d84a TJ |
236 | */ |
237 | template< | |
6ab3bc95 | 238 | class T |
6a93d84a | 239 | > |
6ab3bc95 | 240 | bool string_to(const std::string& s, T& result) |
6a93d84a | 241 | { |
6ab3bc95 RP |
242 | std::istringstream istr(s); |
243 | istr >> result; | |
244 | return istr.eof(); | |
245 | } // eo string_to(const std::string&) | |
6a93d84a TJ |
246 | |
247 | ||
248 | /** | |
3bcc713f GE |
249 | * convert string in hexadecimal notation to a datatype @a T |
250 | * supports strings with and without "0x" notation, e.g. 0xff and FF are both valid | |
251 | * | |
252 | * @param s the hex string which should be converted to @a T. | |
253 | * @return the value of type T. | |
254 | */ | |
255 | template< | |
256 | class T | |
257 | > | |
258 | T hex_string_to(const std::string& s) | |
259 | { | |
260 | std::istringstream istr(s); | |
261 | T result; | |
262 | istr >> std::hex >> result; | |
263 | return result; | |
264 | } // eo string_to(const std::string&) | |
265 | ||
266 | ||
267 | /** | |
268 | * convert string in hexadecimal notation to a datatype @a T | |
269 | * supports strings with and without "0x" notation, e.g. 0xff and FF are both valid | |
270 | * | |
271 | * @param s the hex string which should be converted to @a T. | |
272 | * @param result the resulting value of type @a T. | |
273 | * @return @a true iff the internal string stream was EOF after the conversion. | |
274 | * | |
275 | * @attention: does not return if the conversion was successful. So check for empty strings before. | |
276 | */ | |
277 | template< | |
278 | class T | |
279 | > | |
280 | bool hex_string_to(const std::string& s, T& result) | |
281 | { | |
282 | std::istringstream istr(s); | |
283 | istr >> std::hex >> result; | |
284 | return istr.eof(); | |
285 | } // eo string_to(const std::string&) | |
286 | ||
287 | ||
288 | /** | |
6a93d84a TJ |
289 | * convert a string to another datatype @a T via string stream. |
290 | * | |
291 | * @param v the value (of type @a T) which should be converted to a string. | |
292 | * @return the resulting string. | |
293 | */ | |
294 | template< | |
6ab3bc95 | 295 | class T |
6a93d84a | 296 | > |
6ab3bc95 | 297 | std::string to_string(const T& v) |
6a93d84a | 298 | { |
6ab3bc95 RP |
299 | std::ostringstream ostr; |
300 | ostr << v; | |
301 | return ostr.str(); | |
302 | } // eo to_string(const T&) | |
303 | ||
304 | ||
1a0267e5 CH |
305 | /** |
306 | * Create a string with types shortened in texts describing C++ types | |
307 | * | |
308 | * for example: std::list<some_long_type, std::allocator<some_long_type> > | |
309 | * --> std::list<some_long_type, _alloc_> | |
310 | * | |
311 | * and std::basic_string<char, std::char_traits<char>, std::allocator<char> > | |
312 | * --> std::string | |
313 | */ | |
314 | std::string shorten_stl_types(const std::string &input); | |
315 | ||
1ebab1e3 TJ |
316 | std::string base64_encode(const std::string &input, bool one_line=true); |
317 | std::string base64_decode(const std::string &input, bool one_line=true); | |
1a0267e5 | 318 | |
6ab3bc95 RP |
319 | } // eo namespace I2n |
320 | ||
6a93d84a | 321 | |
e5b21dbb | 322 | #if 0 |
6ab3bc95 RP |
323 | std::string to_lower(const std::string &src); |
324 | std::string to_upper(const std::string &src); | |
e5b21dbb RP |
325 | #else |
326 | // compatibility: import lower/upper funcs from I2n: | |
327 | using I2n::to_lower; | |
328 | using I2n::to_upper; | |
329 | #endif | |
e93545dd | 330 | |
81267544 | 331 | |
8f296924 GMF |
332 | enum UnitBase { |
333 | UnitBase1000, // SI decimal, composed by multiples of 1000 (KB, MB, etc.) | |
334 | UnitBase1024 // IEC binary, composed by multiples of 1024 (KiB, MiB, etc. ) | |
81267544 GMF |
335 | }; |
336 | ||
8f296924 GMF |
337 | enum UnitFormat { |
338 | ShortUnitFormat, // B, KB, MB, ... | |
339 | LongUnitFormat // Byte, KByte, MByte, ... | |
d1ea9075 GMF |
340 | }; |
341 | ||
81267544 GMF |
342 | std::string nice_unit_format( |
343 | const int64_t input, | |
ee037ee3 GMF |
344 | const UnitFormat format = ShortUnitFormat, |
345 | const UnitBase base = UnitBase1024 | |
81267544 | 346 | ); |
e93545dd | 347 | |
5cd64148 CH |
348 | std::string nice_unit_format( |
349 | const double input, | |
350 | const UnitFormat format = ShortUnitFormat, | |
351 | const UnitBase base = UnitBase1024 | |
352 | ); | |
353 | ||
e93545dd GE |
354 | bool replace_all(std::string &base, const std::string *ist, const std::string *soll); |
355 | bool replace_all(std::string &base, const char *ist, const char *soll); | |
356 | bool replace_all(std::string &base, const char *ist, const std::string *soll); | |
357 | bool replace_all(std::string &base, const std::string &ist, const char *soll); | |
358 | bool replace_all(std::string &base, const std::string &ist, const std::string &soll); | |
359 | ||
360 | std::string iso_to_utf8(const std::string& isostring); | |
361 | std::string utf8_to_iso(const std::string& utf8string); | |
13cc4db1 | 362 | std::string utf7imap_to_utf8(const std::string &utf7imapstring); |
6a2b6dd1 | 363 | std::string utf8_to_utf7imap(const std::string &utf8string); |
118e216e TJ |
364 | |
365 | std::string strip_html_tags(const std::string &input); | |
366 | std::string smart_html_entities(const std::string &input); | |
367 | std::string html_entities(std::string str); | |
554f813d | 368 | std::string html_entities_to_console(std::string str); |
118e216e | 369 | |
3f5c5ccd | 370 | typedef std::pair<std::string::size_type, std::string::size_type> CommentZone; |
46dd1321 | 371 | std::vector<CommentZone> find_html_comments(const std::string &str); |
3f5c5ccd CH |
372 | void remove_html_comments(std::string &str); |
373 | void remove_html_comments(std::string &str, const std::vector<CommentZone> &comments); | |
374 | ||
b953bf36 GE |
375 | std::string sanitize_for_logging(const std::string &str, const char replace_with='?'); |
376 | ||
47c07fba GE |
377 | std::string escape(const std::string &s); |
378 | ||
379 | std::string descape(const std::string &s, int startpos, int &endpos); | |
e6da286a | 380 | inline std::string descape(const std::string &s) |
47c07fba | 381 | { |
6ab3bc95 RP |
382 | int endpos; |
383 | return descape(s,0,endpos); | |
47c07fba GE |
384 | } |
385 | ||
386 | std::string escape_shellarg(const std::string &input); | |
387 | ||
e93545dd | 388 | #endif |