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 | ||
6a93d84a | 37 | #include <list> |
0aaf13b5 | 38 | #include <set> |
376ec4fa | 39 | #include <vector> |
e93545dd | 40 | #include <string> |
6a93d84a TJ |
41 | #include <sstream> |
42 | #include <stdexcept> | |
b04b4f7e | 43 | #include <sys/types.h> |
6a93d84a | 44 | |
6ab3bc95 RP |
45 | namespace I2n |
46 | { | |
6a93d84a TJ |
47 | |
48 | /* | |
49 | ** some useful constants: | |
50 | */ | |
51 | ||
6ab3bc95 RP |
52 | extern const std::string Whitespaces; |
53 | extern const std::string LineEndings; | |
6a93d84a TJ |
54 | |
55 | ||
56 | /* | |
57 | ** predicates: | |
58 | */ | |
59 | ||
60 | ||
6ab3bc95 | 61 | bool has_prefix(const std::string& str, const std::string& prefix); |
6a93d84a | 62 | |
6ab3bc95 | 63 | bool has_suffix(const std::string& str, const std::string& suffix); |
6a93d84a TJ |
64 | |
65 | ||
66 | /* | |
6ab3bc95 | 67 | ** tool functions(modifying): |
6a93d84a TJ |
68 | */ |
69 | ||
6ab3bc95 | 70 | std::string trim_mod(std::string& str, const std::string& charlist = Whitespaces); |
6a93d84a | 71 | |
6ab3bc95 | 72 | std::string chomp_mod(std::string& str, const std::string& what= LineEndings ); |
6a93d84a | 73 | |
6ab3bc95 | 74 | std::string to_lower_mod(std::string& str); |
6a93d84a | 75 | |
6ab3bc95 | 76 | std::string to_upper_mod(std::string& str); |
6a93d84a TJ |
77 | |
78 | ||
79 | /* | |
80 | ** tool functions (not modifying): | |
81 | */ | |
82 | ||
6ab3bc95 | 83 | std::string trim(const std::string& str, const std::string& charlist = Whitespaces); |
6a93d84a | 84 | |
6ab3bc95 | 85 | std::string chomp(const std::string& str, const std::string& what= LineEndings ); |
6a93d84a | 86 | |
6ab3bc95 | 87 | std::string to_lower(const std::string& str); |
6a93d84a | 88 | |
6ab3bc95 | 89 | std::string to_upper(const std::string& str); |
6a93d84a TJ |
90 | |
91 | ||
6ab3bc95 | 92 | std::string remove_suffix(const std::string& str, const std::string& suffix); |
6a93d84a | 93 | |
6ab3bc95 | 94 | std::string remove_prefix(const std::string& str, const std::string& prefix); |
6a93d84a TJ |
95 | |
96 | ||
97 | ||
98 | /* | |
99 | ** split and join: | |
100 | */ | |
101 | ||
102 | ||
6ab3bc95 RP |
103 | bool pair_split( |
104 | const std::string& str, | |
105 | std::string& key, | |
106 | std::string& value, | |
107 | char delimiter = '='); | |
6a93d84a TJ |
108 | |
109 | ||
6ab3bc95 RP |
110 | void split_string( |
111 | const std::string& str, | |
112 | std::list< std::string >& result, | |
113 | const std::string& delimiter= "\n", | |
114 | bool omit_empty= false, | |
115 | const std::string& trim_list= std::string() | |
6a93d84a TJ |
116 | ); |
117 | ||
338da253 CH |
118 | void split_string( |
119 | const std::string& str, | |
120 | std::vector< std::string >& result, | |
121 | const std::string& delimiter= "\n", | |
122 | bool omit_empty= false, | |
123 | const std::string& trim_list= std::string() | |
124 | ); | |
125 | ||
6ab3bc95 RP |
126 | std::list< std::string > split_string( |
127 | const std::string& str, | |
128 | const std::string& delimiter = "\n", | |
129 | bool omit_empty= false, | |
130 | const std::string& trim_list= std::string() | |
6a93d84a TJ |
131 | ); |
132 | ||
133 | ||
6ab3bc95 RP |
134 | std::string join_string( |
135 | const std::list< std::string >& parts, | |
136 | const std::string& delimiter = "\n" | |
6a93d84a TJ |
137 | ); |
138 | ||
376ec4fa CH |
139 | std::string join_string( |
140 | const std::vector< std::string >& parts, | |
141 | const std::string& delimiter = "\n" | |
142 | ); | |
143 | ||
4f7a7b9f | 144 | std::string join_string( |
0aaf13b5 PG |
145 | const std::set< std::string >& parts, |
146 | const std::string& delimiter = "\n" | |
147 | ); | |
148 | ||
149 | std::string join_string( | |
4f7a7b9f PG |
150 | const char *const parts [], |
151 | const std::string& delimiter = "\n" | |
152 | ); | |
153 | ||
6a93d84a TJ |
154 | |
155 | /* | |
156 | ** conversions: | |
157 | */ | |
158 | ||
159 | ||
6ab3bc95 | 160 | std::string convert_binary_to_hex(const std::string&str, bool upper_case_digits= false); |
6a93d84a | 161 | |
6ab3bc95 | 162 | std::string convert_hex_to_binary(const std::string& str) throw(std::runtime_error); |
6a93d84a TJ |
163 | |
164 | ||
6a93d84a TJ |
165 | /* |
166 | ** "type conversions": | |
167 | */ | |
168 | ||
169 | ||
170 | /** | |
171 | * convert a datatype @a T to a string via string stream. | |
172 | * | |
1fe2e899 CH |
173 | * This will not report trouble in conversion; for example: |
174 | * string_to<int>("christian") | |
175 | * will return 0 and not throw an error. | |
176 | * Use boost::lexical_cast<T>(string) to get error-checked results. | |
177 | * | |
6a93d84a TJ |
178 | * @param s the string which should be converted to @a T. |
179 | * @return the value of type T. | |
180 | */ | |
181 | template< | |
6ab3bc95 | 182 | class T |
6a93d84a | 183 | > |
6ab3bc95 | 184 | T string_to(const std::string& s) |
6a93d84a | 185 | { |
6ab3bc95 RP |
186 | std::istringstream istr(s); |
187 | T result; | |
188 | istr >> result; | |
189 | return result; | |
190 | } // eo string_to(const std::string&) | |
6a93d84a TJ |
191 | |
192 | ||
193 | /** | |
194 | * convert a datatype @a T to a string via string stream. | |
195 | * | |
196 | * @param s the string which should be converted to @a T. | |
197 | * @param result the resulting value of type @a T. | |
198 | * @return @a true iff the internal string stream was EOF after the conversion. | |
08f2d184 GE |
199 | * |
200 | * @attention: does not return if the conversion was successful. So check for empty strings before. | |
6a93d84a TJ |
201 | */ |
202 | template< | |
6ab3bc95 | 203 | class T |
6a93d84a | 204 | > |
6ab3bc95 | 205 | bool string_to(const std::string& s, T& result) |
6a93d84a | 206 | { |
6ab3bc95 RP |
207 | std::istringstream istr(s); |
208 | istr >> result; | |
209 | return istr.eof(); | |
210 | } // eo string_to(const std::string&) | |
6a93d84a TJ |
211 | |
212 | ||
213 | /** | |
3bcc713f GE |
214 | * convert string in hexadecimal notation to a datatype @a T |
215 | * supports strings with and without "0x" notation, e.g. 0xff and FF are both valid | |
216 | * | |
217 | * @param s the hex string which should be converted to @a T. | |
218 | * @return the value of type T. | |
219 | */ | |
220 | template< | |
221 | class T | |
222 | > | |
223 | T hex_string_to(const std::string& s) | |
224 | { | |
225 | std::istringstream istr(s); | |
226 | T result; | |
227 | istr >> std::hex >> result; | |
228 | return result; | |
229 | } // eo string_to(const std::string&) | |
230 | ||
231 | ||
232 | /** | |
233 | * convert string in hexadecimal notation to a datatype @a T | |
234 | * supports strings with and without "0x" notation, e.g. 0xff and FF are both valid | |
235 | * | |
236 | * @param s the hex string which should be converted to @a T. | |
237 | * @param result the resulting value of type @a T. | |
238 | * @return @a true iff the internal string stream was EOF after the conversion. | |
239 | * | |
240 | * @attention: does not return if the conversion was successful. So check for empty strings before. | |
241 | */ | |
242 | template< | |
243 | class T | |
244 | > | |
245 | bool hex_string_to(const std::string& s, T& result) | |
246 | { | |
247 | std::istringstream istr(s); | |
248 | istr >> std::hex >> result; | |
249 | return istr.eof(); | |
250 | } // eo string_to(const std::string&) | |
251 | ||
252 | ||
253 | /** | |
6a93d84a TJ |
254 | * convert a string to another datatype @a T via string stream. |
255 | * | |
256 | * @param v the value (of type @a T) which should be converted to a string. | |
257 | * @return the resulting string. | |
258 | */ | |
259 | template< | |
6ab3bc95 | 260 | class T |
6a93d84a | 261 | > |
6ab3bc95 | 262 | std::string to_string(const T& v) |
6a93d84a | 263 | { |
6ab3bc95 RP |
264 | std::ostringstream ostr; |
265 | ostr << v; | |
266 | return ostr.str(); | |
267 | } // eo to_string(const T&) | |
268 | ||
269 | ||
1a0267e5 CH |
270 | /** |
271 | * Create a string with types shortened in texts describing C++ types | |
272 | * | |
273 | * for example: std::list<some_long_type, std::allocator<some_long_type> > | |
274 | * --> std::list<some_long_type, _alloc_> | |
275 | * | |
276 | * and std::basic_string<char, std::char_traits<char>, std::allocator<char> > | |
277 | * --> std::string | |
278 | */ | |
279 | std::string shorten_stl_types(const std::string &input); | |
280 | ||
1ebab1e3 TJ |
281 | std::string base64_encode(const std::string &input, bool one_line=true); |
282 | std::string base64_decode(const std::string &input, bool one_line=true); | |
1a0267e5 | 283 | |
6ab3bc95 RP |
284 | } // eo namespace I2n |
285 | ||
6a93d84a | 286 | |
e5b21dbb | 287 | #if 0 |
6ab3bc95 RP |
288 | std::string to_lower(const std::string &src); |
289 | std::string to_upper(const std::string &src); | |
e5b21dbb RP |
290 | #else |
291 | // compatibility: import lower/upper funcs from I2n: | |
292 | using I2n::to_lower; | |
293 | using I2n::to_upper; | |
294 | #endif | |
e93545dd | 295 | |
81267544 | 296 | |
8f296924 GMF |
297 | enum UnitBase { |
298 | UnitBase1000, // SI decimal, composed by multiples of 1000 (KB, MB, etc.) | |
299 | UnitBase1024 // IEC binary, composed by multiples of 1024 (KiB, MiB, etc. ) | |
81267544 GMF |
300 | }; |
301 | ||
8f296924 GMF |
302 | enum UnitFormat { |
303 | ShortUnitFormat, // B, KB, MB, ... | |
304 | LongUnitFormat // Byte, KByte, MByte, ... | |
d1ea9075 GMF |
305 | }; |
306 | ||
81267544 GMF |
307 | std::string nice_unit_format( |
308 | const int64_t input, | |
ee037ee3 GMF |
309 | const UnitFormat format = ShortUnitFormat, |
310 | const UnitBase base = UnitBase1024 | |
81267544 | 311 | ); |
e93545dd | 312 | |
5cd64148 CH |
313 | std::string nice_unit_format( |
314 | const double input, | |
315 | const UnitFormat format = ShortUnitFormat, | |
316 | const UnitBase base = UnitBase1024 | |
317 | ); | |
318 | ||
e93545dd GE |
319 | bool replace_all(std::string &base, const std::string *ist, const std::string *soll); |
320 | bool replace_all(std::string &base, const char *ist, const char *soll); | |
321 | bool replace_all(std::string &base, const char *ist, const std::string *soll); | |
322 | bool replace_all(std::string &base, const std::string &ist, const char *soll); | |
323 | bool replace_all(std::string &base, const std::string &ist, const std::string &soll); | |
324 | ||
325 | std::string iso_to_utf8(const std::string& isostring); | |
326 | std::string utf8_to_iso(const std::string& utf8string); | |
13cc4db1 | 327 | std::string utf7imap_to_utf8(const std::string &utf7imapstring); |
6a2b6dd1 | 328 | std::string utf8_to_utf7imap(const std::string &utf8string); |
118e216e TJ |
329 | |
330 | std::string strip_html_tags(const std::string &input); | |
331 | std::string smart_html_entities(const std::string &input); | |
332 | std::string html_entities(std::string str); | |
554f813d | 333 | std::string html_entities_to_console(std::string str); |
118e216e | 334 | |
3f5c5ccd | 335 | typedef std::pair<std::string::size_type, std::string::size_type> CommentZone; |
46dd1321 | 336 | std::vector<CommentZone> find_html_comments(const std::string &str); |
3f5c5ccd CH |
337 | void remove_html_comments(std::string &str); |
338 | void remove_html_comments(std::string &str, const std::vector<CommentZone> &comments); | |
339 | ||
b953bf36 GE |
340 | std::string sanitize_for_logging(const std::string &str, const char replace_with='?'); |
341 | ||
47c07fba GE |
342 | std::string escape(const std::string &s); |
343 | ||
344 | std::string descape(const std::string &s, int startpos, int &endpos); | |
e6da286a | 345 | inline std::string descape(const std::string &s) |
47c07fba | 346 | { |
6ab3bc95 RP |
347 | int endpos; |
348 | return descape(s,0,endpos); | |
47c07fba GE |
349 | } |
350 | ||
351 | std::string escape_shellarg(const std::string &input); | |
352 | ||
e93545dd | 353 | #endif |