| 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 | */ |
| 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 |
| 32 | */ |
| 33 | |
| 34 | #ifndef __STRINGFUNC_HXX |
| 35 | #define __STRINGFUNC_HXX |
| 36 | |
| 37 | #include <list> |
| 38 | #include <vector> |
| 39 | #include <string> |
| 40 | #include <sstream> |
| 41 | #include <stdexcept> |
| 42 | #include <sys/types.h> |
| 43 | |
| 44 | namespace I2n |
| 45 | { |
| 46 | |
| 47 | /* |
| 48 | ** some useful constants: |
| 49 | */ |
| 50 | |
| 51 | extern const std::string Whitespaces; |
| 52 | extern const std::string LineEndings; |
| 53 | |
| 54 | |
| 55 | /* |
| 56 | ** predicates: |
| 57 | */ |
| 58 | |
| 59 | |
| 60 | bool has_prefix(const std::string& str, const std::string& prefix); |
| 61 | |
| 62 | bool has_suffix(const std::string& str, const std::string& suffix); |
| 63 | |
| 64 | |
| 65 | /* |
| 66 | ** tool functions(modifying): |
| 67 | */ |
| 68 | |
| 69 | std::string trim_mod(std::string& str, const std::string& charlist = Whitespaces); |
| 70 | |
| 71 | std::string chomp_mod(std::string& str, const std::string& what= LineEndings ); |
| 72 | |
| 73 | std::string to_lower_mod(std::string& str); |
| 74 | |
| 75 | std::string to_upper_mod(std::string& str); |
| 76 | |
| 77 | |
| 78 | /* |
| 79 | ** tool functions (not modifying): |
| 80 | */ |
| 81 | |
| 82 | std::string trim(const std::string& str, const std::string& charlist = Whitespaces); |
| 83 | |
| 84 | std::string chomp(const std::string& str, const std::string& what= LineEndings ); |
| 85 | |
| 86 | std::string to_lower(const std::string& str); |
| 87 | |
| 88 | std::string to_upper(const std::string& str); |
| 89 | |
| 90 | |
| 91 | std::string remove_suffix(const std::string& str, const std::string& suffix); |
| 92 | |
| 93 | std::string remove_prefix(const std::string& str, const std::string& prefix); |
| 94 | |
| 95 | |
| 96 | |
| 97 | /* |
| 98 | ** split and join: |
| 99 | */ |
| 100 | |
| 101 | |
| 102 | bool pair_split( |
| 103 | const std::string& str, |
| 104 | std::string& key, |
| 105 | std::string& value, |
| 106 | char delimiter = '='); |
| 107 | |
| 108 | |
| 109 | void split_string( |
| 110 | const std::string& str, |
| 111 | std::list< std::string >& result, |
| 112 | const std::string& delimiter= "\n", |
| 113 | bool omit_empty= false, |
| 114 | const std::string& trim_list= std::string() |
| 115 | ); |
| 116 | |
| 117 | void split_string( |
| 118 | const std::string& str, |
| 119 | std::vector< std::string >& result, |
| 120 | const std::string& delimiter= "\n", |
| 121 | bool omit_empty= false, |
| 122 | const std::string& trim_list= std::string() |
| 123 | ); |
| 124 | |
| 125 | std::list< std::string > split_string( |
| 126 | const std::string& str, |
| 127 | const std::string& delimiter = "\n", |
| 128 | bool omit_empty= false, |
| 129 | const std::string& trim_list= std::string() |
| 130 | ); |
| 131 | |
| 132 | |
| 133 | std::string join_string( |
| 134 | const std::list< std::string >& parts, |
| 135 | const std::string& delimiter = "\n" |
| 136 | ); |
| 137 | |
| 138 | std::string join_string( |
| 139 | const std::vector< std::string >& parts, |
| 140 | const std::string& delimiter = "\n" |
| 141 | ); |
| 142 | |
| 143 | std::string join_string( |
| 144 | const char *const parts [], |
| 145 | const std::string& delimiter = "\n" |
| 146 | ); |
| 147 | |
| 148 | |
| 149 | /* |
| 150 | ** conversions: |
| 151 | */ |
| 152 | |
| 153 | |
| 154 | std::string convert_binary_to_hex(const std::string&str, bool upper_case_digits= false); |
| 155 | |
| 156 | std::string convert_hex_to_binary(const std::string& str) throw(std::runtime_error); |
| 157 | |
| 158 | |
| 159 | /* |
| 160 | ** "type conversions": |
| 161 | */ |
| 162 | |
| 163 | |
| 164 | /** |
| 165 | * convert a datatype @a T to a string via string stream. |
| 166 | * |
| 167 | * This will not report trouble in conversion; for example: |
| 168 | * string_to<int>("christian") |
| 169 | * will return 0 and not throw an error. |
| 170 | * Use boost::lexical_cast<T>(string) to get error-checked results. |
| 171 | * |
| 172 | * @param s the string which should be converted to @a T. |
| 173 | * @return the value of type T. |
| 174 | */ |
| 175 | template< |
| 176 | class T |
| 177 | > |
| 178 | T string_to(const std::string& s) |
| 179 | { |
| 180 | std::istringstream istr(s); |
| 181 | T result; |
| 182 | istr >> result; |
| 183 | return result; |
| 184 | } // eo string_to(const std::string&) |
| 185 | |
| 186 | |
| 187 | /** |
| 188 | * convert a datatype @a T to a string via string stream. |
| 189 | * |
| 190 | * @param s the string which should be converted to @a T. |
| 191 | * @param result the resulting value of type @a T. |
| 192 | * @return @a true iff the internal string stream was EOF after the conversion. |
| 193 | * |
| 194 | * @attention: does not return if the conversion was successful. So check for empty strings before. |
| 195 | */ |
| 196 | template< |
| 197 | class T |
| 198 | > |
| 199 | bool string_to(const std::string& s, T& result) |
| 200 | { |
| 201 | std::istringstream istr(s); |
| 202 | istr >> result; |
| 203 | return istr.eof(); |
| 204 | } // eo string_to(const std::string&) |
| 205 | |
| 206 | |
| 207 | /** |
| 208 | * convert string in hexadecimal notation to a datatype @a T |
| 209 | * supports strings with and without "0x" notation, e.g. 0xff and FF are both valid |
| 210 | * |
| 211 | * @param s the hex string which should be converted to @a T. |
| 212 | * @return the value of type T. |
| 213 | */ |
| 214 | template< |
| 215 | class T |
| 216 | > |
| 217 | T hex_string_to(const std::string& s) |
| 218 | { |
| 219 | std::istringstream istr(s); |
| 220 | T result; |
| 221 | istr >> std::hex >> result; |
| 222 | return result; |
| 223 | } // eo string_to(const std::string&) |
| 224 | |
| 225 | |
| 226 | /** |
| 227 | * convert string in hexadecimal notation to a datatype @a T |
| 228 | * supports strings with and without "0x" notation, e.g. 0xff and FF are both valid |
| 229 | * |
| 230 | * @param s the hex string which should be converted to @a T. |
| 231 | * @param result the resulting value of type @a T. |
| 232 | * @return @a true iff the internal string stream was EOF after the conversion. |
| 233 | * |
| 234 | * @attention: does not return if the conversion was successful. So check for empty strings before. |
| 235 | */ |
| 236 | template< |
| 237 | class T |
| 238 | > |
| 239 | bool hex_string_to(const std::string& s, T& result) |
| 240 | { |
| 241 | std::istringstream istr(s); |
| 242 | istr >> std::hex >> result; |
| 243 | return istr.eof(); |
| 244 | } // eo string_to(const std::string&) |
| 245 | |
| 246 | |
| 247 | /** |
| 248 | * convert a string to another datatype @a T via string stream. |
| 249 | * |
| 250 | * @param v the value (of type @a T) which should be converted to a string. |
| 251 | * @return the resulting string. |
| 252 | */ |
| 253 | template< |
| 254 | class T |
| 255 | > |
| 256 | std::string to_string(const T& v) |
| 257 | { |
| 258 | std::ostringstream ostr; |
| 259 | ostr << v; |
| 260 | return ostr.str(); |
| 261 | } // eo to_string(const T&) |
| 262 | |
| 263 | |
| 264 | /** |
| 265 | * Create a string with types shortened in texts describing C++ types |
| 266 | * |
| 267 | * for example: std::list<some_long_type, std::allocator<some_long_type> > |
| 268 | * --> std::list<some_long_type, _alloc_> |
| 269 | * |
| 270 | * and std::basic_string<char, std::char_traits<char>, std::allocator<char> > |
| 271 | * --> std::string |
| 272 | */ |
| 273 | std::string shorten_stl_types(const std::string &input); |
| 274 | |
| 275 | std::string base64_encode(const std::string &input, bool one_line=true); |
| 276 | std::string base64_decode(const std::string &input, bool one_line=true); |
| 277 | |
| 278 | } // eo namespace I2n |
| 279 | |
| 280 | |
| 281 | #if 0 |
| 282 | std::string to_lower(const std::string &src); |
| 283 | std::string to_upper(const std::string &src); |
| 284 | #else |
| 285 | // compatibility: import lower/upper funcs from I2n: |
| 286 | using I2n::to_lower; |
| 287 | using I2n::to_upper; |
| 288 | #endif |
| 289 | |
| 290 | |
| 291 | enum UnitBase { |
| 292 | UnitBase1000, // SI decimal, composed by multiples of 1000 (KB, MB, etc.) |
| 293 | UnitBase1024 // IEC binary, composed by multiples of 1024 (KiB, MiB, etc. ) |
| 294 | }; |
| 295 | |
| 296 | enum UnitFormat { |
| 297 | ShortUnitFormat, // B, KB, MB, ... |
| 298 | LongUnitFormat // Byte, KByte, MByte, ... |
| 299 | }; |
| 300 | |
| 301 | std::string nice_unit_format( |
| 302 | const int64_t input, |
| 303 | const UnitFormat format = ShortUnitFormat, |
| 304 | const UnitBase base = UnitBase1024 |
| 305 | ); |
| 306 | |
| 307 | std::string nice_unit_format( |
| 308 | const double input, |
| 309 | const UnitFormat format = ShortUnitFormat, |
| 310 | const UnitBase base = UnitBase1024 |
| 311 | ); |
| 312 | |
| 313 | bool replace_all(std::string &base, const std::string *ist, const std::string *soll); |
| 314 | bool replace_all(std::string &base, const char *ist, const char *soll); |
| 315 | bool replace_all(std::string &base, const char *ist, const std::string *soll); |
| 316 | bool replace_all(std::string &base, const std::string &ist, const char *soll); |
| 317 | bool replace_all(std::string &base, const std::string &ist, const std::string &soll); |
| 318 | |
| 319 | std::string iso_to_utf8(const std::string& isostring); |
| 320 | std::string utf8_to_iso(const std::string& utf8string); |
| 321 | std::string utf7imap_to_utf8(const std::string &utf7imapstring); |
| 322 | std::string utf8_to_utf7imap(const std::string &utf8string); |
| 323 | |
| 324 | std::string strip_html_tags(const std::string &input); |
| 325 | std::string smart_html_entities(const std::string &input); |
| 326 | std::string html_entities(std::string str); |
| 327 | std::string html_entities_to_console(std::string str); |
| 328 | |
| 329 | typedef std::pair<std::string::size_type, std::string::size_type> CommentZone; |
| 330 | std::vector<CommentZone> find_html_comments(const std::string &str); |
| 331 | void remove_html_comments(std::string &str); |
| 332 | void remove_html_comments(std::string &str, const std::vector<CommentZone> &comments); |
| 333 | |
| 334 | std::string sanitize_for_logging(const std::string &str, const char replace_with='?'); |
| 335 | |
| 336 | std::string escape(const std::string &s); |
| 337 | |
| 338 | std::string descape(const std::string &s, int startpos, int &endpos); |
| 339 | inline std::string descape(const std::string &s) |
| 340 | { |
| 341 | int endpos; |
| 342 | return descape(s,0,endpos); |
| 343 | } |
| 344 | |
| 345 | std::string escape_shellarg(const std::string &input); |
| 346 | |
| 347 | #endif |