2 The software in this package is distributed under the GNU General
3 Public License version 2 (with a special exception described below).
5 A copy of GNU General Public License (GPL) is included in this distribution,
6 in the file COPYING.GPL.
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.
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.
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.
21 * @brief collection of string tools (/ functions).
23 * contains a collection of miscellaneous functions for dealing with strings.
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.
31 * (c) Copyright 2007-2008 by Intra2net AG
34 #ifndef __STRINGFUNC_HXX
35 #define __STRINGFUNC_HXX
42 #include <sys/types.h>
48 ** some useful constants:
51 extern const std::string Whitespaces;
52 extern const std::string LineEndings;
60 bool has_prefix(const std::string& str, const std::string& prefix);
62 bool has_suffix(const std::string& str, const std::string& suffix);
66 ** tool functions(modifying):
69 std::string trim_mod(std::string& str, const std::string& charlist = Whitespaces);
71 std::string chomp_mod(std::string& str, const std::string& what= LineEndings );
73 std::string to_lower_mod(std::string& str);
75 std::string to_upper_mod(std::string& str);
79 ** tool functions (not modifying):
82 std::string trim(const std::string& str, const std::string& charlist = Whitespaces);
84 std::string chomp(const std::string& str, const std::string& what= LineEndings );
86 std::string to_lower(const std::string& str);
88 std::string to_upper(const std::string& str);
91 std::string remove_suffix(const std::string& str, const std::string& suffix);
93 std::string remove_prefix(const std::string& str, const std::string& prefix);
103 const std::string& str,
106 char delimiter = '=');
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()
117 std::list< std::string > split_string(
118 const std::string& str,
119 const std::string& delimiter = "\n",
120 bool omit_empty= false,
121 const std::string& trim_list= std::string()
125 std::string join_string(
126 const std::list< std::string >& parts,
127 const std::string& delimiter = "\n"
130 std::string join_string(
131 const std::vector< std::string >& parts,
132 const std::string& delimiter = "\n"
141 std::string convert_binary_to_hex(const std::string&str, bool upper_case_digits= false);
143 std::string convert_hex_to_binary(const std::string& str) throw(std::runtime_error);
147 ** "type conversions":
152 * convert a datatype @a T to a string via string stream.
154 * This will not report trouble in conversion; for example:
155 * string_to<int>("christian")
156 * will return 0 and not throw an error.
157 * Use boost::lexical_cast<T>(string) to get error-checked results.
159 * @param s the string which should be converted to @a T.
160 * @return the value of type T.
165 T string_to(const std::string& s)
167 std::istringstream istr(s);
171 } // eo string_to(const std::string&)
175 * convert a datatype @a T to a string via string stream.
177 * @param s the string which should be converted to @a T.
178 * @param result the resulting value of type @a T.
179 * @return @a true iff the internal string stream was EOF after the conversion.
181 * @attention: does not return if the conversion was successful. So check for empty strings before.
186 bool string_to(const std::string& s, T& result)
188 std::istringstream istr(s);
191 } // eo string_to(const std::string&)
195 * convert string in hexadecimal notation to a datatype @a T
196 * supports strings with and without "0x" notation, e.g. 0xff and FF are both valid
198 * @param s the hex string which should be converted to @a T.
199 * @return the value of type T.
204 T hex_string_to(const std::string& s)
206 std::istringstream istr(s);
208 istr >> std::hex >> result;
210 } // eo string_to(const std::string&)
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
217 * @param s the hex string which should be converted to @a T.
218 * @param result the resulting value of type @a T.
219 * @return @a true iff the internal string stream was EOF after the conversion.
221 * @attention: does not return if the conversion was successful. So check for empty strings before.
226 bool hex_string_to(const std::string& s, T& result)
228 std::istringstream istr(s);
229 istr >> std::hex >> result;
231 } // eo string_to(const std::string&)
235 * convert a string to another datatype @a T via string stream.
237 * @param v the value (of type @a T) which should be converted to a string.
238 * @return the resulting string.
243 std::string to_string(const T& v)
245 std::ostringstream ostr;
248 } // eo to_string(const T&)
251 } // eo namespace I2n
255 std::string to_lower(const std::string &src);
256 std::string to_upper(const std::string &src);
258 // compatibility: import lower/upper funcs from I2n:
265 UnitBase1000, // SI decimal, composed by multiples of 1000 (KB, MB, etc.)
266 UnitBase1024 // IEC binary, composed by multiples of 1024 (KiB, MiB, etc. )
270 ShortUnitFormat, // B, KB, MB, ...
271 LongUnitFormat // Byte, KByte, MByte, ...
274 std::string nice_unit_format(
276 const UnitFormat format = ShortUnitFormat,
277 const UnitBase base = UnitBase1024
280 std::string nice_unit_format(
282 const UnitFormat format = ShortUnitFormat,
283 const UnitBase base = UnitBase1024
286 bool replace_all(std::string &base, const std::string *ist, const std::string *soll);
287 bool replace_all(std::string &base, const char *ist, const char *soll);
288 bool replace_all(std::string &base, const char *ist, const std::string *soll);
289 bool replace_all(std::string &base, const std::string &ist, const char *soll);
290 bool replace_all(std::string &base, const std::string &ist, const std::string &soll);
292 std::string iso_to_utf8(const std::string& isostring);
293 std::string utf8_to_iso(const std::string& utf8string);
294 std::string utf7imap_to_utf8(const std::string &utf7imapstring);
295 std::string utf8_to_utf7imap(const std::string &utf8string);
297 std::string strip_html_tags(const std::string &input);
298 std::string smart_html_entities(const std::string &input);
299 std::string html_entities(std::string str);
300 std::string html_entities_to_console(std::string str);
302 typedef std::pair<std::string::size_type, std::string::size_type> CommentZone;
303 void find_html_comments(const std::string &str, std::vector<CommentZone> &result);
304 void remove_html_comments(std::string &str);
305 void remove_html_comments(std::string &str, const std::vector<CommentZone> &comments);
307 std::string sanitize_for_logging(const std::string &str, const char replace_with='?');
309 std::string escape(const std::string &s);
311 std::string descape(const std::string &s, int startpos, int &endpos);
312 inline std::string descape(const std::string &s)
315 return descape(s,0,endpos);
318 std::string escape_shellarg(const std::string &input);