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
45 #include <sys/types.h>
51 ** some useful constants:
54 extern const std::string Whitespaces;
55 extern const std::string LineEndings;
63 bool has_prefix(const std::string& str, const std::string& prefix);
65 bool has_suffix(const std::string& str, const std::string& suffix);
69 ** tool functions(modifying):
72 std::string trim_mod(std::string& str, const std::string& charlist = Whitespaces);
74 std::string chomp_mod(std::string& str, const std::string& what= LineEndings );
76 std::string to_lower_mod(std::string& str);
78 std::string to_upper_mod(std::string& str);
82 ** tool functions (not modifying):
85 std::string trim(const std::string& str, const std::string& charlist = Whitespaces);
87 std::string chomp(const std::string& str, const std::string& what= LineEndings );
89 std::string to_lower(const std::string& str);
91 std::string to_upper(const std::string& str);
94 std::string remove_suffix(const std::string& str, const std::string& suffix);
96 std::string remove_prefix(const std::string& str, const std::string& prefix);
106 const std::string& str,
109 char delimiter = '=');
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()
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()
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()
135 struct concatenator {
138 concatenator (const std::string &delim) : delim (delim) { }
140 inline std::string operator() (const std::string &acc, const std::string &elt) const
141 { return acc + delim + elt; }
144 template<typename Iter>
149 const std::string &delimiter = "\n"
152 if (first == last) { return ""; }
154 const std::string &init = *first++;
155 if (first == last) { return init; }
157 return std::accumulate (first, last, init, concatenator (delimiter));
161 * @brief joins a container of strings into a single string.
163 * This funtion is (basically) the reverse operation of @a split_string.
165 * @param parts the container of strings.
166 * @param delimiter the delimiter to insert between the strings.
167 * @return the joined string.
169 template<typename Cont>
170 inline std::string join_string(
172 const std::string& delimiter = "\n"
174 { return join_string (parts.begin (), parts.end (), delimiter); }
176 std::string join_string(
177 const char *const parts [],
178 const std::string& delimiter = "\n"
186 std::string convert_binary_to_hex(const std::string&str, bool upper_case_digits= false);
188 std::string convert_hex_to_binary(const std::string& str) throw(std::runtime_error);
192 ** "type conversions":
197 * convert a datatype @a T to a string via string stream.
199 * This will not report trouble in conversion; for example:
200 * string_to<int>("christian")
201 * will return 0 and not throw an error.
202 * Use boost::lexical_cast<T>(string) to get error-checked results.
204 * @param s the string which should be converted to @a T.
205 * @return the value of type T.
210 T string_to(const std::string& s)
212 std::istringstream istr(s);
216 } // eo string_to(const std::string&)
220 * convert a datatype @a T to a string via string stream.
222 * @param s the string which should be converted to @a T.
223 * @param result the resulting value of type @a T.
224 * @return @a true iff the internal string stream was EOF after the conversion.
226 * @attention: does not return if the conversion was successful. So check for empty strings before.
231 bool string_to(const std::string& s, T& result)
233 std::istringstream istr(s);
236 } // eo string_to(const std::string&)
240 * convert string in hexadecimal notation to a datatype @a T
241 * supports strings with and without "0x" notation, e.g. 0xff and FF are both valid
243 * @param s the hex string which should be converted to @a T.
244 * @return the value of type T.
249 T hex_string_to(const std::string& s)
251 std::istringstream istr(s);
253 istr >> std::hex >> result;
255 } // eo string_to(const std::string&)
259 * convert string in hexadecimal notation to a datatype @a T
260 * supports strings with and without "0x" notation, e.g. 0xff and FF are both valid
262 * @param s the hex string which should be converted to @a T.
263 * @param result the resulting value of type @a T.
264 * @return @a true iff the internal string stream was EOF after the conversion.
266 * @attention: does not return if the conversion was successful. So check for empty strings before.
271 bool hex_string_to(const std::string& s, T& result)
273 std::istringstream istr(s);
274 istr >> std::hex >> result;
276 } // eo string_to(const std::string&)
280 * convert a string to another datatype @a T via string stream.
282 * @param v the value (of type @a T) which should be converted to a string.
283 * @return the resulting string.
288 std::string to_string(const T& v)
290 std::ostringstream ostr;
293 } // eo to_string(const T&)
297 * Create a string with types shortened in texts describing C++ types
299 * for example: std::list<some_long_type, std::allocator<some_long_type> >
300 * --> std::list<some_long_type, _alloc_>
302 * and std::basic_string<char, std::char_traits<char>, std::allocator<char> >
305 std::string shorten_stl_types(const std::string &input);
307 std::string base64_encode(const std::string &input, bool one_line=true);
308 std::string base64_decode(const std::string &input, bool one_line=true);
310 } // eo namespace I2n
314 std::string to_lower(const std::string &src);
315 std::string to_upper(const std::string &src);
317 // compatibility: import lower/upper funcs from I2n:
324 UnitBase1000, // SI decimal, composed by multiples of 1000 (KB, MB, etc.)
325 UnitBase1024 // IEC binary, composed by multiples of 1024 (KiB, MiB, etc. )
329 ShortUnitFormat, // B, KB, MB, ...
330 LongUnitFormat // Byte, KByte, MByte, ...
333 std::string nice_unit_format(
335 const UnitFormat format = ShortUnitFormat,
336 const UnitBase base = UnitBase1024
339 std::string nice_unit_format(
341 const UnitFormat format = ShortUnitFormat,
342 const UnitBase base = UnitBase1024
345 bool replace_all(std::string &base, const std::string *ist, const std::string *soll);
346 bool replace_all(std::string &base, const char *ist, const char *soll);
347 bool replace_all(std::string &base, const char *ist, const std::string *soll);
348 bool replace_all(std::string &base, const std::string &ist, const char *soll);
349 bool replace_all(std::string &base, const std::string &ist, const std::string &soll);
351 std::string iso_to_utf8(const std::string& isostring);
352 std::string utf8_to_iso(const std::string& utf8string);
353 std::string utf7imap_to_utf8(const std::string &utf7imapstring);
354 std::string utf8_to_utf7imap(const std::string &utf8string);
356 std::string strip_html_tags(const std::string &input);
357 std::string smart_html_entities(const std::string &input);
358 std::string html_entities(std::string str);
359 std::string html_entities_to_console(std::string str);
361 typedef std::pair<std::string::size_type, std::string::size_type> CommentZone;
362 std::vector<CommentZone> find_html_comments(const std::string &str);
363 void remove_html_comments(std::string &str);
364 void remove_html_comments(std::string &str, const std::vector<CommentZone> &comments);
366 std::string sanitize_for_logging(const std::string &str, const char replace_with='?');
368 std::string escape(const std::string &s);
370 std::string descape(const std::string &s, int startpos, int &endpos);
371 inline std::string descape(const std::string &s)
374 return descape(s,0,endpos);
377 std::string escape_shellarg(const std::string &input);