2 * Copyright 2012 Facebook, Inc.
4 * Licensed under the Apache License, Version 2.0 (the "License");
5 * you may not use this file except in compliance with the License.
6 * You may obtain a copy of the License at
8 * http://www.apache.org/licenses/LICENSE-2.0
10 * Unless required by applicable law or agreed to in writing, software
11 * distributed under the License is distributed on an "AS IS" BASIS,
12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13 * See the License for the specific language governing permissions and
14 * limitations under the License.
17 #ifndef FOLLY_BASE_STRING_H_
18 #define FOLLY_BASE_STRING_H_
21 #include <boost/type_traits.hpp>
24 # include <ext/hash_set>
25 # include <ext/hash_map>
28 #include "folly/Conv.h"
29 #include "folly/FBString.h"
30 #include "folly/FBVector.h"
31 #include "folly/Range.h"
32 #include "folly/ScopeGuard.h"
34 // Compatibility function, to make sure toStdString(s) can be called
35 // to convert a std::string or fbstring variable s into type std::string
36 // with very little overhead if s was already std::string
40 std::string toStdString(const folly::fbstring& s) {
41 return std::string(s.data(), s.size());
45 const std::string& toStdString(const std::string& s) {
49 // If called with a temporary, the compiler will select this overload instead
50 // of the above, so we don't return a (lvalue) reference to a temporary.
52 std::string&& toStdString(std::string&& s) {
57 * C-Escape a string, making it suitable for representation as a C string
58 * literal. Appends the result to the output string.
60 * Backslashes all occurrences of backslash and double-quote:
64 * Replaces all non-printable ASCII characters with backslash-octal
68 * Note that we use backslash-octal instead of backslash-hex because the octal
69 * representation is guaranteed to consume no more than 3 characters; "\3760"
70 * represents two characters, one with value 254, and one with value 48 ('0'),
71 * whereas "\xfe0" represents only one character (with value 4064, which leads
72 * to implementation-defined behavior).
74 template <class String>
75 void cEscape(StringPiece str, String& out);
78 * Similar to cEscape above, but returns the escaped string.
80 template <class String>
81 String cEscape(StringPiece str) {
88 * C-Unescape a string; the opposite of cEscape above. Appends the result
89 * to the output string.
91 * Recognizes the standard C escape sequences:
93 * \' \" \? \\ \a \b \f \n \r \t \v
97 * In strict mode (default), throws std::invalid_argument if it encounters
98 * an unrecognized escape sequence. In non-strict mode, it leaves
99 * the escape sequence unchanged.
101 template <class String>
102 void cUnescape(StringPiece str, String& out, bool strict = true);
105 * Similar to cUnescape above, but returns the escaped string.
107 template <class String>
108 String cUnescape(StringPiece str, bool strict = true) {
110 cUnescape(str, out, strict);
115 * stringPrintf is much like printf but deposits its result into a
116 * string. Two signatures are supported: the first simply returns the
117 * resulting string, and the second appends the produced characters to
118 * the specified string and returns a reference to it.
120 std::string stringPrintf(const char* format, ...)
121 __attribute__ ((format (printf, 1, 2)));
123 /** Similar to stringPrintf, with different signiture.
125 void stringPrintf(std::string* out, const char* fmt, ...)
126 __attribute__ ((format (printf, 2, 3)));
128 std::string& stringAppendf(std::string* output, const char* format, ...)
129 __attribute__ ((format (printf, 2, 3)));
132 * Backslashify a string, that is, replace non-printable characters
133 * with C-style (but NOT C compliant) "\xHH" encoding. If hex_style
134 * is false, then shorthand notations like "\0" will be used instead
135 * of "\x00" for the most common backslash cases.
137 * There are two forms, one returning the input string, and one
138 * creating output in the specified output string.
140 * This is mainly intended for printing to a terminal, so it is not
141 * particularly optimized.
143 * Do *not* use this in situations where you expect to be able to feed
144 * the string to a C or C++ compiler, as there are nuances with how C
145 * parses such strings that lead to failures. This is for display
146 * purposed only. If you want a string you can embed for use in C or
147 * C++, use cEscape instead. This function is for display purposes
150 template <class String1, class String2>
151 void backslashify(const String1& input, String2& output, bool hex_style=false);
153 template <class String>
154 String backslashify(const String& input, bool hex_style=false) {
156 backslashify(input, output, hex_style);
161 * Take a string and "humanify" it -- that is, make it look better.
162 * Since "better" is subjective, caveat emptor. The basic approach is
163 * to count the number of unprintable characters. If there are none,
164 * then the output is the input. If there are relatively few, or if
165 * there is a long "enough" prefix of printable characters, use
166 * backslashify. If it is mostly binary, then simply hex encode.
168 * This is an attempt to make a computer smart, and so likely is wrong
171 template <class String1, class String2>
172 void humanify(const String1& input, String2& output);
174 template <class String>
175 String humanify(const String& input) {
177 humanify(input, output);
182 * Same functionality as Python's binascii.hexlify. Returns true
183 * on successful conversion.
185 * If append_output is true, append data to the output rather than
188 template<class InputString, class OutputString>
189 bool hexlify(const InputString& input, OutputString& output,
193 * Same functionality as Python's binascii.unhexlify. Returns true
194 * on successful conversion.
196 template<class InputString, class OutputString>
197 bool unhexlify(const InputString& input, OutputString& output);
200 * A pretty-printer for numbers that appends suffixes of units of the
201 * given type. It prints 4 sig-figs of value with the most
204 * If `addSpace' is true, we put a space between the units suffix and
208 * PRETTY_TIME - s, ms, us, ns, etc.
209 * PRETTY_BYTES_METRIC - kB, MB, GB, etc (goes up by 10^3 = 1000 each time)
210 * PRETTY_BYTES - kB, MB, GB, etc (goes up by 2^10 = 1024 each time)
211 * PRETTY_BYTES_IEC - KiB, MiB, GiB, etc
212 * PRETTY_UNITS_METRIC - k, M, G, etc (goes up by 10^3 = 1000 each time)
213 * PRETTY_UNITS_BINARY - k, M, G, etc (goes up by 2^10 = 1024 each time)
214 * PRETTY_UNITS_BINARY_IEC - Ki, Mi, Gi, etc
216 * @author Mark Rabkin <mrabkin@fb.com>
223 PRETTY_BYTES = PRETTY_BYTES_BINARY,
224 PRETTY_BYTES_BINARY_IEC,
225 PRETTY_BYTES_IEC = PRETTY_BYTES_BINARY_IEC,
229 PRETTY_UNITS_BINARY_IEC,
234 std::string prettyPrint(double val, PrettyType, bool addSpace = true);
237 * Write a hex dump of size bytes starting at ptr to out.
239 * The hex dump is formatted as follows:
241 * for the string "abcdefghijklmnopqrstuvwxyz\x02"
242 00000000 61 62 63 64 65 66 67 68 69 6a 6b 6c 6d 6e 6f 70 |abcdefghijklmnop|
243 00000010 71 72 73 74 75 76 77 78 79 7a 02 |qrstuvwxyz. |
245 * that is, we write 16 bytes per line, both as hex bytes and as printable
246 * characters. Non-printable characters are replaced with '.'
247 * Lines are written to out one by one (one StringPiece at a time) without
250 template <class OutIt>
251 void hexDump(const void* ptr, size_t size, OutIt out);
254 * Return the hex dump of size bytes starting at ptr as a string.
256 std::string hexDump(const void* ptr, size_t size);
259 * Return a fbstring containing the description of the given errno value.
260 * Takes care not to overwrite the actual system errno, so calling
261 * errnoStr(errno) is valid.
263 fbstring errnoStr(int err);
266 * Return the demangled (prettyfied) version of a C++ type.
268 * This function tries to produce a human-readable type, but the type name will
269 * be returned unchanged in case of error or if demangling isn't supported on
272 * Use for debugging -- do not rely on demangle() returning anything useful.
274 * This function may allocate memory (and therefore throw).
276 fbstring demangle(const char* name);
277 inline fbstring demangle(const std::type_info& type) {
278 return demangle(type.name());
282 * Debug string for an exception: include type and what().
284 inline fbstring exceptionStr(const std::exception& e) {
285 return folly::to<fbstring>(demangle(typeid(e)), ": ", e.what());
289 * Split a string into a list of tokens by delimiter.
291 * The split interface here supports different output types, selected
292 * at compile time: StringPiece, fbstring, or std::string. If you are
293 * using a vector to hold the output, it detects the type based on
294 * what your vector contains. If the output vector is not empty, split
295 * will append to the end of the vector.
297 * You can also use splitTo() to write the output to an arbitrary
298 * OutputIterator (e.g. std::inserter() on a std::set<>), in which
299 * case you have to tell the function the type. (Rationale:
300 * OutputIterators don't have a value_type, so we can't detect the
301 * type in splitTo without being told.)
305 * std::vector<folly::StringPiece> v;
306 * folly::split(":", "asd:bsd", v);
308 * std::set<StringPiece> s;
309 * folly::splitTo<StringPiece>(":", "asd:bsd:asd:csd",
310 * std::inserter(s, s.begin()));
312 * Split also takes a flag (ignoreEmpty) that indicates whether adjacent
313 * delimiters should be treated as one single separator (ignoring empty tokens)
314 * or not (generating empty tokens).
317 template<class Delim, class String, class OutputType>
318 void split(const Delim& delimiter,
320 std::vector<OutputType>& out,
321 bool ignoreEmpty = false);
323 template<class Delim, class String, class OutputType>
324 void split(const Delim& delimiter,
326 folly::fbvector<OutputType>& out,
327 bool ignoreEmpty = false);
329 template<class OutputValueType, class Delim, class String,
330 class OutputIterator>
331 void splitTo(const Delim& delimiter,
334 bool ignoreEmpty = false);
337 * Join list of tokens.
339 * Stores a string representation of tokens in the same order with
340 * deliminer between each element.
343 template <class Delim, class Iterator, class String>
344 void join(const Delim& delimiter,
349 template <class Delim, class Container, class String>
350 void join(const Delim& delimiter,
351 const Container& container,
353 join(delimiter, container.begin(), container.end(), output);
358 // Hash functions for string and fbstring usable with e.g. hash_map
360 namespace __gnu_cxx {
363 struct hash<folly::basic_fbstring<C> > : private hash<const C*> {
364 size_t operator()(const folly::basic_fbstring<C> & s) const {
365 return hash<const C*>::operator()(s.c_str());
370 struct hash<std::basic_string<C> > : private hash<const C*> {
371 size_t operator()(const std::basic_string<C> & s) const {
372 return hash<const C*>::operator()(s.c_str());
376 } // namespace __gnu_cxx
379 // Hook into boost's type traits
382 struct has_nothrow_constructor<folly::basic_fbstring<T> > : true_type {
383 enum { value = true };
387 #include "folly/String-inl.h"