1// Protocol Buffers - Google's data interchange format
2// Copyright 2008 Google Inc. All rights reserved.
3// https://developers.google.com/protocol-buffers/
4//
5// Redistribution and use in source and binary forms, with or without
6// modification, are permitted provided that the following conditions are
7// met:
8//
9// * Redistributions of source code must retain the above copyright
10// notice, this list of conditions and the following disclaimer.
11// * Redistributions in binary form must reproduce the above
12// copyright notice, this list of conditions and the following disclaimer
13// in the documentation and/or other materials provided with the
14// distribution.
15// * Neither the name of Google Inc. nor the names of its
16// contributors may be used to endorse or promote products derived from
17// this software without specific prior written permission.
18//
19// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
22// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
23// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
25// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
26// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
27// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
28// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
29// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
30
31// A StringPiece points to part or all of a string, Cord, double-quoted string
32// literal, or other string-like object. A StringPiece does *not* own the
33// string to which it points. A StringPiece is not null-terminated.
34//
35// You can use StringPiece as a function or method parameter. A StringPiece
36// parameter can receive a double-quoted string literal argument, a "const
37// char*" argument, a string argument, or a StringPiece argument with no data
38// copying. Systematic use of StringPiece for arguments reduces data
39// copies and strlen() calls.
40//
41// Prefer passing StringPieces by value:
42// void MyFunction(StringPiece arg);
43// If circumstances require, you may also pass by const reference:
44// void MyFunction(const StringPiece& arg); // not preferred
45// Both of these have the same lifetime semantics. Passing by value
46// generates slightly smaller code. For more discussion, see the thread
47// go/stringpiecebyvalue on c-users.
48//
49// StringPiece is also suitable for local variables if you know that
50// the lifetime of the underlying object is longer than the lifetime
51// of your StringPiece variable.
52//
53// Beware of binding a StringPiece to a temporary:
54// StringPiece sp = obj.MethodReturningString(); // BAD: lifetime problem
55//
56// This code is okay:
57// string str = obj.MethodReturningString(); // str owns its contents
58// StringPiece sp(str); // GOOD, because str outlives sp
59//
60// StringPiece is sometimes a poor choice for a return value and usually a poor
61// choice for a data member. If you do use a StringPiece this way, it is your
62// responsibility to ensure that the object pointed to by the StringPiece
63// outlives the StringPiece.
64//
65// A StringPiece may represent just part of a string; thus the name "Piece".
66// For example, when splitting a string, vector<StringPiece> is a natural data
67// type for the output. For another example, a Cord is a non-contiguous,
68// potentially very long string-like object. The Cord class has an interface
69// that iteratively provides StringPiece objects that point to the
70// successive pieces of a Cord object.
71//
72// A StringPiece is not null-terminated. If you write code that scans a
73// StringPiece, you must check its length before reading any characters.
74// Common idioms that work on null-terminated strings do not work on
75// StringPiece objects.
76//
77// There are several ways to create a null StringPiece:
78// StringPiece()
79// StringPiece(nullptr)
80// StringPiece(nullptr, 0)
81// For all of the above, sp.data() == nullptr, sp.length() == 0,
82// and sp.empty() == true. Also, if you create a StringPiece with
83// a non-null pointer then sp.data() != nullptr. Once created,
84// sp.data() will stay either nullptr or not-nullptr, except if you call
85// sp.clear() or sp.set().
86//
87// Thus, you can use StringPiece(nullptr) to signal an out-of-band value
88// that is different from other StringPiece values. This is similar
89// to the way that const char* p1 = nullptr; is different from
90// const char* p2 = "";.
91//
92// There are many ways to create an empty StringPiece:
93// StringPiece()
94// StringPiece(nullptr)
95// StringPiece(nullptr, 0)
96// StringPiece("")
97// StringPiece("", 0)
98// StringPiece("abcdef", 0)
99// StringPiece("abcdef"+6, 0)
100// For all of the above, sp.length() will be 0 and sp.empty() will be true.
101// For some empty StringPiece values, sp.data() will be nullptr.
102// For some empty StringPiece values, sp.data() will not be nullptr.
103//
104// Be careful not to confuse: null StringPiece and empty StringPiece.
105// The set of empty StringPieces properly includes the set of null StringPieces.
106// That is, every null StringPiece is an empty StringPiece,
107// but some non-null StringPieces are empty Stringpieces too.
108//
109// All empty StringPiece values compare equal to each other.
110// Even a null StringPieces compares equal to a non-null empty StringPiece:
111// StringPiece() == StringPiece("", 0)
112// StringPiece(nullptr) == StringPiece("abc", 0)
113// StringPiece(nullptr, 0) == StringPiece("abcdef"+6, 0)
114//
115// Look carefully at this example:
116// StringPiece("") == nullptr
117// True or false? TRUE, because StringPiece::operator== converts
118// the right-hand side from nullptr to StringPiece(nullptr),
119// and then compares two zero-length spans of characters.
120// However, we are working to make this example produce a compile error.
121//
122// Suppose you want to write:
123// bool TestWhat?(StringPiece sp) { return sp == nullptr; } // BAD
124// Do not do that. Write one of these instead:
125// bool TestNull(StringPiece sp) { return sp.data() == nullptr; }
126// bool TestEmpty(StringPiece sp) { return sp.empty(); }
127// The intent of TestWhat? is unclear. Did you mean TestNull or TestEmpty?
128// Right now, TestWhat? behaves likes TestEmpty.
129// We are working to make TestWhat? produce a compile error.
130// TestNull is good to test for an out-of-band signal.
131// TestEmpty is good to test for an empty StringPiece.
132//
133// Caveats (again):
134// (1) The lifetime of the pointed-to string (or piece of a string)
135// must be longer than the lifetime of the StringPiece.
136// (2) There may or may not be a '\0' character after the end of
137// StringPiece data.
138// (3) A null StringPiece is empty.
139// An empty StringPiece may or may not be a null StringPiece.
140
141#ifndef GOOGLE_PROTOBUF_STUBS_STRINGPIECE_H_
142#define GOOGLE_PROTOBUF_STUBS_STRINGPIECE_H_
143
144#include <assert.h>
145#include <stddef.h>
146#include <string.h>
147#include <iosfwd>
148#include <limits>
149#include <string>
150
151#include <google/protobuf/stubs/common.h>
152#include <google/protobuf/stubs/hash.h>
153
154#include <google/protobuf/port_def.inc>
155
156namespace google {
157namespace protobuf {
158// StringPiece has *two* size types.
159// StringPiece::size_type
160// is unsigned
161// is 32 bits in LP32, 64 bits in LP64, 64 bits in LLP64
162// no future changes intended
163// stringpiece_ssize_type
164// is signed
165// is 32 bits in LP32, 64 bits in LP64, 64 bits in LLP64
166// future changes intended: http://go/64BitStringPiece
167//
168typedef string::difference_type stringpiece_ssize_type;
169
170// STRINGPIECE_CHECK_SIZE protects us from 32-bit overflows.
171// TODO(mec): delete this after stringpiece_ssize_type goes 64 bit.
172#if !defined(NDEBUG)
173#define STRINGPIECE_CHECK_SIZE 1
174#elif defined(_FORTIFY_SOURCE) && _FORTIFY_SOURCE > 0
175#define STRINGPIECE_CHECK_SIZE 1
176#else
177#define STRINGPIECE_CHECK_SIZE 0
178#endif
179
180class PROTOBUF_EXPORT StringPiece {
181 private:
182 const char* ptr_;
183 stringpiece_ssize_type length_;
184
185 // Prevent overflow in debug mode or fortified mode.
186 // sizeof(stringpiece_ssize_type) may be smaller than sizeof(size_t).
187 static stringpiece_ssize_type CheckedSsizeTFromSizeT(size_t size) {
188#if STRINGPIECE_CHECK_SIZE > 0
189#ifdef max
190#undef max
191#endif
192 if (size > static_cast<size_t>(
193 std::numeric_limits<stringpiece_ssize_type>::max())) {
194 // Some people grep for this message in logs
195 // so take care if you ever change it.
196 LogFatalSizeTooBig(size, "size_t to int conversion");
197 }
198#endif
199 return static_cast<stringpiece_ssize_type>(size);
200 }
201
202 // Out-of-line error path.
203 static void LogFatalSizeTooBig(size_t size, const char* details);
204
205 public:
206 // We provide non-explicit singleton constructors so users can pass
207 // in a "const char*" or a "string" wherever a "StringPiece" is
208 // expected.
209 //
210 // Style guide exception granted:
211 // http://goto/style-guide-exception-20978288
212 StringPiece() : ptr_(nullptr), length_(0) {}
213
214 StringPiece(const char* str) // NOLINT(runtime/explicit)
215 : ptr_(str), length_(0) {
216 if (str != nullptr) {
217 length_ = CheckedSsizeTFromSizeT(strlen(str));
218 }
219 }
220
221 template <class Allocator>
222 StringPiece( // NOLINT(runtime/explicit)
223 const std::basic_string<char, std::char_traits<char>, Allocator>& str)
224 : ptr_(str.data()), length_(0) {
225 length_ = CheckedSsizeTFromSizeT(str.size());
226 }
227
228 StringPiece(const char* offset, stringpiece_ssize_type len)
229 : ptr_(offset), length_(len) {
230 assert(len >= 0);
231 }
232
233 // Substring of another StringPiece.
234 // pos must be non-negative and <= x.length().
235 StringPiece(StringPiece x, stringpiece_ssize_type pos);
236 // Substring of another StringPiece.
237 // pos must be non-negative and <= x.length().
238 // len must be non-negative and will be pinned to at most x.length() - pos.
239 StringPiece(StringPiece x,
240 stringpiece_ssize_type pos,
241 stringpiece_ssize_type len);
242
243 // data() may return a pointer to a buffer with embedded NULs, and the
244 // returned buffer may or may not be null terminated. Therefore it is
245 // typically a mistake to pass data() to a routine that expects a NUL
246 // terminated string.
247 const char* data() const { return ptr_; }
248 stringpiece_ssize_type size() const { return length_; }
249 stringpiece_ssize_type length() const { return length_; }
250 bool empty() const { return length_ == 0; }
251
252 void clear() {
253 ptr_ = nullptr;
254 length_ = 0;
255 }
256
257 void set(const char* data, stringpiece_ssize_type len) {
258 assert(len >= 0);
259 ptr_ = data;
260 length_ = len;
261 }
262
263 void set(const char* str) {
264 ptr_ = str;
265 if (str != nullptr)
266 length_ = CheckedSsizeTFromSizeT(strlen(str));
267 else
268 length_ = 0;
269 }
270
271 void set(const void* data, stringpiece_ssize_type len) {
272 ptr_ = reinterpret_cast<const char*>(data);
273 length_ = len;
274 }
275
276 char operator[](stringpiece_ssize_type i) const {
277 assert(0 <= i);
278 assert(i < length_);
279 return ptr_[i];
280 }
281
282 void remove_prefix(stringpiece_ssize_type n) {
283 assert(length_ >= n);
284 ptr_ += n;
285 length_ -= n;
286 }
287
288 void remove_suffix(stringpiece_ssize_type n) {
289 assert(length_ >= n);
290 length_ -= n;
291 }
292
293 // returns {-1, 0, 1}
294 int compare(StringPiece x) const {
295 const stringpiece_ssize_type min_size =
296 length_ < x.length_ ? length_ : x.length_;
297 int r = memcmp(ptr_, x.ptr_, static_cast<size_t>(min_size));
298 if (r < 0) return -1;
299 if (r > 0) return 1;
300 if (length_ < x.length_) return -1;
301 if (length_ > x.length_) return 1;
302 return 0;
303 }
304
305 string as_string() const {
306 return ToString();
307 }
308 // We also define ToString() here, since many other string-like
309 // interfaces name the routine that converts to a C++ string
310 // "ToString", and it's confusing to have the method that does that
311 // for a StringPiece be called "as_string()". We also leave the
312 // "as_string()" method defined here for existing code.
313 string ToString() const {
314 if (ptr_ == nullptr) return string();
315 return string(data(), static_cast<size_type>(size()));
316 }
317
318 operator string() const {
319 return ToString();
320 }
321
322 void CopyToString(string* target) const;
323 void AppendToString(string* target) const;
324
325 bool starts_with(StringPiece x) const {
326 return (length_ >= x.length_) &&
327 (memcmp(ptr_, x.ptr_, static_cast<size_t>(x.length_)) == 0);
328 }
329
330 bool ends_with(StringPiece x) const {
331 return ((length_ >= x.length_) &&
332 (memcmp(ptr_ + (length_-x.length_), x.ptr_,
333 static_cast<size_t>(x.length_)) == 0));
334 }
335
336 // Checks whether StringPiece starts with x and if so advances the beginning
337 // of it to past the match. It's basically a shortcut for starts_with
338 // followed by remove_prefix.
339 bool Consume(StringPiece x);
340 // Like above but for the end of the string.
341 bool ConsumeFromEnd(StringPiece x);
342
343 // standard STL container boilerplate
344 typedef char value_type;
345 typedef const char* pointer;
346 typedef const char& reference;
347 typedef const char& const_reference;
348 typedef size_t size_type;
349 typedef ptrdiff_t difference_type;
350 static const size_type npos;
351 typedef const char* const_iterator;
352 typedef const char* iterator;
353 typedef std::reverse_iterator<const_iterator> const_reverse_iterator;
354 typedef std::reverse_iterator<iterator> reverse_iterator;
355 iterator begin() const { return ptr_; }
356 iterator end() const { return ptr_ + length_; }
357 const_reverse_iterator rbegin() const {
358 return const_reverse_iterator(ptr_ + length_);
359 }
360 const_reverse_iterator rend() const {
361 return const_reverse_iterator(ptr_);
362 }
363 stringpiece_ssize_type max_size() const { return length_; }
364 stringpiece_ssize_type capacity() const { return length_; }
365
366 // cpplint.py emits a false positive [build/include_what_you_use]
367 stringpiece_ssize_type copy(char* buf, size_type n, size_type pos = 0) const; // NOLINT
368
369 bool contains(StringPiece s) const;
370
371 stringpiece_ssize_type find(StringPiece s, size_type pos = 0) const;
372 stringpiece_ssize_type find(char c, size_type pos = 0) const;
373 stringpiece_ssize_type rfind(StringPiece s, size_type pos = npos) const;
374 stringpiece_ssize_type rfind(char c, size_type pos = npos) const;
375
376 stringpiece_ssize_type find_first_of(StringPiece s, size_type pos = 0) const;
377 stringpiece_ssize_type find_first_of(char c, size_type pos = 0) const {
378 return find(c, pos);
379 }
380 stringpiece_ssize_type find_first_not_of(StringPiece s,
381 size_type pos = 0) const;
382 stringpiece_ssize_type find_first_not_of(char c, size_type pos = 0) const;
383 stringpiece_ssize_type find_last_of(StringPiece s,
384 size_type pos = npos) const;
385 stringpiece_ssize_type find_last_of(char c, size_type pos = npos) const {
386 return rfind(c, pos);
387 }
388 stringpiece_ssize_type find_last_not_of(StringPiece s,
389 size_type pos = npos) const;
390 stringpiece_ssize_type find_last_not_of(char c, size_type pos = npos) const;
391
392 StringPiece substr(size_type pos, size_type n = npos) const;
393};
394
395// This large function is defined inline so that in a fairly common case where
396// one of the arguments is a literal, the compiler can elide a lot of the
397// following comparisons.
398inline bool operator==(StringPiece x, StringPiece y) {
399 stringpiece_ssize_type len = x.size();
400 if (len != y.size()) {
401 return false;
402 }
403
404 return x.data() == y.data() || len <= 0 ||
405 memcmp(x.data(), y.data(), static_cast<size_t>(len)) == 0;
406}
407
408inline bool operator!=(StringPiece x, StringPiece y) {
409 return !(x == y);
410}
411
412inline bool operator<(StringPiece x, StringPiece y) {
413 const stringpiece_ssize_type min_size =
414 x.size() < y.size() ? x.size() : y.size();
415 const int r = memcmp(x.data(), y.data(), static_cast<size_t>(min_size));
416 return (r < 0) || (r == 0 && x.size() < y.size());
417}
418
419inline bool operator>(StringPiece x, StringPiece y) {
420 return y < x;
421}
422
423inline bool operator<=(StringPiece x, StringPiece y) {
424 return !(x > y);
425}
426
427inline bool operator>=(StringPiece x, StringPiece y) {
428 return !(x < y);
429}
430
431// allow StringPiece to be logged
432extern std::ostream& operator<<(std::ostream& o, StringPiece piece);
433
434namespace internal {
435// StringPiece is not a POD and can not be used in an union (pre C++11). We
436// need a POD version of it.
437struct StringPiecePod {
438 // Create from a StringPiece.
439 static StringPiecePod CreateFromStringPiece(StringPiece str) {
440 StringPiecePod pod;
441 pod.data_ = str.data();
442 pod.size_ = str.size();
443 return pod;
444 }
445
446 // Cast to StringPiece.
447 operator StringPiece() const { return StringPiece(data_, size_); }
448
449 bool operator==(const char* value) const {
450 return StringPiece(data_, size_) == StringPiece(value);
451 }
452
453 char operator[](stringpiece_ssize_type i) const {
454 assert(0 <= i);
455 assert(i < size_);
456 return data_[i];
457 }
458
459 const char* data() const { return data_; }
460
461 stringpiece_ssize_type size() const {
462 return size_;
463 }
464
465 std::string ToString() const {
466 return std::string(data_, static_cast<size_t>(size_));
467 }
468
469 operator string() const { return ToString(); }
470
471 private:
472 const char* data_;
473 stringpiece_ssize_type size_;
474};
475
476} // namespace internal
477} // namespace protobuf
478} // namespace google
479
480GOOGLE_PROTOBUF_HASH_NAMESPACE_DECLARATION_START
481template<> struct hash<StringPiece> {
482 size_t operator()(const StringPiece& s) const {
483 size_t result = 0;
484 for (const char *str = s.data(), *end = str + s.size(); str < end; str++) {
485 result = 5 * result + static_cast<size_t>(*str);
486 }
487 return result;
488 }
489};
490GOOGLE_PROTOBUF_HASH_NAMESPACE_DECLARATION_END
491
492#include <google/protobuf/port_undef.inc>
493
494#endif // STRINGS_STRINGPIECE_H_
495