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/hash.h>
152
153#include <google/protobuf/port_def.inc>
154
155namespace google {
156namespace protobuf {
157// StringPiece has *two* size types.
158// StringPiece::size_type
159// is unsigned
160// is 32 bits in LP32, 64 bits in LP64, 64 bits in LLP64
161// no future changes intended
162// stringpiece_ssize_type
163// is signed
164// is 32 bits in LP32, 64 bits in LP64, 64 bits in LLP64
165// future changes intended: http://go/64BitStringPiece
166//
167typedef std::string::difference_type stringpiece_ssize_type;
168
169// STRINGPIECE_CHECK_SIZE protects us from 32-bit overflows.
170// TODO(mec): delete this after stringpiece_ssize_type goes 64 bit.
171#if !defined(NDEBUG)
172#define STRINGPIECE_CHECK_SIZE 1
173#elif defined(_FORTIFY_SOURCE) && _FORTIFY_SOURCE > 0
174#define STRINGPIECE_CHECK_SIZE 1
175#else
176#define STRINGPIECE_CHECK_SIZE 0
177#endif
178
179class PROTOBUF_EXPORT StringPiece {
180 private:
181 const char* ptr_;
182 stringpiece_ssize_type length_;
183
184 // Prevent overflow in debug mode or fortified mode.
185 // sizeof(stringpiece_ssize_type) may be smaller than sizeof(size_t).
186 static stringpiece_ssize_type CheckedSsizeTFromSizeT(size_t size) {
187#if STRINGPIECE_CHECK_SIZE > 0
188#ifdef max
189#undef max
190#endif
191 if (size > static_cast<size_t>(
192 std::numeric_limits<stringpiece_ssize_type>::max())) {
193 // Some people grep for this message in logs
194 // so take care if you ever change it.
195 LogFatalSizeTooBig(size, "size_t to int conversion");
196 }
197#endif
198 return static_cast<stringpiece_ssize_type>(size);
199 }
200
201 // Out-of-line error path.
202 static void LogFatalSizeTooBig(size_t size, const char* details);
203
204 public:
205 // We provide non-explicit singleton constructors so users can pass
206 // in a "const char*" or a "string" wherever a "StringPiece" is
207 // expected.
208 //
209 // Style guide exception granted:
210 // http://goto/style-guide-exception-20978288
211 StringPiece() : ptr_(nullptr), length_(0) {}
212
213 StringPiece(const char* str) // NOLINT(runtime/explicit)
214 : ptr_(str), length_(0) {
215 if (str != nullptr) {
216 length_ = CheckedSsizeTFromSizeT(strlen(str));
217 }
218 }
219
220 template <class Allocator>
221 StringPiece( // NOLINT(runtime/explicit)
222 const std::basic_string<char, std::char_traits<char>, Allocator>& str)
223 : ptr_(str.data()), length_(0) {
224 length_ = CheckedSsizeTFromSizeT(str.size());
225 }
226
227 StringPiece(const char* offset, stringpiece_ssize_type len)
228 : ptr_(offset), length_(len) {
229 assert(len >= 0);
230 }
231
232 // Substring of another StringPiece.
233 // pos must be non-negative and <= x.length().
234 StringPiece(StringPiece x, stringpiece_ssize_type pos);
235 // Substring of another StringPiece.
236 // pos must be non-negative and <= x.length().
237 // len must be non-negative and will be pinned to at most x.length() - pos.
238 StringPiece(StringPiece x,
239 stringpiece_ssize_type pos,
240 stringpiece_ssize_type len);
241
242 // data() may return a pointer to a buffer with embedded NULs, and the
243 // returned buffer may or may not be null terminated. Therefore it is
244 // typically a mistake to pass data() to a routine that expects a NUL
245 // terminated string.
246 const char* data() const { return ptr_; }
247 stringpiece_ssize_type size() const { return length_; }
248 stringpiece_ssize_type length() const { return length_; }
249 bool empty() const { return length_ == 0; }
250
251 void clear() {
252 ptr_ = nullptr;
253 length_ = 0;
254 }
255
256 void set(const char* data, stringpiece_ssize_type len) {
257 assert(len >= 0);
258 ptr_ = data;
259 length_ = len;
260 }
261
262 void set(const char* str) {
263 ptr_ = str;
264 if (str != nullptr)
265 length_ = CheckedSsizeTFromSizeT(strlen(str));
266 else
267 length_ = 0;
268 }
269
270 void set(const void* data, stringpiece_ssize_type len) {
271 ptr_ = reinterpret_cast<const char*>(data);
272 length_ = len;
273 }
274
275 char operator[](stringpiece_ssize_type i) const {
276 assert(0 <= i);
277 assert(i < length_);
278 return ptr_[i];
279 }
280
281 void remove_prefix(stringpiece_ssize_type n) {
282 assert(length_ >= n);
283 ptr_ += n;
284 length_ -= n;
285 }
286
287 void remove_suffix(stringpiece_ssize_type n) {
288 assert(length_ >= n);
289 length_ -= n;
290 }
291
292 // returns {-1, 0, 1}
293 int compare(StringPiece x) const {
294 const stringpiece_ssize_type min_size =
295 length_ < x.length_ ? length_ : x.length_;
296 int r = memcmp(ptr_, x.ptr_, static_cast<size_t>(min_size));
297 if (r < 0) return -1;
298 if (r > 0) return 1;
299 if (length_ < x.length_) return -1;
300 if (length_ > x.length_) return 1;
301 return 0;
302 }
303
304 std::string as_string() const { return ToString(); }
305 // We also define ToString() here, since many other string-like
306 // interfaces name the routine that converts to a C++ string
307 // "ToString", and it's confusing to have the method that does that
308 // for a StringPiece be called "as_string()". We also leave the
309 // "as_string()" method defined here for existing code.
310 std::string ToString() const {
311 if (ptr_ == nullptr) return "";
312 return std::string(data(), static_cast<size_type>(size()));
313 }
314
315 explicit operator std::string() const { return ToString(); }
316
317 void CopyToString(std::string* target) const;
318 void AppendToString(std::string* target) const;
319
320 bool starts_with(StringPiece x) const {
321 return (length_ >= x.length_) &&
322 (memcmp(ptr_, x.ptr_, static_cast<size_t>(x.length_)) == 0);
323 }
324
325 bool ends_with(StringPiece x) const {
326 return ((length_ >= x.length_) &&
327 (memcmp(ptr_ + (length_-x.length_), x.ptr_,
328 static_cast<size_t>(x.length_)) == 0));
329 }
330
331 // Checks whether StringPiece starts with x and if so advances the beginning
332 // of it to past the match. It's basically a shortcut for starts_with
333 // followed by remove_prefix.
334 bool Consume(StringPiece x);
335 // Like above but for the end of the string.
336 bool ConsumeFromEnd(StringPiece x);
337
338 // standard STL container boilerplate
339 typedef char value_type;
340 typedef const char* pointer;
341 typedef const char& reference;
342 typedef const char& const_reference;
343 typedef size_t size_type;
344 typedef ptrdiff_t difference_type;
345 static const size_type npos;
346 typedef const char* const_iterator;
347 typedef const char* iterator;
348 typedef std::reverse_iterator<const_iterator> const_reverse_iterator;
349 typedef std::reverse_iterator<iterator> reverse_iterator;
350 iterator begin() const { return ptr_; }
351 iterator end() const { return ptr_ + length_; }
352 const_reverse_iterator rbegin() const {
353 return const_reverse_iterator(ptr_ + length_);
354 }
355 const_reverse_iterator rend() const {
356 return const_reverse_iterator(ptr_);
357 }
358 stringpiece_ssize_type max_size() const { return length_; }
359 stringpiece_ssize_type capacity() const { return length_; }
360
361 // cpplint.py emits a false positive [build/include_what_you_use]
362 stringpiece_ssize_type copy(char* buf, size_type n, size_type pos = 0) const; // NOLINT
363
364 bool contains(StringPiece s) const;
365
366 stringpiece_ssize_type find(StringPiece s, size_type pos = 0) const;
367 stringpiece_ssize_type find(char c, size_type pos = 0) const;
368 stringpiece_ssize_type rfind(StringPiece s, size_type pos = npos) const;
369 stringpiece_ssize_type rfind(char c, size_type pos = npos) const;
370
371 stringpiece_ssize_type find_first_of(StringPiece s, size_type pos = 0) const;
372 stringpiece_ssize_type find_first_of(char c, size_type pos = 0) const {
373 return find(c, pos);
374 }
375 stringpiece_ssize_type find_first_not_of(StringPiece s,
376 size_type pos = 0) const;
377 stringpiece_ssize_type find_first_not_of(char c, size_type pos = 0) const;
378 stringpiece_ssize_type find_last_of(StringPiece s,
379 size_type pos = npos) const;
380 stringpiece_ssize_type find_last_of(char c, size_type pos = npos) const {
381 return rfind(c, pos);
382 }
383 stringpiece_ssize_type find_last_not_of(StringPiece s,
384 size_type pos = npos) const;
385 stringpiece_ssize_type find_last_not_of(char c, size_type pos = npos) const;
386
387 StringPiece substr(size_type pos, size_type n = npos) const;
388};
389
390// This large function is defined inline so that in a fairly common case where
391// one of the arguments is a literal, the compiler can elide a lot of the
392// following comparisons.
393inline bool operator==(StringPiece x, StringPiece y) {
394 stringpiece_ssize_type len = x.size();
395 if (len != y.size()) {
396 return false;
397 }
398
399 return x.data() == y.data() || len <= 0 ||
400 memcmp(x.data(), y.data(), static_cast<size_t>(len)) == 0;
401}
402
403inline bool operator!=(StringPiece x, StringPiece y) {
404 return !(x == y);
405}
406
407inline bool operator<(StringPiece x, StringPiece y) {
408 const stringpiece_ssize_type min_size =
409 x.size() < y.size() ? x.size() : y.size();
410 const int r = memcmp(x.data(), y.data(), static_cast<size_t>(min_size));
411 return (r < 0) || (r == 0 && x.size() < y.size());
412}
413
414inline bool operator>(StringPiece x, StringPiece y) {
415 return y < x;
416}
417
418inline bool operator<=(StringPiece x, StringPiece y) {
419 return !(x > y);
420}
421
422inline bool operator>=(StringPiece x, StringPiece y) {
423 return !(x < y);
424}
425
426// allow StringPiece to be logged
427extern std::ostream& operator<<(std::ostream& o, StringPiece piece);
428
429namespace internal {
430// StringPiece is not a POD and can not be used in an union (pre C++11). We
431// need a POD version of it.
432struct StringPiecePod {
433 // Create from a StringPiece.
434 static StringPiecePod CreateFromStringPiece(StringPiece str) {
435 StringPiecePod pod;
436 pod.data_ = str.data();
437 pod.size_ = str.size();
438 return pod;
439 }
440
441 // Cast to StringPiece.
442 operator StringPiece() const { return StringPiece(data_, size_); }
443
444 bool operator==(const char* value) const {
445 return StringPiece(data_, size_) == StringPiece(value);
446 }
447
448 char operator[](stringpiece_ssize_type i) const {
449 assert(0 <= i);
450 assert(i < size_);
451 return data_[i];
452 }
453
454 const char* data() const { return data_; }
455
456 stringpiece_ssize_type size() const {
457 return size_;
458 }
459
460 std::string ToString() const {
461 return std::string(data_, static_cast<size_t>(size_));
462 }
463
464 explicit operator std::string() const { return ToString(); }
465
466 private:
467 const char* data_;
468 stringpiece_ssize_type size_;
469};
470
471} // namespace internal
472} // namespace protobuf
473} // namespace google
474
475GOOGLE_PROTOBUF_HASH_NAMESPACE_DECLARATION_START
476template<> struct hash<StringPiece> {
477 size_t operator()(const StringPiece& s) const {
478 size_t result = 0;
479 for (const char *str = s.data(), *end = str + s.size(); str < end; str++) {
480 result = 5 * result + static_cast<size_t>(*str);
481 }
482 return result;
483 }
484};
485GOOGLE_PROTOBUF_HASH_NAMESPACE_DECLARATION_END
486
487#include <google/protobuf/port_undef.inc>
488
489#endif // STRINGS_STRINGPIECE_H_
490