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(NULL)
80 //   StringPiece(NULL, 0)
81 // For all of the above, sp.data() == NULL, sp.length() == 0,
82 // and sp.empty() == true.  Also, if you create a StringPiece with
83 // a non-NULL pointer then sp.data() != NULL.  Once created,
84 // sp.data() will stay either NULL or not-NULL, except if you call
85 // sp.clear() or sp.set().
86 //
87 // Thus, you can use StringPiece(NULL) 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 = NULL; is different from
90 // const char* p2 = "";.
91 //
92 // There are many ways to create an empty StringPiece:
93 //   StringPiece()
94 //   StringPiece(NULL)
95 //   StringPiece(NULL, 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 NULL.
102 // For some empty StringPiece values, sp.data() will not be NULL.
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(NULL) == StringPiece("abc", 0)
113 //  StringPiece(NULL, 0) == StringPiece("abcdef"+6, 0)
114 //
115 // Look carefully at this example:
116 //   StringPiece("") == NULL
117 // True or false?  TRUE, because StringPiece::operator== converts
118 // the right-hand side from NULL to StringPiece(NULL),
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 == NULL; }  // BAD
124 // Do not do that.  Write one of these instead:
125 //   bool TestNull(StringPiece sp) { return sp.data() == NULL; }
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 namespace google {
155 namespace protobuf {
156 // StringPiece has *two* size types.
157 // StringPiece::size_type
158 //   is unsigned
159 //   is 32 bits in LP32, 64 bits in LP64, 64 bits in LLP64
160 //   no future changes intended
161 // stringpiece_ssize_type
162 //   is signed
163 //   is 32 bits in LP32, 64 bits in LP64, 64 bits in LLP64
164 //   future changes intended: http://go/64BitStringPiece
165 //
166 typedef string::difference_type stringpiece_ssize_type;
167 
168 // STRINGPIECE_CHECK_SIZE protects us from 32-bit overflows.
169 // TODO(mec): delete this after stringpiece_ssize_type goes 64 bit.
170 #if !defined(NDEBUG)
171 #define STRINGPIECE_CHECK_SIZE 1
172 #elif defined(_FORTIFY_SOURCE) && _FORTIFY_SOURCE > 0
173 #define STRINGPIECE_CHECK_SIZE 1
174 #else
175 #define STRINGPIECE_CHECK_SIZE 0
176 #endif
177 
178 class LIBPROTOBUF_EXPORT StringPiece {
179  private:
180   const char* ptr_;
181   stringpiece_ssize_type length_;
182 
183   // Prevent overflow in debug mode or fortified mode.
184   // sizeof(stringpiece_ssize_type) may be smaller than sizeof(size_t).
CheckedSsizeTFromSizeT(size_t size)185   static stringpiece_ssize_type CheckedSsizeTFromSizeT(size_t size) {
186 #if STRINGPIECE_CHECK_SIZE > 0
187 #ifdef max
188 #undef max
189 #endif
190     if (size > static_cast<size_t>(
191         std::numeric_limits<stringpiece_ssize_type>::max())) {
192       // Some people grep for this message in logs
193       // so take care if you ever change it.
194       LogFatalSizeTooBig(size, "size_t to int conversion");
195     }
196 #endif
197     return static_cast<stringpiece_ssize_type>(size);
198   }
199 
200   // Out-of-line error path.
201   static void LogFatalSizeTooBig(size_t size, const char* details);
202 
203  public:
204   // We provide non-explicit singleton constructors so users can pass
205   // in a "const char*" or a "string" wherever a "StringPiece" is
206   // expected.
207   //
208   // Style guide exception granted:
209   // http://goto/style-guide-exception-20978288
StringPiece()210   StringPiece() : ptr_(NULL), length_(0) {}
211 
StringPiece(const char * str)212   StringPiece(const char* str)  // NOLINT(runtime/explicit)
213       : ptr_(str), length_(0) {
214     if (str != NULL) {
215       length_ = CheckedSsizeTFromSizeT(strlen(str));
216     }
217   }
218 
219   template <class Allocator>
StringPiece(const std::basic_string<char,std::char_traits<char>,Allocator> & str)220   StringPiece(  // NOLINT(runtime/explicit)
221       const std::basic_string<char, std::char_traits<char>, Allocator>& str)
222       : ptr_(str.data()), length_(0) {
223     length_ = CheckedSsizeTFromSizeT(str.size());
224   }
225 #if defined(HAS_GLOBAL_STRING)
226   template <class Allocator>
StringPiece(const basic_string<char,std::char_traits<char>,Allocator> & str)227   StringPiece(  // NOLINT(runtime/explicit)
228       const basic_string<char, std::char_traits<char>, Allocator>& str)
229       : ptr_(str.data()), length_(0) {
230     length_ = CheckedSsizeTFromSizeT(str.size());
231   }
232 #endif
233 
StringPiece(const char * offset,stringpiece_ssize_type len)234   StringPiece(const char* offset, stringpiece_ssize_type len)
235       : ptr_(offset), length_(len) {
236     assert(len >= 0);
237   }
238 
239   // Substring of another StringPiece.
240   // pos must be non-negative and <= x.length().
241   StringPiece(StringPiece x, stringpiece_ssize_type pos);
242   // Substring of another StringPiece.
243   // pos must be non-negative and <= x.length().
244   // len must be non-negative and will be pinned to at most x.length() - pos.
245   StringPiece(StringPiece x,
246               stringpiece_ssize_type pos,
247               stringpiece_ssize_type len);
248 
249   // data() may return a pointer to a buffer with embedded NULs, and the
250   // returned buffer may or may not be null terminated.  Therefore it is
251   // typically a mistake to pass data() to a routine that expects a NUL
252   // terminated string.
data()253   const char* data() const { return ptr_; }
size()254   stringpiece_ssize_type size() const { return length_; }
length()255   stringpiece_ssize_type length() const { return length_; }
empty()256   bool empty() const { return length_ == 0; }
257 
clear()258   void clear() {
259     ptr_ = NULL;
260     length_ = 0;
261   }
262 
set(const char * data,stringpiece_ssize_type len)263   void set(const char* data, stringpiece_ssize_type len) {
264     assert(len >= 0);
265     ptr_ = data;
266     length_ = len;
267   }
268 
set(const char * str)269   void set(const char* str) {
270     ptr_ = str;
271     if (str != NULL)
272       length_ = CheckedSsizeTFromSizeT(strlen(str));
273     else
274       length_ = 0;
275   }
276 
set(const void * data,stringpiece_ssize_type len)277   void set(const void* data, stringpiece_ssize_type len) {
278     ptr_ = reinterpret_cast<const char*>(data);
279     length_ = len;
280   }
281 
282   char operator[](stringpiece_ssize_type i) const {
283     assert(0 <= i);
284     assert(i < length_);
285     return ptr_[i];
286   }
287 
remove_prefix(stringpiece_ssize_type n)288   void remove_prefix(stringpiece_ssize_type n) {
289     assert(length_ >= n);
290     ptr_ += n;
291     length_ -= n;
292   }
293 
remove_suffix(stringpiece_ssize_type n)294   void remove_suffix(stringpiece_ssize_type n) {
295     assert(length_ >= n);
296     length_ -= n;
297   }
298 
299   // returns {-1, 0, 1}
compare(StringPiece x)300   int compare(StringPiece x) const {
301     const stringpiece_ssize_type min_size =
302         length_ < x.length_ ? length_ : x.length_;
303     int r = memcmp(ptr_, x.ptr_, min_size);
304     if (r < 0) return -1;
305     if (r > 0) return 1;
306     if (length_ < x.length_) return -1;
307     if (length_ > x.length_) return 1;
308     return 0;
309   }
310 
as_string()311   string as_string() const {
312     return ToString();
313   }
314   // We also define ToString() here, since many other string-like
315   // interfaces name the routine that converts to a C++ string
316   // "ToString", and it's confusing to have the method that does that
317   // for a StringPiece be called "as_string()".  We also leave the
318   // "as_string()" method defined here for existing code.
ToString()319   string ToString() const {
320     if (ptr_ == NULL) return string();
321     return string(data(), size());
322   }
323 
string()324   operator string() const {
325     return ToString();
326   }
327 
328   void CopyToString(string* target) const;
329   void AppendToString(string* target) const;
330 
starts_with(StringPiece x)331   bool starts_with(StringPiece x) const {
332     return (length_ >= x.length_) && (memcmp(ptr_, x.ptr_, x.length_) == 0);
333   }
334 
ends_with(StringPiece x)335   bool ends_with(StringPiece x) const {
336     return ((length_ >= x.length_) &&
337             (memcmp(ptr_ + (length_-x.length_), x.ptr_, x.length_) == 0));
338   }
339 
340   // Checks whether StringPiece starts with x and if so advances the beginning
341   // of it to past the match.  It's basically a shortcut for starts_with
342   // followed by remove_prefix.
343   bool Consume(StringPiece x);
344   // Like above but for the end of the string.
345   bool ConsumeFromEnd(StringPiece x);
346 
347   // standard STL container boilerplate
348   typedef char value_type;
349   typedef const char* pointer;
350   typedef const char& reference;
351   typedef const char& const_reference;
352   typedef size_t size_type;
353   typedef ptrdiff_t difference_type;
354   static const size_type npos;
355   typedef const char* const_iterator;
356   typedef const char* iterator;
357   typedef std::reverse_iterator<const_iterator> const_reverse_iterator;
358   typedef std::reverse_iterator<iterator> reverse_iterator;
begin()359   iterator begin() const { return ptr_; }
end()360   iterator end() const { return ptr_ + length_; }
rbegin()361   const_reverse_iterator rbegin() const {
362     return const_reverse_iterator(ptr_ + length_);
363   }
rend()364   const_reverse_iterator rend() const {
365     return const_reverse_iterator(ptr_);
366   }
max_size()367   stringpiece_ssize_type max_size() const { return length_; }
capacity()368   stringpiece_ssize_type capacity() const { return length_; }
369 
370   // cpplint.py emits a false positive [build/include_what_you_use]
371   stringpiece_ssize_type copy(char* buf, size_type n, size_type pos = 0) const;  // NOLINT
372 
373   bool contains(StringPiece s) const;
374 
375   stringpiece_ssize_type find(StringPiece s, size_type pos = 0) const;
376   stringpiece_ssize_type find(char c, size_type pos = 0) const;
377   stringpiece_ssize_type rfind(StringPiece s, size_type pos = npos) const;
378   stringpiece_ssize_type rfind(char c, size_type pos = npos) const;
379 
380   stringpiece_ssize_type find_first_of(StringPiece s, size_type pos = 0) const;
381   stringpiece_ssize_type find_first_of(char c, size_type pos = 0) const {
382     return find(c, pos);
383   }
384   stringpiece_ssize_type find_first_not_of(StringPiece s,
385                                            size_type pos = 0) const;
386   stringpiece_ssize_type find_first_not_of(char c, size_type pos = 0) const;
387   stringpiece_ssize_type find_last_of(StringPiece s,
388                                       size_type pos = npos) const;
389   stringpiece_ssize_type find_last_of(char c, size_type pos = npos) const {
390     return rfind(c, pos);
391   }
392   stringpiece_ssize_type find_last_not_of(StringPiece s,
393                                           size_type pos = npos) const;
394   stringpiece_ssize_type find_last_not_of(char c, size_type pos = npos) const;
395 
396   StringPiece substr(size_type pos, size_type n = npos) const;
397 };
398 
399 // This large function is defined inline so that in a fairly common case where
400 // one of the arguments is a literal, the compiler can elide a lot of the
401 // following comparisons.
402 inline bool operator==(StringPiece x, StringPiece y) {
403   stringpiece_ssize_type len = x.size();
404   if (len != y.size()) {
405     return false;
406   }
407 
408   return x.data() == y.data() || len <= 0 ||
409       memcmp(x.data(), y.data(), len) == 0;
410 }
411 
412 inline bool operator!=(StringPiece x, StringPiece y) {
413   return !(x == y);
414 }
415 
416 inline bool operator<(StringPiece x, StringPiece y) {
417   const stringpiece_ssize_type min_size =
418       x.size() < y.size() ? x.size() : y.size();
419   const int r = memcmp(x.data(), y.data(), min_size);
420   return (r < 0) || (r == 0 && x.size() < y.size());
421 }
422 
423 inline bool operator>(StringPiece x, StringPiece y) {
424   return y < x;
425 }
426 
427 inline bool operator<=(StringPiece x, StringPiece y) {
428   return !(x > y);
429 }
430 
431 inline bool operator>=(StringPiece x, StringPiece y) {
432   return !(x < y);
433 }
434 
435 // allow StringPiece to be logged
436 extern std::ostream& operator<<(std::ostream& o, StringPiece piece);
437 
438 namespace internal {
439 // StringPiece is not a POD and can not be used in an union (pre C++11). We
440 // need a POD version of it.
441 struct StringPiecePod {
442   // Create from a StringPiece.
CreateFromStringPieceStringPiecePod443   static StringPiecePod CreateFromStringPiece(StringPiece str) {
444     StringPiecePod pod;
445     pod.data_ = str.data();
446     pod.size_ = str.size();
447     return pod;
448   }
449 
450   // Cast to StringPiece.
StringPieceStringPiecePod451   operator StringPiece() const { return StringPiece(data_, size_); }
452 
453   bool operator==(const char* value) const {
454     return StringPiece(data_, size_) == StringPiece(value);
455   }
456 
457   char operator[](stringpiece_ssize_type i) const {
458     assert(0 <= i);
459     assert(i < size_);
460     return data_[i];
461   }
462 
dataStringPiecePod463   const char* data() const { return data_; }
464 
sizeStringPiecePod465   stringpiece_ssize_type size() const {
466     return size_;
467   }
468 
ToStringStringPiecePod469   std::string ToString() const { return std::string(data_, size_); }
470  private:
471   const char* data_;
472   stringpiece_ssize_type size_;
473 };
474 
475 }  // namespace internal
476 }  // namespace protobuf
477 }  // namespace google
478 
479 GOOGLE_PROTOBUF_HASH_NAMESPACE_DECLARATION_START
480 template<> struct hash<StringPiece> {
481   size_t operator()(const StringPiece& s) const {
482     size_t result = 0;
483     for (const char *str = s.data(), *end = str + s.size(); str < end; str++) {
484       result = 5 * result + *str;
485     }
486     return result;
487   }
488 };
489 GOOGLE_PROTOBUF_HASH_NAMESPACE_DECLARATION_END
490 
491 #endif  // STRINGS_STRINGPIECE_H_
492