Coverage Report

Created: 2025-06-24 06:43

/src/icu/source/common/unicode/bytestream.h
Line
Count
Source (jump to first uncovered line)
1
// © 2016 and later: Unicode, Inc. and others.
2
// License & terms of use: http://www.unicode.org/copyright.html
3
// Copyright (C) 2009-2012, International Business Machines
4
// Corporation and others. All Rights Reserved.
5
//
6
// Copyright 2007 Google Inc. All Rights Reserved.
7
// Author: sanjay@google.com (Sanjay Ghemawat)
8
//
9
// Abstract interface that consumes a sequence of bytes (ByteSink).
10
//
11
// Used so that we can write a single piece of code that can operate
12
// on a variety of output string types.
13
//
14
// Various implementations of this interface are provided:
15
//   ByteSink:
16
//      CheckedArrayByteSink    Write to a flat array, with bounds checking
17
//      StringByteSink          Write to an STL string
18
19
// This code is a contribution of Google code, and the style used here is
20
// a compromise between the original Google code and the ICU coding guidelines.
21
// For example, data types are ICU-ified (size_t,int->int32_t),
22
// and API comments doxygen-ified, but function names and behavior are
23
// as in the original, if possible.
24
// Assertion-style error handling, not available in ICU, was changed to
25
// parameter "pinning" similar to UnicodeString.
26
//
27
// In addition, this is only a partial port of the original Google code,
28
// limited to what was needed so far. The (nearly) complete original code
29
// is in the ICU svn repository at icuhtml/trunk/design/strings/contrib
30
// (see ICU ticket 6765, r25517).
31
32
#ifndef __BYTESTREAM_H__
33
#define __BYTESTREAM_H__
34
35
/**
36
 * \file
37
 * \brief C++ API: Interface for writing bytes, and implementation classes.
38
 */
39
40
#include "unicode/utypes.h"
41
42
#if U_SHOW_CPLUSPLUS_API
43
44
#include "unicode/uobject.h"
45
#include "unicode/std_string.h"
46
47
U_NAMESPACE_BEGIN
48
49
/**
50
 * A ByteSink can be filled with bytes.
51
 * @stable ICU 4.2
52
 */
53
class U_COMMON_API ByteSink : public UMemory {
54
public:
55
  /**
56
   * Default constructor.
57
   * @stable ICU 4.2
58
   */
59
0
  ByteSink() { }
60
  /**
61
   * Virtual destructor.
62
   * @stable ICU 4.2
63
   */
64
  virtual ~ByteSink();
65
66
  /**
67
   * Append "bytes[0,n-1]" to this.
68
   * @param bytes the pointer to the bytes
69
   * @param n the number of bytes; must be non-negative
70
   * @stable ICU 4.2
71
   */
72
  virtual void Append(const char* bytes, int32_t n) = 0;
73
74
  /**
75
   * Appends n bytes to this. Same as Append().
76
   * Call AppendU8() with u8"string literals" which are const char * in C++11
77
   * but const char8_t * in C++20.
78
   * If the compiler does support char8_t as a distinct type,
79
   * then an AppendU8() overload for that is defined and will be chosen.
80
   *
81
   * @param bytes the pointer to the bytes
82
   * @param n the number of bytes; must be non-negative
83
   * @stable ICU 67
84
   */
85
0
  inline void AppendU8(const char* bytes, int32_t n) {
86
0
    Append(bytes, n);
87
0
  }
88
89
#if defined(__cpp_char8_t) || defined(U_IN_DOXYGEN)
90
  /**
91
   * Appends n bytes to this. Same as Append() but for a const char8_t * pointer.
92
   * Call AppendU8() with u8"string literals" which are const char * in C++11
93
   * but const char8_t * in C++20.
94
   * If the compiler does support char8_t as a distinct type,
95
   * then this AppendU8() overload for that is defined and will be chosen.
96
   *
97
   * @param bytes the pointer to the bytes
98
   * @param n the number of bytes; must be non-negative
99
   * @stable ICU 67
100
   */
101
  inline void AppendU8(const char8_t* bytes, int32_t n) {
102
    Append(reinterpret_cast<const char*>(bytes), n);
103
  }
104
#endif
105
106
  /**
107
   * Returns a writable buffer for appending and writes the buffer's capacity to
108
   * *result_capacity. Guarantees *result_capacity>=min_capacity.
109
   * May return a pointer to the caller-owned scratch buffer which must have
110
   * scratch_capacity>=min_capacity.
111
   * The returned buffer is only valid until the next operation
112
   * on this ByteSink.
113
   *
114
   * After writing at most *result_capacity bytes, call Append() with the
115
   * pointer returned from this function and the number of bytes written.
116
   * Many Append() implementations will avoid copying bytes if this function
117
   * returned an internal buffer.
118
   *
119
   * Partial usage example:
120
   *  int32_t capacity;
121
   *  char* buffer = sink->GetAppendBuffer(..., &capacity);
122
   *  ... Write n bytes into buffer, with n <= capacity.
123
   *  sink->Append(buffer, n);
124
   * In many implementations, that call to Append will avoid copying bytes.
125
   *
126
   * If the ByteSink allocates or reallocates an internal buffer, it should use
127
   * the desired_capacity_hint if appropriate.
128
   * If a caller cannot provide a reasonable guess at the desired capacity,
129
   * it should pass desired_capacity_hint=0.
130
   *
131
   * If a non-scratch buffer is returned, the caller may only pass
132
   * a prefix to it to Append().
133
   * That is, it is not correct to pass an interior pointer to Append().
134
   *
135
   * The default implementation always returns the scratch buffer.
136
   *
137
   * @param min_capacity required minimum capacity of the returned buffer;
138
   *                     must be non-negative
139
   * @param desired_capacity_hint desired capacity of the returned buffer;
140
   *                              must be non-negative
141
   * @param scratch default caller-owned buffer
142
   * @param scratch_capacity capacity of the scratch buffer
143
   * @param result_capacity pointer to an integer which will be set to the
144
   *                        capacity of the returned buffer
145
   * @return a buffer with *result_capacity>=min_capacity
146
   * @stable ICU 4.2
147
   */
148
  virtual char* GetAppendBuffer(int32_t min_capacity,
149
                                int32_t desired_capacity_hint,
150
                                char* scratch, int32_t scratch_capacity,
151
                                int32_t* result_capacity);
152
153
  /**
154
   * Flush internal buffers.
155
   * Some byte sinks use internal buffers or provide buffering
156
   * and require calling Flush() at the end of the stream.
157
   * The ByteSink should be ready for further Append() calls after Flush().
158
   * The default implementation of Flush() does nothing.
159
   * @stable ICU 4.2
160
   */
161
  virtual void Flush();
162
163
private:
164
  ByteSink(const ByteSink &) = delete;
165
  ByteSink &operator=(const ByteSink &) = delete;
166
};
167
168
// -------------------------------------------------------------
169
// Some standard implementations
170
171
/** 
172
 * Implementation of ByteSink that writes to a flat byte array,
173
 * with bounds-checking:
174
 * This sink will not write more than capacity bytes to outbuf.
175
 * If more than capacity bytes are Append()ed, then excess bytes are ignored,
176
 * and Overflowed() will return true.
177
 * Overflow does not cause a runtime error.
178
 * @stable ICU 4.2
179
 */
180
class U_COMMON_API CheckedArrayByteSink : public ByteSink {
181
public:
182
  /**
183
   * Constructs a ByteSink that will write to outbuf[0..capacity-1].
184
   * @param outbuf buffer to write to
185
   * @param capacity size of the buffer
186
   * @stable ICU 4.2
187
   */
188
  CheckedArrayByteSink(char* outbuf, int32_t capacity);
189
  /**
190
   * Destructor.
191
   * @stable ICU 4.2
192
   */
193
  virtual ~CheckedArrayByteSink();
194
  /**
195
   * Returns the sink to its original state, without modifying the buffer.
196
   * Useful for reusing both the buffer and the sink for multiple streams.
197
   * Resets the state to NumberOfBytesWritten()=NumberOfBytesAppended()=0
198
   * and Overflowed()=false.
199
   * @return *this
200
   * @stable ICU 4.6
201
   */
202
  virtual CheckedArrayByteSink& Reset();
203
  /**
204
   * Append "bytes[0,n-1]" to this.
205
   * @param bytes the pointer to the bytes
206
   * @param n the number of bytes; must be non-negative
207
   * @stable ICU 4.2
208
   */
209
  virtual void Append(const char* bytes, int32_t n);
210
  /**
211
   * Returns a writable buffer for appending and writes the buffer's capacity to
212
   * *result_capacity. For details see the base class documentation.
213
   * @param min_capacity required minimum capacity of the returned buffer;
214
   *                     must be non-negative
215
   * @param desired_capacity_hint desired capacity of the returned buffer;
216
   *                              must be non-negative
217
   * @param scratch default caller-owned buffer
218
   * @param scratch_capacity capacity of the scratch buffer
219
   * @param result_capacity pointer to an integer which will be set to the
220
   *                        capacity of the returned buffer
221
   * @return a buffer with *result_capacity>=min_capacity
222
   * @stable ICU 4.2
223
   */
224
  virtual char* GetAppendBuffer(int32_t min_capacity,
225
                                int32_t desired_capacity_hint,
226
                                char* scratch, int32_t scratch_capacity,
227
                                int32_t* result_capacity);
228
  /**
229
   * Returns the number of bytes actually written to the sink.
230
   * @return number of bytes written to the buffer
231
   * @stable ICU 4.2
232
   */
233
0
  int32_t NumberOfBytesWritten() const { return size_; }
234
  /**
235
   * Returns true if any bytes were discarded, i.e., if there was an
236
   * attempt to write more than 'capacity' bytes.
237
   * @return true if more than 'capacity' bytes were Append()ed
238
   * @stable ICU 4.2
239
   */
240
0
  UBool Overflowed() const { return overflowed_; }
241
  /**
242
   * Returns the number of bytes appended to the sink.
243
   * If Overflowed() then NumberOfBytesAppended()>NumberOfBytesWritten()
244
   * else they return the same number.
245
   * @return number of bytes written to the buffer
246
   * @stable ICU 4.6
247
   */
248
0
  int32_t NumberOfBytesAppended() const { return appended_; }
249
private:
250
  char* outbuf_;
251
  const int32_t capacity_;
252
  int32_t size_;
253
  int32_t appended_;
254
  UBool overflowed_;
255
256
  CheckedArrayByteSink() = delete;
257
  CheckedArrayByteSink(const CheckedArrayByteSink &) = delete;
258
  CheckedArrayByteSink &operator=(const CheckedArrayByteSink &) = delete;
259
};
260
261
/** 
262
 * Implementation of ByteSink that writes to a "string".
263
 * The StringClass is usually instantiated with a std::string.
264
 * @stable ICU 4.2
265
 */
266
template<typename StringClass>
267
class StringByteSink : public ByteSink {
268
 public:
269
  /**
270
   * Constructs a ByteSink that will append bytes to the dest string.
271
   * @param dest pointer to string object to append to
272
   * @stable ICU 4.2
273
   */
274
  StringByteSink(StringClass* dest) : dest_(dest) { }
275
  /**
276
   * Constructs a ByteSink that reserves append capacity and will append bytes to the dest string.
277
   * 
278
   * @param dest pointer to string object to append to
279
   * @param initialAppendCapacity capacity beyond dest->length() to be reserve()d
280
   * @stable ICU 60
281
   */
282
  StringByteSink(StringClass* dest, int32_t initialAppendCapacity) : dest_(dest) {
283
    if (initialAppendCapacity > 0 &&
284
        (uint32_t)initialAppendCapacity > (dest->capacity() - dest->length())) {
285
      dest->reserve(dest->length() + initialAppendCapacity);
286
    }
287
  }
288
  /**
289
   * Append "bytes[0,n-1]" to this.
290
   * @param data the pointer to the bytes
291
   * @param n the number of bytes; must be non-negative
292
   * @stable ICU 4.2
293
   */
294
  virtual void Append(const char* data, int32_t n) { dest_->append(data, n); }
295
 private:
296
  StringClass* dest_;
297
298
  StringByteSink() = delete;
299
  StringByteSink(const StringByteSink &) = delete;
300
  StringByteSink &operator=(const StringByteSink &) = delete;
301
};
302
303
U_NAMESPACE_END
304
305
#endif /* U_SHOW_CPLUSPLUS_API */
306
307
#endif  // __BYTESTREAM_H__