Home | History | Annotate | Download | only in unicode
      1 // Copyright (C) 2009-2010, International Business Machines
      2 // Corporation and others. All Rights Reserved.
      3 //
      4 // Copyright 2007 Google Inc. All Rights Reserved.
      5 // Author: sanjay (at) google.com (Sanjay Ghemawat)
      6 //
      7 // Abstract interface that consumes a sequence of bytes (ByteSink).
      8 //
      9 // Used so that we can write a single piece of code that can operate
     10 // on a variety of output string types.
     11 //
     12 // Various implementations of this interface are provided:
     13 //   ByteSink:
     14 //      CheckedArrayByteSink    Write to a flat array, with bounds checking
     15 //      StringByteSink          Write to an STL string
     16 
     17 // This code is a contribution of Google code, and the style used here is
     18 // a compromise between the original Google code and the ICU coding guidelines.
     19 // For example, data types are ICU-ified (size_t,int->int32_t),
     20 // and API comments doxygen-ified, but function names and behavior are
     21 // as in the original, if possible.
     22 // Assertion-style error handling, not available in ICU, was changed to
     23 // parameter "pinning" similar to UnicodeString.
     24 //
     25 // In addition, this is only a partial port of the original Google code,
     26 // limited to what was needed so far. The (nearly) complete original code
     27 // is in the ICU svn repository at icuhtml/trunk/design/strings/contrib
     28 // (see ICU ticket 6765, r25517).
     29 
     30 #ifndef __BYTESTREAM_H__
     31 #define __BYTESTREAM_H__
     32 
     33 /**
     34  * \file
     35  * \brief C++ API: Interface for writing bytes, and implementation classes.
     36  */
     37 
     38 #include "unicode/utypes.h"
     39 #include "unicode/uobject.h"
     40 #include "unicode/std_string.h"
     41 
     42 U_NAMESPACE_BEGIN
     43 
     44 /**
     45  * A ByteSink can be filled with bytes.
     46  * @stable ICU 4.2
     47  */
     48 class U_COMMON_API ByteSink : public UMemory {
     49 public:
     50   /**
     51    * Default constructor.
     52    * @stable ICU 4.2
     53    */
     54   ByteSink() { }
     55   /**
     56    * Virtual destructor.
     57    * @stable ICU 4.2
     58    */
     59   virtual ~ByteSink() { }
     60 
     61   /**
     62    * Append "bytes[0,n-1]" to this.
     63    * @param bytes the pointer to the bytes
     64    * @param n the number of bytes; must be non-negative
     65    * @stable ICU 4.2
     66    */
     67   virtual void Append(const char* bytes, int32_t n) = 0;
     68 
     69   /**
     70    * Returns a writable buffer for appending and writes the buffer's capacity to
     71    * *result_capacity. Guarantees *result_capacity>=min_capacity.
     72    * May return a pointer to the caller-owned scratch buffer which must have
     73    * scratch_capacity>=min_capacity.
     74    * The returned buffer is only valid until the next operation
     75    * on this ByteSink.
     76    *
     77    * After writing at most *result_capacity bytes, call Append() with the
     78    * pointer returned from this function and the number of bytes written.
     79    * Many Append() implementations will avoid copying bytes if this function
     80    * returned an internal buffer.
     81    *
     82    * Partial usage example:
     83    *  int32_t capacity;
     84    *  char* buffer = sink->GetAppendBuffer(..., &capacity);
     85    *  ... Write n bytes into buffer, with n <= capacity.
     86    *  sink->Append(buffer, n);
     87    * In many implementations, that call to Append will avoid copying bytes.
     88    *
     89    * If the ByteSink allocates or reallocates an internal buffer, it should use
     90    * the desired_capacity_hint if appropriate.
     91    * If a caller cannot provide a reasonable guess at the desired capacity,
     92    * it should pass desired_capacity_hint=0.
     93    *
     94    * If a non-scratch buffer is returned, the caller may only pass
     95    * a prefix to it to Append().
     96    * That is, it is not correct to pass an interior pointer to Append().
     97    *
     98    * The default implementation always returns the scratch buffer.
     99    *
    100    * @param min_capacity required minimum capacity of the returned buffer;
    101    *                     must be non-negative
    102    * @param desired_capacity_hint desired capacity of the returned buffer;
    103    *                              must be non-negative
    104    * @param scratch default caller-owned buffer
    105    * @param scratch_capacity capacity of the scratch buffer
    106    * @param result_capacity pointer to an integer which will be set to the
    107    *                        capacity of the returned buffer
    108    * @return a buffer with *result_capacity>=min_capacity
    109    * @stable ICU 4.2
    110    */
    111   virtual char* GetAppendBuffer(int32_t min_capacity,
    112                                 int32_t desired_capacity_hint,
    113                                 char* scratch, int32_t scratch_capacity,
    114                                 int32_t* result_capacity);
    115 
    116   /**
    117    * Flush internal buffers.
    118    * Some byte sinks use internal buffers or provide buffering
    119    * and require calling Flush() at the end of the stream.
    120    * The ByteSink should be ready for further Append() calls after Flush().
    121    * The default implementation of Flush() does nothing.
    122    * @stable ICU 4.2
    123    */
    124   virtual void Flush();
    125 
    126 private:
    127   ByteSink(const ByteSink &); // copy constructor not implemented
    128   ByteSink &operator=(const ByteSink &); // assignment operator not implemented
    129 };
    130 
    131 // -------------------------------------------------------------
    132 // Some standard implementations
    133 
    134 /**
    135  * Implementation of ByteSink that writes to a flat byte array,
    136  * with bounds-checking:
    137  * This sink will not write more than capacity bytes to outbuf.
    138  * If more than capacity bytes are Append()ed, then excess bytes are ignored,
    139  * and Overflowed() will return true.
    140  * Overflow does not cause a runtime error.
    141  * @stable ICU 4.2
    142  */
    143 class U_COMMON_API CheckedArrayByteSink : public ByteSink {
    144 public:
    145   /**
    146    * Constructs a ByteSink that will write to outbuf[0..capacity-1].
    147    * @param outbuf buffer to write to
    148    * @param capacity size of the buffer
    149    * @stable ICU 4.2
    150    */
    151   CheckedArrayByteSink(char* outbuf, int32_t capacity);
    152   /**
    153    * Returns the sink to its original state, without modifying the buffer.
    154    * Useful for reusing both the buffer and the sink for multiple streams.
    155    * Resets the state to NumberOfBytesWritten()=NumberOfBytesAppended()=0
    156    * and Overflowed()=FALSE.
    157    * @return *this
    158    * @draft ICU 4.6
    159    */
    160   virtual CheckedArrayByteSink& Reset();
    161   /**
    162    * Append "bytes[0,n-1]" to this.
    163    * @param bytes the pointer to the bytes
    164    * @param n the number of bytes; must be non-negative
    165    * @stable ICU 4.2
    166    */
    167   virtual void Append(const char* bytes, int32_t n);
    168   /**
    169    * Returns a writable buffer for appending and writes the buffer's capacity to
    170    * *result_capacity. For details see the base class documentation.
    171    * @param min_capacity required minimum capacity of the returned buffer;
    172    *                     must be non-negative
    173    * @param desired_capacity_hint desired capacity of the returned buffer;
    174    *                              must be non-negative
    175    * @param scratch default caller-owned buffer
    176    * @param scratch_capacity capacity of the scratch buffer
    177    * @param result_capacity pointer to an integer which will be set to the
    178    *                        capacity of the returned buffer
    179    * @return a buffer with *result_capacity>=min_capacity
    180    * @stable ICU 4.2
    181    */
    182   virtual char* GetAppendBuffer(int32_t min_capacity,
    183                                 int32_t desired_capacity_hint,
    184                                 char* scratch, int32_t scratch_capacity,
    185                                 int32_t* result_capacity);
    186   /**
    187    * Returns the number of bytes actually written to the sink.
    188    * @return number of bytes written to the buffer
    189    * @stable ICU 4.2
    190    */
    191   int32_t NumberOfBytesWritten() const { return size_; }
    192   /**
    193    * Returns true if any bytes were discarded, i.e., if there was an
    194    * attempt to write more than 'capacity' bytes.
    195    * @return TRUE if more than 'capacity' bytes were Append()ed
    196    * @stable ICU 4.2
    197    */
    198   UBool Overflowed() const { return overflowed_; }
    199   /**
    200    * Returns the number of bytes appended to the sink.
    201    * If Overflowed() then NumberOfBytesAppended()>NumberOfBytesWritten()
    202    * else they return the same number.
    203    * @return number of bytes written to the buffer
    204    * @draft ICU 4.6
    205    */
    206   int32_t NumberOfBytesAppended() const { return appended_; }
    207 private:
    208   char* outbuf_;
    209   const int32_t capacity_;
    210   int32_t size_;
    211   int32_t appended_;
    212   UBool overflowed_;
    213   CheckedArrayByteSink(); ///< default constructor not implemented
    214   CheckedArrayByteSink(const CheckedArrayByteSink &); ///< copy constructor not implemented
    215   CheckedArrayByteSink &operator=(const CheckedArrayByteSink &); ///< assignment operator not implemented
    216 };
    217 
    218 #if U_HAVE_STD_STRING
    219 
    220 /**
    221  * Implementation of ByteSink that writes to a "string".
    222  * The StringClass is usually instantiated with a std::string.
    223  * @stable ICU 4.2
    224  */
    225 template<typename StringClass>
    226 class StringByteSink : public ByteSink {
    227  public:
    228   /**
    229    * Constructs a ByteSink that will append bytes to the dest string.
    230    * @param dest pointer to string object to append to
    231    * @stable ICU 4.2
    232    */
    233   StringByteSink(StringClass* dest) : dest_(dest) { }
    234   /**
    235    * Append "bytes[0,n-1]" to this.
    236    * @param data the pointer to the bytes
    237    * @param n the number of bytes; must be non-negative
    238    * @stable ICU 4.2
    239    */
    240   virtual void Append(const char* data, int32_t n) { dest_->append(data, n); }
    241  private:
    242   StringClass* dest_;
    243   StringByteSink(); ///< default constructor not implemented
    244   StringByteSink(const StringByteSink &); ///< copy constructor not implemented
    245   StringByteSink &operator=(const StringByteSink &); ///< assignment operator not implemented
    246 };
    247 
    248 #endif
    249 
    250 U_NAMESPACE_END
    251 
    252 #endif  // __BYTESTREAM_H__
    253