ndn-cpp/c/encoding/binary-xml-encoder.c - ndnSIM/ndn-cxx - Gitiles

 /**
  * Copyright (C) 2013 Regents of the University of California.
  * @author: Jeff Thompson <jefft0@remap.ucla.edu>
  * Derived from BinaryXMLEncoder.js by Meki Cheraoui.
  * See COPYING for copyright and distribution information.
  */

 #include <math.h>
 #include "../util/ndn_memory.h"
 #include "binary-xml.h"
 #include "binary-xml-encoder.h"

 enum {
   ENCODING_LIMIT_1_BYTE = ((1 << ndn_BinaryXml_TT_VALUE_BITS) - 1),
   ENCODING_LIMIT_2_BYTES = ((1 << (ndn_BinaryXml_TT_VALUE_BITS + ndn_BinaryXml_REGULAR_VALUE_BITS)) - 1),
   ENCODING_LIMIT_3_BYTES = ((1 << (ndn_BinaryXml_TT_VALUE_BITS + 2 * ndn_BinaryXml_REGULAR_VALUE_BITS)) - 1)
 };

 /**
  * Call ndn_DynamicUInt8Array_ensureLength to ensure that there is enough room in the output, and copy
  * array to the output.  This does not write a header.
  * @param self pointer to the ndn_BinaryXmlEncoder struct
  * @param array the array to copy
  * @param arrayLength the length of the array
  * @return 0 for success, else an error code
  */
 static ndn_Error writeArray(struct ndn_BinaryXmlEncoder *self, uint8_t *array, unsigned int arrayLength)
 {
   ndn_Error error;
   if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + arrayLength)))
     return error;

   ndn_memcpy(self->output->array + self->offset, array, arrayLength);
   self->offset += arrayLength;

   return NDN_ERROR_success;
 }

 /**
  * Return the number of bytes to encode a header of value x.
  */
 static unsigned int getNHeaderEncodingBytes(unsigned int x)
 {
   // Do a quick check for pre-compiled results.
   if (x <= ENCODING_LIMIT_1_BYTE)
     return 1;
   if (x <= ENCODING_LIMIT_2_BYTES)
     return 2;
   if (x <= ENCODING_LIMIT_3_BYTES)
     return 3;

   unsigned int nBytes = 1;

   // Last byte gives you TT_VALUE_BITS.
   // Remainder each gives you REGULAR_VALUE_BITS.
   x >>= ndn_BinaryXml_TT_VALUE_BITS;
   while (x != 0) {
     ++nBytes;
     x >>= ndn_BinaryXml_REGULAR_VALUE_BITS;
   }

   return nBytes;
 }

 /**
  * Reverse the length bytes in array.
  * @param array
  * @param length
  */
 static void reverse(uint8_t *array, unsigned int length)
 {
   if (length == 0)
     return;

   uint8_t *left = array;
   uint8_t *right = array + length - 1;
   while (left < right) {
     // Swap.
     uint8_t temp = *left;
     *left = *right;
     *right = temp;

     ++left;
     --right;
   }
 }

 /**
  * Write x as an unsigned decimal integer to the output with the digits in reverse order, using ndn_DynamicUInt8Array_ensureLength.
  * This does not write a header.
  * We encode in reverse order, because this is the natural way to encode the digits, and the caller can reverse as needed.
  * @param self pointer to the ndn_BinaryXmlEncoder struct
  * @param x the unsigned int to write
  * @return 0 for success, else an error code
  */
 static ndn_Error encodeReversedUnsignedDecimalInt(struct ndn_BinaryXmlEncoder *self, unsigned int x)
 {
   while (1) {
     ndn_Error error;
     if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + 1)))
       return error;

     self->output->array[self->offset++] = (uint8_t)(x % 10 + '0');
     x /= 10;

     if (x == 0)
       break;
   }

   return NDN_ERROR_success;
 }

 /**
  * Reverse the buffer in self->output->array, then shift it right by the amount needed to prefix a header with type,
  * then encode the header at startOffset.
  * startOffser it the position in self-output.array of the first byte of the buffer and self->offset is the first byte past the end.
  * We reverse and shift in the same function to avoid unnecessary copying if we first reverse then shift.
  * @param self pointer to the ndn_BinaryXmlEncoder struct
  * @param startOffset the offset in self->output->array of the start of the buffer to shift right
  * @param type the header type
  * @return 0 for success, else an error code
  */
 static ndn_Error reverseBufferAndInsertHeader
   (struct ndn_BinaryXmlEncoder *self, unsigned int startOffset, unsigned int type)
 {
   unsigned int nBufferBytes = self->offset - startOffset;
   unsigned int nHeaderBytes = getNHeaderEncodingBytes(nBufferBytes);
   ndn_Error error;
   if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + nHeaderBytes)))
     return error;

   // To reverse and shift at the same time, we first shift nHeaderBytes to the destination while reversing,
   //   then reverse the remaining bytes in place.
   uint8_t *from = self->output->array + startOffset;
   uint8_t *fromEnd = from + nHeaderBytes;
   uint8_t *to = self->output->array + startOffset + nBufferBytes + nHeaderBytes - 1;
   while (from < fromEnd)
     *(to--) = *(from++);
   // Reverse the remaining bytes in place (if any).
   if (nBufferBytes > nHeaderBytes)
     reverse(self->output->array + startOffset + nHeaderBytes, nBufferBytes - nHeaderBytes);

   // Override the offset to force encodeTypeAndValue to encode at startOffset, then fix the offset.
   self->offset = startOffset;
   if ((error = ndn_BinaryXmlEncoder_encodeTypeAndValue(self, type, nBufferBytes)))
     // We don't really expect to get an error, since we have already ensured the length.
     return error;
   self->offset = startOffset + nHeaderBytes + nBufferBytes;

   return NDN_ERROR_success;
 }

 /**
  * Split the absolute value of x, rounded to an integer into 32 bit unsigned integers hi32 and lo32.
  * We need this because not all C compilers support 64 bit long long integers, so we carry around
  * a high precision value as a double, which we assume has more than 32 bits.
  * But we want to do bit-wise operations on integers.
  * @param x the double value
  * @param hi32 output the high 32 bits
  * @param lo32 output the low 32 bits
  */
 static inline void splitAbsDouble(double x, unsigned long *hi32, unsigned long *lo32)
 {
   if (x < 0)
     x = -x;
   x = round(x);

   double twoPower32 = 4294967296.0;
   double lo32Double = fmod(x, twoPower32);
   *lo32 = (unsigned long)lo32Double;
   *hi32 = (unsigned long)((x - lo32Double) / twoPower32);
 }

 ndn_Error ndn_BinaryXmlEncoder_encodeTypeAndValue(struct ndn_BinaryXmlEncoder *self, unsigned int type, unsigned int value)
 {
   if (type > ndn_BinaryXml_UDATA)
     return NDN_ERROR_header_type_is_out_of_range;

   // Encode backwards. Calculate how many bytes we need.
   unsigned int nEncodingBytes = getNHeaderEncodingBytes(value);
   ndn_Error error;
   if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + nEncodingBytes)))
     return error;

   // Bottom 4 bits of value go in last byte with tag.
   self->output->array[self->offset + nEncodingBytes - 1] =
     (ndn_BinaryXml_TT_MASK & type |
     ((ndn_BinaryXml_TT_VALUE_MASK & value) << ndn_BinaryXml_TT_BITS)) |
     ndn_BinaryXml_TT_FINAL; // set top bit for last byte
   value >>= ndn_BinaryXml_TT_VALUE_BITS;

   // Rest of value goes into preceding bytes, 7 bits per byte. (Zero top bit is "more" flag.)
   unsigned int i = self->offset + nEncodingBytes - 2;
   while (value != 0 && i >= self->offset) {
     self->output->array[i] = (value & ndn_BinaryXml_REGULAR_VALUE_MASK);
     value >>= ndn_BinaryXml_REGULAR_VALUE_BITS;
     --i;
   }
   if (value != 0)
     // This should not happen if getNHeaderEncodingBytes is correct.
     return NDN_ERROR_encodeTypeAndValue_miscalculated_N_encoding_bytes;

   self->offset+= nEncodingBytes;

   return NDN_ERROR_success;
 }

 ndn_Error ndn_BinaryXmlEncoder_writeElementClose(struct ndn_BinaryXmlEncoder *self)
 {
   ndn_Error error;
   if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + 1)))
     return error;

   self->output->array[self->offset] = ndn_BinaryXml_CLOSE;
   self->offset += 1;

   return NDN_ERROR_success;
 }

 ndn_Error ndn_BinaryXmlEncoder_writeBlob(struct ndn_BinaryXmlEncoder *self, uint8_t *value, unsigned int valueLength)
 {
   ndn_Error error;
   if ((error = ndn_BinaryXmlEncoder_encodeTypeAndValue(self, ndn_BinaryXml_BLOB, valueLength)))
     return error;

   if ((error = writeArray(self, value, valueLength)))
     return error;

   return NDN_ERROR_success;
 }

 ndn_Error ndn_BinaryXmlEncoder_writeBlobDTagElement(struct ndn_BinaryXmlEncoder *self, unsigned int tag, uint8_t *value, unsigned int valueLength)
 {
   ndn_Error error;
   if ((error = ndn_BinaryXmlEncoder_writeElementStartDTag(self, tag)))
     return error;

   if ((error = ndn_BinaryXmlEncoder_writeBlob(self, value, valueLength)))
     return error;

   if ((error = ndn_BinaryXmlEncoder_writeElementClose(self)))
     return error;

   return NDN_ERROR_success;
 }

 ndn_Error ndn_BinaryXmlEncoder_writeUData(struct ndn_BinaryXmlEncoder *self, uint8_t *value, unsigned int valueLength)
 {
   ndn_Error error;
   if ((error = ndn_BinaryXmlEncoder_encodeTypeAndValue(self, ndn_BinaryXml_UDATA, valueLength)))
     return error;

   if ((error = writeArray(self, value, valueLength)))
     return error;

   return NDN_ERROR_success;
 }

 ndn_Error ndn_BinaryXmlEncoder_writeUDataDTagElement(struct ndn_BinaryXmlEncoder *self, unsigned int tag, uint8_t *value, unsigned int valueLength)
 {
   ndn_Error error;
   if ((error = ndn_BinaryXmlEncoder_writeElementStartDTag(self, tag)))
     return error;

   if ((error = ndn_BinaryXmlEncoder_writeUData(self, value, valueLength)))
     return error;

   if ((error = ndn_BinaryXmlEncoder_writeElementClose(self)))
     return error;

   return NDN_ERROR_success;
 }

 ndn_Error ndn_BinaryXmlEncoder_writeUnsignedDecimalInt(struct ndn_BinaryXmlEncoder *self, unsigned int value)
 {
   // First write the decimal int (to find out how many bytes it is), then shift it forward to make room for the header.
   unsigned int startOffset = self->offset;

   ndn_Error error;
   if ((error = encodeReversedUnsignedDecimalInt(self, value)))
     return error;

   if ((error = reverseBufferAndInsertHeader(self, startOffset, ndn_BinaryXml_UDATA)))
     return error;

   return NDN_ERROR_success;
 }

 ndn_Error ndn_BinaryXmlEncoder_writeUnsignedDecimalIntDTagElement(struct ndn_BinaryXmlEncoder *self, unsigned int tag, unsigned int value)
 {
   ndn_Error error;
   if ((error = ndn_BinaryXmlEncoder_writeElementStartDTag(self, tag)))
     return error;

   if ((error = ndn_BinaryXmlEncoder_writeUnsignedDecimalInt(self, value)))
     return error;

   if ((error = ndn_BinaryXmlEncoder_writeElementClose(self)))
     return error;

   return NDN_ERROR_success;
 }

 ndn_Error ndn_BinaryXmlEncoder_writeAbsDoubleBigEndianBlob(struct ndn_BinaryXmlEncoder *self, double value)
 {
   unsigned long hi32, lo32;
   splitAbsDouble(value, &hi32, &lo32);

   // First encode the big endian backwards, then reverseBufferAndInsertHeader will reverse it.
   unsigned int startOffset = self->offset;

   ndn_Error error;
   while (lo32 != 0) {
     if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + 1)))
       return error;

     self->output->array[self->offset++] = (uint8_t)(lo32 & 0xff);
     lo32 >>= 8;
   }

   if (hi32 != 0) {
     // Pad the lo values out to 4 bytes.
     while (self->offset - startOffset < 4) {
       if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + 1)))
         return error;

       self->output->array[self->offset++] = 0;
     }

     // Encode hi32
     while (hi32 != 0) {
       if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + 1)))
         return error;

       self->output->array[self->offset++] = (uint8_t)(hi32 & 0xff);
       hi32 >>= 8;
     }
   }

   if ((error = reverseBufferAndInsertHeader(self, startOffset, ndn_BinaryXml_BLOB)))
     return error;

   return NDN_ERROR_success;
 }

 ndn_Error ndn_BinaryXmlEncoder_writeTimeMillisecondsDTagElement(struct ndn_BinaryXmlEncoder *self, unsigned int tag, double milliseconds)
 {
   ndn_Error error;
   if ((error = ndn_BinaryXmlEncoder_writeElementStartDTag(self, tag)))
     return error;

   if ((error = ndn_BinaryXmlEncoder_writeAbsDoubleBigEndianBlob(self, (milliseconds / 1000.0) * 4096.0)))
     return error;

   if ((error = ndn_BinaryXmlEncoder_writeElementClose(self)))
     return error;

   return NDN_ERROR_success;
 }
	/**
	* Copyright (C) 2013 Regents of the University of California.
	* @author: Jeff Thompson <jefft0@remap.ucla.edu>
	* Derived from BinaryXMLEncoder.js by Meki Cheraoui.
	* See COPYING for copyright and distribution information.
	*/

	#include <math.h>
	#include "../util/ndn_memory.h"
	#include "binary-xml.h"
	#include "binary-xml-encoder.h"

	enum {
	ENCODING_LIMIT_1_BYTE = ((1 << ndn_BinaryXml_TT_VALUE_BITS) - 1),
	ENCODING_LIMIT_2_BYTES = ((1 << (ndn_BinaryXml_TT_VALUE_BITS + ndn_BinaryXml_REGULAR_VALUE_BITS)) - 1),
	ENCODING_LIMIT_3_BYTES = ((1 << (ndn_BinaryXml_TT_VALUE_BITS + 2 * ndn_BinaryXml_REGULAR_VALUE_BITS)) - 1)
	};

	/**
	* Call ndn_DynamicUInt8Array_ensureLength to ensure that there is enough room in the output, and copy
	* array to the output. This does not write a header.
	* @param self pointer to the ndn_BinaryXmlEncoder struct
	* @param array the array to copy
	* @param arrayLength the length of the array
	* @return 0 for success, else an error code
	*/
	static ndn_Error writeArray(struct ndn_BinaryXmlEncoder self, uint8_t array, unsigned int arrayLength)
	{
	ndn_Error error;
	if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + arrayLength)))
	return error;

	ndn_memcpy(self->output->array + self->offset, array, arrayLength);
	self->offset += arrayLength;

	return NDN_ERROR_success;
	}

	/**
	* Return the number of bytes to encode a header of value x.
	*/
	static unsigned int getNHeaderEncodingBytes(unsigned int x)
	{
	// Do a quick check for pre-compiled results.
	if (x <= ENCODING_LIMIT_1_BYTE)
	return 1;
	if (x <= ENCODING_LIMIT_2_BYTES)
	return 2;
	if (x <= ENCODING_LIMIT_3_BYTES)
	return 3;

	unsigned int nBytes = 1;

	// Last byte gives you TT_VALUE_BITS.
	// Remainder each gives you REGULAR_VALUE_BITS.
	x >>= ndn_BinaryXml_TT_VALUE_BITS;
	while (x != 0) {
	++nBytes;
	x >>= ndn_BinaryXml_REGULAR_VALUE_BITS;
	}

	return nBytes;
	}

	/**
	* Reverse the length bytes in array.
	* @param array
	* @param length
	*/
	static void reverse(uint8_t *array, unsigned int length)
	{
	if (length == 0)
	return;

	uint8_t *left = array;
	uint8_t *right = array + length - 1;
	while (left < right) {
	// Swap.
	uint8_t temp = *left;
	left = right;
	*right = temp;

	++left;
	--right;
	}
	}

	/**
	* Write x as an unsigned decimal integer to the output with the digits in reverse order, using ndn_DynamicUInt8Array_ensureLength.
	* This does not write a header.
	* We encode in reverse order, because this is the natural way to encode the digits, and the caller can reverse as needed.
	* @param self pointer to the ndn_BinaryXmlEncoder struct
	* @param x the unsigned int to write
	* @return 0 for success, else an error code
	*/
	static ndn_Error encodeReversedUnsignedDecimalInt(struct ndn_BinaryXmlEncoder *self, unsigned int x)
	{
	while (1) {
	ndn_Error error;
	if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + 1)))
	return error;

	self->output->array[self->offset++] = (uint8_t)(x % 10 + '0');
	x /= 10;

	if (x == 0)
	break;
	}

	return NDN_ERROR_success;
	}

	/**
	* Reverse the buffer in self->output->array, then shift it right by the amount needed to prefix a header with type,
	* then encode the header at startOffset.
	* startOffser it the position in self-output.array of the first byte of the buffer and self->offset is the first byte past the end.
	* We reverse and shift in the same function to avoid unnecessary copying if we first reverse then shift.
	* @param self pointer to the ndn_BinaryXmlEncoder struct
	* @param startOffset the offset in self->output->array of the start of the buffer to shift right
	* @param type the header type
	* @return 0 for success, else an error code
	*/
	static ndn_Error reverseBufferAndInsertHeader
	(struct ndn_BinaryXmlEncoder *self, unsigned int startOffset, unsigned int type)
	{
	unsigned int nBufferBytes = self->offset - startOffset;
	unsigned int nHeaderBytes = getNHeaderEncodingBytes(nBufferBytes);
	ndn_Error error;
	if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + nHeaderBytes)))
	return error;

	// To reverse and shift at the same time, we first shift nHeaderBytes to the destination while reversing,
	// then reverse the remaining bytes in place.
	uint8_t *from = self->output->array + startOffset;
	uint8_t *fromEnd = from + nHeaderBytes;
	uint8_t *to = self->output->array + startOffset + nBufferBytes + nHeaderBytes - 1;
	while (from < fromEnd)
	(to--) = (from++);
	// Reverse the remaining bytes in place (if any).
	if (nBufferBytes > nHeaderBytes)
	reverse(self->output->array + startOffset + nHeaderBytes, nBufferBytes - nHeaderBytes);

	// Override the offset to force encodeTypeAndValue to encode at startOffset, then fix the offset.
	self->offset = startOffset;
	if ((error = ndn_BinaryXmlEncoder_encodeTypeAndValue(self, type, nBufferBytes)))
	// We don't really expect to get an error, since we have already ensured the length.
	return error;
	self->offset = startOffset + nHeaderBytes + nBufferBytes;

	return NDN_ERROR_success;
	}

	/**
	* Split the absolute value of x, rounded to an integer into 32 bit unsigned integers hi32 and lo32.
	* We need this because not all C compilers support 64 bit long long integers, so we carry around
	* a high precision value as a double, which we assume has more than 32 bits.
	* But we want to do bit-wise operations on integers.
	* @param x the double value
	* @param hi32 output the high 32 bits
	* @param lo32 output the low 32 bits
	*/
	static inline void splitAbsDouble(double x, unsigned long hi32, unsigned long lo32)
	{
	if (x < 0)
	x = -x;
	x = round(x);

	double twoPower32 = 4294967296.0;
	double lo32Double = fmod(x, twoPower32);
	*lo32 = (unsigned long)lo32Double;
	*hi32 = (unsigned long)((x - lo32Double) / twoPower32);
	}

	ndn_Error ndn_BinaryXmlEncoder_encodeTypeAndValue(struct ndn_BinaryXmlEncoder *self, unsigned int type, unsigned int value)
	{
	if (type > ndn_BinaryXml_UDATA)
	return NDN_ERROR_header_type_is_out_of_range;

	// Encode backwards. Calculate how many bytes we need.
	unsigned int nEncodingBytes = getNHeaderEncodingBytes(value);
	ndn_Error error;
	if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + nEncodingBytes)))
	return error;

	// Bottom 4 bits of value go in last byte with tag.
	self->output->array[self->offset + nEncodingBytes - 1] =
	(ndn_BinaryXml_TT_MASK & type \|
	((ndn_BinaryXml_TT_VALUE_MASK & value) << ndn_BinaryXml_TT_BITS)) \|
	ndn_BinaryXml_TT_FINAL; // set top bit for last byte
	value >>= ndn_BinaryXml_TT_VALUE_BITS;

	// Rest of value goes into preceding bytes, 7 bits per byte. (Zero top bit is "more" flag.)
	unsigned int i = self->offset + nEncodingBytes - 2;
	while (value != 0 && i >= self->offset) {
	self->output->array[i] = (value & ndn_BinaryXml_REGULAR_VALUE_MASK);
	value >>= ndn_BinaryXml_REGULAR_VALUE_BITS;
	--i;
	}
	if (value != 0)
	// This should not happen if getNHeaderEncodingBytes is correct.
	return NDN_ERROR_encodeTypeAndValue_miscalculated_N_encoding_bytes;

	self->offset+= nEncodingBytes;

	return NDN_ERROR_success;
	}

	ndn_Error ndn_BinaryXmlEncoder_writeElementClose(struct ndn_BinaryXmlEncoder *self)
	{
	ndn_Error error;
	if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + 1)))
	return error;

	self->output->array[self->offset] = ndn_BinaryXml_CLOSE;
	self->offset += 1;

	return NDN_ERROR_success;
	}

	ndn_Error ndn_BinaryXmlEncoder_writeBlob(struct ndn_BinaryXmlEncoder self, uint8_t value, unsigned int valueLength)
	{
	ndn_Error error;
	if ((error = ndn_BinaryXmlEncoder_encodeTypeAndValue(self, ndn_BinaryXml_BLOB, valueLength)))
	return error;

	if ((error = writeArray(self, value, valueLength)))
	return error;

	return NDN_ERROR_success;
	}

	ndn_Error ndn_BinaryXmlEncoder_writeBlobDTagElement(struct ndn_BinaryXmlEncoder self, unsigned int tag, uint8_t value, unsigned int valueLength)
	{
	ndn_Error error;
	if ((error = ndn_BinaryXmlEncoder_writeElementStartDTag(self, tag)))
	return error;

	if ((error = ndn_BinaryXmlEncoder_writeBlob(self, value, valueLength)))
	return error;

	if ((error = ndn_BinaryXmlEncoder_writeElementClose(self)))
	return error;

	return NDN_ERROR_success;
	}

	ndn_Error ndn_BinaryXmlEncoder_writeUData(struct ndn_BinaryXmlEncoder self, uint8_t value, unsigned int valueLength)
	{
	ndn_Error error;
	if ((error = ndn_BinaryXmlEncoder_encodeTypeAndValue(self, ndn_BinaryXml_UDATA, valueLength)))
	return error;

	if ((error = writeArray(self, value, valueLength)))
	return error;

	return NDN_ERROR_success;
	}

	ndn_Error ndn_BinaryXmlEncoder_writeUDataDTagElement(struct ndn_BinaryXmlEncoder self, unsigned int tag, uint8_t value, unsigned int valueLength)
	{
	ndn_Error error;
	if ((error = ndn_BinaryXmlEncoder_writeElementStartDTag(self, tag)))
	return error;

	if ((error = ndn_BinaryXmlEncoder_writeUData(self, value, valueLength)))
	return error;

	if ((error = ndn_BinaryXmlEncoder_writeElementClose(self)))
	return error;

	return NDN_ERROR_success;
	}

	ndn_Error ndn_BinaryXmlEncoder_writeUnsignedDecimalInt(struct ndn_BinaryXmlEncoder *self, unsigned int value)
	{
	// First write the decimal int (to find out how many bytes it is), then shift it forward to make room for the header.
	unsigned int startOffset = self->offset;

	ndn_Error error;
	if ((error = encodeReversedUnsignedDecimalInt(self, value)))
	return error;

	if ((error = reverseBufferAndInsertHeader(self, startOffset, ndn_BinaryXml_UDATA)))
	return error;

	return NDN_ERROR_success;
	}

	ndn_Error ndn_BinaryXmlEncoder_writeUnsignedDecimalIntDTagElement(struct ndn_BinaryXmlEncoder *self, unsigned int tag, unsigned int value)
	{
	ndn_Error error;
	if ((error = ndn_BinaryXmlEncoder_writeElementStartDTag(self, tag)))
	return error;

	if ((error = ndn_BinaryXmlEncoder_writeUnsignedDecimalInt(self, value)))
	return error;

	if ((error = ndn_BinaryXmlEncoder_writeElementClose(self)))
	return error;

	return NDN_ERROR_success;
	}

	ndn_Error ndn_BinaryXmlEncoder_writeAbsDoubleBigEndianBlob(struct ndn_BinaryXmlEncoder *self, double value)
	{
	unsigned long hi32, lo32;
	splitAbsDouble(value, &hi32, &lo32);

	// First encode the big endian backwards, then reverseBufferAndInsertHeader will reverse it.
	unsigned int startOffset = self->offset;

	ndn_Error error;
	while (lo32 != 0) {
	if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + 1)))
	return error;

	self->output->array[self->offset++] = (uint8_t)(lo32 & 0xff);
	lo32 >>= 8;
	}

	if (hi32 != 0) {
	// Pad the lo values out to 4 bytes.
	while (self->offset - startOffset < 4) {
	if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + 1)))
	return error;

	self->output->array[self->offset++] = 0;
	}

	// Encode hi32
	while (hi32 != 0) {
	if ((error = ndn_DynamicUInt8Array_ensureLength(self->output, self->offset + 1)))
	return error;

	self->output->array[self->offset++] = (uint8_t)(hi32 & 0xff);
	hi32 >>= 8;
	}
	}

	if ((error = reverseBufferAndInsertHeader(self, startOffset, ndn_BinaryXml_BLOB)))
	return error;

	return NDN_ERROR_success;
	}

	ndn_Error ndn_BinaryXmlEncoder_writeTimeMillisecondsDTagElement(struct ndn_BinaryXmlEncoder *self, unsigned int tag, double milliseconds)
	{
	ndn_Error error;
	if ((error = ndn_BinaryXmlEncoder_writeElementStartDTag(self, tag)))
	return error;

	if ((error = ndn_BinaryXmlEncoder_writeAbsDoubleBigEndianBlob(self, (milliseconds / 1000.0) * 4096.0)))
	return error;

	if ((error = ndn_BinaryXmlEncoder_writeElementClose(self)))
	return error;

	return NDN_ERROR_success;
	}