src/qcbor_decode.c

/*==============================================================================
 Copyright (c) 2016-2018, The Linux Foundation.
 Copyright (c) 2018-2020, Laurence Lundblade.
 All rights reserved.

Redistribution and use in source and binary forms, with or without
modification, are permitted provided that the following conditions are
met:
    * Redistributions of source code must retain the above copyright
      notice, this list of conditions and the following disclaimer.
    * Redistributions in binary form must reproduce the above
      copyright notice, this list of conditions and the following
      disclaimer in the documentation and/or other materials provided
      with the distribution.
    * Neither the name of The Linux Foundation nor the names of its
      contributors, nor the name "Laurence Lundblade" may be used to
      endorse or promote products derived from this software without
      specific prior written permission.

THIS SOFTWARE IS PROVIDED "AS IS" AND ANY EXPRESS OR IMPLIED
WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT
ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS
BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN
IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 =============================================================================*/

/*==============================================================================
 FILE:  qcbor_decode.c

 DESCRIPTION:  This file contains the implementation of QCBOR.

 EDIT HISTORY FOR FILE:

 This section contains comments describing changes made to the module.
 Notice that changes are listed in reverse chronological order.

 when       who             what, where, why
 --------   ----            ---------------------------------------------------
 01/28/2020 llundblade      Refine integer signedness to quiet static analysis.
 01/25/2020 llundblade      Cleaner handling of too-long encoded string input.
 01/25/2020 llundblade      Refine use of integer types to quiet static analysis
 01/08/2020 llundblade      Documentation corrections & improved code formatting
 12/30/19   llundblade      Add support for decimal fractions and bigfloats.
 11/07/19   llundblade      Fix long long conversion to double compiler warning
 09/07/19   llundblade      Fix bug decoding empty arrays and maps
 07/31/19   llundblade      Decode error fixes for some not-well-formed CBOR
 07/31/19   llundblade      New error code for better end of data handling
 02/17/19   llundblade      Fixed: QCBORItem.u{Data|Label}Alloc when
                            bAllStrings set
 02/16/19   llundblade      Redesign MemPool to fix memory access alignment bug
 01/10/19   llundblade      Clever type and argument decoder; 250 bytes smaller
 11/9/18    llundblade      Error codes are now enums.
 11/2/18    llundblade      Simplify float decoding and align with preferred
                            float encoding
 10/31/18   llundblade      Switch to one license that is almost BSD-3.
 10/28/18   llundblade      Reworked tag decoding
 10/15/18   llundblade      Indefinite length maps and arrays supported
 10/8/18    llundblade      Indefinite length strings supported
 02/04/17   llundbla        Work on CPUs that don's require pointer alignment
                            by making use of changes in UsefulBuf
 03/01/17   llundbla        More data types; decoding improvements and fixes
 11/13/16   llundbla        Integrate most TZ changes back into github version.
 09/30/16   gkanike         Porting to TZ.
 03/15/16   llundbla        Initial Version.

 =============================================================================*/

#include "qcbor.h"
#include "ieee754.h"


/*
 This casts away the const-ness of a pointer, usually so it can be
 freed or realloced.
 */
#define UNCONST_POINTER(ptr)    ((void *)(ptr))



/*===========================================================================
 DecodeNesting -- Functions for tracking array/map nesting when decoding

 See qcbor.h for definition of the object used here: QCBORDecodeNesting
  ===========================================================================*/

inline static int
IsMapOrArray(uint8_t uDataType)
{
   return uDataType == QCBOR_TYPE_MAP || uDataType == QCBOR_TYPE_ARRAY;
}

inline static int
DecodeNesting_IsNested(const QCBORDecodeNesting *pNesting)
{
   return pNesting->pCurrent != &(pNesting->pMapsAndArrays[0]);
}

inline static int
DecodeNesting_IsIndefiniteLength(const QCBORDecodeNesting *pNesting)
{
   return pNesting->pCurrent->uCount == UINT16_MAX;
}

inline static uint8_t
DecodeNesting_GetLevel(QCBORDecodeNesting *pNesting)
{
   // Check in DecodeNesting_Descend and never having
   // QCBOR_MAX_ARRAY_NESTING > 255 gaurantee cast is safe
   return (uint8_t)(pNesting->pCurrent - &(pNesting->pMapsAndArrays[0]));
}

inline static int
DecodeNesting_TypeIsMap(const QCBORDecodeNesting *pNesting)
{
   if(!DecodeNesting_IsNested(pNesting)) {
      return 0;
   }

   return CBOR_MAJOR_TYPE_MAP == pNesting->pCurrent->uMajorType;
}

// Process a break. This will either ascend the nesting or error out
inline static QCBORError
DecodeNesting_BreakAscend(QCBORDecodeNesting *pNesting)
{
   // breaks must always occur when there is nesting
   if(!DecodeNesting_IsNested(pNesting)) {
      return QCBOR_ERR_BAD_BREAK;
   }

   // breaks can only occur when the map/array is indefinite length
   if(!DecodeNesting_IsIndefiniteLength(pNesting)) {
      return QCBOR_ERR_BAD_BREAK;
   }

   // if all OK, the break reduces the level of nesting
   pNesting->pCurrent--;

   return QCBOR_SUCCESS;
}

// Called on every single item except breaks including open of a map/array
inline static void
DecodeNesting_DecrementCount(QCBORDecodeNesting *pNesting)
{
   while(DecodeNesting_IsNested(pNesting)) {
      // Not at the top level, so there is decrementing to be done.

      if(!DecodeNesting_IsIndefiniteLength(pNesting)) {
         // Decrement the current nesting level if it is not indefinite.
         pNesting->pCurrent->uCount--;
      }

      if(pNesting->pCurrent->uCount != 0) {
         // Did not close out an array or map, so nothing further
         break;
      }

      // Closed out an array or map so level up
      pNesting->pCurrent--;

      // Continue with loop to see if closing out this doesn't close out more
   }
}

// Called on every map/array
inline static QCBORError
DecodeNesting_Descend(QCBORDecodeNesting *pNesting, QCBORItem *pItem)
{
   QCBORError nReturn = QCBOR_SUCCESS;

   if(pItem->val.uCount == 0) {
      // Nothing to do for empty definite lenth arrays. They are just are
      // effectively the same as an item that is not a map or array
      goto Done;
      // Empty indefinite length maps and arrays are handled elsewhere
   }

   // Error out if arrays is too long to handle
   if(pItem->val.uCount != UINT16_MAX && pItem->val.uCount > QCBOR_MAX_ITEMS_IN_ARRAY) {
      nReturn = QCBOR_ERR_ARRAY_TOO_LONG;
      goto Done;
   }

   // Error out if nesting is too deep
   if(pNesting->pCurrent >= &(pNesting->pMapsAndArrays[QCBOR_MAX_ARRAY_NESTING])) {
      nReturn = QCBOR_ERR_ARRAY_NESTING_TOO_DEEP;
      goto Done;
   }

   // The actual descend
   pNesting->pCurrent++;

   // Record a few details for this nesting level
   pNesting->pCurrent->uMajorType = pItem->uDataType;
   pNesting->pCurrent->uCount     = pItem->val.uCount;

Done:
   return nReturn;;
}

inline static void
DecodeNesting_Init(QCBORDecodeNesting *pNesting)
{
   pNesting->pCurrent = &(pNesting->pMapsAndArrays[0]);
}



/*
 This list of built-in tags. Only add tags here that are
 clearly established and useful. Once a tag is added here
 it can't be taken out as that would break backwards compatibility.
 There are only 48 slots available forever.
 */
static const uint16_t spBuiltInTagMap[] = {
   CBOR_TAG_DATE_STRING, // See TAG_MAPPER_FIRST_SIX
   CBOR_TAG_DATE_EPOCH, // See TAG_MAPPER_FIRST_SIX
   CBOR_TAG_POS_BIGNUM, // See TAG_MAPPER_FIRST_SIX
   CBOR_TAG_NEG_BIGNUM, // See TAG_MAPPER_FIRST_SIX
   CBOR_TAG_DECIMAL_FRACTION, // See TAG_MAPPER_FIRST_SIX
   CBOR_TAG_BIGFLOAT, // See TAG_MAPPER_FIRST_SIX
   CBOR_TAG_COSE_ENCRYPTO,
   CBOR_TAG_COSE_MAC0,
   CBOR_TAG_COSE_SIGN1,
   CBOR_TAG_ENC_AS_B64URL,
   CBOR_TAG_ENC_AS_B64,
   CBOR_TAG_ENC_AS_B16,
   CBOR_TAG_CBOR,
   CBOR_TAG_URI,
   CBOR_TAG_B64URL,
   CBOR_TAG_B64,
   CBOR_TAG_REGEX,
   CBOR_TAG_MIME,
   CBOR_TAG_BIN_UUID,
   CBOR_TAG_CWT,
   CBOR_TAG_ENCRYPT,
   CBOR_TAG_MAC,
   CBOR_TAG_SIGN,
   CBOR_TAG_GEO_COORD,
   CBOR_TAG_CBOR_MAGIC
};

// This is used in a bit of cleverness in GetNext_TaggedItem() to
// keep code size down and switch for the internal processing of
// these types. This will break if the first six items in
// spBuiltInTagMap don't have values 0,1,2,3,4,5. That is the
// mapping is 0 to 0, 1 to 1, 2 to 2 and 3 to 3....
#define QCBOR_TAGFLAG_DATE_STRING      (0x01LL << CBOR_TAG_DATE_STRING)
#define QCBOR_TAGFLAG_DATE_EPOCH       (0x01LL << CBOR_TAG_DATE_EPOCH)
#define QCBOR_TAGFLAG_POS_BIGNUM       (0x01LL << CBOR_TAG_POS_BIGNUM)
#define QCBOR_TAGFLAG_NEG_BIGNUM       (0x01LL << CBOR_TAG_NEG_BIGNUM)
#define QCBOR_TAGFLAG_DECIMAL_FRACTION (0x01LL << CBOR_TAG_DECIMAL_FRACTION)
#define QCBOR_TAGFLAG_BIGFLOAT         (0x01LL << CBOR_TAG_BIGFLOAT)

#define TAG_MAPPER_FIRST_SIX (QCBOR_TAGFLAG_DATE_STRING       |\
                               QCBOR_TAGFLAG_DATE_EPOCH       |\
                               QCBOR_TAGFLAG_POS_BIGNUM       |\
                               QCBOR_TAGFLAG_NEG_BIGNUM       |\
                               QCBOR_TAGFLAG_DECIMAL_FRACTION |\
                               QCBOR_TAGFLAG_BIGFLOAT)

#define TAG_MAPPER_FIRST_FOUR (QCBOR_TAGFLAG_DATE_STRING      |\
                               QCBOR_TAGFLAG_DATE_EPOCH       |\
                               QCBOR_TAGFLAG_POS_BIGNUM       |\
                               QCBOR_TAGFLAG_NEG_BIGNUM)

#define TAG_MAPPER_TOTAL_TAG_BITS 64 // Number of bits in a uint64_t
#define TAG_MAPPER_CUSTOM_TAGS_BASE_INDEX (TAG_MAPPER_TOTAL_TAG_BITS - QCBOR_MAX_CUSTOM_TAGS) // 48
#define TAG_MAPPER_MAX_SIZE_BUILT_IN_TAGS (TAG_MAPPER_TOTAL_TAG_BITS - QCBOR_MAX_CUSTOM_TAGS ) // 48

static inline int TagMapper_LookupBuiltIn(uint64_t uTag)
{
   if(sizeof(spBuiltInTagMap)/sizeof(uint16_t) > TAG_MAPPER_MAX_SIZE_BUILT_IN_TAGS) {
      /*
       This is a cross-check to make sure the above array doesn't
       accidentally get made too big.  In normal conditions the above
       test should optimize out as all the values are known at compile
       time.
       */
      return -1;
   }

   if(uTag > UINT16_MAX) {
      // This tag map works only on 16-bit tags
      return -1;
   }

   for(int nTagBitIndex = 0; nTagBitIndex < (int)(sizeof(spBuiltInTagMap)/sizeof(uint16_t)); nTagBitIndex++) {
      if(spBuiltInTagMap[nTagBitIndex] == uTag) {
         return nTagBitIndex;
      }
   }
   return -1; // Indicates no match
}

static inline int TagMapper_LookupCallerConfigured(const QCBORTagListIn *pCallerConfiguredTagMap, uint64_t uTag)
{
   for(int nTagBitIndex = 0; nTagBitIndex < pCallerConfiguredTagMap->uNumTags; nTagBitIndex++) {
      if(pCallerConfiguredTagMap->puTags[nTagBitIndex] == uTag) {
         return nTagBitIndex + TAG_MAPPER_CUSTOM_TAGS_BASE_INDEX;
      }
   }

   return -1; // Indicates no match
}

/*
  Find the tag bit index for a given tag value, or error out

 This and the above functions could probably be optimized and made
 clearer and neater.
 */
static QCBORError
TagMapper_Lookup(const QCBORTagListIn *pCallerConfiguredTagMap,
                 uint64_t uTag,
                 uint8_t *puTagBitIndex)
{
   int nTagBitIndex = TagMapper_LookupBuiltIn(uTag);
   if(nTagBitIndex >= 0) {
      // Cast is safe because TagMapper_LookupBuiltIn never returns > 47
      *puTagBitIndex = (uint8_t)nTagBitIndex;
      return QCBOR_SUCCESS;
   }

   if(pCallerConfiguredTagMap) {
      if(pCallerConfiguredTagMap->uNumTags > QCBOR_MAX_CUSTOM_TAGS) {
         return QCBOR_ERR_TOO_MANY_TAGS;
      }
      nTagBitIndex = TagMapper_LookupCallerConfigured(pCallerConfiguredTagMap, uTag);
      if(nTagBitIndex >= 0) {
         // Cast is safe because TagMapper_LookupBuiltIn never returns > 63

         *puTagBitIndex = (uint8_t)nTagBitIndex;
         return QCBOR_SUCCESS;
      }
   }

   return QCBOR_ERR_BAD_OPT_TAG;
}



/*===========================================================================
   QCBORStringAllocate -- STRING ALLOCATOR INVOCATION

   The following four functions are pretty wrappers for invocation of
   the string allocator supplied by the caller.

  ===========================================================================*/

static inline void
StringAllocator_Free(const QCORInternalAllocator *pMe, void *pMem)
{
   (pMe->pfAllocator)(pMe->pAllocateCxt, pMem, 0);
}

// StringAllocator_Reallocate called with pMem NULL is
// equal to StringAllocator_Allocate()
static inline UsefulBuf
StringAllocator_Reallocate(const QCORInternalAllocator *pMe,
                           void *pMem,
                           size_t uSize)
{
   return (pMe->pfAllocator)(pMe->pAllocateCxt, pMem, uSize);
}

static inline UsefulBuf
StringAllocator_Allocate(const QCORInternalAllocator *pMe, size_t uSize)
{
   return (pMe->pfAllocator)(pMe->pAllocateCxt, NULL, uSize);
}

static inline void
StringAllocator_Destruct(const QCORInternalAllocator *pMe)
{
   if(pMe->pfAllocator) {
      (pMe->pfAllocator)(pMe->pAllocateCxt, NULL, 0);
   }
}



/*===========================================================================
 QCBORDecode -- The main implementation of CBOR decoding

 See qcbor.h for definition of the object used here: QCBORDecodeContext
  ===========================================================================*/
/*
 Public function, see header file
 */
void QCBORDecode_Init(QCBORDecodeContext *me,
                      UsefulBufC EncodedCBOR,
                      QCBORDecodeMode nDecodeMode)
{
   memset(me, 0, sizeof(QCBORDecodeContext));
   UsefulInputBuf_Init(&(me->InBuf), EncodedCBOR);
   // Don't bother with error check on decode mode. If a bad value is
   // passed it will just act as if the default normal mode of 0 was set.
   me->uDecodeMode = nDecodeMode;
   DecodeNesting_Init(&(me->nesting));
}


/*
 Public function, see header file
 */
void QCBORDecode_SetUpAllocator(QCBORDecodeContext *pMe,
                                QCBORStringAllocate pfAllocateFunction,
                                void *pAllocateContext,
                                bool bAllStrings)
{
   pMe->StringAllocator.pfAllocator   = pfAllocateFunction;
   pMe->StringAllocator.pAllocateCxt  = pAllocateContext;
   pMe->bStringAllocateAll            = bAllStrings;
}


/*
 Public function, see header file
 */
void QCBORDecode_SetCallerConfiguredTagList(QCBORDecodeContext *me,
                                            const QCBORTagListIn *pTagList)
{
   me->pCallerConfiguredTagList = pTagList;
}


/*
 This decodes the fundamental part of a CBOR data item, the type and
 number

 This is the Counterpart to InsertEncodedTypeAndNumber().

 This does the network->host byte order conversion. The conversion
 here also results in the conversion for floats in addition to that
 for lengths, tags and integer values.

 This returns:
   pnMajorType -- the major type for the item

   puArgument -- the "number" which is used a the value for integers,
               tags and floats and length for strings and arrays

   pnAdditionalInfo -- Pass this along to know what kind of float or
                       if length is indefinite

 The int type is preferred to uint8_t for some variables as this
 avoids integer promotions, can reduce code size and makes
 static analyzers happier.
 */
inline static QCBORError DecodeTypeAndNumber(UsefulInputBuf *pUInBuf,
                                              int *pnMajorType,
                                              uint64_t *puArgument,
                                              int *pnAdditionalInfo)
{
   QCBORError nReturn;

   // Get the initial byte that every CBOR data item has
   const int nInitialByte = (int)UsefulInputBuf_GetByte(pUInBuf);

   // Break down the initial byte
   const int nTmpMajorType   = nInitialByte >> 5;
   const int nAdditionalInfo = nInitialByte & 0x1f;

   // Where the number or argument accumulates
   uint64_t uArgument;

   if(nAdditionalInfo >= LEN_IS_ONE_BYTE && nAdditionalInfo <= LEN_IS_EIGHT_BYTES) {
      // Need to get 1,2,4 or 8 additional argument bytes Map
      // LEN_IS_ONE_BYTE.. LEN_IS_EIGHT_BYTES to actual length
      static const uint8_t aIterate[] = {1,2,4,8};

      // Loop getting all the bytes in the argument
      uArgument = 0;
      for(int i = aIterate[nAdditionalInfo - LEN_IS_ONE_BYTE]; i; i--) {
         // This shift and add gives the endian conversion
         uArgument = (uArgument << 8) + UsefulInputBuf_GetByte(pUInBuf);
      }
   } else if(nAdditionalInfo >= ADDINFO_RESERVED1 && nAdditionalInfo <= ADDINFO_RESERVED3) {
      // The reserved and thus-far unused additional info values
      nReturn = QCBOR_ERR_UNSUPPORTED;
      goto Done;
   } else {
      // Less than 24, additional info is argument or 31, an indefinite length
      // No more bytes to get
      uArgument = (uint64_t)nAdditionalInfo;
   }

   if(UsefulInputBuf_GetError(pUInBuf)) {
      nReturn = QCBOR_ERR_HIT_END;
      goto Done;
   }

   // All successful if we got here.
   nReturn           = QCBOR_SUCCESS;
   *pnMajorType      = nTmpMajorType;
   *puArgument       = uArgument;
   *pnAdditionalInfo = nAdditionalInfo;

Done:
   return nReturn;
}


/*
 CBOR doesn't explicitly specify two's compliment for integers but all
 CPUs use it these days and the test vectors in the RFC are so. All
 integers in the CBOR structure are positive and the major type
 indicates positive or negative.  CBOR can express positive integers
 up to 2^x - 1 where x is the number of bits and negative integers
 down to 2^x.  Note that negative numbers can be one more away from
 zero than positive.  Stdint, as far as I can tell, uses two's
 compliment to represent negative integers.

 See http://www.unix.org/whitepapers/64bit.html for reasons int isn't
 used carefully here, and in particular why it isn't used in the interface.
 Also see
 https://stackoverflow.com/questions/17489857/why-is-int-typically-32-bit-on-64-bit-compilers

 Int is used for values that need less than 16-bits and would be subject
 to integer promotion and complaining by static analyzers.
 */
inline static QCBORError
DecodeInteger(int nMajorType, uint64_t uNumber, QCBORItem *pDecodedItem)
{
   QCBORError nReturn = QCBOR_SUCCESS;

   if(nMajorType == CBOR_MAJOR_TYPE_POSITIVE_INT) {
      if (uNumber <= INT64_MAX) {
         pDecodedItem->val.int64 = (int64_t)uNumber;
         pDecodedItem->uDataType = QCBOR_TYPE_INT64;

      } else {
         pDecodedItem->val.uint64 = uNumber;
         pDecodedItem->uDataType  = QCBOR_TYPE_UINT64;

      }
   } else {
      if(uNumber <= INT64_MAX) {
         // CBOR's representation of negative numbers lines up with the
         // two-compliment representation. A negative integer has one
         // more in range than a positive integer. INT64_MIN is
         // equal to (-INT64_MAX) - 1.
         pDecodedItem->val.int64 = (-(int64_t)uNumber) - 1;
         pDecodedItem->uDataType = QCBOR_TYPE_INT64;

      } else {
         // C can't represent a negative integer in this range
         // so it is an error.
         nReturn = QCBOR_ERR_INT_OVERFLOW;
      }
   }

   return nReturn;
}

// Make sure #define value line up as DecodeSimple counts on this.
#if QCBOR_TYPE_FALSE != CBOR_SIMPLEV_FALSE
#error QCBOR_TYPE_FALSE macro value wrong
#endif

#if QCBOR_TYPE_TRUE != CBOR_SIMPLEV_TRUE
#error QCBOR_TYPE_TRUE macro value wrong
#endif

#if QCBOR_TYPE_NULL != CBOR_SIMPLEV_NULL
#error QCBOR_TYPE_NULL macro value wrong
#endif

#if QCBOR_TYPE_UNDEF != CBOR_SIMPLEV_UNDEF
#error QCBOR_TYPE_UNDEF macro value wrong
#endif

#if QCBOR_TYPE_BREAK != CBOR_SIMPLE_BREAK
#error QCBOR_TYPE_BREAK macro value wrong
#endif

#if QCBOR_TYPE_DOUBLE != DOUBLE_PREC_FLOAT
#error QCBOR_TYPE_DOUBLE macro value wrong
#endif

#if QCBOR_TYPE_FLOAT != SINGLE_PREC_FLOAT
#error QCBOR_TYPE_FLOAT macro value wrong
#endif

/*
 Decode true, false, floats, break...
 */
inline static QCBORError
DecodeSimple(int nAdditionalInfo, uint64_t uNumber, QCBORItem *pDecodedItem)
{
   QCBORError nReturn = QCBOR_SUCCESS;

   // uAdditionalInfo is 5 bits from the initial byte compile time checks
   // above make sure uAdditionalInfo values line up with uDataType values.
   // DecodeTypeAndNumber never returns a major type > 1f so cast is safe
   pDecodedItem->uDataType = (uint8_t)nAdditionalInfo;

   switch(nAdditionalInfo) {
      // No check for ADDINFO_RESERVED1 - ADDINFO_RESERVED3 as they are
      // caught before this is called.

      case HALF_PREC_FLOAT:
         pDecodedItem->val.dfnum = IEEE754_HalfToDouble((uint16_t)uNumber);
         pDecodedItem->uDataType = QCBOR_TYPE_DOUBLE;
         break;
      case SINGLE_PREC_FLOAT:
         pDecodedItem->val.dfnum = (double)UsefulBufUtil_CopyUint32ToFloat((uint32_t)uNumber);
         pDecodedItem->uDataType = QCBOR_TYPE_DOUBLE;
         break;
      case DOUBLE_PREC_FLOAT:
         pDecodedItem->val.dfnum = UsefulBufUtil_CopyUint64ToDouble(uNumber);
         pDecodedItem->uDataType = QCBOR_TYPE_DOUBLE;
         break;

      case CBOR_SIMPLEV_FALSE: // 20
      case CBOR_SIMPLEV_TRUE:  // 21
      case CBOR_SIMPLEV_NULL:  // 22
      case CBOR_SIMPLEV_UNDEF: // 23
      case CBOR_SIMPLE_BREAK:  // 31
         break; // nothing to do

      case CBOR_SIMPLEV_ONEBYTE: // 24
         if(uNumber <= CBOR_SIMPLE_BREAK) {
            // This takes out f8 00 ... f8 1f which should be encoded as e0 … f7
            nReturn = QCBOR_ERR_BAD_TYPE_7;
            goto Done;
         }
         /* FALLTHROUGH */
         // fall through intentionally

      default: // 0-19
         pDecodedItem->uDataType   = QCBOR_TYPE_UKNOWN_SIMPLE;
         /*
          DecodeTypeAndNumber will make uNumber equal to
          uAdditionalInfo when uAdditionalInfo is < 24 This cast is
          safe because the 2, 4 and 8 byte lengths of uNumber are in
          the double/float cases above
          */
         pDecodedItem->val.uSimple = (uint8_t)uNumber;
         break;
   }

Done:
   return nReturn;
}


/*
 Decode text and byte strings. Call the string allocator if asked to.
 */
inline static QCBORError DecodeBytes(const QCORInternalAllocator *pAllocator,
                                     int nMajorType,
                                     uint64_t uStrLen,
                                     UsefulInputBuf *pUInBuf,
                                     QCBORItem *pDecodedItem)
{
   QCBORError nReturn = QCBOR_SUCCESS;

   // CBOR lengths can be 64 bits, but size_t is not 64 bits on all CPUs.
   // This check makes the casts to size_t below safe.

   // 4 bytes less than the largest sizeof() so this can be tested by
   // putting a SIZE_MAX length in the CBOR test input (no one will
   // care the limit on strings is 4 bytes shorter).
   if(uStrLen > SIZE_MAX-4) {
      nReturn = QCBOR_ERR_STRING_TOO_LONG;
      goto Done;
   }

   const UsefulBufC Bytes = UsefulInputBuf_GetUsefulBuf(pUInBuf, (size_t)uStrLen);
   if(UsefulBuf_IsNULLC(Bytes)) {
      // Failed to get the bytes for this string item
      nReturn = QCBOR_ERR_HIT_END;
      goto Done;
   }

   if(pAllocator) {
      // We are asked to use string allocator to make a copy
      UsefulBuf NewMem = StringAllocator_Allocate(pAllocator, (size_t)uStrLen);
      if(UsefulBuf_IsNULL(NewMem)) {
         nReturn = QCBOR_ERR_STRING_ALLOCATE;
         goto Done;
      }
      pDecodedItem->val.string = UsefulBuf_Copy(NewMem, Bytes);
      pDecodedItem->uDataAlloc = 1;
   } else {
      // Normal case with no string allocator
      pDecodedItem->val.string = Bytes;
   }
   const bool bIsBstr = (nMajorType == CBOR_MAJOR_TYPE_BYTE_STRING);
   // Cast because ternary operator causes promotion to integer
   pDecodedItem->uDataType = (uint8_t)(bIsBstr ? QCBOR_TYPE_BYTE_STRING
                                               : QCBOR_TYPE_TEXT_STRING);

Done:
   return nReturn;
}







// Make sure the constants align as this is assumed by
// the GetAnItem() implementation
#if QCBOR_TYPE_ARRAY != CBOR_MAJOR_TYPE_ARRAY
#error QCBOR_TYPE_ARRAY value not lined up with major type
#endif
#if QCBOR_TYPE_MAP != CBOR_MAJOR_TYPE_MAP
#error QCBOR_TYPE_MAP value not lined up with major type
#endif

/*
 This gets a single data item and decodes it including preceding
 optional tagging. This does not deal with arrays and maps and nesting
 except to decode the data item introducing them. Arrays and maps are
 handled at the next level up in GetNext().

 Errors detected here include: an array that is too long to decode,
 hit end of buffer unexpectedly, a few forms of invalid encoded CBOR
 */
static QCBORError GetNext_Item(UsefulInputBuf *pUInBuf,
                               QCBORItem *pDecodedItem,
                               const QCORInternalAllocator *pAllocator)
{
   QCBORError nReturn;

   /*
    Get the major type and the number. Number could be length of more
    bytes or the value depending on the major type nAdditionalInfo is
    an encoding of the length of the uNumber and is needed to decode
    floats and doubles
   */
   int      nMajorType;
   uint64_t uNumber;
   int      nAdditionalInfo;

   memset(pDecodedItem, 0, sizeof(QCBORItem));

   nReturn = DecodeTypeAndNumber(pUInBuf, &nMajorType, &uNumber, &nAdditionalInfo);

   // Error out here if we got into trouble on the type and number.  The
   // code after this will not work if the type and number is not good.
   if(nReturn) {
      goto Done;
   }

   // At this point the major type and the value are valid. We've got
   // the type and the number that starts every CBOR data item.
   switch (nMajorType) {
      case CBOR_MAJOR_TYPE_POSITIVE_INT: // Major type 0
      case CBOR_MAJOR_TYPE_NEGATIVE_INT: // Major type 1
         if(nAdditionalInfo == LEN_IS_INDEFINITE) {
            nReturn = QCBOR_ERR_BAD_INT;
         } else {
            nReturn = DecodeInteger(nMajorType, uNumber, pDecodedItem);
         }
         break;

      case CBOR_MAJOR_TYPE_BYTE_STRING: // Major type 2
      case CBOR_MAJOR_TYPE_TEXT_STRING: // Major type 3
         if(nAdditionalInfo == LEN_IS_INDEFINITE) {
            const bool bIsBstr = (nMajorType == CBOR_MAJOR_TYPE_BYTE_STRING);
            pDecodedItem->uDataType = (uint8_t)(bIsBstr ? QCBOR_TYPE_BYTE_STRING
                                                        : QCBOR_TYPE_TEXT_STRING);
            pDecodedItem->val.string = (UsefulBufC){NULL, SIZE_MAX};
         } else {
            nReturn = DecodeBytes(pAllocator, nMajorType, uNumber, pUInBuf, pDecodedItem);
         }
         break;

      case CBOR_MAJOR_TYPE_ARRAY: // Major type 4
      case CBOR_MAJOR_TYPE_MAP:   // Major type 5
         // Record the number of items in the array or map
         if(uNumber > QCBOR_MAX_ITEMS_IN_ARRAY) {
            nReturn = QCBOR_ERR_ARRAY_TOO_LONG;
            goto Done;
         }
         if(nAdditionalInfo == LEN_IS_INDEFINITE) {
            pDecodedItem->val.uCount = UINT16_MAX; // Indicate indefinite length
         } else {
            // type conversion OK because of check above
            pDecodedItem->val.uCount = (uint16_t)uNumber;
         }
         // C preproc #if above makes sure constants for major types align
         // DecodeTypeAndNumber never returns a major type > 7 so cast is safe
         pDecodedItem->uDataType  = (uint8_t)nMajorType;
         break;

      case CBOR_MAJOR_TYPE_OPTIONAL: // Major type 6, optional prepended tags
         if(nAdditionalInfo == LEN_IS_INDEFINITE) {
            nReturn = QCBOR_ERR_BAD_INT;
         } else {
            pDecodedItem->val.uTagV = uNumber;
            pDecodedItem->uDataType = QCBOR_TYPE_OPTTAG;
         }
         break;

      case CBOR_MAJOR_TYPE_SIMPLE:
         // Major type 7, float, double, true, false, null...
         nReturn = DecodeSimple(nAdditionalInfo, uNumber, pDecodedItem);
         break;

      default:
         // Never happens because DecodeTypeAndNumber() should never return > 7
         nReturn = QCBOR_ERR_UNSUPPORTED;
         break;
   }

Done:
   return nReturn;
}



/*
 This layer deals with indefinite length strings. It pulls all the
 individual chunk items together into one QCBORItem using the string
 allocator.

 Code Reviewers: THIS FUNCTION DOES A LITTLE POINTER MATH
 */
static inline QCBORError
GetNext_FullItem(QCBORDecodeContext *me, QCBORItem *pDecodedItem)
{
   // Stack usage; int/ptr 2 UsefulBuf 2 QCBORItem  -- 96
   QCBORError nReturn;
   const QCORInternalAllocator *pAllocator = me->StringAllocator.pfAllocator ?
                                                      &(me->StringAllocator) :
                                                      NULL;
   UsefulBufC FullString = NULLUsefulBufC;

   nReturn = GetNext_Item(&(me->InBuf),
                          pDecodedItem,
                          me->bStringAllocateAll ? pAllocator: NULL);
   if(nReturn) {
      goto Done;
   }

   // To reduce code size by removing support for indefinite length strings, the
   // code in this function from here down can be eliminated. Run tests, except
   // indefinite length string tests, to be sure all is OK if this is removed.

   // Only do indefinite length processing on strings
   if(pDecodedItem->uDataType != QCBOR_TYPE_BYTE_STRING &&
      pDecodedItem->uDataType != QCBOR_TYPE_TEXT_STRING) {
      goto Done; // no need to do any work here on non-string types
   }

   // Is this a string with an indefinite length?
   if(pDecodedItem->val.string.len != SIZE_MAX) {
      goto Done; // length is not indefinite, so no work to do here
   }

   // Can't do indefinite length strings without a string allocator
   if(pAllocator == NULL) {
      nReturn = QCBOR_ERR_NO_STRING_ALLOCATOR;
      goto Done;
   }

   // There is an indefinite length string to work on...
   // Track which type of string it is
   const uint8_t uStringType = pDecodedItem->uDataType;

   // Loop getting chunk of indefinite string
   for(;;) {
      // Get item for next chunk
      QCBORItem StringChunkItem;
      // NULL passed to never string alloc chunk of indefinite length strings
      nReturn = GetNext_Item(&(me->InBuf), &StringChunkItem, NULL);
      if(nReturn) {
         break;  // Error getting the next chunk
      }

      // See if it is a marker at end of indefinite length string
      if(StringChunkItem.uDataType == QCBOR_TYPE_BREAK) {
         // String is complete
         pDecodedItem->val.string = FullString;
         pDecodedItem->uDataAlloc = 1;
         break;
      }

      // Match data type of chunk to type at beginning.
      // Also catches error of other non-string types that don't belong.
      // Also catches indefinite length strings inside indefinite length strings
      if(StringChunkItem.uDataType != uStringType ||
         StringChunkItem.val.string.len == SIZE_MAX) {
         nReturn = QCBOR_ERR_INDEFINITE_STRING_CHUNK;
         break;
      }

      // Alloc new buffer or expand previously allocated buffer so it can fit
      // The first time throurgh FullString.ptr is NULL and this is
      // equivalent to StringAllocator_Allocate()
      UsefulBuf NewMem = StringAllocator_Reallocate(pAllocator,
                                                    UNCONST_POINTER(FullString.ptr),
                                                    FullString.len + StringChunkItem.val.string.len);

      if(UsefulBuf_IsNULL(NewMem)) {
         // Allocation of memory for the string failed
         nReturn = QCBOR_ERR_STRING_ALLOCATE;
         break;
      }

      // Copy new string chunk at the end of string so far.
      FullString = UsefulBuf_CopyOffset(NewMem, FullString.len, StringChunkItem.val.string);
   }

Done:
   if(nReturn != QCBOR_SUCCESS && !UsefulBuf_IsNULLC(FullString)) {
      // Getting the item failed, clean up the allocated memory
      StringAllocator_Free(pAllocator, UNCONST_POINTER(FullString.ptr));
   }

   return nReturn;
}


/*
 Gets all optional tag data items preceding a data item that is not an
 optional tag and records them as bits in the tag map.
 */
static QCBORError
GetNext_TaggedItem(QCBORDecodeContext *me,
                   QCBORItem *pDecodedItem,
                   QCBORTagListOut *pTags)
{
   // Stack usage: int/ptr: 3 -- 24
   QCBORError nReturn;
   uint64_t  uTagBits = 0;
   if(pTags) {
      pTags->uNumUsed = 0;
   }

   // Loop fetching items until the item fetched is not a tag
   for(;;) {
      nReturn = GetNext_FullItem(me, pDecodedItem);
      if(nReturn) {
         goto Done; // Error out of the loop
      }

      if(pDecodedItem->uDataType != QCBOR_TYPE_OPTTAG) {
         // Successful exit from loop; maybe got some tags, maybe not
         pDecodedItem->uTagBits = uTagBits;
         break;
      }

      uint8_t uTagBitIndex;
      // Tag was mapped, tag was not mapped, error with tag list
      switch(TagMapper_Lookup(me->pCallerConfiguredTagList, pDecodedItem->val.uTagV, &uTagBitIndex)) {

         case QCBOR_SUCCESS:
            // Successfully mapped the tag
            uTagBits |= 0x01ULL << uTagBitIndex;
            break;

         case QCBOR_ERR_BAD_OPT_TAG:
            // Tag is not recognized. Do nothing
            break;

         default:
            // Error Condition
            goto Done;
      }

      if(pTags) {
         // Caller wants all tags recorded in the provided buffer
         if(pTags->uNumUsed >= pTags->uNumAllocated) {
            nReturn = QCBOR_ERR_TOO_MANY_TAGS;
            goto Done;
         }
         pTags->puTags[pTags->uNumUsed] = pDecodedItem->val.uTagV;
         pTags->uNumUsed++;
      }
   }

Done:
   return nReturn;
}


/*
 This layer takes care of map entries. It combines the label and data
 items into one QCBORItem.
 */
static inline QCBORError
GetNext_MapEntry(QCBORDecodeContext *me,
                 QCBORItem *pDecodedItem,
                 QCBORTagListOut *pTags)
{
   // Stack use: int/ptr 1, QCBORItem  -- 56
   QCBORError nReturn = GetNext_TaggedItem(me, pDecodedItem, pTags);
   if(nReturn)
      goto Done;

   if(pDecodedItem->uDataType == QCBOR_TYPE_BREAK) {
      // Break can't be a map entry
      goto Done;
   }

   if(me->uDecodeMode != QCBOR_DECODE_MODE_MAP_AS_ARRAY) {
      // In a map and caller wants maps decoded, not treated as arrays

      if(DecodeNesting_TypeIsMap(&(me->nesting))) {
         // If in a map and the right decoding mode, get the label

         // Save label in pDecodedItem and get the next which will
         // be the real data
         QCBORItem LabelItem = *pDecodedItem;
         nReturn = GetNext_TaggedItem(me, pDecodedItem, pTags);
         if(nReturn)
            goto Done;

         pDecodedItem->uLabelAlloc = LabelItem.uDataAlloc;

         if(LabelItem.uDataType == QCBOR_TYPE_TEXT_STRING) {
            // strings are always good labels
            pDecodedItem->label.string = LabelItem.val.string;
            pDecodedItem->uLabelType = QCBOR_TYPE_TEXT_STRING;
         } else if (QCBOR_DECODE_MODE_MAP_STRINGS_ONLY == me->uDecodeMode) {
            // It's not a string and we only want strings
            nReturn = QCBOR_ERR_MAP_LABEL_TYPE;
            goto Done;
         } else if(LabelItem.uDataType == QCBOR_TYPE_INT64) {
            pDecodedItem->label.int64 = LabelItem.val.int64;
            pDecodedItem->uLabelType = QCBOR_TYPE_INT64;
         } else if(LabelItem.uDataType == QCBOR_TYPE_UINT64) {
            pDecodedItem->label.uint64 = LabelItem.val.uint64;
            pDecodedItem->uLabelType = QCBOR_TYPE_UINT64;
         } else if(LabelItem.uDataType == QCBOR_TYPE_BYTE_STRING) {
            pDecodedItem->label.string = LabelItem.val.string;
            pDecodedItem->uLabelAlloc = LabelItem.uDataAlloc;
            pDecodedItem->uLabelType = QCBOR_TYPE_BYTE_STRING;
         } else {
            // label is not an int or a string. It is an arrray
            // or a float or such and this implementation doesn't handle that.
            // Also, tags on labels are ignored.
            nReturn = QCBOR_ERR_MAP_LABEL_TYPE;
            goto Done;
         }
      }
   } else {
      if(pDecodedItem->uDataType == QCBOR_TYPE_MAP) {
         // Decoding a map as an array
         pDecodedItem->uDataType = QCBOR_TYPE_MAP_AS_ARRAY;
         pDecodedItem->val.uCount *= 2;
      }
   }

Done:
   return nReturn;
}


/*
 Public function, see header qcbor.h file
 */
QCBORError QCBORDecode_GetNextMapOrArray(QCBORDecodeContext *me,
                                         QCBORItem *pDecodedItem,
                                         QCBORTagListOut *pTags)
{
   // Stack ptr/int: 2, QCBORItem : 64

   // The public entry point for fetching and parsing the next QCBORItem.
   // All the CBOR parsing work is here and in subordinate calls.
   QCBORError nReturn;

   // Check if there are an
   if(UsefulInputBuf_BytesUnconsumed(&(me->InBuf)) == 0 && !DecodeNesting_IsNested(&(me->nesting))) {
      nReturn = QCBOR_ERR_NO_MORE_ITEMS;
      goto Done;
   }

   nReturn = GetNext_MapEntry(me, pDecodedItem, pTags);
   if(nReturn) {
      goto Done;
   }

   // Break ending arrays/maps are always processed at the end of this function.
   // They should never show up here.
   if(pDecodedItem->uDataType == QCBOR_TYPE_BREAK) {
      nReturn = QCBOR_ERR_BAD_BREAK;
      goto Done;
   }

   // Record the nesting level for this data item before processing any of
   // decrementing and descending.
   pDecodedItem->uNestingLevel = DecodeNesting_GetLevel(&(me->nesting));

   // Process the item just received for descent or decrement, and
   // ascent if decrements are enough to close out a definite length array/map
   if(IsMapOrArray(pDecodedItem->uDataType)) {
      // If the new item is array or map, the nesting level descends
      nReturn = DecodeNesting_Descend(&(me->nesting), pDecodedItem);
      // Maps and arrays do count in as items in the map/array that encloses
      // them so a decrement needs to be done for them too, but that is done
      // only when all the items in them have been processed, not when they
      // are opened with the exception of an empty map or array.
       if(pDecodedItem->val.uCount == 0) {
           DecodeNesting_DecrementCount(&(me->nesting));
       }
   } else {
      // Decrement the count of items in the enclosing map/array
      // If the count in the enclosing map/array goes to zero, that
      // triggers a decrement in the map/array above that and
      // an ascend in nesting level.
      DecodeNesting_DecrementCount(&(me->nesting));
   }
   if(nReturn) {
      goto Done;
   }

   // For indefinite length maps/arrays, looking at any and
   // all breaks that might terminate them. The equivalent
   // for definite length maps/arrays happens in
   // DecodeNesting_DecrementCount().
   if(DecodeNesting_IsNested(&(me->nesting)) && DecodeNesting_IsIndefiniteLength(&(me->nesting))) {
      while(UsefulInputBuf_BytesUnconsumed(&(me->InBuf))) {
         // Peek forward one item to see if it is a break.
         QCBORItem Peek;
         size_t uPeek = UsefulInputBuf_Tell(&(me->InBuf));
         nReturn = GetNext_Item(&(me->InBuf), &Peek, NULL);
         if(nReturn) {
            goto Done;
         }
         if(Peek.uDataType != QCBOR_TYPE_BREAK) {
            // It is not a break, rewind so it can be processed normally.
            UsefulInputBuf_Seek(&(me->InBuf), uPeek);
            break;
         }
         // It is a break. Ascend one nesting level.
         // The break is consumed.
         nReturn = DecodeNesting_BreakAscend(&(me->nesting));
         if(nReturn) {
            // break occured outside of an indefinite length array/map
            goto Done;
         }
      }
   }

   // Tell the caller what level is next. This tells them what maps/arrays
   // were closed out and makes it possible for them to reconstruct
   // the tree with just the information returned by GetNext
   pDecodedItem->uNextNestLevel = DecodeNesting_GetLevel(&(me->nesting));

Done:
   if(nReturn != QCBOR_SUCCESS) {
      // Make sure uDataType and uLabelType are QCBOR_TYPE_NONE
      memset(pDecodedItem, 0, sizeof(QCBORItem));
   }
   return nReturn;
}


/*
 Mostly just assign the right data type for the date string.
 */
inline static QCBORError DecodeDateString(QCBORItem *pDecodedItem)
{
   // Stack Use: UsefulBuf 1 16
   if(pDecodedItem->uDataType != QCBOR_TYPE_TEXT_STRING) {
      return QCBOR_ERR_BAD_OPT_TAG;
   }

   const UsefulBufC Temp        = pDecodedItem->val.string;
   pDecodedItem->val.dateString = Temp;
   pDecodedItem->uDataType      = QCBOR_TYPE_DATE_STRING;
   return QCBOR_SUCCESS;
}


/*
 Mostly just assign the right data type for the bignum.
 */
inline static QCBORError DecodeBigNum(QCBORItem *pDecodedItem)
{
   // Stack Use: UsefulBuf 1  -- 16
   if(pDecodedItem->uDataType != QCBOR_TYPE_BYTE_STRING) {
      return QCBOR_ERR_BAD_OPT_TAG;
   }
   const UsefulBufC Temp    = pDecodedItem->val.string;
   pDecodedItem->val.bigNum = Temp;
   const bool bIsPosBigNum = (bool)(pDecodedItem->uTagBits & QCBOR_TAGFLAG_POS_BIGNUM);
   pDecodedItem->uDataType  = (uint8_t)(bIsPosBigNum ? QCBOR_TYPE_POSBIGNUM
                                                     : QCBOR_TYPE_NEGBIGNUM);
   return QCBOR_SUCCESS;
}


/*
 The epoch formatted date. Turns lots of different forms of encoding
 date into uniform one
 */
static QCBORError DecodeDateEpoch(QCBORItem *pDecodedItem)
{
   // Stack usage: 1
   QCBORError nReturn = QCBOR_SUCCESS;

   pDecodedItem->val.epochDate.fSecondsFraction = 0;

   switch (pDecodedItem->uDataType) {

      case QCBOR_TYPE_INT64:
         pDecodedItem->val.epochDate.nSeconds = pDecodedItem->val.int64;
         break;

      case QCBOR_TYPE_UINT64:
         if(pDecodedItem->val.uint64 > INT64_MAX) {
            nReturn = QCBOR_ERR_DATE_OVERFLOW;
            goto Done;
         }
         pDecodedItem->val.epochDate.nSeconds = (int64_t)pDecodedItem->val.uint64;
         break;

      case QCBOR_TYPE_DOUBLE:
      {
         // This comparison needs to be done as a float before
         // conversion to an int64_t to be able to detect doubles
         // that are too large to fit into an int64_t.  A double
         // has 52 bits of preceision. An int64_t has 63. Casting
         // INT64_MAX to a double actually causes a round up which
         // is bad and wrong for the comparison because it will
         // allow conversion of doubles that can't fit into a
         // uint64_t.  To remedy this INT64_MAX - 0x7ff is used as
         // the cutoff point as if that rounds up in conversion to
         // double it will still be less than INT64_MAX. 0x7ff is
         // picked because it has 11 bits set.
         //
         // INT64_MAX seconds is on the order of 10 billion years,
         // and the earth is less than 5 billion years old, so for
         // most uses this conversion error won't occur even though
         // doubles can go much larger.
         //
         // Without the 0x7ff there is a ~30 minute range of time
         // values 10 billion years in the past and in the future
         // where this this code would go wrong.
         const double d = pDecodedItem->val.dfnum;
         if(d > (double)(INT64_MAX - 0x7ff)) {
            nReturn = QCBOR_ERR_DATE_OVERFLOW;
            goto Done;
         }
         pDecodedItem->val.epochDate.nSeconds = (int64_t)d;
         pDecodedItem->val.epochDate.fSecondsFraction = d - (double)pDecodedItem->val.epochDate.nSeconds;
      }
         break;

      default:
         nReturn = QCBOR_ERR_BAD_OPT_TAG;
         goto Done;
   }
   pDecodedItem->uDataType = QCBOR_TYPE_DATE_EPOCH;

Done:
   return nReturn;
}


#ifndef QCBOR_CONFIG_DISABLE_EXP_AND_MANTISSA
/*
 Decode decimal fractions and big floats.

 When called pDecodedItem must be the array that is tagged as a big
 float or decimal fraction, the array that has the two members, the
 exponent and mantissa.

 This will fetch and decode the exponent and mantissa and put the
 result back into pDecodedItem.
 */
inline static QCBORError
QCBORDecode_MantissaAndExponent(QCBORDecodeContext *me, QCBORItem *pDecodedItem)
{
   QCBORError nReturn;

   // --- Make sure it is an array; track nesting level of members ---
   if(pDecodedItem->uDataType != QCBOR_TYPE_ARRAY) {
      nReturn = QCBOR_ERR_BAD_EXP_AND_MANTISSA;
      goto Done;
   }

   // A check for pDecodedItem->val.uCount == 2 would work for
   // definite length arrays, but not for indefnite.  Instead remember
   // the nesting level the two integers must be at, which is one
   // deeper than that of the array.
   const int nNestLevel = pDecodedItem->uNestingLevel + 1;

   // --- Is it a decimal fraction or a bigfloat? ---
   const bool bIsTaggedDecimalFraction = QCBORDecode_IsTagged(me, pDecodedItem, CBOR_TAG_DECIMAL_FRACTION);
   pDecodedItem->uDataType = bIsTaggedDecimalFraction ? QCBOR_TYPE_DECIMAL_FRACTION : QCBOR_TYPE_BIGFLOAT;

   // --- Get the exponent ---
   QCBORItem exponentItem;
   nReturn = QCBORDecode_GetNextMapOrArray(me, &exponentItem, NULL);
   if(nReturn != QCBOR_SUCCESS) {
      goto Done;
   }
   if(exponentItem.uNestingLevel != nNestLevel) {
      // Array is empty or a map/array encountered when expecting an int
      nReturn = QCBOR_ERR_BAD_EXP_AND_MANTISSA;
      goto Done;
   }
   if(exponentItem.uDataType == QCBOR_TYPE_INT64) {
     // Data arriving as an unsigned int < INT64_MAX has been converted
     // to QCBOR_TYPE_INT64 and thus handled here. This is also means
     // that the only data arriving here of type QCBOR_TYPE_UINT64 data
     // will be too large for this to handle and thus an error that will
     // get handled in the next else.
     pDecodedItem->val.expAndMantissa.nExponent = exponentItem.val.int64;
   } else {
      // Wrong type of exponent or a QCBOR_TYPE_UINT64 > INT64_MAX
      nReturn = QCBOR_ERR_BAD_EXP_AND_MANTISSA;
      goto Done;
   }

   // --- Get the mantissa ---
   QCBORItem mantissaItem;
   nReturn = QCBORDecode_GetNextWithTags(me, &mantissaItem, NULL);
   if(nReturn != QCBOR_SUCCESS) {
      goto Done;
   }
   if(mantissaItem.uNestingLevel != nNestLevel) {
      // Mantissa missing or map/array encountered when expecting number
      nReturn = QCBOR_ERR_BAD_EXP_AND_MANTISSA;
      goto Done;
   }
   if(mantissaItem.uDataType == QCBOR_TYPE_INT64) {
      // Data arriving as an unsigned int < INT64_MAX has been converted
      // to QCBOR_TYPE_INT64 and thus handled here. This is also means
      // that the only data arriving here of type QCBOR_TYPE_UINT64 data
      // will be too large for this to handle and thus an error that
      // will get handled in an else below.
      pDecodedItem->val.expAndMantissa.Mantissa.nInt = mantissaItem.val.int64;
   }  else if(mantissaItem.uDataType == QCBOR_TYPE_POSBIGNUM || mantissaItem.uDataType == QCBOR_TYPE_NEGBIGNUM) {
      // Got a good big num mantissa
      pDecodedItem->val.expAndMantissa.Mantissa.bigNum = mantissaItem.val.bigNum;
      // Depends on numbering of QCBOR_TYPE_XXX
      pDecodedItem->uDataType = (uint8_t)(pDecodedItem->uDataType +
                                          mantissaItem.uDataType - QCBOR_TYPE_POSBIGNUM +
                                          1);
   } else {
      // Wrong type of mantissa or a QCBOR_TYPE_UINT64 > INT64_MAX
      nReturn = QCBOR_ERR_BAD_EXP_AND_MANTISSA;
      goto Done;
   }

   // --- Check that array only has the two numbers ---
   if(mantissaItem.uNextNestLevel == nNestLevel) {
      // Extra items in the decimal fraction / big num
      nReturn = QCBOR_ERR_BAD_EXP_AND_MANTISSA;
      goto Done;
   }

Done:

  return nReturn;
}
#endif /* QCBOR_CONFIG_DISABLE_EXP_AND_MANTISSA */


/*
 Public function, see header qcbor.h file
 */
QCBORError
QCBORDecode_GetNextWithTags(QCBORDecodeContext *me,
                            QCBORItem *pDecodedItem,
                            QCBORTagListOut *pTags)
{
   QCBORError nReturn;

   nReturn = QCBORDecode_GetNextMapOrArray(me, pDecodedItem, pTags);
   if(nReturn != QCBOR_SUCCESS) {
      goto Done;
   }

#ifndef QCBOR_CONFIG_DISABLE_EXP_AND_MANTISSA
#define TAG_MAPPER_FIRST_XXX TAG_MAPPER_FIRST_SIX
#else
#define TAG_MAPPER_FIRST_XXX TAG_MAPPER_FIRST_FOUR
#endif

   // Only pay attention to tags this code knows how to decode.
   switch(pDecodedItem->uTagBits & TAG_MAPPER_FIRST_XXX) {
      case 0:
         // No tags at all or none we know about. Nothing to do.
         // This is the pass-through path of this function
         // that will mostly be taken when decoding any item.
         break;

      case QCBOR_TAGFLAG_DATE_STRING:
         nReturn = DecodeDateString(pDecodedItem);
         break;

      case QCBOR_TAGFLAG_DATE_EPOCH:
         nReturn = DecodeDateEpoch(pDecodedItem);
         break;

      case QCBOR_TAGFLAG_POS_BIGNUM:
      case QCBOR_TAGFLAG_NEG_BIGNUM:
         nReturn = DecodeBigNum(pDecodedItem);
         break;

#ifndef QCBOR_CONFIG_DISABLE_EXP_AND_MANTISSA
      case QCBOR_TAGFLAG_DECIMAL_FRACTION:
      case QCBOR_TAGFLAG_BIGFLOAT:
         // For aggregate tagged types, what goes into pTags is only collected
         // from the surrounding data item, not the contents, so pTags is not
         // passed on here.

         nReturn = QCBORDecode_MantissaAndExponent(me, pDecodedItem);
         break;
#endif /* QCBOR_CONFIG_DISABLE_EXP_AND_MANTISSA */

      default:
         // Encountering some mixed-up CBOR like something that
         // is tagged as both a string and integer date.
         nReturn = QCBOR_ERR_BAD_OPT_TAG;
   }

Done:
   if(nReturn != QCBOR_SUCCESS) {
      pDecodedItem->uDataType  = QCBOR_TYPE_NONE;
      pDecodedItem->uLabelType = QCBOR_TYPE_NONE;
   }
   return nReturn;
}


/*
 Public function, see header qcbor.h file
 */
QCBORError QCBORDecode_GetNext(QCBORDecodeContext *me, QCBORItem *pDecodedItem)
{
   return QCBORDecode_GetNextWithTags(me, pDecodedItem, NULL);
}


/*
 Decoding items is done in 5 layered functions, one calling the
 next one down. If a layer has no work to do for a particular item
 it returns quickly.

 - QCBORDecode_GetNext, GetNextWithTags -- The top layer processes
 tagged data items, turning them into the local C representation.
 For the most simple it is just associating a QCBOR_TYPE with the data. For
 the complex ones that an aggregate of data items, there is some further
 decoding and a little bit of recursion.

 - QCBORDecode_GetNextMapOrArray - This manages the beginnings and
 ends of maps and arrays. It tracks descending into and ascending
 out of maps/arrays. It processes all breaks that terminate
 maps and arrays.

 - GetNext_MapEntry -- This handles the combining of two
 items, the label and the data, that make up a map entry.
 It only does work on maps. It combines the label and data
 items into one labeled item.

 - GetNext_TaggedItem -- This decodes type 6 tagging. It turns the
 tags into bit flags associated with the data item. No actual decoding
 of the contents of the tagged item is performed here.

 - GetNext_FullItem -- This assembles the sub-items that make up
 an indefinte length string into one string item. It uses the
 string allocater to create contiguous space for the item. It
 processes all breaks that are part of indefinite length strings.

 - GetNext_Item -- This decodes the atomic data items in CBOR. Each
 atomic data item has a "major type", an integer "argument" and optionally
 some content. For text and byte strings, the content is the bytes
 that make up the string. These are the smallest data items that are
 considered to be well-formed.  The content may also be other data items in
 the case of aggregate types. They are not handled in this layer.

 Roughly this takes 300 bytes of stack for vars. Need to
 evaluate this more carefully and correctly.

 */


/*
 Public function, see header qcbor.h file
 */
int QCBORDecode_IsTagged(QCBORDecodeContext *me,
                         const QCBORItem *pItem,
                         uint64_t uTag)
{
   const QCBORTagListIn *pCallerConfiguredTagMap = me->pCallerConfiguredTagList;

   uint8_t uTagBitIndex;
   // Do not care about errors in pCallerConfiguredTagMap here. They are
   // caught during GetNext() before this is called.
   if(TagMapper_Lookup(pCallerConfiguredTagMap, uTag, &uTagBitIndex)) {
      return 0;
   }

   const uint64_t uTagBit = 0x01ULL << uTagBitIndex;
   return (uTagBit & pItem->uTagBits) != 0;
}


/*
 Public function, see header qcbor.h file
 */
QCBORError QCBORDecode_Finish(QCBORDecodeContext *me)
{
   QCBORError nReturn = QCBOR_SUCCESS;

   // Error out if all the maps/arrays are not closed out
   if(DecodeNesting_IsNested(&(me->nesting))) {
      nReturn = QCBOR_ERR_ARRAY_OR_MAP_STILL_OPEN;
      goto Done;
   }

   // Error out if not all the bytes are consumed
   if(UsefulInputBuf_BytesUnconsumed(&(me->InBuf))) {
      nReturn = QCBOR_ERR_EXTRA_BYTES;
   }

Done:
   // Call the destructor for the string allocator if there is one.
   // Always called, even if there are errors; always have to clean up
   StringAllocator_Destruct(&(me->StringAllocator));

   return nReturn;
}



/*

Decoder errors handled in this file

 - Hit end of input before it was expected while decoding type and
   number QCBOR_ERR_HIT_END

 - negative integer that is too large for C QCBOR_ERR_INT_OVERFLOW

 - Hit end of input while decoding a text or byte string
   QCBOR_ERR_HIT_END

 - Encountered conflicting tags -- e.g., an item is tagged both a date
   string and an epoch date QCBOR_ERR_UNSUPPORTED

 - Encontered an array or mapp that has too many items
   QCBOR_ERR_ARRAY_TOO_LONG

 - Encountered array/map nesting that is too deep
   QCBOR_ERR_ARRAY_NESTING_TOO_DEEP

 - An epoch date > INT64_MAX or < INT64_MIN was encountered
   QCBOR_ERR_DATE_OVERFLOW

 - The type of a map label is not a string or int
   QCBOR_ERR_MAP_LABEL_TYPE

 - Hit end with arrays or maps still open -- QCBOR_ERR_EXTRA_BYTES

 */




/* ===========================================================================
   MemPool -- BUILT-IN SIMPLE STRING ALLOCATOR

   This implements a simple sting allocator for indefinite length
   strings that can be enabled by calling QCBORDecode_SetMemPool(). It
   implements the function type QCBORStringAllocate and allows easy
   use of it.

   This particular allocator is built-in for convenience. The caller
   can implement their own.  All of this following code will get
   dead-stripped if QCBORDecode_SetMemPool() is not called.

   This is a very primitive memory allocator. It does not track
   individual allocations, only a high-water mark. A free or
   reallocation must be of the last chunk allocated.

   The size of the pool and offset to free memory are packed into the
   first 8 bytes of the memory pool so we don't have to keep them in
   the decode context. Since the address of the pool may not be
   aligned, they have to be packed and unpacked as if they were
   serialized data of the wire or such.

   The sizes packed in are uint32_t to be the same on all CPU types
   and simplify the code.
   ========================================================================== */


static inline int
MemPool_Unpack(const void *pMem, uint32_t *puPoolSize, uint32_t *puFreeOffset)
{
   // Use of UsefulInputBuf is overkill, but it is convenient.
   UsefulInputBuf UIB;

   // Just assume the size here. It was checked during SetUp so
   // the assumption is safe.
   UsefulInputBuf_Init(&UIB, (UsefulBufC){pMem, QCBOR_DECODE_MIN_MEM_POOL_SIZE});
   *puPoolSize     = UsefulInputBuf_GetUint32(&UIB);
   *puFreeOffset   = UsefulInputBuf_GetUint32(&UIB);
   return UsefulInputBuf_GetError(&UIB);
}


static inline int
MemPool_Pack(UsefulBuf Pool, uint32_t uFreeOffset)
{
   // Use of UsefulOutBuf is overkill, but convenient. The
   // length check performed here is useful.
   UsefulOutBuf UOB;

   UsefulOutBuf_Init(&UOB, Pool);
   UsefulOutBuf_AppendUint32(&UOB, (uint32_t)Pool.len); // size of pool
   UsefulOutBuf_AppendUint32(&UOB, uFreeOffset); // first free position
   return UsefulOutBuf_GetError(&UOB);
}


/*
 Internal function for an allocation, reallocation free and destuct.

 Having only one function rather than one each per mode saves space in
 QCBORDecodeContext.

 Code Reviewers: THIS FUNCTION DOES POINTER MATH
 */
static UsefulBuf
MemPool_Function(void *pPool, void *pMem, size_t uNewSize)
{
   UsefulBuf ReturnValue = NULLUsefulBuf;

   uint32_t uPoolSize;
   uint32_t uFreeOffset;

   if(uNewSize > UINT32_MAX) {
      // This allocator is only good up to 4GB.  This check should
      // optimize out if sizeof(size_t) == sizeof(uint32_t)
      goto Done;
   }
   const uint32_t uNewSize32 = (uint32_t)uNewSize;

   if(MemPool_Unpack(pPool, &uPoolSize, &uFreeOffset)) {
      goto Done;
   }

   if(uNewSize) {
      if(pMem) {
         // REALLOCATION MODE
         // Calculate pointer to the end of the memory pool.  It is
         // assumed that pPool + uPoolSize won't wrap around by
         // assuming the caller won't pass a pool buffer in that is
         // not in legitimate memory space.
         const void *pPoolEnd = (uint8_t *)pPool + uPoolSize;

         // Check that the pointer for reallocation is in the range of the
         // pool. This also makes sure that pointer math further down
         // doesn't wrap under or over.
         if(pMem >= pPool && pMem < pPoolEnd) {
            // Offset to start of chunk for reallocation. This won't
            // wrap under because of check that pMem >= pPool.  Cast
            // is safe because the pool is always less than UINT32_MAX
            // because of check in QCBORDecode_SetMemPool().
            const uint32_t uMemOffset = (uint32_t)((uint8_t *)pMem - (uint8_t *)pPool);

            // Check to see if the allocation will fit. uPoolSize -
            // uMemOffset will not wrap under because of check that
            // pMem is in the range of the uPoolSize by check above.
            if(uNewSize <= uPoolSize - uMemOffset) {
               ReturnValue.ptr = pMem;
               ReturnValue.len = uNewSize;

               // Addition won't wrap around over because uNewSize was
               // checked to be sure it is less than the pool size.
               uFreeOffset = uMemOffset + uNewSize32;
            }
         }
      } else {
         // ALLOCATION MODE
         // uPoolSize - uFreeOffset will not underflow because this
         // pool implementation makes sure uFreeOffset is always
         // smaller than uPoolSize through this check here and
         // reallocation case.
         if(uNewSize <= uPoolSize - uFreeOffset) {
            ReturnValue.len = uNewSize;
            ReturnValue.ptr = (uint8_t *)pPool + uFreeOffset;
            uFreeOffset    += (uint32_t)uNewSize;
         }
      }
   } else {
      if(pMem) {
         // FREE MODE
         // Cast is safe because of limit on pool size in
         // QCBORDecode_SetMemPool()
         uFreeOffset = (uint32_t)((uint8_t *)pMem - (uint8_t *)pPool);
      } else {
         // DESTRUCT MODE
         // Nothing to do for this allocator
      }
   }

   UsefulBuf Pool = {pPool, uPoolSize};
   MemPool_Pack(Pool, uFreeOffset);

Done:
   return ReturnValue;
}


/*
 Public function, see header qcbor.h file
 */
QCBORError QCBORDecode_SetMemPool(QCBORDecodeContext *pMe,
                                  UsefulBuf Pool,
                                  bool bAllStrings)
{
   // The pool size and free mem offset are packed into the beginning
   // of the pool memory. This compile time check make sure the
   // constant in the header is correct.  This check should optimize
   // down to nothing.
   if(QCBOR_DECODE_MIN_MEM_POOL_SIZE < 2 * sizeof(uint32_t)) {
      return QCBOR_ERR_BUFFER_TOO_SMALL;
   }

   // The pool size and free offset packed in to the beginning of pool
   // memory are only 32-bits. This check will optimize out on 32-bit
   // machines.
   if(Pool.len > UINT32_MAX) {
      return QCBOR_ERR_BUFFER_TOO_LARGE;
   }

   // This checks that the pool buffer given is big enough.
   if(MemPool_Pack(Pool, QCBOR_DECODE_MIN_MEM_POOL_SIZE)) {
      return QCBOR_ERR_BUFFER_TOO_SMALL;
   }

   pMe->StringAllocator.pfAllocator    = MemPool_Function;
   pMe->StringAllocator.pAllocateCxt  = Pool.ptr;
   pMe->bStringAllocateAll             = bAllStrings;

   return QCBOR_SUCCESS;
}