michael@0: /*
michael@0: ******************************************************************************
michael@0: *
michael@0: *   Copyright (C) 2001-2012, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: *
michael@0: ******************************************************************************
michael@0: *   file name:  utrie.cpp
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: *   created on: 2001oct20
michael@0: *   created by: Markus W. Scherer
michael@0: *
michael@0: *   This is a common implementation of a "folded" trie.
michael@0: *   It is a kind of compressed, serializable table of 16- or 32-bit values associated with
michael@0: *   Unicode code points (0..0x10ffff).
michael@0: */
michael@0: 
michael@0: #ifdef UTRIE_DEBUG
michael@0: #   include <stdio.h>
michael@0: #endif
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: #include "cmemory.h"
michael@0: #include "utrie.h"
michael@0: 
michael@0: /* miscellaneous ------------------------------------------------------------ */
michael@0: 
michael@0: #undef ABS
michael@0: #define ABS(x) ((x)>=0 ? (x) : -(x))
michael@0: 
michael@0: static inline UBool
michael@0: equal_uint32(const uint32_t *s, const uint32_t *t, int32_t length) {
michael@0:     while(length>0 && *s==*t) {
michael@0:         ++s;
michael@0:         ++t;
michael@0:         --length;
michael@0:     }
michael@0:     return (UBool)(length==0);
michael@0: }
michael@0: 
michael@0: /* Building a trie ----------------------------------------------------------*/
michael@0: 
michael@0: U_CAPI UNewTrie * U_EXPORT2
michael@0: utrie_open(UNewTrie *fillIn,
michael@0:            uint32_t *aliasData, int32_t maxDataLength,
michael@0:            uint32_t initialValue, uint32_t leadUnitValue,
michael@0:            UBool latin1Linear) {
michael@0:     UNewTrie *trie;
michael@0:     int32_t i, j;
michael@0: 
michael@0:     if( maxDataLength<UTRIE_DATA_BLOCK_LENGTH ||
michael@0:         (latin1Linear && maxDataLength<1024)
michael@0:     ) {
michael@0:         return NULL;
michael@0:     }
michael@0: 
michael@0:     if(fillIn!=NULL) {
michael@0:         trie=fillIn;
michael@0:     } else {
michael@0:         trie=(UNewTrie *)uprv_malloc(sizeof(UNewTrie));
michael@0:         if(trie==NULL) {
michael@0:             return NULL;
michael@0:         }
michael@0:     }
michael@0:     uprv_memset(trie, 0, sizeof(UNewTrie));
michael@0:     trie->isAllocated= (UBool)(fillIn==NULL);
michael@0: 
michael@0:     if(aliasData!=NULL) {
michael@0:         trie->data=aliasData;
michael@0:         trie->isDataAllocated=FALSE;
michael@0:     } else {
michael@0:         trie->data=(uint32_t *)uprv_malloc(maxDataLength*4);
michael@0:         if(trie->data==NULL) {
michael@0:             uprv_free(trie);
michael@0:             return NULL;
michael@0:         }
michael@0:         trie->isDataAllocated=TRUE;
michael@0:     }
michael@0: 
michael@0:     /* preallocate and reset the first data block (block index 0) */
michael@0:     j=UTRIE_DATA_BLOCK_LENGTH;
michael@0: 
michael@0:     if(latin1Linear) {
michael@0:         /* preallocate and reset the first block (number 0) and Latin-1 (U+0000..U+00ff) after that */
michael@0:         /* made sure above that maxDataLength>=1024 */
michael@0: 
michael@0:         /* set indexes to point to consecutive data blocks */
michael@0:         i=0;
michael@0:         do {
michael@0:             /* do this at least for trie->index[0] even if that block is only partly used for Latin-1 */
michael@0:             trie->index[i++]=j;
michael@0:             j+=UTRIE_DATA_BLOCK_LENGTH;
michael@0:         } while(i<(256>>UTRIE_SHIFT));
michael@0:     }
michael@0: 
michael@0:     /* reset the initially allocated blocks to the initial value */
michael@0:     trie->dataLength=j;
michael@0:     while(j>0) {
michael@0:         trie->data[--j]=initialValue;
michael@0:     }
michael@0: 
michael@0:     trie->leadUnitValue=leadUnitValue;
michael@0:     trie->indexLength=UTRIE_MAX_INDEX_LENGTH;
michael@0:     trie->dataCapacity=maxDataLength;
michael@0:     trie->isLatin1Linear=latin1Linear;
michael@0:     trie->isCompacted=FALSE;
michael@0:     return trie;
michael@0: }
michael@0: 
michael@0: U_CAPI UNewTrie * U_EXPORT2
michael@0: utrie_clone(UNewTrie *fillIn, const UNewTrie *other, uint32_t *aliasData, int32_t aliasDataCapacity) {
michael@0:     UNewTrie *trie;
michael@0:     UBool isDataAllocated;
michael@0: 
michael@0:     /* do not clone if other is not valid or already compacted */
michael@0:     if(other==NULL || other->data==NULL || other->isCompacted) {
michael@0:         return NULL;
michael@0:     }
michael@0: 
michael@0:     /* clone data */
michael@0:     if(aliasData!=NULL && aliasDataCapacity>=other->dataCapacity) {
michael@0:         isDataAllocated=FALSE;
michael@0:     } else {
michael@0:         aliasDataCapacity=other->dataCapacity;
michael@0:         aliasData=(uint32_t *)uprv_malloc(other->dataCapacity*4);
michael@0:         if(aliasData==NULL) {
michael@0:             return NULL;
michael@0:         }
michael@0:         isDataAllocated=TRUE;
michael@0:     }
michael@0: 
michael@0:     trie=utrie_open(fillIn, aliasData, aliasDataCapacity,
michael@0:                     other->data[0], other->leadUnitValue,
michael@0:                     other->isLatin1Linear);
michael@0:     if(trie==NULL) {
michael@0:         uprv_free(aliasData);
michael@0:     } else {
michael@0:         uprv_memcpy(trie->index, other->index, sizeof(trie->index));
michael@0:         uprv_memcpy(trie->data, other->data, other->dataLength*4);
michael@0:         trie->dataLength=other->dataLength;
michael@0:         trie->isDataAllocated=isDataAllocated;
michael@0:     }
michael@0: 
michael@0:     return trie;
michael@0: }
michael@0: 
michael@0: U_CAPI void U_EXPORT2
michael@0: utrie_close(UNewTrie *trie) {
michael@0:     if(trie!=NULL) {
michael@0:         if(trie->isDataAllocated) {
michael@0:             uprv_free(trie->data);
michael@0:             trie->data=NULL;
michael@0:         }
michael@0:         if(trie->isAllocated) {
michael@0:             uprv_free(trie);
michael@0:         }
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI uint32_t * U_EXPORT2
michael@0: utrie_getData(UNewTrie *trie, int32_t *pLength) {
michael@0:     if(trie==NULL || pLength==NULL) {
michael@0:         return NULL;
michael@0:     }
michael@0: 
michael@0:     *pLength=trie->dataLength;
michael@0:     return trie->data;
michael@0: }
michael@0: 
michael@0: static int32_t
michael@0: utrie_allocDataBlock(UNewTrie *trie) {
michael@0:     int32_t newBlock, newTop;
michael@0: 
michael@0:     newBlock=trie->dataLength;
michael@0:     newTop=newBlock+UTRIE_DATA_BLOCK_LENGTH;
michael@0:     if(newTop>trie->dataCapacity) {
michael@0:         /* out of memory in the data array */
michael@0:         return -1;
michael@0:     }
michael@0:     trie->dataLength=newTop;
michael@0:     return newBlock;
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * No error checking for illegal arguments.
michael@0:  *
michael@0:  * @return -1 if no new data block available (out of memory in data array)
michael@0:  * @internal
michael@0:  */
michael@0: static int32_t
michael@0: utrie_getDataBlock(UNewTrie *trie, UChar32 c) {
michael@0:     int32_t indexValue, newBlock;
michael@0: 
michael@0:     c>>=UTRIE_SHIFT;
michael@0:     indexValue=trie->index[c];
michael@0:     if(indexValue>0) {
michael@0:         return indexValue;
michael@0:     }
michael@0: 
michael@0:     /* allocate a new data block */
michael@0:     newBlock=utrie_allocDataBlock(trie);
michael@0:     if(newBlock<0) {
michael@0:         /* out of memory in the data array */
michael@0:         return -1;
michael@0:     }
michael@0:     trie->index[c]=newBlock;
michael@0: 
michael@0:     /* copy-on-write for a block from a setRange() */
michael@0:     uprv_memcpy(trie->data+newBlock, trie->data-indexValue, 4*UTRIE_DATA_BLOCK_LENGTH);
michael@0:     return newBlock;
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * @return TRUE if the value was successfully set
michael@0:  */
michael@0: U_CAPI UBool U_EXPORT2
michael@0: utrie_set32(UNewTrie *trie, UChar32 c, uint32_t value) {
michael@0:     int32_t block;
michael@0: 
michael@0:     /* valid, uncompacted trie and valid c? */
michael@0:     if(trie==NULL || trie->isCompacted || (uint32_t)c>0x10ffff) {
michael@0:         return FALSE;
michael@0:     }
michael@0: 
michael@0:     block=utrie_getDataBlock(trie, c);
michael@0:     if(block<0) {
michael@0:         return FALSE;
michael@0:     }
michael@0: 
michael@0:     trie->data[block+(c&UTRIE_MASK)]=value;
michael@0:     return TRUE;
michael@0: }
michael@0: 
michael@0: U_CAPI uint32_t U_EXPORT2
michael@0: utrie_get32(UNewTrie *trie, UChar32 c, UBool *pInBlockZero) {
michael@0:     int32_t block;
michael@0: 
michael@0:     /* valid, uncompacted trie and valid c? */
michael@0:     if(trie==NULL || trie->isCompacted || (uint32_t)c>0x10ffff) {
michael@0:         if(pInBlockZero!=NULL) {
michael@0:             *pInBlockZero=TRUE;
michael@0:         }
michael@0:         return 0;
michael@0:     }
michael@0: 
michael@0:     block=trie->index[c>>UTRIE_SHIFT];
michael@0:     if(pInBlockZero!=NULL) {
michael@0:         *pInBlockZero= (UBool)(block==0);
michael@0:     }
michael@0: 
michael@0:     return trie->data[ABS(block)+(c&UTRIE_MASK)];
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * @internal
michael@0:  */
michael@0: static void
michael@0: utrie_fillBlock(uint32_t *block, UChar32 start, UChar32 limit,
michael@0:                 uint32_t value, uint32_t initialValue, UBool overwrite) {
michael@0:     uint32_t *pLimit;
michael@0: 
michael@0:     pLimit=block+limit;
michael@0:     block+=start;
michael@0:     if(overwrite) {
michael@0:         while(block<pLimit) {
michael@0:             *block++=value;
michael@0:         }
michael@0:     } else {
michael@0:         while(block<pLimit) {
michael@0:             if(*block==initialValue) {
michael@0:                 *block=value;
michael@0:             }
michael@0:             ++block;
michael@0:         }
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI UBool U_EXPORT2
michael@0: utrie_setRange32(UNewTrie *trie, UChar32 start, UChar32 limit, uint32_t value, UBool overwrite) {
michael@0:     /*
michael@0:      * repeat value in [start..limit[
michael@0:      * mark index values for repeat-data blocks by setting bit 31 of the index values
michael@0:      * fill around existing values if any, if(overwrite)
michael@0:      */
michael@0:     uint32_t initialValue;
michael@0:     int32_t block, rest, repeatBlock;
michael@0: 
michael@0:     /* valid, uncompacted trie and valid indexes? */
michael@0:     if( trie==NULL || trie->isCompacted ||
michael@0:         (uint32_t)start>0x10ffff || (uint32_t)limit>0x110000 || start>limit
michael@0:     ) {
michael@0:         return FALSE;
michael@0:     }
michael@0:     if(start==limit) {
michael@0:         return TRUE; /* nothing to do */
michael@0:     }
michael@0: 
michael@0:     initialValue=trie->data[0];
michael@0:     if(start&UTRIE_MASK) {
michael@0:         UChar32 nextStart;
michael@0: 
michael@0:         /* set partial block at [start..following block boundary[ */
michael@0:         block=utrie_getDataBlock(trie, start);
michael@0:         if(block<0) {
michael@0:             return FALSE;
michael@0:         }
michael@0: 
michael@0:         nextStart=(start+UTRIE_DATA_BLOCK_LENGTH)&~UTRIE_MASK;
michael@0:         if(nextStart<=limit) {
michael@0:             utrie_fillBlock(trie->data+block, start&UTRIE_MASK, UTRIE_DATA_BLOCK_LENGTH,
michael@0:                             value, initialValue, overwrite);
michael@0:             start=nextStart;
michael@0:         } else {
michael@0:             utrie_fillBlock(trie->data+block, start&UTRIE_MASK, limit&UTRIE_MASK,
michael@0:                             value, initialValue, overwrite);
michael@0:             return TRUE;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     /* number of positions in the last, partial block */
michael@0:     rest=limit&UTRIE_MASK;
michael@0: 
michael@0:     /* round down limit to a block boundary */
michael@0:     limit&=~UTRIE_MASK;
michael@0: 
michael@0:     /* iterate over all-value blocks */
michael@0:     if(value==initialValue) {
michael@0:         repeatBlock=0;
michael@0:     } else {
michael@0:         repeatBlock=-1;
michael@0:     }
michael@0:     while(start<limit) {
michael@0:         /* get index value */
michael@0:         block=trie->index[start>>UTRIE_SHIFT];
michael@0:         if(block>0) {
michael@0:             /* already allocated, fill in value */
michael@0:             utrie_fillBlock(trie->data+block, 0, UTRIE_DATA_BLOCK_LENGTH, value, initialValue, overwrite);
michael@0:         } else if(trie->data[-block]!=value && (block==0 || overwrite)) {
michael@0:             /* set the repeatBlock instead of the current block 0 or range block */
michael@0:             if(repeatBlock>=0) {
michael@0:                 trie->index[start>>UTRIE_SHIFT]=-repeatBlock;
michael@0:             } else {
michael@0:                 /* create and set and fill the repeatBlock */
michael@0:                 repeatBlock=utrie_getDataBlock(trie, start);
michael@0:                 if(repeatBlock<0) {
michael@0:                     return FALSE;
michael@0:                 }
michael@0: 
michael@0:                 /* set the negative block number to indicate that it is a repeat block */
michael@0:                 trie->index[start>>UTRIE_SHIFT]=-repeatBlock;
michael@0:                 utrie_fillBlock(trie->data+repeatBlock, 0, UTRIE_DATA_BLOCK_LENGTH, value, initialValue, TRUE);
michael@0:             }
michael@0:         }
michael@0: 
michael@0:         start+=UTRIE_DATA_BLOCK_LENGTH;
michael@0:     }
michael@0: 
michael@0:     if(rest>0) {
michael@0:         /* set partial block at [last block boundary..limit[ */
michael@0:         block=utrie_getDataBlock(trie, start);
michael@0:         if(block<0) {
michael@0:             return FALSE;
michael@0:         }
michael@0: 
michael@0:         utrie_fillBlock(trie->data+block, 0, rest, value, initialValue, overwrite);
michael@0:     }
michael@0: 
michael@0:     return TRUE;
michael@0: }
michael@0: 
michael@0: static int32_t
michael@0: _findSameIndexBlock(const int32_t *idx, int32_t indexLength,
michael@0:                     int32_t otherBlock) {
michael@0:     int32_t block, i;
michael@0: 
michael@0:     for(block=UTRIE_BMP_INDEX_LENGTH; block<indexLength; block+=UTRIE_SURROGATE_BLOCK_COUNT) {
michael@0:         for(i=0; i<UTRIE_SURROGATE_BLOCK_COUNT; ++i) {
michael@0:             if(idx[block+i]!=idx[otherBlock+i]) {
michael@0:                 break;
michael@0:             }
michael@0:         }
michael@0:         if(i==UTRIE_SURROGATE_BLOCK_COUNT) {
michael@0:             return block;
michael@0:         }
michael@0:     }
michael@0:     return indexLength;
michael@0: }
michael@0: 
michael@0: /*
michael@0:  * Fold the normalization data for supplementary code points into
michael@0:  * a compact area on top of the BMP-part of the trie index,
michael@0:  * with the lead surrogates indexing this compact area.
michael@0:  *
michael@0:  * Duplicate the index values for lead surrogates:
michael@0:  * From inside the BMP area, where some may be overridden with folded values,
michael@0:  * to just after the BMP area, where they can be retrieved for
michael@0:  * code point lookups.
michael@0:  */
michael@0: static void
michael@0: utrie_fold(UNewTrie *trie, UNewTrieGetFoldedValue *getFoldedValue, UErrorCode *pErrorCode) {
michael@0:     int32_t leadIndexes[UTRIE_SURROGATE_BLOCK_COUNT];
michael@0:     int32_t *idx;
michael@0:     uint32_t value;
michael@0:     UChar32 c;
michael@0:     int32_t indexLength, block;
michael@0: #ifdef UTRIE_DEBUG
michael@0:     int countLeadCUWithData=0;
michael@0: #endif
michael@0: 
michael@0:     idx=trie->index;
michael@0: 
michael@0:     /* copy the lead surrogate indexes into a temporary array */
michael@0:     uprv_memcpy(leadIndexes, idx+(0xd800>>UTRIE_SHIFT), 4*UTRIE_SURROGATE_BLOCK_COUNT);
michael@0: 
michael@0:     /*
michael@0:      * set all values for lead surrogate code *units* to leadUnitValue
michael@0:      * so that, by default, runtime lookups will find no data for associated
michael@0:      * supplementary code points, unless there is data for such code points
michael@0:      * which will result in a non-zero folding value below that is set for
michael@0:      * the respective lead units
michael@0:      *
michael@0:      * the above saved the indexes for surrogate code *points*
michael@0:      * fill the indexes with simplified code from utrie_setRange32()
michael@0:      */
michael@0:     if(trie->leadUnitValue==trie->data[0]) {
michael@0:         block=0; /* leadUnitValue==initialValue, use all-initial-value block */
michael@0:     } else {
michael@0:         /* create and fill the repeatBlock */
michael@0:         block=utrie_allocDataBlock(trie);
michael@0:         if(block<0) {
michael@0:             /* data table overflow */
michael@0:             *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
michael@0:             return;
michael@0:         }
michael@0:         utrie_fillBlock(trie->data+block, 0, UTRIE_DATA_BLOCK_LENGTH, trie->leadUnitValue, trie->data[0], TRUE);
michael@0:         block=-block; /* negative block number to indicate that it is a repeat block */
michael@0:     }
michael@0:     for(c=(0xd800>>UTRIE_SHIFT); c<(0xdc00>>UTRIE_SHIFT); ++c) {
michael@0:         trie->index[c]=block;
michael@0:     }
michael@0: 
michael@0:     /*
michael@0:      * Fold significant index values into the area just after the BMP indexes.
michael@0:      * In case the first lead surrogate has significant data,
michael@0:      * its index block must be used first (in which case the folding is a no-op).
michael@0:      * Later all folded index blocks are moved up one to insert the copied
michael@0:      * lead surrogate indexes.
michael@0:      */
michael@0:     indexLength=UTRIE_BMP_INDEX_LENGTH;
michael@0: 
michael@0:     /* search for any index (stage 1) entries for supplementary code points */
michael@0:     for(c=0x10000; c<0x110000;) {
michael@0:         if(idx[c>>UTRIE_SHIFT]!=0) {
michael@0:             /* there is data, treat the full block for a lead surrogate */
michael@0:             c&=~0x3ff;
michael@0: 
michael@0: #ifdef UTRIE_DEBUG
michael@0:             ++countLeadCUWithData;
michael@0:             /* printf("supplementary data for lead surrogate U+%04lx\n", (long)(0xd7c0+(c>>10))); */
michael@0: #endif
michael@0: 
michael@0:             /* is there an identical index block? */
michael@0:             block=_findSameIndexBlock(idx, indexLength, c>>UTRIE_SHIFT);
michael@0: 
michael@0:             /*
michael@0:              * get a folded value for [c..c+0x400[ and,
michael@0:              * if different from the value for the lead surrogate code point,
michael@0:              * set it for the lead surrogate code unit
michael@0:              */
michael@0:             value=getFoldedValue(trie, c, block+UTRIE_SURROGATE_BLOCK_COUNT);
michael@0:             if(value!=utrie_get32(trie, U16_LEAD(c), NULL)) {
michael@0:                 if(!utrie_set32(trie, U16_LEAD(c), value)) {
michael@0:                     /* data table overflow */
michael@0:                     *pErrorCode=U_MEMORY_ALLOCATION_ERROR;
michael@0:                     return;
michael@0:                 }
michael@0: 
michael@0:                 /* if we did not find an identical index block... */
michael@0:                 if(block==indexLength) {
michael@0:                     /* move the actual index (stage 1) entries from the supplementary position to the new one */
michael@0:                     uprv_memmove(idx+indexLength,
michael@0:                                  idx+(c>>UTRIE_SHIFT),
michael@0:                                  4*UTRIE_SURROGATE_BLOCK_COUNT);
michael@0:                     indexLength+=UTRIE_SURROGATE_BLOCK_COUNT;
michael@0:                 }
michael@0:             }
michael@0:             c+=0x400;
michael@0:         } else {
michael@0:             c+=UTRIE_DATA_BLOCK_LENGTH;
michael@0:         }
michael@0:     }
michael@0: #ifdef UTRIE_DEBUG
michael@0:     if(countLeadCUWithData>0) {
michael@0:         printf("supplementary data for %d lead surrogates\n", countLeadCUWithData);
michael@0:     }
michael@0: #endif
michael@0: 
michael@0:     /*
michael@0:      * index array overflow?
michael@0:      * This is to guarantee that a folding offset is of the form
michael@0:      * UTRIE_BMP_INDEX_LENGTH+n*UTRIE_SURROGATE_BLOCK_COUNT with n=0..1023.
michael@0:      * If the index is too large, then n>=1024 and more than 10 bits are necessary.
michael@0:      *
michael@0:      * In fact, it can only ever become n==1024 with completely unfoldable data and
michael@0:      * the additional block of duplicated values for lead surrogates.
michael@0:      */
michael@0:     if(indexLength>=UTRIE_MAX_INDEX_LENGTH) {
michael@0:         *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     /*
michael@0:      * make space for the lead surrogate index block and
michael@0:      * insert it between the BMP indexes and the folded ones
michael@0:      */
michael@0:     uprv_memmove(idx+UTRIE_BMP_INDEX_LENGTH+UTRIE_SURROGATE_BLOCK_COUNT,
michael@0:                  idx+UTRIE_BMP_INDEX_LENGTH,
michael@0:                  4*(indexLength-UTRIE_BMP_INDEX_LENGTH));
michael@0:     uprv_memcpy(idx+UTRIE_BMP_INDEX_LENGTH,
michael@0:                 leadIndexes,
michael@0:                 4*UTRIE_SURROGATE_BLOCK_COUNT);
michael@0:     indexLength+=UTRIE_SURROGATE_BLOCK_COUNT;
michael@0: 
michael@0: #ifdef UTRIE_DEBUG
michael@0:     printf("trie index count: BMP %ld  all Unicode %ld  folded %ld\n",
michael@0:            UTRIE_BMP_INDEX_LENGTH, (long)UTRIE_MAX_INDEX_LENGTH, indexLength);
michael@0: #endif
michael@0: 
michael@0:     trie->indexLength=indexLength;
michael@0: }
michael@0: 
michael@0: /*
michael@0:  * Set a value in the trie index map to indicate which data block
michael@0:  * is referenced and which one is not.
michael@0:  * utrie_compact() will remove data blocks that are not used at all.
michael@0:  * Set
michael@0:  * - 0 if it is used
michael@0:  * - -1 if it is not used
michael@0:  */
michael@0: static void
michael@0: _findUnusedBlocks(UNewTrie *trie) {
michael@0:     int32_t i;
michael@0: 
michael@0:     /* fill the entire map with "not used" */
michael@0:     uprv_memset(trie->map, 0xff, (UTRIE_MAX_BUILD_TIME_DATA_LENGTH>>UTRIE_SHIFT)*4);
michael@0: 
michael@0:     /* mark each block that _is_ used with 0 */
michael@0:     for(i=0; i<trie->indexLength; ++i) {
michael@0:         trie->map[ABS(trie->index[i])>>UTRIE_SHIFT]=0;
michael@0:     }
michael@0: 
michael@0:     /* never move the all-initial-value block 0 */
michael@0:     trie->map[0]=0;
michael@0: }
michael@0: 
michael@0: static int32_t
michael@0: _findSameDataBlock(const uint32_t *data, int32_t dataLength,
michael@0:                    int32_t otherBlock, int32_t step) {
michael@0:     int32_t block;
michael@0: 
michael@0:     /* ensure that we do not even partially get past dataLength */
michael@0:     dataLength-=UTRIE_DATA_BLOCK_LENGTH;
michael@0: 
michael@0:     for(block=0; block<=dataLength; block+=step) {
michael@0:         if(equal_uint32(data+block, data+otherBlock, UTRIE_DATA_BLOCK_LENGTH)) {
michael@0:             return block;
michael@0:         }
michael@0:     }
michael@0:     return -1;
michael@0: }
michael@0: 
michael@0: /*
michael@0:  * Compact a folded build-time trie.
michael@0:  *
michael@0:  * The compaction
michael@0:  * - removes blocks that are identical with earlier ones
michael@0:  * - overlaps adjacent blocks as much as possible (if overlap==TRUE)
michael@0:  * - moves blocks in steps of the data granularity
michael@0:  * - moves and overlaps blocks that overlap with multiple values in the overlap region
michael@0:  *
michael@0:  * It does not
michael@0:  * - try to move and overlap blocks that are not already adjacent
michael@0:  */
michael@0: static void
michael@0: utrie_compact(UNewTrie *trie, UBool overlap, UErrorCode *pErrorCode) {
michael@0:     int32_t i, start, newStart, overlapStart;
michael@0: 
michael@0:     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     /* valid, uncompacted trie? */
michael@0:     if(trie==NULL) {
michael@0:         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
michael@0:         return;
michael@0:     }
michael@0:     if(trie->isCompacted) {
michael@0:         return; /* nothing left to do */
michael@0:     }
michael@0: 
michael@0:     /* compaction */
michael@0: 
michael@0:     /* initialize the index map with "block is used/unused" flags */
michael@0:     _findUnusedBlocks(trie);
michael@0: 
michael@0:     /* if Latin-1 is preallocated and linear, then do not compact Latin-1 data */
michael@0:     if(trie->isLatin1Linear && UTRIE_SHIFT<=8) {
michael@0:         overlapStart=UTRIE_DATA_BLOCK_LENGTH+256;
michael@0:     } else {
michael@0:         overlapStart=UTRIE_DATA_BLOCK_LENGTH;
michael@0:     }
michael@0: 
michael@0:     newStart=UTRIE_DATA_BLOCK_LENGTH;
michael@0:     for(start=newStart; start<trie->dataLength;) {
michael@0:         /*
michael@0:          * start: index of first entry of current block
michael@0:          * newStart: index where the current block is to be moved
michael@0:          *           (right after current end of already-compacted data)
michael@0:          */
michael@0: 
michael@0:         /* skip blocks that are not used */
michael@0:         if(trie->map[start>>UTRIE_SHIFT]<0) {
michael@0:             /* advance start to the next block */
michael@0:             start+=UTRIE_DATA_BLOCK_LENGTH;
michael@0: 
michael@0:             /* leave newStart with the previous block! */
michael@0:             continue;
michael@0:         }
michael@0: 
michael@0:         /* search for an identical block */
michael@0:         if( start>=overlapStart &&
michael@0:             (i=_findSameDataBlock(trie->data, newStart, start,
michael@0:                             overlap ? UTRIE_DATA_GRANULARITY : UTRIE_DATA_BLOCK_LENGTH))
michael@0:              >=0
michael@0:         ) {
michael@0:             /* found an identical block, set the other block's index value for the current block */
michael@0:             trie->map[start>>UTRIE_SHIFT]=i;
michael@0: 
michael@0:             /* advance start to the next block */
michael@0:             start+=UTRIE_DATA_BLOCK_LENGTH;
michael@0: 
michael@0:             /* leave newStart with the previous block! */
michael@0:             continue;
michael@0:         }
michael@0: 
michael@0:         /* see if the beginning of this block can be overlapped with the end of the previous block */
michael@0:         if(overlap && start>=overlapStart) {
michael@0:             /* look for maximum overlap (modulo granularity) with the previous, adjacent block */
michael@0:             for(i=UTRIE_DATA_BLOCK_LENGTH-UTRIE_DATA_GRANULARITY;
michael@0:                 i>0 && !equal_uint32(trie->data+(newStart-i), trie->data+start, i);
michael@0:                 i-=UTRIE_DATA_GRANULARITY) {}
michael@0:         } else {
michael@0:             i=0;
michael@0:         }
michael@0: 
michael@0:         if(i>0) {
michael@0:             /* some overlap */
michael@0:             trie->map[start>>UTRIE_SHIFT]=newStart-i;
michael@0: 
michael@0:             /* move the non-overlapping indexes to their new positions */
michael@0:             start+=i;
michael@0:             for(i=UTRIE_DATA_BLOCK_LENGTH-i; i>0; --i) {
michael@0:                 trie->data[newStart++]=trie->data[start++];
michael@0:             }
michael@0:         } else if(newStart<start) {
michael@0:             /* no overlap, just move the indexes to their new positions */
michael@0:             trie->map[start>>UTRIE_SHIFT]=newStart;
michael@0:             for(i=UTRIE_DATA_BLOCK_LENGTH; i>0; --i) {
michael@0:                 trie->data[newStart++]=trie->data[start++];
michael@0:             }
michael@0:         } else /* no overlap && newStart==start */ {
michael@0:             trie->map[start>>UTRIE_SHIFT]=start;
michael@0:             newStart+=UTRIE_DATA_BLOCK_LENGTH;
michael@0:             start=newStart;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     /* now adjust the index (stage 1) table */
michael@0:     for(i=0; i<trie->indexLength; ++i) {
michael@0:         trie->index[i]=trie->map[ABS(trie->index[i])>>UTRIE_SHIFT];
michael@0:     }
michael@0: 
michael@0: #ifdef UTRIE_DEBUG
michael@0:     /* we saved some space */
michael@0:     printf("compacting trie: count of 32-bit words %lu->%lu\n",
michael@0:             (long)trie->dataLength, (long)newStart);
michael@0: #endif
michael@0: 
michael@0:     trie->dataLength=newStart;
michael@0: }
michael@0: 
michael@0: /* serialization ------------------------------------------------------------ */
michael@0: 
michael@0: /*
michael@0:  * Default function for the folding value:
michael@0:  * Just store the offset (16 bits) if there is any non-initial-value entry.
michael@0:  *
michael@0:  * The offset parameter is never 0.
michael@0:  * Returning the offset itself is safe for UTRIE_SHIFT>=5 because
michael@0:  * for UTRIE_SHIFT==5 the maximum index length is UTRIE_MAX_INDEX_LENGTH==0x8800
michael@0:  * which fits into 16-bit trie values;
michael@0:  * for higher UTRIE_SHIFT, UTRIE_MAX_INDEX_LENGTH decreases.
michael@0:  *
michael@0:  * Theoretically, it would be safer for all possible UTRIE_SHIFT including
michael@0:  * those of 4 and lower to return offset>>UTRIE_SURROGATE_BLOCK_BITS
michael@0:  * which would always result in a value of 0x40..0x43f
michael@0:  * (start/end 1k blocks of supplementary Unicode code points).
michael@0:  * However, this would be uglier, and would not work for some existing
michael@0:  * binary data file formats.
michael@0:  *
michael@0:  * Also, we do not plan to change UTRIE_SHIFT because it would change binary
michael@0:  * data file formats, and we would probably not make it smaller because of
michael@0:  * the then even larger BMP index length even for empty tries.
michael@0:  */
michael@0: static uint32_t U_CALLCONV
michael@0: defaultGetFoldedValue(UNewTrie *trie, UChar32 start, int32_t offset) {
michael@0:     uint32_t value, initialValue;
michael@0:     UChar32 limit;
michael@0:     UBool inBlockZero;
michael@0: 
michael@0:     initialValue=trie->data[0];
michael@0:     limit=start+0x400;
michael@0:     while(start<limit) {
michael@0:         value=utrie_get32(trie, start, &inBlockZero);
michael@0:         if(inBlockZero) {
michael@0:             start+=UTRIE_DATA_BLOCK_LENGTH;
michael@0:         } else if(value!=initialValue) {
michael@0:             return (uint32_t)offset;
michael@0:         } else {
michael@0:             ++start;
michael@0:         }
michael@0:     }
michael@0:     return 0;
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: utrie_serialize(UNewTrie *trie, void *dt, int32_t capacity,
michael@0:                 UNewTrieGetFoldedValue *getFoldedValue,
michael@0:                 UBool reduceTo16Bits,
michael@0:                 UErrorCode *pErrorCode) {
michael@0:     UTrieHeader *header;
michael@0:     uint32_t *p;
michael@0:     uint16_t *dest16;
michael@0:     int32_t i, length;
michael@0:     uint8_t* data = NULL;
michael@0: 
michael@0:     /* argument check */
michael@0:     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
michael@0:         return 0;
michael@0:     }
michael@0: 
michael@0:     if(trie==NULL || capacity<0 || (capacity>0 && dt==NULL)) {
michael@0:         *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
michael@0:         return 0;
michael@0:     }
michael@0:     if(getFoldedValue==NULL) {
michael@0:         getFoldedValue=defaultGetFoldedValue;
michael@0:     }
michael@0: 
michael@0:     data = (uint8_t*)dt;
michael@0:     /* fold and compact if necessary, also checks that indexLength is within limits */
michael@0:     if(!trie->isCompacted) {
michael@0:         /* compact once without overlap to improve folding */
michael@0:         utrie_compact(trie, FALSE, pErrorCode);
michael@0: 
michael@0:         /* fold the supplementary part of the index array */
michael@0:         utrie_fold(trie, getFoldedValue, pErrorCode);
michael@0: 
michael@0:         /* compact again with overlap for minimum data array length */
michael@0:         utrie_compact(trie, TRUE, pErrorCode);
michael@0: 
michael@0:         trie->isCompacted=TRUE;
michael@0:         if(U_FAILURE(*pErrorCode)) {
michael@0:             return 0;
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     /* is dataLength within limits? */
michael@0:     if( (reduceTo16Bits ? (trie->dataLength+trie->indexLength) : trie->dataLength) >= UTRIE_MAX_DATA_LENGTH) {
michael@0:         *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR;
michael@0:     }
michael@0: 
michael@0:     length=sizeof(UTrieHeader)+2*trie->indexLength;
michael@0:     if(reduceTo16Bits) {
michael@0:         length+=2*trie->dataLength;
michael@0:     } else {
michael@0:         length+=4*trie->dataLength;
michael@0:     }
michael@0: 
michael@0:     if(length>capacity) {
michael@0:         return length; /* preflighting */
michael@0:     }
michael@0: 
michael@0: #ifdef UTRIE_DEBUG
michael@0:     printf("**UTrieLengths(serialize)** index:%6ld  data:%6ld  serialized:%6ld\n",
michael@0:            (long)trie->indexLength, (long)trie->dataLength, (long)length);
michael@0: #endif
michael@0: 
michael@0:     /* set the header fields */
michael@0:     header=(UTrieHeader *)data;
michael@0:     data+=sizeof(UTrieHeader);
michael@0: 
michael@0:     header->signature=0x54726965; /* "Trie" */
michael@0:     header->options=UTRIE_SHIFT | (UTRIE_INDEX_SHIFT<<UTRIE_OPTIONS_INDEX_SHIFT);
michael@0: 
michael@0:     if(!reduceTo16Bits) {
michael@0:         header->options|=UTRIE_OPTIONS_DATA_IS_32_BIT;
michael@0:     }
michael@0:     if(trie->isLatin1Linear) {
michael@0:         header->options|=UTRIE_OPTIONS_LATIN1_IS_LINEAR;
michael@0:     }
michael@0: 
michael@0:     header->indexLength=trie->indexLength;
michael@0:     header->dataLength=trie->dataLength;
michael@0: 
michael@0:     /* write the index (stage 1) array and the 16/32-bit data (stage 2) array */
michael@0:     if(reduceTo16Bits) {
michael@0:         /* write 16-bit index values shifted right by UTRIE_INDEX_SHIFT, after adding indexLength */
michael@0:         p=(uint32_t *)trie->index;
michael@0:         dest16=(uint16_t *)data;
michael@0:         for(i=trie->indexLength; i>0; --i) {
michael@0:             *dest16++=(uint16_t)((*p++ + trie->indexLength)>>UTRIE_INDEX_SHIFT);
michael@0:         }
michael@0: 
michael@0:         /* write 16-bit data values */
michael@0:         p=trie->data;
michael@0:         for(i=trie->dataLength; i>0; --i) {
michael@0:             *dest16++=(uint16_t)*p++;
michael@0:         }
michael@0:     } else {
michael@0:         /* write 16-bit index values shifted right by UTRIE_INDEX_SHIFT */
michael@0:         p=(uint32_t *)trie->index;
michael@0:         dest16=(uint16_t *)data;
michael@0:         for(i=trie->indexLength; i>0; --i) {
michael@0:             *dest16++=(uint16_t)(*p++ >> UTRIE_INDEX_SHIFT);
michael@0:         }
michael@0: 
michael@0:         /* write 32-bit data values */
michael@0:         uprv_memcpy(dest16, trie->data, 4*trie->dataLength);
michael@0:     }
michael@0: 
michael@0:     return length;
michael@0: }
michael@0: 
michael@0: /* inverse to defaultGetFoldedValue() */
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: utrie_defaultGetFoldingOffset(uint32_t data) {
michael@0:     return (int32_t)data;
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: utrie_unserialize(UTrie *trie, const void *data, int32_t length, UErrorCode *pErrorCode) {
michael@0:     const UTrieHeader *header;
michael@0:     const uint16_t *p16;
michael@0:     uint32_t options;
michael@0: 
michael@0:     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
michael@0:         return -1;
michael@0:     }
michael@0: 
michael@0:     /* enough data for a trie header? */
michael@0:     if(length<(int32_t)sizeof(UTrieHeader)) {
michael@0:         *pErrorCode=U_INVALID_FORMAT_ERROR;
michael@0:         return -1;
michael@0:     }
michael@0: 
michael@0:     /* check the signature */
michael@0:     header=(const UTrieHeader *)data;
michael@0:     if(header->signature!=0x54726965) {
michael@0:         *pErrorCode=U_INVALID_FORMAT_ERROR;
michael@0:         return -1;
michael@0:     }
michael@0: 
michael@0:     /* get the options and check the shift values */
michael@0:     options=header->options;
michael@0:     if( (options&UTRIE_OPTIONS_SHIFT_MASK)!=UTRIE_SHIFT ||
michael@0:         ((options>>UTRIE_OPTIONS_INDEX_SHIFT)&UTRIE_OPTIONS_SHIFT_MASK)!=UTRIE_INDEX_SHIFT
michael@0:     ) {
michael@0:         *pErrorCode=U_INVALID_FORMAT_ERROR;
michael@0:         return -1;
michael@0:     }
michael@0:     trie->isLatin1Linear= (UBool)((options&UTRIE_OPTIONS_LATIN1_IS_LINEAR)!=0);
michael@0: 
michael@0:     /* get the length values */
michael@0:     trie->indexLength=header->indexLength;
michael@0:     trie->dataLength=header->dataLength;
michael@0: 
michael@0:     length-=(int32_t)sizeof(UTrieHeader);
michael@0: 
michael@0:     /* enough data for the index? */
michael@0:     if(length<2*trie->indexLength) {
michael@0:         *pErrorCode=U_INVALID_FORMAT_ERROR;
michael@0:         return -1;
michael@0:     }
michael@0:     p16=(const uint16_t *)(header+1);
michael@0:     trie->index=p16;
michael@0:     p16+=trie->indexLength;
michael@0:     length-=2*trie->indexLength;
michael@0: 
michael@0:     /* get the data */
michael@0:     if(options&UTRIE_OPTIONS_DATA_IS_32_BIT) {
michael@0:         if(length<4*trie->dataLength) {
michael@0:             *pErrorCode=U_INVALID_FORMAT_ERROR;
michael@0:             return -1;
michael@0:         }
michael@0:         trie->data32=(const uint32_t *)p16;
michael@0:         trie->initialValue=trie->data32[0];
michael@0:         length=(int32_t)sizeof(UTrieHeader)+2*trie->indexLength+4*trie->dataLength;
michael@0:     } else {
michael@0:         if(length<2*trie->dataLength) {
michael@0:             *pErrorCode=U_INVALID_FORMAT_ERROR;
michael@0:             return -1;
michael@0:         }
michael@0: 
michael@0:         /* the "data16" data is used via the index pointer */
michael@0:         trie->data32=NULL;
michael@0:         trie->initialValue=trie->index[trie->indexLength];
michael@0:         length=(int32_t)sizeof(UTrieHeader)+2*trie->indexLength+2*trie->dataLength;
michael@0:     }
michael@0: 
michael@0:     trie->getFoldingOffset=utrie_defaultGetFoldingOffset;
michael@0: 
michael@0:     return length;
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: utrie_unserializeDummy(UTrie *trie,
michael@0:                        void *data, int32_t length,
michael@0:                        uint32_t initialValue, uint32_t leadUnitValue,
michael@0:                        UBool make16BitTrie,
michael@0:                        UErrorCode *pErrorCode) {
michael@0:     uint16_t *p16;
michael@0:     int32_t actualLength, latin1Length, i, limit;
michael@0:     uint16_t block;
michael@0: 
michael@0:     if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
michael@0:         return -1;
michael@0:     }
michael@0: 
michael@0:     /* calculate the actual size of the dummy trie data */
michael@0: 
michael@0:     /* max(Latin-1, block 0) */
michael@0:     latin1Length= 256; /*UTRIE_SHIFT<=8 ? 256 : UTRIE_DATA_BLOCK_LENGTH;*/
michael@0: 
michael@0:     trie->indexLength=UTRIE_BMP_INDEX_LENGTH+UTRIE_SURROGATE_BLOCK_COUNT;
michael@0:     trie->dataLength=latin1Length;
michael@0:     if(leadUnitValue!=initialValue) {
michael@0:         trie->dataLength+=UTRIE_DATA_BLOCK_LENGTH;
michael@0:     }
michael@0: 
michael@0:     actualLength=trie->indexLength*2;
michael@0:     if(make16BitTrie) {
michael@0:         actualLength+=trie->dataLength*2;
michael@0:     } else {
michael@0:         actualLength+=trie->dataLength*4;
michael@0:     }
michael@0: 
michael@0:     /* enough space for the dummy trie? */
michael@0:     if(length<actualLength) {
michael@0:         *pErrorCode=U_BUFFER_OVERFLOW_ERROR;
michael@0:         return actualLength;
michael@0:     }
michael@0: 
michael@0:     trie->isLatin1Linear=TRUE;
michael@0:     trie->initialValue=initialValue;
michael@0: 
michael@0:     /* fill the index and data arrays */
michael@0:     p16=(uint16_t *)data;
michael@0:     trie->index=p16;
michael@0: 
michael@0:     if(make16BitTrie) {
michael@0:         /* indexes to block 0 */
michael@0:         block=(uint16_t)(trie->indexLength>>UTRIE_INDEX_SHIFT);
michael@0:         limit=trie->indexLength;
michael@0:         for(i=0; i<limit; ++i) {
michael@0:             p16[i]=block;
michael@0:         }
michael@0: 
michael@0:         if(leadUnitValue!=initialValue) {
michael@0:             /* indexes for lead surrogate code units to the block after Latin-1 */
michael@0:             block+=(uint16_t)(latin1Length>>UTRIE_INDEX_SHIFT);
michael@0:             i=0xd800>>UTRIE_SHIFT;
michael@0:             limit=0xdc00>>UTRIE_SHIFT;
michael@0:             for(; i<limit; ++i) {
michael@0:                 p16[i]=block;
michael@0:             }
michael@0:         }
michael@0: 
michael@0:         trie->data32=NULL;
michael@0: 
michael@0:         /* Latin-1 data */
michael@0:         p16+=trie->indexLength;
michael@0:         for(i=0; i<latin1Length; ++i) {
michael@0:             p16[i]=(uint16_t)initialValue;
michael@0:         }
michael@0: 
michael@0:         /* data for lead surrogate code units */
michael@0:         if(leadUnitValue!=initialValue) {
michael@0:             limit=latin1Length+UTRIE_DATA_BLOCK_LENGTH;
michael@0:             for(/* i=latin1Length */; i<limit; ++i) {
michael@0:                 p16[i]=(uint16_t)leadUnitValue;
michael@0:             }
michael@0:         }
michael@0:     } else {
michael@0:         uint32_t *p32;
michael@0: 
michael@0:         /* indexes to block 0 */
michael@0:         uprv_memset(p16, 0, trie->indexLength*2);
michael@0: 
michael@0:         if(leadUnitValue!=initialValue) {
michael@0:             /* indexes for lead surrogate code units to the block after Latin-1 */
michael@0:             block=(uint16_t)(latin1Length>>UTRIE_INDEX_SHIFT);
michael@0:             i=0xd800>>UTRIE_SHIFT;
michael@0:             limit=0xdc00>>UTRIE_SHIFT;
michael@0:             for(; i<limit; ++i) {
michael@0:                 p16[i]=block;
michael@0:             }
michael@0:         }
michael@0: 
michael@0:         trie->data32=p32=(uint32_t *)(p16+trie->indexLength);
michael@0: 
michael@0:         /* Latin-1 data */
michael@0:         for(i=0; i<latin1Length; ++i) {
michael@0:             p32[i]=initialValue;
michael@0:         }
michael@0: 
michael@0:         /* data for lead surrogate code units */
michael@0:         if(leadUnitValue!=initialValue) {
michael@0:             limit=latin1Length+UTRIE_DATA_BLOCK_LENGTH;
michael@0:             for(/* i=latin1Length */; i<limit; ++i) {
michael@0:                 p32[i]=leadUnitValue;
michael@0:             }
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     trie->getFoldingOffset=utrie_defaultGetFoldingOffset;
michael@0: 
michael@0:     return actualLength;
michael@0: }
michael@0: 
michael@0: /* enumeration -------------------------------------------------------------- */
michael@0: 
michael@0: /* default UTrieEnumValue() returns the input value itself */
michael@0: static uint32_t U_CALLCONV
michael@0: enumSameValue(const void * /*context*/, uint32_t value) {
michael@0:     return value;
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * Enumerate all ranges of code points with the same relevant values.
michael@0:  * The values are transformed from the raw trie entries by the enumValue function.
michael@0:  */
michael@0: U_CAPI void U_EXPORT2
michael@0: utrie_enum(const UTrie *trie,
michael@0:            UTrieEnumValue *enumValue, UTrieEnumRange *enumRange, const void *context) {
michael@0:     const uint32_t *data32;
michael@0:     const uint16_t *idx;
michael@0: 
michael@0:     uint32_t value, prevValue, initialValue;
michael@0:     UChar32 c, prev;
michael@0:     int32_t l, i, j, block, prevBlock, nullBlock, offset;
michael@0: 
michael@0:     /* check arguments */
michael@0:     if(trie==NULL || trie->index==NULL || enumRange==NULL) {
michael@0:         return;
michael@0:     }
michael@0:     if(enumValue==NULL) {
michael@0:         enumValue=enumSameValue;
michael@0:     }
michael@0: 
michael@0:     idx=trie->index;
michael@0:     data32=trie->data32;
michael@0: 
michael@0:     /* get the enumeration value that corresponds to an initial-value trie data entry */
michael@0:     initialValue=enumValue(context, trie->initialValue);
michael@0: 
michael@0:     if(data32==NULL) {
michael@0:         nullBlock=trie->indexLength;
michael@0:     } else {
michael@0:         nullBlock=0;
michael@0:     }
michael@0: 
michael@0:     /* set variables for previous range */
michael@0:     prevBlock=nullBlock;
michael@0:     prev=0;
michael@0:     prevValue=initialValue;
michael@0: 
michael@0:     /* enumerate BMP - the main loop enumerates data blocks */
michael@0:     for(i=0, c=0; c<=0xffff; ++i) {
michael@0:         if(c==0xd800) {
michael@0:             /* skip lead surrogate code _units_, go to lead surr. code _points_ */
michael@0:             i=UTRIE_BMP_INDEX_LENGTH;
michael@0:         } else if(c==0xdc00) {
michael@0:             /* go back to regular BMP code points */
michael@0:             i=c>>UTRIE_SHIFT;
michael@0:         }
michael@0: 
michael@0:         block=idx[i]<<UTRIE_INDEX_SHIFT;
michael@0:         if(block==prevBlock) {
michael@0:             /* the block is the same as the previous one, and filled with value */
michael@0:             c+=UTRIE_DATA_BLOCK_LENGTH;
michael@0:         } else if(block==nullBlock) {
michael@0:             /* this is the all-initial-value block */
michael@0:             if(prevValue!=initialValue) {
michael@0:                 if(prev<c) {
michael@0:                     if(!enumRange(context, prev, c, prevValue)) {
michael@0:                         return;
michael@0:                     }
michael@0:                 }
michael@0:                 prevBlock=nullBlock;
michael@0:                 prev=c;
michael@0:                 prevValue=initialValue;
michael@0:             }
michael@0:             c+=UTRIE_DATA_BLOCK_LENGTH;
michael@0:         } else {
michael@0:             prevBlock=block;
michael@0:             for(j=0; j<UTRIE_DATA_BLOCK_LENGTH; ++j) {
michael@0:                 value=enumValue(context, data32!=NULL ? data32[block+j] : idx[block+j]);
michael@0:                 if(value!=prevValue) {
michael@0:                     if(prev<c) {
michael@0:                         if(!enumRange(context, prev, c, prevValue)) {
michael@0:                             return;
michael@0:                         }
michael@0:                     }
michael@0:                     if(j>0) {
michael@0:                         /* the block is not filled with all the same value */
michael@0:                         prevBlock=-1;
michael@0:                     }
michael@0:                     prev=c;
michael@0:                     prevValue=value;
michael@0:                 }
michael@0:                 ++c;
michael@0:             }
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     /* enumerate supplementary code points */
michael@0:     for(l=0xd800; l<0xdc00;) {
michael@0:         /* lead surrogate access */
michael@0:         offset=idx[l>>UTRIE_SHIFT]<<UTRIE_INDEX_SHIFT;
michael@0:         if(offset==nullBlock) {
michael@0:             /* no entries for a whole block of lead surrogates */
michael@0:             if(prevValue!=initialValue) {
michael@0:                 if(prev<c) {
michael@0:                     if(!enumRange(context, prev, c, prevValue)) {
michael@0:                         return;
michael@0:                     }
michael@0:                 }
michael@0:                 prevBlock=nullBlock;
michael@0:                 prev=c;
michael@0:                 prevValue=initialValue;
michael@0:             }
michael@0: 
michael@0:             l+=UTRIE_DATA_BLOCK_LENGTH;
michael@0:             c+=UTRIE_DATA_BLOCK_LENGTH<<10;
michael@0:             continue;
michael@0:         }
michael@0: 
michael@0:         value= data32!=NULL ? data32[offset+(l&UTRIE_MASK)] : idx[offset+(l&UTRIE_MASK)];
michael@0: 
michael@0:         /* enumerate trail surrogates for this lead surrogate */
michael@0:         offset=trie->getFoldingOffset(value);
michael@0:         if(offset<=0) {
michael@0:             /* no data for this lead surrogate */
michael@0:             if(prevValue!=initialValue) {
michael@0:                 if(prev<c) {
michael@0:                     if(!enumRange(context, prev, c, prevValue)) {
michael@0:                         return;
michael@0:                     }
michael@0:                 }
michael@0:                 prevBlock=nullBlock;
michael@0:                 prev=c;
michael@0:                 prevValue=initialValue;
michael@0:             }
michael@0: 
michael@0:             /* nothing else to do for the supplementary code points for this lead surrogate */
michael@0:             c+=0x400;
michael@0:         } else {
michael@0:             /* enumerate code points for this lead surrogate */
michael@0:             i=offset;
michael@0:             offset+=UTRIE_SURROGATE_BLOCK_COUNT;
michael@0:             do {
michael@0:                 /* copy of most of the body of the BMP loop */
michael@0:                 block=idx[i]<<UTRIE_INDEX_SHIFT;
michael@0:                 if(block==prevBlock) {
michael@0:                     /* the block is the same as the previous one, and filled with value */
michael@0:                     c+=UTRIE_DATA_BLOCK_LENGTH;
michael@0:                 } else if(block==nullBlock) {
michael@0:                     /* this is the all-initial-value block */
michael@0:                     if(prevValue!=initialValue) {
michael@0:                         if(prev<c) {
michael@0:                             if(!enumRange(context, prev, c, prevValue)) {
michael@0:                                 return;
michael@0:                             }
michael@0:                         }
michael@0:                         prevBlock=nullBlock;
michael@0:                         prev=c;
michael@0:                         prevValue=initialValue;
michael@0:                     }
michael@0:                     c+=UTRIE_DATA_BLOCK_LENGTH;
michael@0:                 } else {
michael@0:                     prevBlock=block;
michael@0:                     for(j=0; j<UTRIE_DATA_BLOCK_LENGTH; ++j) {
michael@0:                         value=enumValue(context, data32!=NULL ? data32[block+j] : idx[block+j]);
michael@0:                         if(value!=prevValue) {
michael@0:                             if(prev<c) {
michael@0:                                 if(!enumRange(context, prev, c, prevValue)) {
michael@0:                                     return;
michael@0:                                 }
michael@0:                             }
michael@0:                             if(j>0) {
michael@0:                                 /* the block is not filled with all the same value */
michael@0:                                 prevBlock=-1;
michael@0:                             }
michael@0:                             prev=c;
michael@0:                             prevValue=value;
michael@0:                         }
michael@0:                         ++c;
michael@0:                     }
michael@0:                 }
michael@0:             } while(++i<offset);
michael@0:         }
michael@0: 
michael@0:         ++l;
michael@0:     }
michael@0: 
michael@0:     /* deliver last range */
michael@0:     enumRange(context, prev, c, prevValue);
michael@0: }