[apple/icu.git] / icuSources / common / rbbidata.cpp

/*
***************************************************************************
*   Copyright (C) 1999-2005 International Business Machines Corporation   *
*   and others. All rights reserved.                                      *
***************************************************************************
*/

#include "unicode/utypes.h"

#if !UCONFIG_NO_BREAK_ITERATION

#include "unicode/utypes.h"
#include "rbbidata.h"
#include "rbbirb.h"
#include "utrie.h"
#include "udatamem.h"
#include "cmemory.h"
#include "cstring.h"
#include "umutex.h"

#include "uassert.h"


//-----------------------------------------------------------------------------------
//
//   Trie access folding function.  Copied as-is from properties code in uchar.c
//
//-----------------------------------------------------------------------------------
U_CDECL_BEGIN
static int32_t U_CALLCONV
getFoldingOffset(uint32_t data) {
    /* if bit 15 is set, then the folding offset is in bits 14..0 of the 16-bit trie result */
    if(data&0x8000) {
        return (int32_t)(data&0x7fff);
    } else {
        return 0;
    }
}
U_CDECL_END

U_NAMESPACE_BEGIN

//-----------------------------------------------------------------------------
//
//    Constructors.
//
//-----------------------------------------------------------------------------
RBBIDataWrapper::RBBIDataWrapper(const RBBIDataHeader *data, UErrorCode &status) {
    init(data, status);
}

RBBIDataWrapper::RBBIDataWrapper(UDataMemory* udm, UErrorCode &status) {
    const RBBIDataHeader *d = (const RBBIDataHeader *)
        // ((char *)&(udm->pHeader->info) + udm->pHeader->info.size);
        // taking into consideration the padding added in by udata_write
        ((char *)(udm->pHeader) + udm->pHeader->dataHeader.headerSize);
    init(d, status);
    fUDataMem = udm;
}

//-----------------------------------------------------------------------------
//
//    init().   Does most of the work of construction, shared between the
//              constructors.
//
//-----------------------------------------------------------------------------
void RBBIDataWrapper::init(const RBBIDataHeader *data, UErrorCode &status) {
    if (U_FAILURE(status)) {
        return;
    }
    fHeader = data;
    if (fHeader->fMagic != 0xb1a0 || 
        !(fHeader->fFormatVersion[0] == 3 ||         // ICU 3.4 
          *(int32_t *)fHeader->fFormatVersion == 1))  // ICU 3.2 and earlier.
    {
        status = U_INVALID_FORMAT_ERROR;
        return;
    }

    fUDataMem     = NULL;
    fReverseTable = NULL;
    fSafeFwdTable = NULL;
    fSafeRevTable = NULL;
    if (data->fFTableLen != 0) {
        fForwardTable = (RBBIStateTable *)((char *)data + fHeader->fFTable);
    }
    if (data->fRTableLen != 0) {
        fReverseTable = (RBBIStateTable *)((char *)data + fHeader->fRTable);
    }
    if (data->fSFTableLen != 0) {
        fSafeFwdTable = (RBBIStateTable *)((char *)data + fHeader->fSFTable);
    }
    if (data->fSRTableLen != 0) {
        fSafeRevTable = (RBBIStateTable *)((char *)data + fHeader->fSRTable);
    }


    utrie_unserialize(&fTrie,
                       (uint8_t *)data + fHeader->fTrie,
                       fHeader->fTrieLen,
                       &status);
    if (U_FAILURE(status)) {
        return;
    }
    fTrie.getFoldingOffset=getFoldingOffset;


    fRuleSource   = (UChar *)((char *)data + fHeader->fRuleSource);
    fRuleString.setTo(TRUE, fRuleSource, -1);
    U_ASSERT(data->fRuleSourceLen > 0);

    fRuleStatusTable = (int32_t *)((char *)data + fHeader->fStatusTable);
    fStatusMaxIdx    = data->fStatusTableLen / sizeof(int32_t);

    fRefCount = 1;

#ifdef RBBI_DEBUG
    char *debugEnv = getenv("U_RBBIDEBUG");
    if (debugEnv && uprv_strstr(debugEnv, "data")) {this->printData();}
#endif
}


//-----------------------------------------------------------------------------
//
//    Destructor.     Don't call this - use removeReference() instead.
//
//-----------------------------------------------------------------------------
RBBIDataWrapper::~RBBIDataWrapper() {
    U_ASSERT(fRefCount == 0);
    if (fUDataMem) {
        udata_close(fUDataMem);
    } else {
        uprv_free((void *)fHeader);
    }
}


//-----------------------------------------------------------------------------
//
//   Operator ==    Consider two RBBIDataWrappers to be equal if they
//                  refer to the same underlying data.  Although
//                  the data wrappers are normally shared between
//                  iterator instances, it's possible to independently
//                  open the same data twice, and get two instances, which
//                  should still be ==.
//
//-----------------------------------------------------------------------------
UBool RBBIDataWrapper::operator ==(const RBBIDataWrapper &other) const {
    if (fHeader == other.fHeader) {
        return TRUE;
    }
    if (fHeader->fLength != other.fHeader->fLength) {
        return FALSE;
    }
    if (uprv_memcmp(fHeader, other.fHeader, fHeader->fLength) == 0) {
        return TRUE;
    }
    return FALSE;
}

int32_t  RBBIDataWrapper::hashCode() {
    return fHeader->fFTableLen;
}


//-----------------------------------------------------------------------------
//
//    Reference Counting.   A single RBBIDataWrapper object is shared among
//                          however many RulesBasedBreakIterator instances are
//                          referencing the same data.
//
//-----------------------------------------------------------------------------
void RBBIDataWrapper::removeReference() {
    if (umtx_atomic_dec(&fRefCount) == 0) {
        delete this;
    }
}


RBBIDataWrapper *RBBIDataWrapper::addReference() {
   umtx_atomic_inc(&fRefCount);
   return this;
}


//-----------------------------------------------------------------------------
//
//  getRuleSourceString
//
//-----------------------------------------------------------------------------
const UnicodeString &RBBIDataWrapper::getRuleSourceString() const {
    return fRuleString;
}


//-----------------------------------------------------------------------------
//
//  print   -  debugging function to dump the runtime data tables.
//
//-----------------------------------------------------------------------------
#ifdef RBBI_DEBUG
void  RBBIDataWrapper::printTable(const char *heading, const RBBIStateTable *table) {
    uint32_t   c;
    uint32_t   s;

    RBBIDebugPrintf("   %s\n", heading);

    RBBIDebugPrintf("State |  Acc  LA TagIx");
    for (c=0; c<fHeader->fCatCount; c++) {RBBIDebugPrintf("%3d ", c);}
    RBBIDebugPrintf("\n------|---------------"); for (c=0;c<fHeader->fCatCount; c++) {
        RBBIDebugPrintf("----");
    }
    RBBIDebugPrintf("\n");

    if (table == NULL) {
        RBBIDebugPrintf("         N U L L   T A B L E\n\n");
        return;
    }
    for (s=0; s<table->fNumStates; s++) {
        RBBIStateTableRow *row = (RBBIStateTableRow *)
                                  (table->fTableData + (table->fRowLen * s));
        RBBIDebugPrintf("%4d  |  %3d %3d %3d ", s, row->fAccepting, row->fLookAhead, row->fTagIdx);
        for (c=0; c<fHeader->fCatCount; c++)  {
            RBBIDebugPrintf("%3d ", row->fNextState[c]);
        }
        RBBIDebugPrintf("\n");
    }
    RBBIDebugPrintf("\n");
}
#endif


#ifdef RBBI_DEBUG
void  RBBIDataWrapper::printData() {
    RBBIDebugPrintf("RBBI Data at %p\n", (void *)fHeader);
    RBBIDebugPrintf("   Version = {%d %d %d %d}\n", fHeader->fFormatVersion[0], fHeader->fFormatVersion[1],
                                                    fHeader->fFormatVersion[2], fHeader->fFormatVersion[3]);
    RBBIDebugPrintf("   total length of data  = %d\n", fHeader->fLength);
    RBBIDebugPrintf("   number of character categories = %d\n\n", fHeader->fCatCount);

    printTable("Forward State Transition Table", fForwardTable);
    printTable("Reverse State Transition Table", fReverseTable);
    printTable("Safe Forward State Transition Table", fSafeFwdTable);
    printTable("Safe Reverse State Transition Table", fSafeRevTable);

    RBBIDebugPrintf("\nOrignal Rules source:\n");
    for (int32_t c=0; fRuleSource[c] != 0; c++) {
        RBBIDebugPrintf("%c", fRuleSource[c]);
    }
    RBBIDebugPrintf("\n\n");
}
#endif


U_NAMESPACE_END

//-----------------------------------------------------------------------------
//
//  ubrk_swap   -  byte swap and char encoding swap of RBBI data
//
//-----------------------------------------------------------------------------

U_CAPI int32_t U_EXPORT2
ubrk_swap(const UDataSwapper *ds, const void *inData, int32_t length, void *outData,
           UErrorCode *status) {

    if (status == NULL || U_FAILURE(*status)) {
        return 0;
    }
    if(ds==NULL || inData==NULL || length<-1 || (length>0 && outData==NULL)) {
        *status=U_ILLEGAL_ARGUMENT_ERROR;
        return 0;
    }

    //
    //  Check that the data header is for for break data.
    //    (Header contents are defined in genbrk.cpp)
    //
    const UDataInfo *pInfo = (const UDataInfo *)((const char *)inData+4);
    if(!(  pInfo->dataFormat[0]==0x42 &&   /* dataFormat="Brk " */
           pInfo->dataFormat[1]==0x72 &&
           pInfo->dataFormat[2]==0x6b &&
           pInfo->dataFormat[3]==0x20 &&
           pInfo->formatVersion[0]==3  )) {
        udata_printError(ds, "ubrk_swap(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized\n",
                         pInfo->dataFormat[0], pInfo->dataFormat[1],
                         pInfo->dataFormat[2], pInfo->dataFormat[3],
                         pInfo->formatVersion[0]);
        *status=U_UNSUPPORTED_ERROR;
        return 0;
    }

    //
    // Swap the data header.  (This is the generic ICU Data Header, not the RBBI Specific
    //                         RBBIDataHeader).  This swap also conveniently gets us
    //                         the size of the ICU d.h., which lets us locate the start
    //                         of the RBBI specific data.
    //
    int32_t headerSize=udata_swapDataHeader(ds, inData, length, outData, status);


    //
    // Get the RRBI Data Header, and check that it appears to be OK.
    //
    //    Note:  ICU 3.2 and earlier, RBBIDataHeader::fDataFormat was actually 
    //           an int32_t with a value of 1.  Starting with ICU 3.4,
    //           RBBI's fDataFormat matches the dataFormat field from the
    //           UDataInfo header, four int8_t bytes.  The value is {3,1,0,0}
    //
    const uint8_t  *inBytes =(const uint8_t *)inData+headerSize;
    RBBIDataHeader *rbbiDH = (RBBIDataHeader *)inBytes;
    UBool           formatVersionOne = ds->readUInt32(*(int32_t *)rbbiDH->fFormatVersion) == 1;
    if (ds->readUInt32(rbbiDH->fMagic)   != 0xb1a0 ||
        !(formatVersionOne || rbbiDH->fFormatVersion[0] == 3)   ||
        ds->readUInt32(rbbiDH->fLength)  <  sizeof(RBBIDataHeader)) 
    {
        udata_printError(ds, "ubrk_swap(): RBBI Data header is invalid.\n");
        *status=U_UNSUPPORTED_ERROR;
        return 0;
    }

    //
    // Prefight operation?  Just return the size
    //
    int32_t breakDataLength = ds->readUInt32(rbbiDH->fLength);
    int32_t totalSize = headerSize + breakDataLength;
    if (length < 0) {
        return totalSize;
    }

    //
    // Check that length passed in is consistent with length from RBBI data header.
    //
    if (length < totalSize) {
        udata_printError(ds, "ubrk_swap(): too few bytes (%d after ICU Data header) for break data.\n",
                            breakDataLength);
        *status=U_INDEX_OUTOFBOUNDS_ERROR;
        return 0;
        }


    //
    // Swap the Data.  Do the data itself first, then the RBBI Data Header, because
    //                 we need to reference the header to locate the data, and an
    //                 inplace swap of the header leaves it unusable.
    //
    uint8_t         *outBytes = (uint8_t *)outData + headerSize;
    RBBIDataHeader  *outputDH = (RBBIDataHeader *)outBytes;

    int32_t   tableStartOffset;
    int32_t   tableLength;

    //
    // If not swapping in place, zero out the output buffer before starting.
    //    Individual tables and other data items within are aligned to 8 byte boundaries
    //    when originally created.  Any unused space between items needs to be zero.
    //
    if (inBytes != outBytes) {
        uprv_memset(outBytes, 0, breakDataLength);
    }

    //
    // Each state table begins with several 32 bit fields.  Calculate the size
    //   in bytes of these.
    //
    RBBIStateTable *stp = NULL;
    int32_t         topSize = (char *)stp->fTableData - (char *)stp;

    // Forward state table.  
    tableStartOffset = ds->readUInt32(rbbiDH->fFTable);
    tableLength      = ds->readUInt32(rbbiDH->fFTableLen);

    if (tableLength > 0) {
        ds->swapArray32(ds, inBytes+tableStartOffset, topSize, 
                            outBytes+tableStartOffset, status);
        ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize,
                            outBytes+tableStartOffset+topSize, status);
    }
    
    // Reverse state table.  Same layout as forward table, above.
    tableStartOffset = ds->readUInt32(rbbiDH->fRTable);
    tableLength      = ds->readUInt32(rbbiDH->fRTableLen);

    if (tableLength > 0) {
        ds->swapArray32(ds, inBytes+tableStartOffset, topSize, 
                            outBytes+tableStartOffset, status);
        ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize,
                            outBytes+tableStartOffset+topSize, status);
    }

    // Safe Forward state table.  Same layout as forward table, above.
    tableStartOffset = ds->readUInt32(rbbiDH->fSFTable);
    tableLength      = ds->readUInt32(rbbiDH->fSFTableLen);

    if (tableLength > 0) {
        ds->swapArray32(ds, inBytes+tableStartOffset, topSize, 
                            outBytes+tableStartOffset, status);
        ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize,
                            outBytes+tableStartOffset+topSize, status);
    }

    // Safe Reverse state table.  Same layout as forward table, above.
    tableStartOffset = ds->readUInt32(rbbiDH->fSRTable);
    tableLength      = ds->readUInt32(rbbiDH->fSRTableLen);

    if (tableLength > 0) {
        ds->swapArray32(ds, inBytes+tableStartOffset, topSize, 
                            outBytes+tableStartOffset, status);
        ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize,
                            outBytes+tableStartOffset+topSize, status);
    }

    // Trie table for character categories
    utrie_swap(ds, inBytes+ds->readUInt32(rbbiDH->fTrie), ds->readUInt32(rbbiDH->fTrieLen),
                            outBytes+ds->readUInt32(rbbiDH->fTrie), status);

    // Source Rules Text.  It's UChar data
    ds->swapArray16(ds, inBytes+ds->readUInt32(rbbiDH->fRuleSource), ds->readUInt32(rbbiDH->fRuleSourceLen),
                        outBytes+ds->readUInt32(rbbiDH->fRuleSource), status);

    // Table of rule status values.  It's all int_32 values
    ds->swapArray32(ds, inBytes+ds->readUInt32(rbbiDH->fStatusTable), ds->readUInt32(rbbiDH->fStatusTableLen),
                        outBytes+ds->readUInt32(rbbiDH->fStatusTable), status);

    // And, last, the header.
    //   For the old version one format, the entire header consists of int32_t values.
    //   For the newer formats, the fDataFormat field is an array of four bytes.
    //   Swap the whole thing as int32_t, then, for the newer format, re-swap the one field.
    //
    ds->swapArray32(ds, inBytes, sizeof(RBBIDataHeader), outBytes, status);
    if (formatVersionOne == FALSE) {
        ds->swapArray32(ds, outputDH->fFormatVersion, 4, outputDH->fFormatVersion, status);
    }


    return totalSize;
}


#endif /* #if !UCONFIG_NO_BREAK_ITERATION */
Commit	Line	Data
b75a7d8f A	1	/*
b75a7d8f A	2	***************************************************************************
73c04bcf	3	* Copyright (C) 1999-2005 International Business Machines Corporation *
b75a7d8f A	4	* and others. All rights reserved. *
	5	***************************************************************************
	6	*/
	7
	8	#include "unicode/utypes.h"
	9
	10	#if !UCONFIG_NO_BREAK_ITERATION
	11
	12	#include "unicode/utypes.h"
	13	#include "rbbidata.h"
	14	#include "rbbirb.h"
	15	#include "utrie.h"
	16	#include "udatamem.h"
	17	#include "cmemory.h"
	18	#include "cstring.h"
	19	#include "umutex.h"
	20
	21	#include "uassert.h"
	22
	23
	24	//-----------------------------------------------------------------------------------
	25	//
	26	// Trie access folding function. Copied as-is from properties code in uchar.c
	27	//
	28	//-----------------------------------------------------------------------------------
	29	U_CDECL_BEGIN
	30	static int32_t U_CALLCONV
	31	getFoldingOffset(uint32_t data) {
	32	/* if bit 15 is set, then the folding offset is in bits 14..0 of the 16-bit trie result */
	33	if(data&0x8000) {
	34	return (int32_t)(data&0x7fff);
	35	} else {
	36	return 0;
	37	}
	38	}
	39	U_CDECL_END
	40
	41	U_NAMESPACE_BEGIN
	42
	43	//-----------------------------------------------------------------------------
	44	//
	45	// Constructors.
	46	//
	47	//-----------------------------------------------------------------------------
	48	RBBIDataWrapper::RBBIDataWrapper(const RBBIDataHeader *data, UErrorCode &status) {
	49	init(data, status);
	50	}
	51
	52	RBBIDataWrapper::RBBIDataWrapper(UDataMemory* udm, UErrorCode &status) {
	53	const RBBIDataHeader d = (const RBBIDataHeader )
374ca955 A	54	// ((char *)&(udm->pHeader->info) + udm->pHeader->info.size);
	55	// taking into consideration the padding added in by udata_write
	56	((char *)(udm->pHeader) + udm->pHeader->dataHeader.headerSize);
b75a7d8f A	57	init(d, status);
	58	fUDataMem = udm;
	59	}
	60
	61	//-----------------------------------------------------------------------------
	62	//
	63	// init(). Does most of the work of construction, shared between the
	64	// constructors.
	65	//
	66	//-----------------------------------------------------------------------------
	67	void RBBIDataWrapper::init(const RBBIDataHeader *data, UErrorCode &status) {
	68	if (U_FAILURE(status)) {
	69	return;
	70	}
	71	fHeader = data;
73c04bcf A	72	if (fHeader->fMagic != 0xb1a0 \|\|
	73	!(fHeader->fFormatVersion[0] == 3 \|\| // ICU 3.4
	74	(int32_t )fHeader->fFormatVersion == 1)) // ICU 3.2 and earlier.
	75	{
	76	status = U_INVALID_FORMAT_ERROR;
b75a7d8f A	77	return;
	78	}
	79
	80	fUDataMem = NULL;
b75a7d8f	81	fReverseTable = NULL;
374ca955 A	82	fSafeFwdTable = NULL;
	83	fSafeRevTable = NULL;
	84	if (data->fFTableLen != 0) {
	85	fForwardTable = (RBBIStateTable )((char )data + fHeader->fFTable);
	86	}
b75a7d8f A	87	if (data->fRTableLen != 0) {
	88	fReverseTable = (RBBIStateTable )((char )data + fHeader->fRTable);
	89	}
374ca955 A	90	if (data->fSFTableLen != 0) {
	91	fSafeFwdTable = (RBBIStateTable )((char )data + fHeader->fSFTable);
	92	}
	93	if (data->fSRTableLen != 0) {
	94	fSafeRevTable = (RBBIStateTable )((char )data + fHeader->fSRTable);
	95	}
b75a7d8f A	96
	97
	98	utrie_unserialize(&fTrie,
	99	(uint8_t *)data + fHeader->fTrie,
	100	fHeader->fTrieLen,
	101	&status);
	102	if (U_FAILURE(status)) {
	103	return;
	104	}
	105	fTrie.getFoldingOffset=getFoldingOffset;
	106
	107
	108	fRuleSource = (UChar )((char )data + fHeader->fRuleSource);
	109	fRuleString.setTo(TRUE, fRuleSource, -1);
374ca955 A	110	U_ASSERT(data->fRuleSourceLen > 0);
	111
	112	fRuleStatusTable = (int32_t )((char )data + fHeader->fStatusTable);
	113	fStatusMaxIdx = data->fStatusTableLen / sizeof(int32_t);
b75a7d8f A	114
	115	fRefCount = 1;
	116
	117	#ifdef RBBI_DEBUG
	118	char *debugEnv = getenv("U_RBBIDEBUG");
	119	if (debugEnv && uprv_strstr(debugEnv, "data")) {this->printData();}
	120	#endif
	121	}
	122
	123
	124	//-----------------------------------------------------------------------------
	125	//
374ca955	126	// Destructor. Don't call this - use removeReference() instead.
b75a7d8f A	127	//
	128	//-----------------------------------------------------------------------------
	129	RBBIDataWrapper::~RBBIDataWrapper() {
	130	U_ASSERT(fRefCount == 0);
	131	if (fUDataMem) {
	132	udata_close(fUDataMem);
	133	} else {
	134	uprv_free((void *)fHeader);
	135	}
	136	}
	137
	138
	139
	140	//-----------------------------------------------------------------------------
	141	//
	142	// Operator == Consider two RBBIDataWrappers to be equal if they
	143	// refer to the same underlying data. Although
	144	// the data wrappers are normally shared between
	145	// iterator instances, it's possible to independently
	146	// open the same data twice, and get two instances, which
	147	// should still be ==.
	148	//
	149	//-----------------------------------------------------------------------------
	150	UBool RBBIDataWrapper::operator ==(const RBBIDataWrapper &other) const {
	151	if (fHeader == other.fHeader) {
	152	return TRUE;
	153	}
	154	if (fHeader->fLength != other.fHeader->fLength) {
	155	return FALSE;
	156	}
	157	if (uprv_memcmp(fHeader, other.fHeader, fHeader->fLength) == 0) {
	158	return TRUE;
	159	}
	160	return FALSE;
	161	}
	162
	163	int32_t RBBIDataWrapper::hashCode() {
	164	return fHeader->fFTableLen;
	165	}
	166
	167
	168
	169	//-----------------------------------------------------------------------------
	170	//
	171	// Reference Counting. A single RBBIDataWrapper object is shared among
	172	// however many RulesBasedBreakIterator instances are
	173	// referencing the same data.
	174	//
	175	//-----------------------------------------------------------------------------
	176	void RBBIDataWrapper::removeReference() {
	177	if (umtx_atomic_dec(&fRefCount) == 0) {
	178	delete this;
	179	}
	180	}
	181
	182
	183	RBBIDataWrapper *RBBIDataWrapper::addReference() {
	184	umtx_atomic_inc(&fRefCount);
	185	return this;
	186	}
	187
	188
	189
	190	//-----------------------------------------------------------------------------
191	//
192	// getRuleSourceString
193	//
194	//-----------------------------------------------------------------------------
374ca955	195	const UnicodeString &RBBIDataWrapper::getRuleSourceString() const {
b75a7d8f A	196	return fRuleString;
	197	}
	198
	199
	200	//-----------------------------------------------------------------------------
	201	//
	202	// print - debugging function to dump the runtime data tables.
	203	//
	204	//-----------------------------------------------------------------------------
b75a7d8f	205	#ifdef RBBI_DEBUG
374ca955 A	206	void RBBIDataWrapper::printTable(const char heading, const RBBIStateTable table) {
	207	uint32_t c;
	208	uint32_t s;
b75a7d8f	209
374ca955	210	RBBIDebugPrintf(" %s\n", heading);
b75a7d8f	211
374ca955	212	RBBIDebugPrintf("State \| Acc LA TagIx");
b75a7d8f	213	for (c=0; c<fHeader->fCatCount; c++) {RBBIDebugPrintf("%3d ", c);}
374ca955 A	214	RBBIDebugPrintf("\n------\|---------------"); for (c=0;c<fHeader->fCatCount; c++) {
	215	RBBIDebugPrintf("----");
	216	}
b75a7d8f A	217	RBBIDebugPrintf("\n");
b75a7d8f A	218
374ca955 A	219	if (table == NULL) {
	220	RBBIDebugPrintf(" N U L L T A B L E\n\n");
	221	return;
	222	}
	223	for (s=0; s<table->fNumStates; s++) {
b75a7d8f	224	RBBIStateTableRow row = (RBBIStateTableRow )
374ca955 A	225	(table->fTableData + (table->fRowLen * s));
374ca955 A	226	RBBIDebugPrintf("%4d \| %3d %3d %3d ", s, row->fAccepting, row->fLookAhead, row->fTagIdx);
b75a7d8f A	227	for (c=0; c<fHeader->fCatCount; c++) {
	228	RBBIDebugPrintf("%3d ", row->fNextState[c]);
	229	}
	230	RBBIDebugPrintf("\n");
	231	}
374ca955 A	232	RBBIDebugPrintf("\n");
	233	}
	234	#endif
	235
	236
	237	#ifdef RBBI_DEBUG
	238	void RBBIDataWrapper::printData() {
	239	RBBIDebugPrintf("RBBI Data at %p\n", (void *)fHeader);
73c04bcf A	240	RBBIDebugPrintf(" Version = {%d %d %d %d}\n", fHeader->fFormatVersion[0], fHeader->fFormatVersion[1],
73c04bcf A	241	fHeader->fFormatVersion[2], fHeader->fFormatVersion[3]);
374ca955 A	242	RBBIDebugPrintf(" total length of data = %d\n", fHeader->fLength);
	243	RBBIDebugPrintf(" number of character categories = %d\n\n", fHeader->fCatCount);
	244
	245	printTable("Forward State Transition Table", fForwardTable);
	246	printTable("Reverse State Transition Table", fReverseTable);
	247	printTable("Safe Forward State Transition Table", fSafeFwdTable);
	248	printTable("Safe Reverse State Transition Table", fSafeRevTable);
b75a7d8f A	249
b75a7d8f A	250	RBBIDebugPrintf("\nOrignal Rules source:\n");
374ca955	251	for (int32_t c=0; fRuleSource[c] != 0; c++) {
b75a7d8f	252	RBBIDebugPrintf("%c", fRuleSource[c]);
b75a7d8f A	253	}
b75a7d8f A	254	RBBIDebugPrintf("\n\n");
b75a7d8f	255	}
374ca955	256	#endif
b75a7d8f A	257
	258
	259	U_NAMESPACE_END
	260
374ca955 A	261	//-----------------------------------------------------------------------------
	262	//
	263	// ubrk_swap - byte swap and char encoding swap of RBBI data
	264	//
	265	//-----------------------------------------------------------------------------
	266
	267	U_CAPI int32_t U_EXPORT2
	268	ubrk_swap(const UDataSwapper ds, const void inData, int32_t length, void *outData,
	269	UErrorCode *status) {
	270
	271	if (status == NULL \|\| U_FAILURE(*status)) {
	272	return 0;
	273	}
73c04bcf A	274	if(ds==NULL \|\| inData==NULL \|\| length<-1 \|\| (length>0 && outData==NULL)) {
	275	*status=U_ILLEGAL_ARGUMENT_ERROR;
	276	return 0;
	277	}
374ca955 A	278
	279	//
	280	// Check that the data header is for for break data.
	281	// (Header contents are defined in genbrk.cpp)
	282	//
	283	const UDataInfo pInfo = (const UDataInfo )((const char *)inData+4);
	284	if(!( pInfo->dataFormat[0]==0x42 && /* dataFormat="Brk " */
	285	pInfo->dataFormat[1]==0x72 &&
	286	pInfo->dataFormat[2]==0x6b &&
	287	pInfo->dataFormat[3]==0x20 &&
	288	pInfo->formatVersion[0]==3 )) {
	289	udata_printError(ds, "ubrk_swap(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized\n",
	290	pInfo->dataFormat[0], pInfo->dataFormat[1],
	291	pInfo->dataFormat[2], pInfo->dataFormat[3],
	292	pInfo->formatVersion[0]);
	293	*status=U_UNSUPPORTED_ERROR;
	294	return 0;
	295	}
	296
	297	//
	298	// Swap the data header. (This is the generic ICU Data Header, not the RBBI Specific
	299	// RBBIDataHeader). This swap also conveniently gets us
	300	// the size of the ICU d.h., which lets us locate the start
	301	// of the RBBI specific data.
	302	//
	303	int32_t headerSize=udata_swapDataHeader(ds, inData, length, outData, status);
	304
	305
	306	//
	307	// Get the RRBI Data Header, and check that it appears to be OK.
	308	//
73c04bcf A	309	// Note: ICU 3.2 and earlier, RBBIDataHeader::fDataFormat was actually
	310	// an int32_t with a value of 1. Starting with ICU 3.4,
	311	// RBBI's fDataFormat matches the dataFormat field from the
	312	// UDataInfo header, four int8_t bytes. The value is {3,1,0,0}
	313	//
374ca955 A	314	const uint8_t inBytes =(const uint8_t )inData+headerSize;
374ca955 A	315	RBBIDataHeader rbbiDH = (RBBIDataHeader )inBytes;
73c04bcf	316	UBool formatVersionOne = ds->readUInt32((int32_t )rbbiDH->fFormatVersion) == 1;
374ca955	317	if (ds->readUInt32(rbbiDH->fMagic) != 0xb1a0 \|\|
73c04bcf	318	!(formatVersionOne \|\| rbbiDH->fFormatVersion[0] == 3) \|\|
374ca955 A	319	ds->readUInt32(rbbiDH->fLength) < sizeof(RBBIDataHeader))
	320	{
	321	udata_printError(ds, "ubrk_swap(): RBBI Data header is invalid.\n");
	322	*status=U_UNSUPPORTED_ERROR;
	323	return 0;
	324	}
	325
	326	//
	327	// Prefight operation? Just return the size
	328	//
73c04bcf A	329	int32_t breakDataLength = ds->readUInt32(rbbiDH->fLength);
73c04bcf A	330	int32_t totalSize = headerSize + breakDataLength;
374ca955 A	331	if (length < 0) {
	332	return totalSize;
	333	}
	334
	335	//
	336	// Check that length passed in is consistent with length from RBBI data header.
	337	//
73c04bcf A	338	if (length < totalSize) {
	339	udata_printError(ds, "ubrk_swap(): too few bytes (%d after ICU Data header) for break data.\n",
	340	breakDataLength);
	341	*status=U_INDEX_OUTOFBOUNDS_ERROR;
	342	return 0;
374ca955	343	}
374ca955 A	344
	345
	346	//
	347	// Swap the Data. Do the data itself first, then the RBBI Data Header, because
	348	// we need to reference the header to locate the data, and an
	349	// inplace swap of the header leaves it unusable.
	350	//
73c04bcf A	351	uint8_t outBytes = (uint8_t )outData + headerSize;
	352	RBBIDataHeader outputDH = (RBBIDataHeader )outBytes;
	353
374ca955 A	354	int32_t tableStartOffset;
	355	int32_t tableLength;
	356
	357	//
	358	// If not swapping in place, zero out the output buffer before starting.
	359	// Individual tables and other data items within are aligned to 8 byte boundaries
	360	// when originally created. Any unused space between items needs to be zero.
	361	//
	362	if (inBytes != outBytes) {
73c04bcf	363	uprv_memset(outBytes, 0, breakDataLength);
374ca955 A	364	}
	365
	366	//
	367	// Each state table begins with several 32 bit fields. Calculate the size
	368	// in bytes of these.
	369	//
	370	RBBIStateTable *stp = NULL;
	371	int32_t topSize = (char )stp->fTableData - (char )stp;
	372
	373	// Forward state table.
	374	tableStartOffset = ds->readUInt32(rbbiDH->fFTable);
	375	tableLength = ds->readUInt32(rbbiDH->fFTableLen);
	376
	377	if (tableLength > 0) {
	378	ds->swapArray32(ds, inBytes+tableStartOffset, topSize,
	379	outBytes+tableStartOffset, status);
	380	ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize,
	381	outBytes+tableStartOffset+topSize, status);
	382	}
	383
	384	// Reverse state table. Same layout as forward table, above.
	385	tableStartOffset = ds->readUInt32(rbbiDH->fRTable);
	386	tableLength = ds->readUInt32(rbbiDH->fRTableLen);
	387
	388	if (tableLength > 0) {
	389	ds->swapArray32(ds, inBytes+tableStartOffset, topSize,
	390	outBytes+tableStartOffset, status);
	391	ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize,
	392	outBytes+tableStartOffset+topSize, status);
	393	}
	394
	395	// Safe Forward state table. Same layout as forward table, above.
	396	tableStartOffset = ds->readUInt32(rbbiDH->fSFTable);
	397	tableLength = ds->readUInt32(rbbiDH->fSFTableLen);
	398
	399	if (tableLength > 0) {
	400	ds->swapArray32(ds, inBytes+tableStartOffset, topSize,
	401	outBytes+tableStartOffset, status);
	402	ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize,
	403	outBytes+tableStartOffset+topSize, status);
	404	}
	405
	406	// Safe Reverse state table. Same layout as forward table, above.
	407	tableStartOffset = ds->readUInt32(rbbiDH->fSRTable);
	408	tableLength = ds->readUInt32(rbbiDH->fSRTableLen);
	409
	410	if (tableLength > 0) {
	411	ds->swapArray32(ds, inBytes+tableStartOffset, topSize,
	412	outBytes+tableStartOffset, status);
	413	ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize,
	414	outBytes+tableStartOffset+topSize, status);
	415	}
	416
	417	// Trie table for character categories
	418	utrie_swap(ds, inBytes+ds->readUInt32(rbbiDH->fTrie), ds->readUInt32(rbbiDH->fTrieLen),
	419	outBytes+ds->readUInt32(rbbiDH->fTrie), status);
	420
	421	// Source Rules Text. It's UChar data
	422	ds->swapArray16(ds, inBytes+ds->readUInt32(rbbiDH->fRuleSource), ds->readUInt32(rbbiDH->fRuleSourceLen),
	423	outBytes+ds->readUInt32(rbbiDH->fRuleSource), status);
	424
	425	// Table of rule status values. It's all int_32 values
	426	ds->swapArray32(ds, inBytes+ds->readUInt32(rbbiDH->fStatusTable), ds->readUInt32(rbbiDH->fStatusTableLen),
	427	outBytes+ds->readUInt32(rbbiDH->fStatusTable), status);
428
73c04bcf A	429	// And, last, the header.
	430	// For the old version one format, the entire header consists of int32_t values.
	431	// For the newer formats, the fDataFormat field is an array of four bytes.
	432	// Swap the whole thing as int32_t, then, for the newer format, re-swap the one field.
	433	//
	434	ds->swapArray32(ds, inBytes, sizeof(RBBIDataHeader), outBytes, status);
	435	if (formatVersionOne == FALSE) {
	436	ds->swapArray32(ds, outputDH->fFormatVersion, 4, outputDH->fFormatVersion, status);
	437	}
	438
374ca955 A	439
	440	return totalSize;
	441	}
	442
	443
b75a7d8f	444	#endif /* #if !UCONFIG_NO_BREAK_ITERATION */