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

/*
***************************************************************************
*   Copyright (C) 1999-2008 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(const RBBIDataHeader *data, enum EDontAdopt, UErrorCode &status) {
    init(data, status);
    fDontFreeData = TRUE;
}

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;
    }

    fDontFreeData = FALSE;
    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 if (!fDontFreeData) {
        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
U_NAMESPACE_USE

//-----------------------------------------------------------------------------
//
//  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.
    //
    int32_t         topSize = offsetof(RBBIStateTable, fTableData);

    // 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	***************************************************************************
46f4442e	3	* Copyright (C) 1999-2008 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
46f4442e A	52	RBBIDataWrapper::RBBIDataWrapper(const RBBIDataHeader *data, enum EDontAdopt, UErrorCode &status) {
	53	init(data, status);
	54	fDontFreeData = TRUE;
	55	}
	56
b75a7d8f A	57	RBBIDataWrapper::RBBIDataWrapper(UDataMemory* udm, UErrorCode &status) {
b75a7d8f A	58	const RBBIDataHeader d = (const RBBIDataHeader )
374ca955 A	59	// ((char *)&(udm->pHeader->info) + udm->pHeader->info.size);
	60	// taking into consideration the padding added in by udata_write
	61	((char *)(udm->pHeader) + udm->pHeader->dataHeader.headerSize);
b75a7d8f A	62	init(d, status);
	63	fUDataMem = udm;
	64	}
	65
	66	//-----------------------------------------------------------------------------
	67	//
	68	// init(). Does most of the work of construction, shared between the
	69	// constructors.
	70	//
	71	//-----------------------------------------------------------------------------
	72	void RBBIDataWrapper::init(const RBBIDataHeader *data, UErrorCode &status) {
	73	if (U_FAILURE(status)) {
	74	return;
	75	}
	76	fHeader = data;
73c04bcf A	77	if (fHeader->fMagic != 0xb1a0 \|\|
	78	!(fHeader->fFormatVersion[0] == 3 \|\| // ICU 3.4
	79	(int32_t )fHeader->fFormatVersion == 1)) // ICU 3.2 and earlier.
	80	{
	81	status = U_INVALID_FORMAT_ERROR;
b75a7d8f A	82	return;
	83	}
	84
46f4442e	85	fDontFreeData = FALSE;
b75a7d8f	86	fUDataMem = NULL;
b75a7d8f	87	fReverseTable = NULL;
374ca955 A	88	fSafeFwdTable = NULL;
	89	fSafeRevTable = NULL;
	90	if (data->fFTableLen != 0) {
	91	fForwardTable = (RBBIStateTable )((char )data + fHeader->fFTable);
	92	}
b75a7d8f A	93	if (data->fRTableLen != 0) {
	94	fReverseTable = (RBBIStateTable )((char )data + fHeader->fRTable);
	95	}
374ca955 A	96	if (data->fSFTableLen != 0) {
	97	fSafeFwdTable = (RBBIStateTable )((char )data + fHeader->fSFTable);
	98	}
	99	if (data->fSRTableLen != 0) {
	100	fSafeRevTable = (RBBIStateTable )((char )data + fHeader->fSRTable);
	101	}
b75a7d8f A	102
	103
	104	utrie_unserialize(&fTrie,
	105	(uint8_t *)data + fHeader->fTrie,
	106	fHeader->fTrieLen,
	107	&status);
	108	if (U_FAILURE(status)) {
	109	return;
	110	}
	111	fTrie.getFoldingOffset=getFoldingOffset;
	112
	113
	114	fRuleSource = (UChar )((char )data + fHeader->fRuleSource);
	115	fRuleString.setTo(TRUE, fRuleSource, -1);
374ca955 A	116	U_ASSERT(data->fRuleSourceLen > 0);
	117
	118	fRuleStatusTable = (int32_t )((char )data + fHeader->fStatusTable);
	119	fStatusMaxIdx = data->fStatusTableLen / sizeof(int32_t);
b75a7d8f A	120
	121	fRefCount = 1;
	122
	123	#ifdef RBBI_DEBUG
	124	char *debugEnv = getenv("U_RBBIDEBUG");
	125	if (debugEnv && uprv_strstr(debugEnv, "data")) {this->printData();}
	126	#endif
	127	}
	128
	129
	130	//-----------------------------------------------------------------------------
	131	//
374ca955	132	// Destructor. Don't call this - use removeReference() instead.
b75a7d8f A	133	//
	134	//-----------------------------------------------------------------------------
	135	RBBIDataWrapper::~RBBIDataWrapper() {
	136	U_ASSERT(fRefCount == 0);
	137	if (fUDataMem) {
	138	udata_close(fUDataMem);
46f4442e	139	} else if (!fDontFreeData) {
b75a7d8f A	140	uprv_free((void *)fHeader);
	141	}
	142	}
	143
	144
	145
	146	//-----------------------------------------------------------------------------
	147	//
	148	// Operator == Consider two RBBIDataWrappers to be equal if they
	149	// refer to the same underlying data. Although
	150	// the data wrappers are normally shared between
	151	// iterator instances, it's possible to independently
	152	// open the same data twice, and get two instances, which
	153	// should still be ==.
	154	//
	155	//-----------------------------------------------------------------------------
	156	UBool RBBIDataWrapper::operator ==(const RBBIDataWrapper &other) const {
	157	if (fHeader == other.fHeader) {
	158	return TRUE;
	159	}
	160	if (fHeader->fLength != other.fHeader->fLength) {
	161	return FALSE;
	162	}
	163	if (uprv_memcmp(fHeader, other.fHeader, fHeader->fLength) == 0) {
	164	return TRUE;
	165	}
	166	return FALSE;
	167	}
	168
	169	int32_t RBBIDataWrapper::hashCode() {
	170	return fHeader->fFTableLen;
	171	}
	172
	173
	174
	175	//-----------------------------------------------------------------------------
	176	//
	177	// Reference Counting. A single RBBIDataWrapper object is shared among
	178	// however many RulesBasedBreakIterator instances are
	179	// referencing the same data.
	180	//
	181	//-----------------------------------------------------------------------------
	182	void RBBIDataWrapper::removeReference() {
	183	if (umtx_atomic_dec(&fRefCount) == 0) {
	184	delete this;
	185	}
	186	}
	187
	188
	189	RBBIDataWrapper *RBBIDataWrapper::addReference() {
	190	umtx_atomic_inc(&fRefCount);
	191	return this;
	192	}
	193
	194
	195
	196	//-----------------------------------------------------------------------------
	197	//
	198	// getRuleSourceString
	199	//
	200	//-----------------------------------------------------------------------------
374ca955	201	const UnicodeString &RBBIDataWrapper::getRuleSourceString() const {
b75a7d8f A	202	return fRuleString;
	203	}
	204
	205
	206	//-----------------------------------------------------------------------------
	207	//
	208	// print - debugging function to dump the runtime data tables.
	209	//
	210	//-----------------------------------------------------------------------------
b75a7d8f	211	#ifdef RBBI_DEBUG
374ca955 A	212	void RBBIDataWrapper::printTable(const char heading, const RBBIStateTable table) {
	213	uint32_t c;
	214	uint32_t s;
b75a7d8f	215
374ca955	216	RBBIDebugPrintf(" %s\n", heading);
b75a7d8f	217
374ca955	218	RBBIDebugPrintf("State \| Acc LA TagIx");
b75a7d8f	219	for (c=0; c<fHeader->fCatCount; c++) {RBBIDebugPrintf("%3d ", c);}
374ca955 A	220	RBBIDebugPrintf("\n------\|---------------"); for (c=0;c<fHeader->fCatCount; c++) {
	221	RBBIDebugPrintf("----");
	222	}
b75a7d8f A	223	RBBIDebugPrintf("\n");
b75a7d8f A	224
374ca955 A	225	if (table == NULL) {
	226	RBBIDebugPrintf(" N U L L T A B L E\n\n");
	227	return;
	228	}
	229	for (s=0; s<table->fNumStates; s++) {
b75a7d8f	230	RBBIStateTableRow row = (RBBIStateTableRow )
374ca955 A	231	(table->fTableData + (table->fRowLen * s));
374ca955 A	232	RBBIDebugPrintf("%4d \| %3d %3d %3d ", s, row->fAccepting, row->fLookAhead, row->fTagIdx);
b75a7d8f A	233	for (c=0; c<fHeader->fCatCount; c++) {
	234	RBBIDebugPrintf("%3d ", row->fNextState[c]);
	235	}
	236	RBBIDebugPrintf("\n");
	237	}
374ca955 A	238	RBBIDebugPrintf("\n");
	239	}
	240	#endif
	241
	242
	243	#ifdef RBBI_DEBUG
	244	void RBBIDataWrapper::printData() {
	245	RBBIDebugPrintf("RBBI Data at %p\n", (void *)fHeader);
73c04bcf A	246	RBBIDebugPrintf(" Version = {%d %d %d %d}\n", fHeader->fFormatVersion[0], fHeader->fFormatVersion[1],
73c04bcf A	247	fHeader->fFormatVersion[2], fHeader->fFormatVersion[3]);
374ca955 A	248	RBBIDebugPrintf(" total length of data = %d\n", fHeader->fLength);
	249	RBBIDebugPrintf(" number of character categories = %d\n\n", fHeader->fCatCount);
	250
	251	printTable("Forward State Transition Table", fForwardTable);
	252	printTable("Reverse State Transition Table", fReverseTable);
	253	printTable("Safe Forward State Transition Table", fSafeFwdTable);
	254	printTable("Safe Reverse State Transition Table", fSafeRevTable);
b75a7d8f A	255
b75a7d8f A	256	RBBIDebugPrintf("\nOrignal Rules source:\n");
374ca955	257	for (int32_t c=0; fRuleSource[c] != 0; c++) {
b75a7d8f	258	RBBIDebugPrintf("%c", fRuleSource[c]);
b75a7d8f A	259	}
b75a7d8f A	260	RBBIDebugPrintf("\n\n");
b75a7d8f	261	}
374ca955	262	#endif
b75a7d8f A	263
	264
	265	U_NAMESPACE_END
46f4442e	266	U_NAMESPACE_USE
b75a7d8f	267
374ca955 A	268	//-----------------------------------------------------------------------------
	269	//
	270	// ubrk_swap - byte swap and char encoding swap of RBBI data
	271	//
	272	//-----------------------------------------------------------------------------
	273
	274	U_CAPI int32_t U_EXPORT2
	275	ubrk_swap(const UDataSwapper ds, const void inData, int32_t length, void *outData,
	276	UErrorCode *status) {
	277
	278	if (status == NULL \|\| U_FAILURE(*status)) {
	279	return 0;
	280	}
73c04bcf A	281	if(ds==NULL \|\| inData==NULL \|\| length<-1 \|\| (length>0 && outData==NULL)) {
	282	*status=U_ILLEGAL_ARGUMENT_ERROR;
	283	return 0;
	284	}
374ca955 A	285
	286	//
	287	// Check that the data header is for for break data.
	288	// (Header contents are defined in genbrk.cpp)
	289	//
	290	const UDataInfo pInfo = (const UDataInfo )((const char *)inData+4);
	291	if(!( pInfo->dataFormat[0]==0x42 && /* dataFormat="Brk " */
	292	pInfo->dataFormat[1]==0x72 &&
	293	pInfo->dataFormat[2]==0x6b &&
	294	pInfo->dataFormat[3]==0x20 &&
	295	pInfo->formatVersion[0]==3 )) {
	296	udata_printError(ds, "ubrk_swap(): data format %02x.%02x.%02x.%02x (format version %02x) is not recognized\n",
	297	pInfo->dataFormat[0], pInfo->dataFormat[1],
	298	pInfo->dataFormat[2], pInfo->dataFormat[3],
	299	pInfo->formatVersion[0]);
	300	*status=U_UNSUPPORTED_ERROR;
	301	return 0;
	302	}
	303
	304	//
	305	// Swap the data header. (This is the generic ICU Data Header, not the RBBI Specific
	306	// RBBIDataHeader). This swap also conveniently gets us
	307	// the size of the ICU d.h., which lets us locate the start
	308	// of the RBBI specific data.
	309	//
	310	int32_t headerSize=udata_swapDataHeader(ds, inData, length, outData, status);
	311
	312
	313	//
	314	// Get the RRBI Data Header, and check that it appears to be OK.
	315	//
73c04bcf A	316	// Note: ICU 3.2 and earlier, RBBIDataHeader::fDataFormat was actually
	317	// an int32_t with a value of 1. Starting with ICU 3.4,
	318	// RBBI's fDataFormat matches the dataFormat field from the
	319	// UDataInfo header, four int8_t bytes. The value is {3,1,0,0}
	320	//
374ca955 A	321	const uint8_t inBytes =(const uint8_t )inData+headerSize;
374ca955 A	322	RBBIDataHeader rbbiDH = (RBBIDataHeader )inBytes;
73c04bcf	323	UBool formatVersionOne = ds->readUInt32((int32_t )rbbiDH->fFormatVersion) == 1;
374ca955	324	if (ds->readUInt32(rbbiDH->fMagic) != 0xb1a0 \|\|
73c04bcf	325	!(formatVersionOne \|\| rbbiDH->fFormatVersion[0] == 3) \|\|
374ca955 A	326	ds->readUInt32(rbbiDH->fLength) < sizeof(RBBIDataHeader))
	327	{
	328	udata_printError(ds, "ubrk_swap(): RBBI Data header is invalid.\n");
	329	*status=U_UNSUPPORTED_ERROR;
	330	return 0;
	331	}
	332
	333	//
	334	// Prefight operation? Just return the size
	335	//
73c04bcf A	336	int32_t breakDataLength = ds->readUInt32(rbbiDH->fLength);
73c04bcf A	337	int32_t totalSize = headerSize + breakDataLength;
374ca955 A	338	if (length < 0) {
	339	return totalSize;
	340	}
	341
	342	//
	343	// Check that length passed in is consistent with length from RBBI data header.
	344	//
73c04bcf A	345	if (length < totalSize) {
	346	udata_printError(ds, "ubrk_swap(): too few bytes (%d after ICU Data header) for break data.\n",
	347	breakDataLength);
	348	*status=U_INDEX_OUTOFBOUNDS_ERROR;
	349	return 0;
374ca955	350	}
374ca955 A	351
	352
	353	//
	354	// Swap the Data. Do the data itself first, then the RBBI Data Header, because
	355	// we need to reference the header to locate the data, and an
	356	// inplace swap of the header leaves it unusable.
	357	//
73c04bcf A	358	uint8_t outBytes = (uint8_t )outData + headerSize;
	359	RBBIDataHeader outputDH = (RBBIDataHeader )outBytes;
	360
374ca955 A	361	int32_t tableStartOffset;
	362	int32_t tableLength;
	363
	364	//
	365	// If not swapping in place, zero out the output buffer before starting.
	366	// Individual tables and other data items within are aligned to 8 byte boundaries
	367	// when originally created. Any unused space between items needs to be zero.
	368	//
	369	if (inBytes != outBytes) {
73c04bcf	370	uprv_memset(outBytes, 0, breakDataLength);
374ca955 A	371	}
	372
	373	//
	374	// Each state table begins with several 32 bit fields. Calculate the size
	375	// in bytes of these.
	376	//
46f4442e	377	int32_t topSize = offsetof(RBBIStateTable, fTableData);
374ca955 A	378
	379	// Forward state table.
	380	tableStartOffset = ds->readUInt32(rbbiDH->fFTable);
	381	tableLength = ds->readUInt32(rbbiDH->fFTableLen);
	382
	383	if (tableLength > 0) {
	384	ds->swapArray32(ds, inBytes+tableStartOffset, topSize,
	385	outBytes+tableStartOffset, status);
	386	ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize,
	387	outBytes+tableStartOffset+topSize, status);
	388	}
	389
	390	// Reverse state table. Same layout as forward table, above.
	391	tableStartOffset = ds->readUInt32(rbbiDH->fRTable);
	392	tableLength = ds->readUInt32(rbbiDH->fRTableLen);
	393
	394	if (tableLength > 0) {
	395	ds->swapArray32(ds, inBytes+tableStartOffset, topSize,
	396	outBytes+tableStartOffset, status);
	397	ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize,
	398	outBytes+tableStartOffset+topSize, status);
	399	}
	400
	401	// Safe Forward state table. Same layout as forward table, above.
	402	tableStartOffset = ds->readUInt32(rbbiDH->fSFTable);
	403	tableLength = ds->readUInt32(rbbiDH->fSFTableLen);
	404
	405	if (tableLength > 0) {
	406	ds->swapArray32(ds, inBytes+tableStartOffset, topSize,
	407	outBytes+tableStartOffset, status);
	408	ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize,
	409	outBytes+tableStartOffset+topSize, status);
	410	}
	411
	412	// Safe Reverse state table. Same layout as forward table, above.
	413	tableStartOffset = ds->readUInt32(rbbiDH->fSRTable);
	414	tableLength = ds->readUInt32(rbbiDH->fSRTableLen);
	415
	416	if (tableLength > 0) {
	417	ds->swapArray32(ds, inBytes+tableStartOffset, topSize,
	418	outBytes+tableStartOffset, status);
	419	ds->swapArray16(ds, inBytes+tableStartOffset+topSize, tableLength-topSize,
	420	outBytes+tableStartOffset+topSize, status);
	421	}
	422
	423	// Trie table for character categories
	424	utrie_swap(ds, inBytes+ds->readUInt32(rbbiDH->fTrie), ds->readUInt32(rbbiDH->fTrieLen),
	425	outBytes+ds->readUInt32(rbbiDH->fTrie), status);
	426
	427	// Source Rules Text. It's UChar data
	428	ds->swapArray16(ds, inBytes+ds->readUInt32(rbbiDH->fRuleSource), ds->readUInt32(rbbiDH->fRuleSourceLen),
	429	outBytes+ds->readUInt32(rbbiDH->fRuleSource), status);
	430
	431	// Table of rule status values. It's all int_32 values
	432	ds->swapArray32(ds, inBytes+ds->readUInt32(rbbiDH->fStatusTable), ds->readUInt32(rbbiDH->fStatusTableLen),
	433	outBytes+ds->readUInt32(rbbiDH->fStatusTable), status);
	434
73c04bcf A	435	// And, last, the header.
	436	// For the old version one format, the entire header consists of int32_t values.
	437	// For the newer formats, the fDataFormat field is an array of four bytes.
	438	// Swap the whole thing as int32_t, then, for the newer format, re-swap the one field.
	439	//
	440	ds->swapArray32(ds, inBytes, sizeof(RBBIDataHeader), outBytes, status);
	441	if (formatVersionOne == FALSE) {
	442	ds->swapArray32(ds, outputDH->fFormatVersion, 4, outputDH->fFormatVersion, status);
	443	}
	444
374ca955 A	445
	446	return totalSize;
	447	}
	448
	449
b75a7d8f	450	#endif /* #if !UCONFIG_NO_BREAK_ITERATION */