[apple/icu.git] / icuSources / i18n / collationdata.cpp

/*
*******************************************************************************
* Copyright (C) 2012-2015, International Business Machines
* Corporation and others.  All Rights Reserved.
*******************************************************************************
* collationdata.cpp
*
* created on: 2012jul28
* created by: Markus W. Scherer
*/

#include "unicode/utypes.h"

#if !UCONFIG_NO_COLLATION

#include "unicode/ucol.h"
#include "unicode/udata.h"
#include "unicode/uscript.h"
#include "cmemory.h"
#include "collation.h"
#include "collationdata.h"
#include "uassert.h"
#include "utrie2.h"
#include "uvectr32.h"

U_NAMESPACE_BEGIN

uint32_t
CollationData::getIndirectCE32(uint32_t ce32) const {
    U_ASSERT(Collation::isSpecialCE32(ce32));
    int32_t tag = Collation::tagFromCE32(ce32);
    if(tag == Collation::DIGIT_TAG) {
        // Fetch the non-numeric-collation CE32.
        ce32 = ce32s[Collation::indexFromCE32(ce32)];
    } else if(tag == Collation::LEAD_SURROGATE_TAG) {
        ce32 = Collation::UNASSIGNED_CE32;
    } else if(tag == Collation::U0000_TAG) {
        // Fetch the normal ce32 for U+0000.
        ce32 = ce32s[0];
    }
    return ce32;
}

uint32_t
CollationData::getFinalCE32(uint32_t ce32) const {
    if(Collation::isSpecialCE32(ce32)) {
        ce32 = getIndirectCE32(ce32);
    }
    return ce32;
}

int64_t
CollationData::getSingleCE(UChar32 c, UErrorCode &errorCode) const {
    if(U_FAILURE(errorCode)) { return 0; }
    // Keep parallel with CollationDataBuilder::getSingleCE().
    const CollationData *d;
    uint32_t ce32 = getCE32(c);
    if(ce32 == Collation::FALLBACK_CE32) {
        d = base;
        ce32 = base->getCE32(c);
    } else {
        d = this;
    }
    while(Collation::isSpecialCE32(ce32)) {
        switch(Collation::tagFromCE32(ce32)) {
        case Collation::LATIN_EXPANSION_TAG:
        case Collation::BUILDER_DATA_TAG:
        case Collation::PREFIX_TAG:
        case Collation::CONTRACTION_TAG:
        case Collation::HANGUL_TAG:
        case Collation::LEAD_SURROGATE_TAG:
            errorCode = U_UNSUPPORTED_ERROR;
            return 0;
        case Collation::FALLBACK_TAG:
        case Collation::RESERVED_TAG_3:
            errorCode = U_INTERNAL_PROGRAM_ERROR;
            return 0;
        case Collation::LONG_PRIMARY_TAG:
            return Collation::ceFromLongPrimaryCE32(ce32);
        case Collation::LONG_SECONDARY_TAG:
            return Collation::ceFromLongSecondaryCE32(ce32);
        case Collation::EXPANSION32_TAG:
            if(Collation::lengthFromCE32(ce32) == 1) {
                ce32 = d->ce32s[Collation::indexFromCE32(ce32)];
                break;
            } else {
                errorCode = U_UNSUPPORTED_ERROR;
                return 0;
            }
        case Collation::EXPANSION_TAG: {
            if(Collation::lengthFromCE32(ce32) == 1) {
                return d->ces[Collation::indexFromCE32(ce32)];
            } else {
                errorCode = U_UNSUPPORTED_ERROR;
                return 0;
            }
        }
        case Collation::DIGIT_TAG:
            // Fetch the non-numeric-collation CE32 and continue.
            ce32 = d->ce32s[Collation::indexFromCE32(ce32)];
            break;
        case Collation::U0000_TAG:
            U_ASSERT(c == 0);
            // Fetch the normal ce32 for U+0000 and continue.
            ce32 = d->ce32s[0];
            break;
        case Collation::OFFSET_TAG:
            return d->getCEFromOffsetCE32(c, ce32);
        case Collation::IMPLICIT_TAG:
            return Collation::unassignedCEFromCodePoint(c);
        }
    }
    return Collation::ceFromSimpleCE32(ce32);
}

uint32_t
CollationData::getFirstPrimaryForGroup(int32_t script) const {
    int32_t index = getScriptIndex(script);
    return index == 0 ? 0 : (uint32_t)scriptStarts[index] << 16;
}

uint32_t
CollationData::getLastPrimaryForGroup(int32_t script) const {
    int32_t index = getScriptIndex(script);
    if(index == 0) {
        return 0;
    }
    uint32_t limit = scriptStarts[index + 1];
    return (limit << 16) - 1;
}

int32_t
CollationData::getGroupForPrimary(uint32_t p) const {
    p >>= 16;
    if(p < scriptStarts[1] || scriptStarts[scriptStartsLength - 1] <= p) {
        return -1;
    }
    int32_t index = 1;
    while(p >= scriptStarts[index + 1]) { ++index; }
    for(int32_t i = 0; i < numScripts; ++i) {
        if(scriptsIndex[i] == index) {
            return i;
        }
    }
    for(int32_t i = 0; i < MAX_NUM_SPECIAL_REORDER_CODES; ++i) {
        if(scriptsIndex[numScripts + i] == index) {
            return UCOL_REORDER_CODE_FIRST + i;
        }
    }
    return -1;
}

int32_t
CollationData::getScriptIndex(int32_t script) const {
    if(script < 0) {
        return 0;
    } else if(script < numScripts) {
        return scriptsIndex[script];
    } else if(script < UCOL_REORDER_CODE_FIRST) {
        return 0;
    } else {
        script -= UCOL_REORDER_CODE_FIRST;
        if(script < MAX_NUM_SPECIAL_REORDER_CODES) {
            return scriptsIndex[numScripts + script];
        } else {
            return 0;
        }
    }
}

int32_t
CollationData::getEquivalentScripts(int32_t script,
                                    int32_t dest[], int32_t capacity,
                                    UErrorCode &errorCode) const {
    if(U_FAILURE(errorCode)) { return 0; }
    int32_t index = getScriptIndex(script);
    if(index == 0) { return 0; }
    if(script >= UCOL_REORDER_CODE_FIRST) {
        // Special groups have no aliases.
        if(capacity > 0) {
            dest[0] = script;
        } else {
            errorCode = U_BUFFER_OVERFLOW_ERROR;
        }
        return 1;
    }

    int32_t length = 0;
    for(int32_t i = 0; i < numScripts; ++i) {
        if(scriptsIndex[i] == index) {
            if(length < capacity) {
                dest[length] = i;
            }
            ++length;
        }
    }
    if(length > capacity) {
        errorCode = U_BUFFER_OVERFLOW_ERROR;
    }
    return length;
}

void
CollationData::makeReorderRanges(const int32_t *reorder, int32_t length,
                                 UVector32 &ranges, UErrorCode &errorCode) const {
    makeReorderRanges(reorder, length, FALSE, ranges, errorCode);
}

void
CollationData::makeReorderRanges(const int32_t *reorder, int32_t length,
                                 UBool latinMustMove,
                                 UVector32 &ranges, UErrorCode &errorCode) const {
    if(U_FAILURE(errorCode)) { return; }
    ranges.removeAllElements();
    if(length == 0 || (length == 1 && reorder[0] == USCRIPT_UNKNOWN)) {
        return;
    }

    // Maps each script-or-group range to a new lead byte.
    uint8_t table[MAX_NUM_SCRIPT_RANGES];
    uprv_memset(table, 0, sizeof(table));

    {
        // Set "don't care" values for reserved ranges.
        int32_t index = scriptsIndex[
                numScripts + REORDER_RESERVED_BEFORE_LATIN - UCOL_REORDER_CODE_FIRST];
        if(index != 0) {
            table[index] = 0xff;
        }
        index = scriptsIndex[
                numScripts + REORDER_RESERVED_AFTER_LATIN - UCOL_REORDER_CODE_FIRST];
        if(index != 0) {
            table[index] = 0xff;
        }
    }

    // Never reorder special low and high primary lead bytes.
    U_ASSERT(scriptStartsLength >= 2);
    U_ASSERT(scriptStarts[0] == 0);
    int32_t lowStart = scriptStarts[1];
    U_ASSERT(lowStart == ((Collation::MERGE_SEPARATOR_BYTE + 1) << 8));
    int32_t highLimit = scriptStarts[scriptStartsLength - 1];
    U_ASSERT(highLimit == (Collation::TRAIL_WEIGHT_BYTE << 8));

    // Get the set of special reorder codes in the input list.
    // This supports a fixed number of special reorder codes;
    // it works for data with codes beyond UCOL_REORDER_CODE_LIMIT.
    uint32_t specials = 0;
    for(int32_t i = 0; i < length; ++i) {
        int32_t reorderCode = reorder[i] - UCOL_REORDER_CODE_FIRST;
        if(0 <= reorderCode && reorderCode < MAX_NUM_SPECIAL_REORDER_CODES) {
            specials |= (uint32_t)1 << reorderCode;
        }
    }

    // Start the reordering with the special low reorder codes that do not occur in the input.
    for(int32_t i = 0; i < MAX_NUM_SPECIAL_REORDER_CODES; ++i) {
        int32_t index = scriptsIndex[numScripts + i];
        if(index != 0 && (specials & ((uint32_t)1 << i)) == 0) {
            lowStart = addLowScriptRange(table, index, lowStart);
        }
    }

    // Skip the reserved range before Latin if Latin is the first script,
    // so that we do not move it unnecessarily.
    int32_t skippedReserved = 0;
    if(specials == 0 && reorder[0] == USCRIPT_LATIN && !latinMustMove) {
        int32_t index = scriptsIndex[USCRIPT_LATIN];
        U_ASSERT(index != 0);
        int32_t start = scriptStarts[index];
        U_ASSERT(lowStart <= start);
        skippedReserved = start - lowStart;
        lowStart = start;
    }

    // Reorder according to the input scripts, continuing from the bottom of the primary range.
    int32_t originalLength = length;  // length will be decremented if "others" is in the list.
    UBool hasReorderToEnd = FALSE;
    for(int32_t i = 0; i < length;) {
        int32_t script = reorder[i++];
        if(script == USCRIPT_UNKNOWN) {
            // Put the remaining scripts at the top.
            hasReorderToEnd = TRUE;
            while(i < length) {
                script = reorder[--length];
                if(script == USCRIPT_UNKNOWN ||  // Must occur at most once.
                        script == UCOL_REORDER_CODE_DEFAULT) {
                    errorCode = U_ILLEGAL_ARGUMENT_ERROR;
                    return;
                }
                int32_t index = getScriptIndex(script);
                if(index == 0) { continue; }
                if(table[index] != 0) {  // Duplicate or equivalent script.
                    errorCode = U_ILLEGAL_ARGUMENT_ERROR;
                    return;
                }
                highLimit = addHighScriptRange(table, index, highLimit);
            }
            break;
        }
        if(script == UCOL_REORDER_CODE_DEFAULT) {
            // The default code must be the only one in the list, and that is handled by the caller.
            // Otherwise it must not be used.
            errorCode = U_ILLEGAL_ARGUMENT_ERROR;
            return;
        }
        int32_t index = getScriptIndex(script);
        if(index == 0) { continue; }
        if(table[index] != 0) {  // Duplicate or equivalent script.
            errorCode = U_ILLEGAL_ARGUMENT_ERROR;
            return;
        }
        lowStart = addLowScriptRange(table, index, lowStart);
    }

    // Put all remaining scripts into the middle.
    for(int32_t i = 1; i < scriptStartsLength - 1; ++i) {
        int32_t leadByte = table[i];
        if(leadByte != 0) { continue; }
        int32_t start = scriptStarts[i];
        if(!hasReorderToEnd && start > lowStart) {
            // No need to move this script.
            lowStart = start;
        }
        lowStart = addLowScriptRange(table, i, lowStart);
    }
    if(lowStart > highLimit) {
        if((lowStart - (skippedReserved & 0xff00)) <= highLimit) {
            // Try not skipping the before-Latin reserved range.
            makeReorderRanges(reorder, originalLength, TRUE, ranges, errorCode);
            return;
        }
        // We need more primary lead bytes than available, despite the reserved ranges.
        errorCode = U_BUFFER_OVERFLOW_ERROR;
        return;
    }

    // Turn lead bytes into a list of (limit, offset) pairs.
    // Encode each pair in one list element:
    // Upper 16 bits = limit, lower 16 = signed lead byte offset.
    int32_t offset = 0;
    for(int32_t i = 1;; ++i) {
        int32_t nextOffset = offset;
        while(i < scriptStartsLength - 1) {
            int32_t newLeadByte = table[i];
            if(newLeadByte == 0xff) {
                // "Don't care" lead byte for reserved range, continue with current offset.
            } else {
                nextOffset = newLeadByte - (scriptStarts[i] >> 8);
                if(nextOffset != offset) { break; }
            }
            ++i;
        }
        if(offset != 0 || i < scriptStartsLength - 1) {
            ranges.addElement(((int32_t)scriptStarts[i] << 16) | (offset & 0xffff), errorCode);
        }
        if(i == scriptStartsLength - 1) { break; }
        offset = nextOffset;
    }
}

int32_t
CollationData::addLowScriptRange(uint8_t table[], int32_t index, int32_t lowStart) const {
    int32_t start = scriptStarts[index];
    if((start & 0xff) < (lowStart & 0xff)) {
        lowStart += 0x100;
    }
    table[index] = (uint8_t)(lowStart >> 8);
    int32_t limit = scriptStarts[index + 1];
    lowStart = ((lowStart & 0xff00) + ((limit & 0xff00) - (start & 0xff00))) | (limit & 0xff);
    return lowStart;
}

int32_t
CollationData::addHighScriptRange(uint8_t table[], int32_t index, int32_t highLimit) const {
    int32_t limit = scriptStarts[index + 1];
    if((limit & 0xff) > (highLimit & 0xff)) {
        highLimit -= 0x100;
    }
    int32_t start = scriptStarts[index];
    highLimit = ((highLimit & 0xff00) - ((limit & 0xff00) - (start & 0xff00))) | (start & 0xff);
    table[index] = (uint8_t)(highLimit >> 8);
    return highLimit;
}

U_NAMESPACE_END

#endif  // !UCONFIG_NO_COLLATION
Commit	Line	Data
57a6839d A	1	/*
57a6839d A	2	*******************************************************************************
b331163b	3	* Copyright (C) 2012-2015, International Business Machines
57a6839d A	4	* Corporation and others. All Rights Reserved.
	5	*******************************************************************************
	6	* collationdata.cpp
	7	*
	8	* created on: 2012jul28
	9	* created by: Markus W. Scherer
	10	*/
	11
	12	#include "unicode/utypes.h"
	13
	14	#if !UCONFIG_NO_COLLATION
	15
	16	#include "unicode/ucol.h"
	17	#include "unicode/udata.h"
	18	#include "unicode/uscript.h"
	19	#include "cmemory.h"
	20	#include "collation.h"
	21	#include "collationdata.h"
	22	#include "uassert.h"
	23	#include "utrie2.h"
b331163b	24	#include "uvectr32.h"
57a6839d A	25
	26	U_NAMESPACE_BEGIN
	27
	28	uint32_t
	29	CollationData::getIndirectCE32(uint32_t ce32) const {
	30	U_ASSERT(Collation::isSpecialCE32(ce32));
	31	int32_t tag = Collation::tagFromCE32(ce32);
	32	if(tag == Collation::DIGIT_TAG) {
	33	// Fetch the non-numeric-collation CE32.
	34	ce32 = ce32s[Collation::indexFromCE32(ce32)];
	35	} else if(tag == Collation::LEAD_SURROGATE_TAG) {
	36	ce32 = Collation::UNASSIGNED_CE32;
	37	} else if(tag == Collation::U0000_TAG) {
	38	// Fetch the normal ce32 for U+0000.
	39	ce32 = ce32s[0];
	40	}
	41	return ce32;
	42	}
	43
	44	uint32_t
	45	CollationData::getFinalCE32(uint32_t ce32) const {
	46	if(Collation::isSpecialCE32(ce32)) {
	47	ce32 = getIndirectCE32(ce32);
	48	}
	49	return ce32;
	50	}
	51
b331163b A	52	int64_t
	53	CollationData::getSingleCE(UChar32 c, UErrorCode &errorCode) const {
	54	if(U_FAILURE(errorCode)) { return 0; }
	55	// Keep parallel with CollationDataBuilder::getSingleCE().
	56	const CollationData *d;
	57	uint32_t ce32 = getCE32(c);
	58	if(ce32 == Collation::FALLBACK_CE32) {
	59	d = base;
	60	ce32 = base->getCE32(c);
	61	} else {
	62	d = this;
	63	}
	64	while(Collation::isSpecialCE32(ce32)) {
	65	switch(Collation::tagFromCE32(ce32)) {
	66	case Collation::LATIN_EXPANSION_TAG:
	67	case Collation::BUILDER_DATA_TAG:
	68	case Collation::PREFIX_TAG:
	69	case Collation::CONTRACTION_TAG:
	70	case Collation::HANGUL_TAG:
	71	case Collation::LEAD_SURROGATE_TAG:
	72	errorCode = U_UNSUPPORTED_ERROR;
	73	return 0;
	74	case Collation::FALLBACK_TAG:
	75	case Collation::RESERVED_TAG_3:
	76	errorCode = U_INTERNAL_PROGRAM_ERROR;
	77	return 0;
	78	case Collation::LONG_PRIMARY_TAG:
	79	return Collation::ceFromLongPrimaryCE32(ce32);
	80	case Collation::LONG_SECONDARY_TAG:
	81	return Collation::ceFromLongSecondaryCE32(ce32);
	82	case Collation::EXPANSION32_TAG:
	83	if(Collation::lengthFromCE32(ce32) == 1) {
	84	ce32 = d->ce32s[Collation::indexFromCE32(ce32)];
	85	break;
	86	} else {
	87	errorCode = U_UNSUPPORTED_ERROR;
	88	return 0;
	89	}
	90	case Collation::EXPANSION_TAG: {
	91	if(Collation::lengthFromCE32(ce32) == 1) {
	92	return d->ces[Collation::indexFromCE32(ce32)];
	93	} else {
	94	errorCode = U_UNSUPPORTED_ERROR;
	95	return 0;
	96	}
	97	}
	98	case Collation::DIGIT_TAG:
	99	// Fetch the non-numeric-collation CE32 and continue.
	100	ce32 = d->ce32s[Collation::indexFromCE32(ce32)];
	101	break;
	102	case Collation::U0000_TAG:
	103	U_ASSERT(c == 0);
	104	// Fetch the normal ce32 for U+0000 and continue.
	105	ce32 = d->ce32s[0];
	106	break;
	107	case Collation::OFFSET_TAG:
	108	return d->getCEFromOffsetCE32(c, ce32);
	109	case Collation::IMPLICIT_TAG:
	110	return Collation::unassignedCEFromCodePoint(c);
	111	}
	112	}
	113	return Collation::ceFromSimpleCE32(ce32);
	114	}
	115
57a6839d A	116	uint32_t
57a6839d A	117	CollationData::getFirstPrimaryForGroup(int32_t script) const {
b331163b A	118	int32_t index = getScriptIndex(script);
b331163b A	119	return index == 0 ? 0 : (uint32_t)scriptStarts[index] << 16;
57a6839d A	120	}
	121
	122	uint32_t
	123	CollationData::getLastPrimaryForGroup(int32_t script) const {
b331163b A	124	int32_t index = getScriptIndex(script);
b331163b A	125	if(index == 0) {
57a6839d A	126	return 0;
57a6839d A	127	}
b331163b A	128	uint32_t limit = scriptStarts[index + 1];
b331163b A	129	return (limit << 16) - 1;
57a6839d A	130	}
	131
	132	int32_t
	133	CollationData::getGroupForPrimary(uint32_t p) const {
b331163b A	134	p >>= 16;
	135	if(p < scriptStarts[1] \|\| scriptStarts[scriptStartsLength - 1] <= p) {
	136	return -1;
	137	}
	138	int32_t index = 1;
	139	while(p >= scriptStarts[index + 1]) { ++index; }
	140	for(int32_t i = 0; i < numScripts; ++i) {
	141	if(scriptsIndex[i] == index) {
	142	return i;
	143	}
	144	}
	145	for(int32_t i = 0; i < MAX_NUM_SPECIAL_REORDER_CODES; ++i) {
	146	if(scriptsIndex[numScripts + i] == index) {
	147	return UCOL_REORDER_CODE_FIRST + i;
57a6839d A	148	}
	149	}
	150	return -1;
	151	}
	152
	153	int32_t
b331163b A	154	CollationData::getScriptIndex(int32_t script) const {
	155	if(script < 0) {
	156	return 0;
	157	} else if(script < numScripts) {
	158	return scriptsIndex[script];
	159	} else if(script < UCOL_REORDER_CODE_FIRST) {
	160	return 0;
	161	} else {
	162	script -= UCOL_REORDER_CODE_FIRST;
	163	if(script < MAX_NUM_SPECIAL_REORDER_CODES) {
	164	return scriptsIndex[numScripts + script];
	165	} else {
	166	return 0;
57a6839d	167	}
57a6839d	168	}
57a6839d A	169	}
	170
	171	int32_t
	172	CollationData::getEquivalentScripts(int32_t script,
	173	int32_t dest[], int32_t capacity,
	174	UErrorCode &errorCode) const {
	175	if(U_FAILURE(errorCode)) { return 0; }
b331163b A	176	int32_t index = getScriptIndex(script);
	177	if(index == 0) { return 0; }
	178	if(script >= UCOL_REORDER_CODE_FIRST) {
	179	// Special groups have no aliases.
	180	if(capacity > 0) {
	181	dest[0] = script;
	182	} else {
	183	errorCode = U_BUFFER_OVERFLOW_ERROR;
	184	}
	185	return 1;
	186	}
	187
	188	int32_t length = 0;
	189	for(int32_t i = 0; i < numScripts; ++i) {
	190	if(scriptsIndex[i] == index) {
	191	if(length < capacity) {
	192	dest[length] = i;
57a6839d	193	}
b331163b	194	++length;
57a6839d A	195	}
57a6839d A	196	}
b331163b A	197	if(length > capacity) {
	198	errorCode = U_BUFFER_OVERFLOW_ERROR;
	199	}
57a6839d A	200	return length;
	201	}
	202
	203	void
b331163b A	204	CollationData::makeReorderRanges(const int32_t *reorder, int32_t length,
	205	UVector32 &ranges, UErrorCode &errorCode) const {
	206	makeReorderRanges(reorder, length, FALSE, ranges, errorCode);
	207	}
57a6839d	208
b331163b A	209	void
	210	CollationData::makeReorderRanges(const int32_t *reorder, int32_t length,
	211	UBool latinMustMove,
	212	UVector32 &ranges, UErrorCode &errorCode) const {
	213	if(U_FAILURE(errorCode)) { return; }
	214	ranges.removeAllElements();
	215	if(length == 0 \|\| (length == 1 && reorder[0] == USCRIPT_UNKNOWN)) {
	216	return;
57a6839d	217	}
57a6839d	218
b331163b A	219	// Maps each script-or-group range to a new lead byte.
	220	uint8_t table[MAX_NUM_SCRIPT_RANGES];
	221	uprv_memset(table, 0, sizeof(table));
57a6839d	222
b331163b A	223	{
	224	// Set "don't care" values for reserved ranges.
	225	int32_t index = scriptsIndex[
	226	numScripts + REORDER_RESERVED_BEFORE_LATIN - UCOL_REORDER_CODE_FIRST];
	227	if(index != 0) {
	228	table[index] = 0xff;
	229	}
	230	index = scriptsIndex[
	231	numScripts + REORDER_RESERVED_AFTER_LATIN - UCOL_REORDER_CODE_FIRST];
	232	if(index != 0) {
	233	table[index] = 0xff;
	234	}
57a6839d A	235	}
57a6839d A	236
b331163b A	237	// Never reorder special low and high primary lead bytes.
	238	U_ASSERT(scriptStartsLength >= 2);
	239	U_ASSERT(scriptStarts[0] == 0);
	240	int32_t lowStart = scriptStarts[1];
	241	U_ASSERT(lowStart == ((Collation::MERGE_SEPARATOR_BYTE + 1) << 8));
	242	int32_t highLimit = scriptStarts[scriptStartsLength - 1];
	243	U_ASSERT(highLimit == (Collation::TRAIL_WEIGHT_BYTE << 8));
	244
57a6839d	245	// Get the set of special reorder codes in the input list.
b331163b	246	// This supports a fixed number of special reorder codes;
57a6839d A	247	// it works for data with codes beyond UCOL_REORDER_CODE_LIMIT.
	248	uint32_t specials = 0;
	249	for(int32_t i = 0; i < length; ++i) {
	250	int32_t reorderCode = reorder[i] - UCOL_REORDER_CODE_FIRST;
b331163b	251	if(0 <= reorderCode && reorderCode < MAX_NUM_SPECIAL_REORDER_CODES) {
57a6839d A	252	specials \|= (uint32_t)1 << reorderCode;
	253	}
	254	}
	255
	256	// Start the reordering with the special low reorder codes that do not occur in the input.
b331163b A	257	for(int32_t i = 0; i < MAX_NUM_SPECIAL_REORDER_CODES; ++i) {
	258	int32_t index = scriptsIndex[numScripts + i];
	259	if(index != 0 && (specials & ((uint32_t)1 << i)) == 0) {
	260	lowStart = addLowScriptRange(table, index, lowStart);
57a6839d A	261	}
	262	}
	263
b331163b A	264	// Skip the reserved range before Latin if Latin is the first script,
	265	// so that we do not move it unnecessarily.
	266	int32_t skippedReserved = 0;
	267	if(specials == 0 && reorder[0] == USCRIPT_LATIN && !latinMustMove) {
	268	int32_t index = scriptsIndex[USCRIPT_LATIN];
	269	U_ASSERT(index != 0);
	270	int32_t start = scriptStarts[index];
	271	U_ASSERT(lowStart <= start);
	272	skippedReserved = start - lowStart;
	273	lowStart = start;
	274	}
	275
	276	// Reorder according to the input scripts, continuing from the bottom of the primary range.
	277	int32_t originalLength = length; // length will be decremented if "others" is in the list.
	278	UBool hasReorderToEnd = FALSE;
57a6839d A	279	for(int32_t i = 0; i < length;) {
	280	int32_t script = reorder[i++];
	281	if(script == USCRIPT_UNKNOWN) {
	282	// Put the remaining scripts at the top.
b331163b	283	hasReorderToEnd = TRUE;
57a6839d A	284	while(i < length) {
	285	script = reorder[--length];
	286	if(script == USCRIPT_UNKNOWN \|\| // Must occur at most once.
	287	script == UCOL_REORDER_CODE_DEFAULT) {
	288	errorCode = U_ILLEGAL_ARGUMENT_ERROR;
	289	return;
	290	}
b331163b A	291	int32_t index = getScriptIndex(script);
	292	if(index == 0) { continue; }
	293	if(table[index] != 0) { // Duplicate or equivalent script.
57a6839d A	294	errorCode = U_ILLEGAL_ARGUMENT_ERROR;
	295	return;
	296	}
b331163b	297	highLimit = addHighScriptRange(table, index, highLimit);
57a6839d A	298	}
	299	break;
	300	}
	301	if(script == UCOL_REORDER_CODE_DEFAULT) {
	302	// The default code must be the only one in the list, and that is handled by the caller.
	303	// Otherwise it must not be used.
	304	errorCode = U_ILLEGAL_ARGUMENT_ERROR;
	305	return;
	306	}
b331163b A	307	int32_t index = getScriptIndex(script);
	308	if(index == 0) { continue; }
	309	if(table[index] != 0) { // Duplicate or equivalent script.
57a6839d A	310	errorCode = U_ILLEGAL_ARGUMENT_ERROR;
	311	return;
	312	}
b331163b	313	lowStart = addLowScriptRange(table, index, lowStart);
57a6839d A	314	}
	315
	316	// Put all remaining scripts into the middle.
b331163b A	317	for(int32_t i = 1; i < scriptStartsLength - 1; ++i) {
	318	int32_t leadByte = table[i];
	319	if(leadByte != 0) { continue; }
	320	int32_t start = scriptStarts[i];
	321	if(!hasReorderToEnd && start > lowStart) {
	322	// No need to move this script.
	323	lowStart = start;
	324	}
	325	lowStart = addLowScriptRange(table, i, lowStart);
	326	}
	327	if(lowStart > highLimit) {
	328	if((lowStart - (skippedReserved & 0xff00)) <= highLimit) {
	329	// Try not skipping the before-Latin reserved range.
	330	makeReorderRanges(reorder, originalLength, TRUE, ranges, errorCode);
	331	return;
	332	}
	333	// We need more primary lead bytes than available, despite the reserved ranges.
	334	errorCode = U_BUFFER_OVERFLOW_ERROR;
	335	return;
	336	}
	337
	338	// Turn lead bytes into a list of (limit, offset) pairs.
	339	// Encode each pair in one list element:
	340	// Upper 16 bits = limit, lower 16 = signed lead byte offset.
	341	int32_t offset = 0;
	342	for(int32_t i = 1;; ++i) {
	343	int32_t nextOffset = offset;
	344	while(i < scriptStartsLength - 1) {
	345	int32_t newLeadByte = table[i];
	346	if(newLeadByte == 0xff) {
	347	// "Don't care" lead byte for reserved range, continue with current offset.
	348	} else {
	349	nextOffset = newLeadByte - (scriptStarts[i] >> 8);
	350	if(nextOffset != offset) { break; }
	351	}
	352	++i;
	353	}
	354	if(offset != 0 \|\| i < scriptStartsLength - 1) {
	355	ranges.addElement(((int32_t)scriptStarts[i] << 16) \| (offset & 0xffff), errorCode);
	356	}
	357	if(i == scriptStartsLength - 1) { break; }
	358	offset = nextOffset;
	359	}
	360	}
	361
	362	int32_t
	363	CollationData::addLowScriptRange(uint8_t table[], int32_t index, int32_t lowStart) const {
	364	int32_t start = scriptStarts[index];
	365	if((start & 0xff) < (lowStart & 0xff)) {
	366	lowStart += 0x100;
	367	}
	368	table[index] = (uint8_t)(lowStart >> 8);
	369	int32_t limit = scriptStarts[index + 1];
	370	lowStart = ((lowStart & 0xff00) + ((limit & 0xff00) - (start & 0xff00))) \| (limit & 0xff);
	371	return lowStart;
	372	}
	373
	374	int32_t
	375	CollationData::addHighScriptRange(uint8_t table[], int32_t index, int32_t highLimit) const {
	376	int32_t limit = scriptStarts[index + 1];
	377	if((limit & 0xff) > (highLimit & 0xff)) {
	378	highLimit -= 0x100;
57a6839d	379	}
b331163b A	380	int32_t start = scriptStarts[index];
	381	highLimit = ((highLimit & 0xff00) - ((limit & 0xff00) - (start & 0xff00))) \| (start & 0xff);
	382	table[index] = (uint8_t)(highLimit >> 8);
	383	return highLimit;
57a6839d A	384	}
	385
	386	U_NAMESPACE_END
	387
	388	#endif // !UCONFIG_NO_COLLATION