/src/icu/icu4c/source/i18n/ucol_sit.cpp

Source (jump to first uncovered line)
// © 2016 and later: Unicode, Inc. and others.
// License & terms of use: http://www.unicode.org/copyright.html
/*
*******************************************************************************
*   Copyright (C) 2004-2016, International Business Machines
*   Corporation and others.  All Rights Reserved.
*******************************************************************************
*   file name:  ucol_sit.cpp
*   encoding:   UTF-8
*   tab size:   8 (not used)
*   indentation:4
*
* Modification history
* Date        Name      Comments
* 03/12/2004  weiv      Creation
*/

#include "unicode/ustring.h"
#include "unicode/udata.h"
#include "unicode/utf16.h"
#include "utracimp.h"
#include "ucol_imp.h"
#include "ulocimp.h"
#include "cmemory.h"
#include "cstring.h"
#include "uresimp.h"
#include "unicode/coll.h"
#include "unicode/stringpiece.h"
#include "charstr.h"

U_NAMESPACE_USE

#ifdef UCOL_TRACE_SIT
# include <stdio.h>
#endif

#if !UCONFIG_NO_COLLATION

#include "unicode/tblcoll.h"

enum OptionsList {
    UCOL_SIT_LANGUAGE = 0,
    UCOL_SIT_SCRIPT   = 1,
    UCOL_SIT_REGION   = 2,
    UCOL_SIT_VARIANT  = 3,
    UCOL_SIT_KEYWORD  = 4,
    UCOL_SIT_PROVIDER = 5,
    UCOL_SIT_LOCELEMENT_MAX = UCOL_SIT_PROVIDER, /* the last element that's part of LocElements */

    UCOL_SIT_BCP47,
    UCOL_SIT_STRENGTH,
    UCOL_SIT_CASE_LEVEL,
    UCOL_SIT_CASE_FIRST,
    UCOL_SIT_NUMERIC_COLLATION,
    UCOL_SIT_ALTERNATE_HANDLING,
    UCOL_SIT_NORMALIZATION_MODE,
    UCOL_SIT_FRENCH_COLLATION,
    UCOL_SIT_HIRAGANA_QUATERNARY,
    UCOL_SIT_VARIABLE_TOP,
    UCOL_SIT_VARIABLE_TOP_VALUE,
    UCOL_SIT_ITEMS_COUNT
};

/* option starters chars. */
static const char alternateHArg     = 'A';
static const char variableTopValArg = 'B';
static const char caseFirstArg      = 'C';
static const char numericCollArg    = 'D';
static const char caseLevelArg      = 'E';
static const char frenchCollArg     = 'F';
static const char hiraganaQArg      = 'H';
static const char keywordArg        = 'K';
static const char languageArg       = 'L';
static const char normArg           = 'N';
static const char providerArg       = 'P';
static const char regionArg         = 'R';
static const char strengthArg       = 'S';
static const char variableTopArg    = 'T';
static const char variantArg        = 'V';
static const char RFC3066Arg        = 'X';
static const char scriptArg         = 'Z';

static const char collationKeyword[]  = "@collation=";
static const char providerKeyword[]  = "@sp=";


static const int32_t locElementCount = UCOL_SIT_LOCELEMENT_MAX+1;
static const int32_t locElementCapacity = 32;
static const int32_t loc3066Capacity = 256;

/* structure containing specification of a collator. Initialized
 * from a short string. Also used to construct a short string from a
 * collator instance
 */
struct CollatorSpec {
    inline CollatorSpec();

    CharString locElements[locElementCount];
    CharString locale;
    UColAttributeValue options[UCOL_ATTRIBUTE_COUNT];
    uint32_t variableTopValue;
    char16_t variableTopString[locElementCapacity];
    int32_t variableTopStringLen;
    UBool variableTopSet;
    CharString entries[UCOL_SIT_ITEMS_COUNT];
};

CollatorSpec::CollatorSpec() :
locale(),
variableTopValue(0),
variableTopString(),
variableTopSet(false)
 {
    // set collation options to default
    for(int32_t i = 0; i < UCOL_ATTRIBUTE_COUNT; i++) {
        options[i] = UCOL_DEFAULT;
    }
}


/* structure for converting between character attribute
 * representation and real collation attribute value.
 */
struct AttributeConversion {
    char letter;
    UColAttributeValue value;
};

static const AttributeConversion conversions[12] = {
    { '1', UCOL_PRIMARY },
    { '2', UCOL_SECONDARY },
    { '3', UCOL_TERTIARY },
    { '4', UCOL_QUATERNARY },
    { 'D', UCOL_DEFAULT },
    { 'I', UCOL_IDENTICAL },
    { 'L', UCOL_LOWER_FIRST },
    { 'N', UCOL_NON_IGNORABLE },
    { 'O', UCOL_ON },
    { 'S', UCOL_SHIFTED },
    { 'U', UCOL_UPPER_FIRST },
    { 'X', UCOL_OFF }
};


static UColAttributeValue
ucol_sit_letterToAttributeValue(char letter, UErrorCode *status) {
    uint32_t i = 0;
    for(i = 0; i < UPRV_LENGTHOF(conversions); i++) {
        if(conversions[i].letter == letter) {
            return conversions[i].value;
        }
    }
    *status = U_ILLEGAL_ARGUMENT_ERROR;
#ifdef UCOL_TRACE_SIT
    fprintf(stderr, "%s:%d: unknown letter %c: %s\n", __FILE__, __LINE__, letter, u_errorName(*status));
#endif    
    return UCOL_DEFAULT;
}

/* function prototype for functions used to parse a short string */
U_CDECL_BEGIN
typedef const char* U_CALLCONV
ActionFunction(CollatorSpec *spec, uint32_t value1, const char* string,
               UErrorCode *status);
U_CDECL_END

U_CDECL_BEGIN
static const char* U_CALLCONV
_processLocaleElement(CollatorSpec *spec, uint32_t value, const char* string,
                      UErrorCode *status)
{
    do {
        if(value == UCOL_SIT_LANGUAGE || value == UCOL_SIT_KEYWORD || value == UCOL_SIT_PROVIDER) {
            spec->locElements[value].append(uprv_tolower(*string), *status);
        } else {
            spec->locElements[value].append(*string, *status);
        }
    } while(*(++string) != '_' && *string && U_SUCCESS(*status));
    // don't skip the underscore at the end
    return string;
}
U_CDECL_END

U_CDECL_BEGIN
static const char* U_CALLCONV
_processRFC3066Locale(CollatorSpec *spec, uint32_t, const char* string,
                      UErrorCode *status)
{
    char terminator = *string;
    string++;
    const char *end = uprv_strchr(string+1, terminator);
    if(end == nullptr || end - string >= loc3066Capacity) {
        *status = U_BUFFER_OVERFLOW_ERROR;
        return string;
    } else {
        spec->locale.copyFrom(CharString(string, static_cast<int32_t>(end-string), *status), *status);
        return end+1;
    }
}

U_CDECL_END

U_CDECL_BEGIN
static const char* U_CALLCONV
_processCollatorOption(CollatorSpec *spec, uint32_t option, const char* string,
                       UErrorCode *status)
{
    spec->options[option] = ucol_sit_letterToAttributeValue(*string, status);
    if((*(++string) != '_' && *string) || U_FAILURE(*status)) {
#ifdef UCOL_TRACE_SIT
    fprintf(stderr, "%s:%d: unknown collator option at '%s': %s\n", __FILE__, __LINE__, string, u_errorName(*status));
#endif    
        *status = U_ILLEGAL_ARGUMENT_ERROR;
    }
    return string;
}
U_CDECL_END


static char16_t
readHexCodeUnit(const char **string, UErrorCode *status)
{
    char16_t result = 0;
    int32_t value = 0;
    char c;
    int32_t noDigits = 0;
    while((c = **string) != 0 && noDigits < 4) {
        if( c >= '0' && c <= '9') {
            value = c - '0';
        } else if ( c >= 'a' && c <= 'f') {
            value = c - 'a' + 10;
        } else if ( c >= 'A' && c <= 'F') {
            value = c - 'A' + 10;
        } else {
            *status = U_ILLEGAL_ARGUMENT_ERROR;
#ifdef UCOL_TRACE_SIT
            fprintf(stderr, "%s:%d: Bad hex char at '%s': %s\n", __FILE__, __LINE__, *string, u_errorName(*status));
#endif    
            return 0;
        }
        result = (result << 4) | static_cast<char16_t>(value);
        noDigits++;
        (*string)++;
    }
    // if the string was terminated before we read 4 digits, set an error
    if(noDigits < 4) {
        *status = U_ILLEGAL_ARGUMENT_ERROR;
#ifdef UCOL_TRACE_SIT
        fprintf(stderr, "%s:%d: Short (only %d digits, wanted 4) at '%s': %s\n", __FILE__, __LINE__, noDigits,*string, u_errorName(*status));
#endif    
    }
    return result;
}

U_CDECL_BEGIN
static const char* U_CALLCONV
_processVariableTop(CollatorSpec *spec, uint32_t value1, const char* string, UErrorCode *status)
{
    // get four digits
    int32_t i = 0;
    if(!value1) {
        while(U_SUCCESS(*status) && i < locElementCapacity && *string != 0 && *string != '_') {
            spec->variableTopString[i++] = readHexCodeUnit(&string, status);
        }
        spec->variableTopStringLen = i;
        if(i == locElementCapacity && *string != 0 && *string != '_') {
            *status = U_BUFFER_OVERFLOW_ERROR;
        }
    } else {
        spec->variableTopValue = readHexCodeUnit(&string, status);
    }
    if(U_SUCCESS(*status)) {
        spec->variableTopSet = true;
    }
    return string;
}
U_CDECL_END


/* Table for parsing short strings */
struct ShortStringOptions {
    char optionStart;
    ActionFunction *action;
    uint32_t attr;
};

static const ShortStringOptions options[UCOL_SIT_ITEMS_COUNT] =
{
/* 10 ALTERNATE_HANDLING */   {alternateHArg,     _processCollatorOption, UCOL_ALTERNATE_HANDLING }, // alternate  N, S, D
/* 15 VARIABLE_TOP_VALUE */   {variableTopValArg, _processVariableTop,    1 },
/* 08 CASE_FIRST */           {caseFirstArg,      _processCollatorOption, UCOL_CASE_FIRST }, // case first L, U, X, D
/* 09 NUMERIC_COLLATION */    {numericCollArg,    _processCollatorOption, UCOL_NUMERIC_COLLATION }, // codan      O, X, D
/* 07 CASE_LEVEL */           {caseLevelArg,      _processCollatorOption, UCOL_CASE_LEVEL }, // case level O, X, D
/* 12 FRENCH_COLLATION */     {frenchCollArg,     _processCollatorOption, UCOL_FRENCH_COLLATION }, // french     O, X, D
/* 13 HIRAGANA_QUATERNARY] */ {hiraganaQArg,      _processCollatorOption, UCOL_HIRAGANA_QUATERNARY_MODE }, // hiragana   O, X, D
/* 04 KEYWORD */              {keywordArg,        _processLocaleElement,  UCOL_SIT_KEYWORD }, // keyword
/* 00 LANGUAGE */             {languageArg,       _processLocaleElement,  UCOL_SIT_LANGUAGE }, // language
/* 11 NORMALIZATION_MODE */   {normArg,           _processCollatorOption, UCOL_NORMALIZATION_MODE }, // norm       O, X, D
/* 02 REGION */               {regionArg,         _processLocaleElement,  UCOL_SIT_REGION }, // region
/* 06 STRENGTH */             {strengthArg,       _processCollatorOption, UCOL_STRENGTH }, // strength   1, 2, 3, 4, I, D
/* 14 VARIABLE_TOP */         {variableTopArg,    _processVariableTop,    0 },
/* 03 VARIANT */              {variantArg,        _processLocaleElement,  UCOL_SIT_VARIANT }, // variant
/* 05 RFC3066BIS */           {RFC3066Arg,        _processRFC3066Locale,  0 }, // rfc3066bis locale name
/* 01 SCRIPT */               {scriptArg,         _processLocaleElement,  UCOL_SIT_SCRIPT },  // script
/*    PROVIDER */             {providerArg,       _processLocaleElement, UCOL_SIT_PROVIDER }
};


static
const char* ucol_sit_readOption(const char *start, CollatorSpec *spec,
                            UErrorCode *status)
{
  int32_t i = 0;

  for(i = 0; i < UCOL_SIT_ITEMS_COUNT; i++) {
      if(*start == options[i].optionStart) {
          const char* end = options[i].action(spec, options[i].attr, start+1, status);
#ifdef UCOL_TRACE_SIT
          fprintf(stderr, "***Set %d to %s...\n", i, start);
#endif
          // assume 'start' does not go away through all this
          spec->entries[i].copyFrom(CharString(start, static_cast<int32_t>(end - start), *status), *status);
          return end;
      }
  }
  *status = U_ILLEGAL_ARGUMENT_ERROR;
#ifdef UCOL_TRACE_SIT
  fprintf(stderr, "%s:%d: Unknown option at '%s': %s\n", __FILE__, __LINE__, start, u_errorName(*status));
#endif
  return start;
}

static const char*
ucol_sit_readSpecs(CollatorSpec *s, const char *string,
                        UParseError *parseError, UErrorCode *status)
{
    const char *definition = string;
    while(U_SUCCESS(*status) && *string) {
        string = ucol_sit_readOption(string, s, status);
        // advance over '_'
        while(*string && *string == '_') {
            string++;
        }
    }
    if(U_FAILURE(*status)) {
        parseError->offset = static_cast<int32_t>(string - definition);
    }
    return string;
}

static
int32_t ucol_sit_dumpSpecs(CollatorSpec *s, char *destination, int32_t capacity, UErrorCode *status)
{
    int32_t i = 0, j = 0;
    int32_t len = 0;
    char optName;
    if(U_SUCCESS(*status)) {
        for(i = 0; i < UCOL_SIT_ITEMS_COUNT; i++) {
            if(!s->entries[i].isEmpty()) {
                if(len) {
                    if(len < capacity) {
                        uprv_strcat(destination, "_");
                    }
                    len++;
                }
                optName = s->entries[i][0];
                if(optName == languageArg || optName == regionArg || optName == variantArg || optName == keywordArg) {
                    for(j = 0; j < s->entries[i].length(); j++) {
                        if(len + j < capacity) {
                            destination[len+j] = uprv_toupper(s->entries[i][j]);
                        }
                    }
                    len += s->entries[i].length();
                } else {
                    len += s->entries[i].extract(destination + len, capacity - len, *status);
                }
            }
        }
        return len;
    } else {
        return 0;
    }
}

static void
ucol_sit_calculateWholeLocale(CollatorSpec *s, UErrorCode &status) {
    // put the locale together, unless we have a done
    // locale
    if(s->locale.isEmpty()) {
        // first the language
        s->locale.append(s->locElements[UCOL_SIT_LANGUAGE], status);
        // then the script, if present
        if(!s->locElements[UCOL_SIT_SCRIPT].isEmpty()) {
            s->locale.append("_", status);
            s->locale.append(s->locElements[UCOL_SIT_SCRIPT], status);
        }
        // then the region, if present
        if(!s->locElements[UCOL_SIT_REGION].isEmpty()) {
            s->locale.append("_", status);
            s->locale.append(s->locElements[UCOL_SIT_REGION], status);
        } else if(!s->locElements[UCOL_SIT_VARIANT].isEmpty()) { // if there is a variant, we need an underscore
            s->locale.append("_", status);
        }
        // add variant, if there
        if(!s->locElements[UCOL_SIT_VARIANT].isEmpty()) {
            s->locale.append("_", status);
            s->locale.append(s->locElements[UCOL_SIT_VARIANT], status);
        }

        // if there is a collation keyword, add that too
        if(!s->locElements[UCOL_SIT_KEYWORD].isEmpty()) {
            s->locale.append(collationKeyword, status);
            s->locale.append(s->locElements[UCOL_SIT_KEYWORD], status);
        }

        // if there is a provider keyword, add that too
        if(!s->locElements[UCOL_SIT_PROVIDER].isEmpty()) {
            s->locale.append(providerKeyword, status);
            s->locale.append(s->locElements[UCOL_SIT_PROVIDER], status);
        }
    }
}


U_CAPI void U_EXPORT2
ucol_prepareShortStringOpen( const char *definition,
                          UBool,
                          UParseError *parseError,
                          UErrorCode *status)
{
    if(U_FAILURE(*status)) return;

    UParseError internalParseError;

    if(!parseError) {
        parseError = &internalParseError;
    }
    parseError->line = 0;
    parseError->offset = 0;
    parseError->preContext[0] = 0;
    parseError->postContext[0] = 0;


    // first we want to pick stuff out of short string.
    // we'll end up with an UCA version, locale and a bunch of
    // settings

    // analyse the string in order to get everything we need.
    CollatorSpec s;
    ucol_sit_readSpecs(&s, definition, parseError, status);
    ucol_sit_calculateWholeLocale(&s, *status);

    CharString buffer = ulocimp_canonicalize(s.locale.toStringPiece(), *status);

    UResourceBundle *b = ures_open(U_ICUDATA_COLL, buffer.data(), status);
    /* we try to find stuff from keyword */
    UResourceBundle *collations = ures_getByKey(b, "collations", nullptr, status);
    UResourceBundle *collElem = nullptr;
    // if there is a keyword, we pick it up and try to get elements
    CharString keyBuffer = ulocimp_getKeywordValue(buffer.data(), "collation", *status);
    if(keyBuffer.isEmpty()) {
      // no keyword
      // we try to find the default setting, which will give us the keyword value
      UResourceBundle *defaultColl = ures_getByKeyWithFallback(collations, "default", nullptr, status);
      if(U_SUCCESS(*status)) {
        int32_t defaultKeyLen = 0;
        const char16_t *defaultKey = ures_getString(defaultColl, &defaultKeyLen, status);
        keyBuffer.appendInvariantChars(defaultKey, defaultKeyLen, *status);
      } else {
        *status = U_INTERNAL_PROGRAM_ERROR;
        return;
      }
      ures_close(defaultColl);
    }
    collElem = ures_getByKeyWithFallback(collations, keyBuffer.data(), collElem, status);
    ures_close(collElem);
    ures_close(collations);
    ures_close(b);
}


U_CAPI UCollator* U_EXPORT2
ucol_openFromShortString( const char *definition,
                          UBool forceDefaults,
                          UParseError *parseError,
                          UErrorCode *status)
{
    UTRACE_ENTRY_OC(UTRACE_UCOL_OPEN_FROM_SHORT_STRING);
    UTRACE_DATA1(UTRACE_INFO, "short string = \"%s\"", definition);

    if (U_FAILURE(*status)) return nullptr;

    UParseError internalParseError;

    if(!parseError) {
        parseError = &internalParseError;
    }
    parseError->line = 0;
    parseError->offset = 0;
    parseError->preContext[0] = 0;
    parseError->postContext[0] = 0;


    // first we want to pick stuff out of short string.
    // we'll end up with an UCA version, locale and a bunch of
    // settings

    // analyse the string in order to get everything we need.
    const char *string = definition;
    CollatorSpec s;
    string = ucol_sit_readSpecs(&s, definition, parseError, status);
    ucol_sit_calculateWholeLocale(&s, *status);

#ifdef UCOL_TRACE_SIT
    fprintf(stderr, "DEF %s, DATA %s, ERR %s\n", definition, s.locale.data(), u_errorName(*status));
#endif
    CharString buffer = ulocimp_canonicalize(s.locale.toStringPiece(), *status);

    UCollator *result = ucol_open(buffer.data(), status);
    int32_t i = 0;

    for(i = 0; i < UCOL_ATTRIBUTE_COUNT; i++) {
        if(s.options[i] != UCOL_DEFAULT) {
            if(forceDefaults || ucol_getAttribute(result, (UColAttribute)i, status) != s.options[i]) {
                ucol_setAttribute(result, (UColAttribute)i, s.options[i], status);
            }

            if(U_FAILURE(*status)) {
                parseError->offset = (int32_t)(string - definition);
                ucol_close(result);
                return nullptr;
            }

        }
    }
    if(s.variableTopSet) {
        if(s.variableTopString[0]) {
            ucol_setVariableTop(result, s.variableTopString, s.variableTopStringLen, status);
        } else { // we set by value, using 'B'
            ucol_restoreVariableTop(result, s.variableTopValue, status);
        }
    }


    if(U_FAILURE(*status)) { // here it can only be a bogus value
        ucol_close(result);
        result = nullptr;
    }

    UTRACE_EXIT_PTR_STATUS(result, *status);
    return result;
}


U_CAPI int32_t U_EXPORT2
ucol_getShortDefinitionString(const UCollator *coll,
                              const char *locale,
                              char *dst,
                              int32_t capacity,
                              UErrorCode *status)
{
    if(U_FAILURE(*status)) return 0;
    if(coll == nullptr) {
        *status = U_ILLEGAL_ARGUMENT_ERROR;
        return 0;
    }
    return ((icu::Collator*)coll)->internalGetShortDefinitionString(locale,dst,capacity,*status);
}

U_CAPI int32_t U_EXPORT2
ucol_normalizeShortDefinitionString(const char *definition,
                                    char *destination,
                                    int32_t capacity,
                                    UParseError *parseError,
                                    UErrorCode *status)
{

    if(U_FAILURE(*status)) {
        return 0;
    }

    if(destination) {
        uprv_memset(destination, 0, capacity*sizeof(char));
    }

    UParseError pe;
    if(!parseError) {
        parseError = &pe;
    }

    // validate
    CollatorSpec s;
    ucol_sit_readSpecs(&s, definition, parseError, status);
    return ucol_sit_dumpSpecs(&s, destination, capacity, status);
}

/**
 * Get a set containing the contractions defined by the collator. The set includes
 * both the UCA contractions and the contractions defined by the collator
 * @param coll collator
 * @param conts the set to hold the result
 * @param status to hold the error code
 * @return the size of the contraction set
 */
U_CAPI int32_t U_EXPORT2
ucol_getContractions( const UCollator *coll,
                  USet *contractions,
                  UErrorCode *status)
{
  ucol_getContractionsAndExpansions(coll, contractions, nullptr, false, status);
  return uset_getItemCount(contractions);
}

/**
 * Get a set containing the expansions defined by the collator. The set includes
 * both the UCA expansions and the expansions defined by the tailoring
 * @param coll collator
 * @param conts the set to hold the result
 * @param addPrefixes add the prefix contextual elements to contractions
 * @param status to hold the error code
 *
 * @draft ICU 3.4
 */
U_CAPI void U_EXPORT2
ucol_getContractionsAndExpansions( const UCollator *coll,
                  USet *contractions,
                  USet *expansions,
                  UBool addPrefixes,
                  UErrorCode *status)
{
    if(U_FAILURE(*status)) {
        return;
    }
    if(coll == nullptr) {
        *status = U_ILLEGAL_ARGUMENT_ERROR;
        return;
    }
    const icu::RuleBasedCollator *rbc = icu::RuleBasedCollator::rbcFromUCollator(coll);
    if(rbc == nullptr) {
        *status = U_UNSUPPORTED_ERROR;
        return;
    }
    rbc->internalGetContractionsAndExpansions(
            icu::UnicodeSet::fromUSet(contractions),
            icu::UnicodeSet::fromUSet(expansions),
            addPrefixes, *status);
}
#endif

Coverage Report

Created: 2025-06-24 06:54

Line	Count	Source (jump to first uncovered line)
1		// © 2016 and later: Unicode, Inc. and others.
2		// License & terms of use: http://www.unicode.org/copyright.html
3		/*
4		*******************************************************************************
5		* Copyright (C) 2004-2016, International Business Machines
6		* Corporation and others. All Rights Reserved.
7		*******************************************************************************
8		* file name: ucol_sit.cpp
9		* encoding: UTF-8
10		* tab size: 8 (not used)
11		* indentation:4
12		*
13		* Modification history
14		* Date Name Comments
15		* 03/12/2004 weiv Creation
16		*/
17
18		#include "unicode/ustring.h"
19		#include "unicode/udata.h"
20		#include "unicode/utf16.h"
21		#include "utracimp.h"
22		#include "ucol_imp.h"
23		#include "ulocimp.h"
24		#include "cmemory.h"
25		#include "cstring.h"
26		#include "uresimp.h"
27		#include "unicode/coll.h"
28		#include "unicode/stringpiece.h"
29		#include "charstr.h"
30
31		U_NAMESPACE_USE
32
33		#ifdef UCOL_TRACE_SIT
34		# include <stdio.h>
35		#endif
36
37		#if !UCONFIG_NO_COLLATION
38
39		#include "unicode/tblcoll.h"
40
41		enum OptionsList {
42		UCOL_SIT_LANGUAGE = 0,
43		UCOL_SIT_SCRIPT = 1,
44		UCOL_SIT_REGION = 2,
45		UCOL_SIT_VARIANT = 3,
46		UCOL_SIT_KEYWORD = 4,
47		UCOL_SIT_PROVIDER = 5,
48		UCOL_SIT_LOCELEMENT_MAX = UCOL_SIT_PROVIDER, /* the last element that's part of LocElements */
49
50		UCOL_SIT_BCP47,
51		UCOL_SIT_STRENGTH,
52		UCOL_SIT_CASE_LEVEL,
53		UCOL_SIT_CASE_FIRST,
54		UCOL_SIT_NUMERIC_COLLATION,
55		UCOL_SIT_ALTERNATE_HANDLING,
56		UCOL_SIT_NORMALIZATION_MODE,
57		UCOL_SIT_FRENCH_COLLATION,
58		UCOL_SIT_HIRAGANA_QUATERNARY,
59		UCOL_SIT_VARIABLE_TOP,
60		UCOL_SIT_VARIABLE_TOP_VALUE,
61		UCOL_SIT_ITEMS_COUNT
62		};
63
64		/* option starters chars. */
65		static const char alternateHArg = 'A';
66		static const char variableTopValArg = 'B';
67		static const char caseFirstArg = 'C';
68		static const char numericCollArg = 'D';
69		static const char caseLevelArg = 'E';
70		static const char frenchCollArg = 'F';
71		static const char hiraganaQArg = 'H';
72		static const char keywordArg = 'K';
73		static const char languageArg = 'L';
74		static const char normArg = 'N';
75		static const char providerArg = 'P';
76		static const char regionArg = 'R';
77		static const char strengthArg = 'S';
78		static const char variableTopArg = 'T';
79		static const char variantArg = 'V';
80		static const char RFC3066Arg = 'X';
81		static const char scriptArg = 'Z';
82
83		static const char collationKeyword[] = "@collation=";
84		static const char providerKeyword[] = "@sp=";
85
86
87		static const int32_t locElementCount = UCOL_SIT_LOCELEMENT_MAX+1;
88		static const int32_t locElementCapacity = 32;
89		static const int32_t loc3066Capacity = 256;
90
91		/* structure containing specification of a collator. Initialized
92		* from a short string. Also used to construct a short string from a
93		* collator instance
94		*/
95		struct CollatorSpec {
96		inline CollatorSpec();
97
98		CharString locElements[locElementCount];
99		CharString locale;
100		UColAttributeValue options[UCOL_ATTRIBUTE_COUNT];
101		uint32_t variableTopValue;
102		char16_t variableTopString[locElementCapacity];
103		int32_t variableTopStringLen;
104		UBool variableTopSet;
105		CharString entries[UCOL_SIT_ITEMS_COUNT];
106		};
107
108		CollatorSpec::CollatorSpec() :
109	0	locale(),
110	0	variableTopValue(0),
111		variableTopString(),
112	0	variableTopSet(false)
113	0	{
114		// set collation options to default
115	0	for(int32_t i = 0; i < UCOL_ATTRIBUTE_COUNT; i++) {
116	0	options[i] = UCOL_DEFAULT;
117	0	}
118	0	}
119
120
121		/* structure for converting between character attribute
122		* representation and real collation attribute value.
123		*/
124		struct AttributeConversion {
125		char letter;
126		UColAttributeValue value;
127		};
128
129		static const AttributeConversion conversions[12] = {
130		{ '1', UCOL_PRIMARY },
131		{ '2', UCOL_SECONDARY },
132		{ '3', UCOL_TERTIARY },
133		{ '4', UCOL_QUATERNARY },
134		{ 'D', UCOL_DEFAULT },
135		{ 'I', UCOL_IDENTICAL },
136		{ 'L', UCOL_LOWER_FIRST },
137		{ 'N', UCOL_NON_IGNORABLE },
138		{ 'O', UCOL_ON },
139		{ 'S', UCOL_SHIFTED },
140		{ 'U', UCOL_UPPER_FIRST },
141		{ 'X', UCOL_OFF }
142		};
143
144
145		static UColAttributeValue
146	0	ucol_sit_letterToAttributeValue(char letter, UErrorCode *status) {
147	0	uint32_t i = 0;
148	0	for(i = 0; i < UPRV_LENGTHOF(conversions); i++) {
149	0	if(conversions[i].letter == letter) {
150	0	return conversions[i].value;
151	0	}
152	0	}
153	0	*status = U_ILLEGAL_ARGUMENT_ERROR;
154		#ifdef UCOL_TRACE_SIT
155		fprintf(stderr, "%s:%d: unknown letter %c: %s\n", __FILE__, __LINE__, letter, u_errorName(*status));
156		#endif
157	0	return UCOL_DEFAULT;
158	0	}
159
160		/* function prototype for functions used to parse a short string */
161		U_CDECL_BEGIN
162		typedef const char* U_CALLCONV
163		ActionFunction(CollatorSpec spec, uint32_t value1, const char string,
164		UErrorCode *status);
165		U_CDECL_END
166
167		U_CDECL_BEGIN
168		static const char* U_CALLCONV
169		_processLocaleElement(CollatorSpec spec, uint32_t value, const char string,
170		UErrorCode *status)
171	0	{
172	0	do {
173	0	if(value == UCOL_SIT_LANGUAGE \|\| value == UCOL_SIT_KEYWORD \|\| value == UCOL_SIT_PROVIDER) {
174	0	spec->locElements[value].append(uprv_tolower(string), status);
175	0	} else {
176	0	spec->locElements[value].append(string, status);
177	0	}
178	0	} while((++string) != '_' && string && U_SUCCESS(*status));
179		// don't skip the underscore at the end
180	0	return string;
181	0	}
182		U_CDECL_END
183
184		U_CDECL_BEGIN
185		static const char* U_CALLCONV
186		_processRFC3066Locale(CollatorSpec spec, uint32_t, const char string,
187		UErrorCode *status)
188	0	{
189	0	char terminator = *string;
190	0	string++;
191	0	const char *end = uprv_strchr(string+1, terminator);
192	0	if(end == nullptr \|\| end - string >= loc3066Capacity) {
193	0	*status = U_BUFFER_OVERFLOW_ERROR;
194	0	return string;
195	0	} else {
196	0	spec->locale.copyFrom(CharString(string, static_cast<int32_t>(end-string), status), status);
197	0	return end+1;
198	0	}
199	0	}
200
201		U_CDECL_END
202
203		U_CDECL_BEGIN
204		static const char* U_CALLCONV
205		_processCollatorOption(CollatorSpec spec, uint32_t option, const char string,
206		UErrorCode *status)
207	0	{
208	0	spec->options[option] = ucol_sit_letterToAttributeValue(*string, status);
209	0	if(((++string) != '_' && string) \|\| U_FAILURE(*status)) {
210		#ifdef UCOL_TRACE_SIT
211		fprintf(stderr, "%s:%d: unknown collator option at '%s': %s\n", __FILE__, __LINE__, string, u_errorName(*status));
212		#endif
213	0	*status = U_ILLEGAL_ARGUMENT_ERROR;
214	0	}
215	0	return string;
216	0	}
217		U_CDECL_END
218
219
220		static char16_t
221		readHexCodeUnit(const char *string, UErrorCode status)
222	0	{
223	0	char16_t result = 0;
224	0	int32_t value = 0;
225	0	char c;
226	0	int32_t noDigits = 0;
227	0	while((c = **string) != 0 && noDigits < 4) {
228	0	if( c >= '0' && c <= '9') {
229	0	value = c - '0';
230	0	} else if ( c >= 'a' && c <= 'f') {
231	0	value = c - 'a' + 10;
232	0	} else if ( c >= 'A' && c <= 'F') {
233	0	value = c - 'A' + 10;
234	0	} else {
235	0	*status = U_ILLEGAL_ARGUMENT_ERROR;
236		#ifdef UCOL_TRACE_SIT
237		fprintf(stderr, "%s:%d: Bad hex char at '%s': %s\n", __FILE__, __LINE__, string, u_errorName(status));
238		#endif
239	0	return 0;
240	0	}
241	0	result = (result << 4) \| static_cast<char16_t>(value);
242	0	noDigits++;
243	0	(*string)++;
244	0	}
245		// if the string was terminated before we read 4 digits, set an error
246	0	if(noDigits < 4) {
247	0	*status = U_ILLEGAL_ARGUMENT_ERROR;
248		#ifdef UCOL_TRACE_SIT
249		fprintf(stderr, "%s:%d: Short (only %d digits, wanted 4) at '%s': %s\n", __FILE__, __LINE__, noDigits,string, u_errorName(status));
250		#endif
251	0	}
252	0	return result;
253	0	}
254
255		U_CDECL_BEGIN
256		static const char* U_CALLCONV
257		_processVariableTop(CollatorSpec spec, uint32_t value1, const char string, UErrorCode *status)
258	0	{
259		// get four digits
260	0	int32_t i = 0;
261	0	if(!value1) {
262	0	while(U_SUCCESS(status) && i < locElementCapacity && string != 0 && *string != '_') {
263	0	spec->variableTopString[i++] = readHexCodeUnit(&string, status);
264	0	}
265	0	spec->variableTopStringLen = i;
266	0	if(i == locElementCapacity && string != 0 && string != '_') {
267	0	*status = U_BUFFER_OVERFLOW_ERROR;
268	0	}
269	0	} else {
270	0	spec->variableTopValue = readHexCodeUnit(&string, status);
271	0	}
272	0	if(U_SUCCESS(*status)) {
273	0	spec->variableTopSet = true;
274	0	}
275	0	return string;
276	0	}
277		U_CDECL_END
278
279
280		/* Table for parsing short strings */
281		struct ShortStringOptions {
282		char optionStart;
283		ActionFunction *action;
284		uint32_t attr;
285		};
286
287		static const ShortStringOptions options[UCOL_SIT_ITEMS_COUNT] =
288		{
289		/* 10 ALTERNATE_HANDLING */ {alternateHArg, _processCollatorOption, UCOL_ALTERNATE_HANDLING }, // alternate N, S, D
290		/* 15 VARIABLE_TOP_VALUE */ {variableTopValArg, _processVariableTop, 1 },
291		/* 08 CASE_FIRST */ {caseFirstArg, _processCollatorOption, UCOL_CASE_FIRST }, // case first L, U, X, D
292		/* 09 NUMERIC_COLLATION */ {numericCollArg, _processCollatorOption, UCOL_NUMERIC_COLLATION }, // codan O, X, D
293		/* 07 CASE_LEVEL */ {caseLevelArg, _processCollatorOption, UCOL_CASE_LEVEL }, // case level O, X, D
294		/* 12 FRENCH_COLLATION */ {frenchCollArg, _processCollatorOption, UCOL_FRENCH_COLLATION }, // french O, X, D
295		/* 13 HIRAGANA_QUATERNARY] */ {hiraganaQArg, _processCollatorOption, UCOL_HIRAGANA_QUATERNARY_MODE }, // hiragana O, X, D
296		/* 04 KEYWORD */ {keywordArg, _processLocaleElement, UCOL_SIT_KEYWORD }, // keyword
297		/* 00 LANGUAGE */ {languageArg, _processLocaleElement, UCOL_SIT_LANGUAGE }, // language
298		/* 11 NORMALIZATION_MODE */ {normArg, _processCollatorOption, UCOL_NORMALIZATION_MODE }, // norm O, X, D
299		/* 02 REGION */ {regionArg, _processLocaleElement, UCOL_SIT_REGION }, // region
300		/* 06 STRENGTH */ {strengthArg, _processCollatorOption, UCOL_STRENGTH }, // strength 1, 2, 3, 4, I, D
301		/* 14 VARIABLE_TOP */ {variableTopArg, _processVariableTop, 0 },
302		/* 03 VARIANT */ {variantArg, _processLocaleElement, UCOL_SIT_VARIANT }, // variant
303		/* 05 RFC3066BIS */ {RFC3066Arg, _processRFC3066Locale, 0 }, // rfc3066bis locale name
304		/* 01 SCRIPT */ {scriptArg, _processLocaleElement, UCOL_SIT_SCRIPT }, // script
305		/* PROVIDER */ {providerArg, _processLocaleElement, UCOL_SIT_PROVIDER }
306		};
307
308
309		static
310		const char* ucol_sit_readOption(const char start, CollatorSpec spec,
311		UErrorCode *status)
312	0	{
313	0	int32_t i = 0;
314
315	0	for(i = 0; i < UCOL_SIT_ITEMS_COUNT; i++) {
316	0	if(*start == options[i].optionStart) {
317	0	const char* end = options[i].action(spec, options[i].attr, start+1, status);
318		#ifdef UCOL_TRACE_SIT
319		fprintf(stderr, "***Set %d to %s...\n", i, start);
320		#endif
321		// assume 'start' does not go away through all this
322	0	spec->entries[i].copyFrom(CharString(start, static_cast<int32_t>(end - start), status), status);
323	0	return end;
324	0	}
325	0	}
326	0	*status = U_ILLEGAL_ARGUMENT_ERROR;
327		#ifdef UCOL_TRACE_SIT
328		fprintf(stderr, "%s:%d: Unknown option at '%s': %s\n", __FILE__, __LINE__, start, u_errorName(*status));
329		#endif
330	0	return start;
331	0	}
332
333		static const char*
334		ucol_sit_readSpecs(CollatorSpec s, const char string,
335		UParseError parseError, UErrorCode status)
336	0	{
337	0	const char *definition = string;
338	0	while(U_SUCCESS(status) && string) {
339	0	string = ucol_sit_readOption(string, s, status);
340		// advance over '_'
341	0	while(string && string == '_') {
342	0	string++;
343	0	}
344	0	}
345	0	if(U_FAILURE(*status)) {
346	0	parseError->offset = static_cast<int32_t>(string - definition);
347	0	}
348	0	return string;
349	0	}
350
351		static
352		int32_t ucol_sit_dumpSpecs(CollatorSpec s, char destination, int32_t capacity, UErrorCode *status)
353	0	{
354	0	int32_t i = 0, j = 0;
355	0	int32_t len = 0;
356	0	char optName;
357	0	if(U_SUCCESS(*status)) {
358	0	for(i = 0; i < UCOL_SIT_ITEMS_COUNT; i++) {
359	0	if(!s->entries[i].isEmpty()) {
360	0	if(len) {
361	0	if(len < capacity) {
362	0	uprv_strcat(destination, "_");
363	0	}
364	0	len++;
365	0	}
366	0	optName = s->entries[i][0];
367	0	if(optName == languageArg \|\| optName == regionArg \|\| optName == variantArg \|\| optName == keywordArg) {
368	0	for(j = 0; j < s->entries[i].length(); j++) {
369	0	if(len + j < capacity) {
370	0	destination[len+j] = uprv_toupper(s->entries[i][j]);
371	0	}
372	0	}
373	0	len += s->entries[i].length();
374	0	} else {
375	0	len += s->entries[i].extract(destination + len, capacity - len, *status);
376	0	}
377	0	}
378	0	}
379	0	return len;
380	0	} else {
381	0	return 0;
382	0	}
383	0	}
384
385		static void
386	0	ucol_sit_calculateWholeLocale(CollatorSpec *s, UErrorCode &status) {
387		// put the locale together, unless we have a done
388		// locale
389	0	if(s->locale.isEmpty()) {
390		// first the language
391	0	s->locale.append(s->locElements[UCOL_SIT_LANGUAGE], status);
392		// then the script, if present
393	0	if(!s->locElements[UCOL_SIT_SCRIPT].isEmpty()) {
394	0	s->locale.append("_", status);
395	0	s->locale.append(s->locElements[UCOL_SIT_SCRIPT], status);
396	0	}
397		// then the region, if present
398	0	if(!s->locElements[UCOL_SIT_REGION].isEmpty()) {
399	0	s->locale.append("_", status);
400	0	s->locale.append(s->locElements[UCOL_SIT_REGION], status);
401	0	} else if(!s->locElements[UCOL_SIT_VARIANT].isEmpty()) { // if there is a variant, we need an underscore
402	0	s->locale.append("_", status);
403	0	}
404		// add variant, if there
405	0	if(!s->locElements[UCOL_SIT_VARIANT].isEmpty()) {
406	0	s->locale.append("_", status);
407	0	s->locale.append(s->locElements[UCOL_SIT_VARIANT], status);
408	0	}
409
410		// if there is a collation keyword, add that too
411	0	if(!s->locElements[UCOL_SIT_KEYWORD].isEmpty()) {
412	0	s->locale.append(collationKeyword, status);
413	0	s->locale.append(s->locElements[UCOL_SIT_KEYWORD], status);
414	0	}
415
416		// if there is a provider keyword, add that too
417	0	if(!s->locElements[UCOL_SIT_PROVIDER].isEmpty()) {
418	0	s->locale.append(providerKeyword, status);
419	0	s->locale.append(s->locElements[UCOL_SIT_PROVIDER], status);
420	0	}
421	0	}
422	0	}
423
424
425		U_CAPI void U_EXPORT2
426		ucol_prepareShortStringOpen( const char *definition,
427		UBool,
428		UParseError *parseError,
429		UErrorCode *status)
430	0	{
431	0	if(U_FAILURE(*status)) return;
432
433	0	UParseError internalParseError;
434
435	0	if(!parseError) {
436	0	parseError = &internalParseError;
437	0	}
438	0	parseError->line = 0;
439	0	parseError->offset = 0;
440	0	parseError->preContext[0] = 0;
441	0	parseError->postContext[0] = 0;
442
443
444		// first we want to pick stuff out of short string.
445		// we'll end up with an UCA version, locale and a bunch of
446		// settings
447
448		// analyse the string in order to get everything we need.
449	0	CollatorSpec s;
450	0	ucol_sit_readSpecs(&s, definition, parseError, status);
451	0	ucol_sit_calculateWholeLocale(&s, *status);
452
453	0	CharString buffer = ulocimp_canonicalize(s.locale.toStringPiece(), *status);
454
455	0	UResourceBundle *b = ures_open(U_ICUDATA_COLL, buffer.data(), status);
456		/* we try to find stuff from keyword */
457	0	UResourceBundle *collations = ures_getByKey(b, "collations", nullptr, status);
458	0	UResourceBundle *collElem = nullptr;
459		// if there is a keyword, we pick it up and try to get elements
460	0	CharString keyBuffer = ulocimp_getKeywordValue(buffer.data(), "collation", *status);
461	0	if(keyBuffer.isEmpty()) {
462		// no keyword
463		// we try to find the default setting, which will give us the keyword value
464	0	UResourceBundle *defaultColl = ures_getByKeyWithFallback(collations, "default", nullptr, status);
465	0	if(U_SUCCESS(*status)) {
466	0	int32_t defaultKeyLen = 0;
467	0	const char16_t *defaultKey = ures_getString(defaultColl, &defaultKeyLen, status);
468	0	keyBuffer.appendInvariantChars(defaultKey, defaultKeyLen, *status);
469	0	} else {
470	0	*status = U_INTERNAL_PROGRAM_ERROR;
471	0	return;
472	0	}
473	0	ures_close(defaultColl);
474	0	}
475	0	collElem = ures_getByKeyWithFallback(collations, keyBuffer.data(), collElem, status);
476	0	ures_close(collElem);
477	0	ures_close(collations);
478	0	ures_close(b);
479	0	}
480
481
482		U_CAPI UCollator* U_EXPORT2
483		ucol_openFromShortString( const char *definition,
484		UBool forceDefaults,
485		UParseError *parseError,
486		UErrorCode *status)
487	0	{
488	0	UTRACE_ENTRY_OC(UTRACE_UCOL_OPEN_FROM_SHORT_STRING);
489	0	UTRACE_DATA1(UTRACE_INFO, "short string = \"%s\"", definition);
490
491	0	if (U_FAILURE(*status)) return nullptr;
492
493	0	UParseError internalParseError;
494
495	0	if(!parseError) {
496	0	parseError = &internalParseError;
497	0	}
498	0	parseError->line = 0;
499	0	parseError->offset = 0;
500	0	parseError->preContext[0] = 0;
501	0	parseError->postContext[0] = 0;
502
503
504		// first we want to pick stuff out of short string.
505		// we'll end up with an UCA version, locale and a bunch of
506		// settings
507
508		// analyse the string in order to get everything we need.
509	0	const char *string = definition;
510	0	CollatorSpec s;
511	0	string = ucol_sit_readSpecs(&s, definition, parseError, status);
512	0	ucol_sit_calculateWholeLocale(&s, *status);
513
514		#ifdef UCOL_TRACE_SIT
515		fprintf(stderr, "DEF %s, DATA %s, ERR %s\n", definition, s.locale.data(), u_errorName(*status));
516		#endif
517	0	CharString buffer = ulocimp_canonicalize(s.locale.toStringPiece(), *status);
518
519	0	UCollator *result = ucol_open(buffer.data(), status);
520	0	int32_t i = 0;
521
522	0	for(i = 0; i < UCOL_ATTRIBUTE_COUNT; i++) {
523	0	if(s.options[i] != UCOL_DEFAULT) {
524	0	if(forceDefaults \|\| ucol_getAttribute(result, (UColAttribute)i, status) != s.options[i]) {
525	0	ucol_setAttribute(result, (UColAttribute)i, s.options[i], status);
526	0	}
527
528	0	if(U_FAILURE(*status)) {
529	0	parseError->offset = (int32_t)(string - definition);
530	0	ucol_close(result);
531	0	return nullptr;
532	0	}
533
534	0	}
535	0	}
536	0	if(s.variableTopSet) {
537	0	if(s.variableTopString[0]) {
538	0	ucol_setVariableTop(result, s.variableTopString, s.variableTopStringLen, status);
539	0	} else { // we set by value, using 'B'
540	0	ucol_restoreVariableTop(result, s.variableTopValue, status);
541	0	}
542	0	}
543
544
545	0	if(U_FAILURE(*status)) { // here it can only be a bogus value
546	0	ucol_close(result);
547	0	result = nullptr;
548	0	}
549
550	0	UTRACE_EXIT_PTR_STATUS(result, *status);
551	0	return result;
552	0	}
553
554
555		U_CAPI int32_t U_EXPORT2
556		ucol_getShortDefinitionString(const UCollator *coll,
557		const char *locale,
558		char *dst,
559		int32_t capacity,
560		UErrorCode *status)
561	0	{
562	0	if(U_FAILURE(*status)) return 0;
563	0	if(coll == nullptr) {
564	0	*status = U_ILLEGAL_ARGUMENT_ERROR;
565	0	return 0;
566	0	}
567	0	return ((icu::Collator)coll)->internalGetShortDefinitionString(locale,dst,capacity,status);
568	0	}
569
570		U_CAPI int32_t U_EXPORT2
571		ucol_normalizeShortDefinitionString(const char *definition,
572		char *destination,
573		int32_t capacity,
574		UParseError *parseError,
575		UErrorCode *status)
576	0	{
577
578	0	if(U_FAILURE(*status)) {
579	0	return 0;
580	0	}
581
582	0	if(destination) {
583	0	uprv_memset(destination, 0, capacity*sizeof(char));
584	0	}
585
586	0	UParseError pe;
587	0	if(!parseError) {
588	0	parseError = &pe;
589	0	}
590
591		// validate
592	0	CollatorSpec s;
593	0	ucol_sit_readSpecs(&s, definition, parseError, status);
594	0	return ucol_sit_dumpSpecs(&s, destination, capacity, status);
595	0	}
596
597		/**
598		* Get a set containing the contractions defined by the collator. The set includes
599		* both the UCA contractions and the contractions defined by the collator
600		* @param coll collator
601		* @param conts the set to hold the result
602		* @param status to hold the error code
603		* @return the size of the contraction set
604		*/
605		U_CAPI int32_t U_EXPORT2
606		ucol_getContractions( const UCollator *coll,
607		USet *contractions,
608		UErrorCode *status)
609	0	{
610	0	ucol_getContractionsAndExpansions(coll, contractions, nullptr, false, status);
611	0	return uset_getItemCount(contractions);
612	0	}
613
614		/**
615		* Get a set containing the expansions defined by the collator. The set includes
616		* both the UCA expansions and the expansions defined by the tailoring
617		* @param coll collator
618		* @param conts the set to hold the result
619		* @param addPrefixes add the prefix contextual elements to contractions
620		* @param status to hold the error code
621		*
622		* @draft ICU 3.4
623		*/
624		U_CAPI void U_EXPORT2
625		ucol_getContractionsAndExpansions( const UCollator *coll,
626		USet *contractions,
627		USet *expansions,
628		UBool addPrefixes,
629		UErrorCode *status)
630	0	{
631	0	if(U_FAILURE(*status)) {
632	0	return;
633	0	}
634	0	if(coll == nullptr) {
635	0	*status = U_ILLEGAL_ARGUMENT_ERROR;
636	0	return;
637	0	}
638	0	const icu::RuleBasedCollator *rbc = icu::RuleBasedCollator::rbcFromUCollator(coll);
639	0	if(rbc == nullptr) {
640	0	*status = U_UNSUPPORTED_ERROR;
641	0	return;
642	0	}
643	0	rbc->internalGetContractionsAndExpansions(
644	0	icu::UnicodeSet::fromUSet(contractions),
645	0	icu::UnicodeSet::fromUSet(expansions),
646	0	addPrefixes, *status);
647	0	}
648		#endif