/*
*******************************************************************************
-* Copyright (C) 2004-2006, International Business Machines
+* Copyright (C) 2004-2014, International Business Machines
* Corporation and others. All Rights Reserved.
*******************************************************************************
* file name: ucol_sit.cpp
*/
#include "unicode/ustring.h"
-
+#include "unicode/udata.h"
+#include "unicode/utf16.h"
#include "utracimp.h"
#include "ucol_imp.h"
-#include "ucol_tok.h"
-#include "unormimp.h"
#include "cmemory.h"
#include "cstring.h"
+#include "uresimp.h"
+#include "unicode/coll.h"
+
+#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,
- UCOL_SIT_REGION,
- UCOL_SIT_VARIANT,
- UCOL_SIT_KEYWORD,
- UCOL_SIT_RFC3166BIS,
+ 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_ITEMS_COUNT
};
-/* list of locales for packing of a collator to an integer.
- * This list corresponds to ICU 3.0. If more collation bearing
- * locales are added in the future, this won't be a simple array
- * but a mapping allowing forward and reverse lookup would have to
- * be established. Currently, the mapping is from locale name to
- * index.
- */
-static const char* const locales[] = {
-/* 00 - 09 */ "ar", "be", "bg", "ca", "cs", "da", "de", "de__PHONEBOOK", "el", "en",
-/* 10 - 19 */ "en_BE", "eo", "es", "es__TRADITIONAL", "et", "fa", "fa_AF", "fi", "fo", "fr",
-/* 20 - 29 */ "gu", "he", "hi", "hi__DIRECT", "hr", "hu", "is", "it", "ja", "kk",
-/* 30 - 39 */ "kl", "kn", "ko", "lt", "lv", "mk", "mr", "mt", "nb", "nn",
-/* 40 - 49 */ "om", "pa", "pl", "ps", "ro", "root", "ru", "sh", "sk", "sl",
-/* 50 - 59 */ "sq", "sr", "sv", "ta", "te", "th", "tr", "uk", "vi", "zh",
-/* 60 - 64 */ "zh_HK", "zh_MO", "zh_TW", "zh_TW_STROKE", "zh__PINYIN"
-};
-
-static const char* const keywords[] = {
-/* 00 */ "",
-/* 01 */ "direct",
-/* 02 */ "phonebook",
-/* 03 */ "pinyin",
-/* 04 */ "standard",
-/* 05 */ "stroke",
-/* 06 */ "traditional"
-};
-
-
/* option starters chars. */
static const char alternateHArg = 'A';
static const char variableTopValArg = 'B';
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 scriptArg = 'Z';
static const char collationKeyword[] = "@collation=";
+static const char providerKeyword[] = "@sp=";
-static const int32_t locElementCount = 5;
+
+static const int32_t locElementCount = UCOL_SIT_LOCELEMENT_MAX+1;
static const int32_t locElementCapacity = 32;
static const int32_t loc3066Capacity = 256;
+static const int32_t locProviderCapacity = 10;
static const int32_t internalBufferSize = 512;
/* structure containing specification of a collator. Initialized
struct CollatorSpec {
char locElements[locElementCount][locElementCapacity];
char locale[loc3066Capacity];
+ char provider[locProviderCapacity];
UColAttributeValue options[UCOL_ATTRIBUTE_COUNT];
uint32_t variableTopValue;
UChar variableTopString[locElementCapacity];
};
-static char
-ucol_sit_attributeValueToLetter(UColAttributeValue value, UErrorCode *status) {
- uint32_t i = 0;
- for(i = 0; i < sizeof(conversions)/sizeof(conversions[0]); i++) {
- if(conversions[i].value == value) {
- return conversions[i].letter;
- }
- }
- *status = U_ILLEGAL_ARGUMENT_ERROR;
- return 0;
-}
-
-static UColAttributeValue
+static UColAttributeValue
ucol_sit_letterToAttributeValue(char letter, UErrorCode *status) {
uint32_t i = 0;
for(i = 0; i < sizeof(conversions)/sizeof(conversions[0]); i++) {
}
}
*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;
}
U_CDECL_BEGIN
static const char* U_CALLCONV
-_processLocaleElement(CollatorSpec *spec, uint32_t value, const char* string,
- UErrorCode *status)
+_processLocaleElement(CollatorSpec *spec, uint32_t value, const char* string,
+ UErrorCode *status)
{
int32_t len = 0;
do {
- if(value == 0 || value == 4) {
+ if(value == UCOL_SIT_LANGUAGE || value == UCOL_SIT_KEYWORD || value == UCOL_SIT_PROVIDER) {
spec->locElements[value][len++] = uprv_tolower(*string);
} else {
spec->locElements[value][len++] = *string;
U_CDECL_BEGIN
static const char* U_CALLCONV
-_processRFC3066Locale(CollatorSpec *spec, uint32_t, const char* string,
- UErrorCode *status)
+_processRFC3066Locale(CollatorSpec *spec, uint32_t, const char* string,
+ UErrorCode *status)
{
char terminator = *string;
string++;
U_CDECL_BEGIN
static const char* U_CALLCONV
-_processCollatorOption(CollatorSpec *spec, uint32_t option, const char* string,
- UErrorCode *status)
+_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 UChar
-readHexCodeUnit(const char **string, UErrorCode *status)
+static UChar
+readHexCodeUnit(const char **string, UErrorCode *status)
{
UChar result = 0;
int32_t value = 0;
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) | (UChar)value;
// 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)
+_processVariableTop(CollatorSpec *spec, uint32_t value1, const char* string, UErrorCode *status)
{
// get four digits
int32_t i = 0;
spec->variableTopString[i++] = readHexCodeUnit(&string, status);
}
spec->variableTopStringLen = i;
- if(i == locElementCapacity && (*string != 0 || *string != '_')) {
+ if(i == locElementCapacity && *string != 0 && *string != '_') {
*status = U_BUFFER_OVERFLOW_ERROR;
}
} else {
}
if(U_SUCCESS(*status)) {
spec->variableTopSet = TRUE;
- }
+ }
return string;
}
U_CDECL_END
static const ShortStringOptions options[UCOL_SIT_ITEMS_COUNT] =
{
-/* 10 ALTERNATE_HANDLING */ {alternateHArg, _processCollatorOption, UCOL_ALTERNATE_HANDLING }, // alternate N, S, D
+/* 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, 4 }, // keyword
-/* 00 LANGUAGE */ {languageArg, _processLocaleElement, 0 }, // language
+/* 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, 2 }, // region
+/* 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, 3 }, // variant
+/* 03 VARIANT */ {variantArg, _processLocaleElement, UCOL_SIT_VARIANT }, // variant
/* 05 RFC3066BIS */ {RFC3066Arg, _processRFC3066Locale, 0 }, // rfc3066bis locale name
-/* 01 SCRIPT */ {scriptArg, _processLocaleElement, 1 } // script
+/* 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)
+const char* ucol_sit_readOption(const char *start, CollatorSpec *spec,
+ UErrorCode *status)
{
int32_t i = 0;
if(*start == options[i].optionStart) {
spec->entries[i].start = start;
const char* end = options[i].action(spec, options[i].attr, start+1, status);
- spec->entries[i].len = end - start;
+ spec->entries[i].len = (int32_t)(end - start);
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
-void ucol_sit_initCollatorSpecs(CollatorSpec *spec)
+void ucol_sit_initCollatorSpecs(CollatorSpec *spec)
{
// reset everything
uprv_memset(spec, 0, sizeof(CollatorSpec));
}
}
-static const char*
-ucol_sit_readSpecs(CollatorSpec *s, const char *string,
+static const char*
+ucol_sit_readSpecs(CollatorSpec *s, const char *string,
UParseError *parseError, UErrorCode *status)
{
const char *definition = string;
- while(U_SUCCESS(*status) && *string) {
+ while(U_SUCCESS(*status) && *string) {
string = ucol_sit_readOption(string, s, status);
// advance over '_'
while(*string && *string == '_') {
}
}
if(U_FAILURE(*status)) {
- parseError->offset = string - definition;
+ parseError->offset = (int32_t)(string - definition);
}
return string;
}
{
int32_t i = 0, j = 0;
int32_t len = 0;
- char optName;
+ char optName;
if(U_SUCCESS(*status)) {
for(i = 0; i < UCOL_SIT_ITEMS_COUNT; i++) {
if(s->entries[i].start) {
uprv_strcat(destination, "_");
}
len++;
- }
+ }
optName = *(s->entries[i].start);
if(optName == languageArg || optName == regionArg || optName == variantArg || optName == keywordArg) {
for(j = 0; j < s->entries[i].len; j++) {
// locale
if(s->locale[0] == 0) {
// first the language
- uprv_strcat(s->locale, s->locElements[0]);
+ uprv_strcat(s->locale, s->locElements[UCOL_SIT_LANGUAGE]);
// then the script, if present
- if(*(s->locElements[1])) {
+ if(*(s->locElements[UCOL_SIT_SCRIPT])) {
uprv_strcat(s->locale, "_");
- uprv_strcat(s->locale, s->locElements[1]);
+ uprv_strcat(s->locale, s->locElements[UCOL_SIT_SCRIPT]);
}
// then the region, if present
- if(*(s->locElements[2])) {
+ if(*(s->locElements[UCOL_SIT_REGION])) {
uprv_strcat(s->locale, "_");
- uprv_strcat(s->locale, s->locElements[2]);
- } else if(*(s->locElements[3])) { // if there is a variant, we need an underscore
+ uprv_strcat(s->locale, s->locElements[UCOL_SIT_REGION]);
+ } else if(*(s->locElements[UCOL_SIT_VARIANT])) { // if there is a variant, we need an underscore
uprv_strcat(s->locale, "_");
}
// add variant, if there
- if(*(s->locElements[3])) {
+ if(*(s->locElements[UCOL_SIT_VARIANT])) {
uprv_strcat(s->locale, "_");
- uprv_strcat(s->locale, s->locElements[3]);
+ uprv_strcat(s->locale, s->locElements[UCOL_SIT_VARIANT]);
}
// if there is a collation keyword, add that too
- if(*(s->locElements[4])) {
+ if(*(s->locElements[UCOL_SIT_KEYWORD])) {
uprv_strcat(s->locale, collationKeyword);
- uprv_strcat(s->locale, s->locElements[4]);
+ uprv_strcat(s->locale, s->locElements[UCOL_SIT_KEYWORD]);
+ }
+
+ // if there is a provider keyword, add that too
+ if(*(s->locElements[UCOL_SIT_PROVIDER])) {
+ uprv_strcat(s->locale, providerKeyword);
+ uprv_strcat(s->locale, s->locElements[UCOL_SIT_PROVIDER]);
}
}
}
// settings
// analyse the string in order to get everything we need.
- const char *string = definition;
CollatorSpec s;
ucol_sit_initCollatorSpecs(&s);
- string = ucol_sit_readSpecs(&s, definition, parseError, status);
+ ucol_sit_readSpecs(&s, definition, parseError, status);
ucol_sit_calculateWholeLocale(&s);
-
+
char buffer[internalBufferSize];
uprv_memset(buffer, 0, internalBufferSize);
uloc_canonicalize(s.locale, buffer, internalBufferSize, status);
ucol_sit_initCollatorSpecs(&s);
string = ucol_sit_readSpecs(&s, definition, parseError, status);
ucol_sit_calculateWholeLocale(&s);
-
+
char buffer[internalBufferSize];
uprv_memset(buffer, 0, internalBufferSize);
uloc_canonicalize(s.locale, buffer, internalBufferSize, status);
}
if(U_FAILURE(*status)) {
- parseError->offset = string - definition;
+ parseError->offset = (int32_t)(string - definition);
ucol_close(result);
return NULL;
}
}
-static void appendShortStringElement(const char *src, int32_t len, char *result, int32_t *resultSize, int32_t capacity, char arg)
-{
- if(len) {
- if(*resultSize) {
- if(*resultSize < capacity) {
- uprv_strcat(result, "_");
- }
- (*resultSize)++;
- }
- *resultSize += len + 1;
- if(*resultSize < capacity) {
- uprv_strncat(result, &arg, 1);
- uprv_strncat(result, src, len);
- }
- }
-}
-
U_CAPI int32_t U_EXPORT2
ucol_getShortDefinitionString(const UCollator *coll,
const char *locale,
UErrorCode *status)
{
if(U_FAILURE(*status)) return 0;
- char buffer[internalBufferSize];
- uprv_memset(buffer, 0, internalBufferSize*sizeof(char));
- int32_t resultSize = 0;
- char tempbuff[internalBufferSize];
- char locBuff[internalBufferSize];
- uprv_memset(buffer, 0, internalBufferSize*sizeof(char));
- int32_t elementSize = 0;
- UBool isAvailable = 0;
- CollatorSpec s;
- ucol_sit_initCollatorSpecs(&s);
-
- if(!locale) {
- locale = ucol_getLocale(coll, ULOC_VALID_LOCALE, status);
- }
- elementSize = ucol_getFunctionalEquivalent(locBuff, internalBufferSize, "collation", locale, &isAvailable, status);
-
- if(elementSize) {
- // we should probably canonicalize here...
- elementSize = uloc_getLanguage(locBuff, tempbuff, internalBufferSize, status);
- appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, capacity, languageArg);
- elementSize = uloc_getCountry(locBuff, tempbuff, internalBufferSize, status);
- appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, capacity, regionArg);
- elementSize = uloc_getScript(locBuff, tempbuff, internalBufferSize, status);
- appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, capacity, scriptArg);
- elementSize = uloc_getVariant(locBuff, tempbuff, internalBufferSize, status);
- appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, capacity, variantArg);
- elementSize = uloc_getKeywordValue(locBuff, "collation", tempbuff, internalBufferSize, status);
- appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, capacity, keywordArg);
- }
-
- int32_t i = 0;
- UColAttributeValue attribute = UCOL_DEFAULT;
- for(i = 0; i < UCOL_SIT_ITEMS_COUNT; i++) {
- if(options[i].action == _processCollatorOption) {
- attribute = ucol_getAttributeOrDefault(coll, (UColAttribute)options[i].attr, status);
- if(attribute != UCOL_DEFAULT) {
- char letter = ucol_sit_attributeValueToLetter(attribute, status);
- appendShortStringElement(&letter, 1,
- buffer, &resultSize, capacity, options[i].optionStart);
- }
- }
- }
- if(coll->variableTopValueisDefault == FALSE) {
- //s.variableTopValue = ucol_getVariableTop(coll, status);
- elementSize = T_CString_integerToString(tempbuff, coll->variableTopValue, 16);
- appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, capacity, variableTopValArg);
+ if(coll == NULL) {
+ *status = U_ILLEGAL_ARGUMENT_ERROR;
+ return 0;
}
-
- UParseError parseError;
- return ucol_normalizeShortDefinitionString(buffer, dst, capacity, &parseError, status);
+ return ((icu::Collator*)coll)->internalGetShortDefinitionString(locale,dst,capacity,*status);
}
U_CAPI int32_t U_EXPORT2
return ucol_sit_dumpSpecs(&s, destination, capacity, status);
}
-// structure for packing the bits of the attributes in the
-// identifier number.
-// locale is packed separately
-struct bitPacking {
- char letter;
- uint32_t offset;
- uint32_t width;
- UColAttribute attribute;
- UColAttributeValue values[6];
-};
-
-static const bitPacking attributesToBits[UCOL_ATTRIBUTE_COUNT] = {
- /* french */ { frenchCollArg, 29, 2, UCOL_FRENCH_COLLATION, { UCOL_DEFAULT, UCOL_OFF, UCOL_ON }},
- /* alternate */ { alternateHArg, 27, 2, UCOL_ALTERNATE_HANDLING, { UCOL_DEFAULT, UCOL_NON_IGNORABLE, UCOL_SHIFTED }},
- /* case first */ { caseFirstArg, 25, 2, UCOL_CASE_FIRST, { UCOL_DEFAULT, UCOL_OFF, UCOL_LOWER_FIRST, UCOL_UPPER_FIRST }},
- /* case level */ { caseLevelArg, 23, 2, UCOL_CASE_LEVEL, { UCOL_DEFAULT, UCOL_OFF, UCOL_ON }},
- /* normalization */ { normArg, 21, 2, UCOL_NORMALIZATION_MODE, { UCOL_DEFAULT, UCOL_OFF, UCOL_ON }},
- /* strength */ { strengthArg, 18, 3, UCOL_STRENGTH, { UCOL_DEFAULT, UCOL_PRIMARY, UCOL_SECONDARY, UCOL_TERTIARY, UCOL_QUATERNARY, UCOL_IDENTICAL }},
- /* hiragana */ { hiraganaQArg, 16, 2, UCOL_HIRAGANA_QUATERNARY_MODE, { UCOL_DEFAULT, UCOL_OFF, UCOL_ON }},
- /* numeric coll */ { numericCollArg, 14, 2, UCOL_NUMERIC_COLLATION, { UCOL_DEFAULT, UCOL_OFF, UCOL_ON }}
-};
-
-static const uint32_t keywordShift = 9;
-static const uint32_t keywordWidth = 5;
-static const uint32_t localeShift = 0;
-static const uint32_t localeWidth = 7;
-
-
-static uint32_t ucol_sit_putLocaleInIdentifier(uint32_t result, const char* locale, UErrorCode* status) {
- char buffer[internalBufferSize], keywordBuffer[internalBufferSize],
- baseName[internalBufferSize], localeBuffer[internalBufferSize];
- int32_t len = 0, keywordLen = 0,
- baseNameLen = 0, localeLen = 0;
- uint32_t i = 0;
- UBool isAvailable = FALSE;
- if(locale) {
- len = uloc_canonicalize(locale, buffer, internalBufferSize, status);
- localeLen = ucol_getFunctionalEquivalent(localeBuffer, internalBufferSize, "collation", buffer, &isAvailable, status);
- keywordLen = uloc_getKeywordValue(buffer, "collation", keywordBuffer, internalBufferSize, status);
- baseNameLen = uloc_getBaseName(buffer, baseName, internalBufferSize, status);
-
- /*Binary search for the map entry for normal cases */
-
- uint32_t low = 0;
- uint32_t high = sizeof(locales)/sizeof(locales[0]);
- uint32_t mid = high;
- uint32_t oldmid = 0;
- int32_t compVal = 0;
-
-
- while (high > low) /*binary search*/{
-
- mid = (high+low) >> 1; /*Finds median*/
-
- if (mid == oldmid)
- return UCOL_SIT_COLLATOR_NOT_ENCODABLE; // we didn't find it
-
- compVal = uprv_strcmp(baseName, locales[mid]);
- if (compVal < 0){
- high = mid;
- }
- else if (compVal > 0){
- low = mid;
- }
- else /*we found it*/{
- break;
- }
- oldmid = mid;
- }
-
- result |= (mid & ((1 << localeWidth) - 1)) << localeShift;
- }
-
- if(keywordLen) {
- for(i = 1; i < sizeof(keywords)/sizeof(keywords[0]); i++) {
- if(uprv_strcmp(keywords[i], keywordBuffer) == 0) {
- result |= (i & ((1 << keywordWidth) - 1)) << keywordShift;
- break;
- }
- }
- }
- return result;
-}
-
-U_CAPI uint32_t U_EXPORT2
-ucol_collatorToIdentifier(const UCollator *coll,
- const char *locale,
- UErrorCode *status)
-{
- uint32_t result = 0;
- uint32_t i = 0, j = 0;
- UColAttributeValue attrValue = UCOL_DEFAULT;
-
- // if variable top is not default, we need to use strings
- if(coll->variableTopValueisDefault != TRUE) {
- return UCOL_SIT_COLLATOR_NOT_ENCODABLE;
- }
-
- if(locale == NULL) {
- locale = ucol_getLocale(coll, ULOC_VALID_LOCALE, status);
- }
-
- result = ucol_sit_putLocaleInIdentifier(result, locale, status);
-
- for(i = 0; i < sizeof(attributesToBits)/sizeof(attributesToBits[0]); i++) {
- attrValue = ucol_getAttributeOrDefault(coll, attributesToBits[i].attribute, status);
- j = 0;
- while(attributesToBits[i].values[j] != attrValue) {
- j++;
- }
- result |= (j & ((1 << attributesToBits[i].width) - 1)) << attributesToBits[i].offset;
- }
-
- return result;
-}
-
-U_CAPI UCollator* U_EXPORT2
-ucol_openFromIdentifier(uint32_t identifier,
- UBool forceDefaults,
- UErrorCode *status)
-{
- uint32_t i = 0;
- int32_t value = 0, keyword = 0;
- char locale[internalBufferSize];
-
- value = (identifier >> localeShift) & ((1 << localeWidth) - 1);
- keyword = (identifier >> keywordShift) & ((1 << keywordWidth) - 1);
-
- uprv_strcpy(locale, locales[value]);
-
- if(keyword) {
- uprv_strcat(locale, collationKeyword);
- uprv_strcat(locale, keywords[keyword]);
- }
-
- UColAttributeValue attrValue = UCOL_DEFAULT;
-
- UCollator *result = ucol_open(locale, status);
-
- // variable top is not set in the identifier, so we can easily skip that on
-
- for(i = 0; i < sizeof(attributesToBits)/sizeof(attributesToBits[0]); i++) {
- value = (identifier >> attributesToBits[i].offset) & ((1 << attributesToBits[i].width) - 1);
- attrValue = attributesToBits[i].values[value];
- // the collator is all default, so we will set only the values that will differ from
- // the default values.
- if(attrValue != UCOL_DEFAULT) {
- if(forceDefaults ||
- ucol_getAttribute(result, attributesToBits[i].attribute, status) != attrValue) {
- ucol_setAttribute(result, attributesToBits[i].attribute, attrValue, status);
- }
- }
- }
-
- return result;
-}
-
-U_CAPI int32_t U_EXPORT2
-ucol_identifierToShortString(uint32_t identifier,
- char *buffer,
- int32_t capacity,
- UBool forceDefaults,
- UErrorCode *status)
-{
- int32_t locIndex = (identifier >> localeShift) & ((1 << localeWidth) - 1);
- int32_t keywordIndex = (identifier >> keywordShift) & ((1 << keywordWidth) - 1);
- CollatorSpec s;
- ucol_sit_initCollatorSpecs(&s);
- uprv_strcpy(s.locale, locales[locIndex]);
- if(keywordIndex) {
- uprv_strcat(s.locale, collationKeyword);
- uprv_strcat(s.locale, keywords[keywordIndex]);
- }
- UCollator *coll = ucol_openFromIdentifier(identifier, forceDefaults, status);
- int32_t resultLen = ucol_getShortDefinitionString(coll, s.locale, buffer, capacity, status);
- ucol_close(coll);
- return resultLen;
-
-#if 0
- // TODO: Crumy, crumy, crumy... Very hard to currently go algorithmically from
- // identifier to short string. Do rethink
- if(forceDefaults == FALSE) {
- UCollator *coll = ucol_openFromIdentifier(identifier, FALSE, status);
- int32_t resultLen = ucol_getShortDefinitionString(coll, s.locale, buffer, capacity, status);
- ucol_close(coll);
- return resultLen;
- } else { // forceDefaults == TRUE
- char letter;
- UColAttributeValue value;
- int32_t i = 0;
- for(i = 0; i < sizeof(attributesToBits)/sizeof(attributesToBits[0]); i++) {
- value = attributesToBits[i].values[(identifier >> attributesToBits[i].offset) & ((1 << attributesToBits[i].width) - 1)];
- if(value != UCOL_DEFAULT) {
- uprv_strcat(buffer, "_");
- uprv_strncat(buffer, &attributesToBits[i].letter, 1);
- letter = ucol_sit_attributeValueToLetter(value, status);
- uprv_strncat(buffer, &letter, 1);
- }
- }
- return ucol_sit_dumpSpecs(&s, buffer, capacity, status);
- }
-#endif
-}
-
-U_CAPI uint32_t U_EXPORT2
-ucol_shortStringToIdentifier(const char *definition,
- UBool forceDefaults,
- UErrorCode *status)
-{
- UParseError parseError;
- CollatorSpec s;
- uint32_t result = 0;
- uint32_t i = 0, j = 0;
- ucol_sit_initCollatorSpecs(&s);
-
- ucol_sit_readSpecs(&s, definition, &parseError, status);
- ucol_sit_calculateWholeLocale(&s);
-
- char locBuffer[internalBufferSize];
- UBool isAvailable = FALSE;
- UColAttributeValue attrValue = UCOL_DEFAULT;
-
- ucol_getFunctionalEquivalent(locBuffer, internalBufferSize, "collation", s.locale, &isAvailable, status);
-
- if(forceDefaults == FALSE) {
- UCollator *coll = ucol_openFromShortString(definition, FALSE, &parseError, status);
- result = ucol_collatorToIdentifier(coll, locBuffer, status);
- ucol_close(coll);
- } else { // forceDefaults == TRUE
- result = ucol_sit_putLocaleInIdentifier(result, locBuffer, status);
-
- for(i = 0; i < sizeof(attributesToBits)/sizeof(attributesToBits[0]); i++) {
- attrValue = s.options[i];
- j = 0;
- while(attributesToBits[i].values[j] != attrValue) {
- j++;
- }
- result |= (j & ((1 << attributesToBits[i].width) - 1)) << attributesToBits[i].offset;
- }
-
- }
- return result;
-
-}
-
-U_CAPI UColAttributeValue U_EXPORT2
-ucol_getAttributeOrDefault(const UCollator *coll, UColAttribute attr, UErrorCode *status)
-{
- if(U_FAILURE(*status) || coll == NULL) {
- return UCOL_DEFAULT;
- }
- switch(attr) {
- case UCOL_NUMERIC_COLLATION:
- return coll->numericCollationisDefault?UCOL_DEFAULT:coll->numericCollation;
- case UCOL_HIRAGANA_QUATERNARY_MODE:
- return coll->hiraganaQisDefault?UCOL_DEFAULT:coll->hiraganaQ;
- case UCOL_FRENCH_COLLATION: /* attribute for direction of secondary weights*/
- return coll->frenchCollationisDefault?UCOL_DEFAULT:coll->frenchCollation;
- case UCOL_ALTERNATE_HANDLING: /* attribute for handling variable elements*/
- return coll->alternateHandlingisDefault?UCOL_DEFAULT:coll->alternateHandling;
- case UCOL_CASE_FIRST: /* who goes first, lower case or uppercase */
- return coll->caseFirstisDefault?UCOL_DEFAULT:coll->caseFirst;
- case UCOL_CASE_LEVEL: /* do we have an extra case level */
- return coll->caseLevelisDefault?UCOL_DEFAULT:coll->caseLevel;
- case UCOL_NORMALIZATION_MODE: /* attribute for normalization */
- return coll->normalizationModeisDefault?UCOL_DEFAULT:coll->normalizationMode;
- case UCOL_STRENGTH: /* attribute for strength */
- return coll->strengthisDefault?UCOL_DEFAULT:coll->strength;
- case UCOL_ATTRIBUTE_COUNT:
- default:
- *status = U_ILLEGAL_ARGUMENT_ERROR;
- break;
- }
- return UCOL_DEFAULT;
-}
-
-
-struct contContext {
- const UCollator *coll;
- USet *conts;
- USet *expansions;
- USet *removedContractions;
- UBool addPrefixes;
- UErrorCode *status;
-};
-
-
-
-static void
-addSpecial(contContext *context, UChar *buffer, int32_t bufLen,
- uint32_t CE, int32_t leftIndex, int32_t rightIndex, UErrorCode *status)
-{
- const UCollator *coll = context->coll;
- USet *contractions = context->conts;
- USet *expansions = context->expansions;
- UBool addPrefixes = context->addPrefixes;
-
- const UChar *UCharOffset = (UChar *)coll->image+getContractOffset(CE);
- uint32_t newCE = *(coll->contractionCEs + (UCharOffset - coll->contractionIndex));
- // we might have a contraction that ends from previous level
- if(newCE != UCOL_NOT_FOUND) {
- if(isSpecial(CE) && getCETag(CE) == CONTRACTION_TAG && isSpecial(newCE) && getCETag(newCE) == SPEC_PROC_TAG && addPrefixes) {
- addSpecial(context, buffer, bufLen, newCE, leftIndex, rightIndex, status);
- }
- if(contractions && rightIndex-leftIndex > 1) {
- uset_addString(contractions, buffer+leftIndex, rightIndex-leftIndex);
- if(expansions && isSpecial(CE) && getCETag(CE) == EXPANSION_TAG) {
- uset_addString(expansions, buffer+leftIndex, rightIndex-leftIndex);
- }
- }
- }
-
- UCharOffset++;
- // check whether we're doing contraction or prefix
- if(getCETag(CE) == SPEC_PROC_TAG && addPrefixes) {
- if(leftIndex == 0) {
- *status = U_INTERNAL_PROGRAM_ERROR;
- return;
- }
- --leftIndex;
- while(*UCharOffset != 0xFFFF) {
- newCE = *(coll->contractionCEs + (UCharOffset - coll->contractionIndex));
- buffer[leftIndex] = *UCharOffset;
- if(isSpecial(newCE) && (getCETag(newCE) == CONTRACTION_TAG || getCETag(newCE) == SPEC_PROC_TAG)) {
- addSpecial(context, buffer, bufLen, newCE, leftIndex, rightIndex, status);
- } else {
- if(contractions) {
- uset_addString(contractions, buffer+leftIndex, rightIndex-leftIndex);
- }
- if(expansions && isSpecial(newCE) && getCETag(newCE) == EXPANSION_TAG) {
- uset_addString(expansions, buffer+leftIndex, rightIndex-leftIndex);
- }
- }
- UCharOffset++;
- }
- } else if(getCETag(CE) == CONTRACTION_TAG) {
- if(rightIndex == bufLen-1) {
- *status = U_INTERNAL_PROGRAM_ERROR;
- return;
- }
- while(*UCharOffset != 0xFFFF) {
- newCE = *(coll->contractionCEs + (UCharOffset - coll->contractionIndex));
- buffer[rightIndex] = *UCharOffset;
- if(isSpecial(newCE) && (getCETag(newCE) == CONTRACTION_TAG || getCETag(newCE) == SPEC_PROC_TAG)) {
- addSpecial(context, buffer, bufLen, newCE, leftIndex, rightIndex+1, status);
- } else {
- if(contractions) {
- uset_addString(contractions, buffer+leftIndex, rightIndex+1-leftIndex);
- }
- if(expansions && isSpecial(newCE) && getCETag(newCE) == EXPANSION_TAG) {
- uset_addString(expansions, buffer+leftIndex, rightIndex+1-leftIndex);
- }
- }
- UCharOffset++;
- }
- }
-
-}
-
-U_CDECL_BEGIN
-static UBool U_CALLCONV
-_processSpecials(const void *context, UChar32 start, UChar32 limit, uint32_t CE)
-{
- UErrorCode *status = ((contContext *)context)->status;
- USet *expansions = ((contContext *)context)->expansions;
- USet *removed = ((contContext *)context)->removedContractions;
- UBool addPrefixes = ((contContext *)context)->addPrefixes;
- UChar contraction[internalBufferSize];
- if(isSpecial(CE)) {
- if(((getCETag(CE) == SPEC_PROC_TAG && addPrefixes) || getCETag(CE) == CONTRACTION_TAG)) {
- while(start < limit && U_SUCCESS(*status)) {
- // if there are suppressed contractions, we don't
- // want to add them.
- if(removed && uset_contains(removed, start)) {
- start++;
- continue;
- }
- // we start our contraction from middle, since we don't know if it
- // will grow toward right or left
- contraction[internalBufferSize/2] = (UChar)start;
- addSpecial(((contContext *)context), contraction, internalBufferSize, CE, internalBufferSize/2, internalBufferSize/2+1, status);
- start++;
- }
- } else if(expansions && getCETag(CE) == EXPANSION_TAG) {
- while(start < limit && U_SUCCESS(*status)) {
- uset_add(expansions, start++);
- }
- }
- }
- if(U_FAILURE(*status)) {
- return FALSE;
- } else {
- return TRUE;
- }
-}
-
-U_CDECL_END
-
-
-
/**
* 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 conts the set to hold the result
* @param status to hold the error code
* @return the size of the contraction set
- *
- * @draft ICU 3.0
*/
U_CAPI int32_t U_EXPORT2
ucol_getContractions( const UCollator *coll,
*status = U_ILLEGAL_ARGUMENT_ERROR;
return;
}
-
- if(contractions) {
- uset_clear(contractions);
- }
- if(expansions) {
- uset_clear(expansions);
- }
- int32_t rulesLen = 0;
- const UChar* rules = ucol_getRules(coll, &rulesLen);
- UColTokenParser src;
- ucol_tok_initTokenList(&src, rules, rulesLen, coll->UCA, status);
-
- contContext c = { NULL, contractions, expansions, src.removeSet, addPrefixes, status };
-
- // Add the UCA contractions
- c.coll = coll->UCA;
- utrie_enum(&coll->UCA->mapping, NULL, _processSpecials, &c);
-
- // This is collator specific. Add contractions from a collator
- c.coll = coll;
- c.removedContractions = NULL;
- utrie_enum(&coll->mapping, NULL, _processSpecials, &c);
- ucol_tok_closeTokenList(&src);
-}
-
-U_CAPI int32_t U_EXPORT2
-ucol_getUnsafeSet( const UCollator *coll,
- USet *unsafe,
- UErrorCode *status)
-{
- UChar buffer[internalBufferSize];
- int32_t len = 0;
-
- uset_clear(unsafe);
-
- // cccpattern = "[[:^tccc=0:][:^lccc=0:]]", unfortunately variant
- static const UChar cccpattern[25] = { 0x5b, 0x5b, 0x3a, 0x5e, 0x74, 0x63, 0x63, 0x63, 0x3d, 0x30, 0x3a, 0x5d,
- 0x5b, 0x3a, 0x5e, 0x6c, 0x63, 0x63, 0x63, 0x3d, 0x30, 0x3a, 0x5d, 0x5d, 0x00 };
-
- // add chars that fail the fcd check
- uset_applyPattern(unsafe, cccpattern, 24, USET_IGNORE_SPACE, status);
-
- // add Thai/Lao prevowels
- uset_addRange(unsafe, 0xe40, 0xe44);
- uset_addRange(unsafe, 0xec0, 0xec4);
- // add lead/trail surrogates
- uset_addRange(unsafe, 0xd800, 0xdfff);
-
- USet *contractions = uset_open(0,0);
-
- int32_t i = 0, j = 0;
- int32_t contsSize = ucol_getContractions(coll, contractions, status);
- UChar32 c = 0;
- // Contraction set consists only of strings
- // to get unsafe code points, we need to
- // break the strings apart and add them to the unsafe set
- for(i = 0; i < contsSize; i++) {
- len = uset_getItem(contractions, i, NULL, NULL, buffer, internalBufferSize, status);
- if(len > 0) {
- j = 0;
- while(j < len) {
- U16_NEXT(buffer, j, len, c);
- if(j < len) {
- uset_add(unsafe, c);
- }
- }
- }
+ const icu::RuleBasedCollator *rbc = icu::RuleBasedCollator::rbcFromUCollator(coll);
+ if(rbc == NULL) {
+ *status = U_UNSUPPORTED_ERROR;
+ return;
}
-
- uset_close(contractions);
-
- return uset_size(unsafe);
+ rbc->internalGetContractionsAndExpansions(
+ icu::UnicodeSet::fromUSet(contractions),
+ icu::UnicodeSet::fromUSet(expansions),
+ addPrefixes, *status);
}
#endif
projects / apple / icu.git / blobdiff