michael@0: /*
michael@0: *******************************************************************************
michael@0: *   Copyright (C) 2004-2012, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: *******************************************************************************
michael@0: *   file name:  ucol_sit.cpp
michael@0: *   encoding:   US-ASCII
michael@0: *   tab size:   8 (not used)
michael@0: *   indentation:4
michael@0: *
michael@0: * Modification history
michael@0: * Date        Name      Comments
michael@0: * 03/12/2004  weiv      Creation
michael@0: */
michael@0: 
michael@0: #include "unicode/ustring.h"
michael@0: #include "unicode/udata.h"
michael@0: 
michael@0: #include "utracimp.h"
michael@0: #include "ucol_imp.h"
michael@0: #include "ucol_tok.h"
michael@0: #include "cmemory.h"
michael@0: #include "cstring.h"
michael@0: #include "uresimp.h"
michael@0: #include "unicode/coll.h"
michael@0: 
michael@0: #ifdef UCOL_TRACE_SIT
michael@0: # include <stdio.h>
michael@0: #endif
michael@0: 
michael@0: #if !UCONFIG_NO_COLLATION
michael@0: 
michael@0: enum OptionsList {
michael@0:     UCOL_SIT_LANGUAGE = 0,
michael@0:     UCOL_SIT_SCRIPT   = 1,
michael@0:     UCOL_SIT_REGION   = 2,
michael@0:     UCOL_SIT_VARIANT  = 3,
michael@0:     UCOL_SIT_KEYWORD  = 4,
michael@0:     UCOL_SIT_PROVIDER = 5,
michael@0:     UCOL_SIT_LOCELEMENT_MAX = UCOL_SIT_PROVIDER, /* the last element that's part of LocElements */
michael@0: 
michael@0:     UCOL_SIT_BCP47,
michael@0:     UCOL_SIT_STRENGTH,
michael@0:     UCOL_SIT_CASE_LEVEL,
michael@0:     UCOL_SIT_CASE_FIRST,
michael@0:     UCOL_SIT_NUMERIC_COLLATION,
michael@0:     UCOL_SIT_ALTERNATE_HANDLING,
michael@0:     UCOL_SIT_NORMALIZATION_MODE,
michael@0:     UCOL_SIT_FRENCH_COLLATION,
michael@0:     UCOL_SIT_HIRAGANA_QUATERNARY,
michael@0:     UCOL_SIT_VARIABLE_TOP,
michael@0:     UCOL_SIT_VARIABLE_TOP_VALUE,
michael@0:     UCOL_SIT_ITEMS_COUNT
michael@0: };
michael@0: 
michael@0: /* option starters chars. */
michael@0: static const char alternateHArg     = 'A';
michael@0: static const char variableTopValArg = 'B';
michael@0: static const char caseFirstArg      = 'C';
michael@0: static const char numericCollArg    = 'D';
michael@0: static const char caseLevelArg      = 'E';
michael@0: static const char frenchCollArg     = 'F';
michael@0: static const char hiraganaQArg      = 'H';
michael@0: static const char keywordArg        = 'K';
michael@0: static const char languageArg       = 'L';
michael@0: static const char normArg           = 'N';
michael@0: static const char providerArg       = 'P';
michael@0: static const char regionArg         = 'R';
michael@0: static const char strengthArg       = 'S';
michael@0: static const char variableTopArg    = 'T';
michael@0: static const char variantArg        = 'V';
michael@0: static const char RFC3066Arg        = 'X';
michael@0: static const char scriptArg         = 'Z';
michael@0: 
michael@0: static const char collationKeyword[]  = "@collation=";
michael@0: static const char providerKeyword[]  = "@sp=";
michael@0: 
michael@0: 
michael@0: static const int32_t locElementCount = UCOL_SIT_LOCELEMENT_MAX+1;
michael@0: static const int32_t locElementCapacity = 32;
michael@0: static const int32_t loc3066Capacity = 256;
michael@0: static const int32_t locProviderCapacity = 10;
michael@0: static const int32_t internalBufferSize = 512;
michael@0: 
michael@0: /* structure containing specification of a collator. Initialized
michael@0:  * from a short string. Also used to construct a short string from a
michael@0:  * collator instance
michael@0:  */
michael@0: struct CollatorSpec {
michael@0:     char locElements[locElementCount][locElementCapacity];
michael@0:     char locale[loc3066Capacity];
michael@0:     char provider[locProviderCapacity];
michael@0:     UColAttributeValue options[UCOL_ATTRIBUTE_COUNT];
michael@0:     uint32_t variableTopValue;
michael@0:     UChar variableTopString[locElementCapacity];
michael@0:     int32_t variableTopStringLen;
michael@0:     UBool variableTopSet;
michael@0:     struct {
michael@0:         const char *start;
michael@0:         int32_t len;
michael@0:     } entries[UCOL_SIT_ITEMS_COUNT];
michael@0: };
michael@0: 
michael@0: 
michael@0: /* structure for converting between character attribute
michael@0:  * representation and real collation attribute value.
michael@0:  */
michael@0: struct AttributeConversion {
michael@0:     char letter;
michael@0:     UColAttributeValue value;
michael@0: };
michael@0: 
michael@0: static const AttributeConversion conversions[12] = {
michael@0:     { '1', UCOL_PRIMARY },
michael@0:     { '2', UCOL_SECONDARY },
michael@0:     { '3', UCOL_TERTIARY },
michael@0:     { '4', UCOL_QUATERNARY },
michael@0:     { 'D', UCOL_DEFAULT },
michael@0:     { 'I', UCOL_IDENTICAL },
michael@0:     { 'L', UCOL_LOWER_FIRST },
michael@0:     { 'N', UCOL_NON_IGNORABLE },
michael@0:     { 'O', UCOL_ON },
michael@0:     { 'S', UCOL_SHIFTED },
michael@0:     { 'U', UCOL_UPPER_FIRST },
michael@0:     { 'X', UCOL_OFF }
michael@0: };
michael@0: 
michael@0: 
michael@0: static char
michael@0: ucol_sit_attributeValueToLetter(UColAttributeValue value, UErrorCode *status) {
michael@0:     uint32_t i = 0;
michael@0:     for(i = 0; i < sizeof(conversions)/sizeof(conversions[0]); i++) {
michael@0:         if(conversions[i].value == value) {
michael@0:             return conversions[i].letter;
michael@0:         }
michael@0:     }
michael@0:     *status = U_ILLEGAL_ARGUMENT_ERROR;
michael@0: #ifdef UCOL_TRACE_SIT
michael@0:     fprintf(stderr, "%s:%d: unknown UColAttributeValue %d: %s\n", __FILE__, __LINE__, value, u_errorName(*status));
michael@0: #endif    
michael@0:     return 0;
michael@0: }
michael@0: 
michael@0: static UColAttributeValue
michael@0: ucol_sit_letterToAttributeValue(char letter, UErrorCode *status) {
michael@0:     uint32_t i = 0;
michael@0:     for(i = 0; i < sizeof(conversions)/sizeof(conversions[0]); i++) {
michael@0:         if(conversions[i].letter == letter) {
michael@0:             return conversions[i].value;
michael@0:         }
michael@0:     }
michael@0:     *status = U_ILLEGAL_ARGUMENT_ERROR;
michael@0: #ifdef UCOL_TRACE_SIT
michael@0:     fprintf(stderr, "%s:%d: unknown letter %c: %s\n", __FILE__, __LINE__, letter, u_errorName(*status));
michael@0: #endif    
michael@0:     return UCOL_DEFAULT;
michael@0: }
michael@0: 
michael@0: /* function prototype for functions used to parse a short string */
michael@0: U_CDECL_BEGIN
michael@0: typedef const char* U_CALLCONV
michael@0: ActionFunction(CollatorSpec *spec, uint32_t value1, const char* string,
michael@0:                UErrorCode *status);
michael@0: U_CDECL_END
michael@0: 
michael@0: U_CDECL_BEGIN
michael@0: static const char* U_CALLCONV
michael@0: _processLocaleElement(CollatorSpec *spec, uint32_t value, const char* string,
michael@0:                       UErrorCode *status)
michael@0: {
michael@0:     int32_t len = 0;
michael@0:     do {
michael@0:         if(value == UCOL_SIT_LANGUAGE || value == UCOL_SIT_KEYWORD || value == UCOL_SIT_PROVIDER) {
michael@0:             spec->locElements[value][len++] = uprv_tolower(*string);
michael@0:         } else {
michael@0:             spec->locElements[value][len++] = *string;
michael@0:         }
michael@0:     } while(*(++string) != '_' && *string && len < locElementCapacity);
michael@0:     if(len >= locElementCapacity) {
michael@0:         *status = U_BUFFER_OVERFLOW_ERROR;
michael@0:         return string;
michael@0:     }
michael@0:     // don't skip the underscore at the end
michael@0:     return string;
michael@0: }
michael@0: U_CDECL_END
michael@0: 
michael@0: U_CDECL_BEGIN
michael@0: static const char* U_CALLCONV
michael@0: _processRFC3066Locale(CollatorSpec *spec, uint32_t, const char* string,
michael@0:                       UErrorCode *status)
michael@0: {
michael@0:     char terminator = *string;
michael@0:     string++;
michael@0:     const char *end = uprv_strchr(string+1, terminator);
michael@0:     if(end == NULL || end - string >= loc3066Capacity) {
michael@0:         *status = U_BUFFER_OVERFLOW_ERROR;
michael@0:         return string;
michael@0:     } else {
michael@0:         uprv_strncpy(spec->locale, string, end-string);
michael@0:         return end+1;
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CDECL_END
michael@0: 
michael@0: U_CDECL_BEGIN
michael@0: static const char* U_CALLCONV
michael@0: _processCollatorOption(CollatorSpec *spec, uint32_t option, const char* string,
michael@0:                        UErrorCode *status)
michael@0: {
michael@0:     spec->options[option] = ucol_sit_letterToAttributeValue(*string, status);
michael@0:     if((*(++string) != '_' && *string) || U_FAILURE(*status)) {
michael@0: #ifdef UCOL_TRACE_SIT
michael@0:     fprintf(stderr, "%s:%d: unknown collator option at '%s': %s\n", __FILE__, __LINE__, string, u_errorName(*status));
michael@0: #endif    
michael@0:         *status = U_ILLEGAL_ARGUMENT_ERROR;
michael@0:     }
michael@0:     return string;
michael@0: }
michael@0: U_CDECL_END
michael@0: 
michael@0: 
michael@0: static UChar
michael@0: readHexCodeUnit(const char **string, UErrorCode *status)
michael@0: {
michael@0:     UChar result = 0;
michael@0:     int32_t value = 0;
michael@0:     char c;
michael@0:     int32_t noDigits = 0;
michael@0:     while((c = **string) != 0 && noDigits < 4) {
michael@0:         if( c >= '0' && c <= '9') {
michael@0:             value = c - '0';
michael@0:         } else if ( c >= 'a' && c <= 'f') {
michael@0:             value = c - 'a' + 10;
michael@0:         } else if ( c >= 'A' && c <= 'F') {
michael@0:             value = c - 'A' + 10;
michael@0:         } else {
michael@0:             *status = U_ILLEGAL_ARGUMENT_ERROR;
michael@0: #ifdef UCOL_TRACE_SIT
michael@0:             fprintf(stderr, "%s:%d: Bad hex char at '%s': %s\n", __FILE__, __LINE__, *string, u_errorName(*status));
michael@0: #endif    
michael@0:             return 0;
michael@0:         }
michael@0:         result = (result << 4) | (UChar)value;
michael@0:         noDigits++;
michael@0:         (*string)++;
michael@0:     }
michael@0:     // if the string was terminated before we read 4 digits, set an error
michael@0:     if(noDigits < 4) {
michael@0:         *status = U_ILLEGAL_ARGUMENT_ERROR;
michael@0: #ifdef UCOL_TRACE_SIT
michael@0:         fprintf(stderr, "%s:%d: Short (only %d digits, wanted 4) at '%s': %s\n", __FILE__, __LINE__, noDigits,*string, u_errorName(*status));
michael@0: #endif    
michael@0:     }
michael@0:     return result;
michael@0: }
michael@0: 
michael@0: U_CDECL_BEGIN
michael@0: static const char* U_CALLCONV
michael@0: _processVariableTop(CollatorSpec *spec, uint32_t value1, const char* string, UErrorCode *status)
michael@0: {
michael@0:     // get four digits
michael@0:     int32_t i = 0;
michael@0:     if(!value1) {
michael@0:         while(U_SUCCESS(*status) && i < locElementCapacity && *string != 0 && *string != '_') {
michael@0:             spec->variableTopString[i++] = readHexCodeUnit(&string, status);
michael@0:         }
michael@0:         spec->variableTopStringLen = i;
michael@0:         if(i == locElementCapacity && *string != 0 && *string != '_') {
michael@0:             *status = U_BUFFER_OVERFLOW_ERROR;
michael@0:         }
michael@0:     } else {
michael@0:         spec->variableTopValue = readHexCodeUnit(&string, status);
michael@0:     }
michael@0:     if(U_SUCCESS(*status)) {
michael@0:         spec->variableTopSet = TRUE;
michael@0:     }
michael@0:     return string;
michael@0: }
michael@0: U_CDECL_END
michael@0: 
michael@0: 
michael@0: /* Table for parsing short strings */
michael@0: struct ShortStringOptions {
michael@0:     char optionStart;
michael@0:     ActionFunction *action;
michael@0:     uint32_t attr;
michael@0: };
michael@0: 
michael@0: static const ShortStringOptions options[UCOL_SIT_ITEMS_COUNT] =
michael@0: {
michael@0: /* 10 ALTERNATE_HANDLING */   {alternateHArg,     _processCollatorOption, UCOL_ALTERNATE_HANDLING }, // alternate  N, S, D
michael@0: /* 15 VARIABLE_TOP_VALUE */   {variableTopValArg, _processVariableTop,    1 },
michael@0: /* 08 CASE_FIRST */           {caseFirstArg,      _processCollatorOption, UCOL_CASE_FIRST }, // case first L, U, X, D
michael@0: /* 09 NUMERIC_COLLATION */    {numericCollArg,    _processCollatorOption, UCOL_NUMERIC_COLLATION }, // codan      O, X, D
michael@0: /* 07 CASE_LEVEL */           {caseLevelArg,      _processCollatorOption, UCOL_CASE_LEVEL }, // case level O, X, D
michael@0: /* 12 FRENCH_COLLATION */     {frenchCollArg,     _processCollatorOption, UCOL_FRENCH_COLLATION }, // french     O, X, D
michael@0: /* 13 HIRAGANA_QUATERNARY] */ {hiraganaQArg,      _processCollatorOption, UCOL_HIRAGANA_QUATERNARY_MODE }, // hiragana   O, X, D
michael@0: /* 04 KEYWORD */              {keywordArg,        _processLocaleElement,  UCOL_SIT_KEYWORD }, // keyword
michael@0: /* 00 LANGUAGE */             {languageArg,       _processLocaleElement,  UCOL_SIT_LANGUAGE }, // language
michael@0: /* 11 NORMALIZATION_MODE */   {normArg,           _processCollatorOption, UCOL_NORMALIZATION_MODE }, // norm       O, X, D
michael@0: /* 02 REGION */               {regionArg,         _processLocaleElement,  UCOL_SIT_REGION }, // region
michael@0: /* 06 STRENGTH */             {strengthArg,       _processCollatorOption, UCOL_STRENGTH }, // strength   1, 2, 3, 4, I, D
michael@0: /* 14 VARIABLE_TOP */         {variableTopArg,    _processVariableTop,    0 },
michael@0: /* 03 VARIANT */              {variantArg,        _processLocaleElement,  UCOL_SIT_VARIANT }, // variant
michael@0: /* 05 RFC3066BIS */           {RFC3066Arg,        _processRFC3066Locale,  0 }, // rfc3066bis locale name
michael@0: /* 01 SCRIPT */               {scriptArg,         _processLocaleElement,  UCOL_SIT_SCRIPT },  // script
michael@0: /*    PROVIDER */             {providerArg,       _processLocaleElement, UCOL_SIT_PROVIDER }
michael@0: };
michael@0: 
michael@0: 
michael@0: static
michael@0: const char* ucol_sit_readOption(const char *start, CollatorSpec *spec,
michael@0:                             UErrorCode *status)
michael@0: {
michael@0:   int32_t i = 0;
michael@0: 
michael@0:   for(i = 0; i < UCOL_SIT_ITEMS_COUNT; i++) {
michael@0:       if(*start == options[i].optionStart) {
michael@0:           spec->entries[i].start = start;
michael@0:           const char* end = options[i].action(spec, options[i].attr, start+1, status);
michael@0:           spec->entries[i].len = (int32_t)(end - start);
michael@0:           return end;
michael@0:       }
michael@0:   }
michael@0:   *status = U_ILLEGAL_ARGUMENT_ERROR;
michael@0: #ifdef UCOL_TRACE_SIT
michael@0:   fprintf(stderr, "%s:%d: Unknown option at '%s': %s\n", __FILE__, __LINE__, start, u_errorName(*status));
michael@0: #endif
michael@0:   return start;
michael@0: }
michael@0: 
michael@0: static
michael@0: void ucol_sit_initCollatorSpecs(CollatorSpec *spec)
michael@0: {
michael@0:     // reset everything
michael@0:     uprv_memset(spec, 0, sizeof(CollatorSpec));
michael@0:     // set collation options to default
michael@0:     int32_t i = 0;
michael@0:     for(i = 0; i < UCOL_ATTRIBUTE_COUNT; i++) {
michael@0:         spec->options[i] = UCOL_DEFAULT;
michael@0:     }
michael@0: }
michael@0: 
michael@0: static const char*
michael@0: ucol_sit_readSpecs(CollatorSpec *s, const char *string,
michael@0:                         UParseError *parseError, UErrorCode *status)
michael@0: {
michael@0:     const char *definition = string;
michael@0:     while(U_SUCCESS(*status) && *string) {
michael@0:         string = ucol_sit_readOption(string, s, status);
michael@0:         // advance over '_'
michael@0:         while(*string && *string == '_') {
michael@0:             string++;
michael@0:         }
michael@0:     }
michael@0:     if(U_FAILURE(*status)) {
michael@0:         parseError->offset = (int32_t)(string - definition);
michael@0:     }
michael@0:     return string;
michael@0: }
michael@0: 
michael@0: static
michael@0: int32_t ucol_sit_dumpSpecs(CollatorSpec *s, char *destination, int32_t capacity, UErrorCode *status)
michael@0: {
michael@0:     int32_t i = 0, j = 0;
michael@0:     int32_t len = 0;
michael@0:     char optName;
michael@0:     if(U_SUCCESS(*status)) {
michael@0:         for(i = 0; i < UCOL_SIT_ITEMS_COUNT; i++) {
michael@0:             if(s->entries[i].start) {
michael@0:                 if(len) {
michael@0:                     if(len < capacity) {
michael@0:                         uprv_strcat(destination, "_");
michael@0:                     }
michael@0:                     len++;
michael@0:                 }
michael@0:                 optName = *(s->entries[i].start);
michael@0:                 if(optName == languageArg || optName == regionArg || optName == variantArg || optName == keywordArg) {
michael@0:                     for(j = 0; j < s->entries[i].len; j++) {
michael@0:                         if(len + j < capacity) {
michael@0:                             destination[len+j] = uprv_toupper(*(s->entries[i].start+j));
michael@0:                         }
michael@0:                     }
michael@0:                     len += s->entries[i].len;
michael@0:                 } else {
michael@0:                     len += s->entries[i].len;
michael@0:                     if(len < capacity) {
michael@0:                         uprv_strncat(destination,s->entries[i].start, s->entries[i].len);
michael@0:                     }
michael@0:                 }
michael@0:             }
michael@0:         }
michael@0:         return len;
michael@0:     } else {
michael@0:         return 0;
michael@0:     }
michael@0: }
michael@0: 
michael@0: static void
michael@0: ucol_sit_calculateWholeLocale(CollatorSpec *s) {
michael@0:     // put the locale together, unless we have a done
michael@0:     // locale
michael@0:     if(s->locale[0] == 0) {
michael@0:         // first the language
michael@0:         uprv_strcat(s->locale, s->locElements[UCOL_SIT_LANGUAGE]);
michael@0:         // then the script, if present
michael@0:         if(*(s->locElements[UCOL_SIT_SCRIPT])) {
michael@0:             uprv_strcat(s->locale, "_");
michael@0:             uprv_strcat(s->locale, s->locElements[UCOL_SIT_SCRIPT]);
michael@0:         }
michael@0:         // then the region, if present
michael@0:         if(*(s->locElements[UCOL_SIT_REGION])) {
michael@0:             uprv_strcat(s->locale, "_");
michael@0:             uprv_strcat(s->locale, s->locElements[UCOL_SIT_REGION]);
michael@0:         } else if(*(s->locElements[UCOL_SIT_VARIANT])) { // if there is a variant, we need an underscore
michael@0:             uprv_strcat(s->locale, "_");
michael@0:         }
michael@0:         // add variant, if there
michael@0:         if(*(s->locElements[UCOL_SIT_VARIANT])) {
michael@0:             uprv_strcat(s->locale, "_");
michael@0:             uprv_strcat(s->locale, s->locElements[UCOL_SIT_VARIANT]);
michael@0:         }
michael@0: 
michael@0:         // if there is a collation keyword, add that too
michael@0:         if(*(s->locElements[UCOL_SIT_KEYWORD])) {
michael@0:             uprv_strcat(s->locale, collationKeyword);
michael@0:             uprv_strcat(s->locale, s->locElements[UCOL_SIT_KEYWORD]);
michael@0:         }
michael@0: 
michael@0:         // if there is a provider keyword, add that too
michael@0:         if(*(s->locElements[UCOL_SIT_PROVIDER])) {
michael@0:             uprv_strcat(s->locale, providerKeyword);
michael@0:             uprv_strcat(s->locale, s->locElements[UCOL_SIT_PROVIDER]);
michael@0:         }
michael@0:     }
michael@0: }
michael@0: 
michael@0: 
michael@0: U_CAPI void U_EXPORT2
michael@0: ucol_prepareShortStringOpen( const char *definition,
michael@0:                           UBool,
michael@0:                           UParseError *parseError,
michael@0:                           UErrorCode *status)
michael@0: {
michael@0:     if(U_FAILURE(*status)) return;
michael@0: 
michael@0:     UParseError internalParseError;
michael@0: 
michael@0:     if(!parseError) {
michael@0:         parseError = &internalParseError;
michael@0:     }
michael@0:     parseError->line = 0;
michael@0:     parseError->offset = 0;
michael@0:     parseError->preContext[0] = 0;
michael@0:     parseError->postContext[0] = 0;
michael@0: 
michael@0: 
michael@0:     // first we want to pick stuff out of short string.
michael@0:     // we'll end up with an UCA version, locale and a bunch of
michael@0:     // settings
michael@0: 
michael@0:     // analyse the string in order to get everything we need.
michael@0:     CollatorSpec s;
michael@0:     ucol_sit_initCollatorSpecs(&s);
michael@0:     ucol_sit_readSpecs(&s, definition, parseError, status);
michael@0:     ucol_sit_calculateWholeLocale(&s);
michael@0: 
michael@0:     char buffer[internalBufferSize];
michael@0:     uprv_memset(buffer, 0, internalBufferSize);
michael@0:     uloc_canonicalize(s.locale, buffer, internalBufferSize, status);
michael@0: 
michael@0:     UResourceBundle *b = ures_open(U_ICUDATA_COLL, buffer, status);
michael@0:     /* we try to find stuff from keyword */
michael@0:     UResourceBundle *collations = ures_getByKey(b, "collations", NULL, status);
michael@0:     UResourceBundle *collElem = NULL;
michael@0:     char keyBuffer[256];
michael@0:     // if there is a keyword, we pick it up and try to get elements
michael@0:     if(!uloc_getKeywordValue(buffer, "collation", keyBuffer, 256, status)) {
michael@0:       // no keyword. we try to find the default setting, which will give us the keyword value
michael@0:       UResourceBundle *defaultColl = ures_getByKeyWithFallback(collations, "default", NULL, status);
michael@0:       if(U_SUCCESS(*status)) {
michael@0:         int32_t defaultKeyLen = 0;
michael@0:         const UChar *defaultKey = ures_getString(defaultColl, &defaultKeyLen, status);
michael@0:         u_UCharsToChars(defaultKey, keyBuffer, defaultKeyLen);
michael@0:         keyBuffer[defaultKeyLen] = 0;
michael@0:       } else {
michael@0:         *status = U_INTERNAL_PROGRAM_ERROR;
michael@0:         return;
michael@0:       }
michael@0:       ures_close(defaultColl);
michael@0:     }
michael@0:     collElem = ures_getByKeyWithFallback(collations, keyBuffer, collElem, status);
michael@0:     ures_close(collElem);
michael@0:     ures_close(collations);
michael@0:     ures_close(b);
michael@0: }
michael@0: 
michael@0: 
michael@0: U_CAPI UCollator* U_EXPORT2
michael@0: ucol_openFromShortString( const char *definition,
michael@0:                           UBool forceDefaults,
michael@0:                           UParseError *parseError,
michael@0:                           UErrorCode *status)
michael@0: {
michael@0:     UTRACE_ENTRY_OC(UTRACE_UCOL_OPEN_FROM_SHORT_STRING);
michael@0:     UTRACE_DATA1(UTRACE_INFO, "short string = \"%s\"", definition);
michael@0: 
michael@0:     if(U_FAILURE(*status)) return 0;
michael@0: 
michael@0:     UParseError internalParseError;
michael@0: 
michael@0:     if(!parseError) {
michael@0:         parseError = &internalParseError;
michael@0:     }
michael@0:     parseError->line = 0;
michael@0:     parseError->offset = 0;
michael@0:     parseError->preContext[0] = 0;
michael@0:     parseError->postContext[0] = 0;
michael@0: 
michael@0: 
michael@0:     // first we want to pick stuff out of short string.
michael@0:     // we'll end up with an UCA version, locale and a bunch of
michael@0:     // settings
michael@0: 
michael@0:     // analyse the string in order to get everything we need.
michael@0:     const char *string = definition;
michael@0:     CollatorSpec s;
michael@0:     ucol_sit_initCollatorSpecs(&s);
michael@0:     string = ucol_sit_readSpecs(&s, definition, parseError, status);
michael@0:     ucol_sit_calculateWholeLocale(&s);
michael@0: 
michael@0:     char buffer[internalBufferSize];
michael@0:     uprv_memset(buffer, 0, internalBufferSize);
michael@0:     uloc_canonicalize(s.locale, buffer, internalBufferSize, status);
michael@0: 
michael@0:     UCollator *result = ucol_open(buffer, status);
michael@0:     int32_t i = 0;
michael@0: 
michael@0:     for(i = 0; i < UCOL_ATTRIBUTE_COUNT; i++) {
michael@0:         if(s.options[i] != UCOL_DEFAULT) {
michael@0:             if(forceDefaults || ucol_getAttribute(result, (UColAttribute)i, status) != s.options[i]) {
michael@0:                 ucol_setAttribute(result, (UColAttribute)i, s.options[i], status);
michael@0:             }
michael@0: 
michael@0:             if(U_FAILURE(*status)) {
michael@0:                 parseError->offset = (int32_t)(string - definition);
michael@0:                 ucol_close(result);
michael@0:                 return NULL;
michael@0:             }
michael@0: 
michael@0:         }
michael@0:     }
michael@0:     if(s.variableTopSet) {
michael@0:         if(s.variableTopString[0]) {
michael@0:             ucol_setVariableTop(result, s.variableTopString, s.variableTopStringLen, status);
michael@0:         } else { // we set by value, using 'B'
michael@0:             ucol_restoreVariableTop(result, s.variableTopValue, status);
michael@0:         }
michael@0:     }
michael@0: 
michael@0: 
michael@0:     if(U_FAILURE(*status)) { // here it can only be a bogus value
michael@0:         ucol_close(result);
michael@0:         result = NULL;
michael@0:     }
michael@0: 
michael@0:     UTRACE_EXIT_PTR_STATUS(result, *status);
michael@0:     return result;
michael@0: }
michael@0: 
michael@0: 
michael@0: static void appendShortStringElement(const char *src, int32_t len, char *result, int32_t *resultSize, int32_t capacity, char arg)
michael@0: {
michael@0:     if(len) {
michael@0:         if(*resultSize) {
michael@0:             if(*resultSize < capacity) {
michael@0:                 uprv_strcat(result, "_");
michael@0:             }
michael@0:             (*resultSize)++;
michael@0:         }
michael@0:         *resultSize += len + 1;
michael@0:         if(*resultSize < capacity) {
michael@0:             uprv_strncat(result, &arg, 1);
michael@0:             uprv_strncat(result, src, len);
michael@0:         }
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: ucol_getShortDefinitionString(const UCollator *coll,
michael@0:                               const char *locale,
michael@0:                               char *dst,
michael@0:                               int32_t capacity,
michael@0:                               UErrorCode *status)
michael@0: {
michael@0:     if(U_FAILURE(*status)) return 0;
michael@0:     if(coll->delegate != NULL) {
michael@0:       return ((icu::Collator*)coll->delegate)->internalGetShortDefinitionString(locale,dst,capacity,*status);
michael@0:     }
michael@0:     char buffer[internalBufferSize];
michael@0:     uprv_memset(buffer, 0, internalBufferSize*sizeof(char));
michael@0:     int32_t resultSize = 0;
michael@0:     char tempbuff[internalBufferSize];
michael@0:     char locBuff[internalBufferSize];
michael@0:     uprv_memset(buffer, 0, internalBufferSize*sizeof(char));
michael@0:     int32_t elementSize = 0;
michael@0:     UBool isAvailable = 0;
michael@0:     CollatorSpec s;
michael@0:     ucol_sit_initCollatorSpecs(&s);
michael@0: 
michael@0:     if(!locale) {
michael@0:         locale = ucol_getLocaleByType(coll, ULOC_VALID_LOCALE, status);
michael@0:     }
michael@0:     elementSize = ucol_getFunctionalEquivalent(locBuff, internalBufferSize, "collation", locale, &isAvailable, status);
michael@0: 
michael@0:     if(elementSize) {
michael@0:         // we should probably canonicalize here...
michael@0:         elementSize = uloc_getLanguage(locBuff, tempbuff, internalBufferSize, status);
michael@0:         appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, /*capacity*/internalBufferSize, languageArg);
michael@0:         elementSize = uloc_getCountry(locBuff, tempbuff, internalBufferSize, status);
michael@0:         appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, /*capacity*/internalBufferSize, regionArg);
michael@0:         elementSize = uloc_getScript(locBuff, tempbuff, internalBufferSize, status);
michael@0:         appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, /*capacity*/internalBufferSize, scriptArg);
michael@0:         elementSize = uloc_getVariant(locBuff, tempbuff, internalBufferSize, status);
michael@0:         appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, /*capacity*/internalBufferSize, variantArg);
michael@0:         elementSize = uloc_getKeywordValue(locBuff, "collation", tempbuff, internalBufferSize, status);
michael@0:         appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, /*capacity*/internalBufferSize, keywordArg);
michael@0:     }
michael@0: 
michael@0:     int32_t i = 0;
michael@0:     UColAttributeValue attribute = UCOL_DEFAULT;
michael@0:     for(i = 0; i < UCOL_SIT_ITEMS_COUNT; i++) {
michael@0:         if(options[i].action == _processCollatorOption) {
michael@0:             attribute = ucol_getAttributeOrDefault(coll, (UColAttribute)options[i].attr, status);
michael@0:             if(attribute != UCOL_DEFAULT) {
michael@0:                 char letter = ucol_sit_attributeValueToLetter(attribute, status);
michael@0:                 appendShortStringElement(&letter, 1,
michael@0:                     buffer, &resultSize, /*capacity*/internalBufferSize, options[i].optionStart);
michael@0:             }
michael@0:         }
michael@0:     }
michael@0:     if(coll->variableTopValueisDefault == FALSE) {
michael@0:         //s.variableTopValue = ucol_getVariableTop(coll, status);
michael@0:         elementSize = T_CString_integerToString(tempbuff, coll->variableTopValue, 16);
michael@0:         appendShortStringElement(tempbuff, elementSize, buffer, &resultSize, capacity, variableTopValArg);
michael@0:     }
michael@0: 
michael@0:     UParseError parseError;
michael@0:     return ucol_normalizeShortDefinitionString(buffer, dst, capacity, &parseError, status);
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: ucol_normalizeShortDefinitionString(const char *definition,
michael@0:                                     char *destination,
michael@0:                                     int32_t capacity,
michael@0:                                     UParseError *parseError,
michael@0:                                     UErrorCode *status)
michael@0: {
michael@0: 
michael@0:     if(U_FAILURE(*status)) {
michael@0:         return 0;
michael@0:     }
michael@0: 
michael@0:     if(destination) {
michael@0:         uprv_memset(destination, 0, capacity*sizeof(char));
michael@0:     }
michael@0: 
michael@0:     UParseError pe;
michael@0:     if(!parseError) {
michael@0:         parseError = &pe;
michael@0:     }
michael@0: 
michael@0:     // validate
michael@0:     CollatorSpec s;
michael@0:     ucol_sit_initCollatorSpecs(&s);
michael@0:     ucol_sit_readSpecs(&s, definition, parseError, status);
michael@0:     return ucol_sit_dumpSpecs(&s, destination, capacity, status);
michael@0: }
michael@0: 
michael@0: U_CAPI UColAttributeValue  U_EXPORT2
michael@0: ucol_getAttributeOrDefault(const UCollator *coll, UColAttribute attr, UErrorCode *status)
michael@0: {
michael@0:     if(U_FAILURE(*status) || coll == NULL) {
michael@0:       return UCOL_DEFAULT;
michael@0:     }
michael@0:     switch(attr) {
michael@0:     case UCOL_NUMERIC_COLLATION:
michael@0:         return coll->numericCollationisDefault?UCOL_DEFAULT:coll->numericCollation;
michael@0:     case UCOL_HIRAGANA_QUATERNARY_MODE:
michael@0:         return coll->hiraganaQisDefault?UCOL_DEFAULT:coll->hiraganaQ;
michael@0:     case UCOL_FRENCH_COLLATION: /* attribute for direction of secondary weights*/
michael@0:         return coll->frenchCollationisDefault?UCOL_DEFAULT:coll->frenchCollation;
michael@0:     case UCOL_ALTERNATE_HANDLING: /* attribute for handling variable elements*/
michael@0:         return coll->alternateHandlingisDefault?UCOL_DEFAULT:coll->alternateHandling;
michael@0:     case UCOL_CASE_FIRST: /* who goes first, lower case or uppercase */
michael@0:         return coll->caseFirstisDefault?UCOL_DEFAULT:coll->caseFirst;
michael@0:     case UCOL_CASE_LEVEL: /* do we have an extra case level */
michael@0:         return coll->caseLevelisDefault?UCOL_DEFAULT:coll->caseLevel;
michael@0:     case UCOL_NORMALIZATION_MODE: /* attribute for normalization */
michael@0:         return coll->normalizationModeisDefault?UCOL_DEFAULT:coll->normalizationMode;
michael@0:     case UCOL_STRENGTH:         /* attribute for strength */
michael@0:         return coll->strengthisDefault?UCOL_DEFAULT:coll->strength;
michael@0:     case UCOL_ATTRIBUTE_COUNT:
michael@0:     default:
michael@0:         *status = U_ILLEGAL_ARGUMENT_ERROR;
michael@0: #ifdef UCOL_TRACE_SIT
michael@0:         fprintf(stderr, "%s:%d: Unknown attr value '%d': %s\n", __FILE__, __LINE__, (int)attr, u_errorName(*status));
michael@0: #endif
michael@0:         break;
michael@0:     }
michael@0:     return UCOL_DEFAULT;
michael@0: }
michael@0: 
michael@0: 
michael@0: struct contContext {
michael@0:     const UCollator *coll;
michael@0:     USet            *conts;
michael@0:     USet            *expansions;
michael@0:     USet            *removedContractions;
michael@0:     UBool           addPrefixes;
michael@0:     UErrorCode      *status;
michael@0: };
michael@0: 
michael@0: 
michael@0: 
michael@0: static void
michael@0: addSpecial(contContext *context, UChar *buffer, int32_t bufLen,
michael@0:                uint32_t CE, int32_t leftIndex, int32_t rightIndex, UErrorCode *status)
michael@0: {
michael@0:   const UCollator *coll = context->coll;
michael@0:   USet *contractions = context->conts;
michael@0:   USet *expansions = context->expansions;
michael@0:   UBool addPrefixes = context->addPrefixes;
michael@0: 
michael@0:     const UChar *UCharOffset = (UChar *)coll->image+getContractOffset(CE);
michael@0:     uint32_t newCE = *(coll->contractionCEs + (UCharOffset - coll->contractionIndex));
michael@0:     // we might have a contraction that ends from previous level
michael@0:     if(newCE != UCOL_NOT_FOUND) {
michael@0:       if(isSpecial(CE) && getCETag(CE) == CONTRACTION_TAG && isSpecial(newCE) && getCETag(newCE) == SPEC_PROC_TAG && addPrefixes) {
michael@0:         addSpecial(context, buffer, bufLen, newCE, leftIndex, rightIndex, status);
michael@0:       }
michael@0:       if(contractions && rightIndex-leftIndex > 1) {
michael@0:             uset_addString(contractions, buffer+leftIndex, rightIndex-leftIndex);
michael@0:             if(expansions && isSpecial(CE) && getCETag(CE) == EXPANSION_TAG) {
michael@0:               uset_addString(expansions, buffer+leftIndex, rightIndex-leftIndex);
michael@0:             }
michael@0:       }
michael@0:     }
michael@0: 
michael@0:     UCharOffset++;
michael@0:     // check whether we're doing contraction or prefix
michael@0:     if(getCETag(CE) == SPEC_PROC_TAG && addPrefixes) {
michael@0:       if(leftIndex == 0) {
michael@0:           *status = U_INTERNAL_PROGRAM_ERROR;
michael@0:           return;
michael@0:       }
michael@0:       --leftIndex;
michael@0:       while(*UCharOffset != 0xFFFF) {
michael@0:           newCE = *(coll->contractionCEs + (UCharOffset - coll->contractionIndex));
michael@0:           buffer[leftIndex] = *UCharOffset;
michael@0:           if(isSpecial(newCE) && (getCETag(newCE) == CONTRACTION_TAG || getCETag(newCE) == SPEC_PROC_TAG)) {
michael@0:               addSpecial(context, buffer, bufLen, newCE, leftIndex, rightIndex, status);
michael@0:           } else {
michael@0:             if(contractions) {
michael@0:                 uset_addString(contractions, buffer+leftIndex, rightIndex-leftIndex);
michael@0:             }
michael@0:             if(expansions && isSpecial(newCE) && getCETag(newCE) == EXPANSION_TAG) {
michael@0:               uset_addString(expansions, buffer+leftIndex, rightIndex-leftIndex);
michael@0:             }
michael@0:           }
michael@0:           UCharOffset++;
michael@0:       }
michael@0:     } else if(getCETag(CE) == CONTRACTION_TAG) {
michael@0:       if(rightIndex == bufLen-1) {
michael@0:           *status = U_INTERNAL_PROGRAM_ERROR;
michael@0:           return;
michael@0:       }
michael@0:       while(*UCharOffset != 0xFFFF) {
michael@0:           newCE = *(coll->contractionCEs + (UCharOffset - coll->contractionIndex));
michael@0:           buffer[rightIndex] = *UCharOffset;
michael@0:           if(isSpecial(newCE) && (getCETag(newCE) == CONTRACTION_TAG || getCETag(newCE) == SPEC_PROC_TAG)) {
michael@0:               addSpecial(context, buffer, bufLen, newCE, leftIndex, rightIndex+1, status);
michael@0:           } else {
michael@0:             if(contractions) {
michael@0:               uset_addString(contractions, buffer+leftIndex, rightIndex+1-leftIndex);
michael@0:             }
michael@0:             if(expansions && isSpecial(newCE) && getCETag(newCE) == EXPANSION_TAG) {
michael@0:               uset_addString(expansions, buffer+leftIndex, rightIndex+1-leftIndex);
michael@0:             }
michael@0:           }
michael@0:           UCharOffset++;
michael@0:       }
michael@0:     }
michael@0: 
michael@0: }
michael@0: 
michael@0: U_CDECL_BEGIN
michael@0: static UBool U_CALLCONV
michael@0: _processSpecials(const void *context, UChar32 start, UChar32 limit, uint32_t CE)
michael@0: {
michael@0:     UErrorCode *status = ((contContext *)context)->status;
michael@0:     USet *expansions = ((contContext *)context)->expansions;
michael@0:     USet *removed = ((contContext *)context)->removedContractions;
michael@0:     UBool addPrefixes = ((contContext *)context)->addPrefixes;
michael@0:     UChar contraction[internalBufferSize];
michael@0:     if(isSpecial(CE)) {
michael@0:       if(((getCETag(CE) == SPEC_PROC_TAG && addPrefixes) || getCETag(CE) == CONTRACTION_TAG)) {
michael@0:         while(start < limit && U_SUCCESS(*status)) {
michael@0:             // if there are suppressed contractions, we don't
michael@0:             // want to add them.
michael@0:             if(removed && uset_contains(removed, start)) {
michael@0:                 start++;
michael@0:                 continue;
michael@0:             }
michael@0:             // we start our contraction from middle, since we don't know if it
michael@0:             // will grow toward right or left
michael@0:             contraction[internalBufferSize/2] = (UChar)start;
michael@0:             addSpecial(((contContext *)context), contraction, internalBufferSize, CE, internalBufferSize/2, internalBufferSize/2+1, status);
michael@0:             start++;
michael@0:         }
michael@0:       } else if(expansions && getCETag(CE) == EXPANSION_TAG) {
michael@0:         while(start < limit && U_SUCCESS(*status)) {
michael@0:           uset_add(expansions, start++);
michael@0:         }
michael@0:       }
michael@0:     }
michael@0:     if(U_FAILURE(*status)) {
michael@0:         return FALSE;
michael@0:     } else {
michael@0:         return TRUE;
michael@0:     }
michael@0: }
michael@0: 
michael@0: U_CDECL_END
michael@0: 
michael@0: 
michael@0: 
michael@0: /**
michael@0:  * Get a set containing the contractions defined by the collator. The set includes
michael@0:  * both the UCA contractions and the contractions defined by the collator
michael@0:  * @param coll collator
michael@0:  * @param conts the set to hold the result
michael@0:  * @param status to hold the error code
michael@0:  * @return the size of the contraction set
michael@0:  */
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: ucol_getContractions( const UCollator *coll,
michael@0:                   USet *contractions,
michael@0:                   UErrorCode *status)
michael@0: {
michael@0:   ucol_getContractionsAndExpansions(coll, contractions, NULL, FALSE, status);
michael@0:   return uset_getItemCount(contractions);
michael@0: }
michael@0: 
michael@0: /**
michael@0:  * Get a set containing the expansions defined by the collator. The set includes
michael@0:  * both the UCA expansions and the expansions defined by the tailoring
michael@0:  * @param coll collator
michael@0:  * @param conts the set to hold the result
michael@0:  * @param addPrefixes add the prefix contextual elements to contractions
michael@0:  * @param status to hold the error code
michael@0:  *
michael@0:  * @draft ICU 3.4
michael@0:  */
michael@0: U_CAPI void U_EXPORT2
michael@0: ucol_getContractionsAndExpansions( const UCollator *coll,
michael@0:                   USet *contractions,
michael@0:                   USet *expansions,
michael@0:                   UBool addPrefixes,
michael@0:                   UErrorCode *status)
michael@0: {
michael@0:     if(U_FAILURE(*status)) {
michael@0:         return;
michael@0:     }
michael@0:     if(coll == NULL) {
michael@0:         *status = U_ILLEGAL_ARGUMENT_ERROR;
michael@0:         return;
michael@0:     }
michael@0: 
michael@0:     if(contractions) {
michael@0:       uset_clear(contractions);
michael@0:     }
michael@0:     if(expansions) {
michael@0:       uset_clear(expansions);
michael@0:     }
michael@0:     int32_t rulesLen = 0;
michael@0:     const UChar* rules = ucol_getRules(coll, &rulesLen);
michael@0:     UColTokenParser src;
michael@0:     ucol_tok_initTokenList(&src, rules, rulesLen, coll->UCA,
michael@0:                            ucol_tok_getRulesFromBundle, NULL, status);
michael@0: 
michael@0:     contContext c = { NULL, contractions, expansions, src.removeSet, addPrefixes, status };
michael@0: 
michael@0:     // Add the UCA contractions
michael@0:     c.coll = coll->UCA;
michael@0:     utrie_enum(&coll->UCA->mapping, NULL, _processSpecials, &c);
michael@0: 
michael@0:     // This is collator specific. Add contractions from a collator
michael@0:     c.coll = coll;
michael@0:     c.removedContractions =  NULL;
michael@0:     utrie_enum(&coll->mapping, NULL, _processSpecials, &c);
michael@0:     ucol_tok_closeTokenList(&src);
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: ucol_getUnsafeSet( const UCollator *coll,
michael@0:                   USet *unsafe,
michael@0:                   UErrorCode *status)
michael@0: {
michael@0:     UChar buffer[internalBufferSize];
michael@0:     int32_t len = 0;
michael@0: 
michael@0:     uset_clear(unsafe);
michael@0: 
michael@0:     // cccpattern = "[[:^tccc=0:][:^lccc=0:]]", unfortunately variant
michael@0:     static const UChar cccpattern[25] = { 0x5b, 0x5b, 0x3a, 0x5e, 0x74, 0x63, 0x63, 0x63, 0x3d, 0x30, 0x3a, 0x5d,
michael@0:                                     0x5b, 0x3a, 0x5e, 0x6c, 0x63, 0x63, 0x63, 0x3d, 0x30, 0x3a, 0x5d, 0x5d, 0x00 };
michael@0: 
michael@0:     // add chars that fail the fcd check
michael@0:     uset_applyPattern(unsafe, cccpattern, 24, USET_IGNORE_SPACE, status);
michael@0: 
michael@0:     // add Thai/Lao prevowels
michael@0:     uset_addRange(unsafe, 0xe40, 0xe44);
michael@0:     uset_addRange(unsafe, 0xec0, 0xec4);
michael@0:     // add lead/trail surrogates
michael@0:     uset_addRange(unsafe, 0xd800, 0xdfff);
michael@0: 
michael@0:     USet *contractions = uset_open(0,0);
michael@0: 
michael@0:     int32_t i = 0, j = 0;
michael@0:     int32_t contsSize = ucol_getContractions(coll, contractions, status);
michael@0:     UChar32 c = 0;
michael@0:     // Contraction set consists only of strings
michael@0:     // to get unsafe code points, we need to
michael@0:     // break the strings apart and add them to the unsafe set
michael@0:     for(i = 0; i < contsSize; i++) {
michael@0:         len = uset_getItem(contractions, i, NULL, NULL, buffer, internalBufferSize, status);
michael@0:         if(len > 0) {
michael@0:             j = 0;
michael@0:             while(j < len) {
michael@0:                 U16_NEXT(buffer, j, len, c);
michael@0:                 if(j < len) {
michael@0:                     uset_add(unsafe, c);
michael@0:                 }
michael@0:             }
michael@0:         }
michael@0:     }
michael@0: 
michael@0:     uset_close(contractions);
michael@0: 
michael@0:     return uset_size(unsafe);
michael@0: }
michael@0: #endif