michael@0: /*
michael@0: **********************************************************************
michael@0: *   Copyright (C) 2013, International Business Machines
michael@0: *   Corporation and others.  All Rights Reserved.
michael@0: **********************************************************************
michael@0: *
michael@0: * scriptset.cpp
michael@0: *
michael@0: * created on: 2013 Jan 7
michael@0: * created by: Andy Heninger
michael@0: */
michael@0: 
michael@0: #include "unicode/utypes.h"
michael@0: 
michael@0: #include "unicode/uchar.h"
michael@0: #include "unicode/unistr.h"
michael@0: 
michael@0: #include "scriptset.h"
michael@0: #include "uassert.h"
michael@0: 
michael@0: U_NAMESPACE_BEGIN
michael@0: 
michael@0: #define LENGTHOF(array) (int32_t)(sizeof(array)/sizeof((array)[0]))
michael@0: 
michael@0: //----------------------------------------------------------------------------
michael@0: //
michael@0: //  ScriptSet implementation
michael@0: //
michael@0: //----------------------------------------------------------------------------
michael@0: ScriptSet::ScriptSet() {
michael@0:     for (uint32_t i=0; i<LENGTHOF(bits); i++) {
michael@0:         bits[i] = 0;
michael@0:     }
michael@0: }
michael@0: 
michael@0: ScriptSet::~ScriptSet() {
michael@0: }
michael@0: 
michael@0: ScriptSet::ScriptSet(const ScriptSet &other) {
michael@0:     *this = other;
michael@0: }
michael@0:     
michael@0: 
michael@0: ScriptSet & ScriptSet::operator =(const ScriptSet &other) {
michael@0:     for (uint32_t i=0; i<LENGTHOF(bits); i++) {
michael@0:         bits[i] = other.bits[i];
michael@0:     }
michael@0:     return *this;
michael@0: }
michael@0: 
michael@0: 
michael@0: UBool ScriptSet::operator == (const ScriptSet &other) const {
michael@0:     for (uint32_t i=0; i<LENGTHOF(bits); i++) {
michael@0:         if (bits[i] != other.bits[i]) {
michael@0:             return FALSE;
michael@0:         }
michael@0:     }
michael@0:     return TRUE;
michael@0: }
michael@0: 
michael@0: UBool ScriptSet::test(UScriptCode script, UErrorCode &status) const {
michael@0:     if (U_FAILURE(status)) {
michael@0:         return FALSE;
michael@0:     }
michael@0:     if (script < 0 || script >= (int32_t)sizeof(bits) * 8) {
michael@0:         status = U_ILLEGAL_ARGUMENT_ERROR;
michael@0:         return FALSE;
michael@0:     }
michael@0:     uint32_t index = script / 32;
michael@0:     uint32_t bit   = 1 << (script & 31);
michael@0:     return ((bits[index] & bit) != 0);
michael@0: }
michael@0: 
michael@0: 
michael@0: ScriptSet &ScriptSet::set(UScriptCode script, UErrorCode &status) {
michael@0:     if (U_FAILURE(status)) {
michael@0:         return *this;
michael@0:     }
michael@0:     if (script < 0 || script >= (int32_t)sizeof(bits) * 8) {
michael@0:         status = U_ILLEGAL_ARGUMENT_ERROR;
michael@0:         return *this;
michael@0:     }
michael@0:     uint32_t index = script / 32;
michael@0:     uint32_t bit   = 1 << (script & 31);
michael@0:     bits[index] |= bit;
michael@0:     return *this;
michael@0: }
michael@0: 
michael@0: ScriptSet &ScriptSet::reset(UScriptCode script, UErrorCode &status) {
michael@0:     if (U_FAILURE(status)) {
michael@0:         return *this;
michael@0:     }
michael@0:     if (script < 0 || script >= (int32_t)sizeof(bits) * 8) {
michael@0:         status = U_ILLEGAL_ARGUMENT_ERROR;
michael@0:         return *this;
michael@0:     }
michael@0:     uint32_t index = script / 32;
michael@0:     uint32_t bit   = 1 << (script & 31);
michael@0:     bits[index] &= ~bit;
michael@0:     return *this;
michael@0: }
michael@0: 
michael@0: 
michael@0: 
michael@0: ScriptSet &ScriptSet::Union(const ScriptSet &other) {
michael@0:     for (uint32_t i=0; i<LENGTHOF(bits); i++) {
michael@0:         bits[i] |= other.bits[i];
michael@0:     }
michael@0:     return *this;
michael@0: }
michael@0: 
michael@0: ScriptSet &ScriptSet::intersect(const ScriptSet &other) {
michael@0:     for (uint32_t i=0; i<LENGTHOF(bits); i++) {
michael@0:         bits[i] &= other.bits[i];
michael@0:     }
michael@0:     return *this;
michael@0: }
michael@0: 
michael@0: ScriptSet &ScriptSet::intersect(UScriptCode script, UErrorCode &status) {
michael@0:     ScriptSet t;
michael@0:     t.set(script, status);
michael@0:     if (U_SUCCESS(status)) {
michael@0:         this->intersect(t);
michael@0:     }
michael@0:     return *this;
michael@0: }
michael@0:     
michael@0: UBool ScriptSet::intersects(const ScriptSet &other) const {
michael@0:     for (uint32_t i=0; i<LENGTHOF(bits); i++) {
michael@0:         if ((bits[i] & other.bits[i]) != 0) {
michael@0:             return true;
michael@0:         }
michael@0:     }
michael@0:     return false;
michael@0: }
michael@0: 
michael@0: UBool ScriptSet::contains(const ScriptSet &other) const {
michael@0:     ScriptSet t(*this);
michael@0:     t.intersect(other);
michael@0:     return (t == other);
michael@0: }
michael@0: 
michael@0: 
michael@0: ScriptSet &ScriptSet::setAll() {
michael@0:     for (uint32_t i=0; i<LENGTHOF(bits); i++) {
michael@0:         bits[i] = 0xffffffffu;
michael@0:     }
michael@0:     return *this;
michael@0: }
michael@0: 
michael@0: 
michael@0: ScriptSet &ScriptSet::resetAll() {
michael@0:     for (uint32_t i=0; i<LENGTHOF(bits); i++) {
michael@0:         bits[i] = 0;
michael@0:     }
michael@0:     return *this;
michael@0: }
michael@0: 
michael@0: int32_t ScriptSet::countMembers() const {
michael@0:     // This bit counter is good for sparse numbers of '1's, which is
michael@0:     //  very much the case that we will usually have.
michael@0:     int32_t count = 0;
michael@0:     for (uint32_t i=0; i<LENGTHOF(bits); i++) {
michael@0:         uint32_t x = bits[i];
michael@0:         while (x > 0) {
michael@0:             count++;
michael@0:             x &= (x - 1);    // and off the least significant one bit.
michael@0:         }
michael@0:     }
michael@0:     return count;
michael@0: }
michael@0: 
michael@0: int32_t ScriptSet::hashCode() const {
michael@0:     int32_t hash = 0;
michael@0:     for (int32_t i=0; i<LENGTHOF(bits); i++) {
michael@0:         hash ^= bits[i];
michael@0:     }
michael@0:     return hash;
michael@0: }
michael@0: 
michael@0: int32_t ScriptSet::nextSetBit(int32_t fromIndex) const {
michael@0:     // TODO: Wants a better implementation.
michael@0:     if (fromIndex < 0) {
michael@0:         return -1;
michael@0:     }
michael@0:     UErrorCode status = U_ZERO_ERROR;
michael@0:     for (int32_t scriptIndex = fromIndex; scriptIndex < (int32_t)sizeof(bits)*8; scriptIndex++) {
michael@0:         if (test((UScriptCode)scriptIndex, status)) {
michael@0:             return scriptIndex;
michael@0:         }
michael@0:     }
michael@0:     return -1;
michael@0: }
michael@0: 
michael@0: UnicodeString &ScriptSet::displayScripts(UnicodeString &dest) const {
michael@0:     UBool firstTime = TRUE;
michael@0:     for (int32_t i = nextSetBit(0); i >= 0; i = nextSetBit(i + 1)) {
michael@0:         if (!firstTime) {
michael@0:             dest.append((UChar)0x20);
michael@0:         }
michael@0:         firstTime = FALSE;
michael@0:         const char *scriptName = uscript_getShortName((UScriptCode(i)));
michael@0:         dest.append(UnicodeString(scriptName, -1, US_INV));
michael@0:     }
michael@0:     return dest;
michael@0: }
michael@0: 
michael@0: ScriptSet &ScriptSet::parseScripts(const UnicodeString &scriptString, UErrorCode &status) {
michael@0:     resetAll();
michael@0:     if (U_FAILURE(status)) {
michael@0:         return *this;
michael@0:     }
michael@0:     UnicodeString oneScriptName;
michael@0:     for (int32_t i=0; i<scriptString.length();) {
michael@0:         UChar32 c = scriptString.char32At(i);
michael@0:         i = scriptString.moveIndex32(i, 1);
michael@0:         if (!u_isUWhiteSpace(c)) {
michael@0:             oneScriptName.append(c);
michael@0:             if (i < scriptString.length()) {
michael@0:                 continue;
michael@0:             }
michael@0:         }
michael@0:         if (oneScriptName.length() > 0) {
michael@0:             char buf[40];
michael@0:             oneScriptName.extract(0, oneScriptName.length(), buf, sizeof(buf)-1, US_INV);
michael@0:             buf[sizeof(buf)-1] = 0;
michael@0:             int32_t sc = u_getPropertyValueEnum(UCHAR_SCRIPT, buf);
michael@0:             if (sc == UCHAR_INVALID_CODE) {
michael@0:                 status = U_ILLEGAL_ARGUMENT_ERROR;
michael@0:             } else {
michael@0:                 this->set((UScriptCode)sc, status);
michael@0:             }
michael@0:             if (U_FAILURE(status)) {
michael@0:                 return *this;
michael@0:             }
michael@0:             oneScriptName.remove();
michael@0:         }
michael@0:     }
michael@0:     return *this;
michael@0: }
michael@0: 
michael@0: U_NAMESPACE_END
michael@0: 
michael@0: U_CAPI UBool U_EXPORT2
michael@0: uhash_equalsScriptSet(const UElement key1, const UElement key2) {
michael@0:     icu::ScriptSet *s1 = static_cast<icu::ScriptSet *>(key1.pointer);
michael@0:     icu::ScriptSet *s2 = static_cast<icu::ScriptSet *>(key2.pointer);
michael@0:     return (*s1 == *s2);
michael@0: }
michael@0: 
michael@0: U_CAPI int8_t U_EXPORT2
michael@0: uhash_compareScriptSet(UElement key0, UElement key1) {
michael@0:     icu::ScriptSet *s0 = static_cast<icu::ScriptSet *>(key0.pointer);
michael@0:     icu::ScriptSet *s1 = static_cast<icu::ScriptSet *>(key1.pointer);
michael@0:     int32_t diff = s0->countMembers() - s1->countMembers();
michael@0:     if (diff != 0) return diff;
michael@0:     int32_t i0 = s0->nextSetBit(0);
michael@0:     int32_t i1 = s1->nextSetBit(0);
michael@0:     while ((diff = i0-i1) == 0 && i0 > 0) {
michael@0:         i0 = s0->nextSetBit(i0+1);
michael@0:         i1 = s1->nextSetBit(i1+1);
michael@0:     }
michael@0:     return (int8_t)diff;
michael@0: }
michael@0: 
michael@0: U_CAPI int32_t U_EXPORT2
michael@0: uhash_hashScriptSet(const UElement key) {
michael@0:     icu::ScriptSet *s = static_cast<icu::ScriptSet *>(key.pointer);
michael@0:     return s->hashCode();
michael@0: }
michael@0: 
michael@0: U_CAPI void U_EXPORT2
michael@0: uhash_deleteScriptSet(void *obj) {
michael@0:     icu::ScriptSet *s = static_cast<icu::ScriptSet *>(obj);
michael@0:     delete s;
michael@0: }