1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/icu/source/i18n/unicode/sortkey.h Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,337 @@
1.4 +/*
1.5 + *****************************************************************************
1.6 + * Copyright (C) 1996-2013, International Business Machines Corporation and others.
1.7 + * All Rights Reserved.
1.8 + *****************************************************************************
1.9 + *
1.10 + * File sortkey.h
1.11 + *
1.12 + * Created by: Helena Shih
1.13 + *
1.14 + * Modification History:
1.15 + *
1.16 + * Date Name Description
1.17 + *
1.18 + * 6/20/97 helena Java class name change.
1.19 + * 8/18/97 helena Added internal API documentation.
1.20 + * 6/26/98 erm Changed to use byte arrays and memcmp.
1.21 + *****************************************************************************
1.22 + */
1.23 +
1.24 +#ifndef SORTKEY_H
1.25 +#define SORTKEY_H
1.26 +
1.27 +#include "unicode/utypes.h"
1.28 +
1.29 +/**
1.30 + * \file
1.31 + * \brief C++ API: Keys for comparing strings multiple times.
1.32 + */
1.33 +
1.34 +#if !UCONFIG_NO_COLLATION
1.35 +
1.36 +#include "unicode/uobject.h"
1.37 +#include "unicode/unistr.h"
1.38 +#include "unicode/coll.h"
1.39 +
1.40 +U_NAMESPACE_BEGIN
1.41 +
1.42 +/* forward declaration */
1.43 +class RuleBasedCollator;
1.44 +
1.45 +/**
1.46 + *
1.47 + * Collation keys are generated by the Collator class. Use the CollationKey objects
1.48 + * instead of Collator to compare strings multiple times. A CollationKey
1.49 + * preprocesses the comparison information from the Collator object to
1.50 + * make the comparison faster. If you are not going to comparing strings
1.51 + * multiple times, then using the Collator object is generally faster,
1.52 + * since it only processes as much of the string as needed to make a
1.53 + * comparison.
1.54 + * <p> For example (with strength == tertiary)
1.55 + * <p>When comparing "Abernathy" to "Baggins-Smythworthy", Collator
1.56 + * only needs to process a couple of characters, while a comparison
1.57 + * with CollationKeys will process all of the characters. On the other hand,
1.58 + * if you are doing a sort of a number of fields, it is much faster to use
1.59 + * CollationKeys, since you will be comparing strings multiple times.
1.60 + * <p>Typical use of CollationKeys are in databases, where you store a CollationKey
1.61 + * in a hidden field, and use it for sorting or indexing.
1.62 + *
1.63 + * <p>Example of use:
1.64 + * <pre>
1.65 + * \code
1.66 + * UErrorCode success = U_ZERO_ERROR;
1.67 + * Collator* myCollator = Collator::createInstance(success);
1.68 + * CollationKey* keys = new CollationKey [3];
1.69 + * myCollator->getCollationKey("Tom", keys[0], success );
1.70 + * myCollator->getCollationKey("Dick", keys[1], success );
1.71 + * myCollator->getCollationKey("Harry", keys[2], success );
1.72 + *
1.73 + * // Inside body of sort routine, compare keys this way:
1.74 + * CollationKey tmp;
1.75 + * if(keys[0].compareTo( keys[1] ) > 0 ) {
1.76 + * tmp = keys[0]; keys[0] = keys[1]; keys[1] = tmp;
1.77 + * }
1.78 + * //...
1.79 + * \endcode
1.80 + * </pre>
1.81 + * <p>Because Collator::compare()'s algorithm is complex, it is faster to sort
1.82 + * long lists of words by retrieving collation keys with Collator::getCollationKey().
1.83 + * You can then cache the collation keys and compare them using CollationKey::compareTo().
1.84 + * <p>
1.85 + * <strong>Note:</strong> <code>Collator</code>s with different Locale,
1.86 + * CollationStrength and DecompositionMode settings will return different
1.87 + * CollationKeys for the same set of strings. Locales have specific
1.88 + * collation rules, and the way in which secondary and tertiary differences
1.89 + * are taken into account, for example, will result in different CollationKeys
1.90 + * for same strings.
1.91 + * <p>
1.92 +
1.93 + * @see Collator
1.94 + * @see RuleBasedCollator
1.95 + * @version 1.3 12/18/96
1.96 + * @author Helena Shih
1.97 + * @stable ICU 2.0
1.98 + */
1.99 +class U_I18N_API CollationKey : public UObject {
1.100 +public:
1.101 + /**
1.102 + * This creates an empty collation key based on the null string. An empty
1.103 + * collation key contains no sorting information. When comparing two empty
1.104 + * collation keys, the result is Collator::EQUAL. Comparing empty collation key
1.105 + * with non-empty collation key is always Collator::LESS.
1.106 + * @stable ICU 2.0
1.107 + */
1.108 + CollationKey();
1.109 +
1.110 +
1.111 + /**
1.112 + * Creates a collation key based on the collation key values.
1.113 + * @param values the collation key values
1.114 + * @param count number of collation key values, including trailing nulls.
1.115 + * @stable ICU 2.0
1.116 + */
1.117 + CollationKey(const uint8_t* values,
1.118 + int32_t count);
1.119 +
1.120 + /**
1.121 + * Copy constructor.
1.122 + * @param other the object to be copied.
1.123 + * @stable ICU 2.0
1.124 + */
1.125 + CollationKey(const CollationKey& other);
1.126 +
1.127 + /**
1.128 + * Sort key destructor.
1.129 + * @stable ICU 2.0
1.130 + */
1.131 + virtual ~CollationKey();
1.132 +
1.133 + /**
1.134 + * Assignment operator
1.135 + * @param other the object to be copied.
1.136 + * @stable ICU 2.0
1.137 + */
1.138 + const CollationKey& operator=(const CollationKey& other);
1.139 +
1.140 + /**
1.141 + * Compare if two collation keys are the same.
1.142 + * @param source the collation key to compare to.
1.143 + * @return Returns true if two collation keys are equal, false otherwise.
1.144 + * @stable ICU 2.0
1.145 + */
1.146 + UBool operator==(const CollationKey& source) const;
1.147 +
1.148 + /**
1.149 + * Compare if two collation keys are not the same.
1.150 + * @param source the collation key to compare to.
1.151 + * @return Returns TRUE if two collation keys are different, FALSE otherwise.
1.152 + * @stable ICU 2.0
1.153 + */
1.154 + UBool operator!=(const CollationKey& source) const;
1.155 +
1.156 +
1.157 + /**
1.158 + * Test to see if the key is in an invalid state. The key will be in an
1.159 + * invalid state if it couldn't allocate memory for some operation.
1.160 + * @return Returns TRUE if the key is in an invalid, FALSE otherwise.
1.161 + * @stable ICU 2.0
1.162 + */
1.163 + UBool isBogus(void) const;
1.164 +
1.165 + /**
1.166 + * Returns a pointer to the collation key values. The storage is owned
1.167 + * by the collation key and the pointer will become invalid if the key
1.168 + * is deleted.
1.169 + * @param count the output parameter of number of collation key values,
1.170 + * including any trailing nulls.
1.171 + * @return a pointer to the collation key values.
1.172 + * @stable ICU 2.0
1.173 + */
1.174 + const uint8_t* getByteArray(int32_t& count) const;
1.175 +
1.176 +#ifdef U_USE_COLLATION_KEY_DEPRECATES
1.177 + /**
1.178 + * Extracts the collation key values into a new array. The caller owns
1.179 + * this storage and should free it.
1.180 + * @param count the output parameter of number of collation key values,
1.181 + * including any trailing nulls.
1.182 + * @obsolete ICU 2.6. Use getByteArray instead since this API will be removed in that release.
1.183 + */
1.184 + uint8_t* toByteArray(int32_t& count) const;
1.185 +#endif
1.186 +
1.187 +#ifndef U_HIDE_DEPRECATED_API
1.188 + /**
1.189 + * Convenience method which does a string(bit-wise) comparison of the
1.190 + * two collation keys.
1.191 + * @param target target collation key to be compared with
1.192 + * @return Returns Collator::LESS if sourceKey < targetKey,
1.193 + * Collator::GREATER if sourceKey > targetKey and Collator::EQUAL
1.194 + * otherwise.
1.195 + * @deprecated ICU 2.6 use the overload with error code
1.196 + */
1.197 + Collator::EComparisonResult compareTo(const CollationKey& target) const;
1.198 +#endif /* U_HIDE_DEPRECATED_API */
1.199 +
1.200 + /**
1.201 + * Convenience method which does a string(bit-wise) comparison of the
1.202 + * two collation keys.
1.203 + * @param target target collation key to be compared with
1.204 + * @param status error code
1.205 + * @return Returns UCOL_LESS if sourceKey < targetKey,
1.206 + * UCOL_GREATER if sourceKey > targetKey and UCOL_EQUAL
1.207 + * otherwise.
1.208 + * @stable ICU 2.6
1.209 + */
1.210 + UCollationResult compareTo(const CollationKey& target, UErrorCode &status) const;
1.211 +
1.212 + /**
1.213 + * Creates an integer that is unique to the collation key. NOTE: this
1.214 + * is not the same as String.hashCode.
1.215 + * <p>Example of use:
1.216 + * <pre>
1.217 + * . UErrorCode status = U_ZERO_ERROR;
1.218 + * . Collator *myCollation = Collator::createInstance(Locale::US, status);
1.219 + * . if (U_FAILURE(status)) return;
1.220 + * . CollationKey key1, key2;
1.221 + * . UErrorCode status1 = U_ZERO_ERROR, status2 = U_ZERO_ERROR;
1.222 + * . myCollation->getCollationKey("abc", key1, status1);
1.223 + * . if (U_FAILURE(status1)) { delete myCollation; return; }
1.224 + * . myCollation->getCollationKey("ABC", key2, status2);
1.225 + * . if (U_FAILURE(status2)) { delete myCollation; return; }
1.226 + * . // key1.hashCode() != key2.hashCode()
1.227 + * </pre>
1.228 + * @return the hash value based on the string's collation order.
1.229 + * @see UnicodeString#hashCode
1.230 + * @stable ICU 2.0
1.231 + */
1.232 + int32_t hashCode(void) const;
1.233 +
1.234 + /**
1.235 + * ICU "poor man's RTTI", returns a UClassID for the actual class.
1.236 + * @stable ICU 2.2
1.237 + */
1.238 + virtual UClassID getDynamicClassID() const;
1.239 +
1.240 + /**
1.241 + * ICU "poor man's RTTI", returns a UClassID for this class.
1.242 + * @stable ICU 2.2
1.243 + */
1.244 + static UClassID U_EXPORT2 getStaticClassID();
1.245 +
1.246 +private:
1.247 + /**
1.248 + * Replaces the current bytes buffer with a new one of newCapacity
1.249 + * and copies length bytes from the old buffer to the new one.
1.250 + * @return the new buffer, or NULL if the allocation failed
1.251 + */
1.252 + uint8_t *reallocate(int32_t newCapacity, int32_t length);
1.253 + /**
1.254 + * Set a new length for a new sort key in the existing fBytes.
1.255 + */
1.256 + void setLength(int32_t newLength);
1.257 +
1.258 + uint8_t *getBytes() {
1.259 + return (fFlagAndLength >= 0) ? fUnion.fStackBuffer : fUnion.fFields.fBytes;
1.260 + }
1.261 + const uint8_t *getBytes() const {
1.262 + return (fFlagAndLength >= 0) ? fUnion.fStackBuffer : fUnion.fFields.fBytes;
1.263 + }
1.264 + int32_t getCapacity() const {
1.265 + return (fFlagAndLength >= 0) ? (int32_t)sizeof(fUnion) : fUnion.fFields.fCapacity;
1.266 + }
1.267 + int32_t getLength() const { return fFlagAndLength & 0x7fffffff; }
1.268 +
1.269 + /**
1.270 + * Set the CollationKey to a "bogus" or invalid state
1.271 + * @return this CollationKey
1.272 + */
1.273 + CollationKey& setToBogus(void);
1.274 + /**
1.275 + * Resets this CollationKey to an empty state
1.276 + * @return this CollationKey
1.277 + */
1.278 + CollationKey& reset(void);
1.279 +
1.280 + /**
1.281 + * Allow private access to RuleBasedCollator
1.282 + */
1.283 + friend class RuleBasedCollator;
1.284 + friend class CollationKeyByteSink;
1.285 +
1.286 + // Class fields. sizeof(CollationKey) is intended to be 48 bytes
1.287 + // on a machine with 64-bit pointers.
1.288 + // We use a union to maximize the size of the internal buffer,
1.289 + // similar to UnicodeString but not as tight and complex.
1.290 +
1.291 + // (implicit) *vtable;
1.292 + /**
1.293 + * Sort key length and flag.
1.294 + * Bit 31 is set if the buffer is heap-allocated.
1.295 + * Bits 30..0 contain the sort key length.
1.296 + */
1.297 + int32_t fFlagAndLength;
1.298 + /**
1.299 + * Unique hash value of this CollationKey.
1.300 + * Special value 2 if the key is bogus.
1.301 + */
1.302 + mutable int32_t fHashCode;
1.303 + /**
1.304 + * fUnion provides 32 bytes for the internal buffer or for
1.305 + * pointer+capacity.
1.306 + */
1.307 + union StackBufferOrFields {
1.308 + /** fStackBuffer is used iff fFlagAndLength>=0, else fFields is used */
1.309 + uint8_t fStackBuffer[32];
1.310 + struct {
1.311 + uint8_t *fBytes;
1.312 + int32_t fCapacity;
1.313 + } fFields;
1.314 + } fUnion;
1.315 +};
1.316 +
1.317 +inline UBool
1.318 +CollationKey::operator!=(const CollationKey& other) const
1.319 +{
1.320 + return !(*this == other);
1.321 +}
1.322 +
1.323 +inline UBool
1.324 +CollationKey::isBogus() const
1.325 +{
1.326 + return fHashCode == 2; // kBogusHashCode
1.327 +}
1.328 +
1.329 +inline const uint8_t*
1.330 +CollationKey::getByteArray(int32_t &count) const
1.331 +{
1.332 + count = getLength();
1.333 + return getBytes();
1.334 +}
1.335 +
1.336 +U_NAMESPACE_END
1.337 +
1.338 +#endif /* #if !UCONFIG_NO_COLLATION */
1.339 +
1.340 +#endif
