icuSources/i18n/collationdatabuilder.h

   1 /*
   2 *******************************************************************************
   3 * Copyright (C) 2012-2014, International Business Machines
   4 * Corporation and others.  All Rights Reserved.
   5 *******************************************************************************
   6 * collationdatabuilder.h
   7 *
   8 * created on: 2012apr01
   9 * created by: Markus W. Scherer
  10 */
  11
  12 #ifndef __COLLATIONDATABUILDER_H__
  13 #define __COLLATIONDATABUILDER_H__
  14
  15 #include "unicode/utypes.h"
  16
  17 #if !UCONFIG_NO_COLLATION
  18
  19 #include "unicode/uniset.h"
  20 #include "unicode/unistr.h"
  21 #include "unicode/uversion.h"
  22 #include "collation.h"
  23 #include "collationdata.h"
  24 #include "collationsettings.h"
  25 #include "normalizer2impl.h"
  26 #include "utrie2.h"
  27 #include "uvectr32.h"
  28 #include "uvectr64.h"
  29 #include "uvector.h"
  30
  31 U_NAMESPACE_BEGIN
  32
  33 struct ConditionalCE32;
  34
  35 class CollationFastLatinBuilder;
  36 class CopyHelper;
  37 class DataBuilderCollationIterator;
  38 class UCharsTrieBuilder;
  39
  40 /**
  41  * Low-level CollationData builder.
  42  * Takes (character, CE) pairs and builds them into runtime data structures.
  43  * Supports characters with context prefixes and contraction suffixes.
  44  */
  45 class U_I18N_API CollationDataBuilder : public UObject {
  46 public:
  47     /**
  48      * Collation element modifier. Interface class for a modifier
  49      * that changes a tailoring builder's temporary CEs to final CEs.
  50      * Called for every non-special CE32 and every expansion CE.
  51      */
  52     class CEModifier : public UObject {
  53     public:
  54         virtual ~CEModifier();
  55         /** Returns a new CE to replace the non-special input CE32, or else Collation::NO_CE. */
  56         virtual int64_t modifyCE32(uint32_t ce32) const = 0;
  57         /** Returns a new CE to replace the input CE, or else Collation::NO_CE. */
  58         virtual int64_t modifyCE(int64_t ce) const = 0;
  59     };
  60
  61     CollationDataBuilder(UErrorCode &errorCode);
  62
  63     virtual ~CollationDataBuilder();
  64
  65     void initForTailoring(const CollationData *b, UErrorCode &errorCode);
  66
  67     virtual UBool isCompressibleLeadByte(uint32_t b) const;
  68
  69     inline UBool isCompressiblePrimary(uint32_t p) const {
  70         return isCompressibleLeadByte(p >> 24);
  71     }
  72
  73     /**
  74      * @return TRUE if this builder has mappings (e.g., add() has been called)
  75      */
  76     UBool hasMappings() const { return modified; }
  77
  78     /**
  79      * @return TRUE if c has CEs in this builder
  80      */
  81     UBool isAssigned(UChar32 c) const;
  82
  83     /**
  84      * @return the three-byte primary if c maps to a single such CE and has no context data,
  85      * otherwise returns 0.
  86      */
  87     uint32_t getLongPrimaryIfSingleCE(UChar32 c) const;
  88
  89     /**
  90      * @return the single CE for c.
  91      * Sets an error code if c does not have a single CE.
  92      */
  93     int64_t getSingleCE(UChar32 c, UErrorCode &errorCode) const;
  94
  95     void add(const UnicodeString &prefix, const UnicodeString &s,
  96              const int64_t ces[], int32_t cesLength,
  97              UErrorCode &errorCode);
  98
  99     /**
 100      * Encodes the ces as either the returned ce32 by itself,
 101      * or by storing an expansion, with the returned ce32 referring to that.
 102      *
 103      * add(p, s, ces, cesLength) = addCE32(p, s, encodeCEs(ces, cesLength))
 104      */
 105     virtual uint32_t encodeCEs(const int64_t ces[], int32_t cesLength, UErrorCode &errorCode);
 106     void addCE32(const UnicodeString &prefix, const UnicodeString &s,
 107                  uint32_t ce32, UErrorCode &errorCode);
 108
 109     /**
 110      * Sets three-byte-primary CEs for a range of code points in code point order,
 111      * if it is worth doing; otherwise no change is made.
 112      * None of the code points in the range should have complex mappings so far
 113      * (expansions/contractions/prefixes).
 114      * @param start first code point
 115      * @param end last code point (inclusive)
 116      * @param primary primary weight for 'start'
 117      * @param step per-code point primary-weight increment
 118      * @param errorCode ICU in/out error code
 119      * @return TRUE if an OFFSET_TAG range was used for start..end
 120      */
 121     UBool maybeSetPrimaryRange(UChar32 start, UChar32 end,
 122                                uint32_t primary, int32_t step,
 123                                UErrorCode &errorCode);
 124
 125     /**
 126      * Sets three-byte-primary CEs for a range of code points in code point order.
 127      * Sets range values if that is worth doing, or else individual values.
 128      * None of the code points in the range should have complex mappings so far
 129      * (expansions/contractions/prefixes).
 130      * @param start first code point
 131      * @param end last code point (inclusive)
 132      * @param primary primary weight for 'start'
 133      * @param step per-code point primary-weight increment
 134      * @param errorCode ICU in/out error code
 135      * @return the next primary after 'end': start primary incremented by ((end-start)+1)*step
 136      */
 137     uint32_t setPrimaryRangeAndReturnNext(UChar32 start, UChar32 end,
 138                                           uint32_t primary, int32_t step,
 139                                           UErrorCode &errorCode);
 140
 141     /**
 142      * Copies all mappings from the src builder, with modifications.
 143      * This builder here must not be built yet, and should be empty.
 144      */
 145     void copyFrom(const CollationDataBuilder &src, const CEModifier &modifier,
 146                   UErrorCode &errorCode);
 147
 148     void optimize(const UnicodeSet &set, UErrorCode &errorCode);
 149     void suppressContractions(const UnicodeSet &set, UErrorCode &errorCode);
 150
 151     void enableFastLatin() { fastLatinEnabled = TRUE; }
 152     virtual void build(CollationData &data, UErrorCode &errorCode);
 153
 154     /**
 155      * Looks up CEs for s and appends them to the ces array.
 156      * Does not handle normalization: s should be in FCD form.
 157      *
 158      * Does not write completely ignorable CEs.
 159      * Does not write beyond Collation::MAX_EXPANSION_LENGTH.
 160      *
 161      * @return incremented cesLength
 162      */
 163     int32_t getCEs(const UnicodeString &s, int64_t ces[], int32_t cesLength);
 164     int32_t getCEs(const UnicodeString &prefix, const UnicodeString &s,
 165                    int64_t ces[], int32_t cesLength);
 166
 167 protected:
 168     friend class CopyHelper;
 169     friend class DataBuilderCollationIterator;
 170
 171     uint32_t getCE32FromOffsetCE32(UBool fromBase, UChar32 c, uint32_t ce32) const;
 172
 173     int32_t addCE(int64_t ce, UErrorCode &errorCode);
 174     int32_t addCE32(uint32_t ce32, UErrorCode &errorCode);
 175     int32_t addConditionalCE32(const UnicodeString &context, uint32_t ce32, UErrorCode &errorCode);
 176
 177     inline ConditionalCE32 *getConditionalCE32(int32_t index) const {
 178         return static_cast<ConditionalCE32 *>(conditionalCE32s[index]);
 179     }
 180     inline ConditionalCE32 *getConditionalCE32ForCE32(uint32_t ce32) const {
 181         return getConditionalCE32(Collation::indexFromCE32(ce32));
 182     }
 183
 184     static uint32_t makeBuilderContextCE32(int32_t index) {
 185         return Collation::makeCE32FromTagAndIndex(Collation::BUILDER_DATA_TAG, index);
 186     }
 187     static inline UBool isBuilderContextCE32(uint32_t ce32) {
 188         return Collation::hasCE32Tag(ce32, Collation::BUILDER_DATA_TAG);
 189     }
 190
 191     static uint32_t encodeOneCEAsCE32(int64_t ce);
 192     uint32_t encodeOneCE(int64_t ce, UErrorCode &errorCode);
 193     uint32_t encodeExpansion(const int64_t ces[], int32_t length, UErrorCode &errorCode);
 194     uint32_t encodeExpansion32(const int32_t newCE32s[], int32_t length, UErrorCode &errorCode);
 195
 196     uint32_t copyFromBaseCE32(UChar32 c, uint32_t ce32, UBool withContext, UErrorCode &errorCode);
 197     /**
 198      * Copies base contractions to a list of ConditionalCE32.
 199      * Sets cond->next to the index of the first new item
 200      * and returns the index of the last new item.
 201      */
 202     int32_t copyContractionsFromBaseCE32(UnicodeString &context, UChar32 c, uint32_t ce32,
 203                                          ConditionalCE32 *cond, UErrorCode &errorCode);
 204
 205     UBool getJamoCE32s(uint32_t jamoCE32s[], UErrorCode &errorCode);
 206     void setDigitTags(UErrorCode &errorCode);
 207     void setLeadSurrogates(UErrorCode &errorCode);
 208
 209     void buildMappings(CollationData &data, UErrorCode &errorCode);
 210
 211     void clearContexts();
 212     void buildContexts(UErrorCode &errorCode);
 213     uint32_t buildContext(ConditionalCE32 *head, UErrorCode &errorCode);
 214     int32_t addContextTrie(uint32_t defaultCE32, UCharsTrieBuilder &trieBuilder,
 215                            UErrorCode &errorCode);
 216
 217     void buildFastLatinTable(CollationData &data, UErrorCode &errorCode);
 218
 219     int32_t getCEs(const UnicodeString &s, int32_t start, int64_t ces[], int32_t cesLength);
 220
 221     static UChar32 jamoCpFromIndex(int32_t i) {
 222         // 0 <= i < CollationData::JAMO_CE32S_LENGTH = 19 + 21 + 27
 223         if(i < Hangul::JAMO_L_COUNT) { return Hangul::JAMO_L_BASE + i; }
 224         i -= Hangul::JAMO_L_COUNT;
 225         if(i < Hangul::JAMO_V_COUNT) { return Hangul::JAMO_V_BASE + i; }
 226         i -= Hangul::JAMO_V_COUNT;
 227         // i < 27
 228         return Hangul::JAMO_T_BASE + 1 + i;
 229     }
 230
 231     /** @see Collation::BUILDER_DATA_TAG */
 232     static const uint32_t IS_BUILDER_JAMO_CE32 = 0x100;
 233
 234     const Normalizer2Impl &nfcImpl;
 235     const CollationData *base;
 236     const CollationSettings *baseSettings;
 237     UTrie2 *trie;
 238     UVector32 ce32s;
 239     UVector64 ce64s;
 240     UVector conditionalCE32s;  // vector of ConditionalCE32
 241     // Characters that have context (prefixes or contraction suffixes).
 242     UnicodeSet contextChars;
 243     // Serialized UCharsTrie structures for finalized contexts.
 244     UnicodeString contexts;
 245     UnicodeSet unsafeBackwardSet;
 246     UBool modified;
 247
 248     UBool fastLatinEnabled;
 249     CollationFastLatinBuilder *fastLatinBuilder;
 250
 251     DataBuilderCollationIterator *collIter;
 252 };
 253
 254 U_NAMESPACE_END
 255
 256 #endif  // !UCONFIG_NO_COLLATION
 257 #endif  // __COLLATIONDATABUILDER_H__