1 /*
2 *******************************************************************************
3 * Copyright (C) 2010-2015, International Business Machines
4 * Corporation and others.  All Rights Reserved.
5 *******************************************************************************
6 * collationdata.h
7 *
8 * created on: 2010oct27
9 * created by: Markus W. Scherer
10 */
11 
12 #ifndef __COLLATIONDATA_H__
13 #define __COLLATIONDATA_H__
14 
15 #include "unicode/utypes.h"
16 
17 #if !UCONFIG_NO_COLLATION
18 
19 #include "unicode/ucol.h"
20 #include "unicode/uniset.h"
21 #include "collation.h"
22 #include "normalizer2impl.h"
23 #include "utrie2.h"
24 
25 struct UDataMemory;
26 
27 U_NAMESPACE_BEGIN
28 
29 class UVector32;
30 
31 /**
32  * Collation data container.
33  * Immutable data created by a CollationDataBuilder, or loaded from a file,
34  * or deserialized from API-provided binary data.
35  *
36  * Includes data for the collation base (root/default), aliased if this is not the base.
37  */
38 struct U_I18N_API CollationData : public UMemory {
39     // Note: The ucadata.icu loader could discover the reserved ranges by setting an array
40     // parallel with the ranges, and resetting ranges that are indexed.
41     // The reordering builder code could clone the resulting template array.
42     enum {
43         REORDER_RESERVED_BEFORE_LATIN = UCOL_REORDER_CODE_FIRST + 14,
44         REORDER_RESERVED_AFTER_LATIN
45     };
46 
47     enum {
48         MAX_NUM_SPECIAL_REORDER_CODES = 8,
49         /** C++ only, data reader check scriptStartsLength. */
50         MAX_NUM_SCRIPT_RANGES = 256
51     };
52 
CollationDataCollationData53     CollationData(const Normalizer2Impl &nfc)
54             : trie(NULL),
55               ce32s(NULL), ces(NULL), contexts(NULL), base(NULL),
56               jamoCE32s(NULL),
57               nfcImpl(nfc),
58               numericPrimary(0x12000000),
59               ce32sLength(0), cesLength(0), contextsLength(0),
60               compressibleBytes(NULL),
61               unsafeBackwardSet(NULL),
62               fastLatinTable(NULL), fastLatinTableLength(0),
63               numScripts(0), scriptsIndex(NULL), scriptStarts(NULL), scriptStartsLength(0),
64               rootElements(NULL), rootElementsLength(0) {}
65 
getCE32CollationData66     uint32_t getCE32(UChar32 c) const {
67         return UTRIE2_GET32(trie, c);
68     }
69 
getCE32FromSupplementaryCollationData70     uint32_t getCE32FromSupplementary(UChar32 c) const {
71         return UTRIE2_GET32_FROM_SUPP(trie, c);
72     }
73 
isDigitCollationData74     UBool isDigit(UChar32 c) const {
75         return c < 0x660 ? c <= 0x39 && 0x30 <= c :
76                 Collation::hasCE32Tag(getCE32(c), Collation::DIGIT_TAG);
77     }
78 
isUnsafeBackwardCollationData79     UBool isUnsafeBackward(UChar32 c, UBool numeric) const {
80         return unsafeBackwardSet->contains(c) || (numeric && isDigit(c));
81     }
82 
isCompressibleLeadByteCollationData83     UBool isCompressibleLeadByte(uint32_t b) const {
84         return compressibleBytes[b];
85     }
86 
isCompressiblePrimaryCollationData87     inline UBool isCompressiblePrimary(uint32_t p) const {
88         return isCompressibleLeadByte(p >> 24);
89     }
90 
91     /**
92      * Returns the CE32 from two contexts words.
93      * Access to the defaultCE32 for contraction and prefix matching.
94      */
readCE32CollationData95     static uint32_t readCE32(const UChar *p) {
96         return ((uint32_t)p[0] << 16) | p[1];
97     }
98 
99     /**
100      * Returns the CE32 for an indirect special CE32 (e.g., with DIGIT_TAG).
101      * Requires that ce32 is special.
102      */
103     uint32_t getIndirectCE32(uint32_t ce32) const;
104     /**
105      * Returns the CE32 for an indirect special CE32 (e.g., with DIGIT_TAG),
106      * if ce32 is special.
107      */
108     uint32_t getFinalCE32(uint32_t ce32) const;
109 
110     /**
111      * Computes a CE from c's ce32 which has the OFFSET_TAG.
112      */
getCEFromOffsetCE32CollationData113     int64_t getCEFromOffsetCE32(UChar32 c, uint32_t ce32) const {
114         int64_t dataCE = ces[Collation::indexFromCE32(ce32)];
115         return Collation::makeCE(Collation::getThreeBytePrimaryForOffsetData(c, dataCE));
116     }
117 
118     /**
119      * Returns the single CE that c maps to.
120      * Sets U_UNSUPPORTED_ERROR if c does not map to a single CE.
121      */
122     int64_t getSingleCE(UChar32 c, UErrorCode &errorCode) const;
123 
124     /**
125      * Returns the FCD16 value for code point c. c must be >= 0.
126      */
getFCD16CollationData127     uint16_t getFCD16(UChar32 c) const {
128         return nfcImpl.getFCD16(c);
129     }
130 
131     /**
132      * Returns the first primary for the script's reordering group.
133      * @return the primary with only the first primary lead byte of the group
134      *         (not necessarily an actual root collator primary weight),
135      *         or 0 if the script is unknown
136      */
137     uint32_t getFirstPrimaryForGroup(int32_t script) const;
138 
139     /**
140      * Returns the last primary for the script's reordering group.
141      * @return the last primary of the group
142      *         (not an actual root collator primary weight),
143      *         or 0 if the script is unknown
144      */
145     uint32_t getLastPrimaryForGroup(int32_t script) const;
146 
147     /**
148      * Finds the reordering group which contains the primary weight.
149      * @return the first script of the group, or -1 if the weight is beyond the last group
150      */
151     int32_t getGroupForPrimary(uint32_t p) const;
152 
153     int32_t getEquivalentScripts(int32_t script,
154                                  int32_t dest[], int32_t capacity, UErrorCode &errorCode) const;
155 
156     /**
157      * Writes the permutation of primary-weight ranges
158      * for the given reordering of scripts and groups.
159      * The caller checks for illegal arguments and
160      * takes care of [DEFAULT] and memory allocation.
161      *
162      * Each list element will be a (limit, offset) pair as described
163      * for the CollationSettings::reorderRanges.
164      * The list will be empty if no ranges are reordered.
165      */
166     void makeReorderRanges(const int32_t *reorder, int32_t length,
167                            UVector32 &ranges, UErrorCode &errorCode) const;
168 
169     /** @see jamoCE32s */
170     static const int32_t JAMO_CE32S_LENGTH = 19 + 21 + 27;
171 
172     /** Main lookup trie. */
173     const UTrie2 *trie;
174     /**
175      * Array of CE32 values.
176      * At index 0 there must be CE32(U+0000)
177      * to support U+0000's special-tag for NUL-termination handling.
178      */
179     const uint32_t *ce32s;
180     /** Array of CE values for expansions and OFFSET_TAG. */
181     const int64_t *ces;
182     /** Array of prefix and contraction-suffix matching data. */
183     const UChar *contexts;
184     /** Base collation data, or NULL if this data itself is a base. */
185     const CollationData *base;
186     /**
187      * Simple array of JAMO_CE32S_LENGTH=19+21+27 CE32s, one per canonical Jamo L/V/T.
188      * They are normally simple CE32s, rarely expansions.
189      * For fast handling of HANGUL_TAG.
190      */
191     const uint32_t *jamoCE32s;
192     const Normalizer2Impl &nfcImpl;
193     /** The single-byte primary weight (xx000000) for numeric collation. */
194     uint32_t numericPrimary;
195 
196     int32_t ce32sLength;
197     int32_t cesLength;
198     int32_t contextsLength;
199 
200     /** 256 flags for which primary-weight lead bytes are compressible. */
201     const UBool *compressibleBytes;
202     /**
203      * Set of code points that are unsafe for starting string comparison after an identical prefix,
204      * or in backwards CE iteration.
205      */
206     const UnicodeSet *unsafeBackwardSet;
207 
208     /**
209      * Fast Latin table for common-Latin-text string comparisons.
210      * Data structure see class CollationFastLatin.
211      */
212     const uint16_t *fastLatinTable;
213     int32_t fastLatinTableLength;
214 
215     /**
216      * Data for scripts and reordering groups.
217      * Uses include building a reordering permutation table and
218      * providing script boundaries to AlphabeticIndex.
219      */
220     int32_t numScripts;
221     /**
222      * The length of scriptsIndex is numScripts+16.
223      * It maps from a UScriptCode or a special reorder code to an entry in scriptStarts.
224      * 16 special reorder codes (not all used) are mapped starting at numScripts.
225      * Up to MAX_NUM_SPECIAL_REORDER_CODES are codes for special groups like space/punct/digit.
226      * There are special codes at the end for reorder-reserved primary ranges.
227      *
228      * Multiple scripts may share a range and index, for example Hira & Kana.
229      */
230     const uint16_t *scriptsIndex;
231     /**
232      * Start primary weight (top 16 bits only) for a group/script/reserved range
233      * indexed by scriptsIndex.
234      * The first range (separators & terminators) and the last range (trailing weights)
235      * are not reorderable, and no scriptsIndex entry points to them.
236      */
237     const uint16_t *scriptStarts;
238     int32_t scriptStartsLength;
239 
240     /**
241      * Collation elements in the root collator.
242      * Used by the CollationRootElements class. The data structure is described there.
243      * NULL in a tailoring.
244      */
245     const uint32_t *rootElements;
246     int32_t rootElementsLength;
247 
248 private:
249     int32_t getScriptIndex(int32_t script) const;
250     void makeReorderRanges(const int32_t *reorder, int32_t length,
251                            UBool latinMustMove,
252                            UVector32 &ranges, UErrorCode &errorCode) const;
253     int32_t addLowScriptRange(uint8_t table[], int32_t index, int32_t lowStart) const;
254     int32_t addHighScriptRange(uint8_t table[], int32_t index, int32_t highLimit) const;
255 };
256 
257 U_NAMESPACE_END
258 
259 #endif  // !UCONFIG_NO_COLLATION
260 #endif  // __COLLATIONDATA_H__
261