2 *******************************************************************************
\r
3 * Copyright (C) 1996-2010, International Business Machines Corporation and *
\r
4 * others. All Rights Reserved. *
\r
5 *******************************************************************************
\r
8 package com.ibm.icu.impl;
\r
10 import java.io.BufferedInputStream;
\r
11 import java.io.IOException;
\r
12 import java.io.InputStream;
\r
13 import java.util.MissingResourceException;
\r
15 import com.ibm.icu.lang.UCharacter;
\r
16 import com.ibm.icu.lang.UCharacterCategory;
\r
17 import com.ibm.icu.lang.UProperty;
\r
18 import com.ibm.icu.text.UTF16;
\r
19 import com.ibm.icu.text.UnicodeSet;
\r
20 import com.ibm.icu.util.RangeValueIterator;
\r
21 import com.ibm.icu.util.VersionInfo;
\r
24 * <p>Internal class used for Unicode character property database.</p>
\r
25 * <p>This classes store binary data read from uprops.icu.
\r
26 * It does not have the capability to parse the data into more high-level
\r
27 * information. It only returns bytes of information when required.</p>
\r
28 * <p>Due to the form most commonly used for retrieval, array of char is used
\r
29 * to store the binary data.</p>
\r
30 * <p>UCharacterPropertyDB also contains information on accessing indexes to
\r
31 * significant points in the binary data.</p>
\r
32 * <p>Responsibility for molding the binary data into more meaning form lies on
\r
33 * <a href=UCharacter.html>UCharacter</a>.</p>
\r
34 * @author Syn Wee Quek
\r
35 * @since release 2.1, february 1st 2002
\r
38 public final class UCharacterProperty
\r
40 // public data members -----------------------------------------------
\r
43 * public singleton instance
\r
45 public static final UCharacterProperty INSTANCE;
\r
49 INSTANCE = new UCharacterProperty();
\r
51 catch (IOException e) {
\r
52 throw new MissingResourceException(e.getMessage(),"","");
\r
59 public CharTrie m_trie_;
\r
62 * CharTrie index array
\r
64 public char[] m_trieIndex_;
\r
67 * CharTrie data array
\r
69 public char[] m_trieData_;
\r
72 * CharTrie data offset
\r
74 public int m_trieInitialValue_;
\r
78 public VersionInfo m_unicodeVersion_;
\r
80 * Latin capital letter i with dot above
\r
82 public static final char LATIN_CAPITAL_LETTER_I_WITH_DOT_ABOVE_ = 0x130;
\r
84 * Latin small letter i with dot above
\r
86 public static final char LATIN_SMALL_LETTER_DOTLESS_I_ = 0x131;
\r
90 public static final char LATIN_SMALL_LETTER_I_ = 0x69;
\r
92 * Character type mask
\r
94 public static final int TYPE_MASK = 0x1F;
\r
96 // uprops.h enum UPropertySource --------------------------------------- ***
\r
98 /** No source, not a supported property. */
\r
99 public static final int SRC_NONE=0;
\r
100 /** From uchar.c/uprops.icu main trie */
\r
101 public static final int SRC_CHAR=1;
\r
102 /** From uchar.c/uprops.icu properties vectors trie */
\r
103 public static final int SRC_PROPSVEC=2;
\r
104 /** From unames.c/unames.icu */
\r
105 public static final int SRC_NAMES=3;
\r
106 /** From ucase.c/ucase.icu */
\r
107 public static final int SRC_CASE=4;
\r
108 /** From ubidi_props.c/ubidi.icu */
\r
109 public static final int SRC_BIDI=5;
\r
110 /** From uchar.c/uprops.icu main trie as well as properties vectors trie */
\r
111 public static final int SRC_CHAR_AND_PROPSVEC=6;
\r
112 /** From ucase.c/ucase.icu as well as unorm.cpp/unorm.icu */
\r
113 public static final int SRC_CASE_AND_NORM=7;
\r
114 /** From normalizer2impl.cpp/nfc.nrm */
\r
115 public static final int SRC_NFC=8;
\r
116 /** From normalizer2impl.cpp/nfkc.nrm */
\r
117 public static final int SRC_NFKC=9;
\r
118 /** From normalizer2impl.cpp/nfkc_cf.nrm */
\r
119 public static final int SRC_NFKC_CF=10;
\r
120 /** From normalizer2impl.cpp/nfc.nrm canonical iterator data */
\r
121 public static final int SRC_NFC_CANON_ITER=11;
\r
122 /** One more than the highest UPropertySource (SRC_) constant. */
\r
123 public static final int SRC_COUNT=12;
\r
125 // public methods ----------------------------------------------------
\r
128 * Java friends implementation
\r
130 public void setIndexData(CharTrie.FriendAgent friendagent)
\r
132 m_trieIndex_ = friendagent.getPrivateIndex();
\r
133 m_trieData_ = friendagent.getPrivateData();
\r
134 m_trieInitialValue_ = friendagent.getPrivateInitialValue();
\r
138 * Gets the property value at the index.
\r
139 * This is optimized.
\r
140 * Note this is a little different from CharTrie the index m_trieData_
\r
141 * is never negative.
\r
142 * @param ch code point whose property value is to be retrieved
\r
143 * @return property value of code point
\r
145 public final int getProperty(int ch)
\r
147 if (ch < UTF16.LEAD_SURROGATE_MIN_VALUE
\r
148 || (ch > UTF16.LEAD_SURROGATE_MAX_VALUE
\r
149 && ch < UTF16.SUPPLEMENTARY_MIN_VALUE)) {
\r
150 // BMP codepoint 0000..D7FF or DC00..FFFF
\r
152 try { // using try for ch < 0 is faster than using an if statement
\r
153 return m_trieData_[
\r
154 (m_trieIndex_[ch >> Trie.INDEX_STAGE_1_SHIFT_]
\r
155 << Trie.INDEX_STAGE_2_SHIFT_)
\r
156 + (ch & Trie.INDEX_STAGE_3_MASK_)];
\r
157 } catch (ArrayIndexOutOfBoundsException e) {
\r
158 return m_trieInitialValue_;
\r
161 if (ch <= UTF16.LEAD_SURROGATE_MAX_VALUE) {
\r
162 // lead surrogate D800..DBFF
\r
163 return m_trieData_[
\r
164 (m_trieIndex_[Trie.LEAD_INDEX_OFFSET_
\r
165 + (ch >> Trie.INDEX_STAGE_1_SHIFT_)]
\r
166 << Trie.INDEX_STAGE_2_SHIFT_)
\r
167 + (ch & Trie.INDEX_STAGE_3_MASK_)];
\r
169 if (ch <= UTF16.CODEPOINT_MAX_VALUE) {
\r
170 // supplementary code point 10000..10FFFF
\r
171 // look at the construction of supplementary characters
\r
172 // trail forms the ends of it.
\r
173 return m_trie_.getSurrogateValue(
\r
174 UTF16.getLeadSurrogate(ch),
\r
175 (char)(ch & Trie.SURROGATE_MASK_));
\r
177 // ch is out of bounds
\r
178 // return m_dataOffset_ if there is an error, in this case we return
\r
179 // the default value: m_initialValue_
\r
180 // we cannot assume that m_initialValue_ is at offset 0
\r
181 // this is for optimization.
\r
182 return m_trieInitialValue_;
\r
184 // this all is an inlined form of return m_trie_.getCodePointValue(ch);
\r
188 * Gets the unicode additional properties.
\r
189 * C version getUnicodeProperties.
\r
190 * @param codepoint codepoint whose additional properties is to be
\r
192 * @param column The column index.
\r
193 * @return unicode properties
\r
195 public int getAdditional(int codepoint, int column) {
\r
196 if (column == -1) {
\r
197 return getProperty(codepoint);
\r
199 if (column < 0 || column >= m_additionalColumnsCount_) {
\r
202 return m_additionalVectors_[
\r
203 m_additionalTrie_.getCodePointValue(codepoint) + column];
\r
206 static final int MY_MASK = UCharacterProperty.TYPE_MASK
\r
207 & ((1<<UCharacterCategory.UPPERCASE_LETTER) |
\r
208 (1<<UCharacterCategory.LOWERCASE_LETTER) |
\r
209 (1<<UCharacterCategory.TITLECASE_LETTER) |
\r
210 (1<<UCharacterCategory.MODIFIER_LETTER) |
\r
211 (1<<UCharacterCategory.OTHER_LETTER));
\r
215 * <p>Get the "age" of the code point.</p>
\r
216 * <p>The "age" is the Unicode version when the code point was first
\r
217 * designated (as a non-character or for Private Use) or assigned a
\r
219 * <p>This can be useful to avoid emitting code points to receiving
\r
220 * processes that do not accept newer characters.</p>
\r
221 * <p>The data is from the UCD file DerivedAge.txt.</p>
\r
222 * <p>This API does not check the validity of the codepoint.</p>
\r
223 * @param codepoint The code point.
\r
224 * @return the Unicode version number
\r
226 public VersionInfo getAge(int codepoint)
\r
228 int version = getAdditional(codepoint, 0) >> AGE_SHIFT_;
\r
229 return VersionInfo.getInstance(
\r
230 (version >> FIRST_NIBBLE_SHIFT_) & LAST_NIBBLE_MASK_,
\r
231 version & LAST_NIBBLE_MASK_, 0, 0);
\r
234 private static final int GC_CN_MASK = getMask(UCharacter.UNASSIGNED);
\r
235 private static final int GC_CC_MASK = getMask(UCharacter.CONTROL);
\r
236 private static final int GC_CS_MASK = getMask(UCharacter.SURROGATE);
\r
237 private static final int GC_ZS_MASK = getMask(UCharacter.SPACE_SEPARATOR);
\r
238 private static final int GC_ZL_MASK = getMask(UCharacter.LINE_SEPARATOR);
\r
239 private static final int GC_ZP_MASK = getMask(UCharacter.PARAGRAPH_SEPARATOR);
\r
240 /** Mask constant for multiple UCharCategory bits (Z Separators). */
\r
241 private static final int GC_Z_MASK = GC_ZS_MASK|GC_ZL_MASK|GC_ZP_MASK;
\r
244 * Checks if c is in
\r
245 * [^\p{space}\p{gc=Control}\p{gc=Surrogate}\p{gc=Unassigned}]
\r
246 * with space=\p{Whitespace} and Control=Cc.
\r
247 * Implements UCHAR_POSIX_GRAPH.
\r
250 private static final boolean isgraphPOSIX(int c) {
\r
251 /* \p{space}\p{gc=Control} == \p{gc=Z}\p{Control} */
\r
252 /* comparing ==0 returns FALSE for the categories mentioned */
\r
253 return (getMask(UCharacter.getType(c))&
\r
254 (GC_CC_MASK|GC_CS_MASK|GC_CN_MASK|GC_Z_MASK))
\r
258 private static final class BinaryProperties{
\r
261 public BinaryProperties(int column, int mask) {
\r
262 this.column = column;
\r
266 BinaryProperties[] binProps={
\r
268 * column and mask values for binary properties from u_getUnicodeProperties().
\r
269 * Must be in order of corresponding UProperty,
\r
270 * and there must be exactly one entry per binary UProperty.
\r
272 new BinaryProperties( 1, ( 1 << ALPHABETIC_PROPERTY_) ),
\r
273 new BinaryProperties( 1, ( 1 << ASCII_HEX_DIGIT_PROPERTY_) ),
\r
274 new BinaryProperties( SRC_BIDI, 0 ), /* UCHAR_BIDI_CONTROL */
\r
275 new BinaryProperties( SRC_BIDI, 0 ), /* UCHAR_BIDI_MIRRORED */
\r
276 new BinaryProperties( 1, ( 1 << DASH_PROPERTY_) ),
\r
277 new BinaryProperties( 1, ( 1 << DEFAULT_IGNORABLE_CODE_POINT_PROPERTY_) ),
\r
278 new BinaryProperties( 1, ( 1 << DEPRECATED_PROPERTY_) ),
\r
279 new BinaryProperties( 1, ( 1 << DIACRITIC_PROPERTY_) ),
\r
280 new BinaryProperties( 1, ( 1 << EXTENDER_PROPERTY_) ),
\r
281 new BinaryProperties( SRC_NFC, 0 ), /* UCHAR_FULL_COMPOSITION_EXCLUSION */
\r
282 new BinaryProperties( 1, ( 1 << GRAPHEME_BASE_PROPERTY_) ),
\r
283 new BinaryProperties( 1, ( 1 << GRAPHEME_EXTEND_PROPERTY_) ),
\r
284 new BinaryProperties( 1, ( 1 << GRAPHEME_LINK_PROPERTY_) ),
\r
285 new BinaryProperties( 1, ( 1 << HEX_DIGIT_PROPERTY_) ),
\r
286 new BinaryProperties( 1, ( 1 << HYPHEN_PROPERTY_) ),
\r
287 new BinaryProperties( 1, ( 1 << ID_CONTINUE_PROPERTY_) ),
\r
288 new BinaryProperties( 1, ( 1 << ID_START_PROPERTY_) ),
\r
289 new BinaryProperties( 1, ( 1 << IDEOGRAPHIC_PROPERTY_) ),
\r
290 new BinaryProperties( 1, ( 1 << IDS_BINARY_OPERATOR_PROPERTY_) ),
\r
291 new BinaryProperties( 1, ( 1 << IDS_TRINARY_OPERATOR_PROPERTY_) ),
\r
292 new BinaryProperties( SRC_BIDI, 0 ), /* UCHAR_JOIN_CONTROL */
\r
293 new BinaryProperties( 1, ( 1 << LOGICAL_ORDER_EXCEPTION_PROPERTY_) ),
\r
294 new BinaryProperties( SRC_CASE, 0 ), /* UCHAR_LOWERCASE */
\r
295 new BinaryProperties( 1, ( 1 << MATH_PROPERTY_) ),
\r
296 new BinaryProperties( 1, ( 1 << NONCHARACTER_CODE_POINT_PROPERTY_) ),
\r
297 new BinaryProperties( 1, ( 1 << QUOTATION_MARK_PROPERTY_) ),
\r
298 new BinaryProperties( 1, ( 1 << RADICAL_PROPERTY_) ),
\r
299 new BinaryProperties( SRC_CASE, 0 ), /* UCHAR_SOFT_DOTTED */
\r
300 new BinaryProperties( 1, ( 1 << TERMINAL_PUNCTUATION_PROPERTY_) ),
\r
301 new BinaryProperties( 1, ( 1 << UNIFIED_IDEOGRAPH_PROPERTY_) ),
\r
302 new BinaryProperties( SRC_CASE, 0 ), /* UCHAR_UPPERCASE */
\r
303 new BinaryProperties( 1, ( 1 << WHITE_SPACE_PROPERTY_) ),
\r
304 new BinaryProperties( 1, ( 1 << XID_CONTINUE_PROPERTY_) ),
\r
305 new BinaryProperties( 1, ( 1 << XID_START_PROPERTY_) ),
\r
306 new BinaryProperties( SRC_CASE, 0 ), /* UCHAR_CASE_SENSITIVE */
\r
307 new BinaryProperties( 1, ( 1 << S_TERM_PROPERTY_) ),
\r
308 new BinaryProperties( 1, ( 1 << VARIATION_SELECTOR_PROPERTY_) ),
\r
309 new BinaryProperties( SRC_NFC, 0 ), /* UCHAR_NFD_INERT */
\r
310 new BinaryProperties( SRC_NFKC, 0 ), /* UCHAR_NFKD_INERT */
\r
311 new BinaryProperties( SRC_NFC, 0 ), /* UCHAR_NFC_INERT */
\r
312 new BinaryProperties( SRC_NFKC, 0 ), /* UCHAR_NFKC_INERT */
\r
313 new BinaryProperties( SRC_NFC_CANON_ITER, 0 ), /* UCHAR_SEGMENT_STARTER */
\r
314 new BinaryProperties( 1, ( 1 << PATTERN_SYNTAX) ),
\r
315 new BinaryProperties( 1, ( 1 << PATTERN_WHITE_SPACE) ),
\r
316 new BinaryProperties( SRC_CHAR_AND_PROPSVEC, 0 ), /* UCHAR_POSIX_ALNUM */
\r
317 new BinaryProperties( SRC_CHAR, 0 ), /* UCHAR_POSIX_BLANK */
\r
318 new BinaryProperties( SRC_CHAR, 0 ), /* UCHAR_POSIX_GRAPH */
\r
319 new BinaryProperties( SRC_CHAR, 0 ), /* UCHAR_POSIX_PRINT */
\r
320 new BinaryProperties( SRC_CHAR, 0 ), /* UCHAR_POSIX_XDIGIT */
\r
321 new BinaryProperties( SRC_CASE, 0 ), /* UCHAR_CASED */
\r
322 new BinaryProperties( SRC_CASE, 0 ), /* UCHAR_CASE_IGNORABLE */
\r
323 new BinaryProperties( SRC_CASE, 0 ), /* UCHAR_CHANGES_WHEN_LOWERCASED */
\r
324 new BinaryProperties( SRC_CASE, 0 ), /* UCHAR_CHANGES_WHEN_UPPERCASED */
\r
325 new BinaryProperties( SRC_CASE, 0 ), /* UCHAR_CHANGES_WHEN_TITLECASED */
\r
326 new BinaryProperties( SRC_CASE_AND_NORM, 0 ), /* UCHAR_CHANGES_WHEN_CASEFOLDED */
\r
327 new BinaryProperties( SRC_CASE, 0 ), /* UCHAR_CHANGES_WHEN_CASEMAPPED */
\r
328 new BinaryProperties( SRC_NFKC_CF, 0 ), /* UCHAR_CHANGES_WHEN_NFKC_CASEFOLDED */
\r
333 * <p>Check a binary Unicode property for a code point.</p>
\r
334 * <p>Unicode, especially in version 3.2, defines many more properties
\r
335 * than the original set in UnicodeData.txt.</p>
\r
336 * <p>This API is intended to reflect Unicode properties as defined in
\r
337 * the Unicode Character Database (UCD) and Unicode Technical Reports
\r
339 * <p>For details about the properties see
\r
340 * <a href=http://www.unicode.org/>http://www.unicode.org/</a>.</p>
\r
341 * <p>For names of Unicode properties see the UCD file
\r
342 * PropertyAliases.txt.</p>
\r
343 * <p>This API does not check the validity of the codepoint.</p>
\r
344 * <p>Important: If ICU is built with UCD files from Unicode versions
\r
345 * below 3.2, then properties marked with "new" are not or
\r
346 * not fully available.</p>
\r
347 * @param c Code point to test.
\r
348 * @param which selector constant from com.ibm.icu.lang.UProperty,
\r
349 * identifies which binary property to check.
\r
350 * @return true or false according to the binary Unicode property value
\r
351 * for ch. Also false if property is out of bounds or if the
\r
352 * Unicode version does not have data for the property at all, or
\r
353 * not for this code point.
\r
354 * @see com.ibm.icu.lang.UProperty
\r
357 public boolean hasBinaryProperty(int c, int which) {
\r
358 if(which<UProperty.BINARY_START || UProperty.BINARY_LIMIT<=which) {
\r
359 // not a known binary property
\r
362 int mask=binProps[which].mask;
\r
363 int column=binProps[which].column;
\r
365 // systematic, directly stored properties
\r
366 return (getAdditional(c, column) & mask)!=0;
\r
368 if(column==SRC_CASE) {
\r
369 /* case mapping properties */
\r
371 return UCaseProps.getSingleton().hasBinaryProperty(c, which);
\r
372 } catch (IOException e) {
\r
375 } else if(column==SRC_NFC) {
\r
376 /* normalization properties from nfc.nrm */
\r
378 case UProperty.FULL_COMPOSITION_EXCLUSION: {
\r
379 // By definition, Full_Composition_Exclusion is the same as NFC_QC=No.
\r
380 Normalizer2Impl impl=Norm2AllModes.getNFCInstance().impl;
\r
381 return impl.isCompNo(impl.getNorm16(c));
\r
384 // UCHAR_NF[CD]_INERT properties
\r
385 return Norm2AllModes.getN2WithImpl(which-UProperty.NFD_INERT).isInert(c);
\r
387 } else if(column==SRC_NFKC) {
\r
388 /* normalization properties from nfkc.nrm */
\r
389 // UCHAR_NFK[CD]_INERT properties
\r
390 return Norm2AllModes.getN2WithImpl(which-UProperty.NFD_INERT).isInert(c);
\r
391 } else if(column==SRC_NFKC_CF) {
\r
392 // currently only for UCHAR_CHANGES_WHEN_NFKC_CASEFOLDED
\r
393 Normalizer2Impl kcf=Norm2AllModes.getNFKC_CFInstance().impl;
\r
394 String src=UTF16.valueOf(c);
\r
395 StringBuilder dest=new StringBuilder();
\r
396 // Small destCapacity for NFKC_CF(c).
\r
397 Normalizer2Impl.ReorderingBuffer buffer=new Normalizer2Impl.ReorderingBuffer(kcf, dest, 5);
\r
398 kcf.compose(src, 0, src.length(), false, true, buffer);
\r
399 return !Normalizer2Impl.UTF16Plus.equal(dest, src);
\r
400 } else if(column==SRC_NFC_CANON_ITER) {
\r
401 /* normalization properties from nfc.nrm canonical iterator data */
\r
403 return Norm2AllModes.getNFCInstance().impl.
\r
404 ensureCanonIterData().isCanonSegmentStarter(c);
\r
405 } else if(column==SRC_BIDI) {
\r
406 /* bidi/shaping properties */
\r
409 bdp = UBiDiProps.getSingleton();
\r
410 } catch (IOException e) {
\r
414 case UProperty.BIDI_MIRRORED:
\r
415 return bdp.isMirrored(c);
\r
416 case UProperty.BIDI_CONTROL:
\r
417 return bdp.isBidiControl(c);
\r
418 case UProperty.JOIN_CONTROL:
\r
419 return bdp.isJoinControl(c);
\r
423 } else if(column==SRC_CHAR) {
\r
425 case UProperty.POSIX_BLANK:
\r
426 // "horizontal space"
\r
428 return c==9 || c==0x20; /* TAB or SPACE */
\r
431 return UCharacter.getType(c)==UCharacter.SPACE_SEPARATOR;
\r
433 case UProperty.POSIX_GRAPH:
\r
434 return isgraphPOSIX(c);
\r
435 case UProperty.POSIX_PRINT:
\r
437 * Checks if codepoint is in \p{graph}\p{blank} - \p{cntrl}.
\r
439 * The only cntrl character in graph+blank is TAB (in blank).
\r
440 * Here we implement (blank-TAB)=Zs instead of calling u_isblank().
\r
442 return (UCharacter.getType(c)==UCharacter.SPACE_SEPARATOR) || isgraphPOSIX(c);
\r
443 case UProperty.POSIX_XDIGIT:
\r
444 /* check ASCII and Fullwidth ASCII a-fA-F */
\r
446 (c<=0x66 && c>=0x41 && (c<=0x46 || c>=0x61)) ||
\r
447 (c>=0xff21 && c<=0xff46 && (c<=0xff26 || c>=0xff41))
\r
452 return UCharacter.getType(c)==UCharacter.DECIMAL_DIGIT_NUMBER;
\r
456 } else if(column==SRC_CHAR_AND_PROPSVEC) {
\r
458 case UProperty.POSIX_ALNUM:
\r
459 return UCharacter.isUAlphabetic(c) || UCharacter.isDigit(c);
\r
463 } else if(column==SRC_CASE_AND_NORM) {
\r
466 case UProperty.CHANGES_WHEN_CASEFOLDED:
\r
467 nfd=Norm2AllModes.getNFCInstance().impl.getDecomposition(c);
\r
469 /* c has a decomposition */
\r
470 c=nfd.codePointAt(0);
\r
471 if(Character.charCount(c)!=nfd.length()) {
\r
472 /* multiple code points */
\r
476 return false; /* protect against bad input */
\r
479 /* single code point */
\r
481 UCaseProps csp=UCaseProps.getSingleton();
\r
482 UCaseProps.dummyStringBuffer.setLength(0);
\r
483 return csp.toFullFolding(c, UCaseProps.dummyStringBuffer,
\r
484 UCharacter.FOLD_CASE_DEFAULT)>=0;
\r
485 } catch (IOException e) {
\r
489 String folded=UCharacter.foldCase(nfd, true);
\r
490 return !folded.equals(nfd);
\r
501 public final int getSource(int which) {
\r
502 if(which<UProperty.BINARY_START) {
\r
503 return SRC_NONE; /* undefined */
\r
504 } else if(which<UProperty.BINARY_LIMIT) {
\r
505 if(binProps[which].mask!=0) {
\r
506 return SRC_PROPSVEC;
\r
508 return binProps[which].column;
\r
510 } else if(which<UProperty.INT_START) {
\r
511 return SRC_NONE; /* undefined */
\r
512 } else if(which<UProperty.INT_LIMIT) {
\r
514 case UProperty.GENERAL_CATEGORY:
\r
515 case UProperty.NUMERIC_TYPE:
\r
518 case UProperty.CANONICAL_COMBINING_CLASS:
\r
519 case UProperty.NFD_QUICK_CHECK:
\r
520 case UProperty.NFC_QUICK_CHECK:
\r
521 case UProperty.LEAD_CANONICAL_COMBINING_CLASS:
\r
522 case UProperty.TRAIL_CANONICAL_COMBINING_CLASS:
\r
524 case UProperty.NFKD_QUICK_CHECK:
\r
525 case UProperty.NFKC_QUICK_CHECK:
\r
528 case UProperty.BIDI_CLASS:
\r
529 case UProperty.JOINING_GROUP:
\r
530 case UProperty.JOINING_TYPE:
\r
534 return SRC_PROPSVEC;
\r
536 } else if(which<UProperty.STRING_START) {
\r
538 case UProperty.GENERAL_CATEGORY_MASK:
\r
539 case UProperty.NUMERIC_VALUE:
\r
545 } else if(which<UProperty.STRING_LIMIT) {
\r
547 case UProperty.AGE:
\r
548 return SRC_PROPSVEC;
\r
550 case UProperty.BIDI_MIRRORING_GLYPH:
\r
553 case UProperty.CASE_FOLDING:
\r
554 case UProperty.LOWERCASE_MAPPING:
\r
555 case UProperty.SIMPLE_CASE_FOLDING:
\r
556 case UProperty.SIMPLE_LOWERCASE_MAPPING:
\r
557 case UProperty.SIMPLE_TITLECASE_MAPPING:
\r
558 case UProperty.SIMPLE_UPPERCASE_MAPPING:
\r
559 case UProperty.TITLECASE_MAPPING:
\r
560 case UProperty.UPPERCASE_MAPPING:
\r
563 case UProperty.ISO_COMMENT:
\r
564 case UProperty.NAME:
\r
565 case UProperty.UNICODE_1_NAME:
\r
572 return SRC_NONE; /* undefined */
\r
577 * Forms a supplementary code point from the argument character<br>
\r
578 * Note this is for internal use hence no checks for the validity of the
\r
579 * surrogate characters are done
\r
580 * @param lead lead surrogate character
\r
581 * @param trail trailing surrogate character
\r
582 * @return code point of the supplementary character
\r
584 public static int getRawSupplementary(char lead, char trail)
\r
586 return (lead << LEAD_SURROGATE_SHIFT_) + trail + SURROGATE_OFFSET_;
\r
591 * Unicode property names and property value names are compared
\r
592 * "loosely". Property[Value]Aliases.txt say:
\r
594 * "With loose matching of property names, the case distinctions,
\r
595 * whitespace, and '_' are ignored."
\r
599 * This function does just that, for ASCII (char *) name strings.
\r
600 * It is almost identical to ucnv_compareNames() but also ignores
\r
601 * ASCII White_Space characters (U+0009..U+000d).
\r
603 * @param name1 name to compare
\r
604 * @param name2 name to compare
\r
605 * @return 0 if names are equal, < 0 if name1 is less than name2 and > 0
\r
606 * if name1 is greater than name2.
\r
608 /* to be implemented in 2.4
\r
609 * public static int comparePropertyNames(String name1, String name2)
\r
617 // Ignore delimiters '-', '_', and ASCII White_Space
\r
618 if (i1 < name1.length()) {
\r
619 ch1 = name1.charAt(i1 ++);
\r
621 while (ch1 == '-' || ch1 == '_' || ch1 == ' ' || ch1 == '\t'
\r
622 || ch1 == '\n' // synwee what is || ch1 == '\v'
\r
623 || ch1 == '\f' || ch1=='\r') {
\r
624 if (i1 < name1.length()) {
\r
625 ch1 = name1.charAt(i1 ++);
\r
631 if (i2 < name2.length()) {
\r
632 ch2 = name2.charAt(i2 ++);
\r
634 while (ch2 == '-' || ch2 == '_' || ch2 == ' ' || ch2 == '\t'
\r
635 || ch2 == '\n' // synwee what is || ch1 == '\v'
\r
636 || ch2 == '\f' || ch2=='\r') {
\r
637 if (i2 < name2.length()) {
\r
638 ch2 = name2.charAt(i2 ++);
\r
645 // If we reach the ends of both strings then they match
\r
646 if (ch1 == 0 && ch2 == 0) {
\r
650 // Case-insensitive comparison
\r
652 result = Character.toLowerCase(ch1)
\r
653 - Character.toLowerCase(ch2);
\r
663 * Checks if the argument c is to be treated as a white space in ICU
\r
664 * rules. Usually ICU rule white spaces are ignored unless quoted.
\r
665 * Equivalent to test for Pattern_White_Space Unicode property.
\r
666 * Stable set of characters, won't change.
\r
667 * See UAX #31 Identifier and Pattern Syntax: http://www.unicode.org/reports/tr31/
\r
668 * @param c codepoint to check
\r
669 * @return true if c is a ICU white space
\r
671 public static boolean isRuleWhiteSpace(int c)
\r
673 /* "white space" in the sense of ICU rule parsers
\r
674 This is a FIXED LIST that is NOT DEPENDENT ON UNICODE PROPERTIES.
\r
675 See UAX #31 Identifier and Pattern Syntax: http://www.unicode.org/reports/tr31/
\r
676 U+0009..U+000D, U+0020, U+0085, U+200E..U+200F, and U+2028..U+2029
\r
677 Equivalent to test for Pattern_White_Space Unicode property.
\r
679 return (c >= 0x0009 && c <= 0x2029 &&
\r
680 (c <= 0x000D || c == 0x0020 || c == 0x0085 ||
\r
681 c == 0x200E || c == 0x200F || c >= 0x2028));
\r
685 * Get the the maximum values for some enum/int properties.
\r
686 * @return maximum values for the integer properties.
\r
688 public int getMaxValues(int column)
\r
690 // return m_maxBlockScriptValue_;
\r
694 return m_maxBlockScriptValue_;
\r
696 return m_maxJTGValue_;
\r
703 * Gets the type mask
\r
704 * @param type character type
\r
707 public static final int getMask(int type)
\r
712 // protected variables -----------------------------------------------
\r
715 * Extra property trie
\r
717 CharTrie m_additionalTrie_;
\r
719 * Extra property vectors, 1st column for age and second for binary
\r
722 int m_additionalVectors_[];
\r
724 * Number of additional columns
\r
726 int m_additionalColumnsCount_;
\r
728 * Maximum values for block, bits used as in vector word
\r
731 int m_maxBlockScriptValue_;
\r
733 * Maximum values for script, bits used as in vector word
\r
736 int m_maxJTGValue_;
\r
737 // private variables -------------------------------------------------
\r
740 * Default name of the datafile
\r
742 private static final String DATA_FILE_NAME_ = ICUResourceBundle.ICU_BUNDLE+"/uprops.icu";
\r
745 * Default buffer size of datafile
\r
747 private static final int DATA_BUFFER_SIZE_ = 25000;
\r
750 * Shift value for lead surrogate to form a supplementary character.
\r
752 private static final int LEAD_SURROGATE_SHIFT_ = 10;
\r
754 * Offset to add to combined surrogate pair to avoid msking.
\r
756 private static final int SURROGATE_OFFSET_ =
\r
757 UTF16.SUPPLEMENTARY_MIN_VALUE -
\r
758 (UTF16.SURROGATE_MIN_VALUE <<
\r
759 LEAD_SURROGATE_SHIFT_) -
\r
760 UTF16.TRAIL_SURROGATE_MIN_VALUE;
\r
763 // additional properties ----------------------------------------------
\r
766 * Additional properties used in internal trie data
\r
769 * Properties in vector word 1
\r
770 * Each bit encodes one binary property.
\r
771 * The following constants represent the bit number, use 1<<UPROPS_XYZ.
\r
772 * UPROPS_BINARY_1_TOP<=32!
\r
774 * Keep this list of property enums in sync with
\r
775 * propListNames[] in icu/source/tools/genprops/props2.c!
\r
777 * ICU 2.6/uprops format version 3.2 stores full properties instead of "Other_".
\r
779 private static final int WHITE_SPACE_PROPERTY_ = 0;
\r
780 private static final int DASH_PROPERTY_ = 1;
\r
781 private static final int HYPHEN_PROPERTY_ = 2;
\r
782 private static final int QUOTATION_MARK_PROPERTY_ = 3;
\r
783 private static final int TERMINAL_PUNCTUATION_PROPERTY_ = 4;
\r
784 private static final int MATH_PROPERTY_ = 5;
\r
785 private static final int HEX_DIGIT_PROPERTY_ = 6;
\r
786 private static final int ASCII_HEX_DIGIT_PROPERTY_ = 7;
\r
787 private static final int ALPHABETIC_PROPERTY_ = 8;
\r
788 private static final int IDEOGRAPHIC_PROPERTY_ = 9;
\r
789 private static final int DIACRITIC_PROPERTY_ = 10;
\r
790 private static final int EXTENDER_PROPERTY_ = 11;
\r
791 private static final int NONCHARACTER_CODE_POINT_PROPERTY_ = 12;
\r
792 private static final int GRAPHEME_EXTEND_PROPERTY_ = 13;
\r
793 private static final int GRAPHEME_LINK_PROPERTY_ = 14;
\r
794 private static final int IDS_BINARY_OPERATOR_PROPERTY_ = 15;
\r
795 private static final int IDS_TRINARY_OPERATOR_PROPERTY_ = 16;
\r
796 private static final int RADICAL_PROPERTY_ = 17;
\r
797 private static final int UNIFIED_IDEOGRAPH_PROPERTY_ = 18;
\r
798 private static final int DEFAULT_IGNORABLE_CODE_POINT_PROPERTY_ = 19;
\r
799 private static final int DEPRECATED_PROPERTY_ = 20;
\r
800 private static final int LOGICAL_ORDER_EXCEPTION_PROPERTY_ = 21;
\r
801 private static final int XID_START_PROPERTY_ = 22;
\r
802 private static final int XID_CONTINUE_PROPERTY_ = 23;
\r
803 private static final int ID_START_PROPERTY_ = 24;
\r
804 private static final int ID_CONTINUE_PROPERTY_ = 25;
\r
805 private static final int GRAPHEME_BASE_PROPERTY_ = 26;
\r
806 private static final int S_TERM_PROPERTY_ = 27;
\r
807 private static final int VARIATION_SELECTOR_PROPERTY_ = 28;
\r
808 private static final int PATTERN_SYNTAX = 29; /* new in ICU 3.4 and Unicode 4.1 */
\r
809 private static final int PATTERN_WHITE_SPACE = 30;
\r
812 * First nibble shift
\r
814 private static final int FIRST_NIBBLE_SHIFT_ = 0x4;
\r
816 * Second nibble mask
\r
818 private static final int LAST_NIBBLE_MASK_ = 0xF;
\r
822 private static final int AGE_SHIFT_ = 24;
\r
825 // private constructors --------------------------------------------------
\r
829 * @exception IOException thrown when data reading fails or data corrupted
\r
831 private UCharacterProperty() throws IOException
\r
834 InputStream is = ICUData.getRequiredStream(DATA_FILE_NAME_);
\r
835 BufferedInputStream b = new BufferedInputStream(is, DATA_BUFFER_SIZE_);
\r
836 UCharacterPropertyReader reader = new UCharacterPropertyReader(b);
\r
840 m_trie_.putIndexData(this);
\r
843 // private methods -------------------------------------------------------
\r
846 * Compare additional properties to see if it has argument type
\r
847 * @param property 32 bit properties
\r
848 * @param type character type
\r
849 * @return true if property has type
\r
851 /*private boolean compareAdditionalType(int property, int type)
\r
853 return (property & (1 << type)) != 0;
\r
856 // property starts for UnicodeSet -------------------------------------- ***
\r
858 private static final int TAB = 0x0009;
\r
859 //private static final int LF = 0x000a;
\r
860 //private static final int FF = 0x000c;
\r
861 private static final int CR = 0x000d;
\r
862 private static final int U_A = 0x0041;
\r
863 private static final int U_F = 0x0046;
\r
864 private static final int U_Z = 0x005a;
\r
865 private static final int U_a = 0x0061;
\r
866 private static final int U_f = 0x0066;
\r
867 private static final int U_z = 0x007a;
\r
868 private static final int DEL = 0x007f;
\r
869 private static final int NL = 0x0085;
\r
870 private static final int NBSP = 0x00a0;
\r
871 private static final int CGJ = 0x034f;
\r
872 private static final int FIGURESP= 0x2007;
\r
873 private static final int HAIRSP = 0x200a;
\r
874 //private static final int ZWNJ = 0x200c;
\r
875 //private static final int ZWJ = 0x200d;
\r
876 private static final int RLM = 0x200f;
\r
877 private static final int NNBSP = 0x202f;
\r
878 private static final int WJ = 0x2060;
\r
879 private static final int INHSWAP = 0x206a;
\r
880 private static final int NOMDIG = 0x206f;
\r
881 private static final int U_FW_A = 0xff21;
\r
882 private static final int U_FW_F = 0xff26;
\r
883 private static final int U_FW_Z = 0xff3a;
\r
884 private static final int U_FW_a = 0xff41;
\r
885 private static final int U_FW_f = 0xff46;
\r
886 private static final int U_FW_z = 0xff5a;
\r
887 private static final int ZWNBSP = 0xfeff;
\r
889 public UnicodeSet addPropertyStarts(UnicodeSet set) {
\r
890 /* add the start code point of each same-value range of the main trie */
\r
891 TrieIterator propsIter = new TrieIterator(m_trie_);
\r
892 RangeValueIterator.Element propsResult = new RangeValueIterator.Element();
\r
893 while(propsIter.next(propsResult)){
\r
894 set.add(propsResult.start);
\r
897 /* add code points with hardcoded properties, plus the ones following them */
\r
899 /* add for u_isblank() */
\r
903 /* add for IS_THAT_CONTROL_SPACE() */
\r
904 set.add(CR+1); /* range TAB..CR */
\r
910 /* add for u_isIDIgnorable() what was not added above */
\r
911 set.add(DEL); /* range DEL..NBSP-1, NBSP added below */
\r
919 /* add no-break spaces for u_isWhitespace() what was not added above */
\r
923 set.add(FIGURESP+1);
\r
927 /* add for u_charDigitValue() */
\r
928 // TODO remove when UCharacter.getHanNumericValue() is changed to just return
\r
929 // Unicode numeric values
\r
951 /* add for u_digit() */
\r
961 /* add for u_isxdigit() */
\r
967 /* add for UCHAR_DEFAULT_IGNORABLE_CODE_POINT what was not added above */
\r
968 set.add(WJ); /* range WJ..NOMDIG */
\r
972 set.add(0xe0fff+1);
\r
974 /* add for UCHAR_GRAPHEME_BASE and others */
\r
978 return set; // for chaining
\r
981 public void upropsvec_addPropertyStarts(UnicodeSet set) {
\r
982 /* add the start code point of each same-value range of the properties vectors trie */
\r
983 if(m_additionalColumnsCount_>0) {
\r
984 /* if m_additionalColumnsCount_==0 then the properties vectors trie may not be there at all */
\r
985 TrieIterator propsVectorsIter = new TrieIterator(m_additionalTrie_);
\r
986 RangeValueIterator.Element propsVectorsResult = new RangeValueIterator.Element();
\r
987 while(propsVectorsIter.next(propsVectorsResult)){
\r
988 set.add(propsVectorsResult.start);
\r
993 /*----------------------------------------------------------------
\r
995 *----------------------------------------------------------------*/
\r
998 * Return a set of characters for property enumeration.
\r
999 * The set implicitly contains 0x110000 as well, which is one more than the highest
\r
1000 * Unicode code point.
\r
1002 * This set is used as an ordered list - its code points are ordered, and
\r
1003 * consecutive code points (in Unicode code point order) in the set define a range.
\r
1004 * For each two consecutive characters (start, limit) in the set,
\r
1005 * all of the UCD/normalization and related properties for
\r
1006 * all code points start..limit-1 are all the same,
\r
1007 * except for character names and ISO comments.
\r
1009 * All Unicode code points U+0000..U+10ffff are covered by these ranges.
\r
1010 * The ranges define a partition of the Unicode code space.
\r
1011 * ICU uses the inclusions set to enumerate properties for generating
\r
1012 * UnicodeSets containing all code points that have a certain property value.
\r
1014 * The Inclusion List is generated from the UCD. It is generated
\r
1015 * by enumerating the data tries, and code points for hardcoded properties
\r
1016 * are added as well.
\r
1018 * --------------------------------------------------------------------------
\r
1020 * The following are ideas for getting properties-unique code point ranges,
\r
1021 * with possible optimizations beyond the current implementation.
\r
1022 * These optimizations would require more code and be more fragile.
\r
1023 * The current implementation generates one single list (set) for all properties.
\r
1025 * To enumerate properties efficiently, one needs to know ranges of
\r
1026 * repetitive values, so that the value of only each start code point
\r
1027 * can be applied to the whole range.
\r
1028 * This information is in principle available in the uprops.icu/unorm.icu data.
\r
1030 * There are two obstacles:
\r
1032 * 1. Some properties are computed from multiple data structures,
\r
1033 * making it necessary to get repetitive ranges by intersecting
\r
1034 * ranges from multiple tries.
\r
1036 * 2. It is not economical to write code for getting repetitive ranges
\r
1037 * that are precise for each of some 50 properties.
\r
1039 * Compromise ideas:
\r
1041 * - Get ranges per trie, not per individual property.
\r
1042 * Each range contains the same values for a whole group of properties.
\r
1043 * This would generate currently five range sets, two for uprops.icu tries
\r
1044 * and three for unorm.icu tries.
\r
1046 * - Combine sets of ranges for multiple tries to get sufficient sets
\r
1047 * for properties, e.g., the uprops.icu main and auxiliary tries
\r
1048 * for all non-normalization properties.
\r
1050 * Ideas for representing ranges and combining them:
\r
1052 * - A UnicodeSet could hold just the start code points of ranges.
\r
1053 * Multiple sets are easily combined by or-ing them together.
\r
1055 * - Alternatively, a UnicodeSet could hold each even-numbered range.
\r
1056 * All ranges could be enumerated by using each start code point
\r
1057 * (for the even-numbered ranges) as well as each limit (end+1) code point
\r
1058 * (for the odd-numbered ranges).
\r
1059 * It should be possible to combine two such sets by xor-ing them,
\r
1060 * but no more than two.
\r
1062 * The second way to represent ranges may(?!) yield smaller UnicodeSet arrays,
\r
1063 * but the first one is certainly simpler and applicable for combining more than
\r
1066 * It is possible to combine all range sets for all uprops/unorm tries into one
\r
1067 * set that can be used for all properties.
\r
1068 * As an optimization, there could be less-combined range sets for certain
\r
1069 * groups of properties.
\r
1070 * The relationship of which less-combined range set to use for which property
\r
1071 * depends on the implementation of the properties and must be hardcoded
\r
1072 * - somewhat error-prone and higher maintenance but can be tested easily
\r
1073 * by building property sets "the simple way" in test code.
\r
1077 * Do not use a UnicodeSet pattern because that causes infinite recursion;
\r
1078 * UnicodeSet depends on the inclusions set.
\r
1082 * getInclusions() is commented out starting 2005-feb-12 because
\r
1083 * UnicodeSet now calls the uxyz_addPropertyStarts() directly,
\r
1084 * and only for the relevant property source.
\r
1087 public UnicodeSet getInclusions() {
\r
1088 UnicodeSet set = new UnicodeSet();
\r
1089 NormalizerImpl.addPropertyStarts(set);
\r
1090 addPropertyStarts(set);
\r