1 /*
2  * Copyright (c) 2016, 2020, Oracle and/or its affiliates. All rights reserved.
3  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
4  *
5  * This code is free software; you can redistribute it and/or modify it
6  * under the terms of the GNU General Public License version 2 only, as
7  * published by the Free Software Foundation.
8  *
9  * This code is distributed in the hope that it will be useful, but WITHOUT
10  * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
11  * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
12  * version 2 for more details (a copy is included in the LICENSE file that
13  * accompanied this code).
14  *
15  * You should have received a copy of the GNU General Public License version
16  * 2 along with this work; if not, write to the Free Software Foundation,
17  * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
18  *
19  * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
20  * or visit www.oracle.com if you need additional information or have any
21  * questions.
22  */
23 
24 /*
25  * @test
26  * @bug 7071819 8221431 8239383
27  * @summary tests Unicode Extended Grapheme support
28  * @library /lib/testlibrary/java/lang
29  * @run main GraphemeTest
30  */
31 
32 import java.io.IOException;
33 import java.nio.file.Files;
34 import java.nio.file.Path;
35 import java.nio.file.Paths;
36 import java.util.Arrays;
37 import java.util.ArrayList;
38 import java.util.Scanner;
39 import java.util.regex.Pattern;
40 import java.util.regex.Matcher;
41 
42 public class GraphemeTest {
43 
main(String[] args)44     public static void main(String[] args) throws Throwable {
45         testProps(UCDFiles.GRAPHEME_BREAK_PROPERTY);
46         testProps(UCDFiles.EMOJI_DATA);
47     }
48 
testProps(Path path)49     private static void testProps(Path path) throws IOException {
50         Files.lines(path)
51             .map( ln -> ln.replaceFirst("#.*", "") )
52             .filter( ln -> ln.length() != 0 )
53             .forEach(ln -> {
54                     String[] strs = ln.split("\\s+");
55                     int off = strs[0].indexOf("..");
56                     int cp0, cp1;
57                     String expected = strs[2];
58                     if (off != -1) {
59                         cp0 = Integer.parseInt(strs[0], 0, off, 16);
60                         cp1 = Integer.parseInt(strs[0], off + 2, strs[0].length(), 16);
61                     } else {
62                         cp0 = cp1 = Integer.parseInt(strs[0], 16);
63                     }
64                     for (int cp = cp0; cp <=  cp1; cp++) {
65                         // Ignore Emoji* for now (only interested in Extended_Pictographic)
66                         if (expected.startsWith("Emoji")) {
67                             continue;
68                         }
69 
70                         // NOTE:
71                         // #tr29 "plus a few General_Category = Spacing_Mark needed for
72                         // canonical equivalence."
73                         // For "extended grapheme clusters" support, there is no
74                         // need actually to diff "extend" and "spackmark" given GB9, GB9a.
75                         if (!expected.equals(types[getType(cp)])) {
76                             if ("Extend".equals(expected) &&
77                                 "SpacingMark".equals(types[getType(cp)]))
78                                 System.out.printf("[%x]  [%s][%d] -> [%s]%n",
79                                     cp, expected, Character.getType(cp), types[getType(cp)]);
80                             else
81                                 throw new RuntimeException(String.format(
82                                     "cp=[%x], expeced:[%s] result:[%s]%n",
83                                     cp, expected, types[getType(cp)]));
84                         }
85                     }
86                 });
87     }
88 
89     private static final String[] types = {
90         "Other", "CR", "LF", "Control", "Extend", "ZWJ", "Regional_Indicator",
91         "Prepend", "SpacingMark",
92         "L", "V", "T", "LV", "LVT",
93         "Extended_Pictographic" };
94 
95     ///////////////////////////////////////////////////////////////////////////////////////////////////////////////
96 
97     // from java.util.regex.Grapheme.java
98     // types
99     private static final int OTHER = 0;
100     private static final int CR = 1;
101     private static final int LF = 2;
102     private static final int CONTROL = 3;
103     private static final int EXTEND = 4;
104     private static final int ZWJ = 5;
105     private static final int RI = 6;
106     private static final int PREPEND = 7;
107     private static final int SPACINGMARK = 8;
108     private static final int L = 9;
109     private static final int V = 10;
110     private static final int T = 11;
111     private static final int LV = 12;
112     private static final int LVT = 13;
113     private static final int EXTENDED_PICTOGRAPHIC = 14;
114 
115     private static final int FIRST_TYPE = 0;
116     private static final int LAST_TYPE = 14;
117 
118     private static boolean[][] rules;
119     static {
120         rules = new boolean[LAST_TYPE + 1][LAST_TYPE + 1];
121         // GB 999 Any + Any  -> default
122         for (int i = FIRST_TYPE; i <= LAST_TYPE; i++)
123             for (int j = FIRST_TYPE; j <= LAST_TYPE; j++)
124                 rules[i][j] = true;
125         // GB 6 L x (L | V | LV | VT)
126         rules[L][L] = false;
127         rules[L][V] = false;
128         rules[L][LV] = false;
129         rules[L][LVT] = false;
130         // GB 7 (LV | V) x (V | T)
131         rules[LV][V] = false;
132         rules[LV][T] = false;
133         rules[V][V] = false;
134         rules[V][T] = false;
135         // GB 8 (LVT | T) x T
136         rules[LVT][T] = false;
137         rules[T][T] = false;
138         // GB 9 x (Extend|ZWJ)
139         // GB 9a x Spacing Mark
140         // GB 9b Prepend x
141         for (int i = FIRST_TYPE; i <= LAST_TYPE; i++) {
142             rules[i][EXTEND] = false;
143             rules[i][ZWJ] = false;
144             rules[i][SPACINGMARK] = false;
145             rules[PREPEND][i] = false;
146         }
147         // GB 4  (Control | CR | LF) +
148         // GB 5  + (Control | CR | LF)
149         for (int i = FIRST_TYPE; i <= LAST_TYPE; i++)
150             for (int j = CR; j <= CONTROL; j++) {
151                 rules[i][j] = true;
152                 rules[j][i] = true;
153             }
154         // GB 3 CR x LF
155         rules[CR][LF] = false;
156         // GB 11 Exended_Pictographic x (Extend|ZWJ)
157         rules[EXTENDED_PICTOGRAPHIC][EXTEND] = false;
158         rules[EXTENDED_PICTOGRAPHIC][ZWJ] = false;
159     }
160 
161     // Hangul syllables
162     private static final int SYLLABLE_BASE = 0xAC00;
163     private static final int LCOUNT = 19;
164     private static final int VCOUNT = 21;
165     private static final int TCOUNT = 28;
166     private static final int NCOUNT = VCOUNT * TCOUNT; // 588
167     private static final int SCOUNT = LCOUNT * NCOUNT; // 11172
168 
169     // #tr29: SpacingMark exceptions: The following (which have
170     // General_Category = Spacing_Mark and would otherwise be included)
171     // are specifically excluded
isExcludedSpacingMark(int cp)172     private static boolean isExcludedSpacingMark(int cp) {
173        return  cp == 0x102B || cp == 0x102C || cp == 0x1038 ||
174                cp >= 0x1062 && cp <= 0x1064 ||
175                cp >= 0x1062 && cp <= 0x106D ||
176                cp == 0x1083 ||
177                cp >= 0x1087 && cp <= 0x108C ||
178                cp == 0x108F ||
179                cp >= 0x109A && cp <= 0x109C ||
180                cp == 0x1A61 || cp == 0x1A63 || cp == 0x1A64 ||
181                cp == 0xAA7B || cp == 0xAA7D;
182     }
183 
184     @SuppressWarnings("fallthrough")
getType(int cp)185     private static int getType(int cp) {
186         if (isExtendedPictographic(cp)) {
187             return EXTENDED_PICTOGRAPHIC;
188         }
189 
190         int type = Character.getType(cp);
191         switch(type) {
192         case Character.CONTROL:
193             if (cp == 0x000D)
194                 return CR;
195             if (cp == 0x000A)
196                 return LF;
197             return CONTROL;
198         case Character.UNASSIGNED:
199             // NOTE: #tr29 lists "Unassigned and Default_Ignorable_Code_Point" as Control
200             // but GraphemeBreakTest.txt lists u+0378/reserved-0378 as "Other"
201             // so type it as "Other" to make the test happy
202             if (cp == 0x0378)
203                 return OTHER;
204 
205         case Character.LINE_SEPARATOR:
206         case Character.PARAGRAPH_SEPARATOR:
207         case Character.SURROGATE:
208             return CONTROL;
209         case Character.FORMAT:
210             if (cp == 0x200C ||
211                 cp >= 0xE0020 && cp <= 0xE007F)
212                 return EXTEND;
213             if (cp == 0x200D)
214                 return ZWJ;
215             if (cp >= 0x0600 && cp <= 0x0605 ||
216                 cp == 0x06DD || cp == 0x070F || cp == 0x08E2 ||
217                 cp == 0x110BD || cp == 0x110CD)
218                 return PREPEND;
219             return CONTROL;
220         case Character.NON_SPACING_MARK:
221         case Character.ENCLOSING_MARK:
222             // NOTE:
223             // #tr29 "plus a few General_Category = Spacing_Mark needed for
224             // canonical equivalence."
225             // but for "extended grapheme clusters" support, there is no
226             // need actually to diff "extend" and "spackmark" given GB9, GB9a
227             return EXTEND;
228         case  Character.COMBINING_SPACING_MARK:
229             if (isExcludedSpacingMark(cp))
230                 return OTHER;
231             // NOTE:
232             // 0x11720 and 0x11721 are mentioned in #tr29 as
233             // OTHER_LETTER but it appears their category has been updated to
234             // COMBING_SPACING_MARK already (verified in ver.8)
235             return SPACINGMARK;
236         case Character.OTHER_SYMBOL:
237             if (cp >= 0x1F1E6 && cp <= 0x1F1FF)
238                 return RI;
239             return OTHER;
240         case Character.MODIFIER_LETTER:
241         case Character.MODIFIER_SYMBOL:
242             // WARNING:
243             // not mentioned in #tr29 but listed in GraphemeBreakProperty.txt
244             if (cp == 0xFF9E || cp == 0xFF9F ||
245                 cp >= 0x1F3FB && cp <= 0x1F3FF)
246                 return EXTEND;
247             return OTHER;
248         case Character.OTHER_LETTER:
249             if (cp == 0x0E33 || cp == 0x0EB3)
250                 return SPACINGMARK;
251             // hangul jamo
252             if (cp >= 0x1100 && cp <= 0x11FF) {
253                 if (cp <= 0x115F)
254                     return L;
255                 if (cp <= 0x11A7)
256                     return V;
257                 return T;
258             }
259             // hangul syllables
260             int sindex = cp - SYLLABLE_BASE;
261             if (sindex >= 0 && sindex < SCOUNT) {
262 
263                 if (sindex % TCOUNT == 0)
264                     return LV;
265                 return LVT;
266             }
267             //  hangul jamo_extended A
268             if (cp >= 0xA960 && cp <= 0xA97C)
269                 return L;
270             //  hangul jamo_extended B
271             if (cp >= 0xD7B0 && cp <= 0xD7C6)
272                 return V;
273             if (cp >= 0xD7CB && cp <= 0xD7FB)
274                 return T;
275 
276             // Prepend
277             switch (cp) {
278                 case 0x0D4E:
279                 case 0x111C2:
280                 case 0x111C3:
281                 case 0x1193F:
282                 case 0x11941:
283                 case 0x11A3A:
284                 case 0x11A84:
285                 case 0x11A85:
286                 case 0x11A86:
287                 case 0x11A87:
288                 case 0x11A88:
289                 case 0x11A89:
290                 case 0x11D46:
291                     return PREPEND;
292             }
293         }
294         return OTHER;
295     }
296 
297     // from generated java.util.regex.EmojiData.java
isExtendedPictographic(int cp)298     static boolean isExtendedPictographic(int cp) {
299         return
300                 cp == 0x00A9 ||
301                 cp == 0x00AE ||
302                 cp == 0x203C ||
303                 cp == 0x2049 ||
304                 cp == 0x2122 ||
305                 cp == 0x2139 ||
306                (cp >= 0x2194 && cp <= 0x2199) ||
307                 cp == 0x21A9 ||
308                 cp == 0x21AA ||
309                 cp == 0x231A ||
310                 cp == 0x231B ||
311                 cp == 0x2328 ||
312                 cp == 0x2388 ||
313                 cp == 0x23CF ||
314                (cp >= 0x23E9 && cp <= 0x23F3) ||
315                (cp >= 0x23F8 && cp <= 0x23FA) ||
316                 cp == 0x24C2 ||
317                 cp == 0x25AA ||
318                 cp == 0x25AB ||
319                 cp == 0x25B6 ||
320                 cp == 0x25C0 ||
321                (cp >= 0x25FB && cp <= 0x25FE) ||
322                (cp >= 0x2600 && cp <= 0x2605) ||
323                (cp >= 0x2607 && cp <= 0x2612) ||
324                (cp >= 0x2614 && cp <= 0x2685) ||
325                (cp >= 0x2690 && cp <= 0x2705) ||
326                (cp >= 0x2708 && cp <= 0x2712) ||
327                 cp == 0x2714 ||
328                 cp == 0x2716 ||
329                 cp == 0x271D ||
330                 cp == 0x2721 ||
331                 cp == 0x2728 ||
332                 cp == 0x2733 ||
333                 cp == 0x2734 ||
334                 cp == 0x2744 ||
335                 cp == 0x2747 ||
336                 cp == 0x274C ||
337                 cp == 0x274E ||
338                (cp >= 0x2753 && cp <= 0x2755) ||
339                 cp == 0x2757 ||
340                (cp >= 0x2763 && cp <= 0x2767) ||
341                (cp >= 0x2795 && cp <= 0x2797) ||
342                 cp == 0x27A1 ||
343                 cp == 0x27B0 ||
344                 cp == 0x27BF ||
345                 cp == 0x2934 ||
346                 cp == 0x2935 ||
347                (cp >= 0x2B05 && cp <= 0x2B07) ||
348                 cp == 0x2B1B ||
349                 cp == 0x2B1C ||
350                 cp == 0x2B50 ||
351                 cp == 0x2B55 ||
352                 cp == 0x3030 ||
353                 cp == 0x303D ||
354                 cp == 0x3297 ||
355                 cp == 0x3299 ||
356                (cp >= 0x1F000 && cp <= 0x1F0FF) ||
357                (cp >= 0x1F10D && cp <= 0x1F10F) ||
358                 cp == 0x1F12F ||
359                (cp >= 0x1F16C && cp <= 0x1F171) ||
360                 cp == 0x1F17E ||
361                 cp == 0x1F17F ||
362                 cp == 0x1F18E ||
363                (cp >= 0x1F191 && cp <= 0x1F19A) ||
364                (cp >= 0x1F1AD && cp <= 0x1F1E5) ||
365                (cp >= 0x1F201 && cp <= 0x1F20F) ||
366                 cp == 0x1F21A ||
367                 cp == 0x1F22F ||
368                (cp >= 0x1F232 && cp <= 0x1F23A) ||
369                (cp >= 0x1F23C && cp <= 0x1F23F) ||
370                (cp >= 0x1F249 && cp <= 0x1F3FA) ||
371                (cp >= 0x1F400 && cp <= 0x1F53D) ||
372                (cp >= 0x1F546 && cp <= 0x1F64F) ||
373                (cp >= 0x1F680 && cp <= 0x1F6FF) ||
374                (cp >= 0x1F774 && cp <= 0x1F77F) ||
375                (cp >= 0x1F7D5 && cp <= 0x1F7FF) ||
376                (cp >= 0x1F80C && cp <= 0x1F80F) ||
377                (cp >= 0x1F848 && cp <= 0x1F84F) ||
378                (cp >= 0x1F85A && cp <= 0x1F85F) ||
379                (cp >= 0x1F888 && cp <= 0x1F88F) ||
380                (cp >= 0x1F8AE && cp <= 0x1F8FF) ||
381                (cp >= 0x1F90C && cp <= 0x1F93A) ||
382                (cp >= 0x1F93C && cp <= 0x1F945) ||
383                (cp >= 0x1F947 && cp <= 0x1FAFF) ||
384                (cp >= 0x1FC00 && cp <= 0x1FFFD);
385     }
386 }
387