|
1
7
8 package com.ibm.icu.impl;
9
10 import java.util.Comparator  ;
11 import java.util.Iterator ;
12 import java.util.Set ;
13 import java.util.TreeSet ;
14
15 import com.ibm.icu.impl.CollectionUtilities.MultiComparator;
16 import com.ibm.icu.lang.UCharacter;
17 import com.ibm.icu.text.Collator;
18 import com.ibm.icu.text.RuleBasedCollator;
19 import com.ibm.icu.text.Transliterator;
20 import com.ibm.icu.text.UTF16;
21 import com.ibm.icu.text.UnicodeSet;
22 import com.ibm.icu.text.UnicodeSetIterator;
23 import com.ibm.icu.util.ULocale;
24
25
27 public class PrettyPrinter {
28 private static final UnicodeSet patternWhitespace = (UnicodeSet) new UnicodeSet("[[:Cn:][:Default_Ignorable_Code_Point:][:patternwhitespace:]]").freeze();
29 private static final UnicodeSet sortAtEnd = (UnicodeSet) new UnicodeSet("[[:Cn:][:Cs:][:Co:][:Ideographic:]]").freeze();
30
31 private boolean first = true;
32 private StringBuffer target = new StringBuffer ();
33 private int firstCodePoint = -2;
34 private int lastCodePoint = -2;
35 private boolean compressRanges = true;
36 private String lastString = "";
37 private UnicodeSet toQuote = new UnicodeSet(patternWhitespace);
38 private Transliterator quoter = null;
39
40 private Comparator ordering;
41 private Comparator spaceComp = Collator.getInstance(ULocale.ROOT);
42 {
43 setOrdering(Collator.getInstance(ULocale.ROOT));
44 ((RuleBasedCollator)spaceComp).setStrength(RuleBasedCollator.PRIMARY);
45 }
46
47 public Transliterator getQuoter() {
48 return quoter;
49 }
50
51 public PrettyPrinter setQuoter(Transliterator quoter) {
52 this.quoter = quoter;
53 return this; }
55
56 public boolean isCompressRanges() {
57 return compressRanges;
58 }
59
60
64 public PrettyPrinter setCompressRanges(boolean compressRanges) {
65 this.compressRanges = compressRanges;
66 return this;
67 }
68
69 public Comparator getOrdering() {
70 return ordering;
71 }
72
73
77 public PrettyPrinter setOrdering(Comparator ordering) {
78 this.ordering = new MultiComparator(new Comparator [] {ordering, new UTF16.StringComparator(true,false,0)});
79 return this;
80 }
81
82 public Comparator getSpaceComparator() {
83 return spaceComp;
84 }
85
86
90 public PrettyPrinter setSpaceComparator(Comparator spaceComp) {
91 this.spaceComp = spaceComp;
92 return this;
93 }
94
95 public UnicodeSet getToQuote() {
96 return toQuote;
97 }
98
99
103 public PrettyPrinter setToQuote(UnicodeSet toQuote) {
104 toQuote = (UnicodeSet)toQuote.clone();
105 toQuote.addAll(patternWhitespace);
106 this.toQuote = toQuote;
107 return this;
108 }
109
110
115 public String toPattern(UnicodeSet uset) {
116 first = true;
117 UnicodeSet putAtEnd = new UnicodeSet(uset).retainAll(sortAtEnd); Set orderedStrings = new TreeSet (ordering);
120 for (UnicodeSetIterator it = new UnicodeSetIterator(uset); it.nextRange();) {
121 if (it.codepoint == it.IS_STRING) {
122 orderedStrings.add(it.string);
123 } else {
124 for (int i = it.codepoint; i <= it.codepointEnd; ++i) {
125 if (!putAtEnd.contains(i)) {
126 orderedStrings.add(UTF16.valueOf(i));
127 }
128 }
129 }
130 }
131 target.setLength(0);
132 target.append("[");
133 for (Iterator it = orderedStrings.iterator(); it.hasNext();) {
134 appendUnicodeSetItem((String ) it.next());
135 }
136 for (UnicodeSetIterator it = new UnicodeSetIterator(putAtEnd); it.next();) { appendUnicodeSetItem(it.codepoint);
138 }
139 flushLast();
140 target.append("]");
141 String sresult = target.toString();
142
143 return sresult;
153 }
154
155 private PrettyPrinter appendUnicodeSetItem(String s) {
156 int cp;
157 if (UTF16.hasMoreCodePointsThan(s, 1)) {
158 flushLast();
159 addSpace(s);
160 target.append("{");
161 for (int i = 0; i < s.length(); i += UTF16.getCharCount(cp)) {
162 appendQuoted(cp = UTF16.charAt(s, i));
163 }
164 target.append("}");
165 lastString = s;
166 } else {
167 appendUnicodeSetItem(UTF16.charAt(s, 0));
168 }
169 return this;
170 }
171
172 private void appendUnicodeSetItem(int cp) {
173 if (!compressRanges)
174 flushLast();
175 if (cp == lastCodePoint + 1) {
176 lastCodePoint = cp; } else { flushLast();
179 firstCodePoint = lastCodePoint = cp;
180 }
181 }
182
185 private void addSpace(String s) {
186 if (first) {
187 first = false;
188 } else if (spaceComp.compare(s, lastString) != 0) {
189 target.append(' ');
190 } else {
191 int cp = UTF16.charAt(s,0);
192 int type = UCharacter.getType(cp);
193 if (type == UCharacter.NON_SPACING_MARK || type == UCharacter.ENCLOSING_MARK) {
194 target.append(' ');
195 } else if (type == UCharacter.SURROGATE && cp >= UTF16.TRAIL_SURROGATE_MIN_VALUE) {
196 target.append(' '); }
198 }
199 }
200
201 private void flushLast() {
202 if (lastCodePoint >= 0) {
203 addSpace(UTF16.valueOf(firstCodePoint));
204 if (firstCodePoint != lastCodePoint) {
205 appendQuoted(firstCodePoint);
206 target.append(firstCodePoint + 1 == lastCodePoint ? ' ' : '-');
207 }
208 appendQuoted(lastCodePoint);
209 lastString = UTF16.valueOf(lastCodePoint);
210 firstCodePoint = lastCodePoint = -2;
211 }
212 }
213 PrettyPrinter appendQuoted(int codePoint) {
214 if (toQuote.contains(codePoint)) {
215 if (quoter != null) {
216 target.append(quoter.transliterate(UTF16.valueOf(codePoint)));
217 return this;
218 }
219 if (codePoint > 0xFFFF) {
220 target.append("\\U");
221 target.append(Utility.hex(codePoint,8));
222 } else {
223 target.append("\\u");
224 target.append(Utility.hex(codePoint,4));
225 }
226 return this;
227 }
228 switch (codePoint) {
229 case '[': case ']': case '-': case '^': case '&': case '\\': case '{':
236 case '}':
237 case '$':
238 case ':':
239 target.append('\\');
240 break;
241 default:
242 if (patternWhitespace.contains(codePoint)) {
244 target.append('\\');
245 }
246 break;
247 }
248 UTF16.append(target, codePoint);
249 return this;
250 }
251 }
259 |
Popular Tags |
Java API By Example, From Geeks To Geeks.