1 package org.apache.harmony.regex.tests.java.util.regex; 2 3 import junit.framework.TestCase; 4 import java.util.regex.*; 5 6 /** 7 * TODO Type description 8 * 9 */ 10 public class SplitTest extends TestCase { 11 public void testSimple() { 12 Pattern p = Pattern.compile("/"); 13 String[] results = p.split("have/you/done/it/right"); 14 String[] expected = new String[] { "have", "you", "done", "it", "right" }; 15 assertArraysEqual(expected, results); 16 } 17 18 public void testEmptySplits() { 19 // Trailing empty matches are removed. 20 assertArraysEqual(new String[0], "hello".split(".")); 21 assertArraysEqual(new String[] { "1", "2" }, "1:2:".split(":")); 22 // ...including when that results in an empty result. 23 assertArraysEqual(new String[0], ":".split(":")); 24 // ...but not when limit < 0. 25 assertArraysEqual(new String[] { "1", "2", "" }, "1:2:".split(":", -1)); 26 27 // Leading empty matches are retained. 28 assertArraysEqual(new String[] { "", "", "o" }, "hello".split("..")); 29 30 // A separator that doesn't occur in the input gets you the input. 31 assertArraysEqual(new String[] { "hello" }, "hello".split("not-present-in-test")); 32 // ...including when the input is the empty string. 33 // (Perl returns an empty list instead.) 34 assertArraysEqual(new String[] { "" }, "".split("not-present-in-test")); 35 assertArraysEqual(new String[] { "" }, "".split("A?")); 36 37 // The limit argument controls the size of the result. 38 // If l == 0, the result is as long as needed, except trailing empty matches are dropped. 39 // If l < 0, the result is as long as needed, and trailing empty matches are retained. 40 // If l > 0, the result contains the first l matches, plus one string containing the remaining input. 41 // Examples without a trailing separator (and hence without a trailing empty match): 42 assertArraysEqual(new String[] { "a", "b", "c" }, "a,b,c".split(",", 0)); 43 assertArraysEqual(new String[] { "a,b,c" }, "a,b,c".split(",", 1)); 44 assertArraysEqual(new String[] { "a", "b,c" }, "a,b,c".split(",", 2)); 45 assertArraysEqual(new String[] { "a", "b", "c" }, "a,b,c".split(",", 3)); 46 assertArraysEqual(new String[] { "a", "b", "c" }, "a,b,c".split(",", Integer.MAX_VALUE)); 47 // Examples with a trailing separator (and hence possibly with a trailing empty match): 48 assertArraysEqual(new String[] { "a", "b", "c" }, "a,b,c,".split(",", 0)); 49 assertArraysEqual(new String[] { "a,b,c," }, "a,b,c,".split(",", 1)); 50 assertArraysEqual(new String[] { "a", "b,c," }, "a,b,c,".split(",", 2)); 51 assertArraysEqual(new String[] { "a", "b", "c," }, "a,b,c,".split(",", 3)); 52 assertArraysEqual(new String[] { "a", "b", "c", "" }, "a,b,c,".split(",", Integer.MAX_VALUE)); 53 assertArraysEqual(new String[] { "a", "b", "c", "" }, "a,b,c,".split(",", -1)); 54 } 55 56 private void assertArraysEqual(String[] expected, String[] actual) { 57 assertEquals(expected.length, actual.length); 58 for (int i = 0; i < expected.length; i++) { 59 assertEquals(Integer.toString(i), expected[i], actual[i]); 60 } 61 } 62 63 public void testSplit1() throws PatternSyntaxException { 64 Pattern p = Pattern.compile(" "); 65 66 String input = "poodle zoo"; 67 String tokens[]; 68 69 tokens = p.split(input, 1); 70 assertEquals(1, tokens.length); 71 assertTrue(tokens[0].equals(input)); 72 tokens = p.split(input, 2); 73 assertEquals(2, tokens.length); 74 assertEquals("poodle", tokens[0]); 75 assertEquals("zoo", tokens[1]); 76 tokens = p.split(input, 5); 77 assertEquals(2, tokens.length); 78 assertEquals("poodle", tokens[0]); 79 assertEquals("zoo", tokens[1]); 80 tokens = p.split(input, -2); 81 assertEquals(2, tokens.length); 82 assertEquals("poodle", tokens[0]); 83 assertEquals("zoo", tokens[1]); 84 tokens = p.split(input, 0); 85 assertEquals(2, tokens.length); 86 assertEquals("poodle", tokens[0]); 87 assertEquals("zoo", tokens[1]); 88 tokens = p.split(input); 89 assertEquals(2, tokens.length); 90 assertEquals("poodle", tokens[0]); 91 assertEquals("zoo", tokens[1]); 92 93 p = Pattern.compile("d"); 94 95 tokens = p.split(input, 1); 96 assertEquals(1, tokens.length); 97 assertTrue(tokens[0].equals(input)); 98 tokens = p.split(input, 2); 99 assertEquals(2, tokens.length); 100 assertEquals("poo", tokens[0]); 101 assertEquals("le zoo", tokens[1]); 102 tokens = p.split(input, 5); 103 assertEquals(2, tokens.length); 104 assertEquals("poo", tokens[0]); 105 assertEquals("le zoo", tokens[1]); 106 tokens = p.split(input, -2); 107 assertEquals(2, tokens.length); 108 assertEquals("poo", tokens[0]); 109 assertEquals("le zoo", tokens[1]); 110 tokens = p.split(input, 0); 111 assertEquals(2, tokens.length); 112 assertEquals("poo", tokens[0]); 113 assertEquals("le zoo", tokens[1]); 114 tokens = p.split(input); 115 assertEquals(2, tokens.length); 116 assertEquals("poo", tokens[0]); 117 assertEquals("le zoo", tokens[1]); 118 119 p = Pattern.compile("o"); 120 121 tokens = p.split(input, 1); 122 assertEquals(1, tokens.length); 123 assertTrue(tokens[0].equals(input)); 124 tokens = p.split(input, 2); 125 assertEquals(2, tokens.length); 126 assertEquals("p", tokens[0]); 127 assertEquals("odle zoo", tokens[1]); 128 tokens = p.split(input, 5); 129 assertEquals(5, tokens.length); 130 assertEquals("p", tokens[0]); 131 assertTrue(tokens[1].equals("")); 132 assertEquals("dle z", tokens[2]); 133 assertTrue(tokens[3].equals("")); 134 assertTrue(tokens[4].equals("")); 135 tokens = p.split(input, -2); 136 assertEquals(5, tokens.length); 137 assertEquals("p", tokens[0]); 138 assertTrue(tokens[1].equals("")); 139 assertEquals("dle z", tokens[2]); 140 assertTrue(tokens[3].equals("")); 141 assertTrue(tokens[4].equals("")); 142 tokens = p.split(input, 0); 143 assertEquals(3, tokens.length); 144 assertEquals("p", tokens[0]); 145 assertTrue(tokens[1].equals("")); 146 assertEquals("dle z", tokens[2]); 147 tokens = p.split(input); 148 assertEquals(3, tokens.length); 149 assertEquals("p", tokens[0]); 150 assertTrue(tokens[1].equals("")); 151 assertEquals("dle z", tokens[2]); 152 } 153 154 public void testSplit2() { 155 Pattern p = Pattern.compile(""); 156 String s[]; 157 s = p.split("a", -1); 158 assertEquals(3, s.length); 159 assertEquals("", s[0]); 160 assertEquals("a", s[1]); 161 assertEquals("", s[2]); 162 163 s = p.split("", -1); 164 assertEquals(1, s.length); 165 assertEquals("", s[0]); 166 167 s = p.split("abcd", -1); 168 assertEquals(6, s.length); 169 assertEquals("", s[0]); 170 assertEquals("a", s[1]); 171 assertEquals("b", s[2]); 172 assertEquals("c", s[3]); 173 assertEquals("d", s[4]); 174 assertEquals("", s[5]); 175 176 // Regression test for Android 177 assertEquals("GOOG,23,500".split("|").length, 12); 178 } 179 180 181 public void testSplitSupplementaryWithEmptyString() { 182 183 /* 184 * See http://www.unicode.org/reports/tr18/#Supplementary_Characters 185 * We have to treat text as code points not code units. 186 */ 187 Pattern p = Pattern.compile(""); 188 String s[]; 189 s = p.split("a\ud869\uded6b", -1); 190 assertEquals(5, s.length); 191 assertEquals("", s[0]); 192 assertEquals("a", s[1]); 193 assertEquals("\ud869\uded6", s[2]); 194 assertEquals("b", s[3]); 195 assertEquals("", s[4]); 196 } 197 } 198
