1
0
mirror of https://github.com/apache/commons-lang.git synced 2025-02-15 06:25:24 +00:00

[LANG-1238] Add RegexUtils class instead of overloadinh methods in

StringUtils that take a regex to take precompiled Pattern. Sort methods.
This commit is contained in:
Gary Gregory 2018-05-17 17:16:08 -06:00
parent bcc4f82a7a
commit 54acb6e10f

@ -32,24 +32,84 @@
public class RegExUtilsTest {
@Test
public void testReplacePattern_StringStringString() {
assertNull(RegExUtils.replacePattern(null, "", ""));
assertEquals("any", RegExUtils.replacePattern("any", (String) null, ""));
assertEquals("any", RegExUtils.replacePattern("any", "", null));
public void testRemoveAll_StringPattern() {
assertNull(RegExUtils.removeAll(null, Pattern.compile("")));
assertEquals("any", RegExUtils.removeAll("any", (Pattern) null));
assertEquals("zzz", RegExUtils.replacePattern("", "", "zzz"));
assertEquals("zzz", RegExUtils.replacePattern("", ".*", "zzz"));
assertEquals("", RegExUtils.replacePattern("", ".+", "zzz"));
assertEquals("any", RegExUtils.removeAll("any", Pattern.compile("")));
assertEquals("", RegExUtils.removeAll("any", Pattern.compile(".*")));
assertEquals("", RegExUtils.removeAll("any", Pattern.compile(".+")));
assertEquals("", RegExUtils.removeAll("any", Pattern.compile(".?")));
assertEquals("z", RegExUtils.replacePattern("<__>\n<__>", "<.*>", "z"));
assertEquals("z", RegExUtils.replacePattern("<__>\\n<__>", "<.*>", "z"));
assertEquals("X", RegExUtils.replacePattern("<A>\nxy\n</A>", "<A>.*</A>", "X"));
assertEquals("A\nB", RegExUtils.removeAll("A<__>\n<__>B", Pattern.compile("<.*>")));
assertEquals("AB", RegExUtils.removeAll("A<__>\n<__>B", Pattern.compile("(?s)<.*>")));
assertEquals("ABC123", RegExUtils.removeAll("ABCabc123abc", Pattern.compile("[a-z]")));
assertEquals("ABC___123", RegExUtils.replacePattern("ABCabc123", "[a-z]", "_"));
assertEquals("ABC_123", RegExUtils.replacePattern("ABCabc123", "[^A-Z0-9]+", "_"));
assertEquals("ABC123", RegExUtils.replacePattern("ABCabc123", "[^A-Z0-9]+", ""));
assertEquals("Lorem_ipsum_dolor_sit",
RegExUtils.replacePattern("Lorem ipsum dolor sit", "( +)([a-z]+)", "_$2"));
assertEquals("AB", RegExUtils.removeAll("A<__>\n<__>B", Pattern.compile("<.*>", Pattern.DOTALL)));
assertEquals("AB", RegExUtils.removeAll("A<__>\\n<__>B", Pattern.compile("<.*>")));
assertEquals("", RegExUtils.removeAll("<A>x\\ny</A>", Pattern.compile("<A>.*</A>")));
assertEquals("", RegExUtils.removeAll("<A>\nxy\n</A>", Pattern.compile("<A>.*</A>", Pattern.DOTALL)));
}
@Test
public void testRemoveAll_StringString() {
assertNull(RegExUtils.removeAll(null, ""));
assertEquals("any", RegExUtils.removeAll("any", (String) null));
assertEquals("any", RegExUtils.removeAll("any", ""));
assertEquals("", RegExUtils.removeAll("any", ".*"));
assertEquals("", RegExUtils.removeAll("any", ".+"));
assertEquals("", RegExUtils.removeAll("any", ".?"));
assertEquals("A\nB", RegExUtils.removeAll("A<__>\n<__>B", "<.*>"));
assertEquals("AB", RegExUtils.removeAll("A<__>\n<__>B", "(?s)<.*>"));
assertEquals("ABC123", RegExUtils.removeAll("ABCabc123abc", "[a-z]"));
try {
RegExUtils.removeAll("any", "{badRegexSyntax}");
fail("RegExUtils.removeAll expecting PatternSyntaxException");
} catch (final PatternSyntaxException ex) {
// empty
}
}
@Test
public void testRemoveFirst_StringPattern() {
assertNull(RegExUtils.removeFirst(null, Pattern.compile("")));
assertEquals("any", RegExUtils.removeFirst("any", (Pattern) null));
assertEquals("any", RegExUtils.removeFirst("any", Pattern.compile("")));
assertEquals("", RegExUtils.removeFirst("any", Pattern.compile(".*")));
assertEquals("", RegExUtils.removeFirst("any", Pattern.compile(".+")));
assertEquals("bc", RegExUtils.removeFirst("abc", Pattern.compile(".?")));
assertEquals("A\n<__>B", RegExUtils.removeFirst("A<__>\n<__>B", Pattern.compile("<.*>")));
assertEquals("AB", RegExUtils.removeFirst("A<__>\n<__>B", Pattern.compile("(?s)<.*>")));
assertEquals("ABCbc123", RegExUtils.removeFirst("ABCabc123", Pattern.compile("[a-z]")));
assertEquals("ABC123abc", RegExUtils.removeFirst("ABCabc123abc", Pattern.compile("[a-z]+")));
}
@Test
public void testRemoveFirst_StringString() {
assertNull(RegExUtils.removeFirst(null, ""));
assertEquals("any", RegExUtils.removeFirst("any", (String) null));
assertEquals("any", RegExUtils.removeFirst("any", ""));
assertEquals("", RegExUtils.removeFirst("any", ".*"));
assertEquals("", RegExUtils.removeFirst("any", ".+"));
assertEquals("bc", RegExUtils.removeFirst("abc", ".?"));
assertEquals("A\n<__>B", RegExUtils.removeFirst("A<__>\n<__>B", "<.*>"));
assertEquals("AB", RegExUtils.removeFirst("A<__>\n<__>B", "(?s)<.*>"));
assertEquals("ABCbc123", RegExUtils.removeFirst("ABCabc123", "[a-z]"));
assertEquals("ABC123abc", RegExUtils.removeFirst("ABCabc123abc", "[a-z]+"));
try {
RegExUtils.removeFirst("any", "{badRegexSyntax}");
fail("RegExUtils.removeFirst expecting PatternSyntaxException");
} catch (final PatternSyntaxException ex) {
// empty
}
}
@Test
@ -69,6 +129,32 @@ public void testRemovePattern_StringString() {
assertEquals("ABC123", RegExUtils.removePattern("ABCabc123", "[a-z]"));
}
@Test
public void testReplaceAll_StringPatternString() {
assertNull(RegExUtils.replaceAll(null, Pattern.compile(""), ""));
assertEquals("any", RegExUtils.replaceAll("any", (Pattern) null, ""));
assertEquals("any", RegExUtils.replaceAll("any", Pattern.compile(""), null));
assertEquals("zzz", RegExUtils.replaceAll("", Pattern.compile(""), "zzz"));
assertEquals("zzz", RegExUtils.replaceAll("", Pattern.compile(".*"), "zzz"));
assertEquals("", RegExUtils.replaceAll("", Pattern.compile(".+"), "zzz"));
assertEquals("ZZaZZbZZcZZ", RegExUtils.replaceAll("abc", Pattern.compile(""), "ZZ"));
assertEquals("z\nz", RegExUtils.replaceAll("<__>\n<__>", Pattern.compile("<.*>"), "z"));
assertEquals("z", RegExUtils.replaceAll("<__>\n<__>", Pattern.compile("(?s)<.*>"), "z"));
assertEquals("z", RegExUtils.replaceAll("<__>\n<__>", Pattern.compile("<.*>", Pattern.DOTALL), "z"));
assertEquals("z", RegExUtils.replaceAll("<__>\\n<__>", Pattern.compile("<.*>"), "z"));
assertEquals("X", RegExUtils.replaceAll("<A>\nxy\n</A>", Pattern.compile("<A>.*</A>", Pattern.DOTALL), "X"));
assertEquals("ABC___123", RegExUtils.replaceAll("ABCabc123", Pattern.compile("[a-z]"), "_"));
assertEquals("ABC_123", RegExUtils.replaceAll("ABCabc123", Pattern.compile("[^A-Z0-9]+"), "_"));
assertEquals("ABC123", RegExUtils.replaceAll("ABCabc123", Pattern.compile("[^A-Z0-9]+"), ""));
assertEquals("Lorem_ipsum_dolor_sit",
RegExUtils.replaceAll("Lorem ipsum dolor sit", Pattern.compile("( +)([a-z]+)"), "_$2"));
}
@Test
public void testReplaceAll_StringStringString() {
assertNull(RegExUtils.replaceAll(null, "", ""));
@ -99,29 +185,25 @@ public void testReplaceAll_StringStringString() {
}
@Test
public void testReplaceAll_StringPatternString() {
assertNull(RegExUtils.replaceAll(null, Pattern.compile(""), ""));
public void testReplaceFirst_StringPatternString() {
assertNull(RegExUtils.replaceFirst(null, Pattern.compile(""), ""));
assertEquals("any", RegExUtils.replaceAll("any", (Pattern) null, ""));
assertEquals("any", RegExUtils.replaceAll("any", Pattern.compile(""), null));
assertEquals("any", RegExUtils.replaceFirst("any", (Pattern) null, ""));
assertEquals("any", RegExUtils.replaceFirst("any", Pattern.compile(""), null));
assertEquals("zzz", RegExUtils.replaceAll("", Pattern.compile(""), "zzz"));
assertEquals("zzz", RegExUtils.replaceAll("", Pattern.compile(".*"), "zzz"));
assertEquals("", RegExUtils.replaceAll("", Pattern.compile(".+"), "zzz"));
assertEquals("ZZaZZbZZcZZ", RegExUtils.replaceAll("abc", Pattern.compile(""), "ZZ"));
assertEquals("zzz", RegExUtils.replaceFirst("", Pattern.compile(""), "zzz"));
assertEquals("zzz", RegExUtils.replaceFirst("", Pattern.compile(".*"), "zzz"));
assertEquals("", RegExUtils.replaceFirst("", Pattern.compile(".+"), "zzz"));
assertEquals("ZZabc", RegExUtils.replaceFirst("abc", Pattern.compile(""), "ZZ"));
assertEquals("z\nz", RegExUtils.replaceAll("<__>\n<__>", Pattern.compile("<.*>"), "z"));
assertEquals("z", RegExUtils.replaceAll("<__>\n<__>", Pattern.compile("(?s)<.*>"), "z"));
assertEquals("z\n<__>", RegExUtils.replaceFirst("<__>\n<__>", Pattern.compile("<.*>"), "z"));
assertEquals("z", RegExUtils.replaceFirst("<__>\n<__>", Pattern.compile("(?s)<.*>"), "z"));
assertEquals("z", RegExUtils.replaceAll("<__>\n<__>", Pattern.compile("<.*>", Pattern.DOTALL), "z"));
assertEquals("z", RegExUtils.replaceAll("<__>\\n<__>", Pattern.compile("<.*>"), "z"));
assertEquals("X", RegExUtils.replaceAll("<A>\nxy\n</A>", Pattern.compile("<A>.*</A>", Pattern.DOTALL), "X"));
assertEquals("ABC___123", RegExUtils.replaceAll("ABCabc123", Pattern.compile("[a-z]"), "_"));
assertEquals("ABC_123", RegExUtils.replaceAll("ABCabc123", Pattern.compile("[^A-Z0-9]+"), "_"));
assertEquals("ABC123", RegExUtils.replaceAll("ABCabc123", Pattern.compile("[^A-Z0-9]+"), ""));
assertEquals("Lorem_ipsum_dolor_sit",
RegExUtils.replaceAll("Lorem ipsum dolor sit", Pattern.compile("( +)([a-z]+)"), "_$2"));
assertEquals("ABC_bc123", RegExUtils.replaceFirst("ABCabc123", Pattern.compile("[a-z]"), "_"));
assertEquals("ABC_123abc", RegExUtils.replaceFirst("ABCabc123abc", Pattern.compile("[^A-Z0-9]+"), "_"));
assertEquals("ABC123abc", RegExUtils.replaceFirst("ABCabc123abc", Pattern.compile("[^A-Z0-9]+"), ""));
assertEquals("Lorem_ipsum dolor sit",
RegExUtils.replaceFirst("Lorem ipsum dolor sit", Pattern.compile("( +)([a-z]+)"), "_$2"));
}
@Test
@ -154,106 +236,24 @@ public void testReplaceFirst_StringStringString() {
}
@Test
public void testReplaceFirst_StringPatternString() {
assertNull(RegExUtils.replaceFirst(null, Pattern.compile(""), ""));
public void testReplacePattern_StringStringString() {
assertNull(RegExUtils.replacePattern(null, "", ""));
assertEquals("any", RegExUtils.replacePattern("any", (String) null, ""));
assertEquals("any", RegExUtils.replacePattern("any", "", null));
assertEquals("any", RegExUtils.replaceFirst("any", (Pattern) null, ""));
assertEquals("any", RegExUtils.replaceFirst("any", Pattern.compile(""), null));
assertEquals("zzz", RegExUtils.replacePattern("", "", "zzz"));
assertEquals("zzz", RegExUtils.replacePattern("", ".*", "zzz"));
assertEquals("", RegExUtils.replacePattern("", ".+", "zzz"));
assertEquals("zzz", RegExUtils.replaceFirst("", Pattern.compile(""), "zzz"));
assertEquals("zzz", RegExUtils.replaceFirst("", Pattern.compile(".*"), "zzz"));
assertEquals("", RegExUtils.replaceFirst("", Pattern.compile(".+"), "zzz"));
assertEquals("ZZabc", RegExUtils.replaceFirst("abc", Pattern.compile(""), "ZZ"));
assertEquals("z", RegExUtils.replacePattern("<__>\n<__>", "<.*>", "z"));
assertEquals("z", RegExUtils.replacePattern("<__>\\n<__>", "<.*>", "z"));
assertEquals("X", RegExUtils.replacePattern("<A>\nxy\n</A>", "<A>.*</A>", "X"));
assertEquals("z\n<__>", RegExUtils.replaceFirst("<__>\n<__>", Pattern.compile("<.*>"), "z"));
assertEquals("z", RegExUtils.replaceFirst("<__>\n<__>", Pattern.compile("(?s)<.*>"), "z"));
assertEquals("ABC_bc123", RegExUtils.replaceFirst("ABCabc123", Pattern.compile("[a-z]"), "_"));
assertEquals("ABC_123abc", RegExUtils.replaceFirst("ABCabc123abc", Pattern.compile("[^A-Z0-9]+"), "_"));
assertEquals("ABC123abc", RegExUtils.replaceFirst("ABCabc123abc", Pattern.compile("[^A-Z0-9]+"), ""));
assertEquals("Lorem_ipsum dolor sit",
RegExUtils.replaceFirst("Lorem ipsum dolor sit", Pattern.compile("( +)([a-z]+)"), "_$2"));
}
@Test
public void testRemoveAll_StringString() {
assertNull(RegExUtils.removeAll(null, ""));
assertEquals("any", RegExUtils.removeAll("any", (String) null));
assertEquals("any", RegExUtils.removeAll("any", ""));
assertEquals("", RegExUtils.removeAll("any", ".*"));
assertEquals("", RegExUtils.removeAll("any", ".+"));
assertEquals("", RegExUtils.removeAll("any", ".?"));
assertEquals("A\nB", RegExUtils.removeAll("A<__>\n<__>B", "<.*>"));
assertEquals("AB", RegExUtils.removeAll("A<__>\n<__>B", "(?s)<.*>"));
assertEquals("ABC123", RegExUtils.removeAll("ABCabc123abc", "[a-z]"));
try {
RegExUtils.removeAll("any", "{badRegexSyntax}");
fail("RegExUtils.removeAll expecting PatternSyntaxException");
} catch (final PatternSyntaxException ex) {
// empty
}
}
@Test
public void testRemoveAll_StringPattern() {
assertNull(RegExUtils.removeAll(null, Pattern.compile("")));
assertEquals("any", RegExUtils.removeAll("any", (Pattern) null));
assertEquals("any", RegExUtils.removeAll("any", Pattern.compile("")));
assertEquals("", RegExUtils.removeAll("any", Pattern.compile(".*")));
assertEquals("", RegExUtils.removeAll("any", Pattern.compile(".+")));
assertEquals("", RegExUtils.removeAll("any", Pattern.compile(".?")));
assertEquals("A\nB", RegExUtils.removeAll("A<__>\n<__>B", Pattern.compile("<.*>")));
assertEquals("AB", RegExUtils.removeAll("A<__>\n<__>B", Pattern.compile("(?s)<.*>")));
assertEquals("ABC123", RegExUtils.removeAll("ABCabc123abc", Pattern.compile("[a-z]")));
assertEquals("AB", RegExUtils.removeAll("A<__>\n<__>B", Pattern.compile("<.*>", Pattern.DOTALL)));
assertEquals("AB", RegExUtils.removeAll("A<__>\\n<__>B", Pattern.compile("<.*>")));
assertEquals("", RegExUtils.removeAll("<A>x\\ny</A>", Pattern.compile("<A>.*</A>")));
assertEquals("", RegExUtils.removeAll("<A>\nxy\n</A>", Pattern.compile("<A>.*</A>", Pattern.DOTALL)));
}
@Test
public void testRemoveFirst_StringString() {
assertNull(RegExUtils.removeFirst(null, ""));
assertEquals("any", RegExUtils.removeFirst("any", (String) null));
assertEquals("any", RegExUtils.removeFirst("any", ""));
assertEquals("", RegExUtils.removeFirst("any", ".*"));
assertEquals("", RegExUtils.removeFirst("any", ".+"));
assertEquals("bc", RegExUtils.removeFirst("abc", ".?"));
assertEquals("A\n<__>B", RegExUtils.removeFirst("A<__>\n<__>B", "<.*>"));
assertEquals("AB", RegExUtils.removeFirst("A<__>\n<__>B", "(?s)<.*>"));
assertEquals("ABCbc123", RegExUtils.removeFirst("ABCabc123", "[a-z]"));
assertEquals("ABC123abc", RegExUtils.removeFirst("ABCabc123abc", "[a-z]+"));
try {
RegExUtils.removeFirst("any", "{badRegexSyntax}");
fail("RegExUtils.removeFirst expecting PatternSyntaxException");
} catch (final PatternSyntaxException ex) {
// empty
}
}
@Test
public void testRemoveFirst_StringPattern() {
assertNull(RegExUtils.removeFirst(null, Pattern.compile("")));
assertEquals("any", RegExUtils.removeFirst("any", (Pattern) null));
assertEquals("any", RegExUtils.removeFirst("any", Pattern.compile("")));
assertEquals("", RegExUtils.removeFirst("any", Pattern.compile(".*")));
assertEquals("", RegExUtils.removeFirst("any", Pattern.compile(".+")));
assertEquals("bc", RegExUtils.removeFirst("abc", Pattern.compile(".?")));
assertEquals("A\n<__>B", RegExUtils.removeFirst("A<__>\n<__>B", Pattern.compile("<.*>")));
assertEquals("AB", RegExUtils.removeFirst("A<__>\n<__>B", Pattern.compile("(?s)<.*>")));
assertEquals("ABCbc123", RegExUtils.removeFirst("ABCabc123", Pattern.compile("[a-z]")));
assertEquals("ABC123abc", RegExUtils.removeFirst("ABCabc123abc", Pattern.compile("[a-z]+")));
assertEquals("ABC___123", RegExUtils.replacePattern("ABCabc123", "[a-z]", "_"));
assertEquals("ABC_123", RegExUtils.replacePattern("ABCabc123", "[^A-Z0-9]+", "_"));
assertEquals("ABC123", RegExUtils.replacePattern("ABCabc123", "[^A-Z0-9]+", ""));
assertEquals("Lorem_ipsum_dolor_sit",
RegExUtils.replacePattern("Lorem ipsum dolor sit", "( +)([a-z]+)", "_$2"));
}
}