Revision da9ff9c4
Added by Andreas Müller about 5 years ago
| cdmlib-commons/src/main/java/eu/etaxonomy/cdm/common/CdmUtils.java | ||
|---|---|---|
| 467 | 467 |
return false; |
| 468 | 468 |
} |
| 469 | 469 |
|
| 470 |
/** |
|
| 471 |
* Removes all non-word character (i.e. *, @, %, line breaks, etc.) |
|
| 472 |
* from the given string |
|
| 473 |
* @param string the string which should be trimmed |
|
| 474 |
* @return the trimmed string |
|
| 475 |
*/ |
|
| 476 |
static public String trimNonWordCharacters(String string){
|
|
| 477 |
return replaceNonWordCharacters(string, ""); |
|
| 478 |
} |
|
| 479 |
|
|
| 480 |
/** |
|
| 481 |
* Replaces all non-word character (i.e. *, @, %, line breaks, etc.) |
|
| 482 |
* with the given replacement string |
|
| 483 |
* @param string the string which should be trimmed |
|
| 484 |
* @param replacement the replacement for the non-word characters |
|
| 485 |
* @return the trimmed string |
|
| 486 |
*/ |
|
| 487 |
static public String replaceNonWordCharacters(String string, String replacement){
|
|
| 488 |
return string.replaceAll("\\W", replacement);
|
|
| 489 |
} |
|
| 490 |
|
|
| 491 | 470 |
/** |
| 492 | 471 |
* Returns <code>false</code> if string is null, "" or string.trim() is "" |
| 493 | 472 |
* @see isNotEmpty(String string) |
| ... | ... | |
| 686 | 665 |
return true; |
| 687 | 666 |
} |
| 688 | 667 |
|
| 668 |
/** |
|
| 669 |
* Transforms a search string which allows wildcard "*" into a |
|
| 670 |
* java regular expression such that all other characters are handled as normal text. |
|
| 671 |
* @param regEx |
|
| 672 |
* @return |
|
| 673 |
*/ |
|
| 674 |
public static String quoteRegExWithWildcard(String regEx){
|
|
| 675 |
return Pattern.quote(regEx).replace("*", "\\E.*\\Q").replace("\\Q\\E", "");
|
|
| 676 |
} |
|
| 677 |
|
|
| 689 | 678 |
} |
| cdmlib-commons/src/test/java/eu/etaxonomy/cdm/common/CdmUtilsTest.java | ||
|---|---|---|
| 14 | 14 |
import java.io.File; |
| 15 | 15 |
import java.io.IOException; |
| 16 | 16 |
import java.io.InputStream; |
| 17 |
import java.util.regex.Pattern; |
|
| 17 | 18 |
|
| 18 | 19 |
import org.apache.log4j.Level; |
| 19 | 20 |
import org.apache.log4j.Logger; |
| ... | ... | |
| 107 | 108 |
|
| 108 | 109 |
} |
| 109 | 110 |
|
| 111 |
@Test |
|
| 112 |
public void testquoteRegExWithWildcard(){
|
|
| 113 |
String regExBase = ".(*$[ms^"; |
|
| 114 |
String regEx = CdmUtils.quoteRegExWithWildcard(regExBase); |
|
| 115 |
Assert.assertEquals("\\Q.(\\E.*\\Q$[ms^\\E", regEx);
|
|
| 116 |
boolean matches = ".(*$[ms^".matches(regEx); |
|
| 117 |
Assert.assertTrue(matches); |
|
| 118 |
matches = ".(aaaaaa$[ms^".matches(regEx); |
|
| 119 |
Assert.assertTrue(matches); |
|
| 120 |
matches = "b(aaaaaa$[ms^".matches(regEx); |
|
| 121 |
Assert.assertFalse(matches); |
|
| 122 |
|
|
| 123 |
regEx = CdmUtils.quoteRegExWithWildcard("*abc*");
|
|
| 124 |
Assert.assertEquals(".*\\Qabc\\E.*", regEx);
|
|
| 125 |
Assert.assertTrue("abc".matches(regEx));
|
|
| 126 |
Assert.assertTrue("a80/(--e*wabc?äe".matches(regEx));
|
|
| 127 |
|
|
| 128 |
} |
|
| 129 |
|
|
| 110 | 130 |
/** |
| 111 | 131 |
* This test can be used for functional testing of any task but should |
| 112 | 132 |
* never be committed when failing. |
| 113 | 133 |
*/ |
| 114 | 134 |
@Test |
| 115 | 135 |
public void testSomething(){
|
| 116 |
String MCL = "MCL[0-9]{1,3}(\\-[0-9]{1,4}(\\-[0-9]{1,4}(\\-[0-9]{1,3}(\\-[0-9]{1,3})?)?)?)?";
|
|
| 117 |
// String MCL = "a{1,3}";
|
|
| 118 |
String filter = "Acc "+MCL; |
|
| 119 |
|
|
| 120 |
String notes = "Acc: 0x is Hieracium djimilense subsp. neotericum Zahn MCL293-3140-00-630"; |
|
| 121 |
String result; |
|
| 122 |
if (notes.matches("Acc:.*")){
|
|
| 123 |
if (notes.matches("Acc: .*\\$$") || (notes.matches("Acc: .*"+MCL))){
|
|
| 124 |
result = null; |
|
| 125 |
}else if (notes.matches("Acc: .*(\\$|"+MCL+")\\s*\\{.*\\}")){
|
|
| 126 |
notes = notes.substring(notes.indexOf("{")+1, notes.length()-1);
|
|
| 127 |
result = notes; |
|
| 128 |
}else if (notes.matches("Acc: .*(\\$|"+MCL+")\\s*\\[.*\\]")){
|
|
| 129 |
notes = notes.substring(notes.indexOf("[")+1, notes.length()-1);
|
|
| 130 |
result = notes; |
|
| 131 |
}else{
|
|
| 132 |
logger.warn("Namenote: " + notes);
|
|
| 133 |
result = notes; |
|
| 134 |
} |
|
| 135 |
}else if (notes.matches("Syn:.*")){
|
|
| 136 |
if (notes.matches("Syn: .*\\$$") || (notes.matches("Syn: .*"+MCL))){
|
|
| 137 |
result = null; |
|
| 138 |
}else if (notes.matches("Syn: .*(\\$|"+MCL+")\\s*\\{.*\\}")){
|
|
| 139 |
notes = notes.substring(notes.indexOf("{")+1, notes.length()-1);
|
|
| 140 |
result = notes; |
|
| 141 |
}else if (notes.matches("Syn: .*(\\$|"+MCL+")\\s*\\[.*\\]")){
|
|
| 142 |
notes = notes.substring(notes.indexOf("[")+1, notes.length()-1);
|
|
| 143 |
result = notes; |
|
| 144 |
}else{
|
|
| 145 |
logger.warn("Namenote: " + notes);
|
|
| 146 |
result = notes; |
|
| 147 |
} |
|
| 148 |
}else{
|
|
| 149 |
result = notes; |
|
| 150 |
} |
|
| 151 |
System.out.println(result); |
|
| 136 |
String str = ".(*$[ms^"; |
|
| 137 |
String patQuote = Pattern.quote("str");
|
|
| 138 |
// System.out.println(patQuote); |
|
| 139 |
// String matchQuote = Matcher.quoteReplacement(str); |
|
| 140 |
// System.out.println(matchQuote); |
|
| 141 |
// System.out.println(CdmUtils.quoteRegExWithWildcard(str)); |
|
| 152 | 142 |
} |
| 153 | 143 |
|
| 144 |
|
|
| 145 |
|
|
| 146 |
|
|
| 147 |
|
|
| 154 | 148 |
} |
Also available in: Unified diff
ref #8041 correct regex quotation method implemented in CdmUtils