public void testMultiCharEsc2() throws EXIException {
String regex = "\\S";
EXIRegularExpression re = new EXIRegularExpression(regex);
assertTrue(re.isEntireSetOfXMLCharacters());
}
public void testNonUnicode3_1_0() throws Exception {
EXIRegularExpression re = new EXIRegularExpression("\uFFFF");
assertFalse(re.isEntireSetOfXMLCharacters());
Set<Integer> codePoints = re.getCodePoints();
assertTrue("CP Size=" + codePoints.size(), codePoints.size() == 1);
}
public void testHuge1() throws Exception {
EXIRegularExpression re = new EXIRegularExpression("\\p{L}");
assertTrue(re.isEntireSetOfXMLCharacters());
// Set<Integer> codePoints = re.getCodePoints();
// assertTrue(codePoints.size() > 45000);
// assertTrue(re.isEntireSetOfXMLCharacters());
}
public void testProdNumType() throws Exception {
// \d{3}-[A-Z]{2}|\d{7}
EXIRegularExpression re = new EXIRegularExpression("\\d{3}-[A-Z]{2}|\\d{7}");
// multi characters ...
assertTrue(re.isEntireSetOfXMLCharacters());
}
public void testComplEscape2() throws EXIException {
String regex = "\\P{Zs}";
EXIRegularExpression re = new EXIRegularExpression(regex);
assertTrue(re.isEntireSetOfXMLCharacters());
}
public void testPattern22() throws EXIException {
// Pattern in OpenOffice
// ([$]?([^\. ']+|'[^']+'))?\.[$]?[A-Z]+[$]?[0-9]+
String regex = "([$]?([^\\. ']+|'[^']+'))?\\.[$]?[A-Z]+[$]?[0-9]+";
EXIRegularExpression re = new EXIRegularExpression(regex);
assertTrue(re.isEntireSetOfXMLCharacters());
}
public void testPattern17() throws EXIException {
// complexEsc \P{ L }
String regex = "abc\\P{L}def";
EXIRegularExpression re = new EXIRegularExpression(regex);
assertTrue(re.isEntireSetOfXMLCharacters());
}
public void testBasicLatin() throws Exception {
// \p{IsBasicLatin}
EXIRegularExpression re = new EXIRegularExpression("\\p{IsBasicLatin}");
// block ...
assertTrue(re.isEntireSetOfXMLCharacters());
}
public void testPattern15() throws EXIException {
// Number decimal digit
String regex = "\\p{Nd}";
EXIRegularExpression re = new EXIRegularExpression(regex);
assertTrue(re.isEntireSetOfXMLCharacters());
}
public void testPattern16() throws EXIException {
//
String regex = "\\i\\c*";
EXIRegularExpression re = new EXIRegularExpression(regex);
assertTrue(re.isEntireSetOfXMLCharacters());
}
public void testCategoryEscape3() throws EXIException {
String regex = "\\p{IsBasicLatin}";
EXIRegularExpression re = new EXIRegularExpression(regex);
assertTrue(re.isEntireSetOfXMLCharacters());
}
public void testPattern30() throws EXIException {
// <xs:pattern value="\\c"/> --> means '\' followed by 'c'
// Note: Nothing to do with Multiple Character Escape Sequences \c
EXIRegularExpression re = new EXIRegularExpression("\\\\c");
assertFalse(re.isEntireSetOfXMLCharacters());
assertTrue(re.getCodePoints().size() == 2);
}
public void testSubtraction3() throws Exception {
// [\p{Ll}\p{Lu}-[\p{IsBasicLatin}]] matches all uppercase and lowercase
// Unicode letters, except any ASCII letters
EXIRegularExpression re = new EXIRegularExpression("[\\p{Ll}\\p{Lu}-[\\p{IsBasicLatin}]]");
// catEsc, block & ...
assertTrue(re.isEntireSetOfXMLCharacters());
}
@Test
public void testRangeSpecialChars() throws Exception {
// UnicodeData-3.1.0: 03D7 .. 03DA (2 chars)
// UnicodeData-3.2.0: 03D7 .. 03DA (4 chars, 03D8 & 03D9)
EXIRegularExpression re = new EXIRegularExpression("[\u03D7-\u03DA]");
assertFalse(re.isEntireSetOfXMLCharacters());
assertTrue(re.getCodePoints().size() == 4);
}
public void testPattern24() throws EXIException, IOException {
// --0\d --0\d --0\d --0\d --\d1 --0\d --0\d
String regex = "--0\\d --0\\d --0\\d --0\\d --\\d1 --0\\d --0\\d";
EXIRegularExpression re = new EXIRegularExpression(regex);
// multi characters ...
assertTrue(re.isEntireSetOfXMLCharacters());
}
public void testPattern21() throws EXIException {
// Partial Pattern in OpenOffice
// ([$]?([^\. ']+|'[^']+'))
String regex = "([$]?([^\\. ']+|'[^']+'))";
EXIRegularExpression re = new EXIRegularExpression(regex);
// TODO.. what is it ?
assertTrue(re.isEntireSetOfXMLCharacters());
}
public void testMultiCharEsc1() throws EXIException {
String regex = "\\s";
EXIRegularExpression re = new EXIRegularExpression(regex);
assertFalse(re.isEntireSetOfXMLCharacters());
Set<Integer> codePoints = re.getCodePoints();
assertTrue("CP Size=" + codePoints.size(), codePoints.size() == 4);
}
public void testSSN() throws Exception {
EXIRegularExpression re = new EXIRegularExpression("[0-9]{3}-[0-9]{2}-[0-9]{4}");
assertTrue(re.getCodePoints().size() == 11);
RestrictedCharacterSet rcs = new CodePointCharacterSet(re.getCodePoints());
assertTrue(rcs.getCode('5') == 6);
assertTrue(rcs.getCodePoint(3) == '2');
assertTrue(rcs.getCodingLength() == 4);
}
public void testAorBorCorMinus() throws Exception {
EXIRegularExpression re = new EXIRegularExpression("A|B|C|-");
assertTrue(re.getCodePoints().size() == 4);
RestrictedCharacterSet rcs = new CodePointCharacterSet(re.getCodePoints());
assertTrue(rcs.getCode('C') == 3);
assertTrue(rcs.getCodePoint(0) == '-');
assertTrue(rcs.getCodingLength() == 3);
}
public void testMaleFemale() throws Exception {
EXIRegularExpression re = new EXIRegularExpression("male|female");
// aeflm
assertTrue(re.getCodePoints().size() == 5);
RestrictedCharacterSet rcs = new CodePointCharacterSet(re.getCodePoints());
assertTrue(rcs.getCode('a') == 0);
assertTrue(rcs.getCodePoint(3) == 'l');
assertTrue(rcs.getCodingLength() == 3);
}
public void testPattern10() throws EXIException {
// 8 restricted digits
String regex = "[2-4]{8}";
EXIRegularExpression re = new EXIRegularExpression(regex);
RestrictedCharacterSet rcs = new CodePointCharacterSet(re.getCodePoints());
assertTrue(rcs.getCodePoint(0) == '2');
assertTrue(rcs.getCode('4') == 2);
assertTrue(rcs.getCodingLength() == 2);
}
public void testPattern11() throws EXIException {
// 111 restricted digits
String regex = "[3-9]{111}";
EXIRegularExpression re = new EXIRegularExpression(regex);
RestrictedCharacterSet rcs = new CodePointCharacterSet(re.getCodePoints());
assertTrue(rcs.getCodePoint(1) == '4');
assertTrue(rcs.getCode('4') == 1);
assertTrue(rcs.getCodingLength() == 3);
}
public void testRange2() throws Exception {
EXIRegularExpression re = new EXIRegularExpression("[A-Z][B-Z][C-Z]");
assertTrue(re.getCodePoints().size() == 26);
RestrictedCharacterSet rcs = new CodePointCharacterSet(re.getCodePoints());
assertTrue(rcs.getCode('A') == 0);
assertTrue(rcs.getCode('L') == 11);
assertTrue(rcs.getCode('Z') == 25);
assertTrue(rcs.getCodingLength() == 5);
}
public void testRange7() throws EXIException {
// ONE of the following letters: x, y, OR z:
String regex = "[xyz]"; // e.g. "x" "y" "z"
EXIRegularExpression re = new EXIRegularExpression(regex);
RestrictedCharacterSet rcs = new CodePointCharacterSet(re.getCodePoints());
assertTrue(rcs.getCodePoint(0) == 'x');
assertTrue(rcs.getCodePoint(1) == 'y');
assertTrue(rcs.getCodePoint(2) == 'z');
assertTrue(rcs.getCodingLength() == 2);
}
public void testPattern26() throws EXIException {
// "𐿿" NON BMP
String test =
new StringBuilder().append("[ABC").appendCodePoint(0x10FFF).append("]{1}").toString();
// EXIRegularExpression re = new
// EXIRegularExpression("[ABC𐿿]{1}");
EXIRegularExpression re = new EXIRegularExpression(test);
assertTrue(re.isEntireSetOfXMLCharacters());
// // A, B, and C ?
// assertTrue(re.getCodePoints().size() == 3);
}
public void testRange8() throws EXIException {
// zero or more occurrences of lowercase letters from a to z:
String regex = "([a-z])*";
EXIRegularExpression re = new EXIRegularExpression(regex);
RestrictedCharacterSet rcs = new CodePointCharacterSet(re.getCodePoints());
assertTrue(rcs.getCodePoint(0) == 'a');
assertTrue(rcs.getCode('c') == 2);
assertTrue(rcs.getCodePoint(25) == 'z');
assertTrue(rcs.getCodingLength() == 5);
}
public void testPattern13() throws EXIException {
//
String regex = "\\s";
EXIRegularExpression re = new EXIRegularExpression(regex);
RestrictedCharacterSet rcs = new CodePointCharacterSet(re.getCodePoints());
assertTrue(rcs.getCodePoint(0) == '\t');
assertTrue(rcs.getCode(' ') == 3);
assertTrue(rcs.getCode('?') == Constants.NOT_FOUND);
assertTrue(rcs.getCodingLength() == 3);
}
public void testSubtraction2() throws Exception {
// matches any character in the string 0123789
EXIRegularExpression re = new EXIRegularExpression("[0-9-[0-6-[0-3]]]");
Set<Integer> codePoints = re.getCodePoints();
assertTrue(codePoints.size() == 7);
RestrictedCharacterSet rcs = new CodePointCharacterSet(re.getCodePoints());
assertTrue(rcs.getCode('7') == 4);
assertTrue(rcs.getCodePoint(5) == '8');
assertTrue(rcs.getCodingLength() == 3);
}
public void testRange4() throws EXIException {
String regex = "[0-9][0-9][0-9][0-9][0-9]"; // e.g. "12345"
EXIRegularExpression re = new EXIRegularExpression(regex);
assertTrue(re.getCodePoints().size() == 10);
RestrictedCharacterSet rcs = new CodePointCharacterSet(re.getCodePoints());
assertTrue(rcs.getCode('0') == 0);
assertTrue(rcs.getCode('5') == 5);
assertTrue(rcs.getCode('9') == 9);
assertTrue(rcs.getCodingLength() == 4);
}
public void testPattern18() throws EXIException {
// language --> ([a-zA-Z]{1,8})(-[a-zA-Z0-9]{1,8})*
String regex = "([a-zA-Z]{1,8})(-[a-zA-Z0-9]{1,8})*";
EXIRegularExpression re = new EXIRegularExpression(regex);
RestrictedCharacterSet rcs = new CodePointCharacterSet(re.getCodePoints());
assertTrue(rcs.getCodePoint(0) == '-');
assertTrue(rcs.getCodePoint(1) == '0');
assertTrue(rcs.getCode('8') == 9);
assertTrue(rcs.getCode('?') == Constants.NOT_FOUND);
assertTrue(rcs.getCodingLength() == 6);
}
