@Test
public void testTCData() {
final InputStream stream = SplitterTest.class.getResourceAsStream("contents.txt");
String text = convertStreamToString(stream);
List<String> words = wordsToCheck(PlainTextSplitter.getInstance(), text);
assertEquals(0, words.size());
}
@Test
public void testSemicolonSeparatedList() {
String text = "properties;test;properties";
correctListToCheck(PlainTextSplitter.getInstance(), text, "properties", "test", "properties");
}
@Test
public void testWordWithUmlauts() {
String text = "rechtsb\u00FCndig";
correctListToCheck(PlainTextSplitter.getInstance(), text, text);
}
@Test
public void testWordInCapsAfterDelimiter3() {
String text = ";BADD;";
correctListToCheck(PlainTextSplitter.getInstance(), text, "BADD");
}
@Test
public void testWordInCapsBeforeDelimiter() {
String text = "BADD,";
correctListToCheck(PlainTextSplitter.getInstance(), text, "BADD");
}
@Test
public void testCheckXmlIgnored() {
String text = "abcdef" + new String(new char[] {0xDC00}) + "test";
correctListToCheck(PlainTextSplitter.getInstance(), text);
}
@Test
public void testWordWithApostrophe1() {
String text = "don't check";
correctListToCheck(PlainTextSplitter.getInstance(), text, "don't", "check");
}
@Test
public void testWordsInSingleQuotesWithSep() {
String text = "'test-something'";
correctListToCheck(PlainTextSplitter.getInstance(), text, "test", "something");
}
@Test
public void testEmailOnly() {
String text = "shkate123-\[email protected]";
correctListToCheck(PlainTextSplitter.getInstance(), text);
}
@Test
public void testEmail() {
String text = "some text with email ([email protected]) inside";
correctListToCheck(
PlainTextSplitter.getInstance(), text, "some", "text", "with", "email", "inside");
}
@Test
public void testTooShort() {
String text = "bgColor carLight";
correctListToCheck(PlainTextSplitter.getInstance(), text, "Color", "Light");
}
@Test
public void testSpecial() {
String text = "test test ⊃ >";
correctListToCheck(PlainTextSplitter.getInstance(), text, "test", "test");
}
@Test
public void testStringLiterals() {
String text = "test\ntest\n";
correctListToCheck(PlainTextSplitter.getInstance(), text, "test", "test");
}
@Test
public void testWordWithApostrophe5() {
String text = "I'm you're we'll";
correctListToCheck(PlainTextSplitter.getInstance(), text, "you're", "we'll");
}
@Test
public void testWordWithApostrophe3() {
String text = "customer's";
correctListToCheck(PlainTextSplitter.getInstance(), text, "customer's");
}
@Test
public void testWordUpperCasedWithUmlautsInTheBeginning() {
String text = "\u00DCNDIG";
correctListToCheck(PlainTextSplitter.getInstance(), text, text);
}
@Test
public void testUrl() {
String text = "https://www.jetbrains.com/idea";
correctListToCheck(PlainTextSplitter.getInstance(), text);
}
@Test
public void testUrlThenSpaces() {
String text = "https://www.jetbrains.com/idea asdasdasd sdfsdf";
correctListToCheck(PlainTextSplitter.getInstance(), text, "asdasdasd", "sdfsdf");
}
@Test
public void testComplexWordsInQuotes() {
String text = "\"test-customer's'\"";
correctListToCheck(PlainTextSplitter.getInstance(), text, "test", "customer's");
}
@Test
public void testWordAfterDelimiter() {
String text = ",badd";
correctListToCheck(PlainTextSplitter.getInstance(), text, "badd");
}
@Test
public void testHexInPlainText() {
String text = "some text 0xacvfgt";
correctListToCheck(PlainTextSplitter.getInstance(), text, "some", "text");
}
@Test
public void testHexInStringLiteral() {
String text = "qwerty 0x12acfgt test";
correctListToCheck(PlainTextSplitter.getInstance(), text, "qwerty", "test");
}
