in fop-core/src/main/java/org/apache/fop/fo/FOText.java [452:522]
private static int isWordChar(char inputChar) {
switch (Character.getType(inputChar)) {
case Character.COMBINING_SPACING_MARK:
return IS_WORD_CHAR_TRUE;
case Character.CONNECTOR_PUNCTUATION:
return IS_WORD_CHAR_TRUE;
case Character.CONTROL:
return IS_WORD_CHAR_FALSE;
case Character.CURRENCY_SYMBOL:
return IS_WORD_CHAR_TRUE;
case Character.DASH_PUNCTUATION:
if (inputChar == '-') {
return IS_WORD_CHAR_TRUE; //hyphen
}
return IS_WORD_CHAR_FALSE;
case Character.DECIMAL_DIGIT_NUMBER:
return IS_WORD_CHAR_TRUE;
case Character.ENCLOSING_MARK:
return IS_WORD_CHAR_FALSE;
case Character.END_PUNCTUATION:
if (inputChar == '\u2019') {
return IS_WORD_CHAR_MAYBE; //apostrophe, right single quote
}
return IS_WORD_CHAR_FALSE;
case Character.FORMAT:
return IS_WORD_CHAR_FALSE;
case Character.LETTER_NUMBER:
return IS_WORD_CHAR_TRUE;
case Character.LINE_SEPARATOR:
return IS_WORD_CHAR_FALSE;
case Character.LOWERCASE_LETTER:
return IS_WORD_CHAR_TRUE;
case Character.MATH_SYMBOL:
return IS_WORD_CHAR_FALSE;
case Character.MODIFIER_LETTER:
return IS_WORD_CHAR_TRUE;
case Character.MODIFIER_SYMBOL:
return IS_WORD_CHAR_TRUE;
case Character.NON_SPACING_MARK:
return IS_WORD_CHAR_TRUE;
case Character.OTHER_LETTER:
return IS_WORD_CHAR_TRUE;
case Character.OTHER_NUMBER:
return IS_WORD_CHAR_TRUE;
case Character.OTHER_PUNCTUATION:
if (inputChar == '\'') {
return IS_WORD_CHAR_MAYBE; //ASCII apostrophe
}
return IS_WORD_CHAR_FALSE;
case Character.OTHER_SYMBOL:
return IS_WORD_CHAR_TRUE;
case Character.PARAGRAPH_SEPARATOR:
return IS_WORD_CHAR_FALSE;
case Character.PRIVATE_USE:
return IS_WORD_CHAR_FALSE;
case Character.SPACE_SEPARATOR:
return IS_WORD_CHAR_FALSE;
case Character.START_PUNCTUATION:
return IS_WORD_CHAR_FALSE;
case Character.SURROGATE:
return IS_WORD_CHAR_FALSE;
case Character.TITLECASE_LETTER:
return IS_WORD_CHAR_TRUE;
case Character.UNASSIGNED:
return IS_WORD_CHAR_FALSE;
case Character.UPPERCASE_LETTER:
return IS_WORD_CHAR_TRUE;
default:
return IS_WORD_CHAR_FALSE;
}
}