@Test
public void normalize() throws Exception {
String in = null;
// try every charset with simple A-Z, a-z, and 0-9 which should work
// in all charsets
in = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefjhijklmnopqrstuvwxyz01234567890?&@";
for (Map.Entry<String, Charset> entry : CharsetUtil.getCharsetMap().entrySet()) {
Assert.assertEquals(
"Charset " + entry.getKey() + " implementation broken",
in,
CharsetUtil.normalize(in, entry.getValue()));
}
in = "\u20AC"; // euro currency char (only supported in a couple charsets)
Assert.assertEquals("\u20AC", CharsetUtil.normalize(in, CharsetUtil.CHARSET_GSM));
Assert.assertEquals("\u20AC", CharsetUtil.normalize(in, CharsetUtil.CHARSET_PACKED_GSM));
Assert.assertEquals("\u20AC", CharsetUtil.normalize(in, CharsetUtil.CHARSET_AIRWIDE_GSM));
Assert.assertEquals("\u20AC", CharsetUtil.normalize(in, CharsetUtil.CHARSET_VFD2_GSM));
Assert.assertEquals("\u20AC", CharsetUtil.normalize(in, CharsetUtil.CHARSET_VFTR_GSM));
Assert.assertEquals("?", CharsetUtil.normalize(in, CharsetUtil.CHARSET_ISO_8859_1));
Assert.assertEquals("\u20AC", CharsetUtil.normalize(in, CharsetUtil.CHARSET_ISO_8859_15));
Assert.assertEquals("\u20AC", CharsetUtil.normalize(in, CharsetUtil.CHARSET_UCS_2));
Assert.assertEquals("\u20AC", CharsetUtil.normalize(in, CharsetUtil.CHARSET_UTF_8));
in = "\u6025"; // arabic char (only supported in a couple charsets)
Assert.assertEquals("?", CharsetUtil.normalize(in, CharsetUtil.CHARSET_GSM));
Assert.assertEquals("?", CharsetUtil.normalize(in, CharsetUtil.CHARSET_PACKED_GSM));
Assert.assertEquals("?", CharsetUtil.normalize(in, CharsetUtil.CHARSET_AIRWIDE_GSM));
Assert.assertEquals("?", CharsetUtil.normalize(in, CharsetUtil.CHARSET_VFD2_GSM));
Assert.assertEquals("?", CharsetUtil.normalize(in, CharsetUtil.CHARSET_VFTR_GSM));
Assert.assertEquals("?", CharsetUtil.normalize(in, CharsetUtil.CHARSET_ISO_8859_1));
Assert.assertEquals("?", CharsetUtil.normalize(in, CharsetUtil.CHARSET_ISO_8859_15));
Assert.assertEquals("\u6025", CharsetUtil.normalize(in, CharsetUtil.CHARSET_UCS_2));
Assert.assertEquals("\u6025", CharsetUtil.normalize(in, CharsetUtil.CHARSET_UTF_8));
}
