@Test
public void verifyNullByteArray() throws Exception {
for (Map.Entry<String, Charset> entry : CharsetUtil.getCharsetMap().entrySet()) {
// test that the byte array wasn't changed
Assert.assertEquals(
"Charset " + entry.getKey() + " impl bad -- did not return null",
"",
CharsetUtil.decode(null, entry.getValue()));
}
}
@Test
public void verifyDecodeDoesNotChangeByteArray() throws Exception {
for (Map.Entry<String, Charset> entry : CharsetUtil.getCharsetMap().entrySet()) {
byte[] bytes = new byte[] {(byte) 0x40, (byte) 0x5F, (byte) 0x24, (byte) 0x78, (byte) 0x02};
byte[] expectedBytes = Arrays.copyOf(bytes, bytes.length);
String str0 = CharsetUtil.decode(bytes, entry.getValue());
// test that the byte array wasn't changed
Assert.assertArrayEquals(
"Charset " + entry.getKey() + " impl bad -- modified byte array parameter",
expectedBytes,
bytes);
}
}
@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));
}