https://bz.apache.org/bugzilla/show_bug.cgi?id=59675
Bug ID: 59675
Summary: CellFormat with Chinese locale doesn't return date in
Chinese
Product: POI
Version: 3.14-dev
Hardware: PC
OS: Mac OS X 10.1
Status: NEW
Severity: normal
Priority: P2
Component: HSSF
Assignee: [email protected]
Reporter: [email protected]
The exact version we're running with right now is 3.14-beta2.
Originally reported on 53494 as a NPE, but the new code doesn't produce the
right result anyway, so I'm reporting it again.
@Test
public void testCellFormatWithChineseDateFormat() throws Exception
{
// -804 here means Chinese.
CellFormat format = CellFormat.getInstance(
"[$-804]mmmm dd yyyy h:mm AM/PM");
// Substitute whatever way you want to create the Date.
// I left our test utilities and joda in the example.
Date date = new DateTime(1973, 11, 27, 11, 28, 0,
DateTestUtils.SYDNEY_TIME).toDate();
// Since the format said to use Chinese, we expect Chinese.
assertThat(
format.apply(date).text,
is(equalTo("十一月 27 1973 11:28 上午")));
}
This fails because the locale is presumably ignored:
Expected :十一月 27 1973 11:28 上午
Actual :November 27 1973 11:28 AM
Another developer noted that if you add this at the start of the text:
LocaleUtil.setUserLocale(Locale.SIMPLIFIED_CHINESE);
(I would prefer a way to set the user locale which doesn't involve a static
method, though, as static methods have a habit of poisoning later tests which
run in unexpected ways. But setting that aside...)
With this, it does generate Chinese, but with a quirk where it seemingly
replaces the last character with an M:
Expected :十一月 27 1973 11:28 上午
Actual :十一月 27 1973 11:28 上M
I tracked this particular quirk down to CellDateFormatter which has some really
baffling code going on.
} else if(it1.getAttribute(Field.AM_PM) != null) {
if(!doneAm1) {
if(this.showAmPm) {
if(this.amPmUpper) {
toAppendTo.append(Character.toUpperCase(ch1));
if(this.showM) {
toAppendTo.append('M');
}
} else {
toAppendTo.append(Character.toLowerCase(ch1));
if(this.showM) {
toAppendTo.append('m');
}
}
}
doneAm1 = true;
}
}
If I breakpoint before this code runs, the current contents of the buffer is
exactly the expected result, and then this code goes and stamps an M at the
end. (lol)
Ideally though I should not even have to set the user locale to get this test
to pass. The cell format itself contains the locale which should be used to
format the value, so that should be used.
--
You are receiving this mail because:
You are the assignee for the bug.
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
