Re: nsEmbedString

[Darin Fisher]

banli,

nsEmbedString is to be used when you need to pass a string to a XPCOM method expecting a |const nsAString&|. you can used nsEmbedString like this:

{
  const PRUnichar data[] = {'h','e','l','l','o','\0'};
  nsEmbedString s(data);

  nsCOMPtr<nsIFoo> foo = do_CreateInstance(NS_FOO_CONTRACTID);
  foo->SetData(s);
  ...
}

Assuming, nsIFoo is some XPCOM interface with a SetData(const nsAString&) method.

I think the tricky thing is generating the PRUnichar array under Linux. wchar_t is typically not compatible with PRUnichar under Linux. In fact, wchar_t is commonly UTF-32, while PRUnichar is UTF-16. You therefore need to write your own code to convert from UTF-32 to UTF-16.

I think the code bsmedberg showed you would work if you only care about supporting the UCS-2 subset of UTF-16. Under Linux, you can use the iconv function to convert strings from one charset to another.

Attached is some sample code that you can use to convert between wchar_t arrays and nsAString objects. I've not tested this code extensively, so I don't make any promises that it is correct. But, it seems to work for me.

NOTE: This sample code requires a recent Mozilla trunk. It won't work with any existing Mozilla releases. However, it should not be too hard to make it work with an older release.

-Darin





banli wrote:

Under linux, is there some c++  API which can output string of type
nsEmbedString to file? So that I don't need the conversion when I use
"fputws"?

In the cxc book, nsEmbedString and nsEmbedCString are mentioned. In the
string-guide document, it mentions that "nsString or nsSharableString for
member variables". But in the mozilla source code, nsString is under
"/xpcom/string/obsolete" directory. So if I need some string variable to
feed into some interface method to get return value of type of nsAString,
which string type I should choose to declare string variable?  thanks!

"Benjamin D. Smedberg" <[EMAIL PROTECTED]> wrote
news:[EMAIL PROTECTED]

Unfortunately, there is not a way to to UTF16->UTF8 conversions without
linking to non-frozen stuff (yet). Darin and I are trying to figure out
the minimum necessary linkage to accomplish these kinds of tasks.

You can't cast PRUnichar* to wchar_t* because on most *ixes, wchar_t is
a four-byte type. However, you can safely do a funky copy-conversion to
a wchar_t type:

PRUnichar *start = nsStr.get();
PRUnichar *end nsStr.get() + nsStr.Length();
wchar_t *wstart = new wchar_t[nsStr.Length()];
wchar_t *wend = start + nsStr.Length();

for(; start < end; ++start) {
  *wstart = (wchar_t) *start;
  ++wstart;
}
*wstart = 0;

There are also some interfaces in intl (charset conversion) that might
do the trick for you.

--BDS

test_iconv.tar.gz

Description: Unix tar archive