Development/String Classes

    From The Document Foundation Wiki

    Throughout the codebase are two different families of string classes in use. Each of them has different features but efforts are ongoing to limit the use to one string family, with OUString for UTF-16 strings, and OString for 8bit strings. For better or worse, OString/OUString is modelled after the immutable Java String class, except with a contradictory additional operator+=

    Favored Family, C++

    Favored Family, C

    Favored Family, C++ Helpers

    Favored Family, C Helpers

    Namespacing of OUString/OString

    The rtl::OUString and rtl::OString classes are in the rtl namespace, but since they are already in practice namespaced by the O- prefix, for ease of reading it is not necessary to explicitly refer to the rtl namespace when using them in code that is not part of public API (i.e. most of the codebase).

    Conversion Macros between OUString and OString

    • OString aOString = OUStringToOString( aOUString, RTL_TEXTENCODING_UTF8 );
    • OUString aOUString = OStringToOUString( aOString, RTL_TEXTENCODING_UTF8 );

    Accessing the char* Data Buffer

    Use SAL_DEBUG or SAL_INFO for debug output.

    If you want to dump these strings with gdb, see Development/How to debug#Dumping Strings (etc)

    Iterating through an OUString

    Since sal_Unicode is a 16 bits type, using a simple for loop with aString[n] might cause problem UTF-16 codepoints coded on two words. If you need to iterate on an OUString that might contain such codepoints, use rtl::rtl_uString_iterateCodePoints or aString.iterateCodePoints(&index).

    Iterating through a String

    There are many loops over the UTF-16 code units in a String, and usually the loops use index variables of type xub_Strlen (typedef for unsigned short). Since the OUString has a higher capacity these loops must be converted so that the index variable is sal_Int32. It may be even better to use OUString.iterateCodePoints, which iterates over Unicode code points; see previous section.

    Favored Methods Utilizing String Literals

    Both rtl::OString and rtl::OUString have specialized overloads for string literals that treat them efficiently, and string literals are automatically converted to rtl::OString/rtl::OUString when needed, or rtl::OString/rtl::OUString functions have overloads to handle them:

    if (string == "foo")

    functionTakingOUString( "bar" ); // and just "" instead of OUString() for empty string

    if( string.endsWith( "foo" )) ...

    Please note that this auto-conversion does not work in the following cases, and you still need an explicit OUString constructor:

    • bFoo ? "bar" : "baz" [ you need bFoo ? OUString("bar") : OUString("baz"), and so on ]
    • return "foo"
    • aAny << "foo"
    • CPPUNIT_ASSERT_EQUAL("foo")

    The RTL_CONSTASCII_STRINGPARAM and RTL_CONSTASCII_USTRINGPARAM macros are obsolete, and so are the various *AsciiL functions.

    Empty string can be simply written as "" instead of OUString().