This conversion safe? (c++) (Solved)

[JohnOne]

std::string to std::wstring

It works for me, but I've learnt that does not mean a thing in c++.

string sSTRING("A string");
wstring wSTRING(sSTRING.begin(),sSTRING.end());

Edited February 19, 2013 by JohnOne

[jaberwacky]

I can't imagine a good reason for needing to do that.

[Shaggi]

http://www.cplusplus.com/reference/string/basic_string/basic_string/

Specifically:

  Quote

The function template argument InputIterator shall be an input iterator type that points to elements of a type convertible to charT.

where charT = wchar_t for std::wstrings.

The casting and iteration process is safe per-se, but it's not guaranteed to be a valid ascii -> unicode conversion and as of such might output garbage.

This operation is not reversable, as in, you cannot do this from unicode to ascii (it will truncate the wchar_t's to chars).

Edited February 19, 2013 by Shaggi

[JohnOne]

  On 2/19/2013 at 1:45 AM, 'Shaggi said:

http://www.cplusplus.com/reference/string/basic_string/basic_string/

Specifically:

where charT = wchar_t for std::wstrings.

The casting and iteration process is safe per-se, but it's not guaranteed to be a valid ascii -> unicode conversion and as of such might output garbage.

This operation is not reversable, as in, you cannot do this from unicode to ascii (it will truncate the wchar_t's to chars).

Cheers. For the record the other way around from wstring to string also appears to work.

[JohnOne]

  On 2/19/2013 at 1:35 AM, 'jaberwocky6669 said:

I can't imagine a good reason for needing to do that.

Your dreams must be boring.

[Shaggi]

  On 2/19/2013 at 2:01 AM, 'JohnOne said:

Cheers. For the record the other way around from wstring to string also appears to work.

Possibly if your wstring only contains characters from the ascii set, which in both standards are defined as the first 255. As soon as you get past that, you have troubles.

eg. the letter 'Ā', in binary:

0000000100000000

will in a conversion done like this:

wchar_t my_wc = 'Ā';
char my_c = (char)my_wc;

be truncated to:

00000000

Or, null, which is the common string stop nominator (even though it works with std::strings .. you suddenly have unwanted null's in your string)

[JohnOne]

Ah! yes, cheers, I always forget because I have never had a need to go beyond ascii.

Must start taking these things into consideration.

Thanks Shaggi.

[Ascend4nt]

It's already been stated that going from wide to narrow strings is a bad idea, however if you are certain the characters are ascii or some ANSI subset, you can use a conversion function. Check out these C++11 conversion functions, from 'The Standard C++ Library, 2nd Ed':

wstring2string (and vice versa)

Also more appropriately, you can convert wide strings to UTF-8, and keep that in a std::string:

wstring2utf8

You can also go with widen and narrow for individual characters.

[Shaggi]

  On 2/19/2013 at 2:26 AM, 'JohnOne said:

Ah! yes, cheers, I always forget because I have never had a need to go beyond ascii.

Must start taking these things into consideration.

Thanks Shaggi.

No problem. But yes, keep in mind most conversions / copies in C++ are done bitwise and subject to conversion penalties (you may or may not notice for a long time ) unless you define another interface (this takes some getting used to when you're working with pointers and containers...)

Anyway, the compiler should have emitted a warning when you did wstring -> string, depending on the code in the library.. If they did an explicit casting, it might not have done it. All those things that happen without you know it

Edited February 19, 2013 by Shaggi