I am trying to use a DLL with an ANSI C compiler. One of the DLL functions takes a void pointer. In some sample Windows code that was provided with the DLL, the struct that gets passed to the function is defined as having three CString entities. I have told the author of the DLL that they should not be passing MFC classes through their DLL functions. They have told me just to replace the CString declarations in the struct with char arrays and it should be fine. I'm 99% sure that's wrong, but since I don't have VC++ and don't have any experience with MFC, and since I've seen some posts saying LPTSTR can be used in place of CString (What is `CString`?), I'm starting to wonder if I'm wrong.
Can someone please confirm for me that CString and LPTSTR are not interchangeable as arguments to a function? If you can provide the source for the definition of the CString class, that would be helpful so I can send it to the DLL's author and explain that the memory footprint of a char array is not the same as a CString class, and that you can't pass a pointer to a struct that was defined with char arrays and then treat it as a bunch of CString objects.
CString is an alias of the CStringT class template. Objects of this class are really better not to pass to the DLL. The character type of the string class can be TCHAR (for both ANSI and Unicode character strings - see explanation below). The definition of CString (and CStringT) can most likely be found in the atlstr.h header file.
LPTSTR is a regular pointer to a sequence of characters. The data type (TCHAR*) depends on the settings of the development environment: if the "Use Unicode Character set" option is selected, the TCHAR data type will be wchar_t (and LPTSTR will be wchar_t*, respectively). If the "Use Multi-byte character set" is selected, the TCHAR will be defined as char (and LPTSTR will be char*).
So the question about interchangeability between CString and LPTSTR is not so simple. It also depends on how the DLL is written. If the DLL was designed with the same environment settings as the main program, then CString and LPTSTR can really be interchangeable.
Also, remember that CStirng is a class with many methods, while LPTSTR is just a pointer.
Related
I'm learning to write and use DLLs and this is my first attempt at exporting a function from my dll. It works, but this line is what gave me trouble and what I've been able to find regarding the TEXT cast for UNICODE and ANSI I think I need some guidance. As far as I can find this question has not been asked elsewhere on the site so I apologize if anyone finds what I couldn't.
HINSTANCE hInstLibrary = LoadLibrary("MyDLL.dll");
My initial usage, from a short tutorial on explicit linking gives E0167 and C2664 errors regarding LPCWSTR type
HINSTANCE hInstLibrary = LoadLibrary(TEXT("MyDLL.dll"));
Casting the string to TEXT solves the problem, though I'm not sure why and would like to know
HINSTANCE hInstLibrary = LoadLibraryA("MyDLL.dll");
The line I decided to use in the working example. LoadLibraryA() expands LoadLibrary to accept ANSI rather than Wide, which may be the root of my misunderstanding. Why is this necessary when most examples I find show LoadLibrary("NameOfDLL.dll")?
Why does the string not satisfy the standard LoadLibrary() call?
LoadLibrary() is a preprocessor macro. It maps to either LoadLibraryW() or LoadLibraryA() depending on whether UNICODE is defined or not, respectively. LoadLibraryW() takes a const wchar_t* string as input, while LoadLibraryA() takes a const char * string instead.
The string literal "MyDLL.dll" is a const char[10], which decays into a const char *. If UNICODE is defined, LoadLibrary("MyDLL.dll") will fail to compile, as you cannot pass a const char * where a const wchar_t * is expected.
TEXT() is also a preprocessor macro. If UNICODE is defined, it appends an L prefix to the specified literal making the literal use wchar_t, otherwise no prefix is added and the literal uses char instead.
Thus, if UNICODE is defined, then LoadLibrary(TEXT("MyDLL.dll")) is compiled as LoadLibraryW(L"MyDLL.dll"), otherwise it is compiled as LoadLibraryA("MyDLL.dll") instead.
A majority of Win32 APIs that deal with textual data have similar A and W versions, and corresponding UNICODE-aware preprocessor macros. So, when using character/string literals with these APIs, you should always use the TEXT() macro. Otherwise, just use the A and W APIs directly as needed, depending on the type of textual data you are working with.
I'm learning to write and use DLLs and this is my first attempt at exporting a function from my dll. It works, but this line is what gave me trouble and what I've been able to find regarding the TEXT cast for UNICODE and ANSI I think I need some guidance. As far as I can find this question has not been asked elsewhere on the site so I apologize if anyone finds what I couldn't.
HINSTANCE hInstLibrary = LoadLibrary("MyDLL.dll");
My initial usage, from a short tutorial on explicit linking gives E0167 and C2664 errors regarding LPCWSTR type
HINSTANCE hInstLibrary = LoadLibrary(TEXT("MyDLL.dll"));
Casting the string to TEXT solves the problem, though I'm not sure why and would like to know
HINSTANCE hInstLibrary = LoadLibraryA("MyDLL.dll");
The line I decided to use in the working example. LoadLibraryA() expands LoadLibrary to accept ANSI rather than Wide, which may be the root of my misunderstanding. Why is this necessary when most examples I find show LoadLibrary("NameOfDLL.dll")?
Why does the string not satisfy the standard LoadLibrary() call?
LoadLibrary() is a preprocessor macro. It maps to either LoadLibraryW() or LoadLibraryA() depending on whether UNICODE is defined or not, respectively. LoadLibraryW() takes a const wchar_t* string as input, while LoadLibraryA() takes a const char * string instead.
The string literal "MyDLL.dll" is a const char[10], which decays into a const char *. If UNICODE is defined, LoadLibrary("MyDLL.dll") will fail to compile, as you cannot pass a const char * where a const wchar_t * is expected.
TEXT() is also a preprocessor macro. If UNICODE is defined, it appends an L prefix to the specified literal making the literal use wchar_t, otherwise no prefix is added and the literal uses char instead.
Thus, if UNICODE is defined, then LoadLibrary(TEXT("MyDLL.dll")) is compiled as LoadLibraryW(L"MyDLL.dll"), otherwise it is compiled as LoadLibraryA("MyDLL.dll") instead.
A majority of Win32 APIs that deal with textual data have similar A and W versions, and corresponding UNICODE-aware preprocessor macros. So, when using character/string literals with these APIs, you should always use the TEXT() macro. Otherwise, just use the A and W APIs directly as needed, depending on the type of textual data you are working with.
I'm working on a win32 project with CStrings (console application), and I noticed something odd when I want to pass to a function (like strtok_s for example) a LPSTR pointer from a CString with the method GetBuffer(), this last one instead of giving me a LPSTR, it gave me a LPWSTR (a pointer to a wide string)... CString is supposed to store 8 bit chars isn't it ?
I'm obliged in some cases to use CStringA for example to be able for example to use the method Find() because with a CString my input string must be a wide one. But in other another project (windowed program), I don't have this problem, i'm suspecting the headers (when I use afxstr.h "Find" works with a normal string, but not with afxcoll.h...)
Usually I work with std::string that's why I'm lost.
CString is a typdef, declared as (afxstr.h):
typedef ATL::CStringT< TCHAR, StrTraitMFC< TCHAR > > CString;
// Or, when using the MFC DLL
typedef ATL::CStringT< TCHAR, StrTraitMFC_DLL< TCHAR > > CString;
Depending on what TCHAR is, a CString stores either an ANSI (MBCS) or Unicode string. There are also explicit instantiations of the CStringT template: CStringW and CStringA.
Either type has a conversion constructor, taking a constant pointer to the respective other character encoding. In other words: You can construct a CStringW from an ANSI (MBCS) string, as well as a CStringA from a UTF-16LE-encoded Unicode string.
If you need to be explicit about the character encoding, use either CStringW or CStringA.
Full documentation for CString is available at CStringT Class.
since I can get hands on the new RAD Studio Xe4 I thought I'd give it a try.
Unfortunatly, I am not so experienced with C++ and therefore I was wondering why the Code that works perfectly fine in VC++ doesn't work at all in C++ Builder.
Most of the problems are converting different var-types.
For example :
std::string Test = " ";
GetFileAttributes(Test.c_str());
works in VC++ but in C++ Builder it won't compile, telling me "E2034 Cannot convert 'const char *' to 'wchar_t *'.
Am I missing something? What is the reason that doesn't work the same on all compilers the same?
Thanks
Welcome to Windows Unicode/ASCII hell.
The function
GetFileAttributes
is actually a macro defined to either GetFileAttributesA or GetFileAttributesW depending on if you have _UNICODE (or was it UNICODE, or both?) defined when you include the Windows headers. The *A variants take char* and related arguments, the *W functions take wchar_t* and related arguments.
I suggest calling only the wide *W variants directly in new code. This would mean switching to std::wstring for Windows only code and some well-thought out design choices for a cross-platform application.
Your C++ Builder config is set to use UNICODE character set, which means that Win32 APIs are resolved to their wide character versions. Therefore you need to use wide char strings in your C++ code. If you would set your VS config to use UNICODE, you would get the same error.
You can try this:
// wstring = basic_string<wchar_t>
// _T macro ensures that the specified literal is a wide char literal
std::wstring Test = _T(" ");
GetFileAttributes(Test.c_str()); // c_str now returns const wchar_t*, not const char*
See more details about _T/_TEXT macros here: http://docwiki.embarcadero.com/RADStudio/XE3/en/TCHAR_Mapping
You have defined _UNICODE and/or UNICODE in Builder and not defined it in VC.
Most Windows APIs come in 2 flavours the ANSI flavour and the UNICODE flavour.
For, when you call SetWindowText, there really is no SetWindowText functions. Instead there are 2 different functions
- SetWindowTextA which takes an ANSI string
and
- SetWindowTextW which takes a UNICODE string.
If your program is compiled with /DUNICODE /D_UNICODE, SetWindowText maps to SetWindowTextWwhich expects aconst wchar_t *`.
If your program is compiled without these macros defined, it maps to SetWindowTextA which takes a const char *.
The windows headers typically do something like this to make this happen.
#ifdef UNICODE
#define SetWindowText SetWindowTextW
#else
#define SetWindowText SetWindowTextA
#endif
Likewise, there are 2 GetFileAttributes.
DWORD WINAPI GetFileAttributesA(LPCSTR lpFileName);
DWORD WINAPI GetFileAttributesW(LPCWSTR lpFileName);
In VC, you haven't defined UNICODE/_UNICODE & hence you are able to pass string::c_str() which returns a char *.
In Builder, you probably have defined UNICODE/_UNICODE & it expects a wchar_t *.
You may not have done this UNICODE/_UNICODE thing explicitly - may be the IDE is doing it for you - so check the options in the IDE.
You have many ways of fixing this
find the UNICODE/_UNICODE option in the IDE and disable it.
or
use std::w_string - then c_str() will return a wchar_t *
or
Call GetFileAttributesA directly instead of GetFileAttributes - you will need to do this for every other Windows API which comes with these 2 variants.
I have a function which takes two CHAR* as input viz. int _stdcall FileTrans(char* InFile, char* OutFile) in a DLL project.
In the function I'm just calling CopyFile(InFile, OutFile, false); after some process (not related to the files). But it says that it needs both inputs as LPCWSTR. I Googled it but couldn't find anything very interesting.
Like all Windows API functions that accept a string parameter, there are actually two variants of the CopyFile function:
CopyFileA is the ANSI version, which takes narrow (non-Unicode) strings in the system's default character set. Basically, it accepts parameters of type const char*, but the Windows headers use the typedef LPCSTR for this.
CopyFileW is the wide version, which takes Unicode strings. In order to do this, it accepts parameters of type w_char*, but the Windows headers use the typedef LPCWSTR for this (note the additional W in the typedef).
Then, depending on whether the UNICODE preprocessor macro is defined for your project (either in your code before you include the Windows headers, or in your project's properties in Visual Studio), the Windows headers define the unadorned CopyFile as either CopyFileA or CopyFileW. Naturally, if UNICODE is defined, CopyFile will be defined as the Unicode version CopyFileW. Otherwise, it will be defined as CopyFileA. The idea is that the call to the general CopyFile function is automatically resolved at compile time to the correct variant.
Of course, now that you understand all of that, you can mostly forget about it. In modern Windows programming, there is absolutely no reason to call the old ANSI versions of functions or to deal with narrow strings at all. Forget that char* can even be used as a string type—those strings are dead to you. The only strings you're going to be using from now on are Unicode strings, composed of wchar_t characters. Thus the UNICODE symbol should always be defined for your code, and you should only use the W version of Windows API functions.
Looking again at the prototype for the CopyFileW function (the same one you get when you call CopyFile with UNICODE defined), we see:
BOOL WINAPI CopyFile(LPCWSTR lpExistingFileName,
LPCWSTR lpNewFileName,
BOOL bFailIfExists);
Recall that you learned above that LPCWSTR is just a typedef synonym for const wchar_t*, a C-style string that is composed of wide characters. You already know why the parameters are marked const: because the function doesn't modify those values.
And because you also learned above that these are the only types of strings that you should be using anymore, the next step is to modify your FileTrans function to accept wide strings (and make them const if it's not going to modify them):
int _stdcall FileTrans(const wchar_t* InFile, const wchar_t* OutFile);
Now, from inside of FileTrans, you can call CopyFile without any problems because you have the right type of strings.
But a bit of free, extra advice: never use raw C-style strings in C++. Always use the C++ string class, defined in the std namespace by the <string> header.
There are two common variants of this class, std::string and std::wstring. As before, the w refers to wide strings, which are the only type you want to use in Windows. So std::wstring is your new replacement for CHAR* throughout your code base.
Change your declaration of the FileTrans function to look like this:
#include <string>
// ...some other stuff...
int __stdcall FileTrans(const std::wstring& InFile, const std::wstring& OutFile);
Note that I've changed your original CHAR* parameters to constant references to std::wstring objects. Constant references work well here, since you're not going to be changing either of those values inside of the function.
If you're unclear on what constant means, how to use references, or how class types generally work in C++, please consult your favorite C++ book)—this is required knowledge for all C++ programmers. Remember that C++ is not the same language as C and therefore the same idioms do not apply. In many cases, there is a better way to do things, and this is certainly an example of such a case.