# std::codecvt<InternT,ExternT,StateT>::unshift, do_unshift

> C++ Atlas: `cppatlas.com/w/cpp/locale/codecvt/unshift` | Source: https://en.cppreference.com/w/cpp/locale/codecvt/unshift | Domain: Localization

A value of type std::codecvt_base::result, indicating the success status as follows:

## Official Specification Summary (cppreference)

```cpp
// Defined in header <locale>

public:
result unshift( StateT& state, ExternT* to, ExternT* to_end,
                ExternT*& to_next ) const; // (1)

protected:
virtual result do_unshift( StateT& state, ExternT* to, ExternT* to_end,
                           ExternT*& to_next ) const; // (2)
```

1) Public member function, calls the member function `do_unshift` of the most derived class.

2) If the encoding represented by this `codecvt` facet is state-dependent, and `state` represents a conversion state that is not the initial shift state, writes the characters necessary to return to the initial shift state. The characters are written to a character array whose first element is pointed to by `to`. No more than `to_end - to` characters are written. The parameter `to_next` is updated to point one past the last character written.

### Return value

A value of type `std::codecvt_base::result`, indicating the success status as follows:

- `ok`: all necessary characters were written. `state` now represents initial shift state
- `partial`: not enough space in the output buffer. `to_next == to_end`
- `error`: an unspecified error has occurred
- `noconv`: the encoding is not state-dependent, no termination sequence necessary

### Notes

This function is called by `std::basic_filebuf::close()` and in other situations when finalizing a state-dependent multibyte character sequence.

### Example

> Note: This section is incomplete Reason: no example

### Defect reports

The following behavior-changing defect reports were applied retroactively to previously published C++ standards.

| DR | Applied to | Behavior as published | Correct behavior |
| --- | --- | --- | --- |
| LWG 305 | C++98 | `std::codecvt<wchar_t, char, std::mbstate_t>::do_unshift` was required not to write any character | not required |
| LWG 380 | C++98 | the meaning of returning `partial` was 'more characters need to be supplied to complete termination', but no character is supplied | corrected to indicating insufficient buffer space |
| LWG 381 | C++98 | `state` was not required to be valid, and `error` is returned if `state` is invalid | `state` is required to be valid, and returning `error` indicates an error |
| LWG 664 | C++98 | `std::codecvt<char, char, std::mbstate_t>::do_unshift` was required not to write any character | not required |
| LWG 665 | C++98 | `std::codecvt<char, char, std::mbstate_t>::do_unshift` was required to return `noconv` | not required |

### See also

- [`wcrtomb`](https://en.cppreference.com/w/cpp/string/multibyte/wcrtomb): converts a wide character to its multibyte representation, given state (function)
- [`do_out`](https://en.cppreference.com/w/cpp/locale/codecvt/out) [virtual]: converts a string from `InternT` to `ExternT`, such as when writing to file (virtual protected member function)

---
Attribution: content from https://en.cppreference.com/w/cpp/locale/codecvt/unshift (cppreference.com), licensed under CC-BY-SA 3.0 and GFDL. Modified: converted to markdown, study notes added. Offline snapshot of 9 February 2025.