icu_string_conversions.h revision 513209b27ff55e2841eac0e4120199c23acce758 
1// Copyright (c) 2009 The Chromium Authors. All rights reserved.
2// Use of this source code is governed by a BSD-style license that can be
3// found in the LICENSE file.
4
5#ifndef BASE_I18N_ICU_STRING_CONVERSIONS_H_
6#define BASE_I18N_ICU_STRING_CONVERSIONS_H_
7#pragma once
8
9#include <string>
10
11#include "base/string16.h"
12
13namespace base {
14
15// Defines the error handling modes of UTF16ToCodepage, CodepageToUTF16,
16// WideToCodepage and CodepageToWide.
17class OnStringConversionError {
18 public:
19  enum Type {
20    // The function will return failure. The output buffer will be empty.
21    FAIL,
22
23    // The offending characters are skipped and the conversion will proceed as
24    // if they did not exist.
25    SKIP,
26
27    // When converting to Unicode, the offending byte sequences are substituted
28    // by Unicode replacement character (U+FFFD). When converting from Unicode,
29    // this is the same as SKIP.
30    SUBSTITUTE,
31  };
32
33 private:
34  OnStringConversionError();
35};
36
37// Names of codepages (charsets) understood by icu.
38extern const char kCodepageLatin1[];  // a.k.a. ISO 8859-1
39extern const char kCodepageUTF8[];
40extern const char kCodepageUTF16BE[];
41extern const char kCodepageUTF16LE[];
42
43// Converts between UTF-16 strings and the encoding specified.  If the
44// encoding doesn't exist or the encoding fails (when on_error is FAIL),
45// returns false.
46bool UTF16ToCodepage(const string16& utf16,
47                     const char* codepage_name,
48                     OnStringConversionError::Type on_error,
49                     std::string* encoded);
50bool CodepageToUTF16(const std::string& encoded,
51                     const char* codepage_name,
52                     OnStringConversionError::Type on_error,
53                     string16* utf16);
54
55// Converts between wide strings and the encoding specified.  If the
56// encoding doesn't exist or the encoding fails (when on_error is FAIL),
57// returns false.
58bool WideToCodepage(const std::wstring& wide,
59                    const char* codepage_name,
60                    OnStringConversionError::Type on_error,
61                    std::string* encoded);
62bool CodepageToWide(const std::string& encoded,
63                    const char* codepage_name,
64                    OnStringConversionError::Type on_error,
65                    std::wstring* wide);
66
67// Converts from any codepage to UTF-8 and ensures the resulting UTF-8 is
68// normalized.
69bool ConvertToUtf8AndNormalize(const std::string& text,
70                               const std::string& charset,
71                               std::string* result);
72
73}  // namespace base
74
75#endif  // BASE_I18N_ICU_STRING_CONVERSIONS_H_
76