1 // Copyright 2014 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 // A streaming validator for UTF-8. Validation is based on the definition in
6 // RFC-3629. In particular, it does not reject the invalid characters rejected
7 // by base::IsStringUTF8().
8 //
9 // The implementation detects errors on the first possible byte.
10 
11 #ifndef BASE_I18N_STREAMING_UTF8_VALIDATOR_H_
12 #define BASE_I18N_STREAMING_UTF8_VALIDATOR_H_
13 
14 #include <stddef.h>
15 #include <stdint.h>
16 
17 #include <string>
18 
19 #include "base/i18n/base_i18n_export.h"
20 #include "base/macros.h"
21 
22 namespace base {
23 
24 class BASE_I18N_EXPORT StreamingUtf8Validator {
25  public:
26   // The validator exposes 3 states. It starts in state VALID_ENDPOINT. As it
27   // processes characters it alternates between VALID_ENDPOINT and
28   // VALID_MIDPOINT. If it encounters an invalid byte or UTF-8 sequence the
29   // state changes permanently to INVALID.
30   enum State {
31     VALID_ENDPOINT,
32     VALID_MIDPOINT,
33     INVALID
34   };
35 
StreamingUtf8Validator()36   StreamingUtf8Validator() : state_(0u) {}
37   // Trivial destructor intentionally omitted.
38 
39   // Validate |size| bytes starting at |data|. If the concatenation of all calls
40   // to AddBytes() since this object was constructed or reset is a valid UTF-8
41   // string, returns VALID_ENDPOINT. If it could be the prefix of a valid UTF-8
42   // string, returns VALID_MIDPOINT. If an invalid byte or UTF-8 sequence was
43   // present, returns INVALID.
44   State AddBytes(const char* data, size_t size);
45 
46   // Return the object to a freshly-constructed state so that it can be re-used.
47   void Reset();
48 
49   // Validate a complete string using the same criteria. Returns true if the
50   // string only contains complete, valid UTF-8 codepoints.
51   static bool Validate(const std::string& string);
52 
53  private:
54   // The current state of the validator. Value 0 is the initial/valid state.
55   // The state is stored as an offset into |kUtf8ValidatorTables|. The special
56   // state |kUtf8InvalidState| is invalid.
57   uint8_t state_;
58 
59   // This type could be made copyable but there is currently no use-case for
60   // it.
61   DISALLOW_COPY_AND_ASSIGN(StreamingUtf8Validator);
62 };
63 
64 }  // namespace base
65 
66 #endif  // BASE_I18N_STREAMING_UTF8_VALIDATOR_H_
67