base/i18n/streaming_utf8_validator.cc

   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 // This implementation doesn't use ICU. The ICU macros are oriented towards
   6 // character-at-a-time processing, whereas byte-at-a-time processing is easier
   7 // with streaming input.
   8
   9 #include "base/i18n/streaming_utf8_validator.h"
  10
  11 #include "base/i18n/utf8_validator_tables.h"
  12 #include "base/logging.h"
  13
  14 namespace base {
  15 namespace {
  16
  17 uint8 StateTableLookup(uint8 offset) {
  18   // Skip the bounds check on non-debug builds so that it isn't necessary to set
  19   // LOGGING_IS_OFFICIAL_BUILD just to do a performance test.
  20   if (logging::DEBUG_MODE)
  21     DCHECK_LT(offset, internal::kUtf8ValidatorTablesSize);
  22   return internal::kUtf8ValidatorTables[offset];
  23 }
  24
  25 }  // namespace
  26
  27 StreamingUtf8Validator::State StreamingUtf8Validator::AddBytes(const char* data,
  28                                                                size_t size) {
  29   // Copy |state_| into a local variable so that the compiler doesn't have to be
  30   // careful of aliasing.
  31   uint8 state = state_;
  32   for (const char* p = data; p != data + size; ++p) {
  33     if ((*p & 0x80) == 0) {
  34       if (state == 0)
  35         continue;
  36       state = internal::I18N_UTF8_VALIDATOR_INVALID_INDEX;
  37       break;
  38     }
  39     const uint8 shift_amount = StateTableLookup(state);
  40     const uint8 shifted_char = (*p & 0x7F) >> shift_amount;
  41     state = StateTableLookup(state + shifted_char + 1);
  42     // State may be INVALID here, but this code is optimised for the case of
  43     // valid UTF-8 and it is more efficient (by about 2%) to not attempt an
  44     // early loop exit unless we hit an ASCII character.
  45   }
  46   state_ = state;
  47   return state == 0 ? VALID_ENDPOINT
  48       : state == internal::I18N_UTF8_VALIDATOR_INVALID_INDEX
  49       ? INVALID
  50       : VALID_MIDPOINT;
  51 }
  52
  53 void StreamingUtf8Validator::Reset() {
  54   state_ = 0u;
  55 }
  56
  57 bool StreamingUtf8Validator::Validate(const std::string& string) {
  58   return StreamingUtf8Validator().AddBytes(string.data(), string.size()) ==
  59          VALID_ENDPOINT;
  60 }
  61
  62 }  // namespace base