third_party/libaddressinput/chromium/canonicalize_string.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 #include "cpp/src/util/canonicalize_string.h"
   6
   7 #include "base/logging.h"
   8 #include "cpp/include/libaddressinput/util/scoped_ptr.h"
   9 #include "third_party/icu/source/i18n/unicode/coll.h"
  10
  11 namespace i18n {
  12 namespace addressinput {
  13
  14 namespace {
  15
  16 class ChromeStringCanonicalizer : public StringCanonicalizer {
  17  public:
  18   ChromeStringCanonicalizer()
  19       : error_code_(U_ZERO_ERROR),
  20         collator_(icu::Collator::createInstance(error_code_)) {
  21     collator_->setStrength(icu::Collator::PRIMARY);
  22     DCHECK(U_SUCCESS(error_code_));
  23   }
  24
  25   virtual ~ChromeStringCanonicalizer() {}
  26
  27   // StringCanonicalizer implementation.
  28   virtual std::string CanonicalizeString(const std::string& original) {
  29     // Returns a canonical version of the string that can be used for comparing
  30     // strings regardless of diacritics and capitalization.
  31     //    CanonicalizeString("Texas") == CanonicalizeString("T\u00E9xas");
  32     //    CanonicalizeString("Texas") == CanonicalizeString("teXas");
  33     //    CanonicalizeString("Texas") != CanonicalizeString("California");
  34     //
  35     // The output is not human-readable.
  36     //    CanonicalizeString("Texas") != "Texas";
  37     icu::UnicodeString icu_str(
  38         original.c_str(), static_cast<int32_t>(original.length()));
  39     int32_t buffer_size = collator_->getSortKey(icu_str, NULL, 0);
  40     scoped_ptr<uint8_t[]> buffer(new uint8_t[buffer_size]);
  41     DCHECK(buffer.get());
  42     int32_t filled_size =
  43         collator_->getSortKey(icu_str, buffer.get(), buffer_size);
  44     DCHECK_EQ(buffer_size, filled_size);
  45     return std::string(reinterpret_cast<const char*>(buffer.get()));
  46   }
  47
  48  private:
  49   UErrorCode error_code_;
  50   scoped_ptr<icu::Collator> collator_;
  51
  52   DISALLOW_COPY_AND_ASSIGN(ChromeStringCanonicalizer);
  53 };
  54
  55 }  // namespace
  56
  57 // static
  58 scoped_ptr<StringCanonicalizer> StringCanonicalizer::Build() {
  59   return scoped_ptr<StringCanonicalizer>(new ChromeStringCanonicalizer);
  60 }
  61
  62 }  // namespace addressinput
  63 }  // namespace i18n