base/i18n/string_search.cc

   1 // Copyright (c) 2011 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 <stdint.h>
   6
   7 #include "base/i18n/string_search.h"
   8
   9 #include "third_party/icu/source/i18n/unicode/usearch.h"
  10
  11 namespace base {
  12 namespace i18n {
  13
  14 FixedPatternStringSearch::FixedPatternStringSearch(const string16& find_this,
  15                                                    bool case_sensitive)
  16     : find_this_(find_this) {
  17   // usearch_open requires a valid string argument to be searched, even if we
  18   // want to set it by usearch_setText afterwards. So, supplying a dummy text.
  19   const string16& dummy = find_this_;
  20
  21   UErrorCode status = U_ZERO_ERROR;
  22   search_ = usearch_open(find_this_.data(), find_this_.size(), dummy.data(),
  23                          dummy.size(), uloc_getDefault(),
  24                          nullptr,  // breakiter
  25                          &status);
  26   if (U_SUCCESS(status)) {
  27     // http://icu-project.org/apiref/icu4c40/ucol_8h.html#6a967f36248b0a1bc7654f538ee8ba96
  28     // Set comparison level to UCOL_PRIMARY to ignore secondary and tertiary
  29     // differences. Set comparison level to UCOL_TERTIARY to include all
  30     // comparison differences.
  31     // Diacritical differences on the same base letter represent a
  32     // secondary difference.
  33     // Uppercase and lowercase versions of the same character represents a
  34     // tertiary difference.
  35     UCollator* collator = usearch_getCollator(search_);
  36     ucol_setStrength(collator, case_sensitive ? UCOL_TERTIARY : UCOL_PRIMARY);
  37     usearch_reset(search_);
  38   }
  39 }
  40
  41 FixedPatternStringSearch::~FixedPatternStringSearch() {
  42   if (search_)
  43     usearch_close(search_);
  44 }
  45
  46 bool FixedPatternStringSearch::Search(const string16& in_this,
  47                                       size_t* match_index,
  48                                       size_t* match_length,
  49                                       bool forward_search) {
  50   UErrorCode status = U_ZERO_ERROR;
  51   usearch_setText(search_, in_this.data(), in_this.size(), &status);
  52
  53   // Default to basic substring search if usearch fails. According to
  54   // http://icu-project.org/apiref/icu4c/usearch_8h.html, usearch_open will fail
  55   // if either |find_this| or |in_this| are empty. In either case basic
  56   // substring search will give the correct return value.
  57   if (!U_SUCCESS(status)) {
  58     size_t index = in_this.find(find_this_);
  59     if (index == string16::npos)
  60       return false;
  61     if (match_index)
  62       *match_index = index;
  63     if (match_length)
  64       *match_length = find_this_.size();
  65     return true;
  66   }
  67
  68   int32_t index = forward_search ? usearch_first(search_, &status)
  69                                  : usearch_last(search_, &status);
  70   if (!U_SUCCESS(status) || index == USEARCH_DONE)
  71     return false;
  72   if (match_index)
  73     *match_index = static_cast<size_t>(index);
  74   if (match_length)
  75     *match_length = static_cast<size_t>(usearch_getMatchedLength(search_));
  76   return true;
  77 }
  78
  79 FixedPatternStringSearchIgnoringCaseAndAccents::
  80     FixedPatternStringSearchIgnoringCaseAndAccents(const string16& find_this)
  81     : base_search_(find_this, /*case_sensitive=*/false) {}
  82
  83 bool FixedPatternStringSearchIgnoringCaseAndAccents::Search(
  84     const string16& in_this,
  85     size_t* match_index,
  86     size_t* match_length) {
  87   return base_search_.Search(in_this, match_index, match_length,
  88                              /*forward_search=*/true);
  89 }
  90
  91 bool StringSearchIgnoringCaseAndAccents(const string16& find_this,
  92                                         const string16& in_this,
  93                                         size_t* match_index,
  94                                         size_t* match_length) {
  95   return FixedPatternStringSearchIgnoringCaseAndAccents(find_this).Search(
  96       in_this, match_index, match_length);
  97 }
  98
  99 bool StringSearch(const string16& find_this,
 100                   const string16& in_this,
 101                   size_t* match_index,
 102                   size_t* match_length,
 103                   bool case_sensitive,
 104                   bool forward_search) {
 105   return FixedPatternStringSearch(find_this, case_sensitive)
 106       .Search(in_this, match_index, match_length, forward_search);
 107 }
 108
 109 }  // namespace i18n
 110 }  // namespace base