components/url_matcher/regex_set_matcher.cc

   1 // Copyright 2013 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 "components/url_matcher/regex_set_matcher.h"
   6
   7 #include "base/logging.h"
   8 #include "base/stl_util.h"
   9 #include "base/strings/string_util.h"
  10 #include "components/url_matcher/substring_set_matcher.h"
  11 #include "third_party/re2/re2/filtered_re2.h"
  12 #include "third_party/re2/re2/re2.h"
  13
  14 namespace url_matcher {
  15
  16 RegexSetMatcher::RegexSetMatcher() {}
  17
  18 RegexSetMatcher::~RegexSetMatcher() {
  19   DeleteSubstringPatterns();
  20 }
  21
  22 void RegexSetMatcher::AddPatterns(
  23     const std::vector<const StringPattern*>& regex_list) {
  24   if (regex_list.empty())
  25     return;
  26   for (size_t i = 0; i < regex_list.size(); ++i) {
  27     regexes_[regex_list[i]->id()] = regex_list[i];
  28   }
  29
  30   RebuildMatcher();
  31 }
  32
  33 void RegexSetMatcher::ClearPatterns() {
  34   regexes_.clear();
  35   RebuildMatcher();
  36 }
  37
  38 bool RegexSetMatcher::Match(const std::string& text,
  39                             std::set<StringPattern::ID>* matches) const {
  40   size_t old_number_of_matches = matches->size();
  41   if (regexes_.empty())
  42     return false;
  43   if (!filtered_re2_.get()) {
  44     LOG(ERROR) << "RegexSetMatcher was not initialized";
  45     return false;
  46   }
  47
  48   // FilteredRE2 expects lowercase for prefiltering, but we still
  49   // match case-sensitively.
  50   std::vector<RE2ID> atoms(FindSubstringMatches(base::ToLowerASCII(text)));
  51
  52   std::vector<RE2ID> re2_ids;
  53   filtered_re2_->AllMatches(text, atoms, &re2_ids);
  54
  55   for (size_t i = 0; i < re2_ids.size(); ++i) {
  56     StringPattern::ID id = re2_id_map_[re2_ids[i]];
  57     matches->insert(id);
  58   }
  59   return old_number_of_matches != matches->size();
  60 }
  61
  62 bool RegexSetMatcher::IsEmpty() const {
  63   return regexes_.empty();
  64 }
  65
  66 std::vector<RegexSetMatcher::RE2ID> RegexSetMatcher::FindSubstringMatches(
  67     const std::string& text) const {
  68   std::set<int> atoms_set;
  69   substring_matcher_->Match(text, &atoms_set);
  70   return std::vector<RE2ID>(atoms_set.begin(), atoms_set.end());
  71 }
  72
  73 void RegexSetMatcher::RebuildMatcher() {
  74   re2_id_map_.clear();
  75   filtered_re2_.reset(new re2::FilteredRE2());
  76   if (regexes_.empty())
  77     return;
  78
  79   for (RegexMap::iterator it = regexes_.begin(); it != regexes_.end(); ++it) {
  80     RE2ID re2_id;
  81     RE2::ErrorCode error = filtered_re2_->Add(
  82         it->second->pattern(), RE2::DefaultOptions, &re2_id);
  83     if (error == RE2::NoError) {
  84       DCHECK_EQ(static_cast<RE2ID>(re2_id_map_.size()), re2_id);
  85       re2_id_map_.push_back(it->first);
  86     } else {
  87       // Unparseable regexes should have been rejected already in
  88       // URLMatcherFactory::CreateURLMatchesCondition.
  89       LOG(ERROR) << "Could not parse regex (id=" << it->first << ", "
  90                  << it->second->pattern() << ")";
  91     }
  92   }
  93
  94   std::vector<std::string> strings_to_match;
  95   filtered_re2_->Compile(&strings_to_match);
  96
  97   substring_matcher_.reset(new SubstringSetMatcher);
  98   DeleteSubstringPatterns();
  99   // Build SubstringSetMatcher from |strings_to_match|.
 100   // SubstringSetMatcher doesn't own its strings.
 101   for (size_t i = 0; i < strings_to_match.size(); ++i) {
 102     substring_patterns_.push_back(
 103         new StringPattern(strings_to_match[i], i));
 104   }
 105   substring_matcher_->RegisterPatterns(substring_patterns_);
 106 }
 107
 108 void RegexSetMatcher::DeleteSubstringPatterns() {
 109   STLDeleteElements(&substring_patterns_);
 110 }
 111
 112 }  // namespace url_matcher