Factor out UrlPatternIndex from IndexedRuleset.

components/subresource_filter/core/common/indexed_ruleset.h

Index: components/subresource_filter/core/common/indexed_ruleset.h
diff --git a/components/subresource_filter/core/common/indexed_ruleset.h b/components/subresource_filter/core/common/indexed_ruleset.h
index e5206d783fae93e12a56d9ad643647cbc1e228a5..00588f8a0a76514dd8699da5e167fb5ff4c1a881 100644
--- a/components/subresource_filter/core/common/indexed_ruleset.h
+++ b/components/subresource_filter/core/common/indexed_ruleset.h
@@ -5,33 +5,28 @@
 #ifndef COMPONENTS_SUBRESOURCE_FILTER_CORE_COMMON_INDEXED_RULESET_H_
 #define COMPONENTS_SUBRESOURCE_FILTER_CORE_COMMON_INDEXED_RULESET_H_
 
+#include <stddef.h>
 #include <stdint.h>
 
-#include <vector>
-
 #include "base/macros.h"
 #include "base/numerics/safe_conversions.h"
-#include "base/strings/string_piece.h"
-#include "components/subresource_filter/core/common/closed_hash_map.h"
-#include "components/subresource_filter/core/common/flat/rules_generated.h"
-#include "components/subresource_filter/core/common/proto/rules.pb.h"
-#include "components/subresource_filter/core/common/uint64_hasher.h"
-#include "url/gurl.h"
-#include "url/origin.h"
+#include "components/subresource_filter/core/common/flat/indexed_ruleset_generated.h"
+#include "components/subresource_filter/core/common/url_pattern_index.h"
+#include "third_party/flatbuffers/src/include/flatbuffers/flatbuffers.h"
+
+class GURL;
+
+namespace url {
+class Origin;
+}
 
 namespace subresource_filter {
 
 class FirstPartyOrigin;
 
-// The integer type used to represent N-grams.
-using NGram = uint64_t;
-// The hasher used for hashing N-grams.
-using NGramHasher = Uint64Hasher;
-// The hash table probe sequence used both by the ruleset builder and matcher.
-using NGramHashTableProber = DefaultProber<NGram, NGramHasher>;
-
-constexpr size_t kNGramSize = 5;
-static_assert(kNGramSize <= sizeof(NGram), "NGram type is too narrow.");
+namespace proto {
+class UrlRule;
+}
 
 // The class used to construct flat data structures representing the set of URL
 // filtering rules, as well as the index of those. Internally owns a
@@ -66,45 +61,12 @@ class RulesetIndexer {
   size_t size() const { return base::strict_cast<size_t>(builder_.GetSize()); }
 
  private:
-  using MutableUrlRuleList = std::vector<flatbuffers::Offset<flat::UrlRule>>;
-  using MutableNGramIndex =
-      ClosedHashMap<NGram, MutableUrlRuleList, NGramHashTableProber>;
-
-  // Encapsulates a subset of the rules, and an index built on the URL patterns
-  // in these rules. The ruleset is divided into parts according to metadata of
-  // the rules. Currently there are two parts: blacklist and whitelist.
-  struct MutableUrlPatternIndex {
-    // This index contains all non-REGEXP rules that have at least one
-    // acceptable N-gram. For a single rule the N-gram used as an index key is
-    // picked greedily (see GetMostDistinctiveNGram).
-    MutableNGramIndex ngram_index;
-
-    // A fallback list that contains all the rules with no acceptable N-gram,
-    // and all the REGEXP rules.
-    MutableUrlRuleList fallback_rules;
-
-    MutableUrlPatternIndex();
-    ~MutableUrlPatternIndex();
-  };
-
-  // Returns an N-gram of the |pattern| encoded into the NGram integer type. The
-  // N-gram is picked using a greedy heuristic, i.e. the one is chosen which
-  // corresponds to the shortest list of rules within the |index|. If there are
-  // no valid N-grams in the |pattern|, the return value is 0.
-  static NGram GetMostDistinctiveNGram(const MutableNGramIndex& index,
-                                       base::StringPiece pattern);
-
-  // Serialized an |index| built over a part of the ruleset, and returns its
-  // offset in the FlatBuffer.
-  flatbuffers::Offset<flat::UrlPatternIndex> SerializeUrlPatternIndex(
-      const MutableUrlPatternIndex& index);
-
-  MutableUrlPatternIndex blacklist_;
-  MutableUrlPatternIndex whitelist_;
-  MutableUrlPatternIndex activation_;
-
   flatbuffers::FlatBufferBuilder builder_;
 
+  UrlPatternIndexBuilder blacklist_;
+  UrlPatternIndexBuilder whitelist_;
+  UrlPatternIndexBuilder deactivation_;
+
   DISALLOW_COPY_AND_ASSIGN(RulesetIndexer);
 };
 
@@ -142,6 +104,10 @@ class IndexedRulesetMatcher {
  private:
   const flat::IndexedRuleset* root_;
 
+  UrlPatternIndexMatcher blacklist_;
+  UrlPatternIndexMatcher whitelist_;
+  UrlPatternIndexMatcher deactivation_;
+
   DISALLOW_COPY_AND_ASSIGN(IndexedRulesetMatcher);
 };