Implementation of a Matching strategy for URLs in the Declarative WebRequest API.

chrome/browser/extensions/api/declarative/url_matcher.h

+// Copyright (c) 2012 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef CHROME_BROWSER_EXTENSIONS_API_DECLARATIVE_URL_MATCHER_H_
+#define CHROME_BROWSER_EXTENSIONS_API_DECLARATIVE_URL_MATCHER_H_
+#pragma once
+
+#include <set>
+#include <vector>
+
+#include "base/gtest_prod_util.h"
+#include "base/memory/linked_ptr.h"
+#include "chrome/browser/extensions/api/declarative/substring_set_matcher.h"
+#include "chrome/browser/extensions/api/declarative/url_component_patterns.h"
+#include "googleurl/src/gurl.h"
+#include "url_component_patterns.h"
+
+namespace base {
+class DictionaryValue;
+}
+
+namespace extensions {
+
+// This class represents a single URL matching condition, e.g. a match on the
+// host suffix or the containment of a string in the query component of a GURL.
+//
+// This is basically the next layer above UrlComponentPatterns. The difference
+// between a SubstringPattern created by UrlComponentPatterns is that also

[Matt Perry, 2012/02/14 01:38:34]

s/between/from
"is that this* also"

[battre, 2012/02/14 19:32:21]

Done.

+// supports matches to check whether {Host, Path, Query} of a URL contains
+// a string.
+class UrlMatcherCondition {

[Matt Perry, 2012/02/14 01:38:34]

URLMatcherCondition

[battre, 2012/02/14 19:32:21]

Done.

+ public:
+  enum Criterion {
+    HOST_PREFIX,
+    HOST_SUFFIX,
+    HOST_CONTAINS,
+    HOST_EQUALS,
+    PATH_PREFIX,
+    PATH_SUFFIX,
+    PATH_CONTAINS,
+    PATH_EQUALS,
+    QUERY_PREFIX,
+    QUERY_SUFFIX,
+    QUERY_CONTAINS,
+    QUERY_EQUALS,
+    URL_PREFIX,
+    URL_SUFFIX,
+    URL_CONTAINS,
+    URL_EQUALS,
+  };
+
+  UrlMatcherCondition();
+  UrlMatcherCondition(const UrlMatcherCondition& rhs);
+  UrlMatcherCondition& operator=(const UrlMatcherCondition& rhs);
+  UrlMatcherCondition(Criterion criterion, std::string value);

[Matt Perry, 2012/02/14 01:38:34]

pass const-ref

[battre, 2012/02/14 19:32:21]

Done.

+
+  Criterion criterion() const { return criterion_; }
+  const std::string& value() const { return value_; }
+  const SubstringPattern& substring_pattern() const {
+    return substring_pattern_;
+  }
+
+  // Returns whether this UrlMatcherCondition needs to be executed on a
+  // full URL rather than the individual components (see
+  // UrlMatcherConditionSets).
+  bool IsFullUrlCondition() const;
+
+  // Returns whether this condition is fulfilled according to
+  // |matching_substring_patterns| and |url|.
+  bool IsMatch(
+      const std::set<SubstringPattern::ID>& matching_substring_patterns,
+      const GURL& url) const;
+
+  // Initializes the SubstringPattern of this class, this is taken care of
+  // by the UrlMatcher.
+  void BuildSubstringPattern(UrlComponentPatterns* url_component_patterns);
+
+ private:
+  // |criterion_| and |value_| describe together what property a URL needs
+  // to fulfill to be considered a match.
+  Criterion criterion_;
+  std::string value_;
+
+  // This is the UrlMatcherCondition that is used in a SubstringSetMatcher.
+  // It becomes valid after BuildSubstringPattern has been called.
+  SubstringPattern substring_pattern_;
+};
+
+// This class represents a set of conditions that all need to match on a
+// given URL in order to be considered a match.
+class UrlMatcherConditionSet {
+ public:
+  typedef int ID;
+  typedef std::vector<UrlMatcherCondition> Conditions;
+
+  UrlMatcherConditionSet();
+  UrlMatcherConditionSet(const UrlMatcherConditionSet& rhs);
+  UrlMatcherConditionSet& operator=(const UrlMatcherConditionSet& rhs);
+  UrlMatcherConditionSet(ID id,
+                         const std::vector<UrlMatcherCondition>& conditions);
+
+  ID id() const { return id_; }
+  const Conditions& conditions() const { return conditions_; }
+
+  bool IsMatch(
+      const std::set<SubstringPattern::ID>& matching_substring_patterns,
+      const GURL& url) const;
+
+ private:
+  friend class UrlMatcher;
+
+  Conditions& mutable_conditions() { return conditions_; }
+
+  ID id_;
+  Conditions conditions_;
+};
+
+// This class allows matching one URL against a large set of
+// UrlMatcherConditionSets at the same time.
+class UrlMatcher {
+ public:
+  UrlMatcher();
+
+  void AddConditionSets(const std::vector<UrlMatcherConditionSet>& conditions);
+  void RemoveConditionSets(
+      const std::vector<UrlMatcherConditionSet::ID>& condition_ids);
+
+  std::set<UrlMatcherConditionSet::ID> MatchUrl(const GURL& url);
+
+ private:
+  FRIEND_TEST_ALL_PREFIXES(UrlMatcherTest, FullTest);
+
+  void AddConditionSet(const UrlMatcherConditionSet& new_condition_set);
+  void RemoveConditionSet(UrlMatcherConditionSet::ID id);
+  void UpdateSubstringSetMatcher(
+      bool full_url_conditions,
+      std::vector<SubstringPattern>* unregistered_patterns);
+  void UpdateFrequenciesAndTriggers();
+  void RecreateTriggers();
+
+  // Maps a condition ID (as passed to AddConditions()) to the respective
+  // UrlMatcherConditionSet.
+  typedef std::map<UrlMatcherConditionSet::ID, UrlMatcherConditionSet>
+      UrlMatcherConditionSets;
+  UrlMatcherConditionSets url_matcher_condition_sets_;
+
+  // Maps a SubstringPattern ID to the UrlMatcherConditions that need to
+  // be triggered in case of a SubstringPatter match.
+  std::map<SubstringPattern::ID, std::set<UrlMatcherConditionSet::ID> >
+      substring_match_triggers_;
+
+  std::map<SubstringPattern::ID, size_t> substring_pattern_frequencies_;
+
+  SubstringSetMatcher full_url_matcher_;
+  SubstringSetMatcher url_component_matcher_;
+  std::set<SubstringPattern> registered_full_url_patterns_;
+  std::set<SubstringPattern> registered_url_component_patterns_;
+
+  UrlComponentPatterns url_component_patterns_;
+
+  DISALLOW_COPY_AND_ASSIGN(UrlMatcher);
+};
+
+
+}  // namespace extensions
+
+#endif  // CHROME_BROWSER_EXTENSIONS_API_DECLARATIVE_URL_MATCHER_H_