// Copyright (c) 2011 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 NET_BASE_SDCH_MANAGER_H_
#define NET_BASE_SDCH_MANAGER_H_

#include <map>
#include <set>
#include <string>

#include "base/gtest_prod_util.h"
#include "base/memory/ref_counted.h"
#include "base/memory/scoped_ptr.h"
#include "base/observer_list.h"
#include "base/threading/thread_checker.h"
#include "base/time/time.h"
#include "net/base/net_export.h"
#include "url/gurl.h"

namespace net {

class SdchObserver;

// Provides global database of differential decompression dictionaries for the
// SDCH filter (processes sdch enconded content).
//
// The SdchManager maintains a collection of memory resident dictionaries. It
// can find a dictionary (based on a server specification of a hash), store a
// dictionary, and make judgements about what URLs can use, set, etc. a
// dictionary.

// These dictionaries are acquired over the net, and include a header
// (containing metadata) as well as a VCDIFF dictionary (for use by a VCDIFF
// module) to decompress data.
class NET_EXPORT SdchManager {
 public:
  // A list of errors that appeared and were either resolved, or used to turn
  // off sdch encoding.
  enum ProblemCodes {
    MIN_PROBLEM_CODE,

    // Content-encoding correction problems.
    ADDED_CONTENT_ENCODING = 1,
    FIXED_CONTENT_ENCODING = 2,
    FIXED_CONTENT_ENCODINGS = 3,

    // Content decoding errors.
    DECODE_HEADER_ERROR = 4,
    DECODE_BODY_ERROR = 5,

    // More content-encoding correction problems.
    OPTIONAL_GUNZIP_ENCODING_ADDED = 6,

    // Content encoding correction when we're not even tagged as HTML!?!
    BINARY_ADDED_CONTENT_ENCODING = 7,
    BINARY_FIXED_CONTENT_ENCODING = 8,
    BINARY_FIXED_CONTENT_ENCODINGS = 9,

    // Dictionary selection for use problems.
    DICTIONARY_FOUND_HAS_WRONG_DOMAIN = 10,
    DICTIONARY_FOUND_HAS_WRONG_PORT_LIST = 11,
    DICTIONARY_FOUND_HAS_WRONG_PATH = 12,
    DICTIONARY_FOUND_HAS_WRONG_SCHEME = 13,
    DICTIONARY_HASH_NOT_FOUND = 14,
    DICTIONARY_HASH_MALFORMED = 15,

    // Dictionary saving problems.
    DICTIONARY_HAS_NO_HEADER = 20,
    DICTIONARY_HEADER_LINE_MISSING_COLON = 21,
    DICTIONARY_MISSING_DOMAIN_SPECIFIER = 22,
    DICTIONARY_SPECIFIES_TOP_LEVEL_DOMAIN = 23,
    DICTIONARY_DOMAIN_NOT_MATCHING_SOURCE_URL = 24,
    DICTIONARY_PORT_NOT_MATCHING_SOURCE_URL = 25,
    DICTIONARY_HAS_NO_TEXT = 26,
    DICTIONARY_REFERER_URL_HAS_DOT_IN_PREFIX = 27,

    // Dictionary loading problems.
    DICTIONARY_LOAD_ATTEMPT_FROM_DIFFERENT_HOST = 30,
    DICTIONARY_SELECTED_FOR_SSL = 31,
    DICTIONARY_ALREADY_LOADED = 32,
    DICTIONARY_SELECTED_FROM_NON_HTTP = 33,
    DICTIONARY_IS_TOO_LARGE= 34,
    DICTIONARY_COUNT_EXCEEDED = 35,
    DICTIONARY_ALREADY_SCHEDULED_TO_DOWNLOAD = 36,
    DICTIONARY_ALREADY_TRIED_TO_DOWNLOAD = 37,
    DICTIONARY_FETCH_READ_FAILED = 38,

    // Failsafe hack.
    ATTEMPT_TO_DECODE_NON_HTTP_DATA = 40,


    // Content-Encoding problems detected, with no action taken.
    MULTIENCODING_FOR_NON_SDCH_REQUEST = 50,
    SDCH_CONTENT_ENCODE_FOR_NON_SDCH_REQUEST = 51,

    // Dictionary manager issues.
    DOMAIN_BLACKLIST_INCLUDES_TARGET = 61,

    // Problematic decode recovery methods.
    META_REFRESH_RECOVERY = 70,            // Dictionary not found.
    // defunct =  71, // Almost the same as META_REFRESH_UNSUPPORTED.
    // defunct = 72,  // Almost the same as CACHED_META_REFRESH_UNSUPPORTED.
    // defunct = 73,  // PASSING_THROUGH_NON_SDCH plus
                      // RESPONSE_TENTATIVE_SDCH in ../filter/sdch_filter.cc.
    META_REFRESH_UNSUPPORTED = 74,         // Unrecoverable error.
    CACHED_META_REFRESH_UNSUPPORTED = 75,  // As above, but pulled from cache.
    PASSING_THROUGH_NON_SDCH = 76,  // Tagged sdch but missing dictionary-hash.
    INCOMPLETE_SDCH_CONTENT = 77,   // Last window was not completely decoded.
    PASS_THROUGH_404_CODE = 78,     // URL not found message passing through.

    // This next report is very common, and not really an error scenario, but
    // it exercises the error recovery logic.
    PASS_THROUGH_OLD_CACHED = 79,   // Back button got pre-SDCH cached content.

    // Common decoded recovery methods.
    META_REFRESH_CACHED_RECOVERY = 80,  // Probably startup tab loading.
    // defunct = 81, // Now tracked by ResponseCorruptionDetectionCause histo.

    // Non SDCH problems, only accounted for to make stat counting complete
    // (i.e., be able to be sure all dictionary advertisements are accounted
    // for).

    UNFLUSHED_CONTENT = 90,    // Possible error in filter chaining.
    // defunct = 91,           // MISSING_TIME_STATS (Should never happen.)
    CACHE_DECODED = 92,        // No timing stats recorded.
    // defunct = 93,           // OVER_10_MINUTES (No timing stats recorded.)
    UNINITIALIZED = 94,        // Filter never even got initialized.
    PRIOR_TO_DICTIONARY = 95,  // We hadn't even parsed a dictionary selector.
    DECODE_ERROR = 96,         // Something went wrong during decode.

    // Problem during the latency test.
    LATENCY_TEST_DISALLOWED = 100,  // SDCH now failing, but it worked before!

    MAX_PROBLEM_CODE  // Used to bound histogram.
  };

  // Use the following static limits to block DOS attacks until we implement
  // a cached dictionary evicition strategy.
  static const size_t kMaxDictionarySize;
  static const size_t kMaxDictionaryCount;

  // There is one instance of |Dictionary| for each memory-cached SDCH
  // dictionary.
  class NET_EXPORT_PRIVATE Dictionary : public base::RefCounted<Dictionary> {
   public:
    // Sdch filters can get our text to use in decoding compressed data.
    const std::string& text() const { return text_; }

   private:
    friend class base::RefCounted<Dictionary>;
    friend class SdchManager;  // Only manager can construct an instance.
    FRIEND_TEST_ALL_PREFIXES(SdchManagerTest, PathMatch);

    // Construct a vc-diff usable dictionary from the dictionary_text starting
    // at the given offset. The supplied client_hash should be used to
    // advertise the dictionary's availability relative to the suppplied URL.
    Dictionary(const std::string& dictionary_text,
               size_t offset,
               const std::string& client_hash,
               const GURL& url,
               const std::string& domain,
               const std::string& path,
               const base::Time& expiration,
               const std::set<int>& ports);
    virtual ~Dictionary();

    const GURL& url() const { return url_; }
    const std::string& client_hash() const { return client_hash_; }

    // Security method to check if we can advertise this dictionary for use
    // if the |target_url| returns SDCH compressed data.
    bool CanAdvertise(const GURL& target_url);

    // Security methods to check if we can establish a new dictionary with the
    // given data, that arrived in response to get of dictionary_url.
    static bool CanSet(const std::string& domain, const std::string& path,
                       const std::set<int>& ports, const GURL& dictionary_url);

    // Security method to check if we can use a dictionary to decompress a
    // target that arrived with a reference to this dictionary.
    bool CanUse(const GURL& referring_url);

    // Compare paths to see if they "match" for dictionary use.
    static bool PathMatch(const std::string& path,
                          const std::string& restriction);

    // Compare domains to see if the "match" for dictionary use.
    static bool DomainMatch(const GURL& url, const std::string& restriction);


    // The actual text of the dictionary.
    std::string text_;

    // Part of the hash of text_ that the client uses to advertise the fact that
    // it has a specific dictionary pre-cached.
    std::string client_hash_;

    // The GURL that arrived with the text_ in a URL request to specify where
    // this dictionary may be used.
    const GURL url_;

    // Metadate "headers" in before dictionary text contained the following:
    // Each dictionary payload consists of several headers, followed by the text
    // of the dictionary. The following are the known headers.
    const std::string domain_;
    const std::string path_;
    const base::Time expiration_;  // Implied by max-age.
    const std::set<int> ports_;

    DISALLOW_COPY_AND_ASSIGN(Dictionary);
  };

  SdchManager();
  ~SdchManager();

  // Clear data (for browser data removal).
  void ClearData();

  // Record stats on various errors.
  static void SdchErrorRecovery(ProblemCodes problem);

  // Enables or disables SDCH compression.
  static void EnableSdchSupport(bool enabled);

  static bool sdch_enabled() { return g_sdch_enabled_; }

  // Enables or disables SDCH compression over secure connection.
  static void EnableSecureSchemeSupport(bool enabled);

  static bool secure_scheme_supported() { return g_secure_scheme_supported_; }

  // Briefly prevent further advertising of SDCH on this domain (if SDCH is
  // enabled). After enough calls to IsInSupportedDomain() the blacklisting
  // will be removed. Additional blacklists take exponentially more calls
  // to IsInSupportedDomain() before the blacklisting is undone.
  // Used when filter errors are found from a given domain, but it is plausible
  // that the cause is temporary (such as application startup, where cached
  // entries are used, but a dictionary is not yet loaded).
  void BlacklistDomain(const GURL& url, ProblemCodes blacklist_reason);

  // Used when SEVERE filter errors are found from a given domain, to prevent
  // further use of SDCH on that domain.
  void BlacklistDomainForever(const GURL& url, ProblemCodes blacklist_reason);

  // Unit test only, this function resets enabling of sdch, and clears the
  // blacklist.
  void ClearBlacklistings();

  // Unit test only, this function resets the blacklisting count for a domain.
  void ClearDomainBlacklisting(const std::string& domain);

  // Unit test only: indicate how many more times a domain will be blacklisted.
  int BlackListDomainCount(const std::string& domain);

  // Unit test only: Indicate what current blacklist increment is for a domain.
  int BlacklistDomainExponential(const std::string& domain);

  // Check to see if SDCH is enabled (globally), and the given URL is in a
  // supported domain (i.e., not blacklisted, and either the specific supported
  // domain, or all domains were assumed supported). If it is blacklist, reduce
  // by 1 the number of times it will be reported as blacklisted.
  bool IsInSupportedDomain(const GURL& url);

  // Send out appropriate events notifying observers that a Get-Dictionary
  // header has been seen.
  void OnGetDictionary(const GURL& request_url, const GURL& dictionary_url);

  // Find the vcdiff dictionary (the body of the sdch dictionary that appears
  // after the meta-data headers like Domain:...) with the given |server_hash|
  // to use to decompreses data that arrived as SDCH encoded content. Check to
  // be sure the returned |dictionary| can be used for decoding content supplied
  // in response to a request for |referring_url|.
  // Return null in |dictionary| if there is no matching legal dictionary.
  void GetVcdiffDictionary(const std::string& server_hash,
                           const GURL& referring_url,
                           scoped_refptr<Dictionary>* dictionary);

  // Get list of available (pre-cached) dictionaries that we have already loaded
  // into memory. The list is a comma separated list of (client) hashes per
  // the SDCH spec.
  void GetAvailDictionaryList(const GURL& target_url, std::string* list);

  // Construct the pair of hashes for client and server to identify an SDCH
  // dictionary. This is only made public to facilitate unit testing, but is
  // otherwise private
  static void GenerateHash(const std::string& dictionary_text,
                           std::string* client_hash, std::string* server_hash);

  // For Latency testing only, we need to know if we've succeeded in doing a
  // round trip before starting our comparative tests. If ever we encounter
  // problems with SDCH, we opt-out of the test unless/until we perform a
  // complete SDCH decoding.
  bool AllowLatencyExperiment(const GURL& url) const;

  void SetAllowLatencyExperiment(const GURL& url, bool enable);

  // Add an SDCH dictionary to our list of availible
  // dictionaries. This addition will fail if addition is illegal
  // (data in the dictionary is not acceptable from the
  // dictionary_url; dictionary already added, etc.).
  void AddSdchDictionary(const std::string& dictionary_text,
                         const GURL& dictionary_url);

  // Registration for events generated by the SDCH subsystem.
  void AddObserver(SdchObserver* observer);
  void RemoveObserver(SdchObserver* observer);

 private:
  struct BlacklistInfo {
    BlacklistInfo()
        : count(0),
          exponential_count(0),
          reason(MIN_PROBLEM_CODE) {}

    int count;                   // # of times to refuse SDCH advertisement.
    int exponential_count;       // Current exponential backoff ratchet.
    ProblemCodes reason;         // Why domain was blacklisted.

  };
  typedef std::map<std::string, BlacklistInfo> DomainBlacklistInfo;
  typedef std::set<std::string> ExperimentSet;

  // Determines whether a "Get-Dictionary" header is legal (dictionary
  // url has appropriate relationship to referrer url) in the SDCH
  // protocol.  Return true if fetch is legal.
  bool CanFetchDictionary(const GURL& referring_url,
                          const GURL& dictionary_url) const;

  // A map of dictionaries info indexed by the hash that the server provides.
  typedef std::map<std::string, scoped_refptr<Dictionary> > DictionaryMap;

  // Support SDCH compression, by advertising in headers.
  static bool g_sdch_enabled_;

  // Support SDCH compression for HTTPS requests and responses. When supported,
  // HTTPS applicable dictionaries MUST have been acquired securely via HTTPS.
  static bool g_secure_scheme_supported_;

  // A simple implementation of a RFC 3548 "URL safe" base64 encoder.
  static void UrlSafeBase64Encode(const std::string& input,
                                  std::string* output);
  DictionaryMap dictionaries_;

  // List domains where decode failures have required disabling sdch.
  DomainBlacklistInfo blacklisted_domains_;

  // List of hostnames for which a latency experiment is allowed (because a
  // round trip test has recently passed).
  ExperimentSet allow_latency_experiment_;

  // Observers that want to be notified of SDCH events.
  // Assert list is empty on destruction since if there is an observer
  // that hasn't removed itself from the list, that observer probably
  // has a reference to the SdchManager.
  ObserverList<SdchObserver, true> observers_;

  base::ThreadChecker thread_checker_;

  DISALLOW_COPY_AND_ASSIGN(SdchManager);
};

}  // namespace net

#endif  // NET_BASE_SDCH_MANAGER_H_