content/browser/appcache/appcache_update_job.h

   1 // Copyright (c) 2012 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 #ifndef CONTENT_BROWSER_APPCACHE_APPCACHE_UPDATE_JOB_H_
   6 #define CONTENT_BROWSER_APPCACHE_APPCACHE_UPDATE_JOB_H_
   7
   8 #include <deque>
   9 #include <map>
  10 #include <set>
  11 #include <string>
  12 #include <vector>
  13
  14 #include "base/gtest_prod_util.h"
  15 #include "base/memory/ref_counted.h"
  16 #include "base/time/time.h"
  17 #include "content/browser/appcache/appcache.h"
  18 #include "content/browser/appcache/appcache_host.h"
  19 #include "content/browser/appcache/appcache_response.h"
  20 #include "content/browser/appcache/appcache_service_impl.h"
  21 #include "content/browser/appcache/appcache_storage.h"
  22 #include "content/common/appcache_interfaces.h"
  23 #include "content/common/content_export.h"
  24 #include "net/base/completion_callback.h"
  25 #include "net/http/http_response_headers.h"
  26 #include "net/url_request/url_request.h"
  27 #include "url/gurl.h"
  28
  29 namespace content {
  30 FORWARD_DECLARE_TEST(AppCacheGroupTest, QueueUpdate);
  31 class AppCacheGroupTest;
  32 class AppCacheUpdateJobTest;
  33 class HostNotifier;
  34
  35 // Application cache Update algorithm and state.
  36 class CONTENT_EXPORT AppCacheUpdateJob
  37     : public AppCacheStorage::Delegate,
  38       public AppCacheHost::Observer,
  39       public AppCacheServiceImpl::Observer {
  40  public:
  41   // Used for uma stats only for now, so new values are append only.
  42   enum ResultType {
  43     UPDATE_OK, DB_ERROR, DISKCACHE_ERROR, QUOTA_ERROR, REDIRECT_ERROR,
  44     MANIFEST_ERROR, NETWORK_ERROR, SERVER_ERROR, CANCELLED_ERROR,
  45     NUM_UPDATE_JOB_RESULT_TYPES
  46   };
  47
  48   AppCacheUpdateJob(AppCacheServiceImpl* service, AppCacheGroup* group);
  49   virtual ~AppCacheUpdateJob();
  50
  51   // Triggers the update process or adds more info if this update is already
  52   // in progress.
  53   void StartUpdate(AppCacheHost* host, const GURL& new_master_resource);
  54
  55  private:
  56   friend class content::AppCacheGroupTest;
  57   friend class content::AppCacheUpdateJobTest;
  58   class URLFetcher;
  59
  60   // Master entries have multiple hosts, for example, the same page is opened
  61   // in different tabs.
  62   typedef std::vector<AppCacheHost*> PendingHosts;
  63   typedef std::map<GURL, PendingHosts> PendingMasters;
  64   typedef std::map<GURL, URLFetcher*> PendingUrlFetches;
  65   typedef std::map<int64, GURL> LoadingResponses;
  66
  67   static const int kRerunDelayMs = 1000;
  68
  69   // TODO(michaeln): Rework the set of states vs update types vs stored states.
  70   // The NO_UPDATE state is really more of an update type. For all update types
  71   // storing the results is relevant.
  72
  73   enum UpdateType {
  74     UNKNOWN_TYPE,
  75     UPGRADE_ATTEMPT,
  76     CACHE_ATTEMPT,
  77   };
  78
  79   enum InternalUpdateState {
  80     FETCH_MANIFEST,
  81     NO_UPDATE,
  82     DOWNLOADING,
  83
  84     // Every state after this comment indicates the update is terminating.
  85     REFETCH_MANIFEST,
  86     CACHE_FAILURE,
  87     CANCELLED,
  88     COMPLETED,
  89   };
  90
  91   enum StoredState {
  92     UNSTORED,
  93     STORING,
  94     STORED,
  95   };
  96
  97   struct UrlToFetch {
  98     UrlToFetch(const GURL& url, bool checked, AppCacheResponseInfo* info);
  99     ~UrlToFetch();
 100
 101     GURL url;
 102     bool storage_checked;
 103     scoped_refptr<AppCacheResponseInfo> existing_response_info;
 104   };
 105
 106   class URLFetcher : public net::URLRequest::Delegate {
 107    public:
 108     enum FetchType {
 109       MANIFEST_FETCH,
 110       URL_FETCH,
 111       MASTER_ENTRY_FETCH,
 112       MANIFEST_REFETCH,
 113     };
 114     URLFetcher(const GURL& url,
 115                FetchType fetch_type,
 116                AppCacheUpdateJob* job);
 117     virtual ~URLFetcher();
 118     void Start();
 119     FetchType fetch_type() const { return fetch_type_; }
 120     net::URLRequest* request() const { return request_.get(); }
 121     const AppCacheEntry& existing_entry() const { return existing_entry_; }
 122     const std::string& manifest_data() const { return manifest_data_; }
 123     AppCacheResponseWriter* response_writer() const {
 124       return response_writer_.get();
 125     }
 126     void set_existing_response_headers(net::HttpResponseHeaders* headers) {
 127       existing_response_headers_ = headers;
 128     }
 129     void set_existing_entry(const AppCacheEntry& entry) {
 130       existing_entry_ = entry;
 131     }
 132     ResultType result() const { return result_; }
 133     int redirect_response_code() const { return redirect_response_code_; }
 134
 135    private:
 136     // URLRequest::Delegate overrides
 137     virtual void OnReceivedRedirect(net::URLRequest* request,
 138                                     const net::RedirectInfo& redirect_info,
 139                                     bool* defer_redirect) OVERRIDE;
 140     virtual void OnResponseStarted(net::URLRequest* request) OVERRIDE;
 141     virtual void OnReadCompleted(net::URLRequest* request,
 142                                  int bytes_read) OVERRIDE;
 143
 144     void AddConditionalHeaders(const net::HttpResponseHeaders* headers);
 145     void OnWriteComplete(int result);
 146     void ReadResponseData();
 147     bool ConsumeResponseData(int bytes_read);
 148     void OnResponseCompleted();
 149     bool MaybeRetryRequest();
 150
 151     GURL url_;
 152     AppCacheUpdateJob* job_;
 153     FetchType fetch_type_;
 154     int retry_503_attempts_;
 155     scoped_refptr<net::IOBuffer> buffer_;
 156     scoped_ptr<net::URLRequest> request_;
 157     AppCacheEntry existing_entry_;
 158     scoped_refptr<net::HttpResponseHeaders> existing_response_headers_;
 159     std::string manifest_data_;
 160     ResultType result_;
 161     int redirect_response_code_;
 162     scoped_ptr<AppCacheResponseWriter> response_writer_;
 163   };  // class URLFetcher
 164
 165   AppCacheResponseWriter* CreateResponseWriter();
 166
 167   // Methods for AppCacheStorage::Delegate.
 168   virtual void OnResponseInfoLoaded(AppCacheResponseInfo* response_info,
 169                                     int64 response_id) OVERRIDE;
 170   virtual void OnGroupAndNewestCacheStored(AppCacheGroup* group,
 171                                            AppCache* newest_cache,
 172                                            bool success,
 173                                            bool would_exceed_quota) OVERRIDE;
 174   virtual void OnGroupMadeObsolete(AppCacheGroup* group,
 175                                    bool success,
 176                                    int response_code) OVERRIDE;
 177
 178   // Methods for AppCacheHost::Observer.
 179   virtual void OnCacheSelectionComplete(AppCacheHost* host) OVERRIDE {}  // N/A
 180   virtual void OnDestructionImminent(AppCacheHost* host) OVERRIDE;
 181
 182   // Methods for AppCacheServiceImpl::Observer.
 183   virtual void OnServiceReinitialized(
 184       AppCacheStorageReference* old_storage) OVERRIDE;
 185
 186   void HandleCacheFailure(const AppCacheErrorDetails& details,
 187                           ResultType result,
 188                           const GURL& failed_resource_url);
 189
 190   void FetchManifest(bool is_first_fetch);
 191   void HandleManifestFetchCompleted(URLFetcher* fetcher);
 192   void ContinueHandleManifestFetchCompleted(bool changed);
 193
 194   void HandleUrlFetchCompleted(URLFetcher* fetcher);
 195   void HandleMasterEntryFetchCompleted(URLFetcher* fetcher);
 196
 197   void HandleManifestRefetchCompleted(URLFetcher* fetcher);
 198   void OnManifestInfoWriteComplete(int result);
 199   void OnManifestDataWriteComplete(int result);
 200
 201   void StoreGroupAndCache();
 202
 203   void NotifySingleHost(AppCacheHost* host, AppCacheEventID event_id);
 204   void NotifyAllAssociatedHosts(AppCacheEventID event_id);
 205   void NotifyAllProgress(const GURL& url);
 206   void NotifyAllFinalProgress();
 207   void NotifyAllError(const AppCacheErrorDetails& detals);
 208   void LogConsoleMessageToAll(const std::string& message);
 209   void AddAllAssociatedHostsToNotifier(HostNotifier* notifier);
 210
 211   // Checks if manifest is byte for byte identical with the manifest
 212   // in the newest application cache.
 213   void CheckIfManifestChanged();
 214   void OnManifestDataReadComplete(int result);
 215
 216   // Creates the list of files that may need to be fetched and initiates
 217   // fetches. Section 6.9.4 steps 12-17
 218   void BuildUrlFileList(const Manifest& manifest);
 219   void AddUrlToFileList(const GURL& url, int type);
 220   void FetchUrls();
 221   void CancelAllUrlFetches();
 222   bool ShouldSkipUrlFetch(const AppCacheEntry& entry);
 223
 224   // If entry already exists in the cache currently being updated, merge
 225   // the entry type information with the existing entry.
 226   // Returns true if entry exists in cache currently being updated.
 227   bool AlreadyFetchedEntry(const GURL& url, int entry_type);
 228
 229   // TODO(jennb): Delete when update no longer fetches master entries directly.
 230   // Creates the list of master entries that need to be fetched and initiates
 231   // fetches.
 232   void AddMasterEntryToFetchList(AppCacheHost* host, const GURL& url,
 233                                  bool is_new);
 234   void FetchMasterEntries();
 235   void CancelAllMasterEntryFetches(const AppCacheErrorDetails& details);
 236
 237   // Asynchronously loads the entry from the newest complete cache if the
 238   // HTTP caching semantics allow.
 239   // Returns false if immediately obvious that data cannot be loaded from
 240   // newest complete cache.
 241   bool MaybeLoadFromNewestCache(const GURL& url, AppCacheEntry& entry);
 242   void LoadFromNewestCacheFailed(const GURL& url,
 243                                  AppCacheResponseInfo* newest_response_info);
 244
 245   // Does nothing if update process is still waiting for pending master
 246   // entries or URL fetches to complete downloading. Otherwise, completes
 247   // the update process.
 248   void MaybeCompleteUpdate();
 249
 250   // Schedules a rerun of the entire update with the same parameters as
 251   // this update job after a short delay.
 252   void ScheduleUpdateRetry(int delay_ms);
 253
 254   void Cancel();
 255   void ClearPendingMasterEntries();
 256   void DiscardInprogressCache();
 257   void DiscardDuplicateResponses();
 258
 259   void LogHistogramStats(ResultType result, const GURL& failed_resource_url);
 260   void MadeProgress() { last_progress_time_ = base::Time::Now(); }
 261
 262   // Deletes this object after letting the stack unwind.
 263   void DeleteSoon();
 264
 265   bool IsTerminating() { return internal_state_ >= REFETCH_MANIFEST ||
 266                                 stored_state_ != UNSTORED; }
 267
 268   AppCacheServiceImpl* service_;
 269   const GURL manifest_url_;  // here for easier access
 270
 271   // Defined prior to refs to AppCaches and Groups because destruction
 272   // order matters, the disabled_storage_reference_ must outlive those
 273   // objects.
 274   scoped_refptr<AppCacheStorageReference> disabled_storage_reference_;
 275
 276   scoped_refptr<AppCache> inprogress_cache_;
 277
 278   AppCacheGroup* group_;
 279
 280   UpdateType update_type_;
 281   InternalUpdateState internal_state_;
 282   base::Time last_progress_time_;
 283
 284   PendingMasters pending_master_entries_;
 285   size_t master_entries_completed_;
 286
 287   // TODO(jennb): Delete when update no longer fetches master entries directly.
 288   // Helper containers to track which pending master entries have yet to be
 289   // fetched and which are currently being fetched. Master entries that
 290   // are listed in the manifest may be fetched as a regular URL instead of
 291   // as a separate master entry fetch to optimize against duplicate fetches.
 292   std::set<GURL> master_entries_to_fetch_;
 293   PendingUrlFetches master_entry_fetches_;
 294
 295   // URLs of files to fetch along with their flags.
 296   AppCache::EntryMap url_file_list_;
 297   size_t url_fetches_completed_;
 298
 299   // Helper container to track which urls have not been fetched yet. URLs are
 300   // removed when the fetch is initiated. Flag indicates whether an attempt
 301   // to load the URL from storage has already been tried and failed.
 302   std::deque<UrlToFetch> urls_to_fetch_;
 303
 304   // Helper container to track which urls are being loaded from response
 305   // storage.
 306   LoadingResponses loading_responses_;
 307
 308   // Keep track of pending URL requests so we can cancel them if necessary.
 309   URLFetcher* manifest_fetcher_;
 310   PendingUrlFetches pending_url_fetches_;
 311
 312   // Temporary storage of manifest response data for parsing and comparison.
 313   std::string manifest_data_;
 314   scoped_ptr<net::HttpResponseInfo> manifest_response_info_;
 315   scoped_ptr<AppCacheResponseWriter> manifest_response_writer_;
 316   scoped_refptr<net::IOBuffer> read_manifest_buffer_;
 317   std::string loaded_manifest_data_;
 318   scoped_ptr<AppCacheResponseReader> manifest_response_reader_;
 319   bool manifest_has_valid_mime_type_;
 320
 321   // New master entries added to the cache by this job, used to cleanup
 322   // in error conditions.
 323   std::vector<GURL> added_master_entries_;
 324
 325   // Response ids stored by this update job, used to cleanup in
 326   // error conditions.
 327   std::vector<int64> stored_response_ids_;
 328
 329   // In some cases we fetch the same resource multiple times, and then
 330   // have to delete the duplicates upon successful update. These ids
 331   // are also in the stored_response_ids_ collection so we only schedule
 332   // these for deletion on success.
 333   // TODO(michaeln): Rework when we no longer fetches master entries directly.
 334   std::vector<int64> duplicate_response_ids_;
 335
 336   // Whether we've stored the resulting group/cache yet.
 337   StoredState stored_state_;
 338
 339   AppCacheStorage* storage_;
 340
 341   FRIEND_TEST_ALL_PREFIXES(content::AppCacheGroupTest, QueueUpdate);
 342
 343   DISALLOW_COPY_AND_ASSIGN(AppCacheUpdateJob);
 344 };
 345
 346 }  // namespace content
 347
 348 #endif  // CONTENT_BROWSER_APPCACHE_APPCACHE_UPDATE_JOB_H_