code.grnet.gr Git - pithos-ms-client/blob - trunk/Pithos.Core/Agents/PollAgent.cs

   1 #region\r
   2 /* -----------------------------------------------------------------------\r
   3  * <copyright file="PollAgent.cs" company="GRNet">\r
   4  * \r
   5  * Copyright 2011-2012 GRNET S.A. All rights reserved.\r
   6  *\r
   7  * Redistribution and use in source and binary forms, with or\r
   8  * without modification, are permitted provided that the following\r
   9  * conditions are met:\r
  10  *\r
  11  *   1. Redistributions of source code must retain the above\r
  12  *      copyright notice, this list of conditions and the following\r
  13  *      disclaimer.\r
  14  *\r
  15  *   2. Redistributions in binary form must reproduce the above\r
  16  *      copyright notice, this list of conditions and the following\r
  17  *      disclaimer in the documentation and/or other materials\r
  18  *      provided with the distribution.\r
  19  *\r
  20  *\r
  21  * THIS SOFTWARE IS PROVIDED BY GRNET S.A. ``AS IS'' AND ANY EXPRESS\r
  22  * OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED\r
  23  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR\r
  24  * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL GRNET S.A OR\r
  25  * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,\r
  26  * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT\r
  27  * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF\r
  28  * USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED\r
  29  * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT\r
  30  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN\r
  31  * ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE\r
  32  * POSSIBILITY OF SUCH DAMAGE.\r
  33  *\r
  34  * The views and conclusions contained in the software and\r
  35  * documentation are those of the authors and should not be\r
  36  * interpreted as representing official policies, either expressed\r
  37  * or implied, of GRNET S.A.\r
  38  * </copyright>\r
  39  * -----------------------------------------------------------------------\r
  40  */\r
  41 #endregion\r
  42 \r
  43 using System.Collections.Concurrent;\r
  44 using System.ComponentModel.Composition;\r
  45 using System.Diagnostics;\r
  46 using System.Diagnostics.Contracts;\r
  47 using System.IO;\r
  48 using System.Threading;\r
  49 using System.Threading.Tasks;\r
  50 using System.Threading.Tasks.Dataflow;\r
  51 using Castle.ActiveRecord;\r
  52 using Pithos.Interfaces;\r
  53 using Pithos.Network;\r
  54 using log4net;\r
  55 \r
  56 namespace Pithos.Core.Agents\r
  57 {\r
  58     using System;\r
  59     using System.Collections.Generic;\r
  60     using System.Linq;\r
  61     using System.Text;\r
  62 \r
  63     /// <summary>\r
  64     /// PollAgent periodically polls the server to detect object changes. The agent retrieves a listing of all\r
  65     /// objects and compares it with a previously cached version to detect differences. \r
  66     /// New files are downloaded, missing files are deleted from the local file system and common files are compared\r
  67     /// to determine the appropriate action\r
  68     /// </summary>\r
  69     [Export]\r
  70     public class PollAgent\r
  71     {\r
  72         private static readonly ILog Log = LogManager.GetLogger("PollAgent");\r
  73 \r
  74         [System.ComponentModel.Composition.Import]\r
  75         public IStatusKeeper StatusKeeper { get; set; }\r
  76 \r
  77         [System.ComponentModel.Composition.Import]\r
  78         public IPithosSettings Settings { get; set; }\r
  79 \r
  80         [System.ComponentModel.Composition.Import]\r
  81         public NetworkAgent NetworkAgent { get; set; }\r
  82 \r
  83         public IStatusNotification StatusNotification { get; set; }\r
  84 \r
  85         private bool _firstPoll = true;\r
  86 \r
  87         //The Sync Event signals a manual synchronisation\r
  88         private readonly AsyncManualResetEvent _syncEvent = new AsyncManualResetEvent();\r
  89 \r
  90         private ConcurrentDictionary<string, DateTime> _lastSeen = new ConcurrentDictionary<string, DateTime>();\r
  91         private readonly ConcurrentBag<AccountInfo> _accounts = new ConcurrentBag<AccountInfo>();\r
  92 \r
  93 \r
  94         /// <summary>\r
  95         /// Start a manual synchronization\r
  96         /// </summary>\r
  97         public void SynchNow()\r
  98         {            \r
  99             _syncEvent.Set();\r
 100         }\r
 101 \r
 102         /// <summary>\r
 103         /// Remote files are polled periodically. Any changes are processed\r
 104         /// </summary>\r
 105         /// <param name="since"></param>\r
 106         /// <returns></returns>\r
 107         public async Task PollRemoteFiles(DateTime? since = null)\r
 108         {\r
 109             Debug.Assert(Thread.CurrentThread.IsBackground, "Polling Ended up in the main thread!");\r
 110 \r
 111             UpdateStatus(PithosStatus.Syncing);\r
 112             StatusNotification.Notify(new PollNotification());\r
 113 \r
 114             using (log4net.ThreadContext.Stacks["Retrieve Remote"].Push("All accounts"))\r
 115             {\r
 116                 //If this poll fails, we will retry with the same since value\r
 117                 var nextSince = since;\r
 118                 try\r
 119                 {\r
 120                     //Next time we will check for all changes since the current check minus 1 second\r
 121                     //This is done to ensure there are no discrepancies due to clock differences\r
 122                     var current = DateTime.Now.AddSeconds(-1);\r
 123 \r
 124                     var tasks = from accountInfo in _accounts\r
 125                                 select ProcessAccountFiles(accountInfo, since);\r
 126 \r
 127                     await TaskEx.WhenAll(tasks.ToList());\r
 128 \r
 129                     _firstPoll = false;\r
 130                     //Reschedule the poll with the current timestamp as a "since" value\r
 131                     nextSince = current;\r
 132                 }\r
 133                 catch (Exception ex)\r
 134                 {\r
 135                     Log.ErrorFormat("Error while processing accounts\r\n{0}", ex);\r
 136                     //In case of failure retry with the same "since" value\r
 137                 }\r
 138 \r
 139                 UpdateStatus(PithosStatus.InSynch);\r
 140                 //The multiple try blocks are required because we can't have an await call\r
 141                 //inside a finally block\r
 142                 //TODO: Find a more elegant solution for reschedulling in the event of an exception\r
 143                 try\r
 144                 {\r
 145                     //Wait for the polling interval to pass or the Sync event to be signalled\r
 146                     nextSince = await WaitForScheduledOrManualPoll(nextSince);\r
 147                 }\r
 148                 finally\r
 149                 {\r
 150                     //Ensure polling is scheduled even in case of error\r
 151                     TaskEx.Run(() => PollRemoteFiles(nextSince));                        \r
 152                 }\r
 153             }\r
 154         }\r
 155 \r
 156         /// <summary>\r
 157         /// Wait for the polling period to expire or a manual sync request\r
 158         /// </summary>\r
 159         /// <param name="since"></param>\r
 160         /// <returns></returns>\r
 161         private async Task<DateTime?> WaitForScheduledOrManualPoll(DateTime? since)\r
 162         {\r
 163             var sync = _syncEvent.WaitAsync();\r
 164             var wait = TaskEx.Delay(TimeSpan.FromSeconds(Settings.PollingInterval), NetworkAgent.CancellationToken);\r
 165             var signaledTask = await TaskEx.WhenAny(sync, wait);\r
 166 \r
 167             //Wait for network processing to finish before polling\r
 168             var pauseTask=NetworkAgent.ProceedEvent.WaitAsync();\r
 169             await TaskEx.WhenAll(signaledTask, pauseTask);\r
 170 \r
 171             //If polling is signalled by SynchNow, ignore the since tag\r
 172             if (sync.IsCompleted)\r
 173             {\r
 174                 //TODO: Must convert to AutoReset\r
 175                 _syncEvent.Reset();\r
 176                 return null;\r
 177             }\r
 178             return since;\r
 179         }\r
 180 \r
 181         public async Task ProcessAccountFiles(AccountInfo accountInfo, DateTime? since = null)\r
 182         {\r
 183             if (accountInfo == null)\r
 184                 throw new ArgumentNullException("accountInfo");\r
 185             if (String.IsNullOrWhiteSpace(accountInfo.AccountPath))\r
 186                 throw new ArgumentException("The AccountInfo.AccountPath is empty", "accountInfo");\r
 187             Contract.EndContractBlock();\r
 188 \r
 189 \r
 190             using (log4net.ThreadContext.Stacks["Retrieve Remote"].Push(accountInfo.UserName))\r
 191             {\r
 192                 await NetworkAgent.GetDeleteAwaiter();\r
 193 \r
 194                 Log.Info("Scheduled");\r
 195                 var client = new CloudFilesClient(accountInfo);\r
 196 \r
 197                 //We don't need to check the trash container\r
 198                 var containers = client.ListContainers(accountInfo.UserName).Where(c=>c.Name!="trash");\r
 199 \r
 200 \r
 201                 CreateContainerFolders(accountInfo, containers);\r
 202 \r
 203                 try\r
 204                 {\r
 205                     //Wait for any deletions to finish\r
 206                     await NetworkAgent.GetDeleteAwaiter();\r
 207                     //Get the poll time now. We may miss some deletions but it's better to keep a file that was deleted\r
 208                     //than delete a file that was created while we were executing the poll                    \r
 209                     var pollTime = DateTime.Now;\r
 210 \r
 211                     //Get the list of server objects changed since the last check\r
 212                     //The name of the container is passed as state in order to create a dictionary of tasks in a subsequent step\r
 213                     var listObjects = (from container in containers\r
 214                                        select Task<IList<ObjectInfo>>.Factory.StartNew(_ =>\r
 215                                              client.ListObjects(accountInfo.UserName, container.Name, since), container.Name)).ToList();\r
 216                     //BUG: Can't detect difference between no changes or no objects\r
 217                     //ListObjects returns nothing if there are no changes since the last check time (since value)                    \r
 218                     //TODO: Must detect the difference between no server objects and no change\r
 219 \r
 220                     //NOTE: One option is to "mark" all result lists with their container name, or \r
 221                     //rather the url of the container\r
 222                     //Another option \r
 223 \r
 224                     var listShared = Task<IList<ObjectInfo>>.Factory.StartNew(_ => \r
 225                         client.ListSharedObjects(since), "shared");\r
 226                     listObjects.Add(listShared);\r
 227                     var listTasks = await Task.Factory.WhenAll(listObjects.ToArray());\r
 228 \r
 229                     using (log4net.ThreadContext.Stacks["SCHEDULE"].Push("Process Results"))\r
 230                     {\r
 231                         var dict = listTasks.ToDictionary(t => t.AsyncState);\r
 232 \r
 233                         //Get all non-trash objects. Remember, the container name is stored in AsyncState\r
 234                         var remoteObjects = from objectList in listTasks\r
 235                                             where (string)objectList.AsyncState != "trash"\r
 236                                             from obj in objectList.Result\r
 237                                             select obj;\r
 238 \r
 239                         var sharedObjects = dict["shared"].Result;\r
 240 \r
 241                         //DON'T process trashed files\r
 242                         //If some files are deleted and added again to a folder, they will be deleted\r
 243                         //even though they are new.\r
 244                         //We would have to check file dates and hashes to ensure that a trashed file\r
 245                         //can be deleted safely from the local hard drive.\r
 246                         /*\r
 247                         //Items with the same name, hash may be both in the container and the trash\r
 248                         //Don't delete items that exist in the container\r
 249                         var realTrash = from trash in trashObjects\r
 250                                         where\r
 251                                             !remoteObjects.Any(\r
 252                                                 info => info.Name == trash.Name && info.Hash == trash.Hash)\r
 253                                         select trash;\r
 254                         ProcessTrashedFiles(accountInfo, realTrash);\r
 255 */\r
 256 \r
 257                         var cleanRemotes = (from info in remoteObjects.Union(sharedObjects)\r
 258                                             let name = info.Name\r
 259                                             where !name.EndsWith(".ignore", StringComparison.InvariantCultureIgnoreCase) &&\r
 260                                                   !name.StartsWith(FolderConstants.CacheFolder + "/",\r
 261                                                                    StringComparison.InvariantCultureIgnoreCase)\r
 262                                             select info).ToList();\r
 263 \r
 264                         var differencer = _differencer.PostSnapshot(accountInfo, cleanRemotes);\r
 265 \r
 266                         ProcessDeletedFiles(accountInfo, differencer.Deleted.FilterDirectlyBelow(SelectiveUris), pollTime);\r
 267 \r
 268                         // @@@ NEED To add previous state here as well, To compare with previous hash\r
 269 \r
 270                         \r
 271 \r
 272                         //Create a list of actions from the remote files\r
 273                         var allActions = ChangesToActions(accountInfo, differencer.Changed.FilterDirectlyBelow(SelectiveUris))\r
 274                                         .Union(\r
 275                                         CreatesToActions(accountInfo, differencer.Created.FilterDirectlyBelow(SelectiveUris)));\r
 276 \r
 277                         //And remove those that are already being processed by the agent\r
 278                         var distinctActions = allActions\r
 279                             .Except(NetworkAgent.GetEnumerable(), new PithosMonitor.LocalFileComparer())\r
 280                             .ToList();\r
 281 \r
 282                         //Queue all the actions\r
 283                         foreach (var message in distinctActions)\r
 284                         {\r
 285                             NetworkAgent.Post(message);\r
 286                         }\r
 287 \r
 288                         Log.Info("[LISTENER] End Processing");\r
 289                     }\r
 290                 }\r
 291                 catch (Exception ex)\r
 292                 {\r
 293                     Log.ErrorFormat("[FAIL] ListObjects for{0} in ProcessRemoteFiles with {1}", accountInfo.UserName, ex);\r
 294                     return;\r
 295                 }\r
 296 \r
 297                 Log.Info("[LISTENER] Finished");\r
 298 \r
 299             }\r
 300         }\r
 301 \r
 302         AccountsDifferencer _differencer = new AccountsDifferencer();\r
 303         private List<Uri> _selectiveUris=new List<Uri>();\r
 304 \r
 305         /// <summary>\r
 306         /// Deletes local files that are not found in the list of cloud files\r
 307         /// </summary>\r
 308         /// <param name="accountInfo"></param>\r
 309         /// <param name="cloudFiles"></param>\r
 310         /// <param name="pollTime"></param>\r
 311         private void ProcessDeletedFiles(AccountInfo accountInfo, IEnumerable<ObjectInfo> cloudFiles, DateTime pollTime)\r
 312         {\r
 313             if (accountInfo == null)\r
 314                 throw new ArgumentNullException("accountInfo");\r
 315             if (String.IsNullOrWhiteSpace(accountInfo.AccountPath))\r
 316                 throw new ArgumentException("The AccountInfo.AccountPath is empty", "accountInfo");\r
 317             if (cloudFiles == null)\r
 318                 throw new ArgumentNullException("cloudFiles");\r
 319             Contract.EndContractBlock();\r
 320 \r
 321             //On the first run\r
 322             if (_firstPoll)\r
 323             {\r
 324                 //Only consider files that are not being modified, ie they are in the Unchanged state            \r
 325                 var deleteCandidates = FileState.Queryable.Where(state =>\r
 326                     state.FilePath.StartsWith(accountInfo.AccountPath)\r
 327                     && state.FileStatus == FileStatus.Unchanged).ToList();\r
 328 \r
 329 \r
 330                 //TODO: filesToDelete must take into account the Others container            \r
 331                 var filesToDelete = (from deleteCandidate in deleteCandidates\r
 332                                      let localFile = FileInfoExtensions.FromPath(deleteCandidate.FilePath)\r
 333                                      let relativeFilePath = localFile.AsRelativeTo(accountInfo.AccountPath)\r
 334                                      where\r
 335                                          !cloudFiles.Any(r => r.RelativeUrlToFilePath(accountInfo.UserName) == relativeFilePath)\r
 336                                      select localFile).ToList();\r
 337 \r
 338 \r
 339 \r
 340                 //Set the status of missing files to Conflict\r
 341                 foreach (var item in filesToDelete)\r
 342                 {\r
 343                     //Try to acquire a gate on the file, to take into account files that have been dequeued\r
 344                     //and are being processed\r
 345                     using (var gate = NetworkGate.Acquire(item.FullName, NetworkOperation.Deleting))\r
 346                     {\r
 347                         if (gate.Failed)\r
 348                             continue;\r
 349                         StatusKeeper.SetFileState(item.FullName, FileStatus.Conflict, FileOverlayStatus.Deleted);\r
 350                     }\r
 351                 }\r
 352                 UpdateStatus(PithosStatus.HasConflicts);\r
 353                 StatusNotification.NotifyConflicts(filesToDelete, String.Format("{0} local files are missing from Pithos, possibly because they were deleted", filesToDelete.Count));\r
 354                 StatusNotification.NotifyForFiles(filesToDelete, String.Format("{0} files were deleted", filesToDelete.Count), TraceLevel.Info);\r
 355             }\r
 356             else\r
 357             {\r
 358                 var deletedFiles = new List<FileSystemInfo>();\r
 359                 foreach (var objectInfo in cloudFiles)\r
 360                 {\r
 361                     var relativePath = objectInfo.RelativeUrlToFilePath(accountInfo.UserName);\r
 362                     var item = FileAgent.GetFileAgent(accountInfo).GetFileSystemInfo(relativePath);\r
 363                     if (item.Exists)\r
 364                     {\r
 365                         if ((item.Attributes & FileAttributes.ReadOnly) == FileAttributes.ReadOnly)\r
 366                         {\r
 367                             item.Attributes = item.Attributes & ~FileAttributes.ReadOnly;\r
 368 \r
 369                         }\r
 370                         item.Delete();\r
 371                         DateTime lastDate;\r
 372                         _lastSeen.TryRemove(item.FullName, out lastDate);\r
 373                         deletedFiles.Add(item);\r
 374                     }\r
 375                     StatusKeeper.SetFileState(item.FullName, FileStatus.Deleted, FileOverlayStatus.Deleted);\r
 376                 }\r
 377                 StatusNotification.NotifyForFiles(deletedFiles, String.Format("{0} files were deleted", deletedFiles.Count), TraceLevel.Info);\r
 378             }\r
 379 \r
 380         }\r
 381 \r
 382         //Creates an appropriate action for each server file\r
 383         private IEnumerable<CloudAction> ChangesToActions(AccountInfo accountInfo, IEnumerable<ObjectInfo> changes)\r
 384         {\r
 385             if (changes == null)\r
 386                 throw new ArgumentNullException();\r
 387             Contract.EndContractBlock();\r
 388             var fileAgent = FileAgent.GetFileAgent(accountInfo);\r
 389 \r
 390             //In order to avoid multiple iterations over the files, we iterate only once\r
 391             //over the remote files\r
 392             foreach (var objectInfo in changes)\r
 393             {\r
 394                 var relativePath = objectInfo.RelativeUrlToFilePath(accountInfo.UserName);\r
 395                 //and remove any matching objects from the list, adding them to the commonObjects list\r
 396                 if (fileAgent.Exists(relativePath))\r
 397                 {\r
 398                     //If a directory object already exists, we don't need to perform any other action                    \r
 399                     var localFile = fileAgent.GetFileSystemInfo(relativePath);\r
 400                     if (objectInfo.Content_Type == @"application/directory" && localFile is DirectoryInfo)\r
 401                         continue;\r
 402                     using (new SessionScope(FlushAction.Never))\r
 403                     {\r
 404                         var state = StatusKeeper.GetStateByFilePath(localFile.FullName);\r
 405                         _lastSeen[localFile.FullName] = DateTime.Now;\r
 406                         //Common files should be checked on a per-case basis to detect differences, which is newer\r
 407 \r
 408                         yield return new CloudAction(accountInfo, CloudActionType.MustSynch,\r
 409                                                      localFile, objectInfo, state, accountInfo.BlockSize,\r
 410                                                      accountInfo.BlockHash);\r
 411                     }\r
 412                 }\r
 413                 else\r
 414                 {\r
 415                     //Remote files should be downloaded\r
 416                     yield return new CloudDownloadAction(accountInfo, objectInfo);\r
 417                 }\r
 418             }\r
 419         }\r
 420 \r
 421         private IEnumerable<CloudAction> CreatesToActions(AccountInfo accountInfo, IEnumerable<ObjectInfo> creates)\r
 422         {\r
 423             if (creates == null)\r
 424                 throw new ArgumentNullException();\r
 425             Contract.EndContractBlock();\r
 426             var fileAgent = FileAgent.GetFileAgent(accountInfo);\r
 427 \r
 428             //In order to avoid multiple iterations over the files, we iterate only once\r
 429             //over the remote files\r
 430             foreach (var objectInfo in creates)\r
 431             {\r
 432                 var relativePath = objectInfo.RelativeUrlToFilePath(accountInfo.UserName);\r
 433                 //and remove any matching objects from the list, adding them to the commonObjects list\r
 434                 if (fileAgent.Exists(relativePath))\r
 435                 {\r
 436                     //If the object already exists, we probably have a conflict\r
 437                     //If a directory object already exists, we don't need to perform any other action                    \r
 438                     var localFile = fileAgent.GetFileSystemInfo(relativePath);\r
 439                     StatusKeeper.SetFileState(localFile.FullName, FileStatus.Conflict, FileOverlayStatus.Conflict);\r
 440                 }\r
 441                 else\r
 442                 {\r
 443                     //Remote files should be downloaded\r
 444                     yield return new CloudDownloadAction(accountInfo, objectInfo);\r
 445                 }\r
 446             }\r
 447         }\r
 448 \r
 449         private void ProcessTrashedFiles(AccountInfo accountInfo, IEnumerable<ObjectInfo> trashObjects)\r
 450         {\r
 451             var fileAgent = FileAgent.GetFileAgent(accountInfo);\r
 452             foreach (var trashObject in trashObjects)\r
 453             {\r
 454                 var barePath = trashObject.RelativeUrlToFilePath(accountInfo.UserName);\r
 455                 //HACK: Assume only the "pithos" container is used. Must find out what happens when\r
 456                 //deleting a file from a different container\r
 457                 var relativePath = Path.Combine("pithos", barePath);\r
 458                 fileAgent.Delete(relativePath);\r
 459             }\r
 460         }\r
 461 \r
 462         /// <summary>\r
 463         /// Notify the UI to update the visual status\r
 464         /// </summary>\r
 465         /// <param name="status"></param>\r
 466         private void UpdateStatus(PithosStatus status)\r
 467         {\r
 468             try\r
 469             {\r
 470                 StatusKeeper.SetPithosStatus(status);\r
 471                 StatusNotification.Notify(new Notification());\r
 472             }\r
 473             catch (Exception exc)\r
 474             {\r
 475                 //Failure is not critical, just log it\r
 476                 Log.Warn("Error while updating status", exc);\r
 477             }\r
 478         }\r
 479 \r
 480         private static void CreateContainerFolders(AccountInfo accountInfo, IEnumerable<ContainerInfo> containers)\r
 481         {\r
 482             var containerPaths = from container in containers\r
 483                                  let containerPath = Path.Combine(accountInfo.AccountPath, container.Name)\r
 484                                  where container.Name != FolderConstants.TrashContainer && !Directory.Exists(containerPath)\r
 485                                  select containerPath;\r
 486 \r
 487             foreach (var path in containerPaths)\r
 488             {\r
 489                 Directory.CreateDirectory(path);\r
 490             }\r
 491         }\r
 492 \r
 493         public void SetSyncUris(Uri[] uris)\r
 494         {            \r
 495             SelectiveUris=uris.ToList();\r
 496         }\r
 497 \r
 498         protected List<Uri> SelectiveUris\r
 499         {\r
 500             get { return _selectiveUris;}\r
 501             set { _selectiveUris = value; }\r
 502         }\r
 503 \r
 504         public void AddAccount(AccountInfo accountInfo)\r
 505         {\r
 506             if (!_accounts.Contains(accountInfo))\r
 507                 _accounts.Add(accountInfo);\r
 508         }\r
 509     }\r
 510 }\r