beagled/EvolutionMailDriver/EvolutionMailIndexableGenerator.cs

   1
   2 //
   3 // EvolutionMailIndexableGenerator.cs
   4 //
   5 // Copyright (C) 2004 Novell, Inc.
   6 //
   7 //
   8 //
   9 // Permission is hereby granted, free of charge, to any person obtaining a
  10 // copy of this software and associated documentation files (the "Software"),
  11 // to deal in the Software without restriction, including without limitation
  12 // the rights to use, copy, modify, merge, publish, distribute, sublicense,
  13 // and/or sell copies of the Software, and to permit persons to whom the
  14 // Software is furnished to do so, subject to the following conditions:
  15 //
  16 // The above copyright notice and this permission notice shall be included in
  17 // all copies or substantial portions of the Software.
  18 //
  19 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  20 // IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  21 // FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
  22 // AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  23 // LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  24 // FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
  25 // DEALINGS IN THE SOFTWARE.
  26 //
  27
  28 using System;
  29 using System.Collections;
  30 using System.IO;
  31 using System.Runtime.Serialization.Formatters.Binary;
  32 using System.Threading;
  33 using System.Xml;
  34
  35 using Beagle.Util;
  36 using Beagle.Daemon;
  37
  38 using Camel = Beagle.Util.Camel;
  39
  40 namespace Beagle.Daemon.EvolutionMailDriver {
  41
  42         public abstract class EvolutionMailIndexableGenerator : IIndexableGenerator {
  43
  44                 protected static bool Debug = false;
  45
  46                 private static bool gmime_initialized = false;
  47
  48                 private static ArrayList excludes = new ArrayList ();
  49
  50                 static EvolutionMailIndexableGenerator () {
  51                         foreach (ExcludeItem exclude in Conf.Indexing.Excludes)
  52                                 if (exclude.Type == ExcludeType.MailFolder)
  53                                         excludes.Add (exclude);
  54                 }
  55
  56                 protected EvolutionMailQueryable queryable;
  57
  58                 protected string account_name, folder_name;
  59                 protected int count, indexed_count;
  60
  61                 protected EvolutionMailIndexableGenerator (EvolutionMailQueryable queryable)
  62                 {
  63                         this.queryable = queryable;
  64                 }
  65
  66                 protected abstract string GetFolderName (FileSystemInfo info);
  67                 protected abstract bool Setup ();
  68                 protected abstract FileInfo CrawlFile { get; }
  69                 public abstract string GetTarget ();
  70                 public abstract bool HasNextIndexable ();
  71                 public abstract Indexable GetNextIndexable ();
  72                 public abstract void Checkpoint ();
  73
  74
  75                 protected double progress_percent;
  76                 public double ProgressPercent {
  77                         get { return progress_percent; }
  78                 }
  79
  80                 public void PostFlushHook ()
  81                 {
  82                         Checkpoint ();
  83                 }
  84
  85                 protected static void InitializeGMime ()
  86                 {
  87                         if (!gmime_initialized) {
  88                                 GMime.Global.Init ();
  89                                 gmime_initialized = true;
  90                         }
  91                 }
  92
  93                 protected bool IsSpamFolder (string name)
  94                 {
  95                         if (name.ToLower () == "spam" || name.ToLower () == "junk")
  96                                 return true;
  97                         else
  98                                 return false;
  99                 }
 100
 101                 protected bool IgnoreFolder (string path)
 102                 {
 103                         // FIXME: Use System.IO.Path
 104                         foreach (ExcludeItem exclude in excludes) {
 105                                 if (exclude.IsMatch (path))
 106                                         return true;
 107                         }
 108                         return false;
 109                 }
 110
 111                 protected void CrawlFinished ()
 112                 {
 113                         this.queryable.FileAttributesStore.AttachLastWriteTime (this.CrawlFile.FullName, File.GetLastWriteTimeUtc (this.CrawlFile.FullName));
 114                         this.queryable.RemoveGenerator (this);
 115                 }
 116
 117                 public string StatusName {
 118                         get { return this.CrawlFile.FullName; }
 119                 }
 120
 121                 public override bool Equals (object o)
 122                 {
 123                         EvolutionMailIndexableGenerator generator = o as EvolutionMailIndexableGenerator;
 124
 125                         if (generator == null)
 126                                 return false;
 127
 128                         if (Object.ReferenceEquals (this, generator))
 129                                 return true;
 130
 131                         if (this.CrawlFile.FullName == generator.CrawlFile.FullName)
 132                                 return true;
 133                         else
 134                                 return false;
 135                 }
 136
 137                 public override int GetHashCode ()
 138                 {
 139                         return this.CrawlFile.FullName.GetHashCode ();
 140                 }
 141         }
 142
 143         public class EvolutionMailIndexableGeneratorMbox : EvolutionMailIndexableGenerator {
 144                 private FileInfo mbox_info;
 145                 private int mbox_fd = -1;
 146                 private GMime.StreamFs mbox_stream;
 147                 private GMime.Parser mbox_parser;
 148                 private long file_size;
 149
 150                 public EvolutionMailIndexableGeneratorMbox (EvolutionMailQueryable queryable, FileInfo mbox_info) : base (queryable)
 151                 {
 152                         this.mbox_info = mbox_info;
 153                 }
 154
 155                 protected override string GetFolderName (FileSystemInfo info)
 156                 {
 157                         FileInfo file_info = (FileInfo) info;
 158                         DirectoryInfo di;
 159                         string folder_name = "";
 160
 161                         di = file_info.Directory;
 162                         while (di != null) {
 163                                 // Evo uses ".sbd" as the extension on a folder
 164                                 if (di.Extension == ".sbd")
 165                                         folder_name = Path.Combine (Path.GetFileNameWithoutExtension (di.Name), folder_name);
 166                                 else
 167                                         break;
 168
 169                                 di = di.Parent;
 170                         }
 171
 172                         return Path.Combine (folder_name, file_info.Name);
 173                 }
 174
 175                 protected override bool Setup ()
 176                 {
 177                         this.account_name = "local@local";
 178                         this.folder_name = this.GetFolderName (this.mbox_info);
 179
 180                         if (this.IsSpamFolder (this.folder_name))
 181                                 return false;
 182
 183                         if (this.IgnoreFolder (this.mbox_info.FullName))
 184                                 return false;
 185
 186                         return true;
 187                 }
 188
 189                 private long MboxLastOffset {
 190                         get {
 191                                 string offset_str = this.queryable.ReadDataLine ("offset-" + this.folder_name.Replace ('/', '-'));
 192                                 long offset = Convert.ToInt64 (offset_str);
 193
 194                                 return offset;
 195                         }
 196
 197                         set {
 198                                 this.queryable.WriteDataLine ("offset-" + this.folder_name.Replace ('/', '-'), value.ToString ());
 199                         }
 200                 }
 201
 202                 public override bool HasNextIndexable ()
 203                 {
 204                         if (this.account_name == null) {
 205                                 if (!Setup ()) {
 206                                         this.queryable.RemoveGenerator (this);
 207                                         return false;
 208                                 }
 209                         }
 210
 211                         if (this.mbox_fd < 0) {
 212                                 Logger.Log.Debug ("Opening mbox {0}", this.mbox_info.Name);
 213
 214                                 try {
 215                                         InitializeGMime ();
 216                                 } catch (Exception e) {
 217                                         Logger.Log.Warn (e, "Caught exception trying to initalize gmime:");
 218                                         return false;
 219                                 }
 220
 221                                 this.mbox_fd = Mono.Unix.Native.Syscall.open (this.mbox_info.FullName, Mono.Unix.Native.OpenFlags.O_RDONLY);
 222
 223                                 if (this.mbox_fd < 0) {
 224                                         Log.Error ("Unable to open {0}: {1}", this.mbox_info.FullName, Mono.Unix.Native.Stdlib.strerror (Mono.Unix.Native.Stdlib.GetLastError ()));
 225                                         return false;
 226                                 }
 227
 228                                 this.mbox_stream = new GMime.StreamFs (this.mbox_fd);
 229                                 this.mbox_stream.Seek ((int) this.MboxLastOffset);
 230                                 this.mbox_parser = new GMime.Parser (this.mbox_stream);
 231                                 this.mbox_parser.ScanFrom = true;
 232
 233                                 FileInfo info = new FileInfo (this.mbox_info.FullName);
 234                                 this.file_size = info.Length;
 235                         }
 236
 237                         if (this.mbox_parser.Eos ()) {
 238                                 long offset = this.mbox_parser.FromOffset;
 239
 240                                 this.mbox_stream.Close ();
 241
 242                                 this.mbox_fd = -1;
 243                                 this.mbox_stream.Dispose ();
 244                                 this.mbox_stream = null;
 245                                 this.mbox_parser.Dispose ();
 246                                 this.mbox_parser = null;
 247
 248                                 Logger.Log.Debug ("{0}: Finished indexing {1} messages", this.folder_name, this.indexed_count);
 249
 250                                 if (offset >= 0)
 251                                         this.MboxLastOffset = offset;
 252                                 this.CrawlFinished ();
 253
 254                                 return false;
 255                         } else
 256                                 return true;
 257                 }
 258
 259                 public override Indexable GetNextIndexable ()
 260                 {
 261                         using (GMime.Message message = this.mbox_parser.ConstructMessage ()) {
 262                                 // Work around what I think is a bug in GMime: If you
 263                                 // have a zero-byte file or seek to the end of a
 264                                 // file, parser.Eos () will return true until it
 265                                 // actually tries to read something off the wire.
 266                                 // Since parser.ConstructMessage() always returns a
 267                                 // message (which may also be a bug), we'll often get
 268                                 // one empty message which we need to deal with here.
 269                                 //
 270                                 // Check if its empty by seeing if the Headers
 271                                 // property is null or empty.
 272                                 if (message == null || message.Headers == null || message.Headers == "")
 273                                         return null;
 274
 275                                 ++this.count;
 276
 277                                 string x_evolution = message.GetHeader ("X-Evolution");
 278                                 if (x_evolution == null || x_evolution == "") {
 279                                         Logger.Log.Info ("{0}: Message at offset {1} has no X-Evolution header!",
 280                                                          this.folder_name, this.mbox_parser.FromOffset);
 281                                         return null;
 282                                 }
 283
 284                                 // This extracts the UID and flags from the X-Evolution header.
 285                                 // It may also contain user-defined flags and tags, but we don't
 286                                 // support those right now.
 287                                 int separator_idx = x_evolution.IndexOf ('-');
 288
 289                                 string uid_str = x_evolution.Substring (0, separator_idx);
 290                                 string uid = Convert.ToUInt32 (uid_str, 16).ToString (); // ugh.
 291                                 uint flags = Convert.ToUInt32 (x_evolution.Substring (separator_idx + 1, 4), 16);
 292
 293                                 Indexable indexable = this.GMimeMessageToIndexable (uid, message, flags);
 294
 295                                 if (Debug) {
 296                                         Logger.Log.Debug ("Constructed message {0} with uid {1}, flags {2}.  Indexable {3} null",
 297                                                           this.count, uid, flags, indexable == null ? "" : "not");
 298                                 }
 299
 300                                 if (indexable == null)
 301                                         return null;
 302
 303                                 ++this.indexed_count;
 304
 305                                 return indexable;
 306                         }
 307                 }
 308
 309                 private static bool CheckFlags (uint flags, Camel.CamelFlags test)
 310                 {
 311                         return (flags & (uint) test) == (uint) test;
 312                 }
 313
 314                 private Indexable GMimeMessageToIndexable (string uid, GMime.Message message, uint flags)
 315                 {
 316                         // Don't index messages flagged as junk
 317                         if (CheckFlags (flags, Camel.CamelFlags.Junk))
 318                                 return null;
 319
 320                         System.Uri uri = EvolutionMailQueryable.EmailUri (this.account_name, this.folder_name, uid);
 321                         Indexable indexable = new Indexable (uri);
 322
 323                         indexable.Timestamp = message.Date.ToUniversalTime ();
 324                         indexable.HitType = "MailMessage";
 325                         indexable.MimeType = "message/rfc822";
 326                         indexable.CacheContent = false;
 327
 328                         indexable.AddProperty (Property.NewUnsearched ("fixme:client", "evolution"));
 329                         indexable.AddProperty (Property.NewUnsearched ("fixme:account", "Local"));
 330                         indexable.AddProperty (Property.NewUnsearched ("fixme:folder", this.folder_name));
 331
 332                         GMime.InternetAddressList addrs;
 333
 334                         addrs = message.GetRecipients (GMime.Message.RecipientType.To);
 335                         foreach (GMime.InternetAddress ia in addrs) {
 336                                 if (this.folder_name == "Sent" && ia.AddressType != GMime.InternetAddressType.Group)
 337                                         indexable.AddProperty (Property.NewUnsearched ("fixme:sentTo", ia.Addr));
 338                         }
 339                         addrs.Dispose ();
 340
 341                         addrs = message.GetRecipients (GMime.Message.RecipientType.Cc);
 342                         foreach (GMime.InternetAddress ia in addrs) {
 343                                 if (this.folder_name == "Sent"  && ia.AddressType != GMime.InternetAddressType.Group)
 344                                         indexable.AddProperty (Property.NewUnsearched ("fixme:sentTo", ia.Addr));
 345                         }
 346                         addrs.Dispose ();
 347
 348                         addrs = GMime.InternetAddressList.ParseString (GMime.Utils.HeaderDecodePhrase (message.Sender));
 349                         foreach (GMime.InternetAddress ia in addrs) {
 350                                 if (this.folder_name != "Sent"  && ia.AddressType != GMime.InternetAddressType.Group)
 351                                         indexable.AddProperty (Property.NewUnsearched ("fixme:gotFrom", ia.Addr));
 352                         }
 353                         addrs.Dispose ();
 354
 355                         if (this.folder_name == "Sent")
 356                                 indexable.AddProperty (Property.NewFlag ("fixme:isSent"));
 357
 358                         Property flag_prop = Property.NewUnsearched ("fixme:flags", flags);
 359                         flag_prop.IsMutable = true;
 360                         indexable.AddProperty (flag_prop);
 361
 362                         if (CheckFlags (flags, Camel.CamelFlags.Answered))
 363                                 indexable.AddProperty (Property.NewFlag ("fixme:isAnswered"));
 364
 365                         if (CheckFlags (flags, Camel.CamelFlags.Deleted))
 366                                 indexable.AddProperty (Property.NewFlag ("fixme:isDeleted"));
 367
 368                         if (CheckFlags (flags, Camel.CamelFlags.Draft))
 369                                 indexable.AddProperty (Property.NewFlag ("fixme:isDraft"));
 370
 371                         if (CheckFlags (flags, Camel.CamelFlags.Flagged))
 372                                 indexable.AddProperty (Property.NewFlag ("fixme:isFlagged"));
 373
 374                         if (CheckFlags (flags, Camel.CamelFlags.Seen))
 375                                 indexable.AddProperty (Property.NewFlag ("fixme:isSeen"));
 376
 377                         if (CheckFlags (flags, Camel.CamelFlags.AnsweredAll))
 378                                 indexable.AddProperty (Property.NewFlag ("fixme:isAnsweredAll"));
 379
 380                         indexable.SetBinaryStream (message.Stream);
 381
 382                         return indexable;
 383                 }
 384
 385                 public override void Checkpoint ()
 386                 {
 387                         long offset = -1;
 388
 389                         if (this.mbox_parser != null) {
 390                                 offset = this.mbox_parser.FromOffset;
 391
 392                                 if (offset >= 0)
 393                                         this.MboxLastOffset = offset;
 394                         }
 395
 396                         string progress = "";
 397                         if (this.file_size > 0 && offset > 0) {
 398                                 progress = String.Format (" ({0}/{1} bytes {2:###.0}%)",
 399                                                           offset, this.file_size, 100.0 * offset / this.file_size);
 400
 401                                 this.progress_percent = 100.0 * offset / this.file_size;
 402                         }
 403
 404                         Logger.Log.Debug ("{0}: indexed {1} messages{2}",
 405                                           this.folder_name, this.indexed_count, progress);
 406                 }
 407
 408                 public override string GetTarget ()
 409                 {
 410                         return "mbox-file:" + mbox_info.FullName;
 411                 }
 412
 413                 protected override FileInfo CrawlFile {
 414                         get { return this.mbox_info; }
 415                 }
 416         }
 417
 418         public class EvolutionMailIndexableGeneratorImap : EvolutionMailIndexableGenerator {
 419                 private enum ImapBackendType {
 420                         Imap,
 421                         Imap4
 422                 };
 423
 424                 private FileInfo summary_info;
 425                 private string imap_name;
 426                 private ImapBackendType backend_type;
 427                 private Camel.Summary summary;
 428                 private IEnumerator summary_enumerator;
 429                 private ICollection accounts;
 430                 private string folder_cache_name;
 431                 private Hashtable mapping;
 432                 private ArrayList deleted_list;
 433                 private bool delete_mode;
 434                 private int delete_count;
 435
 436                 public EvolutionMailIndexableGeneratorImap (EvolutionMailQueryable queryable, FileInfo summary_info) : base (queryable)
 437                 {
 438                         this.summary_info = summary_info;
 439                 }
 440
 441                 protected override string GetFolderName (FileSystemInfo info)
 442                 {
 443                         DirectoryInfo dir_info = (DirectoryInfo) info;
 444                         string folder_name = "";
 445
 446                         while (dir_info != null) {
 447                                 folder_name = Path.Combine (dir_info.Name, folder_name);
 448
 449                                 dir_info = dir_info.Parent;
 450
 451                                 if (dir_info.Name != "subfolders")
 452                                         break;
 453                                 else
 454                                         dir_info = dir_info.Parent;
 455                         }
 456
 457                         return folder_name;
 458                 }
 459
 460                 protected override bool Setup ()
 461                 {
 462                         string dir_name = summary_info.DirectoryName;
 463                         int imap_start_idx;
 464
 465                         int idx = dir_name.IndexOf (".evolution/mail/imap4/");
 466
 467                         if (idx >= 0) {
 468                                 this.backend_type = ImapBackendType.Imap4;
 469                                 imap_start_idx = idx + 22;
 470                         } else {
 471                                 this.backend_type = ImapBackendType.Imap;
 472                                 imap_start_idx = dir_name.IndexOf (".evolution/mail/imap/") + 21;
 473                         }
 474
 475                         string imap_start = dir_name.Substring (imap_start_idx);
 476                         this.imap_name = imap_start.Substring (0, imap_start.IndexOf ('/'));
 477
 478                         try {
 479                                 this.accounts = (ICollection) GConfThreadHelper.Get ("/apps/evolution/mail/accounts");
 480                         } catch (Exception ex) {
 481                                 Logger.Log.Warn ("Caught exception in Setup(): " + ex.Message);
 482                                 Logger.Log.Warn ("There are no configured evolution accounts, ignoring {0}", this.imap_name);
 483                                 return false;
 484                         }
 485
 486                         // This should only happen if we shut down while waiting for the GConf results to come back.
 487                         if (this.accounts == null)
 488                                 return false;
 489
 490                         foreach (string xml in this.accounts) {
 491                                 XmlDocument xmlDoc = new XmlDocument ();
 492
 493                                 xmlDoc.LoadXml (xml);
 494
 495                                 XmlNode account = xmlDoc.SelectSingleNode ("//account");
 496
 497                                 if (account == null)
 498                                         continue;
 499
 500                                 string uid = null;
 501
 502                                 foreach (XmlAttribute attr in account.Attributes) {
 503                                         if (attr.Name == "uid") {
 504                                                 uid = attr.InnerText;
 505                                                 break;
 506                                         }
 507                                 }
 508
 509                                 if (uid == null)
 510                                         continue;
 511
 512                                 XmlNode imap_url_node = xmlDoc.SelectSingleNode ("//source/url");
 513
 514                                 if (imap_url_node == null)
 515                                         continue;
 516
 517                                 string imap_url = imap_url_node.InnerText;
 518                                 // If there is a semicolon in the username part of the URL, it
 519                                 // indicates that there's an auth scheme there.  We don't care
 520                                 // about that, so remove it.
 521                                 int user_end = imap_url.IndexOf ('@');
 522                                 int semicolon = imap_url.IndexOf (';', 0, user_end + 1);
 523
 524                                 if (semicolon != -1)
 525                                         imap_url = imap_url.Substring (0, semicolon) + imap_url.Substring (user_end);
 526
 527                                 // Escape backslashes, which frequently appear when using IMAP against Exchange servers
 528                                 this.imap_name = this.imap_name.Replace ("\\", "%5c");
 529
 530                                 // Escape out additional @s in the name.  I hate the class libs so much.
 531                                 int lastIdx = this.imap_name.LastIndexOf ('@');
 532                                 if (this.imap_name.IndexOf ('@') != lastIdx) {
 533                                         string toEscape = this.imap_name.Substring (0, lastIdx);
 534                                         this.imap_name = toEscape.Replace ("@", "%40") + this.imap_name.Substring (lastIdx);
 535                                 }
 536
 537                                 string backend_url_prefix;
 538                                 if (this.backend_type == ImapBackendType.Imap)
 539                                         backend_url_prefix = "imap";
 540                                 else
 541                                         backend_url_prefix = "imap4";
 542
 543                                 if (imap_url.StartsWith (backend_url_prefix + "://" + this.imap_name + "/")) {
 544                                         this.account_name = uid;
 545                                         break;
 546                                 }
 547                         }
 548
 549                         if (this.account_name == null) {
 550                                 Logger.Log.Info ("Unable to determine account name for {0}", this.imap_name);
 551                                 return false;
 552                         }
 553
 554                         // Need to check the directory on disk to see if it's a junk/spam folder,
 555                         // since the folder name will be "foo/spam" and not match the check below.
 556                         DirectoryInfo dir_info = new DirectoryInfo (dir_name);
 557                         if (this.IsSpamFolder (dir_info.Name))
 558                                 return false;
 559
 560                         // Check if the folder is listed in the configuration as to be excluded from indexing
 561                         if (this.IgnoreFolder (dir_info.FullName))
 562                                 return false;
 563
 564                         this.folder_name = GetFolderName (new DirectoryInfo (dir_name));
 565
 566                         return true;
 567                 }
 568
 569                 private string FolderCacheName {
 570                         get {
 571                                 if (this.account_name == null || this.folder_name == null)
 572                                         return null;
 573
 574                                 if (this.folder_cache_name == null)
 575                                         this.folder_cache_name = "status-" + this.account_name + "-" + this.folder_name.Replace ('/', '-');
 576
 577                                 return this.folder_cache_name;
 578                         }
 579                 }
 580
 581                 private bool LoadCache ()
 582                 {
 583                         Stream cacheStream;
 584                         BinaryFormatter formatter;
 585
 586                         if (this.FolderCacheName == null) {
 587                                 this.mapping = new Hashtable ();
 588                                 return false;
 589                         }
 590
 591                         try {
 592                                 cacheStream = this.queryable.ReadDataStream (this.FolderCacheName);
 593                                 formatter = new BinaryFormatter ();
 594                                 this.mapping = formatter.Deserialize (cacheStream) as Hashtable;
 595                                 cacheStream.Close ();
 596                                 Logger.Log.Debug ("Successfully loaded previous crawled data from disk: {0}", this.FolderCacheName);
 597
 598                                 return true;
 599                         } catch {
 600                                 this.mapping = new Hashtable ();
 601
 602                                 return false;
 603                         }
 604                 }
 605
 606                 private void SaveCache ()
 607                 {
 608                         Stream cacheStream;
 609                         BinaryFormatter formatter;
 610
 611                         if (this.FolderCacheName == null)
 612                                 return;
 613
 614                         cacheStream = this.queryable.WriteDataStream (this.FolderCacheName);
 615                         formatter = new BinaryFormatter ();
 616                         formatter.Serialize (cacheStream, mapping);
 617                         cacheStream.Close ();
 618                 }
 619
 620                 public override bool HasNextIndexable ()
 621                 {
 622                         if (this.account_name == null) {
 623                                 if (!Setup ()) {
 624                                         this.queryable.RemoveGenerator (this);
 625                                         return false;
 626                                 }
 627                         }
 628
 629                         if (this.mapping == null) {
 630                                 bool cache_loaded = this.LoadCache ();
 631
 632                                 this.deleted_list = new ArrayList (this.mapping.Keys);
 633                                 this.deleted_list.Sort ();
 634                                 Logger.Log.Debug ("Deleted list starting at {0} for {1}", this.deleted_list.Count, this.folder_name);
 635
 636                                 // Check to see if we even need to bother walking the summary
 637                                 if (cache_loaded && this.queryable.FileAttributesStore.IsUpToDate (this.CrawlFile.FullName)) {
 638                                         Logger.Log.Debug ("{0}: summary has not been updated; crawl unncessary", this.folder_name);
 639                                         this.queryable.RemoveGenerator (this);
 640                                         return false;
 641                                 }
 642                         }
 643
 644                         if (this.summary == null) {
 645                                 try {
 646                                         if (this.backend_type == ImapBackendType.Imap)
 647                                                 this.summary = Camel.Summary.LoadImapSummary (this.summary_info.FullName);
 648                                         else
 649                                                 this.summary = Camel.Summary.LoadImap4Summary (this.summary_info.FullName);
 650                                 } catch (Exception e) {
 651                                         Logger.Log.Warn (e, "Unable to index {0}:", this.folder_name);
 652                                         this.queryable.RemoveGenerator (this);
 653                                         return false;
 654                                 }
 655                         }
 656
 657                         if (this.summary_enumerator == null)
 658                                 this.summary_enumerator = this.summary.GetEnumerator ();
 659
 660                         if (this.summary_enumerator.MoveNext ())
 661                                 return true;
 662
 663                         this.delete_mode = true;
 664
 665                         if (this.deleted_list.Count > 0)
 666                                 return true;
 667
 668                         string progress = "";
 669                         if (this.count > 0 && this.summary.header.count > 0) {
 670                                 progress = String.Format ("({0}/{1} {2:###.0}%)",
 671                                                           this.count,
 672                                                           this.summary.header.count,
 673                                                           100.0 * this.count / this.summary.header.count);
 674                         }
 675
 676                         Logger.Log.Debug ("{0}: Finished indexing {1} messages {2}, {3} messages deleted", this.folder_name, this.indexed_count, progress, this.delete_count);
 677
 678                         this.SaveCache ();
 679                         this.CrawlFinished ();
 680
 681                         return false;
 682                 }
 683
 684                 // Kind of nasty, but we need the function.
 685                 [System.Runtime.InteropServices.DllImport("libglib-2.0.so.0")]
 686                 static extern int g_str_hash (string str);
 687
 688                 // Stolen from deep within e-d-s's camel-data-cache.c  Very evil.
 689                 private const int CAMEL_DATA_CACHE_MASK = ((1 << 6) - 1);
 690
 691                 public override Indexable GetNextIndexable ()
 692                 {
 693                         Indexable indexable = null;
 694
 695                         // No more new messages to index, so start on the removals.
 696                         if (this.delete_mode) {
 697                                 string uid = (string) this.deleted_list [0];
 698                                 Uri uri = EvolutionMailQueryable.EmailUri (this.account_name, this.folder_name, uid);
 699
 700                                 indexable = new Indexable (IndexableType.Remove, uri);
 701
 702                                 this.deleted_list.RemoveAt (0);
 703                                 this.mapping.Remove (uid);
 704
 705                                 this.delete_count++;
 706
 707                                 return indexable;
 708                         }
 709
 710                         Camel.MessageInfo mi = (Camel.MessageInfo) this.summary_enumerator.Current;
 711
 712                         ++this.count;
 713
 714                         if (Debug) {
 715                                 Logger.Log.Debug ("Constructed message {0} with uid {1}, flags {2}.",
 716                                                   this.count, mi.uid, mi.flags);
 717                         }
 718
 719                         // Try to load the cached message data off disk
 720                         object flags = this.mapping[mi.uid];
 721
 722                         if (flags == null) {
 723                                 // New, previously unseen message
 724                                 string msg_file;
 725
 726                                 if (this.backend_type == ImapBackendType.Imap)
 727                                         msg_file = Path.Combine (summary_info.DirectoryName, mi.uid + ".");
 728                                 else {
 729                                         // This is taken from e-d-s's camel-data-cache.c.  No doubt
 730                                         // NotZed would scream bloody murder if he saw this here.
 731                                         int hash = (g_str_hash (mi.uid) >> 5) & CAMEL_DATA_CACHE_MASK;
 732                                         string cache_path = String.Format ("cache/{0:x}/{1}", hash, mi.uid);
 733                                         msg_file = Path.Combine (summary_info.DirectoryName, cache_path);
 734                                 }
 735
 736                                 indexable = this.CamelMessageToIndexable (mi, msg_file);
 737
 738                                 if (Debug)
 739                                         Logger.Log.Debug ("Unseen message, indexable {0} null", indexable == null ? "" : "not");
 740
 741                                 this.mapping[mi.uid] = mi.flags;
 742
 743                                 if (indexable != null)
 744                                         ++this.indexed_count;
 745                         } else if ((uint) flags != mi.flags) {
 746                                 // Previously seen message, but flags have changed.
 747                                 Uri uri = CamelMessageUri (mi);
 748                                 indexable = new Indexable (uri);
 749                                 indexable.Type = IndexableType.PropertyChange;
 750
 751                                 Property flag_prop = Property.NewUnsearched ("fixme:flags", mi.flags);
 752                                 flag_prop.IsMutable = true;
 753                                 indexable.AddProperty (flag_prop);
 754
 755                                 if (Debug)
 756                                         Logger.Log.Debug ("Previously seen message, flags changed: {0} -> {1}", flags, mi.flags);
 757
 758                                 ++this.indexed_count;
 759                         } else {
 760                                 if (Debug)
 761                                         Logger.Log.Debug ("Previously seen message, unchanged.");
 762                         }
 763
 764                         if (flags != null)
 765                                 this.deleted_list.Remove (mi.uid);
 766
 767                         return indexable;
 768                 }
 769
 770                 private Uri CamelMessageUri (Camel.MessageInfo message_info)
 771                 {
 772                         return EvolutionMailQueryable.EmailUri (this.account_name, this.folder_name, message_info.uid);
 773                 }
 774
 775                 private Indexable CamelMessageToIndexable (Camel.MessageInfo messageInfo, string msg_file)
 776                 {
 777                         // Don't index messages flagged as junk
 778                         if (messageInfo.IsJunk)
 779                                 return null;
 780
 781                         // Many properties will be set by the filter when
 782                         // processing the cached data, if it's there.  So
 783                         // don't set a number of properties in that case.
 784                         bool have_content = File.Exists (msg_file);
 785
 786                         Uri uri = CamelMessageUri (messageInfo);
 787                         Indexable indexable = new Indexable (uri);
 788
 789                         indexable.Timestamp = messageInfo.SentDate;
 790                         indexable.MimeType = "message/rfc822";
 791                         indexable.HitType = "MailMessage";
 792
 793                         indexable.AddProperty (Property.NewUnsearched ("fixme:account",  this.imap_name));
 794                         indexable.AddProperty (Property.NewUnsearched ("fixme:folder",   this.folder_name));
 795                         indexable.AddProperty (Property.NewUnsearched ("fixme:client", "evolution"));
 796
 797                         if (!have_content) {
 798                                 indexable.AddProperty (Property.New ("dc:title", GMime.Utils.HeaderDecodePhrase (messageInfo.subject)));
 799                                 indexable.AddProperty (Property.NewDate ("fixme:date", messageInfo.SentDate));
 800                         }
 801
 802                         GMime.InternetAddressList addrs;
 803                         addrs = GMime.InternetAddressList.ParseString (messageInfo.to);
 804                         foreach (GMime.InternetAddress ia in addrs) {
 805                                 if (!have_content) {
 806                                         indexable.AddProperty (Property.NewUnsearched ("fixme:to", ia.ToString (false)));
 807                                         if (ia.AddressType != GMime.InternetAddressType.Group)
 808                                                 indexable.AddProperty (Property.New ("fixme:to_address", ia.Addr));
 809
 810                                         indexable.AddProperty (Property.New ("fixme:to_name", ia.Name));
 811                                 }
 812
 813                                 if (this.folder_name == "Sent" && ia.AddressType != GMime.InternetAddressType.Group)
 814                                         indexable.AddProperty (Property.NewUnsearched ("fixme:sentTo", ia.Addr));
 815                         }
 816                         addrs.Dispose ();
 817
 818                         addrs = GMime.InternetAddressList.ParseString (messageInfo.cc);
 819                         foreach (GMime.InternetAddress ia in addrs) {
 820                                 if (!have_content) {
 821                                         indexable.AddProperty (Property.NewUnsearched ("fixme:cc", ia.ToString (false)));
 822                                         if (ia.AddressType != GMime.InternetAddressType.Group)
 823                                                 indexable.AddProperty (Property.New ("fixme:cc_address", ia.Addr));
 824
 825                                         indexable.AddProperty (Property.New ("fixme:cc_name", ia.Name));
 826                                 }
 827
 828                                 if (this.folder_name == "Sent" && ia.AddressType != GMime.InternetAddressType.Group)
 829                                         indexable.AddProperty (Property.NewUnsearched ("fixme:sentTo", ia.Addr));
 830                         }
 831                         addrs.Dispose ();
 832
 833                         addrs = GMime.InternetAddressList.ParseString (messageInfo.from);
 834                         foreach (GMime.InternetAddress ia in addrs) {
 835                                 if (!have_content) {
 836                                         indexable.AddProperty (Property.NewUnsearched ("fixme:from", ia.ToString (false)));
 837                                         if (ia.AddressType != GMime.InternetAddressType.Group)
 838                                                 indexable.AddProperty (Property.New ("fixme:from_address", ia.Addr));
 839
 840                                         indexable.AddProperty (Property.New ("fixme:from_name", ia.Name));
 841                                 }
 842
 843                                 if (this.folder_name != "Sent" && ia.AddressType != GMime.InternetAddressType.Group)
 844                                         indexable.AddProperty (Property.NewUnsearched ("fixme:gotFrom", ia.Addr));
 845                         }
 846                         addrs.Dispose ();
 847
 848                         indexable.AddProperty (Property.NewKeyword ("fixme:mlist", messageInfo.mlist));
 849
 850                         Property flag_prop = Property.NewUnsearched ("fixme:flags", messageInfo.flags);
 851                         flag_prop.IsMutable = true;
 852                         indexable.AddProperty (flag_prop);
 853
 854                         if (this.folder_name == "Sent")
 855                                 indexable.AddProperty (Property.NewFlag ("fixme:isSent"));
 856
 857                         if (messageInfo.IsAnswered)
 858                                 indexable.AddProperty (Property.NewFlag ("fixme:isAnswered"));
 859
 860                         if (messageInfo.IsDeleted)
 861                                 indexable.AddProperty (Property.NewFlag ("fixme:isDeleted"));
 862
 863                         if (messageInfo.IsDraft)
 864                                 indexable.AddProperty (Property.NewFlag ("fixme:isDraft"));
 865
 866                         if (messageInfo.IsFlagged)
 867                                 indexable.AddProperty (Property.NewFlag ("fixme:isFlagged"));
 868
 869                         if (messageInfo.IsSeen)
 870                                 indexable.AddProperty (Property.NewFlag ("fixme:isSeen"));
 871
 872                         if (messageInfo.HasAttachments && !have_content)
 873                                 indexable.AddProperty (Property.NewFlag ("fixme:hasAttachments"));
 874
 875                         if (messageInfo.IsAnsweredAll)
 876                                 indexable.AddProperty (Property.NewFlag ("fixme:isAnsweredAll"));
 877
 878                         if (have_content)
 879                                 indexable.ContentUri = UriFu.PathToFileUri (msg_file);
 880                         else
 881                                 indexable.NoContent = true;
 882
 883                         return indexable;
 884                 }
 885
 886                 public override void Checkpoint ()
 887                 {
 888                         if (this.summary != null) {
 889
 890                                 string progress = "";
 891                                 if (this.count > 0 && this.summary.header.count > 0) {
 892                                         progress = String.Format (" ({0}/{1} {2:###.0}%)",
 893                                                                   this.count,
 894                                                                   this.summary.header.count,
 895                                                                   100.0 * this.count / this.summary.header.count);
 896
 897                                         this.progress_percent = 100.0 * this.count / this.summary.header.count;
 898
 899                                 }
 900
 901                                 Logger.Log.Debug ("{0}: indexed {1} messages{2}",
 902                                                   this.folder_name, this.indexed_count, progress);
 903                         }
 904
 905                         this.SaveCache ();
 906                 }
 907
 908                 public override string GetTarget ()
 909                 {
 910                         return "summary-file:" + summary_info.FullName;
 911                 }
 912
 913                 protected override FileInfo CrawlFile {
 914                         get { return this.summary_info; }
 915                 }
 916         }
 917 }