git.roojs.org Git - app.mailtrimmer/blob - src/strip.vala

   1 /**
   2
   3  ** check left to do:
   4   - range scans on maildir
   5   - see how replacing the links works in the resulting email via thunderbird etc..
   6   - some checksum issues (see dupelicates?? suspect 0byte issues?)  -- seems ok now?
   7
   8
   9
  10
  11   needs to scan 2 things
  12   a) our mailfort email database
  13        point it at the top directory, containing YEAR/MONTH/DAY.... directories.
  14        scan each file (over a year old...)
  15        extract out the attachment, and replace with HTML
  16        DATABASE? - mysql or sqlite? -
  17            filesize / name / date / checksum / mimetype -- into mailfort should be OK.
  18   b) the imap user emails
  19            loop through user's directories
  20            check age of email .. over 1 years..
  21            ?? how to prevent 'repeat' scanning of emails?
  22               ??? hidden '.' files containing last scan date?
  23
  24            check if file exists in our DB.. - replace the link...
  25            otherwise generate a file. + add to DB...
  26
  27    c) retreival system
  28      -> URL -> get file
  29    d) redirect system.
  30      -> URL -> redirect to correct server
  31
  32
  33 More notes on our Mailfort DB sync:
  34 * some of these attachments are already in the database...
  35  - so we need to update the DB..
  36  - probably worth putting the code in a stored procedure..
  37
  38  -- key scenarios
  39    * first scan (and extract)
  40    * rescan (as I messed up the first time - fix the DB...)
  41    * email scan - attachments might not have related messages.
  42
  43
  44  - {id} attachment_init(
  45                 {exim_msg_id}
  46                 {chksum}
  47                 {filename),
  48         )
  49         // creates or returns id (can look for existing messages?
  50         // can do a merge?? - copy 'old' record data into 'new'....  "prefer checksummed"
  51
  52         attachment_update(
  53                 {id}
  54                 {exim_msg_id}
  55                 {mailfort_msg_sig}
  56                 {file_size}
  57                 {created} // message date..
  58                 {chksum}
  59                 {filename),
  60         {mime_type}
  61         )
  62         attachment_update_store(
  63                         {id}
  64                         {stored_filename}
  65         )
  66
  67
  68 */
  69
  70 // valac --pkg gmime --vapi
  71 /*
  72
  73 // http://www.fromdual.com/mysql-vala-program-example << check mysql if this does not work.
  74
  75  valac  -g --vapidir=. --thread  strip.vala   --vapidir=../vapi \
  76      --pkg glib-2.0 --pkg mysql --pkg gio-2.0 --pkg posix --pkg gmime-2.6 \
  77       --Xcc=-lmysqlclient  -v \
  78        -o /tmp/strip
  79 */
  80
  81 public class StripApplication : GLib.Application {
  82
  83         public static string? opt_path = null;
  84         public static string? opt_file = null;
  85         public static string? opt_target_path = null;
  86         public static string? opt_db_host = "127.0.0.1";
  87         public static string? opt_db_name = null;
  88         public static string? opt_db_user = null;
  89         public static string? opt_db_pass = null;
  90
  91         public static int    opt_limit = -1;
  92
  93         public static int    opt_age_newest = 1;
  94         public static int    opt_age_oldest = 6;
  95
  96
  97         public static bool      opt_is_extracting = false;
  98         public static bool      opt_is_replacing = false;
  99         public static bool      opt_scan_maildir  = false;
 100         public static bool      opt_scan_mailfort  = false;
 101         public static bool              opt_dump = false;
 102         public static bool              opt_debug = false;
 103
 104         public static bool opt_debug_sql = false;
 105         public static string? opt_replace_link = null;
 106
 107
 108         public const GLib.OptionEntry[] options = {
 109
 110                 { "debug", 0, 0, OptionArg.NONE, ref opt_debug, "show debug messages for components", null },
 111                 { "debug-sql", 0, 0, OptionArg.NONE, ref opt_debug_sql, "debug the SQL statements", null },
 112
 113                 { "path", 0, 0, OptionArg.STRING, ref opt_path, "Directory where email to be parsed is", null },
 114                 { "file", 0, 0, OptionArg.STRING, ref opt_file, "A specific file to be parsed", null },
 115
 116                 { "target-path", 0, 0, OptionArg.STRING, ref opt_target_path, "Directory where attachments are to be put", null },
 117
 118                 { "link", 0, 0, OptionArg.STRING, ref opt_replace_link, "url for the replement link: eg. http://www.mysite.com/xxxx/%s", null },
 119
 120                 { "host", 0, 0, OptionArg.STRING, ref opt_db_host, "Mysql host (default localhost)", null },
 121                 { "name", 0, 0, OptionArg.STRING, ref opt_db_name, "Mysql database name REQUIRED", null },
 122                 { "user", 0, 0, OptionArg.STRING, ref opt_db_user, "Mysql database user REQUIRED", null },
 123                 { "pass", 0, 0, OptionArg.STRING, ref opt_db_pass, "Mysql database password (default empty)", null },
 124
 125                 { "extract", 0, 0, OptionArg.NONE, ref opt_is_extracting, "Should attachments be extracted (default NO)", null },
 126                 { "replace", 0, 0, OptionArg.NONE, ref opt_is_replacing, "Should attachments be replaced (default NO)", null },
 127                 { "dump", 0, 0, OptionArg.NONE, ref opt_dump, "Print the replaced mail contents to stdout", null },
 128
 129                 { "limit", 0, 0, OptionArg.INT, ref opt_limit, "stop after X number of messages with attachments have been processed", null },
 130                 { "newest", 0, 0, OptionArg.INT, ref opt_age_newest, "do not replace messages newer that X months (default is 1 months)", null },
 131                 { "oldest", 0, 0, OptionArg.INT, ref opt_age_oldest, "do not replace messages older than X (default is 6 months)", null },
 132
 133                 { "scan-maildir", 0, 0, OptionArg.NONE, ref opt_scan_maildir, "scan an maildir tree", null },
 134                 { "scan-mailfort", 0, 0, OptionArg.NONE, ref opt_scan_mailfort, "scan a mailfort tree", null },
 135                 { "scan-mailfort-only", 0, 0, OptionArg.STRING, ref opt_scan_mailfort_only, "scan a mailfort Year/month eg. 2010/08", null },
 136                 { null }
 137         };
 138     public StripApplication( string[] args )
 139     {
 140                  Object(
 141             application_id: "org.roojs.mailstripper",
 142             flags: ApplicationFlags.FLAGS_NONE
 143          );
 144
 145
 146          var opt_context =  new GLib.OptionContext ("Mail Stripper");
 147
 148          try {
 149
 150             opt_context.set_help_enabled (true);
 151             opt_context.add_main_entries (options, null);
 152             opt_context.parse ( ref  args);
 153             //opt_detach = !optx_no_detach;
 154
 155
 156
 157              // options that have to be set.. bee or hive... (or stop all)
 158             if ((!opt_scan_mailfort && !opt_scan_maildir) || (opt_scan_mailfort && opt_scan_maildir))  {
 159                stdout.printf ("You must specify the type of directory tree to scan - either imap or mailfort\n%s",
 160                    opt_context.get_help(true, null));
 161                GLib.Process.exit(Posix.EXIT_FAILURE);
 162             }
 163
 164                          if ((opt_db_name == null || opt_db_name.length < 1 || opt_db_user == null || opt_db_user.length < 1))  {
 165                stdout.printf ("You must specify the database name / user \n%s",
 166                    opt_context.get_help(true, null));
 167                GLib.Process.exit(Posix.EXIT_FAILURE);
 168             }
 169                          if ((opt_path == null || opt_path.length < 1)   )  {
 170                stdout.printf ("You must specify the scan start path\n%s",
 171                    opt_context.get_help(true, null));
 172                GLib.Process.exit(Posix.EXIT_FAILURE);
 173             }
 174                         if (opt_replace_link == null || (opt_replace_link.length < 1))  {
 175                stdout.printf ("You must specify the link to use in the replacement \n%s",
 176                    opt_context.get_help(true, null));
 177                GLib.Process.exit(Posix.EXIT_FAILURE);
 178             }
 179             if ((opt_is_replacing || opt_is_extracting ) && (opt_target_path == null || opt_target_path.length < 1)) {
 180                       stdout.printf ("You must specify a target path to put attachments\n%s",
 181                    opt_context.get_help(true, null));
 182                GLib.Process.exit(Posix.EXIT_FAILURE);
 183             }
 184
 185
 186          } catch (GLib.OptionError e) {
 187             stdout.printf ("error: %s\n", e.message);
 188             stdout.printf ("Run '%s --help' to see a full list of available command line options.\n%s",
 189                       args[0], opt_context.get_help(true, null));
 190             GLib.Process.exit(Posix.EXIT_FAILURE);
 191          }
 192         }
 193
 194     public static int main(string[] args)
 195     {
 196
 197                 var application = new StripApplication(  args);
 198
 199                 GLib.Log.set_always_fatal(LogLevelFlags.LEVEL_ERROR | LogLevelFlags.LEVEL_CRITICAL);
 200
 201            if (opt_debug || opt_debug_sql) {
 202                         GLib.Log.set_handler(null,
 203                         GLib.LogLevelFlags.LEVEL_DEBUG | GLib.LogLevelFlags.LEVEL_WARNING | GLib.LogLevelFlags.LEVEL_INFO,
 204                         (dom, lvl, msg) => {
 205                                         print("%s\n", msg);
 206                                 }
 207                         );
 208                 }
 209
 210         GMime.init(0);
 211                 if (StripApplication.opt_is_replacing) {
 212                         StripApplication.opt_is_extracting = true;
 213                 }
 214
 215                 GLib.debug("scanning folder: %s", opt_path );
 216
 217                 var strip = new Strip( opt_path );
 218
 219
 220                 strip.mysql  = new Mysql.Database();
 221                 if (!strip.mysql.real_connect(
 222                                 opt_db_host,
 223                                 opt_db_user ,
 224                                 opt_db_pass == null ? "" : opt_db_pass, //passwd
 225                                 opt_db_name, //DB
 226                                 3306, // not changable...?
 227                                 null
 228                         )
 229                 ) {
 230                         stdout.printf("ERROR %u: Connection failed: %s\n",
 231                                 strip.mysql.errno(), strip.mysql.error()
 232                         );
 233
 234                         return 1;
 235                 }
 236         if (opt_file != null) {
 237                 strip.base_dir = opt_path;
 238                 strip.scan_file( GLib.Path.get_dirname(opt_file),  GLib.Path.get_basename(opt_file));
 239                 return 0;
 240         }
 241
 242                 strip.scan_dir(opt_path, "");
 243
 244
 245
 246         return 0;
 247     }
 248 }
 249
 250 public class Strip : GLib.Object {
 251
 252
 253
 254         public string base_dir = "";
 255
 256         public Mysql.Database mysql;
 257
 258         int processed = 0;
 259
 260     uint64 used_space_before = 0;
 261     uint64 used_space_after = 0;
 262
 263
 264     public Strip(string base_dir)
 265     {
 266         this.base_dir = base_dir;
 267     }
 268
 269     public void handle_part(GMime.Object parent, GMime.Object mime_obj)
 270     {
 271                 if (mime_obj is GMime.Part) {
 272                    var  p = (GMime.Part)mime_obj;
 273                         var ct = p.get_content_type();
 274                         var cd = p.get_content_disposition();
 275
 276                         var sid = p.get_header("X-strip-id");
 277                     if (sid != null && sid.length > 0) {
 278                         this.update_attachment_db(p);
 279                             GLib.debug("Skip attachment replace - it's already been done");
 280                         return;
 281                         }
 282
 283                         if (cd == null || cd.get_disposition().down() != "attachment") {
 284                                 return;
 285                         }
 286                         if (ct.get_media_type() == "text") {
 287                                 return;
 288                         }
 289                         if (ct.to_string() == "application/pgp-encrypted") {
 290                                 return;
 291                         }
 292                         if (ct.to_string() == "application/pgp-keys") {
 293                                 return;
 294                         }
 295                         if (p.get_filename() == null) {
 296                                 return;
 297                         }
 298                          // print("got part %s\n", ct.to_string());
 299                          if (parent is GMime.Multipart) {
 300
 301                                 this.replace_attachment(((GMime.Multipart)parent), p);
 302                                 // remove it !?
 303
 304                           }
 305
 306
 307                         return;
 308                 }
 309                 if (mime_obj is GMime.Multipart) {
 310
 311
 312                         var  mp = (GMime.Multipart)mime_obj;
 313                         //var ct = mp.get_content_type();
 314
 315                         //print("got multi-part %s\n", ct.to_string());
 316                         for (var i = 0; i< mp.get_count(); i++) {
 317                           var mo = mp.get_part(i);
 318                           this.handle_part(mime_obj,mo);
 319                         }
 320                    // ((GMime.Multipart)mime_obj).foreach((sub_obj) => {
 321                    //     Strip.handle_part(sub_obj);
 322                 //
 323                    // });
 324
 325
 326                         return;
 327                 }
 328
 329                 if (mime_obj is GMime.MessagePart) {
 330                         var msg = ((GMime.MessagePart)mime_obj).get_message();
 331                         msg.foreach((subobj) => {
 332                          this.handle_part(msg,subobj);
 333                     });
 334
 335                         //print("got message-part\n");
 336                         return;
 337                 }
 338
 339                 if (mime_obj is GMime.Message) {
 340                         var mp = ((GMime.Message) mime_obj).get_mime_part();
 341
 342                         if (!(mp is GMime.Multipart)) {
 343                                 //GLib.debug("get mimepart does not return a Multipart?");
 344                                 return;
 345                         }
 346
 347                         var mpc = ((GMime.Multipart)mp).get_count();
 348
 349                         //GLib.debug("Message has %d parts", mpc);
 350                         for (var i =0 ; i < mpc; i++) {
 351                                 //GLib.debug("Getting part %d", i);
 352                                 var submime_obj = ((GMime.Multipart)mp).get_part(i);
 353                         this.handle_part(mp,submime_obj);
 354                     }
 355                         print("got message??\n");
 356                         return;
 357                 }
 358
 359                 print("got something else\n");
 360
 361
 362     }
 363     public void update_attachment_db(GMime.Part attachment)
 364     {
 365         // only called when we have an sid...
 366         var sid = attachment.get_header("X-strip-id");
 367         if (sid == null || sid.length < 1) {
 368                 GLib.debug("Strange - update attachment db called ?");
 369                 return;
 370         }
 371
 372         // initialize it with known data..
 373         // that should wipe out dupes.
 374         var matches = this.query("SELECT count(id)   FROM Attachment WHERE id = %d".printf(
 375                         int.parse(sid)));
 376
 377
 378                 if (matches == "0") {
 379                         // our old mailfort code deleted the crap out of old records...
 380                         // if this occurs we will need to create the record again..
 381                         this.fix_deleted_attachment_db(int.parse(sid),attachment);
 382                         return;
 383
 384                 }
 385
 386
 387         // initialize it with known data..
 388         // that should wipe out dupes.
 389         var filesize = this.query("SELECT filesize FROM Attachment WHERE id = %d".printf(
 390                         int.parse(sid)));
 391
 392                 if (filesize=="") {
 393                    GLib.error("Ignoring record id (missing in database) :%s", sid);
 394                    return;
 395                 }
 396                 if (int.parse(filesize) < 1) {
 397                 GLib.debug("Could not get filesize from id :%s = %s", sid,filesize);
 398                 Posix.exit(0);
 399                 return;
 400         }
 401
 402         var chksum = this.query("SELECT  checksum FROM Attachment WHERE id = %d".printf(
 403                         int.parse(sid)
 404                 ));
 405         var mime_filename = this.query("SELECT  mime_filename FROM Attachment WHERE id = %d".printf(
 406                         int.parse(sid)));
 407
 408         this.query("""
 409              SELECT
 410                  attachment_init(
 411                      '%s', '%s', '%s', %d
 412                  ) as id
 413
 414           """.printf(
 415                           this.mysql_escape(this.active_message_exim_id),
 416                           this.mysql_escape(chksum),
 417                           this.mysql_escape(mime_filename),
 418                           int.parse(filesize)
 419                 ));
 420         this.query("""
 421                  SELECT attachment_update(
 422                       %d, -- in_id INT(11),
 423                       '%s', -- in_mime_type varchar(255),
 424                       '%s', -- in_created DATETIME,
 425                       '%s' -- in_mailfort_sig varchar(64)
 426                  )
 427               """.printf(
 428                         int.parse(sid),
 429                         "", // this will be ignored..
 430                                 this.created_date,
 431                                 this.mysql_escape(this.active_message_x_mailfort_sig)
 432
 433               )
 434                 );
 435                 this.mysql.store_result();
 436
 437
 438
 439     }
 440
 441
 442     public void fix_deleted_attachment_db(int id, GMime.Part attachment)
 443     {
 444
 445         var filename = attachment.get_header("X-strip-content-name");
 446         var file_path  = attachment.get_header("X-strip-path");
 447         var fn =  StripApplication.opt_target_path + "/" + file_path;
 448
 449
 450                 if (!FileUtils.test (fn, FileTest.EXISTS)) {
 451                         GLib.debug("SKIP -- file does not exist");
 452                         return;
 453         }
 454
 455         var chksum = this.md5_file(fn);
 456                 var mime_type = attachment.get_header("X-strip-content-type");
 457
 458                 var fileinfo = File.new_for_path(fn)
 459                                         .query_info(GLib.FileAttribute.STANDARD_SIZE+","+GLib.FileAttribute.TIME_MODIFIED
 460                                                 ,GLib.FileQueryInfoFlags.NONE,null);
 461         var file_size = (int) fileinfo.get_size();
 462
 463
 464                 this.real_query(-1, """
 465
 466
 467                                 INSERT INTO Attachment  (
 468                                         id,
 469
 470                                     msgid ,
 471                                     queue_id ,
 472                                     mime_filename ,
 473                                     mime_type,
 474
 475                                     stored_filename ,
 476                                     mime_charset ,
 477                                     mime_cdisp ,
 478                                     mime_is_cover ,
 479
 480                                     mime_is_multi ,
 481                                     mime_is_mail,
 482                                     mime_size ,
 483                                     filesize,
 484
 485                                     checksum,
 486                                     created
 487
 488                                 ) VALUES (
 489                                         %d,  -- id
 490
 491                                     '%s' , -- msgid
 492                                     0,
 493                                     '%s'  , -- filename
 494                                     '%s',  -- mimetype
 495
 496                                     '%s', -- stored file anme
 497                                     '', -- charset
 498                                     'attachment',
 499                                     0,
 500                                     0,
 501                                     0,
 502                                     %d, -- size
 503                                     %d, -- size
 504
 505                                     '%s', -- checkum
 506                                         '%s' -- created:
 507                                 )
 508
 509
 510                       """.printf(
 511                                 id,
 512                                       this.mysql_escape(this.active_message_exim_id),
 513                                       this.mysql_escape(filename),
 514                                   this.mysql_escape(mime_type),
 515                                   this.mysql_escape(file_path),
 516                                         file_size,
 517                                       file_size,
 518                                       this.mysql_escape(chksum),
 519                                 this.created_date
 520                          ));
 521               // this is done to fix the queue_id or maillog_id ??
 522                  this.query("""
 523                  SELECT attachment_update(
 524                       %d, -- in_id INT(11),
 525                                                 '', -- mime type
 526                       '%s', -- in_created DATETIME,
 527                       '%s' -- in_mailfort_sig varchar(64)
 528                  )
 529               """.printf(
 530                                 id,
 531                                 this.created_date,
 532                                 this.mysql_escape(this.active_message_x_mailfort_sig)
 533
 534               )
 535                 );
 536                 // GLib.error("added attachment?");
 537     }
 538
 539
 540     public void replace_attachment(GMime.Multipart parent, GMime.Part attachment)
 541     {
 542         var sid = attachment.get_header("X-strip-id");
 543         if (sid != null && sid.length > 0) {
 544                 GLib.debug("Skip attachment replace - it's already been done");
 545                 return;
 546         }
 547
 548         var c = attachment.get_content_object();
 549
 550         var filename = attachment.get_filename().replace("/", "-").replace("\n", "").replace("\t", " ");
 551         var fn = GLib.Environment.get_tmp_dir() +
 552                         "/"+ this.active_name + "."+   GLib.Uri.escape_string(filename,"", false);
 553
 554             var outfile = new GMime.StreamFile.for_path(fn, "w");
 555             outfile.set_owner(true);
 556             var file_size = (int) c.write_to_stream(outfile);
 557             var chksum = this.md5_file(fn);
 558             outfile.flush();
 559             outfile = null;
 560
 561         if (file_size == 0) {
 562
 563                 GLib.debug("ERROR - file size of write to stream returned 0?");
 564                 Posix.unlink(fn);
 565                 return;
 566         }
 567
 568
 569
 570
 571         var mime_type= attachment.get_content_type().to_string();
 572         // at this point we have to do our database magic...
 573         //filesize / name / date / checksum / mimetype -- into mailfort should be OK.
 574
 575         var file_id = this.query("""
 576                 SELECT
 577
 578                 attachment_init(
 579                                 '%s', -- in_msgid VARCHAR(32),
 580                                 '%s', -- in_checksum VARCHAR(64),
 581                                 '%s', -- in_mime_filename varchar(255)
 582                                 %d -- filesize
 583                         ) as id
 584
 585           """.printf(
 586                         this.mysql_escape(this.active_message_exim_id),
 587                         chksum,
 588                         this.mysql_escape( GLib.Uri.escape_string(attachment.get_filename(),"", false) ), // what is thsi is invalid?
 589                          file_size)
 590                 );
 591
 592
 593                 if (file_id.length < 1) {
 594                         GLib.debug("ERROR - CALL to attachment_init failed");
 595                 Posix.unlink(fn);
 596                 return;
 597
 598                 }
 599
 600                 if (int.parse(file_id) < 1) {
 601                         GLib.debug("ERROR - CALL to attachment_init failed - returned 0?");
 602                 Posix.unlink(fn);
 603                 return;
 604
 605                 }
 606
 607
 608                 GLib.debug("fn = %s, m5=%s, id= %s", filename, mime_type, this.active_message_id);
 609                 this.query("""
 610
 611                         SELECT attachment_update(
 612                                 %d, -- in_id INT(11),
 613                                 '%s', -- in_mime_type varchar(255),
 614                                 '%s', -- in_created DATETIME,
 615                                 '%s' -- in_mailfort_sig varchar(64)
 616
 617                                 ) as result
 618       """.printf(
 619                 int.parse(file_id),
 620                         this.mysql_escape(mime_type),
 621                         this.created_date,
 622                         this.mysql_escape(this.active_message_x_mailfort_sig)
 623                 ));
 624                  this.mysql.store_result();
 625
 626
 627                 this.used_space_after += file_size;
 628
 629                 var target_fn = "";
 630
 631             if (StripApplication.opt_is_extracting) {
 632                         target_fn = StripApplication.opt_target_path + "/" + this.created_dir +"/"+ file_id  + "-" + GLib.Uri.escape_string(filename,"", false);
 633                 }
 634
 635             var stored =  "/" + this.created_dir +"/"+ file_id  + "-" + GLib.Uri.escape_string(filename,"", false);
 636                  this.query("""
 637
 638                         SELECT attachment_update_store(
 639                                 %d, -- in_id INT(11),
 640                                 '%s'  -- in_store_filename varchar(255),
 641
 642
 643                                 ) as result
 644       """.printf(
 645                 int.parse(file_id),
 646                          this.mysql_escape( stored)
 647                 ));
 648
 649         var rep = new GMime.Part.with_type("text","html");
 650         // we have to set up a redirect server - to redirect hpasite... to their internal service..
 651         rep.set_filename(filename);
 652         string txt = "<html><body>"+
 653             "<a href=\"" + StripApplication.opt_replace_link + "/" +
 654                         file_id + "/" + this.created_dir + "/"+chksum+"/"+ GLib.Uri.escape_string( filename, "", false) +"\">" +
 655             GLib.Uri.escape_string( filename, "", false) + // fixme needs html escaping...
 656             "</a>" +
 657             "</body></html>";
 658
 659         rep.get_content_type().set_parameter("charset", "utf-8");
 660                 rep.set_header("X-strip-id", file_id);
 661                 rep.set_header("X-strip-content-name",  filename);
 662                 rep.set_header("X-strip-path", this.created_dir + "/" + file_id + "-" +  GLib.Uri.escape_string(filename,"", false));
 663                 rep.set_header("X-strip-content-type", mime_type);
 664         var stream =  new GMime.StreamMem.with_buffer(txt.data);
 665         var con = new GMime.DataWrapper.with_stream(stream,GMime.ContentEncoding.DEFAULT);
 666
 667         rep.set_content_object(con);
 668         GLib.debug("Replacing Attachment with HTML");
 669         parent.replace(parent.index_of(attachment), rep);
 670                 this.has_replaced = true;
 671
 672                 if (StripApplication.opt_is_extracting && target_fn.length > 0) {
 673                         var dir = GLib.Path.get_dirname(target_fn);
 674                         if (!FileUtils.test (dir, FileTest.IS_DIR)) {
 675                                 GLib.DirUtils.create_with_parents(dir, 0755);
 676                         }
 677                         GLib.debug("Creating file %s", target_fn);
 678                         if (!FileUtils.test (target_fn, FileTest.EXISTS)) {
 679                                 var from = File.new_for_path (fn);
 680                                 var to =  File.new_for_path (target_fn);
 681                                 from.copy(to, 0, null);
 682
 683                         }
 684                 } else {
 685                         GLib.debug("Skipping extraction %s", target_fn);
 686                 }
 687                 Posix.unlink(fn);
 688
 689
 690
 691     }
 692     public string query(string str)
 693     {
 694             return this.real_query(1, str);
 695     }
 696     public string execute(string str)
 697     {
 698             return this.real_query(0, str);
 699     }
 700     /**
 701     * need_return
 702     0 = no
 703     1 = yes
 704     -1 = don't try.
 705     */
 706     public string real_query(int need_return, string str)
 707     {
 708                 GLib.debug("Before Query : %u  : %s\n", this.mysql.errno(), this.mysql.error());
 709
 710
 711         if (StripApplication.opt_debug_sql) {
 712                 GLib.debug("SQL: %s\n", str);
 713                 }
 714
 715
 716
 717         var rc=  this.mysql.query(str);
 718         if ( rc != 0 ) {
 719
 720                     GLib.debug("ERROR %u: Query failed: %s\n", this.mysql.errno(), this.mysql.error());
 721                                 Posix.exit(1);
 722                 }
 723         var rs = mysql.use_result();
 724                 if (need_return == -1) {
 725                         return "";
 726                 }
 727
 728
 729         //GLib.debug("got %d rows", (int) rs.num_rows());
 730
 731         var got_row = false;
 732                 string[] row;
 733                 string ret = "";
 734                 while( (row = rs.fetch_row()) != null) {
 735                         got_row = true;
 736                         ret = row[0];
 737
 738                 }
 739                 if (need_return == 0) {
 740                 if (StripApplication.opt_debug_sql) {
 741                                 GLib.debug("got %s", got_row ? "=Nothing=" : ret);
 742                         }
 743                         return got_row ? "" : ret;
 744                 }
 745                 if (!got_row) {
 746
 747                          GLib.debug("ERROR : no rows returned");
 748                         Posix.exit(1);
 749                         return "";
 750                 }
 751         if (StripApplication.opt_debug_sql) {
 752                         GLib.debug("got %s", ret);
 753                 }
 754                 return ret;
 755
 756
 757         }
 758
 759     public string mysql_escape(string str)
 760     {
 761             unichar[] value_escaped = new unichar[str.length * 2 + 1];
 762                 this.mysql.real_escape_string ((string) value_escaped, str, str.length);
 763                 return (string) value_escaped;
 764     }
 765
 766     public string  md5_file(string fn) {
 767               Checksum checksum = new Checksum (ChecksumType.MD5);
 768
 769               FileStream stream = FileStream.open (fn, "rb");
 770               uint8 fbuf[100];
 771               size_t size;
 772
 773               while ((size = stream.read (fbuf)) > 0) {
 774                       checksum.update (fbuf, size);
 775               }
 776
 777               unowned string digest = checksum.get_string ();
 778               return digest;
 779     }
 780
 781         string active_path = "";
 782     string active_name = "";
 783     string active_message_id = "";
 784     string active_message_x_mailfort_sig = "";
 785     string active_message_exim_id = "";
 786     bool has_replaced = false;
 787     string created_date = ""; // should be YYYY-mm-dd
 788     string created_dir = ""; // should be YYY/mm/dd
 789
 790     public void scan_file(string path, string name)
 791     {
 792                 GLib.debug("Scan: %s/%s", path,name);
 793
 794                 this.has_replaced = false;
 795         this.active_path = path;
 796         this.active_name = name;
 797         this.active_message_id = "";
 798
 799                 var mailtime = new DateTime.now_local();
 800                 if (StripApplication.opt_scan_mailfort) {
 801                     this.created_dir = this.active_path.substring(this.base_dir.length + 1 );
 802                         this.created_date = this.created_dir.replace("/", "-");
 803                         var bits = this.created_date.split("-");
 804                         mailtime = new DateTime.local(int.parse(bits[0]),int.parse(bits[1]),int.parse(bits[2]),0,0,0);
 805
 806                         var oldest = new  DateTime.now_local();
 807                         oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
 808                         var tspan = mailtime.difference(oldest) / GLib.TimeSpan.DAY;
 809
 810                         if (tspan < 0) {
 811                                 GLib.debug("skip file is %d days older than %d months", (int)tspan, StripApplication.opt_age_oldest);
 812                                 return;
 813                         }
 814
 815                         var newest = new  DateTime.now_local();
 816                         newest = newest.add_months(-1 * StripApplication.opt_age_newest);
 817                         tspan = mailtime.difference(newest) / GLib.TimeSpan.DAY;
 818                         if (tspan > 0) {
 819                                 GLib.debug("skip file is %d days newer than %d months", (int)tspan, StripApplication.opt_age_newest);
 820                                 return;
 821                         }
 822
 823                 }
 824
 825
 826                 var fileinfo = File.new_for_path(path +"/" + name)
 827                                         .query_info(GLib.FileAttribute.STANDARD_SIZE+","+GLib.FileAttribute.TIME_MODIFIED
 828                                                 ,GLib.FileQueryInfoFlags.NONE,null);
 829         var file_size = (int) fileinfo.get_size();
 830                 var mod_time = fileinfo.get_modification_time();
 831
 832
 833
 834                 if (!StripApplication.opt_scan_mailfort) {
 835
 836                 // it's a mail directory...
 837                 // use the last modification time? as the default...
 838                  mailtime = new DateTime.from_timeval_utc(mod_time);
 839                  this.created_dir = mailtime.format("%Y/%m/%d");
 840                          this.created_date =  mailtime.format("%Y-%m-%d %H:%M:%S");
 841
 842         }
 843                 // check on age of file...
 844
 845
 846
 847
 848
 849         this.used_space_before += file_size;
 850
 851         var stream = new GMime.StreamFs.for_path (path +"/" + name,Posix.O_RDONLY, 0);
 852         //stream.set_owner(true);
 853         var parser = new GMime.Parser.with_stream(stream);
 854         var message = parser.construct_message();
 855
 856                 if (message == null) {
 857                         GLib.debug("Could not parse file? %s/%s", path,name);
 858                 this.used_space_after += file_size;
 859                 return;
 860                 }
 861
 862
 863                 // check : - is message over a year old?
 864                 // get various msg info..
 865                 this.active_message_id = message.get_message_id();
 866                 this.active_message_x_mailfort_sig = message.get_header("x-mailfort-sig");
 867                 var recvd = message.get_header("received");
 868                 this.active_message_exim_id = "";
 869                 if (recvd != null && recvd.length > 1) {
 870                         GLib.debug("RECV: %s", recvd);
 871                         var lines = recvd.split("\t");
 872                         for (var i = 0; i < lines.length;i++) {
 873                                 var bits = lines[i].strip().split(" ");
 874                                 if (bits[0] == "id") {
 875                                         this.active_message_exim_id = bits[1].replace(";","");
 876
 877                                 }
 878
 879                                 if (lines[i].contains(";")) {
 880                                         var dbits = lines[i].strip().split(";");
 881                                         GLib.debug("Reading time from : %s", dbits[1]);
 882                                         var timez = GMime.utils_header_decode_date(dbits[1], null);
 883                                         if (timez != 0) {
 884                                                 mailtime = new DateTime.from_unix_utc(timez);
 885                                                 this.created_date = mailtime.format("%Y-%m-%d %H:%M:%S");
 886                                                 GLib.debug("Time is %s",this.created_date);
 887                                                 // if it's not mailfort we can use that date to determine where to store it...
 888                                                 if (!StripApplication.opt_scan_mailfort) {
 889                                                         this.created_dir = mailtime.format("%Y/%m/%d");
 890                                                 }
 891                                         } else {
 892                                                 GLib.debug("Could not read time from headers?");
 893                                         }
 894                                 }
 895
 896                         }
 897                 }
 898
 899                 var oldest = new  DateTime.now_local();
 900                 oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
 901                 var rtspan = mailtime.difference(oldest) / GLib.TimeSpan.DAY;
 902                 GLib.debug("Checking oldest %d days difference", (int)rtspan   );
 903                 if (rtspan < 0) {
 904                         GLib.debug("skip(2) file is %d days older than %d months", (int)rtspan, StripApplication.opt_age_oldest);
 905                         return;
 906                 }
 907                 var newest = new  DateTime.now_local();
 908                 newest = newest.add_months(-1 * StripApplication.opt_age_newest);
 909                 rtspan = mailtime.difference(newest) / GLib.TimeSpan.DAY;
 910                 if (rtspan > 0) {
 911                         GLib.debug("skip(2) file is %d days newer than %d months : %s", (int)rtspan, StripApplication.opt_age_newest,
 912                                 mailtime.format("%Y-%m-%d %H:%M:%S"));
 913                         return;
 914                 }
 915
 916
 917
 918                 /*
 919                 GLib.debug("Message DATA:\n mid: %s\nmailfort: %s \nexim_id: %s",
 920                         this.active_message_id,
 921                         this.active_message_x_mailfort_sig,
 922                         this.active_message_exim_id
 923                 );
 924                  */
 925
 926                 // DATE?
 927
 928                 var mp = message.get_mime_part();
 929
 930                 if (!(mp is GMime.Multipart)) {
 931                         //GLib.debug("get mimepart does not return a Multipart?");
 932                 this.used_space_after += file_size;
 933                         return;
 934                 }
 935
 936                 var mpc = ((GMime.Multipart)mp).get_count();
 937
 938                 //GLib.debug("Message has %d parts", mpc);
 939                 for (var i =0 ; i < mpc; i++) {
 940                         //GLib.debug("Getting part %d", i);
 941                         var mime_obj = ((GMime.Multipart)mp).get_part(i);
 942             this.handle_part(mp,mime_obj);
 943         }
 944
 945         parser= null;
 946
 947       //  stream.set_owner(false);
 948             //stream.close();
 949         stream = null;//.close();
 950
 951
 952                 if (!this.has_replaced) {
 953                         this.used_space_after += file_size;
 954                         GLib.debug("skpping write file - no replacement occured");
 955                         return;
 956                 }
 957                 string tmpfile = "";
 958                 GMime.Stream outstream = new GMime.StreamNull();
 959                 if (StripApplication.opt_is_replacing) {
 960
 961                         tmpfile = GLib.Environment.get_tmp_dir() +"/" + name;
 962                 outstream = new GMime.StreamFile.for_path (tmpfile,"w");
 963                 ((GMime.StreamFile)outstream).set_owner(true);
 964         }
 965                 if (StripApplication.opt_dump) {
 966                         outstream = new GMime.StreamMem();
 967         }
 968
 969         file_size = (int) message.write_to_stream(outstream);
 970         if (StripApplication.opt_is_replacing) {
 971                 ((GMime.StreamFile)outstream).set_owner(false);
 972         }
 973                 if (StripApplication.opt_dump) {
 974                         var ua = ((GMime.StreamMem)outstream).get_byte_array().data;
 975                         print("%s\n", (string) ua);
 976                 }
 977         message = null;
 978         outstream.flush();
 979         outstream.close();
 980         GLib.debug("finished writing output %d", file_size);
 981
 982         //
 983         outstream = null;
 984
 985
 986         this.used_space_after += file_size;
 987
 988
 989         if (StripApplication.opt_is_replacing) {
 990                 Posix.unlink(path +"/" + name);
 991                 GLib.debug("copy tmp file %s to %s" , tmpfile, path +"/" + name);
 992
 993                 // link will not work, as we are doing it accross file systems
 994                         var from = File.new_for_path (tmpfile);
 995                         var nf =  File.new_for_path (path +"/" + name);
 996                         from.copy(nf, 0, null);
 997
 998
 999                 var newfileinfo = nf.query_info(GLib.FileAttribute.TIME_MODIFIED,GLib.FileQueryInfoFlags.NONE,null);
1000                 newfileinfo.set_modification_time(mod_time);
1001                 nf.set_attributes_from_info(newfileinfo,FileQueryInfoFlags.NONE);
1002                 Posix.unlink(tmpfile);
1003                 }
1004         this.processed++;
1005
1006         if (StripApplication.opt_limit > -1 && this.processed >= StripApplication.opt_limit) {
1007                 GLib.debug("Reached replacement limit");
1008                 Posix.exit(1);
1009         }
1010
1011
1012
1013
1014     }
1015
1016
1017     public void scan_dir(string basepath, string subpath)
1018     {
1019
1020
1021         // determine if path is to old to scan..
1022         if (subpath.length > 0 && StripApplication.opt_scan_mailfort) {
1023                         var year =  int.parse(subpath.substring(1,4));  // "/2000"
1024                         var month = subpath.length > 5 ? int.parse(subpath.substring(6,2)) : 999; // "/2000/12"
1025                         var day = subpath.length > 8 ? int.parse(subpath.substring(9,2)) : 999; // "/2000/12/01"
1026
1027                 var oldest = new  DateTime.now_local();
1028                         oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
1029
1030                         //GLib.debug("Checking directory %s is older than min: %d/%d/%d", subpath, oldest.get_year() , oldest.get_month(), oldest.get_day_of_month() );
1031
1032                         if (year < oldest.get_year()) {
1033                                 GLib.debug("Skip directory %s is older than min year: %d", subpath, oldest.get_year());
1034                                 return;
1035                         }
1036                         if (year == oldest.get_year() &&  month < oldest.get_month()) {
1037                                 GLib.debug("Skip directory %s is older than min month: %d/%d", subpath, oldest.get_year() , oldest.get_month() );
1038                                 return;
1039                         }
1040                 if (year == oldest.get_year() &&  month == oldest.get_month() && day < oldest.get_day_of_month()) {
1041                                 GLib.debug("Skip directory %s is older than min day: %d/%d/%d", subpath, oldest.get_year() , oldest.get_month(), oldest.get_day_of_month() );
1042                                 return;
1043                         }
1044
1045                 var newest = new  DateTime.now_local();
1046                         newest = newest.add_months(-1 * StripApplication.opt_age_newest);
1047
1048                         //GLib.debug("Checking directory %s is newer than max: %d/%d/%d", subpath, newest.get_year() , newest.get_month(), newest.get_day_of_month() );
1049
1050                         if (year > newest.get_year()) {
1051                                 GLib.debug("Skip directory %s is newer than max year: %d", subpath, newest.get_year());
1052                                 return;
1053                         }
1054                         if (year == newest.get_year() &&  month != 999 && month > newest.get_month()) {
1055                                 GLib.debug("Skip directory %s is newer than max month: %d/%d", subpath, newest.get_year() , newest.get_month() );
1056                                 return;
1057                         }
1058                 if (year == newest.get_year() &&  month == newest.get_month() &&  day != 999 && day > newest.get_day_of_month()) {
1059                                 GLib.debug("Skip directory %s is newer than max day: %d/%d/%d", subpath, newest.get_year() , newest.get_month(), newest.get_day_of_month() );
1060                                 return;
1061                         }
1062
1063
1064
1065         }
1066
1067
1068         var f = File.new_for_path(basepath + subpath);
1069                 FileEnumerator file_enum;
1070         var cancellable = new Cancellable ();
1071         try {
1072             file_enum = f.enumerate_children(
1073                 FileAttribute.STANDARD_DISPLAY_NAME + "," +   FileAttribute.STANDARD_TYPE,
1074                         FileQueryInfoFlags.NOFOLLOW_SYMLINKS,  // FileQueryInfoFlags.NONE,
1075                         cancellable
1076                 );
1077         } catch (Error e) {
1078                 GLib.debug("Got error scanning dir? %s", e.message);
1079             // FIXME - show error..
1080             return;
1081         }
1082         FileInfo next_file;
1083
1084         while (cancellable.is_cancelled () == false ) {
1085             try {
1086                 next_file = file_enum.next_file (cancellable);
1087             } catch(Error e) {
1088                 GLib.debug("error getting next file? %s", e.message);
1089                 break;
1090             }
1091
1092             if (next_file == null) {
1093                 break;
1094             }
1095
1096
1097                 if (next_file.get_is_symlink()) {
1098                 next_file = null;
1099                 continue;
1100             }
1101
1102             var ds = next_file.get_display_name();
1103             if (next_file.get_file_type() != FileType.DIRECTORY) {
1104
1105
1106
1107                 if (ds[0] == ',') {
1108                         continue;
1109                 }
1110                 // other files to ignore?
1111                 if (Regex.match_simple (".tgz$", ds)) {
1112                         continue;
1113                 }
1114                 this.scan_file(basepath + subpath , ds);
1115                                 if(this.has_replaced) {
1116                          this.report_state("After scanning %s/%s".printf(basepath + subpath , ds));
1117                         }
1118                 continue;
1119             }
1120
1121
1122             //stdout.printf("Monitor.monitor: got file %s : type :%u\n",
1123             //        next_file.get_display_name(), next_file.get_file_type());
1124
1125
1126
1127
1128             // not really needed?? - we are storing attachments in a seperate location now...
1129             if (ds[0] == '.') {
1130                 next_file = null;
1131                 continue;
1132             }
1133             if (ds == "attachments") {
1134                         continue;
1135                 }
1136
1137
1138             var sp = subpath+"/"+next_file.get_display_name();
1139             // skip modules.
1140             //print("got a file : " + sp);
1141
1142             next_file = null;
1143
1144
1145             this.scan_dir(basepath,sp);
1146
1147         }
1148
1149
1150     }
1151     void report_state(string msg)
1152     {
1153         // Saved: 2G  Original 10G : 20%
1154         GLib.debug("Saved : %s (%.1f%%) | Original %s | %s",
1155                         GLib.format_size(this.used_space_before - this.used_space_after),
1156                         100f * ((1f * (this.used_space_before - this.used_space_after)) / (this.used_space_before * 1f)),
1157                         GLib.format_size(this.used_space_before),
1158                         msg
1159                 );
1160
1161         }
1162
1163
1164
1165 }