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         public static string? opt_scan_mailfort_only = "";
  91
  92
  93         public static int    opt_limit = -1;
  94
  95         public static int    opt_age_newest = 1;
  96         public static int    opt_age_oldest = 6;
  97
  98
  99         public static bool      opt_is_extracting = false;
 100         public static bool      opt_is_replacing = false;
 101         public static bool      opt_scan_maildir  = false;
 102         public static bool      opt_scan_mailfort  = false;
 103         public static bool              opt_dump = false;
 104         public static bool              opt_debug = false;
 105
 106         public static bool opt_debug_sql = false;
 107         public static string? opt_replace_link = null;
 108
 109
 110         public const GLib.OptionEntry[] options = {
 111
 112                 { "debug", 0, 0, OptionArg.NONE, ref opt_debug, "show debug messages for components", null },
 113                 { "debug-sql", 0, 0, OptionArg.NONE, ref opt_debug_sql, "debug the SQL statements", null },
 114
 115                 { "path", 0, 0, OptionArg.STRING, ref opt_path, "Directory where email to be parsed is", null },
 116                 { "file", 0, 0, OptionArg.STRING, ref opt_file, "A specific file to be parsed", null },
 117
 118                 { "target-path", 0, 0, OptionArg.STRING, ref opt_target_path, "Directory where attachments are to be put", null },
 119
 120                 { "link", 0, 0, OptionArg.STRING, ref opt_replace_link, "url for the replement link: eg. http://www.mysite.com/xxxx/%s", null },
 121
 122                 { "host", 0, 0, OptionArg.STRING, ref opt_db_host, "Mysql host (default localhost)", null },
 123                 { "name", 0, 0, OptionArg.STRING, ref opt_db_name, "Mysql database name REQUIRED", null },
 124                 { "user", 0, 0, OptionArg.STRING, ref opt_db_user, "Mysql database user REQUIRED", null },
 125                 { "pass", 0, 0, OptionArg.STRING, ref opt_db_pass, "Mysql database password (default empty)", null },
 126
 127                 { "extract", 0, 0, OptionArg.NONE, ref opt_is_extracting, "Should attachments be extracted (default NO)", null },
 128                 { "replace", 0, 0, OptionArg.NONE, ref opt_is_replacing, "Should attachments be replaced (default NO)", null },
 129                 { "dump", 0, 0, OptionArg.NONE, ref opt_dump, "Print the replaced mail contents to stdout", null },
 130
 131                 { "limit", 0, 0, OptionArg.INT, ref opt_limit, "stop after X number of messages with attachments have been processed", null },
 132                 { "newest", 0, 0, OptionArg.INT, ref opt_age_newest, "do not replace messages newer that X months (default is 1 months)", null },
 133                 { "oldest", 0, 0, OptionArg.INT, ref opt_age_oldest, "do not replace messages older than X (default is 6 months)", null },
 134
 135                 { "scan-maildir", 0, 0, OptionArg.NONE, ref opt_scan_maildir, "scan an maildir tree", null },
 136                 { "scan-mailfort", 0, 0, OptionArg.NONE, ref opt_scan_mailfort, "scan a mailfort tree", null },
 137                 { "scan-mailfort-only", 0, 0, OptionArg.STRING, ref opt_scan_mailfort_only, "scan a mailfort Year/month eg. 2010/08", null },
 138                 { null }
 139         };
 140     public StripApplication( string[] args )
 141     {
 142                  Object(
 143             application_id: "org.roojs.mailstripper",
 144             flags: ApplicationFlags.FLAGS_NONE
 145          );
 146
 147
 148          var opt_context =  new GLib.OptionContext ("Mail Stripper");
 149
 150          try {
 151
 152             opt_context.set_help_enabled (true);
 153             opt_context.add_main_entries (options, null);
 154             opt_context.parse ( ref  args);
 155             //opt_detach = !optx_no_detach;
 156
 157
 158
 159              // options that have to be set.. bee or hive... (or stop all)
 160             if ((!opt_scan_mailfort && !opt_scan_maildir) || (opt_scan_mailfort && opt_scan_maildir))  {
 161                stdout.printf ("You must specify the type of directory tree to scan - either imap or mailfort\n%s",
 162                    opt_context.get_help(true, null));
 163                GLib.Process.exit(Posix.EXIT_FAILURE);
 164             }
 165
 166                          if ((opt_db_name == null || opt_db_name.length < 1 || opt_db_user == null || opt_db_user.length < 1))  {
 167                stdout.printf ("You must specify the database name / user \n%s",
 168                    opt_context.get_help(true, null));
 169                GLib.Process.exit(Posix.EXIT_FAILURE);
 170             }
 171                          if ((opt_path == null || opt_path.length < 1)   )  {
 172                stdout.printf ("You must specify the scan start path\n%s",
 173                    opt_context.get_help(true, null));
 174                GLib.Process.exit(Posix.EXIT_FAILURE);
 175             }
 176                         if (opt_replace_link == null || (opt_replace_link.length < 1))  {
 177                stdout.printf ("You must specify the link to use in the replacement \n%s",
 178                    opt_context.get_help(true, null));
 179                GLib.Process.exit(Posix.EXIT_FAILURE);
 180             }
 181             if ((opt_is_replacing || opt_is_extracting ) && (opt_target_path == null || opt_target_path.length < 1)) {
 182                       stdout.printf ("You must specify a target path to put attachments\n%s",
 183                    opt_context.get_help(true, null));
 184                GLib.Process.exit(Posix.EXIT_FAILURE);
 185             }
 186
 187
 188          } catch (GLib.OptionError e) {
 189             stdout.printf ("error: %s\n", e.message);
 190             stdout.printf ("Run '%s --help' to see a full list of available command line options.\n%s",
 191                       args[0], opt_context.get_help(true, null));
 192             GLib.Process.exit(Posix.EXIT_FAILURE);
 193          }
 194         }
 195
 196     public static int main(string[] args)
 197     {
 198
 199                 var application = new StripApplication(  args);
 200
 201                 GLib.Log.set_always_fatal(LogLevelFlags.LEVEL_ERROR | LogLevelFlags.LEVEL_CRITICAL);
 202
 203            if (opt_debug || opt_debug_sql) {
 204                         GLib.Log.set_handler(null,
 205                         GLib.LogLevelFlags.LEVEL_DEBUG | GLib.LogLevelFlags.LEVEL_WARNING | GLib.LogLevelFlags.LEVEL_INFO,
 206                         (dom, lvl, msg) => {
 207                                         print("%s\n", msg);
 208                                 }
 209                         );
 210                 }
 211
 212         GMime.init(0);
 213                 if (StripApplication.opt_is_replacing) {
 214                         StripApplication.opt_is_extracting = true;
 215                 }
 216
 217                 GLib.debug("scanning folder: %s", opt_path );
 218
 219                 var strip = new Strip( opt_path );
 220
 221
 222                 strip.mysql  = new Mysql.Database();
 223                 if (!strip.mysql.real_connect(
 224                                 opt_db_host,
 225                                 opt_db_user ,
 226                                 opt_db_pass == null ? "" : opt_db_pass, //passwd
 227                                 opt_db_name, //DB
 228                                 3306, // not changable...?
 229                                 null
 230                         )
 231                 ) {
 232                         stdout.printf("ERROR %u: Connection failed: %s\n",
 233                                 strip.mysql.errno(), strip.mysql.error()
 234                         );
 235
 236                         return 1;
 237                 }
 238         if (opt_file != null) {
 239                 strip.base_dir = opt_path;
 240                 strip.scan_file( GLib.Path.get_dirname(opt_file),  GLib.Path.get_basename(opt_file));
 241                 return 0;
 242         }
 243
 244                 strip.scan_dir(opt_path, opt_scan_mailfort_only);
 245
 246
 247
 248         return 0;
 249     }
 250 }
 251
 252 public class Strip : GLib.Object {
 253
 254
 255
 256         public string base_dir = "";
 257
 258         public Mysql.Database mysql;
 259
 260         int processed = 0;
 261
 262     uint64 used_space_before = 0;
 263     uint64 used_space_after = 0;
 264
 265
 266     public Strip(string base_dir)
 267     {
 268         this.base_dir = base_dir;
 269     }
 270
 271     public void handle_part(GMime.Object parent, GMime.Object mime_obj)
 272     {
 273                 if (mime_obj is GMime.Part) {
 274                    var  p = (GMime.Part)mime_obj;
 275                         var ct = p.get_content_type();
 276                         var cd = p.get_content_disposition();
 277
 278                         var sid = p.get_header("X-strip-id");
 279                     if (sid != null && sid.length > 0) {
 280                         this.update_attachment_db(p);
 281                             GLib.debug("Skip attachment replace - it's already been done");
 282                         return;
 283                         }
 284
 285                         if (cd == null || cd.get_disposition().down() != "attachment") {
 286                                 return;
 287                         }
 288                         if (ct.get_media_type() == "text") {
 289                                 return;
 290                         }
 291                         if (ct.to_string() == "application/pgp-encrypted") {
 292                                 return;
 293                         }
 294                         if (ct.to_string() == "application/pgp-keys") {
 295                                 return;
 296                         }
 297                         if (p.get_filename() == null) {
 298                                 return;
 299                         }
 300                          // print("got part %s\n", ct.to_string());
 301                          if (parent is GMime.Multipart) {
 302
 303                                 this.replace_attachment(((GMime.Multipart)parent), p);
 304                                 // remove it !?
 305
 306                           }
 307
 308
 309                         return;
 310                 }
 311                 if (mime_obj is GMime.Multipart) {
 312
 313
 314                         var  mp = (GMime.Multipart)mime_obj;
 315                         //var ct = mp.get_content_type();
 316
 317                         //print("got multi-part %s\n", ct.to_string());
 318                         for (var i = 0; i< mp.get_count(); i++) {
 319                           var mo = mp.get_part(i);
 320                           this.handle_part(mime_obj,mo);
 321                         }
 322                    // ((GMime.Multipart)mime_obj).foreach((sub_obj) => {
 323                    //     Strip.handle_part(sub_obj);
 324                 //
 325                    // });
 326
 327
 328                         return;
 329                 }
 330
 331                 if (mime_obj is GMime.MessagePart) {
 332                         var msg = ((GMime.MessagePart)mime_obj).get_message();
 333                         msg.foreach((subobj) => {
 334                          this.handle_part(msg,subobj);
 335                     });
 336
 337                         //print("got message-part\n");
 338                         return;
 339                 }
 340
 341                 if (mime_obj is GMime.Message) {
 342                         var mp = ((GMime.Message) mime_obj).get_mime_part();
 343
 344                         if (!(mp is GMime.Multipart)) {
 345                                 //GLib.debug("get mimepart does not return a Multipart?");
 346                                 return;
 347                         }
 348
 349                         var mpc = ((GMime.Multipart)mp).get_count();
 350
 351                         //GLib.debug("Message has %d parts", mpc);
 352                         for (var i =0 ; i < mpc; i++) {
 353                                 //GLib.debug("Getting part %d", i);
 354                                 var submime_obj = ((GMime.Multipart)mp).get_part(i);
 355                         this.handle_part(mp,submime_obj);
 356                     }
 357                         print("got message??\n");
 358                         return;
 359                 }
 360
 361                 print("got something else\n");
 362
 363
 364     }
 365     public void update_attachment_db(GMime.Part attachment)
 366     {
 367         // only called when we have an sid...
 368         var sid = attachment.get_header("X-strip-id");
 369         if (sid == null || sid.length < 1) {
 370                 GLib.debug("Strange - update attachment db called ?");
 371                 return;
 372         }
 373
 374         // initialize it with known data..
 375         // that should wipe out dupes.
 376         var matches = this.query("SELECT count(id)   FROM Attachment WHERE id = %d".printf(
 377                         int.parse(sid)));
 378
 379
 380                 if (matches == "0") {
 381                         // our old mailfort code deleted the crap out of old records...
 382                         // if this occurs we will need to create the record again..
 383                         this.fix_deleted_attachment_db(int.parse(sid),attachment);
 384                         return;
 385
 386                 }
 387
 388
 389         // initialize it with known data..
 390         // that should wipe out dupes.
 391         var filesize = this.query("SELECT filesize FROM Attachment WHERE id = %d".printf(
 392                         int.parse(sid)));
 393
 394                 if (filesize=="") {
 395                    GLib.error("Ignoring record id (missing in database) :%s", sid);
 396                    return;
 397                 }
 398                 if (int.parse(filesize) < 1) {
 399                 GLib.debug("Could not get filesize from id :%s = %s", sid,filesize);
 400                 Posix.exit(0);
 401                 return;
 402         }
 403
 404         var chksum = this.query("SELECT  checksum FROM Attachment WHERE id = %d".printf(
 405                         int.parse(sid)
 406                 ));
 407         var mime_filename = this.query("SELECT  mime_filename FROM Attachment WHERE id = %d".printf(
 408                         int.parse(sid)));
 409
 410         this.query("""
 411              SELECT
 412                  attachment_init(
 413                      '%s', '%s', '%s', %d
 414                  ) as id
 415
 416           """.printf(
 417                           this.mysql_escape(this.active_message_exim_id),
 418                           this.mysql_escape(chksum),
 419                           this.mysql_escape(mime_filename),
 420                           int.parse(filesize)
 421                 ));
 422         this.query("""
 423                  SELECT attachment_update(
 424                       %d, -- in_id INT(11),
 425                       '%s', -- in_mime_type varchar(255),
 426                       '%s', -- in_created DATETIME,
 427                       '%s' -- in_mailfort_sig varchar(64)
 428                  )
 429               """.printf(
 430                         int.parse(sid),
 431                         "", // this will be ignored..
 432                                 this.created_date,
 433                                 this.mysql_escape(this.active_message_x_mailfort_sig)
 434
 435               )
 436                 );
 437                 this.mysql.store_result();
 438
 439
 440
 441     }
 442
 443
 444     public void fix_deleted_attachment_db(int id, GMime.Part attachment)
 445     {
 446
 447         var filename = attachment.get_header("X-strip-content-name");
 448         var file_path  = attachment.get_header("X-strip-path");
 449         var fn =  StripApplication.opt_target_path + "/" + file_path;
 450
 451
 452                 if (!FileUtils.test (fn, FileTest.EXISTS)) {
 453                         GLib.debug("SKIP -- file does not exist");
 454                         return;
 455         }
 456
 457         var chksum = this.md5_file(fn);
 458                 var mime_type = attachment.get_header("X-strip-content-type");
 459
 460                 var fileinfo = File.new_for_path(fn)
 461                                         .query_info(GLib.FileAttribute.STANDARD_SIZE+","+GLib.FileAttribute.TIME_MODIFIED
 462                                                 ,GLib.FileQueryInfoFlags.NONE,null);
 463         var file_size = (int) fileinfo.get_size();
 464
 465
 466                 this.real_query(-1, """
 467
 468
 469                                 INSERT INTO Attachment  (
 470                                         id,
 471
 472                                     msgid ,
 473                                     queue_id ,
 474                                     mime_filename ,
 475                                     mime_type,
 476
 477                                     stored_filename ,
 478                                     mime_charset ,
 479                                     mime_cdisp ,
 480                                     mime_is_cover ,
 481
 482                                     mime_is_multi ,
 483                                     mime_is_mail,
 484                                     mime_size ,
 485                                     filesize,
 486
 487                                     checksum,
 488                                     created
 489
 490                                 ) VALUES (
 491                                         %d,  -- id
 492
 493                                     '%s' , -- msgid
 494                                     0,
 495                                     '%s'  , -- filename
 496                                     '%s',  -- mimetype
 497
 498                                     '%s', -- stored file anme
 499                                     '', -- charset
 500                                     'attachment',
 501                                     0,
 502                                     0,
 503                                     0,
 504                                     %d, -- size
 505                                     %d, -- size
 506
 507                                     '%s', -- checkum
 508                                         '%s' -- created:
 509                                 )
 510
 511
 512                       """.printf(
 513                                 id,
 514                                       this.mysql_escape(this.active_message_exim_id),
 515                                       this.mysql_escape(filename),
 516                                   this.mysql_escape(mime_type),
 517                                   this.mysql_escape(file_path),
 518                                         file_size,
 519                                       file_size,
 520                                       this.mysql_escape(chksum),
 521                                 this.created_date
 522                          ));
 523               // this is done to fix the queue_id or maillog_id ??
 524                  this.query("""
 525                  SELECT attachment_update(
 526                       %d, -- in_id INT(11),
 527                                                 '', -- mime type
 528                       '%s', -- in_created DATETIME,
 529                       '%s' -- in_mailfort_sig varchar(64)
 530                  )
 531               """.printf(
 532                                 id,
 533                                 this.created_date,
 534                                 this.mysql_escape(this.active_message_x_mailfort_sig)
 535
 536               )
 537                 );
 538                 // GLib.error("added attachment?");
 539     }
 540
 541
 542     public void replace_attachment(GMime.Multipart parent, GMime.Part attachment)
 543     {
 544         var sid = attachment.get_header("X-strip-id");
 545         if (sid != null && sid.length > 0) {
 546                 GLib.debug("Skip attachment replace - it's already been done");
 547                 return;
 548         }
 549
 550         var c = attachment.get_content_object();
 551
 552         var filename = attachment.get_filename().replace("/", "-").replace("\n", "").replace("\t", " ");
 553         var fn = GLib.Environment.get_tmp_dir() +
 554                         "/"+ this.active_name + "."+   GLib.Uri.escape_string(filename,"", false);
 555
 556             var outfile = new GMime.StreamFile.for_path(fn, "w");
 557             outfile.set_owner(true);
 558             var file_size = (int) c.write_to_stream(outfile);
 559             var chksum = this.md5_file(fn);
 560             outfile.flush();
 561             outfile = null;
 562
 563         if (file_size == 0) {
 564
 565                 GLib.debug("ERROR - file size of write to stream returned 0?");
 566                 Posix.unlink(fn);
 567                 return;
 568         }
 569
 570
 571
 572
 573         var mime_type= attachment.get_content_type().to_string();
 574         // at this point we have to do our database magic...
 575         //filesize / name / date / checksum / mimetype -- into mailfort should be OK.
 576
 577         var file_id = this.query("""
 578                 SELECT
 579
 580                 attachment_init(
 581                                 '%s', -- in_msgid VARCHAR(32),
 582                                 '%s', -- in_checksum VARCHAR(64),
 583                                 '%s', -- in_mime_filename varchar(255)
 584                                 %d -- filesize
 585                         ) as id
 586
 587           """.printf(
 588                         this.mysql_escape(this.active_message_exim_id),
 589                         chksum,
 590                         this.mysql_escape( GLib.Uri.escape_string(attachment.get_filename(),"", false) ), // what is thsi is invalid?
 591                          file_size)
 592                 );
 593
 594
 595                 if (file_id.length < 1) {
 596                         GLib.debug("ERROR - CALL to attachment_init failed");
 597                 Posix.unlink(fn);
 598                 return;
 599
 600                 }
 601
 602                 if (int.parse(file_id) < 1) {
 603                         GLib.debug("ERROR - CALL to attachment_init failed - returned 0?");
 604                 Posix.unlink(fn);
 605                 return;
 606
 607                 }
 608
 609
 610                 GLib.debug("fn = %s, m5=%s, id= %s", filename, mime_type, this.active_message_id);
 611                 this.query("""
 612
 613                         SELECT attachment_update(
 614                                 %d, -- in_id INT(11),
 615                                 '%s', -- in_mime_type varchar(255),
 616                                 '%s', -- in_created DATETIME,
 617                                 '%s' -- in_mailfort_sig varchar(64)
 618
 619                                 ) as result
 620       """.printf(
 621                 int.parse(file_id),
 622                         this.mysql_escape(mime_type),
 623                         this.created_date,
 624                         this.mysql_escape(this.active_message_x_mailfort_sig)
 625                 ));
 626                  this.mysql.store_result();
 627
 628
 629                 this.used_space_after += file_size;
 630
 631                 var target_fn = "";
 632
 633             if (StripApplication.opt_is_extracting) {
 634                         target_fn = StripApplication.opt_target_path + "/" + this.created_dir +"/"+ file_id  + "-" + GLib.Uri.escape_string(filename,"", false);
 635                 }
 636
 637             var stored =  "/" + this.created_dir +"/"+ file_id  + "-" + GLib.Uri.escape_string(filename,"", false);
 638                  this.query("""
 639
 640                         SELECT attachment_update_store(
 641                                 %d, -- in_id INT(11),
 642                                 '%s'  -- in_store_filename varchar(255),
 643
 644
 645                                 ) as result
 646       """.printf(
 647                 int.parse(file_id),
 648                          this.mysql_escape( stored)
 649                 ));
 650
 651         var rep = new GMime.Part.with_type("text","html");
 652         // we have to set up a redirect server - to redirect hpasite... to their internal service..
 653         rep.set_filename(filename);
 654         string txt = "<html><body>"+
 655             "<a href=\"" + StripApplication.opt_replace_link + "/" +
 656                         file_id + "/" + this.created_dir + "/"+chksum+"/"+ GLib.Uri.escape_string( filename, "", false) +"\">" +
 657             GLib.Uri.escape_string( filename, "", false) + // fixme needs html escaping...
 658             "</a>" +
 659             "</body></html>";
 660
 661         rep.get_content_type().set_parameter("charset", "utf-8");
 662                 rep.set_header("X-strip-id", file_id);
 663                 rep.set_header("X-strip-content-name",  filename);
 664                 rep.set_header("X-strip-path", this.created_dir + "/" + file_id + "-" +  GLib.Uri.escape_string(filename,"", false));
 665                 rep.set_header("X-strip-content-type", mime_type);
 666         var stream =  new GMime.StreamMem.with_buffer(txt.data);
 667         var con = new GMime.DataWrapper.with_stream(stream,GMime.ContentEncoding.DEFAULT);
 668
 669         rep.set_content_object(con);
 670         GLib.debug("Replacing Attachment with HTML");
 671         parent.replace(parent.index_of(attachment), rep);
 672                 this.has_replaced = true;
 673
 674                 if (StripApplication.opt_is_extracting && target_fn.length > 0) {
 675                         var dir = GLib.Path.get_dirname(target_fn);
 676                         if (!FileUtils.test (dir, FileTest.IS_DIR)) {
 677                                 GLib.DirUtils.create_with_parents(dir, 0755);
 678                         }
 679                         GLib.debug("Creating file %s", target_fn);
 680                         if (!FileUtils.test (target_fn, FileTest.EXISTS)) {
 681                                 var from = File.new_for_path (fn);
 682                                 var to =  File.new_for_path (target_fn);
 683                                 from.copy(to, 0, null);
 684
 685                         }
 686                 } else {
 687                         GLib.debug("Skipping extraction %s", target_fn);
 688                 }
 689                 Posix.unlink(fn);
 690
 691
 692
 693     }
 694     public string query(string str)
 695     {
 696             return this.real_query(1, str);
 697     }
 698     public string execute(string str)
 699     {
 700             return this.real_query(0, str);
 701     }
 702     /**
 703     * need_return
 704     0 = no
 705     1 = yes
 706     -1 = don't try.
 707     */
 708     public string real_query(int need_return, string str)
 709     {
 710                 GLib.debug("Before Query : %u  : %s\n", this.mysql.errno(), this.mysql.error());
 711
 712
 713         if (StripApplication.opt_debug_sql) {
 714                 GLib.debug("SQL: %s\n", str);
 715                 }
 716
 717
 718
 719         var rc=  this.mysql.query(str);
 720         if ( rc != 0 ) {
 721
 722                     GLib.debug("ERROR %u: Query failed: %s\n", this.mysql.errno(), this.mysql.error());
 723                                 Posix.exit(1);
 724                 }
 725         var rs = mysql.use_result();
 726                 if (need_return == -1) {
 727                         return "";
 728                 }
 729
 730
 731         //GLib.debug("got %d rows", (int) rs.num_rows());
 732
 733         var got_row = false;
 734                 string[] row;
 735                 string ret = "";
 736                 while( (row = rs.fetch_row()) != null) {
 737                         got_row = true;
 738                         ret = row[0];
 739
 740                 }
 741                 if (need_return == 0) {
 742                 if (StripApplication.opt_debug_sql) {
 743                                 GLib.debug("got %s", got_row ? "=Nothing=" : ret);
 744                         }
 745                         return got_row ? "" : ret;
 746                 }
 747                 if (!got_row) {
 748
 749                          GLib.debug("ERROR : no rows returned");
 750                         Posix.exit(1);
 751                         return "";
 752                 }
 753         if (StripApplication.opt_debug_sql) {
 754                         GLib.debug("got %s", ret);
 755                 }
 756                 return ret;
 757
 758
 759         }
 760
 761     public string mysql_escape(string str)
 762     {
 763             unichar[] value_escaped = new unichar[str.length * 2 + 1];
 764                 this.mysql.real_escape_string ((string) value_escaped, str, str.length);
 765                 return (string) value_escaped;
 766     }
 767
 768     public string  md5_file(string fn) {
 769               Checksum checksum = new Checksum (ChecksumType.MD5);
 770
 771               FileStream stream = FileStream.open (fn, "rb");
 772               uint8 fbuf[100];
 773               size_t size;
 774
 775               while ((size = stream.read (fbuf)) > 0) {
 776                       checksum.update (fbuf, size);
 777               }
 778
 779               unowned string digest = checksum.get_string ();
 780               return digest;
 781     }
 782
 783         string active_path = "";
 784     string active_name = "";
 785     string active_message_id = "";
 786     string active_message_x_mailfort_sig = "";
 787     string active_message_exim_id = "";
 788     bool has_replaced = false;
 789     string created_date = ""; // should be YYYY-mm-dd
 790     string created_dir = ""; // should be YYY/mm/dd
 791
 792     public void scan_file(string path, string name)
 793     {
 794                 GLib.debug("Scan: %s/%s", path,name);
 795
 796                 this.has_replaced = false;
 797         this.active_path = path;
 798         this.active_name = name;
 799         this.active_message_id = "";
 800
 801                 var mailtime = new DateTime.now_local();
 802                 if (StripApplication.opt_scan_mailfort) {
 803                     this.created_dir = this.active_path.substring(this.base_dir.length + 1 );
 804                         this.created_date = this.created_dir.replace("/", "-");
 805                         var bits = this.created_date.split("-");
 806                         mailtime = new DateTime.local(int.parse(bits[0]),int.parse(bits[1]),int.parse(bits[2]),0,0,0);
 807
 808                         var oldest = new  DateTime.now_local();
 809                         oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
 810                         var tspan = mailtime.difference(oldest) / GLib.TimeSpan.DAY;
 811
 812                         if (tspan < 0) {
 813                                 GLib.debug("skip file is %d days older than %d months", (int)tspan, StripApplication.opt_age_oldest);
 814                                 return;
 815                         }
 816
 817                         var newest = new  DateTime.now_local();
 818                         newest = newest.add_months(-1 * StripApplication.opt_age_newest);
 819                         tspan = mailtime.difference(newest) / GLib.TimeSpan.DAY;
 820                         if (tspan > 0) {
 821                                 GLib.debug("skip file is %d days newer than %d months", (int)tspan, StripApplication.opt_age_newest);
 822                                 return;
 823                         }
 824
 825                 }
 826
 827
 828                 var fileinfo = File.new_for_path(path +"/" + name)
 829                                         .query_info(GLib.FileAttribute.STANDARD_SIZE+","+GLib.FileAttribute.TIME_MODIFIED
 830                                                 ,GLib.FileQueryInfoFlags.NONE,null);
 831         var file_size = (int) fileinfo.get_size();
 832                 var mod_time = fileinfo.get_modification_time();
 833
 834
 835
 836                 if (!StripApplication.opt_scan_mailfort) {
 837
 838                 // it's a mail directory...
 839                 // use the last modification time? as the default...
 840                  mailtime = new DateTime.from_timeval_utc(mod_time);
 841                  this.created_dir = mailtime.format("%Y/%m/%d");
 842                          this.created_date =  mailtime.format("%Y-%m-%d %H:%M:%S");
 843
 844         }
 845                 // check on age of file...
 846
 847
 848
 849
 850
 851         this.used_space_before += file_size;
 852
 853         var stream = new GMime.StreamFs.for_path (path +"/" + name,Posix.O_RDONLY, 0);
 854         //stream.set_owner(true);
 855         var parser = new GMime.Parser.with_stream(stream);
 856         var message = parser.construct_message();
 857
 858                 if (message == null) {
 859                         GLib.debug("Could not parse file? %s/%s", path,name);
 860                 this.used_space_after += file_size;
 861                 return;
 862                 }
 863
 864
 865                 // check : - is message over a year old?
 866                 // get various msg info..
 867                 this.active_message_id = message.get_message_id();
 868                 this.active_message_x_mailfort_sig = message.get_header("x-mailfort-sig");
 869                 var recvd = message.get_header("received");
 870                 this.active_message_exim_id = "";
 871                 if (recvd != null && recvd.length > 1) {
 872                         GLib.debug("RECV: %s", recvd);
 873                         var lines = recvd.split("\t");
 874                         for (var i = 0; i < lines.length;i++) {
 875                                 var bits = lines[i].strip().split(" ");
 876                                 if (bits[0] == "id") {
 877                                         this.active_message_exim_id = bits[1].replace(";","");
 878
 879                                 }
 880
 881                                 if (lines[i].contains(";")) {
 882                                         var dbits = lines[i].strip().split(";");
 883                                         GLib.debug("Reading time from : %s", dbits[1]);
 884                                         var timez = GMime.utils_header_decode_date(dbits[1], null);
 885                                         if (timez != 0) {
 886                                                 mailtime = new DateTime.from_unix_utc(timez);
 887                                                 this.created_date = mailtime.format("%Y-%m-%d %H:%M:%S");
 888                                                 GLib.debug("Time is %s",this.created_date);
 889                                                 // if it's not mailfort we can use that date to determine where to store it...
 890                                                 if (!StripApplication.opt_scan_mailfort) {
 891                                                         this.created_dir = mailtime.format("%Y/%m/%d");
 892                                                 }
 893                                         } else {
 894                                                 GLib.debug("Could not read time from headers?");
 895                                         }
 896                                 }
 897
 898                         }
 899                 }
 900
 901                 var oldest = new  DateTime.now_local();
 902                 oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
 903                 var rtspan = mailtime.difference(oldest) / GLib.TimeSpan.DAY;
 904                 GLib.debug("Checking oldest %d days difference", (int)rtspan   );
 905                 if (rtspan < 0) {
 906                         GLib.debug("skip(2) file is %d days older than %d months", (int)rtspan, StripApplication.opt_age_oldest);
 907                         return;
 908                 }
 909                 var newest = new  DateTime.now_local();
 910                 newest = newest.add_months(-1 * StripApplication.opt_age_newest);
 911                 rtspan = mailtime.difference(newest) / GLib.TimeSpan.DAY;
 912                 if (rtspan > 0) {
 913                         GLib.debug("skip(2) file is %d days newer than %d months : %s", (int)rtspan, StripApplication.opt_age_newest,
 914                                 mailtime.format("%Y-%m-%d %H:%M:%S"));
 915                         return;
 916                 }
 917
 918
 919
 920                 /*
 921                 GLib.debug("Message DATA:\n mid: %s\nmailfort: %s \nexim_id: %s",
 922                         this.active_message_id,
 923                         this.active_message_x_mailfort_sig,
 924                         this.active_message_exim_id
 925                 );
 926                  */
 927
 928                 // DATE?
 929
 930                 var mp = message.get_mime_part();
 931
 932                 if (!(mp is GMime.Multipart)) {
 933                         //GLib.debug("get mimepart does not return a Multipart?");
 934                 this.used_space_after += file_size;
 935                         return;
 936                 }
 937
 938                 var mpc = ((GMime.Multipart)mp).get_count();
 939
 940                 //GLib.debug("Message has %d parts", mpc);
 941                 for (var i =0 ; i < mpc; i++) {
 942                         //GLib.debug("Getting part %d", i);
 943                         var mime_obj = ((GMime.Multipart)mp).get_part(i);
 944             this.handle_part(mp,mime_obj);
 945         }
 946
 947         parser= null;
 948
 949       //  stream.set_owner(false);
 950             //stream.close();
 951         stream = null;//.close();
 952
 953
 954                 if (!this.has_replaced) {
 955                         this.used_space_after += file_size;
 956                         GLib.debug("skpping write file - no replacement occured");
 957                         return;
 958                 }
 959                 string tmpfile = "";
 960                 GMime.Stream outstream = new GMime.StreamNull();
 961                 if (StripApplication.opt_is_replacing) {
 962
 963                         tmpfile = GLib.Environment.get_tmp_dir() +"/" + name;
 964                 outstream = new GMime.StreamFile.for_path (tmpfile,"w");
 965                 ((GMime.StreamFile)outstream).set_owner(true);
 966         }
 967                 if (StripApplication.opt_dump) {
 968                         outstream = new GMime.StreamMem();
 969         }
 970
 971         file_size = (int) message.write_to_stream(outstream);
 972         if (StripApplication.opt_is_replacing) {
 973                 ((GMime.StreamFile)outstream).set_owner(false);
 974         }
 975                 if (StripApplication.opt_dump) {
 976                         var ua = ((GMime.StreamMem)outstream).get_byte_array().data;
 977                         print("%s\n", (string) ua);
 978                 }
 979         message = null;
 980         outstream.flush();
 981         outstream.close();
 982         GLib.debug("finished writing output %d", file_size);
 983
 984         //
 985         outstream = null;
 986
 987
 988         this.used_space_after += file_size;
 989
 990
 991         if (StripApplication.opt_is_replacing) {
 992                 Posix.unlink(path +"/" + name);
 993                 GLib.debug("copy tmp file %s to %s" , tmpfile, path +"/" + name);
 994
 995                 // link will not work, as we are doing it accross file systems
 996                         var from = File.new_for_path (tmpfile);
 997                         var nf =  File.new_for_path (path +"/" + name);
 998                         from.copy(nf, 0, null);
 999
1000
1001                 var newfileinfo = nf.query_info(GLib.FileAttribute.TIME_MODIFIED,GLib.FileQueryInfoFlags.NONE,null);
1002                 newfileinfo.set_modification_time(mod_time);
1003                 nf.set_attributes_from_info(newfileinfo,FileQueryInfoFlags.NONE);
1004                 Posix.unlink(tmpfile);
1005                 }
1006         this.processed++;
1007
1008         if (StripApplication.opt_limit > -1 && this.processed >= StripApplication.opt_limit) {
1009                 GLib.debug("Reached replacement limit");
1010                 Posix.exit(1);
1011         }
1012
1013
1014
1015
1016     }
1017
1018
1019     public void scan_dir(string basepath, string subpath)
1020     {
1021
1022
1023         // determine if path is to old to scan..
1024         if (subpath.length > 0 && StripApplication.opt_scan_mailfort) {
1025                         var year =  int.parse(subpath.substring(1,4));  // "/2000"
1026                         var month = subpath.length > 5 ? int.parse(subpath.substring(6,2)) : 999; // "/2000/12"
1027                         var day = subpath.length > 8 ? int.parse(subpath.substring(9,2)) : 999; // "/2000/12/01"
1028
1029                 var oldest = new  DateTime.now_local();
1030                         oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
1031
1032                         //GLib.debug("Checking directory %s is older than min: %d/%d/%d", subpath, oldest.get_year() , oldest.get_month(), oldest.get_day_of_month() );
1033
1034                         if (year < oldest.get_year()) {
1035                                 GLib.debug("Skip directory %s is older than min year: %d", subpath, oldest.get_year());
1036                                 return;
1037                         }
1038                         if (year == oldest.get_year() &&  month < oldest.get_month()) {
1039                                 GLib.debug("Skip directory %s is older than min month: %d/%d", subpath, oldest.get_year() , oldest.get_month() );
1040                                 return;
1041                         }
1042                 if (year == oldest.get_year() &&  month == oldest.get_month() && day < oldest.get_day_of_month()) {
1043                                 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() );
1044                                 return;
1045                         }
1046
1047                 var newest = new  DateTime.now_local();
1048                         newest = newest.add_months(-1 * StripApplication.opt_age_newest);
1049
1050                         //GLib.debug("Checking directory %s is newer than max: %d/%d/%d", subpath, newest.get_year() , newest.get_month(), newest.get_day_of_month() );
1051
1052                         if (year > newest.get_year()) {
1053                                 GLib.debug("Skip directory %s is newer than max year: %d", subpath, newest.get_year());
1054                                 return;
1055                         }
1056                         if (year == newest.get_year() &&  month != 999 && month > newest.get_month()) {
1057                                 GLib.debug("Skip directory %s is newer than max month: %d/%d", subpath, newest.get_year() , newest.get_month() );
1058                                 return;
1059                         }
1060                 if (year == newest.get_year() &&  month == newest.get_month() &&  day != 999 && day > newest.get_day_of_month()) {
1061                                 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() );
1062                                 return;
1063                         }
1064
1065
1066
1067         }
1068
1069
1070         var f = File.new_for_path(basepath + subpath);
1071                 FileEnumerator file_enum;
1072         var cancellable = new Cancellable ();
1073         try {
1074             file_enum = f.enumerate_children(
1075                 FileAttribute.STANDARD_DISPLAY_NAME + "," +   FileAttribute.STANDARD_TYPE,
1076                         FileQueryInfoFlags.NOFOLLOW_SYMLINKS,  // FileQueryInfoFlags.NONE,
1077                         cancellable
1078                 );
1079         } catch (Error e) {
1080                 GLib.debug("Got error scanning dir? %s", e.message);
1081             // FIXME - show error..
1082             return;
1083         }
1084         FileInfo next_file;
1085
1086         while (cancellable.is_cancelled () == false ) {
1087             try {
1088                 next_file = file_enum.next_file (cancellable);
1089             } catch(Error e) {
1090                 GLib.debug("error getting next file? %s", e.message);
1091                 break;
1092             }
1093
1094             if (next_file == null) {
1095                 break;
1096             }
1097
1098
1099                 if (next_file.get_is_symlink()) {
1100                 next_file = null;
1101                 continue;
1102             }
1103
1104             var ds = next_file.get_display_name();
1105             if (next_file.get_file_type() != FileType.DIRECTORY) {
1106
1107
1108
1109                 if (ds[0] == ',') {
1110                         continue;
1111                 }
1112                 // other files to ignore?
1113                 if (Regex.match_simple (".tgz$", ds)) {
1114                         continue;
1115                 }
1116                 this.scan_file(basepath + subpath , ds);
1117                                 if(this.has_replaced) {
1118                          this.report_state("After scanning %s/%s".printf(basepath + subpath , ds));
1119                         }
1120                 continue;
1121             }
1122
1123
1124             //stdout.printf("Monitor.monitor: got file %s : type :%u\n",
1125             //        next_file.get_display_name(), next_file.get_file_type());
1126
1127
1128
1129
1130             // not really needed?? - we are storing attachments in a seperate location now...
1131             if (ds[0] == '.') {
1132                 next_file = null;
1133                 continue;
1134             }
1135             if (ds == "attachments") {
1136                         continue;
1137                 }
1138
1139
1140             var sp = subpath+"/"+next_file.get_display_name();
1141             // skip modules.
1142             //print("got a file : " + sp);
1143
1144             next_file = null;
1145
1146
1147             this.scan_dir(basepath,sp);
1148
1149         }
1150
1151
1152     }
1153     void report_state(string msg)
1154     {
1155         // Saved: 2G  Original 10G : 20%
1156         GLib.debug("Saved : %s (%.1f%%) | Original %s | %s",
1157                         GLib.format_size(this.used_space_before - this.used_space_after),
1158                         100f * ((1f * (this.used_space_before - this.used_space_after)) / (this.used_space_before * 1f)),
1159                         GLib.format_size(this.used_space_before),
1160                         msg
1161                 );
1162
1163         }
1164
1165
1166
1167 }