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

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