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