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

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