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

   1 /**
   2
   3  ** check left to do:
   4   - range scans on maildir
   5   - see how replacing the links works in the resulting email via thunderbird etc..
   6   - some checksum issues (see dupelicates?? suspect 0byte issues?)  -- seems ok now?
   7
   8
   9
  10
  11   needs to scan 2 things
  12   a) our mailfort email database
  13        point it at the top directory, containing YEAR/MONTH/DAY.... directories.
  14        scan each file (over a year old...)
  15        extract out the attachment, and replace with HTML
  16        DATABASE? - mysql or sqlite? -
  17            filesize / name / date / checksum / mimetype -- into mailfort should be OK.
  18   b) the imap user emails
  19            loop through user's directories
  20            check age of email .. over 1 years..
  21            ?? how to prevent 'repeat' scanning of emails?
  22               ??? hidden '.' files containing last scan date?
  23
  24            check if file exists in our DB.. - replace the link...
  25            otherwise generate a file. + add to DB...
  26
  27    c) retreival system
  28      -> URL -> get file
  29    d) redirect system.
  30      -> URL -> redirect to correct server
  31
  32
  33 More notes on our Mailfort DB sync:
  34 * some of these attachments are already in the database...
  35  - so we need to update the DB..
  36  - probably worth putting the code in a stored procedure..
  37
  38  -- key scenarios
  39    * first scan (and extract)
  40    * rescan (as I messed up the first time - fix the DB...)
  41    * email scan - attachments might not have related messages.
  42
  43
  44  - {id} attachment_init(
  45                 {exim_msg_id}
  46                 {chksum}
  47                 {filename),
  48         )
  49         // creates or returns id (can look for existing messages?
  50         // can do a merge?? - copy 'old' record data into 'new'....  "prefer checksummed"
  51
  52         attachment_update(
  53                 {id}
  54                 {exim_msg_id}
  55                 {mailfort_msg_sig}
  56                 {file_size}
  57                 {created} // message date..
  58                 {chksum}
  59                 {filename),
  60         {mime_type}
  61         )
  62         attachment_update_store(
  63                         {id}
  64                         {stored_filename}
  65         )
  66
  67
  68 */
  69
  70 // valac --pkg gmime --vapi
  71 /*
  72
  73 // http://www.fromdual.com/mysql-vala-program-example << check mysql if this does not work.
  74
  75  valac  -g --vapidir=. --thread  strip.vala   --vapidir=../vapi \
  76      --pkg glib-2.0 --pkg mysql --pkg gio-2.0 --pkg posix --pkg gmime-2.6 \
  77       --Xcc=-lmysqlclient  -v \
  78        -o /tmp/strip
  79 */
  80
  81 public class StripApplication : GLib.Application {
  82
  83         public static string? opt_path = null;
  84         public static string? opt_file = null;
  85         public static string? opt_target_path = null;
  86         public static string? opt_db_host = "127.0.0.1";
  87         public static string? opt_db_name = null;
  88         public static string? opt_db_user = null;
  89         public static string? opt_db_pass = null;
  90         public static string? opt_scan_mailfort_only = "";
  91
  92
  93         public static int    opt_limit = -1;
  94
  95         public static int    opt_age_newest = 1;
  96         public static int    opt_age_oldest = 6;
  97
  98
  99         public static bool      opt_is_extracting = false;
 100         public static bool      opt_is_replacing = false;
 101         public static bool      opt_scan_maildir  = false;
 102         public static bool      opt_scan_mailfort  = false;
 103         public static bool              opt_dump = false;
 104         public static bool              opt_debug = false;
 105
 106         public static bool opt_debug_sql = false;
 107         public static string? opt_replace_link = null;
 108
 109
 110         public const GLib.OptionEntry[] options = {
 111
 112                 { "debug", 0, 0, OptionArg.NONE, ref opt_debug, "show debug messages for components", null },
 113                 { "debug-sql", 0, 0, OptionArg.NONE, ref opt_debug_sql, "debug the SQL statements", null },
 114
 115                 { "path", 0, 0, OptionArg.STRING, ref opt_path, "Directory where email to be parsed is", null },
 116                 { "file", 0, 0, OptionArg.STRING, ref opt_file, "A specific file to be parsed", null },
 117
 118                 { "target-path", 0, 0, OptionArg.STRING, ref opt_target_path, "Directory where attachments are to be put", null },
 119
 120                 { "link", 0, 0, OptionArg.STRING, ref opt_replace_link, "url for the replement link: eg. http://www.mysite.com/xxxx/%s", null },
 121
 122                 { "host", 0, 0, OptionArg.STRING, ref opt_db_host, "Mysql host (default localhost)", null },
 123                 { "name", 0, 0, OptionArg.STRING, ref opt_db_name, "Mysql database name REQUIRED", null },
 124                 { "user", 0, 0, OptionArg.STRING, ref opt_db_user, "Mysql database user REQUIRED", null },
 125                 { "pass", 0, 0, OptionArg.STRING, ref opt_db_pass, "Mysql database password (default empty)", null },
 126
 127                 { "extract", 0, 0, OptionArg.NONE, ref opt_is_extracting, "Should attachments be extracted (default NO)", null },
 128                 { "replace", 0, 0, OptionArg.NONE, ref opt_is_replacing, "Should attachments be replaced (default NO)", null },
 129                 { "dump", 0, 0, OptionArg.NONE, ref opt_dump, "Print the replaced mail contents to stdout", null },
 130
 131                 { "limit", 0, 0, OptionArg.INT, ref opt_limit, "stop after X number of messages with attachments have been processed", null },
 132                 { "newest", 0, 0, OptionArg.INT, ref opt_age_newest, "do not replace messages newer that X months (default is 1 months)", null },
 133                 { "oldest", 0, 0, OptionArg.INT, ref opt_age_oldest, "do not replace messages older than X (default is 6 months)", null },
 134
 135                 { "scan-maildir", 0, 0, OptionArg.NONE, ref opt_scan_maildir, "scan an maildir tree", null },
 136                 { "stamp", 0, 0, OptionArg.NONE, ref opt_stamp, "create and honour directory stamps - flag that a folder has been scanned already", null },
 137                 { "scan-mailfort", 0, 0, OptionArg.NONE, ref opt_scan_mailfort, "scan a mailfort tree", null },
 138                 { "scan-mailfort-only", 0, 0, OptionArg.STRING, ref opt_scan_mailfort_only, "scan a mailfort Year/month eg. /2010/08", null },
 139                 { null }
 140         };
 141     public StripApplication( string[] args )
 142     {
 143                  Object(
 144             application_id: "org.roojs.mailstripper",
 145             flags: ApplicationFlags.FLAGS_NONE
 146          );
 147
 148
 149          var opt_context =  new GLib.OptionContext ("Mail Stripper");
 150
 151          try {
 152
 153             opt_context.set_help_enabled (true);
 154             opt_context.add_main_entries (options, null);
 155             opt_context.parse ( ref  args);
 156             //opt_detach = !optx_no_detach;
 157
 158
 159
 160              // options that have to be set.. bee or hive... (or stop all)
 161             if ((!opt_scan_mailfort && !opt_scan_maildir) || (opt_scan_mailfort && opt_scan_maildir))  {
 162                stdout.printf ("You must specify the type of directory tree to scan - either imap or mailfort\n%s",
 163                    opt_context.get_help(true, null));
 164                GLib.Process.exit(Posix.EXIT_FAILURE);
 165             }
 166
 167                          if ((opt_db_name == null || opt_db_name.length < 1 || opt_db_user == null || opt_db_user.length < 1))  {
 168                stdout.printf ("You must specify the database name / user \n%s",
 169                    opt_context.get_help(true, null));
 170                GLib.Process.exit(Posix.EXIT_FAILURE);
 171             }
 172                          if ((opt_path == null || opt_path.length < 1)   )  {
 173                stdout.printf ("You must specify the scan start path\n%s",
 174                    opt_context.get_help(true, null));
 175                GLib.Process.exit(Posix.EXIT_FAILURE);
 176             }
 177                         if (opt_replace_link == null || (opt_replace_link.length < 1))  {
 178                stdout.printf ("You must specify the link to use in the replacement \n%s",
 179                    opt_context.get_help(true, null));
 180                GLib.Process.exit(Posix.EXIT_FAILURE);
 181             }
 182             if ((opt_is_replacing || opt_is_extracting ) && (opt_target_path == null || opt_target_path.length < 1)) {
 183                       stdout.printf ("You must specify a target path to put attachments\n%s",
 184                    opt_context.get_help(true, null));
 185                GLib.Process.exit(Posix.EXIT_FAILURE);
 186             }
 187
 188
 189          } catch (GLib.OptionError e) {
 190             stdout.printf ("error: %s\n", e.message);
 191             stdout.printf ("Run '%s --help' to see a full list of available command line options.\n%s",
 192                       args[0], opt_context.get_help(true, null));
 193             GLib.Process.exit(Posix.EXIT_FAILURE);
 194          }
 195         }
 196
 197     public static int main(string[] args)
 198     {
 199
 200                 var application = new StripApplication(  args);
 201
 202                 GLib.Log.set_always_fatal(LogLevelFlags.LEVEL_ERROR | LogLevelFlags.LEVEL_CRITICAL);
 203
 204            if (opt_debug || opt_debug_sql) {
 205                         GLib.Log.set_handler(null,
 206                         GLib.LogLevelFlags.LEVEL_DEBUG | GLib.LogLevelFlags.LEVEL_WARNING | GLib.LogLevelFlags.LEVEL_INFO,
 207                         (dom, lvl, msg) => {
 208                                         print("%s\n", msg);
 209                                 }
 210                         );
 211                 }
 212
 213         GMime.init(0);
 214                 if (StripApplication.opt_is_replacing) {
 215                         StripApplication.opt_is_extracting = true;
 216                 }
 217
 218                 GLib.debug("scanning folder: %s", opt_path );
 219
 220                 var strip = new Strip( opt_path );
 221
 222
 223                 strip.mysql  = new Mysql.Database();
 224                 if (!strip.mysql.real_connect(
 225                                 opt_db_host,
 226                                 opt_db_user ,
 227                                 opt_db_pass == null ? "" : opt_db_pass, //passwd
 228                                 opt_db_name, //DB
 229                                 3306, // not changable...?
 230                                 null
 231                         )
 232                 ) {
 233                         stdout.printf("ERROR %u: Connection failed: %s\n",
 234                                 strip.mysql.errno(), strip.mysql.error()
 235                         );
 236
 237                         return 1;
 238                 }
 239         if (opt_file != null) {
 240                 strip.base_dir = opt_path;
 241                 strip.scan_file( GLib.Path.get_dirname(opt_file),  GLib.Path.get_basename(opt_file));
 242                 return 0;
 243         }
 244
 245                 strip.scan_dir(opt_path, opt_scan_mailfort_only);
 246
 247
 248
 249         return 0;
 250     }
 251 }
 252
 253 public class Strip : GLib.Object {
 254
 255
 256
 257         public string base_dir = "";
 258
 259         public Mysql.Database mysql;
 260
 261         int processed = 0;
 262
 263     uint64 used_space_before = 0;
 264     uint64 used_space_after = 0;
 265
 266
 267     public Strip(string base_dir)
 268     {
 269         this.base_dir = base_dir;
 270     }
 271
 272     public void handle_part(GMime.Object parent, GMime.Object mime_obj)
 273     {
 274                 if (mime_obj is GMime.Part) {
 275                    var  p = (GMime.Part)mime_obj;
 276                         var ct = p.get_content_type();
 277                         var cd = p.get_content_disposition();
 278
 279                         var sid = p.get_header("X-strip-id");
 280                     if (sid != null && sid.length > 0) {
 281                         this.update_attachment_db(p);
 282                             GLib.debug("Skip attachment replace - it's already been done");
 283                         return;
 284                         }
 285
 286                         if (cd == null || cd.get_disposition().down() != "attachment") {
 287                                 return;
 288                         }
 289                         if (ct.get_media_type() == "text") {
 290                                 return;
 291                         }
 292                         if (ct.to_string() == "application/pgp-encrypted") {
 293                                 return;
 294                         }
 295                         if (ct.to_string() == "application/pgp-keys") {
 296                                 return;
 297                         }
 298                         if (p.get_filename() == null) {
 299                                 return;
 300                         }
 301                          // print("got part %s\n", ct.to_string());
 302                          if (parent is GMime.Multipart) {
 303
 304                                 this.replace_attachment(((GMime.Multipart)parent), p);
 305                                 // remove it !?
 306
 307                           }
 308
 309
 310                         return;
 311                 }
 312                 if (mime_obj is GMime.Multipart) {
 313
 314
 315                         var  mp = (GMime.Multipart)mime_obj;
 316                         //var ct = mp.get_content_type();
 317
 318                         //print("got multi-part %s\n", ct.to_string());
 319                         for (var i = 0; i< mp.get_count(); i++) {
 320                           var mo = mp.get_part(i);
 321                           this.handle_part(mime_obj,mo);
 322                         }
 323                    // ((GMime.Multipart)mime_obj).foreach((sub_obj) => {
 324                    //     Strip.handle_part(sub_obj);
 325                 //
 326                    // });
 327
 328
 329                         return;
 330                 }
 331
 332                 if (mime_obj is GMime.MessagePart) {
 333                         var msg = ((GMime.MessagePart)mime_obj).get_message();
 334                         msg.foreach((subobj) => {
 335                          this.handle_part(msg,subobj);
 336                     });
 337
 338                         //print("got message-part\n");
 339                         return;
 340                 }
 341
 342                 if (mime_obj is GMime.Message) {
 343                         var mp = ((GMime.Message) mime_obj).get_mime_part();
 344
 345                         if (!(mp is GMime.Multipart)) {
 346                                 //GLib.debug("get mimepart does not return a Multipart?");
 347                                 return;
 348                         }
 349
 350                         var mpc = ((GMime.Multipart)mp).get_count();
 351
 352                         //GLib.debug("Message has %d parts", mpc);
 353                         for (var i =0 ; i < mpc; i++) {
 354                                 //GLib.debug("Getting part %d", i);
 355                                 var submime_obj = ((GMime.Multipart)mp).get_part(i);
 356                         this.handle_part(mp,submime_obj);
 357                     }
 358                         print("got message??\n");
 359                         return;
 360                 }
 361
 362                 print("got something else\n");
 363
 364
 365     }
 366     public void update_attachment_db(GMime.Part attachment)
 367     {
 368         // only called when we have an sid...
 369         var sid = attachment.get_header("X-strip-id");
 370         if (sid == null || sid.length < 1) {
 371                 GLib.debug("Strange - update attachment db called ?");
 372                 return;
 373         }
 374
 375         // initialize it with known data..
 376         // that should wipe out dupes.
 377         var matches = this.query("SELECT count(id)   FROM Attachment WHERE id = %d".printf(
 378                         int.parse(sid)));
 379
 380
 381                 if (matches == "0") {
 382                         // our old mailfort code deleted the crap out of old records...
 383                         // if this occurs we will need to create the record again..
 384                         this.fix_deleted_attachment_db(int.parse(sid),attachment);
 385                         return;
 386
 387                 }
 388
 389
 390         // initialize it with known data..
 391         // that should wipe out dupes.
 392         var filesize = this.query("SELECT filesize FROM Attachment WHERE id = %d".printf(
 393                         int.parse(sid)));
 394
 395                 if (filesize=="") {
 396                    GLib.error("Ignoring record id (missing in database) :%s", sid);
 397                    return;
 398                 }
 399                 if (int.parse(filesize) < 1) {
 400                 GLib.debug("Could not get filesize from id :%s = %s", sid,filesize);
 401                 Posix.exit(0);
 402                 return;
 403         }
 404
 405         var chksum = this.query("SELECT  checksum FROM Attachment WHERE id = %d".printf(
 406                         int.parse(sid)
 407                 ));
 408         var mime_filename = this.query("SELECT  mime_filename FROM Attachment WHERE id = %d".printf(
 409                         int.parse(sid)));
 410
 411         this.query("""
 412              SELECT
 413                  attachment_init(
 414                      '%s', '%s', '%s', %d
 415                  ) as id
 416
 417           """.printf(
 418                           this.mysql_escape(this.active_message_exim_id),
 419                           this.mysql_escape(chksum),
 420                           this.mysql_escape(mime_filename),
 421                           int.parse(filesize)
 422                 ));
 423         this.query("""
 424                  SELECT attachment_update(
 425                       %d, -- in_id INT(11),
 426                       '%s', -- in_mime_type varchar(255),
 427                       '%s', -- in_created DATETIME,
 428                       '%s' -- in_mailfort_sig varchar(64)
 429                  )
 430               """.printf(
 431                         int.parse(sid),
 432                         "", // this will be ignored..
 433                                 this.created_date,
 434                                 this.mysql_escape(this.active_message_x_mailfort_sig)
 435
 436               )
 437                 );
 438                 this.mysql.store_result();
 439
 440
 441
 442     }
 443
 444
 445     public void fix_deleted_attachment_db(int id, GMime.Part attachment)
 446     {
 447
 448         var filename = attachment.get_header("X-strip-content-name");
 449         var file_path  = attachment.get_header("X-strip-path");
 450         var fn =  StripApplication.opt_target_path + "/" + file_path;
 451
 452
 453                 if (!FileUtils.test (fn, FileTest.EXISTS)) {
 454                         GLib.debug("SKIP -- file does not exist");
 455                         return;
 456         }
 457
 458         var chksum = this.md5_file(fn);
 459                 var mime_type = attachment.get_header("X-strip-content-type");
 460
 461                 var fileinfo = File.new_for_path(fn)
 462                                         .query_info(GLib.FileAttribute.STANDARD_SIZE+","+GLib.FileAttribute.TIME_MODIFIED
 463                                                 ,GLib.FileQueryInfoFlags.NONE,null);
 464         var file_size = (int) fileinfo.get_size();
 465
 466
 467                 this.real_query(-1, """
 468
 469
 470                                 INSERT INTO Attachment  (
 471                                         id,
 472
 473                                     msgid ,
 474                                     queue_id ,
 475                                     mime_filename ,
 476                                     mime_type,
 477
 478                                     stored_filename ,
 479                                     mime_charset ,
 480                                     mime_cdisp ,
 481                                     mime_is_cover ,
 482
 483                                     mime_is_multi ,
 484                                     mime_is_mail,
 485                                     mime_size ,
 486                                     filesize,
 487
 488                                     checksum,
 489                                     created
 490
 491                                 ) VALUES (
 492                                         %d,  -- id
 493
 494                                     '%s' , -- msgid
 495                                     0,
 496                                     '%s'  , -- filename
 497                                     '%s',  -- mimetype
 498
 499                                     '%s', -- stored file anme
 500                                     '', -- charset
 501                                     'attachment',
 502                                     0,
 503                                     0,
 504                                     0,
 505                                     %d, -- size
 506                                     %d, -- size
 507
 508                                     '%s', -- checkum
 509                                         '%s' -- created:
 510                                 )
 511
 512
 513                       """.printf(
 514                                 id,
 515                                       this.mysql_escape(this.active_message_exim_id),
 516                                       this.mysql_escape(filename),
 517                                   this.mysql_escape(mime_type),
 518                                   this.mysql_escape(file_path),
 519                                         file_size,
 520                                       file_size,
 521                                       this.mysql_escape(chksum),
 522                                 this.created_date
 523                          ));
 524               // this is done to fix the queue_id or maillog_id ??
 525                  this.query("""
 526                  SELECT attachment_update(
 527                       %d, -- in_id INT(11),
 528                                                 '', -- mime type
 529                       '%s', -- in_created DATETIME,
 530                       '%s' -- in_mailfort_sig varchar(64)
 531                  )
 532               """.printf(
 533                                 id,
 534                                 this.created_date,
 535                                 this.mysql_escape(this.active_message_x_mailfort_sig)
 536
 537               )
 538                 );
 539                 // GLib.error("added attachment?");
 540     }
 541
 542
 543     public void replace_attachment(GMime.Multipart parent, GMime.Part attachment)
 544     {
 545         var sid = attachment.get_header("X-strip-id");
 546         if (sid != null && sid.length > 0) {
 547                 GLib.debug("Skip attachment replace - it's already been done");
 548                 return;
 549         }
 550
 551         var c = attachment.get_content_object();
 552
 553         var filename = attachment.get_filename().replace("/", "-").replace("\n", "").replace("\t", " ");
 554         var fn = GLib.Environment.get_tmp_dir() +
 555                         "/"+ this.active_name + "."+   GLib.Uri.escape_string(filename,"", false);
 556
 557             var outfile = new GMime.StreamFile.for_path(fn, "w");
 558             outfile.set_owner(true);
 559             var file_size = (int) c.write_to_stream(outfile);
 560             var chksum = this.md5_file(fn);
 561             outfile.flush();
 562             outfile = null;
 563
 564         if (file_size == 0) {
 565
 566                 GLib.debug("ERROR - file size of write to stream returned 0?");
 567                 Posix.unlink(fn);
 568                 return;
 569         }
 570
 571
 572
 573
 574         var mime_type= attachment.get_content_type().to_string();
 575         // at this point we have to do our database magic...
 576         //filesize / name / date / checksum / mimetype -- into mailfort should be OK.
 577
 578         var file_id = this.query("""
 579                 SELECT
 580
 581                 attachment_init(
 582                                 '%s', -- in_msgid VARCHAR(32),
 583                                 '%s', -- in_checksum VARCHAR(64),
 584                                 '%s', -- in_mime_filename varchar(255)
 585                                 %d -- filesize
 586                         ) as id
 587
 588           """.printf(
 589                         this.mysql_escape(this.active_message_exim_id),
 590                         chksum,
 591                         this.mysql_escape( GLib.Uri.escape_string(attachment.get_filename(),"", false) ), // what is thsi is invalid?
 592                          file_size)
 593                 );
 594
 595
 596                 if (file_id.length < 1) {
 597                         GLib.debug("ERROR - CALL to attachment_init failed");
 598                 Posix.unlink(fn);
 599                 return;
 600
 601                 }
 602
 603                 if (int.parse(file_id) < 1) {
 604                         GLib.debug("ERROR - CALL to attachment_init failed - returned 0?");
 605                 Posix.unlink(fn);
 606                 return;
 607
 608                 }
 609
 610
 611                 GLib.debug("fn = %s, m5=%s, id= %s", filename, mime_type, this.active_message_id);
 612                 this.query("""
 613
 614                         SELECT attachment_update(
 615                                 %d, -- in_id INT(11),
 616                                 '%s', -- in_mime_type varchar(255),
 617                                 '%s', -- in_created DATETIME,
 618                                 '%s' -- in_mailfort_sig varchar(64)
 619
 620                                 ) as result
 621       """.printf(
 622                 int.parse(file_id),
 623                         this.mysql_escape(mime_type),
 624                         this.created_date,
 625                         this.mysql_escape(this.active_message_x_mailfort_sig)
 626                 ));
 627                  this.mysql.store_result();
 628
 629
 630                 this.used_space_after += file_size;
 631
 632                 var target_fn = "";
 633
 634             if (StripApplication.opt_is_extracting) {
 635                         target_fn = StripApplication.opt_target_path + "/" + this.created_dir +"/"+ file_id  + "-" + GLib.Uri.escape_string(filename,"", false);
 636                 }
 637
 638             var stored =  "/" + this.created_dir +"/"+ file_id  + "-" + GLib.Uri.escape_string(filename,"", false);
 639                  this.query("""
 640
 641                         SELECT attachment_update_store(
 642                                 %d, -- in_id INT(11),
 643                                 '%s'  -- in_store_filename varchar(255),
 644
 645
 646                                 ) as result
 647       """.printf(
 648                 int.parse(file_id),
 649                          this.mysql_escape( stored)
 650                 ));
 651
 652         var rep = new GMime.Part.with_type("text","html");
 653         // we have to set up a redirect server - to redirect hpasite... to their internal service..
 654         rep.set_filename(filename);
 655         string txt = "<html><body>"+
 656             "<a href=\"" + StripApplication.opt_replace_link + "/" +
 657                         file_id + "/" + this.created_dir + "/"+chksum+"/"+ GLib.Uri.escape_string( filename, "", false) +"\">" +
 658             GLib.Uri.escape_string( filename, "", false) + // fixme needs html escaping...
 659             "</a>" +
 660             "</body></html>";
 661
 662         rep.get_content_type().set_parameter("charset", "utf-8");
 663                 rep.set_header("X-strip-id", file_id);
 664                 rep.set_header("X-strip-content-name",  filename);
 665                 rep.set_header("X-strip-path", this.created_dir + "/" + file_id + "-" +  GLib.Uri.escape_string(filename,"", false));
 666                 rep.set_header("X-strip-content-type", mime_type);
 667         var stream =  new GMime.StreamMem.with_buffer(txt.data);
 668         var con = new GMime.DataWrapper.with_stream(stream,GMime.ContentEncoding.DEFAULT);
 669
 670         rep.set_content_object(con);
 671         GLib.debug("Replacing Attachment with HTML");
 672         parent.replace(parent.index_of(attachment), rep);
 673                 this.has_replaced = true;
 674
 675                 if (StripApplication.opt_is_extracting && target_fn.length > 0) {
 676                         var dir = GLib.Path.get_dirname(target_fn);
 677                         if (!FileUtils.test (dir, FileTest.IS_DIR)) {
 678                                 GLib.DirUtils.create_with_parents(dir, 0755);
 679                         }
 680                         GLib.debug("Creating file %s", target_fn);
 681                         if (!FileUtils.test (target_fn, FileTest.EXISTS)) {
 682                                 var from = File.new_for_path (fn);
 683                                 var to =  File.new_for_path (target_fn);
 684                                 from.copy(to, 0, null);
 685
 686                         }
 687                 } else {
 688                         GLib.debug("Skipping extraction %s", target_fn);
 689                 }
 690                 Posix.unlink(fn);
 691
 692
 693
 694     }
 695     public string query(string str)
 696     {
 697             return this.real_query(1, str);
 698     }
 699     public string execute(string str)
 700     {
 701             return this.real_query(0, str);
 702     }
 703     /**
 704     * need_return
 705     0 = no
 706     1 = yes
 707     -1 = don't try.
 708     */
 709     public string real_query(int need_return, string str)
 710     {
 711                 GLib.debug("Before Query : %u  : %s\n", this.mysql.errno(), this.mysql.error());
 712
 713
 714         if (StripApplication.opt_debug_sql) {
 715                 GLib.debug("SQL: %s\n", str);
 716                 }
 717
 718
 719
 720         var rc=  this.mysql.query(str);
 721         if ( rc != 0 ) {
 722
 723                     GLib.debug("ERROR %u: Query failed: %s\n", this.mysql.errno(), this.mysql.error());
 724                                 Posix.exit(1);
 725                 }
 726         var rs = mysql.use_result();
 727                 if (need_return == -1) {
 728                         return "";
 729                 }
 730
 731
 732         //GLib.debug("got %d rows", (int) rs.num_rows());
 733
 734         var got_row = false;
 735                 string[] row;
 736                 string ret = "";
 737                 while( (row = rs.fetch_row()) != null) {
 738                         got_row = true;
 739                         ret = row[0];
 740
 741                 }
 742                 if (need_return == 0) {
 743                 if (StripApplication.opt_debug_sql) {
 744                                 GLib.debug("got %s", got_row ? "=Nothing=" : ret);
 745                         }
 746                         return got_row ? "" : ret;
 747                 }
 748                 if (!got_row) {
 749
 750                          GLib.debug("ERROR : no rows returned");
 751                         Posix.exit(1);
 752                         return "";
 753                 }
 754         if (StripApplication.opt_debug_sql) {
 755                         GLib.debug("got %s", ret);
 756                 }
 757                 return ret;
 758
 759
 760         }
 761
 762     public string mysql_escape(string str)
 763     {
 764             unichar[] value_escaped = new unichar[str.length * 2 + 1];
 765                 this.mysql.real_escape_string ((string) value_escaped, str, str.length);
 766                 return (string) value_escaped;
 767     }
 768
 769     public string  md5_file(string fn) {
 770               Checksum checksum = new Checksum (ChecksumType.MD5);
 771
 772               FileStream stream = FileStream.open (fn, "rb");
 773               uint8 fbuf[100];
 774               size_t size;
 775
 776               while ((size = stream.read (fbuf)) > 0) {
 777                       checksum.update (fbuf, size);
 778               }
 779
 780               unowned string digest = checksum.get_string ();
 781               return digest;
 782     }
 783
 784         string active_path = "";
 785     string active_name = "";
 786     string active_message_id = "";
 787     string active_message_x_mailfort_sig = "";
 788     string active_message_exim_id = "";
 789     bool has_replaced = false;
 790     string created_date = ""; // should be YYYY-mm-dd
 791     string created_dir = ""; // should be YYY/mm/dd
 792
 793     public void scan_file(string path, string name)
 794     {
 795                 GLib.debug("Scan: %s/%s", path,name);
 796
 797                 this.has_replaced = false;
 798         this.active_path = path;
 799         this.active_name = name;
 800         this.active_message_id = "";
 801
 802                 var mailtime = new DateTime.now_local();
 803                 if (StripApplication.opt_scan_mailfort) {
 804                     this.created_dir = this.active_path.substring(this.base_dir.length + 1 );
 805                         this.created_date = this.created_dir.replace("/", "-");
 806                         var bits = this.created_date.split("-");
 807                         mailtime = new DateTime.local(int.parse(bits[0]),int.parse(bits[1]),int.parse(bits[2]),0,0,0);
 808
 809                         var oldest = new  DateTime.now_local();
 810                         oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
 811                         var tspan = mailtime.difference(oldest) / GLib.TimeSpan.DAY;
 812
 813                         if (tspan < 0) {
 814                                 GLib.debug("skip file is %d days older than %d months", (int)tspan, StripApplication.opt_age_oldest);
 815                                 return;
 816                         }
 817
 818                         var newest = new  DateTime.now_local();
 819                         newest = newest.add_months(-1 * StripApplication.opt_age_newest);
 820                         tspan = mailtime.difference(newest) / GLib.TimeSpan.DAY;
 821                         if (tspan > 0) {
 822                                 GLib.debug("skip file is %d days newer than %d months", (int)tspan, StripApplication.opt_age_newest);
 823                                 return;
 824                         }
 825
 826                 }
 827
 828
 829                 var fileinfo = File.new_for_path(path +"/" + name)
 830                                         .query_info(GLib.FileAttribute.STANDARD_SIZE+","+GLib.FileAttribute.TIME_MODIFIED
 831                                                 ,GLib.FileQueryInfoFlags.NONE,null);
 832         var file_size = (int) fileinfo.get_size();
 833                 var mod_time = fileinfo.get_modification_time();
 834
 835
 836
 837                 if (!StripApplication.opt_scan_mailfort) {
 838
 839                 // it's a mail directory...
 840                 // use the last modification time? as the default...
 841                  mailtime = new DateTime.from_timeval_utc(mod_time);
 842                  this.created_dir = mailtime.format("%Y/%m/%d");
 843                          this.created_date =  mailtime.format("%Y-%m-%d %H:%M:%S");
 844
 845         }
 846                 // check on age of file...
 847
 848
 849
 850
 851
 852         this.used_space_before += file_size;
 853
 854         var stream = new GMime.StreamFs.for_path (path +"/" + name,Posix.O_RDONLY, 0);
 855         //stream.set_owner(true);
 856         var parser = new GMime.Parser.with_stream(stream);
 857         var message = parser.construct_message();
 858
 859                 if (message == null) {
 860                         GLib.debug("Could not parse file? %s/%s", path,name);
 861                 this.used_space_after += file_size;
 862                 return;
 863                 }
 864
 865
 866                 // check : - is message over a year old?
 867                 // get various msg info..
 868                 this.active_message_id = message.get_message_id();
 869                 this.active_message_x_mailfort_sig = message.get_header("x-mailfort-sig");
 870                 var recvd = message.get_header("received");
 871                 this.active_message_exim_id = "";
 872                 if (recvd != null && recvd.length > 1) {
 873                         GLib.debug("RECV: %s", recvd);
 874                         var lines = recvd.split("\t");
 875                         for (var i = 0; i < lines.length;i++) {
 876                                 var bits = lines[i].strip().split(" ");
 877                                 if (bits[0] == "id") {
 878                                         this.active_message_exim_id = bits[1].replace(";","");
 879
 880                                 }
 881
 882                                 if (lines[i].contains(";")) {
 883                                         var dbits = lines[i].strip().split(";");
 884                                         GLib.debug("Reading time from : %s", dbits[1]);
 885                                         var timez = GMime.utils_header_decode_date(dbits[1], null);
 886                                         if (timez != 0) {
 887                                                 mailtime = new DateTime.from_unix_utc(timez);
 888                                                 this.created_date = mailtime.format("%Y-%m-%d %H:%M:%S");
 889                                                 GLib.debug("Time is %s",this.created_date);
 890                                                 // if it's not mailfort we can use that date to determine where to store it...
 891                                                 if (!StripApplication.opt_scan_mailfort) {
 892                                                         this.created_dir = mailtime.format("%Y/%m/%d");
 893                                                 }
 894                                         } else {
 895                                                 GLib.debug("Could not read time from headers?");
 896                                         }
 897                                 }
 898
 899                         }
 900                 }
 901
 902                 var oldest = new  DateTime.now_local();
 903                 oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
 904                 var rtspan = mailtime.difference(oldest) / GLib.TimeSpan.DAY;
 905                 GLib.debug("Checking oldest %d days difference", (int)rtspan   );
 906                 if (rtspan < 0) {
 907                         GLib.debug("skip(2) file is %d days older than %d months", (int)rtspan, StripApplication.opt_age_oldest);
 908                         return;
 909                 }
 910                 var newest = new  DateTime.now_local();
 911                 newest = newest.add_months(-1 * StripApplication.opt_age_newest);
 912                 rtspan = mailtime.difference(newest) / GLib.TimeSpan.DAY;
 913                 if (rtspan > 0) {
 914                         GLib.debug("skip(2) file is %d days newer than %d months : %s", (int)rtspan, StripApplication.opt_age_newest,
 915                                 mailtime.format("%Y-%m-%d %H:%M:%S"));
 916                         return;
 917                 }
 918
 919
 920
 921                 /*
 922                 GLib.debug("Message DATA:\n mid: %s\nmailfort: %s \nexim_id: %s",
 923                         this.active_message_id,
 924                         this.active_message_x_mailfort_sig,
 925                         this.active_message_exim_id
 926                 );
 927                  */
 928
 929                 // DATE?
 930
 931                 var mp = message.get_mime_part();
 932
 933                 if (!(mp is GMime.Multipart)) {
 934                         //GLib.debug("get mimepart does not return a Multipart?");
 935                 this.used_space_after += file_size;
 936                         return;
 937                 }
 938
 939                 var mpc = ((GMime.Multipart)mp).get_count();
 940
 941                 //GLib.debug("Message has %d parts", mpc);
 942                 for (var i =0 ; i < mpc; i++) {
 943                         //GLib.debug("Getting part %d", i);
 944                         var mime_obj = ((GMime.Multipart)mp).get_part(i);
 945             this.handle_part(mp,mime_obj);
 946         }
 947
 948         parser= null;
 949
 950       //  stream.set_owner(false);
 951             //stream.close();
 952         stream = null;//.close();
 953
 954
 955                 if (!this.has_replaced) {
 956                         this.used_space_after += file_size;
 957                         GLib.debug("skpping write file - no replacement occured");
 958                         return;
 959                 }
 960                 string tmpfile = "";
 961                 GMime.Stream outstream = new GMime.StreamNull();
 962                 if (StripApplication.opt_is_replacing) {
 963
 964                         tmpfile = GLib.Environment.get_tmp_dir() +"/" + name;
 965                 outstream = new GMime.StreamFile.for_path (tmpfile,"w");
 966                 ((GMime.StreamFile)outstream).set_owner(true);
 967         }
 968                 if (StripApplication.opt_dump) {
 969                         outstream = new GMime.StreamMem();
 970         }
 971
 972         file_size = (int) message.write_to_stream(outstream);
 973         if (StripApplication.opt_is_replacing) {
 974                 ((GMime.StreamFile)outstream).set_owner(false);
 975         }
 976                 if (StripApplication.opt_dump) {
 977                         var ua = ((GMime.StreamMem)outstream).get_byte_array().data;
 978                         print("%s\n", (string) ua);
 979                 }
 980         message = null;
 981         outstream.flush();
 982         outstream.close();
 983         GLib.debug("finished writing output %d", file_size);
 984
 985         //
 986         outstream = null;
 987
 988
 989         this.used_space_after += file_size;
 990
 991
 992         if (StripApplication.opt_is_replacing) {
 993                 Posix.unlink(path +"/" + name);
 994                 GLib.debug("copy tmp file %s to %s" , tmpfile, path +"/" + name);
 995
 996                 // link will not work, as we are doing it accross file systems
 997                         var from = File.new_for_path (tmpfile);
 998                         var nf =  File.new_for_path (path +"/" + name);
 999                         from.copy(nf, 0, null);
1000
1001
1002                 var newfileinfo = nf.query_info(GLib.FileAttribute.TIME_MODIFIED,GLib.FileQueryInfoFlags.NONE,null);
1003                 newfileinfo.set_modification_time(mod_time);
1004                 nf.set_attributes_from_info(newfileinfo,FileQueryInfoFlags.NONE);
1005                 Posix.unlink(tmpfile);
1006                 }
1007         this.processed++;
1008
1009         if (StripApplication.opt_limit > -1 && this.processed >= StripApplication.opt_limit) {
1010                 GLib.debug("Reached replacement limit");
1011                 Posix.exit(1);
1012         }
1013
1014
1015
1016
1017     }
1018
1019
1020     public void scan_dir(string basepath, string subpath)
1021     {
1022
1023
1024         // determine if path is to old to scan..
1025         if (subpath.length > 0 && StripApplication.opt_scan_mailfort) {
1026                         var year =  int.parse(subpath.substring(1,4));  // "/2000"
1027                         var month = subpath.length > 5 ? int.parse(subpath.substring(6,2)) : 999; // "/2000/12"
1028                         var day = subpath.length > 8 ? int.parse(subpath.substring(9,2)) : 999; // "/2000/12/01"
1029
1030                 var oldest = new  DateTime.now_local();
1031                         oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
1032
1033                         //GLib.debug("Checking directory %s is older than min: %d/%d/%d", subpath, oldest.get_year() , oldest.get_month(), oldest.get_day_of_month() );
1034
1035                         if (year < oldest.get_year()) {
1036                                 GLib.debug("Skip directory %s is older than min year: %d", subpath, oldest.get_year());
1037                                 return;
1038                         }
1039                         if (year == oldest.get_year() &&  month < oldest.get_month()) {
1040                                 GLib.debug("Skip directory %s is older than min month: %d/%d", subpath, oldest.get_year() , oldest.get_month() );
1041                                 return;
1042                         }
1043                 if (year == oldest.get_year() &&  month == oldest.get_month() && day < oldest.get_day_of_month()) {
1044                                 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() );
1045                                 return;
1046                         }
1047
1048                 var newest = new  DateTime.now_local();
1049                         newest = newest.add_months(-1 * StripApplication.opt_age_newest);
1050
1051                         //GLib.debug("Checking directory %s is newer than max: %d/%d/%d", subpath, newest.get_year() , newest.get_month(), newest.get_day_of_month() );
1052
1053                         if (year > newest.get_year()) {
1054                                 GLib.debug("Skip directory %s is newer than max year: %d", subpath, newest.get_year());
1055                                 return;
1056                         }
1057                         if (year == newest.get_year() &&  month != 999 && month > newest.get_month()) {
1058                                 GLib.debug("Skip directory %s is newer than max month: %d/%d", subpath, newest.get_year() , newest.get_month() );
1059                                 return;
1060                         }
1061                 if (year == newest.get_year() &&  month == newest.get_month() &&  day != 999 && day > newest.get_day_of_month()) {
1062                                 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() );
1063                                 return;
1064                         }
1065
1066
1067
1068         }
1069
1070
1071         var f = File.new_for_path(basepath + subpath);
1072                 FileEnumerator file_enum;
1073         var cancellable = new Cancellable ();
1074         try {
1075             file_enum = f.enumerate_children(
1076                 FileAttribute.STANDARD_DISPLAY_NAME + "," +   FileAttribute.STANDARD_TYPE,
1077                         FileQueryInfoFlags.NOFOLLOW_SYMLINKS,  // FileQueryInfoFlags.NONE,
1078                         cancellable
1079                 );
1080         } catch (Error e) {
1081                 GLib.debug("Got error scanning dir? %s", e.message);
1082             // FIXME - show error..
1083             return;
1084         }
1085         FileInfo next_file;
1086
1087         while (cancellable.is_cancelled () == false ) {
1088             try {
1089                 next_file = file_enum.next_file (cancellable);
1090             } catch(Error e) {
1091                 GLib.debug("error getting next file? %s", e.message);
1092                 break;
1093             }
1094
1095             if (next_file == null) {
1096                 break;
1097             }
1098
1099
1100                 if (next_file.get_is_symlink()) {
1101                 next_file = null;
1102                 continue;
1103             }
1104
1105             var ds = next_file.get_display_name();
1106             if (next_file.get_file_type() != FileType.DIRECTORY) {
1107
1108
1109
1110                 if (ds[0] == ',') {
1111                         continue;
1112                 }
1113                 // other files to ignore?
1114                 if (Regex.match_simple (".tgz$", ds)) {
1115                         continue;
1116                 }
1117                 this.scan_file(basepath + subpath , ds);
1118                                 if(this.has_replaced) {
1119                          this.report_state("After scanning %s/%s".printf(basepath + subpath , ds));
1120                         }
1121                 continue;
1122             }
1123
1124
1125             //stdout.printf("Monitor.monitor: got file %s : type :%u\n",
1126             //        next_file.get_display_name(), next_file.get_file_type());
1127
1128
1129
1130
1131             // not really needed?? - we are storing attachments in a seperate location now...
1132             if (ds[0] == '.') {
1133                 next_file = null;
1134                 continue;
1135             }
1136             if (ds == "attachments") {
1137                         continue;
1138                 }
1139
1140
1141             var sp = subpath+"/"+next_file.get_display_name();
1142             // skip modules.
1143             //print("got a file : " + sp);
1144
1145             next_file = null;
1146
1147
1148             this.scan_dir(basepath,sp);
1149
1150         }
1151
1152
1153     }
1154     void report_state(string msg)
1155     {
1156         // Saved: 2G  Original 10G : 20%
1157         GLib.debug("Saved : %s (%.1f%%) | Original %s | %s",
1158                         GLib.format_size(this.used_space_before - this.used_space_after),
1159                         100f * ((1f * (this.used_space_before - this.used_space_after)) / (this.used_space_before * 1f)),
1160                         GLib.format_size(this.used_space_before),
1161                         msg
1162                 );
1163
1164         }
1165
1166
1167
1168 }