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         public static bool              opt_stamp = false;
 106
 107         public static bool opt_debug_sql = false;
 108         public static string? opt_replace_link = null;
 109
 110
 111         public const GLib.OptionEntry[] options = {
 112
 113                 { "debug", 0, 0, OptionArg.NONE, ref opt_debug, "show debug messages for components", null },
 114                 { "debug-sql", 0, 0, OptionArg.NONE, ref opt_debug_sql, "debug the SQL statements", null },
 115
 116                 { "path", 0, 0, OptionArg.STRING, ref opt_path, "Directory where email to be parsed is", null },
 117                 { "file", 0, 0, OptionArg.STRING, ref opt_file, "A specific file to be parsed", null },
 118
 119                 { "target-path", 0, 0, OptionArg.STRING, ref opt_target_path, "Directory where attachments are to be put", null },
 120
 121                 { "link", 0, 0, OptionArg.STRING, ref opt_replace_link, "url for the replement link: eg. http://www.mysite.com/xxxx/%s", null },
 122
 123                 { "host", 0, 0, OptionArg.STRING, ref opt_db_host, "Mysql host (default localhost)", null },
 124                 { "name", 0, 0, OptionArg.STRING, ref opt_db_name, "Mysql database name REQUIRED", null },
 125                 { "user", 0, 0, OptionArg.STRING, ref opt_db_user, "Mysql database user REQUIRED", null },
 126                 { "pass", 0, 0, OptionArg.STRING, ref opt_db_pass, "Mysql database password (default empty)", null },
 127
 128                 { "extract", 0, 0, OptionArg.NONE, ref opt_is_extracting, "Should attachments be extracted (default NO)", null },
 129                 { "replace", 0, 0, OptionArg.NONE, ref opt_is_replacing, "Should attachments be replaced (default NO)", null },
 130                 { "dump", 0, 0, OptionArg.NONE, ref opt_dump, "Print the replaced mail contents to stdout", null },
 131
 132                 { "limit", 0, 0, OptionArg.INT, ref opt_limit, "stop after X number of messages with attachments have been processed", null },
 133                 { "newest", 0, 0, OptionArg.INT, ref opt_age_newest, "do not replace messages newer that X months (default is 1 months)", null },
 134                 { "oldest", 0, 0, OptionArg.INT, ref opt_age_oldest, "do not replace messages older than X (default is 6 months)", null },
 135
 136                 { "scan-maildir", 0, 0, OptionArg.NONE, ref opt_scan_maildir, "scan an maildir tree", null },
 137                 { "stamp", 0, 0, OptionArg.NONE, ref opt_stamp, "create and honour directory stamps - flag that a folder has been scanned already", null },
 138                 { "scan-mailfort", 0, 0, OptionArg.NONE, ref opt_scan_mailfort, "scan a mailfort tree", null },
 139                 { "scan-mailfort-only", 0, 0, OptionArg.STRING, ref opt_scan_mailfort_only, "scan a mailfort Year/month eg. /2010/08", null },
 140                 { null }
 141         };
 142     public StripApplication( string[] args )
 143     {
 144                  Object(
 145             application_id: "org.roojs.mailstripper",
 146             flags: ApplicationFlags.FLAGS_NONE
 147          );
 148
 149
 150          var opt_context =  new GLib.OptionContext ("Mail Stripper");
 151
 152          try {
 153
 154             opt_context.set_help_enabled (true);
 155             opt_context.add_main_entries (options, null);
 156             opt_context.parse ( ref  args);
 157             //opt_detach = !optx_no_detach;
 158
 159
 160
 161              // options that have to be set.. bee or hive... (or stop all)
 162             if ((!opt_scan_mailfort && !opt_scan_maildir) || (opt_scan_mailfort && opt_scan_maildir))  {
 163                stdout.printf ("You must specify the type of directory tree to scan - either imap or mailfort\n%s",
 164                    opt_context.get_help(true, null));
 165                GLib.Process.exit(Posix.EXIT_FAILURE);
 166             }
 167
 168                          if ((opt_db_name == null || opt_db_name.length < 1 || opt_db_user == null || opt_db_user.length < 1))  {
 169                stdout.printf ("You must specify the database name / user \n%s",
 170                    opt_context.get_help(true, null));
 171                GLib.Process.exit(Posix.EXIT_FAILURE);
 172             }
 173                          if ((opt_path == null || opt_path.length < 1)   )  {
 174                stdout.printf ("You must specify the scan start path\n%s",
 175                    opt_context.get_help(true, null));
 176                GLib.Process.exit(Posix.EXIT_FAILURE);
 177             }
 178                         if (opt_replace_link == null || (opt_replace_link.length < 1))  {
 179                stdout.printf ("You must specify the link to use in the replacement \n%s",
 180                    opt_context.get_help(true, null));
 181                GLib.Process.exit(Posix.EXIT_FAILURE);
 182             }
 183             if ((opt_is_replacing || opt_is_extracting ) && (opt_target_path == null || opt_target_path.length < 1)) {
 184                       stdout.printf ("You must specify a target path to put attachments\n%s",
 185                    opt_context.get_help(true, null));
 186                GLib.Process.exit(Posix.EXIT_FAILURE);
 187             }
 188
 189
 190          } catch (GLib.OptionError e) {
 191             stdout.printf ("error: %s\n", e.message);
 192             stdout.printf ("Run '%s --help' to see a full list of available command line options.\n%s",
 193                       args[0], opt_context.get_help(true, null));
 194             GLib.Process.exit(Posix.EXIT_FAILURE);
 195          }
 196         }
 197
 198     public static int main(string[] args)
 199     {
 200
 201                 var application = new StripApplication(  args);
 202
 203                 GLib.Log.set_always_fatal(LogLevelFlags.LEVEL_ERROR | LogLevelFlags.LEVEL_CRITICAL);
 204
 205            if (opt_debug || opt_debug_sql) {
 206                         GLib.Log.set_handler(null,
 207                         GLib.LogLevelFlags.LEVEL_DEBUG | GLib.LogLevelFlags.LEVEL_WARNING | GLib.LogLevelFlags.LEVEL_INFO,
 208                         (dom, lvl, msg) => {
 209                                         print("%s\n", msg);
 210                                 }
 211                         );
 212                 }
 213
 214         GMime.init(0);
 215                 if (StripApplication.opt_is_replacing) {
 216                         StripApplication.opt_is_extracting = true;
 217                 }
 218
 219                 GLib.debug("scanning folder: %s", opt_path );
 220
 221                 var strip = new Strip( opt_path );
 222
 223
 224                 strip.mysql  = new Mysql.Database();
 225                 if (!strip.mysql.real_connect(
 226                                 opt_db_host,
 227                                 opt_db_user ,
 228                                 opt_db_pass == null ? "" : opt_db_pass, //passwd
 229                                 opt_db_name, //DB
 230                                 3306, // not changable...?
 231                                 null
 232                         )
 233                 ) {
 234                         stdout.printf("ERROR %u: Connection failed: %s\n",
 235                                 strip.mysql.errno(), strip.mysql.error()
 236                         );
 237
 238                         return 1;
 239                 }
 240         if (opt_file != null) {
 241                 strip.base_dir = opt_path;
 242                 strip.scan_file( GLib.Path.get_dirname(opt_file),  GLib.Path.get_basename(opt_file));
 243                 return 0;
 244         }
 245
 246                 strip.scan_dir(opt_path, opt_scan_mailfort_only);
 247
 248
 249
 250         return 0;
 251     }
 252 }
 253
 254 public class Strip : GLib.Object {
 255
 256
 257
 258         public string base_dir = "";
 259
 260         public Mysql.Database mysql;
 261
 262         int processed = 0;
 263
 264     uint64 used_space_before = 0;
 265     uint64 used_space_after = 0;
 266
 267
 268     public Strip(string base_dir)
 269     {
 270         this.base_dir = base_dir;
 271     }
 272
 273     public void handle_part(GMime.Object parent, GMime.Object mime_obj)
 274     {
 275                 if (mime_obj is GMime.Part) {
 276                    var  p = (GMime.Part)mime_obj;
 277                         var ct = p.get_content_type();
 278                         var cd = p.get_content_disposition();
 279
 280                         var sid = p.get_header("X-strip-id");
 281                     if (sid != null && sid.length > 0) {
 282                         this.update_attachment_db(p);
 283                             GLib.debug("Skip attachment replace - it's already been done");
 284                         return;
 285                         }
 286
 287                         if (cd == null || cd.get_disposition().down() != "attachment") {
 288                                 return;
 289                         }
 290                         if (ct.get_media_type() == "text") {
 291                                 return;
 292                         }
 293                         if (ct.to_string() == "application/pgp-encrypted") {
 294                                 return;
 295                         }
 296                         if (ct.to_string() == "application/pgp-keys") {
 297                                 return;
 298                         }
 299                         if (p.get_filename() == null) {
 300                                 return;
 301                         }
 302                          // print("got part %s\n", ct.to_string());
 303                          if (parent is GMime.Multipart) {
 304
 305                                 this.replace_attachment(((GMime.Multipart)parent), p);
 306                                 // remove it !?
 307
 308                           }
 309
 310
 311                         return;
 312                 }
 313                 if (mime_obj is GMime.Multipart) {
 314
 315
 316                         var  mp = (GMime.Multipart)mime_obj;
 317                         //var ct = mp.get_content_type();
 318
 319                         //print("got multi-part %s\n", ct.to_string());
 320                         for (var i = 0; i< mp.get_count(); i++) {
 321                           var mo = mp.get_part(i);
 322                           this.handle_part(mime_obj,mo);
 323                         }
 324                    // ((GMime.Multipart)mime_obj).foreach((sub_obj) => {
 325                    //     Strip.handle_part(sub_obj);
 326                 //
 327                    // });
 328
 329
 330                         return;
 331                 }
 332
 333                 if (mime_obj is GMime.MessagePart) {
 334                         var msg = ((GMime.MessagePart)mime_obj).get_message();
 335                         msg.foreach((subobj) => {
 336                          this.handle_part(msg,subobj);
 337                     });
 338
 339                         //print("got message-part\n");
 340                         return;
 341                 }
 342
 343                 if (mime_obj is GMime.Message) {
 344                         var mp = ((GMime.Message) mime_obj).get_mime_part();
 345
 346                         if (!(mp is GMime.Multipart)) {
 347                                 //GLib.debug("get mimepart does not return a Multipart?");
 348                                 return;
 349                         }
 350
 351                         var mpc = ((GMime.Multipart)mp).get_count();
 352
 353                         //GLib.debug("Message has %d parts", mpc);
 354                         for (var i =0 ; i < mpc; i++) {
 355                                 //GLib.debug("Getting part %d", i);
 356                                 var submime_obj = ((GMime.Multipart)mp).get_part(i);
 357                         this.handle_part(mp,submime_obj);
 358                     }
 359                         print("got message??\n");
 360                         return;
 361                 }
 362
 363                 print("got something else\n");
 364
 365
 366     }
 367     public void update_attachment_db(GMime.Part attachment)
 368     {
 369         // only called when we have an sid...
 370         var sid = attachment.get_header("X-strip-id");
 371         if (sid == null || sid.length < 1) {
 372                 GLib.debug("Strange - update attachment db called ?");
 373                 return;
 374         }
 375
 376         // initialize it with known data..
 377         // that should wipe out dupes.
 378         var matches = this.query("SELECT count(id)   FROM Attachment WHERE id = %d".printf(
 379                         int.parse(sid)));
 380
 381
 382                 if (matches == "0") {
 383                         // our old mailfort code deleted the crap out of old records...
 384                         // if this occurs we will need to create the record again..
 385                         this.fix_deleted_attachment_db(int.parse(sid),attachment);
 386                         return;
 387
 388                 }
 389
 390
 391         // initialize it with known data..
 392         // that should wipe out dupes.
 393         var filesize = this.query("SELECT filesize FROM Attachment WHERE id = %d".printf(
 394                         int.parse(sid)));
 395
 396                 if (filesize=="") {
 397                    GLib.error("Ignoring record id (missing in database) :%s", sid);
 398                    return;
 399                 }
 400                 if (int.parse(filesize) < 1) {
 401                 GLib.debug("Could not get filesize from id :%s = %s", sid,filesize);
 402                 Posix.exit(0);
 403                 return;
 404         }
 405
 406         var chksum = this.query("SELECT  checksum FROM Attachment WHERE id = %d".printf(
 407                         int.parse(sid)
 408                 ));
 409         var mime_filename = this.query("SELECT  mime_filename FROM Attachment WHERE id = %d".printf(
 410                         int.parse(sid)));
 411
 412         this.query("""
 413              SELECT
 414                  attachment_init(
 415                      '%s', '%s', '%s', %d
 416                  ) as id
 417
 418           """.printf(
 419                           this.mysql_escape(this.active_message_exim_id),
 420                           this.mysql_escape(chksum),
 421                           this.mysql_escape(mime_filename),
 422                           int.parse(filesize)
 423                 ));
 424         this.query("""
 425                  SELECT attachment_update(
 426                       %d, -- in_id INT(11),
 427                       '%s', -- in_mime_type varchar(255),
 428                       '%s', -- in_created DATETIME,
 429                       '%s' -- in_mailfort_sig varchar(64)
 430                  )
 431               """.printf(
 432                         int.parse(sid),
 433                         "", // this will be ignored..
 434                                 this.created_date,
 435                                 this.mysql_escape(this.active_message_x_mailfort_sig)
 436
 437               )
 438                 );
 439                 this.mysql.store_result();
 440
 441
 442
 443     }
 444
 445
 446     public void fix_deleted_attachment_db(int id, GMime.Part attachment)
 447     {
 448
 449         var filename = attachment.get_header("X-strip-content-name");
 450         var file_path  = attachment.get_header("X-strip-path");
 451         var fn =  StripApplication.opt_target_path + "/" + file_path;
 452
 453
 454                 if (!FileUtils.test (fn, FileTest.EXISTS)) {
 455                         GLib.debug("SKIP -- file does not exist");
 456                         return;
 457         }
 458
 459         var chksum = this.md5_file(fn);
 460                 var mime_type = attachment.get_header("X-strip-content-type");
 461
 462                 var fileinfo = File.new_for_path(fn)
 463                                         .query_info(GLib.FileAttribute.STANDARD_SIZE+","+GLib.FileAttribute.TIME_MODIFIED
 464                                                 ,GLib.FileQueryInfoFlags.NONE,null);
 465         var file_size = (int) fileinfo.get_size();
 466
 467
 468                 this.real_query(-1, """
 469
 470
 471                                 INSERT INTO Attachment  (
 472                                         id,
 473
 474                                     msgid ,
 475                                     queue_id ,
 476                                     mime_filename ,
 477                                     mime_type,
 478
 479                                     stored_filename ,
 480                                     mime_charset ,
 481                                     mime_cdisp ,
 482                                     mime_is_cover ,
 483
 484                                     mime_is_multi ,
 485                                     mime_is_mail,
 486                                     mime_size ,
 487                                     filesize,
 488
 489                                     checksum,
 490                                     created
 491
 492                                 ) VALUES (
 493                                         %d,  -- id
 494
 495                                     '%s' , -- msgid
 496                                     0,
 497                                     '%s'  , -- filename
 498                                     '%s',  -- mimetype
 499
 500                                     '%s', -- stored file anme
 501                                     '', -- charset
 502                                     'attachment',
 503                                     0,
 504                                     0,
 505                                     0,
 506                                     %d, -- size
 507                                     %d, -- size
 508
 509                                     '%s', -- checkum
 510                                         '%s' -- created:
 511                                 )
 512
 513
 514                       """.printf(
 515                                 id,
 516                                       this.mysql_escape(this.active_message_exim_id),
 517                                       this.mysql_escape(filename),
 518                                   this.mysql_escape(mime_type),
 519                                   this.mysql_escape(file_path),
 520                                         file_size,
 521                                       file_size,
 522                                       this.mysql_escape(chksum),
 523                                 this.created_date
 524                          ));
 525               // this is done to fix the queue_id or maillog_id ??
 526                  this.query("""
 527                  SELECT attachment_update(
 528                       %d, -- in_id INT(11),
 529                                                 '', -- mime type
 530                       '%s', -- in_created DATETIME,
 531                       '%s' -- in_mailfort_sig varchar(64)
 532                  )
 533               """.printf(
 534                                 id,
 535                                 this.created_date,
 536                                 this.mysql_escape(this.active_message_x_mailfort_sig)
 537
 538               )
 539                 );
 540                 // GLib.error("added attachment?");
 541     }
 542
 543
 544     public void replace_attachment(GMime.Multipart parent, GMime.Part attachment)
 545     {
 546         var sid = attachment.get_header("X-strip-id");
 547         if (sid != null && sid.length > 0) {
 548                 GLib.debug("Skip attachment replace - it's already been done");
 549                 return;
 550         }
 551
 552         var c = attachment.get_content_object();
 553
 554         var filename = attachment.get_filename().replace("/", "-").replace("\n", "").replace("\t", " ");
 555         var fn = GLib.Environment.get_tmp_dir() +
 556                         "/"+ this.active_name + "."+   GLib.Uri.escape_string(filename,"", false);
 557
 558             var outfile = new GMime.StreamFile.for_path(fn, "w");
 559             outfile.set_owner(true);
 560             var file_size = (int) c.write_to_stream(outfile);
 561             var chksum = this.md5_file(fn);
 562             outfile.flush();
 563             outfile = null;
 564
 565         if (file_size == 0) {
 566
 567                 GLib.debug("ERROR - file size of write to stream returned 0?");
 568                 Posix.unlink(fn);
 569                 return;
 570         }
 571
 572
 573
 574
 575         var mime_type= attachment.get_content_type().to_string();
 576         // at this point we have to do our database magic...
 577         //filesize / name / date / checksum / mimetype -- into mailfort should be OK.
 578
 579         var file_id = this.query("""
 580                 SELECT
 581
 582                 attachment_init(
 583                                 '%s', -- in_msgid VARCHAR(32),
 584                                 '%s', -- in_checksum VARCHAR(64),
 585                                 '%s', -- in_mime_filename varchar(255)
 586                                 %d -- filesize
 587                         ) as id
 588
 589           """.printf(
 590                         this.mysql_escape(this.active_message_exim_id),
 591                         chksum,
 592                         this.mysql_escape( GLib.Uri.escape_string(attachment.get_filename(),"", false) ), // what is thsi is invalid?
 593                          file_size)
 594                 );
 595
 596
 597                 if (file_id.length < 1) {
 598                         GLib.debug("ERROR - CALL to attachment_init failed");
 599                 Posix.unlink(fn);
 600                 return;
 601
 602                 }
 603
 604                 if (int.parse(file_id) < 1) {
 605                         GLib.debug("ERROR - CALL to attachment_init failed - returned 0?");
 606                 Posix.unlink(fn);
 607                 return;
 608
 609                 }
 610
 611
 612                 GLib.debug("fn = %s, m5=%s, id= %s", filename, mime_type, this.active_message_id);
 613                 this.query("""
 614
 615                         SELECT attachment_update(
 616                                 %d, -- in_id INT(11),
 617                                 '%s', -- in_mime_type varchar(255),
 618                                 '%s', -- in_created DATETIME,
 619                                 '%s' -- in_mailfort_sig varchar(64)
 620
 621                                 ) as result
 622       """.printf(
 623                 int.parse(file_id),
 624                         this.mysql_escape(mime_type),
 625                         this.created_date,
 626                         this.mysql_escape(this.active_message_x_mailfort_sig)
 627                 ));
 628                  this.mysql.store_result();
 629
 630
 631                 this.used_space_after += file_size;
 632
 633                 var target_fn = "";
 634
 635             if (StripApplication.opt_is_extracting) {
 636                         target_fn = StripApplication.opt_target_path + "/" + this.created_dir +"/"+ file_id  + "-" + GLib.Uri.escape_string(filename,"", false);
 637                 }
 638
 639             var stored =  "/" + this.created_dir +"/"+ file_id  + "-" + GLib.Uri.escape_string(filename,"", false);
 640                  this.query("""
 641
 642                         SELECT attachment_update_store(
 643                                 %d, -- in_id INT(11),
 644                                 '%s'  -- in_store_filename varchar(255),
 645
 646
 647                                 ) as result
 648       """.printf(
 649                 int.parse(file_id),
 650                          this.mysql_escape( stored)
 651                 ));
 652
 653         var rep = new GMime.Part.with_type("text","html");
 654         // we have to set up a redirect server - to redirect hpasite... to their internal service..
 655         rep.set_filename(filename);
 656         string txt = "<html><body>"+
 657             "<a href=\"" + StripApplication.opt_replace_link + "/" +
 658                         file_id + "/" + this.created_dir + "/"+chksum+"/"+ GLib.Uri.escape_string( filename, "", false) +"\">" +
 659             GLib.Uri.escape_string( filename, "", false) + // fixme needs html escaping...
 660             "</a>" +
 661             "</body></html>";
 662
 663         rep.get_content_type().set_parameter("charset", "utf-8");
 664                 rep.set_header("X-strip-id", file_id);
 665                 rep.set_header("X-strip-content-name",  filename);
 666                 rep.set_header("X-strip-path", this.created_dir + "/" + file_id + "-" +  GLib.Uri.escape_string(filename,"", false));
 667                 rep.set_header("X-strip-content-type", mime_type);
 668         var stream =  new GMime.StreamMem.with_buffer(txt.data);
 669         var con = new GMime.DataWrapper.with_stream(stream,GMime.ContentEncoding.DEFAULT);
 670
 671         rep.set_content_object(con);
 672         GLib.debug("Replacing Attachment with HTML");
 673         parent.replace(parent.index_of(attachment), rep);
 674                 this.has_replaced = true;
 675
 676                 if (StripApplication.opt_is_extracting && target_fn.length > 0) {
 677                         var dir = GLib.Path.get_dirname(target_fn);
 678                         if (!FileUtils.test (dir, FileTest.IS_DIR)) {
 679                                 GLib.DirUtils.create_with_parents(dir, 0755);
 680                         }
 681                         GLib.debug("Creating file %s", target_fn);
 682                         if (!FileUtils.test (target_fn, FileTest.EXISTS)) {
 683                                 var from = File.new_for_path (fn);
 684                                 var to =  File.new_for_path (target_fn);
 685                                 from.copy(to, 0, null);
 686
 687                         }
 688                 } else {
 689                         GLib.debug("Skipping extraction %s", target_fn);
 690                 }
 691                 Posix.unlink(fn);
 692
 693
 694
 695     }
 696     public string query(string str)
 697     {
 698             return this.real_query(1, str);
 699     }
 700     public string execute(string str)
 701     {
 702             return this.real_query(0, str);
 703     }
 704     /**
 705     * need_return
 706     0 = no
 707     1 = yes
 708     -1 = don't try.
 709     */
 710     public string real_query(int need_return, string str)
 711     {
 712                 GLib.debug("Before Query : %u  : %s\n", this.mysql.errno(), this.mysql.error());
 713
 714
 715         if (StripApplication.opt_debug_sql) {
 716                 GLib.debug("SQL: %s\n", str);
 717                 }
 718
 719
 720
 721         var rc=  this.mysql.query(str);
 722         if ( rc != 0 ) {
 723
 724                     GLib.debug("ERROR %u: Query failed: %s\n", this.mysql.errno(), this.mysql.error());
 725                                 Posix.exit(1);
 726                 }
 727         var rs = mysql.use_result();
 728                 if (need_return == -1) {
 729                         return "";
 730                 }
 731
 732
 733         //GLib.debug("got %d rows", (int) rs.num_rows());
 734
 735         var got_row = false;
 736                 string[] row;
 737                 string ret = "";
 738                 while( (row = rs.fetch_row()) != null) {
 739                         got_row = true;
 740                         ret = row[0];
 741
 742                 }
 743                 if (need_return == 0) {
 744                 if (StripApplication.opt_debug_sql) {
 745                                 GLib.debug("got %s", got_row ? "=Nothing=" : ret);
 746                         }
 747                         return got_row ? "" : ret;
 748                 }
 749                 if (!got_row) {
 750
 751                          GLib.debug("ERROR : no rows returned");
 752                         Posix.exit(1);
 753                         return "";
 754                 }
 755         if (StripApplication.opt_debug_sql) {
 756                         GLib.debug("got %s", ret);
 757                 }
 758                 return ret;
 759
 760
 761         }
 762
 763     public string mysql_escape(string str)
 764     {
 765             unichar[] value_escaped = new unichar[str.length * 2 + 1];
 766                 this.mysql.real_escape_string ((string) value_escaped, str, str.length);
 767                 return (string) value_escaped;
 768     }
 769
 770     public string  md5_file(string fn) {
 771               Checksum checksum = new Checksum (ChecksumType.MD5);
 772
 773               FileStream stream = FileStream.open (fn, "rb");
 774               uint8 fbuf[100];
 775               size_t size;
 776
 777               while ((size = stream.read (fbuf)) > 0) {
 778                       checksum.update (fbuf, size);
 779               }
 780
 781               unowned string digest = checksum.get_string ();
 782               return digest;
 783     }
 784
 785         string active_path = "";
 786     string active_name = "";
 787     string active_message_id = "";
 788     string active_message_x_mailfort_sig = "";
 789     string active_message_exim_id = "";
 790     bool has_replaced = false;
 791     string created_date = ""; // should be YYYY-mm-dd
 792     string created_dir = ""; // should be YYY/mm/dd
 793
 794     public void scan_file(string path, string name)
 795     {
 796                 GLib.debug("Scan: %s/%s", path,name);
 797
 798                 this.has_replaced = false;
 799         this.active_path = path;
 800         this.active_name = name;
 801         this.active_message_id = "";
 802
 803                 var mailtime = new DateTime.now_local();
 804                 if (StripApplication.opt_scan_mailfort) {
 805                     this.created_dir = this.active_path.substring(this.base_dir.length + 1 );
 806                         this.created_date = this.created_dir.replace("/", "-");
 807                         var bits = this.created_date.split("-");
 808                         mailtime = new DateTime.local(int.parse(bits[0]),int.parse(bits[1]),int.parse(bits[2]),0,0,0);
 809
 810                         var oldest = new  DateTime.now_local();
 811                         oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
 812                         var tspan = mailtime.difference(oldest) / GLib.TimeSpan.DAY;
 813
 814                         if (tspan < 0) {
 815                                 GLib.debug("skip file is %d days older than %d months", (int)tspan, StripApplication.opt_age_oldest);
 816                                 return;
 817                         }
 818
 819                         var newest = new  DateTime.now_local();
 820                         newest = newest.add_months(-1 * StripApplication.opt_age_newest);
 821                         tspan = mailtime.difference(newest) / GLib.TimeSpan.DAY;
 822                         if (tspan > 0) {
 823                                 GLib.debug("skip file is %d days newer than %d months", (int)tspan, StripApplication.opt_age_newest);
 824                                 return;
 825                         }
 826
 827                 }
 828
 829
 830                 var fileinfo = File.new_for_path(path +"/" + name)
 831                                         .query_info(GLib.FileAttribute.STANDARD_SIZE+","+GLib.FileAttribute.TIME_MODIFIED
 832                                                 ,GLib.FileQueryInfoFlags.NONE,null);
 833         var file_size = (int) fileinfo.get_size();
 834                 var mod_time = fileinfo.get_modification_time();
 835
 836
 837
 838                 if (!StripApplication.opt_scan_mailfort) {
 839
 840                 // it's a mail directory...
 841                 // use the last modification time? as the default...
 842                  mailtime = new DateTime.from_timeval_utc(mod_time);
 843                  this.created_dir = mailtime.format("%Y/%m/%d");
 844                          this.created_date =  mailtime.format("%Y-%m-%d %H:%M:%S");
 845
 846         }
 847                 // check on age of file...
 848
 849
 850
 851
 852
 853         this.used_space_before += file_size;
 854
 855         var stream = new GMime.StreamFs.for_path (path +"/" + name,Posix.O_RDONLY, 0);
 856         //stream.set_owner(true);
 857         var parser = new GMime.Parser.with_stream(stream);
 858         var message = parser.construct_message();
 859
 860                 if (message == null) {
 861                         GLib.debug("Could not parse file? %s/%s", path,name);
 862                 this.used_space_after += file_size;
 863                 return;
 864                 }
 865
 866
 867                 // check : - is message over a year old?
 868                 // get various msg info..
 869                 this.active_message_id = message.get_message_id();
 870                 this.active_message_x_mailfort_sig = message.get_header("x-mailfort-sig");
 871                 var recvd = message.get_header("received");
 872                 this.active_message_exim_id = "";
 873                 if (recvd != null && recvd.length > 1) {
 874                         GLib.debug("RECV: %s", recvd);
 875                         var lines = recvd.split("\t");
 876                         for (var i = 0; i < lines.length;i++) {
 877                                 var bits = lines[i].strip().split(" ");
 878                                 if (bits[0] == "id") {
 879                                         this.active_message_exim_id = bits[1].replace(";","");
 880
 881                                 }
 882
 883                                 if (lines[i].contains(";")) {
 884                                         var dbits = lines[i].strip().split(";");
 885                                         GLib.debug("Reading time from : %s", dbits[1]);
 886                                         var timez = GMime.utils_header_decode_date(dbits[1], null);
 887                                         if (timez != 0) {
 888                                                 mailtime = new DateTime.from_unix_utc(timez);
 889                                                 this.created_date = mailtime.format("%Y-%m-%d %H:%M:%S");
 890                                                 GLib.debug("Time is %s",this.created_date);
 891                                                 // if it's not mailfort we can use that date to determine where to store it...
 892                                                 if (!StripApplication.opt_scan_mailfort) {
 893                                                         this.created_dir = mailtime.format("%Y/%m/%d");
 894                                                 }
 895                                         } else {
 896                                                 GLib.debug("Could not read time from headers?");
 897                                         }
 898                                 }
 899
 900                         }
 901                 }
 902
 903                 var oldest = new  DateTime.now_local();
 904                 oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
 905                 var rtspan = mailtime.difference(oldest) / GLib.TimeSpan.DAY;
 906                 GLib.debug("Checking oldest %d days difference", (int)rtspan   );
 907                 if (rtspan < 0) {
 908                         GLib.debug("skip(2) file is %d days older than %d months", (int)rtspan, StripApplication.opt_age_oldest);
 909                         return;
 910                 }
 911                 var newest = new  DateTime.now_local();
 912                 newest = newest.add_months(-1 * StripApplication.opt_age_newest);
 913                 rtspan = mailtime.difference(newest) / GLib.TimeSpan.DAY;
 914                 if (rtspan > 0) {
 915                         GLib.debug("skip(2) file is %d days newer than %d months : %s", (int)rtspan, StripApplication.opt_age_newest,
 916                                 mailtime.format("%Y-%m-%d %H:%M:%S"));
 917                         return;
 918                 }
 919
 920
 921
 922                 /*
 923                 GLib.debug("Message DATA:\n mid: %s\nmailfort: %s \nexim_id: %s",
 924                         this.active_message_id,
 925                         this.active_message_x_mailfort_sig,
 926                         this.active_message_exim_id
 927                 );
 928                  */
 929
 930                 // DATE?
 931
 932                 var mp = message.get_mime_part();
 933
 934                 if (!(mp is GMime.Multipart)) {
 935                         //GLib.debug("get mimepart does not return a Multipart?");
 936                 this.used_space_after += file_size;
 937                         return;
 938                 }
 939
 940                 var mpc = ((GMime.Multipart)mp).get_count();
 941
 942                 //GLib.debug("Message has %d parts", mpc);
 943                 for (var i =0 ; i < mpc; i++) {
 944                         //GLib.debug("Getting part %d", i);
 945                         var mime_obj = ((GMime.Multipart)mp).get_part(i);
 946             this.handle_part(mp,mime_obj);
 947         }
 948
 949         parser= null;
 950
 951       //  stream.set_owner(false);
 952             //stream.close();
 953         stream = null;//.close();
 954
 955
 956                 if (!this.has_replaced) {
 957                         this.used_space_after += file_size;
 958                         GLib.debug("skpping write file - no replacement occured");
 959                         return;
 960                 }
 961                 string tmpfile = "";
 962                 GMime.Stream outstream = new GMime.StreamNull();
 963                 if (StripApplication.opt_is_replacing) {
 964
 965                         tmpfile = GLib.Environment.get_tmp_dir() +"/" + name;
 966                 outstream = new GMime.StreamFile.for_path (tmpfile,"w");
 967                 ((GMime.StreamFile)outstream).set_owner(true);
 968         }
 969                 if (StripApplication.opt_dump) {
 970                         outstream = new GMime.StreamMem();
 971         }
 972
 973         file_size = (int) message.write_to_stream(outstream);
 974         if (StripApplication.opt_is_replacing) {
 975                 ((GMime.StreamFile)outstream).set_owner(false);
 976         }
 977                 if (StripApplication.opt_dump) {
 978                         var ua = ((GMime.StreamMem)outstream).get_byte_array().data;
 979                         print("%s\n", (string) ua);
 980                 }
 981         message = null;
 982         outstream.flush();
 983         outstream.close();
 984         GLib.debug("finished writing output %d", file_size);
 985
 986         //
 987         outstream = null;
 988
 989
 990         this.used_space_after += file_size;
 991
 992
 993         if (StripApplication.opt_is_replacing) {
 994                 Posix.unlink(path +"/" + name);
 995                 GLib.debug("copy tmp file %s to %s" , tmpfile, path +"/" + name);
 996
 997                 // link will not work, as we are doing it accross file systems
 998                         var from = File.new_for_path (tmpfile);
 999                         var nf =  File.new_for_path (path +"/" + name);
1000                         from.copy(nf, 0, null);
1001
1002
1003                 var newfileinfo = nf.query_info(GLib.FileAttribute.TIME_MODIFIED,GLib.FileQueryInfoFlags.NONE,null);
1004                 newfileinfo.set_modification_time(mod_time);
1005                 nf.set_attributes_from_info(newfileinfo,FileQueryInfoFlags.NONE);
1006                 Posix.unlink(tmpfile);
1007                 }
1008         this.processed++;
1009
1010         if (StripApplication.opt_limit > -1 && this.processed >= StripApplication.opt_limit) {
1011                 GLib.debug("Reached replacement limit");
1012                 Posix.exit(1);
1013         }
1014
1015
1016
1017
1018     }
1019
1020
1021     public void scan_dir(string basepath, string subpath)
1022     {
1023
1024
1025         // determine if path is to old to scan..
1026         if (subpath.length > 0 && StripApplication.opt_scan_mailfort) {
1027                         var year =  int.parse(subpath.substring(1,4));  // "/2000"
1028                         var month = subpath.length > 5 ? int.parse(subpath.substring(6,2)) : 999; // "/2000/12"
1029                         var day = subpath.length > 8 ? int.parse(subpath.substring(9,2)) : 999; // "/2000/12/01"
1030
1031                 var oldest = new  DateTime.now_local();
1032                         oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
1033
1034                         //GLib.debug("Checking directory %s is older than min: %d/%d/%d", subpath, oldest.get_year() , oldest.get_month(), oldest.get_day_of_month() );
1035
1036                         if (year < oldest.get_year()) {
1037                                 GLib.debug("Skip directory %s is older than min year: %d", subpath, oldest.get_year());
1038                                 return;
1039                         }
1040                         if (year == oldest.get_year() &&  month < oldest.get_month()) {
1041                                 GLib.debug("Skip directory %s is older than min month: %d/%d", subpath, oldest.get_year() , oldest.get_month() );
1042                                 return;
1043                         }
1044                 if (year == oldest.get_year() &&  month == oldest.get_month() && day < oldest.get_day_of_month()) {
1045                                 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() );
1046                                 return;
1047                         }
1048
1049                 var newest = new  DateTime.now_local();
1050                         newest = newest.add_months(-1 * StripApplication.opt_age_newest);
1051
1052                         //GLib.debug("Checking directory %s is newer than max: %d/%d/%d", subpath, newest.get_year() , newest.get_month(), newest.get_day_of_month() );
1053
1054                         if (year > newest.get_year()) {
1055                                 GLib.debug("Skip directory %s is newer than max year: %d", subpath, newest.get_year());
1056                                 return;
1057                         }
1058                         if (year == newest.get_year() &&  month != 999 && month > newest.get_month()) {
1059                                 GLib.debug("Skip directory %s is newer than max month: %d/%d", subpath, newest.get_year() , newest.get_month() );
1060                                 return;
1061                         }
1062                 if (year == newest.get_year() &&  month == newest.get_month() &&  day != 999 && day > newest.get_day_of_month()) {
1063                                 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() );
1064                                 return;
1065                         }
1066
1067
1068
1069         }
1070
1071         if (GLib.FileUtils.test( basepath + subpath + ".strip-done-stamp", GLib.FileTest.EXISTS )) {
1072                 return;
1073                 }
1074
1075
1076         var f = File.new_for_path(basepath + subpath);
1077                 FileEnumerator file_enum;
1078         var cancellable = new Cancellable ();
1079         try {
1080             file_enum = f.enumerate_children(
1081                 FileAttribute.STANDARD_DISPLAY_NAME + "," +   FileAttribute.STANDARD_TYPE,
1082                         FileQueryInfoFlags.NOFOLLOW_SYMLINKS,  // FileQueryInfoFlags.NONE,
1083                         cancellable
1084                 );
1085         } catch (Error e) {
1086                 GLib.debug("Got error scanning dir? %s", e.message);
1087             // FIXME - show error..
1088             return;
1089         }
1090         FileInfo next_file;
1091
1092         while (cancellable.is_cancelled () == false ) {
1093             try {
1094                 next_file = file_enum.next_file (cancellable);
1095             } catch(Error e) {
1096                 GLib.debug("error getting next file? %s", e.message);
1097                 break;
1098             }
1099
1100             if (next_file == null) {
1101                 break;
1102             }
1103
1104
1105                 if (next_file.get_is_symlink()) {
1106                 next_file = null;
1107                 continue;
1108             }
1109
1110             var ds = next_file.get_display_name();
1111             if (next_file.get_file_type() != FileType.DIRECTORY) {
1112
1113
1114
1115                 if (ds[0] == ',') {
1116                         continue;
1117                 }
1118                 // other files to ignore?
1119                 if (Regex.match_simple (".tgz$", ds)) {
1120                         continue;
1121                 }
1122                 this.scan_file(basepath + subpath , ds);
1123                                 if(this.has_replaced) {
1124                          this.report_state("After scanning %s/%s".printf(basepath + subpath , ds));
1125                         }
1126                 continue;
1127             }
1128
1129
1130             //stdout.printf("Monitor.monitor: got file %s : type :%u\n",
1131             //        next_file.get_display_name(), next_file.get_file_type());
1132
1133
1134
1135
1136             // not really needed?? - we are storing attachments in a seperate location now...
1137             if (ds[0] == '.') {
1138                 next_file = null;
1139                 continue;
1140             }
1141             if (ds == "attachments") {
1142                         continue;
1143                 }
1144
1145
1146             var sp = subpath+"/"+next_file.get_display_name();
1147             // skip modules.
1148             //print("got a file : " + sp);
1149
1150             next_file = null;
1151
1152
1153             this.scan_dir(basepath,sp);
1154
1155         }
1156
1157         // completed this folder
1158
1159         if (opt_stamp) {
1160                 GLib.FileUtils.set_contents (basepath + subpath + ".strip-done-stamp", "Stripper done");
1161         }
1162
1163
1164
1165     }
1166     void report_state(string msg)
1167     {
1168         // Saved: 2G  Original 10G : 20%
1169         GLib.debug("Saved : %s (%.1f%%) | Original %s | %s",
1170                         GLib.format_size(this.used_space_before - this.used_space_after),
1171                         100f * ((1f * (this.used_space_before - this.used_space_after)) / (this.used_space_before * 1f)),
1172                         GLib.format_size(this.used_space_before),
1173                         msg
1174                 );
1175
1176         }
1177
1178
1179
1180 }