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             if (outfile == null) {
 560                 return; // something has gone wrong??
 561         }
 562             outfile.set_owner(true);
 563             var file_size = (int) c.write_to_stream(outfile);
 564             var chksum = this.md5_file(fn);
 565             outfile.flush();
 566             outfile = null;
 567
 568         if (file_size == 0) {
 569
 570                 GLib.debug("ERROR - file size of write to stream returned 0?");
 571                 Posix.unlink(fn);
 572                 return;
 573         }
 574
 575
 576
 577
 578         var mime_type= attachment.get_content_type().to_string();
 579         // at this point we have to do our database magic...
 580         //filesize / name / date / checksum / mimetype -- into mailfort should be OK.
 581
 582         var file_id = this.query("""
 583                 SELECT
 584
 585                 attachment_init(
 586                                 '%s', -- in_msgid VARCHAR(32),
 587                                 '%s', -- in_checksum VARCHAR(64),
 588                                 '%s', -- in_mime_filename varchar(255)
 589                                 %d -- filesize
 590                         ) as id
 591
 592           """.printf(
 593                         this.mysql_escape(this.active_message_exim_id),
 594                         chksum,
 595                         this.mysql_escape( GLib.Uri.escape_string(attachment.get_filename(),"", false) ), // what is thsi is invalid?
 596                          file_size)
 597                 );
 598
 599
 600                 if (file_id.length < 1) {
 601                         GLib.debug("ERROR - CALL to attachment_init failed");
 602                 Posix.unlink(fn);
 603                 return;
 604
 605                 }
 606
 607                 if (int.parse(file_id) < 1) {
 608                         GLib.debug("ERROR - CALL to attachment_init failed - returned 0?");
 609                 Posix.unlink(fn);
 610                 return;
 611
 612                 }
 613
 614
 615                 GLib.debug("fn = %s, m5=%s, id= %s", filename, mime_type, this.active_message_id);
 616                 this.query("""
 617
 618                         SELECT attachment_update(
 619                                 %d, -- in_id INT(11),
 620                                 '%s', -- in_mime_type varchar(255),
 621                                 '%s', -- in_created DATETIME,
 622                                 '%s' -- in_mailfort_sig varchar(64)
 623
 624                                 ) as result
 625       """.printf(
 626                 int.parse(file_id),
 627                         this.mysql_escape(mime_type),
 628                         this.created_date,
 629                         this.mysql_escape(this.active_message_x_mailfort_sig)
 630                 ));
 631                  this.mysql.store_result();
 632
 633
 634                 this.used_space_after += file_size;
 635
 636                 var target_fn = "";
 637
 638             if (StripApplication.opt_is_extracting) {
 639                         target_fn = StripApplication.opt_target_path + "/" + this.created_dir +"/"+ file_id  + "-" + GLib.Uri.escape_string(filename,"", false);
 640                 }
 641
 642             var stored =  "/" + this.created_dir +"/"+ file_id  + "-" + GLib.Uri.escape_string(filename,"", false);
 643                  this.query("""
 644
 645                         SELECT attachment_update_store(
 646                                 %d, -- in_id INT(11),
 647                                 '%s'  -- in_store_filename varchar(255),
 648
 649
 650                                 ) as result
 651       """.printf(
 652                 int.parse(file_id),
 653                          this.mysql_escape( stored)
 654                 ));
 655
 656         var rep = new GMime.Part.with_type("text","html");
 657         // we have to set up a redirect server - to redirect hpasite... to their internal service..
 658         rep.set_filename(filename);
 659         string txt = "<html><body>"+
 660             "<a href=\"" + StripApplication.opt_replace_link + "/" +
 661                         file_id + "/" + this.created_dir + "/"+chksum+"/"+ GLib.Uri.escape_string( filename, "", false) +"\">" +
 662             GLib.Uri.escape_string( filename, "", false) + // fixme needs html escaping...
 663             "</a>" +
 664             "</body></html>";
 665
 666         rep.get_content_type().set_parameter("charset", "utf-8");
 667                 rep.set_header("X-strip-id", file_id);
 668                 rep.set_header("X-strip-content-name",  filename);
 669                 rep.set_header("X-strip-path", this.created_dir + "/" + file_id + "-" +  GLib.Uri.escape_string(filename,"", false));
 670                 rep.set_header("X-strip-content-type", mime_type);
 671         var stream =  new GMime.StreamMem.with_buffer(txt.data);
 672         var con = new GMime.DataWrapper.with_stream(stream,GMime.ContentEncoding.DEFAULT);
 673
 674         rep.set_content_object(con);
 675         GLib.debug("Replacing Attachment with HTML");
 676         parent.replace(parent.index_of(attachment), rep);
 677                 this.has_replaced = true;
 678
 679                 if (StripApplication.opt_is_extracting && target_fn.length > 0) {
 680                         var dir = GLib.Path.get_dirname(target_fn);
 681                         if (!FileUtils.test (dir, FileTest.IS_DIR)) {
 682                                 GLib.DirUtils.create_with_parents(dir, 0755);
 683                         }
 684                         GLib.debug("Creating file %s", target_fn);
 685                         if (!FileUtils.test (target_fn, FileTest.EXISTS)) {
 686                                 var from = File.new_for_path (fn);
 687                                 var to =  File.new_for_path (target_fn);
 688                                 from.copy(to, 0, null);
 689
 690                         }
 691                 } else {
 692                         GLib.debug("Skipping extraction %s", target_fn);
 693                 }
 694                 Posix.unlink(fn);
 695
 696
 697
 698     }
 699     public string query(string str)
 700     {
 701             return this.real_query(1, str);
 702     }
 703     public string execute(string str)
 704     {
 705             return this.real_query(0, str);
 706     }
 707     /**
 708     * need_return
 709     0 = no
 710     1 = yes
 711     -1 = don't try.
 712     */
 713     public string real_query(int need_return, string str)
 714     {
 715                 GLib.debug("Before Query : %u  : %s\n", this.mysql.errno(), this.mysql.error());
 716
 717
 718         if (StripApplication.opt_debug_sql) {
 719                 GLib.debug("SQL: %s\n", str);
 720                 }
 721
 722
 723
 724         var rc=  this.mysql.query(str);
 725         if ( rc != 0 ) {
 726
 727                     GLib.debug("ERROR %u: Query failed: %s\n", this.mysql.errno(), this.mysql.error());
 728                                 Posix.exit(1);
 729                 }
 730         var rs = mysql.use_result();
 731                 if (need_return == -1) {
 732                         return "";
 733                 }
 734
 735
 736         //GLib.debug("got %d rows", (int) rs.num_rows());
 737
 738         var got_row = false;
 739                 string[] row;
 740                 string ret = "";
 741                 while( (row = rs.fetch_row()) != null) {
 742                         got_row = true;
 743                         ret = row[0];
 744
 745                 }
 746                 if (need_return == 0) {
 747                 if (StripApplication.opt_debug_sql) {
 748                                 GLib.debug("got %s", got_row ? "=Nothing=" : ret);
 749                         }
 750                         return got_row ? "" : ret;
 751                 }
 752                 if (!got_row) {
 753
 754                          GLib.debug("ERROR : no rows returned");
 755                         Posix.exit(1);
 756                         return "";
 757                 }
 758         if (StripApplication.opt_debug_sql) {
 759                         GLib.debug("got %s", ret);
 760                 }
 761                 return ret;
 762
 763
 764         }
 765
 766     public string mysql_escape(string str)
 767     {
 768             unichar[] value_escaped = new unichar[str.length * 2 + 1];
 769                 this.mysql.real_escape_string ((string) value_escaped, str, str.length);
 770                 return (string) value_escaped;
 771     }
 772
 773     public string  md5_file(string fn) {
 774               Checksum checksum = new Checksum (ChecksumType.MD5);
 775
 776               FileStream stream = FileStream.open (fn, "rb");
 777               uint8 fbuf[100];
 778               size_t size;
 779
 780               while ((size = stream.read (fbuf)) > 0) {
 781                       checksum.update (fbuf, size);
 782               }
 783
 784               unowned string digest = checksum.get_string ();
 785               return digest;
 786     }
 787
 788         string active_path = "";
 789     string active_name = "";
 790     string active_message_id = "";
 791     string active_message_x_mailfort_sig = "";
 792     string active_message_exim_id = "";
 793     bool has_replaced = false;
 794     string created_date = ""; // should be YYYY-mm-dd
 795     string created_dir = ""; // should be YYY/mm/dd
 796
 797     public void scan_file(string path, string name)
 798     {
 799                 GLib.debug("Scan: %s/%s", path,name);
 800
 801                 this.has_replaced = false;
 802         this.active_path = path;
 803         this.active_name = name;
 804         this.active_message_id = "";
 805
 806                 var mailtime = new DateTime.now_local();
 807                 if (StripApplication.opt_scan_mailfort) {
 808                     this.created_dir = this.active_path.substring(this.base_dir.length + 1 );
 809                         this.created_date = this.created_dir.replace("/", "-");
 810                         var bits = this.created_date.split("-");
 811                         mailtime = new DateTime.local(int.parse(bits[0]),int.parse(bits[1]),int.parse(bits[2]),0,0,0);
 812
 813                         var oldest = new  DateTime.now_local();
 814                         oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
 815                         var tspan = mailtime.difference(oldest) / GLib.TimeSpan.DAY;
 816
 817                         if (tspan < 0) {
 818                                 GLib.debug("skip file is %d days older than %d months", (int)tspan, StripApplication.opt_age_oldest);
 819                                 return;
 820                         }
 821
 822                         var newest = new  DateTime.now_local();
 823                         newest = newest.add_months(-1 * StripApplication.opt_age_newest);
 824                         tspan = mailtime.difference(newest) / GLib.TimeSpan.DAY;
 825                         if (tspan > 0) {
 826                                 GLib.debug("skip file is %d days newer than %d months", (int)tspan, StripApplication.opt_age_newest);
 827                                 return;
 828                         }
 829
 830                 }
 831
 832
 833                 var fileinfo = File.new_for_path(path +"/" + name)
 834                                         .query_info(GLib.FileAttribute.STANDARD_SIZE+","+GLib.FileAttribute.TIME_MODIFIED
 835                                                 ,GLib.FileQueryInfoFlags.NONE,null);
 836         var file_size = (int) fileinfo.get_size();
 837                 var mod_time = fileinfo.get_modification_time();
 838
 839
 840
 841                 if (!StripApplication.opt_scan_mailfort) {
 842
 843                 // it's a mail directory...
 844                 // use the last modification time? as the default...
 845                  mailtime = new DateTime.from_timeval_utc(mod_time);
 846                  this.created_dir = mailtime.format("%Y/%m/%d");
 847                          this.created_date =  mailtime.format("%Y-%m-%d %H:%M:%S");
 848
 849         }
 850                 // check on age of file...
 851
 852
 853
 854
 855
 856         this.used_space_before += file_size;
 857
 858         var stream = new GMime.StreamFs.for_path (path +"/" + name,Posix.O_RDONLY, 0);
 859         //stream.set_owner(true);
 860         var parser = new GMime.Parser.with_stream(stream);
 861         var message = parser.construct_message();
 862
 863                 if (message == null) {
 864                         GLib.debug("Could not parse file? %s/%s", path,name);
 865                 this.used_space_after += file_size;
 866                 return;
 867                 }
 868
 869
 870                 // check : - is message over a year old?
 871                 // get various msg info..
 872                 this.active_message_id = message.get_message_id();
 873                 this.active_message_x_mailfort_sig = message.get_header("x-mailfort-sig");
 874                 var recvd = message.get_header("received");
 875                 this.active_message_exim_id = "";
 876                 if (recvd != null && recvd.length > 1) {
 877                         GLib.debug("RECV: %s", recvd);
 878                         var lines = recvd.split("\t");
 879                         for (var i = 0; i < lines.length;i++) {
 880                                 var bits = lines[i].strip().split(" ");
 881                                 if (bits[0] == "id") {
 882                                         this.active_message_exim_id = bits[1].replace(";","");
 883
 884                                 }
 885
 886                                 if (lines[i].contains(";")) {
 887                                         var dbits = lines[i].strip().split(";");
 888                                         GLib.debug("Reading time from : %s", dbits[1]);
 889                                         var timez = GMime.utils_header_decode_date(dbits[1], null);
 890                                         if (timez != 0) {
 891                                                 mailtime = new DateTime.from_unix_utc(timez);
 892                                                 this.created_date = mailtime.format("%Y-%m-%d %H:%M:%S");
 893                                                 GLib.debug("Time is %s",this.created_date);
 894                                                 // if it's not mailfort we can use that date to determine where to store it...
 895                                                 if (!StripApplication.opt_scan_mailfort) {
 896                                                         this.created_dir = mailtime.format("%Y/%m/%d");
 897                                                 }
 898                                         } else {
 899                                                 GLib.debug("Could not read time from headers?");
 900                                         }
 901                                 }
 902
 903                         }
 904                 }
 905
 906                 var oldest = new  DateTime.now_local();
 907                 oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
 908                 var rtspan = mailtime.difference(oldest) / GLib.TimeSpan.DAY;
 909                 GLib.debug("Checking oldest %d days difference", (int)rtspan   );
 910                 if (rtspan < 0) {
 911                         GLib.debug("skip(2) file is %d days older than %d months", (int)rtspan, StripApplication.opt_age_oldest);
 912                         return;
 913                 }
 914                 var newest = new  DateTime.now_local();
 915                 newest = newest.add_months(-1 * StripApplication.opt_age_newest);
 916                 rtspan = mailtime.difference(newest) / GLib.TimeSpan.DAY;
 917                 if (rtspan > 0) {
 918                         GLib.debug("skip(2) file is %d days newer than %d months : %s", (int)rtspan, StripApplication.opt_age_newest,
 919                                 mailtime.format("%Y-%m-%d %H:%M:%S"));
 920                         return;
 921                 }
 922
 923
 924
 925                 /*
 926                 GLib.debug("Message DATA:\n mid: %s\nmailfort: %s \nexim_id: %s",
 927                         this.active_message_id,
 928                         this.active_message_x_mailfort_sig,
 929                         this.active_message_exim_id
 930                 );
 931                  */
 932
 933                 // DATE?
 934
 935                 var mp = message.get_mime_part();
 936
 937                 if (!(mp is GMime.Multipart)) {
 938                         //GLib.debug("get mimepart does not return a Multipart?");
 939                 this.used_space_after += file_size;
 940                         return;
 941                 }
 942
 943                 var mpc = ((GMime.Multipart)mp).get_count();
 944
 945                 //GLib.debug("Message has %d parts", mpc);
 946                 for (var i =0 ; i < mpc; i++) {
 947                         //GLib.debug("Getting part %d", i);
 948                         var mime_obj = ((GMime.Multipart)mp).get_part(i);
 949             this.handle_part(mp,mime_obj);
 950         }
 951
 952         parser= null;
 953
 954       //  stream.set_owner(false);
 955             //stream.close();
 956         stream = null;//.close();
 957
 958
 959                 if (!this.has_replaced) {
 960                         this.used_space_after += file_size;
 961                         GLib.debug("skpping write file - no replacement occured");
 962                         return;
 963                 }
 964                 string tmpfile = "";
 965                 GMime.Stream outstream = new GMime.StreamNull();
 966                 if (StripApplication.opt_is_replacing) {
 967
 968                         tmpfile = GLib.Environment.get_tmp_dir() +"/" + name;
 969                 outstream = new GMime.StreamFile.for_path (tmpfile,"w");
 970                 ((GMime.StreamFile)outstream).set_owner(true);
 971         }
 972                 if (StripApplication.opt_dump) {
 973                         outstream = new GMime.StreamMem();
 974         }
 975
 976         file_size = (int) message.write_to_stream(outstream);
 977         if (StripApplication.opt_is_replacing) {
 978                 ((GMime.StreamFile)outstream).set_owner(false);
 979         }
 980                 if (StripApplication.opt_dump) {
 981                         var ua = ((GMime.StreamMem)outstream).get_byte_array().data;
 982                         print("%s\n", (string) ua);
 983                 }
 984         message = null;
 985         outstream.flush();
 986         outstream.close();
 987         GLib.debug("finished writing output %d", file_size);
 988
 989         //
 990         outstream = null;
 991
 992
 993         this.used_space_after += file_size;
 994
 995
 996         if (StripApplication.opt_is_replacing) {
 997                 Posix.unlink(path +"/" + name);
 998                 GLib.debug("copy tmp file %s to %s" , tmpfile, path +"/" + name);
 999
1000                 // link will not work, as we are doing it accross file systems
1001                         var from = File.new_for_path (tmpfile);
1002                         var nf =  File.new_for_path (path +"/" + name);
1003                         from.copy(nf, 0, null);
1004
1005
1006                 var newfileinfo = nf.query_info(GLib.FileAttribute.TIME_MODIFIED,GLib.FileQueryInfoFlags.NONE,null);
1007                 newfileinfo.set_modification_time(mod_time);
1008                 nf.set_attributes_from_info(newfileinfo,FileQueryInfoFlags.NONE);
1009                 Posix.unlink(tmpfile);
1010                 }
1011         this.processed++;
1012
1013         if (StripApplication.opt_limit > -1 && this.processed >= StripApplication.opt_limit) {
1014                 GLib.debug("Reached replacement limit");
1015                 Posix.exit(1);
1016         }
1017
1018
1019
1020
1021     }
1022
1023
1024     public void scan_dir(string basepath, string subpath)
1025     {
1026
1027
1028         // determine if path is to old to scan..
1029         if (subpath.length > 0 && StripApplication.opt_scan_mailfort) {
1030                         var year =  int.parse(subpath.substring(1,4));  // "/2000"
1031                         var month = subpath.length > 5 ? int.parse(subpath.substring(6,2)) : 999; // "/2000/12"
1032                         var day = subpath.length > 8 ? int.parse(subpath.substring(9,2)) : 999; // "/2000/12/01"
1033
1034                 var oldest = new  DateTime.now_local();
1035                         oldest = oldest.add_months(-1 * StripApplication.opt_age_oldest);
1036
1037                         //GLib.debug("Checking directory %s is older than min: %d/%d/%d", subpath, oldest.get_year() , oldest.get_month(), oldest.get_day_of_month() );
1038
1039                         if (year < oldest.get_year()) {
1040                                 GLib.debug("Skip directory %s is older than min year: %d", subpath, oldest.get_year());
1041                                 return;
1042                         }
1043                         if (year == oldest.get_year() &&  month < oldest.get_month()) {
1044                                 GLib.debug("Skip directory %s is older than min month: %d/%d", subpath, oldest.get_year() , oldest.get_month() );
1045                                 return;
1046                         }
1047                 if (year == oldest.get_year() &&  month == oldest.get_month() && day < oldest.get_day_of_month()) {
1048                                 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() );
1049                                 return;
1050                         }
1051
1052                 var newest = new  DateTime.now_local();
1053                         newest = newest.add_months(-1 * StripApplication.opt_age_newest);
1054
1055                         //GLib.debug("Checking directory %s is newer than max: %d/%d/%d", subpath, newest.get_year() , newest.get_month(), newest.get_day_of_month() );
1056
1057                         if (year > newest.get_year()) {
1058                                 GLib.debug("Skip directory %s is newer than max year: %d", subpath, newest.get_year());
1059                                 return;
1060                         }
1061                         if (year == newest.get_year() &&  month != 999 && month > newest.get_month()) {
1062                                 GLib.debug("Skip directory %s is newer than max month: %d/%d", subpath, newest.get_year() , newest.get_month() );
1063                                 return;
1064                         }
1065                 if (year == newest.get_year() &&  month == newest.get_month() &&  day != 999 && day > newest.get_day_of_month()) {
1066                                 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() );
1067                                 return;
1068                         }
1069
1070
1071
1072         }
1073
1074         if (StripApplication.opt_stamp && GLib.FileUtils.test( basepath + subpath + ".strip-done-stamp", GLib.FileTest.EXISTS )) {
1075                 return;
1076                 }
1077
1078
1079         var f = File.new_for_path(basepath + subpath);
1080                 FileEnumerator file_enum;
1081         var cancellable = new Cancellable ();
1082         try {
1083             file_enum = f.enumerate_children(
1084                 FileAttribute.STANDARD_DISPLAY_NAME + "," +   FileAttribute.STANDARD_TYPE,
1085                         FileQueryInfoFlags.NOFOLLOW_SYMLINKS,  // FileQueryInfoFlags.NONE,
1086                         cancellable
1087                 );
1088         } catch (Error e) {
1089                 GLib.debug("Got error scanning dir? %s", e.message);
1090             // FIXME - show error..
1091             return;
1092         }
1093         FileInfo next_file;
1094
1095         while (cancellable.is_cancelled () == false ) {
1096             try {
1097                 next_file = file_enum.next_file (cancellable);
1098             } catch(Error e) {
1099                 GLib.debug("error getting next file? %s", e.message);
1100                 break;
1101             }
1102
1103             if (next_file == null) {
1104                 break;
1105             }
1106
1107
1108                 if (next_file.get_is_symlink()) {
1109                 next_file = null;
1110                 continue;
1111             }
1112
1113             var ds = next_file.get_display_name();
1114             if (next_file.get_file_type() != FileType.DIRECTORY) {
1115
1116
1117
1118                 if (ds[0] == ',') {
1119                         continue;
1120                 }
1121                 // other files to ignore?
1122                 if (Regex.match_simple (".tgz$", ds)) {
1123                         continue;
1124                 }
1125                 this.scan_file(basepath + subpath , ds);
1126                                 if(this.has_replaced) {
1127                          this.report_state("After scanning %s/%s".printf(basepath + subpath , ds));
1128                         }
1129                 continue;
1130             }
1131
1132
1133             //stdout.printf("Monitor.monitor: got file %s : type :%u\n",
1134             //        next_file.get_display_name(), next_file.get_file_type());
1135
1136
1137
1138
1139             // not really needed?? - we are storing attachments in a seperate location now...
1140             if (ds[0] == '.') {
1141                 next_file = null;
1142                 continue;
1143             }
1144             if (ds == "attachments") {
1145                         continue;
1146                 }
1147
1148
1149             var sp = subpath+"/"+next_file.get_display_name();
1150             // skip modules.
1151             //print("got a file : " + sp);
1152
1153             next_file = null;
1154
1155
1156             this.scan_dir(basepath,sp);
1157
1158         }
1159
1160         // completed this folder
1161
1162         if (StripApplication.opt_stamp) {
1163                 GLib.FileUtils.set_contents (basepath + subpath + ".strip-done-stamp", "Stripper done");
1164         }
1165
1166
1167
1168     }
1169     void report_state(string msg)
1170     {
1171         // Saved: 2G  Original 10G : 20%
1172         GLib.debug("Saved : %s (%.1f%%) | Original %s | %s",
1173                         GLib.format_size(this.used_space_before - this.used_space_after),
1174                         100f * ((1f * (this.used_space_before - this.used_space_after)) / (this.used_space_before * 1f)),
1175                         GLib.format_size(this.used_space_before),
1176                         msg
1177                 );
1178
1179         }
1180
1181
1182
1183 }