src/addrharvest.c

   1 /*
   2  * Sylpheed -- a GTK+ based, lightweight, and fast e-mail client
   3  * Copyright (C) 2002-2012 Match Grun and the Claws Mail team
   4  *
   5  * This program is free software; you can redistribute it and/or modify
   6  * it under the terms of the GNU General Public License as published by
   7  * the Free Software Foundation; either version 3 of the License, or
   8  * (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  * GNU General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU General Public License
  16  * along with this program. If not, see <http://www.gnu.org/licenses/>.
  17  *
  18  */
  19
  20 /*
  21  * Functions for an E-Mail address harvester.
  22  */
  23
  24 #include <sys/stat.h>
  25 #include <dirent.h>
  26 #include <glib.h>
  27 #include <string.h>
  28
  29 #include "proctypes.h"
  30 #include "utils.h"
  31 #include "mgutils.h"
  32 #include "addrharvest.h"
  33 #include "codeconv.h"
  34 #include "addritem.h"
  35 #ifdef USE_ALT_ADDRBOOK
  36         #include "addressbook-dbus.h"
  37 #endif
  38
  39 /* Mail header names of interest */
  40 static gchar *_headerFrom_     = HEADER_FROM;
  41 static gchar *_headerReplyTo_  = HEADER_REPLY_TO;
  42 static gchar *_headerSender_   = HEADER_SENDER;
  43 static gchar *_headerErrorsTo_ = HEADER_ERRORS_TO;
  44 static gchar *_headerCC_       = HEADER_CC;
  45 static gchar *_headerTo_       = HEADER_TO;
  46
  47 #define ADDR_BUFFSIZE    1024
  48 #define MSG_BUFFSIZE     2048
  49 #define MSGNUM_BUFFSIZE  32
  50 #define DFL_FOLDER_SIZE  20
  51
  52 /* Noise strings included by some other E-Mail clients */
  53 #define REM_NAME_STRING  "(Email)"
  54 #define REM_NAME_STRING2 "(Email 2)"
  55
  56 /* Directories to ignore */
  57 #define DIR_IGNORE ".\t..\t.sylpheed_mark\t.sylpheed_claws_cache"
  58
  59 /*
  60  * Header entry.
  61  */
  62 struct _HeaderEntry {
  63         gchar      *header;
  64         gboolean   selected;
  65         ItemFolder *folder;
  66         gint       count;
  67 };
  68
  69 #ifdef USE_ALT_ADDRBOOK
  70 typedef enum {
  71     FIRST = 0,
  72     LAST,
  73 } Namepart;
  74
  75 #endif
  76
  77 /*
  78  * Build header table entry.
  79  * Enter: harvester Harvester object.
  80  *        name      Header name.
  81  */
  82 static void addrharvest_build_entry(
  83                 AddressHarvester* harvester, gchar *name )
  84 {
  85         HeaderEntry *entry;
  86
  87         entry = g_new0( HeaderEntry, 1 );
  88         entry->header = name;
  89         entry->selected = FALSE;
  90         entry->folder = NULL;
  91         entry->count = 0;
  92         harvester->headerTable = g_list_append( harvester->headerTable, entry );
  93 }
  94
  95 /*
  96  * Free key in table.
  97  */
  98 static gint addrharvest_free_table_vis( gpointer key, gpointer value, gpointer data ) {
  99         g_free( key );
 100         key = NULL;
 101         value = NULL;
 102         return TRUE;
 103 }
 104
 105 /*
 106  * Free lookup table.
 107  */
 108 static void addrharvest_free_table( AddressHarvester* harvester ) {
 109         GList *node;
 110         HeaderEntry *entry;
 111
 112         /* Free header list */
 113         node = harvester->headerTable;
 114         while( node ) {
 115                 entry = ( HeaderEntry * ) node->data;
 116                 entry->header = NULL;
 117                 entry->selected = FALSE;
 118                 entry->folder = NULL;
 119                 entry->count = 0;
 120                 g_free( entry );
 121                 node = g_list_next( node );
 122         }
 123         g_list_free( harvester->headerTable );
 124         harvester->headerTable = NULL;
 125
 126         /* Free duplicate table */
 127         g_hash_table_foreach_remove( harvester->dupTable, addrharvest_free_table_vis, NULL );
 128         g_hash_table_destroy( harvester->dupTable );
 129         harvester->dupTable = NULL;
 130 }
 131
 132 /*
 133 * Create new object.
 134 * Return: Harvester.
 135 */
 136 AddressHarvester *addrharvest_create( void ) {
 137         AddressHarvester *harvester;
 138
 139         harvester = g_new0( AddressHarvester, 1 );
 140         harvester->path = NULL;
 141         harvester->dupTable = g_hash_table_new( g_str_hash, g_str_equal );
 142         harvester->folderSize = DFL_FOLDER_SIZE;
 143         harvester->retVal = MGU_SUCCESS;
 144
 145         /* Build header table */
 146         harvester->headerTable = NULL;
 147         addrharvest_build_entry( harvester, _headerFrom_ );
 148         addrharvest_build_entry( harvester, _headerReplyTo_ );
 149         addrharvest_build_entry( harvester, _headerSender_ );
 150         addrharvest_build_entry( harvester, _headerErrorsTo_ );
 151         addrharvest_build_entry( harvester, _headerCC_ );
 152         addrharvest_build_entry( harvester, _headerTo_ );
 153
 154         return harvester;
 155 }
 156
 157 /*
 158 * Properties...
 159 */
 160 /*
 161  * Specify path to folder that will be harvested.
 162  * Entry: harvester Harvester object.
 163  *        value     Full directory path.
 164  */
 165 void addrharvest_set_path( AddressHarvester* harvester, const gchar *value ) {
 166         cm_return_if_fail( harvester != NULL );
 167         harvester->path = mgu_replace_string( harvester->path, value );
 168         g_strstrip( harvester->path );
 169 }
 170
 171 /*
 172  * Specify maximum folder size.
 173  * Entry: harvester Harvester object.
 174  *        value     Folder size.
 175  */
 176 void addrharvest_set_folder_size(
 177         AddressHarvester* harvester, const gint value )
 178 {
 179         cm_return_if_fail( harvester != NULL );
 180         if( value > 0 ) {
 181                 harvester->folderSize = value;
 182         }
 183 }
 184
 185 /*
 186  * Specify folder recursion.
 187  * Entry: harvester Harvester object.
 188  *        value     TRUE to process sub-folders, FALSE to process folder only.
 189  */
 190 void addrharvest_set_recurse(
 191         AddressHarvester* harvester, const gboolean value )
 192 {
 193         cm_return_if_fail( harvester != NULL );
 194         harvester->folderRecurse = value;
 195 }
 196
 197 /*
 198  * Search (case insensitive) for header entry with specified name.
 199  * Enter: harvester Harvester.
 200  *        name      Header name.
 201  * Return: Header, or NULL if not found.
 202  */
 203 static HeaderEntry *addrharvest_find(
 204         AddressHarvester* harvester, const gchar *name ) {
 205         HeaderEntry *retVal;
 206         GList *node;
 207
 208         retVal = NULL;
 209         node = harvester->headerTable;
 210         while( node ) {
 211                 HeaderEntry *entry;
 212
 213                 entry = node->data;
 214                 if (g_ascii_strncasecmp(entry->header, name,
 215                                         strlen(entry->header)) == 0 ) {
 216                         retVal = entry;
 217                         break;
 218                 }
 219                 node = g_list_next( node );
 220         }
 221         return retVal;
 222 }
 223
 224 /*
 225  * Set selection for specified heaader.
 226  * Enter: harvester Harvester.
 227  *        name      Header name.
 228  *        value     Value to set.
 229  */
 230 void addrharvest_set_header(
 231         AddressHarvester* harvester, const gchar *name, const gboolean value )
 232 {
 233         HeaderEntry *entry;
 234
 235         cm_return_if_fail( harvester != NULL );
 236         entry = addrharvest_find( harvester, name );
 237         if( entry != NULL ) {
 238                 entry->selected = value;
 239         }
 240 }
 241
 242 /*
 243  * Get address count
 244  * Enter: harvester Harvester.
 245  *        name      Header name.
 246  * Return: Address count, or -1 if header not found.
 247  */
 248 gint addrharvest_get_count( AddressHarvester* harvester, const gchar *name ) {
 249         HeaderEntry *entry;
 250         gint count;
 251
 252         count = -1;
 253         cm_return_val_if_fail( harvester != NULL, count );
 254         entry = addrharvest_find( harvester, name );
 255         if( entry != NULL ) {
 256                 count = entry->count;
 257         }
 258         return count;
 259 }
 260
 261 /*
 262 * Free up object by releasing internal memory.
 263 * Enter: harvester Harvester.
 264 */
 265 void addrharvest_free( AddressHarvester *harvester ) {
 266         cm_return_if_fail( harvester != NULL );
 267
 268         /* Free internal stuff */
 269         addrharvest_free_table( harvester );
 270         g_free( harvester->path );
 271
 272         /* Clear pointers */
 273         harvester->path = NULL;
 274         harvester->retVal = MGU_SUCCESS;
 275         harvester->headerTable = NULL;
 276
 277         harvester->folderSize = 0;
 278
 279         /* Now release object */
 280         g_free( harvester );
 281 }
 282
 283 #ifdef USE_ALT_ADDRBOOK
 284 static gchar* get_namepart(const gchar* name, Namepart namepart) {
 285     gchar *pos, *part = NULL;
 286     gchar *token = g_strdup(name);
 287
 288     pos = g_strrstr(token, " ");
 289     if (namepart == FIRST) {
 290         if (pos) {
 291             *pos = '\0';
 292             part = g_strdup(token);
 293             *pos = ' ';
 294         }
 295     }
 296     else {
 297         if (! pos)
 298             part = g_strdup(token);
 299         else {
 300             pos +=1;
 301             part = g_strdup(pos);
 302         }
 303     }
 304     g_free(token);
 305     return part;
 306 }
 307 #endif
 308
 309 /*
 310  * Insert address into cache.
 311  * Enter: harvester Harvester object.
 312  *        entry     Header object.
 313  *        cache     Address cache to load.
 314  *        name      Name.
 315  *        address   eMail address.
 316  */
 317 static void addrharvest_insert_cache(
 318                 AddressHarvester *harvester, HeaderEntry *entry,
 319                 AddressCache *cache, const gchar *name,
 320                 const gchar *address )
 321 {
 322 #ifndef USE_ALT_ADDRBOOK
 323         ItemPerson *person;
 324         ItemFolder *folder;
 325         gchar *folderName;
 326         gboolean newFolder;
 327         gint cnt;
 328         gchar *key, *value;
 329
 330         newFolder = FALSE;
 331         folder = entry->folder;
 332         if( folder == NULL ) {
 333                 newFolder = TRUE;       /* No folder yet */
 334         }
 335         if( entry->count % harvester->folderSize == 0 ) {
 336                 newFolder = TRUE;       /* Folder is full */
 337         }
 338 #else
 339     ContactEntry* person;
 340     gchar* key;
 341 #endif
 342
 343         /* Insert address */
 344         key = g_utf8_strdown( address, -1 );
 345         person = g_hash_table_lookup( harvester->dupTable, key );
 346 #ifndef USE_ALT_ADDRBOOK
 347         if( person ) {
 348                 /* Update existing person to use longest name */
 349                 value = ADDRITEM_NAME(person);
 350                 if( strlen( name ) > strlen( value ) ) {
 351                         addritem_person_set_common_name( person, name );
 352                 }
 353                 g_free( key );
 354         }
 355         else {
 356                 /* Folder if required */
 357                 if( newFolder ) {
 358                         cnt = 1 + ( entry->count / harvester->folderSize );
 359                         folderName =g_strdup_printf( "%s (%d)",
 360                                         entry->header, cnt );
 361                         folder = addritem_create_item_folder();
 362                         addritem_folder_set_name( folder, folderName );
 363                         addritem_folder_set_remarks( folder, "" );
 364                         addrcache_id_folder( cache, folder );
 365                         addrcache_add_folder( cache, folder );
 366                         entry->folder = folder;
 367                         g_free( folderName );
 368                 }
 369
 370                 /* Insert entry */
 371                 person = addrcache_add_contact(
 372                                 cache, folder, name, address, "" );
 373                 g_hash_table_insert( harvester->dupTable, key, person );
 374                 entry->count++;
 375         }
 376         addritem_parse_first_last( person );
 377 #else
 378         if (! person) {
 379                 person = g_new0(ContactEntry, 1);
 380                 person->first_name = get_namepart(name, FIRST);
 381                 person->last_name = get_namepart(name, LAST);
 382                 person->email = g_strdup(address);
 383                 g_hash_table_insert(harvester->dupTable, key, person);
 384                 entry->count++;
 385         }
 386 #endif
 387 }
 388
 389 /*
 390  * Remove specified string from name.
 391  * Enter: name Name.
 392  *        str  String to remove.
 393  */
 394 static void addrharvest_del_email( gchar *name, gchar *str ) {
 395         gchar *p;
 396         gint lenn, lenr;
 397
 398         lenr = strlen( str );
 399         while((p = strcasestr( name, str )) != NULL) {
 400                 lenn = strlen( p );
 401                 memmove( p, p + lenr, lenn );
 402         }
 403 }
 404
 405 /*
 406  * Find position of at (@) character in buffer.
 407  * Enter:  buffer Start of buffer.
 408  * Return: Position of at character, or NULL if not found.
 409  * Note: This function searches for the last occurrence of an 'at' character
 410  * prior to a valid delimiter character for the end of address. This enables
 411  * an address to be found where it is also used as the name of the
 412  * recipient. For example:
 413  *     "axle.rose@netscape.com" <axle.rose@netscape.com>
 414  * The last occurrence of the at character is detected.
 415  */
 416 static gchar *addrharvest_find_at( const gchar *buffer ) {
 417         gchar *atCh;
 418         gchar *p;
 419
 420         atCh = strchr( buffer, '@' );
 421         if( atCh ) {
 422                 /* Search forward for another one */
 423                 p = atCh + 1;
 424                 while( *p ) {
 425                         if( *p == '>' ) {
 426                                 break;
 427                         }
 428                         if( *p == ',' ) {
 429                                 break;
 430                         }
 431                         if( *p == '\n' ) {
 432                                 break;
 433                         }
 434                         if( *p == '@' ) {
 435                                 atCh = p;
 436                                 break;
 437                         }
 438                         p++;
 439                 }
 440         }
 441         return atCh;
 442 }
 443
 444 /*
 445  * Find start and end of address string.
 446  * Enter: buf Start address of buffer to process (not modified).
 447  *        atp Pointer to email at (@) character.
 448  *        bp  Pointer to start of email address (returned).
 449  *        ep  Pointer to end of email address (returned).
 450  */
 451 static void addrharvest_find_address(
 452                 const gchar *buf, const gchar *atp, const gchar **bp,
 453                 const gchar **ep )
 454 {
 455         const gchar *p;
 456
 457         /* Find first non-separator char */
 458         *bp = NULL;
 459         p = buf;
 460         while( TRUE ) {
 461                 if( strchr( ",; \n\r", *p ) == NULL ) break;
 462                 p++;
 463         }
 464         *bp = p;
 465
 466         /* Search forward for end of address */
 467         *ep = NULL;
 468         p = atp + 1;
 469         while( TRUE ) {
 470                 if( strchr( ",;", *p ) ) break;
 471                 p++;
 472         }
 473         *ep = p;
 474 }
 475
 476 /*
 477  * Extract E-Mail address from buffer. If found, address is removed from
 478  * buffer.
 479  * Enter:  buffer Address buffer.
 480  * Return: E-Mail address, or NULL if none found. Must g_free() when done.
 481  */
 482 static gchar *addrharvest_extract_address( gchar *buffer ) {
 483         gchar *addr;
 484         gchar *atCh, *p, *bp, *ep;
 485         gint len;
 486
 487         addr = NULL;
 488         atCh = addrharvest_find_at( buffer );
 489         if( atCh ) {
 490                 /* Search back for start of address */
 491                 bp = NULL;
 492                 p = atCh;
 493                 while( p >= buffer ) {
 494                         bp = p;
 495                         if( *p == '<' ) {
 496                                 *p = ' ';
 497                                 bp++;
 498                                 break;
 499                         }
 500                         p--;
 501                 }
 502
 503                 /* Search fwd for end */
 504                 ep = NULL;
 505                 ep = p = atCh;
 506                 while( *p ) {
 507                         if( *p == '>' ) {
 508                                 *p = ' ';
 509                                 break;
 510                         }
 511                         else if( *p == ' ' ) {
 512                                 break;
 513                         }
 514                         ep = p;
 515                         p++;
 516                 }
 517
 518                 /* Extract email */
 519                 if( bp != NULL ) {
 520                         len = ( ep - bp );
 521                         if( len > 0 ) {
 522                                 addr = g_strndup( bp, len + 1 );
 523                                 memmove( bp, ep, len );
 524                                 *bp = ' ';
 525                         }
 526                 }
 527         }
 528         return addr;
 529 }
 530
 531 /*
 532  * Parse address from header buffer creating address in cache.
 533  * Enter: harvester Harvester object.
 534  *        entry     Header object.
 535  *        cache     Address cache to load.
 536  *        hdrBuf    Pointer to header buffer.
 537  */
 538 static void addrharvest_parse_address(
 539                 AddressHarvester *harvester, HeaderEntry *entry,
 540                 AddressCache *cache, const gchar *hdrBuf )
 541 {
 542         gchar buffer[ ADDR_BUFFSIZE + 2 ];
 543         const gchar *bp;
 544         const gchar *ep;
 545         gchar *atCh, *email, *name;
 546         gint bufLen;
 547
 548         /* Search for an address */
 549         while((atCh = addrharvest_find_at( hdrBuf )) != NULL) {
 550                 /* Find addres string */
 551                 addrharvest_find_address( hdrBuf, atCh, &bp, &ep );
 552
 553                 /* Copy into buffer */
 554                 bufLen = ( size_t ) ( ep - bp );
 555                 if( bufLen > ADDR_BUFFSIZE -1 ) {
 556                         bufLen = ADDR_BUFFSIZE - 1;
 557                 }
 558                 strncpy( buffer, bp, bufLen );
 559                 buffer[ bufLen ] = '\0';
 560                 buffer[ bufLen + 1 ] = '\0';
 561                 buffer[ bufLen + 2 ] = '\0';
 562
 563                 /* Extract address from buffer */
 564                 email = addrharvest_extract_address( buffer );
 565                 if( email ) {
 566                         /* Unescape characters */
 567                         mgu_str_unescape( buffer );
 568
 569                         /* Remove noise characaters */
 570                         addrharvest_del_email( buffer, REM_NAME_STRING );
 571                         addrharvest_del_email( buffer, REM_NAME_STRING2 );
 572
 573                         /* Remove leading trailing quotes and spaces */
 574                         mgu_str_ltc2space( buffer, '\"', '\"' );
 575                         mgu_str_ltc2space( buffer, '\'', '\'' );
 576                         mgu_str_ltc2space( buffer, '\"', '\"' );
 577                         mgu_str_ltc2space( buffer, '(', ')' );
 578                         g_strstrip( buffer );
 579
 580                         if( g_ascii_strcasecmp( buffer, email ) == 0 )
 581                                 name = g_strdup("");
 582                         else
 583                                 name = conv_unmime_header(buffer, NULL, TRUE);
 584
 585                         /* Insert into address book */
 586 #ifndef USE_ALT_ADDRBOOK
 587                         addrharvest_insert_cache(
 588                                 harvester, entry, cache, name, email );
 589 #else
 590                         addrharvest_insert_cache(
 591                                 harvester, entry, NULL, name, email);
 592 #endif
 593                         g_free( email );
 594                         g_free( name );
 595                 }
 596                 hdrBuf = ep;
 597         }
 598 }
 599
 600 /*
 601  * Test whether buffer contains a header that appears in header list.
 602  * Enter: listHdr Header list.
 603  *        buf     Header buffer.
 604  * Return: TRUE if header in list.
 605  */
 606 static gboolean addrharvest_check_hdr( GList *listHdr, gchar *buf ) {
 607         gboolean retVal;
 608         GList *node;
 609         gchar *p, *hdr, *nhdr;
 610         gint len;
 611
 612         retVal = FALSE;
 613         p = strchr( buf, ':' );
 614         if( p ) {
 615                 len = ( size_t ) ( p - buf );
 616                 hdr = g_strndup( buf, len );
 617                 node = listHdr;
 618                 while( node ) {
 619                         nhdr = node->data;
 620                         if (g_ascii_strncasecmp(nhdr, hdr, strlen(nhdr)) == 0 ) {
 621                                 retVal = TRUE;
 622                                 break;
 623                         }
 624                         node = g_list_next( node );
 625                 }
 626                 g_free( hdr );
 627         }
 628         return retVal;
 629 }
 630
 631 /*
 632  * Read header into a linked list of lines.
 633  * Enter:  fp      File to read.
 634  *         listHdr List of header lines of interest.
 635  *         done    End of headers or end of file reached.
 636  * Return: Linked list of lines.
 637  */
 638 static GSList *addrharvest_get_header( FILE *fp, GList *listHdr, gboolean *done ) {
 639         GSList *list;
 640         gchar buf[ MSG_BUFFSIZE + 2 ];
 641         gint ch;
 642         gboolean foundHdr;
 643
 644         list = NULL;
 645
 646         /* Read line */
 647         if( fgets( buf, MSG_BUFFSIZE, fp ) == NULL ) {
 648                 *done = TRUE;
 649                 return list;
 650         }
 651
 652         /* Test for end of headers */
 653         if( buf[0] == '\r' || buf[0] == '\n' ) {
 654                 *done = TRUE;
 655                 return list;
 656         }
 657
 658         /* Test whether required header */
 659         foundHdr = addrharvest_check_hdr( listHdr, buf );
 660
 661         /* Read all header lines. Only add reqd ones to list */
 662         while( TRUE ) {
 663                 gchar *p;
 664
 665                 if( foundHdr ) {
 666                         p = g_strdup( buf );
 667                         list = g_slist_append( list, p );
 668                 }
 669
 670                 /* Read first character */
 671                 ch = fgetc( fp );
 672                 if( ch == ' ' || ch == '\t' ) {
 673                         /* Continuation character - read into buffer */
 674                         if( fgets( buf, MSG_BUFFSIZE, fp ) == NULL ) {
 675                                 break;
 676                         }
 677                 }
 678                 else {
 679                         if( ch == EOF ) {
 680                                 *done = TRUE;
 681                         }
 682                         else {
 683                                 /* Push back character for next header */
 684                                 ungetc( ch, fp );
 685                         }
 686                         break;
 687                 }
 688         }
 689
 690         return list;
 691 }
 692
 693 /*
 694  * Read specified file into address book.
 695  * Enter:  harvester Harvester object.
 696  *         fileName  File to read.
 697  *         cache     Address cache to load.
 698  * Return: Status.
 699  */
 700 static gint addrharvest_readfile(
 701                 AddressHarvester *harvester, const gchar *fileName,
 702                 AddressCache *cache, GList *listHdr )
 703 {
 704         gint retVal;
 705         FILE *msgFile;
 706         gchar *buf, *addr, *p;
 707         HeaderEntry *entry;
 708         GSList *list;
 709         gboolean done;
 710
 711         msgFile = g_fopen( fileName, "rb" );
 712         if( ! msgFile ) {
 713                 /* Cannot open file */
 714                 retVal = MGU_OPEN_FILE;
 715                 return retVal;
 716         }
 717
 718         done = FALSE;
 719         while( TRUE ) {
 720                 list = addrharvest_get_header( msgFile, listHdr, &done );
 721                 if( done ) break;
 722
 723                 if( list == NULL ) {
 724                         continue;
 725                 }
 726
 727                 buf = mgu_list_coalesce( list );
 728                 mgu_free_list( list );
 729
 730                 if(( p = strchr( buf, ':' ) ) != NULL ) {
 731                         addr = p + 1;
 732                         *p = '\0';
 733
 734                         entry = addrharvest_find( harvester, buf );
 735                         if( entry && entry->selected ) {
 736                                 /* Sanitize control characters */
 737                                 p = addr;
 738                                 while( *p ) {
 739                                         if( *p == '\r' || *p == '\n' || *p == '\t' )
 740                                                 *p = ' ';
 741                                         p++;
 742                                 }
 743                                 addrharvest_parse_address(
 744                                         harvester, entry, cache, addr );
 745                         }
 746                 }
 747                 g_free( buf );
 748         }
 749
 750         fclose( msgFile );
 751         return MGU_SUCCESS;
 752 }
 753
 754 /*
 755  * Read all files in specified directory into address book. Directories are
 756  * traversed recursively if necessary.
 757  * Enter:  harvester Harvester object.
 758  *         cache     Address cache to load.
 759  *         msgList   List of message numbers, or NULL to process folder.
 760  *         dir       Directory to process.
 761  */
 762 static void addrharvest_harvest_dir(
 763         AddressHarvester *harvester, AddressCache *cache, GList *listHdr,
 764         gchar *dir )
 765 {
 766         GDir *dp;
 767         const gchar *d;
 768         gchar *fullname;
 769         GError *error = NULL;
 770         gint num;
 771
 772         debug_print("Harvesting addresses from dir '%s'\n", dir);
 773
 774         if( ( dp = g_dir_open( dir, 0, &error ) ) == NULL ) {
 775                 debug_print("opening '%s' failed: %d (%s)\n", dir,
 776                                 error->code, error->message);
 777                 g_error_free(error);
 778                 return;
 779         }
 780
 781         /* Process directory */
 782         while( (d = g_dir_read_name( dp )) != NULL ) {
 783                 fullname = g_strconcat(dir, G_DIR_SEPARATOR_S, d, NULL);
 784                 if( g_file_test(fullname, G_FILE_TEST_IS_DIR) ) {
 785                         if( harvester->folderRecurse ) {
 786                                 if( strstr( DIR_IGNORE, d ) != NULL ) {
 787                                         g_free(fullname);
 788                                         continue;
 789                                 }
 790
 791                                 addrharvest_harvest_dir(
 792                                         harvester, cache, listHdr, (gchar *)fullname );
 793                         }
 794                 }
 795                 if( g_file_test(fullname, G_FILE_TEST_IS_REGULAR) ) {
 796                         if( ( num = to_number( d ) ) >= 0 ) {
 797                                 addrharvest_readfile(
 798                                         harvester, fullname, cache, listHdr );
 799                         }
 800                 }
 801                 g_free(fullname);
 802         }
 803         g_dir_close( dp );
 804 }
 805
 806 /*
 807  * Read list of files in specified directory into address book.
 808  * Enter:  harvester Harvester object.
 809  *         cache     Address cache to load.
 810  *         msgList   List of message numbers, or NULL to process folder.
 811  */
 812 static void addrharvest_harvest_list(
 813         AddressHarvester *harvester, AddressCache *cache, GList *listHdr,
 814         GList *msgList )
 815 {
 816         gint num;
 817         GList *node;
 818         gchar *fullname;
 819
 820         if (!g_file_test(harvester->path, G_FILE_TEST_EXISTS | G_FILE_TEST_IS_DIR)) {
 821                 debug_print("'%s' doesn't exist or is not a dir\n", harvester->path);
 822                 return;
 823         }
 824
 825         /* Process message list */
 826         node = msgList;
 827         while( node ) {
 828                 num = GPOINTER_TO_UINT( node->data );
 829                 fullname = g_strdup_printf("%s%c%d",
 830                                 harvester->path, G_DIR_SEPARATOR, num);
 831                 addrharvest_readfile( harvester, fullname, cache, listHdr );
 832                 g_free(fullname);
 833                 node = g_list_next( node );
 834         }
 835 }
 836
 837 /*
 838  * ============================================================================
 839  * Read all files in specified directory into address book.
 840  * Enter:  harvester Harvester object.
 841  *         cache     Address cache to load.
 842  *         msgList   List of message numbers, or NULL to process folder.
 843  * Return: Status.
 844  * ============================================================================
 845  */
 846 gint addrharvest_harvest(
 847         AddressHarvester *harvester, AddressCache *cache, GList *msgList )
 848 {
 849         gint retVal;
 850         GList *node;
 851         GList *listHdr;
 852
 853         retVal = MGU_BAD_ARGS;
 854         cm_return_val_if_fail( harvester != NULL, retVal );
 855 #ifndef USE_ALT_ADDRBOOK
 856         cm_return_val_if_fail( cache != NULL, retVal );
 857 #endif
 858         cm_return_val_if_fail( harvester->path != NULL, retVal );
 859
 860 #ifndef USE_ALT_ADDRBOOK
 861         /* Clear cache */
 862         addrcache_clear( cache );
 863         cache->dataRead = FALSE;
 864 #endif
 865         /* Build list of headers of interest */
 866         listHdr = NULL;
 867         node = harvester->headerTable;
 868         while( node ) {
 869                 HeaderEntry *entry;
 870
 871                 entry = node->data;
 872                 if( entry->selected ) {
 873                         gchar *p;
 874
 875                         p = g_utf8_strdown( entry->header, -1 );
 876                         listHdr = g_list_append( listHdr, p );
 877                 }
 878                 node = g_list_next( node );
 879         }
 880
 881         /* Process directory/files */
 882         if( msgList == NULL ) {
 883                 addrharvest_harvest_dir( harvester, cache, listHdr, harvester->path );
 884         }
 885         else {
 886                 addrharvest_harvest_list( harvester, cache, listHdr, msgList );
 887         }
 888         mgu_free_dlist( listHdr );
 889
 890 #ifndef USE_ALT_ADDRBOOK
 891         /* Mark cache */
 892         cache->modified = FALSE;
 893         cache->dataRead = TRUE;
 894 #endif
 895         return retVal;
 896 }
 897
 898 /*
 899  * ============================================================================
 900  * Test whether any headers have been selected for processing.
 901  * Enter:  harvester Harvester object.
 902  * Return: TRUE if a header was selected, FALSE if none were selected.
 903  * ============================================================================
 904  */
 905 gboolean addrharvest_check_header( AddressHarvester *harvester ) {
 906         gboolean retVal;
 907         GList *node;
 908
 909         retVal = FALSE;
 910         cm_return_val_if_fail( harvester != NULL, retVal );
 911
 912         node = harvester->headerTable;
 913         while( node ) {
 914                 HeaderEntry *entry;
 915
 916                 entry = ( HeaderEntry * ) node->data;
 917                 if( entry->selected ) return TRUE;
 918                 node = g_list_next( node );
 919         }
 920         return retVal;
 921 }
 922
 923 /*
 924  * ============================================================================
 925  * End of Source.
 926  * ============================================================================
 927  */
 928
 929