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 #ifdef HAVE_CONFIG_H
  25 #  include "config.h"
  26 #include "claws-features.h"
  27 #endif
  28
  29 #include <sys/stat.h>
  30 #include <dirent.h>
  31 #include <glib.h>
  32 #include <string.h>
  33
  34 #include "proctypes.h"
  35 #include "utils.h"
  36 #include "mgutils.h"
  37 #include "addrharvest.h"
  38 #include "codeconv.h"
  39 #include "addritem.h"
  40 #ifdef USE_ALT_ADDRBOOK
  41         #include "addressbook-dbus.h"
  42 #endif
  43 #include "file-utils.h"
  44
  45 /* Mail header names of interest */
  46 static gchar *_headerFrom_     = HEADER_FROM;
  47 static gchar *_headerReplyTo_  = HEADER_REPLY_TO;
  48 static gchar *_headerSender_   = HEADER_SENDER;
  49 static gchar *_headerErrorsTo_ = HEADER_ERRORS_TO;
  50 static gchar *_headerCC_       = HEADER_CC;
  51 static gchar *_headerTo_       = HEADER_TO;
  52
  53 #define ADDR_BUFFSIZE    1024
  54 #define MSG_BUFFSIZE     2048
  55 #define MSGNUM_BUFFSIZE  32
  56 #define DFL_FOLDER_SIZE  20
  57
  58 /* Noise strings included by some other E-Mail clients */
  59 #define REM_NAME_STRING  "(Email)"
  60 #define REM_NAME_STRING2 "(Email 2)"
  61
  62 /* Directories to ignore */
  63 #define DIR_IGNORE ".\t.."
  64
  65 /*
  66  * Header entry.
  67  */
  68 struct _HeaderEntry {
  69         gchar      *header;
  70         gboolean   selected;
  71         ItemFolder *folder;
  72         gint       count;
  73 };
  74
  75 #ifdef USE_ALT_ADDRBOOK
  76 typedef enum {
  77     FIRST = 0,
  78     LAST,
  79 } Namepart;
  80
  81 #endif
  82
  83 /*
  84  * Build header table entry.
  85  * Enter: harvester Harvester object.
  86  *        name      Header name.
  87  */
  88 static void addrharvest_build_entry(
  89                 AddressHarvester* harvester, gchar *name )
  90 {
  91         HeaderEntry *entry;
  92
  93         entry = g_new0( HeaderEntry, 1 );
  94         entry->header = name;
  95         entry->selected = FALSE;
  96         entry->folder = NULL;
  97         entry->count = 0;
  98         harvester->headerTable = g_list_append( harvester->headerTable, entry );
  99 }
 100
 101 /*
 102  * Free key in table.
 103  */
 104 static gint addrharvest_free_table_vis( gpointer key, gpointer value, gpointer data ) {
 105         g_free( key );
 106         key = NULL;
 107         value = NULL;
 108         return TRUE;
 109 }
 110
 111 /*
 112  * Free lookup table.
 113  */
 114 static void addrharvest_free_table( AddressHarvester* harvester ) {
 115         GList *node;
 116         HeaderEntry *entry;
 117
 118         /* Free header list */
 119         node = harvester->headerTable;
 120         while( node ) {
 121                 entry = ( HeaderEntry * ) node->data;
 122                 entry->header = NULL;
 123                 entry->selected = FALSE;
 124                 entry->folder = NULL;
 125                 entry->count = 0;
 126                 g_free( entry );
 127                 node = g_list_next( node );
 128         }
 129         g_list_free( harvester->headerTable );
 130         harvester->headerTable = NULL;
 131
 132         /* Free duplicate table */
 133         g_hash_table_foreach_remove( harvester->dupTable, addrharvest_free_table_vis, NULL );
 134         g_hash_table_destroy( harvester->dupTable );
 135         harvester->dupTable = NULL;
 136 }
 137
 138 /*
 139 * Create new object.
 140 * Return: Harvester.
 141 */
 142 AddressHarvester *addrharvest_create( void ) {
 143         AddressHarvester *harvester;
 144
 145         harvester = g_new0( AddressHarvester, 1 );
 146         harvester->path = NULL;
 147         harvester->dupTable = g_hash_table_new( g_str_hash, g_str_equal );
 148         harvester->folderSize = DFL_FOLDER_SIZE;
 149         harvester->retVal = MGU_SUCCESS;
 150
 151         /* Build header table */
 152         harvester->headerTable = NULL;
 153         addrharvest_build_entry( harvester, _headerFrom_ );
 154         addrharvest_build_entry( harvester, _headerReplyTo_ );
 155         addrharvest_build_entry( harvester, _headerSender_ );
 156         addrharvest_build_entry( harvester, _headerErrorsTo_ );
 157         addrharvest_build_entry( harvester, _headerCC_ );
 158         addrharvest_build_entry( harvester, _headerTo_ );
 159
 160         return harvester;
 161 }
 162
 163 /*
 164 * Properties...
 165 */
 166 /*
 167  * Specify path to folder that will be harvested.
 168  * Entry: harvester Harvester object.
 169  *        value     Full directory path.
 170  */
 171 void addrharvest_set_path( AddressHarvester* harvester, const gchar *value ) {
 172         cm_return_if_fail( harvester != NULL );
 173         harvester->path = mgu_replace_string( harvester->path, value );
 174         g_strstrip( harvester->path );
 175 }
 176
 177 /*
 178  * Specify maximum folder size.
 179  * Entry: harvester Harvester object.
 180  *        value     Folder size.
 181  */
 182 void addrharvest_set_folder_size(
 183         AddressHarvester* harvester, const gint value )
 184 {
 185         cm_return_if_fail( harvester != NULL );
 186         if( value > 0 ) {
 187                 harvester->folderSize = value;
 188         }
 189 }
 190
 191 /*
 192  * Specify folder recursion.
 193  * Entry: harvester Harvester object.
 194  *        value     TRUE to process sub-folders, FALSE to process folder only.
 195  */
 196 void addrharvest_set_recurse(
 197         AddressHarvester* harvester, const gboolean value )
 198 {
 199         cm_return_if_fail( harvester != NULL );
 200         harvester->folderRecurse = value;
 201 }
 202
 203 /*
 204  * Search (case insensitive) for header entry with specified name.
 205  * Enter: harvester Harvester.
 206  *        name      Header name.
 207  * Return: Header, or NULL if not found.
 208  */
 209 static HeaderEntry *addrharvest_find(
 210         AddressHarvester* harvester, const gchar *name ) {
 211         HeaderEntry *retVal;
 212         GList *node;
 213
 214         retVal = NULL;
 215         node = harvester->headerTable;
 216         while( node ) {
 217                 HeaderEntry *entry;
 218
 219                 entry = node->data;
 220                 if (g_ascii_strncasecmp(entry->header, name,
 221                                         strlen(entry->header)) == 0 ) {
 222                         retVal = entry;
 223                         break;
 224                 }
 225                 node = g_list_next( node );
 226         }
 227         return retVal;
 228 }
 229
 230 /*
 231  * Set selection for specified heaader.
 232  * Enter: harvester Harvester.
 233  *        name      Header name.
 234  *        value     Value to set.
 235  */
 236 void addrharvest_set_header(
 237         AddressHarvester* harvester, const gchar *name, const gboolean value )
 238 {
 239         HeaderEntry *entry;
 240
 241         cm_return_if_fail( harvester != NULL );
 242         entry = addrharvest_find( harvester, name );
 243         if( entry != NULL ) {
 244                 entry->selected = value;
 245         }
 246 }
 247
 248 /*
 249  * Get address count
 250  * Enter: harvester Harvester.
 251  *        name      Header name.
 252  * Return: Address count, or -1 if header not found.
 253  */
 254 gint addrharvest_get_count( AddressHarvester* harvester, const gchar *name ) {
 255         HeaderEntry *entry;
 256         gint count;
 257
 258         count = -1;
 259         cm_return_val_if_fail( harvester != NULL, count );
 260         entry = addrharvest_find( harvester, name );
 261         if( entry != NULL ) {
 262                 count = entry->count;
 263         }
 264         return count;
 265 }
 266
 267 /*
 268 * Free up object by releasing internal memory.
 269 * Enter: harvester Harvester.
 270 */
 271 void addrharvest_free( AddressHarvester *harvester ) {
 272         cm_return_if_fail( harvester != NULL );
 273
 274         /* Free internal stuff */
 275         addrharvest_free_table( harvester );
 276         g_free( harvester->path );
 277
 278         /* Clear pointers */
 279         harvester->path = NULL;
 280         harvester->retVal = MGU_SUCCESS;
 281         harvester->headerTable = NULL;
 282
 283         harvester->folderSize = 0;
 284
 285         /* Now release object */
 286         g_free( harvester );
 287 }
 288
 289 #ifdef USE_ALT_ADDRBOOK
 290 static gchar* get_namepart(const gchar* name, Namepart namepart) {
 291     gchar *pos, *part = NULL;
 292     gchar *token = g_strdup(name);
 293
 294     pos = g_strrstr(token, " ");
 295     if (namepart == FIRST) {
 296         if (pos) {
 297             *pos = '\0';
 298             part = g_strdup(token);
 299             *pos = ' ';
 300         }
 301     }
 302     else {
 303         if (! pos)
 304             part = g_strdup(token);
 305         else {
 306             pos +=1;
 307             part = g_strdup(pos);
 308         }
 309     }
 310     g_free(token);
 311     return part;
 312 }
 313 #endif
 314
 315 /*
 316  * Insert address into cache.
 317  * Enter: harvester Harvester object.
 318  *        entry     Header object.
 319  *        cache     Address cache to load.
 320  *        name      Name.
 321  *        address   eMail address.
 322  */
 323 static void addrharvest_insert_cache(
 324                 AddressHarvester *harvester, HeaderEntry *entry,
 325                 AddressCache *cache, const gchar *name,
 326                 const gchar *address )
 327 {
 328 #ifndef USE_ALT_ADDRBOOK
 329         ItemPerson *person;
 330         ItemFolder *folder;
 331         gchar *folderName;
 332         gboolean newFolder;
 333         gint cnt;
 334         gchar *key, *value;
 335
 336         newFolder = FALSE;
 337         folder = entry->folder;
 338         if( folder == NULL ) {
 339                 newFolder = TRUE;       /* No folder yet */
 340         }
 341         if( entry->count % harvester->folderSize == 0 ) {
 342                 newFolder = TRUE;       /* Folder is full */
 343         }
 344 #else
 345     ContactEntry* person;
 346     gchar* key;
 347 #endif
 348
 349         /* Insert address */
 350         key = g_utf8_strdown( address, -1 );
 351         person = g_hash_table_lookup( harvester->dupTable, key );
 352 #ifndef USE_ALT_ADDRBOOK
 353         if( person ) {
 354                 /* Update existing person to use longest name */
 355                 value = ADDRITEM_NAME(person);
 356                 if( strlen( name ) > strlen( value ) ) {
 357                         addritem_person_set_common_name( person, name );
 358                 }
 359                 g_free( key );
 360         }
 361         else {
 362                 /* Folder if required */
 363                 if( newFolder ) {
 364                         cnt = 1 + ( entry->count / harvester->folderSize );
 365                         folderName =g_strdup_printf( "%s (%d)",
 366                                         entry->header, cnt );
 367                         folder = addritem_create_item_folder();
 368                         addritem_folder_set_name( folder, folderName );
 369                         addritem_folder_set_remarks( folder, "" );
 370                         addrcache_id_folder( cache, folder );
 371                         addrcache_add_folder( cache, folder );
 372                         entry->folder = folder;
 373                         g_free( folderName );
 374                 }
 375
 376                 /* Insert entry */
 377                 person = addrcache_add_contact(
 378                                 cache, folder, name, address, "" );
 379                 g_hash_table_insert( harvester->dupTable, key, person );
 380                 entry->count++;
 381         }
 382         addritem_parse_first_last( person );
 383 #else
 384         if (! person) {
 385                 person = g_new0(ContactEntry, 1);
 386                 person->first_name = get_namepart(name, FIRST);
 387                 person->last_name = get_namepart(name, LAST);
 388                 person->email = g_strdup(address);
 389                 g_hash_table_insert(harvester->dupTable, key, person);
 390                 entry->count++;
 391         }
 392 #endif
 393 }
 394
 395 /*
 396  * Remove specified string from name.
 397  * Enter: name Name.
 398  *        str  String to remove.
 399  */
 400 static void addrharvest_del_email( gchar *name, gchar *str ) {
 401         gchar *p;
 402         gint lenn, lenr;
 403
 404         lenr = strlen( str );
 405         while((p = strcasestr( name, str )) != NULL) {
 406                 lenn = strlen( p );
 407                 memmove( p, p + lenr, lenn );
 408         }
 409 }
 410
 411 /*
 412  * Find position of at (@) character in buffer.
 413  * Enter:  buffer Start of buffer.
 414  * Return: Position of at character, or NULL if not found.
 415  * Note: This function searches for the last occurrence of an 'at' character
 416  * prior to a valid delimiter character for the end of address. This enables
 417  * an address to be found where it is also used as the name of the
 418  * recipient. For example:
 419  *     "axle.rose@netscape.com" <axle.rose@netscape.com>
 420  * The last occurrence of the at character is detected.
 421  */
 422 static gchar *addrharvest_find_at( const gchar *buffer ) {
 423         gchar *atCh;
 424         gchar *p;
 425
 426         atCh = strchr( buffer, '@' );
 427         if( atCh ) {
 428                 /* Search forward for another one */
 429                 p = atCh + 1;
 430                 while( *p ) {
 431                         if( *p == '>' ) {
 432                                 break;
 433                         }
 434                         if( *p == ',' ) {
 435                                 break;
 436                         }
 437                         if( *p == '\n' ) {
 438                                 break;
 439                         }
 440                         if( *p == '@' ) {
 441                                 atCh = p;
 442                                 break;
 443                         }
 444                         p++;
 445                 }
 446         }
 447         return atCh;
 448 }
 449
 450 /*
 451  * Find start and end of address string.
 452  * Enter: buf Start address of buffer to process (not modified).
 453  *        atp Pointer to email at (@) character.
 454  *        bp  Pointer to start of email address (returned).
 455  *        ep  Pointer to end of email address (returned).
 456  */
 457 static void addrharvest_find_address(
 458                 const gchar *buf, const gchar *atp, const gchar **bp,
 459                 const gchar **ep )
 460 {
 461         const gchar *p;
 462
 463         /* Find first non-separator char */
 464         *bp = NULL;
 465         p = buf;
 466         while( TRUE ) {
 467                 if( strchr( ",; \n\r", *p ) == NULL ) break;
 468                 p++;
 469         }
 470         *bp = p;
 471
 472         /* Search forward for end of address */
 473         *ep = NULL;
 474         p = atp + 1;
 475         while( TRUE ) {
 476                 if( strchr( ",;", *p ) ) break;
 477                 p++;
 478         }
 479         *ep = p;
 480 }
 481
 482 /*
 483  * Extract E-Mail address from buffer. If found, address is removed from
 484  * buffer.
 485  * Enter:  buffer Address buffer.
 486  * Return: E-Mail address, or NULL if none found. Must g_free() when done.
 487  */
 488 static gchar *addrharvest_extract_address( gchar *buffer ) {
 489         gchar *addr;
 490         gchar *atCh, *p, *bp, *ep;
 491         gint len;
 492
 493         addr = NULL;
 494         atCh = addrharvest_find_at( buffer );
 495         if( atCh ) {
 496                 /* Search back for start of address */
 497                 bp = NULL;
 498                 p = atCh;
 499                 while( p >= buffer ) {
 500                         bp = p;
 501                         if( *p == '<' ) {
 502                                 *p = ' ';
 503                                 bp++;
 504                                 break;
 505                         }
 506                         p--;
 507                 }
 508
 509                 /* Search fwd for end */
 510                 ep = NULL;
 511                 ep = p = atCh;
 512                 while( *p ) {
 513                         if( *p == '>' ) {
 514                                 *p = ' ';
 515                                 break;
 516                         }
 517                         else if( *p == ' ' ) {
 518                                 break;
 519                         }
 520                         ep = p;
 521                         p++;
 522                 }
 523
 524                 /* Extract email */
 525                 if( bp != NULL ) {
 526                         len = ( ep - bp );
 527                         if( len > 0 ) {
 528                                 addr = g_strndup( bp, len + 1 );
 529                                 memmove( bp, ep, len );
 530                                 *bp = ' ';
 531                         }
 532                 }
 533         }
 534         return addr;
 535 }
 536
 537 /*
 538  * Parse address from header buffer creating address in cache.
 539  * Enter: harvester Harvester object.
 540  *        entry     Header object.
 541  *        cache     Address cache to load.
 542  *        hdrBuf    Pointer to header buffer.
 543  */
 544 static void addrharvest_parse_address(
 545                 AddressHarvester *harvester, HeaderEntry *entry,
 546                 AddressCache *cache, const gchar *hdrBuf )
 547 {
 548         gchar buffer[ ADDR_BUFFSIZE + 2 ];
 549         const gchar *bp;
 550         const gchar *ep;
 551         gchar *atCh, *email, *name;
 552         gint bufLen;
 553
 554         /* Search for an address */
 555         while((atCh = addrharvest_find_at( hdrBuf )) != NULL) {
 556                 /* Find addres string */
 557                 addrharvest_find_address( hdrBuf, atCh, &bp, &ep );
 558
 559                 /* Copy into buffer */
 560                 bufLen = ( size_t ) ( ep - bp );
 561                 if( bufLen > ADDR_BUFFSIZE -1 ) {
 562                         bufLen = ADDR_BUFFSIZE - 1;
 563                 }
 564                 strncpy( buffer, bp, bufLen );
 565                 buffer[ bufLen ] = '\0';
 566                 buffer[ bufLen + 1 ] = '\0';
 567                 buffer[ bufLen + 2 ] = '\0';
 568
 569                 /* Extract address from buffer */
 570                 email = addrharvest_extract_address( buffer );
 571                 if( email ) {
 572                         /* Unescape characters */
 573                         mgu_str_unescape( buffer );
 574
 575                         /* Remove noise characaters */
 576                         addrharvest_del_email( buffer, REM_NAME_STRING );
 577                         addrharvest_del_email( buffer, REM_NAME_STRING2 );
 578
 579                         /* Remove leading trailing quotes and spaces */
 580                         mgu_str_ltc2space( buffer, '\"', '\"' );
 581                         mgu_str_ltc2space( buffer, '\'', '\'' );
 582                         mgu_str_ltc2space( buffer, '\"', '\"' );
 583                         mgu_str_ltc2space( buffer, '(', ')' );
 584                         g_strstrip( buffer );
 585
 586                         if( g_ascii_strcasecmp( buffer, email ) == 0 )
 587                                 name = g_strdup("");
 588                         else
 589                                 name = conv_unmime_header(buffer, NULL, TRUE);
 590
 591                         /* Insert into address book */
 592 #ifndef USE_ALT_ADDRBOOK
 593                         addrharvest_insert_cache(
 594                                 harvester, entry, cache, name, email );
 595 #else
 596                         addrharvest_insert_cache(
 597                                 harvester, entry, NULL, name, email);
 598 #endif
 599                         g_free( email );
 600                         g_free( name );
 601                 }
 602                 hdrBuf = ep;
 603         }
 604 }
 605
 606 /*
 607  * Test whether buffer contains a header that appears in header list.
 608  * Enter: listHdr Header list.
 609  *        buf     Header buffer.
 610  * Return: TRUE if header in list.
 611  */
 612 static gboolean addrharvest_check_hdr( GList *listHdr, gchar *buf ) {
 613         gboolean retVal;
 614         GList *node;
 615         gchar *p, *hdr, *nhdr;
 616         gint len;
 617
 618         retVal = FALSE;
 619         p = strchr( buf, ':' );
 620         if( p ) {
 621                 len = ( size_t ) ( p - buf );
 622                 hdr = g_strndup( buf, len );
 623                 node = listHdr;
 624                 while( node ) {
 625                         nhdr = node->data;
 626                         if (g_ascii_strncasecmp(nhdr, hdr, strlen(nhdr)) == 0 ) {
 627                                 retVal = TRUE;
 628                                 break;
 629                         }
 630                         node = g_list_next( node );
 631                 }
 632                 g_free( hdr );
 633         }
 634         return retVal;
 635 }
 636
 637 /*
 638  * Read header into a linked list of lines.
 639  * Enter:  fp      File to read.
 640  *         listHdr List of header lines of interest.
 641  *         done    End of headers or end of file reached.
 642  * Return: Linked list of lines.
 643  */
 644 static GSList *addrharvest_get_header( FILE *fp, GList *listHdr, gboolean *done ) {
 645         GSList *list;
 646         gchar buf[ MSG_BUFFSIZE + 2 ];
 647         gint ch;
 648         gboolean foundHdr;
 649
 650         list = NULL;
 651
 652         /* Read line */
 653         if( claws_fgets( buf, MSG_BUFFSIZE, fp ) == NULL ) {
 654                 *done = TRUE;
 655                 return list;
 656         }
 657
 658         /* Test for end of headers */
 659         if( buf[0] == '\r' || buf[0] == '\n' ) {
 660                 *done = TRUE;
 661                 return list;
 662         }
 663
 664         /* Test whether required header */
 665         foundHdr = addrharvest_check_hdr( listHdr, buf );
 666
 667         /* Read all header lines. Only add reqd ones to list */
 668         while( TRUE ) {
 669                 gchar *p;
 670
 671                 if( foundHdr ) {
 672                         p = g_strdup( buf );
 673                         list = g_slist_append( list, p );
 674                 }
 675
 676                 /* Read first character */
 677                 ch = fgetc( fp );
 678                 if( ch == ' ' || ch == '\t' ) {
 679                         /* Continuation character - read into buffer */
 680                         if( claws_fgets( buf, MSG_BUFFSIZE, fp ) == NULL ) {
 681                                 break;
 682                         }
 683                 }
 684                 else {
 685                         if( ch == EOF ) {
 686                                 *done = TRUE;
 687                         }
 688                         else {
 689                                 /* Push back character for next header */
 690                                 ungetc( ch, fp );
 691                         }
 692                         break;
 693                 }
 694         }
 695
 696         return list;
 697 }
 698
 699 /*
 700  * Read specified file into address book.
 701  * Enter:  harvester Harvester object.
 702  *         fileName  File to read.
 703  *         cache     Address cache to load.
 704  * Return: Status.
 705  */
 706 static gint addrharvest_readfile(
 707                 AddressHarvester *harvester, const gchar *fileName,
 708                 AddressCache *cache, GList *listHdr )
 709 {
 710         gint retVal;
 711         FILE *msgFile;
 712         gchar *buf, *addr, *p;
 713         HeaderEntry *entry;
 714         GSList *list;
 715         gboolean done;
 716
 717         msgFile = claws_fopen( fileName, "rb" );
 718         if( ! msgFile ) {
 719                 /* Cannot open file */
 720                 retVal = MGU_OPEN_FILE;
 721                 return retVal;
 722         }
 723
 724         done = FALSE;
 725         while( TRUE ) {
 726                 list = addrharvest_get_header( msgFile, listHdr, &done );
 727                 if( done ) break;
 728
 729                 if( list == NULL ) {
 730                         continue;
 731                 }
 732
 733                 buf = mgu_list_coalesce( list );
 734                 g_slist_free_full( list, g_free );
 735
 736                 if(( p = strchr( buf, ':' ) ) != NULL ) {
 737                         addr = p + 1;
 738                         *p = '\0';
 739
 740                         entry = addrharvest_find( harvester, buf );
 741                         if( entry && entry->selected ) {
 742                                 /* Sanitize control characters */
 743                                 p = addr;
 744                                 while( *p ) {
 745                                         if( *p == '\r' || *p == '\n' || *p == '\t' )
 746                                                 *p = ' ';
 747                                         p++;
 748                                 }
 749                                 addrharvest_parse_address(
 750                                         harvester, entry, cache, addr );
 751                         }
 752                 }
 753                 g_free( buf );
 754         }
 755
 756         claws_fclose( msgFile );
 757         return MGU_SUCCESS;
 758 }
 759
 760 /*
 761  * Read all files in specified directory into address book. Directories are
 762  * traversed recursively if necessary.
 763  * Enter:  harvester Harvester object.
 764  *         cache     Address cache to load.
 765  *         msgList   List of message numbers, or NULL to process folder.
 766  *         dir       Directory to process.
 767  */
 768 static void addrharvest_harvest_dir(
 769         AddressHarvester *harvester, AddressCache *cache, GList *listHdr,
 770         gchar *dir )
 771 {
 772         GDir *dp;
 773         const gchar *d;
 774         gchar *fullname;
 775         GError *error = NULL;
 776         gint num;
 777
 778         debug_print("Harvesting addresses from dir '%s'\n", dir);
 779
 780         if( ( dp = g_dir_open( dir, 0, &error ) ) == NULL ) {
 781                 debug_print("opening '%s' failed: %d (%s)\n", dir,
 782                                 error->code, error->message);
 783                 g_error_free(error);
 784                 return;
 785         }
 786
 787         /* Process directory */
 788         while( (d = g_dir_read_name( dp )) != NULL ) {
 789                 fullname = g_strconcat(dir, G_DIR_SEPARATOR_S, d, NULL);
 790                 if( g_file_test(fullname, G_FILE_TEST_IS_DIR) ) {
 791                         if( harvester->folderRecurse ) {
 792                                 if( strstr( DIR_IGNORE, d ) != NULL ) {
 793                                         g_free(fullname);
 794                                         continue;
 795                                 }
 796
 797                                 addrharvest_harvest_dir(
 798                                         harvester, cache, listHdr, (gchar *)fullname );
 799                         }
 800                 }
 801                 if( g_file_test(fullname, G_FILE_TEST_IS_REGULAR) ) {
 802                         if( ( num = to_number( d ) ) >= 0 ) {
 803                                 addrharvest_readfile(
 804                                         harvester, fullname, cache, listHdr );
 805                         }
 806                 }
 807                 g_free(fullname);
 808         }
 809         g_dir_close( dp );
 810 }
 811
 812 /*
 813  * Read list of files in specified directory into address book.
 814  * Enter:  harvester Harvester object.
 815  *         cache     Address cache to load.
 816  *         msgList   List of message numbers, or NULL to process folder.
 817  */
 818 static void addrharvest_harvest_list(
 819         AddressHarvester *harvester, AddressCache *cache, GList *listHdr,
 820         GList *msgList )
 821 {
 822         gint num;
 823         GList *node;
 824         gchar *fullname;
 825
 826         if (!g_file_test(harvester->path, G_FILE_TEST_EXISTS | G_FILE_TEST_IS_DIR)) {
 827                 debug_print("'%s' doesn't exist or is not a dir\n", harvester->path);
 828                 return;
 829         }
 830
 831         /* Process message list */
 832         node = msgList;
 833         while( node ) {
 834                 num = GPOINTER_TO_UINT( node->data );
 835                 fullname = g_strdup_printf("%s%c%d",
 836                                 harvester->path, G_DIR_SEPARATOR, num);
 837                 addrharvest_readfile( harvester, fullname, cache, listHdr );
 838                 g_free(fullname);
 839                 node = g_list_next( node );
 840         }
 841 }
 842
 843 /*
 844  * ============================================================================
 845  * Read all files in specified directory into address book.
 846  * Enter:  harvester Harvester object.
 847  *         cache     Address cache to load.
 848  *         msgList   List of message numbers, or NULL to process folder.
 849  * Return: Status.
 850  * ============================================================================
 851  */
 852 gint addrharvest_harvest(
 853         AddressHarvester *harvester, AddressCache *cache, GList *msgList )
 854 {
 855         gint retVal;
 856         GList *node;
 857         GList *listHdr;
 858
 859         retVal = MGU_BAD_ARGS;
 860         cm_return_val_if_fail( harvester != NULL, retVal );
 861 #ifndef USE_ALT_ADDRBOOK
 862         cm_return_val_if_fail( cache != NULL, retVal );
 863 #endif
 864         cm_return_val_if_fail( harvester->path != NULL, retVal );
 865
 866 #ifndef USE_ALT_ADDRBOOK
 867         /* Clear cache */
 868         addrcache_clear( cache );
 869         cache->dataRead = FALSE;
 870 #endif
 871         /* Build list of headers of interest */
 872         listHdr = NULL;
 873         node = harvester->headerTable;
 874         while( node ) {
 875                 HeaderEntry *entry;
 876
 877                 entry = node->data;
 878                 if( entry->selected ) {
 879                         gchar *p;
 880
 881                         p = g_utf8_strdown( entry->header, -1 );
 882                         listHdr = g_list_append( listHdr, p );
 883                 }
 884                 node = g_list_next( node );
 885         }
 886
 887         /* Process directory/files */
 888         if( msgList == NULL ) {
 889                 addrharvest_harvest_dir( harvester, cache, listHdr, harvester->path );
 890         }
 891         else {
 892                 addrharvest_harvest_list( harvester, cache, listHdr, msgList );
 893         }
 894         g_list_free_full( listHdr, g_free );
 895
 896 #ifndef USE_ALT_ADDRBOOK
 897         /* Mark cache */
 898         cache->modified = FALSE;
 899         cache->dataRead = TRUE;
 900 #endif
 901         return retVal;
 902 }
 903
 904 /*
 905  * ============================================================================
 906  * Test whether any headers have been selected for processing.
 907  * Enter:  harvester Harvester object.
 908  * Return: TRUE if a header was selected, FALSE if none were selected.
 909  * ============================================================================
 910  */
 911 gboolean addrharvest_check_header( AddressHarvester *harvester ) {
 912         gboolean retVal;
 913         GList *node;
 914
 915         retVal = FALSE;
 916         cm_return_val_if_fail( harvester != NULL, retVal );
 917
 918         node = harvester->headerTable;
 919         while( node ) {
 920                 HeaderEntry *entry;
 921
 922                 entry = ( HeaderEntry * ) node->data;
 923                 if( entry->selected ) return TRUE;
 924                 node = g_list_next( node );
 925         }
 926         return retVal;
 927 }
 928
 929 /*
 930  * ============================================================================
 931  * End of Source.
 932  * ============================================================================
 933  */
 934
 935