src/addrharvest.c

   1 /*
   2  * Sylpheed -- a GTK+ based, lightweight, and fast e-mail client
   3  * Copyright (C) 2002-2006 Match Grun and the Sylpheed-Claws 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 2 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, write to the Free Software
  17  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
  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 "utils.h"
  30 #include "mgutils.h"
  31 #include "addrharvest.h"
  32 #include "codeconv.h"
  33 #include "addritem.h"
  34
  35 /* Mail header names of interest */
  36 static gchar *_headerFrom_     = HEADER_FROM;
  37 static gchar *_headerReplyTo_  = HEADER_REPLY_TO;
  38 static gchar *_headerSender_   = HEADER_SENDER;
  39 static gchar *_headerErrorsTo_ = HEADER_ERRORS_TO;
  40 static gchar *_headerCC_       = HEADER_CC;
  41 static gchar *_headerTo_       = HEADER_TO;
  42
  43 #define ADDR_BUFFSIZE    1024
  44 #define MSG_BUFFSIZE     2048
  45 #define MSGNUM_BUFFSIZE  32
  46 #define DFL_FOLDER_SIZE  20
  47
  48 /* Noise strings included by some other E-Mail clients */
  49 #define REM_NAME_STRING  "(Email)"
  50 #define REM_NAME_STRING2 "(Email 2)"
  51
  52 /* Directories to ignore */
  53 #define DIR_IGNORE ".\t..\t.sylpheed_mark\t.sylpheed_cache"
  54
  55 /*
  56  * Header entry.
  57  */
  58 typedef struct _HeaderEntry HeaderEntry;
  59 struct _HeaderEntry {
  60         gchar      *header;
  61         gboolean   selected;
  62         ItemFolder *folder;
  63         gint       count;
  64 };
  65
  66 /*
  67  * Build header table entry.
  68  * Enter: harvester Harvester object.
  69  *        name      Header name.
  70  */
  71 static void addrharvest_build_entry(
  72                 AddressHarvester* harvester, gchar *name )
  73 {
  74         HeaderEntry *entry;
  75
  76         entry = g_new0( HeaderEntry, 1 );
  77         entry->header = name;
  78         entry->selected = FALSE;
  79         entry->folder = NULL;
  80         entry->count = 0;
  81         harvester->headerTable = g_list_append( harvester->headerTable, entry );
  82 }
  83
  84 /*
  85  * Free key in table.
  86  */
  87 static gint addrharvest_free_table_vis( gpointer key, gpointer value, gpointer data ) {
  88         g_free( key );
  89         key = NULL;
  90         value = NULL;
  91         return TRUE;
  92 }
  93
  94 /*
  95  * Free lookup table.
  96  */
  97 static void addrharvest_free_table( AddressHarvester* harvester ) {
  98         GList *node;
  99         HeaderEntry *entry;
 100
 101         /* Free header list */
 102         node = harvester->headerTable;
 103         while( node ) {
 104                 entry = ( HeaderEntry * ) node->data;
 105                 entry->header = NULL;
 106                 entry->selected = FALSE;
 107                 entry->folder = NULL;
 108                 entry->count = 0;
 109                 g_free( entry );
 110                 node = g_list_next( node );
 111         }
 112         g_list_free( harvester->headerTable );
 113         harvester->headerTable = NULL;
 114
 115         /* Free duplicate table */
 116         g_hash_table_foreach_remove( harvester->dupTable, addrharvest_free_table_vis, NULL );
 117         g_hash_table_destroy( harvester->dupTable );
 118         harvester->dupTable = NULL;
 119 }
 120
 121 /*
 122 * Create new object.
 123 * Return: Harvester.
 124 */
 125 AddressHarvester *addrharvest_create( void ) {
 126         AddressHarvester *harvester;
 127
 128         harvester = g_new0( AddressHarvester, 1 );
 129         harvester->path = NULL;
 130         harvester->dupTable = g_hash_table_new( g_str_hash, g_str_equal );
 131         harvester->folderSize = DFL_FOLDER_SIZE;
 132         harvester->retVal = MGU_SUCCESS;
 133
 134         /* Build header table */
 135         harvester->headerTable = NULL;
 136         addrharvest_build_entry( harvester, _headerFrom_ );
 137         addrharvest_build_entry( harvester, _headerReplyTo_ );
 138         addrharvest_build_entry( harvester, _headerSender_ );
 139         addrharvest_build_entry( harvester, _headerErrorsTo_ );
 140         addrharvest_build_entry( harvester, _headerCC_ );
 141         addrharvest_build_entry( harvester, _headerTo_ );
 142
 143         return harvester;
 144 }
 145
 146 /*
 147 * Properties...
 148 */
 149 /*
 150  * Specify path to folder that will be harvested.
 151  * Entry: harvester Harvester object.
 152  *        value     Full directory path.
 153  */
 154 void addrharvest_set_path( AddressHarvester* harvester, const gchar *value ) {
 155         g_return_if_fail( harvester != NULL );
 156         harvester->path = mgu_replace_string( harvester->path, value );
 157         g_strstrip( harvester->path );
 158 }
 159
 160 /*
 161  * Specify maximum folder size.
 162  * Entry: harvester Harvester object.
 163  *        value     Folder size.
 164  */
 165 void addrharvest_set_folder_size(
 166         AddressHarvester* harvester, const gint value )
 167 {
 168         g_return_if_fail( harvester != NULL );
 169         if( value > 0 ) {
 170                 harvester->folderSize = value;
 171         }
 172 }
 173
 174 /*
 175  * Specify folder recursion.
 176  * Entry: harvester Harvester object.
 177  *        value     TRUE to process sub-folders, FALSE to process folder only.
 178  */
 179 void addrharvest_set_recurse(
 180         AddressHarvester* harvester, const gboolean value )
 181 {
 182         g_return_if_fail( harvester != NULL );
 183         harvester->folderRecurse = value;
 184 }
 185
 186 /*
 187  * Search (case insensitive) for header entry with specified name.
 188  * Enter: harvester Harvester.
 189  *        name      Header name.
 190  * Return: Header, or NULL if not found.
 191  */
 192 static HeaderEntry *addrharvest_find(
 193         AddressHarvester* harvester, const gchar *name ) {
 194         HeaderEntry *retVal;
 195         GList *node;
 196
 197         retVal = NULL;
 198         node = harvester->headerTable;
 199         while( node ) {
 200                 HeaderEntry *entry;
 201
 202                 entry = node->data;
 203                 if (g_ascii_strncasecmp(entry->header, name,
 204                                         strlen(entry->header)) == 0 ) {
 205                         retVal = entry;
 206                         break;
 207                 }
 208                 node = g_list_next( node );
 209         }
 210         return retVal;
 211 }
 212
 213 /*
 214  * Set selection for specified heaader.
 215  * Enter: harvester Harvester.
 216  *        name      Header name.
 217  *        value     Value to set.
 218  */
 219 void addrharvest_set_header(
 220         AddressHarvester* harvester, const gchar *name, const gboolean value )
 221 {
 222         HeaderEntry *entry;
 223
 224         g_return_if_fail( harvester != NULL );
 225         entry = addrharvest_find( harvester, name );
 226         if( entry != NULL ) {
 227                 entry->selected = value;
 228         }
 229 }
 230
 231 /*
 232  * Get address count
 233  * Enter: harvester Harvester.
 234  *        name      Header name.
 235  * Return: Address count, or -1 if header not found.
 236  */
 237 gint addrharvest_get_count( AddressHarvester* harvester, const gchar *name ) {
 238         HeaderEntry *entry;
 239         gint count;
 240
 241         count = -1;
 242         g_return_val_if_fail( harvester != NULL, count );
 243         entry = addrharvest_find( harvester, name );
 244         if( entry != NULL ) {
 245                 count = entry->count;
 246         }
 247         return count;
 248 }
 249
 250 /*
 251 * Free up object by releasing internal memory.
 252 * Enter: harvester Harvester.
 253 */
 254 void addrharvest_free( AddressHarvester *harvester ) {
 255         g_return_if_fail( harvester != NULL );
 256
 257         /* Free internal stuff */
 258         addrharvest_free_table( harvester );
 259         g_free( harvester->path );
 260
 261         /* Clear pointers */
 262         harvester->path = NULL;
 263         harvester->retVal = MGU_SUCCESS;
 264         harvester->headerTable = NULL;
 265
 266         harvester->folderSize = 0;
 267
 268         /* Now release object */
 269         g_free( harvester );
 270 }
 271
 272 /*
 273 * Display object to specified stream.
 274 * Enter: harvester Harvester.
 275 *        stream    Output stream.
 276 */
 277 void addrharvest_print( AddressHarvester *harvester, FILE *stream ) {
 278         GList *node;
 279         HeaderEntry *entry;
 280
 281         g_return_if_fail( harvester != NULL );
 282         fprintf( stream, "Address Harvester:\n" );
 283         fprintf( stream, " file path: '%s'\n", harvester->path );
 284         fprintf( stream, "max folder: %d'\n", harvester->folderSize );
 285
 286         node = harvester->headerTable;
 287         while( node ) {
 288                 entry = node->data;
 289                 fprintf( stream, "   header: %s", entry->header );
 290                 fprintf( stream, "\t: %s", entry->selected ? "yes" : "no" );
 291                 fprintf( stream, "\t: %d\n", entry->count );
 292                 node = g_list_next( node );
 293         }
 294         fprintf( stream, "  ret val: %d\n", harvester->retVal );
 295 }
 296
 297 /*
 298  * Insert address into cache.
 299  * Enter: harvester Harvester object.
 300  *        entry     Header object.
 301  *        cache     Address cache to load.
 302  *        name      Name.
 303  *        address   eMail address.
 304  */
 305 static void addrharvest_insert_cache(
 306                 AddressHarvester *harvester, HeaderEntry *entry,
 307                 AddressCache *cache, const gchar *name,
 308                 const gchar *address )
 309 {
 310         ItemPerson *person;
 311         ItemFolder *folder;
 312         gchar *folderName;
 313         gboolean newFolder;
 314         gint cnt;
 315         gchar *key, *value;
 316
 317         newFolder = FALSE;
 318         folder = entry->folder;
 319         if( folder == NULL ) {
 320                 newFolder = TRUE;       /* No folder yet */
 321         }
 322         if( entry->count % harvester->folderSize == 0 ) {
 323                 newFolder = TRUE;       /* Folder is full */
 324         }
 325
 326         /* Insert address */
 327         key = g_strdup( address );
 328         g_strdown( key );
 329         person = g_hash_table_lookup( harvester->dupTable, key );
 330         if( person ) {
 331                 /* Update existing person to use longest name */
 332                 value = ADDRITEM_NAME(person);
 333                 if( strlen( name ) > strlen( value ) ) {
 334                         addritem_person_set_common_name( person, name );
 335                 }
 336                 g_free( key );
 337         }
 338         else {
 339                 /* Folder if required */
 340                 if( newFolder ) {
 341                         cnt = 1 + ( entry->count / harvester->folderSize );
 342                         folderName =g_strdup_printf( "%s (%d)",
 343                                         entry->header, cnt );
 344                         folder = addritem_create_item_folder();
 345                         addritem_folder_set_name( folder, folderName );
 346                         addritem_folder_set_remarks( folder, "" );
 347                         addrcache_id_folder( cache, folder );
 348                         addrcache_add_folder( cache, folder );
 349                         entry->folder = folder;
 350                         g_free( folderName );
 351                 }
 352
 353                 /* Insert entry */
 354                 person = addrcache_add_contact(
 355                                 cache, folder, name, address, "" );
 356                 g_hash_table_insert( harvester->dupTable, key, person );
 357                 entry->count++;
 358         }
 359         addritem_parse_first_last( person );
 360 }
 361
 362 /*
 363  * Remove specified string from name.
 364  * Enter: name Name.
 365  *        str  String to remove.
 366  */
 367 static void addrharvest_del_email( gchar *name, gchar *str ) {
 368         gchar *p;
 369         gint lenn, lenr;
 370
 371         lenr = strlen( str );
 372         while((p = strcasestr( name, str )) != NULL) {
 373                 lenn = strlen( p );
 374                 memmove( p, p + lenr, lenn );
 375         }
 376 }
 377
 378 /*
 379  * Find position of at (@) character in buffer.
 380  * Enter:  buffer Start of buffer.
 381  * Return: Position of at character, or NULL if not found.
 382  * Note: This function searches for the last occurrence of an 'at' character
 383  * prior to a valid delimiter character for the end of address. This enables
 384  * an address to be found where it is also used as the name of the
 385  * recipient. For example:
 386  *     "axle.rose@netscape.com" <axle.rose@netscape.com>
 387  * The last occurrence of the at character is detected.
 388  */
 389 static gchar *addrharvest_find_at( const gchar *buffer ) {
 390         gchar *atCh;
 391         gchar *p;
 392
 393         atCh = strchr( buffer, '@' );
 394         if( atCh ) {
 395                 /* Search forward for another one */
 396                 p = atCh + 1;
 397                 while( *p ) {
 398                         if( *p == '>' ) {
 399                                 break;
 400                         }
 401                         if( *p == ',' ) {
 402                                 break;
 403                         }
 404                         if( *p == '\n' ) {
 405                                 break;
 406                         }
 407                         if( *p == '@' ) {
 408                                 atCh = p;
 409                                 break;
 410                         }
 411                         p++;
 412                 }
 413         }
 414         return atCh;
 415 }
 416
 417 /*
 418  * Find start and end of address string.
 419  * Enter: buf Start address of buffer to process (not modified).
 420  *        atp Pointer to email at (@) character.
 421  *        bp  Pointer to start of email address (returned).
 422  *        ep  Pointer to end of email address (returned).
 423  */
 424 static void addrharvest_find_address(
 425                 const gchar *buf, const gchar *atp, const gchar **bp,
 426                 const gchar **ep )
 427 {
 428         const gchar *p;
 429
 430         /* Find first non-separator char */
 431         *bp = NULL;
 432         p = buf;
 433         while( TRUE ) {
 434                 if( strchr( ",; \n\r", *p ) == NULL ) break;
 435                 p++;
 436         }
 437         *bp = p;
 438
 439         /* Search forward for end of address */
 440         *ep = NULL;
 441         p = atp + 1;
 442         while( TRUE ) {
 443                 if( strchr( ",;", *p ) ) break;
 444                 p++;
 445         }
 446         *ep = p;
 447 }
 448
 449 /*
 450  * Extract E-Mail address from buffer. If found, address is removed from
 451  * buffer.
 452  * Enter:  buffer Address buffer.
 453  * Return: E-Mail address, or NULL if none found. Must g_free() when done.
 454  */
 455 static gchar *addrharvest_extract_address( gchar *buffer ) {
 456         gchar *addr;
 457         gchar *atCh, *p, *bp, *ep;
 458         gint len;
 459
 460         addr = NULL;
 461         atCh = addrharvest_find_at( buffer );
 462         if( atCh ) {
 463                 /* Search back for start of address */
 464                 bp = NULL;
 465                 p = atCh;
 466                 while( p >= buffer ) {
 467                         bp = p;
 468                         if( *p == '<' ) {
 469                                 *p = ' ';
 470                                 bp++;
 471                                 break;
 472                         }
 473                         p--;
 474                 }
 475
 476                 /* Search fwd for end */
 477                 ep = NULL;
 478                 ep = p = atCh;
 479                 while( *p ) {
 480                         if( *p == '>' ) {
 481                                 *p = ' ';
 482                                 break;
 483                         }
 484                         else if( *p == ' ' ) {
 485                                 break;
 486                         }
 487                         ep = p;
 488                         p++;
 489                 }
 490
 491                 /* Extract email */
 492                 if( bp != NULL ) {
 493                         len = ( ep - bp );
 494                         if( len > 0 ) {
 495                                 addr = g_strndup( bp, len + 1 );
 496                                 memmove( bp, ep, len );
 497                                 *bp = ' ';
 498                         }
 499                 }
 500         }
 501         return addr;
 502 }
 503
 504 /*
 505  * Parse address from header buffer creating address in cache.
 506  * Enter: harvester Harvester object.
 507  *        entry     Header object.
 508  *        cache     Address cache to load.
 509  *        hdrBuf    Pointer to header buffer.
 510  */
 511 static void addrharvest_parse_address(
 512                 AddressHarvester *harvester, HeaderEntry *entry,
 513                 AddressCache *cache, const gchar *hdrBuf )
 514 {
 515         gchar buffer[ ADDR_BUFFSIZE + 2 ];
 516         const gchar *bp;
 517         const gchar *ep;
 518         gchar *atCh, *email, *name;
 519         gint bufLen;
 520
 521         /* Search for an address */
 522         while((atCh = addrharvest_find_at( hdrBuf )) != NULL) {
 523                 /* Find addres string */
 524                 addrharvest_find_address( hdrBuf, atCh, &bp, &ep );
 525
 526                 /* Copy into buffer */
 527                 bufLen = ( size_t ) ( ep - bp );
 528                 if( bufLen > ADDR_BUFFSIZE ) {
 529                         bufLen = ADDR_BUFFSIZE;
 530                 }
 531                 strncpy( buffer, bp, bufLen );
 532                 buffer[ bufLen ] = '\0';
 533                 buffer[ bufLen + 1 ] = '\0';
 534                 buffer[ bufLen + 2 ] = '\0';
 535
 536                 /* Extract address from buffer */
 537                 email = addrharvest_extract_address( buffer );
 538                 if( email ) {
 539                         /* Unescape characters */
 540                         mgu_str_unescape( buffer );
 541
 542                         /* Remove noise characaters */
 543                         addrharvest_del_email( buffer, REM_NAME_STRING );
 544                         addrharvest_del_email( buffer, REM_NAME_STRING2 );
 545
 546                         /* Remove leading trailing quotes and spaces */
 547                         mgu_str_ltc2space( buffer, '\"', '\"' );
 548                         mgu_str_ltc2space( buffer, '\'', '\'' );
 549                         mgu_str_ltc2space( buffer, '\"', '\"' );
 550                         mgu_str_ltc2space( buffer, '(', ')' );
 551                         g_strstrip( buffer );
 552
 553                         if( g_ascii_strcasecmp( buffer, email ) == 0 ) {
 554                                 name = g_strdup("");
 555                         }
 556                         else {
 557                                 name = buffer;
 558                                 name = conv_unmime_header(buffer, NULL);
 559                         }
 560
 561                         /* Insert into address book */
 562                         addrharvest_insert_cache(
 563                                 harvester, entry, cache, name, email );
 564                         g_free( email );
 565                         g_free( name );
 566                 }
 567                 hdrBuf = ep;
 568         }
 569 }
 570
 571 /*
 572  * Test whether buffer contains a header that appears in header list.
 573  * Enter: listHdr Header list.
 574  *        buf     Header buffer.
 575  * Return: TRUE if header in list.
 576  */
 577 static gboolean addrharvest_check_hdr( GList *listHdr, gchar *buf ) {
 578         gboolean retVal;
 579         GList *node;
 580         gchar *p, *hdr, *nhdr;
 581         gint len;
 582
 583         retVal = FALSE;
 584         p = strchr( buf, ':' );
 585         if( p ) {
 586                 len = ( size_t ) ( p - buf );
 587                 hdr = g_strndup( buf, len );
 588                 node = listHdr;
 589                 while( node ) {
 590                         nhdr = node->data;
 591                         if (g_ascii_strncasecmp(nhdr, hdr, strlen(nhdr)) == 0 ) {
 592                                 retVal = TRUE;
 593                                 break;
 594                         }
 595                         node = g_list_next( node );
 596                 }
 597                 g_free( hdr );
 598         }
 599         return retVal;
 600 }
 601
 602 /*
 603  * Read header into a linked list of lines.
 604  * Enter:  fp      File to read.
 605  *         listHdr List of header lines of interest.
 606  *         done    End of headers or end of file reached.
 607  * Return: Linked list of lines.
 608  */
 609 static GSList *addrharvest_get_header( FILE *fp, GList *listHdr, gboolean *done ) {
 610         GSList *list;
 611         gchar buf[ MSG_BUFFSIZE + 2 ];
 612         gint ch;
 613         gboolean foundHdr;
 614
 615         list = NULL;
 616
 617         /* Read line */
 618         if( fgets( buf, MSG_BUFFSIZE, fp ) == NULL ) {
 619                 *done = TRUE;
 620                 return list;
 621         }
 622
 623         /* Test for end of headers */
 624         if( buf[0] == '\r' || buf[0] == '\n' ) {
 625                 *done = TRUE;
 626                 return list;
 627         }
 628
 629         /* Test whether required header */
 630         foundHdr = addrharvest_check_hdr( listHdr, buf );
 631
 632         /* Read all header lines. Only add reqd ones to list */
 633         while( TRUE ) {
 634                 gchar *p;
 635
 636                 if( foundHdr ) {
 637                         p = g_strdup( buf );
 638                         list = g_slist_append( list, p );
 639                 }
 640
 641                 /* Read first character */
 642                 ch = fgetc( fp );
 643                 if( ch == ' ' || ch == '\t' ) {
 644                         /* Continuation character - read into buffer */
 645                         if( fgets( buf, MSG_BUFFSIZE, fp ) == NULL ) {
 646                                 break;
 647                         }
 648                 }
 649                 else {
 650                         if( ch == EOF ) {
 651                                 *done = TRUE;
 652                         }
 653                         else {
 654                                 /* Push back character for next header */
 655                                 ungetc( ch, fp );
 656                         }
 657                         break;
 658                 }
 659         }
 660
 661         return list;
 662 }
 663
 664 /*
 665  * Read specified file into address book.
 666  * Enter:  harvester Harvester object.
 667  *         fileName  File to read.
 668  *         cache     Address cache to load.
 669  * Return: Status.
 670  */
 671 static gint addrharvest_readfile(
 672                 AddressHarvester *harvester, const gchar *fileName,
 673                 AddressCache *cache, GList *listHdr )
 674 {
 675         gint retVal;
 676         FILE *msgFile;
 677         gchar *buf, *addr, *p;
 678         HeaderEntry *entry;
 679         GSList *list;
 680         gboolean done;
 681
 682         msgFile = g_fopen( fileName, "rb" );
 683         if( ! msgFile ) {
 684                 /* Cannot open file */
 685                 retVal = MGU_OPEN_FILE;
 686                 return retVal;
 687         }
 688
 689         done = FALSE;
 690         while( TRUE ) {
 691                 list = addrharvest_get_header( msgFile, listHdr, &done );
 692                 if( done ) break;
 693
 694                 if( list == NULL ) {
 695                         continue;
 696                 }
 697
 698                 buf = mgu_list_coalesce( list );
 699                 mgu_free_list( list );
 700
 701                 if(( p = strchr( buf, ':' ) ) != NULL ) {
 702                         addr = p + 1;
 703                         *p = '\0';
 704
 705                         entry = addrharvest_find( harvester, buf );
 706                         if( entry && entry->selected ) {
 707                                 /* Sanitize control characters */
 708                                 p = addr;
 709                                 while( *p ) {
 710                                         if( *p == '\r' || *p == '\n' || *p == '\t' )
 711                                                 *p = ' ';
 712                                         p++;
 713                                 }
 714                                 addrharvest_parse_address(
 715                                         harvester, entry, cache, addr );
 716                         }
 717                 }
 718                 g_free( buf );
 719         }
 720
 721         fclose( msgFile );
 722         return MGU_SUCCESS;
 723 }
 724
 725 /*
 726  * Read all files in specified directory into address book. Directories are
 727  * traversed recursively if necessary.
 728  * Enter:  harvester Harvester object.
 729  *         cache     Address cache to load.
 730  *         msgList   List of message numbers, or NULL to process folder.
 731  *         dir       Directory to process.
 732  */
 733 static void addrharvest_harvest_dir(
 734         AddressHarvester *harvester, AddressCache *cache, GList *listHdr,
 735         gchar *dir )
 736 {
 737         DIR *dp;
 738         struct dirent *d;
 739         struct stat s;
 740         gint num;
 741
 742         if( ( dp = opendir( dir ) ) == NULL ) {
 743                 return;
 744         }
 745
 746         /* Process directory */
 747         chdir( dir );
 748         while( ( d = readdir( dp ) ) != NULL ) {
 749                 stat( d->d_name, &s );
 750                 if( S_ISDIR( s.st_mode ) ) {
 751                         if( harvester->folderRecurse ) {
 752                                 if( strstr( DIR_IGNORE, d->d_name ) != NULL )
 753                                         continue;
 754                                 addrharvest_harvest_dir(
 755                                         harvester, cache, listHdr, d->d_name );
 756                         }
 757                 }
 758                 if( S_ISREG( s.st_mode ) ) {
 759                         if( ( num = to_number( d->d_name ) ) >= 0 ) {
 760                                 addrharvest_readfile(
 761                                         harvester, d->d_name, cache, listHdr );
 762                         }
 763                 }
 764         }
 765         chdir( ".." );
 766         closedir( dp );
 767 }
 768
 769 /*
 770  * Read list of files in specified directory into address book.
 771  * Enter:  harvester Harvester object.
 772  *         cache     Address cache to load.
 773  *         msgList   List of message numbers, or NULL to process folder.
 774  */
 775 static void addrharvest_harvest_list(
 776         AddressHarvester *harvester, AddressCache *cache, GList *listHdr,
 777         GList *msgList )
 778 {
 779         DIR *dp;
 780         gint num;
 781         GList *node;
 782         gchar msgNum[ MSGNUM_BUFFSIZE ];
 783
 784         if( ( dp = opendir( harvester->path ) ) == NULL ) {
 785                 return;
 786         }
 787
 788         /* Process message list */
 789         chdir( harvester->path );
 790         node = msgList;
 791         while( node ) {
 792                 num = GPOINTER_TO_UINT( node->data );
 793                 sprintf( msgNum, "%d", num );
 794                 addrharvest_readfile( harvester, msgNum, cache, listHdr );
 795                 node = g_list_next( node );
 796         }
 797         closedir( dp );
 798 }
 799
 800 /*
 801  * ============================================================================
 802  * Read all files in specified directory into address book.
 803  * Enter:  harvester Harvester object.
 804  *         cache     Address cache to load.
 805  *         msgList   List of message numbers, or NULL to process folder.
 806  * Return: Status.
 807  * ============================================================================
 808  */
 809 gint addrharvest_harvest(
 810         AddressHarvester *harvester, AddressCache *cache, GList *msgList )
 811 {
 812         gint retVal;
 813         GList *node;
 814         GList *listHdr;
 815
 816         retVal = MGU_BAD_ARGS;
 817         g_return_val_if_fail( harvester != NULL, retVal );
 818         g_return_val_if_fail( cache != NULL, retVal );
 819         g_return_val_if_fail( harvester->path != NULL, retVal );
 820
 821         /* Clear cache */
 822         addrcache_clear( cache );
 823         cache->dataRead = FALSE;
 824
 825         /* Build list of headers of interest */
 826         listHdr = NULL;
 827         node = harvester->headerTable;
 828         while( node ) {
 829                 HeaderEntry *entry;
 830
 831                 entry = node->data;
 832                 if( entry->selected ) {
 833                         gchar *p;
 834
 835                         p = g_strdup( entry->header );
 836                         g_strdown( p );
 837                         listHdr = g_list_append( listHdr, p );
 838                 }
 839                 node = g_list_next( node );
 840         }
 841
 842         /* Process directory/files */
 843         if( msgList == NULL ) {
 844                 addrharvest_harvest_dir( harvester, cache, listHdr, harvester->path );
 845         }
 846         else {
 847                 addrharvest_harvest_list( harvester, cache, listHdr, msgList );
 848         }
 849         mgu_free_dlist( listHdr );
 850
 851         /* Mark cache */
 852         cache->modified = FALSE;
 853         cache->dataRead = TRUE;
 854
 855         return retVal;
 856 }
 857
 858 /*
 859  * ============================================================================
 860  * Test whether any headers have been selected for processing.
 861  * Enter:  harvester Harvester object.
 862  * Return: TRUE if a header was selected, FALSE if none were selected.
 863  * ============================================================================
 864  */
 865 gboolean addrharvest_check_header( AddressHarvester *harvester ) {
 866         gboolean retVal;
 867         GList *node;
 868
 869         retVal = FALSE;
 870         g_return_val_if_fail( harvester != NULL, retVal );
 871
 872         node = harvester->headerTable;
 873         while( node ) {
 874                 HeaderEntry *entry;
 875
 876                 entry = ( HeaderEntry * ) node->data;
 877                 if( entry->selected ) return TRUE;
 878                 node = g_list_next( node );
 879         }
 880         return retVal;
 881 }
 882
 883 /*
 884  * ============================================================================
 885  * End of Source.
 886  * ============================================================================
 887  */
 888
 889