src/addrharvest.c

   1 /*
   2  * Sylpheed -- a GTK+ based, lightweight, and fast e-mail client
   3  * Copyright (C) 2002-2007 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 "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_claws_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  * Insert address into cache.
 274  * Enter: harvester Harvester object.
 275  *        entry     Header object.
 276  *        cache     Address cache to load.
 277  *        name      Name.
 278  *        address   eMail address.
 279  */
 280 static void addrharvest_insert_cache(
 281                 AddressHarvester *harvester, HeaderEntry *entry,
 282                 AddressCache *cache, const gchar *name,
 283                 const gchar *address )
 284 {
 285         ItemPerson *person;
 286         ItemFolder *folder;
 287         gchar *folderName;
 288         gboolean newFolder;
 289         gint cnt;
 290         gchar *key, *value;
 291
 292         newFolder = FALSE;
 293         folder = entry->folder;
 294         if( folder == NULL ) {
 295                 newFolder = TRUE;       /* No folder yet */
 296         }
 297         if( entry->count % harvester->folderSize == 0 ) {
 298                 newFolder = TRUE;       /* Folder is full */
 299         }
 300
 301         /* Insert address */
 302         key = g_utf8_strdown( address, -1 );
 303         person = g_hash_table_lookup( harvester->dupTable, key );
 304         if( person ) {
 305                 /* Update existing person to use longest name */
 306                 value = ADDRITEM_NAME(person);
 307                 if( strlen( name ) > strlen( value ) ) {
 308                         addritem_person_set_common_name( person, name );
 309                 }
 310                 g_free( key );
 311         }
 312         else {
 313                 /* Folder if required */
 314                 if( newFolder ) {
 315                         cnt = 1 + ( entry->count / harvester->folderSize );
 316                         folderName =g_strdup_printf( "%s (%d)",
 317                                         entry->header, cnt );
 318                         folder = addritem_create_item_folder();
 319                         addritem_folder_set_name( folder, folderName );
 320                         addritem_folder_set_remarks( folder, "" );
 321                         addrcache_id_folder( cache, folder );
 322                         addrcache_add_folder( cache, folder );
 323                         entry->folder = folder;
 324                         g_free( folderName );
 325                 }
 326
 327                 /* Insert entry */
 328                 person = addrcache_add_contact(
 329                                 cache, folder, name, address, "" );
 330                 g_hash_table_insert( harvester->dupTable, key, person );
 331                 entry->count++;
 332         }
 333         addritem_parse_first_last( person );
 334 }
 335
 336 /*
 337  * Remove specified string from name.
 338  * Enter: name Name.
 339  *        str  String to remove.
 340  */
 341 static void addrharvest_del_email( gchar *name, gchar *str ) {
 342         gchar *p;
 343         gint lenn, lenr;
 344
 345         lenr = strlen( str );
 346         while((p = strcasestr( name, str )) != NULL) {
 347                 lenn = strlen( p );
 348                 memmove( p, p + lenr, lenn );
 349         }
 350 }
 351
 352 /*
 353  * Find position of at (@) character in buffer.
 354  * Enter:  buffer Start of buffer.
 355  * Return: Position of at character, or NULL if not found.
 356  * Note: This function searches for the last occurrence of an 'at' character
 357  * prior to a valid delimiter character for the end of address. This enables
 358  * an address to be found where it is also used as the name of the
 359  * recipient. For example:
 360  *     "axle.rose@netscape.com" <axle.rose@netscape.com>
 361  * The last occurrence of the at character is detected.
 362  */
 363 static gchar *addrharvest_find_at( const gchar *buffer ) {
 364         gchar *atCh;
 365         gchar *p;
 366
 367         atCh = strchr( buffer, '@' );
 368         if( atCh ) {
 369                 /* Search forward for another one */
 370                 p = atCh + 1;
 371                 while( *p ) {
 372                         if( *p == '>' ) {
 373                                 break;
 374                         }
 375                         if( *p == ',' ) {
 376                                 break;
 377                         }
 378                         if( *p == '\n' ) {
 379                                 break;
 380                         }
 381                         if( *p == '@' ) {
 382                                 atCh = p;
 383                                 break;
 384                         }
 385                         p++;
 386                 }
 387         }
 388         return atCh;
 389 }
 390
 391 /*
 392  * Find start and end of address string.
 393  * Enter: buf Start address of buffer to process (not modified).
 394  *        atp Pointer to email at (@) character.
 395  *        bp  Pointer to start of email address (returned).
 396  *        ep  Pointer to end of email address (returned).
 397  */
 398 static void addrharvest_find_address(
 399                 const gchar *buf, const gchar *atp, const gchar **bp,
 400                 const gchar **ep )
 401 {
 402         const gchar *p;
 403
 404         /* Find first non-separator char */
 405         *bp = NULL;
 406         p = buf;
 407         while( TRUE ) {
 408                 if( strchr( ",; \n\r", *p ) == NULL ) break;
 409                 p++;
 410         }
 411         *bp = p;
 412
 413         /* Search forward for end of address */
 414         *ep = NULL;
 415         p = atp + 1;
 416         while( TRUE ) {
 417                 if( strchr( ",;", *p ) ) break;
 418                 p++;
 419         }
 420         *ep = p;
 421 }
 422
 423 /*
 424  * Extract E-Mail address from buffer. If found, address is removed from
 425  * buffer.
 426  * Enter:  buffer Address buffer.
 427  * Return: E-Mail address, or NULL if none found. Must g_free() when done.
 428  */
 429 static gchar *addrharvest_extract_address( gchar *buffer ) {
 430         gchar *addr;
 431         gchar *atCh, *p, *bp, *ep;
 432         gint len;
 433
 434         addr = NULL;
 435         atCh = addrharvest_find_at( buffer );
 436         if( atCh ) {
 437                 /* Search back for start of address */
 438                 bp = NULL;
 439                 p = atCh;
 440                 while( p >= buffer ) {
 441                         bp = p;
 442                         if( *p == '<' ) {
 443                                 *p = ' ';
 444                                 bp++;
 445                                 break;
 446                         }
 447                         p--;
 448                 }
 449
 450                 /* Search fwd for end */
 451                 ep = NULL;
 452                 ep = p = atCh;
 453                 while( *p ) {
 454                         if( *p == '>' ) {
 455                                 *p = ' ';
 456                                 break;
 457                         }
 458                         else if( *p == ' ' ) {
 459                                 break;
 460                         }
 461                         ep = p;
 462                         p++;
 463                 }
 464
 465                 /* Extract email */
 466                 if( bp != NULL ) {
 467                         len = ( ep - bp );
 468                         if( len > 0 ) {
 469                                 addr = g_strndup( bp, len + 1 );
 470                                 memmove( bp, ep, len );
 471                                 *bp = ' ';
 472                         }
 473                 }
 474         }
 475         return addr;
 476 }
 477
 478 /*
 479  * Parse address from header buffer creating address in cache.
 480  * Enter: harvester Harvester object.
 481  *        entry     Header object.
 482  *        cache     Address cache to load.
 483  *        hdrBuf    Pointer to header buffer.
 484  */
 485 static void addrharvest_parse_address(
 486                 AddressHarvester *harvester, HeaderEntry *entry,
 487                 AddressCache *cache, const gchar *hdrBuf )
 488 {
 489         gchar buffer[ ADDR_BUFFSIZE + 2 ];
 490         const gchar *bp;
 491         const gchar *ep;
 492         gchar *atCh, *email, *name;
 493         gint bufLen;
 494
 495         /* Search for an address */
 496         while((atCh = addrharvest_find_at( hdrBuf )) != NULL) {
 497                 /* Find addres string */
 498                 addrharvest_find_address( hdrBuf, atCh, &bp, &ep );
 499
 500                 /* Copy into buffer */
 501                 bufLen = ( size_t ) ( ep - bp );
 502                 if( bufLen > ADDR_BUFFSIZE ) {
 503                         bufLen = ADDR_BUFFSIZE;
 504                 }
 505                 strncpy( buffer, bp, bufLen );
 506                 buffer[ bufLen ] = '\0';
 507                 buffer[ bufLen + 1 ] = '\0';
 508                 buffer[ bufLen + 2 ] = '\0';
 509
 510                 /* Extract address from buffer */
 511                 email = addrharvest_extract_address( buffer );
 512                 if( email ) {
 513                         /* Unescape characters */
 514                         mgu_str_unescape( buffer );
 515
 516                         /* Remove noise characaters */
 517                         addrharvest_del_email( buffer, REM_NAME_STRING );
 518                         addrharvest_del_email( buffer, REM_NAME_STRING2 );
 519
 520                         /* Remove leading trailing quotes and spaces */
 521                         mgu_str_ltc2space( buffer, '\"', '\"' );
 522                         mgu_str_ltc2space( buffer, '\'', '\'' );
 523                         mgu_str_ltc2space( buffer, '\"', '\"' );
 524                         mgu_str_ltc2space( buffer, '(', ')' );
 525                         g_strstrip( buffer );
 526
 527                         if( g_ascii_strcasecmp( buffer, email ) == 0 ) {
 528                                 name = g_strdup("");
 529                         }
 530                         else {
 531                                 name = buffer;
 532                                 name = conv_unmime_header(buffer, NULL);
 533                         }
 534
 535                         /* Insert into address book */
 536                         addrharvest_insert_cache(
 537                                 harvester, entry, cache, name, email );
 538                         g_free( email );
 539                         g_free( name );
 540                 }
 541                 hdrBuf = ep;
 542         }
 543 }
 544
 545 /*
 546  * Test whether buffer contains a header that appears in header list.
 547  * Enter: listHdr Header list.
 548  *        buf     Header buffer.
 549  * Return: TRUE if header in list.
 550  */
 551 static gboolean addrharvest_check_hdr( GList *listHdr, gchar *buf ) {
 552         gboolean retVal;
 553         GList *node;
 554         gchar *p, *hdr, *nhdr;
 555         gint len;
 556
 557         retVal = FALSE;
 558         p = strchr( buf, ':' );
 559         if( p ) {
 560                 len = ( size_t ) ( p - buf );
 561                 hdr = g_strndup( buf, len );
 562                 node = listHdr;
 563                 while( node ) {
 564                         nhdr = node->data;
 565                         if (g_ascii_strncasecmp(nhdr, hdr, strlen(nhdr)) == 0 ) {
 566                                 retVal = TRUE;
 567                                 break;
 568                         }
 569                         node = g_list_next( node );
 570                 }
 571                 g_free( hdr );
 572         }
 573         return retVal;
 574 }
 575
 576 /*
 577  * Read header into a linked list of lines.
 578  * Enter:  fp      File to read.
 579  *         listHdr List of header lines of interest.
 580  *         done    End of headers or end of file reached.
 581  * Return: Linked list of lines.
 582  */
 583 static GSList *addrharvest_get_header( FILE *fp, GList *listHdr, gboolean *done ) {
 584         GSList *list;
 585         gchar buf[ MSG_BUFFSIZE + 2 ];
 586         gint ch;
 587         gboolean foundHdr;
 588
 589         list = NULL;
 590
 591         /* Read line */
 592         if( fgets( buf, MSG_BUFFSIZE, fp ) == NULL ) {
 593                 *done = TRUE;
 594                 return list;
 595         }
 596
 597         /* Test for end of headers */
 598         if( buf[0] == '\r' || buf[0] == '\n' ) {
 599                 *done = TRUE;
 600                 return list;
 601         }
 602
 603         /* Test whether required header */
 604         foundHdr = addrharvest_check_hdr( listHdr, buf );
 605
 606         /* Read all header lines. Only add reqd ones to list */
 607         while( TRUE ) {
 608                 gchar *p;
 609
 610                 if( foundHdr ) {
 611                         p = g_strdup( buf );
 612                         list = g_slist_append( list, p );
 613                 }
 614
 615                 /* Read first character */
 616                 ch = fgetc( fp );
 617                 if( ch == ' ' || ch == '\t' ) {
 618                         /* Continuation character - read into buffer */
 619                         if( fgets( buf, MSG_BUFFSIZE, fp ) == NULL ) {
 620                                 break;
 621                         }
 622                 }
 623                 else {
 624                         if( ch == EOF ) {
 625                                 *done = TRUE;
 626                         }
 627                         else {
 628                                 /* Push back character for next header */
 629                                 ungetc( ch, fp );
 630                         }
 631                         break;
 632                 }
 633         }
 634
 635         return list;
 636 }
 637
 638 /*
 639  * Read specified file into address book.
 640  * Enter:  harvester Harvester object.
 641  *         fileName  File to read.
 642  *         cache     Address cache to load.
 643  * Return: Status.
 644  */
 645 static gint addrharvest_readfile(
 646                 AddressHarvester *harvester, const gchar *fileName,
 647                 AddressCache *cache, GList *listHdr )
 648 {
 649         gint retVal;
 650         FILE *msgFile;
 651         gchar *buf, *addr, *p;
 652         HeaderEntry *entry;
 653         GSList *list;
 654         gboolean done;
 655
 656         msgFile = g_fopen( fileName, "rb" );
 657         if( ! msgFile ) {
 658                 /* Cannot open file */
 659                 retVal = MGU_OPEN_FILE;
 660                 return retVal;
 661         }
 662
 663         done = FALSE;
 664         while( TRUE ) {
 665                 list = addrharvest_get_header( msgFile, listHdr, &done );
 666                 if( done ) break;
 667
 668                 if( list == NULL ) {
 669                         continue;
 670                 }
 671
 672                 buf = mgu_list_coalesce( list );
 673                 mgu_free_list( list );
 674
 675                 if(( p = strchr( buf, ':' ) ) != NULL ) {
 676                         addr = p + 1;
 677                         *p = '\0';
 678
 679                         entry = addrharvest_find( harvester, buf );
 680                         if( entry && entry->selected ) {
 681                                 /* Sanitize control characters */
 682                                 p = addr;
 683                                 while( *p ) {
 684                                         if( *p == '\r' || *p == '\n' || *p == '\t' )
 685                                                 *p = ' ';
 686                                         p++;
 687                                 }
 688                                 addrharvest_parse_address(
 689                                         harvester, entry, cache, addr );
 690                         }
 691                 }
 692                 g_free( buf );
 693         }
 694
 695         fclose( msgFile );
 696         return MGU_SUCCESS;
 697 }
 698
 699 /*
 700  * Read all files in specified directory into address book. Directories are
 701  * traversed recursively if necessary.
 702  * Enter:  harvester Harvester object.
 703  *         cache     Address cache to load.
 704  *         msgList   List of message numbers, or NULL to process folder.
 705  *         dir       Directory to process.
 706  */
 707 static void addrharvest_harvest_dir(
 708         AddressHarvester *harvester, AddressCache *cache, GList *listHdr,
 709         gchar *dir )
 710 {
 711         DIR *dp;
 712         struct dirent *d;
 713         struct stat s;
 714         gint num;
 715
 716         if( ( dp = opendir( dir ) ) == NULL ) {
 717                 return;
 718         }
 719
 720         /* Process directory */
 721         chdir( dir );
 722         while( ( d = readdir( dp ) ) != NULL ) {
 723                 g_stat( d->d_name, &s );
 724                 if( S_ISDIR( s.st_mode ) ) {
 725                         if( harvester->folderRecurse ) {
 726                                 if( strstr( DIR_IGNORE, d->d_name ) != NULL )
 727                                         continue;
 728                                 addrharvest_harvest_dir(
 729                                         harvester, cache, listHdr, d->d_name );
 730                         }
 731                 }
 732                 if( S_ISREG( s.st_mode ) ) {
 733                         if( ( num = to_number( d->d_name ) ) >= 0 ) {
 734                                 addrharvest_readfile(
 735                                         harvester, d->d_name, cache, listHdr );
 736                         }
 737                 }
 738         }
 739         chdir( ".." );
 740         closedir( dp );
 741 }
 742
 743 /*
 744  * Read list of files in specified directory into address book.
 745  * Enter:  harvester Harvester object.
 746  *         cache     Address cache to load.
 747  *         msgList   List of message numbers, or NULL to process folder.
 748  */
 749 static void addrharvest_harvest_list(
 750         AddressHarvester *harvester, AddressCache *cache, GList *listHdr,
 751         GList *msgList )
 752 {
 753         DIR *dp;
 754         gint num;
 755         GList *node;
 756         gchar msgNum[ MSGNUM_BUFFSIZE ];
 757
 758         if( ( dp = opendir( harvester->path ) ) == NULL ) {
 759                 return;
 760         }
 761
 762         /* Process message list */
 763         chdir( harvester->path );
 764         node = msgList;
 765         while( node ) {
 766                 num = GPOINTER_TO_UINT( node->data );
 767                 sprintf( msgNum, "%d", num );
 768                 addrharvest_readfile( harvester, msgNum, cache, listHdr );
 769                 node = g_list_next( node );
 770         }
 771         closedir( dp );
 772 }
 773
 774 /*
 775  * ============================================================================
 776  * Read all files in specified directory into address book.
 777  * Enter:  harvester Harvester object.
 778  *         cache     Address cache to load.
 779  *         msgList   List of message numbers, or NULL to process folder.
 780  * Return: Status.
 781  * ============================================================================
 782  */
 783 gint addrharvest_harvest(
 784         AddressHarvester *harvester, AddressCache *cache, GList *msgList )
 785 {
 786         gint retVal;
 787         GList *node;
 788         GList *listHdr;
 789
 790         retVal = MGU_BAD_ARGS;
 791         g_return_val_if_fail( harvester != NULL, retVal );
 792         g_return_val_if_fail( cache != NULL, retVal );
 793         g_return_val_if_fail( harvester->path != NULL, retVal );
 794
 795         /* Clear cache */
 796         addrcache_clear( cache );
 797         cache->dataRead = FALSE;
 798
 799         /* Build list of headers of interest */
 800         listHdr = NULL;
 801         node = harvester->headerTable;
 802         while( node ) {
 803                 HeaderEntry *entry;
 804
 805                 entry = node->data;
 806                 if( entry->selected ) {
 807                         gchar *p;
 808
 809                         p = g_utf8_strdown( entry->header, -1 );
 810                         listHdr = g_list_append( listHdr, p );
 811                 }
 812                 node = g_list_next( node );
 813         }
 814
 815         /* Process directory/files */
 816         if( msgList == NULL ) {
 817                 addrharvest_harvest_dir( harvester, cache, listHdr, harvester->path );
 818         }
 819         else {
 820                 addrharvest_harvest_list( harvester, cache, listHdr, msgList );
 821         }
 822         mgu_free_dlist( listHdr );
 823
 824         /* Mark cache */
 825         cache->modified = FALSE;
 826         cache->dataRead = TRUE;
 827
 828         return retVal;
 829 }
 830
 831 /*
 832  * ============================================================================
 833  * Test whether any headers have been selected for processing.
 834  * Enter:  harvester Harvester object.
 835  * Return: TRUE if a header was selected, FALSE if none were selected.
 836  * ============================================================================
 837  */
 838 gboolean addrharvest_check_header( AddressHarvester *harvester ) {
 839         gboolean retVal;
 840         GList *node;
 841
 842         retVal = FALSE;
 843         g_return_val_if_fail( harvester != NULL, retVal );
 844
 845         node = harvester->headerTable;
 846         while( node ) {
 847                 HeaderEntry *entry;
 848
 849                 entry = ( HeaderEntry * ) node->data;
 850                 if( entry->selected ) return TRUE;
 851                 node = g_list_next( node );
 852         }
 853         return retVal;
 854 }
 855
 856 /*
 857  * ============================================================================
 858  * End of Source.
 859  * ============================================================================
 860  */
 861
 862