1 /* File Transfer Protocol support.
2 Copyright (C) 1995, 1996, 1997, 1998, 2000, 2001
3 Free Software Foundation, Inc.
5 This file is part of GNU Wget.
7 GNU Wget is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 2 of the License, or
10 (at your option) any later version.
12 GNU Wget is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License
18 along with Wget; if not, write to the Free Software
19 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
21 In addition, as a special exception, the Free Software Foundation
22 gives permission to link the code of its release of Wget with the
23 OpenSSL project's "OpenSSL" library (or with modified versions of it
24 that use the same license as the "OpenSSL" library), and distribute
25 the linked executables. You must obey the GNU General Public License
26 in all respects for all of the code used other than "OpenSSL". If you
27 modify this file, you may extend this exception to your version of the
28 file, but you are not obligated to do so. If you do not wish to do
29 so, delete this exception statement from your version. */
43 #include <sys/types.h>
55 #include "convert.h" /* for downloaded_file */
56 #include "recur.h" /* for INFINITE_RECURSION */
62 extern LARGE_INT total_downloaded_bytes;
64 /* File where the "ls -al" listing will be saved. */
65 #define LIST_FILENAME ".listing"
67 extern char ftp_last_respline[];
69 extern FILE *output_stream;
70 extern int output_stream_regular;
74 int st; /* connection status */
75 int cmd; /* command code */
76 int csock; /* control connection socket */
77 double dltime; /* time of the download in msecs */
78 enum stype rs; /* remote system reported by ftp server */
79 char *id; /* initial directory */
80 char *target; /* target file name */
81 struct url *proxy; /* FTWK-style proxy */
85 /* Look for regexp "( *[0-9]+ *byte" (literal parenthesis) anywhere in
86 the string S, and return the number converted to wgint, if found, 0
89 ftp_expected_bytes (const char *s)
95 while (*s && *s != '(')
99 ++s; /* skip the '(' */
100 res = str_to_wgint (s, (char **) &s, 10);
103 while (*s && ISSPACE (*s))
107 if (TOLOWER (*s) != 'b')
109 if (strncasecmp (s, "byte", 4))
119 * This function sets up a passive data connection with the FTP server.
120 * It is merely a wrapper around ftp_epsv, ftp_lpsv and ftp_pasv.
123 ftp_do_pasv (int csock, ip_address *addr, int *port)
127 /* We need to determine the address family and need to call
128 getpeername, so while we're at it, store the address to ADDR.
129 ftp_pasv and ftp_lpsv can simply override it. */
130 if (!socket_ip_address (csock, addr, ENDPOINT_PEER))
133 /* If our control connection is over IPv6, then we first try EPSV and then
134 * LPSV if the former is not supported. If the control connection is over
135 * IPv4, we simply issue the good old PASV request. */
139 if (!opt.server_response)
140 logputs (LOG_VERBOSE, "==> PASV ... ");
141 err = ftp_pasv (csock, addr, port);
144 if (!opt.server_response)
145 logputs (LOG_VERBOSE, "==> EPSV ... ");
146 err = ftp_epsv (csock, addr, port);
148 /* If EPSV is not supported try LPSV */
149 if (err == FTPNOPASV)
151 if (!opt.server_response)
152 logputs (LOG_VERBOSE, "==> LPSV ... ");
153 err = ftp_lpsv (csock, addr, port);
164 * This function sets up an active data connection with the FTP server.
165 * It is merely a wrapper around ftp_eprt, ftp_lprt and ftp_port.
168 ftp_do_port (int csock, int *local_sock)
173 if (!socket_ip_address (csock, &cip, ENDPOINT_PEER))
176 /* If our control connection is over IPv6, then we first try EPRT and then
177 * LPRT if the former is not supported. If the control connection is over
178 * IPv4, we simply issue the good old PORT request. */
182 if (!opt.server_response)
183 logputs (LOG_VERBOSE, "==> PORT ... ");
184 err = ftp_port (csock, local_sock);
187 if (!opt.server_response)
188 logputs (LOG_VERBOSE, "==> EPRT ... ");
189 err = ftp_eprt (csock, local_sock);
191 /* If EPRT is not supported try LPRT */
192 if (err == FTPPORTERR)
194 if (!opt.server_response)
195 logputs (LOG_VERBOSE, "==> LPRT ... ");
196 err = ftp_lprt (csock, local_sock);
207 ftp_do_pasv (int csock, ip_address *addr, int *port)
209 if (!opt.server_response)
210 logputs (LOG_VERBOSE, "==> PASV ... ");
211 return ftp_pasv (csock, addr, port);
215 ftp_do_port (int csock, int *local_sock)
217 if (!opt.server_response)
218 logputs (LOG_VERBOSE, "==> PORT ... ");
219 return ftp_port (csock, local_sock);
224 print_length (wgint size, wgint start, int authoritative)
226 logprintf (LOG_VERBOSE, _("Length: %s"), with_thousand_seps (size));
228 logprintf (LOG_VERBOSE, " (%s)", human_readable (size));
232 logprintf (LOG_VERBOSE, _(", %s (%s) remaining"),
233 with_thousand_seps (size - start),
234 human_readable (size - start));
236 logprintf (LOG_VERBOSE, _(", %s remaining"),
237 with_thousand_seps (size - start));
239 logputs (LOG_VERBOSE, !authoritative ? _(" (unauthoritative)\n") : "\n");
242 /* Retrieves a file with denoted parameters through opening an FTP
243 connection to the server. It always closes the data connection,
244 and closes the control connection in case of error. */
246 getftp (struct url *u, wgint *len, wgint restval, ccon *con)
248 int csock, dtsock, local_sock, res;
249 uerr_t err = RETROK; /* appease the compiler */
251 char *user, *passwd, *respline;
254 int pasv_mode_open = 0;
255 wgint expected_bytes = 0;
260 assert (con != NULL);
261 assert (con->target != NULL);
263 /* Debug-check of the sanity of the request by making sure that LIST
264 and RETR are never both requested (since we can handle only one
266 assert (!((cmd & DO_LIST) && (cmd & DO_RETR)));
267 /* Make sure that at least *something* is requested. */
268 assert ((cmd & (DO_LIST | DO_CWD | DO_RETR | DO_LOGIN)) != 0);
272 search_netrc (u->host, (const char **)&user, (const char **)&passwd, 1);
273 user = user ? user : (opt.ftp_user ? opt.ftp_user : opt.user);
274 if (!user) user = "anonymous";
275 passwd = passwd ? passwd : (opt.ftp_passwd ? opt.ftp_passwd : opt.passwd);
276 if (!passwd) passwd = "-wget@";
282 if (!(cmd & DO_LOGIN))
284 else /* cmd & DO_LOGIN */
287 char *host = con->proxy ? con->proxy->host : u->host;
288 int port = con->proxy ? con->proxy->port : u->port;
289 char *logname = user;
293 /* If proxy is in use, log in as username@target-site. */
294 logname = concat_strings (user, "@", u->host, (char *) 0);
297 /* Login to the server: */
299 /* First: Establish the control connection. */
301 csock = connect_to_host (host, port);
305 return (retryable_socket_connect_error (errno)
306 ? CONERROR : CONIMPOSSIBLE);
308 if (cmd & LEAVE_PENDING)
313 /* Second: Login with proper USER/PASS sequence. */
314 logprintf (LOG_VERBOSE, _("Logging in as %s ... "), escnonprint (user));
315 if (opt.server_response)
316 logputs (LOG_ALWAYS, "\n");
317 err = ftp_login (csock, logname, passwd);
322 /* FTPRERR, FTPSRVERR, WRITEFAILED, FTPLOGREFUSED, FTPLOGINC */
326 logputs (LOG_VERBOSE, "\n");
327 logputs (LOG_NOTQUIET, _("\
328 Error in server response, closing control connection.\n"));
333 logputs (LOG_VERBOSE, "\n");
334 logputs (LOG_NOTQUIET, _("Error in server greeting.\n"));
339 logputs (LOG_VERBOSE, "\n");
340 logputs (LOG_NOTQUIET,
341 _("Write failed, closing control connection.\n"));
346 logputs (LOG_VERBOSE, "\n");
347 logputs (LOG_NOTQUIET, _("The server refuses login.\n"));
350 return FTPLOGREFUSED;
352 logputs (LOG_VERBOSE, "\n");
353 logputs (LOG_NOTQUIET, _("Login incorrect.\n"));
358 if (!opt.server_response)
359 logputs (LOG_VERBOSE, _("Logged in!\n"));
364 /* Third: Get the system type */
365 if (!opt.server_response)
366 logprintf (LOG_VERBOSE, "==> SYST ... ");
367 err = ftp_syst (csock, &con->rs);
372 logputs (LOG_VERBOSE, "\n");
373 logputs (LOG_NOTQUIET, _("\
374 Error in server response, closing control connection.\n"));
379 logputs (LOG_VERBOSE, "\n");
380 logputs (LOG_NOTQUIET,
381 _("Server error, can't determine system type.\n"));
384 /* Everything is OK. */
389 if (!opt.server_response && err != FTPSRVERR)
390 logputs (LOG_VERBOSE, _("done. "));
392 /* Fourth: Find the initial ftp directory */
394 if (!opt.server_response)
395 logprintf (LOG_VERBOSE, "==> PWD ... ");
396 err = ftp_pwd (csock, &con->id);
401 logputs (LOG_VERBOSE, "\n");
402 logputs (LOG_NOTQUIET, _("\
403 Error in server response, closing control connection.\n"));
408 /* PWD unsupported -- assume "/". */
409 xfree_null (con->id);
410 con->id = xstrdup ("/");
413 /* Everything is OK. */
418 /* VMS will report something like "PUB$DEVICE:[INITIAL.FOLDER]".
419 Convert it to "/INITIAL/FOLDER" */
420 if (con->rs == ST_VMS)
422 char *path = strchr (con->id, '[');
423 char *pathend = path ? strchr (path + 1, ']') : NULL;
424 if (!path || !pathend)
425 DEBUGP (("Initial VMS directory not in the form [...]!\n"));
428 char *idir = con->id;
429 DEBUGP (("Preprocessing the initial VMS directory\n"));
430 DEBUGP ((" old = '%s'\n", con->id));
431 /* We do the conversion in-place by copying the stuff
432 between [ and ] to the beginning, and changing dots
433 to slashes at the same time. */
435 for (++path; path < pathend; path++, idir++)
436 *idir = *path == '.' ? '/' : *path;
438 DEBUGP ((" new = '%s'\n\n", con->id));
441 if (!opt.server_response)
442 logputs (LOG_VERBOSE, _("done.\n"));
444 /* Fifth: Set the FTP type. */
445 type_char = ftp_process_type (u->params);
446 if (!opt.server_response)
447 logprintf (LOG_VERBOSE, "==> TYPE %c ... ", type_char);
448 err = ftp_type (csock, type_char);
449 /* FTPRERR, WRITEFAILED, FTPUNKNOWNTYPE */
453 logputs (LOG_VERBOSE, "\n");
454 logputs (LOG_NOTQUIET, _("\
455 Error in server response, closing control connection.\n"));
460 logputs (LOG_VERBOSE, "\n");
461 logputs (LOG_NOTQUIET,
462 _("Write failed, closing control connection.\n"));
467 logputs (LOG_VERBOSE, "\n");
468 logprintf (LOG_NOTQUIET,
469 _("Unknown type `%c', closing control connection.\n"),
475 /* Everything is OK. */
480 if (!opt.server_response)
481 logputs (LOG_VERBOSE, _("done. "));
487 logputs (LOG_VERBOSE, _("==> CWD not needed.\n"));
490 char *target = u->dir;
492 DEBUGP (("changing working directory\n"));
494 /* Change working directory. To change to a non-absolute
495 Unix directory, we need to prepend initial directory
496 (con->id) to it. Absolute directories "just work".
498 A relative directory is one that does not begin with '/'
499 and, on non-Unix OS'es, one that doesn't begin with
502 This is not done for OS400, which doesn't use
503 "/"-delimited directories, nor does it support directory
504 hierarchies. "CWD foo" followed by "CWD bar" leaves us
505 in "bar", not in "foo/bar", as would be customary
509 && !(con->rs != ST_UNIX
510 && ISALPHA (target[0])
512 && con->rs != ST_OS400)
514 int idlen = strlen (con->id);
517 /* Strip trailing slash(es) from con->id. */
518 while (idlen > 0 && con->id[idlen - 1] == '/')
520 p = ntarget = (char *)alloca (idlen + 1 + strlen (u->dir) + 1);
521 memcpy (p, con->id, idlen);
526 DEBUGP (("Prepended initial PWD to relative path:\n"));
527 DEBUGP ((" pwd: '%s'\n old: '%s'\n new: '%s'\n",
528 con->id, target, ntarget));
532 /* If the FTP host runs VMS, we will have to convert the absolute
533 directory path in UNIX notation to absolute directory path in
534 VMS notation as VMS FTP servers do not like UNIX notation of
535 absolute paths. "VMS notation" is [dir.subdir.subsubdir]. */
537 if (con->rs == ST_VMS)
540 char *ntarget = (char *)alloca (strlen (target) + 2);
541 /* We use a converted initial dir, so directories in
542 TARGET will be separated with slashes, something like
543 "/INITIAL/FOLDER/DIR/SUBDIR". Convert that to
544 "[INITIAL.FOLDER.DIR.SUBDIR]". */
545 strcpy (ntarget, target);
546 assert (*ntarget == '/');
548 for (tmpp = ntarget + 1; *tmpp; tmpp++)
553 DEBUGP (("Changed file name to VMS syntax:\n"));
554 DEBUGP ((" Unix: '%s'\n VMS: '%s'\n", target, ntarget));
558 if (!opt.server_response)
559 logprintf (LOG_VERBOSE, "==> CWD %s ... ", escnonprint (target));
560 err = ftp_cwd (csock, target);
561 /* FTPRERR, WRITEFAILED, FTPNSFOD */
565 logputs (LOG_VERBOSE, "\n");
566 logputs (LOG_NOTQUIET, _("\
567 Error in server response, closing control connection.\n"));
572 logputs (LOG_VERBOSE, "\n");
573 logputs (LOG_NOTQUIET,
574 _("Write failed, closing control connection.\n"));
579 logputs (LOG_VERBOSE, "\n");
580 logprintf (LOG_NOTQUIET, _("No such directory `%s'.\n\n"),
581 escnonprint (u->dir));
590 if (!opt.server_response)
591 logputs (LOG_VERBOSE, _("done.\n"));
594 else /* do not CWD */
595 logputs (LOG_VERBOSE, _("==> CWD not required.\n"));
597 if ((cmd & DO_RETR) && restval && *len == 0)
601 if (!opt.server_response)
602 logprintf (LOG_VERBOSE, "==> SIZE %s ... ", escnonprint (u->file));
605 err = ftp_size (csock, u->file, len);
611 logputs (LOG_VERBOSE, "\n");
612 logputs (LOG_NOTQUIET, _("\
613 Error in server response, closing control connection.\n"));
618 /* Everything is OK. */
623 if (!opt.server_response)
624 logputs (LOG_VERBOSE, _("done.\n"));
627 /* If anything is to be retrieved, PORT (or PASV) must be sent. */
628 if (cmd & (DO_LIST | DO_RETR))
630 if (opt.ftp_pasv > 0)
632 ip_address passive_addr;
634 err = ftp_do_pasv (csock, &passive_addr, &passive_port);
635 /* FTPRERR, WRITEFAILED, FTPNOPASV, FTPINVPASV */
639 logputs (LOG_VERBOSE, "\n");
640 logputs (LOG_NOTQUIET, _("\
641 Error in server response, closing control connection.\n"));
646 logputs (LOG_VERBOSE, "\n");
647 logputs (LOG_NOTQUIET,
648 _("Write failed, closing control connection.\n"));
653 logputs (LOG_VERBOSE, "\n");
654 logputs (LOG_NOTQUIET, _("Cannot initiate PASV transfer.\n"));
657 logputs (LOG_VERBOSE, "\n");
658 logputs (LOG_NOTQUIET, _("Cannot parse PASV response.\n"));
667 DEBUGP (("trying to connect to %s port %d\n",
668 pretty_print_address (&passive_addr),
670 dtsock = connect_to_ip (&passive_addr, passive_port, NULL);
673 int save_errno = errno;
676 logprintf (LOG_VERBOSE, _("couldn't connect to %s port %d: %s\n"),
677 pretty_print_address (&passive_addr), passive_port,
678 strerror (save_errno));
679 return (retryable_socket_connect_error (save_errno)
680 ? CONERROR : CONIMPOSSIBLE);
683 pasv_mode_open = 1; /* Flag to avoid accept port */
684 if (!opt.server_response)
685 logputs (LOG_VERBOSE, _("done. "));
689 if (!pasv_mode_open) /* Try to use a port command if PASV failed */
691 err = ftp_do_port (csock, &local_sock);
692 /* FTPRERR, WRITEFAILED, bindport (FTPSYSERR), HOSTERR,
697 logputs (LOG_VERBOSE, "\n");
698 logputs (LOG_NOTQUIET, _("\
699 Error in server response, closing control connection.\n"));
703 fd_close (local_sock);
706 logputs (LOG_VERBOSE, "\n");
707 logputs (LOG_NOTQUIET,
708 _("Write failed, closing control connection.\n"));
712 fd_close (local_sock);
715 logputs (LOG_VERBOSE, "\n");
716 logprintf (LOG_NOTQUIET, "socket: %s\n", strerror (errno));
720 fd_close (local_sock);
723 logputs (LOG_VERBOSE, "\n");
724 logprintf (LOG_NOTQUIET, _("Bind error (%s).\n"),
729 logputs (LOG_VERBOSE, "\n");
730 logputs (LOG_NOTQUIET, _("Invalid PORT.\n"));
734 fd_close (local_sock);
741 if (!opt.server_response)
742 logputs (LOG_VERBOSE, _("done. "));
744 } /* cmd & (DO_LIST | DO_RETR) */
746 /* Restart if needed. */
747 if (restval && (cmd & DO_RETR))
749 if (!opt.server_response)
750 logprintf (LOG_VERBOSE, "==> REST %s ... ",
751 number_to_static_string (restval));
752 err = ftp_rest (csock, restval);
754 /* FTPRERR, WRITEFAILED, FTPRESTFAIL */
758 logputs (LOG_VERBOSE, "\n");
759 logputs (LOG_NOTQUIET, _("\
760 Error in server response, closing control connection.\n"));
764 fd_close (local_sock);
767 logputs (LOG_VERBOSE, "\n");
768 logputs (LOG_NOTQUIET,
769 _("Write failed, closing control connection.\n"));
773 fd_close (local_sock);
776 logputs (LOG_VERBOSE, _("\nREST failed, starting from scratch.\n"));
784 if (err != FTPRESTFAIL && !opt.server_response)
785 logputs (LOG_VERBOSE, _("done. "));
786 } /* restval && cmd & DO_RETR */
790 /* If we're in spider mode, don't really retrieve anything. The
791 fact that we got to this point should be proof enough that
792 the file exists, vaguely akin to HTTP's concept of a "HEAD"
799 fd_close (local_sock);
805 if (!opt.server_response)
808 logputs (LOG_VERBOSE, "\n");
809 logprintf (LOG_VERBOSE, "==> RETR %s ... ", escnonprint (u->file));
813 err = ftp_retr (csock, u->file);
814 /* FTPRERR, WRITEFAILED, FTPNSFOD */
818 logputs (LOG_VERBOSE, "\n");
819 logputs (LOG_NOTQUIET, _("\
820 Error in server response, closing control connection.\n"));
824 fd_close (local_sock);
827 logputs (LOG_VERBOSE, "\n");
828 logputs (LOG_NOTQUIET,
829 _("Write failed, closing control connection.\n"));
833 fd_close (local_sock);
836 logputs (LOG_VERBOSE, "\n");
837 logprintf (LOG_NOTQUIET, _("No such file `%s'.\n\n"),
838 escnonprint (u->file));
840 fd_close (local_sock);
848 if (!opt.server_response)
849 logputs (LOG_VERBOSE, _("done.\n"));
850 expected_bytes = ftp_expected_bytes (ftp_last_respline);
855 if (!opt.server_response)
856 logputs (LOG_VERBOSE, "==> LIST ... ");
857 /* As Maciej W. Rozycki (macro@ds2.pg.gda.pl) says, `LIST'
858 without arguments is better than `LIST .'; confirmed by
860 err = ftp_list (csock, NULL);
861 /* FTPRERR, WRITEFAILED */
865 logputs (LOG_VERBOSE, "\n");
866 logputs (LOG_NOTQUIET, _("\
867 Error in server response, closing control connection.\n"));
871 fd_close (local_sock);
874 logputs (LOG_VERBOSE, "\n");
875 logputs (LOG_NOTQUIET,
876 _("Write failed, closing control connection.\n"));
880 fd_close (local_sock);
883 logputs (LOG_VERBOSE, "\n");
884 logprintf (LOG_NOTQUIET, _("No such file or directory `%s'.\n\n"),
887 fd_close (local_sock);
894 if (!opt.server_response)
895 logputs (LOG_VERBOSE, _("done.\n"));
896 expected_bytes = ftp_expected_bytes (ftp_last_respline);
897 } /* cmd & DO_LIST */
899 if (!(cmd & (DO_LIST | DO_RETR)) || (opt.spider && !(cmd & DO_LIST)))
902 /* Some FTP servers return the total length of file after REST
903 command, others just return the remaining size. */
904 if (*len && restval && expected_bytes
905 && (expected_bytes == *len - restval))
907 DEBUGP (("Lying FTP server found, adjusting.\n"));
908 expected_bytes = *len;
911 /* If no transmission was required, then everything is OK. */
912 if (!pasv_mode_open) /* we are not using pasive mode so we need
915 /* Wait for the server to connect to the address we're waiting
917 dtsock = accept_connection (local_sock);
920 logprintf (LOG_NOTQUIET, "accept: %s\n", strerror (errno));
925 /* Open the file -- if output_stream is set, use it instead. */
926 if (!output_stream || con->cmd & DO_LIST)
928 mkalldirs (con->target);
930 rotate_backups (con->target);
933 fp = fopen (con->target, "ab");
934 else if (opt.noclobber || opt.always_rest || opt.timestamping || opt.dirstruct
935 || opt.output_document)
936 fp = fopen (con->target, "wb");
939 fp = fopen_excl (con->target, 1);
940 if (!fp && errno == EEXIST)
942 /* We cannot just invent a new name and use it (which is
943 what functions like unique_create typically do)
944 because we told the user we'd use this name.
945 Instead, return and retry the download. */
946 logprintf (LOG_NOTQUIET, _("%s has sprung into existence.\n"),
951 fd_close (local_sock);
952 return FOPEN_EXCL_ERR;
957 logprintf (LOG_NOTQUIET, "%s: %s\n", con->target, strerror (errno));
961 fd_close (local_sock);
970 print_length (*len, restval, 1);
971 expected_bytes = *len; /* for get_contents/show_progress */
973 else if (expected_bytes)
974 print_length (expected_bytes, restval, 0);
976 /* Get the contents of the document. */
978 if (restval && rest_failed)
979 flags |= rb_skip_startpos;
982 res = fd_read_body (dtsock, fp,
983 expected_bytes ? expected_bytes - restval : 0,
984 restval, &rd_size, len, &con->dltime, flags);
986 tms = time_str (NULL);
987 tmrate = retr_rate (rd_size, con->dltime, 0);
988 /* Close data connection socket. */
990 fd_close (local_sock);
991 /* Close the local file. */
993 /* Close or flush the file. We have to be careful to check for
994 error here. Checking the result of fwrite() is not enough --
995 errors could go unnoticed! */
997 if (!output_stream || con->cmd & DO_LIST)
998 flush_res = fclose (fp);
1000 flush_res = fflush (fp);
1001 if (flush_res == EOF)
1005 /* If get_contents couldn't write to fp, bail out. */
1008 logprintf (LOG_NOTQUIET, _("%s: %s, closing control connection.\n"),
1009 con->target, strerror (errno));
1016 logprintf (LOG_NOTQUIET, _("%s (%s) - Data connection: %s; "),
1017 tms, tmrate, strerror (errno));
1018 if (opt.server_response)
1019 logputs (LOG_ALWAYS, "\n");
1022 /* Get the server to tell us if everything is retrieved. */
1023 err = ftp_response (csock, &respline);
1027 /* The control connection is decidedly closed. Print the time
1028 only if it hasn't already been printed. */
1030 logprintf (LOG_NOTQUIET, "%s (%s) - ", tms, tmrate);
1031 logputs (LOG_NOTQUIET, _("Control connection closed.\n"));
1032 /* If there is an error on the control connection, close it, but
1033 return FTPRETRINT, since there is a possibility that the
1034 whole file was retrieved nevertheless (but that is for
1035 ftp_loop_internal to decide). */
1039 } /* err != FTPOK */
1040 /* If retrieval failed for any reason, return FTPRETRINT, but do not
1041 close socket, since the control connection is still alive. If
1042 there is something wrong with the control connection, it will
1043 become apparent later. */
1044 if (*respline != '2')
1048 logprintf (LOG_NOTQUIET, "%s (%s) - ", tms, tmrate);
1049 logputs (LOG_NOTQUIET, _("Data transfer aborted.\n"));
1056 /* What now? The data connection was erroneous, whereas the
1057 response says everything is OK. We shall play it safe. */
1061 if (!(cmd & LEAVE_PENDING))
1063 /* Closing the socket is faster than sending 'QUIT' and the
1064 effect is the same. */
1068 /* If it was a listing, and opt.server_response is true,
1070 if (opt.server_response && (con->cmd & DO_LIST))
1072 mkalldirs (con->target);
1073 fp = fopen (con->target, "r");
1075 logprintf (LOG_ALWAYS, "%s: %s\n", con->target, strerror (errno));
1079 /* The lines are being read with read_whole_line because of
1080 no-buffering on opt.lfile. */
1081 while ((line = read_whole_line (fp)) != NULL)
1083 logprintf (LOG_ALWAYS, "%s\n", escnonprint (line));
1088 } /* con->cmd & DO_LIST && server_response */
1090 return RETRFINISHED;
1093 /* A one-file FTP loop. This is the part where FTP retrieval is
1094 retried, and retried, and retried, and...
1096 This loop either gets commands from con, or (if ON_YOUR_OWN is
1097 set), makes them up to retrieve the file given by the URL. */
1099 ftp_loop_internal (struct url *u, struct fileinfo *f, ccon *con)
1102 wgint restval, len = 0;
1104 char *tmrate = NULL;
1109 con->target = url_file_name (u);
1111 if (opt.noclobber && file_exists_p (con->target))
1113 logprintf (LOG_VERBOSE,
1114 _("File `%s' already there, not retrieving.\n"), con->target);
1115 /* If the file is there, we suppose it's retrieved OK. */
1119 /* Remove it if it's a link. */
1120 remove_link (con->target);
1121 if (!opt.output_document)
1124 locf = opt.output_document;
1128 if (con->st & ON_YOUR_OWN)
1129 con->st = ON_YOUR_OWN;
1131 orig_lp = con->cmd & LEAVE_PENDING ? 1 : 0;
1136 /* Increment the pass counter. */
1138 sleep_between_retrievals (count);
1139 if (con->st & ON_YOUR_OWN)
1142 con->cmd |= (DO_RETR | LEAVE_PENDING);
1143 if (con->csock != -1)
1144 con->cmd &= ~ (DO_LOGIN | DO_CWD);
1146 con->cmd |= (DO_LOGIN | DO_CWD);
1148 else /* not on your own */
1150 if (con->csock != -1)
1151 con->cmd &= ~DO_LOGIN;
1153 con->cmd |= DO_LOGIN;
1154 if (con->st & DONE_CWD)
1155 con->cmd &= ~DO_CWD;
1160 /* Decide whether or not to restart. */
1163 restval = len; /* start where the previous run left off */
1164 else if (opt.always_rest
1165 && stat (locf, &st) == 0
1166 && S_ISREG (st.st_mode))
1167 restval = st.st_size;
1169 /* Get the current time string. */
1170 tms = time_str (NULL);
1171 /* Print fetch message, if opt.verbose. */
1174 char *hurl = url_string (u, 1);
1178 sprintf (tmp, _("(try:%2d)"), count);
1179 logprintf (LOG_VERBOSE, "--%s-- %s\n %s => `%s'\n",
1180 tms, hurl, tmp, locf);
1182 ws_changetitle (hurl);
1186 /* Send getftp the proper length, if fileinfo was provided. */
1191 err = getftp (u, &len, restval, con);
1193 if (con->csock != -1)
1194 con->st &= ~DONE_CWD;
1196 con->st |= DONE_CWD;
1200 case HOSTERR: case CONIMPOSSIBLE: case FWRITEERR: case FOPENERR:
1201 case FTPNSFOD: case FTPLOGINC: case FTPNOPASV: case CONTNOTSUPPORTED:
1202 /* Fatal errors, give up. */
1204 case CONSOCKERR: case CONERROR: case FTPSRVERR: case FTPRERR:
1205 case WRITEFAILED: case FTPUNKNOWNTYPE: case FTPSYSERR:
1206 case FTPPORTERR: case FTPLOGREFUSED: case FTPINVPASV:
1207 case FOPEN_EXCL_ERR:
1208 printwhat (count, opt.ntry);
1209 /* non-fatal errors */
1210 if (err == FOPEN_EXCL_ERR)
1212 /* Re-determine the file name. */
1213 xfree_null (con->target);
1214 con->target = url_file_name (u);
1219 /* If the control connection was closed, the retrieval
1220 will be considered OK if f->size == len. */
1221 if (!f || len != f->size)
1223 printwhat (count, opt.ntry);
1234 tms = time_str (NULL);
1236 tmrate = retr_rate (len - restval, con->dltime, 0);
1238 /* If we get out of the switch above without continue'ing, we've
1239 successfully downloaded a file. Remember this fact. */
1240 downloaded_file (FILE_DOWNLOADED_NORMALLY, locf);
1242 if (con->st & ON_YOUR_OWN)
1244 fd_close (con->csock);
1248 logprintf (LOG_VERBOSE, _("%s (%s) - `%s' saved [%s]\n\n"),
1249 tms, tmrate, locf, number_to_static_string (len));
1250 if (!opt.verbose && !opt.quiet)
1252 /* Need to hide the password from the URL. The `if' is here
1253 so that we don't do the needless allocation every
1255 char *hurl = url_string (u, 1);
1256 logprintf (LOG_NONVERBOSE, "%s URL: %s [%s] -> \"%s\" [%d]\n",
1257 tms, hurl, number_to_static_string (len), locf, count);
1261 if ((con->cmd & DO_LIST))
1262 /* This is a directory listing file. */
1264 if (!opt.remove_listing)
1265 /* --dont-remove-listing was specified, so do count this towards the
1266 number of bytes and files downloaded. */
1268 total_downloaded_bytes += len;
1272 /* Deletion of listing files is not controlled by --delete-after, but
1273 by the more specific option --dont-remove-listing, and the code
1274 to do this deletion is in another function. */
1276 else if (!opt.spider)
1277 /* This is not a directory listing file. */
1279 /* Unlike directory listing files, don't pretend normal files weren't
1280 downloaded if they're going to be deleted. People seeding proxies,
1281 for instance, may want to know how many bytes and files they've
1282 downloaded through it. */
1283 total_downloaded_bytes += len;
1286 if (opt.delete_after)
1289 Removing file due to --delete-after in ftp_loop_internal():\n"));
1290 logprintf (LOG_VERBOSE, _("Removing %s.\n"), locf);
1292 logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
1296 /* Restore the original leave-pendingness. */
1298 con->cmd |= LEAVE_PENDING;
1300 con->cmd &= ~LEAVE_PENDING;
1302 } while (!opt.ntry || (count < opt.ntry));
1304 if (con->csock != -1 && (con->st & ON_YOUR_OWN))
1306 fd_close (con->csock);
1312 /* Return the directory listing in a reusable format. The directory
1313 is specifed in u->dir. */
1315 ftp_get_listing (struct url *u, ccon *con, struct fileinfo **f)
1318 char *uf; /* url file name */
1319 char *lf; /* list file name */
1320 char *old_target = con->target;
1322 con->st &= ~ON_YOUR_OWN;
1323 con->cmd |= (DO_LIST | LEAVE_PENDING);
1324 con->cmd &= ~DO_RETR;
1326 /* Find the listing file name. We do it by taking the file name of
1327 the URL and replacing the last component with the listing file
1329 uf = url_file_name (u);
1330 lf = file_merge (uf, LIST_FILENAME);
1332 DEBUGP ((_("Using `%s' as listing tmp file.\n"), lf));
1335 err = ftp_loop_internal (u, NULL, con);
1336 con->target = old_target;
1339 *f = ftp_parse_ls (lf, con->rs);
1342 if (opt.remove_listing)
1345 logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
1347 logprintf (LOG_VERBOSE, _("Removed `%s'.\n"), lf);
1350 con->cmd &= ~DO_LIST;
1354 static uerr_t ftp_retrieve_dirs PARAMS ((struct url *, struct fileinfo *,
1356 static uerr_t ftp_retrieve_glob PARAMS ((struct url *, ccon *, int));
1357 static struct fileinfo *delelement PARAMS ((struct fileinfo *,
1358 struct fileinfo **));
1359 static void freefileinfo PARAMS ((struct fileinfo *f));
1361 /* Retrieve a list of files given in struct fileinfo linked list. If
1362 a file is a symbolic link, do not retrieve it, but rather try to
1363 set up a similar link on the local disk, if the symlinks are
1366 If opt.recursive is set, after all files have been retrieved,
1367 ftp_retrieve_dirs will be called to retrieve the directories. */
1369 ftp_retrieve_list (struct url *u, struct fileinfo *f, ccon *con)
1371 static int depth = 0;
1373 struct fileinfo *orig;
1378 /* Increase the depth. */
1380 if (opt.reclevel != INFINITE_RECURSION && depth > opt.reclevel)
1382 DEBUGP ((_("Recursion depth %d exceeded max. depth %d.\n"),
1383 depth, opt.reclevel));
1391 con->st &= ~ON_YOUR_OWN;
1392 if (!(con->st & DONE_CWD))
1395 con->cmd &= ~DO_CWD;
1396 con->cmd |= (DO_RETR | LEAVE_PENDING);
1399 con->cmd |= DO_LOGIN;
1401 con->cmd &= ~DO_LOGIN;
1403 err = RETROK; /* in case it's not used */
1407 char *old_target, *ofile;
1409 if (opt.quota && total_downloaded_bytes > opt.quota)
1414 old_target = con->target;
1416 ofile = xstrdup (u->file);
1417 url_set_file (u, f->name);
1419 con->target = url_file_name (u);
1423 if (opt.timestamping && f->type == FT_PLAINFILE)
1426 /* If conversion of HTML files retrieved via FTP is ever implemented,
1427 we'll need to stat() <file>.orig here when -K has been specified.
1428 I'm not implementing it now since files on an FTP server are much
1429 more likely than files on an HTTP server to legitimately have a
1431 if (!stat (con->target, &st))
1435 /* Else, get it from the file. */
1436 local_size = st.st_size;
1439 /* Modification time granularity is 2 seconds for Windows, so
1440 increase local time by 1 second for later comparison. */
1443 /* Compare file sizes only for servers that tell us correct
1444 values. Assumme sizes being equal for servers that lie
1446 cor_val = (con->rs == ST_UNIX || con->rs == ST_WINNT);
1447 eq_size = cor_val ? (local_size == f->size) : 1 ;
1448 if (f->tstamp <= tml && eq_size)
1450 /* Remote file is older, file sizes can be compared and
1452 logprintf (LOG_VERBOSE, _("\
1453 Remote file no newer than local file `%s' -- not retrieving.\n"), con->target);
1458 /* Remote file is newer or sizes cannot be matched */
1459 logprintf (LOG_VERBOSE, _("\
1460 Remote file is newer than local file `%s' -- retrieving.\n\n"),
1465 /* Sizes do not match */
1466 logprintf (LOG_VERBOSE, _("\
1467 The sizes do not match (local %s) -- retrieving.\n\n"),
1468 number_to_static_string (local_size));
1471 } /* opt.timestamping && f->type == FT_PLAINFILE */
1475 /* If opt.retr_symlinks is defined, we treat symlinks as
1476 if they were normal files. There is currently no way
1477 to distinguish whether they might be directories, and
1479 if (!opt.retr_symlinks)
1483 logputs (LOG_NOTQUIET,
1484 _("Invalid name of the symlink, skipping.\n"));
1488 /* Check whether we already have the correct
1490 int rc = lstat (con->target, &st);
1493 size_t len = strlen (f->linkto) + 1;
1494 if (S_ISLNK (st.st_mode))
1496 char *link_target = (char *)alloca (len);
1497 size_t n = readlink (con->target, link_target, len);
1499 && (memcmp (link_target, f->linkto, n) == 0))
1501 logprintf (LOG_VERBOSE, _("\
1502 Already have correct symlink %s -> %s\n\n"),
1503 con->target, escnonprint (f->linkto));
1509 logprintf (LOG_VERBOSE, _("Creating symlink %s -> %s\n"),
1510 con->target, escnonprint (f->linkto));
1511 /* Unlink before creating symlink! */
1512 unlink (con->target);
1513 if (symlink (f->linkto, con->target) == -1)
1514 logprintf (LOG_NOTQUIET, "symlink: %s\n", strerror (errno));
1515 logputs (LOG_VERBOSE, "\n");
1516 } /* have f->linkto */
1517 #else /* not HAVE_SYMLINK */
1518 logprintf (LOG_NOTQUIET,
1519 _("Symlinks not supported, skipping symlink `%s'.\n"),
1521 #endif /* not HAVE_SYMLINK */
1523 else /* opt.retr_symlinks */
1526 err = ftp_loop_internal (u, f, con);
1527 } /* opt.retr_symlinks */
1531 logprintf (LOG_NOTQUIET, _("Skipping directory `%s'.\n"),
1532 escnonprint (f->name));
1535 /* Call the retrieve loop. */
1537 err = ftp_loop_internal (u, f, con);
1540 logprintf (LOG_NOTQUIET, _("%s: unknown/unsupported file type.\n"),
1541 escnonprint (f->name));
1545 /* Set the time-stamp information to the local file. Symlinks
1546 are not to be stamped because it sets the stamp on the
1548 if (!(f->type == FT_SYMLINK && !opt.retr_symlinks)
1551 && file_exists_p (con->target))
1553 /* #### This code repeats in http.c and ftp.c. Move it to a
1555 const char *fl = NULL;
1556 if (opt.output_document)
1558 if (output_stream_regular)
1559 fl = opt.output_document;
1564 touch (fl, f->tstamp);
1566 else if (f->tstamp == -1)
1567 logprintf (LOG_NOTQUIET, _("%s: corrupt time-stamp.\n"), con->target);
1569 if (f->perms && f->type == FT_PLAINFILE && dlthis)
1571 if (opt.preserve_perm)
1572 chmod (con->target, f->perms);
1575 DEBUGP (("Unrecognized permissions for %s.\n", con->target));
1577 xfree (con->target);
1578 con->target = old_target;
1580 url_set_file (u, ofile);
1583 /* Break on fatals. */
1584 if (err == QUOTEXC || err == HOSTERR || err == FWRITEERR)
1586 con->cmd &= ~ (DO_CWD | DO_LOGIN);
1590 /* We do not want to call ftp_retrieve_dirs here */
1591 if (opt.recursive &&
1592 !(opt.reclevel != INFINITE_RECURSION && depth >= opt.reclevel))
1593 err = ftp_retrieve_dirs (u, orig, con);
1594 else if (opt.recursive)
1595 DEBUGP ((_("Will not retrieve dirs since depth is %d (max %d).\n"),
1596 depth, opt.reclevel));
1601 /* Retrieve the directories given in a file list. This function works
1602 by simply going through the linked list and calling
1603 ftp_retrieve_glob on each directory entry. The function knows
1604 about excluded directories. */
1606 ftp_retrieve_dirs (struct url *u, struct fileinfo *f, ccon *con)
1608 char *container = NULL;
1609 int container_size = 0;
1611 for (; f; f = f->next)
1614 char *odir, *newdir;
1616 if (opt.quota && total_downloaded_bytes > opt.quota)
1618 if (f->type != FT_DIRECTORY)
1621 /* Allocate u->dir off stack, but reallocate only if a larger
1622 string is needed. It's a pity there's no "realloca" for an
1623 item on the bottom of the stack. */
1624 size = strlen (u->dir) + 1 + strlen (f->name) + 1;
1625 if (size > container_size)
1626 container = (char *)alloca (size);
1631 || (*odir == '/' && *(odir + 1) == '\0'))
1632 /* If ODIR is empty or just "/", simply append f->name to
1633 ODIR. (In the former case, to preserve u->dir being
1634 relative; in the latter case, to avoid double slash.) */
1635 sprintf (newdir, "%s%s", odir, f->name);
1637 /* Else, use a separator. */
1638 sprintf (newdir, "%s/%s", odir, f->name);
1640 DEBUGP (("Composing new CWD relative to the initial directory.\n"));
1641 DEBUGP ((" odir = '%s'\n f->name = '%s'\n newdir = '%s'\n\n",
1642 odir, f->name, newdir));
1643 if (!accdir (newdir, ALLABS))
1645 logprintf (LOG_VERBOSE, _("\
1646 Not descending to `%s' as it is excluded/not-included.\n"),
1647 escnonprint (newdir));
1651 con->st &= ~DONE_CWD;
1653 odir = xstrdup (u->dir); /* because url_set_dir will free
1655 url_set_dir (u, newdir);
1656 ftp_retrieve_glob (u, con, GLOB_GETALL);
1657 url_set_dir (u, odir);
1660 /* Set the time-stamp? */
1663 if (opt.quota && total_downloaded_bytes > opt.quota)
1669 /* Return non-zero if S has a leading '/' or contains '../' */
1671 has_insecure_name_p (const char *s)
1676 if (strstr (s, "../") != 0)
1682 /* A near-top-level function to retrieve the files in a directory.
1683 The function calls ftp_get_listing, to get a linked list of files.
1684 Then it weeds out the file names that do not match the pattern.
1685 ftp_retrieve_list is called with this updated list as an argument.
1687 If the argument ACTION is GLOB_GETONE, just download the file (but
1688 first get the listing, so that the time-stamp is heeded); if it's
1689 GLOB_GLOBALL, use globbing; if it's GLOB_GETALL, download the whole
1692 ftp_retrieve_glob (struct url *u, ccon *con, int action)
1694 struct fileinfo *f, *start;
1697 con->cmd |= LEAVE_PENDING;
1699 res = ftp_get_listing (u, con, &start);
1702 /* First: weed out that do not conform the global rules given in
1703 opt.accepts and opt.rejects. */
1704 if (opt.accepts || opt.rejects)
1709 if (f->type != FT_DIRECTORY && !acceptable (f->name))
1711 logprintf (LOG_VERBOSE, _("Rejecting `%s'.\n"),
1712 escnonprint (f->name));
1713 f = delelement (f, &start);
1719 /* Remove all files with possible harmful names */
1723 if (has_insecure_name_p (f->name))
1725 logprintf (LOG_VERBOSE, _("Rejecting `%s'.\n"),
1726 escnonprint (f->name));
1727 f = delelement (f, &start);
1732 /* Now weed out the files that do not match our globbing pattern.
1733 If we are dealing with a globbing pattern, that is. */
1734 if (*u->file && (action == GLOB_GLOBALL || action == GLOB_GETONE))
1741 matchres = fnmatch (u->file, f->name, 0);
1744 logprintf (LOG_NOTQUIET, "%s: %s\n", con->target,
1748 if (matchres == FNM_NOMATCH)
1749 f = delelement (f, &start); /* delete the element from the list */
1751 f = f->next; /* leave the element in the list */
1755 freefileinfo (start);
1756 return RETRBADPATTERN;
1761 /* Just get everything. */
1762 ftp_retrieve_list (u, start, con);
1766 if (action == GLOB_GLOBALL)
1769 /* #### This message SUCKS. We should see what was the
1770 reason that nothing was retrieved. */
1771 logprintf (LOG_VERBOSE, _("No matches on pattern `%s'.\n"),
1772 escnonprint (u->file));
1774 else /* GLOB_GETONE or GLOB_GETALL */
1776 /* Let's try retrieving it anyway. */
1777 con->st |= ON_YOUR_OWN;
1778 res = ftp_loop_internal (u, NULL, con);
1782 freefileinfo (start);
1783 if (opt.quota && total_downloaded_bytes > opt.quota)
1786 /* #### Should we return `res' here? */
1790 /* The wrapper that calls an appropriate routine according to contents
1791 of URL. Inherently, its capabilities are limited on what can be
1792 encoded into a URL. */
1794 ftp_loop (struct url *u, int *dt, struct url *proxy)
1796 ccon con; /* FTP connection */
1804 con.st = ON_YOUR_OWN;
1809 /* If the file name is empty, the user probably wants a directory
1810 index. We'll provide one, properly HTML-ized. Unless
1811 opt.htmlify is 0, of course. :-) */
1812 if (!*u->file && !opt.recursive)
1815 res = ftp_get_listing (u, &con, &f);
1819 if (opt.htmlify && !opt.spider)
1821 char *filename = (opt.output_document
1822 ? xstrdup (opt.output_document)
1823 : (con.target ? xstrdup (con.target)
1824 : url_file_name (u)));
1825 res = ftp_index (filename, u, f);
1826 if (res == FTPOK && opt.verbose)
1828 if (!opt.output_document)
1832 if (stat (filename, &st) == 0)
1836 logprintf (LOG_NOTQUIET,
1837 _("Wrote HTML-ized index to `%s' [%s].\n"),
1838 filename, number_to_static_string (sz));
1841 logprintf (LOG_NOTQUIET,
1842 _("Wrote HTML-ized index to `%s'.\n"),
1855 /* Treat the URL as a pattern if the file name part of the
1856 URL path contains wildcards. (Don't check for u->file
1857 because it is unescaped and therefore doesn't leave users
1858 the option to escape literal '*' as %2A.) */
1859 char *file_part = strrchr (u->path, '/');
1861 file_part = u->path;
1862 ispattern = has_wildcards_p (file_part);
1864 if (ispattern || opt.recursive || opt.timestamping)
1866 /* ftp_retrieve_glob is a catch-all function that gets called
1867 if we need globbing, time-stamping or recursion. Its
1868 third argument is just what we really need. */
1869 res = ftp_retrieve_glob (u, &con,
1870 ispattern ? GLOB_GLOBALL : GLOB_GETONE);
1873 res = ftp_loop_internal (u, NULL, &con);
1879 /* If a connection was left, quench it. */
1880 if (con.csock != -1)
1881 fd_close (con.csock);
1882 xfree_null (con.id);
1884 xfree_null (con.target);
1889 /* Delete an element from the fileinfo linked list. Returns the
1890 address of the next element, or NULL if the list is exhausted. It
1891 can modify the start of the list. */
1892 static struct fileinfo *
1893 delelement (struct fileinfo *f, struct fileinfo **start)
1895 struct fileinfo *prev = f->prev;
1896 struct fileinfo *next = f->next;
1899 xfree_null (f->linkto);
1911 /* Free the fileinfo linked list of files. */
1913 freefileinfo (struct fileinfo *f)
1917 struct fileinfo *next = f->next;