blob: 3fd1df124f749c187daf7d283c734437c0f73cac [file] [log] [blame]
Eric Andersen96700832000-09-04 15:15:55 +00001/* vi: set sw=4 ts=4: */
2/*
Eric Andersen79757c92001-04-05 21:45:54 +00003 * wget - retrieve a file using HTTP or FTP
Eric Andersen96700832000-09-04 15:15:55 +00004 *
Eric Andersen4e573f42000-11-14 23:29:24 +00005 * Chip Rosenthal Covad Communications <chip@laserlink.net>
Eric Andersenb520e082000-10-03 00:21:45 +00006 *
Eric Andersen96700832000-09-04 15:15:55 +00007 */
8
Eric Andersen96700832000-09-04 15:15:55 +00009#include <stdio.h>
Eric Andersendff9d542001-01-26 02:04:49 +000010#include <errno.h>
Eric Andersen96700832000-09-04 15:15:55 +000011#include <stdlib.h>
12#include <unistd.h>
13#include <ctype.h>
14#include <string.h>
Eric Andersenb520e082000-10-03 00:21:45 +000015#include <unistd.h>
16#include <signal.h>
17#include <sys/ioctl.h>
Eric Andersen96700832000-09-04 15:15:55 +000018
Eric Andersenb520e082000-10-03 00:21:45 +000019#include <sys/time.h>
Eric Andersen96700832000-09-04 15:15:55 +000020#include <sys/types.h>
21#include <sys/stat.h>
22#include <sys/socket.h>
23#include <netinet/in.h>
24#include <arpa/inet.h>
25#include <netdb.h>
26
Eric Andersen50ae3102001-05-15 17:51:37 +000027#ifndef _GNU_SOURCE
28#define _GNU_SOURCE
29#endif
30#include <getopt.h>
31
Eric Andersencbe31da2001-02-20 06:14:08 +000032#include "busybox.h"
33
Eric Andersen79757c92001-04-05 21:45:54 +000034struct host_info {
35 char *host;
36 int port;
37 char *path;
38 int is_ftp;
39 char *user;
40};
41
42static void parse_url(char *url, struct host_info *h);
Glenn L McGrathffccf6e2003-12-20 01:47:18 +000043static FILE *open_socket(struct sockaddr_in *s_in);
Eric Andersen3e6ff902001-03-09 21:24:12 +000044static char *gethdr(char *buf, size_t bufsiz, FILE *fp, int *istrunc);
Eric Andersen79757c92001-04-05 21:45:54 +000045static int ftpcmd(char *s1, char *s2, FILE *fp, char *buf);
Eric Andersen96700832000-09-04 15:15:55 +000046
Eric Andersenb520e082000-10-03 00:21:45 +000047/* Globals (can be accessed from signal handlers */
48static off_t filesize = 0; /* content-length of the file */
Eric Andersen6d7fa432001-04-10 18:17:05 +000049static int chunked = 0; /* chunked transfer encoding */
Eric Andersenbdfd0d72001-10-24 05:00:29 +000050#ifdef CONFIG_FEATURE_WGET_STATUSBAR
Eric Andersen20aab262001-07-19 22:28:02 +000051static void progressmeter(int flag);
Eric Andersenb520e082000-10-03 00:21:45 +000052static char *curfile; /* Name of current file being transferred. */
53static struct timeval start; /* Time a transfer started. */
Mark Whitley30ac01c2001-04-17 18:13:16 +000054static volatile unsigned long statbytes = 0; /* Number of bytes transferred so far. */
Eric Andersenb520e082000-10-03 00:21:45 +000055/* For progressmeter() -- number of seconds before xfer considered "stalled" */
Mark Whitley59ab0252001-01-23 22:30:04 +000056static const int STALLTIME = 5;
Eric Andersenb520e082000-10-03 00:21:45 +000057#endif
Eric Andersen7d697012001-01-24 20:28:35 +000058
Eric Andersen3e6ff902001-03-09 21:24:12 +000059static void close_and_delete_outfile(FILE* output, char *fname_out, int do_continue)
Eric Andersen7d697012001-01-24 20:28:35 +000060{
61 if (output != stdout && do_continue==0) {
62 fclose(output);
63 unlink(fname_out);
64 }
65}
Eric Andersen96700832000-09-04 15:15:55 +000066
Matt Kraai854125f2001-05-09 19:15:46 +000067/* Read NMEMB elements of SIZE bytes into PTR from STREAM. Returns the
68 * number of elements read, and a short count if an eof or non-interrupt
69 * error is encountered. */
70static size_t safe_fread(void *ptr, size_t size, size_t nmemb, FILE *stream)
71{
72 size_t ret = 0;
73
74 do {
75 clearerr(stream);
76 ret += fread((char *)ptr + (ret * size), size, nmemb - ret, stream);
77 } while (ret < nmemb && ferror(stream) && errno == EINTR);
78
79 return ret;
80}
81
82/* Write NMEMB elements of SIZE bytes from PTR to STREAM. Returns the
83 * number of elements written, and a short count if an eof or non-interrupt
84 * error is encountered. */
85static size_t safe_fwrite(void *ptr, size_t size, size_t nmemb, FILE *stream)
86{
87 size_t ret = 0;
88
89 do {
90 clearerr(stream);
91 ret += fwrite((char *)ptr + (ret * size), size, nmemb - ret, stream);
92 } while (ret < nmemb && ferror(stream) && errno == EINTR);
93
94 return ret;
95}
96
97/* Read a line or SIZE - 1 bytes into S, whichever is less, from STREAM.
98 * Returns S, or NULL if an eof or non-interrupt error is encountered. */
99static char *safe_fgets(char *s, int size, FILE *stream)
100{
101 char *ret;
102
103 do {
104 clearerr(stream);
105 ret = fgets(s, size, stream);
106 } while (ret == NULL && ferror(stream) && errno == EINTR);
107
108 return ret;
109}
110
Eric Andersen79757c92001-04-05 21:45:54 +0000111#define close_delete_and_die(s...) { \
112 close_and_delete_outfile(output, fname_out, do_continue); \
Manuel Novoa III cad53642003-03-19 09:13:01 +0000113 bb_error_msg_and_die(s); }
Eric Andersen79757c92001-04-05 21:45:54 +0000114
115
Eric Andersenbdfd0d72001-10-24 05:00:29 +0000116#ifdef CONFIG_FEATURE_WGET_AUTHENTICATION
Eric Andersen79757c92001-04-05 21:45:54 +0000117/*
118 * Base64-encode character string
119 * oops... isn't something similar in uuencode.c?
120 * It would be better to use already existing code
121 */
122char *base64enc(char *p, char *buf, int len) {
123
124 char al[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"
125 "0123456789+/";
126 char *s = buf;
127
128 while(*p) {
129 if (s >= buf+len-4)
Manuel Novoa III cad53642003-03-19 09:13:01 +0000130 bb_error_msg_and_die("buffer overflow");
Eric Andersen79757c92001-04-05 21:45:54 +0000131 *(s++) = al[(*p >> 2) & 0x3F];
132 *(s++) = al[((*p << 4) & 0x30) | ((*(p+1) >> 4) & 0x0F)];
133 *s = *(s+1) = '=';
134 *(s+2) = 0;
135 if (! *(++p)) break;
136 *(s++) = al[((*p << 2) & 0x3C) | ((*(p+1) >> 6) & 0x03)];
137 if (! *(++p)) break;
138 *(s++) = al[*(p++) & 0x3F];
139 }
140
141 return buf;
142}
143#endif
144
Glenn L McGrath514aeab2003-12-19 12:08:56 +0000145#define WGET_OPT_CONTINUE 1
146#define WGET_OPT_QUIET 2
147#define WGET_OPT_PASSIVE 4
148#define WGET_OPT_OUTNAME 8
149#define WGET_OPT_HEADER 16
150#define WGET_OPT_PREFIX 32
151#define WGET_OPT_PROXY 64
152
153static const struct option wget_long_options[] = {
154 { "continue", 0, NULL, 'c' },
155 { "quiet", 0, NULL, 'q' },
156 { "passive-ftp", 0, NULL, 139 },
157 { "output-document", 1, NULL, 'O' },
158 { "header", 1, NULL, 131 },
159 { "directory-prefix",1, NULL, 'P' },
160 { "proxy", 1, NULL, 'Y' },
161 { 0, 0, 0, 0 }
162};
163
Eric Andersen96700832000-09-04 15:15:55 +0000164int wget_main(int argc, char **argv)
165{
Eric Andersen79757c92001-04-05 21:45:54 +0000166 int n, try=5, status;
Glenn L McGrath514aeab2003-12-19 12:08:56 +0000167 unsigned long opt;
Eric Andersen79757c92001-04-05 21:45:54 +0000168 int port;
Robert Griebld7760112002-05-14 23:36:45 +0000169 char *proxy = 0;
Eric Andersen8071c022001-06-21 19:45:06 +0000170 char *dir_prefix=NULL;
Eric Andersen29edd002000-12-09 16:55:35 +0000171 char *s, buf[512];
172 struct stat sbuf;
Eric Andersen50ae3102001-05-15 17:51:37 +0000173 char extra_headers[1024];
174 char *extra_headers_ptr = extra_headers;
175 int extra_headers_left = sizeof(extra_headers);
Eric Andersen79757c92001-04-05 21:45:54 +0000176 struct host_info server, target;
Eric Andersene6dc4392003-10-31 09:31:46 +0000177 struct sockaddr_in s_in;
Glenn L McGrath514aeab2003-12-19 12:08:56 +0000178 llist_t *headers_llist = NULL;
Eric Andersen79757c92001-04-05 21:45:54 +0000179
180 FILE *sfp = NULL; /* socket to web/ftp server */
181 FILE *dfp = NULL; /* socket to ftp server (data) */
Eric Andersen96700832000-09-04 15:15:55 +0000182 char *fname_out = NULL; /* where to direct output (-O) */
183 int do_continue = 0; /* continue a prev transfer (-c) */
184 long beg_range = 0L; /* range at which continue begins */
185 int got_clen = 0; /* got content-length: from server */
Eric Andersen29edd002000-12-09 16:55:35 +0000186 FILE *output; /* socket to web server */
187 int quiet_flag = FALSE; /* Be verry, verry quiet... */
Robert Griebld7760112002-05-14 23:36:45 +0000188 int noproxy = 0; /* Use proxies if env vars are set */
Glenn L McGrath514aeab2003-12-19 12:08:56 +0000189 char *proxy_flag = "on"; /* Use proxies if env vars are set */
Eric Andersen29edd002000-12-09 16:55:35 +0000190
Eric Andersen96700832000-09-04 15:15:55 +0000191 /*
192 * Crack command line.
193 */
Glenn L McGrath514aeab2003-12-19 12:08:56 +0000194 bb_opt_complementaly = "\203*";
195 bb_applet_long_options = wget_long_options;
196 opt = bb_getopt_ulflags(argc, argv, "cq\213O:\203:P:Y:", &fname_out, &headers_llist, &dir_prefix, &proxy_flag);
197 if (opt & WGET_OPT_CONTINUE) {
198 ++do_continue;
199 }
200 if (opt & WGET_OPT_QUIET) {
201 quiet_flag = TRUE;
202 }
203 if (strcmp(proxy_flag, "on") == 0) {
204 /* Use the proxy if necessary. */
205 proxy = getenv(target.is_ftp ? "ftp_proxy" : "http_proxy");
206 if (proxy)
207 parse_url(bb_xstrdup(proxy), &server);
208 }
209 if (opt & WGET_OPT_HEADER) {
210 while (headers_llist) {
211 int arglen = strlen(headers_llist->data);
212 if (extra_headers_left - arglen - 2 <= 0)
213 bb_error_msg_and_die("extra_headers buffer too small(need %i)", extra_headers_left - arglen);
214 strcpy(extra_headers_ptr, headers_llist->data);
215 extra_headers_ptr += arglen;
216 extra_headers_left -= ( arglen + 2 );
217 *extra_headers_ptr++ = '\r';
218 *extra_headers_ptr++ = '\n';
219 *(extra_headers_ptr + 1) = 0;
220 headers_llist = headers_llist->link;
Eric Andersen96700832000-09-04 15:15:55 +0000221 }
222 }
Eric Andersen96700832000-09-04 15:15:55 +0000223 if (argc - optind != 1)
Manuel Novoa III cad53642003-03-19 09:13:01 +0000224 bb_show_usage();
Eric Andersen25b669c2000-10-02 23:19:38 +0000225
Eric Andersen79757c92001-04-05 21:45:54 +0000226 parse_url(argv[optind], &target);
227 server.host = target.host;
228 server.port = target.port;
229
Eric Andersen96700832000-09-04 15:15:55 +0000230 /*
Eric Andersenf3b2b522000-12-07 22:42:11 +0000231 * Use the proxy if necessary.
Eric Andersen96700832000-09-04 15:15:55 +0000232 */
Robert Griebld7760112002-05-14 23:36:45 +0000233 if (!noproxy) {
234 proxy = getenv(target.is_ftp ? "ftp_proxy" : "http_proxy");
235 if (proxy)
Manuel Novoa III cad53642003-03-19 09:13:01 +0000236 parse_url(bb_xstrdup(proxy), &server);
Robert Griebld7760112002-05-14 23:36:45 +0000237 }
Eric Andersen29edd002000-12-09 16:55:35 +0000238
239 /* Guess an output filename */
240 if (!fname_out) {
241 fname_out =
Eric Andersenbdfd0d72001-10-24 05:00:29 +0000242#ifdef CONFIG_FEATURE_WGET_STATUSBAR
Eric Andersen29edd002000-12-09 16:55:35 +0000243 curfile =
244#endif
Manuel Novoa III cad53642003-03-19 09:13:01 +0000245 bb_get_last_path_component(target.path);
Eric Andersen29edd002000-12-09 16:55:35 +0000246 if (fname_out==NULL || strlen(fname_out)<1) {
247 fname_out =
Eric Andersenbdfd0d72001-10-24 05:00:29 +0000248#ifdef CONFIG_FEATURE_WGET_STATUSBAR
Eric Andersen29edd002000-12-09 16:55:35 +0000249 curfile =
250#endif
251 "index.html";
252 }
Matt Kraai0382eb82001-07-19 19:13:55 +0000253 if (dir_prefix != NULL)
254 fname_out = concat_path_file(dir_prefix, fname_out);
Eric Andersenbdfd0d72001-10-24 05:00:29 +0000255#ifdef CONFIG_FEATURE_WGET_STATUSBAR
Eric Andersen29edd002000-12-09 16:55:35 +0000256 } else {
Manuel Novoa III cad53642003-03-19 09:13:01 +0000257 curfile = bb_get_last_path_component(fname_out);
Eric Andersen29edd002000-12-09 16:55:35 +0000258#endif
259 }
Eric Andersen7d697012001-01-24 20:28:35 +0000260 if (do_continue && !fname_out)
Manuel Novoa III cad53642003-03-19 09:13:01 +0000261 bb_error_msg_and_die("cannot specify continue (-c) without a filename (-O)");
Eric Andersen29edd002000-12-09 16:55:35 +0000262
Eric Andersen96700832000-09-04 15:15:55 +0000263
264 /*
Eric Andersen29edd002000-12-09 16:55:35 +0000265 * Open the output file stream.
Eric Andersen96700832000-09-04 15:15:55 +0000266 */
Matt Kraai65317ea2001-04-11 20:03:01 +0000267 if (strcmp(fname_out, "-") == 0) {
Randolph Chung02553a22000-12-07 03:53:47 +0000268 output = stdout;
Eric Andersen95a349f2001-05-13 00:55:54 +0000269 quiet_flag = TRUE;
Matt Kraai65317ea2001-04-11 20:03:01 +0000270 } else {
Manuel Novoa III cad53642003-03-19 09:13:01 +0000271 output = bb_xfopen(fname_out, (do_continue ? "a" : "w"));
Eric Andersen96700832000-09-04 15:15:55 +0000272 }
273
274 /*
275 * Determine where to start transfer.
276 */
277 if (do_continue) {
Eric Andersenb520e082000-10-03 00:21:45 +0000278 if (fstat(fileno(output), &sbuf) < 0)
Manuel Novoa III cad53642003-03-19 09:13:01 +0000279 bb_perror_msg_and_die("fstat()");
Eric Andersen96700832000-09-04 15:15:55 +0000280 if (sbuf.st_size > 0)
281 beg_range = sbuf.st_size;
282 else
283 do_continue = 0;
284 }
285
Eric Andersene6dc4392003-10-31 09:31:46 +0000286 /* We want to do exactly _one_ DNS lookup, since some
287 * sites (i.e. ftp.us.debian.org) use round-robin DNS
288 * and we want to connect to only one IP... */
Glenn L McGrathffccf6e2003-12-20 01:47:18 +0000289 bb_lookup_host(&s_in, server.host);
290 s_in.sin_port = server.port;
Eric Andersene6dc4392003-10-31 09:31:46 +0000291 if (quiet_flag==FALSE) {
292 fprintf(stdout, "Connecting to %s[%s]:%d\n",
Glenn L McGrathffccf6e2003-12-20 01:47:18 +0000293 server.host, inet_ntoa(s_in.sin_addr), ntohs(server.port));
Eric Andersene6dc4392003-10-31 09:31:46 +0000294 }
295
Eric Andersen79757c92001-04-05 21:45:54 +0000296 if (proxy || !target.is_ftp) {
297 /*
298 * HTTP session
299 */
300 do {
Glenn L McGrathe7bdfcc2003-08-28 22:03:19 +0000301 got_clen = chunked = 0;
302
Eric Andersen79757c92001-04-05 21:45:54 +0000303 if (! --try)
304 close_delete_and_die("too many redirections");
Eric Andersen7d697012001-01-24 20:28:35 +0000305
Eric Andersen79757c92001-04-05 21:45:54 +0000306 /*
307 * Open socket to http server
308 */
309 if (sfp) fclose(sfp);
Glenn L McGrathffccf6e2003-12-20 01:47:18 +0000310 sfp = open_socket(&s_in);
Eric Andersen79757c92001-04-05 21:45:54 +0000311
312 /*
313 * Send HTTP request.
314 */
315 if (proxy) {
Glenn L McGrathcc20ebc2003-09-10 23:52:15 +0000316 const char *format = "GET %stp://%s:%d/%s HTTP/1.1\r\n";
317#ifdef CONFIG_FEATURE_WGET_IP6_LITERAL
318 if (strchr (target.host, ':'))
319 format = "GET %stp://[%s]:%d/%s HTTP/1.1\r\n";
320#endif
321 fprintf(sfp, format,
Eric Andersen79757c92001-04-05 21:45:54 +0000322 target.is_ftp ? "f" : "ht", target.host,
323 target.port, target.path);
324 } else {
Eric Andersen6d7fa432001-04-10 18:17:05 +0000325 fprintf(sfp, "GET /%s HTTP/1.1\r\n", target.path);
Eric Andersen79757c92001-04-05 21:45:54 +0000326 }
Eric Andersen96700832000-09-04 15:15:55 +0000327
Eric Andersen79757c92001-04-05 21:45:54 +0000328 fprintf(sfp, "Host: %s\r\nUser-Agent: Wget\r\n", target.host);
329
Eric Andersenbdfd0d72001-10-24 05:00:29 +0000330#ifdef CONFIG_FEATURE_WGET_AUTHENTICATION
Eric Andersen79757c92001-04-05 21:45:54 +0000331 if (target.user) {
332 fprintf(sfp, "Authorization: Basic %s\r\n",
333 base64enc(target.user, buf, sizeof(buf)));
334 }
335 if (proxy && server.user) {
336 fprintf(sfp, "Proxy-Authorization: Basic %s\r\n",
337 base64enc(server.user, buf, sizeof(buf)));
338 }
339#endif
340
Eric Andersenb520e082000-10-03 00:21:45 +0000341 if (do_continue)
Eric Andersen79757c92001-04-05 21:45:54 +0000342 fprintf(sfp, "Range: bytes=%ld-\r\n", beg_range);
Eric Andersen50ae3102001-05-15 17:51:37 +0000343 if(extra_headers_left < sizeof(extra_headers))
Eric Andersen9abfe852001-05-15 20:11:49 +0000344 fputs(extra_headers,sfp);
Eric Andersen79757c92001-04-05 21:45:54 +0000345 fprintf(sfp,"Connection: close\r\n\r\n");
346
347 /*
348 * Retrieve HTTP response line and check for "200" status code.
349 */
Glenn L McGrathe7bdfcc2003-08-28 22:03:19 +0000350read_response:
351 if (fgets(buf, sizeof(buf), sfp) == NULL)
Eric Andersen79757c92001-04-05 21:45:54 +0000352 close_delete_and_die("no response from server");
353
354 for (s = buf ; *s != '\0' && !isspace(*s) ; ++s)
355 ;
356 for ( ; isspace(*s) ; ++s)
357 ;
358 switch (status = atoi(s)) {
359 case 0:
Eric Andersen6d7fa432001-04-10 18:17:05 +0000360 case 100:
361 while (gethdr(buf, sizeof(buf), sfp, &n) != NULL);
362 goto read_response;
Eric Andersen79757c92001-04-05 21:45:54 +0000363 case 200:
364 if (do_continue && output != stdout)
365 output = freopen(fname_out, "w", output);
366 do_continue = 0;
367 break;
368 case 300: /* redirection */
369 case 301:
370 case 302:
371 case 303:
372 break;
373 case 206:
374 if (do_continue)
375 break;
376 /*FALLTHRU*/
377 default:
378 chomp(buf);
379 close_delete_and_die("server returned error %d: %s", atoi(s), buf);
380 }
381
382 /*
383 * Retrieve HTTP headers.
384 */
385 while ((s = gethdr(buf, sizeof(buf), sfp, &n)) != NULL) {
386 if (strcasecmp(buf, "content-length") == 0) {
387 filesize = atol(s);
388 got_clen = 1;
389 continue;
390 }
Eric Andersen6d7fa432001-04-10 18:17:05 +0000391 if (strcasecmp(buf, "transfer-encoding") == 0) {
392 if (strcasecmp(s, "chunked") == 0) {
393 chunked = got_clen = 1;
394 } else {
Eric Andersen79757c92001-04-05 21:45:54 +0000395 close_delete_and_die("server wants to do %s transfer encoding", s);
Eric Andersen6d7fa432001-04-10 18:17:05 +0000396 }
397 }
Eric Andersen79757c92001-04-05 21:45:54 +0000398 if (strcasecmp(buf, "location") == 0) {
399 if (s[0] == '/')
Manuel Novoa III cad53642003-03-19 09:13:01 +0000400 target.path = bb_xstrdup(s+1);
Eric Andersen79757c92001-04-05 21:45:54 +0000401 else {
Manuel Novoa III cad53642003-03-19 09:13:01 +0000402 parse_url(bb_xstrdup(s), &target);
Eric Andersen79757c92001-04-05 21:45:54 +0000403 if (!proxy) {
404 server.host = target.host;
405 server.port = target.port;
406 }
Glenn L McGrath58a2e0e2004-01-17 23:07:14 +0000407 bb_lookup_host(&s_in, server.host);
408 s_in.sin_port = server.port;
409 break;
Eric Andersen79757c92001-04-05 21:45:54 +0000410 }
411 }
412 }
413 } while(status >= 300);
414
415 dfp = sfp;
416 }
417 else
418 {
419 /*
420 * FTP session
421 */
422 if (! target.user)
Manuel Novoa III cad53642003-03-19 09:13:01 +0000423 target.user = bb_xstrdup("anonymous:busybox@");
Eric Andersen79757c92001-04-05 21:45:54 +0000424
Glenn L McGrathffccf6e2003-12-20 01:47:18 +0000425 sfp = open_socket(&s_in);
Eric Andersen79757c92001-04-05 21:45:54 +0000426 if (ftpcmd(NULL, NULL, sfp, buf) != 220)
427 close_delete_and_die("%s", buf+4);
428
429 /*
430 * Splitting username:password pair,
431 * trying to log in
432 */
433 s = strchr(target.user, ':');
434 if (s)
435 *(s++) = '\0';
436 switch(ftpcmd("USER ", target.user, sfp, buf)) {
437 case 230:
Eric Andersenb520e082000-10-03 00:21:45 +0000438 break;
Eric Andersen79757c92001-04-05 21:45:54 +0000439 case 331:
440 if (ftpcmd("PASS ", s, sfp, buf) == 230)
441 break;
442 /* FALLTHRU (failed login) */
443 default:
444 close_delete_and_die("ftp login: %s", buf+4);
445 }
446
447 ftpcmd("CDUP", NULL, sfp, buf);
448 ftpcmd("TYPE I", NULL, sfp, buf);
449
450 /*
451 * Querying file size
452 */
453 if (ftpcmd("SIZE /", target.path, sfp, buf) == 213) {
454 filesize = atol(buf+4);
Eric Andersen96700832000-09-04 15:15:55 +0000455 got_clen = 1;
Eric Andersen96700832000-09-04 15:15:55 +0000456 }
Eric Andersen79757c92001-04-05 21:45:54 +0000457
458 /*
459 * Entering passive mode
460 */
461 if (ftpcmd("PASV", NULL, sfp, buf) != 227)
462 close_delete_and_die("PASV: %s", buf+4);
463 s = strrchr(buf, ',');
464 *s = 0;
465 port = atoi(s+1);
466 s = strrchr(buf, ',');
467 port += atoi(s+1) * 256;
Glenn L McGrathffccf6e2003-12-20 01:47:18 +0000468 s_in.sin_port = htons(port);
469 dfp = open_socket(&s_in);
Eric Andersen79757c92001-04-05 21:45:54 +0000470
471 if (do_continue) {
472 sprintf(buf, "REST %ld", beg_range);
473 if (ftpcmd(buf, NULL, sfp, buf) != 350) {
474 if (output != stdout)
475 output = freopen(fname_out, "w", output);
476 do_continue = 0;
477 } else
478 filesize -= beg_range;
Eric Andersen96700832000-09-04 15:15:55 +0000479 }
Eric Andersen79757c92001-04-05 21:45:54 +0000480
481 if (ftpcmd("RETR /", target.path, sfp, buf) > 150)
482 close_delete_and_die("RETR: %s", buf+4);
483
Eric Andersen96700832000-09-04 15:15:55 +0000484 }
485
Eric Andersen79757c92001-04-05 21:45:54 +0000486
Eric Andersen96700832000-09-04 15:15:55 +0000487 /*
Eric Andersen79757c92001-04-05 21:45:54 +0000488 * Retrieve file
Eric Andersen96700832000-09-04 15:15:55 +0000489 */
Eric Andersen6d7fa432001-04-10 18:17:05 +0000490 if (chunked) {
491 fgets(buf, sizeof(buf), dfp);
492 filesize = strtol(buf, (char **) NULL, 16);
493 }
Eric Andersenbdfd0d72001-10-24 05:00:29 +0000494#ifdef CONFIG_FEATURE_WGET_STATUSBAR
Glenn L McGrath1bca5ed2000-12-09 08:12:06 +0000495 if (quiet_flag==FALSE)
496 progressmeter(-1);
Eric Andersenb520e082000-10-03 00:21:45 +0000497#endif
Mark Whitley30ac01c2001-04-17 18:13:16 +0000498 do {
Glenn L McGrath23365972003-08-29 06:25:04 +0000499 while ((filesize > 0 || !got_clen) && (n = safe_fread(buf, 1, ((chunked || got_clen) && (filesize < sizeof(buf)) ? filesize : sizeof(buf)), dfp)) > 0) {
Glenn L McGrath83e4a5b2003-08-28 21:55:22 +0000500 if (safe_fwrite(buf, 1, n, output) != n) {
Manuel Novoa III cad53642003-03-19 09:13:01 +0000501 bb_perror_msg_and_die("write error");
Glenn L McGrath83e4a5b2003-08-28 21:55:22 +0000502 }
Eric Andersenbdfd0d72001-10-24 05:00:29 +0000503#ifdef CONFIG_FEATURE_WGET_STATUSBAR
Glenn L McGrath83e4a5b2003-08-28 21:55:22 +0000504 statbytes+=n;
Eric Andersenb520e082000-10-03 00:21:45 +0000505#endif
Glenn L McGrath83e4a5b2003-08-28 21:55:22 +0000506 if (got_clen) {
507 filesize -= n;
508 }
509 }
Eric Andersen79757c92001-04-05 21:45:54 +0000510
Eric Andersen6d7fa432001-04-10 18:17:05 +0000511 if (chunked) {
Matt Kraai854125f2001-05-09 19:15:46 +0000512 safe_fgets(buf, sizeof(buf), dfp); /* This is a newline */
513 safe_fgets(buf, sizeof(buf), dfp);
Eric Andersen6d7fa432001-04-10 18:17:05 +0000514 filesize = strtol(buf, (char **) NULL, 16);
Glenn L McGrath83e4a5b2003-08-28 21:55:22 +0000515 if (filesize==0) {
516 chunked = 0; /* all done! */
517 }
Eric Andersen6d7fa432001-04-10 18:17:05 +0000518 }
519
Glenn L McGrath83e4a5b2003-08-28 21:55:22 +0000520 if (n == 0 && ferror(dfp)) {
521 bb_perror_msg_and_die("network read error");
522 }
Eric Andersen6d7fa432001-04-10 18:17:05 +0000523 } while (chunked);
Eric Andersenbdfd0d72001-10-24 05:00:29 +0000524#ifdef CONFIG_FEATURE_WGET_STATUSBAR
Mark Whitley30ac01c2001-04-17 18:13:16 +0000525 if (quiet_flag==FALSE)
526 progressmeter(1);
527#endif
Eric Andersen79757c92001-04-05 21:45:54 +0000528 if (!proxy && target.is_ftp) {
529 fclose(dfp);
530 if (ftpcmd(NULL, NULL, sfp, buf) != 226)
Manuel Novoa III cad53642003-03-19 09:13:01 +0000531 bb_error_msg_and_die("ftp error: %s", buf+4);
Eric Andersen79757c92001-04-05 21:45:54 +0000532 ftpcmd("QUIT", NULL, sfp, buf);
533 }
Eric Andersen79757c92001-04-05 21:45:54 +0000534 exit(EXIT_SUCCESS);
Eric Andersen96700832000-09-04 15:15:55 +0000535}
536
537
Eric Andersen79757c92001-04-05 21:45:54 +0000538void parse_url(char *url, struct host_info *h)
Eric Andersen96700832000-09-04 15:15:55 +0000539{
Glenn L McGrathcc20ebc2003-09-10 23:52:15 +0000540 char *cp, *sp, *up, *pp;
Eric Andersen96700832000-09-04 15:15:55 +0000541
Eric Andersen79757c92001-04-05 21:45:54 +0000542 if (strncmp(url, "http://", 7) == 0) {
Glenn L McGrath036dbaa2004-01-17 05:03:31 +0000543 h->port = bb_lookup_port("http", "tcp", 80);
Eric Andersen79757c92001-04-05 21:45:54 +0000544 h->host = url + 7;
545 h->is_ftp = 0;
546 } else if (strncmp(url, "ftp://", 6) == 0) {
Glenn L McGrath036dbaa2004-01-17 05:03:31 +0000547 h->port = bb_lookup_port("ftp", "tfp", 21);
Eric Andersen79757c92001-04-05 21:45:54 +0000548 h->host = url + 6;
549 h->is_ftp = 1;
550 } else
Manuel Novoa III cad53642003-03-19 09:13:01 +0000551 bb_error_msg_and_die("not an http or ftp url: %s", url);
Eric Andersen96700832000-09-04 15:15:55 +0000552
Eric Andersen79757c92001-04-05 21:45:54 +0000553 sp = strchr(h->host, '/');
Matt Kraaia9711a52001-01-03 16:15:15 +0000554 if (sp != NULL) {
555 *sp++ = '\0';
Eric Andersen79757c92001-04-05 21:45:54 +0000556 h->path = sp;
Matt Kraaia9711a52001-01-03 16:15:15 +0000557 } else
Manuel Novoa III cad53642003-03-19 09:13:01 +0000558 h->path = bb_xstrdup("");
Eric Andersen79757c92001-04-05 21:45:54 +0000559
560 up = strrchr(h->host, '@');
561 if (up != NULL) {
562 h->user = h->host;
563 *up++ = '\0';
564 h->host = up;
565 } else
566 h->user = NULL;
567
Glenn L McGrathcc20ebc2003-09-10 23:52:15 +0000568 pp = h->host;
569
570#ifdef CONFIG_FEATURE_WGET_IP6_LITERAL
571 if (h->host[0] == '[') {
572 char *ep;
573
574 ep = h->host + 1;
Eric Andersen6231f092003-09-11 08:25:11 +0000575 while (*ep == ':' || isxdigit (*ep))
Glenn L McGrathcc20ebc2003-09-10 23:52:15 +0000576 ep++;
577 if (*ep == ']') {
578 h->host++;
579 *ep = '\0';
580 pp = ep + 1;
581 }
582 }
583#endif
584
585 cp = strchr(pp, ':');
Eric Andersen79757c92001-04-05 21:45:54 +0000586 if (cp != NULL) {
587 *cp++ = '\0';
Glenn L McGrathf980bd52003-12-27 00:21:47 +0000588 h->port = htons(atoi(cp));
Eric Andersen79757c92001-04-05 21:45:54 +0000589 }
Eric Andersen96700832000-09-04 15:15:55 +0000590}
591
592
Glenn L McGrathffccf6e2003-12-20 01:47:18 +0000593FILE *open_socket(struct sockaddr_in *s_in)
Eric Andersen96700832000-09-04 15:15:55 +0000594{
Eric Andersen96700832000-09-04 15:15:55 +0000595 FILE *fp;
596
Glenn L McGrathffccf6e2003-12-20 01:47:18 +0000597 fp = fdopen(xconnect(s_in), "r+");
598 if (fp == NULL)
Manuel Novoa III cad53642003-03-19 09:13:01 +0000599 bb_perror_msg_and_die("fdopen()");
Eric Andersen96700832000-09-04 15:15:55 +0000600
601 return fp;
602}
603
604
605char *gethdr(char *buf, size_t bufsiz, FILE *fp, int *istrunc)
606{
607 char *s, *hdrval;
608 int c;
609
610 *istrunc = 0;
611
612 /* retrieve header line */
613 if (fgets(buf, bufsiz, fp) == NULL)
614 return NULL;
615
616 /* see if we are at the end of the headers */
617 for (s = buf ; *s == '\r' ; ++s)
618 ;
619 if (s[0] == '\n')
620 return NULL;
621
622 /* convert the header name to lower case */
623 for (s = buf ; isalnum(*s) || *s == '-' ; ++s)
624 *s = tolower(*s);
625
626 /* verify we are at the end of the header name */
627 if (*s != ':')
Manuel Novoa III cad53642003-03-19 09:13:01 +0000628 bb_error_msg_and_die("bad header line: %s", buf);
Eric Andersen96700832000-09-04 15:15:55 +0000629
630 /* locate the start of the header value */
631 for (*s++ = '\0' ; *s == ' ' || *s == '\t' ; ++s)
632 ;
633 hdrval = s;
634
635 /* locate the end of header */
636 while (*s != '\0' && *s != '\r' && *s != '\n')
637 ++s;
638
639 /* end of header found */
640 if (*s != '\0') {
641 *s = '\0';
642 return hdrval;
643 }
644
Eric Andersen5d638842000-09-14 21:46:30 +0000645 /* Rats! The buffer isn't big enough to hold the entire header value. */
Eric Andersen96700832000-09-04 15:15:55 +0000646 while (c = getc(fp), c != EOF && c != '\n')
647 ;
648 *istrunc = 1;
649 return hdrval;
650}
651
Eric Andersen79757c92001-04-05 21:45:54 +0000652static int ftpcmd(char *s1, char *s2, FILE *fp, char *buf)
653{
654 char *p;
655
656 if (s1) {
657 if (!s2) s2="";
Eric Andersen3f1cf452003-03-11 18:03:39 +0000658 fprintf(fp, "%s%s\r\n", s1, s2);
Eric Andersen79757c92001-04-05 21:45:54 +0000659 fflush(fp);
660 }
661
662 do {
663 p = fgets(buf, 510, fp);
664 if (!p)
Manuel Novoa III cad53642003-03-19 09:13:01 +0000665 bb_perror_msg_and_die("fgets()");
Eric Andersen79757c92001-04-05 21:45:54 +0000666 } while (! isdigit(buf[0]) || buf[3] != ' ');
667
668 return atoi(buf);
669}
670
Eric Andersenbdfd0d72001-10-24 05:00:29 +0000671#ifdef CONFIG_FEATURE_WGET_STATUSBAR
Eric Andersen4e573f42000-11-14 23:29:24 +0000672/* Stuff below is from BSD rcp util.c, as added to openshh.
673 * Original copyright notice is retained at the end of this file.
674 *
675 */
Eric Andersenb520e082000-10-03 00:21:45 +0000676
677
Eric Andersen3e6ff902001-03-09 21:24:12 +0000678static int
Eric Andersenb520e082000-10-03 00:21:45 +0000679getttywidth(void)
680{
Eric Andersen8efe9672003-09-15 08:33:45 +0000681 int width=0;
682 get_terminal_width_height(0, &width, NULL);
683 return (width);
Eric Andersenb520e082000-10-03 00:21:45 +0000684}
685
Eric Andersen3e6ff902001-03-09 21:24:12 +0000686static void
Eric Andersenb520e082000-10-03 00:21:45 +0000687updateprogressmeter(int ignore)
688{
689 int save_errno = errno;
690
691 progressmeter(0);
692 errno = save_errno;
693}
694
Eric Andersen3e6ff902001-03-09 21:24:12 +0000695static void
Eric Andersenb520e082000-10-03 00:21:45 +0000696alarmtimer(int wait)
697{
698 struct itimerval itv;
699
700 itv.it_value.tv_sec = wait;
701 itv.it_value.tv_usec = 0;
702 itv.it_interval = itv.it_value;
703 setitimer(ITIMER_REAL, &itv, NULL);
704}
705
706
Eric Andersen3e6ff902001-03-09 21:24:12 +0000707static void
Eric Andersenb520e082000-10-03 00:21:45 +0000708progressmeter(int flag)
709{
710 static const char prefixes[] = " KMGTP";
711 static struct timeval lastupdate;
Mark Whitley30ac01c2001-04-17 18:13:16 +0000712 static off_t lastsize, totalsize;
Eric Andersenb520e082000-10-03 00:21:45 +0000713 struct timeval now, td, wait;
714 off_t cursize, abbrevsize;
715 double elapsed;
716 int ratio, barlength, i, remaining;
717 char buf[256];
718
719 if (flag == -1) {
720 (void) gettimeofday(&start, (struct timezone *) 0);
721 lastupdate = start;
722 lastsize = 0;
Mark Whitley30ac01c2001-04-17 18:13:16 +0000723 totalsize = filesize; /* as filesize changes.. */
Eric Andersenb520e082000-10-03 00:21:45 +0000724 }
725
726 (void) gettimeofday(&now, (struct timezone *) 0);
727 cursize = statbytes;
Mark Whitley30ac01c2001-04-17 18:13:16 +0000728 if (totalsize != 0 && !chunked) {
729 ratio = 100.0 * cursize / totalsize;
Eric Andersenb520e082000-10-03 00:21:45 +0000730 ratio = MAX(ratio, 0);
731 ratio = MIN(ratio, 100);
732 } else
733 ratio = 100;
734
735 snprintf(buf, sizeof(buf), "\r%-20.20s %3d%% ", curfile, ratio);
736
737 barlength = getttywidth() - 51;
738 if (barlength > 0) {
739 i = barlength * ratio / 100;
740 snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf),
741 "|%.*s%*s|", i,
742 "*****************************************************************************"
743 "*****************************************************************************",
744 barlength - i, "");
745 }
746 i = 0;
747 abbrevsize = cursize;
748 while (abbrevsize >= 100000 && i < sizeof(prefixes)) {
749 i++;
750 abbrevsize >>= 10;
751 }
752 snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf), " %5d %c%c ",
753 (int) abbrevsize, prefixes[i], prefixes[i] == ' ' ? ' ' :
754 'B');
755
756 timersub(&now, &lastupdate, &wait);
757 if (cursize > lastsize) {
758 lastupdate = now;
759 lastsize = cursize;
760 if (wait.tv_sec >= STALLTIME) {
761 start.tv_sec += wait.tv_sec;
762 start.tv_usec += wait.tv_usec;
763 }
764 wait.tv_sec = 0;
765 }
766 timersub(&now, &start, &td);
767 elapsed = td.tv_sec + (td.tv_usec / 1000000.0);
768
Mark Whitley30ac01c2001-04-17 18:13:16 +0000769 if (wait.tv_sec >= STALLTIME) {
Eric Andersenb520e082000-10-03 00:21:45 +0000770 snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf),
771 " - stalled -");
Mark Whitley30ac01c2001-04-17 18:13:16 +0000772 } else if (statbytes <= 0 || elapsed <= 0.0 || cursize > totalsize || chunked) {
773 snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf),
774 " --:-- ETA");
Eric Andersenb520e082000-10-03 00:21:45 +0000775 } else {
Mark Whitley30ac01c2001-04-17 18:13:16 +0000776 remaining = (int) (totalsize / (statbytes / elapsed) - elapsed);
Eric Andersenb520e082000-10-03 00:21:45 +0000777 i = remaining / 3600;
778 if (i)
779 snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf),
780 "%2d:", i);
781 else
782 snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf),
783 " ");
784 i = remaining % 3600;
785 snprintf(buf + strlen(buf), sizeof(buf) - strlen(buf),
786 "%02d:%02d ETA", i / 60, i % 60);
787 }
Randolph Chungda7b8292000-12-07 03:55:35 +0000788 write(fileno(stderr), buf, strlen(buf));
Eric Andersenb520e082000-10-03 00:21:45 +0000789
790 if (flag == -1) {
791 struct sigaction sa;
792 sa.sa_handler = updateprogressmeter;
793 sigemptyset(&sa.sa_mask);
794 sa.sa_flags = SA_RESTART;
795 sigaction(SIGALRM, &sa, NULL);
796 alarmtimer(1);
797 } else if (flag == 1) {
798 alarmtimer(0);
799 statbytes = 0;
Mark Whitley30ac01c2001-04-17 18:13:16 +0000800 putc('\n', stderr);
Eric Andersenb520e082000-10-03 00:21:45 +0000801 }
802}
803#endif
Eric Andersen4e573f42000-11-14 23:29:24 +0000804
Eric Andersenbdfd0d72001-10-24 05:00:29 +0000805/* Original copyright notice which applies to the CONFIG_FEATURE_WGET_STATUSBAR stuff,
Eric Andersen4e573f42000-11-14 23:29:24 +0000806 * much of which was blatently stolen from openssh. */
807
808/*-
809 * Copyright (c) 1992, 1993
810 * The Regents of the University of California. All rights reserved.
811 *
812 * Redistribution and use in source and binary forms, with or without
813 * modification, are permitted provided that the following conditions
814 * are met:
815 * 1. Redistributions of source code must retain the above copyright
816 * notice, this list of conditions and the following disclaimer.
817 * 2. Redistributions in binary form must reproduce the above copyright
818 * notice, this list of conditions and the following disclaimer in the
819 * documentation and/or other materials provided with the distribution.
820 *
821 * 3. <BSD Advertising Clause omitted per the July 22, 1999 licensing change
822 * ftp://ftp.cs.berkeley.edu/pub/4bsd/README.Impt.License.Change>
823 *
824 * 4. Neither the name of the University nor the names of its contributors
825 * may be used to endorse or promote products derived from this software
826 * without specific prior written permission.
827 *
828 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
829 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
830 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
831 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
832 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
833 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
834 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
835 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
836 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
837 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
838 * SUCH DAMAGE.
839 *
Glenn L McGrath58a2e0e2004-01-17 23:07:14 +0000840 * $Id: wget.c,v 1.66 2004/01/17 23:07:14 bug1 Exp $
Eric Andersen4e573f42000-11-14 23:29:24 +0000841 */
842
843
844
Eric Andersen96700832000-09-04 15:15:55 +0000845/*
846Local Variables:
847c-file-style: "linux"
848c-basic-offset: 4
849tab-width: 4
850End:
851*/