2 Copyright (C) 1995, 1996, 1997, 1998, 2000, 2001 Free Software Foundation, Inc.
4 This file is part of GNU Wget.
6 GNU Wget is free software; you can redistribute it and/or modify
7 it under the terms of the GNU General Public License as published by
8 the Free Software Foundation; either version 2 of the License, or
9 (at your option) any later version.
11 GNU Wget is distributed in the hope that it will be useful,
12 but WITHOUT ANY WARRANTY; without even the implied warranty of
13 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
14 GNU General Public License for more details.
16 You should have received a copy of the GNU General Public License
17 along with Wget; if not, write to the Free Software
18 Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
20 In addition, as a special exception, the Free Software Foundation
21 gives permission to link the code of its release of Wget with the
22 OpenSSL project's "OpenSSL" library (or with modified versions of it
23 that use the same license as the "OpenSSL" library), and distribute
24 the linked executables. You must obey the GNU General Public License
25 in all respects for all of the code used other than "OpenSSL". If you
26 modify this file, you may extend this exception to your version of the
27 file, but you are not obligated to do so. If you do not wish to do
28 so, delete this exception statement from your version. */
34 #include <sys/types.h>
37 #endif /* HAVE_UNISTD_H */
43 #endif /* HAVE_STRING_H */
58 # include "gen_sslfunc.h" /* for ssl_iread */
65 /* See the comment in gethttp() why this is needed. */
66 int global_download_count;
75 limit_bandwidth_reset (void)
78 limit_data.dltime = 0;
81 /* Limit the bandwidth by pausing the download for an amount of time.
82 BYTES is the number of bytes received from the network, DELTA is
83 how long it took to receive them, DLTIME the current download time,
84 TIMER the timer, and ADJUSTMENT the previous. */
87 limit_bandwidth (long bytes, long delta)
91 limit_data.bytes += bytes;
92 limit_data.dltime += delta;
94 expected = (long)(1000.0 * limit_data.bytes / opt.limit_rate);
96 if (expected > limit_data.dltime)
98 long slp = expected - limit_data.dltime;
101 DEBUGP (("deferring a %ld ms sleep (%ld/%ld) until later.\n",
102 slp, limit_data.bytes, limit_data.dltime));
105 DEBUGP (("sleeping %ld ms\n", slp));
109 limit_data.bytes = 0;
110 limit_data.dltime = 0;
113 #define MIN(i, j) ((i) <= (j) ? (i) : (j))
115 /* Reads the contents of file descriptor FD, until it is closed, or a
116 read error occurs. The data is read in 8K chunks, and stored to
117 stream fp, which should have been open for writing. If BUF is
118 non-NULL and its file descriptor is equal to FD, flush RBUF first.
119 This function will *not* use the rbuf_* functions!
121 The EXPECTED argument is passed to show_progress() unchanged, but
124 If opt.verbose is set, the progress is also shown. RESTVAL
125 represents a value from which to start downloading (which will be
126 shown accordingly). If RESTVAL is non-zero, the stream should have
127 been open for appending.
129 The function exits and returns codes of 0, -1 and -2 if the
130 connection was closed, there was a read error, or if it could not
131 write to the output stream, respectively.
133 IMPORTANT: The function flushes the contents of the buffer in
134 rbuf_flush() before actually reading from fd. If you wish to read
135 from fd immediately, flush or discard the buffer. */
137 get_contents (int fd, FILE *fp, long *len, long restval, long expected,
138 struct rbuf *rbuf, int use_expected, long *elapsed)
142 void *progress = NULL;
143 struct wget_timer *timer = wtimer_allocate ();
144 long dltime = 0, last_dltime = 0;
149 progress = progress_create (restval, expected);
151 if (rbuf && RBUF_FD (rbuf) == fd)
154 while ((res = rbuf_flush (rbuf, c, sizeof (c))) != 0)
156 fwrite (c, sizeof (char), res, fp);
168 progress_update (progress, sz, 0);
172 limit_bandwidth_reset ();
173 wtimer_reset (timer);
175 /* Read from fd while there is available data.
177 Normally, if expected is 0, it means that it is not known how
178 much data is expected. However, if use_expected is specified,
179 then expected being zero means exactly that. */
180 while (!use_expected || (*len < expected))
182 int amount_to_read = (use_expected
183 ? MIN (expected - *len, sizeof (c))
187 res = ssl_iread (rbuf->ssl, c, amount_to_read);
189 #endif /* HAVE_SSL */
190 res = iread (fd, c, amount_to_read);
194 fwrite (c, sizeof (char), res, fp);
195 /* Always flush the contents of the network packet. This
196 should not be adverse to performance, as the network
197 packets typically won't be too tiny anyway. */
205 /* If bandwidth is not limited, one call to wtimer_elapsed
207 dltime = wtimer_elapsed (timer);
210 limit_bandwidth (res, dltime - last_dltime);
211 dltime = wtimer_elapsed (timer);
212 last_dltime = dltime;
216 progress_update (progress, res, dltime);
227 progress_finish (progress, dltime);
230 wtimer_delete (timer);
235 /* Return a printed representation of the download rate, as
236 appropriate for the speed. If PAD is non-zero, strings will be
237 padded to the width of 7 characters (xxxx.xx). */
239 retr_rate (long bytes, long msecs, int pad)
242 static char *rate_names[] = {"B/s", "KB/s", "MB/s", "GB/s" };
245 double dlrate = calc_rate (bytes, msecs, &units);
246 sprintf (res, pad ? "%7.2f %s" : "%.2f %s", dlrate, rate_names[units]);
251 /* Calculate the download rate and trim it as appropriate for the
252 speed. Appropriate means that if rate is greater than 1K/s,
253 kilobytes are used, and if rate is greater than 1MB/s, megabytes
256 UNITS is zero for B/s, one for KB/s, two for MB/s, and three for
259 calc_rate (long bytes, long msecs, int *units)
267 /* If elapsed time is 0, it means we're under the granularity of
268 the timer. This often happens on systems that use time() for
270 msecs = wtimer_granularity ();
272 dlrate = (double)1000 * bytes / msecs;
275 else if (dlrate < 1024.0 * 1024.0)
276 *units = 1, dlrate /= 1024.0;
277 else if (dlrate < 1024.0 * 1024.0 * 1024.0)
278 *units = 2, dlrate /= (1024.0 * 1024.0);
280 /* Maybe someone will need this one day. More realistically, it
281 will get tickled by buggy timers. */
282 *units = 3, dlrate /= (1024.0 * 1024.0 * 1024.0);
287 /* Maximum number of allowed redirections. 20 was chosen as a
288 "reasonable" value, which is low enough to not cause havoc, yet
289 high enough to guarantee that normal retrievals will not be hurt by
292 #define MAX_REDIRECTIONS 20
294 #define SUSPEND_POST_DATA do { \
295 post_data_suspended = 1; \
296 saved_post_data = opt.post_data; \
297 saved_post_file_name = opt.post_file_name; \
298 opt.post_data = NULL; \
299 opt.post_file_name = NULL; \
302 #define RESTORE_POST_DATA do { \
303 if (post_data_suspended) \
305 opt.post_data = saved_post_data; \
306 opt.post_file_name = saved_post_file_name; \
307 post_data_suspended = 0; \
311 /* Retrieve the given URL. Decides which loop to call -- HTTP, FTP,
314 /* #### This function should be rewritten so it doesn't return from
318 retrieve_url (const char *origurl, char **file, char **newloc,
319 const char *refurl, int *dt)
323 int location_changed, dummy;
324 char *mynewloc, *proxy;
325 struct url *u, *proxy_url;
326 int up_error_code; /* url parse error code */
328 int redirection_count = 0;
330 int post_data_suspended = 0;
331 char *saved_post_data = NULL;
332 char *saved_post_file_name = NULL;
334 /* If dt is NULL, just ignore it. */
337 url = xstrdup (origurl);
343 u = url_parse (url, &up_error_code);
346 logprintf (LOG_NOTQUIET, "%s: %s.\n", url, url_error (up_error_code));
352 refurl = opt.referer;
361 proxy = getproxy (u);
364 /* Parse the proxy URL. */
365 proxy_url = url_parse (proxy, &up_error_code);
368 logprintf (LOG_NOTQUIET, _("Error parsing proxy URL %s: %s.\n"),
369 proxy, url_error (up_error_code));
374 if (proxy_url->scheme != SCHEME_HTTP && proxy_url->scheme != u->scheme)
376 logprintf (LOG_NOTQUIET, _("Error in proxy URL %s: Must be HTTP.\n"), proxy);
377 url_free (proxy_url);
384 if (u->scheme == SCHEME_HTTP
386 || u->scheme == SCHEME_HTTPS
388 || (proxy_url && proxy_url->scheme == SCHEME_HTTP))
390 result = http_loop (u, &mynewloc, &local_file, refurl, dt, proxy_url);
392 else if (u->scheme == SCHEME_FTP)
394 /* If this is a redirection, we must not allow recursive FTP
395 retrieval, so we save recursion to oldrec, and restore it
397 int oldrec = opt.recursive;
398 if (redirection_count)
400 result = ftp_loop (u, dt, proxy_url);
401 opt.recursive = oldrec;
403 /* There is a possibility of having HTTP being redirected to
404 FTP. In these cases we must decide whether the text is HTML
405 according to the suffix. The HTML suffixes are `.html',
406 `.htm' and a few others, case-insensitive. */
407 if (redirection_count && local_file && u->scheme == SCHEME_FTP)
409 if (has_html_suffix_p (local_file))
416 url_free (proxy_url);
420 location_changed = (result == NEWLOCATION);
421 if (location_changed)
423 char *construced_newloc;
424 struct url *newloc_parsed;
426 assert (mynewloc != NULL);
431 /* The HTTP specs only allow absolute URLs to appear in
432 redirects, but a ton of boneheaded webservers and CGIs out
433 there break the rules and use relative URLs, and popular
434 browsers are lenient about this, so wget should be too. */
435 construced_newloc = uri_merge (url, mynewloc);
437 mynewloc = construced_newloc;
439 /* Now, see if this new location makes sense. */
440 newloc_parsed = url_parse (mynewloc, &up_error_code);
443 logprintf (LOG_NOTQUIET, "%s: %s.\n", mynewloc,
444 url_error (up_error_code));
452 /* Now mynewloc will become newloc_parsed->url, because if the
453 Location contained relative paths like .././something, we
454 don't want that propagating as url. */
456 mynewloc = xstrdup (newloc_parsed->url);
458 /* Check for max. number of redirections. */
459 if (++redirection_count > MAX_REDIRECTIONS)
461 logprintf (LOG_NOTQUIET, _("%d redirections exceeded.\n"),
463 url_free (newloc_parsed);
476 /* If we're being redirected from POST, we don't want to POST
477 again. Many requests answer POST with a redirection to an
478 index page; that redirection is clearly a GET. We "suspend"
479 POST data for the duration of the redirections, and restore
480 it when we're done. */
481 if (!post_data_suspended)
491 register_download (u->url, local_file);
492 if (redirection_count && 0 != strcmp (origurl, u->url))
493 register_redirection (origurl, u->url);
495 register_html (u->url, local_file);
500 *file = local_file ? local_file : NULL;
502 FREE_MAYBE (local_file);
506 if (redirection_count)
520 ++global_download_count;
526 /* Find the URLs in the file and call retrieve_url() for each of
527 them. If HTML is non-zero, treat the file as HTML, and construct
528 the URLs accordingly.
530 If opt.recursive is set, call recursive_retrieve() for each file. */
532 retrieve_from_file (const char *file, int html, int *count)
535 struct urlpos *url_list, *cur_url;
537 url_list = (html ? get_urls_html (file, NULL, NULL)
538 : get_urls_file (file));
539 status = RETROK; /* Suppose everything is OK. */
540 *count = 0; /* Reset the URL count. */
542 for (cur_url = url_list; cur_url; cur_url = cur_url->next, ++*count)
544 char *filename = NULL, *new_file = NULL;
547 if (cur_url->ignore_when_downloading)
550 if (downloaded_exceeds_quota ())
555 if (opt.recursive && cur_url->url->scheme != SCHEME_FTP)
556 status = retrieve_tree (cur_url->url->url);
558 status = retrieve_url (cur_url->url->url, &filename, &new_file, NULL, &dt);
560 if (filename && opt.delete_after && file_exists_p (filename))
562 DEBUGP (("Removing file due to --delete-after in"
563 " retrieve_from_file():\n"));
564 logprintf (LOG_VERBOSE, _("Removing %s.\n"), filename);
565 if (unlink (filename))
566 logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
570 FREE_MAYBE (new_file);
571 FREE_MAYBE (filename);
574 /* Free the linked list of URL-s. */
575 free_urlpos (url_list);
580 /* Print `giving up', or `retrying', depending on the impending
581 action. N1 and N2 are the attempt number and the attempt limit. */
583 printwhat (int n1, int n2)
585 logputs (LOG_VERBOSE, (n1 == n2) ? _("Giving up.\n\n") : _("Retrying.\n\n"));
588 /* Increment opt.downloaded by BY_HOW_MUCH. If an overflow occurs,
589 set opt.downloaded_overflow to 1. */
591 downloaded_increase (unsigned long by_how_much)
594 if (opt.downloaded_overflow)
596 old = opt.downloaded;
597 opt.downloaded += by_how_much;
598 if (opt.downloaded < old) /* carry flag, where are you when I
602 opt.downloaded_overflow = 1;
603 opt.downloaded = ~((VERY_LONG_TYPE)0);
607 /* Return non-zero if the downloaded amount of bytes exceeds the
608 desired quota. If quota is not set or if the amount overflowed, 0
611 downloaded_exceeds_quota (void)
615 if (opt.downloaded_overflow)
616 /* We don't really know. (Wildly) assume not. */
619 return opt.downloaded > opt.quota;
622 /* If opt.wait or opt.waitretry are specified, and if certain
623 conditions are met, sleep the appropriate number of seconds. See
624 the documentation of --wait and --waitretry for more information.
626 COUNT is the count of current retrieval, beginning with 1. */
629 sleep_between_retrievals (int count)
631 static int first_retrieval = 1;
635 /* Don't sleep before the very first retrieval. */
640 if (opt.waitretry && count > 1)
642 /* If opt.waitretry is specified and this is a retry, wait for
643 COUNT-1 number of seconds, or for opt.waitretry seconds. */
644 if (count <= opt.waitretry)
647 sleep (opt.waitretry);
651 if (!opt.random_wait || count > 1)
652 /* If random-wait is not specified, or if we are sleeping
653 between retries of the same download, sleep the fixed
658 /* Sleep a random amount of time averaging in opt.wait
659 seconds. The sleeping amount ranges from 0 to
660 opt.wait*2, inclusive. */
661 int waitsecs = random_number (opt.wait * 2 + 1);
663 DEBUGP (("sleep_between_retrievals: norm=%ld,fuzz=%ld,sleep=%d\n",
664 opt.wait, waitsecs - opt.wait, waitsecs));