2 Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
3 2004, 2005, 2006, 2007, 2008 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 3 of the License, or (at
10 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, see <http://www.gnu.org/licenses/>.
20 Additional permission under GNU GPL version 3 section 7
22 If you modify this program, or any covered work, by linking or
23 combining it with the OpenSSL project's OpenSSL library (or a
24 modified version of that library), containing parts covered by the
25 terms of the OpenSSL or SSLeay licenses, the Free Software Foundation
26 grants you additional permission to convey the resulting work.
27 Corresponding Source for a non-source form of such a combination
28 shall include the source code for the parts of OpenSSL used as well
29 as that of the covered work. */
37 #endif /* HAVE_UNISTD_H */
57 /* Total size of downloaded files. Used to enforce quota. */
58 SUM_SIZE_INT total_downloaded_bytes;
60 /* Total download time in seconds. */
61 double total_download_time;
63 /* If non-NULL, the stream to which output should be written. This
64 stream is initialized when `-O' is used. */
67 /* Whether output_document is a regular file we can manipulate,
68 i.e. not `-' or a device file. */
69 bool output_stream_regular;
78 limit_bandwidth_reset (void)
83 /* Limit the bandwidth by pausing the download for an amount of time.
84 BYTES is the number of bytes received from the network, and TIMER
85 is the timer that started at the beginning of download. */
88 limit_bandwidth (wgint bytes, struct ptimer *timer)
90 double delta_t = ptimer_read (timer) - limit_data.chunk_start;
93 limit_data.chunk_bytes += bytes;
95 /* Calculate the amount of time we expect downloading the chunk
96 should take. If in reality it took less time, sleep to
97 compensate for the difference. */
98 expected = (double) limit_data.chunk_bytes / opt.limit_rate;
100 if (expected > delta_t)
102 double slp = expected - delta_t + limit_data.sleep_adjust;
106 DEBUGP (("deferring a %.2f ms sleep (%s/%.2f).\n",
107 slp * 1000, number_to_static_string (limit_data.chunk_bytes),
111 DEBUGP (("\nsleeping %.2f ms for %s bytes, adjust %.2f ms\n",
112 slp * 1000, number_to_static_string (limit_data.chunk_bytes),
113 limit_data.sleep_adjust));
115 t0 = ptimer_read (timer);
117 t1 = ptimer_measure (timer);
119 /* Due to scheduling, we probably slept slightly longer (or
120 shorter) than desired. Calculate the difference between the
121 desired and the actual sleep, and adjust the next sleep by
123 limit_data.sleep_adjust = slp - (t1 - t0);
124 /* If sleep_adjust is very large, it's likely due to suspension
125 and not clock inaccuracy. Don't enforce those. */
126 if (limit_data.sleep_adjust > 0.5)
127 limit_data.sleep_adjust = 0.5;
128 else if (limit_data.sleep_adjust < -0.5)
129 limit_data.sleep_adjust = -0.5;
132 limit_data.chunk_bytes = 0;
133 limit_data.chunk_start = ptimer_read (timer);
137 # define MIN(i, j) ((i) <= (j) ? (i) : (j))
140 /* Write data in BUF to OUT. However, if *SKIP is non-zero, skip that
141 amount of data and decrease SKIP. Increment *TOTAL by the amount
145 write_data (FILE *out, const char *buf, int bufsize, wgint *skip,
164 fwrite (buf, 1, bufsize, out);
167 /* Immediately flush the downloaded data. This should not hinder
168 performance: fast downloads will arrive in large 16K chunks
169 (which stdio would write out immediately anyway), and slow
170 downloads wouldn't be limited by disk speed. */
173 Perhaps it shouldn't hinder performance, but it sure does, at least
174 on VMS (more than 2X). Rather than speculate on what it should or
175 shouldn't do, it might make more sense to test it. Even better, it
176 might be nice to explain what possible benefit it could offer, as
177 it appears to be a clear invitation to poor performance with no
178 actual justification. (Also, why 16K? Anyone test other values?)
182 #endif /* ndef __VMS */
183 return !ferror (out);
186 /* Read the contents of file descriptor FD until it the connection
187 terminates or a read error occurs. The data is read in portions of
188 up to 16K and written to OUT as it arrives. If opt.verbose is set,
189 the progress is shown.
191 TOREAD is the amount of data expected to arrive, normally only used
192 by the progress gauge.
194 STARTPOS is the position from which the download starts, used by
195 the progress gauge. If QTYREAD is non-NULL, the value it points to
196 is incremented by the amount of data read from the network. If
197 QTYWRITTEN is non-NULL, the value it points to is incremented by
198 the amount of data written to disk. The time it took to download
199 the data is stored to ELAPSED.
201 The function exits and returns the amount of data read. In case of
202 error while reading data, -1 is returned. In case of error while
203 writing data, -2 is returned. */
206 fd_read_body (int fd, FILE *out, wgint toread, wgint startpos,
207 wgint *qtyread, wgint *qtywritten, double *elapsed, int flags)
211 static char dlbuf[16384];
212 int dlbufsize = sizeof (dlbuf);
214 struct ptimer *timer = NULL;
215 double last_successful_read_tm = 0;
217 /* The progress gauge, set according to the user preferences. */
218 void *progress = NULL;
220 /* Non-zero if the progress gauge is interactive, i.e. if it can
221 continually update the display. When true, smaller timeout
222 values are used so that the gauge can update the display when
223 data arrives slowly. */
224 bool progress_interactive = false;
226 bool exact = !!(flags & rb_read_exactly);
229 /* How much data we've read/written. */
231 wgint sum_written = 0;
233 if (flags & rb_skip_startpos)
238 /* If we're skipping STARTPOS bytes, pass 0 as the INITIAL
239 argument to progress_create because the indicator doesn't
240 (yet) know about "skipping" data. */
241 wgint start = skip ? 0 : startpos;
242 progress = progress_create (start, start + toread);
243 progress_interactive = progress_interactive_p (progress);
247 limit_bandwidth_reset ();
249 /* A timer is needed for tracking progress, for throttling, and for
250 tracking elapsed time. If either of these are requested, start
252 if (progress || opt.limit_rate || elapsed)
254 timer = ptimer_new ();
255 last_successful_read_tm = 0;
258 /* Use a smaller buffer for low requested bandwidths. For example,
259 with --limit-rate=2k, it doesn't make sense to slurp in 16K of
260 data and then sleep for 8s. With buffer size equal to the limit,
261 we never have to sleep for more than one second. */
262 if (opt.limit_rate && opt.limit_rate < dlbufsize)
263 dlbufsize = opt.limit_rate;
265 /* Read from FD while there is data to read. Normally toread==0
266 means that it is unknown how much data is to arrive. However, if
267 EXACT is set, then toread==0 means what it says: that no data
269 while (!exact || (sum_read < toread))
271 int rdsize = exact ? MIN (toread - sum_read, dlbufsize) : dlbufsize;
272 double tmout = opt.read_timeout;
273 if (progress_interactive)
275 /* For interactive progress gauges, always specify a ~1s
276 timeout, so that the gauge can be updated regularly even
277 when the data arrives very slowly or stalls. */
279 if (opt.read_timeout)
282 waittm = ptimer_read (timer) - last_successful_read_tm;
283 if (waittm + tmout > opt.read_timeout)
285 /* Don't let total idle time exceed read timeout. */
286 tmout = opt.read_timeout - waittm;
289 /* We've already exceeded the timeout. */
290 ret = -1, errno = ETIMEDOUT;
296 ret = fd_read (fd, dlbuf, rdsize, tmout);
298 if (progress_interactive && ret < 0 && errno == ETIMEDOUT)
299 ret = 0; /* interactive timeout, handled above */
301 break; /* EOF or read error */
303 if (progress || opt.limit_rate)
305 ptimer_measure (timer);
307 last_successful_read_tm = ptimer_read (timer);
313 if (!write_data (out, dlbuf, ret, &skip, &sum_written))
321 limit_bandwidth (ret, timer);
324 progress_update (progress, ret, ptimer_read (timer));
326 if (toread > 0 && !opt.quiet)
327 ws_percenttitle (100.0 *
328 (startpos + sum_read) / (startpos + toread));
336 progress_finish (progress, ptimer_read (timer));
339 *elapsed = ptimer_read (timer);
341 ptimer_destroy (timer);
344 *qtyread += sum_read;
346 *qtywritten += sum_written;
351 /* Read a hunk of data from FD, up until a terminator. The hunk is
352 limited by whatever the TERMINATOR callback chooses as its
353 terminator. For example, if terminator stops at newline, the hunk
354 will consist of a line of data; if terminator stops at two
355 newlines, it can be used to read the head of an HTTP response.
356 Upon determining the boundary, the function returns the data (up to
357 the terminator) in malloc-allocated storage.
359 In case of read error, NULL is returned. In case of EOF and no
360 data read, NULL is returned and errno set to 0. In case of having
361 read some data, but encountering EOF before seeing the terminator,
362 the data that has been read is returned, but it will (obviously)
363 not contain the terminator.
365 The TERMINATOR function is called with three arguments: the
366 beginning of the data read so far, the beginning of the current
367 block of peeked-at data, and the length of the current block.
368 Depending on its needs, the function is free to choose whether to
369 analyze all data or just the newly arrived data. If TERMINATOR
370 returns NULL, it means that the terminator has not been seen.
371 Otherwise it should return a pointer to the charactre immediately
372 following the terminator.
374 The idea is to be able to read a line of input, or otherwise a hunk
375 of text, such as the head of an HTTP request, without crossing the
376 boundary, so that the next call to fd_read etc. reads the data
377 after the hunk. To achieve that, this function does the following:
379 1. Peek at incoming data.
381 2. Determine whether the peeked data, along with the previously
382 read data, includes the terminator.
384 2a. If yes, read the data until the end of the terminator, and
387 2b. If no, read the peeked data and goto 1.
389 The function is careful to assume as little as possible about the
390 implementation of peeking. For example, every peek is followed by
391 a read. If the read returns a different amount of data, the
392 process is retried until all data arrives safely.
394 SIZEHINT is the buffer size sufficient to hold all the data in the
395 typical case (it is used as the initial buffer size). MAXSIZE is
396 the maximum amount of memory this function is allowed to allocate,
397 or 0 if no upper limit is to be enforced.
399 This function should be used as a building block for other
400 functions -- see fd_read_line as a simple example. */
403 fd_read_hunk (int fd, hunk_terminator_t terminator, long sizehint, long maxsize)
405 long bufsize = sizehint;
406 char *hunk = xmalloc (bufsize);
407 int tail = 0; /* tail position in HUNK */
409 assert (!maxsize || maxsize >= bufsize);
414 int pklen, rdlen, remain;
416 /* First, peek at the available data. */
418 pklen = fd_peek (fd, hunk + tail, bufsize - 1 - tail, -1);
424 end = terminator (hunk, hunk + tail, pklen);
427 /* The data contains the terminator: we'll drain the data up
428 to the end of the terminator. */
429 remain = end - (hunk + tail);
430 assert (remain >= 0);
433 /* No more data needs to be read. */
437 if (bufsize - 1 < tail + remain)
439 bufsize = tail + remain + 1;
440 hunk = xrealloc (hunk, bufsize);
444 /* No terminator: simply read the data we know is (or should
448 /* Now, read the data. Note that we make no assumptions about
449 how much data we'll get. (Some TCP stacks are notorious for
450 read returning less data than the previous MSG_PEEK.) */
452 rdlen = fd_read (fd, hunk + tail, remain, 0);
465 /* EOF without anything having been read */
471 /* EOF seen: return the data we've read. */
474 if (end && rdlen == remain)
475 /* The terminator was seen and the remaining data drained --
476 we got what we came for. */
479 /* Keep looping until all the data arrives. */
481 if (tail == bufsize - 1)
483 /* Double the buffer size, but refuse to allocate more than
485 if (maxsize && bufsize >= maxsize)
492 if (maxsize && bufsize > maxsize)
494 hunk = xrealloc (hunk, bufsize);
500 line_terminator (const char *start, const char *peeked, int peeklen)
502 const char *p = memchr (peeked, '\n', peeklen);
504 /* p+1 because the line must include '\n' */
509 /* The maximum size of the single line we agree to accept. This is
510 not meant to impose an arbitrary limit, but to protect the user
511 from Wget slurping up available memory upon encountering malicious
512 or buggy server output. Define it to 0 to remove the limit. */
513 #define FD_READ_LINE_MAX 4096
515 /* Read one line from FD and return it. The line is allocated using
516 malloc, but is never larger than FD_READ_LINE_MAX.
518 If an error occurs, or if no data can be read, NULL is returned.
519 In the former case errno indicates the error condition, and in the
520 latter case, errno is NULL. */
523 fd_read_line (int fd)
525 return fd_read_hunk (fd, line_terminator, 128, FD_READ_LINE_MAX);
528 /* Return a printed representation of the download rate, along with
529 the units appropriate for the download speed. */
532 retr_rate (wgint bytes, double secs)
535 static const char *rate_names[] = {"B/s", "KB/s", "MB/s", "GB/s" };
538 double dlrate = calc_rate (bytes, secs, &units);
539 /* Use more digits for smaller numbers (regardless of unit used),
540 e.g. "1022", "247", "12.5", "2.38". */
541 sprintf (res, "%.*f %s",
542 dlrate >= 99.95 ? 0 : dlrate >= 9.995 ? 1 : 2,
543 dlrate, rate_names[units]);
548 /* Calculate the download rate and trim it as appropriate for the
549 speed. Appropriate means that if rate is greater than 1K/s,
550 kilobytes are used, and if rate is greater than 1MB/s, megabytes
553 UNITS is zero for B/s, one for KB/s, two for MB/s, and three for
557 calc_rate (wgint bytes, double secs, int *units)
565 /* If elapsed time is exactly zero, it means we're under the
566 resolution of the timer. This can easily happen on systems
567 that use time() for the timer. Since the interval lies between
568 0 and the timer's resolution, assume half the resolution. */
569 secs = ptimer_resolution () / 2.0;
571 dlrate = bytes / secs;
574 else if (dlrate < 1024.0 * 1024.0)
575 *units = 1, dlrate /= 1024.0;
576 else if (dlrate < 1024.0 * 1024.0 * 1024.0)
577 *units = 2, dlrate /= (1024.0 * 1024.0);
579 /* Maybe someone will need this, one day. */
580 *units = 3, dlrate /= (1024.0 * 1024.0 * 1024.0);
586 #define SUSPEND_POST_DATA do { \
587 post_data_suspended = true; \
588 saved_post_data = opt.post_data; \
589 saved_post_file_name = opt.post_file_name; \
590 opt.post_data = NULL; \
591 opt.post_file_name = NULL; \
594 #define RESTORE_POST_DATA do { \
595 if (post_data_suspended) \
597 opt.post_data = saved_post_data; \
598 opt.post_file_name = saved_post_file_name; \
599 post_data_suspended = false; \
603 static char *getproxy (struct url *);
605 /* Retrieve the given URL. Decides which loop to call -- HTTP, FTP,
608 /* #### This function should be rewritten so it doesn't return from
612 retrieve_url (struct url * orig_parsed, const char *origurl, char **file,
613 char **newloc, const char *refurl, int *dt, bool recursive,
618 bool location_changed;
619 bool iri_fallbacked = 0;
621 char *mynewloc, *proxy;
622 struct url *u = orig_parsed, *proxy_url;
623 int up_error_code; /* url parse error code */
625 int redirection_count = 0;
627 bool post_data_suspended = false;
628 char *saved_post_data = NULL;
629 char *saved_post_file_name = NULL;
631 /* If dt is NULL, use local storage. */
637 url = xstrdup (origurl);
644 refurl = opt.referer;
647 /* (also for IRI fallbacking) */
654 proxy = getproxy (u);
657 struct iri *pi = iri_new ();
658 set_uri_encoding (pi, opt.locale, true);
659 pi->utf8_encode = false;
661 /* Parse the proxy URL. */
662 proxy_url = url_parse (proxy, &up_error_code, NULL, true);
665 char *error = url_error (proxy, up_error_code);
666 logprintf (LOG_NOTQUIET, _("Error parsing proxy URL %s: %s.\n"),
673 if (proxy_url->scheme != SCHEME_HTTP && proxy_url->scheme != u->scheme)
675 logprintf (LOG_NOTQUIET, _("Error in proxy URL %s: Must be HTTP.\n"), proxy);
676 url_free (proxy_url);
683 if (u->scheme == SCHEME_HTTP
685 || u->scheme == SCHEME_HTTPS
687 || (proxy_url && proxy_url->scheme == SCHEME_HTTP))
689 result = http_loop (u, &mynewloc, &local_file, refurl, dt, proxy_url, iri);
691 else if (u->scheme == SCHEME_FTP)
693 /* If this is a redirection, temporarily turn off opt.ftp_glob
694 and opt.recursive, both being undesirable when following
696 bool oldrec = recursive, glob = opt.ftp_glob;
697 if (redirection_count)
698 oldrec = glob = false;
700 result = ftp_loop (u, dt, proxy_url, recursive, glob);
703 /* There is a possibility of having HTTP being redirected to
704 FTP. In these cases we must decide whether the text is HTML
705 according to the suffix. The HTML suffixes are `.html',
706 `.htm' and a few others, case-insensitive. */
707 if (redirection_count && local_file && u->scheme == SCHEME_FTP)
709 if (has_html_suffix_p (local_file))
716 url_free (proxy_url);
720 location_changed = (result == NEWLOCATION);
721 if (location_changed)
723 char *construced_newloc;
724 struct url *newloc_parsed;
726 assert (mynewloc != NULL);
731 /* The HTTP specs only allow absolute URLs to appear in
732 redirects, but a ton of boneheaded webservers and CGIs out
733 there break the rules and use relative URLs, and popular
734 browsers are lenient about this, so wget should be too. */
735 construced_newloc = uri_merge (url, mynewloc);
737 mynewloc = construced_newloc;
739 /* Reset UTF-8 encoding state, keep the URI encoding and reset
740 the content encoding. */
741 iri->utf8_encode = opt.enable_iri;
742 set_content_encoding (iri, NULL);
743 xfree_null (iri->orig_url);
745 /* Now, see if this new location makes sense. */
746 newloc_parsed = url_parse (mynewloc, &up_error_code, iri, true);
749 char *error = url_error (mynewloc, up_error_code);
750 logprintf (LOG_NOTQUIET, "%s: %s.\n", escnonprint_uri (mynewloc),
752 if (orig_parsed != u)
763 /* Now mynewloc will become newloc_parsed->url, because if the
764 Location contained relative paths like .././something, we
765 don't want that propagating as url. */
767 mynewloc = xstrdup (newloc_parsed->url);
769 /* Check for max. number of redirections. */
770 if (++redirection_count > opt.max_redirect)
772 logprintf (LOG_NOTQUIET, _("%d redirections exceeded.\n"),
774 url_free (newloc_parsed);
775 if (orig_parsed != u)
787 if (orig_parsed != u)
793 /* If we're being redirected from POST, we don't want to POST
794 again. Many requests answer POST with a redirection to an
795 index page; that redirection is clearly a GET. We "suspend"
796 POST data for the duration of the redirections, and restore
797 it when we're done. */
798 if (!post_data_suspended)
804 /* Try to not encode in UTF-8 if fetching failed */
805 if (!(*dt & RETROKF) && iri->utf8_encode)
807 iri->utf8_encode = false;
808 if (orig_parsed != u)
812 u = url_parse (origurl, NULL, iri, true);
815 DEBUGP (("[IRI fallbacking to non-utf8 for %s\n", quote (url)));
816 url = xstrdup (u->url);
821 DEBUGP (("[Couldn't fallback to non-utf8 for %s\n", quote (url)));
824 if (local_file && *dt & RETROKF)
826 register_download (u->url, local_file);
827 if (redirection_count && 0 != strcmp (origurl, u->url))
828 register_redirection (origurl, u->url);
830 register_html (u->url, local_file);
833 register_download (u->url, local_file);
834 if (redirection_count && 0 != strcmp (origurl, u->url))
835 register_redirection (origurl, u->url);
837 register_html (u->url, local_file);
839 register_css (u->url, local_file);
844 *file = local_file ? local_file : NULL;
846 xfree_null (local_file);
848 if (orig_parsed != u)
853 if (redirection_count || iri_fallbacked)
872 /* Find the URLs in the file and call retrieve_url() for each of them.
873 If HTML is true, treat the file as HTML, and construct the URLs
876 If opt.recursive is set, call retrieve_tree() for each file. */
879 retrieve_from_file (const char *file, bool html, int *count)
882 struct urlpos *url_list, *cur_url;
883 struct iri *iri = iri_new();
885 char *input_file = NULL;
886 const char *url = file;
888 status = RETROK; /* Suppose everything is OK. */
889 *count = 0; /* Reset the URL count. */
891 /* sXXXav : Assume filename and links in the file are in the locale */
892 set_uri_encoding (iri, opt.locale, true);
893 set_content_encoding (iri, opt.locale);
895 if (url_has_scheme (url))
899 struct url * url_parsed = url_parse(url, &url_err, iri, true);
903 char *error = url_error (url, url_err);
904 logprintf (LOG_NOTQUIET, "%s: %s.\n", url, error);
910 opt.base_href = xstrdup (url);
912 status = retrieve_url (url_parsed, url, &input_file, NULL, NULL, &dt,
914 if (status != RETROK)
920 /* If we have a found a content encoding, use it.
921 * ( == is okay, because we're checking for identical object) */
922 if (iri->content_encoding != opt.locale)
923 set_uri_encoding (iri, iri->content_encoding, false);
925 /* Reset UTF-8 encode status */
926 iri->utf8_encode = opt.enable_iri;
927 xfree_null (iri->orig_url);
928 iri->orig_url = NULL;
931 input_file = (char *) file;
933 url_list = (html ? get_urls_html (input_file, NULL, NULL, iri)
934 : get_urls_file (input_file));
936 for (cur_url = url_list; cur_url; cur_url = cur_url->next, ++*count)
938 char *filename = NULL, *new_file = NULL;
940 struct iri *tmpiri = iri_dup (iri);
941 struct url *parsed_url = NULL;
943 if (cur_url->ignore_when_downloading)
946 if (opt.quota && total_downloaded_bytes > opt.quota)
952 /* Need to reparse the url, since it didn't have iri information. */
954 parsed_url = url_parse (cur_url->url->url, NULL, tmpiri, true);
956 if ((opt.recursive || opt.page_requisites)
957 && (cur_url->url->scheme != SCHEME_FTP || getproxy (cur_url->url)))
959 int old_follow_ftp = opt.follow_ftp;
961 /* Turn opt.follow_ftp on in case of recursive FTP retrieval */
962 if (cur_url->url->scheme == SCHEME_FTP)
965 status = retrieve_tree (parsed_url ? parsed_url : cur_url->url,
968 opt.follow_ftp = old_follow_ftp;
971 status = retrieve_url (parsed_url ? parsed_url : cur_url->url,
972 cur_url->url->url, &filename,
973 &new_file, NULL, &dt, opt.recursive, tmpiri);
976 url_free (parsed_url);
978 if (filename && opt.delete_after && file_exists_p (filename))
981 Removing file due to --delete-after in retrieve_from_file():\n"));
982 logprintf (LOG_VERBOSE, _("Removing %s.\n"), filename);
983 if (unlink (filename))
984 logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
988 xfree_null (new_file);
989 xfree_null (filename);
993 /* Free the linked list of URL-s. */
994 free_urlpos (url_list);
1001 /* Print `giving up', or `retrying', depending on the impending
1002 action. N1 and N2 are the attempt number and the attempt limit. */
1004 printwhat (int n1, int n2)
1006 logputs (LOG_VERBOSE, (n1 == n2) ? _("Giving up.\n\n") : _("Retrying.\n\n"));
1009 /* If opt.wait or opt.waitretry are specified, and if certain
1010 conditions are met, sleep the appropriate number of seconds. See
1011 the documentation of --wait and --waitretry for more information.
1013 COUNT is the count of current retrieval, beginning with 1. */
1016 sleep_between_retrievals (int count)
1018 static bool first_retrieval = true;
1020 if (first_retrieval)
1022 /* Don't sleep before the very first retrieval. */
1023 first_retrieval = false;
1027 if (opt.waitretry && count > 1)
1029 /* If opt.waitretry is specified and this is a retry, wait for
1030 COUNT-1 number of seconds, or for opt.waitretry seconds. */
1031 if (count <= opt.waitretry)
1034 xsleep (opt.waitretry);
1038 if (!opt.random_wait || count > 1)
1039 /* If random-wait is not specified, or if we are sleeping
1040 between retries of the same download, sleep the fixed
1045 /* Sleep a random amount of time averaging in opt.wait
1046 seconds. The sleeping amount ranges from 0.5*opt.wait to
1048 double waitsecs = (0.5 + random_float ()) * opt.wait;
1049 DEBUGP (("sleep_between_retrievals: avg=%f,sleep=%f\n",
1050 opt.wait, waitsecs));
1056 /* Free the linked list of urlpos. */
1058 free_urlpos (struct urlpos *l)
1062 struct urlpos *next = l->next;
1065 xfree_null (l->local_name);
1071 /* Rotate FNAME opt.backups times */
1073 rotate_backups(const char *fname)
1075 int maxlen = strlen (fname) + 1 + numdigit (opt.backups) + 1;
1076 char *from = (char *)alloca (maxlen);
1077 char *to = (char *)alloca (maxlen);
1081 if (stat (fname, &sb) == 0)
1082 if (S_ISREG (sb.st_mode) == 0)
1085 for (i = opt.backups; i > 1; i--)
1087 sprintf (from, "%s.%d", fname, i - 1);
1088 sprintf (to, "%s.%d", fname, i);
1092 sprintf (to, "%s.%d", fname, 1);
1096 static bool no_proxy_match (const char *, const char **);
1098 /* Return the URL of the proxy appropriate for url U. */
1101 getproxy (struct url *u)
1104 char *rewritten_url;
1105 static char rewritten_storage[1024];
1109 if (no_proxy_match (u->host, (const char **)opt.no_proxy))
1115 proxy = opt.http_proxy ? opt.http_proxy : getenv ("http_proxy");
1119 proxy = opt.https_proxy ? opt.https_proxy : getenv ("https_proxy");
1123 proxy = opt.ftp_proxy ? opt.ftp_proxy : getenv ("ftp_proxy");
1125 case SCHEME_INVALID:
1128 if (!proxy || !*proxy)
1131 /* Handle shorthands. `rewritten_storage' is a kludge to allow
1132 getproxy() to return static storage. */
1133 rewritten_url = rewrite_shorthand_url (proxy);
1136 strncpy (rewritten_storage, rewritten_url, sizeof (rewritten_storage));
1137 rewritten_storage[sizeof (rewritten_storage) - 1] = '\0';
1138 proxy = rewritten_storage;
1144 /* Returns true if URL would be downloaded through a proxy. */
1147 url_uses_proxy (struct url * u)
1152 ret = getproxy (u) != NULL;
1156 /* Should a host be accessed through proxy, concerning no_proxy? */
1158 no_proxy_match (const char *host, const char **no_proxy)
1163 return sufmatch (no_proxy, host);
1166 /* Set the file parameter to point to the local file string. */
1168 set_local_file (const char **file, const char *default_file)
1170 if (opt.output_document)
1172 if (output_stream_regular)
1173 *file = opt.output_document;
1176 *file = default_file;