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. */
172 return !ferror (out);
175 /* Read the contents of file descriptor FD until it the connection
176 terminates or a read error occurs. The data is read in portions of
177 up to 16K and written to OUT as it arrives. If opt.verbose is set,
178 the progress is shown.
180 TOREAD is the amount of data expected to arrive, normally only used
181 by the progress gauge.
183 STARTPOS is the position from which the download starts, used by
184 the progress gauge. If QTYREAD is non-NULL, the value it points to
185 is incremented by the amount of data read from the network. If
186 QTYWRITTEN is non-NULL, the value it points to is incremented by
187 the amount of data written to disk. The time it took to download
188 the data is stored to ELAPSED.
190 The function exits and returns the amount of data read. In case of
191 error while reading data, -1 is returned. In case of error while
192 writing data, -2 is returned. */
195 fd_read_body (int fd, FILE *out, wgint toread, wgint startpos,
196 wgint *qtyread, wgint *qtywritten, double *elapsed, int flags)
200 static char dlbuf[16384];
201 int dlbufsize = sizeof (dlbuf);
203 struct ptimer *timer = NULL;
204 double last_successful_read_tm = 0;
206 /* The progress gauge, set according to the user preferences. */
207 void *progress = NULL;
209 /* Non-zero if the progress gauge is interactive, i.e. if it can
210 continually update the display. When true, smaller timeout
211 values are used so that the gauge can update the display when
212 data arrives slowly. */
213 bool progress_interactive = false;
215 bool exact = !!(flags & rb_read_exactly);
218 /* How much data we've read/written. */
220 wgint sum_written = 0;
222 if (flags & rb_skip_startpos)
227 /* If we're skipping STARTPOS bytes, pass 0 as the INITIAL
228 argument to progress_create because the indicator doesn't
229 (yet) know about "skipping" data. */
230 wgint start = skip ? 0 : startpos;
231 progress = progress_create (start, start + toread);
232 progress_interactive = progress_interactive_p (progress);
236 limit_bandwidth_reset ();
238 /* A timer is needed for tracking progress, for throttling, and for
239 tracking elapsed time. If either of these are requested, start
241 if (progress || opt.limit_rate || elapsed)
243 timer = ptimer_new ();
244 last_successful_read_tm = 0;
247 /* Use a smaller buffer for low requested bandwidths. For example,
248 with --limit-rate=2k, it doesn't make sense to slurp in 16K of
249 data and then sleep for 8s. With buffer size equal to the limit,
250 we never have to sleep for more than one second. */
251 if (opt.limit_rate && opt.limit_rate < dlbufsize)
252 dlbufsize = opt.limit_rate;
254 /* Read from FD while there is data to read. Normally toread==0
255 means that it is unknown how much data is to arrive. However, if
256 EXACT is set, then toread==0 means what it says: that no data
258 while (!exact || (sum_read < toread))
260 int rdsize = exact ? MIN (toread - sum_read, dlbufsize) : dlbufsize;
261 double tmout = opt.read_timeout;
262 if (progress_interactive)
264 /* For interactive progress gauges, always specify a ~1s
265 timeout, so that the gauge can be updated regularly even
266 when the data arrives very slowly or stalls. */
268 if (opt.read_timeout)
271 waittm = ptimer_read (timer) - last_successful_read_tm;
272 if (waittm + tmout > opt.read_timeout)
274 /* Don't let total idle time exceed read timeout. */
275 tmout = opt.read_timeout - waittm;
278 /* We've already exceeded the timeout. */
279 ret = -1, errno = ETIMEDOUT;
285 ret = fd_read (fd, dlbuf, rdsize, tmout);
287 if (progress_interactive && ret < 0 && errno == ETIMEDOUT)
288 ret = 0; /* interactive timeout, handled above */
290 break; /* EOF or read error */
292 if (progress || opt.limit_rate)
294 ptimer_measure (timer);
296 last_successful_read_tm = ptimer_read (timer);
302 if (!write_data (out, dlbuf, ret, &skip, &sum_written))
310 limit_bandwidth (ret, timer);
313 progress_update (progress, ret, ptimer_read (timer));
315 if (toread > 0 && !opt.quiet)
316 ws_percenttitle (100.0 *
317 (startpos + sum_read) / (startpos + toread));
325 progress_finish (progress, ptimer_read (timer));
328 *elapsed = ptimer_read (timer);
330 ptimer_destroy (timer);
333 *qtyread += sum_read;
335 *qtywritten += sum_written;
340 /* Read a hunk of data from FD, up until a terminator. The hunk is
341 limited by whatever the TERMINATOR callback chooses as its
342 terminator. For example, if terminator stops at newline, the hunk
343 will consist of a line of data; if terminator stops at two
344 newlines, it can be used to read the head of an HTTP response.
345 Upon determining the boundary, the function returns the data (up to
346 the terminator) in malloc-allocated storage.
348 In case of read error, NULL is returned. In case of EOF and no
349 data read, NULL is returned and errno set to 0. In case of having
350 read some data, but encountering EOF before seeing the terminator,
351 the data that has been read is returned, but it will (obviously)
352 not contain the terminator.
354 The TERMINATOR function is called with three arguments: the
355 beginning of the data read so far, the beginning of the current
356 block of peeked-at data, and the length of the current block.
357 Depending on its needs, the function is free to choose whether to
358 analyze all data or just the newly arrived data. If TERMINATOR
359 returns NULL, it means that the terminator has not been seen.
360 Otherwise it should return a pointer to the charactre immediately
361 following the terminator.
363 The idea is to be able to read a line of input, or otherwise a hunk
364 of text, such as the head of an HTTP request, without crossing the
365 boundary, so that the next call to fd_read etc. reads the data
366 after the hunk. To achieve that, this function does the following:
368 1. Peek at incoming data.
370 2. Determine whether the peeked data, along with the previously
371 read data, includes the terminator.
373 2a. If yes, read the data until the end of the terminator, and
376 2b. If no, read the peeked data and goto 1.
378 The function is careful to assume as little as possible about the
379 implementation of peeking. For example, every peek is followed by
380 a read. If the read returns a different amount of data, the
381 process is retried until all data arrives safely.
383 SIZEHINT is the buffer size sufficient to hold all the data in the
384 typical case (it is used as the initial buffer size). MAXSIZE is
385 the maximum amount of memory this function is allowed to allocate,
386 or 0 if no upper limit is to be enforced.
388 This function should be used as a building block for other
389 functions -- see fd_read_line as a simple example. */
392 fd_read_hunk (int fd, hunk_terminator_t terminator, long sizehint, long maxsize)
394 long bufsize = sizehint;
395 char *hunk = xmalloc (bufsize);
396 int tail = 0; /* tail position in HUNK */
398 assert (!maxsize || maxsize >= bufsize);
403 int pklen, rdlen, remain;
405 /* First, peek at the available data. */
407 pklen = fd_peek (fd, hunk + tail, bufsize - 1 - tail, -1);
413 end = terminator (hunk, hunk + tail, pklen);
416 /* The data contains the terminator: we'll drain the data up
417 to the end of the terminator. */
418 remain = end - (hunk + tail);
419 assert (remain >= 0);
422 /* No more data needs to be read. */
426 if (bufsize - 1 < tail + remain)
428 bufsize = tail + remain + 1;
429 hunk = xrealloc (hunk, bufsize);
433 /* No terminator: simply read the data we know is (or should
437 /* Now, read the data. Note that we make no assumptions about
438 how much data we'll get. (Some TCP stacks are notorious for
439 read returning less data than the previous MSG_PEEK.) */
441 rdlen = fd_read (fd, hunk + tail, remain, 0);
454 /* EOF without anything having been read */
460 /* EOF seen: return the data we've read. */
463 if (end && rdlen == remain)
464 /* The terminator was seen and the remaining data drained --
465 we got what we came for. */
468 /* Keep looping until all the data arrives. */
470 if (tail == bufsize - 1)
472 /* Double the buffer size, but refuse to allocate more than
474 if (maxsize && bufsize >= maxsize)
481 if (maxsize && bufsize > maxsize)
483 hunk = xrealloc (hunk, bufsize);
489 line_terminator (const char *start, const char *peeked, int peeklen)
491 const char *p = memchr (peeked, '\n', peeklen);
493 /* p+1 because the line must include '\n' */
498 /* The maximum size of the single line we agree to accept. This is
499 not meant to impose an arbitrary limit, but to protect the user
500 from Wget slurping up available memory upon encountering malicious
501 or buggy server output. Define it to 0 to remove the limit. */
502 #define FD_READ_LINE_MAX 4096
504 /* Read one line from FD and return it. The line is allocated using
505 malloc, but is never larger than FD_READ_LINE_MAX.
507 If an error occurs, or if no data can be read, NULL is returned.
508 In the former case errno indicates the error condition, and in the
509 latter case, errno is NULL. */
512 fd_read_line (int fd)
514 return fd_read_hunk (fd, line_terminator, 128, FD_READ_LINE_MAX);
517 /* Return a printed representation of the download rate, along with
518 the units appropriate for the download speed. */
521 retr_rate (wgint bytes, double secs)
524 static const char *rate_names[] = {"B/s", "KB/s", "MB/s", "GB/s" };
527 double dlrate = calc_rate (bytes, secs, &units);
528 /* Use more digits for smaller numbers (regardless of unit used),
529 e.g. "1022", "247", "12.5", "2.38". */
530 sprintf (res, "%.*f %s",
531 dlrate >= 99.95 ? 0 : dlrate >= 9.995 ? 1 : 2,
532 dlrate, rate_names[units]);
537 /* Calculate the download rate and trim it as appropriate for the
538 speed. Appropriate means that if rate is greater than 1K/s,
539 kilobytes are used, and if rate is greater than 1MB/s, megabytes
542 UNITS is zero for B/s, one for KB/s, two for MB/s, and three for
546 calc_rate (wgint bytes, double secs, int *units)
554 /* If elapsed time is exactly zero, it means we're under the
555 resolution of the timer. This can easily happen on systems
556 that use time() for the timer. Since the interval lies between
557 0 and the timer's resolution, assume half the resolution. */
558 secs = ptimer_resolution () / 2.0;
560 dlrate = bytes / secs;
563 else if (dlrate < 1024.0 * 1024.0)
564 *units = 1, dlrate /= 1024.0;
565 else if (dlrate < 1024.0 * 1024.0 * 1024.0)
566 *units = 2, dlrate /= (1024.0 * 1024.0);
568 /* Maybe someone will need this, one day. */
569 *units = 3, dlrate /= (1024.0 * 1024.0 * 1024.0);
575 #define SUSPEND_POST_DATA do { \
576 post_data_suspended = true; \
577 saved_post_data = opt.post_data; \
578 saved_post_file_name = opt.post_file_name; \
579 opt.post_data = NULL; \
580 opt.post_file_name = NULL; \
583 #define RESTORE_POST_DATA do { \
584 if (post_data_suspended) \
586 opt.post_data = saved_post_data; \
587 opt.post_file_name = saved_post_file_name; \
588 post_data_suspended = false; \
592 static char *getproxy (struct url *);
594 /* Retrieve the given URL. Decides which loop to call -- HTTP, FTP,
597 /* #### This function should be rewritten so it doesn't return from
601 retrieve_url (struct url * orig_parsed, const char *origurl, char **file,
602 char **newloc, const char *refurl, int *dt, bool recursive,
607 bool location_changed;
609 char *mynewloc, *proxy;
610 struct url *u = orig_parsed, *proxy_url;
611 int up_error_code; /* url parse error code */
613 int redirection_count = 0;
615 bool post_data_suspended = false;
616 char *saved_post_data = NULL;
617 char *saved_post_file_name = NULL;
619 /* If dt is NULL, use local storage. */
625 url = xstrdup (origurl);
632 DEBUGP (("[IRI Retrieving %s with %s (UTF-8=%d)\n", quote_n (0, url),
633 iri->uri_encoding ? quote_n (1, iri->uri_encoding) : "None",
637 refurl = opt.referer;
646 proxy = getproxy (u);
649 struct iri *pi = iri_new ();
650 set_uri_encoding (pi, opt.locale, true);
651 pi->utf8_encode = false;
653 /* Parse the proxy URL. */
654 proxy_url = url_parse (proxy, &up_error_code, NULL, true);
657 char *error = url_error (proxy, up_error_code);
658 logprintf (LOG_NOTQUIET, _("Error parsing proxy URL %s: %s.\n"),
665 if (proxy_url->scheme != SCHEME_HTTP && proxy_url->scheme != u->scheme)
667 logprintf (LOG_NOTQUIET, _("Error in proxy URL %s: Must be HTTP.\n"), proxy);
668 url_free (proxy_url);
675 if (u->scheme == SCHEME_HTTP
677 || u->scheme == SCHEME_HTTPS
679 || (proxy_url && proxy_url->scheme == SCHEME_HTTP))
681 result = http_loop (u, &mynewloc, &local_file, refurl, dt, proxy_url, iri);
683 else if (u->scheme == SCHEME_FTP)
685 /* If this is a redirection, temporarily turn off opt.ftp_glob
686 and opt.recursive, both being undesirable when following
688 bool oldrec = recursive, glob = opt.ftp_glob;
689 if (redirection_count)
690 oldrec = glob = false;
692 result = ftp_loop (u, dt, proxy_url, recursive, glob);
695 /* There is a possibility of having HTTP being redirected to
696 FTP. In these cases we must decide whether the text is HTML
697 according to the suffix. The HTML suffixes are `.html',
698 `.htm' and a few others, case-insensitive. */
699 if (redirection_count && local_file && u->scheme == SCHEME_FTP)
701 if (has_html_suffix_p (local_file))
708 url_free (proxy_url);
712 location_changed = (result == NEWLOCATION);
713 if (location_changed)
715 char *construced_newloc;
716 struct url *newloc_parsed;
718 assert (mynewloc != NULL);
723 /* The HTTP specs only allow absolute URLs to appear in
724 redirects, but a ton of boneheaded webservers and CGIs out
725 there break the rules and use relative URLs, and popular
726 browsers are lenient about this, so wget should be too. */
727 construced_newloc = uri_merge (url, mynewloc);
729 mynewloc = construced_newloc;
731 /* Reset UTF-8 encoding state, keep the URI encoding and reset
732 the content encoding. */
733 iri->utf8_encode = opt.enable_iri;
734 set_content_encoding (iri, NULL);
735 xfree_null (iri->orig_url);
737 /* Now, see if this new location makes sense. */
738 newloc_parsed = url_parse (mynewloc, &up_error_code, iri, true);
741 char *error = url_error (mynewloc, up_error_code);
742 logprintf (LOG_NOTQUIET, "%s: %s.\n", escnonprint_uri (mynewloc),
744 if (orig_parsed != u)
755 /* Now mynewloc will become newloc_parsed->url, because if the
756 Location contained relative paths like .././something, we
757 don't want that propagating as url. */
759 mynewloc = xstrdup (newloc_parsed->url);
761 /* Check for max. number of redirections. */
762 if (++redirection_count > opt.max_redirect)
764 logprintf (LOG_NOTQUIET, _("%d redirections exceeded.\n"),
766 url_free (newloc_parsed);
767 if (orig_parsed != u)
779 if (orig_parsed != u)
785 /* If we're being redirected from POST, we don't want to POST
786 again. Many requests answer POST with a redirection to an
787 index page; that redirection is clearly a GET. We "suspend"
788 POST data for the duration of the redirections, and restore
789 it when we're done. */
790 if (!post_data_suspended)
796 /* Try to not encode in UTF-8 if fetching failed */
797 if (!(*dt & RETROKF) && iri->utf8_encode)
799 iri->utf8_encode = false;
800 if (orig_parsed != u)
804 u = url_parse (origurl, NULL, iri, true);
807 DEBUGP (("[IRI fallbacking to non-utf8 for %s\n", quote (url)));
811 DEBUGP (("[Couldn't fallback to non-utf8 for %s\n", quote (url)));
814 if (local_file && *dt & RETROKF)
816 register_download (u->url, local_file);
817 if (redirection_count && 0 != strcmp (origurl, u->url))
818 register_redirection (origurl, u->url);
820 register_html (u->url, local_file);
823 register_download (u->url, local_file);
824 if (redirection_count && 0 != strcmp (origurl, u->url))
825 register_redirection (origurl, u->url);
827 register_html (u->url, local_file);
829 register_css (u->url, local_file);
834 *file = local_file ? local_file : NULL;
836 xfree_null (local_file);
838 if (orig_parsed != u)
843 if (redirection_count)
862 /* Find the URLs in the file and call retrieve_url() for each of them.
863 If HTML is true, treat the file as HTML, and construct the URLs
866 If opt.recursive is set, call retrieve_tree() for each file. */
869 retrieve_from_file (const char *file, bool html, int *count)
872 struct urlpos *url_list, *cur_url;
873 struct iri *iri = iri_new();
875 char *input_file = NULL;
876 const char *url = file;
878 status = RETROK; /* Suppose everything is OK. */
879 *count = 0; /* Reset the URL count. */
881 /* sXXXav : Assume filename and links in the file are in the locale */
882 set_uri_encoding (iri, opt.locale, true);
883 set_content_encoding (iri, opt.locale);
885 if (url_has_scheme (url))
889 struct url * url_parsed = url_parse(url, &url_err, iri, true);
893 char *error = url_error (url, url_err);
894 logprintf (LOG_NOTQUIET, "%s: %s.\n", url, error);
900 opt.base_href = xstrdup (url);
902 status = retrieve_url (url_parsed, url, &input_file, NULL, NULL, &dt,
904 if (status != RETROK)
910 /* If we have a found a content encoding, use it.
911 * ( == is okay, because we're checking for identical object) */
912 if (iri->content_encoding != opt.locale)
913 set_uri_encoding (iri, iri->content_encoding, false);
915 /* Reset UTF-8 encode status */
916 iri->utf8_encode = opt.enable_iri;
917 xfree_null (iri->orig_url);
918 iri->orig_url = NULL;
921 input_file = (char *) file;
923 url_list = (html ? get_urls_html (input_file, NULL, NULL, iri)
924 : get_urls_file (input_file));
926 for (cur_url = url_list; cur_url; cur_url = cur_url->next, ++*count)
928 char *filename = NULL, *new_file = NULL;
930 struct iri *tmpiri = iri_dup (iri);
931 struct url *parsed_url = NULL;
933 if (cur_url->ignore_when_downloading)
936 if (opt.quota && total_downloaded_bytes > opt.quota)
942 /* Need to reparse the url, since it didn't have iri information. */
944 parsed_url = url_parse (cur_url->url->url, NULL, tmpiri, true);
946 if ((opt.recursive || opt.page_requisites)
947 && (cur_url->url->scheme != SCHEME_FTP || getproxy (cur_url->url)))
949 int old_follow_ftp = opt.follow_ftp;
951 /* Turn opt.follow_ftp on in case of recursive FTP retrieval */
952 if (cur_url->url->scheme == SCHEME_FTP)
955 status = retrieve_tree (parsed_url ? parsed_url : cur_url->url,
958 opt.follow_ftp = old_follow_ftp;
961 status = retrieve_url (parsed_url ? parsed_url : cur_url->url,
962 cur_url->url->url, &filename,
963 &new_file, NULL, &dt, opt.recursive, tmpiri);
966 url_free (parsed_url);
968 if (filename && opt.delete_after && file_exists_p (filename))
971 Removing file due to --delete-after in retrieve_from_file():\n"));
972 logprintf (LOG_VERBOSE, _("Removing %s.\n"), filename);
973 if (unlink (filename))
974 logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
978 xfree_null (new_file);
979 xfree_null (filename);
983 /* Free the linked list of URL-s. */
984 free_urlpos (url_list);
991 /* Print `giving up', or `retrying', depending on the impending
992 action. N1 and N2 are the attempt number and the attempt limit. */
994 printwhat (int n1, int n2)
996 logputs (LOG_VERBOSE, (n1 == n2) ? _("Giving up.\n\n") : _("Retrying.\n\n"));
999 /* If opt.wait or opt.waitretry are specified, and if certain
1000 conditions are met, sleep the appropriate number of seconds. See
1001 the documentation of --wait and --waitretry for more information.
1003 COUNT is the count of current retrieval, beginning with 1. */
1006 sleep_between_retrievals (int count)
1008 static bool first_retrieval = true;
1010 if (first_retrieval)
1012 /* Don't sleep before the very first retrieval. */
1013 first_retrieval = false;
1017 if (opt.waitretry && count > 1)
1019 /* If opt.waitretry is specified and this is a retry, wait for
1020 COUNT-1 number of seconds, or for opt.waitretry seconds. */
1021 if (count <= opt.waitretry)
1024 xsleep (opt.waitretry);
1028 if (!opt.random_wait || count > 1)
1029 /* If random-wait is not specified, or if we are sleeping
1030 between retries of the same download, sleep the fixed
1035 /* Sleep a random amount of time averaging in opt.wait
1036 seconds. The sleeping amount ranges from 0.5*opt.wait to
1038 double waitsecs = (0.5 + random_float ()) * opt.wait;
1039 DEBUGP (("sleep_between_retrievals: avg=%f,sleep=%f\n",
1040 opt.wait, waitsecs));
1046 /* Free the linked list of urlpos. */
1048 free_urlpos (struct urlpos *l)
1052 struct urlpos *next = l->next;
1055 xfree_null (l->local_name);
1061 /* Rotate FNAME opt.backups times */
1063 rotate_backups(const char *fname)
1065 int maxlen = strlen (fname) + 1 + numdigit (opt.backups) + 1;
1066 char *from = (char *)alloca (maxlen);
1067 char *to = (char *)alloca (maxlen);
1071 if (stat (fname, &sb) == 0)
1072 if (S_ISREG (sb.st_mode) == 0)
1075 for (i = opt.backups; i > 1; i--)
1077 sprintf (from, "%s.%d", fname, i - 1);
1078 sprintf (to, "%s.%d", fname, i);
1082 sprintf (to, "%s.%d", fname, 1);
1086 static bool no_proxy_match (const char *, const char **);
1088 /* Return the URL of the proxy appropriate for url U. */
1091 getproxy (struct url *u)
1094 char *rewritten_url;
1095 static char rewritten_storage[1024];
1099 if (no_proxy_match (u->host, (const char **)opt.no_proxy))
1105 proxy = opt.http_proxy ? opt.http_proxy : getenv ("http_proxy");
1109 proxy = opt.https_proxy ? opt.https_proxy : getenv ("https_proxy");
1113 proxy = opt.ftp_proxy ? opt.ftp_proxy : getenv ("ftp_proxy");
1115 case SCHEME_INVALID:
1118 if (!proxy || !*proxy)
1121 /* Handle shorthands. `rewritten_storage' is a kludge to allow
1122 getproxy() to return static storage. */
1123 rewritten_url = rewrite_shorthand_url (proxy);
1126 strncpy (rewritten_storage, rewritten_url, sizeof (rewritten_storage));
1127 rewritten_storage[sizeof (rewritten_storage) - 1] = '\0';
1128 proxy = rewritten_storage;
1134 /* Returns true if URL would be downloaded through a proxy. */
1137 url_uses_proxy (struct url * u)
1142 ret = getproxy (u) != NULL;
1146 /* Should a host be accessed through proxy, concerning no_proxy? */
1148 no_proxy_match (const char *host, const char **no_proxy)
1153 return sufmatch (no_proxy, host);
1156 /* Set the file parameter to point to the local file string. */
1158 set_local_file (const char **file, const char *default_file)
1160 if (opt.output_document)
1162 if (output_stream_regular)
1163 *file = opt.output_document;
1166 *file = default_file;