2 Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
3 2005, 2006, 2007, 2008, 2009, 2010 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 */
58 /* Total size of downloaded files. Used to enforce quota. */
59 SUM_SIZE_INT total_downloaded_bytes;
61 /* Total download time in seconds. */
62 double total_download_time;
64 /* If non-NULL, the stream to which output should be written. This
65 stream is initialized when `-O' is used. */
68 /* Whether output_document is a regular file we can manipulate,
69 i.e. not `-' or a device file. */
70 bool output_stream_regular;
79 limit_bandwidth_reset (void)
84 /* Limit the bandwidth by pausing the download for an amount of time.
85 BYTES is the number of bytes received from the network, and TIMER
86 is the timer that started at the beginning of download. */
89 limit_bandwidth (wgint bytes, struct ptimer *timer)
91 double delta_t = ptimer_read (timer) - limit_data.chunk_start;
94 limit_data.chunk_bytes += bytes;
96 /* Calculate the amount of time we expect downloading the chunk
97 should take. If in reality it took less time, sleep to
98 compensate for the difference. */
99 expected = (double) limit_data.chunk_bytes / opt.limit_rate;
101 if (expected > delta_t)
103 double slp = expected - delta_t + limit_data.sleep_adjust;
107 DEBUGP (("deferring a %.2f ms sleep (%s/%.2f).\n",
108 slp * 1000, number_to_static_string (limit_data.chunk_bytes),
112 DEBUGP (("\nsleeping %.2f ms for %s bytes, adjust %.2f ms\n",
113 slp * 1000, number_to_static_string (limit_data.chunk_bytes),
114 limit_data.sleep_adjust));
116 t0 = ptimer_read (timer);
118 t1 = ptimer_measure (timer);
120 /* Due to scheduling, we probably slept slightly longer (or
121 shorter) than desired. Calculate the difference between the
122 desired and the actual sleep, and adjust the next sleep by
124 limit_data.sleep_adjust = slp - (t1 - t0);
125 /* If sleep_adjust is very large, it's likely due to suspension
126 and not clock inaccuracy. Don't enforce those. */
127 if (limit_data.sleep_adjust > 0.5)
128 limit_data.sleep_adjust = 0.5;
129 else if (limit_data.sleep_adjust < -0.5)
130 limit_data.sleep_adjust = -0.5;
133 limit_data.chunk_bytes = 0;
134 limit_data.chunk_start = ptimer_read (timer);
138 # define MIN(i, j) ((i) <= (j) ? (i) : (j))
141 /* Write data in BUF to OUT. However, if *SKIP is non-zero, skip that
142 amount of data and decrease SKIP. Increment *TOTAL by the amount
146 write_data (FILE *out, const char *buf, int bufsize, wgint *skip,
165 fwrite (buf, 1, bufsize, out);
168 /* Immediately flush the downloaded data. This should not hinder
169 performance: fast downloads will arrive in large 16K chunks
170 (which stdio would write out immediately anyway), and slow
171 downloads wouldn't be limited by disk speed. */
174 Perhaps it shouldn't hinder performance, but it sure does, at least
175 on VMS (more than 2X). Rather than speculate on what it should or
176 shouldn't do, it might make more sense to test it. Even better, it
177 might be nice to explain what possible benefit it could offer, as
178 it appears to be a clear invitation to poor performance with no
179 actual justification. (Also, why 16K? Anyone test other values?)
183 #endif /* ndef __VMS */
184 return !ferror (out);
187 /* Read the contents of file descriptor FD until it the connection
188 terminates or a read error occurs. The data is read in portions of
189 up to 16K and written to OUT as it arrives. If opt.verbose is set,
190 the progress is shown.
192 TOREAD is the amount of data expected to arrive, normally only used
193 by the progress gauge.
195 STARTPOS is the position from which the download starts, used by
196 the progress gauge. If QTYREAD is non-NULL, the value it points to
197 is incremented by the amount of data read from the network. If
198 QTYWRITTEN is non-NULL, the value it points to is incremented by
199 the amount of data written to disk. The time it took to download
200 the data is stored to ELAPSED.
202 The function exits and returns the amount of data read. In case of
203 error while reading data, -1 is returned. In case of error while
204 writing data, -2 is returned. */
207 fd_read_body (int fd, FILE *out, wgint toread, wgint startpos,
208 wgint *qtyread, wgint *qtywritten, double *elapsed, int flags)
212 static char dlbuf[16384];
213 int dlbufsize = sizeof (dlbuf);
215 struct ptimer *timer = NULL;
216 double last_successful_read_tm = 0;
218 /* The progress gauge, set according to the user preferences. */
219 void *progress = NULL;
221 /* Non-zero if the progress gauge is interactive, i.e. if it can
222 continually update the display. When true, smaller timeout
223 values are used so that the gauge can update the display when
224 data arrives slowly. */
225 bool progress_interactive = false;
227 bool exact = !!(flags & rb_read_exactly);
229 /* Used only by HTTP/HTTPS chunked transfer encoding. */
230 bool chunked = flags & rb_chunked_transfer_encoding;
233 /* How much data we've read/written. */
235 wgint sum_written = 0;
236 wgint remaining_chunk_size = 0;
238 if (flags & rb_skip_startpos)
243 /* If we're skipping STARTPOS bytes, pass 0 as the INITIAL
244 argument to progress_create because the indicator doesn't
245 (yet) know about "skipping" data. */
246 wgint start = skip ? 0 : startpos;
247 progress = progress_create (start, start + toread);
248 progress_interactive = progress_interactive_p (progress);
252 limit_bandwidth_reset ();
254 /* A timer is needed for tracking progress, for throttling, and for
255 tracking elapsed time. If either of these are requested, start
257 if (progress || opt.limit_rate || elapsed)
259 timer = ptimer_new ();
260 last_successful_read_tm = 0;
263 /* Use a smaller buffer for low requested bandwidths. For example,
264 with --limit-rate=2k, it doesn't make sense to slurp in 16K of
265 data and then sleep for 8s. With buffer size equal to the limit,
266 we never have to sleep for more than one second. */
267 if (opt.limit_rate && opt.limit_rate < dlbufsize)
268 dlbufsize = opt.limit_rate;
270 /* Read from FD while there is data to read. Normally toread==0
271 means that it is unknown how much data is to arrive. However, if
272 EXACT is set, then toread==0 means what it says: that no data
274 while (!exact || (sum_read < toread))
277 double tmout = opt.read_timeout;
281 if (remaining_chunk_size == 0)
283 char *line = fd_read_line (fd);
291 remaining_chunk_size = strtol (line, &endl, 16);
292 if (remaining_chunk_size == 0)
295 if (fd_read_line (fd) == NULL)
301 rdsize = MIN (remaining_chunk_size, dlbufsize);
304 rdsize = exact ? MIN (toread - sum_read, dlbufsize) : dlbufsize;
306 if (progress_interactive)
308 /* For interactive progress gauges, always specify a ~1s
309 timeout, so that the gauge can be updated regularly even
310 when the data arrives very slowly or stalls. */
312 if (opt.read_timeout)
315 waittm = ptimer_read (timer) - last_successful_read_tm;
316 if (waittm + tmout > opt.read_timeout)
318 /* Don't let total idle time exceed read timeout. */
319 tmout = opt.read_timeout - waittm;
322 /* We've already exceeded the timeout. */
323 ret = -1, errno = ETIMEDOUT;
329 ret = fd_read (fd, dlbuf, rdsize, tmout);
331 if (progress_interactive && ret < 0 && errno == ETIMEDOUT)
332 ret = 0; /* interactive timeout, handled above */
334 break; /* EOF or read error */
336 if (progress || opt.limit_rate || elapsed)
338 ptimer_measure (timer);
340 last_successful_read_tm = ptimer_read (timer);
346 if (!write_data (out, dlbuf, ret, &skip, &sum_written))
353 remaining_chunk_size -= ret;
354 if (remaining_chunk_size == 0)
355 if (fd_read_line (fd) == NULL)
364 limit_bandwidth (ret, timer);
367 progress_update (progress, ret, ptimer_read (timer));
369 if (toread > 0 && !opt.quiet)
370 ws_percenttitle (100.0 *
371 (startpos + sum_read) / (startpos + toread));
379 progress_finish (progress, ptimer_read (timer));
382 *elapsed = ptimer_read (timer);
384 ptimer_destroy (timer);
387 *qtyread += sum_read;
389 *qtywritten += sum_written;
394 /* Read a hunk of data from FD, up until a terminator. The hunk is
395 limited by whatever the TERMINATOR callback chooses as its
396 terminator. For example, if terminator stops at newline, the hunk
397 will consist of a line of data; if terminator stops at two
398 newlines, it can be used to read the head of an HTTP response.
399 Upon determining the boundary, the function returns the data (up to
400 the terminator) in malloc-allocated storage.
402 In case of read error, NULL is returned. In case of EOF and no
403 data read, NULL is returned and errno set to 0. In case of having
404 read some data, but encountering EOF before seeing the terminator,
405 the data that has been read is returned, but it will (obviously)
406 not contain the terminator.
408 The TERMINATOR function is called with three arguments: the
409 beginning of the data read so far, the beginning of the current
410 block of peeked-at data, and the length of the current block.
411 Depending on its needs, the function is free to choose whether to
412 analyze all data or just the newly arrived data. If TERMINATOR
413 returns NULL, it means that the terminator has not been seen.
414 Otherwise it should return a pointer to the charactre immediately
415 following the terminator.
417 The idea is to be able to read a line of input, or otherwise a hunk
418 of text, such as the head of an HTTP request, without crossing the
419 boundary, so that the next call to fd_read etc. reads the data
420 after the hunk. To achieve that, this function does the following:
422 1. Peek at incoming data.
424 2. Determine whether the peeked data, along with the previously
425 read data, includes the terminator.
427 2a. If yes, read the data until the end of the terminator, and
430 2b. If no, read the peeked data and goto 1.
432 The function is careful to assume as little as possible about the
433 implementation of peeking. For example, every peek is followed by
434 a read. If the read returns a different amount of data, the
435 process is retried until all data arrives safely.
437 SIZEHINT is the buffer size sufficient to hold all the data in the
438 typical case (it is used as the initial buffer size). MAXSIZE is
439 the maximum amount of memory this function is allowed to allocate,
440 or 0 if no upper limit is to be enforced.
442 This function should be used as a building block for other
443 functions -- see fd_read_line as a simple example. */
446 fd_read_hunk (int fd, hunk_terminator_t terminator, long sizehint, long maxsize)
448 long bufsize = sizehint;
449 char *hunk = xmalloc (bufsize);
450 int tail = 0; /* tail position in HUNK */
452 assert (!maxsize || maxsize >= bufsize);
457 int pklen, rdlen, remain;
459 /* First, peek at the available data. */
461 pklen = fd_peek (fd, hunk + tail, bufsize - 1 - tail, -1);
467 end = terminator (hunk, hunk + tail, pklen);
470 /* The data contains the terminator: we'll drain the data up
471 to the end of the terminator. */
472 remain = end - (hunk + tail);
473 assert (remain >= 0);
476 /* No more data needs to be read. */
480 if (bufsize - 1 < tail + remain)
482 bufsize = tail + remain + 1;
483 hunk = xrealloc (hunk, bufsize);
487 /* No terminator: simply read the data we know is (or should
491 /* Now, read the data. Note that we make no assumptions about
492 how much data we'll get. (Some TCP stacks are notorious for
493 read returning less data than the previous MSG_PEEK.) */
495 rdlen = fd_read (fd, hunk + tail, remain, 0);
508 /* EOF without anything having been read */
514 /* EOF seen: return the data we've read. */
517 if (end && rdlen == remain)
518 /* The terminator was seen and the remaining data drained --
519 we got what we came for. */
522 /* Keep looping until all the data arrives. */
524 if (tail == bufsize - 1)
526 /* Double the buffer size, but refuse to allocate more than
528 if (maxsize && bufsize >= maxsize)
535 if (maxsize && bufsize > maxsize)
537 hunk = xrealloc (hunk, bufsize);
543 line_terminator (const char *start, const char *peeked, int peeklen)
545 const char *p = memchr (peeked, '\n', peeklen);
547 /* p+1 because the line must include '\n' */
552 /* The maximum size of the single line we agree to accept. This is
553 not meant to impose an arbitrary limit, but to protect the user
554 from Wget slurping up available memory upon encountering malicious
555 or buggy server output. Define it to 0 to remove the limit. */
556 #define FD_READ_LINE_MAX 4096
558 /* Read one line from FD and return it. The line is allocated using
559 malloc, but is never larger than FD_READ_LINE_MAX.
561 If an error occurs, or if no data can be read, NULL is returned.
562 In the former case errno indicates the error condition, and in the
563 latter case, errno is NULL. */
566 fd_read_line (int fd)
568 return fd_read_hunk (fd, line_terminator, 128, FD_READ_LINE_MAX);
571 /* Return a printed representation of the download rate, along with
572 the units appropriate for the download speed. */
575 retr_rate (wgint bytes, double secs)
578 static const char *rate_names[] = {"B/s", "KB/s", "MB/s", "GB/s" };
581 double dlrate = calc_rate (bytes, secs, &units);
582 /* Use more digits for smaller numbers (regardless of unit used),
583 e.g. "1022", "247", "12.5", "2.38". */
584 sprintf (res, "%.*f %s",
585 dlrate >= 99.95 ? 0 : dlrate >= 9.995 ? 1 : 2,
586 dlrate, rate_names[units]);
591 /* Calculate the download rate and trim it as appropriate for the
592 speed. Appropriate means that if rate is greater than 1K/s,
593 kilobytes are used, and if rate is greater than 1MB/s, megabytes
596 UNITS is zero for B/s, one for KB/s, two for MB/s, and three for
600 calc_rate (wgint bytes, double secs, int *units)
608 /* If elapsed time is exactly zero, it means we're under the
609 resolution of the timer. This can easily happen on systems
610 that use time() for the timer. Since the interval lies between
611 0 and the timer's resolution, assume half the resolution. */
612 secs = ptimer_resolution () / 2.0;
614 dlrate = bytes / secs;
617 else if (dlrate < 1024.0 * 1024.0)
618 *units = 1, dlrate /= 1024.0;
619 else if (dlrate < 1024.0 * 1024.0 * 1024.0)
620 *units = 2, dlrate /= (1024.0 * 1024.0);
622 /* Maybe someone will need this, one day. */
623 *units = 3, dlrate /= (1024.0 * 1024.0 * 1024.0);
629 #define SUSPEND_POST_DATA do { \
630 post_data_suspended = true; \
631 saved_post_data = opt.post_data; \
632 saved_post_file_name = opt.post_file_name; \
633 opt.post_data = NULL; \
634 opt.post_file_name = NULL; \
637 #define RESTORE_POST_DATA do { \
638 if (post_data_suspended) \
640 opt.post_data = saved_post_data; \
641 opt.post_file_name = saved_post_file_name; \
642 post_data_suspended = false; \
646 static char *getproxy (struct url *);
648 /* Retrieve the given URL. Decides which loop to call -- HTTP, FTP,
651 /* #### This function should be rewritten so it doesn't return from
655 retrieve_url (struct url * orig_parsed, const char *origurl, char **file,
656 char **newloc, const char *refurl, int *dt, bool recursive,
657 struct iri *iri, bool register_status)
661 bool location_changed;
662 bool iri_fallbacked = 0;
664 char *mynewloc, *proxy;
665 struct url *u = orig_parsed, *proxy_url;
666 int up_error_code; /* url parse error code */
668 int redirection_count = 0;
670 bool post_data_suspended = false;
671 char *saved_post_data = NULL;
672 char *saved_post_file_name = NULL;
674 /* If dt is NULL, use local storage. */
680 url = xstrdup (origurl);
687 refurl = opt.referer;
690 /* (also for IRI fallbacking) */
697 proxy = getproxy (u);
700 struct iri *pi = iri_new ();
701 set_uri_encoding (pi, opt.locale, true);
702 pi->utf8_encode = false;
704 /* Parse the proxy URL. */
705 proxy_url = url_parse (proxy, &up_error_code, NULL, true);
708 char *error = url_error (proxy, up_error_code);
709 logprintf (LOG_NOTQUIET, _("Error parsing proxy URL %s: %s.\n"),
717 if (proxy_url->scheme != SCHEME_HTTP && proxy_url->scheme != u->scheme)
719 logprintf (LOG_NOTQUIET, _("Error in proxy URL %s: Must be HTTP.\n"), proxy);
720 url_free (proxy_url);
728 if (u->scheme == SCHEME_HTTP
730 || u->scheme == SCHEME_HTTPS
732 || (proxy_url && proxy_url->scheme == SCHEME_HTTP))
734 result = http_loop (u, &mynewloc, &local_file, refurl, dt, proxy_url, iri);
736 else if (u->scheme == SCHEME_FTP)
738 /* If this is a redirection, temporarily turn off opt.ftp_glob
739 and opt.recursive, both being undesirable when following
741 bool oldrec = recursive, glob = opt.ftp_glob;
742 if (redirection_count)
743 oldrec = glob = false;
745 result = ftp_loop (u, &local_file, dt, proxy_url, recursive, glob);
748 /* There is a possibility of having HTTP being redirected to
749 FTP. In these cases we must decide whether the text is HTML
750 according to the suffix. The HTML suffixes are `.html',
751 `.htm' and a few others, case-insensitive. */
752 if (redirection_count && local_file && u->scheme == SCHEME_FTP)
754 if (has_html_suffix_p (local_file))
761 url_free (proxy_url);
765 location_changed = (result == NEWLOCATION);
766 if (location_changed)
768 char *construced_newloc;
769 struct url *newloc_parsed;
771 assert (mynewloc != NULL);
776 /* The HTTP specs only allow absolute URLs to appear in
777 redirects, but a ton of boneheaded webservers and CGIs out
778 there break the rules and use relative URLs, and popular
779 browsers are lenient about this, so wget should be too. */
780 construced_newloc = uri_merge (url, mynewloc);
782 mynewloc = construced_newloc;
784 /* Reset UTF-8 encoding state, keep the URI encoding and reset
785 the content encoding. */
786 iri->utf8_encode = opt.enable_iri;
787 set_content_encoding (iri, NULL);
788 xfree_null (iri->orig_url);
790 /* Now, see if this new location makes sense. */
791 newloc_parsed = url_parse (mynewloc, &up_error_code, iri, true);
794 char *error = url_error (mynewloc, up_error_code);
795 logprintf (LOG_NOTQUIET, "%s: %s.\n", escnonprint_uri (mynewloc),
797 if (orig_parsed != u)
808 /* Now mynewloc will become newloc_parsed->url, because if the
809 Location contained relative paths like .././something, we
810 don't want that propagating as url. */
812 mynewloc = xstrdup (newloc_parsed->url);
814 /* Check for max. number of redirections. */
815 if (++redirection_count > opt.max_redirect)
817 logprintf (LOG_NOTQUIET, _("%d redirections exceeded.\n"),
819 url_free (newloc_parsed);
820 if (orig_parsed != u)
833 if (orig_parsed != u)
839 /* If we're being redirected from POST, we don't want to POST
840 again. Many requests answer POST with a redirection to an
841 index page; that redirection is clearly a GET. We "suspend"
842 POST data for the duration of the redirections, and restore
843 it when we're done. */
844 if (!post_data_suspended)
850 /* Try to not encode in UTF-8 if fetching failed */
851 if (!(*dt & RETROKF) && iri->utf8_encode)
853 iri->utf8_encode = false;
854 if (orig_parsed != u)
858 u = url_parse (origurl, NULL, iri, true);
861 DEBUGP (("[IRI fallbacking to non-utf8 for %s\n", quote (url)));
862 url = xstrdup (u->url);
867 DEBUGP (("[Couldn't fallback to non-utf8 for %s\n", quote (url)));
870 if (local_file && *dt & RETROKF)
872 register_download (u->url, local_file);
873 if (redirection_count && 0 != strcmp (origurl, u->url))
874 register_redirection (origurl, u->url);
876 register_html (u->url, local_file);
879 register_download (u->url, local_file);
880 if (redirection_count && 0 != strcmp (origurl, u->url))
881 register_redirection (origurl, u->url);
883 register_html (u->url, local_file);
885 register_css (u->url, local_file);
890 *file = local_file ? local_file : NULL;
892 xfree_null (local_file);
894 if (orig_parsed != u)
899 if (redirection_count || iri_fallbacked)
917 inform_exit_status (result);
921 /* Find the URLs in the file and call retrieve_url() for each of them.
922 If HTML is true, treat the file as HTML, and construct the URLs
925 If opt.recursive is set, call retrieve_tree() for each file. */
928 retrieve_from_file (const char *file, bool html, int *count)
931 struct urlpos *url_list, *cur_url;
932 struct iri *iri = iri_new();
934 char *input_file, *url_file = NULL;
935 const char *url = file;
937 status = RETROK; /* Suppose everything is OK. */
938 *count = 0; /* Reset the URL count. */
940 /* sXXXav : Assume filename and links in the file are in the locale */
941 set_uri_encoding (iri, opt.locale, true);
942 set_content_encoding (iri, opt.locale);
944 if (url_valid_scheme (url))
948 struct url * url_parsed = url_parse(url, &url_err, iri, true);
952 char *error = url_error (url, url_err);
953 logprintf (LOG_NOTQUIET, "%s: %s.\n", url, error);
959 opt.base_href = xstrdup (url);
961 status = retrieve_url (url_parsed, url, &url_file, NULL, NULL, &dt,
963 url_free (url_parsed);
965 if (!url_file || (status != RETROK))
971 /* If we have a found a content encoding, use it.
972 * ( == is okay, because we're checking for identical object) */
973 if (iri->content_encoding != opt.locale)
974 set_uri_encoding (iri, iri->content_encoding, false);
976 /* Reset UTF-8 encode status */
977 iri->utf8_encode = opt.enable_iri;
978 xfree_null (iri->orig_url);
979 iri->orig_url = NULL;
981 input_file = url_file;
984 input_file = (char *) file;
986 url_list = (html ? get_urls_html (input_file, NULL, NULL, iri)
987 : get_urls_file (input_file));
989 xfree_null (url_file);
991 for (cur_url = url_list; cur_url; cur_url = cur_url->next, ++*count)
993 char *filename = NULL, *new_file = NULL;
995 struct iri *tmpiri = iri_dup (iri);
996 struct url *parsed_url = NULL;
998 if (cur_url->ignore_when_downloading)
1001 if (opt.quota && total_downloaded_bytes > opt.quota)
1007 /* Need to reparse the url, since it didn't have iri information. */
1009 parsed_url = url_parse (cur_url->url->url, NULL, tmpiri, true);
1011 if ((opt.recursive || opt.page_requisites)
1012 && (cur_url->url->scheme != SCHEME_FTP || getproxy (cur_url->url)))
1014 int old_follow_ftp = opt.follow_ftp;
1016 /* Turn opt.follow_ftp on in case of recursive FTP retrieval */
1017 if (cur_url->url->scheme == SCHEME_FTP)
1020 status = retrieve_tree (parsed_url ? parsed_url : cur_url->url,
1023 opt.follow_ftp = old_follow_ftp;
1026 status = retrieve_url (parsed_url ? parsed_url : cur_url->url,
1027 cur_url->url->url, &filename,
1028 &new_file, NULL, &dt, opt.recursive, tmpiri,
1032 url_free (parsed_url);
1034 if (filename && opt.delete_after && file_exists_p (filename))
1037 Removing file due to --delete-after in retrieve_from_file():\n"));
1038 logprintf (LOG_VERBOSE, _("Removing %s.\n"), filename);
1039 if (unlink (filename))
1040 logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
1044 xfree_null (new_file);
1045 xfree_null (filename);
1049 /* Free the linked list of URL-s. */
1050 free_urlpos (url_list);
1057 /* Print `giving up', or `retrying', depending on the impending
1058 action. N1 and N2 are the attempt number and the attempt limit. */
1060 printwhat (int n1, int n2)
1062 logputs (LOG_VERBOSE, (n1 == n2) ? _("Giving up.\n\n") : _("Retrying.\n\n"));
1065 /* If opt.wait or opt.waitretry are specified, and if certain
1066 conditions are met, sleep the appropriate number of seconds. See
1067 the documentation of --wait and --waitretry for more information.
1069 COUNT is the count of current retrieval, beginning with 1. */
1072 sleep_between_retrievals (int count)
1074 static bool first_retrieval = true;
1076 if (first_retrieval)
1078 /* Don't sleep before the very first retrieval. */
1079 first_retrieval = false;
1083 if (opt.waitretry && count > 1)
1085 /* If opt.waitretry is specified and this is a retry, wait for
1086 COUNT-1 number of seconds, or for opt.waitretry seconds. */
1087 if (count <= opt.waitretry)
1090 xsleep (opt.waitretry);
1094 if (!opt.random_wait || count > 1)
1095 /* If random-wait is not specified, or if we are sleeping
1096 between retries of the same download, sleep the fixed
1101 /* Sleep a random amount of time averaging in opt.wait
1102 seconds. The sleeping amount ranges from 0.5*opt.wait to
1104 double waitsecs = (0.5 + random_float ()) * opt.wait;
1105 DEBUGP (("sleep_between_retrievals: avg=%f,sleep=%f\n",
1106 opt.wait, waitsecs));
1112 /* Free the linked list of urlpos. */
1114 free_urlpos (struct urlpos *l)
1118 struct urlpos *next = l->next;
1121 xfree_null (l->local_name);
1127 /* Rotate FNAME opt.backups times */
1129 rotate_backups(const char *fname)
1131 int maxlen = strlen (fname) + 1 + numdigit (opt.backups) + 1;
1132 char *from = (char *)alloca (maxlen);
1133 char *to = (char *)alloca (maxlen);
1137 if (stat (fname, &sb) == 0)
1138 if (S_ISREG (sb.st_mode) == 0)
1141 for (i = opt.backups; i > 1; i--)
1143 sprintf (from, "%s.%d", fname, i - 1);
1144 sprintf (to, "%s.%d", fname, i);
1148 sprintf (to, "%s.%d", fname, 1);
1152 static bool no_proxy_match (const char *, const char **);
1154 /* Return the URL of the proxy appropriate for url U. */
1157 getproxy (struct url *u)
1160 char *rewritten_url;
1161 static char rewritten_storage[1024];
1165 if (no_proxy_match (u->host, (const char **)opt.no_proxy))
1171 proxy = opt.http_proxy ? opt.http_proxy : getenv ("http_proxy");
1175 proxy = opt.https_proxy ? opt.https_proxy : getenv ("https_proxy");
1179 proxy = opt.ftp_proxy ? opt.ftp_proxy : getenv ("ftp_proxy");
1181 case SCHEME_INVALID:
1184 if (!proxy || !*proxy)
1187 /* Handle shorthands. `rewritten_storage' is a kludge to allow
1188 getproxy() to return static storage. */
1189 rewritten_url = rewrite_shorthand_url (proxy);
1192 strncpy (rewritten_storage, rewritten_url, sizeof (rewritten_storage));
1193 rewritten_storage[sizeof (rewritten_storage) - 1] = '\0';
1194 proxy = rewritten_storage;
1200 /* Returns true if URL would be downloaded through a proxy. */
1203 url_uses_proxy (struct url * u)
1208 ret = getproxy (u) != NULL;
1212 /* Should a host be accessed through proxy, concerning no_proxy? */
1214 no_proxy_match (const char *host, const char **no_proxy)
1219 return sufmatch (no_proxy, host);
1222 /* Set the file parameter to point to the local file string. */
1224 set_local_file (const char **file, const char *default_file)
1226 if (opt.output_document)
1228 if (output_stream_regular)
1229 *file = opt.output_document;
1232 *file = default_file;
1235 /* Return true for an input file's own URL, false otherwise. */
1237 input_file_url (const char *input_file)
1239 static bool first = true;
1242 && url_has_scheme (input_file)