2 Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004,
3 2005, 2006, 2007, 2008, 2009, 2010, 2011 Free Software Foundation,
6 This file is part of GNU Wget.
8 GNU Wget is free software; you can redistribute it and/or modify
9 it under the terms of the GNU General Public License as published by
10 the Free Software Foundation; either version 3 of the License, or (at
11 your option) any later version.
13 GNU Wget is distributed in the hope that it will be useful,
14 but WITHOUT ANY WARRANTY; without even the implied warranty of
15 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 GNU General Public License for more details.
18 You should have received a copy of the GNU General Public License
19 along with Wget. If not, see <http://www.gnu.org/licenses/>.
21 Additional permission under GNU GPL version 3 section 7
23 If you modify this program, or any covered work, by linking or
24 combining it with the OpenSSL project's OpenSSL library (or a
25 modified version of that library), containing parts covered by the
26 terms of the OpenSSL or SSLeay licenses, the Free Software Foundation
27 grants you additional permission to convey the resulting work.
28 Corresponding Source for a non-source form of such a combination
29 shall include the source code for the parts of OpenSSL used as well
30 as that of the covered work. */
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
142 of data written. If OUT2 is not NULL, also write BUF to OUT2.
143 In case of error writing to OUT, -1 is returned. In case of error
144 writing to OUT2, -2 is returned. In case of any other error,
148 write_data (FILE *out, FILE *out2, const char *buf, int bufsize,
149 wgint *skip, wgint *written)
151 if (out == NULL && out2 == NULL)
168 fwrite (buf, 1, bufsize, out);
170 fwrite (buf, 1, bufsize, out2);
173 /* Immediately flush the downloaded data. This should not hinder
174 performance: fast downloads will arrive in large 16K chunks
175 (which stdio would write out immediately anyway), and slow
176 downloads wouldn't be limited by disk speed. */
179 Perhaps it shouldn't hinder performance, but it sure does, at least
180 on VMS (more than 2X). Rather than speculate on what it should or
181 shouldn't do, it might make more sense to test it. Even better, it
182 might be nice to explain what possible benefit it could offer, as
183 it appears to be a clear invitation to poor performance with no
184 actual justification. (Also, why 16K? Anyone test other values?)
191 #endif /* ndef __VMS */
192 if (out != NULL && ferror (out))
194 else if (out2 != NULL && ferror (out2))
200 /* Read the contents of file descriptor FD until it the connection
201 terminates or a read error occurs. The data is read in portions of
202 up to 16K and written to OUT as it arrives. If opt.verbose is set,
203 the progress is shown.
205 TOREAD is the amount of data expected to arrive, normally only used
206 by the progress gauge.
208 STARTPOS is the position from which the download starts, used by
209 the progress gauge. If QTYREAD is non-NULL, the value it points to
210 is incremented by the amount of data read from the network. If
211 QTYWRITTEN is non-NULL, the value it points to is incremented by
212 the amount of data written to disk. The time it took to download
213 the data is stored to ELAPSED.
215 If OUT2 is non-NULL, the contents is also written to OUT2.
217 The function exits and returns the amount of data read. In case of
218 error while reading data, -1 is returned. In case of error while
219 writing data to OUT, -2 is returned. In case of error while writing
220 data to OUT2, -3 is returned. */
223 fd_read_body (int fd, FILE *out, wgint toread, wgint startpos,
224 wgint *qtyread, wgint *qtywritten, double *elapsed, int flags,
229 #define max(a,b) ((a) > (b) ? (a) : (b))
230 int dlbufsize = max (BUFSIZ, 8 * 1024);
231 char *dlbuf = xmalloc (dlbufsize);
233 struct ptimer *timer = NULL;
234 double last_successful_read_tm = 0;
236 /* The progress gauge, set according to the user preferences. */
237 void *progress = NULL;
239 /* Non-zero if the progress gauge is interactive, i.e. if it can
240 continually update the display. When true, smaller timeout
241 values are used so that the gauge can update the display when
242 data arrives slowly. */
243 bool progress_interactive = false;
245 bool exact = !!(flags & rb_read_exactly);
247 /* Used only by HTTP/HTTPS chunked transfer encoding. */
248 bool chunked = flags & rb_chunked_transfer_encoding;
251 /* How much data we've read/written. */
253 wgint sum_written = 0;
254 wgint remaining_chunk_size = 0;
256 if (flags & rb_skip_startpos)
261 /* If we're skipping STARTPOS bytes, pass 0 as the INITIAL
262 argument to progress_create because the indicator doesn't
263 (yet) know about "skipping" data. */
264 wgint start = skip ? 0 : startpos;
265 progress = progress_create (start, start + toread);
266 progress_interactive = progress_interactive_p (progress);
270 limit_bandwidth_reset ();
272 /* A timer is needed for tracking progress, for throttling, and for
273 tracking elapsed time. If either of these are requested, start
275 if (progress || opt.limit_rate || elapsed)
277 timer = ptimer_new ();
278 last_successful_read_tm = 0;
281 /* Use a smaller buffer for low requested bandwidths. For example,
282 with --limit-rate=2k, it doesn't make sense to slurp in 16K of
283 data and then sleep for 8s. With buffer size equal to the limit,
284 we never have to sleep for more than one second. */
285 if (opt.limit_rate && opt.limit_rate < dlbufsize)
286 dlbufsize = opt.limit_rate;
288 /* Read from FD while there is data to read. Normally toread==0
289 means that it is unknown how much data is to arrive. However, if
290 EXACT is set, then toread==0 means what it says: that no data
292 while (!exact || (sum_read < toread))
295 double tmout = opt.read_timeout;
299 if (remaining_chunk_size == 0)
301 char *line = fd_read_line (fd);
309 remaining_chunk_size = strtol (line, &endl, 16);
310 if (remaining_chunk_size == 0)
313 if (fd_read_line (fd) == NULL)
319 rdsize = MIN (remaining_chunk_size, dlbufsize);
322 rdsize = exact ? MIN (toread - sum_read, dlbufsize) : dlbufsize;
324 if (progress_interactive)
326 /* For interactive progress gauges, always specify a ~1s
327 timeout, so that the gauge can be updated regularly even
328 when the data arrives very slowly or stalls. */
330 if (opt.read_timeout)
333 waittm = ptimer_read (timer) - last_successful_read_tm;
334 if (waittm + tmout > opt.read_timeout)
336 /* Don't let total idle time exceed read timeout. */
337 tmout = opt.read_timeout - waittm;
340 /* We've already exceeded the timeout. */
341 ret = -1, errno = ETIMEDOUT;
347 ret = fd_read (fd, dlbuf, rdsize, tmout);
349 if (progress_interactive && ret < 0 && errno == ETIMEDOUT)
350 ret = 0; /* interactive timeout, handled above */
352 break; /* EOF or read error */
354 if (progress || opt.limit_rate || elapsed)
356 ptimer_measure (timer);
358 last_successful_read_tm = ptimer_read (timer);
364 int write_res = write_data (out, out2, dlbuf, ret, &skip, &sum_written);
367 ret = (write_res == -3) ? -3 : -2;
372 remaining_chunk_size -= ret;
373 if (remaining_chunk_size == 0)
374 if (fd_read_line (fd) == NULL)
383 limit_bandwidth (ret, timer);
386 progress_update (progress, ret, ptimer_read (timer));
388 if (toread > 0 && !opt.quiet)
389 ws_percenttitle (100.0 *
390 (startpos + sum_read) / (startpos + toread));
398 progress_finish (progress, ptimer_read (timer));
401 *elapsed = ptimer_read (timer);
403 ptimer_destroy (timer);
406 *qtyread += sum_read;
408 *qtywritten += sum_written;
415 /* Read a hunk of data from FD, up until a terminator. The hunk is
416 limited by whatever the TERMINATOR callback chooses as its
417 terminator. For example, if terminator stops at newline, the hunk
418 will consist of a line of data; if terminator stops at two
419 newlines, it can be used to read the head of an HTTP response.
420 Upon determining the boundary, the function returns the data (up to
421 the terminator) in malloc-allocated storage.
423 In case of read error, NULL is returned. In case of EOF and no
424 data read, NULL is returned and errno set to 0. In case of having
425 read some data, but encountering EOF before seeing the terminator,
426 the data that has been read is returned, but it will (obviously)
427 not contain the terminator.
429 The TERMINATOR function is called with three arguments: the
430 beginning of the data read so far, the beginning of the current
431 block of peeked-at data, and the length of the current block.
432 Depending on its needs, the function is free to choose whether to
433 analyze all data or just the newly arrived data. If TERMINATOR
434 returns NULL, it means that the terminator has not been seen.
435 Otherwise it should return a pointer to the charactre immediately
436 following the terminator.
438 The idea is to be able to read a line of input, or otherwise a hunk
439 of text, such as the head of an HTTP request, without crossing the
440 boundary, so that the next call to fd_read etc. reads the data
441 after the hunk. To achieve that, this function does the following:
443 1. Peek at incoming data.
445 2. Determine whether the peeked data, along with the previously
446 read data, includes the terminator.
448 2a. If yes, read the data until the end of the terminator, and
451 2b. If no, read the peeked data and goto 1.
453 The function is careful to assume as little as possible about the
454 implementation of peeking. For example, every peek is followed by
455 a read. If the read returns a different amount of data, the
456 process is retried until all data arrives safely.
458 SIZEHINT is the buffer size sufficient to hold all the data in the
459 typical case (it is used as the initial buffer size). MAXSIZE is
460 the maximum amount of memory this function is allowed to allocate,
461 or 0 if no upper limit is to be enforced.
463 This function should be used as a building block for other
464 functions -- see fd_read_line as a simple example. */
467 fd_read_hunk (int fd, hunk_terminator_t terminator, long sizehint, long maxsize)
469 long bufsize = sizehint;
470 char *hunk = xmalloc (bufsize);
471 int tail = 0; /* tail position in HUNK */
473 assert (!maxsize || maxsize >= bufsize);
478 int pklen, rdlen, remain;
480 /* First, peek at the available data. */
482 pklen = fd_peek (fd, hunk + tail, bufsize - 1 - tail, -1);
488 end = terminator (hunk, hunk + tail, pklen);
491 /* The data contains the terminator: we'll drain the data up
492 to the end of the terminator. */
493 remain = end - (hunk + tail);
494 assert (remain >= 0);
497 /* No more data needs to be read. */
501 if (bufsize - 1 < tail + remain)
503 bufsize = tail + remain + 1;
504 hunk = xrealloc (hunk, bufsize);
508 /* No terminator: simply read the data we know is (or should
512 /* Now, read the data. Note that we make no assumptions about
513 how much data we'll get. (Some TCP stacks are notorious for
514 read returning less data than the previous MSG_PEEK.) */
516 rdlen = fd_read (fd, hunk + tail, remain, 0);
529 /* EOF without anything having been read */
535 /* EOF seen: return the data we've read. */
538 if (end && rdlen == remain)
539 /* The terminator was seen and the remaining data drained --
540 we got what we came for. */
543 /* Keep looping until all the data arrives. */
545 if (tail == bufsize - 1)
547 /* Double the buffer size, but refuse to allocate more than
549 if (maxsize && bufsize >= maxsize)
556 if (maxsize && bufsize > maxsize)
558 hunk = xrealloc (hunk, bufsize);
564 line_terminator (const char *start, const char *peeked, int peeklen)
566 const char *p = memchr (peeked, '\n', peeklen);
568 /* p+1 because the line must include '\n' */
573 /* The maximum size of the single line we agree to accept. This is
574 not meant to impose an arbitrary limit, but to protect the user
575 from Wget slurping up available memory upon encountering malicious
576 or buggy server output. Define it to 0 to remove the limit. */
577 #define FD_READ_LINE_MAX 4096
579 /* Read one line from FD and return it. The line is allocated using
580 malloc, but is never larger than FD_READ_LINE_MAX.
582 If an error occurs, or if no data can be read, NULL is returned.
583 In the former case errno indicates the error condition, and in the
584 latter case, errno is NULL. */
587 fd_read_line (int fd)
589 return fd_read_hunk (fd, line_terminator, 128, FD_READ_LINE_MAX);
592 /* Return a printed representation of the download rate, along with
593 the units appropriate for the download speed. */
596 retr_rate (wgint bytes, double secs)
599 static const char *rate_names[] = {"B/s", "KB/s", "MB/s", "GB/s" };
602 double dlrate = calc_rate (bytes, secs, &units);
603 /* Use more digits for smaller numbers (regardless of unit used),
604 e.g. "1022", "247", "12.5", "2.38". */
605 sprintf (res, "%.*f %s",
606 dlrate >= 99.95 ? 0 : dlrate >= 9.995 ? 1 : 2,
607 dlrate, rate_names[units]);
612 /* Calculate the download rate and trim it as appropriate for the
613 speed. Appropriate means that if rate is greater than 1K/s,
614 kilobytes are used, and if rate is greater than 1MB/s, megabytes
617 UNITS is zero for B/s, one for KB/s, two for MB/s, and three for
621 calc_rate (wgint bytes, double secs, int *units)
629 /* If elapsed time is exactly zero, it means we're under the
630 resolution of the timer. This can easily happen on systems
631 that use time() for the timer. Since the interval lies between
632 0 and the timer's resolution, assume half the resolution. */
633 secs = ptimer_resolution () / 2.0;
635 dlrate = bytes / secs;
638 else if (dlrate < 1024.0 * 1024.0)
639 *units = 1, dlrate /= 1024.0;
640 else if (dlrate < 1024.0 * 1024.0 * 1024.0)
641 *units = 2, dlrate /= (1024.0 * 1024.0);
643 /* Maybe someone will need this, one day. */
644 *units = 3, dlrate /= (1024.0 * 1024.0 * 1024.0);
650 #define SUSPEND_POST_DATA do { \
651 post_data_suspended = true; \
652 saved_post_data = opt.post_data; \
653 saved_post_file_name = opt.post_file_name; \
654 opt.post_data = NULL; \
655 opt.post_file_name = NULL; \
658 #define RESTORE_POST_DATA do { \
659 if (post_data_suspended) \
661 opt.post_data = saved_post_data; \
662 opt.post_file_name = saved_post_file_name; \
663 post_data_suspended = false; \
667 static char *getproxy (struct url *);
669 /* Retrieve the given URL. Decides which loop to call -- HTTP, FTP,
672 /* #### This function should be rewritten so it doesn't return from
676 retrieve_url (struct url * orig_parsed, const char *origurl, char **file,
677 char **newloc, const char *refurl, int *dt, bool recursive,
678 struct iri *iri, bool register_status)
682 bool location_changed;
683 bool iri_fallbacked = 0;
685 char *mynewloc, *proxy;
686 struct url *u = orig_parsed, *proxy_url;
687 int up_error_code; /* url parse error code */
689 int redirection_count = 0;
691 bool post_data_suspended = false;
692 char *saved_post_data = NULL;
693 char *saved_post_file_name = NULL;
695 /* If dt is NULL, use local storage. */
701 url = xstrdup (origurl);
708 refurl = opt.referer;
711 /* (also for IRI fallbacking) */
718 proxy = getproxy (u);
721 struct iri *pi = iri_new ();
722 set_uri_encoding (pi, opt.locale, true);
723 pi->utf8_encode = false;
725 /* Parse the proxy URL. */
726 proxy_url = url_parse (proxy, &up_error_code, NULL, true);
729 char *error = url_error (proxy, up_error_code);
730 logprintf (LOG_NOTQUIET, _("Error parsing proxy URL %s: %s.\n"),
738 if (proxy_url->scheme != SCHEME_HTTP && proxy_url->scheme != u->scheme)
740 logprintf (LOG_NOTQUIET, _("Error in proxy URL %s: Must be HTTP.\n"), proxy);
741 url_free (proxy_url);
749 if (u->scheme == SCHEME_HTTP
751 || u->scheme == SCHEME_HTTPS
753 || (proxy_url && proxy_url->scheme == SCHEME_HTTP))
755 result = http_loop (u, orig_parsed, &mynewloc, &local_file, refurl, dt,
758 else if (u->scheme == SCHEME_FTP)
760 /* If this is a redirection, temporarily turn off opt.ftp_glob
761 and opt.recursive, both being undesirable when following
763 bool oldrec = recursive, glob = opt.ftp_glob;
764 if (redirection_count)
765 oldrec = glob = false;
767 result = ftp_loop (u, &local_file, dt, proxy_url, recursive, glob);
770 /* There is a possibility of having HTTP being redirected to
771 FTP. In these cases we must decide whether the text is HTML
772 according to the suffix. The HTML suffixes are `.html',
773 `.htm' and a few others, case-insensitive. */
774 if (redirection_count && local_file && u->scheme == SCHEME_FTP)
776 if (has_html_suffix_p (local_file))
783 url_free (proxy_url);
787 location_changed = (result == NEWLOCATION || result == NEWLOCATION_KEEP_POST);
788 if (location_changed)
790 char *construced_newloc;
791 struct url *newloc_parsed;
793 assert (mynewloc != NULL);
798 /* The HTTP specs only allow absolute URLs to appear in
799 redirects, but a ton of boneheaded webservers and CGIs out
800 there break the rules and use relative URLs, and popular
801 browsers are lenient about this, so wget should be too. */
802 construced_newloc = uri_merge (url, mynewloc);
804 mynewloc = construced_newloc;
806 /* Reset UTF-8 encoding state, keep the URI encoding and reset
807 the content encoding. */
808 iri->utf8_encode = opt.enable_iri;
809 set_content_encoding (iri, NULL);
810 xfree_null (iri->orig_url);
812 /* Now, see if this new location makes sense. */
813 newloc_parsed = url_parse (mynewloc, &up_error_code, iri, true);
816 char *error = url_error (mynewloc, up_error_code);
817 logprintf (LOG_NOTQUIET, "%s: %s.\n", escnonprint_uri (mynewloc),
819 if (orig_parsed != u)
830 /* Now mynewloc will become newloc_parsed->url, because if the
831 Location contained relative paths like .././something, we
832 don't want that propagating as url. */
834 mynewloc = xstrdup (newloc_parsed->url);
836 /* Check for max. number of redirections. */
837 if (++redirection_count > opt.max_redirect)
839 logprintf (LOG_NOTQUIET, _("%d redirections exceeded.\n"),
841 url_free (newloc_parsed);
842 if (orig_parsed != u)
855 if (orig_parsed != u)
861 /* If we're being redirected from POST, and we received a
862 redirect code different than 307, we don't want to POST
863 again. Many requests answer POST with a redirection to an
864 index page; that redirection is clearly a GET. We "suspend"
865 POST data for the duration of the redirections, and restore
868 RFC2616 HTTP/1.1 introduces code 307 Temporary Redirect
869 specifically to preserve the method of the request.
871 if (result != NEWLOCATION_KEEP_POST && !post_data_suspended)
877 /* Try to not encode in UTF-8 if fetching failed */
878 if (!(*dt & RETROKF) && iri->utf8_encode)
880 iri->utf8_encode = false;
881 if (orig_parsed != u)
885 u = url_parse (origurl, NULL, iri, true);
888 DEBUGP (("[IRI fallbacking to non-utf8 for %s\n", quote (url)));
889 url = xstrdup (u->url);
894 DEBUGP (("[Couldn't fallback to non-utf8 for %s\n", quote (url)));
897 if (local_file && u && *dt & RETROKF)
899 register_download (u->url, local_file);
901 if (!opt.spider && redirection_count && 0 != strcmp (origurl, u->url))
902 register_redirection (origurl, u->url);
905 register_html (u->url, local_file);
908 register_css (u->url, local_file);
912 *file = local_file ? local_file : NULL;
914 xfree_null (local_file);
916 if (orig_parsed != u)
921 if (redirection_count || iri_fallbacked)
939 inform_exit_status (result);
943 /* Find the URLs in the file and call retrieve_url() for each of them.
944 If HTML is true, treat the file as HTML, and construct the URLs
947 If opt.recursive is set, call retrieve_tree() for each file. */
950 retrieve_from_file (const char *file, bool html, int *count)
953 struct urlpos *url_list, *cur_url;
954 struct iri *iri = iri_new();
956 char *input_file, *url_file = NULL;
957 const char *url = file;
959 status = RETROK; /* Suppose everything is OK. */
960 *count = 0; /* Reset the URL count. */
962 /* sXXXav : Assume filename and links in the file are in the locale */
963 set_uri_encoding (iri, opt.locale, true);
964 set_content_encoding (iri, opt.locale);
966 if (url_valid_scheme (url))
970 struct url *url_parsed = url_parse (url, &url_err, iri, true);
973 char *error = url_error (url, url_err);
974 logprintf (LOG_NOTQUIET, "%s: %s.\n", url, error);
980 opt.base_href = xstrdup (url);
982 status = retrieve_url (url_parsed, url, &url_file, NULL, NULL, &dt,
984 url_free (url_parsed);
986 if (!url_file || (status != RETROK))
992 /* If we have a found a content encoding, use it.
993 * ( == is okay, because we're checking for identical object) */
994 if (iri->content_encoding != opt.locale)
995 set_uri_encoding (iri, iri->content_encoding, false);
997 /* Reset UTF-8 encode status */
998 iri->utf8_encode = opt.enable_iri;
999 xfree_null (iri->orig_url);
1000 iri->orig_url = NULL;
1002 input_file = url_file;
1005 input_file = (char *) file;
1007 url_list = (html ? get_urls_html (input_file, NULL, NULL, iri)
1008 : get_urls_file (input_file));
1010 xfree_null (url_file);
1012 for (cur_url = url_list; cur_url; cur_url = cur_url->next, ++*count)
1014 char *filename = NULL, *new_file = NULL;
1016 struct iri *tmpiri = iri_dup (iri);
1017 struct url *parsed_url = NULL;
1019 if (cur_url->ignore_when_downloading)
1022 if (opt.quota && total_downloaded_bytes > opt.quota)
1028 parsed_url = url_parse (cur_url->url->url, NULL, tmpiri, true);
1030 if ((opt.recursive || opt.page_requisites)
1031 && (cur_url->url->scheme != SCHEME_FTP || getproxy (cur_url->url)))
1033 int old_follow_ftp = opt.follow_ftp;
1035 /* Turn opt.follow_ftp on in case of recursive FTP retrieval */
1036 if (cur_url->url->scheme == SCHEME_FTP)
1039 status = retrieve_tree (parsed_url ? parsed_url : cur_url->url,
1042 opt.follow_ftp = old_follow_ftp;
1045 status = retrieve_url (parsed_url ? parsed_url : cur_url->url,
1046 cur_url->url->url, &filename,
1047 &new_file, NULL, &dt, opt.recursive, tmpiri,
1051 url_free (parsed_url);
1053 if (filename && opt.delete_after && file_exists_p (filename))
1056 Removing file due to --delete-after in retrieve_from_file():\n"));
1057 logprintf (LOG_VERBOSE, _("Removing %s.\n"), filename);
1058 if (unlink (filename))
1059 logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
1063 xfree_null (new_file);
1064 xfree_null (filename);
1068 /* Free the linked list of URL-s. */
1069 free_urlpos (url_list);
1076 /* Print `giving up', or `retrying', depending on the impending
1077 action. N1 and N2 are the attempt number and the attempt limit. */
1079 printwhat (int n1, int n2)
1081 logputs (LOG_VERBOSE, (n1 == n2) ? _("Giving up.\n\n") : _("Retrying.\n\n"));
1084 /* If opt.wait or opt.waitretry are specified, and if certain
1085 conditions are met, sleep the appropriate number of seconds. See
1086 the documentation of --wait and --waitretry for more information.
1088 COUNT is the count of current retrieval, beginning with 1. */
1091 sleep_between_retrievals (int count)
1093 static bool first_retrieval = true;
1095 if (first_retrieval)
1097 /* Don't sleep before the very first retrieval. */
1098 first_retrieval = false;
1102 if (opt.waitretry && count > 1)
1104 /* If opt.waitretry is specified and this is a retry, wait for
1105 COUNT-1 number of seconds, or for opt.waitretry seconds. */
1106 if (count <= opt.waitretry)
1109 xsleep (opt.waitretry);
1113 if (!opt.random_wait || count > 1)
1114 /* If random-wait is not specified, or if we are sleeping
1115 between retries of the same download, sleep the fixed
1120 /* Sleep a random amount of time averaging in opt.wait
1121 seconds. The sleeping amount ranges from 0.5*opt.wait to
1123 double waitsecs = (0.5 + random_float ()) * opt.wait;
1124 DEBUGP (("sleep_between_retrievals: avg=%f,sleep=%f\n",
1125 opt.wait, waitsecs));
1131 /* Free the linked list of urlpos. */
1133 free_urlpos (struct urlpos *l)
1137 struct urlpos *next = l->next;
1140 xfree_null (l->local_name);
1146 /* Rotate FNAME opt.backups times */
1148 rotate_backups(const char *fname)
1150 int maxlen = strlen (fname) + 1 + numdigit (opt.backups) + 1;
1151 char *from = (char *)alloca (maxlen);
1152 char *to = (char *)alloca (maxlen);
1156 if (stat (fname, &sb) == 0)
1157 if (S_ISREG (sb.st_mode) == 0)
1160 for (i = opt.backups; i > 1; i--)
1162 sprintf (from, "%s.%d", fname, i - 1);
1163 sprintf (to, "%s.%d", fname, i);
1167 sprintf (to, "%s.%d", fname, 1);
1171 static bool no_proxy_match (const char *, const char **);
1173 /* Return the URL of the proxy appropriate for url U. */
1176 getproxy (struct url *u)
1179 char *rewritten_url;
1180 static char rewritten_storage[1024];
1184 if (no_proxy_match (u->host, (const char **)opt.no_proxy))
1190 proxy = opt.http_proxy ? opt.http_proxy : getenv ("http_proxy");
1194 proxy = opt.https_proxy ? opt.https_proxy : getenv ("https_proxy");
1198 proxy = opt.ftp_proxy ? opt.ftp_proxy : getenv ("ftp_proxy");
1200 case SCHEME_INVALID:
1203 if (!proxy || !*proxy)
1206 /* Handle shorthands. `rewritten_storage' is a kludge to allow
1207 getproxy() to return static storage. */
1208 rewritten_url = rewrite_shorthand_url (proxy);
1211 strncpy (rewritten_storage, rewritten_url, sizeof (rewritten_storage));
1212 rewritten_storage[sizeof (rewritten_storage) - 1] = '\0';
1213 proxy = rewritten_storage;
1219 /* Returns true if URL would be downloaded through a proxy. */
1222 url_uses_proxy (struct url * u)
1227 ret = getproxy (u) != NULL;
1231 /* Should a host be accessed through proxy, concerning no_proxy? */
1233 no_proxy_match (const char *host, const char **no_proxy)
1238 return sufmatch (no_proxy, host);
1241 /* Set the file parameter to point to the local file string. */
1243 set_local_file (const char **file, const char *default_file)
1245 if (opt.output_document)
1247 if (output_stream_regular)
1248 *file = opt.output_document;
1251 *file = default_file;
1254 /* Return true for an input file's own URL, false otherwise. */
1256 input_file_url (const char *input_file)
1258 static bool first = true;
1261 && url_has_scheme (input_file)