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);
312 if (remaining_chunk_size == 0)
315 line = fd_read_line (fd);
324 rdsize = MIN (remaining_chunk_size, dlbufsize);
327 rdsize = exact ? MIN (toread - sum_read, dlbufsize) : dlbufsize;
329 if (progress_interactive)
331 /* For interactive progress gauges, always specify a ~1s
332 timeout, so that the gauge can be updated regularly even
333 when the data arrives very slowly or stalls. */
335 if (opt.read_timeout)
338 waittm = ptimer_read (timer) - last_successful_read_tm;
339 if (waittm + tmout > opt.read_timeout)
341 /* Don't let total idle time exceed read timeout. */
342 tmout = opt.read_timeout - waittm;
345 /* We've already exceeded the timeout. */
346 ret = -1, errno = ETIMEDOUT;
352 ret = fd_read (fd, dlbuf, rdsize, tmout);
354 if (progress_interactive && ret < 0 && errno == ETIMEDOUT)
355 ret = 0; /* interactive timeout, handled above */
357 break; /* EOF or read error */
359 if (progress || opt.limit_rate || elapsed)
361 ptimer_measure (timer);
363 last_successful_read_tm = ptimer_read (timer);
369 int write_res = write_data (out, out2, dlbuf, ret, &skip, &sum_written);
372 ret = (write_res == -3) ? -3 : -2;
377 remaining_chunk_size -= ret;
378 if (remaining_chunk_size == 0)
380 char *line = fd_read_line (fd);
393 limit_bandwidth (ret, timer);
396 progress_update (progress, ret, ptimer_read (timer));
398 if (toread > 0 && !opt.quiet)
399 ws_percenttitle (100.0 *
400 (startpos + sum_read) / (startpos + toread));
408 progress_finish (progress, ptimer_read (timer));
411 *elapsed = ptimer_read (timer);
413 ptimer_destroy (timer);
416 *qtyread += sum_read;
418 *qtywritten += sum_written;
425 /* Read a hunk of data from FD, up until a terminator. The hunk is
426 limited by whatever the TERMINATOR callback chooses as its
427 terminator. For example, if terminator stops at newline, the hunk
428 will consist of a line of data; if terminator stops at two
429 newlines, it can be used to read the head of an HTTP response.
430 Upon determining the boundary, the function returns the data (up to
431 the terminator) in malloc-allocated storage.
433 In case of read error, NULL is returned. In case of EOF and no
434 data read, NULL is returned and errno set to 0. In case of having
435 read some data, but encountering EOF before seeing the terminator,
436 the data that has been read is returned, but it will (obviously)
437 not contain the terminator.
439 The TERMINATOR function is called with three arguments: the
440 beginning of the data read so far, the beginning of the current
441 block of peeked-at data, and the length of the current block.
442 Depending on its needs, the function is free to choose whether to
443 analyze all data or just the newly arrived data. If TERMINATOR
444 returns NULL, it means that the terminator has not been seen.
445 Otherwise it should return a pointer to the charactre immediately
446 following the terminator.
448 The idea is to be able to read a line of input, or otherwise a hunk
449 of text, such as the head of an HTTP request, without crossing the
450 boundary, so that the next call to fd_read etc. reads the data
451 after the hunk. To achieve that, this function does the following:
453 1. Peek at incoming data.
455 2. Determine whether the peeked data, along with the previously
456 read data, includes the terminator.
458 2a. If yes, read the data until the end of the terminator, and
461 2b. If no, read the peeked data and goto 1.
463 The function is careful to assume as little as possible about the
464 implementation of peeking. For example, every peek is followed by
465 a read. If the read returns a different amount of data, the
466 process is retried until all data arrives safely.
468 SIZEHINT is the buffer size sufficient to hold all the data in the
469 typical case (it is used as the initial buffer size). MAXSIZE is
470 the maximum amount of memory this function is allowed to allocate,
471 or 0 if no upper limit is to be enforced.
473 This function should be used as a building block for other
474 functions -- see fd_read_line as a simple example. */
477 fd_read_hunk (int fd, hunk_terminator_t terminator, long sizehint, long maxsize)
479 long bufsize = sizehint;
480 char *hunk = xmalloc (bufsize);
481 int tail = 0; /* tail position in HUNK */
483 assert (!maxsize || maxsize >= bufsize);
488 int pklen, rdlen, remain;
490 /* First, peek at the available data. */
492 pklen = fd_peek (fd, hunk + tail, bufsize - 1 - tail, -1);
498 end = terminator (hunk, hunk + tail, pklen);
501 /* The data contains the terminator: we'll drain the data up
502 to the end of the terminator. */
503 remain = end - (hunk + tail);
504 assert (remain >= 0);
507 /* No more data needs to be read. */
511 if (bufsize - 1 < tail + remain)
513 bufsize = tail + remain + 1;
514 hunk = xrealloc (hunk, bufsize);
518 /* No terminator: simply read the data we know is (or should
522 /* Now, read the data. Note that we make no assumptions about
523 how much data we'll get. (Some TCP stacks are notorious for
524 read returning less data than the previous MSG_PEEK.) */
526 rdlen = fd_read (fd, hunk + tail, remain, 0);
539 /* EOF without anything having been read */
545 /* EOF seen: return the data we've read. */
548 if (end && rdlen == remain)
549 /* The terminator was seen and the remaining data drained --
550 we got what we came for. */
553 /* Keep looping until all the data arrives. */
555 if (tail == bufsize - 1)
557 /* Double the buffer size, but refuse to allocate more than
559 if (maxsize && bufsize >= maxsize)
566 if (maxsize && bufsize > maxsize)
568 hunk = xrealloc (hunk, bufsize);
574 line_terminator (const char *start, const char *peeked, int peeklen)
576 const char *p = memchr (peeked, '\n', peeklen);
578 /* p+1 because the line must include '\n' */
583 /* The maximum size of the single line we agree to accept. This is
584 not meant to impose an arbitrary limit, but to protect the user
585 from Wget slurping up available memory upon encountering malicious
586 or buggy server output. Define it to 0 to remove the limit. */
587 #define FD_READ_LINE_MAX 4096
589 /* Read one line from FD and return it. The line is allocated using
590 malloc, but is never larger than FD_READ_LINE_MAX.
592 If an error occurs, or if no data can be read, NULL is returned.
593 In the former case errno indicates the error condition, and in the
594 latter case, errno is NULL. */
597 fd_read_line (int fd)
599 return fd_read_hunk (fd, line_terminator, 128, FD_READ_LINE_MAX);
602 /* Return a printed representation of the download rate, along with
603 the units appropriate for the download speed. */
606 retr_rate (wgint bytes, double secs)
609 static const char *rate_names[] = {"B/s", "KB/s", "MB/s", "GB/s" };
612 double dlrate = calc_rate (bytes, secs, &units);
613 /* Use more digits for smaller numbers (regardless of unit used),
614 e.g. "1022", "247", "12.5", "2.38". */
615 sprintf (res, "%.*f %s",
616 dlrate >= 99.95 ? 0 : dlrate >= 9.995 ? 1 : 2,
617 dlrate, rate_names[units]);
622 /* Calculate the download rate and trim it as appropriate for the
623 speed. Appropriate means that if rate is greater than 1K/s,
624 kilobytes are used, and if rate is greater than 1MB/s, megabytes
627 UNITS is zero for B/s, one for KB/s, two for MB/s, and three for
631 calc_rate (wgint bytes, double secs, int *units)
639 /* If elapsed time is exactly zero, it means we're under the
640 resolution of the timer. This can easily happen on systems
641 that use time() for the timer. Since the interval lies between
642 0 and the timer's resolution, assume half the resolution. */
643 secs = ptimer_resolution () / 2.0;
645 dlrate = bytes / secs;
648 else if (dlrate < 1024.0 * 1024.0)
649 *units = 1, dlrate /= 1024.0;
650 else if (dlrate < 1024.0 * 1024.0 * 1024.0)
651 *units = 2, dlrate /= (1024.0 * 1024.0);
653 /* Maybe someone will need this, one day. */
654 *units = 3, dlrate /= (1024.0 * 1024.0 * 1024.0);
660 #define SUSPEND_POST_DATA do { \
661 post_data_suspended = true; \
662 saved_post_data = opt.post_data; \
663 saved_post_file_name = opt.post_file_name; \
664 opt.post_data = NULL; \
665 opt.post_file_name = NULL; \
668 #define RESTORE_POST_DATA do { \
669 if (post_data_suspended) \
671 opt.post_data = saved_post_data; \
672 opt.post_file_name = saved_post_file_name; \
673 post_data_suspended = false; \
677 static char *getproxy (struct url *);
679 /* Retrieve the given URL. Decides which loop to call -- HTTP, FTP,
682 /* #### This function should be rewritten so it doesn't return from
686 retrieve_url (struct url * orig_parsed, const char *origurl, char **file,
687 char **newloc, const char *refurl, int *dt, bool recursive,
688 struct iri *iri, bool register_status)
692 bool location_changed;
693 bool iri_fallbacked = 0;
695 char *mynewloc, *proxy;
696 struct url *u = orig_parsed, *proxy_url;
697 int up_error_code; /* url parse error code */
699 int redirection_count = 0;
701 bool post_data_suspended = false;
702 char *saved_post_data = NULL;
703 char *saved_post_file_name = NULL;
705 /* If dt is NULL, use local storage. */
711 url = xstrdup (origurl);
718 refurl = opt.referer;
721 /* (also for IRI fallbacking) */
728 proxy = getproxy (u);
731 struct iri *pi = iri_new ();
732 set_uri_encoding (pi, opt.locale, true);
733 pi->utf8_encode = false;
735 /* Parse the proxy URL. */
736 proxy_url = url_parse (proxy, &up_error_code, NULL, true);
739 char *error = url_error (proxy, up_error_code);
740 logprintf (LOG_NOTQUIET, _("Error parsing proxy URL %s: %s.\n"),
748 if (proxy_url->scheme != SCHEME_HTTP && proxy_url->scheme != u->scheme)
750 logprintf (LOG_NOTQUIET, _("Error in proxy URL %s: Must be HTTP.\n"), proxy);
751 url_free (proxy_url);
759 if (u->scheme == SCHEME_HTTP
761 || u->scheme == SCHEME_HTTPS
763 || (proxy_url && proxy_url->scheme == SCHEME_HTTP))
765 result = http_loop (u, orig_parsed, &mynewloc, &local_file, refurl, dt,
768 else if (u->scheme == SCHEME_FTP)
770 /* If this is a redirection, temporarily turn off opt.ftp_glob
771 and opt.recursive, both being undesirable when following
773 bool oldrec = recursive, glob = opt.ftp_glob;
774 if (redirection_count)
775 oldrec = glob = false;
777 result = ftp_loop (u, &local_file, dt, proxy_url, recursive, glob);
780 /* There is a possibility of having HTTP being redirected to
781 FTP. In these cases we must decide whether the text is HTML
782 according to the suffix. The HTML suffixes are `.html',
783 `.htm' and a few others, case-insensitive. */
784 if (redirection_count && local_file && u->scheme == SCHEME_FTP)
786 if (has_html_suffix_p (local_file))
793 url_free (proxy_url);
797 location_changed = (result == NEWLOCATION || result == NEWLOCATION_KEEP_POST);
798 if (location_changed)
800 char *construced_newloc;
801 struct url *newloc_parsed;
803 assert (mynewloc != NULL);
808 /* The HTTP specs only allow absolute URLs to appear in
809 redirects, but a ton of boneheaded webservers and CGIs out
810 there break the rules and use relative URLs, and popular
811 browsers are lenient about this, so wget should be too. */
812 construced_newloc = uri_merge (url, mynewloc);
814 mynewloc = construced_newloc;
816 /* Reset UTF-8 encoding state, keep the URI encoding and reset
817 the content encoding. */
818 iri->utf8_encode = opt.enable_iri;
819 set_content_encoding (iri, NULL);
820 xfree_null (iri->orig_url);
822 /* Now, see if this new location makes sense. */
823 newloc_parsed = url_parse (mynewloc, &up_error_code, iri, true);
826 char *error = url_error (mynewloc, up_error_code);
827 logprintf (LOG_NOTQUIET, "%s: %s.\n", escnonprint_uri (mynewloc),
829 if (orig_parsed != u)
840 /* Now mynewloc will become newloc_parsed->url, because if the
841 Location contained relative paths like .././something, we
842 don't want that propagating as url. */
844 mynewloc = xstrdup (newloc_parsed->url);
846 /* Check for max. number of redirections. */
847 if (++redirection_count > opt.max_redirect)
849 logprintf (LOG_NOTQUIET, _("%d redirections exceeded.\n"),
851 url_free (newloc_parsed);
852 if (orig_parsed != u)
865 if (orig_parsed != u)
871 /* If we're being redirected from POST, and we received a
872 redirect code different than 307, we don't want to POST
873 again. Many requests answer POST with a redirection to an
874 index page; that redirection is clearly a GET. We "suspend"
875 POST data for the duration of the redirections, and restore
878 RFC2616 HTTP/1.1 introduces code 307 Temporary Redirect
879 specifically to preserve the method of the request.
881 if (result != NEWLOCATION_KEEP_POST && !post_data_suspended)
887 /* Try to not encode in UTF-8 if fetching failed */
888 if (!(*dt & RETROKF) && iri->utf8_encode)
890 iri->utf8_encode = false;
891 if (orig_parsed != u)
895 u = url_parse (origurl, NULL, iri, true);
898 DEBUGP (("[IRI fallbacking to non-utf8 for %s\n", quote (url)));
899 url = xstrdup (u->url);
904 DEBUGP (("[Couldn't fallback to non-utf8 for %s\n", quote (url)));
907 if (local_file && u && *dt & RETROKF)
909 register_download (u->url, local_file);
911 if (!opt.spider && redirection_count && 0 != strcmp (origurl, u->url))
912 register_redirection (origurl, u->url);
915 register_html (u->url, local_file);
918 register_css (u->url, local_file);
922 *file = local_file ? local_file : NULL;
924 xfree_null (local_file);
926 if (orig_parsed != u)
931 if (redirection_count || iri_fallbacked)
949 inform_exit_status (result);
953 /* Find the URLs in the file and call retrieve_url() for each of them.
954 If HTML is true, treat the file as HTML, and construct the URLs
957 If opt.recursive is set, call retrieve_tree() for each file. */
960 retrieve_from_file (const char *file, bool html, int *count)
963 struct urlpos *url_list, *cur_url;
964 struct iri *iri = iri_new();
966 char *input_file, *url_file = NULL;
967 const char *url = file;
969 status = RETROK; /* Suppose everything is OK. */
970 *count = 0; /* Reset the URL count. */
972 /* sXXXav : Assume filename and links in the file are in the locale */
973 set_uri_encoding (iri, opt.locale, true);
974 set_content_encoding (iri, opt.locale);
976 if (url_valid_scheme (url))
980 struct url *url_parsed = url_parse (url, &url_err, iri, true);
983 char *error = url_error (url, url_err);
984 logprintf (LOG_NOTQUIET, "%s: %s.\n", url, error);
990 opt.base_href = xstrdup (url);
992 status = retrieve_url (url_parsed, url, &url_file, NULL, NULL, &dt,
994 url_free (url_parsed);
996 if (!url_file || (status != RETROK))
1002 /* If we have a found a content encoding, use it.
1003 * ( == is okay, because we're checking for identical object) */
1004 if (iri->content_encoding != opt.locale)
1005 set_uri_encoding (iri, iri->content_encoding, false);
1007 /* Reset UTF-8 encode status */
1008 iri->utf8_encode = opt.enable_iri;
1009 xfree_null (iri->orig_url);
1010 iri->orig_url = NULL;
1012 input_file = url_file;
1015 input_file = (char *) file;
1017 url_list = (html ? get_urls_html (input_file, NULL, NULL, iri)
1018 : get_urls_file (input_file));
1020 xfree_null (url_file);
1022 for (cur_url = url_list; cur_url; cur_url = cur_url->next, ++*count)
1024 char *filename = NULL, *new_file = NULL;
1026 struct iri *tmpiri = iri_dup (iri);
1027 struct url *parsed_url = NULL;
1029 if (cur_url->ignore_when_downloading)
1032 if (opt.quota && total_downloaded_bytes > opt.quota)
1038 parsed_url = url_parse (cur_url->url->url, NULL, tmpiri, true);
1040 if ((opt.recursive || opt.page_requisites)
1041 && (cur_url->url->scheme != SCHEME_FTP || getproxy (cur_url->url)))
1043 int old_follow_ftp = opt.follow_ftp;
1045 /* Turn opt.follow_ftp on in case of recursive FTP retrieval */
1046 if (cur_url->url->scheme == SCHEME_FTP)
1049 status = retrieve_tree (parsed_url ? parsed_url : cur_url->url,
1052 opt.follow_ftp = old_follow_ftp;
1055 status = retrieve_url (parsed_url ? parsed_url : cur_url->url,
1056 cur_url->url->url, &filename,
1057 &new_file, NULL, &dt, opt.recursive, tmpiri,
1061 url_free (parsed_url);
1063 if (filename && opt.delete_after && file_exists_p (filename))
1066 Removing file due to --delete-after in retrieve_from_file():\n"));
1067 logprintf (LOG_VERBOSE, _("Removing %s.\n"), filename);
1068 if (unlink (filename))
1069 logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
1073 xfree_null (new_file);
1074 xfree_null (filename);
1078 /* Free the linked list of URL-s. */
1079 free_urlpos (url_list);
1086 /* Print `giving up', or `retrying', depending on the impending
1087 action. N1 and N2 are the attempt number and the attempt limit. */
1089 printwhat (int n1, int n2)
1091 logputs (LOG_VERBOSE, (n1 == n2) ? _("Giving up.\n\n") : _("Retrying.\n\n"));
1094 /* If opt.wait or opt.waitretry are specified, and if certain
1095 conditions are met, sleep the appropriate number of seconds. See
1096 the documentation of --wait and --waitretry for more information.
1098 COUNT is the count of current retrieval, beginning with 1. */
1101 sleep_between_retrievals (int count)
1103 static bool first_retrieval = true;
1105 if (first_retrieval)
1107 /* Don't sleep before the very first retrieval. */
1108 first_retrieval = false;
1112 if (opt.waitretry && count > 1)
1114 /* If opt.waitretry is specified and this is a retry, wait for
1115 COUNT-1 number of seconds, or for opt.waitretry seconds. */
1116 if (count <= opt.waitretry)
1119 xsleep (opt.waitretry);
1123 if (!opt.random_wait || count > 1)
1124 /* If random-wait is not specified, or if we are sleeping
1125 between retries of the same download, sleep the fixed
1130 /* Sleep a random amount of time averaging in opt.wait
1131 seconds. The sleeping amount ranges from 0.5*opt.wait to
1133 double waitsecs = (0.5 + random_float ()) * opt.wait;
1134 DEBUGP (("sleep_between_retrievals: avg=%f,sleep=%f\n",
1135 opt.wait, waitsecs));
1141 /* Free the linked list of urlpos. */
1143 free_urlpos (struct urlpos *l)
1147 struct urlpos *next = l->next;
1150 xfree_null (l->local_name);
1156 /* Rotate FNAME opt.backups times */
1158 rotate_backups(const char *fname)
1160 int maxlen = strlen (fname) + 1 + numdigit (opt.backups) + 1;
1161 char *from = (char *)alloca (maxlen);
1162 char *to = (char *)alloca (maxlen);
1166 if (stat (fname, &sb) == 0)
1167 if (S_ISREG (sb.st_mode) == 0)
1170 for (i = opt.backups; i > 1; i--)
1172 sprintf (from, "%s.%d", fname, i - 1);
1173 sprintf (to, "%s.%d", fname, i);
1177 sprintf (to, "%s.%d", fname, 1);
1181 static bool no_proxy_match (const char *, const char **);
1183 /* Return the URL of the proxy appropriate for url U. */
1186 getproxy (struct url *u)
1189 char *rewritten_url;
1190 static char rewritten_storage[1024];
1194 if (no_proxy_match (u->host, (const char **)opt.no_proxy))
1200 proxy = opt.http_proxy ? opt.http_proxy : getenv ("http_proxy");
1204 proxy = opt.https_proxy ? opt.https_proxy : getenv ("https_proxy");
1208 proxy = opt.ftp_proxy ? opt.ftp_proxy : getenv ("ftp_proxy");
1210 case SCHEME_INVALID:
1213 if (!proxy || !*proxy)
1216 /* Handle shorthands. `rewritten_storage' is a kludge to allow
1217 getproxy() to return static storage. */
1218 rewritten_url = rewrite_shorthand_url (proxy);
1221 strncpy (rewritten_storage, rewritten_url, sizeof (rewritten_storage));
1222 rewritten_storage[sizeof (rewritten_storage) - 1] = '\0';
1223 proxy = rewritten_storage;
1229 /* Returns true if URL would be downloaded through a proxy. */
1232 url_uses_proxy (struct url * u)
1237 ret = getproxy (u) != NULL;
1241 /* Should a host be accessed through proxy, concerning no_proxy? */
1243 no_proxy_match (const char *host, const char **no_proxy)
1248 return sufmatch (no_proxy, host);
1251 /* Set the file parameter to point to the local file string. */
1253 set_local_file (const char **file, const char *default_file)
1255 if (opt.output_document)
1257 if (output_stream_regular)
1258 *file = opt.output_document;
1261 *file = default_file;
1264 /* Return true for an input file's own URL, false otherwise. */
1266 input_file_url (const char *input_file)
1268 static bool first = true;
1271 && url_has_scheme (input_file)