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. Return 1 if the whole BUF was
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.
216 OUT2 will get an exact copy of the response: if this is a chunked
217 response, everything -- including the chunk headers -- is written
218 to OUT2. (OUT will only get the unchunked response.)
220 The function exits and returns the amount of data read. In case of
221 error while reading data, -1 is returned. In case of error while
222 writing data to OUT, -2 is returned. In case of error while writing
223 data to OUT2, -3 is returned. */
226 fd_read_body (int fd, FILE *out, wgint toread, wgint startpos,
227 wgint *qtyread, wgint *qtywritten, double *elapsed, int flags,
232 #define max(a,b) ((a) > (b) ? (a) : (b))
233 int dlbufsize = max (BUFSIZ, 8 * 1024);
234 char *dlbuf = xmalloc (dlbufsize);
236 struct ptimer *timer = NULL;
237 double last_successful_read_tm = 0;
239 /* The progress gauge, set according to the user preferences. */
240 void *progress = NULL;
242 /* Non-zero if the progress gauge is interactive, i.e. if it can
243 continually update the display. When true, smaller timeout
244 values are used so that the gauge can update the display when
245 data arrives slowly. */
246 bool progress_interactive = false;
248 bool exact = !!(flags & rb_read_exactly);
250 /* Used only by HTTP/HTTPS chunked transfer encoding. */
251 bool chunked = flags & rb_chunked_transfer_encoding;
254 /* How much data we've read/written. */
256 wgint sum_written = 0;
257 wgint remaining_chunk_size = 0;
259 if (flags & rb_skip_startpos)
264 /* If we're skipping STARTPOS bytes, pass 0 as the INITIAL
265 argument to progress_create because the indicator doesn't
266 (yet) know about "skipping" data. */
267 wgint start = skip ? 0 : startpos;
268 progress = progress_create (start, start + toread);
269 progress_interactive = progress_interactive_p (progress);
273 limit_bandwidth_reset ();
275 /* A timer is needed for tracking progress, for throttling, and for
276 tracking elapsed time. If either of these are requested, start
278 if (progress || opt.limit_rate || elapsed)
280 timer = ptimer_new ();
281 last_successful_read_tm = 0;
284 /* Use a smaller buffer for low requested bandwidths. For example,
285 with --limit-rate=2k, it doesn't make sense to slurp in 16K of
286 data and then sleep for 8s. With buffer size equal to the limit,
287 we never have to sleep for more than one second. */
288 if (opt.limit_rate && opt.limit_rate < dlbufsize)
289 dlbufsize = opt.limit_rate;
291 /* Read from FD while there is data to read. Normally toread==0
292 means that it is unknown how much data is to arrive. However, if
293 EXACT is set, then toread==0 means what it says: that no data
295 while (!exact || (sum_read < toread))
298 double tmout = opt.read_timeout;
302 if (remaining_chunk_size == 0)
304 char *line = fd_read_line (fd);
311 else if (out2 != NULL)
312 fwrite (line, 1, strlen (line), out2);
314 remaining_chunk_size = strtol (line, &endl, 16);
317 if (remaining_chunk_size == 0)
320 line = fd_read_line (fd);
326 fwrite (line, 1, strlen (line), out2);
333 rdsize = MIN (remaining_chunk_size, dlbufsize);
336 rdsize = exact ? MIN (toread - sum_read, dlbufsize) : dlbufsize;
338 if (progress_interactive)
340 /* For interactive progress gauges, always specify a ~1s
341 timeout, so that the gauge can be updated regularly even
342 when the data arrives very slowly or stalls. */
344 if (opt.read_timeout)
347 waittm = ptimer_read (timer) - last_successful_read_tm;
348 if (waittm + tmout > opt.read_timeout)
350 /* Don't let total idle time exceed read timeout. */
351 tmout = opt.read_timeout - waittm;
354 /* We've already exceeded the timeout. */
355 ret = -1, errno = ETIMEDOUT;
361 ret = fd_read (fd, dlbuf, rdsize, tmout);
363 if (progress_interactive && ret < 0 && errno == ETIMEDOUT)
364 ret = 0; /* interactive timeout, handled above */
366 break; /* EOF or read error */
368 if (progress || opt.limit_rate || elapsed)
370 ptimer_measure (timer);
372 last_successful_read_tm = ptimer_read (timer);
378 int write_res = write_data (out, out2, dlbuf, ret, &skip, &sum_written);
381 ret = (write_res == -3) ? -3 : -2;
386 remaining_chunk_size -= ret;
387 if (remaining_chunk_size == 0)
389 char *line = fd_read_line (fd);
398 fwrite (line, 1, strlen (line), out2);
406 limit_bandwidth (ret, timer);
409 progress_update (progress, ret, ptimer_read (timer));
411 if (toread > 0 && !opt.quiet)
412 ws_percenttitle (100.0 *
413 (startpos + sum_read) / (startpos + toread));
421 progress_finish (progress, ptimer_read (timer));
424 *elapsed = ptimer_read (timer);
426 ptimer_destroy (timer);
429 *qtyread += sum_read;
431 *qtywritten += sum_written;
438 /* Read a hunk of data from FD, up until a terminator. The hunk is
439 limited by whatever the TERMINATOR callback chooses as its
440 terminator. For example, if terminator stops at newline, the hunk
441 will consist of a line of data; if terminator stops at two
442 newlines, it can be used to read the head of an HTTP response.
443 Upon determining the boundary, the function returns the data (up to
444 the terminator) in malloc-allocated storage.
446 In case of read error, NULL is returned. In case of EOF and no
447 data read, NULL is returned and errno set to 0. In case of having
448 read some data, but encountering EOF before seeing the terminator,
449 the data that has been read is returned, but it will (obviously)
450 not contain the terminator.
452 The TERMINATOR function is called with three arguments: the
453 beginning of the data read so far, the beginning of the current
454 block of peeked-at data, and the length of the current block.
455 Depending on its needs, the function is free to choose whether to
456 analyze all data or just the newly arrived data. If TERMINATOR
457 returns NULL, it means that the terminator has not been seen.
458 Otherwise it should return a pointer to the charactre immediately
459 following the terminator.
461 The idea is to be able to read a line of input, or otherwise a hunk
462 of text, such as the head of an HTTP request, without crossing the
463 boundary, so that the next call to fd_read etc. reads the data
464 after the hunk. To achieve that, this function does the following:
466 1. Peek at incoming data.
468 2. Determine whether the peeked data, along with the previously
469 read data, includes the terminator.
471 2a. If yes, read the data until the end of the terminator, and
474 2b. If no, read the peeked data and goto 1.
476 The function is careful to assume as little as possible about the
477 implementation of peeking. For example, every peek is followed by
478 a read. If the read returns a different amount of data, the
479 process is retried until all data arrives safely.
481 SIZEHINT is the buffer size sufficient to hold all the data in the
482 typical case (it is used as the initial buffer size). MAXSIZE is
483 the maximum amount of memory this function is allowed to allocate,
484 or 0 if no upper limit is to be enforced.
486 This function should be used as a building block for other
487 functions -- see fd_read_line as a simple example. */
490 fd_read_hunk (int fd, hunk_terminator_t terminator, long sizehint, long maxsize)
492 long bufsize = sizehint;
493 char *hunk = xmalloc (bufsize);
494 int tail = 0; /* tail position in HUNK */
496 assert (!maxsize || maxsize >= bufsize);
501 int pklen, rdlen, remain;
503 /* First, peek at the available data. */
505 pklen = fd_peek (fd, hunk + tail, bufsize - 1 - tail, -1);
511 end = terminator (hunk, hunk + tail, pklen);
514 /* The data contains the terminator: we'll drain the data up
515 to the end of the terminator. */
516 remain = end - (hunk + tail);
517 assert (remain >= 0);
520 /* No more data needs to be read. */
524 if (bufsize - 1 < tail + remain)
526 bufsize = tail + remain + 1;
527 hunk = xrealloc (hunk, bufsize);
531 /* No terminator: simply read the data we know is (or should
535 /* Now, read the data. Note that we make no assumptions about
536 how much data we'll get. (Some TCP stacks are notorious for
537 read returning less data than the previous MSG_PEEK.) */
539 rdlen = fd_read (fd, hunk + tail, remain, 0);
552 /* EOF without anything having been read */
558 /* EOF seen: return the data we've read. */
561 if (end && rdlen == remain)
562 /* The terminator was seen and the remaining data drained --
563 we got what we came for. */
566 /* Keep looping until all the data arrives. */
568 if (tail == bufsize - 1)
570 /* Double the buffer size, but refuse to allocate more than
572 if (maxsize && bufsize >= maxsize)
579 if (maxsize && bufsize > maxsize)
581 hunk = xrealloc (hunk, bufsize);
587 line_terminator (const char *start, const char *peeked, int peeklen)
589 const char *p = memchr (peeked, '\n', peeklen);
591 /* p+1 because the line must include '\n' */
596 /* The maximum size of the single line we agree to accept. This is
597 not meant to impose an arbitrary limit, but to protect the user
598 from Wget slurping up available memory upon encountering malicious
599 or buggy server output. Define it to 0 to remove the limit. */
600 #define FD_READ_LINE_MAX 4096
602 /* Read one line from FD and return it. The line is allocated using
603 malloc, but is never larger than FD_READ_LINE_MAX.
605 If an error occurs, or if no data can be read, NULL is returned.
606 In the former case errno indicates the error condition, and in the
607 latter case, errno is NULL. */
610 fd_read_line (int fd)
612 return fd_read_hunk (fd, line_terminator, 128, FD_READ_LINE_MAX);
615 /* Return a printed representation of the download rate, along with
616 the units appropriate for the download speed. */
619 retr_rate (wgint bytes, double secs)
622 static const char *rate_names[] = {"B/s", "KB/s", "MB/s", "GB/s" };
623 static const char *rate_names_bits[] = {"b/s", "Kb/s", "Mb/s", "Gb/s" };
626 double dlrate = calc_rate (bytes, secs, &units);
627 /* Use more digits for smaller numbers (regardless of unit used),
628 e.g. "1022", "247", "12.5", "2.38". */
629 sprintf (res, "%.*f %s",
630 dlrate >= 99.95 ? 0 : dlrate >= 9.995 ? 1 : 2,
631 dlrate, !opt.report_bps ? rate_names[units]: rate_names_bits[units]);
636 /* Calculate the download rate and trim it as appropriate for the
637 speed. Appropriate means that if rate is greater than 1K/s,
638 kilobytes are used, and if rate is greater than 1MB/s, megabytes
641 UNITS is zero for B/s, one for KB/s, two for MB/s, and three for
645 calc_rate (wgint bytes, double secs, int *units)
648 double bibyte = 1000.0;
658 /* If elapsed time is exactly zero, it means we're under the
659 resolution of the timer. This can easily happen on systems
660 that use time() for the timer. Since the interval lies between
661 0 and the timer's resolution, assume half the resolution. */
662 secs = ptimer_resolution () / 2.0;
664 dlrate = convert_to_bits (bytes) / secs;
667 else if (dlrate < (bibyte * bibyte))
668 *units = 1, dlrate /= bibyte;
669 else if (dlrate < (bibyte * bibyte * bibyte))
670 *units = 2, dlrate /= (bibyte * bibyte);
673 /* Maybe someone will need this, one day. */
674 *units = 3, dlrate /= (bibyte * bibyte * bibyte);
680 #define SUSPEND_METHOD do { \
681 method_suspended = true; \
682 saved_body_data = opt.body_data; \
683 saved_body_file_name = opt.body_file; \
684 saved_method = opt.method; \
685 opt.body_data = NULL; \
686 opt.body_file = NULL; \
690 #define RESTORE_METHOD do { \
691 if (method_suspended) \
693 opt.body_data = saved_body_data; \
694 opt.body_file = saved_body_file_name; \
695 opt.method = saved_method; \
696 method_suspended = false; \
700 static char *getproxy (struct url *);
702 /* Retrieve the given URL. Decides which loop to call -- HTTP, FTP,
705 /* #### This function should be rewritten so it doesn't return from
709 retrieve_url (struct url * orig_parsed, const char *origurl, char **file,
710 char **newloc, const char *refurl, int *dt, bool recursive,
711 struct iri *iri, bool register_status)
715 bool location_changed;
716 bool iri_fallbacked = 0;
718 char *mynewloc, *proxy;
719 struct url *u = orig_parsed, *proxy_url;
720 int up_error_code; /* url parse error code */
722 int redirection_count = 0;
724 bool method_suspended = false;
725 char *saved_body_data = NULL;
726 char *saved_method = NULL;
727 char *saved_body_file_name = NULL;
729 /* If dt is NULL, use local storage. */
735 url = xstrdup (origurl);
742 refurl = opt.referer;
745 /* (also for IRI fallbacking) */
752 proxy = getproxy (u);
755 struct iri *pi = iri_new ();
756 set_uri_encoding (pi, opt.locale, true);
757 pi->utf8_encode = false;
759 /* Parse the proxy URL. */
760 proxy_url = url_parse (proxy, &up_error_code, NULL, true);
763 char *error = url_error (proxy, up_error_code);
764 logprintf (LOG_NOTQUIET, _("Error parsing proxy URL %s: %s.\n"),
772 if (proxy_url->scheme != SCHEME_HTTP && proxy_url->scheme != u->scheme)
774 logprintf (LOG_NOTQUIET, _("Error in proxy URL %s: Must be HTTP.\n"), proxy);
775 url_free (proxy_url);
783 if (u->scheme == SCHEME_HTTP
785 || u->scheme == SCHEME_HTTPS
787 || (proxy_url && proxy_url->scheme == SCHEME_HTTP))
789 result = http_loop (u, orig_parsed, &mynewloc, &local_file, refurl, dt,
792 else if (u->scheme == SCHEME_FTP)
794 /* If this is a redirection, temporarily turn off opt.ftp_glob
795 and opt.recursive, both being undesirable when following
797 bool oldrec = recursive, glob = opt.ftp_glob;
798 if (redirection_count)
799 oldrec = glob = false;
801 result = ftp_loop (u, &local_file, dt, proxy_url, recursive, glob);
804 /* There is a possibility of having HTTP being redirected to
805 FTP. In these cases we must decide whether the text is HTML
806 according to the suffix. The HTML suffixes are `.html',
807 `.htm' and a few others, case-insensitive. */
808 if (redirection_count && local_file && u->scheme == SCHEME_FTP)
810 if (has_html_suffix_p (local_file))
817 url_free (proxy_url);
821 location_changed = (result == NEWLOCATION || result == NEWLOCATION_KEEP_POST);
822 if (location_changed)
824 char *construced_newloc;
825 struct url *newloc_parsed;
827 assert (mynewloc != NULL);
832 /* The HTTP specs only allow absolute URLs to appear in
833 redirects, but a ton of boneheaded webservers and CGIs out
834 there break the rules and use relative URLs, and popular
835 browsers are lenient about this, so wget should be too. */
836 construced_newloc = uri_merge (url, mynewloc);
838 mynewloc = construced_newloc;
840 /* Reset UTF-8 encoding state, keep the URI encoding and reset
841 the content encoding. */
842 iri->utf8_encode = opt.enable_iri;
843 set_content_encoding (iri, NULL);
844 xfree_null (iri->orig_url);
845 iri->orig_url = NULL;
847 /* Now, see if this new location makes sense. */
848 newloc_parsed = url_parse (mynewloc, &up_error_code, iri, true);
851 char *error = url_error (mynewloc, up_error_code);
852 logprintf (LOG_NOTQUIET, "%s: %s.\n", escnonprint_uri (mynewloc),
854 if (orig_parsed != u)
865 /* Now mynewloc will become newloc_parsed->url, because if the
866 Location contained relative paths like .././something, we
867 don't want that propagating as url. */
869 mynewloc = xstrdup (newloc_parsed->url);
871 /* Check for max. number of redirections. */
872 if (++redirection_count > opt.max_redirect)
874 logprintf (LOG_NOTQUIET, _("%d redirections exceeded.\n"),
876 url_free (newloc_parsed);
877 if (orig_parsed != u)
890 if (orig_parsed != u)
896 /* If we're being redirected from POST, and we received a
897 redirect code different than 307, we don't want to POST
898 again. Many requests answer POST with a redirection to an
899 index page; that redirection is clearly a GET. We "suspend"
900 POST data for the duration of the redirections, and restore
903 RFC2616 HTTP/1.1 introduces code 307 Temporary Redirect
904 specifically to preserve the method of the request.
906 if (result != NEWLOCATION_KEEP_POST && !method_suspended)
912 /* Try to not encode in UTF-8 if fetching failed */
913 if (!(*dt & RETROKF) && iri->utf8_encode)
915 iri->utf8_encode = false;
916 if (orig_parsed != u)
920 u = url_parse (origurl, NULL, iri, true);
923 DEBUGP (("[IRI fallbacking to non-utf8 for %s\n", quote (url)));
924 url = xstrdup (u->url);
929 DEBUGP (("[Couldn't fallback to non-utf8 for %s\n", quote (url)));
932 if (local_file && u && *dt & RETROKF)
934 register_download (u->url, local_file);
936 if (!opt.spider && redirection_count && 0 != strcmp (origurl, u->url))
937 register_redirection (origurl, u->url);
940 register_html (local_file);
943 register_css (local_file);
947 *file = local_file ? local_file : NULL;
949 xfree_null (local_file);
951 if (orig_parsed != u)
956 if (redirection_count || iri_fallbacked)
974 inform_exit_status (result);
978 /* Find the URLs in the file and call retrieve_url() for each of them.
979 If HTML is true, treat the file as HTML, and construct the URLs
982 If opt.recursive is set, call retrieve_tree() for each file. */
985 retrieve_from_file (const char *file, bool html, int *count)
988 struct urlpos *url_list, *cur_url;
989 struct iri *iri = iri_new();
991 char *input_file, *url_file = NULL;
992 const char *url = file;
994 status = RETROK; /* Suppose everything is OK. */
995 *count = 0; /* Reset the URL count. */
997 /* sXXXav : Assume filename and links in the file are in the locale */
998 set_uri_encoding (iri, opt.locale, true);
999 set_content_encoding (iri, opt.locale);
1001 if (url_valid_scheme (url))
1005 struct url *url_parsed = url_parse (url, &url_err, iri, true);
1008 char *error = url_error (url, url_err);
1009 logprintf (LOG_NOTQUIET, "%s: %s.\n", url, error);
1015 opt.base_href = xstrdup (url);
1017 status = retrieve_url (url_parsed, url, &url_file, NULL, NULL, &dt,
1019 url_free (url_parsed);
1021 if (!url_file || (status != RETROK))
1027 /* If we have a found a content encoding, use it.
1028 * ( == is okay, because we're checking for identical object) */
1029 if (iri->content_encoding != opt.locale)
1030 set_uri_encoding (iri, iri->content_encoding, false);
1032 /* Reset UTF-8 encode status */
1033 iri->utf8_encode = opt.enable_iri;
1034 xfree_null (iri->orig_url);
1035 iri->orig_url = NULL;
1037 input_file = url_file;
1040 input_file = (char *) file;
1042 url_list = (html ? get_urls_html (input_file, NULL, NULL, iri)
1043 : get_urls_file (input_file));
1045 xfree_null (url_file);
1047 for (cur_url = url_list; cur_url; cur_url = cur_url->next, ++*count)
1049 char *filename = NULL, *new_file = NULL;
1051 struct iri *tmpiri = iri_dup (iri);
1052 struct url *parsed_url = NULL;
1054 if (cur_url->ignore_when_downloading)
1057 if (opt.quota && total_downloaded_bytes > opt.quota)
1063 parsed_url = url_parse (cur_url->url->url, NULL, tmpiri, true);
1065 if ((opt.recursive || opt.page_requisites)
1066 && (cur_url->url->scheme != SCHEME_FTP || getproxy (cur_url->url)))
1068 int old_follow_ftp = opt.follow_ftp;
1070 /* Turn opt.follow_ftp on in case of recursive FTP retrieval */
1071 if (cur_url->url->scheme == SCHEME_FTP)
1074 status = retrieve_tree (parsed_url ? parsed_url : cur_url->url,
1077 opt.follow_ftp = old_follow_ftp;
1080 status = retrieve_url (parsed_url ? parsed_url : cur_url->url,
1081 cur_url->url->url, &filename,
1082 &new_file, NULL, &dt, opt.recursive, tmpiri,
1086 url_free (parsed_url);
1088 if (filename && opt.delete_after && file_exists_p (filename))
1091 Removing file due to --delete-after in retrieve_from_file():\n"));
1092 logprintf (LOG_VERBOSE, _("Removing %s.\n"), filename);
1093 if (unlink (filename))
1094 logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
1098 xfree_null (new_file);
1099 xfree_null (filename);
1103 /* Free the linked list of URL-s. */
1104 free_urlpos (url_list);
1111 /* Print `giving up', or `retrying', depending on the impending
1112 action. N1 and N2 are the attempt number and the attempt limit. */
1114 printwhat (int n1, int n2)
1116 logputs (LOG_VERBOSE, (n1 == n2) ? _("Giving up.\n\n") : _("Retrying.\n\n"));
1119 /* If opt.wait or opt.waitretry are specified, and if certain
1120 conditions are met, sleep the appropriate number of seconds. See
1121 the documentation of --wait and --waitretry for more information.
1123 COUNT is the count of current retrieval, beginning with 1. */
1126 sleep_between_retrievals (int count)
1128 static bool first_retrieval = true;
1130 if (first_retrieval)
1132 /* Don't sleep before the very first retrieval. */
1133 first_retrieval = false;
1137 if (opt.waitretry && count > 1)
1139 /* If opt.waitretry is specified and this is a retry, wait for
1140 COUNT-1 number of seconds, or for opt.waitretry seconds. */
1141 if (count <= opt.waitretry)
1144 xsleep (opt.waitretry);
1148 if (!opt.random_wait || count > 1)
1149 /* If random-wait is not specified, or if we are sleeping
1150 between retries of the same download, sleep the fixed
1155 /* Sleep a random amount of time averaging in opt.wait
1156 seconds. The sleeping amount ranges from 0.5*opt.wait to
1158 double waitsecs = (0.5 + random_float ()) * opt.wait;
1159 DEBUGP (("sleep_between_retrievals: avg=%f,sleep=%f\n",
1160 opt.wait, waitsecs));
1166 /* Free the linked list of urlpos. */
1168 free_urlpos (struct urlpos *l)
1172 struct urlpos *next = l->next;
1175 xfree_null (l->local_name);
1181 /* Rotate FNAME opt.backups times */
1183 rotate_backups(const char *fname)
1185 int maxlen = strlen (fname) + 1 + numdigit (opt.backups) + 1;
1186 char *from = (char *)alloca (maxlen);
1187 char *to = (char *)alloca (maxlen);
1191 if (stat (fname, &sb) == 0)
1192 if (S_ISREG (sb.st_mode) == 0)
1195 for (i = opt.backups; i > 1; i--)
1197 sprintf (from, "%s.%d", fname, i - 1);
1198 sprintf (to, "%s.%d", fname, i);
1202 sprintf (to, "%s.%d", fname, 1);
1206 static bool no_proxy_match (const char *, const char **);
1208 /* Return the URL of the proxy appropriate for url U. */
1211 getproxy (struct url *u)
1214 char *rewritten_url;
1215 static char rewritten_storage[1024];
1219 if (no_proxy_match (u->host, (const char **)opt.no_proxy))
1225 proxy = opt.http_proxy ? opt.http_proxy : getenv ("http_proxy");
1229 proxy = opt.https_proxy ? opt.https_proxy : getenv ("https_proxy");
1233 proxy = opt.ftp_proxy ? opt.ftp_proxy : getenv ("ftp_proxy");
1235 case SCHEME_INVALID:
1238 if (!proxy || !*proxy)
1241 /* Handle shorthands. `rewritten_storage' is a kludge to allow
1242 getproxy() to return static storage. */
1243 rewritten_url = rewrite_shorthand_url (proxy);
1246 strncpy (rewritten_storage, rewritten_url, sizeof (rewritten_storage));
1247 rewritten_storage[sizeof (rewritten_storage) - 1] = '\0';
1248 proxy = rewritten_storage;
1254 /* Returns true if URL would be downloaded through a proxy. */
1257 url_uses_proxy (struct url * u)
1262 ret = getproxy (u) != NULL;
1266 /* Should a host be accessed through proxy, concerning no_proxy? */
1268 no_proxy_match (const char *host, const char **no_proxy)
1273 return sufmatch (no_proxy, host);
1276 /* Set the file parameter to point to the local file string. */
1278 set_local_file (const char **file, const char *default_file)
1280 if (opt.output_document)
1282 if (output_stream_regular)
1283 *file = opt.output_document;
1286 *file = default_file;
1289 /* Return true for an input file's own URL, false otherwise. */
1291 input_file_url (const char *input_file)
1293 static bool first = true;
1296 && url_has_scheme (input_file)