2 Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
3 2004, 2005, 2006, 2007, 2008 Free Software Foundation, Inc.
5 This file is part of GNU Wget.
7 GNU Wget is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 3 of the License, or (at
10 your option) any later version.
12 GNU Wget is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License
18 along with Wget. If not, see <http://www.gnu.org/licenses/>.
20 Additional permission under GNU GPL version 3 section 7
22 If you modify this program, or any covered work, by linking or
23 combining it with the OpenSSL project's OpenSSL library (or a
24 modified version of that library), containing parts covered by the
25 terms of the OpenSSL or SSLeay licenses, the Free Software Foundation
26 grants you additional permission to convey the resulting work.
27 Corresponding Source for a non-source form of such a combination
28 shall include the source code for the parts of OpenSSL used as well
29 as that of the covered work. */
37 #endif /* HAVE_UNISTD_H */
56 /* Total size of downloaded files. Used to enforce quota. */
57 SUM_SIZE_INT total_downloaded_bytes;
59 /* Total download time in seconds. */
60 double total_download_time;
62 /* If non-NULL, the stream to which output should be written. This
63 stream is initialized when `-O' is used. */
66 /* Whether output_document is a regular file we can manipulate,
67 i.e. not `-' or a device file. */
68 bool output_stream_regular;
77 limit_bandwidth_reset (void)
82 /* Limit the bandwidth by pausing the download for an amount of time.
83 BYTES is the number of bytes received from the network, and TIMER
84 is the timer that started at the beginning of download. */
87 limit_bandwidth (wgint bytes, struct ptimer *timer)
89 double delta_t = ptimer_read (timer) - limit_data.chunk_start;
92 limit_data.chunk_bytes += bytes;
94 /* Calculate the amount of time we expect downloading the chunk
95 should take. If in reality it took less time, sleep to
96 compensate for the difference. */
97 expected = (double) limit_data.chunk_bytes / opt.limit_rate;
99 if (expected > delta_t)
101 double slp = expected - delta_t + limit_data.sleep_adjust;
105 DEBUGP (("deferring a %.2f ms sleep (%s/%.2f).\n",
106 slp * 1000, number_to_static_string (limit_data.chunk_bytes),
110 DEBUGP (("\nsleeping %.2f ms for %s bytes, adjust %.2f ms\n",
111 slp * 1000, number_to_static_string (limit_data.chunk_bytes),
112 limit_data.sleep_adjust));
114 t0 = ptimer_read (timer);
116 t1 = ptimer_measure (timer);
118 /* Due to scheduling, we probably slept slightly longer (or
119 shorter) than desired. Calculate the difference between the
120 desired and the actual sleep, and adjust the next sleep by
122 limit_data.sleep_adjust = slp - (t1 - t0);
123 /* If sleep_adjust is very large, it's likely due to suspension
124 and not clock inaccuracy. Don't enforce those. */
125 if (limit_data.sleep_adjust > 0.5)
126 limit_data.sleep_adjust = 0.5;
127 else if (limit_data.sleep_adjust < -0.5)
128 limit_data.sleep_adjust = -0.5;
131 limit_data.chunk_bytes = 0;
132 limit_data.chunk_start = ptimer_read (timer);
136 # define MIN(i, j) ((i) <= (j) ? (i) : (j))
139 /* Write data in BUF to OUT. However, if *SKIP is non-zero, skip that
140 amount of data and decrease SKIP. Increment *TOTAL by the amount
144 write_data (FILE *out, const char *buf, int bufsize, wgint *skip,
163 fwrite (buf, 1, bufsize, out);
166 /* Immediately flush the downloaded data. This should not hinder
167 performance: fast downloads will arrive in large 16K chunks
168 (which stdio would write out immediately anyway), and slow
169 downloads wouldn't be limited by disk speed. */
171 return !ferror (out);
174 /* Read the contents of file descriptor FD until it the connection
175 terminates or a read error occurs. The data is read in portions of
176 up to 16K and written to OUT as it arrives. If opt.verbose is set,
177 the progress is shown.
179 TOREAD is the amount of data expected to arrive, normally only used
180 by the progress gauge.
182 STARTPOS is the position from which the download starts, used by
183 the progress gauge. If QTYREAD is non-NULL, the value it points to
184 is incremented by the amount of data read from the network. If
185 QTYWRITTEN is non-NULL, the value it points to is incremented by
186 the amount of data written to disk. The time it took to download
187 the data is stored to ELAPSED.
189 The function exits and returns the amount of data read. In case of
190 error while reading data, -1 is returned. In case of error while
191 writing data, -2 is returned. */
194 fd_read_body (int fd, FILE *out, wgint toread, wgint startpos,
195 wgint *qtyread, wgint *qtywritten, double *elapsed, int flags)
199 static char dlbuf[16384];
200 int dlbufsize = sizeof (dlbuf);
202 struct ptimer *timer = NULL;
203 double last_successful_read_tm = 0;
205 /* The progress gauge, set according to the user preferences. */
206 void *progress = NULL;
208 /* Non-zero if the progress gauge is interactive, i.e. if it can
209 continually update the display. When true, smaller timeout
210 values are used so that the gauge can update the display when
211 data arrives slowly. */
212 bool progress_interactive = false;
214 bool exact = !!(flags & rb_read_exactly);
217 /* How much data we've read/written. */
219 wgint sum_written = 0;
221 if (flags & rb_skip_startpos)
226 /* If we're skipping STARTPOS bytes, pass 0 as the INITIAL
227 argument to progress_create because the indicator doesn't
228 (yet) know about "skipping" data. */
229 wgint start = skip ? 0 : startpos;
230 progress = progress_create (start, start + toread);
231 progress_interactive = progress_interactive_p (progress);
235 limit_bandwidth_reset ();
237 /* A timer is needed for tracking progress, for throttling, and for
238 tracking elapsed time. If either of these are requested, start
240 if (progress || opt.limit_rate || elapsed)
242 timer = ptimer_new ();
243 last_successful_read_tm = 0;
246 /* Use a smaller buffer for low requested bandwidths. For example,
247 with --limit-rate=2k, it doesn't make sense to slurp in 16K of
248 data and then sleep for 8s. With buffer size equal to the limit,
249 we never have to sleep for more than one second. */
250 if (opt.limit_rate && opt.limit_rate < dlbufsize)
251 dlbufsize = opt.limit_rate;
253 /* Read from FD while there is data to read. Normally toread==0
254 means that it is unknown how much data is to arrive. However, if
255 EXACT is set, then toread==0 means what it says: that no data
257 while (!exact || (sum_read < toread))
259 int rdsize = exact ? MIN (toread - sum_read, dlbufsize) : dlbufsize;
260 double tmout = opt.read_timeout;
261 if (progress_interactive)
263 /* For interactive progress gauges, always specify a ~1s
264 timeout, so that the gauge can be updated regularly even
265 when the data arrives very slowly or stalls. */
267 if (opt.read_timeout)
270 waittm = ptimer_read (timer) - last_successful_read_tm;
271 if (waittm + tmout > opt.read_timeout)
273 /* Don't let total idle time exceed read timeout. */
274 tmout = opt.read_timeout - waittm;
277 /* We've already exceeded the timeout. */
278 ret = -1, errno = ETIMEDOUT;
284 ret = fd_read (fd, dlbuf, rdsize, tmout);
286 if (progress_interactive && ret < 0 && errno == ETIMEDOUT)
287 ret = 0; /* interactive timeout, handled above */
289 break; /* EOF or read error */
291 if (progress || opt.limit_rate)
293 ptimer_measure (timer);
295 last_successful_read_tm = ptimer_read (timer);
301 if (!write_data (out, dlbuf, ret, &skip, &sum_written))
309 limit_bandwidth (ret, timer);
312 progress_update (progress, ret, ptimer_read (timer));
314 if (toread > 0 && !opt.quiet)
315 ws_percenttitle (100.0 *
316 (startpos + sum_read) / (startpos + toread));
324 progress_finish (progress, ptimer_read (timer));
327 *elapsed = ptimer_read (timer);
329 ptimer_destroy (timer);
332 *qtyread += sum_read;
334 *qtywritten += sum_written;
339 /* Read a hunk of data from FD, up until a terminator. The hunk is
340 limited by whatever the TERMINATOR callback chooses as its
341 terminator. For example, if terminator stops at newline, the hunk
342 will consist of a line of data; if terminator stops at two
343 newlines, it can be used to read the head of an HTTP response.
344 Upon determining the boundary, the function returns the data (up to
345 the terminator) in malloc-allocated storage.
347 In case of read error, NULL is returned. In case of EOF and no
348 data read, NULL is returned and errno set to 0. In case of having
349 read some data, but encountering EOF before seeing the terminator,
350 the data that has been read is returned, but it will (obviously)
351 not contain the terminator.
353 The TERMINATOR function is called with three arguments: the
354 beginning of the data read so far, the beginning of the current
355 block of peeked-at data, and the length of the current block.
356 Depending on its needs, the function is free to choose whether to
357 analyze all data or just the newly arrived data. If TERMINATOR
358 returns NULL, it means that the terminator has not been seen.
359 Otherwise it should return a pointer to the charactre immediately
360 following the terminator.
362 The idea is to be able to read a line of input, or otherwise a hunk
363 of text, such as the head of an HTTP request, without crossing the
364 boundary, so that the next call to fd_read etc. reads the data
365 after the hunk. To achieve that, this function does the following:
367 1. Peek at incoming data.
369 2. Determine whether the peeked data, along with the previously
370 read data, includes the terminator.
372 2a. If yes, read the data until the end of the terminator, and
375 2b. If no, read the peeked data and goto 1.
377 The function is careful to assume as little as possible about the
378 implementation of peeking. For example, every peek is followed by
379 a read. If the read returns a different amount of data, the
380 process is retried until all data arrives safely.
382 SIZEHINT is the buffer size sufficient to hold all the data in the
383 typical case (it is used as the initial buffer size). MAXSIZE is
384 the maximum amount of memory this function is allowed to allocate,
385 or 0 if no upper limit is to be enforced.
387 This function should be used as a building block for other
388 functions -- see fd_read_line as a simple example. */
391 fd_read_hunk (int fd, hunk_terminator_t terminator, long sizehint, long maxsize)
393 long bufsize = sizehint;
394 char *hunk = xmalloc (bufsize);
395 int tail = 0; /* tail position in HUNK */
397 assert (!maxsize || maxsize >= bufsize);
402 int pklen, rdlen, remain;
404 /* First, peek at the available data. */
406 pklen = fd_peek (fd, hunk + tail, bufsize - 1 - tail, -1);
412 end = terminator (hunk, hunk + tail, pklen);
415 /* The data contains the terminator: we'll drain the data up
416 to the end of the terminator. */
417 remain = end - (hunk + tail);
418 assert (remain >= 0);
421 /* No more data needs to be read. */
425 if (bufsize - 1 < tail + remain)
427 bufsize = tail + remain + 1;
428 hunk = xrealloc (hunk, bufsize);
432 /* No terminator: simply read the data we know is (or should
436 /* Now, read the data. Note that we make no assumptions about
437 how much data we'll get. (Some TCP stacks are notorious for
438 read returning less data than the previous MSG_PEEK.) */
440 rdlen = fd_read (fd, hunk + tail, remain, 0);
453 /* EOF without anything having been read */
459 /* EOF seen: return the data we've read. */
462 if (end && rdlen == remain)
463 /* The terminator was seen and the remaining data drained --
464 we got what we came for. */
467 /* Keep looping until all the data arrives. */
469 if (tail == bufsize - 1)
471 /* Double the buffer size, but refuse to allocate more than
473 if (maxsize && bufsize >= maxsize)
480 if (maxsize && bufsize > maxsize)
482 hunk = xrealloc (hunk, bufsize);
488 line_terminator (const char *start, const char *peeked, int peeklen)
490 const char *p = memchr (peeked, '\n', peeklen);
492 /* p+1 because the line must include '\n' */
497 /* The maximum size of the single line we agree to accept. This is
498 not meant to impose an arbitrary limit, but to protect the user
499 from Wget slurping up available memory upon encountering malicious
500 or buggy server output. Define it to 0 to remove the limit. */
501 #define FD_READ_LINE_MAX 4096
503 /* Read one line from FD and return it. The line is allocated using
504 malloc, but is never larger than FD_READ_LINE_MAX.
506 If an error occurs, or if no data can be read, NULL is returned.
507 In the former case errno indicates the error condition, and in the
508 latter case, errno is NULL. */
511 fd_read_line (int fd)
513 return fd_read_hunk (fd, line_terminator, 128, FD_READ_LINE_MAX);
516 /* Return a printed representation of the download rate, along with
517 the units appropriate for the download speed. */
520 retr_rate (wgint bytes, double secs)
523 static const char *rate_names[] = {"B/s", "KB/s", "MB/s", "GB/s" };
526 double dlrate = calc_rate (bytes, secs, &units);
527 /* Use more digits for smaller numbers (regardless of unit used),
528 e.g. "1022", "247", "12.5", "2.38". */
529 sprintf (res, "%.*f %s",
530 dlrate >= 99.95 ? 0 : dlrate >= 9.995 ? 1 : 2,
531 dlrate, rate_names[units]);
536 /* Calculate the download rate and trim it as appropriate for the
537 speed. Appropriate means that if rate is greater than 1K/s,
538 kilobytes are used, and if rate is greater than 1MB/s, megabytes
541 UNITS is zero for B/s, one for KB/s, two for MB/s, and three for
545 calc_rate (wgint bytes, double secs, int *units)
553 /* If elapsed time is exactly zero, it means we're under the
554 resolution of the timer. This can easily happen on systems
555 that use time() for the timer. Since the interval lies between
556 0 and the timer's resolution, assume half the resolution. */
557 secs = ptimer_resolution () / 2.0;
559 dlrate = bytes / secs;
562 else if (dlrate < 1024.0 * 1024.0)
563 *units = 1, dlrate /= 1024.0;
564 else if (dlrate < 1024.0 * 1024.0 * 1024.0)
565 *units = 2, dlrate /= (1024.0 * 1024.0);
567 /* Maybe someone will need this, one day. */
568 *units = 3, dlrate /= (1024.0 * 1024.0 * 1024.0);
574 #define SUSPEND_POST_DATA do { \
575 post_data_suspended = true; \
576 saved_post_data = opt.post_data; \
577 saved_post_file_name = opt.post_file_name; \
578 opt.post_data = NULL; \
579 opt.post_file_name = NULL; \
582 #define RESTORE_POST_DATA do { \
583 if (post_data_suspended) \
585 opt.post_data = saved_post_data; \
586 opt.post_file_name = saved_post_file_name; \
587 post_data_suspended = false; \
591 static char *getproxy (struct url *);
593 /* Retrieve the given URL. Decides which loop to call -- HTTP, FTP,
596 /* #### This function should be rewritten so it doesn't return from
600 retrieve_url (struct url * orig_parsed, const char *origurl, char **file,
601 char **newloc, const char *refurl, int *dt, bool recursive,
606 bool location_changed;
608 char *mynewloc, *proxy;
609 struct url *u = orig_parsed, *proxy_url;
610 int up_error_code; /* url parse error code */
612 int redirection_count = 0;
614 bool post_data_suspended = false;
615 char *saved_post_data = NULL;
616 char *saved_post_file_name = NULL;
618 /* If dt is NULL, use local storage. */
624 url = xstrdup (origurl);
631 DEBUGP (("[IRI Retrieving %s with %s (UTF-8=%d)\n", quote_n (0, url),
632 iri->uri_encoding ? quote_n (1, iri->uri_encoding) : "None",
636 refurl = opt.referer;
645 proxy = getproxy (u);
648 struct iri *pi = iri_new ();
649 set_uri_encoding (pi, opt.locale, true);
650 pi->utf8_encode = false;
652 /* Parse the proxy URL. */
653 proxy_url = url_parse (proxy, &up_error_code, NULL, true);
656 char *error = url_error (proxy, up_error_code);
657 logprintf (LOG_NOTQUIET, _("Error parsing proxy URL %s: %s.\n"),
664 if (proxy_url->scheme != SCHEME_HTTP && proxy_url->scheme != u->scheme)
666 logprintf (LOG_NOTQUIET, _("Error in proxy URL %s: Must be HTTP.\n"), proxy);
667 url_free (proxy_url);
674 if (u->scheme == SCHEME_HTTP
676 || u->scheme == SCHEME_HTTPS
678 || (proxy_url && proxy_url->scheme == SCHEME_HTTP))
680 result = http_loop (u, &mynewloc, &local_file, refurl, dt, proxy_url, iri);
682 else if (u->scheme == SCHEME_FTP)
684 /* If this is a redirection, temporarily turn off opt.ftp_glob
685 and opt.recursive, both being undesirable when following
687 bool oldrec = recursive, glob = opt.ftp_glob;
688 if (redirection_count)
689 oldrec = glob = false;
691 result = ftp_loop (u, dt, proxy_url, recursive, glob);
694 /* There is a possibility of having HTTP being redirected to
695 FTP. In these cases we must decide whether the text is HTML
696 according to the suffix. The HTML suffixes are `.html',
697 `.htm' and a few others, case-insensitive. */
698 if (redirection_count && local_file && u->scheme == SCHEME_FTP)
700 if (has_html_suffix_p (local_file))
707 url_free (proxy_url);
711 location_changed = (result == NEWLOCATION);
712 if (location_changed)
714 char *construced_newloc;
715 struct url *newloc_parsed;
717 assert (mynewloc != NULL);
722 /* The HTTP specs only allow absolute URLs to appear in
723 redirects, but a ton of boneheaded webservers and CGIs out
724 there break the rules and use relative URLs, and popular
725 browsers are lenient about this, so wget should be too. */
726 construced_newloc = uri_merge (url, mynewloc);
728 mynewloc = construced_newloc;
730 /* Reset UTF-8 encoding state, keep the URI encoding and reset
731 the content encoding. */
732 iri->utf8_encode = opt.enable_iri;
733 set_content_encoding (iri, NULL);
734 xfree_null (iri->orig_url);
736 /* Now, see if this new location makes sense. */
737 newloc_parsed = url_parse (mynewloc, &up_error_code, iri, true);
740 char *error = url_error (mynewloc, up_error_code);
741 logprintf (LOG_NOTQUIET, "%s: %s.\n", escnonprint_uri (mynewloc),
743 if (orig_parsed != u)
754 /* Now mynewloc will become newloc_parsed->url, because if the
755 Location contained relative paths like .././something, we
756 don't want that propagating as url. */
758 mynewloc = xstrdup (newloc_parsed->url);
760 /* Check for max. number of redirections. */
761 if (++redirection_count > opt.max_redirect)
763 logprintf (LOG_NOTQUIET, _("%d redirections exceeded.\n"),
765 url_free (newloc_parsed);
766 if (orig_parsed != u)
778 if (orig_parsed != u)
784 /* If we're being redirected from POST, we don't want to POST
785 again. Many requests answer POST with a redirection to an
786 index page; that redirection is clearly a GET. We "suspend"
787 POST data for the duration of the redirections, and restore
788 it when we're done. */
789 if (!post_data_suspended)
795 /* Try to not encode in UTF-8 if fetching failed */
796 if (!(*dt & RETROKF) && iri->utf8_encode)
798 iri->utf8_encode = false;
799 if (orig_parsed != u)
803 u = url_parse (origurl, NULL, iri, true);
806 DEBUGP (("[IRI fallbacking to non-utf8 for %s\n", quote (url)));
810 DEBUGP (("[Couldn't fallback to non-utf8 for %s\n", quote (url)));
813 if (local_file && *dt & RETROKF)
815 register_download (u->url, local_file);
816 if (redirection_count && 0 != strcmp (origurl, u->url))
817 register_redirection (origurl, u->url);
819 register_html (u->url, local_file);
822 register_download (u->url, local_file);
823 if (redirection_count && 0 != strcmp (origurl, u->url))
824 register_redirection (origurl, u->url);
826 register_html (u->url, local_file);
828 register_css (u->url, local_file);
833 *file = local_file ? local_file : NULL;
835 xfree_null (local_file);
837 if (orig_parsed != u)
842 if (redirection_count)
861 /* Find the URLs in the file and call retrieve_url() for each of them.
862 If HTML is true, treat the file as HTML, and construct the URLs
865 If opt.recursive is set, call retrieve_tree() for each file. */
868 retrieve_from_file (const char *file, bool html, int *count)
871 struct urlpos *url_list, *cur_url;
872 struct iri *iri = iri_new();
874 char *input_file = NULL;
875 const char *url = file;
877 status = RETROK; /* Suppose everything is OK. */
878 *count = 0; /* Reset the URL count. */
880 /* sXXXav : Assume filename and links in the file are in the locale */
881 set_uri_encoding (iri, opt.locale, true);
882 set_content_encoding (iri, opt.locale);
884 if (url_has_scheme (url))
888 struct url * url_parsed = url_parse(url, &url_err, NULL, true);
892 char *error = url_error (url, url_err);
893 logprintf (LOG_NOTQUIET, "%s: %s.\n", url, error);
899 opt.base_href = xstrdup (url);
901 status = retrieve_url (url_parsed, url, &input_file, NULL, NULL, &dt,
903 if (status != RETROK)
909 /* If we have a found a content encoding, use it */
910 if (iri->content_encoding)
911 set_uri_encoding (iri, iri->content_encoding, false);
914 input_file = (char *) file;
916 url_list = (html ? get_urls_html (input_file, NULL, NULL, iri)
917 : get_urls_file (input_file));
919 for (cur_url = url_list; cur_url; cur_url = cur_url->next, ++*count)
921 char *filename = NULL, *new_file = NULL;
924 if (cur_url->ignore_when_downloading)
927 if (opt.quota && total_downloaded_bytes > opt.quota)
933 /* Reset UTF-8 encode status */
934 iri->utf8_encode = opt.enable_iri;
935 xfree_null (iri->orig_url);
936 iri->orig_url = NULL;
938 if ((opt.recursive || opt.page_requisites)
939 && (cur_url->url->scheme != SCHEME_FTP || getproxy (cur_url->url)))
941 int old_follow_ftp = opt.follow_ftp;
943 /* Turn opt.follow_ftp on in case of recursive FTP retrieval */
944 if (cur_url->url->scheme == SCHEME_FTP)
947 status = retrieve_tree (cur_url->url, iri);
949 opt.follow_ftp = old_follow_ftp;
952 status = retrieve_url (cur_url->url, cur_url->url->url, &filename,
953 &new_file, NULL, &dt, opt.recursive, iri);
955 if (filename && opt.delete_after && file_exists_p (filename))
958 Removing file due to --delete-after in retrieve_from_file():\n"));
959 logprintf (LOG_VERBOSE, _("Removing %s.\n"), filename);
960 if (unlink (filename))
961 logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
965 xfree_null (new_file);
966 xfree_null (filename);
969 /* Free the linked list of URL-s. */
970 free_urlpos (url_list);
977 /* Print `giving up', or `retrying', depending on the impending
978 action. N1 and N2 are the attempt number and the attempt limit. */
980 printwhat (int n1, int n2)
982 logputs (LOG_VERBOSE, (n1 == n2) ? _("Giving up.\n\n") : _("Retrying.\n\n"));
985 /* If opt.wait or opt.waitretry are specified, and if certain
986 conditions are met, sleep the appropriate number of seconds. See
987 the documentation of --wait and --waitretry for more information.
989 COUNT is the count of current retrieval, beginning with 1. */
992 sleep_between_retrievals (int count)
994 static bool first_retrieval = true;
998 /* Don't sleep before the very first retrieval. */
999 first_retrieval = false;
1003 if (opt.waitretry && count > 1)
1005 /* If opt.waitretry is specified and this is a retry, wait for
1006 COUNT-1 number of seconds, or for opt.waitretry seconds. */
1007 if (count <= opt.waitretry)
1010 xsleep (opt.waitretry);
1014 if (!opt.random_wait || count > 1)
1015 /* If random-wait is not specified, or if we are sleeping
1016 between retries of the same download, sleep the fixed
1021 /* Sleep a random amount of time averaging in opt.wait
1022 seconds. The sleeping amount ranges from 0.5*opt.wait to
1024 double waitsecs = (0.5 + random_float ()) * opt.wait;
1025 DEBUGP (("sleep_between_retrievals: avg=%f,sleep=%f\n",
1026 opt.wait, waitsecs));
1032 /* Free the linked list of urlpos. */
1034 free_urlpos (struct urlpos *l)
1038 struct urlpos *next = l->next;
1041 xfree_null (l->local_name);
1047 /* Rotate FNAME opt.backups times */
1049 rotate_backups(const char *fname)
1051 int maxlen = strlen (fname) + 1 + numdigit (opt.backups) + 1;
1052 char *from = (char *)alloca (maxlen);
1053 char *to = (char *)alloca (maxlen);
1057 if (stat (fname, &sb) == 0)
1058 if (S_ISREG (sb.st_mode) == 0)
1061 for (i = opt.backups; i > 1; i--)
1063 sprintf (from, "%s.%d", fname, i - 1);
1064 sprintf (to, "%s.%d", fname, i);
1068 sprintf (to, "%s.%d", fname, 1);
1072 static bool no_proxy_match (const char *, const char **);
1074 /* Return the URL of the proxy appropriate for url U. */
1077 getproxy (struct url *u)
1080 char *rewritten_url;
1081 static char rewritten_storage[1024];
1085 if (no_proxy_match (u->host, (const char **)opt.no_proxy))
1091 proxy = opt.http_proxy ? opt.http_proxy : getenv ("http_proxy");
1095 proxy = opt.https_proxy ? opt.https_proxy : getenv ("https_proxy");
1099 proxy = opt.ftp_proxy ? opt.ftp_proxy : getenv ("ftp_proxy");
1101 case SCHEME_INVALID:
1104 if (!proxy || !*proxy)
1107 /* Handle shorthands. `rewritten_storage' is a kludge to allow
1108 getproxy() to return static storage. */
1109 rewritten_url = rewrite_shorthand_url (proxy);
1112 strncpy (rewritten_storage, rewritten_url, sizeof (rewritten_storage));
1113 rewritten_storage[sizeof (rewritten_storage) - 1] = '\0';
1114 proxy = rewritten_storage;
1120 /* Returns true if URL would be downloaded through a proxy. */
1123 url_uses_proxy (struct url * u)
1128 ret = getproxy (u) != NULL;
1132 /* Should a host be accessed through proxy, concerning no_proxy? */
1134 no_proxy_match (const char *host, const char **no_proxy)
1139 return sufmatch (no_proxy, host);
1142 /* Set the file parameter to point to the local file string. */
1144 set_local_file (const char **file, const char *default_file)
1146 if (opt.output_document)
1148 if (output_stream_regular)
1149 *file = opt.output_document;
1152 *file = default_file;