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
145 write_data (FILE *out, const char *buf, int bufsize, wgint *skip,
164 fwrite (buf, 1, bufsize, out);
167 /* Immediately flush the downloaded data. This should not hinder
168 performance: fast downloads will arrive in large 16K chunks
169 (which stdio would write out immediately anyway), and slow
170 downloads wouldn't be limited by disk speed. */
173 Perhaps it shouldn't hinder performance, but it sure does, at least
174 on VMS (more than 2X). Rather than speculate on what it should or
175 shouldn't do, it might make more sense to test it. Even better, it
176 might be nice to explain what possible benefit it could offer, as
177 it appears to be a clear invitation to poor performance with no
178 actual justification. (Also, why 16K? Anyone test other values?)
182 #endif /* ndef __VMS */
183 return !ferror (out);
186 /* Read the contents of file descriptor FD until it the connection
187 terminates or a read error occurs. The data is read in portions of
188 up to 16K and written to OUT as it arrives. If opt.verbose is set,
189 the progress is shown.
191 TOREAD is the amount of data expected to arrive, normally only used
192 by the progress gauge.
194 STARTPOS is the position from which the download starts, used by
195 the progress gauge. If QTYREAD is non-NULL, the value it points to
196 is incremented by the amount of data read from the network. If
197 QTYWRITTEN is non-NULL, the value it points to is incremented by
198 the amount of data written to disk. The time it took to download
199 the data is stored to ELAPSED.
201 The function exits and returns the amount of data read. In case of
202 error while reading data, -1 is returned. In case of error while
203 writing data, -2 is returned. */
206 fd_read_body (int fd, FILE *out, wgint toread, wgint startpos,
207 wgint *qtyread, wgint *qtywritten, double *elapsed, int flags)
210 #define max(a,b) ((a) > (b) ? (a) : (b))
211 int dlbufsize = max (BUFSIZ, 8 * 1024);
212 char *dlbuf = xmalloc (dlbufsize);
214 struct ptimer *timer = NULL;
215 double last_successful_read_tm = 0;
217 /* The progress gauge, set according to the user preferences. */
218 void *progress = NULL;
220 /* Non-zero if the progress gauge is interactive, i.e. if it can
221 continually update the display. When true, smaller timeout
222 values are used so that the gauge can update the display when
223 data arrives slowly. */
224 bool progress_interactive = false;
226 bool exact = !!(flags & rb_read_exactly);
228 /* Used only by HTTP/HTTPS chunked transfer encoding. */
229 bool chunked = flags & rb_chunked_transfer_encoding;
232 /* How much data we've read/written. */
234 wgint sum_written = 0;
235 wgint remaining_chunk_size = 0;
237 if (flags & rb_skip_startpos)
242 /* If we're skipping STARTPOS bytes, pass 0 as the INITIAL
243 argument to progress_create because the indicator doesn't
244 (yet) know about "skipping" data. */
245 wgint start = skip ? 0 : startpos;
246 progress = progress_create (start, start + toread);
247 progress_interactive = progress_interactive_p (progress);
251 limit_bandwidth_reset ();
253 /* A timer is needed for tracking progress, for throttling, and for
254 tracking elapsed time. If either of these are requested, start
256 if (progress || opt.limit_rate || elapsed)
258 timer = ptimer_new ();
259 last_successful_read_tm = 0;
262 /* Use a smaller buffer for low requested bandwidths. For example,
263 with --limit-rate=2k, it doesn't make sense to slurp in 16K of
264 data and then sleep for 8s. With buffer size equal to the limit,
265 we never have to sleep for more than one second. */
266 if (opt.limit_rate && opt.limit_rate < dlbufsize)
267 dlbufsize = opt.limit_rate;
269 /* Read from FD while there is data to read. Normally toread==0
270 means that it is unknown how much data is to arrive. However, if
271 EXACT is set, then toread==0 means what it says: that no data
273 while (!exact || (sum_read < toread))
276 double tmout = opt.read_timeout;
280 if (remaining_chunk_size == 0)
282 char *line = fd_read_line (fd);
290 remaining_chunk_size = strtol (line, &endl, 16);
291 if (remaining_chunk_size == 0)
294 if (fd_read_line (fd) == NULL)
300 rdsize = MIN (remaining_chunk_size, dlbufsize);
303 rdsize = exact ? MIN (toread - sum_read, dlbufsize) : dlbufsize;
305 if (progress_interactive)
307 /* For interactive progress gauges, always specify a ~1s
308 timeout, so that the gauge can be updated regularly even
309 when the data arrives very slowly or stalls. */
311 if (opt.read_timeout)
314 waittm = ptimer_read (timer) - last_successful_read_tm;
315 if (waittm + tmout > opt.read_timeout)
317 /* Don't let total idle time exceed read timeout. */
318 tmout = opt.read_timeout - waittm;
321 /* We've already exceeded the timeout. */
322 ret = -1, errno = ETIMEDOUT;
328 ret = fd_read (fd, dlbuf, rdsize, tmout);
330 if (progress_interactive && ret < 0 && errno == ETIMEDOUT)
331 ret = 0; /* interactive timeout, handled above */
333 break; /* EOF or read error */
335 if (progress || opt.limit_rate || elapsed)
337 ptimer_measure (timer);
339 last_successful_read_tm = ptimer_read (timer);
345 if (!write_data (out, dlbuf, ret, &skip, &sum_written))
352 remaining_chunk_size -= ret;
353 if (remaining_chunk_size == 0)
354 if (fd_read_line (fd) == NULL)
363 limit_bandwidth (ret, timer);
366 progress_update (progress, ret, ptimer_read (timer));
368 if (toread > 0 && !opt.quiet)
369 ws_percenttitle (100.0 *
370 (startpos + sum_read) / (startpos + toread));
378 progress_finish (progress, ptimer_read (timer));
381 *elapsed = ptimer_read (timer);
383 ptimer_destroy (timer);
386 *qtyread += sum_read;
388 *qtywritten += sum_written;
395 /* Read a hunk of data from FD, up until a terminator. The hunk is
396 limited by whatever the TERMINATOR callback chooses as its
397 terminator. For example, if terminator stops at newline, the hunk
398 will consist of a line of data; if terminator stops at two
399 newlines, it can be used to read the head of an HTTP response.
400 Upon determining the boundary, the function returns the data (up to
401 the terminator) in malloc-allocated storage.
403 In case of read error, NULL is returned. In case of EOF and no
404 data read, NULL is returned and errno set to 0. In case of having
405 read some data, but encountering EOF before seeing the terminator,
406 the data that has been read is returned, but it will (obviously)
407 not contain the terminator.
409 The TERMINATOR function is called with three arguments: the
410 beginning of the data read so far, the beginning of the current
411 block of peeked-at data, and the length of the current block.
412 Depending on its needs, the function is free to choose whether to
413 analyze all data or just the newly arrived data. If TERMINATOR
414 returns NULL, it means that the terminator has not been seen.
415 Otherwise it should return a pointer to the charactre immediately
416 following the terminator.
418 The idea is to be able to read a line of input, or otherwise a hunk
419 of text, such as the head of an HTTP request, without crossing the
420 boundary, so that the next call to fd_read etc. reads the data
421 after the hunk. To achieve that, this function does the following:
423 1. Peek at incoming data.
425 2. Determine whether the peeked data, along with the previously
426 read data, includes the terminator.
428 2a. If yes, read the data until the end of the terminator, and
431 2b. If no, read the peeked data and goto 1.
433 The function is careful to assume as little as possible about the
434 implementation of peeking. For example, every peek is followed by
435 a read. If the read returns a different amount of data, the
436 process is retried until all data arrives safely.
438 SIZEHINT is the buffer size sufficient to hold all the data in the
439 typical case (it is used as the initial buffer size). MAXSIZE is
440 the maximum amount of memory this function is allowed to allocate,
441 or 0 if no upper limit is to be enforced.
443 This function should be used as a building block for other
444 functions -- see fd_read_line as a simple example. */
447 fd_read_hunk (int fd, hunk_terminator_t terminator, long sizehint, long maxsize)
449 long bufsize = sizehint;
450 char *hunk = xmalloc (bufsize);
451 int tail = 0; /* tail position in HUNK */
453 assert (!maxsize || maxsize >= bufsize);
458 int pklen, rdlen, remain;
460 /* First, peek at the available data. */
462 pklen = fd_peek (fd, hunk + tail, bufsize - 1 - tail, -1);
468 end = terminator (hunk, hunk + tail, pklen);
471 /* The data contains the terminator: we'll drain the data up
472 to the end of the terminator. */
473 remain = end - (hunk + tail);
474 assert (remain >= 0);
477 /* No more data needs to be read. */
481 if (bufsize - 1 < tail + remain)
483 bufsize = tail + remain + 1;
484 hunk = xrealloc (hunk, bufsize);
488 /* No terminator: simply read the data we know is (or should
492 /* Now, read the data. Note that we make no assumptions about
493 how much data we'll get. (Some TCP stacks are notorious for
494 read returning less data than the previous MSG_PEEK.) */
496 rdlen = fd_read (fd, hunk + tail, remain, 0);
509 /* EOF without anything having been read */
515 /* EOF seen: return the data we've read. */
518 if (end && rdlen == remain)
519 /* The terminator was seen and the remaining data drained --
520 we got what we came for. */
523 /* Keep looping until all the data arrives. */
525 if (tail == bufsize - 1)
527 /* Double the buffer size, but refuse to allocate more than
529 if (maxsize && bufsize >= maxsize)
536 if (maxsize && bufsize > maxsize)
538 hunk = xrealloc (hunk, bufsize);
544 line_terminator (const char *start, const char *peeked, int peeklen)
546 const char *p = memchr (peeked, '\n', peeklen);
548 /* p+1 because the line must include '\n' */
553 /* The maximum size of the single line we agree to accept. This is
554 not meant to impose an arbitrary limit, but to protect the user
555 from Wget slurping up available memory upon encountering malicious
556 or buggy server output. Define it to 0 to remove the limit. */
557 #define FD_READ_LINE_MAX 4096
559 /* Read one line from FD and return it. The line is allocated using
560 malloc, but is never larger than FD_READ_LINE_MAX.
562 If an error occurs, or if no data can be read, NULL is returned.
563 In the former case errno indicates the error condition, and in the
564 latter case, errno is NULL. */
567 fd_read_line (int fd)
569 return fd_read_hunk (fd, line_terminator, 128, FD_READ_LINE_MAX);
572 /* Return a printed representation of the download rate, along with
573 the units appropriate for the download speed. */
576 retr_rate (wgint bytes, double secs)
579 static const char *rate_names[] = {"B/s", "KB/s", "MB/s", "GB/s" };
582 double dlrate = calc_rate (bytes, secs, &units);
583 /* Use more digits for smaller numbers (regardless of unit used),
584 e.g. "1022", "247", "12.5", "2.38". */
585 sprintf (res, "%.*f %s",
586 dlrate >= 99.95 ? 0 : dlrate >= 9.995 ? 1 : 2,
587 dlrate, rate_names[units]);
592 /* Calculate the download rate and trim it as appropriate for the
593 speed. Appropriate means that if rate is greater than 1K/s,
594 kilobytes are used, and if rate is greater than 1MB/s, megabytes
597 UNITS is zero for B/s, one for KB/s, two for MB/s, and three for
601 calc_rate (wgint bytes, double secs, int *units)
609 /* If elapsed time is exactly zero, it means we're under the
610 resolution of the timer. This can easily happen on systems
611 that use time() for the timer. Since the interval lies between
612 0 and the timer's resolution, assume half the resolution. */
613 secs = ptimer_resolution () / 2.0;
615 dlrate = bytes / secs;
618 else if (dlrate < 1024.0 * 1024.0)
619 *units = 1, dlrate /= 1024.0;
620 else if (dlrate < 1024.0 * 1024.0 * 1024.0)
621 *units = 2, dlrate /= (1024.0 * 1024.0);
623 /* Maybe someone will need this, one day. */
624 *units = 3, dlrate /= (1024.0 * 1024.0 * 1024.0);
630 #define SUSPEND_POST_DATA do { \
631 post_data_suspended = true; \
632 saved_post_data = opt.post_data; \
633 saved_post_file_name = opt.post_file_name; \
634 opt.post_data = NULL; \
635 opt.post_file_name = NULL; \
638 #define RESTORE_POST_DATA do { \
639 if (post_data_suspended) \
641 opt.post_data = saved_post_data; \
642 opt.post_file_name = saved_post_file_name; \
643 post_data_suspended = false; \
647 static char *getproxy (struct url *);
649 /* Retrieve the given URL. Decides which loop to call -- HTTP, FTP,
652 /* #### This function should be rewritten so it doesn't return from
656 retrieve_url (struct url * orig_parsed, const char *origurl, char **file,
657 char **newloc, const char *refurl, int *dt, bool recursive,
658 struct iri *iri, bool register_status)
662 bool location_changed;
663 bool iri_fallbacked = 0;
665 char *mynewloc, *proxy;
666 struct url *u = orig_parsed, *proxy_url;
667 int up_error_code; /* url parse error code */
669 int redirection_count = 0;
671 bool post_data_suspended = false;
672 char *saved_post_data = NULL;
673 char *saved_post_file_name = NULL;
675 /* If dt is NULL, use local storage. */
681 url = xstrdup (origurl);
688 refurl = opt.referer;
691 /* (also for IRI fallbacking) */
698 proxy = getproxy (u);
701 struct iri *pi = iri_new ();
702 set_uri_encoding (pi, opt.locale, true);
703 pi->utf8_encode = false;
705 /* Parse the proxy URL. */
706 proxy_url = url_parse (proxy, &up_error_code, NULL, true);
709 char *error = url_error (proxy, up_error_code);
710 logprintf (LOG_NOTQUIET, _("Error parsing proxy URL %s: %s.\n"),
718 if (proxy_url->scheme != SCHEME_HTTP && proxy_url->scheme != u->scheme)
720 logprintf (LOG_NOTQUIET, _("Error in proxy URL %s: Must be HTTP.\n"), proxy);
721 url_free (proxy_url);
729 if (u->scheme == SCHEME_HTTP
731 || u->scheme == SCHEME_HTTPS
733 || (proxy_url && proxy_url->scheme == SCHEME_HTTP))
735 result = http_loop (u, orig_parsed, &mynewloc, &local_file, refurl, dt,
738 else if (u->scheme == SCHEME_FTP)
740 /* If this is a redirection, temporarily turn off opt.ftp_glob
741 and opt.recursive, both being undesirable when following
743 bool oldrec = recursive, glob = opt.ftp_glob;
744 if (redirection_count)
745 oldrec = glob = false;
747 result = ftp_loop (u, &local_file, dt, proxy_url, recursive, glob);
750 /* There is a possibility of having HTTP being redirected to
751 FTP. In these cases we must decide whether the text is HTML
752 according to the suffix. The HTML suffixes are `.html',
753 `.htm' and a few others, case-insensitive. */
754 if (redirection_count && local_file && u->scheme == SCHEME_FTP)
756 if (has_html_suffix_p (local_file))
763 url_free (proxy_url);
767 location_changed = (result == NEWLOCATION || result == NEWLOCATION_KEEP_POST);
768 if (location_changed)
770 char *construced_newloc;
771 struct url *newloc_parsed;
773 assert (mynewloc != NULL);
778 /* The HTTP specs only allow absolute URLs to appear in
779 redirects, but a ton of boneheaded webservers and CGIs out
780 there break the rules and use relative URLs, and popular
781 browsers are lenient about this, so wget should be too. */
782 construced_newloc = uri_merge (url, mynewloc);
784 mynewloc = construced_newloc;
786 /* Reset UTF-8 encoding state, keep the URI encoding and reset
787 the content encoding. */
788 iri->utf8_encode = opt.enable_iri;
789 set_content_encoding (iri, NULL);
790 xfree_null (iri->orig_url);
792 /* Now, see if this new location makes sense. */
793 newloc_parsed = url_parse (mynewloc, &up_error_code, iri, true);
796 char *error = url_error (mynewloc, up_error_code);
797 logprintf (LOG_NOTQUIET, "%s: %s.\n", escnonprint_uri (mynewloc),
799 if (orig_parsed != u)
810 /* Now mynewloc will become newloc_parsed->url, because if the
811 Location contained relative paths like .././something, we
812 don't want that propagating as url. */
814 mynewloc = xstrdup (newloc_parsed->url);
816 /* Check for max. number of redirections. */
817 if (++redirection_count > opt.max_redirect)
819 logprintf (LOG_NOTQUIET, _("%d redirections exceeded.\n"),
821 url_free (newloc_parsed);
822 if (orig_parsed != u)
835 if (orig_parsed != u)
841 /* If we're being redirected from POST, and we received a
842 redirect code different than 307, we don't want to POST
843 again. Many requests answer POST with a redirection to an
844 index page; that redirection is clearly a GET. We "suspend"
845 POST data for the duration of the redirections, and restore
848 RFC2616 HTTP/1.1 introduces code 307 Temporary Redirect
849 specifically to preserve the method of the request.
851 if (result != NEWLOCATION_KEEP_POST && !post_data_suspended)
857 /* Try to not encode in UTF-8 if fetching failed */
858 if (!(*dt & RETROKF) && iri->utf8_encode)
860 iri->utf8_encode = false;
861 if (orig_parsed != u)
865 u = url_parse (origurl, NULL, iri, true);
868 DEBUGP (("[IRI fallbacking to non-utf8 for %s\n", quote (url)));
869 url = xstrdup (u->url);
874 DEBUGP (("[Couldn't fallback to non-utf8 for %s\n", quote (url)));
877 if (local_file && u && *dt & RETROKF)
879 register_download (u->url, local_file);
881 if (!opt.spider && redirection_count && 0 != strcmp (origurl, u->url))
882 register_redirection (origurl, u->url);
885 register_html (u->url, local_file);
888 register_css (u->url, local_file);
892 *file = local_file ? local_file : NULL;
894 xfree_null (local_file);
896 if (orig_parsed != u)
901 if (redirection_count || iri_fallbacked)
919 inform_exit_status (result);
923 /* Find the URLs in the file and call retrieve_url() for each of them.
924 If HTML is true, treat the file as HTML, and construct the URLs
927 If opt.recursive is set, call retrieve_tree() for each file. */
930 retrieve_from_file (const char *file, bool html, int *count)
933 struct urlpos *url_list, *cur_url;
934 struct iri *iri = iri_new();
936 char *input_file, *url_file = NULL;
937 const char *url = file;
939 status = RETROK; /* Suppose everything is OK. */
940 *count = 0; /* Reset the URL count. */
942 /* sXXXav : Assume filename and links in the file are in the locale */
943 set_uri_encoding (iri, opt.locale, true);
944 set_content_encoding (iri, opt.locale);
946 if (url_valid_scheme (url))
950 struct url *url_parsed = url_parse (url, &url_err, iri, true);
953 char *error = url_error (url, url_err);
954 logprintf (LOG_NOTQUIET, "%s: %s.\n", url, error);
960 opt.base_href = xstrdup (url);
962 status = retrieve_url (url_parsed, url, &url_file, NULL, NULL, &dt,
964 url_free (url_parsed);
966 if (!url_file || (status != RETROK))
972 /* If we have a found a content encoding, use it.
973 * ( == is okay, because we're checking for identical object) */
974 if (iri->content_encoding != opt.locale)
975 set_uri_encoding (iri, iri->content_encoding, false);
977 /* Reset UTF-8 encode status */
978 iri->utf8_encode = opt.enable_iri;
979 xfree_null (iri->orig_url);
980 iri->orig_url = NULL;
982 input_file = url_file;
985 input_file = (char *) file;
987 url_list = (html ? get_urls_html (input_file, NULL, NULL, iri)
988 : get_urls_file (input_file));
990 xfree_null (url_file);
992 for (cur_url = url_list; cur_url; cur_url = cur_url->next, ++*count)
994 char *filename = NULL, *new_file = NULL;
996 struct iri *tmpiri = iri_dup (iri);
997 struct url *parsed_url = NULL;
999 if (cur_url->ignore_when_downloading)
1002 if (opt.quota && total_downloaded_bytes > opt.quota)
1008 parsed_url = url_parse (cur_url->url->url, NULL, tmpiri, true);
1010 if ((opt.recursive || opt.page_requisites)
1011 && (cur_url->url->scheme != SCHEME_FTP || getproxy (cur_url->url)))
1013 int old_follow_ftp = opt.follow_ftp;
1015 /* Turn opt.follow_ftp on in case of recursive FTP retrieval */
1016 if (cur_url->url->scheme == SCHEME_FTP)
1019 status = retrieve_tree (parsed_url ? parsed_url : cur_url->url,
1022 opt.follow_ftp = old_follow_ftp;
1025 status = retrieve_url (parsed_url ? parsed_url : cur_url->url,
1026 cur_url->url->url, &filename,
1027 &new_file, NULL, &dt, opt.recursive, tmpiri,
1031 url_free (parsed_url);
1033 if (filename && opt.delete_after && file_exists_p (filename))
1036 Removing file due to --delete-after in retrieve_from_file():\n"));
1037 logprintf (LOG_VERBOSE, _("Removing %s.\n"), filename);
1038 if (unlink (filename))
1039 logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
1043 xfree_null (new_file);
1044 xfree_null (filename);
1048 /* Free the linked list of URL-s. */
1049 free_urlpos (url_list);
1056 /* Print `giving up', or `retrying', depending on the impending
1057 action. N1 and N2 are the attempt number and the attempt limit. */
1059 printwhat (int n1, int n2)
1061 logputs (LOG_VERBOSE, (n1 == n2) ? _("Giving up.\n\n") : _("Retrying.\n\n"));
1064 /* If opt.wait or opt.waitretry are specified, and if certain
1065 conditions are met, sleep the appropriate number of seconds. See
1066 the documentation of --wait and --waitretry for more information.
1068 COUNT is the count of current retrieval, beginning with 1. */
1071 sleep_between_retrievals (int count)
1073 static bool first_retrieval = true;
1075 if (first_retrieval)
1077 /* Don't sleep before the very first retrieval. */
1078 first_retrieval = false;
1082 if (opt.waitretry && count > 1)
1084 /* If opt.waitretry is specified and this is a retry, wait for
1085 COUNT-1 number of seconds, or for opt.waitretry seconds. */
1086 if (count <= opt.waitretry)
1089 xsleep (opt.waitretry);
1093 if (!opt.random_wait || count > 1)
1094 /* If random-wait is not specified, or if we are sleeping
1095 between retries of the same download, sleep the fixed
1100 /* Sleep a random amount of time averaging in opt.wait
1101 seconds. The sleeping amount ranges from 0.5*opt.wait to
1103 double waitsecs = (0.5 + random_float ()) * opt.wait;
1104 DEBUGP (("sleep_between_retrievals: avg=%f,sleep=%f\n",
1105 opt.wait, waitsecs));
1111 /* Free the linked list of urlpos. */
1113 free_urlpos (struct urlpos *l)
1117 struct urlpos *next = l->next;
1120 xfree_null (l->local_name);
1126 /* Rotate FNAME opt.backups times */
1128 rotate_backups(const char *fname)
1130 int maxlen = strlen (fname) + 1 + numdigit (opt.backups) + 1;
1131 char *from = (char *)alloca (maxlen);
1132 char *to = (char *)alloca (maxlen);
1136 if (stat (fname, &sb) == 0)
1137 if (S_ISREG (sb.st_mode) == 0)
1140 for (i = opt.backups; i > 1; i--)
1142 sprintf (from, "%s.%d", fname, i - 1);
1143 sprintf (to, "%s.%d", fname, i);
1147 sprintf (to, "%s.%d", fname, 1);
1151 static bool no_proxy_match (const char *, const char **);
1153 /* Return the URL of the proxy appropriate for url U. */
1156 getproxy (struct url *u)
1159 char *rewritten_url;
1160 static char rewritten_storage[1024];
1164 if (no_proxy_match (u->host, (const char **)opt.no_proxy))
1170 proxy = opt.http_proxy ? opt.http_proxy : getenv ("http_proxy");
1174 proxy = opt.https_proxy ? opt.https_proxy : getenv ("https_proxy");
1178 proxy = opt.ftp_proxy ? opt.ftp_proxy : getenv ("ftp_proxy");
1180 case SCHEME_INVALID:
1183 if (!proxy || !*proxy)
1186 /* Handle shorthands. `rewritten_storage' is a kludge to allow
1187 getproxy() to return static storage. */
1188 rewritten_url = rewrite_shorthand_url (proxy);
1191 strncpy (rewritten_storage, rewritten_url, sizeof (rewritten_storage));
1192 rewritten_storage[sizeof (rewritten_storage) - 1] = '\0';
1193 proxy = rewritten_storage;
1199 /* Returns true if URL would be downloaded through a proxy. */
1202 url_uses_proxy (struct url * u)
1207 ret = getproxy (u) != NULL;
1211 /* Should a host be accessed through proxy, concerning no_proxy? */
1213 no_proxy_match (const char *host, const char **no_proxy)
1218 return sufmatch (no_proxy, host);
1221 /* Set the file parameter to point to the local file string. */
1223 set_local_file (const char **file, const char *default_file)
1225 if (opt.output_document)
1227 if (output_stream_regular)
1228 *file = opt.output_document;
1231 *file = default_file;
1234 /* Return true for an input file's own URL, false otherwise. */
1236 input_file_url (const char *input_file)
1238 static bool first = true;
1241 && url_has_scheme (input_file)