1 /* File retrieval.
2 Copyright (C) 1996-2011, 2014-2015, 2018-2021 Free Software
3 Foundation, Inc.
4
5 This file is part of GNU Wget.
6
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.
11
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.
16
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/>.
19
20 Additional permission under GNU GPL version 3 section 7
21
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. */
30
31 #include "wget.h"
32
33 #include <stdio.h>
34 #include <stdlib.h>
35 #include <unistd.h>
36 #include <errno.h>
37 #include <string.h>
38 #include <assert.h>
39 #ifdef VMS
40 # include <unixio.h> /* For delete(). */
41 #endif
42
43 #ifdef HAVE_LIBZ
44 # include <zlib.h>
45 #endif
46
47 #include "exits.h"
48 #include "utils.h"
49 #include "retr.h"
50 #include "progress.h"
51 #include "url.h"
52 #include "recur.h"
53 #include "ftp.h"
54 #include "http.h"
55 #include "host.h"
56 #include "connect.h"
57 #include "hash.h"
58 #include "convert.h"
59 #include "ptimer.h"
60 #include "html-url.h"
61 #include "iri.h"
62 #include "hsts.h"
63
64 /* Total size of downloaded files. Used to enforce quota. */
65 wgint total_downloaded_bytes;
66
67 /* Total download time in seconds. */
68 double total_download_time;
69
70 /* If non-NULL, the stream to which output should be written. This
71 stream is initialized when `-O' is used. */
72 FILE *output_stream;
73
74 /* Whether output_document is a regular file we can manipulate,
75 i.e. not `-' or a device file. */
76 bool output_stream_regular;
77
78 static struct {
79 wgint chunk_bytes;
80 double chunk_start;
81 double sleep_adjust;
82 } limit_data;
83
84 static void
limit_bandwidth_reset(void)85 limit_bandwidth_reset (void)
86 {
87 xzero (limit_data);
88 }
89
90 #ifdef HAVE_LIBZ
91 static voidpf
zalloc(voidpf opaque,unsigned int items,unsigned int size)92 zalloc (voidpf opaque, unsigned int items, unsigned int size)
93 {
94 (void) opaque;
95 return (voidpf) xcalloc (items, size);
96 }
97
98 static void
zfree(voidpf opaque,voidpf address)99 zfree (voidpf opaque, voidpf address)
100 {
101 (void) opaque;
102 xfree (address);
103 }
104 #endif
105
106 /* Limit the bandwidth by pausing the download for an amount of time.
107 BYTES is the number of bytes received from the network, and TIMER
108 is the timer that started at the beginning of download. */
109
110 static void
limit_bandwidth(wgint bytes,struct ptimer * timer)111 limit_bandwidth (wgint bytes, struct ptimer *timer)
112 {
113 double delta_t = ptimer_read (timer) - limit_data.chunk_start;
114 double expected;
115
116 limit_data.chunk_bytes += bytes;
117
118 /* Calculate the amount of time we expect downloading the chunk
119 should take. If in reality it took less time, sleep to
120 compensate for the difference. */
121 expected = (double) limit_data.chunk_bytes / opt.limit_rate;
122
123 if (expected > delta_t)
124 {
125 double slp = expected - delta_t + limit_data.sleep_adjust;
126 double t0, t1;
127 if (slp < 0.2)
128 {
129 DEBUGP (("deferring a %.2f ms sleep (%s/%.2f).\n",
130 slp * 1000, number_to_static_string (limit_data.chunk_bytes),
131 delta_t));
132 return;
133 }
134 DEBUGP (("\nsleeping %.2f ms for %s bytes, adjust %.2f ms\n",
135 slp * 1000, number_to_static_string (limit_data.chunk_bytes),
136 limit_data.sleep_adjust));
137
138 t0 = ptimer_read (timer);
139 xsleep (slp);
140 t1 = ptimer_measure (timer);
141
142 /* Due to scheduling, we probably slept slightly longer (or
143 shorter) than desired. Calculate the difference between the
144 desired and the actual sleep, and adjust the next sleep by
145 that amount. */
146 limit_data.sleep_adjust = slp - (t1 - t0);
147 /* If sleep_adjust is very large, it's likely due to suspension
148 and not clock inaccuracy. Don't enforce those. */
149 if (limit_data.sleep_adjust > 0.5)
150 limit_data.sleep_adjust = 0.5;
151 else if (limit_data.sleep_adjust < -0.5)
152 limit_data.sleep_adjust = -0.5;
153 }
154
155 limit_data.chunk_bytes = 0;
156 limit_data.chunk_start = ptimer_read (timer);
157 }
158
159 /* Write data in BUF to OUT. However, if *SKIP is non-zero, skip that
160 amount of data and decrease SKIP. Increment *TOTAL by the amount
161 of data written. If OUT2 is not NULL, also write BUF to OUT2.
162 In case of error writing to OUT, -2 is returned. In case of error
163 writing to OUT2, -3 is returned. Return 1 if the whole BUF was
164 skipped. */
165
166 static int
write_data(FILE * out,FILE * out2,const char * buf,int bufsize,wgint * skip,wgint * written)167 write_data (FILE *out, FILE *out2, const char *buf, int bufsize,
168 wgint *skip, wgint *written)
169 {
170 if (out == NULL && out2 == NULL)
171 return 1;
172
173 if (skip)
174 {
175 if (*skip > bufsize)
176 {
177 *skip -= bufsize;
178 return 1;
179 }
180 if (*skip)
181 {
182 buf += *skip;
183 bufsize -= *skip;
184 *skip = 0;
185 if (bufsize == 0)
186 return 1;
187 }
188 }
189
190 if (out)
191 fwrite (buf, 1, bufsize, out);
192 if (out2)
193 fwrite (buf, 1, bufsize, out2);
194
195 if (written)
196 *written += bufsize;
197
198 /* Immediately flush the downloaded data. This should not hinder
199 performance: fast downloads will arrive in large 16K chunks
200 (which stdio would write out immediately anyway), and slow
201 downloads wouldn't be limited by disk speed. */
202
203 /* 2005-04-20 SMS.
204 Perhaps it shouldn't hinder performance, but it sure does, at least
205 on VMS (more than 2X). Rather than speculate on what it should or
206 shouldn't do, it might make more sense to test it. Even better, it
207 might be nice to explain what possible benefit it could offer, as
208 it appears to be a clear invitation to poor performance with no
209 actual justification. (Also, why 16K? Anyone test other values?)
210 */
211 #ifndef __VMS
212 if (out)
213 fflush (out);
214 if (out2)
215 fflush (out2);
216 #endif /* ndef __VMS */
217
218 if (out && ferror (out))
219 return -2;
220 else if (out2 && ferror (out2))
221 return -3;
222
223 return 0;
224 }
225
226 /* Read the contents of file descriptor FD until it the connection
227 terminates or a read error occurs. The data is read in portions of
228 up to 16K and written to OUT as it arrives. If opt.verbose is set,
229 the progress is shown.
230
231 TOREAD is the amount of data expected to arrive, normally only used
232 by the progress gauge.
233
234 STARTPOS is the position from which the download starts, used by
235 the progress gauge. If QTYREAD is non-NULL, the value it points to
236 is incremented by the amount of data read from the network. If
237 QTYWRITTEN is non-NULL, the value it points to is incremented by
238 the amount of data written to disk. The time it took to download
239 the data is stored to ELAPSED.
240
241 If OUT2 is non-NULL, the contents is also written to OUT2.
242 OUT2 will get an exact copy of the response: if this is a chunked
243 response, everything -- including the chunk headers -- is written
244 to OUT2. (OUT will only get the unchunked response.)
245
246 The function exits and returns the amount of data read. In case of
247 error while reading data, -1 is returned. In case of error while
248 writing data to OUT, -2 is returned. In case of error while writing
249 data to OUT2, -3 is returned. */
250
251 int
fd_read_body(const char * downloaded_filename,int fd,FILE * out,wgint toread,wgint startpos,wgint * qtyread,wgint * qtywritten,double * elapsed,int flags,FILE * out2)252 fd_read_body (const char *downloaded_filename, int fd, FILE *out, wgint toread, wgint startpos,
253
254 wgint *qtyread, wgint *qtywritten, double *elapsed, int flags,
255 FILE *out2)
256 {
257 int ret = 0;
258 #undef max
259 #define max(a,b) ((a) > (b) ? (a) : (b))
260 int dlbufsize = max (BUFSIZ, 8 * 1024);
261 char *dlbuf = xmalloc (dlbufsize);
262
263 struct ptimer *timer = NULL;
264 double last_successful_read_tm = 0;
265
266 /* The progress gauge, set according to the user preferences. */
267 void *progress = NULL;
268
269 /* Non-zero if the progress gauge is interactive, i.e. if it can
270 continually update the display. When true, smaller timeout
271 values are used so that the gauge can update the display when
272 data arrives slowly. */
273 bool progress_interactive = false;
274
275 bool exact = !!(flags & rb_read_exactly);
276
277 /* Used only by HTTP/HTTPS chunked transfer encoding. */
278 bool chunked = flags & rb_chunked_transfer_encoding;
279 wgint skip = 0;
280
281 /* How much data we've read/written. */
282 wgint sum_read = 0;
283 wgint sum_written = 0;
284 wgint remaining_chunk_size = 0;
285
286 #ifdef HAVE_LIBZ
287 /* try to minimize the number of calls to inflate() and write_data() per
288 call to fd_read() */
289 unsigned int gzbufsize = dlbufsize * 4;
290 char *gzbuf = NULL;
291 z_stream gzstream;
292
293 if (flags & rb_compressed_gzip)
294 {
295 gzbuf = xmalloc (gzbufsize);
296 if (gzbuf != NULL)
297 {
298 gzstream.zalloc = zalloc;
299 gzstream.zfree = zfree;
300 gzstream.opaque = Z_NULL;
301 gzstream.next_in = Z_NULL;
302 gzstream.avail_in = 0;
303
304 #define GZIP_DETECT 32 /* gzip format detection */
305 #define GZIP_WINDOW 15 /* logarithmic window size (default: 15) */
306 ret = inflateInit2 (&gzstream, GZIP_DETECT | GZIP_WINDOW);
307 if (ret != Z_OK)
308 {
309 xfree (gzbuf);
310 errno = (ret == Z_MEM_ERROR) ? ENOMEM : EINVAL;
311 ret = -1;
312 goto out;
313 }
314 }
315 else
316 {
317 errno = ENOMEM;
318 ret = -1;
319 goto out;
320 }
321 }
322 #endif
323
324 if (flags & rb_skip_startpos)
325 skip = startpos;
326
327 if (opt.show_progress)
328 {
329 const char *filename_progress;
330 /* If we're skipping STARTPOS bytes, pass 0 as the INITIAL
331 argument to progress_create because the indicator doesn't
332 (yet) know about "skipping" data. */
333 wgint start = skip ? 0 : startpos;
334 if (opt.dir_prefix)
335 filename_progress = downloaded_filename + strlen (opt.dir_prefix) + 1;
336 else
337 filename_progress = downloaded_filename;
338 progress = progress_create (filename_progress, start, start + toread);
339 progress_interactive = progress_interactive_p (progress);
340 }
341
342 if (opt.limit_rate)
343 limit_bandwidth_reset ();
344
345 /* A timer is needed for tracking progress, for throttling, and for
346 tracking elapsed time. If either of these are requested, start
347 the timer. */
348 if (progress || opt.limit_rate || elapsed)
349 {
350 timer = ptimer_new ();
351 last_successful_read_tm = 0;
352 }
353
354 /* Use a smaller buffer for low requested bandwidths. For example,
355 with --limit-rate=2k, it doesn't make sense to slurp in 16K of
356 data and then sleep for 8s. With buffer size equal to the limit,
357 we never have to sleep for more than one second. */
358 if (opt.limit_rate && opt.limit_rate < dlbufsize)
359 dlbufsize = opt.limit_rate;
360
361 /* Read from FD while there is data to read. Normally toread==0
362 means that it is unknown how much data is to arrive. However, if
363 EXACT is set, then toread==0 means what it says: that no data
364 should be read. */
365 while (!exact || (sum_read < toread))
366 {
367 int rdsize;
368 double tmout = opt.read_timeout;
369
370 if (chunked)
371 {
372 if (remaining_chunk_size == 0)
373 {
374 char *line = fd_read_line (fd);
375 char *endl;
376 if (line == NULL)
377 {
378 ret = -1;
379 break;
380 }
381 else if (out2 != NULL)
382 fwrite (line, 1, strlen (line), out2);
383
384 remaining_chunk_size = strtol (line, &endl, 16);
385 xfree (line);
386
387 if (remaining_chunk_size < 0)
388 {
389 ret = -1;
390 break;
391 }
392
393 if (remaining_chunk_size == 0)
394 {
395 ret = 0;
396 line = fd_read_line (fd);
397 if (line == NULL)
398 ret = -1;
399 else
400 {
401 if (out2 != NULL)
402 fwrite (line, 1, strlen (line), out2);
403 xfree (line);
404 }
405 break;
406 }
407 }
408
409 rdsize = MIN (remaining_chunk_size, dlbufsize);
410 }
411 else
412 rdsize = exact ? MIN (toread - sum_read, dlbufsize) : dlbufsize;
413
414 if (progress_interactive)
415 {
416 /* For interactive progress gauges, always specify a ~1s
417 timeout, so that the gauge can be updated regularly even
418 when the data arrives very slowly or stalls. */
419 tmout = 0.95;
420 /* avoid wrong 'interactive timeout' */
421 errno = 0;
422 if (opt.read_timeout)
423 {
424 double waittm;
425 waittm = ptimer_read (timer) - last_successful_read_tm;
426 if (waittm + tmout > opt.read_timeout)
427 {
428 /* Don't let total idle time exceed read timeout. */
429 tmout = opt.read_timeout - waittm;
430 /* if 0 fd_read can be 'blocked read' */
431 if (tmout <= 0)
432 {
433 /* We've already exceeded the timeout. */
434 ret = -1, errno = ETIMEDOUT;
435 break;
436 }
437 }
438 }
439 }
440 ret = fd_read (fd, dlbuf, rdsize, tmout);
441
442 if (progress_interactive && ret < 0 && errno == ETIMEDOUT)
443 ret = 0; /* interactive timeout, handled above */
444 else if (ret <= 0)
445 break; /* EOF or read error */
446
447 if (progress || opt.limit_rate || elapsed)
448 {
449 ptimer_measure (timer);
450 if (ret > 0)
451 last_successful_read_tm = ptimer_read (timer);
452 }
453
454 if (ret > 0)
455 {
456 int write_res;
457
458 sum_read += ret;
459
460 #ifdef HAVE_LIBZ
461 if (gzbuf != NULL)
462 {
463 int err;
464 int towrite;
465
466 /* Write original data to WARC file */
467 write_res = write_data (NULL, out2, dlbuf, ret, NULL, NULL);
468 if (write_res < 0)
469 {
470 ret = write_res;
471 goto out;
472 }
473
474 gzstream.avail_in = ret;
475 gzstream.next_in = (unsigned char *) dlbuf;
476
477 do
478 {
479 gzstream.avail_out = gzbufsize;
480 gzstream.next_out = (unsigned char *) gzbuf;
481
482 err = inflate (&gzstream, Z_NO_FLUSH);
483
484 switch (err)
485 {
486 case Z_MEM_ERROR:
487 errno = ENOMEM;
488 ret = -1;
489 goto out;
490 case Z_NEED_DICT:
491 case Z_DATA_ERROR:
492 errno = EINVAL;
493 ret = -1;
494 goto out;
495 case Z_STREAM_END:
496 if (exact && sum_read != toread)
497 {
498 DEBUGP(("zlib stream ended unexpectedly after %"PRId64"/%"PRId64
499 " bytes\n", sum_read, toread));
500 }
501 }
502
503 towrite = gzbufsize - gzstream.avail_out;
504 write_res = write_data (out, NULL, gzbuf, towrite, &skip,
505 &sum_written);
506 if (write_res < 0)
507 {
508 ret = write_res;
509 goto out;
510 }
511 }
512 while (gzstream.avail_out == 0);
513 }
514 else
515 #endif
516 {
517 write_res = write_data (out, out2, dlbuf, ret, &skip,
518 &sum_written);
519 if (write_res < 0)
520 {
521 ret = write_res;
522 goto out;
523 }
524 }
525
526 if (chunked)
527 {
528 remaining_chunk_size -= ret;
529 if (remaining_chunk_size == 0)
530 {
531 char *line = fd_read_line (fd);
532 if (line == NULL)
533 {
534 ret = -1;
535 break;
536 }
537 else
538 {
539 if (out2 != NULL)
540 fwrite (line, 1, strlen (line), out2);
541 xfree (line);
542 }
543 }
544 }
545 }
546
547 if (opt.limit_rate)
548 limit_bandwidth (ret, timer);
549
550 if (progress)
551 progress_update (progress, ret, ptimer_read (timer));
552 #ifdef WINDOWS
553 if (toread > 0 && opt.show_progress)
554 ws_percenttitle (100.0 *
555 (startpos + sum_read) / (startpos + toread));
556 #endif
557 }
558 if (ret < -1)
559 ret = -1;
560
561 out:
562 if (progress)
563 progress_finish (progress, ptimer_read (timer));
564
565 if (timer)
566 {
567 if (elapsed)
568 *elapsed = ptimer_read (timer);
569 ptimer_destroy (timer);
570 }
571
572 #ifdef HAVE_LIBZ
573 if (gzbuf != NULL)
574 {
575 int err = inflateEnd (&gzstream);
576 if (ret >= 0)
577 {
578 /* with compression enabled, ret must be 0 if successful */
579 if (err == Z_OK)
580 ret = 0;
581 else
582 {
583 errno = EINVAL;
584 ret = -1;
585 }
586 }
587 xfree (gzbuf);
588
589 if (gzstream.total_in != (uLong) sum_read)
590 {
591 DEBUGP(("zlib read size differs from raw read size (%lu/%"PRId64")\n",
592 gzstream.total_in, sum_read));
593 }
594 }
595 #endif
596
597 if (qtyread)
598 *qtyread += sum_read;
599 if (qtywritten)
600 *qtywritten += sum_written;
601
602 xfree (dlbuf);
603
604 return ret;
605 }
606
607 /* Read a hunk of data from FD, up until a terminator. The hunk is
608 limited by whatever the TERMINATOR callback chooses as its
609 terminator. For example, if terminator stops at newline, the hunk
610 will consist of a line of data; if terminator stops at two
611 newlines, it can be used to read the head of an HTTP response.
612 Upon determining the boundary, the function returns the data (up to
613 the terminator) in malloc-allocated storage.
614
615 In case of read error, NULL is returned. In case of EOF and no
616 data read, NULL is returned and errno set to 0. In case of having
617 read some data, but encountering EOF before seeing the terminator,
618 the data that has been read is returned, but it will (obviously)
619 not contain the terminator.
620
621 The TERMINATOR function is called with three arguments: the
622 beginning of the data read so far, the beginning of the current
623 block of peeked-at data, and the length of the current block.
624 Depending on its needs, the function is free to choose whether to
625 analyze all data or just the newly arrived data. If TERMINATOR
626 returns NULL, it means that the terminator has not been seen.
627 Otherwise it should return a pointer to the charactre immediately
628 following the terminator.
629
630 The idea is to be able to read a line of input, or otherwise a hunk
631 of text, such as the head of an HTTP request, without crossing the
632 boundary, so that the next call to fd_read etc. reads the data
633 after the hunk. To achieve that, this function does the following:
634
635 1. Peek at incoming data.
636
637 2. Determine whether the peeked data, along with the previously
638 read data, includes the terminator.
639
640 2a. If yes, read the data until the end of the terminator, and
641 exit.
642
643 2b. If no, read the peeked data and goto 1.
644
645 The function is careful to assume as little as possible about the
646 implementation of peeking. For example, every peek is followed by
647 a read. If the read returns a different amount of data, the
648 process is retried until all data arrives safely.
649
650 SIZEHINT is the buffer size sufficient to hold all the data in the
651 typical case (it is used as the initial buffer size). MAXSIZE is
652 the maximum amount of memory this function is allowed to allocate,
653 or 0 if no upper limit is to be enforced.
654
655 This function should be used as a building block for other
656 functions -- see fd_read_line as a simple example. */
657
658 char *
fd_read_hunk(int fd,hunk_terminator_t terminator,long sizehint,long maxsize)659 fd_read_hunk (int fd, hunk_terminator_t terminator, long sizehint, long maxsize)
660 {
661 long bufsize = sizehint;
662 char *hunk = xmalloc (bufsize);
663 int tail = 0; /* tail position in HUNK */
664
665 assert (!maxsize || maxsize >= bufsize);
666
667 while (1)
668 {
669 const char *end;
670 int pklen, rdlen, remain;
671
672 /* First, peek at the available data. */
673
674 pklen = fd_peek (fd, hunk + tail, bufsize - 1 - tail, -1);
675 if (pklen < 0)
676 {
677 xfree (hunk);
678 return NULL;
679 }
680 end = terminator (hunk, hunk + tail, pklen);
681 if (end)
682 {
683 /* The data contains the terminator: we'll drain the data up
684 to the end of the terminator. */
685 remain = end - (hunk + tail);
686 assert (remain >= 0);
687 if (remain == 0)
688 {
689 /* No more data needs to be read. */
690 hunk[tail] = '\0';
691 return hunk;
692 }
693 if (bufsize - 1 < tail + remain)
694 {
695 bufsize = tail + remain + 1;
696 hunk = xrealloc (hunk, bufsize);
697 }
698 }
699 else
700 /* No terminator: simply read the data we know is (or should
701 be) available. */
702 remain = pklen;
703
704 /* Now, read the data. Note that we make no assumptions about
705 how much data we'll get. (Some TCP stacks are notorious for
706 read returning less data than the previous MSG_PEEK.) */
707
708 rdlen = fd_read (fd, hunk + tail, remain, 0);
709 if (rdlen < 0)
710 {
711 xfree (hunk);
712 return NULL;
713 }
714 tail += rdlen;
715 hunk[tail] = '\0';
716
717 if (rdlen == 0)
718 {
719 if (tail == 0)
720 {
721 /* EOF without anything having been read */
722 xfree (hunk);
723 errno = 0;
724 return NULL;
725 }
726 else
727 /* EOF seen: return the data we've read. */
728 return hunk;
729 }
730 if (end && rdlen == remain)
731 /* The terminator was seen and the remaining data drained --
732 we got what we came for. */
733 return hunk;
734
735 /* Keep looping until all the data arrives. */
736
737 if (tail == bufsize - 1)
738 {
739 /* Double the buffer size, but refuse to allocate more than
740 MAXSIZE bytes. */
741 if (maxsize && bufsize >= maxsize)
742 {
743 xfree (hunk);
744 errno = ENOMEM;
745 return NULL;
746 }
747 bufsize <<= 1;
748 if (maxsize && bufsize > maxsize)
749 bufsize = maxsize;
750 hunk = xrealloc (hunk, bufsize);
751 }
752 }
753 }
754
755 static const char *
line_terminator(const char * start _GL_UNUSED,const char * peeked,int peeklen)756 line_terminator (const char *start _GL_UNUSED, const char *peeked, int peeklen)
757 {
758 const char *p = memchr (peeked, '\n', peeklen);
759 if (p)
760 /* p+1 because the line must include '\n' */
761 return p + 1;
762 return NULL;
763 }
764
765 /* The maximum size of the single line we agree to accept. This is
766 not meant to impose an arbitrary limit, but to protect the user
767 from Wget slurping up available memory upon encountering malicious
768 or buggy server output. Define it to 0 to remove the limit. */
769 #define FD_READ_LINE_MAX 4096
770
771 /* Read one line from FD and return it. The line is allocated using
772 malloc, but is never larger than FD_READ_LINE_MAX.
773
774 If an error occurs, or if no data can be read, NULL is returned.
775 In the former case errno indicates the error condition, and in the
776 latter case, errno is NULL. */
777
778 char *
fd_read_line(int fd)779 fd_read_line (int fd)
780 {
781 return fd_read_hunk (fd, line_terminator, 128, FD_READ_LINE_MAX);
782 }
783
784 /* Return a printed representation of the download rate, along with
785 the units appropriate for the download speed. */
786
787 const char *
retr_rate(wgint bytes,double secs)788 retr_rate (wgint bytes, double secs)
789 {
790 static char res[20];
791 static const char *rate_names[] = {"B/s", "KB/s", "MB/s", "GB/s" };
792 static const char *rate_names_bits[] = {"b/s", "Kb/s", "Mb/s", "Gb/s" };
793 int units;
794
795 double dlrate = calc_rate (bytes, secs, &units);
796 /* Use more digits for smaller numbers (regardless of unit used),
797 e.g. "1022", "247", "12.5", "2.38". */
798 snprintf (res, sizeof(res), "%.*f %s",
799 dlrate >= 99.95 ? 0 : dlrate >= 9.995 ? 1 : 2,
800 dlrate, !opt.report_bps ? rate_names[units]: rate_names_bits[units]);
801
802 return res;
803 }
804
805 /* Calculate the download rate and trim it as appropriate for the
806 speed. Appropriate means that if rate is greater than 1K/s,
807 kilobytes are used, and if rate is greater than 1MB/s, megabytes
808 are used.
809
810 UNITS is zero for B/s, one for KB/s, two for MB/s, and three for
811 GB/s. */
812
813 double
calc_rate(wgint bytes,double secs,int * units)814 calc_rate (wgint bytes, double secs, int *units)
815 {
816 double dlrate;
817 double bibyte;
818
819 if (!opt.report_bps)
820 bibyte = 1024.0;
821 else
822 bibyte = 1000.0;
823
824 if (secs == 0)
825 /* If elapsed time is exactly zero, it means we're under the
826 resolution of the timer. This can easily happen on systems
827 that use time() for the timer. Since the interval lies between
828 0 and the timer's resolution, assume half the resolution. */
829 secs = ptimer_resolution () / 2.0;
830
831 dlrate = secs ? convert_to_bits (bytes) / secs : 0;
832 if (dlrate < bibyte)
833 *units = 0;
834 else if (dlrate < (bibyte * bibyte))
835 *units = 1, dlrate /= bibyte;
836 else if (dlrate < (bibyte * bibyte * bibyte))
837 *units = 2, dlrate /= (bibyte * bibyte);
838 else if (dlrate < (bibyte * bibyte * bibyte * bibyte))
839 *units = 3, dlrate /= (bibyte * bibyte * bibyte);
840 else {
841 *units = 4, dlrate /= (bibyte * bibyte * bibyte * bibyte);
842 if (dlrate > 99.99)
843 dlrate = 99.99; // upper limit 99.99TB/s
844 }
845
846 return dlrate;
847 }
848
849
850 #define SUSPEND_METHOD do { \
851 method_suspended = true; \
852 saved_body_data = opt.body_data; \
853 saved_body_file_name = opt.body_file; \
854 saved_method = opt.method; \
855 opt.body_data = NULL; \
856 opt.body_file = NULL; \
857 opt.method = NULL; \
858 } while (0)
859
860 #define RESTORE_METHOD do { \
861 if (method_suspended) \
862 { \
863 opt.body_data = saved_body_data; \
864 opt.body_file = saved_body_file_name; \
865 opt.method = saved_method; \
866 method_suspended = false; \
867 } \
868 } while (0)
869
870 static char *getproxy (struct url *);
871
872 /* Retrieve the given URL. Decides which loop to call -- HTTP, FTP,
873 FTP, proxy, etc. */
874
875 /* #### This function should be rewritten so it doesn't return from
876 multiple points. */
877
878 uerr_t
retrieve_url(struct url * orig_parsed,const char * origurl,char ** file,char ** newloc,const char * refurl,int * dt,bool recursive,struct iri * iri,bool register_status)879 retrieve_url (struct url * orig_parsed, const char *origurl, char **file,
880 char **newloc, const char *refurl, int *dt, bool recursive,
881 struct iri *iri, bool register_status)
882 {
883 uerr_t result;
884 char *url;
885 bool location_changed;
886 bool iri_fallbacked = 0;
887 int dummy;
888 char *mynewloc, *proxy;
889 struct url *u = orig_parsed, *proxy_url;
890 int up_error_code; /* url parse error code */
891 char *local_file = NULL;
892 int redirection_count = 0;
893
894 bool method_suspended = false;
895 char *saved_body_data = NULL;
896 char *saved_method = NULL;
897 char *saved_body_file_name = NULL;
898
899 /* If dt is NULL, use local storage. */
900 if (!dt)
901 {
902 dt = &dummy;
903 dummy = 0;
904 }
905 url = xstrdup (origurl);
906 if (newloc)
907 *newloc = NULL;
908 if (file)
909 *file = NULL;
910
911 if (!refurl)
912 refurl = opt.referer;
913
914 redirected:
915 /* (also for IRI fallbacking) */
916
917 result = NOCONERROR;
918 mynewloc = NULL;
919 xfree(local_file);
920 proxy_url = NULL;
921
922 proxy = getproxy (u);
923 if (proxy)
924 {
925 struct iri *pi = iri_new ();
926 set_uri_encoding (pi, opt.locale, true);
927 pi->utf8_encode = false;
928
929 /* Parse the proxy URL. */
930 proxy_url = url_parse (proxy, &up_error_code, pi, true);
931 if (!proxy_url)
932 {
933 char *error = url_error (proxy, up_error_code);
934 logprintf (LOG_NOTQUIET, _("Error parsing proxy URL %s: %s.\n"),
935 proxy, error);
936 xfree (url);
937 xfree (error);
938 xfree (proxy);
939 iri_free (pi);
940 RESTORE_METHOD;
941 result = PROXERR;
942 if (orig_parsed != u)
943 url_free (u);
944 goto bail;
945 }
946 if (proxy_url->scheme != SCHEME_HTTP && proxy_url->scheme != u->scheme)
947 {
948 logprintf (LOG_NOTQUIET, _("Error in proxy URL %s: Must be HTTP.\n"), proxy);
949 url_free (proxy_url);
950 xfree (url);
951 xfree (proxy);
952 iri_free (pi);
953 RESTORE_METHOD;
954 result = PROXERR;
955 if (orig_parsed != u)
956 url_free (u);
957 goto bail;
958 }
959 iri_free(pi);
960 xfree (proxy);
961 }
962
963 if (u->scheme == SCHEME_HTTP
964 #ifdef HAVE_SSL
965 || u->scheme == SCHEME_HTTPS
966 #endif
967 || (proxy_url && proxy_url->scheme == SCHEME_HTTP))
968 {
969 #ifdef HAVE_HSTS
970 #ifdef TESTING
971 /* we don't link against main.o when we're testing */
972 hsts_store_t hsts_store = NULL;
973 #else
974 extern hsts_store_t hsts_store;
975 #endif
976
977 if (opt.hsts && hsts_store)
978 {
979 if (hsts_match (hsts_store, u))
980 logprintf (LOG_VERBOSE, "URL transformed to HTTPS due to an HSTS policy\n");
981 }
982 #endif
983 result = http_loop (u, orig_parsed, &mynewloc, &local_file, refurl, dt,
984 proxy_url, iri);
985 }
986 else if (u->scheme == SCHEME_FTP
987 #ifdef HAVE_SSL
988 || u->scheme == SCHEME_FTPS
989 #endif
990 )
991 {
992 /* If this is a redirection, temporarily turn off opt.ftp_glob
993 and opt.recursive, both being undesirable when following
994 redirects. */
995 bool oldrec = recursive, glob = opt.ftp_glob;
996 if (redirection_count)
997 oldrec = glob = false;
998
999 result = ftp_loop (u, orig_parsed, &local_file, dt, proxy_url,
1000 recursive, glob);
1001 recursive = oldrec;
1002
1003 /* There is a possibility of having HTTP being redirected to
1004 FTP. In these cases we must decide whether the text is HTML
1005 according to the suffix. The HTML suffixes are `.html',
1006 `.htm' and a few others, case-insensitive. */
1007 if (redirection_count && local_file && (u->scheme == SCHEME_FTP
1008 #ifdef HAVE_SSL
1009 || u->scheme == SCHEME_FTPS
1010 #endif
1011 ))
1012 {
1013 if (has_html_suffix_p (local_file))
1014 *dt |= TEXTHTML;
1015 }
1016 }
1017
1018 if (proxy_url)
1019 {
1020 url_free (proxy_url);
1021 proxy_url = NULL;
1022 }
1023
1024 location_changed = (result == NEWLOCATION || result == NEWLOCATION_KEEP_POST);
1025 if (location_changed)
1026 {
1027 char *construced_newloc;
1028 struct url *newloc_parsed;
1029
1030 assert (mynewloc != NULL);
1031
1032 xfree (local_file);
1033
1034 /* The HTTP specs only allow absolute URLs to appear in
1035 redirects, but a ton of boneheaded webservers and CGIs out
1036 there break the rules and use relative URLs, and popular
1037 browsers are lenient about this, so wget should be too. */
1038 construced_newloc = uri_merge (url, mynewloc ? mynewloc : "");
1039 xfree (mynewloc);
1040 mynewloc = construced_newloc;
1041
1042 #ifdef ENABLE_IRI
1043 /* Reset UTF-8 encoding state, set the URI encoding and reset
1044 the content encoding. */
1045 iri->utf8_encode = opt.enable_iri;
1046 if (opt.encoding_remote)
1047 set_uri_encoding (iri, opt.encoding_remote, true);
1048 set_content_encoding (iri, NULL);
1049 xfree (iri->orig_url);
1050 #endif
1051
1052 /* Now, see if this new location makes sense. */
1053 newloc_parsed = url_parse (mynewloc, &up_error_code, iri, true);
1054 if (!newloc_parsed)
1055 {
1056 char *error = url_error (mynewloc, up_error_code);
1057 logprintf (LOG_NOTQUIET, "%s: %s.\n", escnonprint_uri (mynewloc),
1058 error);
1059 if (orig_parsed != u)
1060 {
1061 url_free (u);
1062 }
1063 xfree (url);
1064 xfree (mynewloc);
1065 xfree (error);
1066 RESTORE_METHOD;
1067 goto bail;
1068 }
1069
1070 /* Now mynewloc will become newloc_parsed->url, because if the
1071 Location contained relative paths like .././something, we
1072 don't want that propagating as url. */
1073 xfree (mynewloc);
1074 mynewloc = xstrdup (newloc_parsed->url);
1075
1076 /* Check for max. number of redirections. */
1077 if (++redirection_count > opt.max_redirect)
1078 {
1079 logprintf (LOG_NOTQUIET, _("%d redirections exceeded.\n"),
1080 opt.max_redirect);
1081 url_free (newloc_parsed);
1082 if (orig_parsed != u)
1083 {
1084 url_free (u);
1085 }
1086 xfree (url);
1087 xfree (mynewloc);
1088 RESTORE_METHOD;
1089 result = WRONGCODE;
1090 goto bail;
1091 }
1092
1093 xfree (url);
1094 url = mynewloc;
1095 if (orig_parsed != u)
1096 {
1097 url_free (u);
1098 }
1099 u = newloc_parsed;
1100
1101 /* If we're being redirected from POST, and we received a
1102 redirect code different than 307, we don't want to POST
1103 again. Many requests answer POST with a redirection to an
1104 index page; that redirection is clearly a GET. We "suspend"
1105 POST data for the duration of the redirections, and restore
1106 it when we're done.
1107
1108 RFC2616 HTTP/1.1 introduces code 307 Temporary Redirect
1109 specifically to preserve the method of the request.
1110 */
1111 if (result != NEWLOCATION_KEEP_POST && !method_suspended)
1112 SUSPEND_METHOD;
1113
1114 goto redirected;
1115 }
1116 else
1117 {
1118 xfree(mynewloc);
1119 }
1120
1121 /* Try to not encode in UTF-8 if fetching failed */
1122 if (!(*dt & RETROKF) && iri->utf8_encode)
1123 {
1124 iri->utf8_encode = false;
1125 if (orig_parsed != u)
1126 {
1127 url_free (u);
1128 }
1129 u = url_parse (origurl, NULL, iri, true);
1130 if (u)
1131 {
1132 if (strcmp(u->url, orig_parsed->url))
1133 {
1134 DEBUGP (("[IRI fallbacking to non-utf8 for %s\n", quote (url)));
1135 xfree (url);
1136 url = xstrdup (u->url);
1137 iri_fallbacked = 1;
1138 goto redirected;
1139 }
1140 else
1141 DEBUGP (("[Needn't fallback to non-utf8 for %s\n", quote (url)));
1142 }
1143 else
1144 DEBUGP (("[Couldn't fallback to non-utf8 for %s\n", quote (url)));
1145 }
1146
1147 if (local_file && u && (*dt & RETROKF || opt.content_on_error))
1148 {
1149 register_download (u->url, local_file);
1150
1151 if (!opt.spider && redirection_count && 0 != strcmp (origurl, u->url))
1152 register_redirection (origurl, u->url);
1153
1154 if (*dt & TEXTHTML)
1155 register_html (local_file);
1156
1157 if (*dt & TEXTCSS)
1158 register_css (local_file);
1159 }
1160
1161 if (file)
1162 *file = local_file ? local_file : NULL;
1163 else
1164 xfree (local_file);
1165
1166 if (orig_parsed != u)
1167 url_free (u);
1168
1169 if (redirection_count || iri_fallbacked)
1170 {
1171 if (newloc)
1172 *newloc = url;
1173 else
1174 xfree (url);
1175 }
1176 else
1177 {
1178 if (newloc)
1179 *newloc = NULL;
1180 xfree (url);
1181 }
1182
1183 RESTORE_METHOD;
1184
1185 bail:
1186 if (register_status)
1187 inform_exit_status (result);
1188
1189 return result;
1190 }
1191
1192 /* Find the URLs in the file and call retrieve_url() for each of them.
1193 If HTML is true, treat the file as HTML, and construct the URLs
1194 accordingly.
1195
1196 If opt.recursive is set, call retrieve_tree() for each file. */
1197
1198 uerr_t
retrieve_from_file(const char * file,bool html,int * count)1199 retrieve_from_file (const char *file, bool html, int *count)
1200 {
1201 uerr_t status;
1202 struct urlpos *url_list, *cur_url;
1203 struct iri *iri = iri_new();
1204
1205 char *input_file, *url_file = NULL;
1206 const char *url = file;
1207
1208 status = RETROK; /* Suppose everything is OK. */
1209 *count = 0; /* Reset the URL count. */
1210
1211 /* sXXXav : Assume filename and links in the file are in the locale */
1212 set_uri_encoding (iri, opt.locale, true);
1213 set_content_encoding (iri, opt.locale);
1214
1215 if (url_valid_scheme (url))
1216 {
1217 int dt,url_err;
1218 struct url *url_parsed = url_parse (url, &url_err, iri, true);
1219 if (!url_parsed)
1220 {
1221 char *error = url_error (url, url_err);
1222 logprintf (LOG_NOTQUIET, "%s: %s.\n", url, error);
1223 xfree (error);
1224 iri_free (iri);
1225 return URLERROR;
1226 }
1227
1228 if (!opt.base_href)
1229 opt.base_href = xstrdup (url);
1230
1231 status = retrieve_url (url_parsed, url, &url_file, NULL, NULL, &dt,
1232 false, iri, true);
1233 url_free (url_parsed);
1234
1235 if (!url_file || (status != RETROK))
1236 return status;
1237
1238 if (dt & TEXTHTML)
1239 html = true;
1240
1241 #ifdef ENABLE_IRI
1242 /* If we have a found a content encoding, use it.
1243 * ( == is okay, because we're checking for identical object) */
1244 if (iri->content_encoding != opt.locale)
1245 set_uri_encoding (iri, iri->content_encoding, false);
1246 #endif
1247
1248 /* Reset UTF-8 encode status */
1249 iri->utf8_encode = opt.enable_iri;
1250 xfree (iri->orig_url);
1251
1252 input_file = url_file;
1253 }
1254 else
1255 input_file = (char *) file;
1256
1257 url_list = (html ? get_urls_html (input_file, NULL, NULL, iri)
1258 : get_urls_file (input_file));
1259
1260 xfree (url_file);
1261
1262 for (cur_url = url_list; cur_url; cur_url = cur_url->next, ++*count)
1263 {
1264 char *filename = NULL, *new_file = NULL, *proxy;
1265 int dt = 0;
1266 struct iri *tmpiri = iri_dup (iri);
1267 struct url *parsed_url = NULL;
1268
1269 if (cur_url->ignore_when_downloading)
1270 continue;
1271
1272 if (opt.quota && total_downloaded_bytes > opt.quota)
1273 {
1274 status = QUOTEXC;
1275 break;
1276 }
1277
1278 parsed_url = url_parse (cur_url->url->url, NULL, tmpiri, true);
1279
1280 proxy = getproxy (cur_url->url);
1281 if ((opt.recursive || opt.page_requisites)
1282 && ((cur_url->url->scheme != SCHEME_FTP
1283 #ifdef HAVE_SSL
1284 && cur_url->url->scheme != SCHEME_FTPS
1285 #endif
1286 ) || proxy))
1287 {
1288 int old_follow_ftp = opt.follow_ftp;
1289
1290 /* Turn opt.follow_ftp on in case of recursive FTP retrieval */
1291 if (cur_url->url->scheme == SCHEME_FTP
1292 #ifdef HAVE_SSL
1293 || cur_url->url->scheme == SCHEME_FTPS
1294 #endif
1295 )
1296 opt.follow_ftp = 1;
1297
1298 status = retrieve_tree (parsed_url ? parsed_url : cur_url->url,
1299 tmpiri);
1300
1301 opt.follow_ftp = old_follow_ftp;
1302 }
1303 else
1304 status = retrieve_url (parsed_url ? parsed_url : cur_url->url,
1305 cur_url->url->url, &filename,
1306 &new_file, NULL, &dt, opt.recursive, tmpiri,
1307 true);
1308 xfree (proxy);
1309
1310 if (parsed_url)
1311 url_free (parsed_url);
1312
1313 if (filename && opt.delete_after && file_exists_p (filename, NULL))
1314 {
1315 DEBUGP (("\
1316 Removing file due to --delete-after in retrieve_from_file():\n"));
1317 logprintf (LOG_VERBOSE, _("Removing %s.\n"), filename);
1318 if (unlink (filename))
1319 logprintf (LOG_NOTQUIET, "Failed to unlink %s: (%d) %s\n", filename, errno, strerror (errno));
1320 dt &= ~RETROKF;
1321 }
1322
1323 xfree (new_file);
1324 xfree (filename);
1325 iri_free (tmpiri);
1326 }
1327
1328 /* Free the linked list of URL-s. */
1329 free_urlpos (url_list);
1330
1331 iri_free (iri);
1332
1333 return status;
1334 }
1335
1336 /* Print `giving up', or `retrying', depending on the impending
1337 action. N1 and N2 are the attempt number and the attempt limit. */
1338 void
printwhat(int n1,int n2)1339 printwhat (int n1, int n2)
1340 {
1341 logputs (LOG_VERBOSE, (n1 == n2) ? _("Giving up.\n\n") : _("Retrying.\n\n"));
1342 }
1343
1344 /* If opt.wait or opt.waitretry are specified, and if certain
1345 conditions are met, sleep the appropriate number of seconds. See
1346 the documentation of --wait and --waitretry for more information.
1347
1348 COUNT is the count of current retrieval, beginning with 1. */
1349
1350 void
sleep_between_retrievals(int count)1351 sleep_between_retrievals (int count)
1352 {
1353 static bool first_retrieval = true;
1354
1355 if (first_retrieval)
1356 {
1357 /* Don't sleep before the very first retrieval. */
1358 first_retrieval = false;
1359 return;
1360 }
1361
1362 if (opt.waitretry && count > 1)
1363 {
1364 /* If opt.waitretry is specified and this is a retry, wait for
1365 COUNT-1 number of seconds, or for opt.waitretry seconds. */
1366 if (count <= opt.waitretry)
1367 xsleep (count - 1);
1368 else
1369 xsleep (opt.waitretry);
1370 }
1371 else if (opt.wait)
1372 {
1373 if (!opt.random_wait || count > 1)
1374 /* If random-wait is not specified, or if we are sleeping
1375 between retries of the same download, sleep the fixed
1376 interval. */
1377 xsleep (opt.wait);
1378 else
1379 {
1380 /* Sleep a random amount of time averaging in opt.wait
1381 seconds. The sleeping amount ranges from 0.5*opt.wait to
1382 1.5*opt.wait. */
1383 double waitsecs = (0.5 + random_float ()) * opt.wait;
1384 DEBUGP (("sleep_between_retrievals: avg=%f,sleep=%f\n",
1385 opt.wait, waitsecs));
1386 xsleep (waitsecs);
1387 }
1388 }
1389 }
1390
1391 /* Free the linked list of urlpos. */
1392 void
free_urlpos(struct urlpos * l)1393 free_urlpos (struct urlpos *l)
1394 {
1395 while (l)
1396 {
1397 struct urlpos *next = l->next;
1398 if (l->url)
1399 url_free (l->url);
1400 xfree (l->local_name);
1401 xfree (l);
1402 l = next;
1403 }
1404 }
1405
1406 /* Rotate FNAME opt.backups times */
1407 void
rotate_backups(const char * fname)1408 rotate_backups(const char *fname)
1409 {
1410 #ifdef __VMS
1411 # define SEP "_"
1412 # define AVS ";*" /* All-version suffix. */
1413 # define AVSL (sizeof (AVS) - 1)
1414 #else
1415 # define SEP "."
1416 # define AVSL 0
1417 #endif
1418 #define FILE_BUF_SIZE 1024
1419
1420 /* avoid alloca() here */
1421 char from[FILE_BUF_SIZE], to[FILE_BUF_SIZE];
1422 struct stat sb;
1423 bool overflow;
1424 int i;
1425
1426 if (stat (fname, &sb) == 0)
1427 if (S_ISREG (sb.st_mode) == 0)
1428 return;
1429
1430 for (i = opt.backups; i > 1; i--)
1431 {
1432 #ifdef VMS
1433 /* Delete (all versions of) any existing max-suffix file, to avoid
1434 * creating multiple versions of it. (On VMS, rename() will
1435 * create a new version of an existing destination file, not
1436 * destroy/overwrite it.)
1437 */
1438 if (i == opt.backups)
1439 {
1440 if (((unsigned) snprintf (to, sizeof (to), "%s%s%d%s", fname, SEP, i, AVS)) >= sizeof (to))
1441 logprintf (LOG_NOTQUIET, "Failed to delete %s: File name truncation\n", to);
1442 else
1443 delete (to);
1444 }
1445 #endif
1446 overflow = (unsigned) snprintf (to, FILE_BUF_SIZE, "%s%s%d", fname, SEP, i) >= FILE_BUF_SIZE;
1447 overflow |= (unsigned) snprintf (from, FILE_BUF_SIZE, "%s%s%d", fname, SEP, i - 1) >= FILE_BUF_SIZE;
1448
1449 if (overflow)
1450 errno = ENAMETOOLONG;
1451 if (overflow || rename (from, to))
1452 logprintf (LOG_NOTQUIET, "Failed to rename %s to %s: (%d) %s\n",
1453 from, to, errno, strerror (errno));
1454 }
1455
1456 overflow = (unsigned) snprintf (to, FILE_BUF_SIZE, "%s%s%d", fname, SEP, 1) >= FILE_BUF_SIZE;
1457 if (overflow)
1458 errno = ENAMETOOLONG;
1459 if (overflow || rename(fname, to))
1460 logprintf (LOG_NOTQUIET, "Failed to rename %s to %s: (%d) %s\n",
1461 fname, to, errno, strerror (errno));
1462
1463 #undef FILE_BUF_SIZE
1464 }
1465
1466 static bool no_proxy_match (const char *, const char **);
1467
1468 /* Return the URL of the proxy appropriate for url U. */
1469
1470 static char *
getproxy(struct url * u)1471 getproxy (struct url *u)
1472 {
1473 char *proxy = NULL;
1474 char *rewritten_url;
1475
1476 if (!opt.use_proxy)
1477 return NULL;
1478 if (no_proxy_match (u->host, (const char **)opt.no_proxy))
1479 return NULL;
1480
1481 switch (u->scheme)
1482 {
1483 case SCHEME_HTTP:
1484 proxy = opt.http_proxy ? opt.http_proxy : getenv ("http_proxy");
1485 break;
1486 #ifdef HAVE_SSL
1487 case SCHEME_HTTPS:
1488 proxy = opt.https_proxy ? opt.https_proxy : getenv ("https_proxy");
1489 break;
1490 case SCHEME_FTPS:
1491 proxy = opt.ftp_proxy ? opt.ftp_proxy : getenv ("ftps_proxy");
1492 break;
1493 #endif
1494 case SCHEME_FTP:
1495 proxy = opt.ftp_proxy ? opt.ftp_proxy : getenv ("ftp_proxy");
1496 break;
1497 case SCHEME_INVALID:
1498 break;
1499 }
1500 if (!proxy || !*proxy)
1501 return NULL;
1502
1503 /* Handle shorthands. `rewritten_storage' is a kludge to allow
1504 getproxy() to return static storage. */
1505 rewritten_url = rewrite_shorthand_url (proxy);
1506 if (rewritten_url)
1507 return rewritten_url;
1508
1509 return strdup(proxy);
1510 }
1511
1512 /* Returns true if URL would be downloaded through a proxy. */
1513
1514 bool
url_uses_proxy(struct url * u)1515 url_uses_proxy (struct url * u)
1516 {
1517 bool ret;
1518 char *proxy;
1519
1520 if (!u)
1521 return false;
1522 proxy = getproxy (u);
1523 ret = proxy != NULL;
1524 xfree (proxy);
1525 return ret;
1526 }
1527
1528 /* Should a host be accessed through proxy, concerning no_proxy? */
1529 static bool
no_proxy_match(const char * host,const char ** no_proxy)1530 no_proxy_match (const char *host, const char **no_proxy)
1531 {
1532 if (!no_proxy)
1533 return false;
1534 else
1535 return sufmatch (no_proxy, host);
1536 }
1537
1538 /* Set the file parameter to point to the local file string. */
1539 void
set_local_file(const char ** file,const char * default_file)1540 set_local_file (const char **file, const char *default_file)
1541 {
1542 if (opt.output_document)
1543 {
1544 if (output_stream_regular)
1545 *file = opt.output_document;
1546 }
1547 else
1548 *file = default_file;
1549 }
1550
1551 /* Return true for an input file's own URL, false otherwise. */
1552 bool
input_file_url(const char * input_file)1553 input_file_url (const char *input_file)
1554 {
1555 static bool first = true;
1556
1557 if (input_file
1558 && url_has_scheme (input_file)
1559 && first)
1560 {
1561 first = false;
1562 return true;
1563 }
1564 else
1565 return false;
1566 }
1567