1 /* Handling of recursive HTTP retrieving.
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
10 (at 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. */
38 #endif /* HAVE_UNISTD_H */
55 /* Functions for maintaining the URL queue. */
57 struct queue_element {
58 const char *url; /* the URL to download */
59 const char *referer; /* the referring document */
60 int depth; /* the depth */
61 bool html_allowed; /* whether the document is allowed to
62 be treated as HTML. */
63 struct iri *iri; /* sXXXav */
64 bool css_allowed; /* whether the document is allowed to
66 struct queue_element *next; /* next element in queue */
70 struct queue_element *head;
71 struct queue_element *tail;
75 /* Create a URL queue. */
77 static struct url_queue *
80 struct url_queue *queue = xnew0 (struct url_queue);
84 /* Delete a URL queue. */
87 url_queue_delete (struct url_queue *queue)
92 /* Enqueue a URL in the queue. The queue is FIFO: the items will be
93 retrieved ("dequeued") from the queue in the order they were placed
97 url_enqueue (struct url_queue *queue, struct iri *i,
98 const char *url, const char *referer, int depth,
99 bool html_allowed, bool css_allowed)
101 struct queue_element *qel = xnew (struct queue_element);
104 qel->referer = referer;
106 qel->html_allowed = html_allowed;
107 qel->css_allowed = css_allowed;
111 if (queue->count > queue->maxcount)
112 queue->maxcount = queue->count;
114 DEBUGP (("Enqueuing %s at depth %d\n", url, depth));
115 DEBUGP (("Queue count %d, maxcount %d.\n", queue->count, queue->maxcount));
118 DEBUGP (("[IRI Enqueuing %s with %s\n", quote (url),
119 i->uri_encoding ? quote (i->uri_encoding) : "None"));
122 queue->tail->next = qel;
126 queue->head = queue->tail;
129 /* Take a URL out of the queue. Return true if this operation
130 succeeded, or false if the queue is empty. */
133 url_dequeue (struct url_queue *queue, struct iri **i,
134 const char **url, const char **referer, int *depth,
135 bool *html_allowed, bool *css_allowed)
137 struct queue_element *qel = queue->head;
142 queue->head = queue->head->next;
148 *referer = qel->referer;
150 *html_allowed = qel->html_allowed;
151 *css_allowed = qel->css_allowed;
155 DEBUGP (("Dequeuing %s at depth %d\n", qel->url, qel->depth));
156 DEBUGP (("Queue count %d, maxcount %d.\n", queue->count, queue->maxcount));
162 static bool download_child_p (const struct urlpos *, struct url *, int,
163 struct url *, struct hash_table *, struct iri *);
164 static bool descend_redirect_p (const char *, const char *, int,
165 struct url *, struct hash_table *, struct iri *);
168 /* Retrieve a part of the web beginning with START_URL. This used to
169 be called "recursive retrieval", because the old function was
170 recursive and implemented depth-first search. retrieve_tree on the
171 other hand implements breadth-search traversal of the tree, which
172 results in much nicer ordering of downloads.
174 The algorithm this function uses is simple:
176 1. put START_URL in the queue.
177 2. while there are URLs in the queue:
179 3. get next URL from the queue.
181 5. if the URL is HTML and its depth does not exceed maximum depth,
182 get the list of URLs embedded therein.
183 6. for each of those URLs do the following:
185 7. if the URL is not one of those downloaded before, and if it
186 satisfies the criteria specified by the various command-line
187 options, add it to the queue. */
190 retrieve_tree (const char *start_url, struct iri *pi)
192 uerr_t status = RETROK;
194 /* The queue of URLs we need to load. */
195 struct url_queue *queue;
197 /* The URLs we do not wish to enqueue, because they are already in
198 the queue, but haven't been downloaded yet. */
199 struct hash_table *blacklist;
202 struct url *start_url_parsed;
203 struct iri *i = iri_new ();
205 #define COPYSTR(x) (x) ? xstrdup(x) : NULL;
206 /* Duplicate pi struct if not NULL */
209 i->uri_encoding = COPYSTR (pi->uri_encoding);
210 i->content_encoding = COPYSTR (pi->content_encoding);
211 i->utf8_encode = pi->utf8_encode;
214 set_uri_encoding (i, opt.locale, true);
217 start_url_parsed = url_parse (start_url, &up_error_code, i);
218 if (!start_url_parsed)
220 char *error = url_error (start_url, up_error_code);
221 logprintf (LOG_NOTQUIET, "%s: %s.\n", start_url, error);
226 queue = url_queue_new ();
227 blacklist = make_string_hash_table (0);
229 /* Enqueue the starting URL. Use start_url_parsed->url rather than
230 just URL so we enqueue the canonical form of the URL. */
231 url_enqueue (queue, i, xstrdup (start_url_parsed->url), NULL, 0, true,
233 string_set_add (blacklist, start_url_parsed->url);
237 bool descend = false;
238 char *url, *referer, *file = NULL;
240 bool html_allowed, css_allowed;
242 bool dash_p_leaf_HTML = false;
244 if (opt.quota && total_downloaded_bytes > opt.quota)
246 if (status == FWRITEERR)
249 /* Get the next URL from the queue... */
251 if (!url_dequeue (queue, (struct iri **) &i,
252 (const char **)&url, (const char **)&referer,
253 &depth, &html_allowed, &css_allowed))
256 /* ...and download it. Note that this download is in most cases
257 unconditional, as download_child_p already makes sure a file
258 doesn't get enqueued twice -- and yet this check is here, and
259 not in download_child_p. This is so that if you run `wget -r
260 URL1 URL2', and a random URL is encountered once under URL1
261 and again under URL2, but at a different (possibly smaller)
262 depth, we want the URL's children to be taken into account
264 if (dl_url_file_map && hash_table_contains (dl_url_file_map, url))
266 file = xstrdup (hash_table_get (dl_url_file_map, url));
268 DEBUGP (("Already downloaded \"%s\", reusing it from \"%s\".\n",
271 /* this sucks, needs to be combined! */
273 && downloaded_html_set
274 && string_set_contains (downloaded_html_set, file))
280 && downloaded_css_set
281 && string_set_contains (downloaded_css_set, file))
290 char *redirected = NULL;
292 status = retrieve_url (url, &file, &redirected, referer, &dt,
295 if (html_allowed && file && status == RETROK
296 && (dt & RETROKF) && (dt & TEXTHTML))
302 /* a little different, css_allowed can override content type
303 lots of web servers serve css with an incorrect content type
305 if (file && status == RETROK
307 ((dt & TEXTCSS) || css_allowed))
315 /* We have been redirected, possibly to another host, or
316 different path, or wherever. Check whether we really
317 want to follow it. */
320 if (!descend_redirect_p (redirected, url, depth,
321 start_url_parsed, blacklist, i))
324 /* Make sure that the old pre-redirect form gets
326 string_set_add (blacklist, url);
336 visited_url (url, referer);
340 && depth >= opt.reclevel && opt.reclevel != INFINITE_RECURSION)
342 if (opt.page_requisites
343 && (depth == opt.reclevel || depth == opt.reclevel + 1))
345 /* When -p is specified, we are allowed to exceed the
346 maximum depth, but only for the "inline" links,
347 i.e. those that are needed to display the page.
348 Originally this could exceed the depth at most by
349 one, but we allow one more level so that the leaf
350 pages that contain frames can be loaded
352 dash_p_leaf_HTML = true;
356 /* Either -p wasn't specified or it was and we've
357 already spent the two extra (pseudo-)levels that it
358 affords us, so we need to bail out. */
359 DEBUGP (("Not descending further; at depth %d, max. %d.\n",
360 depth, opt.reclevel));
365 /* If the downloaded document was HTML or CSS, parse it and enqueue the
366 links it contains. */
370 bool meta_disallow_follow = false;
371 struct urlpos *children
372 = is_css ? get_urls_css_file (file, url) :
373 get_urls_html (file, url, &meta_disallow_follow, i);
375 if (opt.use_robots && meta_disallow_follow)
377 free_urlpos (children);
383 struct urlpos *child = children;
384 struct url *url_parsed = url_parse (url, NULL, i);
386 char *referer_url = url;
387 bool strip_auth = (url_parsed != NULL
388 && url_parsed->user != NULL);
389 assert (url_parsed != NULL);
391 /* Strip auth info if present */
393 referer_url = url_string (url_parsed, URL_AUTH_HIDE);
395 for (; child; child = child->next)
397 if (child->ignore_when_downloading)
399 if (dash_p_leaf_HTML && !child->link_inline_p)
401 if (download_child_p (child, url_parsed, depth, start_url_parsed,
405 set_uri_encoding (ci, i->content_encoding, false);
406 url_enqueue (queue, ci, xstrdup (child->url->url),
407 xstrdup (referer_url), depth + 1,
408 child->link_expect_html,
409 child->link_expect_css);
410 /* We blacklist the URL we have enqueued, because we
411 don't want to enqueue (and hence download) the
413 string_set_add (blacklist, child->url->url);
419 url_free (url_parsed);
420 free_urlpos (children);
426 || opt.spider /* opt.recursive is implicitely true */
427 || !acceptable (file)))
429 /* Either --delete-after was specified, or we loaded this
430 (otherwise unneeded because of --spider or rejected by -R)
431 HTML file just to harvest its hyperlinks -- in either case,
432 delete the local file. */
433 DEBUGP (("Removing file due to %s in recursive_retrieve():\n",
434 opt.delete_after ? "--delete-after" :
435 (opt.spider ? "--spider" :
436 "recursive rejection criteria")));
437 logprintf (LOG_VERBOSE,
438 (opt.delete_after || opt.spider
439 ? _("Removing %s.\n")
440 : _("Removing %s since it should be rejected.\n")),
443 logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
444 logputs (LOG_VERBOSE, "\n");
445 register_delete_file (file);
449 xfree_null (referer);
454 /* If anything is left of the queue due to a premature exit, free it
461 while (url_dequeue (queue, (struct iri **)&d6,
462 (const char **)&d1, (const char **)&d2, &d3, &d4, &d5))
469 url_queue_delete (queue);
471 if (start_url_parsed)
472 url_free (start_url_parsed);
473 string_set_free (blacklist);
475 if (opt.quota && total_downloaded_bytes > opt.quota)
477 else if (status == FWRITEERR)
483 /* Based on the context provided by retrieve_tree, decide whether a
484 URL is to be descended to. This is only ever called from
485 retrieve_tree, but is in a separate function for clarity.
487 The most expensive checks (such as those for robots) are memoized
488 by storing these URLs to BLACKLIST. This may or may not help. It
489 will help if those URLs are encountered many times. */
492 download_child_p (const struct urlpos *upos, struct url *parent, int depth,
493 struct url *start_url_parsed, struct hash_table *blacklist,
496 struct url *u = upos->url;
497 const char *url = u->url;
498 bool u_scheme_like_http;
500 DEBUGP (("Deciding whether to enqueue \"%s\".\n", url));
502 if (string_set_contains (blacklist, url))
506 char *referrer = url_string (parent, URL_AUTH_HIDE_PASSWD);
507 DEBUGP (("download_child_p: parent->url is: %s\n", quote (parent->url)));
508 visited_url (url, referrer);
511 DEBUGP (("Already on the black list.\n"));
515 /* Several things to check for:
516 1. if scheme is not http, and we don't load it
517 2. check for relative links (if relative_only is set)
519 4. check for no-parent
520 5. check for excludes && includes
522 7. check for same host (if spanhost is unset), with possible
523 gethostbyname baggage
524 8. check for robots.txt
526 Addendum: If the URL is FTP, and it is to be loaded, only the
527 domain and suffix settings are "stronger".
529 Note that .html files will get loaded regardless of suffix rules
530 (but that is remedied later with unlink) unless the depth equals
533 More time- and memory- consuming tests should be put later on
536 /* Determine whether URL under consideration has a HTTP-like scheme. */
537 u_scheme_like_http = schemes_are_similar_p (u->scheme, SCHEME_HTTP);
539 /* 1. Schemes other than HTTP are normally not recursed into. */
540 if (!u_scheme_like_http && !(u->scheme == SCHEME_FTP && opt.follow_ftp))
542 DEBUGP (("Not following non-HTTP schemes.\n"));
546 /* 2. If it is an absolute link and they are not followed, throw it
548 if (u_scheme_like_http)
549 if (opt.relative_only && !upos->link_relative_p)
551 DEBUGP (("It doesn't really look like a relative link.\n"));
555 /* 3. If its domain is not to be accepted/looked-up, chuck it
557 if (!accept_domain (u))
559 DEBUGP (("The domain was not accepted.\n"));
563 /* 4. Check for parent directory.
565 If we descended to a different host or changed the scheme, ignore
566 opt.no_parent. Also ignore it for documents needed to display
567 the parent page when in -p mode. */
569 && schemes_are_similar_p (u->scheme, start_url_parsed->scheme)
570 && 0 == strcasecmp (u->host, start_url_parsed->host)
571 && u->port == start_url_parsed->port
572 && !(opt.page_requisites && upos->link_inline_p))
574 if (!subdir_p (start_url_parsed->dir, u->dir))
576 DEBUGP (("Going to \"%s\" would escape \"%s\" with no_parent on.\n",
577 u->dir, start_url_parsed->dir));
582 /* 5. If the file does not match the acceptance list, or is on the
583 rejection list, chuck it out. The same goes for the directory
584 exclusion and inclusion lists. */
585 if (opt.includes || opt.excludes)
587 if (!accdir (u->dir))
589 DEBUGP (("%s (%s) is excluded/not-included.\n", url, u->dir));
594 /* 6. Check for acceptance/rejection rules. We ignore these rules
595 for directories (no file name to match) and for non-leaf HTMLs,
596 which can lead to other files that do need to be downloaded. (-p
597 automatically implies non-leaf because with -p we can, if
598 necesary, overstep the maximum depth to get the page requisites.) */
599 if (u->file[0] != '\0'
600 && !(has_html_suffix_p (u->file)
601 /* The exception only applies to non-leaf HTMLs (but -p
602 always implies non-leaf because we can overstep the
603 maximum depth to get the requisites): */
605 opt.reclevel == INFINITE_RECURSION
607 || depth < opt.reclevel - 1
608 /* -p, which implies non-leaf (see above) */
609 || opt.page_requisites)))
611 if (!acceptable (u->file))
613 DEBUGP (("%s (%s) does not match acc/rej rules.\n",
620 if (schemes_are_similar_p (u->scheme, parent->scheme))
621 if (!opt.spanhost && 0 != strcasecmp (parent->host, u->host))
623 DEBUGP (("This is not the same hostname as the parent's (%s and %s).\n",
624 u->host, parent->host));
629 if (opt.use_robots && u_scheme_like_http)
631 struct robot_specs *specs = res_get_specs (u->host, u->port);
635 if (res_retrieve_file (url, &rfile, iri))
637 specs = res_parse_from_file (rfile);
639 /* Delete the robots.txt file if we chose to either delete the
640 files after downloading or we're just running a spider. */
641 if (opt.delete_after || opt.spider)
643 logprintf (LOG_VERBOSE, "Removing %s.\n", rfile);
645 logprintf (LOG_NOTQUIET, "unlink: %s\n",
653 /* If we cannot get real specs, at least produce
654 dummy ones so that we can register them and stop
655 trying to retrieve them. */
656 specs = res_parse ("", 0);
658 res_register_specs (u->host, u->port, specs);
661 /* Now that we have (or don't have) robots.txt specs, we can
662 check what they say. */
663 if (!res_match_path (specs, u->path))
665 DEBUGP (("Not following %s because robots.txt forbids it.\n", url));
666 string_set_add (blacklist, url);
671 /* The URL has passed all the tests. It can be placed in the
673 DEBUGP (("Decided to load it.\n"));
678 DEBUGP (("Decided NOT to load it.\n"));
683 /* This function determines whether we will consider downloading the
684 children of a URL whose download resulted in a redirection,
685 possibly to another host, etc. It is needed very rarely, and thus
686 it is merely a simple-minded wrapper around download_child_p. */
689 descend_redirect_p (const char *redirected, const char *original, int depth,
690 struct url *start_url_parsed, struct hash_table *blacklist,
693 struct url *orig_parsed, *new_parsed;
697 orig_parsed = url_parse (original, NULL, NULL);
698 assert (orig_parsed != NULL);
700 new_parsed = url_parse (redirected, NULL, NULL);
701 assert (new_parsed != NULL);
703 upos = xnew0 (struct urlpos);
704 upos->url = new_parsed;
706 success = download_child_p (upos, orig_parsed, depth,
707 start_url_parsed, blacklist, iri);
709 url_free (orig_parsed);
710 url_free (new_parsed);
714 DEBUGP (("Redirection \"%s\" failed the test.\n", redirected));
719 /* vim:set sts=2 sw=2 cino+={s: */