/* Handling of recursive HTTP retrieving.
- Copyright (C) 1995, 1996, 1997, 2000, 2001 Free Software Foundation, Inc.
+ Copyright (C) 1996-2006 Free Software Foundation, Inc.
This file is part of GNU Wget.
GNU Wget is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
-the Free Software Foundation; either version 2 of the License, or
-(at your option) any later version.
+the Free Software Foundation; either version 3 of the License, or
+ (at your option) any later version.
GNU Wget is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
-along with Wget; if not, write to the Free Software
-Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
+along with Wget. If not, see <http://www.gnu.org/licenses/>.
In addition, as a special exception, the Free Software Foundation
gives permission to link the code of its release of Wget with the
#include <stdio.h>
#include <stdlib.h>
-#ifdef HAVE_STRING_H
-# include <string.h>
-#else
-# include <strings.h>
-#endif /* HAVE_STRING_H */
+#include <string.h>
#ifdef HAVE_UNISTD_H
# include <unistd.h>
#endif /* HAVE_UNISTD_H */
#include <errno.h>
#include <assert.h>
-#include <sys/types.h>
#include "wget.h"
#include "url.h"
#include "utils.h"
#include "retr.h"
#include "ftp.h"
-#include "fnmatch.h"
#include "host.h"
#include "hash.h"
#include "res.h"
-
-#ifndef errno
-extern int errno;
-#endif
-
-extern char *version_string;
-
-static struct hash_table *dl_file_url_map;
-static struct hash_table *dl_url_file_map;
-
-/* List of HTML files downloaded in this Wget run, used for link
- conversion after Wget is done. The list and the set contain the
- same information, except the list maintains the order. Perhaps I
- should get rid of the list, it's there for historical reasons. */
-static slist *downloaded_html_list;
-static struct hash_table *downloaded_html_set;
-
-static void register_delete_file PARAMS ((const char *));
+#include "convert.h"
+#include "spider.h"
\f
/* Functions for maintaining the URL queue. */
struct queue_element {
- const char *url;
- const char *referer;
- int depth;
- struct queue_element *next;
+ const char *url; /* the URL to download */
+ const char *referer; /* the referring document */
+ int depth; /* the depth */
+ bool html_allowed; /* whether the document is allowed to
+ be treated as HTML. */
+
+ struct queue_element *next; /* next element in queue */
};
struct url_queue {
static struct url_queue *
url_queue_new (void)
{
- struct url_queue *queue = xmalloc (sizeof (*queue));
- memset (queue, '\0', sizeof (*queue));
+ struct url_queue *queue = xnew0 (struct url_queue);
return queue;
}
static void
url_enqueue (struct url_queue *queue,
- const char *url, const char *referer, int depth)
+ const char *url, const char *referer, int depth, bool html_allowed)
{
- struct queue_element *qel = xmalloc (sizeof (*qel));
+ struct queue_element *qel = xnew (struct queue_element);
qel->url = url;
qel->referer = referer;
qel->depth = depth;
+ qel->html_allowed = html_allowed;
qel->next = NULL;
++queue->count;
queue->head = queue->tail;
}
-/* Take a URL out of the queue. Return 1 if this operation succeeded,
- or 0 if the queue is empty. */
+/* Take a URL out of the queue. Return true if this operation
+ succeeded, or false if the queue is empty. */
-static int
+static bool
url_dequeue (struct url_queue *queue,
- const char **url, const char **referer, int *depth)
+ const char **url, const char **referer, int *depth,
+ bool *html_allowed)
{
struct queue_element *qel = queue->head;
if (!qel)
- return 0;
+ return false;
queue->head = queue->head->next;
if (!queue->head)
*url = qel->url;
*referer = qel->referer;
*depth = qel->depth;
+ *html_allowed = qel->html_allowed;
--queue->count;
DEBUGP (("Queue count %d, maxcount %d.\n", queue->count, queue->maxcount));
xfree (qel);
- return 1;
+ return true;
}
\f
-static int download_child_p PARAMS ((const struct urlpos *, struct url *, int,
- struct url *, struct hash_table *));
-static int descend_redirect_p PARAMS ((const char *, const char *, int,
- struct url *, struct hash_table *));
+static bool download_child_p (const struct urlpos *, struct url *, int,
+ struct url *, struct hash_table *);
+static bool descend_redirect_p (const char *, const char *, int,
+ struct url *, struct hash_table *);
/* Retrieve a part of the web beginning with START_URL. This used to
/* Enqueue the starting URL. Use start_url_parsed->url rather than
just URL so we enqueue the canonical form of the URL. */
- url_enqueue (queue, xstrdup (start_url_parsed->url), NULL, 0);
+ url_enqueue (queue, xstrdup (start_url_parsed->url), NULL, 0, true);
string_set_add (blacklist, start_url_parsed->url);
while (1)
{
- int descend = 0;
+ bool descend = false;
char *url, *referer, *file = NULL;
int depth;
- boolean dash_p_leaf_HTML = FALSE;
+ bool html_allowed;
+ bool dash_p_leaf_HTML = false;
- if (downloaded_exceeds_quota ())
+ if (opt.quota && total_downloaded_bytes > opt.quota)
break;
if (status == FWRITEERR)
break;
if (!url_dequeue (queue,
(const char **)&url, (const char **)&referer,
- &depth))
+ &depth, &html_allowed))
break;
/* ...and download it. Note that this download is in most cases
DEBUGP (("Already downloaded \"%s\", reusing it from \"%s\".\n",
url, file));
- if (downloaded_html_set
+ if (html_allowed
+ && downloaded_html_set
&& string_set_contains (downloaded_html_set, file))
- descend = 1;
+ descend = true;
}
else
{
int dt = 0;
char *redirected = NULL;
- int oldrec = opt.recursive;
- opt.recursive = 0;
- status = retrieve_url (url, &file, &redirected, referer, &dt);
- opt.recursive = oldrec;
+ status = retrieve_url (url, &file, &redirected, referer, &dt, false);
- if (file && status == RETROK
+ if (html_allowed && file && status == RETROK
&& (dt & RETROKF) && (dt & TEXTHTML))
- descend = 1;
+ descend = true;
if (redirected)
{
{
if (!descend_redirect_p (redirected, url, depth,
start_url_parsed, blacklist))
- descend = 0;
+ descend = false;
else
/* Make sure that the old pre-redirect form gets
blacklisted. */
}
}
+ if (opt.spider)
+ {
+ visited_url (url, referer);
+ }
+
if (descend
&& depth >= opt.reclevel && opt.reclevel != INFINITE_RECURSION)
{
one, but we allow one more level so that the leaf
pages that contain frames can be loaded
correctly. */
- dash_p_leaf_HTML = TRUE;
+ dash_p_leaf_HTML = true;
}
else
{
affords us, so we need to bail out. */
DEBUGP (("Not descending further; at depth %d, max. %d.\n",
depth, opt.reclevel));
- descend = 0;
+ descend = false;
}
}
if (descend)
{
- int meta_disallow_follow = 0;
+ bool meta_disallow_follow = false;
struct urlpos *children
= get_urls_html (file, url, &meta_disallow_follow);
blacklist))
{
url_enqueue (queue, xstrdup (child->url->url),
- xstrdup (url), depth + 1);
+ xstrdup (url), depth + 1,
+ child->link_expect_html);
/* We blacklist the URL we have enqueued, because we
don't want to enqueue (and hence download) the
same URL twice. */
}
}
- if (opt.delete_after || (file && !acceptable (file)))
+ if (file
+ && (opt.delete_after
+ || opt.spider /* opt.recursive is implicitely true */
+ || !acceptable (file)))
{
/* Either --delete-after was specified, or we loaded this
- otherwise rejected (e.g. by -R) HTML file just so we
- could harvest its hyperlinks -- in either case, delete
- the local file. */
+ (otherwise unneeded because of --spider or rejected by -R)
+ HTML file just to harvest its hyperlinks -- in either case,
+ delete the local file. */
DEBUGP (("Removing file due to %s in recursive_retrieve():\n",
opt.delete_after ? "--delete-after" :
- "recursive rejection criteria"));
+ (opt.spider ? "--spider" :
+ "recursive rejection criteria")));
logprintf (LOG_VERBOSE,
- (opt.delete_after
+ (opt.delete_after || opt.spider
? _("Removing %s.\n")
: _("Removing %s since it should be rejected.\n")),
file);
if (unlink (file))
logprintf (LOG_NOTQUIET, "unlink: %s\n", strerror (errno));
+ logputs (LOG_VERBOSE, "\n");
register_delete_file (file);
}
xfree (url);
- FREE_MAYBE (referer);
- FREE_MAYBE (file);
+ xfree_null (referer);
+ xfree_null (file);
}
/* If anything is left of the queue due to a premature exit, free it
{
char *d1, *d2;
int d3;
- while (url_dequeue (queue, (const char **)&d1, (const char **)&d2, &d3))
+ bool d4;
+ while (url_dequeue (queue,
+ (const char **)&d1, (const char **)&d2, &d3, &d4))
{
xfree (d1);
- FREE_MAYBE (d2);
+ xfree_null (d2);
}
}
url_queue_delete (queue);
url_free (start_url_parsed);
string_set_free (blacklist);
- if (downloaded_exceeds_quota ())
+ if (opt.quota && total_downloaded_bytes > opt.quota)
return QUOTEXC;
else if (status == FWRITEERR)
return FWRITEERR;
by storing these URLs to BLACKLIST. This may or may not help. It
will help if those URLs are encountered many times. */
-static int
+static bool
download_child_p (const struct urlpos *upos, struct url *parent, int depth,
struct url *start_url_parsed, struct hash_table *blacklist)
{
struct url *u = upos->url;
const char *url = u->url;
- int u_scheme_like_http;
+ bool u_scheme_like_http;
DEBUGP (("Deciding whether to enqueue \"%s\".\n", url));
if (string_set_contains (blacklist, url))
{
+ if (opt.spider)
+ {
+ char *referrer = url_string (parent, true);
+ DEBUGP (("download_child_p: parent->url is: `%s'\n", parent->url));
+ visited_url (url, referrer);
+ xfree (referrer);
+ }
DEBUGP (("Already on the black list.\n"));
goto out;
}
&& u->port == start_url_parsed->port
&& !(opt.page_requisites && upos->link_inline_p))
{
- if (!frontcmp (start_url_parsed->dir, u->dir))
+ if (!subdir_p (start_url_parsed->dir, u->dir))
{
DEBUGP (("Going to \"%s\" would escape \"%s\" with no_parent on.\n",
u->dir, start_url_parsed->dir));
exclusion and inclusion lists. */
if (opt.includes || opt.excludes)
{
- if (!accdir (u->dir, ALLABS))
+ if (!accdir (u->dir))
{
DEBUGP (("%s (%s) is excluded/not-included.\n", url, u->dir));
goto out;
}
/* 6. Check for acceptance/rejection rules. We ignore these rules
- for directories (no file name to match) and for HTML documents,
- which might lead to other files that do need to be downloaded.
- That is, unless we've exhausted the recursion depth anyway. */
+ for directories (no file name to match) and for non-leaf HTMLs,
+ which can lead to other files that do need to be downloaded. (-p
+ automatically implies non-leaf because with -p we can, if
+ necesary, overstep the maximum depth to get the page requisites.) */
if (u->file[0] != '\0'
&& !(has_html_suffix_p (u->file)
- && depth != INFINITE_RECURSION
- && depth < opt.reclevel - 1))
+ /* The exception only applies to non-leaf HTMLs (but -p
+ always implies non-leaf because we can overstep the
+ maximum depth to get the requisites): */
+ && (/* non-leaf */
+ opt.reclevel == INFINITE_RECURSION
+ /* also non-leaf */
+ || depth < opt.reclevel - 1
+ /* -p, which implies non-leaf (see above) */
+ || opt.page_requisites)))
{
if (!acceptable (u->file))
{
download queue. */
DEBUGP (("Decided to load it.\n"));
- return 1;
+ return true;
out:
DEBUGP (("Decided NOT to load it.\n"));
- return 0;
+ return false;
}
/* This function determines whether we will consider downloading the
possibly to another host, etc. It is needed very rarely, and thus
it is merely a simple-minded wrapper around download_child_p. */
-static int
+static bool
descend_redirect_p (const char *redirected, const char *original, int depth,
struct url *start_url_parsed, struct hash_table *blacklist)
{
struct url *orig_parsed, *new_parsed;
struct urlpos *upos;
- int success;
+ bool success;
orig_parsed = url_parse (original, NULL);
assert (orig_parsed != NULL);
new_parsed = url_parse (redirected, NULL);
assert (new_parsed != NULL);
- upos = xmalloc (sizeof (struct urlpos));
- memset (upos, 0, sizeof (*upos));
+ upos = xnew0 (struct urlpos);
upos->url = new_parsed;
success = download_child_p (upos, orig_parsed, depth,
return success;
}
-
-\f
-#define ENSURE_TABLES_EXIST do { \
- if (!dl_file_url_map) \
- dl_file_url_map = make_string_hash_table (0); \
- if (!dl_url_file_map) \
- dl_url_file_map = make_string_hash_table (0); \
-} while (0)
-
-/* Return 1 if S1 and S2 are the same, except for "/index.html". The
- three cases in which it returns one are (substitute any substring
- for "foo"):
-
- m("foo/index.html", "foo/") ==> 1
- m("foo/", "foo/index.html") ==> 1
- m("foo", "foo/index.html") ==> 1
- m("foo", "foo/" ==> 1
- m("foo", "foo") ==> 1 */
-
-static int
-match_except_index (const char *s1, const char *s2)
-{
- int i;
- const char *lng;
-
- /* Skip common substring. */
- for (i = 0; *s1 && *s2 && *s1 == *s2; s1++, s2++, i++)
- ;
- if (i == 0)
- /* Strings differ at the very beginning -- bail out. We need to
- check this explicitly to avoid `lng - 1' reading outside the
- array. */
- return 0;
-
- if (!*s1 && !*s2)
- /* Both strings hit EOF -- strings are equal. */
- return 1;
- else if (*s1 && *s2)
- /* Strings are randomly different, e.g. "/foo/bar" and "/foo/qux". */
- return 0;
- else if (*s1)
- /* S1 is the longer one. */
- lng = s1;
- else
- /* S2 is the longer one. */
- lng = s2;
-
- /* foo */ /* foo/ */
- /* foo/index.html */ /* or */ /* foo/index.html */
- /* ^ */ /* ^ */
-
- if (*lng != '/')
- /* The right-hand case. */
- --lng;
-
- if (*lng == '/' && *(lng + 1) == '\0')
- /* foo */
- /* foo/ */
- return 1;
-
- return 0 == strcmp (lng, "/index.html");
-}
-
-static int
-dissociate_urls_from_file_mapper (void *key, void *value, void *arg)
-{
- char *mapping_url = (char *)key;
- char *mapping_file = (char *)value;
- char *file = (char *)arg;
-
- if (0 == strcmp (mapping_file, file))
- {
- hash_table_remove (dl_url_file_map, mapping_url);
- xfree (mapping_url);
- xfree (mapping_file);
- }
-
- /* Continue mapping. */
- return 0;
-}
-
-/* Remove all associations from various URLs to FILE from dl_url_file_map. */
-
-static void
-dissociate_urls_from_file (const char *file)
-{
- hash_table_map (dl_url_file_map, dissociate_urls_from_file_mapper,
- (char *)file);
-}
-
-/* Register that URL has been successfully downloaded to FILE. This
- is used by the link conversion code to convert references to URLs
- to references to local files. It is also being used to check if a
- URL has already been downloaded. */
-
-void
-register_download (const char *url, const char *file)
-{
- char *old_file, *old_url;
-
- ENSURE_TABLES_EXIST;
-
- /* With some forms of retrieval, it is possible, although not likely
- or particularly desirable. If both are downloaded, the second
- download will override the first one. When that happens,
- dissociate the old file name from the URL. */
-
- if (hash_table_get_pair (dl_file_url_map, file, &old_file, &old_url))
- {
- if (0 == strcmp (url, old_url))
- /* We have somehow managed to download the same URL twice.
- Nothing to do. */
- return;
-
- if (match_except_index (url, old_url)
- && !hash_table_contains (dl_url_file_map, url))
- /* The two URLs differ only in the "index.html" ending. For
- example, one is "http://www.server.com/", and the other is
- "http://www.server.com/index.html". Don't remove the old
- one, just add the new one as a non-canonical entry. */
- goto url_only;
-
- hash_table_remove (dl_file_url_map, file);
- xfree (old_file);
- xfree (old_url);
-
- /* Remove all the URLs that point to this file. Yes, there can
- be more than one such URL, because we store redirections as
- multiple entries in dl_url_file_map. For example, if URL1
- redirects to URL2 which gets downloaded to FILE, we map both
- URL1 and URL2 to FILE in dl_url_file_map. (dl_file_url_map
- only points to URL2.) When another URL gets loaded to FILE,
- we want both URL1 and URL2 dissociated from it.
-
- This is a relatively expensive operation because it performs
- a linear search of the whole hash table, but it should be
- called very rarely, only when two URLs resolve to the same
- file name, *and* the "<file>.1" extensions are turned off.
- In other words, almost never. */
- dissociate_urls_from_file (file);
- }
-
- hash_table_put (dl_file_url_map, xstrdup (file), xstrdup (url));
-
- url_only:
- /* A URL->FILE mapping is not possible without a FILE->URL mapping.
- If the latter were present, it should have been removed by the
- above `if'. So we could write:
-
- assert (!hash_table_contains (dl_url_file_map, url));
-
- The above is correct when running in recursive mode where the
- same URL always resolves to the same file. But if you do
- something like:
-
- wget URL URL
-
- then the first URL will resolve to "FILE", and the other to
- "FILE.1". In that case, FILE.1 will not be found in
- dl_file_url_map, but URL will still point to FILE in
- dl_url_file_map. */
- if (hash_table_get_pair (dl_url_file_map, url, &old_url, &old_file))
- {
- hash_table_remove (dl_url_file_map, url);
- xfree (old_url);
- xfree (old_file);
- }
-
- hash_table_put (dl_url_file_map, xstrdup (url), xstrdup (file));
-}
-
-/* Register that FROM has been redirected to TO. This assumes that TO
- is successfully downloaded and already registered using
- register_download() above. */
-
-void
-register_redirection (const char *from, const char *to)
-{
- char *file;
-
- ENSURE_TABLES_EXIST;
-
- file = hash_table_get (dl_url_file_map, to);
- assert (file != NULL);
- if (!hash_table_contains (dl_url_file_map, from))
- hash_table_put (dl_url_file_map, xstrdup (from), xstrdup (file));
-}
-
-/* Register that the file has been deleted. */
-
-static void
-register_delete_file (const char *file)
-{
- char *old_url, *old_file;
-
- ENSURE_TABLES_EXIST;
-
- if (!hash_table_get_pair (dl_file_url_map, file, &old_file, &old_url))
- return;
-
- hash_table_remove (dl_file_url_map, file);
- xfree (old_file);
- xfree (old_url);
- dissociate_urls_from_file (file);
-}
-
-/* Register that FILE is an HTML file that has been downloaded. */
-
-void
-register_html (const char *url, const char *file)
-{
- if (!downloaded_html_set)
- downloaded_html_set = make_string_hash_table (0);
- else if (hash_table_contains (downloaded_html_set, file))
- return;
-
- /* The set and the list should use the same copy of FILE, but the
- slist interface insists on strduping the string it gets. Oh
- well. */
- string_set_add (downloaded_html_set, file);
- downloaded_html_list = slist_prepend (downloaded_html_list, file);
-}
-
-/* This function is called when the retrieval is done to convert the
- links that have been downloaded. It has to be called at the end of
- the retrieval, because only then does Wget know conclusively which
- URLs have been downloaded, and which not, so it can tell which
- direction to convert to.
-
- The "direction" means that the URLs to the files that have been
- downloaded get converted to the relative URL which will point to
- that file. And the other URLs get converted to the remote URL on
- the server.
-
- All the downloaded HTMLs are kept in downloaded_html_files, and
- downloaded URLs in urls_downloaded. All the information is
- extracted from these two lists. */
-
-void
-convert_all_links (void)
-{
- slist *html;
- long msecs;
- int file_count = 0;
-
- struct wget_timer *timer = wtimer_new ();
-
- /* Destructively reverse downloaded_html_files to get it in the right order.
- recursive_retrieve() used slist_prepend() consistently. */
- downloaded_html_list = slist_nreverse (downloaded_html_list);
-
- for (html = downloaded_html_list; html; html = html->next)
- {
- struct urlpos *urls, *cur_url;
- char *url;
- char *file = html->string;
-
- /* Determine the URL of the HTML file. get_urls_html will need
- it. */
- url = hash_table_get (dl_file_url_map, file);
- if (!url)
- {
- DEBUGP (("Apparently %s has been removed.\n", file));
- continue;
- }
-
- DEBUGP (("Scanning %s (from %s)\n", file, url));
-
- /* Parse the HTML file... */
- urls = get_urls_html (file, url, NULL);
-
- /* We don't respect meta_disallow_follow here because, even if
- the file is not followed, we might still want to convert the
- links that have been followed from other files. */
-
- for (cur_url = urls; cur_url; cur_url = cur_url->next)
- {
- char *local_name;
- struct url *u = cur_url->url;
-
- if (cur_url->link_base_p)
- {
- /* Base references have been resolved by our parser, so
- we turn the base URL into an empty string. (Perhaps
- we should remove the tag entirely?) */
- cur_url->convert = CO_NULLIFY_BASE;
- continue;
- }
-
- /* We decide the direction of conversion according to whether
- a URL was downloaded. Downloaded URLs will be converted
- ABS2REL, whereas non-downloaded will be converted REL2ABS. */
- local_name = hash_table_get (dl_url_file_map, u->url);
-
- /* Decide on the conversion type. */
- if (local_name)
- {
- /* We've downloaded this URL. Convert it to relative
- form. We do this even if the URL already is in
- relative form, because our directory structure may
- not be identical to that on the server (think `-nd',
- `--cut-dirs', etc.) */
- cur_url->convert = CO_CONVERT_TO_RELATIVE;
- cur_url->local_name = xstrdup (local_name);
- DEBUGP (("will convert url %s to local %s\n", u->url, local_name));
- }
- else
- {
- /* We haven't downloaded this URL. If it's not already
- complete (including a full host name), convert it to
- that form, so it can be reached while browsing this
- HTML locally. */
- if (!cur_url->link_complete_p)
- cur_url->convert = CO_CONVERT_TO_COMPLETE;
- cur_url->local_name = NULL;
- DEBUGP (("will convert url %s to complete\n", u->url));
- }
- }
-
- /* Convert the links in the file. */
- convert_links (file, urls);
- ++file_count;
-
- /* Free the data. */
- free_urlpos (urls);
- }
-
- msecs = wtimer_elapsed (timer);
- wtimer_delete (timer);
- logprintf (LOG_VERBOSE, _("Converted %d files in %.2f seconds.\n"),
- file_count, (double)msecs / 1000);
-}
-
-/* Cleanup the data structures associated with recursive retrieving
- (the variables above). */
-void
-recursive_cleanup (void)
-{
- if (dl_file_url_map)
- {
- free_keys_and_values (dl_file_url_map);
- hash_table_destroy (dl_file_url_map);
- dl_file_url_map = NULL;
- }
- if (dl_url_file_map)
- {
- free_keys_and_values (dl_url_file_map);
- hash_table_destroy (dl_url_file_map);
- dl_url_file_map = NULL;
- }
- if (downloaded_html_set)
- string_set_free (downloaded_html_set);
- slist_free (downloaded_html_list);
- downloaded_html_list = NULL;
-}
projects / wget / blobdiff